Two components of blood-brain barrier disruption in the rat


  • A. S. Easton,

    1. Vascular Biology Research Centre, Physiology Group, Biomedical Sciences Division, King's College London, Campden Hill Road, London W8 7AH, UK
    Search for more papers by this author
  • M. H. Sarker,

    1. Vascular Biology Research Centre, Physiology Group, Biomedical Sciences Division, King's College London, Campden Hill Road, London W8 7AH, UK
    Search for more papers by this author
  • P. A. Fraser

    Corresponding author
    1. Vascular Biology Research Centre, Physiology Group, Biomedical Sciences Division, King's College London, Campden Hill Road, London W8 7AH, UK
    • To whom correspondence should be addressed.

    Search for more papers by this author


  • 1Permeability of pial venular capillaries to Lucifer Yellow (Ply) was measured using the single microvessel occlusion technique.
  • 2P ly was extremely low, when measured shortly after the removal of the meninges, consistent with an intact blood-brain barrier, but rose spontaneously to (1.65 ± 060)× 10−6cms−1 (mean ±s.d.) within 20–-60 min. This first phase of spontaneous disruption lasted 44–164 min. A second phase started when Ply rose sharply, and was characterized by rapid permeability fluctuations with a mean of (12.31 ± 15.14) × 1CT−6 cm s−1.
  • 3The first phase could be mimicked by applying the divalent cation ionophore A23187 in the presence of Ca2+, when Ply rose by (1.47±0–25) × 10−6 cm s−1 (mean±s.e.m.). Application of histamine (10 μm) to tight vessels increased Ply by (2.41 ± 0.22) × 10−6 cm s−1.
  • 4Substances that raised intraendothelial cAMP of vessels during the first phase of disruption reduced Ply to the initial blood-brain barrier level.
  • 5The second phase could be prevented by applying catalase. Similar high and fluctuating Ply values could be produced reversibly by applying arachidonic acid or NH4C1.
  • 6This is the first report of two distinct types of permeability increase in the cerebral microvasculature, and reasons for this are discussed.

The relative restriction to fluid and solute movement which is ordinarily present between blood and brain breaks down in a number of pathological states, such as trauma, stroke and epilepsy. It is possible to disrupt the blood-brain barrier in a number of different ways experimentally, but different laboratories have reported a large range in values for the increased permeability (see Greenwood, 1992). It is possible that methodological differences are responsible for these discrepancies, and convection of tracer molecules is one variable that is not often controlled. Experiments on single pial microvessels have been designed to measure diffusive permeability alone (see Easton & Fraser, 1994a), but even these techniques have given different results. Thus Olesen & Crone (1986), who estimated permeability from trans-microvascular electrical resistance in frogs, showed that in response to bradykinin, serotonin, ATP, ADP, arachidonic acid and oxygen free radicals the endothelial resistance fell from 1000 to 600 Ω cm2, which is equivalent to a small increase in sucrose permeability from 0.45 to 0.75 × 10−6 cm s−1. Similar small increases were observed by Butt & Jones (1992) in rat pial vessels after superfusion of histamine and bradykinin. These reports contrast with a much larger increase in permeability (from unmeasurably small to 20 × 10−6 cm s−1) of the traumatically disrupted pial microvessels of the frog, measured from the rate of escape of a fluorescent fluid-phase marker using the single vessel occlusion technique (Fraser & Dallas, 1993). Similar large increases were found during the superfusion of hyper-osmolar sucrose (Fraser & Dallas, 1990) and these observations compared well with whole brain studies in the rat (Rapoport, Fredericks, Ohno & Pettigrew, 1980). On the other hand, Butt, Jones & Abbott (1990), using electrical resistance measurements, reported that superfusion with hyperosmotic mannitol resulted in only small changes, similar to those found with histamine. There is also agreement between single microvessel and whole organ studies of the effects of histamine (Gross, Teasdale, Graham, Angerson & Murray Harper, 1982; Butt & Jones, 1992), so it is possible that despite differences in technique being responsible for different permeability values, different stimuli may cause different degrees of blood-brain barrier disruption.

The experiments described here were carried out in an attempt to establish whether the different results obtained in single microvessels were due to methodology alone, or if they reflect distinct types of permeability increase. We have studied permeability changes that occur spontaneously after removing the dura and arachnoid and found that there are qualitative differences that occur over time. A preliminary account of some of the results has been reported previously (Easton & Fraser 1992).


The methods have been fully described previously (Easton & Fraser 1994a), and the following is a brief summary. The experiments were performed on Wistar rats of both sexes aged between 23 and 33 days within the guidlines given by The Animals (Scientific Proceedures) Act 1986. Anaesthesia was induced by intraperitoneal injection of 0.72 ml kg−1 urethane (25% w/v in 0.9% NaCl solution) and maintained where necessary with intravenous doses of 0.07 ml kg−1. At the end of the experiment the animal was killed with an overdose of anaesthetic, prior to decapitation. Body temperature was maintained at 37±1 °C, and tracheal cannulation and orthograde cannulation of the left common carotid artery were performed routinely. The animal was placed prone with the head in a small steriotaxic frame. A section of the frontoparietal bones on the left side, approximately 5 mm × 5 mm between the coronal and lambdoid sutures, was thinned with a dental drill and removed. The cerebral surface was exposed by cutting away the overlying meninges, which in these young rats leaves the pial microvessels devoid of any continuous kyer of overlying tissue (Butt et al. 1990). The rat was then placed on the modified stage of a microscope (ACM; Zeiss) and the exposed cerebral surface illuminated with a 50 W mercury vapour lamp through a × 20 objective lens (NA 0.5; Cooke) fitted with a water immersion cap to eliminate the air–water meniscus. The surface of the brain was constantly superfused by a pump at the rate of 1–2 ml min−1 with artificial cerebrospinal fluid (CSF) warmed to 37±1 °C and delivered through a fine plastic tube attached to the side of the water immersion cap.

Measurements were made by introducing the fluorescent dye Lucifer Yellow (457 kDa, dilithium salt, 5 mM in 0.9% NaCl) into the cerebral microcirculation by bolus injections of between 50 and 200 μl in the ipsilateral carotid artery. A venular capillary with a straight unbranching section of 200–400 μm was selected, and a glass occluding probe (10–30 μm diameter) was lowered to trap the dye as it passed through the vessel, so forming an occluded segment. The microscope image was passed via an image intensifier camera (Panasonic WV-1900, Japan), a video-timer (For-A, Japan) through a video-densitometer and displayed on a television monitor. The video-densitometer operated by integrating the video signal between variable electronic gates used to create a bright cursor on the television monitor. The cursor was placed over a section of the vessel at least 30 μm from the occluding probe, where the vessel would not suffer from geometric distortion (Easton & Fraser 1994a), and at least 100 μm from the open end. The fluorescence was proportional to the concentration of dye and the level of illumination was such that there was no significant photolysis (Fraser & Dallas, 1990). The light signal in the region defined by the cursor was relayed by the video-densitometer to a chart recorder (Lectromed M19, Welwyn Garden City, UK). The axial concentration profile of an occluded microvessel was determined from images captured in an image processor (DIS 3000; Digital Imaging Systems, Newport, UK) and measurement of the fluorescence intensity along the length of the vessel using the Inspector image processor (Matrox; Denval, Quebec, Canada).

Permeability was calculated from the rate of decrease in fluorescence from the measured portion of the occluded segment. The rate at which the intravascular concentration of a small polar molecule, such as Lucifer Yellow, decreases is independent of the hydrostatic pressure (Fraser & Dallas, 1993). Thus the concentration of dye early in the occlusion, before axial volume flux distorts the uniform axial concentration of dye in the region of measurement, will be:

display math(1)

where Ct and Co are the concentrations at times t and o, and k is 4P/d, with P representing the permeability and d the diameter of the vessel.

Solutions and chemicals

Lucifer Yellow (CH, dilithium salt) and urethane were obtained from Sigma. Artificial CSF was buffered to pH 7.40 ± 0.05 and contained (mm): NaCl, 110.5; KC1, 4.7; CaCl2, 1.25; KH2 PO4, 1.1; MgSO4.7H2O, 1.25; NaHCO3, 25; and Hepes, 15. These chemicals were from BDH Chemicals (Poole, UK) except MgSO4.7H20 which was from Fisons (Loughborough, UK) and Hepes which was from Sigma. Histamine (free base), isoprenaline (dl form, bitartrate salt), catalase and dibutyryl cyclic AMP (monosodium salt) were from Sigma. Human α-calcitonin gene-related peptide (CGRP) was kindly given by Dr Susan Brain of King's College London. A23187 was from Sigma and was dissolved in dimethyl sulphoxide (DMSO) and diluted to give a 10 μm solution. Arachidonic acid (sodium salt; Cascade Biochem., Reading, UK) was stored in ethanol at −20 °C. Aliquots of this were gassed with nitrogen to remove the ethanol, and the residue was dissolved in 100 mm sodium carbonate, which was diluted to the required concentration in artificial CSF.

Drugs were applied on the brain side of the microvessels either by dissolving them in the superfusing solution, or focally through a micropipette with a 3 μm diameter tip placed between 2 and 5 μm from the abluminal wall. In the latter case a trace of the fluorescent dye sulforhodamine B (Aldrich, Gillingham, Dorset) was included in the pipette so that the distribution and relative concentration of the drug could be seen easily by changing to the appropriate filter set. Measurements of permeability were made over a period of at least 30 s and started within 30 s of a drug being applied or removed.


There was sufficient autofluorescence in the fluorescein filter set for the cerebral surface to be viewed directly under the microscope. The long, straight arteries and arterioles were easily distinguished from the venular capillaries which were shorter and more tortuous. A venular capillary with an unbranching section of at least 200 μm was selected and the glass occluding probe positioned at one end of the vessel. A bolus of fluorescent dye-containing solution was injected through the carotid arterial cannula, and during its transit through the microcirculation the arterioles filled rapidly, and the venular capillaries filled later. The occluding probe was lowered on to the venular capillary, thereby forming a segment containing dye, while dye was cleared from the rest of the microcirculation.

The rate of change in dye concentration for a small molecule such as Lucifer Yellow will depend on diffusion both axially to the open end of the occluded segment, and radially across the endothelium when a permeable pathway is present (Fraser & Dallas, 1993). Our previous observations on leaky cerebral microvessels showed that radial diffusion was much more important than that along the vessel axis. Thus the rate of concentration decrease showed no change when one probe of a doubly occluded venular capillary was removed (Easton & Eraser, 1994a). This was further investigated here since a number of the following experiments concern microvessels with much lower permeabilities than those reported on previously, and hence it would be expected that axial diffusion would be relatively more important. Measurements of the fluorescence along the axis of a microvessel that showed no sign of transmural diffusion (i.e. an intact blood-brain barrier) were made during a single occlusion. This profile was compared with that calculated from the equation CtC0= erf{x/(2√DT)}, where C is the concentration (at times t and 0), erf is the error function, x is the distance from the open end, D the diffusion coefficient for Lucifer Yellow (taken as 5 × 10−6 cm s−2) and T the time from the beginning of the occlusion (see Crank, 1956). The boundary conditions are that: (i) concentration at the capillary entrance is held at zero, (ii) there is no diffusion across the capillary wall, and (iii) the capillary is infinitely long. In an occluded capillary this last condition is broken, and this will result in the dye concentration being overestimated. Fluorescence profiles were measured on images captured shortly after the occlusion started, and 33 s later. The theoretical curve was fitted to the earlier profile by adjusting the variable T, and the second curve was generated from the fitted T value + 33 s. Figure 1B shows that this second curve grossly underestimates the dye remaining in the bulk of the capillary, although the fit is reasonably good close to the open end. It is possible that a few cells, trapped in the microvessel in the dye–blood mixture, effectively produced short lengths of vessel in which the cross-sectional area available for axial diffusion along was reduced. Under these circumstances Lucifer Yellow concentration changes more than 100μm from the open entrance will be largely governed by diffusion across the wall. Later during the same occlusion when this vessel became leaky, the axial fluorescence profile remained flat in this region, and well above the predicted value for unhindered diffusion to the open end.

Figure 1.

Dye concentration changes along the vessel axis

A, Lucifer Yellow dye, following intracarotid bolus injection, in an occluded venular capillary that did not leak dye across the wall. The dye concentration was measured along the white dashed line. B, the normalized axial dye fluorescence shortly after the beginning of the occlusion (upper trace) and 33 s later (lower trace). The upper smooth curve was fitted from the standard one-dimensional diffusion equation (T= 1 s; see text) and the lower curve represents the axial concentration at T= 34 s. The assumptions that underlie the equation would be expected to overestimate the concentrations at the later time, and hence the large underestimate that is apparent at distances further than 40 μm from the open end indicate that axial diffusion is being hindered, possibly by blood cells trapped in the microvessel. C, the measured axial fluorescence during the same occlusion at 137 and 173 s from the start. The vessel had become leaky, but the one-dimensional diffusion curves still grossly underestimate the remaining fluorescence.

Permeability changes with time

Generally, the rate of fall in fluorescence was very low in occlusions carried out shortly after the exposure of the cerebral surface. This rate was so low that it could not really be distinguished from the random oscillations in the signal. Permeability to Lucifer Yellow (Ply) was measured in five single cerebral venular capillaries (diameter range, 14.5–27 μm) in five rats at frequent intervals from 2 to over 5h (Fig. 2). Permeability was initially very low, and corresponds to the tight state of the intact blood-brain barrier. Permeability later rose in two distinct phases, the first of which occurred between 20 and 54min after the exposure of the pial surface and lasted for between 44 and 164min. It was characterized by a relatively stable, low permeability to Lucifer Yellow, about one-tenth of that found in skeletal muscle capillaries (Ply= (1.65±0–603) × 10−6 cm s−1; mean ± S.D., 38 measurements in the 5 vessels). This first phase was considered to have finished, and the second to have started, when Ply increased abruptly to over 5 × 10−6 cm s−1 and fluctuated thereafter.

Figure 2.

Examples of spontaneous progressive permeability changes in five pial venular capillaries timed from the removal of the meninges

Diameters of vessels were 27, 20, 15, 27 and 17 μm in panels ae, respectively. The points give the Ply values derived from the monoexponential fit to the occlusion data, and the bars are the standard error of the fit. Initially permeability was below the detection level of the technique, and later increased in two distinct phases. The first phase was a small increased permeability which stayed below 5 × 10−6 cm s, ranging from 0.66 × 10−6 to 4.45 × 10−6 cm s−1. The second phase was characterized by a much larger increase that fluctuated from values for tight vessels to those of mesenteric capillaries.

Raised and fluctuating permeability was a feature of the second phase of increase: Ply was (12.31±15.14) × 10−6cms−1 (mean±S.D. of 135 measurements in the 5 vessels). There was no discernible pattern in these fluctuations and they were sometimes so rapid that it was possible to detect several different rates of dye loss during a single occlusion (Fig. 3A The onset of this second phase of permeability increase could be prevented for over 4 h by including the hydrogen peroxide scavenger catalase (100 U ml−1) in the superfusing solution: this was examined in four rats and Fig. 3A shows one of these.

Figure 3.

Second phase of disruption

A, rapid fluctuations in permeability are a characteristic of the second phase of spontaneous disruption, and were sometimes observed during the course of an individual occlusion. In this example monoexponential curves have been fitted to indicate the different rates of fall in dye concentration and the numbers above the curve are the calculated values of Ply× 10−6 cm s−1 (vessel diameter, 18μm 53 min after removal of meninges). B, inclusion of the hydrogen peroxide scavenger catalase (100 U ml−1) in the superfusing solution postponed the second phase of disruption for over 4 h (vessel diameter, 16 μm).

Mimicking and reversing the first phase of disruption

Raised intracellular [Ca2+] has often been implicated in increased permeability of microvessels generally, and particularly in reducing the trans-pial microvascular electrical resistance in frogs (Olesen, 1987). This was tested on the present preparation by applying the divalent cation ionophore A23187 (10 μm; dissolved in DMSO which had not effect on fluorescence) on to the occluded microvessel through a 3 μm diameter micropipette applied closely to the abluminal surface. Ply increased reversibly in the presence of calcium in the superfusing medium, but not at all when the calcium concentration was nominally zero. The degree of Ply increase did not change when the calcium concentration was increased from 2 to 10 mm (see Fig. 4).

Figure 4.

A23187-induced PLY increase requires Ca2+

A, A23187 (10 μm) reversibly increased Ply of a ‘tight’ vessel by an amount that was comparable with the first phase of spontaneous disruption (vessel diameter, 13 μm). The numbers below the curve indicate the Ply value × 10−6 cm s−1. B, summary of experiments in which A23187 was applied with different superfusate Ca2+ concentrations. When [Ca2+] was nominally zero there was no change in Ply((0.096±0.229) × 10−6 cm s−1, mean ±s.e.m., 5 vessels), whereas with 2 or 10 mM A23187 Ply increased by (1.47±0.25) × 10−6 cm s−1 (P < 0.001, n= 11) and (1.38 ± 0.24) × 10−6 cm s−1 (P < 0.002, n= 6), respectively. A total of 11 vessels were used with diameters ranging from 10 to 18 μm, and all of them were either ‘tight’ or in the first phase. The initial permeability had no effect on the Ply increase with any of the different levels of calcium in the superfusate.

Addition of histamine (10 μm) to the solution that superfused the brain surface resulted in previously ‘tight’ microvessels becoming permeable at levels similar to those found in the first phase of spontaneous disruption, i.e. Plyincreased from (0.13 ± 0.03) × 10−6 to (2.12±0.23) × 10−6 cms−1 (4 vessels from 4 rats; P < 0.001). This increase in permeability was reversible, (see Fig. 5). Previous reports have indicated that the response to histamine in these vessels is mediated via the H2 and not the H1 receptor. We examined this by carrying out a series of experiments (see Fig. 6) using 0.5 μm histamine in combination with either the H2 antagonist cimetidine or the H1 antagonist mepyramine, and our results confirmed that it is the H2 receptor and not the H1 receptor that is responsible for the permeability increase.

Figure 5.

Histamine increases permeability reversibly

A, histamine (10μm: horizontal open bars) applied to a ‘tight’ pial venular capillary (diameter, 13μm), produced a rapid permeability increase, that reversed even after 15min continuous application. B, summary of 4 similar experiments on vessels with diameters ranging from 13 to 19μm. The mean permeability increase was (2.41±0.219) × 10−6 cm s−1 (P < 0.001).

Figure 6.

Experiments to identify the histamine receptor responsible for the increased permeability

Venular capillaries of diameter between 9 and 20 μm were exposed to 0.5 μm histamine (H), to 2 μm cimetidine (C), to 3 nm mepyramine (M), and to combined histamine and cimetidine (H + C), and combined histamine and mepyramine (H + M). The numbers in parentheses indicate the number of vessels to which each treatment was applied. The statistical significance of each mean permeability increase is indicated: **P < 0.002; ***P < 0.001.

It has been reported that the permeability (measured by transendothelial resistance) of cultured cerebral endothelial monolayers decreases when intracellular cyclic AMP levels are raised (Rubin et al. 1991). This was investigated here by measuring changes in Ply in microvessels that had been exposed sufficiently long for them to be in the first phase of spontaneous increase. Each of the following substances were applied to four vessels with diameters ranging between 12 and 27 μm from four rats. Dibutyryl-cAMP (250 μm) was added to the superfusate and reduced Ply from (0.79±0.02) × 10−6 to (0.01 plusmn; 0.01) × 10−6 cm s−1, (P 0.05, paired t test). Similar effects were obtained with isoprenaline (5 μm) and CGRP (50 pm) where Ply fell from (2.20± 0.02) × 10−6 to (0.20±0.02) × 10−6 cms−1, (P 0.025), and from (2.25±0.56) × 10−6 to (0.16±0.01) x10−6 cm s−1, (P 0.025), respectively. Figure 7 summarizes the results obtained in these experiments.

Figure 7.

The effects of substances thought to raise intracellular cAMP on pial venular capillary permeability

Each substance was tested on 4 vessels (from 4 rats) in the first phase of spontaneous permeability increase. Dibutyryl-cAMP (dBu-cAMP, 0.25 mm), isoprenaline (5 μm) and CGRP (5 pm) reduced permeability to that of the tight blood–brain barrier, or close to it. inline image, measurements taken during the intervention; □, controls immediately before and 2 min after the intervention. The statistical significance of each mean permeability decrease is indicated: *P < 0.05; **P < 0.025.

The second phase of disruption

Since attempts to mimic the second phase of disruption by procedures that were designed to raise [Ca2+]1 failed (see above), other mediators are likely to be involved. Disruption of the blood-brain barrier underlies the development of vasogenic cerebral oedema, and patients with this condition have been shown to have very high levels of arachidonic acid in cerebrospinal fluid (Chan & Fishman, 1984). It seems possible that this fatty acid has a direct effect on cerebro-microvascular permeability (Onishi, Posner & Shapiro, 1992). We examined this by applying 1 mM arachidonic acid focally through a micropipette to occluded microvessels. Mean Ply rose after the addition of arachidonic acid from (1.97 ± 0.317) x10−6 to (13.7 ± 1.98) ± 10−6 cms−1 (7 vessels from 5 rats, P < 0.001), and this effect was fully reversed within 2 min on all occasions. The largest rise in Ply was from 1.35x10−6 to 23.73x10−6 cm s−1. Sometimes rapid changes in permeability were also seen in the course of an occlusion, and when this occurred the first value was reported. Oleic acid (2 him) had no effect on permeability: Ply was (1.73±0.61) × 10−6cms−1 before and (1.71±0.50) × 10−6 cm s−1 during application in four vessels from four rats. These findings are summarized in Fig. 8.

Figure 8.

A, arachidonic acid (3 mm) application to a ‘tight’ vessel (diameter, 17 μm) resulted in an immediate fall in fluorescence which indicates a permeability increase to 21 × 10−6 cm s−1. B, the rapid permeability fluctuations observed in some second phase occlusions were occasionally mimicked. In this occlusion experiment on a 13 μm microvessel the rate of fluorescence decrease in the first 25s indicated that the permeability was in the first phase of spontaneous disruption. When arachidonic acid was applied permeability rose sharply, then fell and rose again. The numbers below the curve indicate the Plyvalue × 10−6 cm s−1. C, bar chart summarizing experiments with 1 mm arachidonic acid. Oleic acid had no effect on the permeability of these microvessels. inline image measurements taken during the intervention; □, controls immediately before and 2 min after the intervention.

It has been suggested that polyamine synthesis is an important factor in the disruption of the blood-brain barrier (Koenig, Goldstone & Lu, 1989), and it is possible that this will result in raised pH1. In our search for mechanisms of mimicking the second phase of disruption we applied an isosmotic superfusate in which NH4C1 (50 mM) substituted for NaCl and the pH buffered to 7.4. Rapid transient rises in Ply were seen in four microvessels from four rats, similar to the second-phase spontaneous permeability increase: Ply rose to peak values of 22.4 × 10−6, 37.2 × 10−6, 41.9 × 10−6 and 13.8 × 10−6 cm s. These results are illustrated in Fig. 9.

Figure 9.

Manipulation of pH1

Application of an isotonic buffered superfusate in which 50 mM NH4C1 was substituted for NaCl to tight pial capillaries (horizontal open bars) resulted in a permeability increase directly to second phase levels; 4 vessels from 4 rats with diameters ranging from 16 to 24 μm.


These experiments show that there is a qualitative difference in the types of permeability increase in pial microvessels. These vessels are part of the blood-brain barrier and initially have a very low permeability, but following the removal of a relatively large area of meninges their permeability increased spontaneously in two distinct phases, despite the continuous superfusion of an artificial cerebrospinal fluid. Immediately after meningeal removal the vessels were tight, but after a few minutes there was a small permeability increase which could be mimicked by procedures that were calculated to raise endothelial [Ca2]1 After about an hour there were large fluctuating increases, some aspects of which could be imitated by the application of arachidonic acid or ammonium chloride. The observation that a 5-fold increase in [Ca2+]i in the presence of A23187 caused no greater increase in permeability is an indication that separate mechanisms are involved in the two phases of spontaneous permeability increase, and that the second phase is not a mere extension of the events of the first. We have carried out some experiments to ascertain the mechanism by which arachidonic acid mediates this large permeability increase (Easton & Eraser, 1994b) and it appears that oxygen free radicals are necessary. This contention is supported by the present finding that the second phase of permeability increase could be prevented with catalase.

The rapid changes in the rate at which fluorescent signal decreases during the second phase could possibly have been brought about by an undetected slippage of dye past the occluding probe, or by a loss and replacement of vascular cells at the capillary entrance. We do not consider that either these mechanisms played an important role. Firstly, there were occlusions in which more than one rate of dye loss was observed (see Fig. 3A) while the vessel was blocked at both ends by occluding probes. Secondly, when an occlusion did slip so that there was an obvious movement of cells within the vessel, dye concentration did fall, but at a rate rather lower than was seen in rapid transient high permeability measurements.

The idea that there are two sequential phases of disruption has not been reported before, and this is likely to reflect the greater sensitivity and spatial resolution of the method used here. Some aspects of the technique will be discussed before the present findings are related to other studies that have examined disruption of the blood-brain barrier.

Comparison with other single cerebral microvessel measurements

The advantage of single vessel techniques is that they can separate diffusional permeability from complications resulting from connective transport. There are a number of reports on permeability changes using the electrical resistance method in pial microvessels. There are, however, additional advantages in resolving higher permeabilities with the micro-occlusion technique due to its greater spatial resolution.

Normal blood-brain barrier

The lower limit of the permeability measurement in the present experiments depends upon the resolution of the rate of fall of fluorescence. This was about 3 % of the signal over the course of a minute, so for a 15 /tm diameter microvessel this is equivalent to a permeability to Lucifer Yellow (Ply) of slightly under 0.2 × 10−6 cm s−1. This may be compared with permeability estimates made from the maximum electrical resistance measurements on single cerebral microvessels, since P = Dp/R, where p/R is the ratio of the resistivity of mammalian extracellular fluid to the resistance measurement. These values of R range between 1500 and 3000 Q, cm2 (Butt & Jones, 1992), and taking D and p to be 5 × 10−6cms−2 and 60 × 2 cm, respectively, giving a PLY range of 0.2 × 10−6 to 0.1 × 10−6 cm s−1. This agrees with measures of sucrose permeability across the blood-brain barrier of the rat frontal lobe using a whole organ technique (0.054 × 10−6 cm s−1; Takasato, Bapoport & Smith, 1984). This indicates that the low Ply values recorded within 30 min of removing the meninges are likely to be from normal tight vessels.

First phase of disruption

Published results show that stimuli that would be expected to raise [Ca2+]i and which result in permeability increases that correspond to the first phase of disruption, give values of similar order of magnitude with the electrical resistance method and the single occlusion technique (Olesen, 1987). The resistance values are somewhat lower, however, and this probably reflects the fact that permeability and resistance are inversely related. Hence simple averages of resistance will necessarily underestimate permeability. This also applies to the effects of hyperosmotic solutions: Fraser & Dallas (1990) calculated a sucrose permeability of 12 × 10−6 cms−1 in frog pial microvessels disrupted by hyperosmotic shock (equivalent to a resistance of 38 Q, cm2), which compares reasonably well with the mean resistance of 98 Ω cm2 in rat pial venules disrupted by hyperosmotic mannitol (Butt et al. 1990).

Second phase of disruption

In contrast, permeability increases with the occlusion method during the application of arachidonic acid, ammonium chloride and prolonged exposure of the cerebral surface were more than an order of magnitude greater than those reported with the resistance method. Many permeability changes have been shown to be confined to discrete sections of microvessels (e.g. Luthert, Greenwood, Pratt & Lantos, 1987), which we have confirmed by showing patchy permeability increases along the length of leaky pial microvessels (Fraser & Dallas, 1990). The measurement of resistance relies on cable theory, which requires a uniform voltage dissipation over the length of the microvessel (at least 200 /tm: Crone & Olesen, 1982). This will distort the sampling of data because leaky vessels with a patchy permeability would have to be discarded (Butt et al. 1990). The microvessel occlusion technique, however, allows patchy permeability to be resolved to within 10 μm, and can measure the changes to within 10s. The electrical method can, in some instances (Crone & Olesen, 1982), give changes in measurements to less than 1 s, although in at least one variation of the method data were gathered between a pair of electrodes 250 μm apart and averaged over 2–15 s to give mean resistance measures. It usually took 30–60 s for an occlusion measurement to be made, but transitions in permeability could be detected more rapidly, so long as they occurred within the course of an occlusion (see Figs 3 A and 7A). These rapid transitions in permeability were not reported with the resistance method, probably because the changes were confined to only a small segment of the total measured length.

Whole brain studies

The majority of studies of barrier disruption have used whole brain techniques which produce measures of the number of leaky sites of dye extravasation, and quantitative changes in permeability–surface area product (PS) to radioisotopes. Techniques that depend upon macromolecular escape will probably be complicated by convective effects brought about by changes in intravascular pressure as well as in the area available for diffusion across the vessel wall (Schilling & Wahl, 1994). Increased cerebral venous hydrostatic pressure, however, can itself result in disruption of the blood-brain barrier (Häggendal & Johansson, 1972), a complication that was avoided in the present study by applying substances focally. Some studies, in which cerebro-vascular PS increases were quantified, have shown significant differences depending on the stimulus used. Thus, infusion of 10 μm histamine in the rat resulted in a 2–3 times permeability increase to [14C]sucrose and amino-isobutyric acid (Gross et al. 1982), topical application of 600 μm arachidonic acid gave about a 5-fold increase in albumin permeability (Kontos et al. 1985), while 20-fold increases in sucrose permeability after infusion of hyper-osmolar arabinose were found by Rapoport et al. (1980). The permeability values from these stimuli are somewhat lower than those from micro-occlusion measurements (e.g. 2 × 10−6 cm s−1 compared with 12 × 10−6 cm s−1; Rapoport et al. 1980 cf. Fraser & Dallas, 1990, assuming a micro-vascular surface area of 100 cm2 g−1), but this is consistent with the patchy disruption of the microvessels observed in these capillaries. Gross et al. (1982) found the regional permeability to sucrose during continuous histamine infusion was equivalent to 0.2 × 10−6 cm s−1, which is rather less than the 2.0 × 10−6cm s−1 measured in the present studies. These differences could be explained by histamine degradation by leucocytes or endothelial ecto-enzymes (Plaut & Lichtenstein, 1978), but it is also possible that the pial venular capillaries are rather more responsive to inflammatory mediators than other pial microvessels (Butt et al. 1990), or parenchymal vessels (see Gross et al. 1982).


The increased sensitivity and resolution given by the single microvessel occlusion technique has allowed us to resolve three distinct ranges of permeability that occur in individual pial venular capillaries as they pass from the intact blood-brain barrier and become inflamed. The results presented here suggest that the first phase of spontaneous increase is an early step in inflammation consequent on removing the meninges. There are many inflammatory mediators which raise endothelial [Ca2+]i and also increase permeability, and histamine is one such example. This permeability increase is limited in these brain venular capillaries and is not sufficiently great to lead to cerebral oedema. The much larger increase seen in the second phase was probably due to generation of free radicals, since its spontaneous development could be prevented by catalase. Permeability decreases could be brought about by the application of substances that raise intracellular cAMP concentration, a phenomenon also found in many other vascular beds. These observations point towards there being controlled increases in cerebrovascular permeability, which may fulfill physiological functions, as well as the much larger permeability increases, which are probably pathological.


A. S. E. was supported by an MRC–Pfizer Collaborative studentship, and M.H. S. was a Commonwealth Staff Scholar.