- Email: [email protected]
Observation of the blood-brain barrier function and vasomotor response in rat microcirculation using intravital fluorescence microscopy
EXPERIMENTAL
NEUROLOGY
117,
247-253 (19%)
Observation of the Blood-Brain Barrier Function and Vasomotor Response in Rat Microcirculation Using lntravital Fluorescence Microscopy SHINGO Department of Surgical
KAWAMDRA
AND NOBUYUKI
YASUI
Neurology, Research Institute for Brain and Blood Vessels-AKZTA,
Akita 010, Japan
aim of this paper is to show that the experimental animals had normal BBB functions as well as physiological vasomotor responses in the pial microvessels. Na+-fluorescein (MW 376) was used as a BBB and vessel marker.
The blood-brain barrier function and the vasomotor response in rat microcirculation was studied using a closed cranial window technique and an intravital fluorescence microscope with an attempt to reduce both the light intensity and the amount of 2% Na+-fluorescein to as little as possible. The light intensities in the focus were ~2.6 mW/cm’, and with the low-light TV camera we were able to reduce the total amount of the dye to ~0.7 ml during 6-h experiments. Observing the brain surface every 30 min showed that the barrier function remained intact for up to 6.0 h following the first iv administration of the dye. This study has reconfirmed that the pial vessels have physiological vasomotor responses (e.g., CO2 reactivity). The osmotic barrier disruption started at a small venule 10 min after the start of superfusion with mannitol (1000 mOsm/liter). In conclusion, the microscope system and the small animal model used in this study were useful in studying the blood-brain barrier function and the vasomotor response simultaneously under various experimental . conditions. o 1992 Academic press, IIN.
MATERIALS
AND
METHODS
General Preparation Twenty-two male Sprague-Dawley rats (250-300 g) were used. Anesthesia was induced ‘by 4% halothane, which was reduced to 1.5% during catheter insertion into a tail artery and bilateral femoral veins. The animals were first allowed to breathe spontaneously with a nitrous oxide/oxygen mixture (70/30%). Following tracheotomy, cr-chloralose (120 mg/kg, iv) was administered, and then halothane and nitrous oxide were discontinued. Supplemental doses of a-chloralose (15-30 mg/kg) were given to maintain anesthesia when necessary. The animals were immobilized with pancronium bromide (1.2 mg/kg/h, iv) and ventilated artificially with room air and supplemental oxygen. Ventilation was adjusted to maintain arterial blood pH at 7.42 f 0.03 (mean & SD; range, 7.37-7.47), P,CO, at 38.3 f 2.4 (33.7-42.6) mmHg, and P,O, at 124 +- 11 (107-143) mmHg. Arterial blood was collected anaerobically for blood-gas analyses and hematocrit measurements. Mean arterial blood pressure (MABP) in the tail artery, ICP, and air-way pressure were recorded continuously. Fluid and drugs were administered into the left femoral vein, and Na+-fluorescein was administered into the right femoral vein. Rectal temperature was maintained at 37.5”C by heating.
INTRODUCTION Rat microcirculation has been studied previously under a closed cranial window because of advantages both in using small animals and in the preservation of intracranial physiology (lo), although only a few investigators have used rats (4,6). Details of a closed cranial window technique in rats were described by Morii et al. (9). This technique was modified by Baethmann and his co-workers (16) in studies of both the blood-brain barrier (BBB) function and the vasomotor control in association with measurements or manipulations of intracranial pressure (ICP). Improved experimental procedures have allowed the prevention of BBB disruption. An intact BBB is indispensible in studies on the barrier function and on vasomotor reactivity of brain microcirculation under physiological and pathological conditions. Recently, we have used an improved intravital fluorescence microscope to investigate rat microcirculation using the modified closed cranial window technique. The
Operative
Technique
Details have been described previously (7), but will be mentioned briefly. A 4-cm skin incision was made from the forehead to the neck. A window (5 x 6 mm) was made over the left parietal brain using a dental drill, and a thin bone layer over the dura was left intact. Three catheters (PE 50) were attached to the window for superfusion of the brain surface and for ICP measurements. They were fixed in dental cement. A plexiglass funnel was fixed to the skin above the window and filled 247 All
Copyright 0 1992 rights of reproduction
0014-4886/92 $5.00 by Academic Press, Inc. in any form reserved.
248
KAWAMURA
AND
YASUI
with paraffin oil. This exerted a pressure of 3 mmHg on the brain surface. Under the oil, the remaining bone layer of the window was removed, and the dura was opened. Following infusion of a small amount of artificial cerebrospinal fluid (CSF) over the brain, the window was closed with a glass cover (13 mm 4) with cyanoacrylate. The funnel was then removed, and the distal end of the outflow catheter of the window was adjusted to a height that maintained an ICP of 5-7 mmHg. Four animals (18%) were discarded from this study due to technical failures in the preparation. Superfusion
with Artificial
The cortical surface was superfused with artificial CSF at a rate of 5 ml/h. The CSF was composed as follows (mM): glucose, 3.3; Na+, 158; K+, 3.2; Ca’+, 1.5; Cl-, 142; Mg2+, 1.33; and HC03-, 24.5, yielding an osmolality of 306 mOsm/liter. Solutions were prepared freshly prior to experiments and bubbled with a mixture of humidified CO, (6%), 0, (lo%), with the balance of N, at 37.5”C, yielding a pH of 7.34 f 0.01 (range, 7.327.37), a PCO, of 44.8 f 1.1 (42.7-47.5) mmHg, and a PO, of 88 -t 4 (80-95) mmHg. In studies on the hypertonic opening of the BBB using a method described previously (7), mannitol (D-Mannit, Kanto Chemical Co., Inc., Tokyo, Japan) was added to the CSF at an osmolality of 1000 mOsm/liter. An infusion pump (Instamatic, B. Braun Melsungen, Germany) was used with an airtight glass syringe as a reservoir. Intravital
Fluorescence
Video
CSF
Microscopy
Intravital fluorescence microscopy was carried out using the fluorescence video-photo-microscope K90 system (Fig. l), consisting of the vertical fluorescence illuminator BH2-RFC and the trinocular tube BH2TR30 (Kyoei-Olympus, Tokyo, Japan). Continuous epiillumination of the brain surface was made only during observation (lo-15 min for each observation). The light source was a 100 W high-pressure mercury lamp attached to the illuminator. Light intensity was reduced through two sets of ND-25 filters (Kyoei-Olympus). The band-pass filter BHB-DMB (Kyoei-Olympus) was used to obtain an excitation source with a wavelength of 455-490 nm. The fluorescence emission of the brain surface, after iv injection of 2% Na+-fluorescein solution, was studied. For visual observation and taking photographs, wavelengths below 515 nm were excluded. Ten ~1 of Na+-fluorescein were administered for a total of 0.3-0.7 ml (23-64 pl/lOO g body wt/hr) during each experiment. Objective lenses of 2X, 10X, and 20X with long working distances (0.08, 0.30, and 0.40 numerical aperture, respectively) were used. These lenses, in association with the trinocular tube, gave magnifications of 6.25X to 62.5X. For quantitative studies of vasomotor
Recorder
+
Lz!G!I= Brain Surface
FIG. 1. Schematic photo-microscope K90
presentation system.
of
the
fluorescence
video-
responses, the 20x objective lens was used to measure internal diameters of pial vessels. The images were processed with a video timer (VTG-33, For.A Company, Ltd., Tokyo, Japan) and recorded on a videotape by a S-VHS videocassette recorder (HR-SlOOOO, Victor, Tokyo, Japan) at a rate of 30 frames/set. A low-light TV camera (C2741, Hamamatsu Photonics, Hamamatsu, Japan) and a TV monitor (VM-RBOOS, Victor) provided final magnifications of 60X to 600X. Regarding past studies on the BBB function, a different form of fluorescence microscopical equipment was used previously (7). In this (present) study, a photographic camera was attached to the trinocular tube in order to perform video recording and picture taking simultaneously. The microphotographs of the brain surface were taken on Kodak Ektachrome P800/1600 35mm film using a C-35AD-4 camera with an automatic exposure control unit PM-CBSP (Kyoei-Olympus). Lack of extravasation of Na+-fluorescein from pial vessels into the surrounding parenchyma was first confirmed 30 min after the first bolus of the dye, and all experiments were then started. The light intensities at the point of the focus were measured in vitro using a detector (Spectra Radio-meter USR-BOA, Ushio, Inc., Tokyo, Japan), with measurements as follows: 0.019,0.88, and 2.5 mW/cm’ at magnifications of 2X, 10X, and 20X, respectively.
BBB AND VASOMOTOR
249
RESPONSE
IN RAT
Studies of Vasomotor Response
Blood-Brain
Barrier
Vasomotor response was studied off-line. Drawings of the vascular network on the brain surface were made on tracing papers, by hand, from microphotographs taken through the 2~ objective lens. Manual measurements of the diameters of randomly selected arterioles and venules were made by way of the TV monitor, using sliding calipers (resolution, 0.05 mm). All measured locations were marked on the drawings in order to provide source material for independent investigators. In three animals, the effect of continuous superfusion of pial vessels with artificial CSF was studied for 5 h. The vessels were observed every 30 min, and sequential changes in the diameters of 63 arterioles (31 f 14 pm; range, 12-69 pm) and 49 venules (35 f 18 pm; range, 15-97 pm) were evaluated. Five animals were used to assessthe CO, reactivity of pial vessels. CO, was put into the respirator 5 min prior to observation. The CO, inhalation process was performed twice for each animal, and the time used for each experiment was 1.5-2.5 h. CSF superfusion was interrupted during the CO, inhalation. Arterial blood analyses were repeated during both normo- and hypercapnic states. Inspired CO, content was monitored using a CO, analyzer (Respina-lH31, San-ei, Tokyo, Japan) and reached 5.0-5.2% during hypercapnia. CO, reactivity was studied in 228 arterioles (32 f 20 pm; range, 8-111 pm) and 174 veins (44 + 28 ym; range, lo-152 pm).
The BBB function was studied in a total of 18 rats for up to 6.0 h following the first administration of Na+-fluorescein; 5 animals were studied for 6.0 h. Extravasation of the marker was not observed during the entire observation period, demonstrating the maintenance of a normal BBB function. The tightness of the barrier can be demonstrated by fluorescence microphotographs taken at the end of each experiment. Neither the adherence of platelets or of other blood cells to the luminal surface, nor the presence of cell aggregates in vessels, is visible. Back-and-forth movements (or stasis) of the blood flow in the smaller venules of all the animals were often observed as being the physiological blood stream in branch venules (5). When seen in photographs (Fig. 2), the physiological stasis of venous blood may look like brightly fluorescing platelet aggregates (14, 15). In testing of the osmotic BBB opening, seven animals were administered with hypertonic mannitol (1000 mOsm/liter) dissolved in artificial CSF. The pH of the solutions was 7.28 + 0.02, the PCO, was 44.7 * 1.0 mmHg, and the POZ was 93 -t 2 mmHg. The pial arterioles dilated markedly within 5 min following the start of the mannitol superfusion (not shown). The BBB began to disrupt at a small venule approximately 10 min after the start of superfusion. The gross and diffuse opening of the barrier was established 5-10 min later (Fig. 3).
Vascular Reactivity Statistical
Analysis
The Student t test was used for a comparison of the paired physiological data. Regression analysis was performed to compare control diameters of vessels with the percentage of change in diameter during CO, inhalation. When the effect of continuous CSF superfusion on vessel diameters was analyzed, the diameter of each vessel was compared with its diameter while at rest, using Dunnett’s multiple comparison (randomized block design). A P level < 0.05 was considered significant.
RESULTS
Physiological
Data
MABP had a range of 81-111 (mean f SD, 95 + 7) mmHg in all animals throughout the experiments. ICP was between 5-7 mmHg. The initial hematocrit (Hct) was 42 f 2% (range, 37-45%). Due to blood sampling, Hct fell to 40 -t 2% (35-44%) within 3 h following the experiments and 36 f 2% (33-39%) following 6-h experiments. Except for the experiments with hypercapnia, P,CO, was held constant at 35.7 + 2.1 mmHg, arterial pH at 7.44 f 0.02, and P,O, at 125 k 10 mmHg.
The effect of continuous superfusion of the brain surface with buffered artificial CSF for 5 h at a rate of 5 ml/h was studied to assessthe constancy of both vasomotor homeostasis and BBB. Extravasation of Na+-fluorescein from the vessels was not observed. The arteriolar diameters were not affected until 3.0 h following the beginning of control measurements, although from 3.5 h, the arterioles showed gradually progressive dilation; they dilated + 20 f 16% at 5 h (Fig. 4a). The diameters of venules remained constant throughout the entire period (Fig. 4b). The response of pial vessels to CO, inhalation was studied in 10 experiments. P,CO, increased from 39-l+2.4 to 56.3 f 2.4 mmHg with an accompanying pH decrease from 7.40 + 0.02 to 7.27 f 0.03, while P,O, (120 +12 mmHg before and 119 f 12 mmHg after) and MABP (97 f 10 mmHg before and 96 I~I 10 mmHg after) were unchanged. Arterioles dilated + 19.3 + 16.0% per 17.2 mmHg P,CO, increase (mean, l.l%/mmHg). As shown in Fig. 5a, the small arterioles had a greater dilation than the large vessels. According to regression analysis, dilation of the arterioles was inversely related to the vessel diameter. Arterioles with diameters > 40 pm dilated + 0.5 to + 0.9% per mmHg P,CO, increase, while arterioles < 40 pm 4 dilated + 1.0 to + 2.5%/mmHg. On
250
KAWAMURA
AND YASUI
BBB AND VASOMOTOR -‘50-
RESPONSE -lm-
a
a ~~,oo-.-~-~-~-~-~-~~~~~~~
_
* : pcO.05 VS. Resting Diameter
•++++++/~~+
6
50-
5oL 1 0
FIG. means
b
e Z t B
z 5& u
(40m
251
IN RAT
I 1
I 2
4. Effect of continuous superfusion f SD. t P < 0.05 vs resting diameters
I 3
I Tie
I 5 1 hour 1
of brain surface with (Dunnett’s multiple
the other hand, venules had no significant hypercapnia (Fig. 5b).
1 0 buffered artificial comparison).
response to
DISCUSSION
Closed cranial window techniques benefit from a maintenance of the intracranial physiology: a CO, loss from the brain and CSF into ambient air is prevented, and ICP can be controlled under various conditions (10). Brain herniation through the created skull defect is prevented also. The cranial window technique used in this study was established to study the BBB function and the vasomotor response simultaneously in rat brain microcirculation. Opening of the skull and dura, as well as closure of the window, has been performed under an oil column before. This prevents the brain swelling out of the skull defect. Thus, this technique makes it possible to preserve a normal BBB function while allowing the presence of the small molecular-weight indicator (Na+-fluorescein). Olesen (11) has demonstrated that opening of the skull and dura under atmospheric pressure conditions induces BBB disruption in rats. However, evidence observed by histochemical techniques suggests that the BBB in rodents under normal conditions is permeable to blood-borne peroxidase, indicating that the BBB is not “absolute” (2, 3). The methods to observe the BBB function in our study may be less sensitive than the peroxidase histochemical techniques (2, 3), and therefore, it is possible that we could not detect the leakiness of pial surface vessels to the exogenous BBB tracer.
CSF
1 on 33 arterioles
2
3
I I 4 5 Time [ hour ]
(a) and 43 venules
(b). Bars indicate
Exposure of the brain to extremely high light intensity in the presence of a high concentration of Na+-fluorescein may cause damage to the endothelium of the pial vessels, as manifested by platelet aggregation and vascular leakage (18, 19). Similar findings were observed in other vascular beds (13). Therefore, when we made the new fluorescence microscope, we planned to reduce the light intensity and the dose of the fluorescent dye to as little as possible. We used the 20X objective lens for quantitative assessments of the vasomotor response (light intensity, 2.5 mW/cm2). Using a low-light TV camera with high sensitivity, the total amount of 2% Na+-fluorescein used was reduced to ~0.7 ml for 6 h (mean + SD, 37 + 10 pl/lOO g body wt/hr). Under these conditions, we never observed either formation of platelet aggregates, or a BBB opening to Na+-fluorescein. When viewed on microphotographs, smaller venules appear to have fluorescing platelet aggregates. However, we want to stress that this appearance actually indicates a back-and-forth movement of the venous blood flow that is normally seen in venules (5) when seen on the TV monitor. Previously, Rosenblum and El-Sabban (14) utilized light plus a bolus iv injection of Na+-fluorescein in order to induce endothelium damage in mice. They used a dose of the dye that was approximately 22 times more and a light intensity at least 560 times greater than that used in our study. The anesthetic agents used in our study may have a significant effect upon cerebral microcirculation, although we believe that they have minor problems, compared with light and dye.
FIG. 2. Fluorescence microphotographs showing a physiological venous they are not actually the aggregates. Bars indicate 100 pm (left) and 50 pm FIG. 3. Hypertonic opening of the blood-brain barrier. Bars indicate 100 the intravascular space, indicating intact BBB. (b) Twenty minutes after the observed, showing complete BBB disruption.
blood stasis. There seem to be fluorescing platelet aggregates, (right). Fm. (a) Under control conditions, Na+-fluorescein is restricted superfusion. Diffise extravasation of the dye into parenchyma
but to is
KAWAMURA
252
AND YASUI =200; 180
0
20
40
60
80
100
120
140
Control
Diamrtrr
FIG. 5. COx reactivity in 228 arterioles (a) and 174 venules in the diameters of each vessel during hypercapnia are plotted
160
’
b
‘0
20
40
60
80
100
urn)
(b). Control diameters on the ordinate.
The superfusion of the brain surface with hyperosmolar mannitol solutions with an osmolarity of 1000 mOsm/liter induced arteriolar dilation. Then, extravasation of Na+-fluorescein began, starting from a venule. These findings concur with observations by Wahl et al. (21) in cats. Perivascular osmolarity undoubtedly plays a major role in the arteriolar dilation (20), while a slight decrease in the pH of the solutions may play a minor role (19). The mannitol-CSF solution with an osmolarity of 2000 mOsm/liter has been used previously (7). In this (present) study, we used a lower osmolar solution in order to study any differences. Five minutes following superfusion with the higher osmolar solution, extravasation of the fluorescence indicator immediately occurred (7); the BBB disruption started about 10 min later, progressed slowly, and became complete after 15 min following the superfusion with the lower osmolar solution (shown in this study). Taken together, the hyperosmolar BBB opening with mannitol may be considered dose dependent. Because of the slowly progressing BBB opening, we have been able to observe the primary site of the BBB disruption. The reason why venules are the primary site of the BBB disruption is unclear. The initial extravasation during acutely induced hypertension takes place in the small venules probably due either to venular congestion by an overload of the venous outflow (l), or because of increases in the pial venous pressure (8). However, the mechanism of the BBB disruption may be different during superfusion with hyperosmolar solutions and during acute hypertension (8). The injurious potential of exposing the cerebral surface to the excitatory light, which could affect the BBB and the vasomotor response, has been reported by Wahl et al. (21) who observed extravasation and arteriolar dilation after continuous light exposure (21 mW/cm2). We, however, are the first to observe the sequential changes in diameters of rat pial vessels which were exposed by light (2.5 mW/cm2) every 30 min for 5.0 h. The
are indicated
on the abscissa,
120
140
Control
Diameter
and percentage
160 (pm)
of changes
arteriolar diameters remained constant for 3.0 h and this reconfirmed the results of Schiirer et al. (17). We found that the arteriolar diameters began a gradually progressive increase after 3.5 h. The latter vasomotor response is probably due to a toxic effect caused by light exposure to brain microcirculation. Our observations of CO, reactivity in the pial vessels concur with previous studies (7), as well as with the results of others who worked with cats (12) and rats (9). Thus, we have reconfirmed that the closed cranial window preparation (7) causes a physiological vasomotor response in pial arterioles and venules when CO, inhalation occurs. In conclusion, the microscope system and the small animal model used in this study are useful in studying the blood-brain barrier function and the vasomotor response simultaneously under various experimental conditions. However, the arterioles did dilate progressively after 3.5 h, suggesting an injurious potential of light exposure on the brain microcirculation. ACKNOWLEDGMENTS The technical assistance of Yozo Ito secretarial assistance of Kimio Yoshioka gratefully acknowledged.
and Ryoetsu and Yoshitaka
Sato and Tozawa
the are
REFERENCES AUER, L. 1978. Origin and localization of Evans blue extravasations in acutely-induced hypertension in cats. Eur. Neurol. 17: 211-215. BALIN, B. J., R. D. BROADWELL, M. SALCMAN, AND M. EL-KALLINY. 1986. Avenues for entry of peripherally administered protein to the central nervous system in mouse, rat, and squirrel monkey. J. Comp. Neurol. 251: 260-280. BROADWELL, R. D. 1989. Transcytosis of macromolecules through the blood-brain barrier: A cell biological perspective and critical appraisal. Acta Neuropathol. 79: 117-128. ELLIS, E. F., E. P. WEI, C. S. COCKFULL, AND H. A. KONTOS.
BBB
AND
VASOMOTOR
1983. The effects of PGF, on in uiuo cerebral arteriolar diameter in cats and rats. Prostaglmdins 26: 917-923. 5. FORBES, H. S. 1928. The cerebral circulation. I. Observation and measurement of pial vessels. Arch. Neural. Psychiatry 19: 751761. 6. JOHANSSON, B. B., L. M. AUER, AND I. SAYAMA. 1985. Reaction of pial arteries and veins to hypercapnia in hypertensive and normotensive rats. Stroke 16: 320-323. 7. KAWAMURA, S., L. SCH~~RER, A. GOETZ, 0. KEMPSKI, B. SCHMUCKER, AND A. BAETHMANN. 1990. An improved closed cranial window technique for investigation of blood-brain barrier function and cerebral vasomotor control in the rat. Int. J. Microcirc. Clin.
Exp.
9: 369-383.
8. MAYHAN, W. G., AND D. D. HEISTAD. 1986. Role of veins and cerebral venous pressure in disruption of the blood-brain barrier. Circ. Res. 59: 216-220. 9. MORII, S., A. C. NGAI, AND H. R. WINN. 1986. Reactivity of rat pial arterioles and venules to adenosine and carbon dioxide: With detailed description of the closed cranial window technique in rats. J. Cereb. Blood Flow Metab. 6: 34-41. 10. NAVARI, R. M., E. P. WEI, H. A. KONTOS, AND J. L. PATTERSON, JR. 1978. Comparison of the open skull and cranial window preparations in the study of the cerebral microcirculation. Microvasc. Res. 16:304-315. 11. OLESEN, S-P. 1987. Leakiness of rat brain microvessels to fluorescent probes following craniotomy. Acta Physiol. Stand. 130: 63-68. 12. RAPER, A. J., H. A. KONTOS, AND J. L. PATTERSON, JR. 1971. Response of pial precapillary vessels to changes in arterial carbon dioxide tension. Circ. Res. 28: 518-523.
RESPONSE
IN
RAT
253
13. REED, M. W. R., AND F. N. MILLER. 1988. Importance of light dose in fluorescent microscopy. Microuasc. Res. 36: 104-107. 14. ROSENBLUM, W. I., AND F. EL-SABBAN. 19’77. Platelet aggregation in the cerebral microcirculation: Effect of aspirin and other agents. Circ. Res. 40: 320-328. 15. ROSENBLUM, W. I. 1978. Fluorescence induced in platelet aggregates as a guide to luminal contours in the presence of platelet aggregation. Microvase. Res. 15: 103-106. 16. SCH~~RER,L., S. KAWAMURA, B. SCHMUCKJXR, AND A. BAETHMANN. 1988. Vasomotoric and blood-brain barrier effects of rat brain surface vessels during hypercapnia and superfusion with Na+-arachidonate. Int. J. Microcirc. Clin. Exp. 7: S82. 17. SCH~R, L., S. KAWAMIJRA, B. SCHMUCKER, A. GOETZ, AND A. BAJXTHMANN. 1989. Vasomotor response and blood-brain barrier function of rat brain studied in a closed cranial window preparation. In Intracranial Pressure VII (J. T. Hoff, and A. L. Betz, Eds.), pp. 947-949. Springer-Verlag, Berlin. 18. UNTERBERG, A., M. WAHL, AND A. BAETHMANN. 1988. Studies of pial vascular permiability. Stroke 19: 1306-1307. 19. WAHL, M., P. DEETJEN, K. THURAU, D. H. INGVAR, AND N. A. LASSEN. 1970. Micropuncture evaluation of the importance of perivascular pH for the arteriolar diameter on the brain surface. PflzXgersArch. 316:152-163. WAHL, M., W. KUSCHINSKY, 0. BOSSE, AND K. THURAU. 1973. 20. Dependency of pial arterial and arteriolar diameter on perivascular osmolarity in the cat. Circ. Res. 32: 162-169. 21. WAHL, M., A. UNTERBERG, AND A. BAETHMANN. 1985. Intravital fluorescence microscopy for the study of blood-brain-barrier function. Int. J. Microcir. Clin. Exp. 4: 3-18.
NEUROLOGY
117,
247-253 (19%)
Observation of the Blood-Brain Barrier Function and Vasomotor Response in Rat Microcirculation Using lntravital Fluorescence Microscopy SHINGO Department of Surgical
KAWAMDRA
AND NOBUYUKI
YASUI
Neurology, Research Institute for Brain and Blood Vessels-AKZTA,
Akita 010, Japan
aim of this paper is to show that the experimental animals had normal BBB functions as well as physiological vasomotor responses in the pial microvessels. Na+-fluorescein (MW 376) was used as a BBB and vessel marker.
The blood-brain barrier function and the vasomotor response in rat microcirculation was studied using a closed cranial window technique and an intravital fluorescence microscope with an attempt to reduce both the light intensity and the amount of 2% Na+-fluorescein to as little as possible. The light intensities in the focus were ~2.6 mW/cm’, and with the low-light TV camera we were able to reduce the total amount of the dye to ~0.7 ml during 6-h experiments. Observing the brain surface every 30 min showed that the barrier function remained intact for up to 6.0 h following the first iv administration of the dye. This study has reconfirmed that the pial vessels have physiological vasomotor responses (e.g., CO2 reactivity). The osmotic barrier disruption started at a small venule 10 min after the start of superfusion with mannitol (1000 mOsm/liter). In conclusion, the microscope system and the small animal model used in this study were useful in studying the blood-brain barrier function and the vasomotor response simultaneously under various experimental . conditions. o 1992 Academic press, IIN.
MATERIALS
AND
METHODS
General Preparation Twenty-two male Sprague-Dawley rats (250-300 g) were used. Anesthesia was induced ‘by 4% halothane, which was reduced to 1.5% during catheter insertion into a tail artery and bilateral femoral veins. The animals were first allowed to breathe spontaneously with a nitrous oxide/oxygen mixture (70/30%). Following tracheotomy, cr-chloralose (120 mg/kg, iv) was administered, and then halothane and nitrous oxide were discontinued. Supplemental doses of a-chloralose (15-30 mg/kg) were given to maintain anesthesia when necessary. The animals were immobilized with pancronium bromide (1.2 mg/kg/h, iv) and ventilated artificially with room air and supplemental oxygen. Ventilation was adjusted to maintain arterial blood pH at 7.42 f 0.03 (mean & SD; range, 7.37-7.47), P,CO, at 38.3 f 2.4 (33.7-42.6) mmHg, and P,O, at 124 +- 11 (107-143) mmHg. Arterial blood was collected anaerobically for blood-gas analyses and hematocrit measurements. Mean arterial blood pressure (MABP) in the tail artery, ICP, and air-way pressure were recorded continuously. Fluid and drugs were administered into the left femoral vein, and Na+-fluorescein was administered into the right femoral vein. Rectal temperature was maintained at 37.5”C by heating.
INTRODUCTION Rat microcirculation has been studied previously under a closed cranial window because of advantages both in using small animals and in the preservation of intracranial physiology (lo), although only a few investigators have used rats (4,6). Details of a closed cranial window technique in rats were described by Morii et al. (9). This technique was modified by Baethmann and his co-workers (16) in studies of both the blood-brain barrier (BBB) function and the vasomotor control in association with measurements or manipulations of intracranial pressure (ICP). Improved experimental procedures have allowed the prevention of BBB disruption. An intact BBB is indispensible in studies on the barrier function and on vasomotor reactivity of brain microcirculation under physiological and pathological conditions. Recently, we have used an improved intravital fluorescence microscope to investigate rat microcirculation using the modified closed cranial window technique. The
Operative
Technique
Details have been described previously (7), but will be mentioned briefly. A 4-cm skin incision was made from the forehead to the neck. A window (5 x 6 mm) was made over the left parietal brain using a dental drill, and a thin bone layer over the dura was left intact. Three catheters (PE 50) were attached to the window for superfusion of the brain surface and for ICP measurements. They were fixed in dental cement. A plexiglass funnel was fixed to the skin above the window and filled 247 All
Copyright 0 1992 rights of reproduction
0014-4886/92 $5.00 by Academic Press, Inc. in any form reserved.
248
KAWAMURA
AND
YASUI
with paraffin oil. This exerted a pressure of 3 mmHg on the brain surface. Under the oil, the remaining bone layer of the window was removed, and the dura was opened. Following infusion of a small amount of artificial cerebrospinal fluid (CSF) over the brain, the window was closed with a glass cover (13 mm 4) with cyanoacrylate. The funnel was then removed, and the distal end of the outflow catheter of the window was adjusted to a height that maintained an ICP of 5-7 mmHg. Four animals (18%) were discarded from this study due to technical failures in the preparation. Superfusion
with Artificial
The cortical surface was superfused with artificial CSF at a rate of 5 ml/h. The CSF was composed as follows (mM): glucose, 3.3; Na+, 158; K+, 3.2; Ca’+, 1.5; Cl-, 142; Mg2+, 1.33; and HC03-, 24.5, yielding an osmolality of 306 mOsm/liter. Solutions were prepared freshly prior to experiments and bubbled with a mixture of humidified CO, (6%), 0, (lo%), with the balance of N, at 37.5”C, yielding a pH of 7.34 f 0.01 (range, 7.327.37), a PCO, of 44.8 f 1.1 (42.7-47.5) mmHg, and a PO, of 88 -t 4 (80-95) mmHg. In studies on the hypertonic opening of the BBB using a method described previously (7), mannitol (D-Mannit, Kanto Chemical Co., Inc., Tokyo, Japan) was added to the CSF at an osmolality of 1000 mOsm/liter. An infusion pump (Instamatic, B. Braun Melsungen, Germany) was used with an airtight glass syringe as a reservoir. Intravital
Fluorescence
Video
CSF
Microscopy
Intravital fluorescence microscopy was carried out using the fluorescence video-photo-microscope K90 system (Fig. l), consisting of the vertical fluorescence illuminator BH2-RFC and the trinocular tube BH2TR30 (Kyoei-Olympus, Tokyo, Japan). Continuous epiillumination of the brain surface was made only during observation (lo-15 min for each observation). The light source was a 100 W high-pressure mercury lamp attached to the illuminator. Light intensity was reduced through two sets of ND-25 filters (Kyoei-Olympus). The band-pass filter BHB-DMB (Kyoei-Olympus) was used to obtain an excitation source with a wavelength of 455-490 nm. The fluorescence emission of the brain surface, after iv injection of 2% Na+-fluorescein solution, was studied. For visual observation and taking photographs, wavelengths below 515 nm were excluded. Ten ~1 of Na+-fluorescein were administered for a total of 0.3-0.7 ml (23-64 pl/lOO g body wt/hr) during each experiment. Objective lenses of 2X, 10X, and 20X with long working distances (0.08, 0.30, and 0.40 numerical aperture, respectively) were used. These lenses, in association with the trinocular tube, gave magnifications of 6.25X to 62.5X. For quantitative studies of vasomotor
Recorder
+
Lz!G!I= Brain Surface
FIG. 1. Schematic photo-microscope K90
presentation system.
of
the
fluorescence
video-
responses, the 20x objective lens was used to measure internal diameters of pial vessels. The images were processed with a video timer (VTG-33, For.A Company, Ltd., Tokyo, Japan) and recorded on a videotape by a S-VHS videocassette recorder (HR-SlOOOO, Victor, Tokyo, Japan) at a rate of 30 frames/set. A low-light TV camera (C2741, Hamamatsu Photonics, Hamamatsu, Japan) and a TV monitor (VM-RBOOS, Victor) provided final magnifications of 60X to 600X. Regarding past studies on the BBB function, a different form of fluorescence microscopical equipment was used previously (7). In this (present) study, a photographic camera was attached to the trinocular tube in order to perform video recording and picture taking simultaneously. The microphotographs of the brain surface were taken on Kodak Ektachrome P800/1600 35mm film using a C-35AD-4 camera with an automatic exposure control unit PM-CBSP (Kyoei-Olympus). Lack of extravasation of Na+-fluorescein from pial vessels into the surrounding parenchyma was first confirmed 30 min after the first bolus of the dye, and all experiments were then started. The light intensities at the point of the focus were measured in vitro using a detector (Spectra Radio-meter USR-BOA, Ushio, Inc., Tokyo, Japan), with measurements as follows: 0.019,0.88, and 2.5 mW/cm’ at magnifications of 2X, 10X, and 20X, respectively.
BBB AND VASOMOTOR
249
RESPONSE
IN RAT
Studies of Vasomotor Response
Blood-Brain
Barrier
Vasomotor response was studied off-line. Drawings of the vascular network on the brain surface were made on tracing papers, by hand, from microphotographs taken through the 2~ objective lens. Manual measurements of the diameters of randomly selected arterioles and venules were made by way of the TV monitor, using sliding calipers (resolution, 0.05 mm). All measured locations were marked on the drawings in order to provide source material for independent investigators. In three animals, the effect of continuous superfusion of pial vessels with artificial CSF was studied for 5 h. The vessels were observed every 30 min, and sequential changes in the diameters of 63 arterioles (31 f 14 pm; range, 12-69 pm) and 49 venules (35 f 18 pm; range, 15-97 pm) were evaluated. Five animals were used to assessthe CO, reactivity of pial vessels. CO, was put into the respirator 5 min prior to observation. The CO, inhalation process was performed twice for each animal, and the time used for each experiment was 1.5-2.5 h. CSF superfusion was interrupted during the CO, inhalation. Arterial blood analyses were repeated during both normo- and hypercapnic states. Inspired CO, content was monitored using a CO, analyzer (Respina-lH31, San-ei, Tokyo, Japan) and reached 5.0-5.2% during hypercapnia. CO, reactivity was studied in 228 arterioles (32 f 20 pm; range, 8-111 pm) and 174 veins (44 + 28 ym; range, lo-152 pm).
The BBB function was studied in a total of 18 rats for up to 6.0 h following the first administration of Na+-fluorescein; 5 animals were studied for 6.0 h. Extravasation of the marker was not observed during the entire observation period, demonstrating the maintenance of a normal BBB function. The tightness of the barrier can be demonstrated by fluorescence microphotographs taken at the end of each experiment. Neither the adherence of platelets or of other blood cells to the luminal surface, nor the presence of cell aggregates in vessels, is visible. Back-and-forth movements (or stasis) of the blood flow in the smaller venules of all the animals were often observed as being the physiological blood stream in branch venules (5). When seen in photographs (Fig. 2), the physiological stasis of venous blood may look like brightly fluorescing platelet aggregates (14, 15). In testing of the osmotic BBB opening, seven animals were administered with hypertonic mannitol (1000 mOsm/liter) dissolved in artificial CSF. The pH of the solutions was 7.28 + 0.02, the PCO, was 44.7 * 1.0 mmHg, and the POZ was 93 -t 2 mmHg. The pial arterioles dilated markedly within 5 min following the start of the mannitol superfusion (not shown). The BBB began to disrupt at a small venule approximately 10 min after the start of superfusion. The gross and diffuse opening of the barrier was established 5-10 min later (Fig. 3).
Vascular Reactivity Statistical
Analysis
The Student t test was used for a comparison of the paired physiological data. Regression analysis was performed to compare control diameters of vessels with the percentage of change in diameter during CO, inhalation. When the effect of continuous CSF superfusion on vessel diameters was analyzed, the diameter of each vessel was compared with its diameter while at rest, using Dunnett’s multiple comparison (randomized block design). A P level < 0.05 was considered significant.
RESULTS
Physiological
Data
MABP had a range of 81-111 (mean f SD, 95 + 7) mmHg in all animals throughout the experiments. ICP was between 5-7 mmHg. The initial hematocrit (Hct) was 42 f 2% (range, 37-45%). Due to blood sampling, Hct fell to 40 -t 2% (35-44%) within 3 h following the experiments and 36 f 2% (33-39%) following 6-h experiments. Except for the experiments with hypercapnia, P,CO, was held constant at 35.7 + 2.1 mmHg, arterial pH at 7.44 f 0.02, and P,O, at 125 k 10 mmHg.
The effect of continuous superfusion of the brain surface with buffered artificial CSF for 5 h at a rate of 5 ml/h was studied to assessthe constancy of both vasomotor homeostasis and BBB. Extravasation of Na+-fluorescein from the vessels was not observed. The arteriolar diameters were not affected until 3.0 h following the beginning of control measurements, although from 3.5 h, the arterioles showed gradually progressive dilation; they dilated + 20 f 16% at 5 h (Fig. 4a). The diameters of venules remained constant throughout the entire period (Fig. 4b). The response of pial vessels to CO, inhalation was studied in 10 experiments. P,CO, increased from 39-l+2.4 to 56.3 f 2.4 mmHg with an accompanying pH decrease from 7.40 + 0.02 to 7.27 f 0.03, while P,O, (120 +12 mmHg before and 119 f 12 mmHg after) and MABP (97 f 10 mmHg before and 96 I~I 10 mmHg after) were unchanged. Arterioles dilated + 19.3 + 16.0% per 17.2 mmHg P,CO, increase (mean, l.l%/mmHg). As shown in Fig. 5a, the small arterioles had a greater dilation than the large vessels. According to regression analysis, dilation of the arterioles was inversely related to the vessel diameter. Arterioles with diameters > 40 pm dilated + 0.5 to + 0.9% per mmHg P,CO, increase, while arterioles < 40 pm 4 dilated + 1.0 to + 2.5%/mmHg. On
250
KAWAMURA
AND YASUI
BBB AND VASOMOTOR -‘50-
RESPONSE -lm-
a
a ~~,oo-.-~-~-~-~-~-~~~~~~~
_
* : pcO.05 VS. Resting Diameter
•++++++/~~+
6
50-
5oL 1 0
FIG. means
b
e Z t B
z 5& u
(40m
251
IN RAT
I 1
I 2
4. Effect of continuous superfusion f SD. t P < 0.05 vs resting diameters
I 3
I Tie
I 5 1 hour 1
of brain surface with (Dunnett’s multiple
the other hand, venules had no significant hypercapnia (Fig. 5b).
1 0 buffered artificial comparison).
response to
DISCUSSION
Closed cranial window techniques benefit from a maintenance of the intracranial physiology: a CO, loss from the brain and CSF into ambient air is prevented, and ICP can be controlled under various conditions (10). Brain herniation through the created skull defect is prevented also. The cranial window technique used in this study was established to study the BBB function and the vasomotor response simultaneously in rat brain microcirculation. Opening of the skull and dura, as well as closure of the window, has been performed under an oil column before. This prevents the brain swelling out of the skull defect. Thus, this technique makes it possible to preserve a normal BBB function while allowing the presence of the small molecular-weight indicator (Na+-fluorescein). Olesen (11) has demonstrated that opening of the skull and dura under atmospheric pressure conditions induces BBB disruption in rats. However, evidence observed by histochemical techniques suggests that the BBB in rodents under normal conditions is permeable to blood-borne peroxidase, indicating that the BBB is not “absolute” (2, 3). The methods to observe the BBB function in our study may be less sensitive than the peroxidase histochemical techniques (2, 3), and therefore, it is possible that we could not detect the leakiness of pial surface vessels to the exogenous BBB tracer.
CSF
1 on 33 arterioles
2
3
I I 4 5 Time [ hour ]
(a) and 43 venules
(b). Bars indicate
Exposure of the brain to extremely high light intensity in the presence of a high concentration of Na+-fluorescein may cause damage to the endothelium of the pial vessels, as manifested by platelet aggregation and vascular leakage (18, 19). Similar findings were observed in other vascular beds (13). Therefore, when we made the new fluorescence microscope, we planned to reduce the light intensity and the dose of the fluorescent dye to as little as possible. We used the 20X objective lens for quantitative assessments of the vasomotor response (light intensity, 2.5 mW/cm2). Using a low-light TV camera with high sensitivity, the total amount of 2% Na+-fluorescein used was reduced to ~0.7 ml for 6 h (mean + SD, 37 + 10 pl/lOO g body wt/hr). Under these conditions, we never observed either formation of platelet aggregates, or a BBB opening to Na+-fluorescein. When viewed on microphotographs, smaller venules appear to have fluorescing platelet aggregates. However, we want to stress that this appearance actually indicates a back-and-forth movement of the venous blood flow that is normally seen in venules (5) when seen on the TV monitor. Previously, Rosenblum and El-Sabban (14) utilized light plus a bolus iv injection of Na+-fluorescein in order to induce endothelium damage in mice. They used a dose of the dye that was approximately 22 times more and a light intensity at least 560 times greater than that used in our study. The anesthetic agents used in our study may have a significant effect upon cerebral microcirculation, although we believe that they have minor problems, compared with light and dye.
FIG. 2. Fluorescence microphotographs showing a physiological venous they are not actually the aggregates. Bars indicate 100 pm (left) and 50 pm FIG. 3. Hypertonic opening of the blood-brain barrier. Bars indicate 100 the intravascular space, indicating intact BBB. (b) Twenty minutes after the observed, showing complete BBB disruption.
blood stasis. There seem to be fluorescing platelet aggregates, (right). Fm. (a) Under control conditions, Na+-fluorescein is restricted superfusion. Diffise extravasation of the dye into parenchyma
but to is
KAWAMURA
252
AND YASUI =200; 180
0
20
40
60
80
100
120
140
Control
Diamrtrr
FIG. 5. COx reactivity in 228 arterioles (a) and 174 venules in the diameters of each vessel during hypercapnia are plotted
160
’
b
‘0
20
40
60
80
100
urn)
(b). Control diameters on the ordinate.
The superfusion of the brain surface with hyperosmolar mannitol solutions with an osmolarity of 1000 mOsm/liter induced arteriolar dilation. Then, extravasation of Na+-fluorescein began, starting from a venule. These findings concur with observations by Wahl et al. (21) in cats. Perivascular osmolarity undoubtedly plays a major role in the arteriolar dilation (20), while a slight decrease in the pH of the solutions may play a minor role (19). The mannitol-CSF solution with an osmolarity of 2000 mOsm/liter has been used previously (7). In this (present) study, we used a lower osmolar solution in order to study any differences. Five minutes following superfusion with the higher osmolar solution, extravasation of the fluorescence indicator immediately occurred (7); the BBB disruption started about 10 min later, progressed slowly, and became complete after 15 min following the superfusion with the lower osmolar solution (shown in this study). Taken together, the hyperosmolar BBB opening with mannitol may be considered dose dependent. Because of the slowly progressing BBB opening, we have been able to observe the primary site of the BBB disruption. The reason why venules are the primary site of the BBB disruption is unclear. The initial extravasation during acutely induced hypertension takes place in the small venules probably due either to venular congestion by an overload of the venous outflow (l), or because of increases in the pial venous pressure (8). However, the mechanism of the BBB disruption may be different during superfusion with hyperosmolar solutions and during acute hypertension (8). The injurious potential of exposing the cerebral surface to the excitatory light, which could affect the BBB and the vasomotor response, has been reported by Wahl et al. (21) who observed extravasation and arteriolar dilation after continuous light exposure (21 mW/cm2). We, however, are the first to observe the sequential changes in diameters of rat pial vessels which were exposed by light (2.5 mW/cm2) every 30 min for 5.0 h. The
are indicated
on the abscissa,
120
140
Control
Diameter
and percentage
160 (pm)
of changes
arteriolar diameters remained constant for 3.0 h and this reconfirmed the results of Schiirer et al. (17). We found that the arteriolar diameters began a gradually progressive increase after 3.5 h. The latter vasomotor response is probably due to a toxic effect caused by light exposure to brain microcirculation. Our observations of CO, reactivity in the pial vessels concur with previous studies (7), as well as with the results of others who worked with cats (12) and rats (9). Thus, we have reconfirmed that the closed cranial window preparation (7) causes a physiological vasomotor response in pial arterioles and venules when CO, inhalation occurs. In conclusion, the microscope system and the small animal model used in this study are useful in studying the blood-brain barrier function and the vasomotor response simultaneously under various experimental conditions. However, the arterioles did dilate progressively after 3.5 h, suggesting an injurious potential of light exposure on the brain microcirculation. ACKNOWLEDGMENTS The technical assistance of Yozo Ito secretarial assistance of Kimio Yoshioka gratefully acknowledged.
and Ryoetsu and Yoshitaka
Sato and Tozawa
the are
REFERENCES AUER, L. 1978. Origin and localization of Evans blue extravasations in acutely-induced hypertension in cats. Eur. Neurol. 17: 211-215. BALIN, B. J., R. D. BROADWELL, M. SALCMAN, AND M. EL-KALLINY. 1986. Avenues for entry of peripherally administered protein to the central nervous system in mouse, rat, and squirrel monkey. J. Comp. Neurol. 251: 260-280. BROADWELL, R. D. 1989. Transcytosis of macromolecules through the blood-brain barrier: A cell biological perspective and critical appraisal. Acta Neuropathol. 79: 117-128. ELLIS, E. F., E. P. WEI, C. S. COCKFULL, AND H. A. KONTOS.
BBB
AND
VASOMOTOR
1983. The effects of PGF, on in uiuo cerebral arteriolar diameter in cats and rats. Prostaglmdins 26: 917-923. 5. FORBES, H. S. 1928. The cerebral circulation. I. Observation and measurement of pial vessels. Arch. Neural. Psychiatry 19: 751761. 6. JOHANSSON, B. B., L. M. AUER, AND I. SAYAMA. 1985. Reaction of pial arteries and veins to hypercapnia in hypertensive and normotensive rats. Stroke 16: 320-323. 7. KAWAMURA, S., L. SCH~~RER, A. GOETZ, 0. KEMPSKI, B. SCHMUCKER, AND A. BAETHMANN. 1990. An improved closed cranial window technique for investigation of blood-brain barrier function and cerebral vasomotor control in the rat. Int. J. Microcirc. Clin.
Exp.
9: 369-383.
8. MAYHAN, W. G., AND D. D. HEISTAD. 1986. Role of veins and cerebral venous pressure in disruption of the blood-brain barrier. Circ. Res. 59: 216-220. 9. MORII, S., A. C. NGAI, AND H. R. WINN. 1986. Reactivity of rat pial arterioles and venules to adenosine and carbon dioxide: With detailed description of the closed cranial window technique in rats. J. Cereb. Blood Flow Metab. 6: 34-41. 10. NAVARI, R. M., E. P. WEI, H. A. KONTOS, AND J. L. PATTERSON, JR. 1978. Comparison of the open skull and cranial window preparations in the study of the cerebral microcirculation. Microvasc. Res. 16:304-315. 11. OLESEN, S-P. 1987. Leakiness of rat brain microvessels to fluorescent probes following craniotomy. Acta Physiol. Stand. 130: 63-68. 12. RAPER, A. J., H. A. KONTOS, AND J. L. PATTERSON, JR. 1971. Response of pial precapillary vessels to changes in arterial carbon dioxide tension. Circ. Res. 28: 518-523.
RESPONSE
IN
RAT
253
13. REED, M. W. R., AND F. N. MILLER. 1988. Importance of light dose in fluorescent microscopy. Microuasc. Res. 36: 104-107. 14. ROSENBLUM, W. I., AND F. EL-SABBAN. 19’77. Platelet aggregation in the cerebral microcirculation: Effect of aspirin and other agents. Circ. Res. 40: 320-328. 15. ROSENBLUM, W. I. 1978. Fluorescence induced in platelet aggregates as a guide to luminal contours in the presence of platelet aggregation. Microvase. Res. 15: 103-106. 16. SCH~~RER,L., S. KAWAMURA, B. SCHMUCKJXR, AND A. BAETHMANN. 1988. Vasomotoric and blood-brain barrier effects of rat brain surface vessels during hypercapnia and superfusion with Na+-arachidonate. Int. J. Microcirc. Clin. Exp. 7: S82. 17. SCH~R, L., S. KAWAMIJRA, B. SCHMUCKER, A. GOETZ, AND A. BAJXTHMANN. 1989. Vasomotor response and blood-brain barrier function of rat brain studied in a closed cranial window preparation. In Intracranial Pressure VII (J. T. Hoff, and A. L. Betz, Eds.), pp. 947-949. Springer-Verlag, Berlin. 18. UNTERBERG, A., M. WAHL, AND A. BAETHMANN. 1988. Studies of pial vascular permiability. Stroke 19: 1306-1307. 19. WAHL, M., P. DEETJEN, K. THURAU, D. H. INGVAR, AND N. A. LASSEN. 1970. Micropuncture evaluation of the importance of perivascular pH for the arteriolar diameter on the brain surface. PflzXgersArch. 316:152-163. WAHL, M., W. KUSCHINSKY, 0. BOSSE, AND K. THURAU. 1973. 20. Dependency of pial arterial and arteriolar diameter on perivascular osmolarity in the cat. Circ. Res. 32: 162-169. 21. WAHL, M., A. UNTERBERG, AND A. BAETHMANN. 1985. Intravital fluorescence microscopy for the study of blood-brain-barrier function. Int. J. Microcir. Clin. Exp. 4: 3-18.