Sustained elevation of calcium induces Ca(2+)/calmodulin-dependent protein kinase II clusters in hippocampal neurons.

Treatment of cultured hippocampal neurons with the mitochondrial uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP) in the absence of glucose mimics ischemic energy depletion and induces formation of Ca(2+)/calmodulin-dependent protein kinase II (CaMKII) clusters, spherical structures with diameters of 75-175 nm [Dosemeci et al., J. Neurosci. 20 (2000) 3076-3084]. The demonstration that CaMKII clustering occurs in the intact, adult rat brain upon interruption of blood flow indicates that clustering is not confined to cell cultures. Application of N-methyl-D-aspartate (250 microM, 15 min) to hippocampal cultures also induces cluster formation, suggesting a role for Ca(2+). Indeed, intracellular Ca(2+) monitored with Fluo3-AM by confocal microscopy reaches a sustained high level within 5 min of CCCP treatment. The appearance of immunolabeled CaMKII clusters, detected by electron microscopy, follows the onset of the sustained increase in intracellular Ca(2+). Moreover, CaMKII does not cluster when the rise in intracellular Ca(2+) is prevented by the omission of extracellular Ca(2+) during CCCP treatment, confirming that clustering is Ca(2+)-dependent. A lag period of 1-2 min between the onset of high intracellular Ca(2+) levels and the formation of CaMKII clusters suggests that a sustained increase in Ca(2+) level is necessary for the clustering. CaMKII clusters disappear within 2 h of returning the cultures to normal incubation conditions, at which time no significant cell death is detected. These results indicate that pathological conditions that promote sustained episodes of Ca(2+) overload result in a transitory clustering of CaMKII into spherical structures. CaMKII clustering may represent a cellular defense mechanism to sequester a portion of the CaMKII pool, thereby preventing excessive protein phosphorylation.

Permeable endothelium and the interstitial space of brain.

1. Fenestrated vessels can be reversibly induced in brain by agents that stimulate urokinase production. This plasminogen activator, like vascular endothelial growth factor and metalloproteinases, is secreted by tumor cells and may account for induction of fenestrated vessels. Why only some of the brain's barrier vessels are converted to fenestrated vessels is unknown. 2. The structures responsible for the filtering of solutes by fenestrated vessels may be the same as those of continuous, less permeable vessels: the glycocalyx on the surfaces of the endothelial cells and the subendothelial basal lamina. 3. Solutes leaving the cerebral ventricles immediately enter the interstitial clefts between the cells lining the ventricles. A fraction of a variety of solutes, injected into CSF compartments, is retained by subendothelial basal lamina, from which the solutes may be released in a regulated way. 4. The brain's CSF and interstitial clefts are the conduits for nonsynaptic volume transmission of diffusible signals, e.g., ions, neurotransmitters, and hormones. This type of transmission could be abetted by a parallel, cell-to-cell volume transmission mediated by gap junctions between astrocytes bordering CSF compartments and parenchymal astrocytes. 5. The width and contents of the interstitial clefts in fetal brain permit cell migration and outgrowth of neurites. The contents of the narrower and different interstitial clefts of mature brain permit solute convection but must be enzymatically degraded in order for cells to migrate through it.

Chemical induction of fenestrae in vessels of the blood-brain barrier.

The endothelium responsible for the blood-brain barrier to hydrophilic solutes has been converted here, by chemical means, to the fenestrated, permeable type of vessel (FV). During development, some brain vessels are reported to be transiently fenestrated. Endothelial fenestrae of adrenal glands are known to be reinduced in vitro by retinoic acid (RA) or phorbol myristate acetate (PMA). Could fenestrae be likewise reinduced in brain barrier vessels? When RA or PMA were infused continuously by an osmotic pump for 28 days into the cerebral cortex of rats, some brain vessels in the lesion cavity created by the reagents were FV. There were no FV in adjacent brain. When 100 microM RA was infused, about 20% of vessels in the cyst were FV, as were about 29% after infusion of 150 ng/ml PMA. Fenestra development depended on concentration and time. Reversibility of fenestra formation was complete at 1-2 months after delivery of RA or PMA had ceased. It is proposed that the RA and PMA effect is mediated by the plasminogen activator urokinase, in as much as both RA and PMA stimulate its production. This notion is supported by preliminary experiments in which urokinase infusion into brain also produced fenestrae. It is further suggested that the reversible induction of fenestrae in the mature brain by RA, PMA, and, perhaps, a variety of other conversion factors may be confined to a subset of brain barrier vessels that must be regenerating and of the kind that were temporarily fenestrated during fetal life.

Pathway across blood-brain barrier opened by the bradykinin agonist, RMP-7.

The route taken by lanthanum (MW 139) across cerebral endothelium was delineated when the blood-brain barrier was opened by RMP-7, a novel bradykinin agonist. Balb C mice were infused through a jugular vein with LaCl3 with or without RMP-7 (5 micrograms/kg). Ten minutes later, the brains were fixed with aldehydes and processed for electron microscopy. The patency of the junctions between endothelial cells was estimated by counting the number of junctions penetrated by LaCl3. Tracer penetrated the junctions in about 25% of microvessels in vehicle infused, control mice and about 58% in the RMP-7 group, where more junctions per vessel were also penetrated. The LaCl3 then penetrated the basal lamina in about 20% of all microvessels in the RMP-7 group, versus 0.50% in the control group. From the basal lamina, the tracer entered perivascular spaces in about 13% of all microvessels in the RMP-7 group and about 0.07% in the controls. Very few endocytic pits or vesicles in the RMP-7 group were labeled, so LaCl3 did not cross endothelium by transcytosis. The increased number of tight junctions penetrated by tracer and its spread into periendothelial basal lamina and interstitial clefts indicated, therefore, a paracellular route of exudation in the RMP-7 treated animals.

Characterization of synaptic vesicles and related neuronal features in nerve growth factor and ras oncogene differentiated PC12 cells.

PC12 cells can differentiate into neuron-like cells after treatment with either nerve growth factor (NGF) or transduction with a retrovirus which expresses the K-ras oncogene. The concomitant treatment of NGF plus ras differentiates PC12 cells further than either agent alone with respect to neurite outgrowth, acetylcholinesterase levels, and most strikingly, the number of synaptic vesicle (SV) clusters. These SV clusters in PC12 cell neurites closely resemble those in the presynaptic terminals of neurons. Such SV clusters have not been described in cell lines previously. The SV clusters from all three differentiated groups (NGF, ras, and NGF plus ras) were similar in size, shape, and configuration, except that the ones in the doubly treated group occur in higher frequency and have more vesicles. The synaptic nature of these vesicle clusters was demonstrated by their regulated depletion after potassium stimulation. Furthermore, these vesicle clusters stained positively for two SV-associated proteins, synapsin I and synaptophysin, by EM immunocytochemistry (ICC). Such SV clusters in a cell line are very useful for characterizing the regulated release of SVs and the distribution of SV-related antigens in intact cells. Analysis by SDS-gel electrophoresis and immunoblotting indicated that synapsin I levels are higher in all three differentiated groups compared to untreated cells; whereas synaptophysin levels are lower in cells exposed to NGF alone or with NGF and ras double treatment. Possible convergence and/or divergence on the mechanisms of NGF and ras differentiation in PC12 cells are discussed.

Penetration of solutes, viruses, and cells across the blood-brain barrier.

The aspects presented here of how solutes, viruses and cells are able to cross the BBB indicate that there must be an active interaction of endothelium with viruses and immune system cells before they can penetrate the brain and spinal cord. The axoplasmic pathway taken by lectin-solute conjugates is similar but not identical to that followed by viral particles during their retrograde or anterograde transit through the axoplasm. Both the conjugates and virus are transferred to other neurons transsynaptically but the receptor mediated transfer utilized by viruses is far more specific. Cranial nerves are involved in both the entry and egress of antigens into and out of the brain. Antigen, generated within the CNS, may be able to escape from the brain to lymphoid tissue by passing into the fluid around a cranial nerve, thence via the lymph into lymph nodes to initiate an immune response involving the CNS.

Astroglial membrane structure is affected by agents that raise cyclic AMP and by phosphatidylcholine phospholipase C.

The role of signal transduction mechanisms in the production of the characteristic orthogonal arrays of particle assemblies in the astroglial plasma membrane was investigated in vitro by freeze-fracture electron microscopy. Agents which raise cellular cAMP levels and subsequently activate protein kinase A, such as forskolin (50 microM), isoproterenol (10 microM) and 8-bromo-cAMP (1 mM), increased the density, the number of assemblies per unit area of cleaved cell membrane, and the frequency of astrocytes with assemblies. Agents that lead to the activation of protein kinase C, such as phorbol 12,13-myristate acetate (at 50 nM) and choline-dependent phospholipase C (at 0.01-0.1 U ml-1), did not affect the assembly concentration. Thus, protein kinase A but not protein kinase C appears to be involved in the production of assemblies or their insertion into the astroglial plasma membrane. Although choline-dependent phospholipase C did not affect the astroglial assemblies, it caused the non-assembly, background particles to aggregate. A choline-dependent phospholipase C from a different source (B. cereus) was also active though at a higher concentration. Phospholipases of different specificities, such as phospholipase A2, phospholipase D or inositol-dependent phospholipase C were inactive over a wide range of concentrations. Two other astroglia derived cells, Müller cells and cells of the C6 glioma cell line, were also similarly affected by choline-dependent phospholipase C, while six other cells types including neurons, endothelial cells and fibroblasts were unaffected. It appears that phosphatidylcholine plays a significant role in determining the membrane structure of astrocytes. In a search for a means of isolating the assemblies, the binding of three lectins: ConA, WGA and PNA, conjugated to gold, was tested by label-fracture to ascertain whether the assemblies have an external oligosaccharide component. None of the lectins bound specifically to assemblies.

Some neuronal properties of PC12 cells differentiated by the K-ras oncogene.

When infected with a virus containing the Kirsten-ras oncogene, rat phaeochromocytoma or PC12 cells elaborated neurites and ceased mitosis, that is, they underwent neuronal differentiation. Such differentiated cells could be replaced and maintained up to 20 weeks in vitro without the need of an exogenous, continuous supply of nerve growth factor (NGF). The neurites of K-ras infected PC12 cells, filled with microtubules and actin which was concentrated within the growth cones, resembled those of primary neurons in vitro. As in the NGF-primed PC12 cells, two types of secretory vesicles were present in the K-ras-infected PC12 neurites: large (100 nm), dense core granules, and small (45 nm), clear vesicles. Compared to naive PC12 cells, K-ras infected PC12 cells had (a) higher activities of acetylcholinesterase and choline acetyltransferase, two enzymes involved in acetylcholine metabolism; (b) enhanced activity of tyrosine hydroxylase, the rate-limiting enzyme in catecholamine biosynthesis; (c) a higher, evoked norepinephrine release; and (d) similar levels of sodium-dependent uptake of both choline and norepinephrine. Although the total content of catecholamines in K-ras-differentiated PC12 cells was less than that of naïve cells, both norepinephrine and dopamine were present in substantial amounts and norepinephrine was released after stimulation. According to their enzymatic activity, these cells can also synthesize acetylcholine and thus have potential as donors for the intracerebral replacement of either catecholaminergic or cholinergic neurotransmitters.

Astrocytic orthogonal arrays of intramembranous particle assemblies are modulated by brain endothelial cells in vitro.

Solo astroglial cultures have randomly distributed, intramembranous, orthogonal arrays of particles (assemblies) which are only revealed by freeze-fracture electron microscopy. Co-culturing astrocytes with brain endothelial cells brought about localized, tightly packed assembly aggregates and greatly increased the overall assembly density. Cytosol homogenates of freeze-thawed brain endothelial cells caused a transient increase in astroglial assembly numbers. These results, taken together with the fact that astrocytes in vivo have the highest concentration of perivascular sites, suggest that brain endothelial cells influence the distribution and concentration of astrogial assemblies both in vivo and in vitro through cellular interactions. Meningeal cells and fibroblasts also augmented the astroglial assembly densities in co-culture, while neuronal cells (cerebellar granule cells and PC12 cells primed with nerve growth factor) and other control cell types did not affect assembly number in co-culture with astrocytes. Moreover, brain endothelial cells did not induce any formation of assemblies in the membranes of two transformed astroglial cell lines.

Regrowth of damaged neurosecretory axons to fenestrated vessels of implanted peripheral tissues.

A major target of neurosecretory axons (NSA) is the basal lamina around fenestrated blood vessels (FBV) in the neural lobe of the pituitary gland. We have posed the question of whether there is neurovascular specificity. Do mature, regenerating NSA terminate selectively on the FBV of the neural lobe compared with the FBV of other tissues that normally are not innervated by NSA? Three types of tissue were transplanted between inbred Fisher rats. Fragments, about 1 mm3, of pineal, adrenal medulla, and neural lobe were grafted bilaterally to the hypothalamic, retro-chiasmatic area, which includes bundles of NSA from supraoptic and paraventricular neurosecretory nuclei exclusively, but no FBV. Two and 4 weeks later, the grafts were prepared for the immunohistochemical localization of NSA and for electron microscopy. NSA-FBV proximity was measured, and the number of NSA, FBV, and of NSA-FBV associations was counted per surface area of each graft. Regenerating NSA can associate as closely with FBV of other tissues as they can with the FBV of the neural lobe. There does not appear to be specificity with respect to the closeness of association between neurosecretory terminals and fenestrated capillaries. However, the number of these associations is greater in neural lobe grafts than in adrenal or pineal grafts at 4 weeks. The number of FBV is also greatest in neural lobe grafts at this time, an increase that would provide a greater opportunity for NSA-FBV associations.

Release from live choroid plexus of apical fragments and electrophoretic characterization of their synthetic products.

Protein synthesis and secretion by the choroid plexus (CP) has been implicated as a major source of certain proteins in cerebrospinal fluid (CSF), such as transthyretin. The suggestion that proteins are elaborated from CP through apocrine secretion has been borne out by the presence of newly labeled proteins in apical protrusions from CP (Agnew et al.: Cell and Tissue Research 208:261-281, 1980a). When the protrusions (aposomes) separate from the cells, they continue to incorporate labeled amino acids (Gudeman et al.: Tissue and Cell 19:101-109, 1987). In the present work the formation of aposomes in live CP explants indicated that these spheroids were not the result of fixation. Aposomes were also identified within rat CSF by immunohistochemistry with monoclonal directed against aposomes as well as with anti-transthyretin serum. The protein product of aposomes was characterized by 2-dimensional SDS-PAGE and compared to the protein products of whole CP tissue. Paradoxically, transthyretin, a heavily labeled protein in the tissue, was virtually undetected in the aposome synthetic profile. However, four other proteins were expressed in relatively equivalent amounts by the aposomes. The presence of mRNA in aposomes was detected with a poly dT probe, and the presence of actin was revealed by phalloidin staining of aposomes. These studies provide a more comprehensive definition of aposomes, but the functions of their secreted proteins remains to be determined.

Cerebral vessels cryofixed after hyperosmosis or cold injury in normothermic and hypothermic frogs.

Three purported means by which large solutes may penetrate the blood-brain barrier are: permeabilized tight junctions; vesicular transport; or channel formation across cerebral blood vessels. The role of vesicular transport has been questioned, in part, because many cytoplasmic vesicles are induced by aldehyde fixation. Cryofixation reduces this artefact and was used to see structural changes in frog cerebral endothelium made permeable to plasma solutes after perivascular exposure to hyperosmotic (3 M) urea, or injury with a cold probe (-50 degrees C). Some control and experimental frogs were made hypothermic so as to inhibit endocytosis and autolytic changes. The brains of some untreated controls were immerse-fixed in aldehydes. Other controls and all other brains of normothermic or hypothermic animals were rapidly frozen, then substituted with acetone-fixative. The interendothelial tight junctions separate partially or completely, after hyperosmotic exposure, in one third of the junctions. Blood-borne ferritin and Evans blue pass through some of the patent junctions. Junctional opening is caused by cell shrinkage, because the perimeter/area ratio of individual endothelial cells in the hyperosmotic group is significantly greater than in the control, due to a decreased area. Large 0.08-0.32-micron-wide invaginations or pits of the endothelial cell membrane characterize both cryofixed and aldehyde-fixed vessels. The pits often appear as isolated vesicles in the cytoplasm, but serial sections reveal that many communicate with either the capillary lumen or subendothelial space. No series of pits opened onto both lumen and space to form a transendothelial channel. The number of vesicles in aldehyde-fixed specimens is about 4 times greater (P less than 0.01) and in the cold injured, cryofixed brain capillary, about two times greater (P less than 0.01), than in the cryofixed control. Hyperosmotic exposure does not increase the number of pits. The permeabilization of anuran cerebral endothelium by hyperosmotic treatment or cold injury is thus by means of an intercellular rather than a transcellular route.

Development of membrane interactions between brain endothelial cells and astrocytes in vitro.

To ascertain whether there is a mutual influence on the structure of their cell membranes, brain endothelial cells and their closest neighbor, astrocytes, were grown alone or together in vitro and freeze-fractured. When cultured separately, the brain endothelial cells had a low frequency of short, fragmented tight junctions. Many gap junctions, which are absent from mature brain capillaries in vivo, intercalated among the tight junctional strands, or were separate from them. The separately cultured astrocytes had low concentrations of randomly distributed assemblies (1-30/micron2) in their membranes. When the two cell types were co-cultured, the endothelial tight junctions were greatly enhanced in frequency, length, width and complexity, and the gap junctional area enclosed by the tight junctional strands were markedly reduced. Thus, the in vitro endothelial junctional complex resembled their in vivo counterpart, the tight junctions of brain capillaries, when co-cultured with astrocytes. Reciprocally, brain endothelial cells induced the astrocytic membrane assemblies to increase in concentrations by approximately 5 fold, and sometimes to form aggregates with very high concentrations (400/micron 2) which approached the concentration of the perivascular astrocytic membranes in vivo. Substituting astrocytes with fibroblasts or smooth muscle cells in co-cultures did not enhance the tight junctions in the brain endothelium. On the other hand, substituting brain endothelium with endothelium from pulmonary artery or aorta in co-cultures did not increase the concentration or induce aggregation of the assemblies in the astrocytes. Thus, the two close neighbors in vivo, brain endothelium and astrocytes, interact specifically in vitro to induce development of membrane specializations which resemble those at the site of the blood-brain barrier.

Local blood flow and vascular permeability of autonomic ganglion-transplants in the brain.

The purpose of this study was to determine whether the blood vessels of transplanted neural tissue retain their functional characteristics. Quantitative autoradiography was used to measure local blood flow (F) with iodoantipyrine and the blood-to-tissue transfer constant (K) with alpha-aminoisobutyric acid in superior cervical ganglion (SCG) allografted to the surface of ventricle IV and into the cerebellum of the same rat. The F of the intraparenchymal grafts was slightly lower than that of the intraventricular grafts; F decreased between 1 and 4 weeks in SCG grafts at both sites. The permeability-surface area (PS) product of the microvessels and extraction fraction of AIB were calculated from these results and indicated restricted transvascular passage of the amino acid in both the in situ and grafted SCG. Surface area (S) and average length (L) of the microvessels were determined morphometrically and their permeability (P) was calculated from these data. Although K and PS decreased in the grafts compared to in situ SCG, a comparable decrease in S indicated that P was similar for the microvessels of both in situ and 1-week-old SCG transplants: 3.5-4.3 x 10(-6) cm/s. Between 1 and 4 weeks after transplantation, the P of the microvessels decreased to approximately 1.6-2.3 x 10(-6) cm/s without any change in S. Thus, the blood vessels of SCG grafts within or upon the brain initially retain the functional attributes of in situ SCG microvessels, but the average permeability of the graft microvessels decreases to approximately one half of the initial value by 4 weeks after transplantation.

Tight junctions of brain endothelium in vitro are enhanced by astroglia.

The belts of endothelial tight junctions, which impede diffusion between blood and brain, were reduced to fragmentary, small junctions in subcultured brain endothelium. When cocultured with the capillaries' nearest neighbor, the astrocytes, these endothelial tight junctions were enhanced in length, width, and complexity, as seen by en face views of the cell membranes with freeze-fracture electron microscopy. Gap junctions, common in brain endothelium in vitro but absent in mature brain capillaries in vivo, were markedly diminished in area from among the enhanced tight junctions of the cocultures. Thus, astrocytes in vitro play a role in the formation, extent, and configuration of the junctional complexes in brain endothelium, whose diffusion barrier may likewise be influenced by astrocytes in vivo.

Focal circumvention of blood-brain barrier with grafts of muscle, skin and autonomic ganglia.

The efficacy with which circulating horseradish peroxidase (HRP) spreads from transplants into the brain's interstitial spaces (IS), was assessed by 3 factors: graft type, site and age. Pieces of skeletal muscle, skin or entire superior cervical ganglion (SCG) were inserted into the IV ventricle (ventricular) or substance of the brain (parenchymal). The age of the grafts, i.e. the intervals after transplantation, were 1, 3, 6 and 12 months. Generally, HRP spread into the IS to about the same extent from ventricular muscle and skin autografts--1 mm, but less from parenchymal SCG allografts--0.5 mm. The spread from all grafts--ventricular and parenchymal--diminished with time. Exudation distance from muscle was the same as that from skin grafts for the first 6 months, but by 1 year, the penetration was significantly greater from muscle than from skin transplants. The flow of HRP was more extensive from parenchymal SCG grafts than from parenchymal muscle and skin grafts at 6 and 12 months. In some of the 6 and 12 month old parenchymal grafts of muscle and skin, no detectable HRP was extravasated. HRP consistently penetrated the brain more deeply from ventricular skin and muscle grafts than from parenchymal ones because more tissue mass survived in ventricular than in parenchymal autografts. Though care was taken not to damage the brain surface during ventricular insertion, there was a consistent, vigorous, collateral sprouting of, as yet unidentified, cranial nerves. These sprouts innervated muscle and skin autografts which, consequently, were able to survive for at least 1 year and contained vessels permeable to HRP. Allografts of muscle between inbred strains did not become innervated, survived for only 2 months and contained the central, barrier type of vessels, but not their intrinsic, permeable type. Thus, it is the muscle cell or its basal lamina within muscle grafts that determines the type of surviving vessel. In SCG allografts, even when all their ganglion cells had disappeared, leaving only connective tissue, Schwann cells and their basal lamina, the ganglion's capillaries survived and remained permeable to HRP. Therefore, the characteristics of the SCG vessels are determined by the Schwann cell-fibroblast milieu rather than the neuronal one.

Alterations of the blood-brain barrier after transplantation of autonomic ganglia into the mammalian central nervous system.

Autonomic (superior cervical) ganglia, the vessels of which are freely permeable to macromolecules, from mature rat donors (allografts or autografts) were transplanted to different sites in the central nervous system (CNS). Minimal trauma was caused by grafts into the IVth ventricle while grafts to intraparenchymal locations such as cerebral cortex and spinal cord were necessarily traumatic and produced glial scarring. Postoperative periods were between 4 weeks and 30 months. A potentially significant aspect of neural transplantation is the functional vascular connections between host and graft. It is highly likely that grafting procedures alter the blood-brain barrier (BBB) in the recipient brain. In order to determine permanent BBB changes, the glycoprotein horseradish peroxidase (HRP) (M.W. 40,000) was injected intravascularly for circulation periods ranging between 50 seconds and 90 minutes. Protein exudation was monitored by using the chromogens DAB and the highly sensitive TMB. All autonomic ganglia transplants, regardless of postoperative or HRP circulation times, were permeable to the injected protein; no qualitative differences were found between allografts and autografts. The blood-borne protein traversed the autonomic graft and infiltrated into the host brain for distances between 200 micron in intraparenchymal grafts to over 1 mm in intraventricular grafts; a smaller exudate was found in the intraparenchymal model than in the intraventricular site probably due to glial scarring that impeded the protein movement in the interstitial spaces. Significantly, TMB demonstrated that the systemic protein entered the cerebrospinal fluid. HRP was detected on the ventricular floor and in the perivascular spaces of the microvasculature. Transplantation of an autonomic ganglion into the brain provides a biological portal that bypasses normal barriers to macromolecules. The vascular and extracellular confluences between host and graft could provide direct access for systematically administered substances to enter brain regions where they, normally, would be excluded.

Muscle grafts as entries for blood-borne proteins into the extracellular space of the brain.

Nuchal muscle autografts of two different sizes were transplanted into rat brain parenchyma (intraparenchymal, 1.5 X 1.5 X 1 mm) and onto the surface of the brain stem (intraventricular, 2 X 2 X 1 and 1.5 X 1.5 X 1 mm). The vasculature of the transplants retained its permeability to proteins. Exogenous, intravenously injected horseradish peroxidase (HRP) and endogenous immunoglobulins (IgG) crossed the vessels of the grafts to enter the surrounding brain tissue 1 and 3 months after transplantation. HRP infiltrated about 0.46 to 4.6 mm into the extracellular spaces around the grafts 60 minutes after its intravenous injection. The penetration of HRP depended on the size and age of the graft. Infiltration was greater in 1-month-old rats with slightly larger intraventricular grafts than in those with smaller grafts. There was a tendency for the penetration of HRP to be greater from 1-month-old grafts than from 3-month-old grafts, but the difference was not statistically significant, except for the horizontal vector of spread in the intraparenchymal group. Although endogenous IgG infiltrated the surrounding brain tissue, its penetration was very limited in comparison with that of HRP. The results suggest that muscle grafts could be used as a readily available and accessible means of circumventing the blood-brain barrier selectively and focally.