Evidence for a blood-thymus barrier using electron-opaque tracers.

In order to verify the existence of a blood-thymus barrier to circulating macromolecules, the permeability of the vessels of the thymus was analyzed in young adult mice using electron opaque tracers of different molecular dimensions (horseradish peroxidase, cytochrome c, catalase, ferritin, colloidal lanthanum). Results show that although blood-borne macromolecules do penetrate the thymus, their parenchyma] distribution is limited to the medulla of the lobe by several factors: (a) the differential permeability of the various segments of the vascular tree; (b) the spatial segregation of these segments within the lobe; (c) the strategic location of parenchymal macrophages along the vessels. The cortex is exclusively supplied by capillaries, which have impermeable endothelial junctions. Although a small amount of tracer is transported by plasmalemmal vesicles through the capillary endothelium, this tracer is promptly sequestrated by macrophages stretched out in a continuous row along the cortical capillaries and it does not reach the intercellular clefts between cortical lymphocytes and reticular cells. The medulla contains all the leaky vessels, namely postcapillary venules and arterioles. Across the walls of the venules, large quantities of all injected tracers escape through the clefts between migrating lymphocytes and endothelial cells; also the arterioles have a small number of endothelial junctions which are permeable to peroxidase, but do not allow passage of tracers of higher molecular weight. The tracers released by the leaky vessels penetrate the intercellular clefts of the medulla, but they never reach the cortical parenchyma, even at long time intervals after the injection. Therefore, a blood-thymus barrier to circulating macromolecules does exist, but is limited to the cortex. Medullary lymphocytes are freely exposed to blood-borne substances.

Glomerular permeability. Ultrastructural cytochemical studies using peroxidases as protein tracers.

1. Glomerular permeability was studied by ultrastructural cytochemistry, using as protein tracers two intravenously injected peroxidases of differing molecular weight. 2. Horseradish peroxidase (molecular weight 40,000) passed rapidly through the endothelial fenestrae, across the basement membrane, and through the epithelial slits into the urinary space. Human myeloperoxidase (molecular weight 160,000 to 180,000) also passed rapidly through the endothelial fenestrae and across the basement membrane, but was impeded at the level of the epithelial slits. Both proteins were taken up in large amounts by the mesangial cells. 3. The present findings indicate that the epithelial slits are the primary filtration barrier responsible for the differential permeability to proteins of varying molecular size. 4. The observations also support the concept that an important function of the mesangial cells is the incorporation and disposal of glomerular filtration residues.

Ligand-induced movement of lymphocyte membrane macromolecules. V. Capping, cell movement, and microtubular function in normal and lectin-treated lymphocytes.

Capping of surface Ig by anti-Ig antibodies involves a membrane perturbation requiring an energy-dependent step. Lymphocytes treated with anti-Ig are stimulated to move. Previously, we had shown that movement was not essential for capping, although it influenced the localization of the cap. We have investigated the role of cell movement and of microtubular proteins in this phenomenon. Treatment of B lymphocytes with colchicine does not affect capping of Ig nor does it affect the increase in translational movement produced by anti-Ig antibodies. Treatment of lymphocytes with cytochalasin B stops translational movement and may affect capping to some degree under appropriate circumstances. Lymphocytes treated with both drugs are impaired in capping. We surmise that there may be two cytoplasmic events regulating directly or indirectly capping: one associated with the process of translational movement, the other associated with the activity of microtubules. Lymphocytes treated with concanavalin A do not cap Ig. Colchicine reverses this inhibition. Certain experimental procedures antagonize the colchicine effect, the most striking of which is the use of cytochalasin B. Colchicine appears to increase movement of the Con A-treated lymphocyte, and this increased movement appears responsible for the accumulation of complexes to the posterior part of the cell. Con A inhibits patching of Ig by anti-Ig, and this is not reversed by colchicine.

Ligand-induced movement of lymphocyte membrane macromolecules. II. Mapping of surface moieties.

Anti-immunoglobulin (Ig) coupled to ferritin or hemocyanin was used to map the distribution of Ig molecules on lymphocytes derived from bone marrow (B lymphocytes) by freeze-etching. The labeled anti-Ig was distributed all over the membrane in the form of random interconnected patches forming a lacy, continuous network. This was the pattern of lymphocytes labeled at 4 degrees C with the anti-Ig. After warming at 37 degrees C, the labeled molecules concentrated into a single area of the cell (forming the cap) and were rapidly internalized in small vesicles Freeze-etching showed close packing of the labeled molecules in the cap area. There was evidence that in the cap area the Ig molecules were exfoliated from the plane of the membrane, suggesting that the Ig may be superficial to the bilipid layer, or weakly anchored to the membrane. Similar studies were made using antibodies to histocompatibility antigens. Thymocytes were labeled with anti-H-2 and ferritin anti-Ig at 4 degrees C. Freeze-etching showed large patches scattered over the membrane and separated from each other by several thousand angstroms. This distribution may, in part, explain why H-2 antigens do not readily form a cap; the large patches are beyond the reach of even a double ligand (sandwich) reaction. The antigens that reacted with heterologous anti-lymphocyte globulin (ALG) were found in small noninterconnected clusters a few hundred angstroms apart. Such clusters presumably cannot be linked by a single antibody but can by a sandwich (ligand to ligand-antigen) reaction. In previous studies it was found that ALG antigens form a cap only after a sandwich reaction. Finally, the receptors for concanavalin A (Con A) were found in a lacy, irregular interconnected, random network. The spatial distribution of these moieties on the membrane may, in great part, determine their movement after reaction with one or two ligands.

Ligand-induced movement of lymphocyte surface macromolecules. IV. Stimulation of cell motility by anti-Ig and lack of relationship to capping.

Microscopic examination of spleen lymphocytes discloses a small number moving at random at a given time. The majority of lymphocytes with this spontaneous movement are thymic derived. Addition of anti-Ig antibodies stimulates random movement of B lymphocytes. This movement depends upon a bivalent antibody and a metabolically active cell. The movement is inhibited by DFP, suggesting the involvement of a serine esterase. Also the anti-Ig stimulated movement of the lymphocyte is inhibited by cytochalasin B or by not allowing the cells to settle onto a surface. Lymphocytes treated with DFP or cytochalasin B, or untreated lymphocytes in suspension, capped the anti-Ig-Ig complexes. Hence, one can dissociate the surface capping of anti-Ig-Ig complexes from cell movement. We postulate that capping may result from superficial movements of the surface and/or from membrane flow, both of which are not related to actual translation of the cell on a surface. Four effects have now been observed following combination of a ligand with the antigen receptor on the B lymphocytes: redistribution on the surface of the complexes; pinocytosis and catabolism; shedding into the extracellular environment; and stimulation of translational movement.

Glomerular permeability. Ultrastructural studies in experimental nephrosis using horseradish peroxidase as a tracer.

Wistar/Furth rats were made nephrotic by daily administration of amino-nucleoside of puromycin, and the ultrastructural localization of horseradish peroxidase (mol wt 40,000) in the renal glomerulus was studied from 1 min to 20 hr after intravenous injection of the tracer. In control rats, peroxidase permeated the endothelial fenestrae, the basement membrane, and the epithelial slits, and was present in tubular lumina. Nephrotic glomeruli showed relatively normal basement membranes, extensive fusion of foot processes with formation of "close" intercellular junctions, and large vacuoles and pockets in epithelial cells. On serial sections some of the epithelial vacuoles communicated on one side with the extracellular space overlying basement membrane, and on the other side with the urinary space. In nephrotic animals, peroxidase permeated the basement membrane and the close junctions, and was present in many of the vacuoles and pockets as early as 1 min after injection. Only small numbers of peroxidase-positive vacuoles remained in. epithelial cells 1 hr or more after injection of the tracer. It is suggested that the epithelial pockets and vacuoles form pathways across which leaking proteins can be transferred across the epithelium into the urinary space. Epithelial vacuoles may also be absorption droplets designed to "conserve" leaking proteins, but this function was not prominent in our experiments with peroxidase.

An ultrastructural study of glomerular permeability in aminonucleoside nephrosis using catalase as a tracer protein.

Beef liver catalase (mol wt 240,000) was injected intravenously into normal rats and rats made nephrotic with aminonucleoside of puromycin. The localization of the tracer in the kidneys was then studied by ultrastructural cytochemistry, 3 min-12 hr after injection. Passage of catalase into the urinary space in normal rats was restricted by the basement membrane and by the epithelial slit pore. Nephrotic glomeruli showed extensive fusion of foot processes and formation of pockets and vacuoles in the fused epithelium; within 3 min after injection, catalase appeared in basal pockets, epithelial vacuoles, and the urinary space. Residual slit pores and close junctions in fused epithelium were impermeable to catalase. These studies indicate that alteration of the epithelial cells and basement membrane is responsible for protein leakage in aminonucleoside nephrosis.

Ligand-induced movement of lymphocyte membrane macromolecules. 3. Relationship between the formation and fate of anti-Ig-surface Ig complexes and cell metabolism.

Spleen lymphocytes were studied for the movement and interiorization of complexes of anti-Ig-surface Ig. The movement of the complex into a small, compact zone of the cell membrane (forming a cap) was inhibited by drugs that inhibited glycolysis and oxidative phosphorylation, but not by drugs that affected protein synthesis. Dead lymphocytes did not form caps. Freeze-etching techniques revealed that inhibited lymphocytes showed formation of multiple small complexes over the entire cell surface. Inhibitors of glycolysis and of oxidative phosphorylation also inhibited the interiorization and catabolism of radioiodinated anti-Ig. We hypothesize that cross-linking of all the surface Ig triggers the membrane movements that are required to pull the lattice into one zone of the cell.

An ultrastructural study of lymphocytes with surface-bound immunoglobulin.

This report is on a radioautographic study of lymphocytes exposed to (125)I-labeled anti-Ig in an attempt to identify surface-bound Ig molecules. The results as studied by ultrastructural radioautography confirmed the presence of surface-bound Ig on a certain population of lymphocytes. The specificity of the anti-Ig was determined by using appropriate controls that included the use of an absorbed anti-Ig and anti-hemocyanin antibody. The labeling pattern resulting from the interaction of labeled anti-Ig and Ig was found to be specifically associated with the cell surface and random in its distribution. Morphological differences were not apparent between labeled and nonlabeled lymphocytes in the spleen and lymph nodes. In the thymus, most lymphocytes did not exhibit detectable Ig. The few thymic lymphocytes that were labeled had unique morphological characteristics that included fewer ribosomes, many of which were monoribosomes. Relative to the amount in their cytoplasmic organelles, plasma cells had surface Ig but to a lesser degree than lymphocytes. Finally, macrophages were nonspecifically labeled and contained antibody on their membranes as well as intracellularly.

Ligand-induced movement of lymphocyte membrane macromolecules. I. Analysis by immunofluorescence and ultrastructural radioautography.

The fate of different complexes on the membrane of thymocytes and spleen lymphocytes was studied with the use of both immunofluorescence and ultrastructural radioautography. The complexes of anti-immunoglobulin (Ig) with the surface Ig of B lymphocytes were present all around the membrane at 4 degrees C; an increase in temperature produced a rapid aggregation of the complex into a cap which was readily interiorized in vesicles. Ultrastructural details of this process were given. The movement of the complexes depended upon the amount of anti-Ig and the temperature. The complexes of anti-lymphocyte antibody with surface antigen(s) did not result in formation of a single large aggregate (or cap) unless an anti-antibody was brought into the reaction. The caps formed by this trilayered complex were not interiorized. Concanavalin A (Con A) bound to cell surface carbohydrate moieties and the complexes of Con A readily formed a cap and were interiorized. Finally, antibodies to H-2 determinants did not form in most instances a single cap aggregate even when anti-antibodies were used. With time the H-2 complexes tended to form several large aggregates with some endocytosis.

An ultrastructural study of glomerular permeability using catalase and peroxidase as tracer proteins.

Mice were injected intravenously with beef liver catalase (mol wt 240,000) and very small doses of horseradish peroxidase (mol wt 40,000) and the site of localization of these enzymes in the kidney was studied by ultrastructural cytochemistry. 1 min after injection, catalase was present in glomerular capillary lumina and there was minimal permeation of the basement membrane. After 5-180 min, staining of the basement membrane increased progressively but was usually less than that in capillary lumina. At all time intervals the inner (sub-endothelial) layer of the basement membrane contained more reaction product than the lamina densa and the outer (subepithelial) layer. Catalase permeated the entire thickness of the basement membrane and extended up to the slit pore but not beyond the level of the slit diaphragm and was not seen in the urinary space or tubular lumina. Horseradish peroxidase permeated the whole thickness of the basement membrane within 2 min after injection; however, gradients of staining from the inner to outer layers of the basement membrane were frequently seen. The findings with both enzymes indicate that (a) the basement membrane restricts the passage of proteins over a wide range of molecular size with increasing impediment for larger molecules and (b) the slit pore functions as an additional barrier for molecules that cross the basement membrane.

Altered functional properties of the renal glomerulus in autologous immune complex nephritis: an ultrastructural tracer study.

The altered functional properties of the glomerular capillary wall in a model of autologous immune complex disease (Heymann's nephritis) was studied by electron microscopy using intravenously injected protein tracers of varying molecular weight. There was an increase in the permeability of the glomerular basement membrane (GBM) itself to large molecules; this change was focal and was found in those areas where the GBM contained immune complex deposits. Both ferritin and catalase, tracers normally restricted from passing the glomerular filter, were present in the urinary space within minutes of injection. No evidence was obtained for increased glomerular epithelial transport in this disease. Foot process swelling and "close" junction formation was moderate, even in animals with marked degrees of proteinuria. Indirect evidence, therefore, makes an alteration in the slit pore complex likely. In addition, there was immediate and selective concentration of all tracers within deposits, though ferritin was partially excluded from some deposits. This phenomenon may be of significance in the perpetuation of the disease.

Cutaneous basophil hypersensitivity. II. A light and electron microscopic description.

Delayed onset erythematous skin reactions elicited in guinea pigs early in the course of sensitization with azobenzenearsonate-protein conjugates or with protein antigens in incomplete Freund's adjuvant or in saline were found to have a characteristic morphology which sets them apart from delayed hypersensitivity and the classic antibody mediated reactions. The principle feature was massive dermal infiltration with basophilic leukocytes. Mononuclear cells of several types including activated and small lymphocytes, monocytes, macrophages, and blast cells were also present. Such reactions have in the past been designated Jones-Mote hypersensitivity, but we prefer the descriptive term cutaneous basophil hypersensitivity (CBH) for the reasons given. Occasional basophils extruded their granules, and individual granules, retaining their characteristic ultrastructure, were commonly seen in the interstitium. However, intercellular junctions between endothelial cells were closed except during cell emigration and there was no morphologic evidence of an histamine-like effect. The majority of basophils, moreover, did not degranulate but underwent nuclear pyknosis and cytoplasmic degeneration and were phagocytosed by macrophages. Phagocytosed basophil granules retained their ultrastructure. Skin tests performed at late intervals after sensitization had a different time course and morphology. Animals sensitized with protein antigens in complete Freund's adjuvant developed delayed hypersensitivity; however, reactions elicited in such animals at early (but not late) intervals after sensitization contained a prominent basophil component. We interpret such reactions to be a mixture of delayed hypersensitivity and cutaneous basophil hypersensitivity. The function of the basophil in CBH and its relation to the mononuclear cells which accompany it are unknown, and various possibilities are discussed. We conclude that cutaneous basophil hypersensitivity is a distinct immunologic and morphologic entity, occurring early in the course of sensitization with protein antigens incorporated in any of several vehicles. The mechanism of the reaction is presently unknown, and a general hypothesis to explain its pathogenesis has been proposed.

The ultrastructural basis of capillary permeability studied with peroxidase as a tracer.

The transendothelial passage of horseradish peroxidase, injected intravenously into mice, was studied at the ultrastructural level in capillaries of cardiac and skeletal muscle. Peroxidase appeared to permeate endothelial intercellular clefts and cell junctions. Abnormal peroxidase-induced vascular leakage was excluded. Neutral lanthanum tracer gave similar results. The endothelial cell junctions were considered to be maculae occludentes, with gaps of about 40 A in width between the maculae, rather than zonulae occludentes. Some observations in favor of concurrent vesicular transport of peroxidase were also made. It is concluded that the endothelial cell junctions are most likely to be the morphological equivalent of the small pore system proposed by physiologists for the passage of small, lipid-insoluble molecules across the endothelium.

Porous substructure of the glomerular slit diaphragm in the rat and mouse.

The highly ordered, isoporous substructure of the glomerular slit diaphragm was revealed in rat and mouse kidneys fixed by perfusion with tannic acid and glutaraldehyde. The slit diaphragm was similar in both animal species and appeared as a continuous junctional band, 300-450 A wide, consistently present within all slits formed by the epithelial foot processes. The diaphragm exhibited a zipper-like substructure with alternating, periodic cross bridges extending from the podocyte plasma membranes to a central filament which ran parallel to and equidistant from the cell membranes. The dimensions and spacing of the cross bridges defined a uniform population of rectangular pores approximately 40 by 140 A in cross section and 70 A in length. The total area of the pores was calculated to be about 2-3% of the total surface area of the glomerular capillaries. Physiological data indicate that the glomerular filter functions as if it were an isoporous membrane which excludes proteins larger than serum albumin. The similarity between the dimensions of the pores in the slit diaphragm and estimates for the size and shape of serum albumin supports the conclusion from tracer experiments that the slit diaphragm may serve as the principal filtration barrier to plasma proteins in the kidney.

Specializations in filopodial membranes at points of attachment to the substrate.

A mouse cell line (LM), which grows predominantly as spindle-shaped cells with numerous filopodia, was employed in this study. These filopodial projections appear to be important as sites of attachment to the substratum in LM cells. Morphologically the filopodia are slender projections from the cell body which usually attach to the substrate at their distal ends (filopodial footpads). Freeze-fracture of monolayer cultures in situ preserves the spatial relationship of filopodial processes to that of the cell body. Examination of these freeze-fracture preparations reveals a striking difference in the density of intramembrane particles (IMP) in the filopodial-footpad plasmalemma compared with the plasmalemma of the cell body (number of IMP in footpad > cell body). Additionally, there is a marked difference in the number of filipin-sterol complexes on the cell body, compared with the filopodial footpad, implying a difference in the cholesterol content in these regions (filipin-sterol complexes in footpad < cell body). These data suggest a structural and functional specialization of the filopodial-footpad plasma membrane which may be related to cell adhesion.

Fine structural localization of a blood-brain barrier to exogenous peroxidase.

Horseradish peroxidase was administered to mice by intravenous injection, and its distribution in cerebral cortex studied with a recently available technique for localizing peroxidase with the electron microscope. Brains were fixed by either immersion or vascular perfusion 10-60 min after administration of various doses of peroxidase. Exogenous peroxidase was localized in the lumina of blood vessels and in some micropinocytotic vesicles within endothelial cells; none was found beyond the vascular endothelium. Micropinocytotic vesicles were few in number and did not appear to transport peroxidase while tight junctions between endothelial cells were probably responsible for preventing its intercellular passage. Our findings therefore localize, at a fine structural level, a "barrier" to the passage of peroxidase at the endothelium of vessels in the cerebral cortex. The significance of these findings is discussed, particularly with reference to a recent study in which similar techniques were applied to capillaries in heart and skeletal muscle.

Cytochemical localization of two glycolytic dehydrogenases in white skeletal muscle.

The cytochemical localization, by conventional methods, of lactate and glyceraldehyde-3-phosphate dehydrogenases is limited, firstly, by the solubility of these enzymes in aqueous media and, secondly, by the dependence of the final electron flow from reduced nicotinamide-adenine dinucleotide (NADH) to the tetrazolium on tissue diaphorase activity: localization is therefore that of the diaphorase, which in rabbit adductor magnus is mitochondrial. NADH has been found to have great affinity to bind in the sarcoplasmic reticulum, and, therefore, if it is generated freely in the incubation media containing 2,2',5,5'-tetra-p-nitrophenyl-3,3'-(3,3'-dimethoxy-4,4'-phenylene)-ditetrazolium chloride (TNBT) and N-methyl phenazonium methyl sulfate (PMS), it can bind there and cause a false staining. Since such a production of NADH can readily occur in the incubation media for glycolytic dehydrogenases due to diffusion of these soluble enzymes from tissue sections, the prevention of enzyme solubilization is extremely important. Fixation in formaldehyde prevented such enzyme diffusion, while at the same time sufficient activity persisted to allow for adequate staining. The incubation media contained PMS, so that the staining system was largely independent of tissue diaphorase activity. Application of these methods to adductor magnus of rabbit revealed by light microscopy, for both enzymes, a fine network which was shown by electron microscopy to represent staining of the sarcoplasmic reticulum. Mitochondria also reacted. These findings add further support for the notion that the sarcoplasmic reticulum is probably involved in glycolytic activity.

The ultrastructural basis of alveolar-capillary membrane permeability to peroxidase used as a tracer.

The permeability of the alveolar-capillary membrane to a small molecular weight protein, horseradish peroxidase (HRP), was investigated by means of ultrastructural cytochemistry. Mice were injected intravenously with HRP and sacrificed at varying intervals. Experiments with intranasally instilled HRP were also carried out. The tissue was fixed in formaldehyde-glutaraldehyde fixative. Frozen sections were cut, incubated in Graham and Karnovsky's medium for demonstrating HRP activity, postfixed in OsO4, and processed for electron microscopy. 90 sec after injection, HRP had passed through endothelial junctions into underlying basement membranes, but was stopped from entering the alveolar space by zonulae occludentes between epithelial cells. HRP was demonstrated in pinocytotic vesicles of both endothelial and epithelial cells, but the role of these vesicles in net protein transport appeared to be minimal. Intranasally instilled HRP was similarly prevented from permeating the underlying basement membrane by epithelial zonulae occludentes. Pulmonary endothelial intercellular clefts stained with uranyl acetate appeared to contain maculae occludentes rather than zonulae occludentes. HRP did not alter the ultrastructure of these junctions.

Ultrastructural studies on the permeability of the mesothelium to horseradish peroxidase.

Peritoneal mesothelium was exposed for 2-60 min to solutions of horseradish peroxidase by incubation in vitro, or after intraperitoneal injection in vivo. Peroxidase was localized, with the electron microscope in the intercellular clefts of the mesothelium, often along their entire lengths, in vesicles adjoining or contiguous with the clefts, and along the peritoneal and basal surfaces of the cell, and also in intracytoplasmic vacuoles. The intercellular junctions of peroxidase-treated mesothelium did not differ from those of controls: open and closed junctions were present in both groups. Intercellular localization was also obtained when the mesothelium was exposed to peroxidase during or after fixation. Although intracellular absorption of peroxidase and its incorporation into larger vacuoles were observed, there was no clearcut evidence of vesicular transport across the mesothelium in these experiments. These findings are consistent with physiologic data which postulate that mesothelial transport can be accounted for, at least in part, by passive diffusion through a system of pores, and they suggest that these pores are located in the intercellular clefts.