Electron microscope study of blood–brain barrier opening induced by immunological targeting of the endothelial barrier antigen

Electron microscope study of blood–brain barrier opening induced by immunological targeting of the endothelial barrier antigen

Brain Research 934 (2002) 140–151 www.elsevier.com / locate / bres Research report Electron microscope study of blood–brain barrier opening induced ...

3MB Sizes 4 Downloads 54 Views

Brain Research 934 (2002) 140–151 www.elsevier.com / locate / bres

Research report

Electron microscope study of blood–brain barrier opening induced by immunological targeting of the endothelial barrier antigen 1

Mounir N. Ghabriel*, Chunni Zhu , Chris Leigh Department of Anatomical Sciences, Medical School, Adelaide University, Adelaide, South Australia 5005, Australia Accepted 20 December 2001

Abstract Barrier vessels in the central nervous system are lined with endothelial cells which constitute the blood–brain barrier (BBB) and show selective expression of certain biochemical markers. One of these, the endothelial barrier antigen (EBA), is specific to the rat. The exact role of EBA in the BBB is not known, although several studies have shown a correlation between the reduction in EBA expression in endothelial cells and the opening of the BBB. However, in these studies it was not possible to determine if EBA reduction was a primary event or was secondary to opening of the BBB. A recent light microscope study demonstrated that immunological targeting of EBA in vivo, by intravenous injection of a monoclonal antibody (anti-EBA), leads to acute and widespread opening of the BBB. In the current study we have employed this model together with tracer application and immunoperoxidase electron microscopy to determine the site of binding of the injected antibody and the route of opening of the BBB. The results showed that (a) the anti-EBA injected in vivo became bound to brain endothelial cells, principally to luminal membranes. (b) Endothelial cells showed widened intercellular junctions and increased cytoplasmic vesicles and vacuoles. (c) Many perivascular astrocytic processes were swollen. (d) The macromolecular tracer HRP was present in vesicles, vacuoles, widened paracellular clefts, the perivascular space and brain parenchyma. In conclusion, the in vivo targeting of EBA leads to opening of the BBB apparently via paracellular and transcellular routes. This model is useful for the study of vascular permeability in the CNS and experimental manipulation of the BBB. It may have a potential application in experimental studies on drug delivery throughout the CNS.  2002 Elsevier Science B.V. All rights reserved. Theme: Cellular and molecular biology Topic: Blood–brain barrier Keywords: Blood–brain barrier; Endothelial cell; Endothelial barrier antigen; Immunoelectron microscopy; Horseradish peroxidase

1. Introduction The endothelial barrier antigen (EBA) is a protein expressed by endothelial cells of the rat blood–brain barrier (BBB) [33]. EBA is detectable by immunostaining of tissue sections using a monoclonal antibody (anti-EBA) [33]. At sites known to lack BBB such as the area postrema, pineal gland, median eminence and choroid plexus the antigen is absent or weakly expressed in some endothelial cells [31,33]. Body organs, which have no blood tissue barrier, such as the intestine, kidney, liver, *Corresponding author. Tel.: 161-8-8303-5481; fax: 161-8-83034398. E-mail address: [email protected] (M.N. Ghabriel). 1 Present address: Department of Medicine, UCLA School of Medicine, LA, CA 90024 USA.

thyroid and pancreas, do not express the antigen [33]. Thus EBA appears to be specific to endothelial cells of barrier vessels although its exact function is not known. In pathological conditions such as experimental allergic encephalomyelitis [34], stab-wound injury of the brain [31], spinal cord trauma [27] and Clostridium perfringens type D epsilon prototoxin administration [36] a reduction in the expression of EBA was reported. The acute phases of the above pathological conditions are accompanied by opening of the BBB and the development of central nervous system (CNS) oedema. Also leakage of endogenous albumin in the brain, associated with neuronal and astroglial damage, was accompanied by a reduction in EBA expression in endothelial cells [12]. During recovery from the above experimental conditions, restoration of barrier properties of CNS vessels and clearance of the oedema are paralleled by re-expression of EBA. Thus, studies on CNS pathological

0006-8993 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 02 )02416-2

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

conditions have advocated a putative role for EBA in barrier vessels. However, it was not known from the above studies, if the reduction of EBA expression in endothelial cells is a cause or a consequence of opening of the barrier and the development of oedema. EBA has been shown by immunoelectron microscopy to be located at the luminal membrane of endothelial cells [13,14,31,34,36]. The luminal membrane is the primary blood–brain interface and is accessible to factors circulating in the blood. In a recent light microscope study we immunologically targeted EBA in brain endothelial cells by intravenous injection of a monoclonal antibody, antiEBA [10]. Light microscopy of brains from experimental animals receiving such an injection showed opening of the BBB to endogenous and exogenous protein tracers, and the injected anti-EBA became localised to brain vessels [10]. However, the resolution of the light microscope did not permit accurate cellular localisation of binding sites of the injected antibody [10]. Furthermore, the route of extravasation of tracers was not identifiable by light microscopy. Thus the aims of the current study are to address these aspects employing immunological targeting of EBA with a monoclonal antibody in vivo, immunoelectron microscopy and the application of horseradish peroxidase (HRP) in vivo as an exogenous tracer.

2. Materials and methods

2.1. Animals, intravenous injection of antibodies and perfusion Thirty-seven Sprague–Dawley adult male rats (250–400 g) were used in two groups. Group 1 was used for the immunoelectron microscopic detection of EBA. Group 2 was used for electron microscopic assessment of HRP permeability. Each group included experimental and control animals. Animals were anaesthetised (i.p. injection of pentobarbitone sodium, 60 mg / kg, or urethane, 1.3 g / kg, Sigma, St. Louis, MO, USA) and the femoral vein exposed. In experimental animals (n515), a monoclonal antibody to EBA (anti-EBA, IgM, SMI 71, Sternberger Monoclonals, USA) was injected slowly into the femoral vein (20 ml, diluted in 0.4 ml phosphate-buffered saline, PBS). Control animals received femoral vein injections of either an equal volume of the vehicle (n55), or an equal volume of the vehicle containing one of three control antibodies. These included (a) 20 ml of a monoclonal antibody to human cytokeratin 8 (clone 35bH11, IgM, Dako, USA) [11] (n55); (b) 20 ml of a monoclonal antibody to Salmonella bacterial antigen (SAL-3, IgM, gift from Dr. L.K. Ashman, Division of Haematology, IMVS, Adelaide, Australia) [26] (n55); (c) 20 ml of monoclonal antibody to rat endothelial cell-specific antigen (RECA-1, clone HIS52, IgG1, Serotec Oxford, UK) [7] (n57). Antibodies were obtained as mouse ascites fluid (Anti-

141

EBA, 10 mg / ml) or culture supernatant (anti-cytokeratin, 511 mg / ml; SAL-3, 2 mg / ml and RECA-1, 10–50 mg / ml). The duration of injection of antibodies ranged from 5 to 12 min with a further circulation time of 5 min. After the circulation time, the left ventricle was cannulated, the right atrium cut, and the circulation flushed with Dulbecco’s phosphate-buffered saline (pH 7.4), containing 1% minimum essential medium and gassed with 95% O 2 –5% CO 2 for 30 s, followed by the fixative under constant pressure of 90 mmHg. The rate was adjusted to allow a fast flow for the first 300 ml of fixative, followed by 400 ml at a slow rate. The brain was removed, divided in the median plane, immersed for 3 h in the same fixative, then cut into 50-mm sections in the sagittal plane using a Vibratome.

2.2. Group 1: preembedding immunoelectron microscopic detection of EBA The first group of rats, experimental (n55) and control (n510), was used for the immunoelectron microscopic detection of EBA. The fixative used in this group was 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. Vibratome sections were washed in PBS, and the endogenous peroxidase activity blocked (0.3% H 2 O 2 in distilled water for 30 min). Sections were incubated for 1 h in 10% normal horse serum (NHS) before incubation overnight at 4 8C in anti-EBA monoclonal primary antibody (1:3000 in 1% NHS). Sections were incubated in a biotinylated ratadsorbed horse anti-mouse IgG secondary antibody (Vector Labs., USA, 1:200 in 1% NHS). The secondary antibody was detected using the avidin–biotin peroxidase method (ABC kit, Vector Labs.) or peroxidase-conjugated streptavidin (Rockland, Gilbertsville, PA, USA, 1:1000 in 1% NHS). The peroxidase reaction product was developed, by incubation in 3,39-diaminobenzidine tetrahydrochloride (DAB) using the glucose oxidase method with nickel enhancement [32]. In the immunostaining protocol for EBA, Vibratome sections in each staining session were divided into three batches. The first batch was treated with the primary and secondary antibodies, tertiary reagent and DAB. In batch 2 the primary antibody was omitted. In batch 3 both the primary and secondary antibodies were omitted. Batches 2 and 3 acted as internal controls for the staining procedure. After immunostaining, Vibratome sections were osmicated (1% osmium tetroxide, 1h), dehydrated and infiltrated with Taab resin (TAAB9 Laboratories Equipment, Berkshire, UK). Rectangles of tissue (232 mm) from the frontal cortex, central white matter and pons were embedded in Taab resin. Ultrathin sections with or without heavy metal staining (uranyl acetate and lead citrate) were examined in a Philips CM100 electron microscope. Images were taken on photographic cut films or digitally using a slow-scan digital camera (MegaViewII, Soft Imaging System, ¨ Munster, Germany). Digital images were stored on elec-

142

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

tronic media, viewed directly on computer screen and printed on Epson photo paper using an Epson printer.

2.3. Group 2: electron microscopic assessment of HRP permeability The second group of rats was used for the HRP permeability study. Experimental (n510) and control (n5 12) animals were injected i.p. with an antihistaminic drug to prevent any allergic reaction to HRP (promethazine, 1 ml of 1% solution in distilled water / kg) [6,16]. The femoral vein was exposed and antibodies or PBS were injected as above. Following the circulation period of antibodies or PBS, animals were injected with HRP (type II, Sigma, 200 mg / kg, in 0.4 ml of saline) over a period of 5 min. The tracer was allowed to circulate for a further period of 2, 5 or 15 min (total exposure time to HRP 7–20 min). Cardiac perfusion was carried out as above. The fixative used was 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, with or without 2% glutaraldehyde. In two experimental animals a very short perfusion time (1 min) was used to instantly kill the animal under anaesthesia with minimal washing of HRP from the vascular bed. The liver and brain were rapidly removed from all animals, the brain was sectioned in the median plane and fixation was achieved by immersion overnight in the fixative. Vibratome sections (50 mm) were cut from the brain and liver and treated with DAB for histochemical development of HRP reaction product. In some sections the reaction product was enhanced with nickel [32]. Following osmication and dehydration of Vibratome sections, 232 mm rectangles were cut from liver, cerebral cortex, white matter and brainstem, and embedded in Taab resin. Ultrathin sections were examined by electron microscopy as above.

3. Results

3.1. HRP permeability Ultrathin sections from the cerebral cortex, white matter, brainstem and liver were examined by electron microscopy for the presence and distribution of HRP. Animals injected with control antibodies, antihuman cytokeratin, RECA-1, or SAL-3 (Fig. 1A), or with saline (Fig. 1B) did not show any HRP leakage in the brain. Occasional red blood cells seen in the lumina of blood vessels showed dense black reaction product of endogenous peroxidase, indicating the successful histochemical development of the reaction product. Also liver tissue from these animals showed HRP among liver cells, indicating the successful injection and circulation of the tracer. Even in the animals with longer exposure period to HRP, no reaction product was seen in these regions of the brain (Fig. 1A and B). In the grey and

white matter vessels, endothelial cell cytoplasmic vesicles occasionally contained HRP (Fig. 1A). Unlike control material, sections obtained from the cerebral cortex, white matter and brainstem in experimental animals injected with anti-EBA showed leakage of HRP in the brain (Figs. 1C, D and 2). In the animals with longer exposure time to HRP, electron-dense reaction product was widely dispersed in brain parenchyma among axons and glial cells (Fig. 1C and D), indicating a severely compromised BBB. The tracer was taken up by many neuronal cell bodies and processes. The reaction product heavily impregnated the perivascular space and the basal laminae of endothelial cells and pericytes of many cerebral vessels (Figs. 1C and D and 2A). HRP was present in endothelial cell vesicles (Figs. 1C and D and 2A). In animals with the long perfusion time the lumina of blood vessels were free of HRP due to flushing of the circulation during perfusion fixation (Figs. 1 and 2A). In most endothelial cell junctions HRP was seen in the abluminal part of the junction and was continuous with the HRP in the perivascular space (Figs. 1D and 2A). The luminal ends of these junctions did not contain HRP (Figs. 1D and 2A). In a few junctions however, the tracer was seen throughout the length of the junction (Fig. 2A). In the shorter exposure times to HRP (7 and 10 min) the tracer was seen in endothelial cell cytoplasmic vesicles and endothelial cell junctions with only moderate presence in the perivascular space and the parenchyma immediately adjacent to vascular profiles. Macrophages seen occasionally in the lumina of blood vessels contained granules of HRP reaction product. Some pericytes were laden with HRP. In the brains fixed by immersion after the short perfusion time (Fig. 2B and C), microvessels were collapsed with many folds in endothelial cells and the lumina contained red blood cells. HRP was seen as a thin rim around red blood cells (Fig. 2B and C). In these animals many vessels showed HRP throughout endothelial cell junctions, in cytoplasmic vesicles, the perivascular space and parenchyma. Despite the presence of HRP in relation to many vessels as described above, in each section examined several vessels were seen with no associated HRP, suggesting heterogeneity of the vascular bed, variation in vessels’ perfusion in vivo, or rapid clearance of HRP.

3.2. Immunoelectron microscopy for EBA in experimental animals EBA immunoreactivity was visible as electron-dense peroxidase reaction product (Fig. 3). The most significant finding in EBA labelling was seen in EM images obtained from experimental animals injected in vivo with anti-EBA but using Vibratome sections from batch 2, where the anti-EBA primary antibody was omitted from the staining procedure. These sections showed immunoreactivity for EBA in brain endothelial cells (Fig. 3). In the majority of

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

143

Fig. 1. Electron micrographs showing the distribution of HRP in control animals injected in vivo with antibody to Salmonella bacterial antigen (A) or phosphate-buffered saline (B), and experimental animals injected in vivo with anti-EBA (C and D), with long perfusion fixation, hence the vascular bed is free of HRP. In control animals HRP is seen in endothelial cell vesicles (large arrows, A), but the perivascular space and parenchyma are free of HRP (A and B). In experimental animals HRP is seen in the parenchyma among neuronal processes (arrowheads, C and D). HRP is also seen in endothelial cytoplasmic vesicles (large arrows, C and D) and in the perivascular space at the planes of the basal laminae of endothelial cells and pericytes (small arrows, C and D); (C and D) show swelling of perivascular astrocytic processes (asterisks) which appear to indent the vascular profile. (D) Endothelial cell junction (double arrow) the outer part of which contains HRP continuous with that in the perivascular space. All images are from sections unstained with heavy metals. Scale bars, 1 mm.

vessels strong immunoreactivity for EBA was seen at the luminal membrane (Fig. 3A–C) and cytoplasm of endothelial cells (Fig. 3B). A minority of endothelial cell nuclei showed labelling of the marginal chromatin on the luminal side of the nucleus (Fig. 3A and C). In a minority of vessels the staining was either absent or weak (Fig. 3D). Batch 1 of sections from experimental animals, treated

with the primary antibody (anti-EBA) in the staining procedure, showed labelling of the luminal membrane, cytoplasm and the nuclei, except that the labelling appeared denser and affected almost all vessels (results not shown). EM sections obtained from Vibratome sections in batch 3, where both the primary and secondary antibodies were omitted, did not show any immunoreactivity.

144

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

Fig. 2. Electron micrographs showing HRP distribution in experimental animals injected in vivo with anti-EBA, with long perfusion fixation time (A) or short perfusion time followed by immersion fixation (B and C). In (A) the vessel lumen is free from HRP but contains two platelet-profiles (P). HRP is seen in the perivascular space at the planes of the basal laminae of endothelial cells and pericytes (arrows) and in endothelial cell vesicles (V). Some vesicles appear continuous with the luminal or abluminal membrane, or in the centre of the cytoplasm. HRP is also seen within endothelial cell junctions, partially or completely filling the intercellular cleft, between the abluminal and adluminal ends (arrowheads). Asterisks indicate swelling of perivascular astrocytic processes; (B and C) show brain microvessels containing red blood cells (R) and HRP reaction product forming a thin rim between the red blood cells and the luminal membrane of endothelial cells (white arrows). HRP is also seen in endothelial cell junctions between the abluminal and luminal ends (arrowheads) and in the perivascular space (black arrow, C); (C) shows part of the wall of a partially collapsed microvessel and displays swelling of perivascular astrocytic processes (asterisks). All images are from sections not stained with heavy metals. Scale bars: A and B, 1 mm; C, 0.5 mm.

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

145

Fig. 3. Electron micrographs of brain microvessels from experimental animals injected in vivo with anti-EBA monoclonal antibody. Although sections were not exposed to the anti-EBA primary antibody in the staining procedure, but only to the secondary antibody and tertiary reagents, they showed labelling for EBA, indicating the formation of antigen–antibody complexes in endothelial cells in vivo. The labelling appears to be located predominantly at the luminal membrane of endothelial cells (small arrows). Also in some regions the labelling is seen in the whole thickness of endothelial cell cytoplasm (large arrows, B) and in the nuclear chromatin on the luminal side of the nucleus (asterisks, A and C). Cytoplasmic vesicles (V, B) appear unlabelled. (D) Example of a minority of vessels, which showed weak labelling (small arrows), or absence of labelling (arrowheads) at the luminal membrane of endothelial cells. Immunostaining with nickel enhancement. No sections were stained with heavy metals. Scale bars: A, 2 mm; B and D, 1 mm; C, 0.5 mm.

146

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

3.3. Immunoelectron microscopy for EBA in control animals Control animals injected with saline, anti-cytokeratin, SAL-3 (Fig. 4A) or RECA-1 (Fig. 4B) antibodies showed normal EBA expression in endothelial cells of microvessels in the grey and white matter, when the primary antibody (anti-EBA) was included in the staining protocol (batch 1 of sections). Thus the labelling was similar to experimental animals in batch 1. Strong labelling for EBA was seen in the luminal membrane and cytoplasm of endothelial cells (Fig. 4A and B). Some endothelial cells showed labelling of the nuclear membrane and chromatin on the side of the nucleus facing the lumen (Fig. 4A and B). When the primary and secondary antibodies were omitted from the staining protocol (batch 3) no immunoreactivity for EBA was seen. Thus batch 3, of control and experimental animals were also similar. On the other hand, EBA labelling in controls in batch 2, where the primary antibody was omitted, did not show any staining. Thus batch 2 was dissimilar to that of experimental animals.

3.4. Ultrastructure of endothelial cells EM sections from control animals examined for the distribution of HRP and for the detection of labelling for EBA also provided opportunities for the examination of the ultrastructure of endothelial cells. The majority of

microvessels had a circular or slightly oval profile with smooth contour, thus excluding any gross response to the injected control antibodies or saline, and endothelial cells had normal ultrastructure (Fig. 1A and B). In experimental animals however, many vessels showed ultrastructural changes, such as swelling of perivascular astrocytic processes (Figs. 1C and D, 2A and C). Endothelial cells showed increased cytoplasmic vesicles and vacuoles and crater-like invaginations of the luminal membrane (Fig. 5A and B). Large vacuoles were predominantly seen near endothelial cell junctions but no endothelial cell gaps were seen. The luminal membrane showed occasional blebbing (Fig. 5C). Endothelial cell junctions showed widening at multiple foci and regions with large vacuoles (Fig 5D and E).

4. Discussion Employing electron microscopy in this study yielded three main findings. Firstly, the in vivo systemically administered anti-EBA was seen to localise to endothelial cells of brain vessels. Secondly, anti-EBA administration resulted in instant opening of the BBB as detected by leakage of HRP into the perivascular space and parenchyma. Thirdly, HRP escaped from the blood to the brain apparently via altered paracellular clefts and endothelial cell cytoplasmic vesicles.

Fig. 4. Electron micrographs of brain microvessels displaying immunolabelling for EBA in control animals injected with anti-Salmonella bacterial antigen (A) or RECA-1 (B). These sections were exposed to the primary anti-EBA, secondary antibody and tertiary reagents in the staining procedure. The vessels show normal labelling for EBA at the luminal membrane (small arrows) and cytoplasm (large arrows), with endothelial cell vesicles (V) not labelled. Also the nuclei are labelled at their luminal side (asterisks). Both sections are not stained with heavy metals. Scale bars, 1 mm.

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

147

Fig. 5. Electron micrographs of brain microvessels demonstrating ultrastructural changes in experimental animals injected in vivo with anti-EBA followed by HRP for 7 min (C and E), or 10 min (A, B and D). In contrast to the control animals in (1A) and (1B), the luminal membrane shows undulations and invaginations (small arrows, A and B), and the cytoplasm shows large vacuoles (large arrows, A and B). (C) Luminal membrane blebbing (arrows) near an endothelial cell junction (arrowhead). (D and E) Endothelial cell junctions (arrowheads) with multiple focal dilatations (D) and large vacuoles (arrows, D and E). Minimal amount of HRP (white arrows) is seen in the perivascular space (A, C and E) and between perivascular astrocytic processes (A). No sections were stained with heavy metals. Scale bars: A, B and D, 1 mm; C and E, 0.5 mm.

4.1. EBA labelling The endothelial barrier antigen (EBA) is a protein strongly expressed by endothelial cells of barrier vessels in the CNS [13,14,31,33,34]. Immunoelectron microscopy, using a monoclonal antibody (anti-EBA), pre- and postembedding methods, and peroxidase and gold probes, have shown that EBA is localised to luminal membranes of endothelial cells [13,14,31,34,36]. Intravenously administered antibodies are able to interact with specific target proteins on endothelial cell membranes. This approach was

employed in previous studies on the transferrin receptor [4] and RECA-1 protein [7]. We recently carried out a light microscope study on the effect of intravenous injection of anti-EBA on the BBB [10]. In that study Vibratome sections immunostained for EBA, without applying the primary antibody (anti-EBA) in the staining protocol, showed immunoreactivity to EBA in brain vessels. This finding indicated that the anti-EBA injected in vivo acted as a substitute for the exposure of tissues to the primary antibody in the staining procedure. Although brain vessels were labelled, cellular localisation of EBA was not feasible

148

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

with the light microscope resolution [10]. The current study employed the same approach combined with immunoelectron microscopy and revealed that the intravenously injected anti-EBA was localised principally to endothelial cell luminal membrane. Presumably the antiEBA identified EBA molecules on endothelial cells and formed antigen–antibody complexes. The anti-EBA injected in vivo appears to form a strong binding to the targeted EBA, since it was not dissociated by vascular flushing with buffer and perfusion fixation. The cytoplasmic and nuclear localisation of EBA, observed in the current study, has not been reported by other laboratories. Previous studies employing electron microscopy with peroxidase and gold probes have reported the localisation of EBA only to luminal membranes of endothelial cells [13,14,31,34]. The cytoplasmic labelling of endothelial cells seen in experimental animals in the current study may be interpreted as representing internalisation by endothelial cells of the presumed antigen–antibody complexes formed in vivo, prior to perfusion fixation of the tissues. However, this interpretation does not explain the cytoplasmic labelling for EBA seen in control animals, injected in vivo with saline or control antibodies. In these animals, anti-EBA was applied to the tissue after fixation, with no possibility of internalisation by endothelial cells. We considered the possibility that cytoplasmic and nuclear labelling in experimental and control animals is nonspecific, as IgM molecules have a tendency to bind nonspecifically to histones and highly charged proteins [Ludwig Sternberger, personal communication]. However, we concluded that nonspecific labelling in the current study is unlikely for two main reasons. (a) The nuclei and cytoplasm of neurons, neuroglia and pericytes, in the same sections which showed labelling of endothelial cells, were unlabelled. (b) The control antibodies anti-cytokeratin and SAL-3 are also IgM molecules, but did not show such nuclear or cytoplasmic labelling in control material. Biotinylated secondary antibodies in conjunction with avidin may also be a source of nonspecific results, by labelling naturally occurring biotin [Ludwig Sternberger, personal communication]. However, in the current study the same secondary antibody (biotinylated rat-adsorbed horse anti-mouse) was used in all experimental and control animals. Sections from control animals exposed to the secondary antibody alone (without prior exposure to antiEBA primary antibody) did not show any labelling. Therefore, nonspecific labelling arising from the biotinylated secondary antibody is also unlikely. We concluded therefore that the cytoplasmic and nuclear localisation of EBA seen in the current study may represent normal distribution of EBA that has not been recognised previously. Cytoplasmic labelling may indicate a pool of EBA the function of which is to replenish the membrane content of EBA molecules. In control animals injected with saline, anti-cytokeratin or SAL-3 antibodies, EBA labelling was similar to ex-

perimental animals when anti EBA was added to the staining protocol (batch 1 of sections). Endothelial cells showed luminal membrane, cytoplasmic and nuclear staining. This indicates that injections of these control antibodies or saline in vivo did not alter the normal expression of EBA in endothelial cells. However, when the primary anti-EBA was omitted from the staining protocol (batch 2 of sections) no labelling for EBA in controls was seen, unlike experimental animals. This indicates that the labelling seen in experimental animals was due to the anti-EBA injected in vivo. The control animals injected with RECA-1 in vivo, but not exposed to the anti-EBA primary antibody in the staining protocol (batch 2) showed unexpected results. We expected to see reaction product on luminal membranes indicative of labelling for RECA-1 antigen for the following reasons. (a) RECA-1 antigen is expressed by brain endothelial cells [7]. (b) RECA-1 antibody was injected in vivo. (c) Since RECA-1 was raised in the mouse, we expected that the secondary antibody used (horse antimouse) would label brain endothelial cells. A previous study by Duijvestijn et al. [7] employed in vivo application of RECA-1 and immunostaining, with omission of the step of primary antibody incubation, and reported endothelial cell labelling. The difference between that study and the current study may be related to (a) the dose of the injected RECA-1. Here we injected 20 ml, while the dose used in the study by Duijvestijn et al. [7] was not specified. (b) The length of survival time, since the RECA-1 antibody– antigen complexes may be transient or may be internalised [7] with possible breakdown of the complexes. (c) Susceptibility of the RECA-1 antibody–antigen complexes to processing solutions and embedding media. In the study of Duijvestijn et al. [7] although paraformaldehyde fixative was used, as in the current study, tissues were dehydrated in acetone and frozen sections were used, or plastic embedded sections were used at low temperature. Our previous light microscope study of this subgroup did not reveal any RECA-1 labelling of blood vessels in Vibratome or wax sections [10]. We concluded that the concentration of the RECA-1 might not have been high enough to be detected in batch 2, with omission of the primary RECA-1 antibody in the staining procedure and / or the processing schedule may have resulted in the breakdown of RECA-1 antigen–antibody complexes.

4.2. Increased vascular permeability In the current study pronounced vascular permeability to HRP was noted in experimental animals where the tracer was seen in the parenchyma and perivascular space. None of the control animals injected with control antibodies or saline showed HRP leakage. Therefore, the results indicate that the leakage was not due to an allergic response to the exposure to foreign proteins. Furthermore, since one of the control antibodies (RECA-1) is specific to rat endothelial

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

cells, including brain endothelial cells, the increased vascular permeability seen in experimental animals is unlikely to be due to a general effect of antibodies to endothelial cells. Absence of HRP leakage in control animals also excludes a histamine response to the administration of HRP. In the current study all animals injected with HRP were pretreated with an antihistaminic drug [6,16]. It is likely therefore that the increased permeability is due a specific effect of anti-EBA on the BBB. Vascular permeability has been a focus for scientific research and interest for many decades, and was recently reviewed [20]. Earlier investigations hypothesised that the structural correlate of increased vascular permeability, induced by histamine and serotonin, is separation of intercellular junctions between adjacent endothelial cells of postcapillary venules [17,18]. Subsequent studies either supported or criticised this hypothesis [see 20]. The apparently conflicting reports in the literature over the last forty years may in fact be a reflection of several factors. (a) The presence of more than one possible route for increased permeability. (b) Variability between species. (c) The different responses shown by different vascular beds. (d) The variety of permeability-inducing factors. (e) Different tracers / molecules used as markers for permeability. Temperature-induced vascular permeability in frog mesenteric vessels predominantly occurs via trans-endothelial cell gaps located less than 2 mm from an intact endothelial cell junction [22]. Similar gaps located close to endothelial cell junctions were seen in vascular permeability induced by the Ca 21 ionophore A23187 [23]. Experiments on frog mesenteric and rat lung vessels demonstrated the formation of trans-endothelial cell gaps in response to an increase in the capillary intraluminal pressure [8,24]. In histamineinduced vascular permeability although trans-endothelial cell gaps were seen, more gaps occurred between endothelial cells [19]. In the current study no trans-endothelial cell gaps were seen, although large vacuoles were seen near endothelial cell junctions. Thus in the current model of anti-EBA injection, it is unlikely that the increased permeability was due to gap formation. HRP has been used extensively in experimental studies of vascular permeability. In normal barrier vessels of the brain, HRP is thought to be taken up by endothelial cells non-specifically into vesicles by fluid-phase endocytosis then directed to endosomes and secondary lysosomes for degradation, and not transported across endothelial cells [3,5]. However, insignificant amounts may be transported by endothelial cells of some arterioles [see 35]. In the current study of control animals injected with saline or control antibodies, HRP was not seen in the brain parenchyma or the perivascular space, but only in some vesicles associated with the luminal membrane or in the cytoplasm of endothelial cells, indicating an intact barrier. However, in experimental animals injected with anti-EBA, HRP was seen in brain parenchyma, the perivascular space and in endothelial cell vesicles associated with the luminal mem-

149

brane, the cytoplasm and the abluminal membrane. This suggests that the increased vesicles may be involved in the transport of HRP across the BBB. Other studies have suggested the transfer of HRP across endothelial cells of the BBB in other pathological condition such as spinal cord trauma [25,30] and in the spinal cord of scrapieinfected mice [15]. However, some authors believe that abluminal membrane-associated vesicles may represent static invaginations of the abluminal membrane filled with HRP that had entered the perivascular space by another route [3]. Endothelial cells of certain segments of brain arterioles have been shown to be leaky to HRP under normal condition [35], and in hypertension [21], seizures and hypoxia [28]. Having entered the perivascular space of the arterioles, HRP may spread at the plane of the basal lamina to reach the perivascular space of capillaries. However, one cannot dismiss that some of the abluminal membrane vesicles represent cytoplasmic HRP-filled vesicles that have fused with the abluminal membrane [5] and thus the transcellular vesicular route may be a possible mechanism for the breakdown of the BBB and transport of HRP in the current model. In the current study electron microscopy demonstrated increased cytoplasmic vesicles and the presence of large endothelial cell vacuoles, both of which contained HRP. Large vacuoles, which traverse the thickness of endothelial cells, have been reported in increased vascular permeability induced by the Ca 21 ionophore A23187, histamine and vascular endothelial growth factor [9,20]. The Ca 21 ionophore A23187 is thought to increase permeability by raising the intracellular Ca 21 . Natural mediators are also thought to act by increasing Ca 21 influx. In the current study of targeting EBA molecules, since anti-EBA was bound to endothelial cell membranes it was presumed that it formed antigen–antibody complexes with EBA molecules in endothelial cell membranes. Whether these presumed complexes resulted in Ca 21 influx into endothelial cells is not known. In the normal BBB, endothelial cells are joined by tight junctions, which occlude the paracellular cleft [2]. Pathological widening of the endothelial cell junctions, as a mechanism for enhanced vascular permeability, has been reported previously. Studies on vascular permeability using substance P administration or sensory nerve stimulation in the rat demonstrated the formation of gaps at endothelial cell junctions in tracheal venules [1]. Also osmotic opening of the BBB is thought to occur via widened paracellular clefts [29]. In the current study in experimental animals subjected to the long perfusion fixation, HRP was seen in the abluminal part but not in the luminal part of many junctions. Two alternative interpretations were considered. (a) The tracer in the abluminal part of the junctions may represent back filling from the pool of tracer in the perivascular space that had escaped from the vascular bed via a different endothelial route. This interpretation implies that the paracellular route was not open. (b) Tight junc-

150

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

tions between endothelial cells were open and allowed HRP passage to the perivascular space but during the long perfusion time the tracer was washed out of the luminal ends of the junctions. Indeed the lumina of almost all vessels were free of any tracer or blood elements following the long perfusion, and only occasional vessels showed some platelets or red blood cells. In an attempt to clarify this, two further experimental animals were subjected to short perfusion. Examination of material obtained from the brains of these two animals showed HRP in the lumen and throughout the length of many endothelial cell junctions. Thus widening of endothelial cell junctions appears to be a mechanism for opening the BBB in this model. EBA has not been reported at endothelial cell junctions [13,14,31,34,36]. In the current study experimental animals showed opening of endothelial cell junctions, which appeared to provide a route for vascular permeability to HRP. Also this study demonstrated an abundance of endothelial cytoplasmic vesicles and vacuoles which contained HRP. It is not clear from the current study how the interaction between the anti-EBA and EBA on luminal membranes resulted in opening of endothelial cell junctions and the formation of numerous vesicles. The mechanisms by which the anti-EBA injected in vivo ‘neutralised’ EBA molecules on endothelial cells or interfered with the function of EBA and possible direct or indirect effects via signalling mechanisms require further investigation.

Acknowledgements We would like to thank Drs. Gerald Allt and Ludwig Sternberger for fruitful discussion. C. Zhu was supported by an Adelaide University grant and was a holder of an International Postgraduate Research Scholarship.

References [1] P. Baluk, A. Hirata, G. Thurston, T. Fujiwara, C.R. Neal, C.C. Michel, D.M. McDonald, Endothelial gaps: time course of formation and closure in inflamed venules of rats, Am. J. Physiol. 272 (1997) L155–L170. [2] M.W. Brightman, T.S. Reese, Junctions between intimately apposed cell membranes in the vertebrate brain, J. Cell Biol. 40 (1969) 648–677. [3] R.D. Broadwell, Endothelial cell biology and the enigma of transcytosis through the blood–brain barrier, in: L.R. Drewes, A.L. Betz (Eds.), Frontiers in Cerebral Vascular Biology: Transport and Its Regulation, Plenum Press, New York, 1993, pp. 137–141. [4] R.D. Broadwell, B.J. Baker-Cairns, P.M. Friden, C. Oliver, J.C. Villegas, Transcytosis of protein through the mammalian cerebral epithelium and endothelium. III. Receptor-mediated transcytosis through the blood–brain barrier of blood-borne transferrin and antibody against the transferrin receptor, Exp. Neurol. 142 (1996) 47–65. [5] R.D. Broadwell, B.J. Balin, M. Salcman, Transcytotic pathway for blood-borne protein through the blood–brain barrier, Proc. Natl. Acad. Sci. USA 85 (1988) 632–636.

[6] R.S. Cotran, M.J. Karnovsky, Ultrastructural studies on the permeability of the mesothelium to horseradish peroxidase, J. Cell Biol. 37 (1968) 123–137. [7] A.M. Duijvestijn, H. Van Goor, F. Klatter, G.D. Majoor, E. Van Bussel, P.J.C. Van Breda Vriesman, Antibodies defining rat endothelial cells: RECA-1, a pan-endothelial cell-specific monoclonal antibody, Lab. Invest. 66 (1992) 459–466. [8] A.R. Elliott, Z. Fu, K. Tsukimoto, R. Prediletto, O. MathieuCostello, J.B. West, Short-term reversibility of ultrastructural changes in pulmonary capillaries caused by stress failure, J. Appl. Physiol. 73 (1992) 1150–1158. [9] D. Feng, J.A. Nagy, J. Hipp, H.F. Dvorak, A.M. Dvorak, Vesiculovacuolar organelles and the regulation of venule permeability to macromolecules by vascular permeability factor, histamine and serotonin, J. Exp. Med. 183 (1996) 1981–1986. [10] M.N. Ghabriel, C. Zhu, G. Hermanis, G. Allt, Immunological targeting of the endothelial barrier antigen (EBA) in vivo leads to opening of the blood–brain barrier, Brain Res. 878 (2000) 127–135. [11] A.M. Gown, A.M. Vogel, Monoclonal antibodies to human intermediate filament proteins. II. Distribution of filament proteins in normal human tissues, Am. J. Pathol. 114 (1984) 309–321. [12] J.M. Krum, K.L. Kenyon, J.M. Rosenstein, Expression of blood– brain barrier characteristics following neuronal loss and astroglial damage after administration of anti-Thy-1 immunotoxin, Exp. Neurol. 146 (1997) 33–45. [13] J.G. Lawrenson, M.N. Ghabriel, A.R. Reid, T.N. Gajree, G. Allt, Differential expression of an endothelial barrier antigen between the CNS and the PNS, J. Anat. 186 (1995) 217–221. [14] J.G. Lawrenson, M.N. Ghabriel, A.R. Reid, T.N. Gajree, G. Allt, Distribution of a putative endothelial barrier antigen in the ocular and orbital tissues of the rat, Br. J. Ophthalmol. 79 (1995) 462–466. [15] A.S. Lossinsky, R.C. Moretz, R.I. Carp, H.M. Wisniewski, Ultrastructural observations of spinal cord lesions and blood–brain barrier changes in scrapie-infected mice, Acta Neuropathol. 73 (1987) 43–52. [16] N. Ma, N.H. Hunt, M.C. Madigan, T. Chan-Ling, Correlation between enhanced vascular permeability, up-regulation of cellular adhesion molecules and monocyte adhesion to the endothelium in the retina during the development of fatal murine cerebral malaria, Am. J. Pathol. 149 (1996) 1745–1762. [17] G. Majno, G.E. Palade, Studies on inflammation. I. The effect of histamine and serotonin on vascular permeability: an electron microscopic study, J. Biophys. Biochem. Cytol. 11 (1961) 571–605. [18] G. Majno, G.E. Palade, G.I. Schoefl, Studies on inflammation. II. The site of action of histamine and serotonin along the vascular tree: a topographic study, J. Biophys. Biochem. Cytol. 11 (1961) 607– 626. [19] C.C. Michel, C.R. Neal, Pathways through microvascular endothelium of normal and increased permeability, in: G.V.R. Born, C.J. Schwartz (Eds.), Vascular Endothelium: Physiology, Pathology and Therapeutic Opportunities, Schattauer, Stuttgart, 1997, pp. 37–48. [20] C.C. Michel, C.R. Neal, Openings through endothelial cells associated with increased microvascular permeability, Microcirculation 6 (1999) 45–54. [21] S. Nag, S.I. Harik, Cerebrovascular permeability to horseradish peroxidase in hypertensive rats: effects of unilateral locus ceruleus lesion, Acta Neuropathol. 73 (1987) 247–253. [22] C.R. Neal, C.C. Michel, Transcellular openings through microvascular walls in acutely inflamed frog mesentery, Exp. Physiol. 77 (1992) 917–920. [23] C.R. Neal, C.C. Michel, Transcellular gaps in microvascular walls of frog and rat when permeability is increased by perfusion with the ionophore A 23187, J. Physiol. 488 (1995) 427–437. [24] C.R. Neal, C.C. Michel, Openings in frog microvascular endothelium induced by high intravascular pressures, J. Physiol. 492 (1996) 39–52. [25] L.J. Noble, J.R. Wrathall, Blood–spinal cord barrier disruption

M.N. Ghabriel et al. / Brain Research 934 (2002) 140 – 151

[26]

[27]

[28]

[29]

[30]

[31]

proximal to a spinal cord transection in the rat: time course and pathways associated with protein leakage, Exp. Neurol. 99 (1988) 567–578. C.G. O’Connor, L.K. Ashman, Application of the nitrocellulose transfer technique and alkaline phosphatase conjugated anti-immunoglobulin for determination of the specificity of monoclonal antibodies to protein mixtures, J. Immunol. Methods 54 (1982) 267–271. M. Perdiki, M. Farooque, A. Holtz, G.L. Li, Y. Olsson, Expression of endothelial barrier antigen immunoreactivity in blood vessels following compression trauma to rat spinal cord. Temporal evolution and relation to the degree of the impact, Acta Neuropathol. 96 (1998) 8–12. C.K. Petito, D.E. Levy, The importance of cerebral arterioles in alterations of the blood–brain barrier, Lab. Invest. 43 (1980) 262– 268. S.I. Rapoport, Osmotic opening of the blood–brain barrier: principles, mechanism, and therapeutic applications, Cell. Mol. Neurobiol. 20 (2000) 217–230. M. Risling, H. Linda, S. Cullheim, P. Franson, A persistent defect in the blood–brain barrier after ventral funiculus lesion in adult cats: implications for CNS regeneration?, Brain Res. 494 (1989) 13–21. J.M. Rosenstein, J.M. Krum, L.A. Sternberger, M.T. Pulley, N.H.

[32]

[33]

[34]

[35]

[36]

151

Sternberger, Immunocytochemical expression of the endothelial barrier antigen (EBA) during brain angiogenesis, Dev. Brain Res. 66 (1992) 47–54. S. Shu, G. Ju, L. Fan, The glucose oxidase-DAB-nickel method in peroxidase histochemistry of the nervous system, Neurosci. Lett. 85 (1988) 169–171. N.H. Sternberger, L.A. Sternberger, Blood–brain barrier protein recognized by monoclonal antibody, Proc. Natl. Acad. Sci. USA 84 (1987) 8169–8173. N.H. Sternberger, L.A. Sternberger, M.W. Kies, C.R. Shear, Cell surface endothelial proteins altered in experimental allergic encephalomyelitis, J. Neuroimmunol. 21 (1989) 241–248. E. Westergaard, Ultrastructural permeability properties of cerebral microvasculature under normal and experimental conditions after application of tracers, in: J. Cervos-Navarro, R. Ferszt (Eds.), Advances in Neurology. Brain Edema, Vol. 28, Raven Press, New York, 1980, pp. 55–74. C. Zhu, M.N. Ghabriel, P.C. Blumbergs, P.L. Reilly, J. Manavis, J. Youssef, S. Hatami, J.W. Finnie, Clostridium perfringens prototoxininduced alteration of endothelial barrier antigen (EBA) immunoreactivity at the blood–brain barrier (BBB), Exp. Neurol. 169 (2001) 72–82.