- Email: [email protected]
The three-dimensional organization of plasmalemmal vesicular profiles in the endothelium of rat heart capillaries
MICROVASCULAR
RESEARCH
25, 3.58-368 (1983)
The Three-Dimensional Organization of Plasmalemmal Vesicular Profiles in the Endothelium of Rat Heart Capillaries MAGNUS Depurtment
BUNDGAARD,
PIA HAGMAN,
AND CHRISTIAN
CRONE
of Medical Physiology A, The Panum Institute, Blegdomsve~ 3, DK-2200 Copenhagen
N, Denmurk
Receilaed J~rly 29, 1982 The organization of plasmalemmal vesicular profiles in the endothelium of rat heart capillaries has been reinvestigated. Judged from random thin sections approximately 50% of the vesicles appeared free in the cytoplasm, the rest opening to the surfaces of the endothelial cells-a distribution which corroborates previous studies. However, threedimensional reconstructions based on ultrathin serial sections (thickness -12 nm) gave a very different picture. All plasmalemmal vesicular profiles (921 from 5 capillaries) were parts of the surface membrane either as caveolae or as more complex racemose invaginations. This organization has previously been observed in frog mesenteric capillaries ((M. Bundgaard, J. Frekjaer-Jensen, and C. Crone, 1979, Proc. Nat. Acud. Sci. USA 76, 6439-6442) and (J. Frekjaer-Jensen, 1980, J. Ultrastmct. Res. 73, 9-20)). It is therefore proposed that absence or extreme rarity of free plasmalemmal vesicles is a general feature of capillary endothelia. Consequently, we suggest that the term “endothelial. plasmalemmal vesicles” be replaced by “endothelial plasmalemmnl invoginations.” The results imply that transendothelial vesicular transport is unlikely to occur and that this membrane system performs other-as yet unknown-functions.
INTRODUCTION The endothelial cells of most blood capillaries contain a conspicuous high number of plasmalemmal vesicular profiles. They are commonly called “endothelial plasmalemmal vesicles,” a term which includes both single profiles fused with the plasmalemma (caveolae) and profiles which appear as free vesicles in the cytoplasm (Bruns and Palade, 1968a). The function of the vesicles is assumed to be transendothelial transport of macromolecules (Bruns and Palade, 1968b), which implies that the vesicles shuttle between the luminal and abluminal plasmalemma (Palade and Bruns, 1968). This picture of the endothelial plasmalemmal vesicles is under revision. Independent experiments with cationized ferritin (Simionescu and Simionescu, 1981) and native ferritin (Clough and Michel, 1981) have led to the view that the caveolae are permanent structures and that macromolecular transport occurs via vesicles which shuttle between stationary caveolae. Another revision was based on an accidental observation we made on frog mesenteric capillaries (Bund358 0026-2862183 $3.00 CopyrIght 0 1983 by Academic Press. lnc All rights of reproduction in any form reserved. Printed in U.S.A.
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
359
gaard et al., 1979). Tannic acid was added to the tissue after fixation for electron microscopy and the labeling of the plasmalemmal vesicular profiles with tannic acid indicated that all the profiles communicated with the extracellular space. This led to the proposal that not only the caveolae, but also apparently free vesicles are part of more or less permanent invaginations of the plasmalemmaan organization which implies a questioning of the role of plasmalemmal vesicles in transendothelial transport. Three-dimensional reconstructions based on ultrathin serial sections of the same capillaries confirmed this interpretation (FrokjaerJensen, 1980). Our examinations of the organization of endothelial plasmalemmal vesicles have now been extended to a well characterized mammalian endothelium. We show by serial sectioning that our revised concept of the vesicles also applies to the endothelium of heart capillaries. This supports the view that the organization of plasmalemmal vesicular profiles in caveolae and racemose invaginations of the surface membrane is a general feature of capillary endothelia. We propose that the term “endothelial plasmalemmal vesicles” be replaced by “endothelial plasmalemmal invaginations.” MATERIALS
AND METHODS
The heart of an adult male rat (Wistar; 350 g) was used in the present investigation. The animal was anesthetized with pentobarbital (17.5 mg i.p.). Fixation for electron microscopy. The heart was excised as quickly as possible and immersed in a fixative solution containing 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3). Five to ten minutes later small pieces of tissue were selected from the surface of the left ventricle while the heart was covered by fixative. After 15hr aldehyde fixation and subsequent buffer rinse the tissue was postfixed for 2 hr in a reduced osmium solution according to the method described by Karnovsky (1971). In order to maximize membrane contrast the tissue blocks were immersed in 1% tannic acid (No. 1764, Mallinckrodt Inc., St. Louis, MO.) for 45 min (Simionescu and Simionescu, 1976) and stained in 2% aqueous uranylacetate at 60” for 15 hr. After dehydration in a graded series of ethanol followed by propylene oxide, the blocks were embedded in Epon resins. Serial sectioning and analysis of the material. Thin sections were cut with a diamond knife mounted on an LKB-ultratome. One successful series, including 16 consecutive sections of an average thickness about 12 nm, was obtained. The ribbon was collected on Formvar-covered slot grids and poststained with 5% uranylacetate in 50% methanol at 40” for 45 min and with lead citrate a.m. Reynolds (1963) for 10 min. The sections were examined in a Zeiss 10 B electron microscope operated at 60 kV. The instrumental magnification was calibrated with a carbon replica of an optical grating (568 lines/mm). Nine cross-sectioned capillary profiles were present in the sections. The technical quality of five profiles turned out to be satisfactory throughout the series. Micrographs of the wall of these capillary profiles in 16 consecutive sections were taken at an instrumental magnification of 31,500 x . The photographic prints which had a final magnification of 78,500 x , were assembled into montages of the capillary profiles-altogether 80 montages.
360
BUNDGAARD,
HAGMAN,
AND
CRONE
Only vesicles and arrangements of fused vesicles which were totally contained in the series were included in the examination. Each vesicular profile in the endothelium of the five capillaries in the eighth section of the series was given a number, which was written on the electron micrographs. After these vesicles had been followed and numbered in the neighboring sections the remaining unlabeled vesicles in section 6 and 10 were numbered and followed. In this way the complete three-dimensional organization of 921 plasmalemmal vesicles was accounted for. The outer diameter of 141 vesicles, chosen randomly among the numbered vesicles, was measured with a calibrated magnifying glass on prints from the sections where the vesicles had their largest diameter. RESULTS Preliminary experiments showed that the three-dimensional organization of plasmalemmal vesicles cannot be determined from consecutive sections of conventional thickness (50-70 nm), where crucial structures such as the stricture between fused vesicles and the neck regions of the caveolae often are lost or difficult to identify with certainty due to overlapping electron-dense material. Further, in projections of sections with a thickness similar to the vesicular diameter (270 nm), the number of free vesicles will inevitably be overestimated (Bundgaard, 1983). With decreasing thickness topological resolution is increased. When a vesicle is represented in five to six consecutive sections it can be followed unequivocally from section to section and a stricture or a neck region will appear well defined in two to three sections. In average a vesicle was represented in five to six consecutive sections in the present series. The average outer diameter of the vesicles was 68 nm (SD = 7.5 nm; n = 141). Consequently the average section thickness was approximately 12 nm. Organization of the plasmalemmal vesicles. The three-dimensional organization of a total number of 921 plasmalemmal vesicles in the endothelium of five capillaries were reconstructed. All these vesicles opened to the extracellular space either directly as simple caveolae or indirectly as parts of racemose invaginations of the plasmalemma. The result is in striking contrast to the finding that 52% of the plasmalemmal vesicles appear free in projections of random individual thin sections (Table 1). How the spatial relations between vesicles can be resolved in ultrathin, consecutive sections is illustrated in Fig. I. The distribution between caveolae and racemose invaginations could not be determined precisely. The 16 consecutive sections constitute only 0.2 pm of capillary length, and more elaborate invaginations were therefore not all fully contained within the series. It is, however, evident that caveolae dominate the picture. Of the plasmalemmal vesicles which were traced in this examination, 74% were caveolae. The racemose invaginations were preferentially observed in the thicker parts (>0.2 pm) of the endothelial cells. Usually 2-3 vesicles were included in the complexes but racemose invaginations made up by as much as 10 vesicles were reconstructed. Transendothelial vesicular channels which can be identified with certainty, if present in ultrathin serial sections, were not encountered. Organization of other endothelial vesicles. In addition to plasmalemmal vesicles,
Organization of endothelial plasmalemmal vesicles reconstructed from consecutive. ultrathin sections
Apparent distribution of plasmalemmal vesicles obtained from random thin sections
COMPARATIVE
36 I3
Rat heart Rat heart 48 28
16 I8 45 IO
Frog mesenterp Rat heart
% Open
to
ENDOTHIXIUM
25 52
68 65 39 80
52 72
39 35
16 17 I6 IO
to 9 Free in the the the lumen cytoplasm interstitium
% Open
Rabbit lung Frog mesentery Frog mesentery Frog mesentery
Tissue
I
VESICLES OF THF. CAPILLARY
TABLE DATA ON PI ASMALEMMAL
Frokjazr-Jensen Present study
(1980)
Mazzone and Kornblau (1981) Clough and Michel (1981) Clough (1982) Bundgaard and FrokjaxJensen (1982) Simionescu et nl. (1974) Present study
Reference
F
? 21 F: $ i
z z E s
Fi i2
2ir:
E
E
$
2
362
BUNDGAARD,
HAGMAN,
AND
CRONE
FIG. I. Electron micrographs showing a segment of a capillary endothelium in 12 consecutive sections of the rat heart. The capillary lumen is to the left. Notice the characteristic endothelial projection (Smith e/ (I/., 1971) which starts to appear in the first section and gradually grows throughout the series. A vesicle, called number I, starts to appear in section (a). In the next two sections it is a typical vesicular profile, apparently free in the cytoplasm. In section (d) there is an electron-dense spot at the lower edge of vesicle I. The spot grows into a vesicular profile (called vesicle 2) in section (e) where vesicle I and 2 are clearly fused. In sections (f-h) vesicle I disappears and a new
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
363
which are not “true” vesicles but invaginations of the plasmalemma, the endothelial cytoplasm contains vesicles similar to those found in other animal cells types. The morphology of these true vesicles differed from the morphology of plasmalemmal vesicles. They were generally smaller (about 40 nm in diameter), their content was usually remarkably electron lucent (Fig. I), and they were located in close proximity to smooth-surfaced cisternae. The three-dimensional reconstructions revealed that many circular membrane profiles with electron-lucent content were parts of structures of irregular shapes. For instance, some profiles, 40-60 nm in diameter, turned out to be tubular structures which could be followed throughout the cntirc scrics of sections. Other circular membrane profiles had the same diameter (60-80 nm) as plasmalemmal vesicles. However, in most cases they represented structures distinct from plasmalemmal vesicles. For example, the diameter of the profiles decreased abruptly to 40 nm and then remained constant in the following six to eight sections. illustrating a spherical structure with a “tail”. In other cases a circular profile gradually flattened and fused with a process of the smooth-surfaced cisternae of the cell (Fig. 2). Altogether, of the numerous membrane profiles with the characteristic electron-lucent content, encountered in the consecutive sections, six profiles turned out to represent vesicles with diameters similar to the diameter of plasmalemmal vesicles. Endothelial cells also contain coated pits and coated vesicles, probably involved in receptor-mediated uptake of macromolecules as seen in other cell types (Goldstein et [I/., 1979). In this investigation only four coated pits were encountcrcd. DISCUSSION Our three-dimensional reconstructions of segments of the capillary endothelium in the rat heart have demonstrated that the characteristic plasmalemmal vesicles are not “true” vesicles but parts of invaginations of the plasmalemma. Since the findings corroborate our previous studies on frog mesenteric capillaries it seems likely that this organization of plasmalemmal vesicles is a general feature of capillary endothelia. The present results also illustrate that it may be hazardous to guess about the three-dimensional organization of organelles from projections of individual thin sections. Serial sectioning is a prerequisite for assessment of structures in three dimensions, a requirement which is rarely met. Further, the thickness of the consecutive sections must be rigorously adjusted to the dimensions of the structures under investigation in order to obtain a reliable reconstruction: vesicle (number 3) which is fused with vesicle 2 develops. In section (h) vesicle 2 approaches the luminal plasmalemma and in section (i) vesicle 2 opens to the lumen. In the following sections vesicle 2 disappears, vesicle 3 appears free and ends as a barely visible dark \pot in the last section. Thus. vesicles I and 3. which appeared free in the cytoplasm in scvcral sections. are fused with vesicle 2, which opens to the lumen. The curved arrow in section (a) indicates a complex arrangement of fused vesicles which clearly opens to the interstitium. In the following sections the complex is segregated into apparently free vesicles. The straight arrows in sections (g-j) mark a vesicle which is jkcr in the cytoplasm. Such vesicles, which differ from plasmalemmal vesicles by their smaller diameter C-40 nm) and electron-lucent content, were regularly observed in relation to smooth-surfaced cisternae. Bar. 100 nm.
364
FIG. 2 Ellect se:ctions of a rat tPle samet size an’ is COtlSplicuoL Isly is fused with al
BUNDGAARD,
HAGMAN,
AND CRONE
I micrographs showing a segment of the capillary endotheliul m in six consec :utive ,art capillary. In sections (a) and (b) the arrow marks a me:mbr ane profile with nape as that of the surrounding plasmalemmal vesicles. The c:ontr :nt of the plrofile xtron lucent. In sections (c-e) the profile gradually flattens and in section (f) it :ess of a smooth-surfaced cisterna (arrows). Bar, 100 nm.
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
365
if the sections are too thick a false picture of the true geometry is created (Bundgaard, 1983). Since these technical requirements have been ignored in most previous studies on endothelial plasmalemmal vesicles major revisions of the available information on this organelle are required. The possibility that fixation for electron microscopy influences the organization of plasmalemmal vesicles must be considered. Experiments have clearly shown that plasmalemmal vesicles are present in the endothelium of rapidly frozen lung capillaries (Mazzone and Kornblau, 1981). The endothelial plasmalemmal vesicles as such are consequently not fixation artifacts. However, aldehyde fixation does influence the population of vesicles. In rapidly frozen endothelium the diameter of the vesicles, their volume density, and the relative number of apparently free vesicles were smaller than in capillary endothelia fixed by perfusion with aldehydes. None of these differences indicate that the vesicular invaginations are created by vesicle-vesicle and vesicle-plasmalemma fusions during the initial stages of the aldehyde fixation. On the contrary, if such fusions occurred the percentage of vesicles which appear free in the cytoplasm should be highest in the rapidly frozen endothelium. The capillary endothelium contains-like other animal cells-different classes of true vesicles which are involved in transport processes within the cells or between the cell and its environment (Farquhar, 1978). Our demonstration of some free vesicles within the endothelial cytoplasm could consequently be expected even if the plasmalemmal vesicles are parts of invaginations. Further, all the true vesicles encountered could be differentiated from plasmalemmal vesicles by their size, shape, or by their content. It cannot be ruled out, however, that the six free vesicles seen, with diameters as plasmalemmal vesicles, but with more electron-lucent content, represent plasmalemmal vesicles which have pinched off from the invaginations and subsequently modified the composition of their content. Even with this possibility in mind it remains clear that free plasmalemmal vesicles are very rare-in contrast to previous figures based on examination of random individual thin sections (cf. Table I). The possibility that the number of free plasmalemmal vesicles is very small has previously been mentioned by Karnovsky and Shea (1970). They found that the endothelium of heart capillaries in the mouse contained very few unlabeled vesicles after lanthanum staining en bloc. The labeled vesicles without obvious openings to the cell surface were assumed to communicate with the extracellular space out of the plane of sectioning. The unlabeled vesicular profiles were considered to be plasmalemmal vesicles and the authors suggested a model of vesicular transport in which it was taken into account that the number of freely movable vesicles might be much smaller than generally believed. Frokjaer-Jensen (1980) found less than 1% free vesicles and Chien et al. (1982) in arterial endothelium also found a low figure (8%). There are reasons to believe that the latter figure is an overestimate. Studies from Simionescu’s laboratory have also successively come to lower figures for free (unattached) vesicles-lo% is mentioned (Simionescu and Simionescu, 1981). Thus, several recent studies argue against the early views that a large fraction of “vesicles” are freely movable and support the notion we put forward on circumstantial evidence (Bundgaard et ul., 1979). The absence or rarity of free plasmalemmal vesicles implies that speculations about their role in transendothelial transport of macromolecules need to be
366
BUNDGAARD,
HAGMAN,
AND
CRONE
reconsidered. An organization of the vesicles in more or less stationary invaginations is incompatible with a moving front of transporting vesicles (Simionescu et al., 1973, 1975). We have, therefore, reinterpreted previous tracer studies on capillary endothelia in the light of our revised picture of the vesicular organization (Bundgaard et ul., 1979; Bundgaard, 1983). This led to the conclusion that earlier observations which were interpreted as indicative of vesicular transport are equally compatible with diffusion of the tracers into vesicular invaginations. Other investigations directly support the view that the plasmalemmal vesicles communicate either with the pericapillary space or the lumen. Thus, recent experiments with cationized ferritin (CF), which binds to the outer surface of plasma membranes, indicate that vesicular transport may be absent in capillary endothelia (Simionescu and Simionescu, 1981). Labeling of the luminal aspect of an epithelium with CF should lead to focal labelings of the abluminal plasmalemma if vesicles shuttle between the two cell surfaces. This was not seen. Transport by coated pinocytotic vesicles across the rat choroid plexus epithelium and guinea pig yolk sac epithelium has been demonstrated by the same technique (van Deurs et al., 1981; King, 1982). Coated vesicles in vascular endothelium were observed by Bruns and Palade ( 1968a). Our observations on the organization of “endothelial, plasmalemmal vesicles” call for a change of terminology. We have demonstrated that these structures are not true vesicles and the word, vesicles, should consequently be avoided. We suggest that these characteristic elements of the endothelial cells be called “endothelial plusmalemmal invaginations,” which can be subdivided into caveolae and more complex racernose invuginations. The implication of the present findings is that our understanding of the functional significance of the plasmalemmal invaginations is still as limited as when the structures were discovered (Palade, 1953). Their generally accepted function, vesicular transport, must be seriously questioned. The fact, that pericytes (Rhodin, 1968; Weibel, 1974) and smooth muscle cells (Somlyo, 1980) are provided with structures similar to endothelial plasmalemmal invaginations speaks against transcellular transport functions. The function of the plasmalemmal invaginations may have nothing to do with permeability but rather be related to metabolic functions of the endothelium. It has clearly been shown in several studies that angiotensinconverting enzyme is present along the luminal plasma membrane, including caveolae, of lung capillary endothelium (Ryan, 1982). Also, the invaginations may function as a membrane reserve for cell types capable of significant variations of their shape (Wolff, 1977). Which particular function that is performed by the extensive system of additional plasmalemmal membrane remains to be elucidated. REFERENCES BRUNS, R. R., AND PALADE, G. E. (1968a). Studies on blood capillaries. I. General organization of blood capillaries in muscle. J. Cell Bid. 37, 244-276. BRUNS, R. R., AND PALADE, G. E. (1968b). Studies on blood capillaries. II. Transport of ferritin molecules across the wall of muscle capillaries. .I. Cell Bid. 37, 277-299. BUNDGAARD, M. (1980). Transport pathways in capillaries-in search of pores. Annu. Rev. Physid. 42, 325-336. BUNDGAARD, M. (1983). Vesicular transport in capillary endothelium-does it occur? Fed. Proc., in press.
PLASMALEMMAL
VESICLES IN HEART CAPILLARIES
367
BUNWAARD, M., AND FRGKJAER-JENSEN,J. (1982). Functional aspects of the ultrastructure of terminal blood vessels: A quantitative study on consecutive segments of the frog mesenteric microvasculature. Microvasc. Res. 23, l-30. BUNDGAARD, M., FRQKJAER-JENSEN,J., AND CRONE. C. (1979). Endothelial plasmalemmal vesicles as elements in a system of branching invaginations from the cell surface. Proc. Nar. Acad. Sci. USA 76, 6439-6442. CHI~N, S., LAUFLR, L., AND HANDLEY, D. A. (1982). Vesicle distribution in the arterial endothelium determined with ruthenium red as an extracellular marker. J. Ulfraslruct. Res. 79, 198-206. CLOUGH, G. (1982). The steady-state transport of cationized ferritin by endothelial cell vesicles. J. Physiol. 328, 389-401. CLOUGH, G., AND MICHEL, C. C. (1981). The role of vesicles in the transport of ferritin through frog endothelium. J. Physiol. 315, 127-142. CRONE, C., AND LEVITT, D. G. (1983). Exchange of small solutes through the capillary walls. In “Handbook of Physiology” (E. M. Kenkin. and C. C. Michel, eds.), Sect. 2, Vol. 3, Amer. Physiol. Sot., Washington, D.C., in press. FARQUHAR, M. G. (1978). Traffic of products and membranes through the Golgi complex. In “Transport of Macromolecules in Cellular Systems” (S. C. Silverstein, ed.), pp. 341-362. Abakon Vertagsgesellschaft, Berlin. FR~KJAER-JENSEN, J. (1980). Three-dimensional organization of plasmalcmmal vesicles in endothelial cells. An analysis by serial sectioning of frog mesenteric capillaries. J. U/trustruct. RPS. 73, 920. GOLDSTEIN, J. L., ANDERSON, R. G. W., AND BKOWN. M. S. (1979). Coated pits, coated vesicles and receptor-mediated endocytosis. Nature (London) 279, 679-685. KARNOVSKY, M. J. (1971). Use of ferrocyanide-reduced osmium tetroxide in electron microscopy. Proceedings qf rhe 11th Annual Meeting of rhe Atrwricun Socirry for Cell Biology. 146a. KARNOVSKY. M. J., AND SHEA, A. M. (1970). Transcapillary transport by pinocytosis. Microvusc,. Res. 2, 353-360. KING, B. F. (1982). The role of coated vesicles in selective transfer across yolk sac epithelium. J. Ultrustruct. Res. 79, 273-284. MAZZONE, R. W., AND KORNBLAU, S. M. (1981). Pinocytotic vesicles in the endothelium of rapidly frozen rabbit lung. Microwsc. Res. 21, 193-21 I. PALADE, Cr. E. (1953). Fine structure of blood capillaries. J. Appl. Phys. 24, 1424 (Abstract). PALADE, G. E., AND BRLJNS,R. R. (1968). Structural modulations of plasmalemmal vesicles. J. Ce// Biol. 37, 633-653. PALADF, G. E., SIMIONFSCIJ, M., AND SIMIONESCU, N. (1979). Structural aspects of the permeability of the microvascular endothelium. Acfu Physiol. Sund. S14ppl. 463, 1 l-32. REYNOLDS, E. S. (1963). The use of lead citrate at high pH as an electron-opaque stain in electron microscopy. J. Cell Biol. 17, 208-213. RHODIN, J. A. G. (1968). Ultrastructure of mammalian venous capillaries, venules. and small collecting veins. J. Ultrastrrrcf. Res. 25, 452-500. RYAN, U. S. (1982). Structural bases for metabolic activity. Annrr. Re19. Physiol. 44, 223-239. SIMIONESCU, N.. AND SIMIONESCU, M. (1976). Galloylglucoses of low molecular weight as mordant in electron microscopy. I-II. J. Cell B&l. 70, 608-633. SIMIONESCU.N.. AND SIMIONESCU,M. (1981). Hydrophilic pathways of capillary endothelium, a dynamic system. In “Water- Transpor-t acr.oss Epithelia” (H. H. Ussing. N. B. Bindslev, and 0. StenKnudsen, eds.), pp. 228-247. Munksgaard, Copenhagen, SIMIONESCLI, N.. SIMIONESCLI. M.. AND PAI ADF. G. E. (1973). Permeability of muscle capillaries to exogenous myoglobin. J. Cell Bid. 57, 424-451. SIMIONESTU. N., SIMIONESCU. M., AND PAI.ADE. G. E. (1974). Morphometric data on the endothelium of blood capillaries. .I. Cell Biol. 60, 128-157. SIMIONESCU, N., SIMIONES~U. M., AND PAI.ADF, G. E. (1975). Permeability of muscle capillaries to small hemepeptides. Evidence for the existence of transendothelial channels, J. Ce// B;o/. 64, 586 607. SMITH, U., RYAN, J. W.. MICHIE, D. D., AND SMITH. D. S. (1971). Endothelial projections as revealed by scanning electron microscopy. Scirnc,c 173, 925-927. SOMI YO, A. V. (1980). tiltrastructure of vascular smooth muscle. /n “Handbook of Physiology” (D. F. Bohr, A. V. Somlyo, H. V. Sparks, and S. R. Geiger-. eds.), Sect. 2, Vol. 2. pp. 33-68, Amer. Physiol. Sot. Washington, D.C.
368
BUNDGAARD,
HAGMAN,
AND
CRONE
WAGNER, R. C., AND CASLEY-SMITH, J. R. (1981). Endothelial vesicles. Microvasc. Res. 21, 267-298. VAN DEURS, B., BULOW, F., AND M@LLER. M. (1981). Vesicular transport of cationized ferritin by
the epithelium of the rat choroid plexus. J. Ceil Bid. 89, 131-139. E. R. (1974). On pericytes, particularly their existence on lung capillaries. Microvnsr. RPS.
WEIBEL,
8, 218-235.
WOLFF,J. R. (1977). Ultrastructure of the terminal vascular bed as related to function. In “Microcirculation” (G. Kaiey, and B. M. Altura, eds.), pp. 95-130. Vol. I. Univ. Park Press, Baltimore.
RESEARCH
25, 3.58-368 (1983)
The Three-Dimensional Organization of Plasmalemmal Vesicular Profiles in the Endothelium of Rat Heart Capillaries MAGNUS Depurtment
BUNDGAARD,
PIA HAGMAN,
AND CHRISTIAN
CRONE
of Medical Physiology A, The Panum Institute, Blegdomsve~ 3, DK-2200 Copenhagen
N, Denmurk
Receilaed J~rly 29, 1982 The organization of plasmalemmal vesicular profiles in the endothelium of rat heart capillaries has been reinvestigated. Judged from random thin sections approximately 50% of the vesicles appeared free in the cytoplasm, the rest opening to the surfaces of the endothelial cells-a distribution which corroborates previous studies. However, threedimensional reconstructions based on ultrathin serial sections (thickness -12 nm) gave a very different picture. All plasmalemmal vesicular profiles (921 from 5 capillaries) were parts of the surface membrane either as caveolae or as more complex racemose invaginations. This organization has previously been observed in frog mesenteric capillaries ((M. Bundgaard, J. Frekjaer-Jensen, and C. Crone, 1979, Proc. Nat. Acud. Sci. USA 76, 6439-6442) and (J. Frekjaer-Jensen, 1980, J. Ultrastmct. Res. 73, 9-20)). It is therefore proposed that absence or extreme rarity of free plasmalemmal vesicles is a general feature of capillary endothelia. Consequently, we suggest that the term “endothelial. plasmalemmal vesicles” be replaced by “endothelial plasmalemmnl invoginations.” The results imply that transendothelial vesicular transport is unlikely to occur and that this membrane system performs other-as yet unknown-functions.
INTRODUCTION The endothelial cells of most blood capillaries contain a conspicuous high number of plasmalemmal vesicular profiles. They are commonly called “endothelial plasmalemmal vesicles,” a term which includes both single profiles fused with the plasmalemma (caveolae) and profiles which appear as free vesicles in the cytoplasm (Bruns and Palade, 1968a). The function of the vesicles is assumed to be transendothelial transport of macromolecules (Bruns and Palade, 1968b), which implies that the vesicles shuttle between the luminal and abluminal plasmalemma (Palade and Bruns, 1968). This picture of the endothelial plasmalemmal vesicles is under revision. Independent experiments with cationized ferritin (Simionescu and Simionescu, 1981) and native ferritin (Clough and Michel, 1981) have led to the view that the caveolae are permanent structures and that macromolecular transport occurs via vesicles which shuttle between stationary caveolae. Another revision was based on an accidental observation we made on frog mesenteric capillaries (Bund358 0026-2862183 $3.00 CopyrIght 0 1983 by Academic Press. lnc All rights of reproduction in any form reserved. Printed in U.S.A.
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
359
gaard et al., 1979). Tannic acid was added to the tissue after fixation for electron microscopy and the labeling of the plasmalemmal vesicular profiles with tannic acid indicated that all the profiles communicated with the extracellular space. This led to the proposal that not only the caveolae, but also apparently free vesicles are part of more or less permanent invaginations of the plasmalemmaan organization which implies a questioning of the role of plasmalemmal vesicles in transendothelial transport. Three-dimensional reconstructions based on ultrathin serial sections of the same capillaries confirmed this interpretation (FrokjaerJensen, 1980). Our examinations of the organization of endothelial plasmalemmal vesicles have now been extended to a well characterized mammalian endothelium. We show by serial sectioning that our revised concept of the vesicles also applies to the endothelium of heart capillaries. This supports the view that the organization of plasmalemmal vesicular profiles in caveolae and racemose invaginations of the surface membrane is a general feature of capillary endothelia. We propose that the term “endothelial plasmalemmal vesicles” be replaced by “endothelial plasmalemmal invaginations.” MATERIALS
AND METHODS
The heart of an adult male rat (Wistar; 350 g) was used in the present investigation. The animal was anesthetized with pentobarbital (17.5 mg i.p.). Fixation for electron microscopy. The heart was excised as quickly as possible and immersed in a fixative solution containing 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3). Five to ten minutes later small pieces of tissue were selected from the surface of the left ventricle while the heart was covered by fixative. After 15hr aldehyde fixation and subsequent buffer rinse the tissue was postfixed for 2 hr in a reduced osmium solution according to the method described by Karnovsky (1971). In order to maximize membrane contrast the tissue blocks were immersed in 1% tannic acid (No. 1764, Mallinckrodt Inc., St. Louis, MO.) for 45 min (Simionescu and Simionescu, 1976) and stained in 2% aqueous uranylacetate at 60” for 15 hr. After dehydration in a graded series of ethanol followed by propylene oxide, the blocks were embedded in Epon resins. Serial sectioning and analysis of the material. Thin sections were cut with a diamond knife mounted on an LKB-ultratome. One successful series, including 16 consecutive sections of an average thickness about 12 nm, was obtained. The ribbon was collected on Formvar-covered slot grids and poststained with 5% uranylacetate in 50% methanol at 40” for 45 min and with lead citrate a.m. Reynolds (1963) for 10 min. The sections were examined in a Zeiss 10 B electron microscope operated at 60 kV. The instrumental magnification was calibrated with a carbon replica of an optical grating (568 lines/mm). Nine cross-sectioned capillary profiles were present in the sections. The technical quality of five profiles turned out to be satisfactory throughout the series. Micrographs of the wall of these capillary profiles in 16 consecutive sections were taken at an instrumental magnification of 31,500 x . The photographic prints which had a final magnification of 78,500 x , were assembled into montages of the capillary profiles-altogether 80 montages.
360
BUNDGAARD,
HAGMAN,
AND
CRONE
Only vesicles and arrangements of fused vesicles which were totally contained in the series were included in the examination. Each vesicular profile in the endothelium of the five capillaries in the eighth section of the series was given a number, which was written on the electron micrographs. After these vesicles had been followed and numbered in the neighboring sections the remaining unlabeled vesicles in section 6 and 10 were numbered and followed. In this way the complete three-dimensional organization of 921 plasmalemmal vesicles was accounted for. The outer diameter of 141 vesicles, chosen randomly among the numbered vesicles, was measured with a calibrated magnifying glass on prints from the sections where the vesicles had their largest diameter. RESULTS Preliminary experiments showed that the three-dimensional organization of plasmalemmal vesicles cannot be determined from consecutive sections of conventional thickness (50-70 nm), where crucial structures such as the stricture between fused vesicles and the neck regions of the caveolae often are lost or difficult to identify with certainty due to overlapping electron-dense material. Further, in projections of sections with a thickness similar to the vesicular diameter (270 nm), the number of free vesicles will inevitably be overestimated (Bundgaard, 1983). With decreasing thickness topological resolution is increased. When a vesicle is represented in five to six consecutive sections it can be followed unequivocally from section to section and a stricture or a neck region will appear well defined in two to three sections. In average a vesicle was represented in five to six consecutive sections in the present series. The average outer diameter of the vesicles was 68 nm (SD = 7.5 nm; n = 141). Consequently the average section thickness was approximately 12 nm. Organization of the plasmalemmal vesicles. The three-dimensional organization of a total number of 921 plasmalemmal vesicles in the endothelium of five capillaries were reconstructed. All these vesicles opened to the extracellular space either directly as simple caveolae or indirectly as parts of racemose invaginations of the plasmalemma. The result is in striking contrast to the finding that 52% of the plasmalemmal vesicles appear free in projections of random individual thin sections (Table 1). How the spatial relations between vesicles can be resolved in ultrathin, consecutive sections is illustrated in Fig. I. The distribution between caveolae and racemose invaginations could not be determined precisely. The 16 consecutive sections constitute only 0.2 pm of capillary length, and more elaborate invaginations were therefore not all fully contained within the series. It is, however, evident that caveolae dominate the picture. Of the plasmalemmal vesicles which were traced in this examination, 74% were caveolae. The racemose invaginations were preferentially observed in the thicker parts (>0.2 pm) of the endothelial cells. Usually 2-3 vesicles were included in the complexes but racemose invaginations made up by as much as 10 vesicles were reconstructed. Transendothelial vesicular channels which can be identified with certainty, if present in ultrathin serial sections, were not encountered. Organization of other endothelial vesicles. In addition to plasmalemmal vesicles,
Organization of endothelial plasmalemmal vesicles reconstructed from consecutive. ultrathin sections
Apparent distribution of plasmalemmal vesicles obtained from random thin sections
COMPARATIVE
36 I3
Rat heart Rat heart 48 28
16 I8 45 IO
Frog mesenterp Rat heart
% Open
to
ENDOTHIXIUM
25 52
68 65 39 80
52 72
39 35
16 17 I6 IO
to 9 Free in the the the lumen cytoplasm interstitium
% Open
Rabbit lung Frog mesentery Frog mesentery Frog mesentery
Tissue
I
VESICLES OF THF. CAPILLARY
TABLE DATA ON PI ASMALEMMAL
Frokjazr-Jensen Present study
(1980)
Mazzone and Kornblau (1981) Clough and Michel (1981) Clough (1982) Bundgaard and FrokjaxJensen (1982) Simionescu et nl. (1974) Present study
Reference
F
? 21 F: $ i
z z E s
Fi i2
2ir:
E
E
$
2
362
BUNDGAARD,
HAGMAN,
AND
CRONE
FIG. I. Electron micrographs showing a segment of a capillary endothelium in 12 consecutive sections of the rat heart. The capillary lumen is to the left. Notice the characteristic endothelial projection (Smith e/ (I/., 1971) which starts to appear in the first section and gradually grows throughout the series. A vesicle, called number I, starts to appear in section (a). In the next two sections it is a typical vesicular profile, apparently free in the cytoplasm. In section (d) there is an electron-dense spot at the lower edge of vesicle I. The spot grows into a vesicular profile (called vesicle 2) in section (e) where vesicle I and 2 are clearly fused. In sections (f-h) vesicle I disappears and a new
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
363
which are not “true” vesicles but invaginations of the plasmalemma, the endothelial cytoplasm contains vesicles similar to those found in other animal cells types. The morphology of these true vesicles differed from the morphology of plasmalemmal vesicles. They were generally smaller (about 40 nm in diameter), their content was usually remarkably electron lucent (Fig. I), and they were located in close proximity to smooth-surfaced cisternae. The three-dimensional reconstructions revealed that many circular membrane profiles with electron-lucent content were parts of structures of irregular shapes. For instance, some profiles, 40-60 nm in diameter, turned out to be tubular structures which could be followed throughout the cntirc scrics of sections. Other circular membrane profiles had the same diameter (60-80 nm) as plasmalemmal vesicles. However, in most cases they represented structures distinct from plasmalemmal vesicles. For example, the diameter of the profiles decreased abruptly to 40 nm and then remained constant in the following six to eight sections. illustrating a spherical structure with a “tail”. In other cases a circular profile gradually flattened and fused with a process of the smooth-surfaced cisternae of the cell (Fig. 2). Altogether, of the numerous membrane profiles with the characteristic electron-lucent content, encountered in the consecutive sections, six profiles turned out to represent vesicles with diameters similar to the diameter of plasmalemmal vesicles. Endothelial cells also contain coated pits and coated vesicles, probably involved in receptor-mediated uptake of macromolecules as seen in other cell types (Goldstein et [I/., 1979). In this investigation only four coated pits were encountcrcd. DISCUSSION Our three-dimensional reconstructions of segments of the capillary endothelium in the rat heart have demonstrated that the characteristic plasmalemmal vesicles are not “true” vesicles but parts of invaginations of the plasmalemma. Since the findings corroborate our previous studies on frog mesenteric capillaries it seems likely that this organization of plasmalemmal vesicles is a general feature of capillary endothelia. The present results also illustrate that it may be hazardous to guess about the three-dimensional organization of organelles from projections of individual thin sections. Serial sectioning is a prerequisite for assessment of structures in three dimensions, a requirement which is rarely met. Further, the thickness of the consecutive sections must be rigorously adjusted to the dimensions of the structures under investigation in order to obtain a reliable reconstruction: vesicle (number 3) which is fused with vesicle 2 develops. In section (h) vesicle 2 approaches the luminal plasmalemma and in section (i) vesicle 2 opens to the lumen. In the following sections vesicle 2 disappears, vesicle 3 appears free and ends as a barely visible dark \pot in the last section. Thus. vesicles I and 3. which appeared free in the cytoplasm in scvcral sections. are fused with vesicle 2, which opens to the lumen. The curved arrow in section (a) indicates a complex arrangement of fused vesicles which clearly opens to the interstitium. In the following sections the complex is segregated into apparently free vesicles. The straight arrows in sections (g-j) mark a vesicle which is jkcr in the cytoplasm. Such vesicles, which differ from plasmalemmal vesicles by their smaller diameter C-40 nm) and electron-lucent content, were regularly observed in relation to smooth-surfaced cisternae. Bar. 100 nm.
364
FIG. 2 Ellect se:ctions of a rat tPle samet size an’ is COtlSplicuoL Isly is fused with al
BUNDGAARD,
HAGMAN,
AND CRONE
I micrographs showing a segment of the capillary endotheliul m in six consec :utive ,art capillary. In sections (a) and (b) the arrow marks a me:mbr ane profile with nape as that of the surrounding plasmalemmal vesicles. The c:ontr :nt of the plrofile xtron lucent. In sections (c-e) the profile gradually flattens and in section (f) it :ess of a smooth-surfaced cisterna (arrows). Bar, 100 nm.
PLASMALEMMAL
VESICLES
IN
HEART
CAPILLARIES
365
if the sections are too thick a false picture of the true geometry is created (Bundgaard, 1983). Since these technical requirements have been ignored in most previous studies on endothelial plasmalemmal vesicles major revisions of the available information on this organelle are required. The possibility that fixation for electron microscopy influences the organization of plasmalemmal vesicles must be considered. Experiments have clearly shown that plasmalemmal vesicles are present in the endothelium of rapidly frozen lung capillaries (Mazzone and Kornblau, 1981). The endothelial plasmalemmal vesicles as such are consequently not fixation artifacts. However, aldehyde fixation does influence the population of vesicles. In rapidly frozen endothelium the diameter of the vesicles, their volume density, and the relative number of apparently free vesicles were smaller than in capillary endothelia fixed by perfusion with aldehydes. None of these differences indicate that the vesicular invaginations are created by vesicle-vesicle and vesicle-plasmalemma fusions during the initial stages of the aldehyde fixation. On the contrary, if such fusions occurred the percentage of vesicles which appear free in the cytoplasm should be highest in the rapidly frozen endothelium. The capillary endothelium contains-like other animal cells-different classes of true vesicles which are involved in transport processes within the cells or between the cell and its environment (Farquhar, 1978). Our demonstration of some free vesicles within the endothelial cytoplasm could consequently be expected even if the plasmalemmal vesicles are parts of invaginations. Further, all the true vesicles encountered could be differentiated from plasmalemmal vesicles by their size, shape, or by their content. It cannot be ruled out, however, that the six free vesicles seen, with diameters as plasmalemmal vesicles, but with more electron-lucent content, represent plasmalemmal vesicles which have pinched off from the invaginations and subsequently modified the composition of their content. Even with this possibility in mind it remains clear that free plasmalemmal vesicles are very rare-in contrast to previous figures based on examination of random individual thin sections (cf. Table I). The possibility that the number of free plasmalemmal vesicles is very small has previously been mentioned by Karnovsky and Shea (1970). They found that the endothelium of heart capillaries in the mouse contained very few unlabeled vesicles after lanthanum staining en bloc. The labeled vesicles without obvious openings to the cell surface were assumed to communicate with the extracellular space out of the plane of sectioning. The unlabeled vesicular profiles were considered to be plasmalemmal vesicles and the authors suggested a model of vesicular transport in which it was taken into account that the number of freely movable vesicles might be much smaller than generally believed. Frokjaer-Jensen (1980) found less than 1% free vesicles and Chien et al. (1982) in arterial endothelium also found a low figure (8%). There are reasons to believe that the latter figure is an overestimate. Studies from Simionescu’s laboratory have also successively come to lower figures for free (unattached) vesicles-lo% is mentioned (Simionescu and Simionescu, 1981). Thus, several recent studies argue against the early views that a large fraction of “vesicles” are freely movable and support the notion we put forward on circumstantial evidence (Bundgaard et ul., 1979). The absence or rarity of free plasmalemmal vesicles implies that speculations about their role in transendothelial transport of macromolecules need to be
366
BUNDGAARD,
HAGMAN,
AND
CRONE
reconsidered. An organization of the vesicles in more or less stationary invaginations is incompatible with a moving front of transporting vesicles (Simionescu et al., 1973, 1975). We have, therefore, reinterpreted previous tracer studies on capillary endothelia in the light of our revised picture of the vesicular organization (Bundgaard et ul., 1979; Bundgaard, 1983). This led to the conclusion that earlier observations which were interpreted as indicative of vesicular transport are equally compatible with diffusion of the tracers into vesicular invaginations. Other investigations directly support the view that the plasmalemmal vesicles communicate either with the pericapillary space or the lumen. Thus, recent experiments with cationized ferritin (CF), which binds to the outer surface of plasma membranes, indicate that vesicular transport may be absent in capillary endothelia (Simionescu and Simionescu, 1981). Labeling of the luminal aspect of an epithelium with CF should lead to focal labelings of the abluminal plasmalemma if vesicles shuttle between the two cell surfaces. This was not seen. Transport by coated pinocytotic vesicles across the rat choroid plexus epithelium and guinea pig yolk sac epithelium has been demonstrated by the same technique (van Deurs et al., 1981; King, 1982). Coated vesicles in vascular endothelium were observed by Bruns and Palade ( 1968a). Our observations on the organization of “endothelial, plasmalemmal vesicles” call for a change of terminology. We have demonstrated that these structures are not true vesicles and the word, vesicles, should consequently be avoided. We suggest that these characteristic elements of the endothelial cells be called “endothelial plusmalemmal invaginations,” which can be subdivided into caveolae and more complex racernose invuginations. The implication of the present findings is that our understanding of the functional significance of the plasmalemmal invaginations is still as limited as when the structures were discovered (Palade, 1953). Their generally accepted function, vesicular transport, must be seriously questioned. The fact, that pericytes (Rhodin, 1968; Weibel, 1974) and smooth muscle cells (Somlyo, 1980) are provided with structures similar to endothelial plasmalemmal invaginations speaks against transcellular transport functions. The function of the plasmalemmal invaginations may have nothing to do with permeability but rather be related to metabolic functions of the endothelium. It has clearly been shown in several studies that angiotensinconverting enzyme is present along the luminal plasma membrane, including caveolae, of lung capillary endothelium (Ryan, 1982). Also, the invaginations may function as a membrane reserve for cell types capable of significant variations of their shape (Wolff, 1977). Which particular function that is performed by the extensive system of additional plasmalemmal membrane remains to be elucidated. REFERENCES BRUNS, R. R., AND PALADE, G. E. (1968a). Studies on blood capillaries. I. General organization of blood capillaries in muscle. J. Cell Bid. 37, 244-276. BRUNS, R. R., AND PALADE, G. E. (1968b). Studies on blood capillaries. II. Transport of ferritin molecules across the wall of muscle capillaries. .I. Cell Bid. 37, 277-299. BUNDGAARD, M. (1980). Transport pathways in capillaries-in search of pores. Annu. Rev. Physid. 42, 325-336. BUNDGAARD, M. (1983). Vesicular transport in capillary endothelium-does it occur? Fed. Proc., in press.
PLASMALEMMAL
VESICLES IN HEART CAPILLARIES
367
BUNWAARD, M., AND FRGKJAER-JENSEN,J. (1982). Functional aspects of the ultrastructure of terminal blood vessels: A quantitative study on consecutive segments of the frog mesenteric microvasculature. Microvasc. Res. 23, l-30. BUNDGAARD, M., FRQKJAER-JENSEN,J., AND CRONE. C. (1979). Endothelial plasmalemmal vesicles as elements in a system of branching invaginations from the cell surface. Proc. Nar. Acad. Sci. USA 76, 6439-6442. CHI~N, S., LAUFLR, L., AND HANDLEY, D. A. (1982). Vesicle distribution in the arterial endothelium determined with ruthenium red as an extracellular marker. J. Ulfraslruct. Res. 79, 198-206. CLOUGH, G. (1982). The steady-state transport of cationized ferritin by endothelial cell vesicles. J. Physiol. 328, 389-401. CLOUGH, G., AND MICHEL, C. C. (1981). The role of vesicles in the transport of ferritin through frog endothelium. J. Physiol. 315, 127-142. CRONE, C., AND LEVITT, D. G. (1983). Exchange of small solutes through the capillary walls. In “Handbook of Physiology” (E. M. Kenkin. and C. C. Michel, eds.), Sect. 2, Vol. 3, Amer. Physiol. Sot., Washington, D.C., in press. FARQUHAR, M. G. (1978). Traffic of products and membranes through the Golgi complex. In “Transport of Macromolecules in Cellular Systems” (S. C. Silverstein, ed.), pp. 341-362. Abakon Vertagsgesellschaft, Berlin. FR~KJAER-JENSEN, J. (1980). Three-dimensional organization of plasmalcmmal vesicles in endothelial cells. An analysis by serial sectioning of frog mesenteric capillaries. J. U/trustruct. RPS. 73, 920. GOLDSTEIN, J. L., ANDERSON, R. G. W., AND BKOWN. M. S. (1979). Coated pits, coated vesicles and receptor-mediated endocytosis. Nature (London) 279, 679-685. KARNOVSKY, M. J. (1971). Use of ferrocyanide-reduced osmium tetroxide in electron microscopy. Proceedings qf rhe 11th Annual Meeting of rhe Atrwricun Socirry for Cell Biology. 146a. KARNOVSKY. M. J., AND SHEA, A. M. (1970). Transcapillary transport by pinocytosis. Microvusc,. Res. 2, 353-360. KING, B. F. (1982). The role of coated vesicles in selective transfer across yolk sac epithelium. J. Ultrustruct. Res. 79, 273-284. MAZZONE, R. W., AND KORNBLAU, S. M. (1981). Pinocytotic vesicles in the endothelium of rapidly frozen rabbit lung. Microwsc. Res. 21, 193-21 I. PALADE, Cr. E. (1953). Fine structure of blood capillaries. J. Appl. Phys. 24, 1424 (Abstract). PALADE, G. E., AND BRLJNS,R. R. (1968). Structural modulations of plasmalemmal vesicles. J. Ce// Biol. 37, 633-653. PALADF, G. E., SIMIONFSCIJ, M., AND SIMIONESCU, N. (1979). Structural aspects of the permeability of the microvascular endothelium. Acfu Physiol. Sund. S14ppl. 463, 1 l-32. REYNOLDS, E. S. (1963). The use of lead citrate at high pH as an electron-opaque stain in electron microscopy. J. Cell Biol. 17, 208-213. RHODIN, J. A. G. (1968). Ultrastructure of mammalian venous capillaries, venules. and small collecting veins. J. Ultrastrrrcf. Res. 25, 452-500. RYAN, U. S. (1982). Structural bases for metabolic activity. Annrr. Re19. Physiol. 44, 223-239. SIMIONESCU, N.. AND SIMIONESCU, M. (1976). Galloylglucoses of low molecular weight as mordant in electron microscopy. I-II. J. Cell B&l. 70, 608-633. SIMIONESCU.N.. AND SIMIONESCU,M. (1981). Hydrophilic pathways of capillary endothelium, a dynamic system. In “Water- Transpor-t acr.oss Epithelia” (H. H. Ussing. N. B. Bindslev, and 0. StenKnudsen, eds.), pp. 228-247. Munksgaard, Copenhagen, SIMIONESCLI, N.. SIMIONESCLI. M.. AND PAI ADF. G. E. (1973). Permeability of muscle capillaries to exogenous myoglobin. J. Cell Bid. 57, 424-451. SIMIONESTU. N., SIMIONESCU. M., AND PAI.ADE. G. E. (1974). Morphometric data on the endothelium of blood capillaries. .I. Cell Biol. 60, 128-157. SIMIONESCU, N., SIMIONES~U. M., AND PAI.ADF, G. E. (1975). Permeability of muscle capillaries to small hemepeptides. Evidence for the existence of transendothelial channels, J. Ce// B;o/. 64, 586 607. SMITH, U., RYAN, J. W.. MICHIE, D. D., AND SMITH. D. S. (1971). Endothelial projections as revealed by scanning electron microscopy. Scirnc,c 173, 925-927. SOMI YO, A. V. (1980). tiltrastructure of vascular smooth muscle. /n “Handbook of Physiology” (D. F. Bohr, A. V. Somlyo, H. V. Sparks, and S. R. Geiger-. eds.), Sect. 2, Vol. 2. pp. 33-68, Amer. Physiol. Sot. Washington, D.C.
368
BUNDGAARD,
HAGMAN,
AND
CRONE
WAGNER, R. C., AND CASLEY-SMITH, J. R. (1981). Endothelial vesicles. Microvasc. Res. 21, 267-298. VAN DEURS, B., BULOW, F., AND M@LLER. M. (1981). Vesicular transport of cationized ferritin by
the epithelium of the rat choroid plexus. J. Ceil Bid. 89, 131-139. E. R. (1974). On pericytes, particularly their existence on lung capillaries. Microvnsr. RPS.
WEIBEL,
8, 218-235.
WOLFF,J. R. (1977). Ultrastructure of the terminal vascular bed as related to function. In “Microcirculation” (G. Kaiey, and B. M. Altura, eds.), pp. 95-130. Vol. I. Univ. Park Press, Baltimore.