- Email: [email protected]
Anionic sites on the luminal surface of fenestrated and continuous capillaries of the CNS
Brain Research, 363 (1986) 265-271 Elsevier
265
BRE 11375
Anionic Sites on the Luminal Surface of Fenestrated and Continuous Capillaries of the CNS JAMES W. SCHMIDLEY t and STEVEN L. WISSIG2
1Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44106 and 2Department of Anatomy, University of California School of Medicine, San Francisco, CA 94143 ( U.S.A. ) (Accepted May 14th, 1985)
Key words: endothelium - - fenestration - - capillaries - - cationic ferritin - - blood-brain barrier
We injected intravenously cationic ferritin, pI 8.4, and studied patterns of labeling of the luminal surface of capillaries in the CNS of rats. Cationic ferritin consistently labeled the luminal aspect of the diaphragms of fenestrated capillaries in the choroid plexus, median eminence and pineal. No other feature of these endothelial cells was consistently labeled. Diaphragms spanning the mouths of vesicles open to the lumen were not labeled. Occasional ferritin molecules were seen in the BL of these fenestrated vessels. The luminal surfaces of endothelial cells of the continuous capillaries of the brain, and of reactive capillaries proliferating in a region of cold injury necrosis, were not labeled. Ferritin was not seen in cytoplasmic structures or the basal lamina of these vessels. The findings in the fenestrated capillaries of CNS are in agreement with those reported for other fenestrated endothelia. The absence of labeling of the continuous capillaries of the blood-brain barrier differs from findings reported for other continuous capillaries, with the exception of the blood-air barrier portions of lung capillaries.
INTRODUCTION
molecules on the PM of BBB capillaries, 'reactive' capillaries in regions of cold-induced brain injury,
In recent years, several groups of investigators have used cationic ferritin (CF) as a probe for anionic molecules on the plasma m e m b r a n e (PM) of endothelial cellsl,S,6,12q< 16-18. These studies have revealed a heterogenous distribution of CF-binding
and fenestrated capillaries in the CP, median eminence (ME), and pineal gland of the rat.
sites on the PM of continuous and fenestrated endothelial cells (ECs). The regions (or 'microdo-
CF (pI 8.4) was obtained from Miles Laboratories, Elkhart, IN, as an 11.0 mg/ml solution in sterile 0.15 M
mains') of PM which label heavily with CF presumably represent areas of m e m b r a n e rich in anionic molecules, such as sulfated glycosaminoglycans, sialoglycoproteins or other acidic proteins 15A6. The distribution of CF labeling on fenestrated endothelial cells of capillaries in the choroid plexus6,12 (CP) and area postrema6 (one of the circumventricular organs) has been studied, but continuous ( b l o o d - b r a i n barrier, BBB) capillaries of the CNS and 'reactive' capillaries proliferating in regions of brain necrosis 8,9 have
NaC1. It was dialyzed against 0.067 M phosphate buffer containing 37 g sodium-ethylenediamine tetraacetate per liter, p H 7.0, for 24 h, then against Gey's balanced salt solution (GBSS), for an additional 24 h. The solution was concentrated to 27 mg/ml using an Amicon Diaflo ultrafiltration system. S p r a g u e - D a w l e y rats of either sex were used. After induction of surgical anesthesia with pentobarbital, 50 mg/kg intraperitoneally, a right parasagittal scalp incision was made, and the skull was exposed. The right hemispheres of two rats were lesioned by the application of a 3 m m diameter brass probe,
not been investigated. We used CF to study the distribution of anionic
MATERIALS AND METHODS
Correspondence: J.W. Schmidley, Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44106, U.S.A. 0006-8993/86/$03.50© 1986 Elsevier Science Publishers B.V. (Biomedical Division)
266 +
. . . . . . . . . . . . . . .
~
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
.+:J] + +
' ......
.
++
+ .
+" " + < "
"~.
__
5 L
..
;~ - y + ,+~++ ' .....+
........ .......
@+
s717+. . . .
.
+
.......
; ~ 7 . ~ =
Figs. 1, 2. Choroid plexus. Clusters of CF label the diaphragms of several endothelial fenestrae (closed arrows). The smaller collections of CF molecules (open arrows) on the luminal surface of the EC probably represent the 'edges' of clusters of CF molecules located on diaphragms out of the plane of section. Single molecules of CF (arrowheads) are seen in the plasma of the unperfused lumen of the vessel. In this and all succeeding figures: L, lumen; B, basal lamina; E, erythrocyte in lumen of vessel. Bar: 0.1 pm.
267 cooled to -65 °C with dry ice-acetone, to the intact skull for 2 min. The scalp was then closed with wound clips and the animals recovered uneventfully. Ten days after the cold lesions, these animals, plus 2 normal animals were anesthetized with pentobarbital, 50 mg/kg intraperitoneally, and given CF intravenously in doses ranging from 4 to 11 mg/100 g body weight. Two minutes following the end of the injection, which took approximately 30 s to complete, the animals were fixed for electron microscopy by transcardiac perfusion of 10 ml of GBSS with 4% polyvinylpyrollidinone (PVP), followed by 2% glutaraldehyde, 1% paraformaldehyde in 100 mM sodiumphosphate buffer, pH 7.4, containing 7.3 mM KC1 and 4% PVP. The brain was allowed to fix in situ for 3 h, then 1 x 1 mm blocks of normal and lesioned cerebral cortex, cerebellum, corpus striatum, median eminence, pancreas, pineal and choroid plexus were removed and postfixed in 1% OsO4 in 100 mM phosphate buffer with or without 1.5% potassium ferricyanide, for 1-3 h. Some of the blocks were further postfixed in 1% aqueous uranyl acetate for 1 h. The blocks were then rinsed with water, dehydrated in acetone and propylene oxide and embedded in a Polybed-Araldite mixture. Thick (1 ~m) sections were stained with toluidine blue, and appropriate areas selected for ultrathin sectioning. Grids were stained with 0.6% lead citrate 20 and examined using a Siemens Elmiskop I electron microscope. RESULTS
Fenestrated capillaries in the CNS Except for the presence of CF molecules, the ultrastructural appearance of these vessels was as previously described2-4,6,7,10-12.19. In CP, the most conspicuous feature was heavy labeling of the diaphragms of endothelial fenestrae (Figs. 1, 2). The luminal surface of these structures was invariably marked by a roughly triangular 'stack' of CF molecules, usually 4-5 rows deep. This labeling was strict-
ly confined to the fenestral diaphragm and did not extend onto adjacent PM. Tangential sections gave an en face view of CF clustered on what were probably fenestral diaphragms (Fig. 3). Rare patches of the luminal PM of the ECs were labeled with a single or double layer of CF molecules. These labeled patches were not consistently associated with any other ultrastructural feature of the EC, such as tight junctions, or openings of vesicles, (Figs. 1, 2). We postulate that these represented clusters of CF molecules adherent to a diaphragm, which, because of an oblique plane of the ultrathin section, was not visualized. Occasional cytoplasmic membrane-bound vesicles contained a cluster of CF molecules, which did not appear to be adherent to the vesicular membrane. An occasional single molecule, or small cluster of several molecules of CF was seen in the basal lamina (BL). No feature of the BL or abluminal surface of the EC was preferentially labeled. 'CF molecules remaining in the lumen of unperfused vessels were usually single (Figs. 1, 3) although rare large clusters of CF molecules were also seen. In the fenestrated capillaries of the ME and pineal, the distribution of CF was similar in most respects to that described for CP (Figs. 4, 5). The ECs of capillaries in these regions contained vesicles, open to either the luminal or abluminal front of the cell, the mouths of which were spanned by diaphragms. These diaphragms, in particular those of vesicles open to the luminal surface, were never labeled with CF (Fig. 6). In the ME only, some of the fenestral diaphragms were not labeled with CF (Fig. 7). In pineal only, rare 'gaps' in ECs, seemingly plugged with large masses of CF molecules, were seen (Fig. 8). The endothelium adjacent to these gaps did not appear to be damaged.
Continuous capillaries in the CNS A rare single molecule of CF, or a short row of molecules was seen on the luminal PM of ECs in the cerebellum, caudate-putamen, and cerebral cortex of the normal animals and of the non-lesioned re-
Fig. 3. Tangential view of capillary wall shows collections of CF molecules on diaphragms of fenestrae seen 'en face'. Bar: 0.1/~m. Fig. 4. Median eminence. Clusters of CF label the diaphragms of 3 fenestrae (closed arrows). The collections of CF marked by open arrows probably represent the edges of clusters of CF molecules on diaphragms out of the plane of section. Bar: 0.1/~m. Fig. 5. Pineal. Clusters of CF label the diaphragms of two endothelial fenestrae. Bar: 0.2 ~tm.
268
7
/ •
Fig. 6. Median eminence. Several vesicles, with mouths spanned by diaphragms, are seen open to the luminal and abluminal surfaces of the vessel. These structures were never the site of CF deposits. Bar: 0.2~m. Fig. 7. Median eminence. CF labels only one of 3 diaphragms in this segment of the EC. Bar: 0.2/~m.
269 gions of the brains of those animals subjected to cold injury. CF was not seen in the BL, in vesicles, or in any other structure within the EC or elsewhere in the neuropil (not illustrated). The ultrastructural appearance of these capillaries was as described previously3.4,11,19. The region of cold injury had an appearance similar to that described by Mitchell et al. 8,9 The capillaries proliferating within the necrotic zone had an increased number of vesicles compared with BBB capillaries, but neither these structures nor any other ultrastructural feature of the EC was consistently labeled by CF (Fig. 9). As was the case for capillaries in normal brain, an occasional single molecule or small patch of CF molecules was present on the luminal PM. CF was not seen in the BL. DISCUSSION Our experiments show that the diaphragms of endothelial fenestrae in the choroid plexus and other circumventricular organs (CVOs), like their counterparts in fenestrated visceral capillaries, are heavily labeled after i.v. administration of CF 17,18. The remainder of the luminal surface of the capillaries of these regions was the site of only occasional sparse deposits of CF. In the ME, some of the fenestral diaphragms were not labeled by CF. Small amounts of CF reached the BL of all fenestrated capillary beds examined. Although others have used CF as a probe for anionic 'microdomains' on the luminal or abluminal surface of ECs, including two studies on fenestrated capillaries in the CNS 6,12, methodological variations make comparisons difficult. The factors to be considered in comparing and contrasting our results with those of others include: pI and dose of CF used, whether the CF was injected intravenously or perfused in a balanced salt solution, length of time the CF circulated before fixation, and whether the tissue was fixed by vascular perfusion or immersion. Peress and Tompkins 12 administered ferritin preparations ranging in pI from 3.9-5.1 to 10-12, at a dose of 15
mg/100 g b. wt., and fixed choroid plexus 15 rain later by immersion in glutaraldehyde. They found that high concentrations of CF with a pI of 7.9-10 had reached the endothelial BL of the choroid plexus after 15 min circulation, and described aggregates of tracer within, adjacent to, or on endothelial fenestrations. Dermietzel et al. 6 gave ferritin preparations ranging in pI from 4.5 to 9.3, but used 40 mg/100 g b. wt. They fixed by vascular perfusion or by superfusing the IVth ventricle, after the ferritin had circulated for 30 min. This group, studying CP and area postrema, obtained results very similar to ours with CF preparations of similar pI (7.8 and 8.8). They noted identical results with both methods of fixation. In both studies, ferritin preparations of lower pI did not selectively label any feature of the EC and reached the BL to a minimal extent or not at all, while more highly cationic preparations (pI 9.3-12) were endocytosed and transported to multivesicular bodies, and reached the BL in higher concentrations. Studies of the fenestrated capillaries of several non-CNS organs, using techniques similar to ours (similar doses of CF with pI 7.4-8.4 and fixation within 2 min or less of the i.v. administration of the CF), have demonstrated that the diaphragms of most endothelial cells label heavily with CF. Simionescu et al. 18found approximately 80% of diaphragms in pancreatic capillaries, fixed within 2 min of tracer administration, were labeled. They also found, as we did, that diaphragms spanning the mouths of vesicles were not labeled. Bankston and Milicil found some capillary beds in which virtually all fenestral diaphragms were labeled by CF and others in which none of the diaphragms was labeled. Both groups of investigators noted large aggregates on the diaphragms, identical to the 'stacks' of CF molecules we have illustrated. Simionescu et al. have further demonstrated that the anionic molecules on the fenestral diaphragms of pancreatic capillaries are probably heparan sulfate-containing proteoglycans (HSPG)a6. Further investigation of other fenestrated endothelia, including those in the CNS, should reveal whether the negative charge on diaphragms is contributed
Fig. 8. Pineal. An extremely large cluster of CF molecules plugs a "gap"in the EC. CF molecules are clustered on two other fenestrae in, and one fenestration out of, the plane of section. Bar: 0.1/zm. Fig. 9. Region of cold injury. CF molecules are not seen on the surface of the EC, in vesicles or in the BL. Bar: 0.5 pm.
270 by HSPG, or similar molecules. From a functional standpoint, the capillaries of CP and CVOs are permeable to horseradish peroxidase and microperoxidase2-4,10,19. Although it has often been assumed the fenestral diaphragms would allow this large protein molecule to pass freely, the available evidence does not necessarily support this conclusion. As pointed out by Simionescu et al. 15, fenestrated endothelia are actually less permeable to protein than continuous endothelia, while having higher permeability to H 2 0 and small molecules. This may be because negatively charged fenestral diaphragms exclude plasma proteins (which are mostly anionic) on an electrostatic basis. Continuous capillaries have not been as intensively studied with CF. Simionescu and Simionescu 14 noted plasmalemmal CF binding only on the thicker vesicle-containing portions of ECs of continuous capillaries of the lung. The thinner, vesicle-free air-blood barrier portions of the ECs were devoid of CF labeling, as are BBB ECs. Pietra et al. 13 found heavy labeling of luminal PM and luminal vesicles in the continuous capillaries of the lung, fixed after 3 min of exposure to CF. Their findings are in marked contrast to our finding that the continuous capillaries of the BBB are virtually not labeled by CF, but because they used an isolated, perfused lung preparation, a definitive conclusion that the luminal PM of lung capillaries is more strongly anionic than that of the BBB is not warranted. We should note, however, that Simionescu et a1,18 found that labeling of fenestrated ECs in pancreas was the same regardless of whether the tracer was injected intravenously into an intact animal or perfused through the vasculature of the intact pancreas in a physiological saline solution.
REFERENCES 1 Bankston, P.W. and Milici, A.J., A survey of the binding of polycationic ferritin in several fenestrated capillary beds: indication of heterogeneity in the luminal glycocalyx of fenestral diaphragms, Microvasc. Res., 26 (1983) 36-48. 2 Becket, N.H., Novikoff, A.B. and Zimmerman, H.M., Fine structure observations of the uptake of intravenously injected peroxidase by the rat choroid plexus, J. Histochem. Cytochem., 15 (1967) 160-165. 3 Brightman, M.W., Reese, T.S. and Feder, N., Assessment with the electron microscope of the permeability to peroxidase of cerebral endothelium and epithelium in mice and
Cavallo et al. 5 studied the continuous capillaries of skeletal muscle, both in the normal state and after injury or exposure to serotonin. They noted irregularly spaced patches of CF on the luminal surface, which increased in size and number after thermal injury or exposure to serotonin. Their results also stand in contrast to ours for the continuous capillaries of the CNS, despite similarities in experimental conditions (they used CF with a pI of 7.7-8.5, allowed it to circulate for 10 min, and gave 30 mg per 'adult' rat, weight unspecified). It is possible that the reactive capillaries in the cold lesion, if studied at earlier times after injury, would have shown an increased labeling with CF, as did the continuous capillaries of the cremaster studied by Cavallo et al. Our findings in the CP, pineal and ME confirm those reported by Dermietzel et al.6, in CP and another CVO, the area postrema. Our finding that CF does not label any feature of the PM of continuous capillaries of the BBB or of reactive capillaries in a region of cold injury stands in contrast to the findings for fenestrated endothelia, and other continuous endothelia, and suggests yet another way in which the BBB differs from other endothelia. ACKNOWLEDGEMENTS This work was carried out during the tenure of a Clinician Scientist Award from the American Heart Association (J.W.S.) and with funds contributed in part by A H A , California affiliate. The work was also supported by N I H Grant NS-14543. We thank Debra Crumrine and Patricia Blue for excellent technical and photographic assistance, and Betty Freed for typing.
sharks. In C. Crone and N.A. Lassen (Eds.), Capillary Permeability, Proceedings of the 2nd Alfred Benzon Symposium, Munksgaard, Copenhagen, 1970, pp. 468-476. 4 Brightman, M.W., Intracerebral movement of proteins injected into blood and CSF of mice. In A. Lajtha and D.H. Ford (Eds.), Brain Barrier Systems, Progress in Brain Research, Vol. 29, Elsevier, Amsterdam, 1968, pp. 19-40, 5 Cavallo, T., Graves, K. and Granholm, N.A., Endothelial and perivascular anionic sites during immediate transient vascular leakage, VirchowsArch. A, 388 (1980) 1-12. 6 Dermietzel, R., Thurauf, N. and Kalweit, P., Surface charges associated with fenestrated brain capillaries. II. In vivo studies on the role of molecular charge in endothelial
271 permeability, ]. Ulstrastruct. Res., 84 (1983) 111-119. 7 Weindl, A. and Joynt, R.J., The median eminence as a circumventricular organ. In K.M. Knigge, D.E. Scott and A. Weindl (Eds.), Brain-Endocrine Interaction: Median Eminence: Structure and Function, Karger, Basel, 1972, p. 280. 8 Mitchell, J., Weller, R.O. and Evans, H., Capillary regeneration following thermal lesions of the mouse cerebral cortex. An ultrastructural study, Acta Neuropath., 44 (1978) 167-171. 9 Mitchell, J., Weller, R.O. and Evans, H., Reestablishment of the blood-brain barrier to peroxidase following cold injury to mouse cortex, Acta Neuropath., 46 (1979) 45-49. 10 Moller, M., Van Deurs, B. and Westergaard, E., Vascular permeability to proteins and peptides in the mouse pineal gland, Cell Tissue Res., 195 (1978) 1-15. 11 Peters, A., Palay, S. and Webster H. deF., The Fine Structure of the Nervous System, W.B. Saunders, Philadelphia, 1976. 12 Peress, N.S. and Tompkins, D., Effect of molecular charge on choroid plexus permeability: tracer studies with cationized ferritins, Cell Tissue Res., 219 (1981) 425-431. 13 Pietra, G.G., Sampson, P., Larken, P., Hansen-Flaschem, J. and Fishman, A.P., Transcapillary movement of cationized ferritin in the isolated perfused rat lung, Lab. Invest., 49 (1983) 54-61.
14 Simionescu, D. and Simionescu, M., Differentiated distribution of the cell surface charge on the alveolar-capillary unit. Characteristic paucity of anionic sites on the airblood barrier, Microvasc. Res., 25 (1983) 85-100. 15 Simionescu, M., Simionescu, N. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. Distribution of lectin receptors, J. Cell Biol., 94 (1982) 406-413. 16 Simionescu, M., Simionescu, N., Silbert, .I.E. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. II. Partial characterization of their anionic sites, J. Cell Biol., 90 (1981) 614-621. 17 Simionescu, N., Cellular aspects of transcapillary exchange, Physiol. Rev., 63 (1983) 1536-1579. 18 Simionescu, N., Simionescu, M. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. I. Preferential distribution of anionic sites, J. Cell Biol., 90 (1981) 605-613. 19 Van Deurs, B., Structural aspects of brain barriers, with special reference to the permeability of the cerebral endothelium and choroidal epithelium, b~t. Rev. Cytol., 65 (1980) 117-191. 20 Venable, J.H. and Coggeshall, R.E., A simplified lead citrate stain for use in electron microscopy, J. Cell Biol., 25 (1965) 407-408.
265
BRE 11375
Anionic Sites on the Luminal Surface of Fenestrated and Continuous Capillaries of the CNS JAMES W. SCHMIDLEY t and STEVEN L. WISSIG2
1Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44106 and 2Department of Anatomy, University of California School of Medicine, San Francisco, CA 94143 ( U.S.A. ) (Accepted May 14th, 1985)
Key words: endothelium - - fenestration - - capillaries - - cationic ferritin - - blood-brain barrier
We injected intravenously cationic ferritin, pI 8.4, and studied patterns of labeling of the luminal surface of capillaries in the CNS of rats. Cationic ferritin consistently labeled the luminal aspect of the diaphragms of fenestrated capillaries in the choroid plexus, median eminence and pineal. No other feature of these endothelial cells was consistently labeled. Diaphragms spanning the mouths of vesicles open to the lumen were not labeled. Occasional ferritin molecules were seen in the BL of these fenestrated vessels. The luminal surfaces of endothelial cells of the continuous capillaries of the brain, and of reactive capillaries proliferating in a region of cold injury necrosis, were not labeled. Ferritin was not seen in cytoplasmic structures or the basal lamina of these vessels. The findings in the fenestrated capillaries of CNS are in agreement with those reported for other fenestrated endothelia. The absence of labeling of the continuous capillaries of the blood-brain barrier differs from findings reported for other continuous capillaries, with the exception of the blood-air barrier portions of lung capillaries.
INTRODUCTION
molecules on the PM of BBB capillaries, 'reactive' capillaries in regions of cold-induced brain injury,
In recent years, several groups of investigators have used cationic ferritin (CF) as a probe for anionic molecules on the plasma m e m b r a n e (PM) of endothelial cellsl,S,6,12q< 16-18. These studies have revealed a heterogenous distribution of CF-binding
and fenestrated capillaries in the CP, median eminence (ME), and pineal gland of the rat.
sites on the PM of continuous and fenestrated endothelial cells (ECs). The regions (or 'microdo-
CF (pI 8.4) was obtained from Miles Laboratories, Elkhart, IN, as an 11.0 mg/ml solution in sterile 0.15 M
mains') of PM which label heavily with CF presumably represent areas of m e m b r a n e rich in anionic molecules, such as sulfated glycosaminoglycans, sialoglycoproteins or other acidic proteins 15A6. The distribution of CF labeling on fenestrated endothelial cells of capillaries in the choroid plexus6,12 (CP) and area postrema6 (one of the circumventricular organs) has been studied, but continuous ( b l o o d - b r a i n barrier, BBB) capillaries of the CNS and 'reactive' capillaries proliferating in regions of brain necrosis 8,9 have
NaC1. It was dialyzed against 0.067 M phosphate buffer containing 37 g sodium-ethylenediamine tetraacetate per liter, p H 7.0, for 24 h, then against Gey's balanced salt solution (GBSS), for an additional 24 h. The solution was concentrated to 27 mg/ml using an Amicon Diaflo ultrafiltration system. S p r a g u e - D a w l e y rats of either sex were used. After induction of surgical anesthesia with pentobarbital, 50 mg/kg intraperitoneally, a right parasagittal scalp incision was made, and the skull was exposed. The right hemispheres of two rats were lesioned by the application of a 3 m m diameter brass probe,
not been investigated. We used CF to study the distribution of anionic
MATERIALS AND METHODS
Correspondence: J.W. Schmidley, Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44106, U.S.A. 0006-8993/86/$03.50© 1986 Elsevier Science Publishers B.V. (Biomedical Division)
266 +
. . . . . . . . . . . . . . .
~
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
.+:J] + +
' ......
.
++
+ .
+" " + < "
"~.
__
5 L
..
;~ - y + ,+~++ ' .....+
........ .......
@+
s717+. . . .
.
+
.......
; ~ 7 . ~ =
Figs. 1, 2. Choroid plexus. Clusters of CF label the diaphragms of several endothelial fenestrae (closed arrows). The smaller collections of CF molecules (open arrows) on the luminal surface of the EC probably represent the 'edges' of clusters of CF molecules located on diaphragms out of the plane of section. Single molecules of CF (arrowheads) are seen in the plasma of the unperfused lumen of the vessel. In this and all succeeding figures: L, lumen; B, basal lamina; E, erythrocyte in lumen of vessel. Bar: 0.1 pm.
267 cooled to -65 °C with dry ice-acetone, to the intact skull for 2 min. The scalp was then closed with wound clips and the animals recovered uneventfully. Ten days after the cold lesions, these animals, plus 2 normal animals were anesthetized with pentobarbital, 50 mg/kg intraperitoneally, and given CF intravenously in doses ranging from 4 to 11 mg/100 g body weight. Two minutes following the end of the injection, which took approximately 30 s to complete, the animals were fixed for electron microscopy by transcardiac perfusion of 10 ml of GBSS with 4% polyvinylpyrollidinone (PVP), followed by 2% glutaraldehyde, 1% paraformaldehyde in 100 mM sodiumphosphate buffer, pH 7.4, containing 7.3 mM KC1 and 4% PVP. The brain was allowed to fix in situ for 3 h, then 1 x 1 mm blocks of normal and lesioned cerebral cortex, cerebellum, corpus striatum, median eminence, pancreas, pineal and choroid plexus were removed and postfixed in 1% OsO4 in 100 mM phosphate buffer with or without 1.5% potassium ferricyanide, for 1-3 h. Some of the blocks were further postfixed in 1% aqueous uranyl acetate for 1 h. The blocks were then rinsed with water, dehydrated in acetone and propylene oxide and embedded in a Polybed-Araldite mixture. Thick (1 ~m) sections were stained with toluidine blue, and appropriate areas selected for ultrathin sectioning. Grids were stained with 0.6% lead citrate 20 and examined using a Siemens Elmiskop I electron microscope. RESULTS
Fenestrated capillaries in the CNS Except for the presence of CF molecules, the ultrastructural appearance of these vessels was as previously described2-4,6,7,10-12.19. In CP, the most conspicuous feature was heavy labeling of the diaphragms of endothelial fenestrae (Figs. 1, 2). The luminal surface of these structures was invariably marked by a roughly triangular 'stack' of CF molecules, usually 4-5 rows deep. This labeling was strict-
ly confined to the fenestral diaphragm and did not extend onto adjacent PM. Tangential sections gave an en face view of CF clustered on what were probably fenestral diaphragms (Fig. 3). Rare patches of the luminal PM of the ECs were labeled with a single or double layer of CF molecules. These labeled patches were not consistently associated with any other ultrastructural feature of the EC, such as tight junctions, or openings of vesicles, (Figs. 1, 2). We postulate that these represented clusters of CF molecules adherent to a diaphragm, which, because of an oblique plane of the ultrathin section, was not visualized. Occasional cytoplasmic membrane-bound vesicles contained a cluster of CF molecules, which did not appear to be adherent to the vesicular membrane. An occasional single molecule, or small cluster of several molecules of CF was seen in the basal lamina (BL). No feature of the BL or abluminal surface of the EC was preferentially labeled. 'CF molecules remaining in the lumen of unperfused vessels were usually single (Figs. 1, 3) although rare large clusters of CF molecules were also seen. In the fenestrated capillaries of the ME and pineal, the distribution of CF was similar in most respects to that described for CP (Figs. 4, 5). The ECs of capillaries in these regions contained vesicles, open to either the luminal or abluminal front of the cell, the mouths of which were spanned by diaphragms. These diaphragms, in particular those of vesicles open to the luminal surface, were never labeled with CF (Fig. 6). In the ME only, some of the fenestral diaphragms were not labeled with CF (Fig. 7). In pineal only, rare 'gaps' in ECs, seemingly plugged with large masses of CF molecules, were seen (Fig. 8). The endothelium adjacent to these gaps did not appear to be damaged.
Continuous capillaries in the CNS A rare single molecule of CF, or a short row of molecules was seen on the luminal PM of ECs in the cerebellum, caudate-putamen, and cerebral cortex of the normal animals and of the non-lesioned re-
Fig. 3. Tangential view of capillary wall shows collections of CF molecules on diaphragms of fenestrae seen 'en face'. Bar: 0.1/~m. Fig. 4. Median eminence. Clusters of CF label the diaphragms of 3 fenestrae (closed arrows). The collections of CF marked by open arrows probably represent the edges of clusters of CF molecules on diaphragms out of the plane of section. Bar: 0.1/~m. Fig. 5. Pineal. Clusters of CF label the diaphragms of two endothelial fenestrae. Bar: 0.2 ~tm.
268
7
/ •
Fig. 6. Median eminence. Several vesicles, with mouths spanned by diaphragms, are seen open to the luminal and abluminal surfaces of the vessel. These structures were never the site of CF deposits. Bar: 0.2~m. Fig. 7. Median eminence. CF labels only one of 3 diaphragms in this segment of the EC. Bar: 0.2/~m.
269 gions of the brains of those animals subjected to cold injury. CF was not seen in the BL, in vesicles, or in any other structure within the EC or elsewhere in the neuropil (not illustrated). The ultrastructural appearance of these capillaries was as described previously3.4,11,19. The region of cold injury had an appearance similar to that described by Mitchell et al. 8,9 The capillaries proliferating within the necrotic zone had an increased number of vesicles compared with BBB capillaries, but neither these structures nor any other ultrastructural feature of the EC was consistently labeled by CF (Fig. 9). As was the case for capillaries in normal brain, an occasional single molecule or small patch of CF molecules was present on the luminal PM. CF was not seen in the BL. DISCUSSION Our experiments show that the diaphragms of endothelial fenestrae in the choroid plexus and other circumventricular organs (CVOs), like their counterparts in fenestrated visceral capillaries, are heavily labeled after i.v. administration of CF 17,18. The remainder of the luminal surface of the capillaries of these regions was the site of only occasional sparse deposits of CF. In the ME, some of the fenestral diaphragms were not labeled by CF. Small amounts of CF reached the BL of all fenestrated capillary beds examined. Although others have used CF as a probe for anionic 'microdomains' on the luminal or abluminal surface of ECs, including two studies on fenestrated capillaries in the CNS 6,12, methodological variations make comparisons difficult. The factors to be considered in comparing and contrasting our results with those of others include: pI and dose of CF used, whether the CF was injected intravenously or perfused in a balanced salt solution, length of time the CF circulated before fixation, and whether the tissue was fixed by vascular perfusion or immersion. Peress and Tompkins 12 administered ferritin preparations ranging in pI from 3.9-5.1 to 10-12, at a dose of 15
mg/100 g b. wt., and fixed choroid plexus 15 rain later by immersion in glutaraldehyde. They found that high concentrations of CF with a pI of 7.9-10 had reached the endothelial BL of the choroid plexus after 15 min circulation, and described aggregates of tracer within, adjacent to, or on endothelial fenestrations. Dermietzel et al. 6 gave ferritin preparations ranging in pI from 4.5 to 9.3, but used 40 mg/100 g b. wt. They fixed by vascular perfusion or by superfusing the IVth ventricle, after the ferritin had circulated for 30 min. This group, studying CP and area postrema, obtained results very similar to ours with CF preparations of similar pI (7.8 and 8.8). They noted identical results with both methods of fixation. In both studies, ferritin preparations of lower pI did not selectively label any feature of the EC and reached the BL to a minimal extent or not at all, while more highly cationic preparations (pI 9.3-12) were endocytosed and transported to multivesicular bodies, and reached the BL in higher concentrations. Studies of the fenestrated capillaries of several non-CNS organs, using techniques similar to ours (similar doses of CF with pI 7.4-8.4 and fixation within 2 min or less of the i.v. administration of the CF), have demonstrated that the diaphragms of most endothelial cells label heavily with CF. Simionescu et al. 18found approximately 80% of diaphragms in pancreatic capillaries, fixed within 2 min of tracer administration, were labeled. They also found, as we did, that diaphragms spanning the mouths of vesicles were not labeled. Bankston and Milicil found some capillary beds in which virtually all fenestral diaphragms were labeled by CF and others in which none of the diaphragms was labeled. Both groups of investigators noted large aggregates on the diaphragms, identical to the 'stacks' of CF molecules we have illustrated. Simionescu et al. have further demonstrated that the anionic molecules on the fenestral diaphragms of pancreatic capillaries are probably heparan sulfate-containing proteoglycans (HSPG)a6. Further investigation of other fenestrated endothelia, including those in the CNS, should reveal whether the negative charge on diaphragms is contributed
Fig. 8. Pineal. An extremely large cluster of CF molecules plugs a "gap"in the EC. CF molecules are clustered on two other fenestrae in, and one fenestration out of, the plane of section. Bar: 0.1/zm. Fig. 9. Region of cold injury. CF molecules are not seen on the surface of the EC, in vesicles or in the BL. Bar: 0.5 pm.
270 by HSPG, or similar molecules. From a functional standpoint, the capillaries of CP and CVOs are permeable to horseradish peroxidase and microperoxidase2-4,10,19. Although it has often been assumed the fenestral diaphragms would allow this large protein molecule to pass freely, the available evidence does not necessarily support this conclusion. As pointed out by Simionescu et al. 15, fenestrated endothelia are actually less permeable to protein than continuous endothelia, while having higher permeability to H 2 0 and small molecules. This may be because negatively charged fenestral diaphragms exclude plasma proteins (which are mostly anionic) on an electrostatic basis. Continuous capillaries have not been as intensively studied with CF. Simionescu and Simionescu 14 noted plasmalemmal CF binding only on the thicker vesicle-containing portions of ECs of continuous capillaries of the lung. The thinner, vesicle-free air-blood barrier portions of the ECs were devoid of CF labeling, as are BBB ECs. Pietra et al. 13 found heavy labeling of luminal PM and luminal vesicles in the continuous capillaries of the lung, fixed after 3 min of exposure to CF. Their findings are in marked contrast to our finding that the continuous capillaries of the BBB are virtually not labeled by CF, but because they used an isolated, perfused lung preparation, a definitive conclusion that the luminal PM of lung capillaries is more strongly anionic than that of the BBB is not warranted. We should note, however, that Simionescu et a1,18 found that labeling of fenestrated ECs in pancreas was the same regardless of whether the tracer was injected intravenously into an intact animal or perfused through the vasculature of the intact pancreas in a physiological saline solution.
REFERENCES 1 Bankston, P.W. and Milici, A.J., A survey of the binding of polycationic ferritin in several fenestrated capillary beds: indication of heterogeneity in the luminal glycocalyx of fenestral diaphragms, Microvasc. Res., 26 (1983) 36-48. 2 Becket, N.H., Novikoff, A.B. and Zimmerman, H.M., Fine structure observations of the uptake of intravenously injected peroxidase by the rat choroid plexus, J. Histochem. Cytochem., 15 (1967) 160-165. 3 Brightman, M.W., Reese, T.S. and Feder, N., Assessment with the electron microscope of the permeability to peroxidase of cerebral endothelium and epithelium in mice and
Cavallo et al. 5 studied the continuous capillaries of skeletal muscle, both in the normal state and after injury or exposure to serotonin. They noted irregularly spaced patches of CF on the luminal surface, which increased in size and number after thermal injury or exposure to serotonin. Their results also stand in contrast to ours for the continuous capillaries of the CNS, despite similarities in experimental conditions (they used CF with a pI of 7.7-8.5, allowed it to circulate for 10 min, and gave 30 mg per 'adult' rat, weight unspecified). It is possible that the reactive capillaries in the cold lesion, if studied at earlier times after injury, would have shown an increased labeling with CF, as did the continuous capillaries of the cremaster studied by Cavallo et al. Our findings in the CP, pineal and ME confirm those reported by Dermietzel et al.6, in CP and another CVO, the area postrema. Our finding that CF does not label any feature of the PM of continuous capillaries of the BBB or of reactive capillaries in a region of cold injury stands in contrast to the findings for fenestrated endothelia, and other continuous endothelia, and suggests yet another way in which the BBB differs from other endothelia. ACKNOWLEDGEMENTS This work was carried out during the tenure of a Clinician Scientist Award from the American Heart Association (J.W.S.) and with funds contributed in part by A H A , California affiliate. The work was also supported by N I H Grant NS-14543. We thank Debra Crumrine and Patricia Blue for excellent technical and photographic assistance, and Betty Freed for typing.
sharks. In C. Crone and N.A. Lassen (Eds.), Capillary Permeability, Proceedings of the 2nd Alfred Benzon Symposium, Munksgaard, Copenhagen, 1970, pp. 468-476. 4 Brightman, M.W., Intracerebral movement of proteins injected into blood and CSF of mice. In A. Lajtha and D.H. Ford (Eds.), Brain Barrier Systems, Progress in Brain Research, Vol. 29, Elsevier, Amsterdam, 1968, pp. 19-40, 5 Cavallo, T., Graves, K. and Granholm, N.A., Endothelial and perivascular anionic sites during immediate transient vascular leakage, VirchowsArch. A, 388 (1980) 1-12. 6 Dermietzel, R., Thurauf, N. and Kalweit, P., Surface charges associated with fenestrated brain capillaries. II. In vivo studies on the role of molecular charge in endothelial
271 permeability, ]. Ulstrastruct. Res., 84 (1983) 111-119. 7 Weindl, A. and Joynt, R.J., The median eminence as a circumventricular organ. In K.M. Knigge, D.E. Scott and A. Weindl (Eds.), Brain-Endocrine Interaction: Median Eminence: Structure and Function, Karger, Basel, 1972, p. 280. 8 Mitchell, J., Weller, R.O. and Evans, H., Capillary regeneration following thermal lesions of the mouse cerebral cortex. An ultrastructural study, Acta Neuropath., 44 (1978) 167-171. 9 Mitchell, J., Weller, R.O. and Evans, H., Reestablishment of the blood-brain barrier to peroxidase following cold injury to mouse cortex, Acta Neuropath., 46 (1979) 45-49. 10 Moller, M., Van Deurs, B. and Westergaard, E., Vascular permeability to proteins and peptides in the mouse pineal gland, Cell Tissue Res., 195 (1978) 1-15. 11 Peters, A., Palay, S. and Webster H. deF., The Fine Structure of the Nervous System, W.B. Saunders, Philadelphia, 1976. 12 Peress, N.S. and Tompkins, D., Effect of molecular charge on choroid plexus permeability: tracer studies with cationized ferritins, Cell Tissue Res., 219 (1981) 425-431. 13 Pietra, G.G., Sampson, P., Larken, P., Hansen-Flaschem, J. and Fishman, A.P., Transcapillary movement of cationized ferritin in the isolated perfused rat lung, Lab. Invest., 49 (1983) 54-61.
14 Simionescu, D. and Simionescu, M., Differentiated distribution of the cell surface charge on the alveolar-capillary unit. Characteristic paucity of anionic sites on the airblood barrier, Microvasc. Res., 25 (1983) 85-100. 15 Simionescu, M., Simionescu, N. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. Distribution of lectin receptors, J. Cell Biol., 94 (1982) 406-413. 16 Simionescu, M., Simionescu, N., Silbert, .I.E. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. II. Partial characterization of their anionic sites, J. Cell Biol., 90 (1981) 614-621. 17 Simionescu, N., Cellular aspects of transcapillary exchange, Physiol. Rev., 63 (1983) 1536-1579. 18 Simionescu, N., Simionescu, M. and Palade, G.E., Differentiated microdomains on the luminal surface of capillary endothelium. I. Preferential distribution of anionic sites, J. Cell Biol., 90 (1981) 605-613. 19 Van Deurs, B., Structural aspects of brain barriers, with special reference to the permeability of the cerebral endothelium and choroidal epithelium, b~t. Rev. Cytol., 65 (1980) 117-191. 20 Venable, J.H. and Coggeshall, R.E., A simplified lead citrate stain for use in electron microscopy, J. Cell Biol., 25 (1965) 407-408.