SUMMARY1. This study reports the first results of measurements of filtration coefficient (LP) and osmotic reflection coefficient to sucrose (oauk) in single brain microvessels.2. Microvessels on the surface of frog brain were cannulated with a micropipette and perfused with an artificial cerebrospinal fluid (CSF) containing the low molecular weight impermeant dye carboxyfluorescein (MW 376). The superfusing solution was a similar CSF which could be made hypertonic by the addition of 40-125 mmol 1I sucrose.3. Vessels were assessed for dye retention using video-intensified microscopy after occlusion with a glass microneedle. Only six vessels out of a total of ninety-five were tight under the experimental conditions used. Those vessels which were tight were occluded while an osmotic load was applied across them. When this load was 50 mosmol 11 and less, the steady-state dye concentration within the vessel lumen was similar to that predicted assuming the endothelium behaves as a perfect semipermeable membrane, with concentration polarization of solute.4. The product L or was estimated in two ways: (i) from the fitted monoexponential function that described the rising dye concentration within the occluded segment, and (ii) from the initial rate of increase in dye concentration. The two values obtained were similar and it was concluded that 0(NaC1 = -suc = 1, and the best estimate for filtration coefficient Lp = 2-0 x 10-cm (cmH2O s)-1.5. At the osmotic loads of 100 mosmol 1I1 and more, the initial rate of increase estimate of Lp o was less than half of the whole curve estimate, the axial dye distributions were dissimilar from those predicted by a mathematical model based on the perfect semipermeable membrane, and the steady-state concentration was less than 70 % of that expected. These findings are consistent with a diffusive pathway having opened. The model was modified to include patches of vessel wall which had developed leaks and a good fit to the data was obtained with a sucrose permeability and an Lp similar to skeletal muscle endothelium.6. The possibility that water passes through a paracellular pathway across the intact blood-brain barrier is discussed. It is concluded that this pathway could not

SUMMARY1. This study reports the results of varying the hydrostatic pressure on measurements of permeability coefficient to the low molecular weight impermeant dye carboxyfluorescein (MW = 376) in single leaky cerebral microvessels. A mathematical model, that solved the convective diffusion equations used to analyse the measurements, showed that the measurements were consistent with the leakiness being due to 22 nm wide parallel-sided slits between endothelial cells.2. Microvessels on the surface of the frog's brain were cannulated with a micropipette and perfused with an artificial cerebrospinal fluid containing the dye. Vessels were occluded with a glass microneedle and the rate of change in dye concentration in a 12 ,tm length section was measured using video-intensified microscopy.3. It was found that the rate of dye loss at all points along the occluded microvessel segment could be accounted for by a model for convection and diffusion, and that changes in dye concentration at a point remote from the segment entrance can give a good measure of diffusive permeability.4. When series of measurements were carried out on a single vessel, permeability rose over the course of 20 min. Mean permeability for all measurements was 3-01 x 10-5 cm sec-1, n = 64 (mode, 2-0; range, 0 48-9-6). The hydrostatic pressure applied during the perfusion had no effect on the measured permeability.5. The dye concentration along the vessel axis was measured at the steady state and was shown to respond to changes in hydrostatic perfusion pressure in a way predicted by the model. This indicates that hydrostatically driven bulk flow can be important, and thus convection may account for effects previously ascribed to vesicular transeytosis.6. The possible anatomical basis for the porous pathway is discussed in the light of recent observations on the presence of 0-5 gum perijunctional gaps, the possibility of transendothelial channels, and the unzipping of tight junctions to leave a 22 nm wide slit.