Hyperosmotic solutions of 3 M urea, either infused into one internal carotid artery or applied topically to the pia mater of rabbits, results in the opening of endothelial tight junctions through which horseradish peroxidase passes from blood to extracellular fluid of the brain. The evidence for this opening of the blood-brain barrier to protein is the entry of peroxidase into the extracellular pools between successive tight junctions. In animals not receiving 3 M urea, the interjunctional pools are inaccessible to proteins. Having passed through the endothelial junctions, the peroxidase spreads along the extracellular channels of the perivascular neuropil for approximately 100 P in 90 seconds. Most of the affected vessels are capillaries, though larger vessels are rendered leaky as well. Calyciform cisterns, that lie beneath shallow notches in the endothelium of untreated rabbits, appear to be enlarged after the administration of 3 M urea. It is undetermined whether these few endothelial cisterns and vesicles are involved in carrying protein from blood to the cerebral extracellular fluid.The injection into the internal carotid artery or pial application of solutions containing hyperosmolar concentrations of various substances causes a reversible opening of the blood-brain barrier in rab-

Portions of certain arterioles on the surface of the brain and within it can transfer protein from blood to perivascular basement membrane. One to 50 mg of horseradish peroxidase (MW: 40,000) were injected intravenously into normal, adult mice. Three to 30 minutes later, their brains were fixed by vascular perfusion of aldehydes, incubated, and processed for electron microscopy.Segments of some cerebral arterioles, with an average diameter of 15-30 p, transported peroxidase. The pial arterioles were mainly situated within sulci, the parenchymal ones in the ventral part of the diencephalon and adjacent brain stem. The arteriolar segments were, commonly, bifurcations and short, straight portions. Similar vessels transferred protein in normal rats and hamsters. Neither increasing the dose of peroxidase above 10 mg nor the circulation time from 3 to 30 minutes altered the number of transferring vessels in mice.The endothelid cells were, as in cerebral capillaries, linked by tight junctions that prevented the extracellular passage of peroxidase. When peroxidase was injected 45 minutes after fixation, it did not reach the basement membrane. Moreover, ferritin, too large a molecule (MW: 900,000) to pass through these endothelid junctions, was also transported across the same s e p e n t s 3 minutes after 100-500 mg had been injected intravenously. These observations indicate that vesicular transport rather than passive, extracellular movement was the means of transferring proteins from blood to perivascular basement membrane. It is probable that serum albumin and globulins are likewise transferred in normal rodents.

This review paper deals with the transport of the protein tracer horseradish peroxidase across cerebral vessels under normal and various experimental conditions. Electronmicroscopical investigations have revealed that, under normal conditions, a minor vesicular transfer of intravenously injected peroxidase occurs across the endothelium in segments of arterioles, capillaries and venules, especially in arterioles with a diameter about 15-30 mu. This normally occurring vesicular transport is susceptible to various experimental conditions. Thus the transfer of tracer increases when a hypertonic solution is injected into the internal carotid artery presumably due to vesicular transport. Extensive acute hypertension of short duration also increases the vesicular transfer of peroxidase from blood to brain. Identical observations are obtained when the hypertension is evoked by intravenous injection of phentolamine and by electrically induced seizures. During the postischemic period, one hour after release of the occlusion of an internal carotid artery in the Mongolian gerbil the vesicular transport of peroxidase is increased across the endothelium of cerebral vessels. The explanation may be release of serotonin from blood platelets during the occlusion. The serotonin could then increase the blood pressure locally in the brain resulting in an enhanced permeability. Serotonin, after perfusion through the cerebral ventricular system, is also able to increase the normally occurring vesicular transfer. The most likely mechanism behind this phenomenon seems at the moment to be local hypertension evoked by serotonin-induced vasoconstriction of arterioles. Finally, the enhanced vesicular transport across cerebral endothelium caused by porto-caval anastomosis is mentioned and the possible role of disturbances in the metabolism of amines as responsible for the extravasation is discussed.

Under normal conditions a slight vesicular transfer of intravenously injected horseradish peroxidase (HRP) occurs across the endothelium of cerebral vessels, especially short segments of arterioles. The vesicular transport can be notably increased by chemically induced acute hypertension. In the present investigation 4 groups of animals received HRP, and the permeability of the cerebral endothelium was studied semimacroscopically, light microscopically and electron microscopically. The rats in group 1 were given 10 electroshocks. This caused a significant rise in the blood pressure (BP). Furthermore, a noticeable extravasation of HRP was observed, especially across the endothelium in cerebral arterioles. From the basement membranes of the vessels reaction product could be followed into the extracellular spaces of the neighbouring neuropil. Group 2 comprises rats that were given 10 electroshocks preceded by transection of the cervical part of the spinal cord. The BP remained at normal level and the permeability was unaltered. The animals in group 3 received only 1 electroshock. Usually, the BP was markedly increased and this was accompanied by enhanced permeability across the vessels of the brain. Group 4 consists of control animals, injected with HRP and treated as groups 1 and 3 with the difference that electrical stimulation was not performed. A general feature was that no endothelial damage was observed and that reaction product was not found between neighbouring endothelial cells from the first luminal to the first abluminal tight junction. Based on the observations it seems reasonable to assume that the increased permeability of tracer that occurs after 10 electroshocks or only one is caused by the acute hypertension evoked by the electrical stimulation; furthermore, the transfer is concluded to be vesicular transport.