Trans-glial Channels in Ventral Nerve Roots of Crayfish RICHARD R. SHIVERS 1 AND MILTON W. BRIGHTMAN Laboratory of Neuropathology a n d Neuroanatomical Sciences, National Institute of Neurological Diseases and Stroke, National Institutes of Health, Bethesdn, Maryland 20014
ABSTRACT The sheath around the roots of the sixth abdominal ganglion in the yentral nerve cord of the crayfish consists of concentric layers of thin glial processes alternating with wide clefts containing filamentous connective tissue. Regions of each glial lamella are perforated by single, short, tubular channels: the trans-dial channels. In thin plastic sections examined in the electron microscope, the channels appear as slits that are 240 A wide and 450-550 A long which traverse glial lamellae less than 1,500 A thick. Branched tubular channels cross glial sheets that are thicker than 1,500 A. The thickest glial wrap is adaxonal; it closely encapsulates individual axons and its cell membrane is separated from the axolemma by a collagen-free space of only 150 A. The adaxonal glial cytoplasm contains unique, three-dimensional networks of interconnected tubules. Separate tubular lattices occur along these thicker processes. In replicas of freeze-fractured sheaths, the outer half of the plasma membrane belonging to the thin glial sheets exhibits many volcano-like protrusions which represent cross fractures through the necks of trans-glial channels. Corresponding depressions o n the inner half of these membranes are sites where the plasma membrane invaginates to form the channels. Although some channels are randomly dispersed, others are lineraly positioned i n restricted areas across successive glial layers. The number of channels is far more readily appreciated i n replicas than in thin sections. The average frequency of channels is 16 per ~2 (range 8 to 33) in normal roots and does not differ significantly from the average of 13 per pz i n proximal stumps of roots fixed three to four weeks after the roots were cut. The channels are not precisely aligned from one glial layer to the next but do appear to coincide approximately with the adaxonal tubular lattice. The combination of trans-glial channels and adaxonal tubular lattices may provide a complex conduit that could facilitate a rapid, passive flow of electrolytes and nutrients across the nerve sheath to the axonal surface. Horseradish peroxidase solutions bathing the ventral roots enter the trans-glial channels, extracellular clefts and finally the tubular lattices. This distribution supports the proposed role of the channels in a rapid extracellular passage of solutes. The channel profiles have a range of forms consistent with the supposition that they are not static but continually reforming. There are indications that, proximal to the cut, the areas of glial plasma membrane with channel profiles contain more junctional complexes between regenerating cells than between glial cells of normal sheaths. The channel profiles and aggregates of particles belonging to junctions are closely associated when they occupy the same region of the membrane.
Glial or supportive cells of Crustacean nervous systems have been implicated in a variety of roles in maintenance of the neuronal components. Glial cells comprise the major element of the highly ordered sheath which surrounds roots of ventral ganglia in crayfish (Somers and Nunnemacher, ’70). Because of their predominance, the glial sheets have been suggested to function in part as selective barriers to J.
COMP.
NEUR.,167: 1-26.
nutrients and electrolytes which must traverse the sheath to axons contained within (Brandt et al., ’65; Geren and Schmitt, ’54; Holtzman et al., ’70; Malzone et al., ’66). Experiments with electron opaque “tracers” have suggested that the tracers reach the surface of axons by a combination of I Present address; Department of Zoology, University of Western Ontario, London, Ontario, C a n a d a .
1
2
RICHARD R. SHIVERS A N D MILTON W. BRIGHTMAN
meters. This gap was of sufficient width that resection of a piece of the nerve root was unnecessary (Hoy, '69). Animals were then returned to crushed ice for 30 minutes before being placed in shallow pans of water. Only the proximal stumps of interrupted ganglionic roc,ts and their growing axonal and glial processes were prepared for freeze-fracture and routine thin sectioning three to eight weeks following nerve section. Unoperated crayfish were anesthetized in crushed ice for 30 minutes prior to dissection. The ventral musculature was opened and the exposed sixth ganglion and its roots were flooded for several minutes with ice-cold fixative. When the tissue had begun to harden and turn yellow, the ganglion and roots, which were at least 4 mm long, were cut free and placed in fresh fixative. Fixation for routine electron microscopical studies was carried out in ice-cold 6 % glutaraldehyde (from Fisher 50 % ) (Sabatini et al., '63) buffered to pH 7.5 with 0.2 M sodium phosphate. All solutions had sucrose added at a concentration of 3 % . Aldehyde fixation was for a minumum of two hours; longer periods did not appear to improve the quality of fixation. Postfixation for one to three hours in 2 % O s 0 4 buffered to pH 7.5 with sodium phosphate was followed by a one and one-half hour wash in buffer. Following osmication, the tissues were stained en bloc with aqueous uranyl acetate (Heuser and Reese, '73), dehydrated through graded methanol to METHODS propylene oxide, and embedded in English Adult crayfish (Orconectes uirilis) were Araldite (Ciba Ltd, Duxford, Cambridge, obtained from Nasco Supply (Ft. Atkinson, England). Thin sections were stained with Wisc.) and kept in shallow pans of aerated 10% uranyl acetate (Heuser and Reese, tap water. They were fed weekly on horse- '73) and lead citrate (Reynolds, '63). (All meat and maintained at room temperature. tissues were examined on an AEI-6B elecFor parts of the present investigation, tron microscope.) roots of the sixth abdominal ganglion were Tissue preparation for freeze-fracture used. The crayfish were anesthetized for was begun by fixing the tissues in ice-cold 30 minutes in crushed ice and then a shal- 6 % glutaraldehyde (described above) for low incision was made at the posterior one hour. The nerve roots were then rinsed margin along the ventral surface of the briefly in 0.2 M sodium cacodylate buffer fifth abdominal segment. The sixth abdom- (pH 7.35) and placed in a solution of 20% inal ganglion of the ventral nerve cord was glycerol in 0.2 M sodium cacodylate (pH carefully exposed and its roots were cut. 7.35) overnight in the refrigerator. PrepNerve root stumps attached to the gan- aration of the material for rapid freezing glion (proximal stumps) were a minimum consisted of placing short (less than 2 mm) of 4 mm long; the distal stumps retracted pieces of nerve roots in a drop of 20% glyfrom the cut leaving a gap of several milli- cerol on gold specimen discs (Balzer's, routes: (1) seepage through the connective tissue spaces between thin glial cytoplasmic processes of the nerve sheath (extracellular clefts); and (2) passage across the adaxonal glial cytoplasmic layer through a fine anastamosing system of tubules (the tubular lattice), thereby appearing in the intercellular space adjacent to the axon plasma membrane (Abbott, '70; Holtzman et al., '70; Kristensson et al., '72; Lane and Abbott, '75; Shivers, '70). From these and other studies it has been proposed that a system of passageways exists across the glial sheath of crayfish nerves which functions as a transport route for essential elements for nerve maintenance (Dehrenzo, '68; Holtzman et al., '70; Peterson and Pepe, '61). The present study on crayfish ventral ganglion roots using freeze-fracture techniques has revealed a hitherto unreported system of trans-glial channels. The channels are inconspicuous in thin sections because of the highly undulating nature of the glial cell membrane and its indentations, in places, as shallow pits and elongated tubules. The channels are far more obvious in freeze-fracture planes that present broad vistas of internal faces within the plasma membrane. In such planes, the distribution and frequency of the channels are readily appreciated. The ubiquity and large numbers of channels suggest that they comprise the primary route between the exterior of the nerve sheath and the surface of axons.
TRANS-GLIAL CHANNELS
Lichtenstein). The discs were plunged into Freon 22 cooled with liquid nitrogen (LN,) for ten seconds and quickly transferred to LN, for storage until fracturing. Tissues were fractured at - 111 "C in a Balzer Apparatus, shadowed with a thin layer of platinum and lightly coated with carbon. Specimens were not etched. Replicas were cleaned in Chlorox for two hours, rinsed in four changes of distilled water and picked up on bare or parlodoin-coated 200 mesh grids. In order to follow the extracellular movement of protein, the sixth abdominal ganglion and roots were removed from several animals and placed in a medium of crayfish ringer (Van Harreveld, '36) containing 1 mg/ml of horseradish peroxidase (Type VI, Sigma Chemical Co.,St. Louis, Mo.). Nerves were immersed for one hour and then fixed for eight hours at room temperature in 3% glutaraldehyde (from Taab 2 5 % ) buffered to pH 7.35 with 0.2 M sodium cacodylate. Following aldehyde fixation tissues were washed overnight in refrigerated buffer and then incubated for two hours on crushed ice (Graham and Karnovsky, '66) to demonstrate peroxidatic activity. The tissue was then processed for electron microscopy as above. Enumeration of channel profiles in electron micrographs of replicas was made using a Hewlett-Packard Model 9810A computer coupled with a Hewlett-Packard Model 9800 digitizer. Programs were written to facilitate simultaneous counting of profiles and area density calculations. These operations were carried out on both normal replicas and replicas of growing nerve and glial processes that were proximal to the transection. OBSERVATIONS
A. Glial arrangement Roots of the sixth abdominal ganglion in the crayfish consist of varying numbers of sensory and motor axons which are encapsulated by a multilamellate sheath. This sheath is composed of closely packed, narrow layers of glial cytoplasm which are usually separated from each other by wider extracellular clefts filled with differing amounts of connective tissue filaments about 100 A thick (fig. 1). The clefts are usually three to four times wider than
3
the glial processes. In a few places, multiple glial layers are separated by only a narrow cleft 150-200 A wide (fig. 1). Axons may be wrapped by 17 alternating layers of glial sheets and connective tissue spaces (fig. 1). The layers of glial cytoplasm are high1 attenuated and may be as thin as 180 with no organelles or much wider, as in the adaxonal glial processes (Ad), which usually contain a complex array of organelles. There is always a glial process immediately adjacent to the axon (fig. 1) with a narrow space of about 150 8, between the two. The arrangement of glial lamellae and extracellular filament-filled clefts can be clarified by using freeze-fracture techniques. The fracture plane passes from one layer of the sheath to the next, exposing the inner surface of the outer leaflet (B-face) or the outer surface of the cytoplasmic leaflet (A-face) of the glial plasma membrane (Branton, '66) (fig. 2). The outer half of the glial plasmalemma is characterized by a sparse population of particles (70-110 A in diameter). The inner half of the glial plasma membrane has a very dense array of randomly distributed particles of the same size.
k:
B. Trans-glial channels The thin glial sheets viewed in thin plastic sections are randomly traversed by short more or less stroaightchannels about 240 8, wide and 500 A long (figs. 4, 5, 11). Each channel appears as a single profile that extends from one side of a glial process to the other side, to form a passageway across the sheet. The channels are about as long as the glial lamella is thick, although a few run diagonally for a short distance or may branch (fig. 5). The channels are inconspicuous in thin sections because the glial cell membranes often undulate and, in places, are indented by shallow pits or deeper tubules that do not traverse the entire thickness of the process (fig. 17). It is likely that some of the tubular indentations are portions of channels. The frequency and distribution of the channels is more readily appreciated in replicas of frozen, cleaved specimens. The openings of the channels appear as small depressions on the inner half (A-face) of the glial plasmalemma and as volcano-like protrusions on the outer half (B-face) of
4
RICHARD R. SHIVERS AND MILTON W. BRIGHTMAN
the glial cell membranes (figs. 3, 10, 1215). The profiles of deprpsions and protrusions are about 240 A wide and thus correspond in size to the channels seen in thin sections. The profiles appear individually or in clusters. It is suggested that the depressions on the inner (A) half of glial membranes represent the invaginations of the channels from the membrane surface. The protrusions on the outer (B) half represent cross-fractures through the necks of the channels as they extend toward the opposite side o f the cell. The channels may be fractured in such a way as to show that they pass entirely through the thin glial process (figs. 3, 10). The depressions on the inner half of the cell membrane directly abut against the protrusions on the outer half of the same membrane (fig. 3). The channels appear tubular in thin sections and, in replicas, the depressions and elevations corresponding to the channels, therefore appear hollow (fig. 10). The distribution of trans-glial channels is variable and can better be determined in replicated membrane faces. The channels occur in successive, concentric glial processes including the inner one next to the lattice-containing adaxonal glial sheet. The channel openings are usually scattered at random (fig. 2), but some are arranged in ordered clusters (figs. 13-15). Thus, patches of membrane containing channel openings alternate with areas devoid of these depressions and protuberances. It is emphasized that the channel-free areas may be quite extensive. In a given sheet, the perforated zones may be coincidental with those of an adjacent glial process (figs. 2 , 13-15). Nevertheless, individual openings in one cell’s membrane are usually not in strict register with the openings of an adjacent glial sheet (figs. 2 , 11, 12). In regenerating sheaths, the channels are especially well ordered, the depressions and protuberances forming rows that are linear (figs. 10, 13, 15) or nearly circular (fig. 14). Adjacent circular clusters may have sharp boundaries separated by strips of membrane devoid of channel openings (fig. 14). Within a glial membrane, the areas between individual openings may be regular (fig. 14) but are more often highly variable in width and contour (figs. 2, 11, 12). Neighboring channels may be as close together
as 450 A or as far apart as 3,000 A. The zone immediately surrounding a junction may be perforated by channels (fig. 15) but is usually not (fig. 10). Although the channels are generally uniform in morphology and size (figs. 12, 15), there is some variation. Some protrusions are much higher than others and have a rim lined with discrete particles; others appear flat with smoother rims (figs. 12, 15). In regenerating sheaths, the individual protrusions in the outer half of the membrane appear to be more variable in height (fig. 15). The frequency of channels in the glial processes of intact nerve roots is 16 per p z . This density does not differ significantly from the 13 per p 2 found in roots that had been cut three to four weeks prior to fixation. Only those regions of the cell membrane containing channel profiles were selected for area measurements. These values, therefore, do not include the large surface of membrane that does not bear channel configurations. In order to determine whether the channels do indeed serve as passageways across successive glial sheets, nerve roots were immersed in a medium containing peroxidase. These roots were either intact or were proximal to a cut that had been made three weeks before the immersion. After 60 minutes, the protein had entered the channels and the extracellular clefts with which they communicated (figs. 4, 11). This pattern is the same in both intact and cut sheaths. The peroxidase-occupied channels were slightly narrower t h q those in replicas but were about 470 A long. The length therefore, is comparable in both thin sections and in replicas.
c. Tubular lattice The glial process closest to the axons is wider than the other glial sheets and contains a rich array of organelles, including a well-developed system of anastamosing tubules (figs. 1, 2). This tubular lattice is of variable complexity that ranges from slightly branched tubules (figs. 2, 6, 7) in narrower portions of the glid process to highly branched systems in wider- parts (fig. 8). The tubules are about 250 A wide (fig. 8). The three dimensional view and, consequently, the considerable surface area of the lattice is apparent in freeze-fracture
5
TRANS-GLIAL CHANNELS
replicas (fig. 9). These images confirm the regenerating sheaths (figs. 13-15) than in observation that the lattice is made up of intact ones (fig. 16). However, some junca series of invaginations of the glial plasma tional aggregates occur separately from membrane (Holtzman et al., '70). Each channels (figs. 10, 13, 14). As in normal lattice appears to be separate from neigh- sheaths, the junction aggregates of regenboring ones and it has not been ascertained erating glial processes are discoid, sharply whether there is any regular spacipg be- circumscribed and variable in overall ditween adjacent lattices. The 150 A wide ameter (figs. 10, 13-15). extracellular cleft between the adaxonal DISCUSSION glial layer and the axons is much narrower than the clefts between successive The system of trans-glial channels across layers of peripheral glia. successive glial sheets around the ventral In freeze-fracture replicas, the lattice nerve roots in crayfish may provide a short appears as a complex array of volcano-like route for large molecules moving from protrusionso (fig. 9). The protrusions are blood and extracellular spaces to the surabout 240 A wide and correspond therefore face of neurons and their processes. At in diameter to the lattice tubules in thin first inspection of thin sections, the glialplastic sections (figs. 6 4 ) . The protrusions connective-tissue sheath around the nerve represent cross-fractures through the necks roots appears to present a formidable obof the lattice tubules where their mem- stacle to such passage. If there were no branes become continuous with the plasma transverse channels but only the clefts membrane. Inasmuch as only a few frac- between the end-to-end abutments of adtured lattices have been encountered, it is jacent glial processes, a molecule would suggested that the fracture plane usually have to migrate for relatively long distances skips the adaxonal glial layer to pass step- before it could move across one glial sheet wise through the thinner, more peripheral into the collagen-containing extracellular members of the sheath. space between that sheet and a neighboring one. Instead, the cross-channels proD. Junctions vide many short, straight routes across the The most common junction encountered sheath. in the glial capsule is the hemi-desmosome The frequency of transverse channels, in (fig. 1). In thin plastic sections, this type the segments of glial plasma membrane of junction is recognized, as in other phyla, where they occur, is variable but about the by a patch of granulo-filamentous, dense same in normal and regenerating sheaths. material that inserts into a segment of Their frequency is not evident in thin cytoplasmic leaflet belonging to the cell plastic sections where only one or a few membrane (fig. 17). The most common channel profiles can be encountered in arrangement of particles in frozen-cleaved any one field. If the channels were rancell membranes consists of a disc$id ag- doml dispersed in plastic sections about gregate of large particles 90-110 A in di- 600 thick, then there would be about 16 ameter that occur on the outer half (B- such sections in a slice 1 p thick. As there face) of glial plasma membranes and of a are about 16 channels per p2 in the replicorresponding patch of complementary pits cated surface of a glial sheet, one would on the inner half (A-face) (figs. 10, 13-15). then expect to see one channel profile, or Since the hemi-desmosomes in thin sections in segments where they are not randomly and the particle aggregates in replicas are dispersed, a few more profiles in a given the most frequently encountered structures thin section. associated with cell junctions, it is inferred Although the entire membrane of each that the two are equivalent. glial process is not perforated by channels, In replicas, the discoid aggregates may limited segments are. The single channels, be located alongside channel profiles (figs. being about 250 A wide, would be expected 13-16). The aggregates, presumably hemi- to allow the non-selective, passive flow of desmosomes, appear to be the only junc- large and small molecules across segments tions in close proximity to the trans-glial of an individual lial lamella. The short channels. The association between junc- length, about 450 to 550 A, of the single tions and channels is more frequent in channels and their straight course would
K
1
6
RICHARD R. SHIVERS A N D MILTON W. BRIGHTMAN
enable the flow to be a rapid one across a given process. The approximate alignment of the channels between one glial sheet and successive neighbors would in turn, shorten the traverse time across the entire multi-laminated sheath. It has been recognized by Holtzman et al. (’70) that the glial sheath does permit the migration of peroxidase from perineural fluid into the tubular lattice of the adaxonal glial cell. The passage was assumed to take place by way of the extracellular collagen-occupied spaces designated as “channels,” a term restricted in the present account to the single, short tubules perforating thin glial processes. It was implied that large substances crossed the glial sheath by seeping through the long extracellular diffusion path around the ends of adjacent glial cells (Holtzman et al., ’70; Kristensson et al., ’72; Lane and Abbott, ’75). The authors did not mention, nor have we found, any signs of vesicular transport across the glial sheets. If these glial processes behave like those of mammalian brain, it would be expected that a cell process less than about 300 d; thick is incapable of forming pinocytotic pits (Brightman, ’65). Although thicker portions of the glial processes do contain peroxidase labelled vesicles (Lane and Abbott, ’75), an observation that we have confirmed, it appears that such uptake is for incorporation into the cell rather than for transfer across it. Contrary to this inference is the possibility that protein can be transferred directly from the adaxonal glial processes to the underlying axon. Proteins, newly synthesized i n Schwann cells, appear to be transferred to the adjacent giant axon of squids (Lasek et al., ’74). The collagen-containing extracellular spaces or clefts and the transverse channels together make up a continuous passageway. The channels however, do not appear to be simply branches of the extracellular clefts. The clefts are 200-3,000 wide and contain filaments, probably collagenous, whereas the trans-glial tubules do not. Furthermore, the variations in the size of the channel openings in replicas suggest that they may be labile, perhaps closing and reforming under certain conditions as yet undefined.
All clefts and channels are ultimately penetrated by tracer protein after exposure of ganglionic root segments to peroxidasecontaining medium for periods of one hour or longer. This observation supports the suggestion that trans-glial channels serve as a short-cut route across the sheath. Unlike insect nerve sheaths which are impermeable to tracers (Treherne and Pichon, ’72), Crustacean nerve sheaths are leaky (Holtzman et al., ’70; Kristensson et al., ’72; Lane and Abbott, ’75; Shivers, ’70). We propose that not only is the crayfish nerve sheath permeable to molecules (including tracers) but that the trans-glial channels and tubular lattices represent a conduit which permits rapid solute flow across the sheath. Further studies are necessary to determine the rate at which molecules move across the sheath via the trans-glial channel system and whether such molecules are selectively transported by this route. Several variations in the basic morphology of trans-glial channel profiles in replicas are reported in the present paper. These are interpreted as representing different stages in the formation of the channels as they invaginate from the plasma membranes of glial processes. Evidence for turnover is difficult to obtain from thin sectioned material since only few channels are encountered in a given thin section. Investigation of the channels must therefore be done with freeze-fracture techniques which more accurately and consistently preserve the channels (Moor, ’72; Moor and Muhlethaler, ’63). Increased rate of formation or increased numbers of trans-glial channels may be a part of the processes occurring during regeneration. Experimentally induced nerve growth results in a pronounced glial reaction characterized by increased cytoplasmic complexity (Nordlander and Singer, ’72; Shivers and Brightman, ’74). Rapid growth and reformation of the trans-glial channels would be anticipated as part of the reestablishment of a complete nerve root sheath. The appearance in regenerating glial processes of a more frequent close association of trans-glial channel profiles with aggregations of particles assumed to be junctions suggests that a functional rela-
TRANS-GLIAL CHANNELS
7
tionship may exist between them. Devel- trations in this paper, and to Dr. Lise oping tissues often exhibit formation of Prescott for her invaluable advice regardnew intercellular junctional complexes ing freeze-fracture techniques. (Gilula, '73; Revel et al., '73),and although LITERATURE CITED recent studies of regeneration in crayfish motor nerves indicated no striking increase Abbott, N. J. 1970 Absence of a blood-brain barrier in a crustacean, Carcinus maenas L. Nain junctions between glial cells (Nordture (London), 225: 291-293. lander and Singer, '72), the present study Brandt, P. W., A. R. Freeman, J. P. Reuben and has shown rather extensive junctional H. Grundfest 1965 Variations between axon complexes in new glial processes (Shivers and Schwann membranes induced in lobster nerve fibers by currents and fluxes. Biol. Bull., and Brightman, '74) which will be de129: 400. scribed in detail in a separate report. The Branton, D. 1966 Fracture faces of frozen memmost common junction in the sheath is a branes. Proc. Nat. Acad. Sci. (U.S.A.), 55: 1048type of hemi-desmosome which appears as 1056. a dense plaque in the cytoplasm adjacent Brightman, M. W. 1965 The distribution within the brain of ferritin injected into cerebrospinal to the glial surface abutting on intercellufluid compartments. 11. Parenchymal distribular clefts. Similar structures have been tion. Am. J. Anat., 117: 193-220. reported by Overton and Mapp ('74) in DeLorenzo, A. J. D., M. Brzin and W. 0. Dettbarn 1968 Fine structure and organization of nerve regenerating notocord. Changes in the fibers and giant axons in Homarus americanus. number, distribution and appearance of J. Ultrastruct. Res., 24: 367. trans-glial channels and junctions could Flower, N. E. 1972 A new junctional structure reflect changes which do not interrelate i n the epithelia of insects of the order Distyoptera. J. Cell Sci., 10: 683. the two structures but instead, represent Geren, B. B., and F. 0. Schmitt 1954 The strucalterations in glial cells during growth. ture of the Schwann cell and its relation to the It is interesting to note that both the axon i n certain invertebrate nerve fibers. Proc. Nat. Acad. Sci. ( U . S . A . ) , 4 0 : 863. particle aggregates constituting junctional complexes and the volcano-like profiles of Gilula, N. B. 1972 Cell junctions of the crayfish hepatopancreas. J. Ultrastruct. Res., 38: 215. trans-glial channels occur on the B-face 1973 Development of cell junctions. of glial plasma membranes. This is differh e r . Zool., 1 3 : 1109-1117. ent from most other systems in which such Graham, R. C., and M. J. Karnovsky 1966 The early stages of absorption of injected horseradish structures are usually present on A-faces peroxidase in the proximal tubules of mouse of membranes (Gilula, '73). The A-face of kidney. Ultrastructural correlates by a new techmembranes in the present study contains nique. J. Histochem. Cytochem., 14: 291-302. instead, round depressions which are the Heuser, J. E., and T. S. Reese 1973 Evidence for recycling of synaptic vesicle membrane durcomplementary images of the protrusions ing transmitter release at the frog neuromuscuand clusters of pits corresponding to the larjunction. J . Cell Biol., 57: 3 1 5 3 4 4 . junctional particle aggregates. This dis- Holtzman, E., A. R. Freeman and L. A. Kashner 1970 A cytochemical and electron microscope tribution is similar to that seen in some study of channels i n the Schwann cells surroundother invertebrate systems (Flower, '72; ing lobster giant axons. J . Cell Biol., 44; 438Gilula, '72). 444. Trans-glial channels have not been re- Hoy, R. R. 1969 Degeneration a n d regeneration in abdominal flexor motor neurons in the crayported in nerve sheaths of other invertefish. J. Exp. Zool., 72: 219-232. brates or vertebrates. Peracchia ('73b, '74) described structures similar to the channels Kristensson, K., E. Stromberg, R. Elofsson and Y. Olsson 1972 Distribution of protein tracers reported here but regarded them only as in the nervous system of the crayfish (Astacus part of the adaxonal tubular lattice. Addiastacus L.) following systemic and local application. J. Neurocytol., I : 3 5 4 8 . tional studies are needed to determine the distribution of trans-glial channels in Lane, N. J., and N . J. Abbott 1975 The organization of the nervous system i n the crayfish other crustacean nerve sheaths and to Procambarus clarkii, with emphasis on the bloodassign a detailed functional role to them. brain interface. Cell Tiss. Res., 156; 173-187. ACKNOWLEDGMENTS
The authors would like to express their appreciation to Mrs. Clara Mowry for her excellent help in preparation of the illus-
Lasek, R. J., H. Gainer and R. J. Przybylski 1974 Transfer of newly synthesized proteins from Schwann cells to the squid giant axon. Proc. Nat. Acad. Sci. (U.S.A.), 71: 1188-1192. Malzone, W. F., G. H. Collins and R. R. Cowden 1966 Neurological relationships i n the thoracic
RICHARD R. SHIVERS AND MILTON W. BRIGHTMAN
8
ganglion of the fiddler crab, Uca. J . Comp. Neur., 127: 5 1 1 5 3 0 . Moor, H. 1972 The principles of freeze-etching. Revista Microsc. Elect., 1 : 269-275. Moor, H., and K. Muhlethaler 1963 Fine structure in frozen-etched yeast cells. J . Cell Biol., 1 7 : 609-628. Nordlander, R. H., and M. Singer 1972 Electron microscopy of severed motor fibers i n the crayfish. Z. Zellforsch., 126: 157-181. Overton, J., and F. E. Mapp 1974 The fine structure of regenerating notocord in anuran tadpoles. J. Exp. Zool., 187: 103-120. Peracchia, C. 1973b Low resistance junctions i n crayfish. 11. Structural details and further evidence for intercellular channels by freezefracture and negative staining. J . Cell Biol., 57: 66-76. 1974 Excitable membrane ultrastructure I . Freeze-fracture of crayfish axons. J . Cell Biol., 61 : 107-122. Peterson, R. P., and F. A. Pepe 1961 The fine structure of inhibitory synapses in the crayfish. J. Biophys. Biochem. Cytol., I 1 ; 157. Revel, J.-P., P. Yip and L. L. Chang 1973 Cell
junctions in the early chick embryo - a freezeetch study. Develop. Biol., 35: 302-317. Reynolds, E. S. 1963 The use of lead citrate at high pH as a n electron opaque stain in electron microscopy. J. Cell Biol., 1 7 : 200-211. Sabatini, D. D., K. Bensch and R. J. Barrnett 1963 Cytochemistry and electron microscopy. The preservation of cellular ultrastructure and enzyme activity by aldehyde fixation. J. Cell Biol., 1 7 : 19-58. Shivers, R. R. 1970 Fine structure of crayfish optic ganglia vascularization and permeability. J . Cell Biol., 47: 191A. Shivers, R. R., and M. W. Brightman 1974 Unpublished Observations. Somers, M . E., and R. E. Nunnemacher 1970 Microanatomy of the ganglionic roots of the abdominal cord of the crayfish, Orconectes virilis (Hagen). J. Comp. Neur., 138: 209-218. Treherne, J. E., and Y. Pichon 1972 The insect blood-brain barrier. In: Advances in Insect Physiology. Vol. 9. Academic Press, pp. 257313. Van Harreveld, A. 1936 A physiological solution for freshwater crustaceans. Proc. Soc. Exptl. Biol. Med. Sci., 34: 428.
Abbreviations A, A-face Ad, Adaxonal glia Ax, Axon B, B-face C, Trans-glial channel Ct, Extracellular cleft; connective tissue layer D, Round depression F, Filaments
G , Glial cytoplasm H, Hemi-desmosome J, Junctional complex L, Adaxonal tubular lattice N, Nucleus V, Volcano-like protrusion Circled Arrows indicate direction of platinum shadowing i n replicas.
PLATE 1 EXPLANATION O F FIGURE
1
Normal ventral ganglion roots contain axons (Ax) encased in a multilamellate glial (G)connective tissue (Ct) sheath. Only adaxonal glial layers (Ad) adjacent to axons contain numerous organelles and specialized tubular networks - the tubular lattices (L). Hemi-desmosomes (H) attach the glial lamellae to adjacent connective tissue layers. X 12,500.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 1
9
PLATE 2 EXPLANATION
2
10
OF FIGURE
The lamellar arrangement of glial cells i n the root sheath is more obvious in freeze-fracture. Exposed intramembrane surfaces contain either volcano-like protrusions (V) on B-faces o r round depressions (D) on A-faces. Extracellular clefts (Ct) filled with filamentous material (F) are interposed between very narrow lamellae of glial cytoplasm (G). The rectangular area shows a portion of glial cytoplasm containing a slightly branched tubular network. Occasionally, glial lamellae do not alternate with extracellular clefts (circled areas. See also circled area of fig. 1 ) . Normal root sheath. X 37,600.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W . Brightman
PLATE 2
11
PLATE 3 EXPLANATION OF FIGURES
12
3
Fractures exposing intact trans-glial channels clearly show the relationship between volcano-like projections on B-faces (B,) a n d round depressions o n A-faces (A,) (arrows). Note that the plasma membrane of adjacent glial lamella (A2) also exhibits channel profiles. Trans-glial channels appear in thin section a s small, round tubules with electronlucent cores (arrows, inset). Normal root sheath. X 75,000. Inset x 60,000.
4
Trans-glial channels (C) traverse glial lamellae and connect adjacent extracellular clefts (Ct). Clefts a n d channels contain HRP reaction product. Three weeks post-operation. X 152,000.
5
Two forms of trans-glial channels are seen. In glial lamellae less t h a n 1,500 A wide, channels (arrows) are short a n d single. In wider lamellae the channels appear branched (circled area). Normal root sheath. X 96,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W . Brightman
PLATE 3
PLATE 4 EXPLANATION OF FIGURES
14
6
Glial processes wider than 1,500 A exhibit a moderately-branched (arrows) trans-glial tubular network. Usually glial layers such as these are found in a n adaxonal position. Normal root sheath. X 99,000.
7
Adaxonal tubular lattice which is slightly branched consists mainly of elongate tubular profiles (between arrows) traversing the process. Normal root sheath. X 45,000.
8
Adaxonal glial cells possess large, highly anastamosed networks of tubules (arrows) -the tubular lattices. I n such cases, a regular pattern of construction of the lattice is suggested. Normal root sheath. X 114,000.
9
Tubular lattices c a n be seen in freeze-fracture replicas and exhibit a n extensive honeycomb appearance because of their complex branched and interconnected tubules (arrows). Compare this view of the lattice with that in figure 8. Four weeks post-operation. x 45,000.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W. Brightman
PLATE 4
15
PLATE 5 EXPLANATION
10
16
OF F I G U R E
Volcano-like projections (V) on the B-face (B2) of glial membranes are seen i n both cross-fracture and in lateral view thus revealing their continuity (arrows) with the round depressions (I)) on the corresponding A-face (A2) of the plasma membrane on the other side of the process. The main glial membrane face here is flanked on either side by an extracellular cleft (Ct) and the A-face of an adjacent glial membrane (Al) is seen i n the upper left of the figure. Junctional complexes (J) are present in some areas of the membrane surface. Three weeks postoperation. X 52,500.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W . Brightman
PLATE 5
PLATE 6 EXPLANATION O F FIGURES
18
11
Trans-glial channels do not appear to be precisely aligned between adjacent glial lamellae however, they are in register to the extent that areas of membrane which possess them coincide across the sheath. No regular spacing of channels occurs (fig. 12). The extracellular clefts (Ct) and trans-glial channels (arrows) are filled with HRP reaction product. Three weeks post-operation. X 90,000.
12
Trans-glial channels in freeze-fracture replicas closely resemble those prepared for thin section (fig. 11). The A-face of the plasma membrane on one side of the cell is seen overlying the cytoplasm (G) and B-face of the plasma membrane of the opposite side of the cell. Note the transglial channels extend from one side of the cell to the other through the cytoplasm (brackets). Variation in form of the volcano-like projections on the B-face is appearent. Arrowheads identify fully-formed channel profiles, black arrows point to immature profiles and circled areas delimit forming channels. Normal root sheath. X 88,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 6
19
PLATE 7 EXPLANATION O F FIGURE
13
20
Fracture planes across the glial sheath show channel profiles i n corresponding areas of membrane of glial cells adjacent to one another. Aggregations of particles representing junctional complexes occur (dotted circles) i n close association with trans-dial channels: esuecially in regenerating glial sheaths. Three weeks post-operation. X 37,500.
TRANS-GLIAL C H A N N E L S Richard R. Shivers and Milton W. Brightman
PLATE 7
21
PLATE 8 EXPLANATION
14
22
O F FIGURE
This replica provides an extensive membrane surface for examination. The B-face is covered with channel profiles which exhibit some clustering -with adjacent areas of bare membrane surface. Interspersed among the protrusions (V) are junctional complexes (") which occur i n great abundance near the top of the figure. Fractures of trans-glial channels showing longitudinal profiles are identified with arrows. Three weeks post-operation. x 25,500.
TRANS-GLIAL C H A N N E L S Richard R. Shivers and Milton W. Brightman
PLATE 8
23
PLATE 9 EXPLANATION
15
24
O F FIGURE
Relationships between trans-glial channels and junctional complexes are seen in this high magnification replica which includes membrane surfaces from several glial processes. Both A- and B-faces are shown here and the B-faces contain volcano-like protuberances (V) and particle aggregates of junctions (circled), while the A-faces exhibit the complementary surfaces of these -round depressions (D) and clusters of small pits (circled). A spectrum of forms of the volcano-like projections is seen and identified as i n figure 12. Three weeks post-operation. X 40,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 9
25
TKANS-GLIAL C H A N N E L S R i c h a r d R . Shivers a n d Milton W . B r i g h t m a n
EXPLANATION O F FIGURES
26
16
Junctional particle aggregates (broken circle) are often closely associated with trans-glial channels. It is suggested that the junctional complexes are hemi-desmosomes which c a n be compared to those in figure 1 7 . Normal root sheath. X 90,000.
17
Trans-glial channels (C) are closely associated with hemi-desmosome type junctions ( H ) which are numerous in layers of the sheath. Compare this figure with figure 16. Normal root sheath. X 90,000.
P L A T E 10
ABSTRACT The sheath around the roots of the sixth abdominal ganglion in the yentral nerve cord of the crayfish consists of concentric layers of thin glial processes alternating with wide clefts containing filamentous connective tissue. Regions of each glial lamella are perforated by single, short, tubular channels: the trans-dial channels. In thin plastic sections examined in the electron microscope, the channels appear as slits that are 240 A wide and 450-550 A long which traverse glial lamellae less than 1,500 A thick. Branched tubular channels cross glial sheets that are thicker than 1,500 A. The thickest glial wrap is adaxonal; it closely encapsulates individual axons and its cell membrane is separated from the axolemma by a collagen-free space of only 150 A. The adaxonal glial cytoplasm contains unique, three-dimensional networks of interconnected tubules. Separate tubular lattices occur along these thicker processes. In replicas of freeze-fractured sheaths, the outer half of the plasma membrane belonging to the thin glial sheets exhibits many volcano-like protrusions which represent cross fractures through the necks of trans-glial channels. Corresponding depressions o n the inner half of these membranes are sites where the plasma membrane invaginates to form the channels. Although some channels are randomly dispersed, others are lineraly positioned i n restricted areas across successive glial layers. The number of channels is far more readily appreciated i n replicas than in thin sections. The average frequency of channels is 16 per ~2 (range 8 to 33) in normal roots and does not differ significantly from the average of 13 per pz i n proximal stumps of roots fixed three to four weeks after the roots were cut. The channels are not precisely aligned from one glial layer to the next but do appear to coincide approximately with the adaxonal tubular lattice. The combination of trans-glial channels and adaxonal tubular lattices may provide a complex conduit that could facilitate a rapid, passive flow of electrolytes and nutrients across the nerve sheath to the axonal surface. Horseradish peroxidase solutions bathing the ventral roots enter the trans-glial channels, extracellular clefts and finally the tubular lattices. This distribution supports the proposed role of the channels in a rapid extracellular passage of solutes. The channel profiles have a range of forms consistent with the supposition that they are not static but continually reforming. There are indications that, proximal to the cut, the areas of glial plasma membrane with channel profiles contain more junctional complexes between regenerating cells than between glial cells of normal sheaths. The channel profiles and aggregates of particles belonging to junctions are closely associated when they occupy the same region of the membrane.
Glial or supportive cells of Crustacean nervous systems have been implicated in a variety of roles in maintenance of the neuronal components. Glial cells comprise the major element of the highly ordered sheath which surrounds roots of ventral ganglia in crayfish (Somers and Nunnemacher, ’70). Because of their predominance, the glial sheets have been suggested to function in part as selective barriers to J.
COMP.
NEUR.,167: 1-26.
nutrients and electrolytes which must traverse the sheath to axons contained within (Brandt et al., ’65; Geren and Schmitt, ’54; Holtzman et al., ’70; Malzone et al., ’66). Experiments with electron opaque “tracers” have suggested that the tracers reach the surface of axons by a combination of I Present address; Department of Zoology, University of Western Ontario, London, Ontario, C a n a d a .
1
2
RICHARD R. SHIVERS A N D MILTON W. BRIGHTMAN
meters. This gap was of sufficient width that resection of a piece of the nerve root was unnecessary (Hoy, '69). Animals were then returned to crushed ice for 30 minutes before being placed in shallow pans of water. Only the proximal stumps of interrupted ganglionic roc,ts and their growing axonal and glial processes were prepared for freeze-fracture and routine thin sectioning three to eight weeks following nerve section. Unoperated crayfish were anesthetized in crushed ice for 30 minutes prior to dissection. The ventral musculature was opened and the exposed sixth ganglion and its roots were flooded for several minutes with ice-cold fixative. When the tissue had begun to harden and turn yellow, the ganglion and roots, which were at least 4 mm long, were cut free and placed in fresh fixative. Fixation for routine electron microscopical studies was carried out in ice-cold 6 % glutaraldehyde (from Fisher 50 % ) (Sabatini et al., '63) buffered to pH 7.5 with 0.2 M sodium phosphate. All solutions had sucrose added at a concentration of 3 % . Aldehyde fixation was for a minumum of two hours; longer periods did not appear to improve the quality of fixation. Postfixation for one to three hours in 2 % O s 0 4 buffered to pH 7.5 with sodium phosphate was followed by a one and one-half hour wash in buffer. Following osmication, the tissues were stained en bloc with aqueous uranyl acetate (Heuser and Reese, '73), dehydrated through graded methanol to METHODS propylene oxide, and embedded in English Adult crayfish (Orconectes uirilis) were Araldite (Ciba Ltd, Duxford, Cambridge, obtained from Nasco Supply (Ft. Atkinson, England). Thin sections were stained with Wisc.) and kept in shallow pans of aerated 10% uranyl acetate (Heuser and Reese, tap water. They were fed weekly on horse- '73) and lead citrate (Reynolds, '63). (All meat and maintained at room temperature. tissues were examined on an AEI-6B elecFor parts of the present investigation, tron microscope.) roots of the sixth abdominal ganglion were Tissue preparation for freeze-fracture used. The crayfish were anesthetized for was begun by fixing the tissues in ice-cold 30 minutes in crushed ice and then a shal- 6 % glutaraldehyde (described above) for low incision was made at the posterior one hour. The nerve roots were then rinsed margin along the ventral surface of the briefly in 0.2 M sodium cacodylate buffer fifth abdominal segment. The sixth abdom- (pH 7.35) and placed in a solution of 20% inal ganglion of the ventral nerve cord was glycerol in 0.2 M sodium cacodylate (pH carefully exposed and its roots were cut. 7.35) overnight in the refrigerator. PrepNerve root stumps attached to the gan- aration of the material for rapid freezing glion (proximal stumps) were a minimum consisted of placing short (less than 2 mm) of 4 mm long; the distal stumps retracted pieces of nerve roots in a drop of 20% glyfrom the cut leaving a gap of several milli- cerol on gold specimen discs (Balzer's, routes: (1) seepage through the connective tissue spaces between thin glial cytoplasmic processes of the nerve sheath (extracellular clefts); and (2) passage across the adaxonal glial cytoplasmic layer through a fine anastamosing system of tubules (the tubular lattice), thereby appearing in the intercellular space adjacent to the axon plasma membrane (Abbott, '70; Holtzman et al., '70; Kristensson et al., '72; Lane and Abbott, '75; Shivers, '70). From these and other studies it has been proposed that a system of passageways exists across the glial sheath of crayfish nerves which functions as a transport route for essential elements for nerve maintenance (Dehrenzo, '68; Holtzman et al., '70; Peterson and Pepe, '61). The present study on crayfish ventral ganglion roots using freeze-fracture techniques has revealed a hitherto unreported system of trans-glial channels. The channels are inconspicuous in thin sections because of the highly undulating nature of the glial cell membrane and its indentations, in places, as shallow pits and elongated tubules. The channels are far more obvious in freeze-fracture planes that present broad vistas of internal faces within the plasma membrane. In such planes, the distribution and frequency of the channels are readily appreciated. The ubiquity and large numbers of channels suggest that they comprise the primary route between the exterior of the nerve sheath and the surface of axons.
TRANS-GLIAL CHANNELS
Lichtenstein). The discs were plunged into Freon 22 cooled with liquid nitrogen (LN,) for ten seconds and quickly transferred to LN, for storage until fracturing. Tissues were fractured at - 111 "C in a Balzer Apparatus, shadowed with a thin layer of platinum and lightly coated with carbon. Specimens were not etched. Replicas were cleaned in Chlorox for two hours, rinsed in four changes of distilled water and picked up on bare or parlodoin-coated 200 mesh grids. In order to follow the extracellular movement of protein, the sixth abdominal ganglion and roots were removed from several animals and placed in a medium of crayfish ringer (Van Harreveld, '36) containing 1 mg/ml of horseradish peroxidase (Type VI, Sigma Chemical Co.,St. Louis, Mo.). Nerves were immersed for one hour and then fixed for eight hours at room temperature in 3% glutaraldehyde (from Taab 2 5 % ) buffered to pH 7.35 with 0.2 M sodium cacodylate. Following aldehyde fixation tissues were washed overnight in refrigerated buffer and then incubated for two hours on crushed ice (Graham and Karnovsky, '66) to demonstrate peroxidatic activity. The tissue was then processed for electron microscopy as above. Enumeration of channel profiles in electron micrographs of replicas was made using a Hewlett-Packard Model 9810A computer coupled with a Hewlett-Packard Model 9800 digitizer. Programs were written to facilitate simultaneous counting of profiles and area density calculations. These operations were carried out on both normal replicas and replicas of growing nerve and glial processes that were proximal to the transection. OBSERVATIONS
A. Glial arrangement Roots of the sixth abdominal ganglion in the crayfish consist of varying numbers of sensory and motor axons which are encapsulated by a multilamellate sheath. This sheath is composed of closely packed, narrow layers of glial cytoplasm which are usually separated from each other by wider extracellular clefts filled with differing amounts of connective tissue filaments about 100 A thick (fig. 1). The clefts are usually three to four times wider than
3
the glial processes. In a few places, multiple glial layers are separated by only a narrow cleft 150-200 A wide (fig. 1). Axons may be wrapped by 17 alternating layers of glial sheets and connective tissue spaces (fig. 1). The layers of glial cytoplasm are high1 attenuated and may be as thin as 180 with no organelles or much wider, as in the adaxonal glial processes (Ad), which usually contain a complex array of organelles. There is always a glial process immediately adjacent to the axon (fig. 1) with a narrow space of about 150 8, between the two. The arrangement of glial lamellae and extracellular filament-filled clefts can be clarified by using freeze-fracture techniques. The fracture plane passes from one layer of the sheath to the next, exposing the inner surface of the outer leaflet (B-face) or the outer surface of the cytoplasmic leaflet (A-face) of the glial plasma membrane (Branton, '66) (fig. 2). The outer half of the glial plasmalemma is characterized by a sparse population of particles (70-110 A in diameter). The inner half of the glial plasma membrane has a very dense array of randomly distributed particles of the same size.
k:
B. Trans-glial channels The thin glial sheets viewed in thin plastic sections are randomly traversed by short more or less stroaightchannels about 240 8, wide and 500 A long (figs. 4, 5, 11). Each channel appears as a single profile that extends from one side of a glial process to the other side, to form a passageway across the sheet. The channels are about as long as the glial lamella is thick, although a few run diagonally for a short distance or may branch (fig. 5). The channels are inconspicuous in thin sections because the glial cell membranes often undulate and, in places, are indented by shallow pits or deeper tubules that do not traverse the entire thickness of the process (fig. 17). It is likely that some of the tubular indentations are portions of channels. The frequency and distribution of the channels is more readily appreciated in replicas of frozen, cleaved specimens. The openings of the channels appear as small depressions on the inner half (A-face) of the glial plasmalemma and as volcano-like protrusions on the outer half (B-face) of
4
RICHARD R. SHIVERS AND MILTON W. BRIGHTMAN
the glial cell membranes (figs. 3, 10, 1215). The profiles of deprpsions and protrusions are about 240 A wide and thus correspond in size to the channels seen in thin sections. The profiles appear individually or in clusters. It is suggested that the depressions on the inner (A) half of glial membranes represent the invaginations of the channels from the membrane surface. The protrusions on the outer (B) half represent cross-fractures through the necks of the channels as they extend toward the opposite side o f the cell. The channels may be fractured in such a way as to show that they pass entirely through the thin glial process (figs. 3, 10). The depressions on the inner half of the cell membrane directly abut against the protrusions on the outer half of the same membrane (fig. 3). The channels appear tubular in thin sections and, in replicas, the depressions and elevations corresponding to the channels, therefore appear hollow (fig. 10). The distribution of trans-glial channels is variable and can better be determined in replicated membrane faces. The channels occur in successive, concentric glial processes including the inner one next to the lattice-containing adaxonal glial sheet. The channel openings are usually scattered at random (fig. 2), but some are arranged in ordered clusters (figs. 13-15). Thus, patches of membrane containing channel openings alternate with areas devoid of these depressions and protuberances. It is emphasized that the channel-free areas may be quite extensive. In a given sheet, the perforated zones may be coincidental with those of an adjacent glial process (figs. 2 , 13-15). Nevertheless, individual openings in one cell’s membrane are usually not in strict register with the openings of an adjacent glial sheet (figs. 2 , 11, 12). In regenerating sheaths, the channels are especially well ordered, the depressions and protuberances forming rows that are linear (figs. 10, 13, 15) or nearly circular (fig. 14). Adjacent circular clusters may have sharp boundaries separated by strips of membrane devoid of channel openings (fig. 14). Within a glial membrane, the areas between individual openings may be regular (fig. 14) but are more often highly variable in width and contour (figs. 2, 11, 12). Neighboring channels may be as close together
as 450 A or as far apart as 3,000 A. The zone immediately surrounding a junction may be perforated by channels (fig. 15) but is usually not (fig. 10). Although the channels are generally uniform in morphology and size (figs. 12, 15), there is some variation. Some protrusions are much higher than others and have a rim lined with discrete particles; others appear flat with smoother rims (figs. 12, 15). In regenerating sheaths, the individual protrusions in the outer half of the membrane appear to be more variable in height (fig. 15). The frequency of channels in the glial processes of intact nerve roots is 16 per p z . This density does not differ significantly from the 13 per p 2 found in roots that had been cut three to four weeks prior to fixation. Only those regions of the cell membrane containing channel profiles were selected for area measurements. These values, therefore, do not include the large surface of membrane that does not bear channel configurations. In order to determine whether the channels do indeed serve as passageways across successive glial sheets, nerve roots were immersed in a medium containing peroxidase. These roots were either intact or were proximal to a cut that had been made three weeks before the immersion. After 60 minutes, the protein had entered the channels and the extracellular clefts with which they communicated (figs. 4, 11). This pattern is the same in both intact and cut sheaths. The peroxidase-occupied channels were slightly narrower t h q those in replicas but were about 470 A long. The length therefore, is comparable in both thin sections and in replicas.
c. Tubular lattice The glial process closest to the axons is wider than the other glial sheets and contains a rich array of organelles, including a well-developed system of anastamosing tubules (figs. 1, 2). This tubular lattice is of variable complexity that ranges from slightly branched tubules (figs. 2, 6, 7) in narrower portions of the glid process to highly branched systems in wider- parts (fig. 8). The tubules are about 250 A wide (fig. 8). The three dimensional view and, consequently, the considerable surface area of the lattice is apparent in freeze-fracture
5
TRANS-GLIAL CHANNELS
replicas (fig. 9). These images confirm the regenerating sheaths (figs. 13-15) than in observation that the lattice is made up of intact ones (fig. 16). However, some junca series of invaginations of the glial plasma tional aggregates occur separately from membrane (Holtzman et al., '70). Each channels (figs. 10, 13, 14). As in normal lattice appears to be separate from neigh- sheaths, the junction aggregates of regenboring ones and it has not been ascertained erating glial processes are discoid, sharply whether there is any regular spacipg be- circumscribed and variable in overall ditween adjacent lattices. The 150 A wide ameter (figs. 10, 13-15). extracellular cleft between the adaxonal DISCUSSION glial layer and the axons is much narrower than the clefts between successive The system of trans-glial channels across layers of peripheral glia. successive glial sheets around the ventral In freeze-fracture replicas, the lattice nerve roots in crayfish may provide a short appears as a complex array of volcano-like route for large molecules moving from protrusionso (fig. 9). The protrusions are blood and extracellular spaces to the surabout 240 A wide and correspond therefore face of neurons and their processes. At in diameter to the lattice tubules in thin first inspection of thin sections, the glialplastic sections (figs. 6 4 ) . The protrusions connective-tissue sheath around the nerve represent cross-fractures through the necks roots appears to present a formidable obof the lattice tubules where their mem- stacle to such passage. If there were no branes become continuous with the plasma transverse channels but only the clefts membrane. Inasmuch as only a few frac- between the end-to-end abutments of adtured lattices have been encountered, it is jacent glial processes, a molecule would suggested that the fracture plane usually have to migrate for relatively long distances skips the adaxonal glial layer to pass step- before it could move across one glial sheet wise through the thinner, more peripheral into the collagen-containing extracellular members of the sheath. space between that sheet and a neighboring one. Instead, the cross-channels proD. Junctions vide many short, straight routes across the The most common junction encountered sheath. in the glial capsule is the hemi-desmosome The frequency of transverse channels, in (fig. 1). In thin plastic sections, this type the segments of glial plasma membrane of junction is recognized, as in other phyla, where they occur, is variable but about the by a patch of granulo-filamentous, dense same in normal and regenerating sheaths. material that inserts into a segment of Their frequency is not evident in thin cytoplasmic leaflet belonging to the cell plastic sections where only one or a few membrane (fig. 17). The most common channel profiles can be encountered in arrangement of particles in frozen-cleaved any one field. If the channels were rancell membranes consists of a disc$id ag- doml dispersed in plastic sections about gregate of large particles 90-110 A in di- 600 thick, then there would be about 16 ameter that occur on the outer half (B- such sections in a slice 1 p thick. As there face) of glial plasma membranes and of a are about 16 channels per p2 in the replicorresponding patch of complementary pits cated surface of a glial sheet, one would on the inner half (A-face) (figs. 10, 13-15). then expect to see one channel profile, or Since the hemi-desmosomes in thin sections in segments where they are not randomly and the particle aggregates in replicas are dispersed, a few more profiles in a given the most frequently encountered structures thin section. associated with cell junctions, it is inferred Although the entire membrane of each that the two are equivalent. glial process is not perforated by channels, In replicas, the discoid aggregates may limited segments are. The single channels, be located alongside channel profiles (figs. being about 250 A wide, would be expected 13-16). The aggregates, presumably hemi- to allow the non-selective, passive flow of desmosomes, appear to be the only junc- large and small molecules across segments tions in close proximity to the trans-glial of an individual lial lamella. The short channels. The association between junc- length, about 450 to 550 A, of the single tions and channels is more frequent in channels and their straight course would
K
1
6
RICHARD R. SHIVERS A N D MILTON W. BRIGHTMAN
enable the flow to be a rapid one across a given process. The approximate alignment of the channels between one glial sheet and successive neighbors would in turn, shorten the traverse time across the entire multi-laminated sheath. It has been recognized by Holtzman et al. (’70) that the glial sheath does permit the migration of peroxidase from perineural fluid into the tubular lattice of the adaxonal glial cell. The passage was assumed to take place by way of the extracellular collagen-occupied spaces designated as “channels,” a term restricted in the present account to the single, short tubules perforating thin glial processes. It was implied that large substances crossed the glial sheath by seeping through the long extracellular diffusion path around the ends of adjacent glial cells (Holtzman et al., ’70; Kristensson et al., ’72; Lane and Abbott, ’75). The authors did not mention, nor have we found, any signs of vesicular transport across the glial sheets. If these glial processes behave like those of mammalian brain, it would be expected that a cell process less than about 300 d; thick is incapable of forming pinocytotic pits (Brightman, ’65). Although thicker portions of the glial processes do contain peroxidase labelled vesicles (Lane and Abbott, ’75), an observation that we have confirmed, it appears that such uptake is for incorporation into the cell rather than for transfer across it. Contrary to this inference is the possibility that protein can be transferred directly from the adaxonal glial processes to the underlying axon. Proteins, newly synthesized i n Schwann cells, appear to be transferred to the adjacent giant axon of squids (Lasek et al., ’74). The collagen-containing extracellular spaces or clefts and the transverse channels together make up a continuous passageway. The channels however, do not appear to be simply branches of the extracellular clefts. The clefts are 200-3,000 wide and contain filaments, probably collagenous, whereas the trans-glial tubules do not. Furthermore, the variations in the size of the channel openings in replicas suggest that they may be labile, perhaps closing and reforming under certain conditions as yet undefined.
All clefts and channels are ultimately penetrated by tracer protein after exposure of ganglionic root segments to peroxidasecontaining medium for periods of one hour or longer. This observation supports the suggestion that trans-glial channels serve as a short-cut route across the sheath. Unlike insect nerve sheaths which are impermeable to tracers (Treherne and Pichon, ’72), Crustacean nerve sheaths are leaky (Holtzman et al., ’70; Kristensson et al., ’72; Lane and Abbott, ’75; Shivers, ’70). We propose that not only is the crayfish nerve sheath permeable to molecules (including tracers) but that the trans-glial channels and tubular lattices represent a conduit which permits rapid solute flow across the sheath. Further studies are necessary to determine the rate at which molecules move across the sheath via the trans-glial channel system and whether such molecules are selectively transported by this route. Several variations in the basic morphology of trans-glial channel profiles in replicas are reported in the present paper. These are interpreted as representing different stages in the formation of the channels as they invaginate from the plasma membranes of glial processes. Evidence for turnover is difficult to obtain from thin sectioned material since only few channels are encountered in a given thin section. Investigation of the channels must therefore be done with freeze-fracture techniques which more accurately and consistently preserve the channels (Moor, ’72; Moor and Muhlethaler, ’63). Increased rate of formation or increased numbers of trans-glial channels may be a part of the processes occurring during regeneration. Experimentally induced nerve growth results in a pronounced glial reaction characterized by increased cytoplasmic complexity (Nordlander and Singer, ’72; Shivers and Brightman, ’74). Rapid growth and reformation of the trans-glial channels would be anticipated as part of the reestablishment of a complete nerve root sheath. The appearance in regenerating glial processes of a more frequent close association of trans-glial channel profiles with aggregations of particles assumed to be junctions suggests that a functional rela-
TRANS-GLIAL CHANNELS
7
tionship may exist between them. Devel- trations in this paper, and to Dr. Lise oping tissues often exhibit formation of Prescott for her invaluable advice regardnew intercellular junctional complexes ing freeze-fracture techniques. (Gilula, '73; Revel et al., '73),and although LITERATURE CITED recent studies of regeneration in crayfish motor nerves indicated no striking increase Abbott, N. J. 1970 Absence of a blood-brain barrier in a crustacean, Carcinus maenas L. Nain junctions between glial cells (Nordture (London), 225: 291-293. lander and Singer, '72), the present study Brandt, P. W., A. R. Freeman, J. P. Reuben and has shown rather extensive junctional H. Grundfest 1965 Variations between axon complexes in new glial processes (Shivers and Schwann membranes induced in lobster nerve fibers by currents and fluxes. Biol. Bull., and Brightman, '74) which will be de129: 400. scribed in detail in a separate report. The Branton, D. 1966 Fracture faces of frozen memmost common junction in the sheath is a branes. Proc. Nat. Acad. Sci. (U.S.A.), 55: 1048type of hemi-desmosome which appears as 1056. a dense plaque in the cytoplasm adjacent Brightman, M. W. 1965 The distribution within the brain of ferritin injected into cerebrospinal to the glial surface abutting on intercellufluid compartments. 11. Parenchymal distribular clefts. Similar structures have been tion. Am. J. Anat., 117: 193-220. reported by Overton and Mapp ('74) in DeLorenzo, A. J. D., M. Brzin and W. 0. Dettbarn 1968 Fine structure and organization of nerve regenerating notocord. Changes in the fibers and giant axons in Homarus americanus. number, distribution and appearance of J. Ultrastruct. Res., 24: 367. trans-glial channels and junctions could Flower, N. E. 1972 A new junctional structure reflect changes which do not interrelate i n the epithelia of insects of the order Distyoptera. J. Cell Sci., 10: 683. the two structures but instead, represent Geren, B. B., and F. 0. Schmitt 1954 The strucalterations in glial cells during growth. ture of the Schwann cell and its relation to the It is interesting to note that both the axon i n certain invertebrate nerve fibers. Proc. Nat. Acad. Sci. ( U . S . A . ) , 4 0 : 863. particle aggregates constituting junctional complexes and the volcano-like profiles of Gilula, N. B. 1972 Cell junctions of the crayfish hepatopancreas. J. Ultrastruct. Res., 38: 215. trans-glial channels occur on the B-face 1973 Development of cell junctions. of glial plasma membranes. This is differh e r . Zool., 1 3 : 1109-1117. ent from most other systems in which such Graham, R. C., and M. J. Karnovsky 1966 The early stages of absorption of injected horseradish structures are usually present on A-faces peroxidase in the proximal tubules of mouse of membranes (Gilula, '73). The A-face of kidney. Ultrastructural correlates by a new techmembranes in the present study contains nique. J. Histochem. Cytochem., 14: 291-302. instead, round depressions which are the Heuser, J. E., and T. S. Reese 1973 Evidence for recycling of synaptic vesicle membrane durcomplementary images of the protrusions ing transmitter release at the frog neuromuscuand clusters of pits corresponding to the larjunction. J . Cell Biol., 57: 3 1 5 3 4 4 . junctional particle aggregates. This dis- Holtzman, E., A. R. Freeman and L. A. Kashner 1970 A cytochemical and electron microscope tribution is similar to that seen in some study of channels i n the Schwann cells surroundother invertebrate systems (Flower, '72; ing lobster giant axons. J . Cell Biol., 44; 438Gilula, '72). 444. Trans-glial channels have not been re- Hoy, R. R. 1969 Degeneration a n d regeneration in abdominal flexor motor neurons in the crayported in nerve sheaths of other invertefish. J. Exp. Zool., 72: 219-232. brates or vertebrates. Peracchia ('73b, '74) described structures similar to the channels Kristensson, K., E. Stromberg, R. Elofsson and Y. Olsson 1972 Distribution of protein tracers reported here but regarded them only as in the nervous system of the crayfish (Astacus part of the adaxonal tubular lattice. Addiastacus L.) following systemic and local application. J. Neurocytol., I : 3 5 4 8 . tional studies are needed to determine the distribution of trans-glial channels in Lane, N. J., and N . J. Abbott 1975 The organization of the nervous system i n the crayfish other crustacean nerve sheaths and to Procambarus clarkii, with emphasis on the bloodassign a detailed functional role to them. brain interface. Cell Tiss. Res., 156; 173-187. ACKNOWLEDGMENTS
The authors would like to express their appreciation to Mrs. Clara Mowry for her excellent help in preparation of the illus-
Lasek, R. J., H. Gainer and R. J. Przybylski 1974 Transfer of newly synthesized proteins from Schwann cells to the squid giant axon. Proc. Nat. Acad. Sci. (U.S.A.), 71: 1188-1192. Malzone, W. F., G. H. Collins and R. R. Cowden 1966 Neurological relationships i n the thoracic
RICHARD R. SHIVERS AND MILTON W. BRIGHTMAN
8
ganglion of the fiddler crab, Uca. J . Comp. Neur., 127: 5 1 1 5 3 0 . Moor, H. 1972 The principles of freeze-etching. Revista Microsc. Elect., 1 : 269-275. Moor, H., and K. Muhlethaler 1963 Fine structure in frozen-etched yeast cells. J . Cell Biol., 1 7 : 609-628. Nordlander, R. H., and M. Singer 1972 Electron microscopy of severed motor fibers i n the crayfish. Z. Zellforsch., 126: 157-181. Overton, J., and F. E. Mapp 1974 The fine structure of regenerating notocord in anuran tadpoles. J. Exp. Zool., 187: 103-120. Peracchia, C. 1973b Low resistance junctions i n crayfish. 11. Structural details and further evidence for intercellular channels by freezefracture and negative staining. J . Cell Biol., 57: 66-76. 1974 Excitable membrane ultrastructure I . Freeze-fracture of crayfish axons. J . Cell Biol., 61 : 107-122. Peterson, R. P., and F. A. Pepe 1961 The fine structure of inhibitory synapses in the crayfish. J. Biophys. Biochem. Cytol., I 1 ; 157. Revel, J.-P., P. Yip and L. L. Chang 1973 Cell
junctions in the early chick embryo - a freezeetch study. Develop. Biol., 35: 302-317. Reynolds, E. S. 1963 The use of lead citrate at high pH as a n electron opaque stain in electron microscopy. J. Cell Biol., 1 7 : 200-211. Sabatini, D. D., K. Bensch and R. J. Barrnett 1963 Cytochemistry and electron microscopy. The preservation of cellular ultrastructure and enzyme activity by aldehyde fixation. J. Cell Biol., 1 7 : 19-58. Shivers, R. R. 1970 Fine structure of crayfish optic ganglia vascularization and permeability. J . Cell Biol., 47: 191A. Shivers, R. R., and M. W. Brightman 1974 Unpublished Observations. Somers, M . E., and R. E. Nunnemacher 1970 Microanatomy of the ganglionic roots of the abdominal cord of the crayfish, Orconectes virilis (Hagen). J. Comp. Neur., 138: 209-218. Treherne, J. E., and Y. Pichon 1972 The insect blood-brain barrier. In: Advances in Insect Physiology. Vol. 9. Academic Press, pp. 257313. Van Harreveld, A. 1936 A physiological solution for freshwater crustaceans. Proc. Soc. Exptl. Biol. Med. Sci., 34: 428.
Abbreviations A, A-face Ad, Adaxonal glia Ax, Axon B, B-face C, Trans-glial channel Ct, Extracellular cleft; connective tissue layer D, Round depression F, Filaments
G , Glial cytoplasm H, Hemi-desmosome J, Junctional complex L, Adaxonal tubular lattice N, Nucleus V, Volcano-like protrusion Circled Arrows indicate direction of platinum shadowing i n replicas.
PLATE 1 EXPLANATION O F FIGURE
1
Normal ventral ganglion roots contain axons (Ax) encased in a multilamellate glial (G)connective tissue (Ct) sheath. Only adaxonal glial layers (Ad) adjacent to axons contain numerous organelles and specialized tubular networks - the tubular lattices (L). Hemi-desmosomes (H) attach the glial lamellae to adjacent connective tissue layers. X 12,500.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 1
9
PLATE 2 EXPLANATION
2
10
OF FIGURE
The lamellar arrangement of glial cells i n the root sheath is more obvious in freeze-fracture. Exposed intramembrane surfaces contain either volcano-like protrusions (V) on B-faces o r round depressions (D) on A-faces. Extracellular clefts (Ct) filled with filamentous material (F) are interposed between very narrow lamellae of glial cytoplasm (G). The rectangular area shows a portion of glial cytoplasm containing a slightly branched tubular network. Occasionally, glial lamellae do not alternate with extracellular clefts (circled areas. See also circled area of fig. 1 ) . Normal root sheath. X 37,600.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W . Brightman
PLATE 2
11
PLATE 3 EXPLANATION OF FIGURES
12
3
Fractures exposing intact trans-glial channels clearly show the relationship between volcano-like projections on B-faces (B,) a n d round depressions o n A-faces (A,) (arrows). Note that the plasma membrane of adjacent glial lamella (A2) also exhibits channel profiles. Trans-glial channels appear in thin section a s small, round tubules with electronlucent cores (arrows, inset). Normal root sheath. X 75,000. Inset x 60,000.
4
Trans-glial channels (C) traverse glial lamellae and connect adjacent extracellular clefts (Ct). Clefts a n d channels contain HRP reaction product. Three weeks post-operation. X 152,000.
5
Two forms of trans-glial channels are seen. In glial lamellae less t h a n 1,500 A wide, channels (arrows) are short a n d single. In wider lamellae the channels appear branched (circled area). Normal root sheath. X 96,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W . Brightman
PLATE 3
PLATE 4 EXPLANATION OF FIGURES
14
6
Glial processes wider than 1,500 A exhibit a moderately-branched (arrows) trans-glial tubular network. Usually glial layers such as these are found in a n adaxonal position. Normal root sheath. X 99,000.
7
Adaxonal tubular lattice which is slightly branched consists mainly of elongate tubular profiles (between arrows) traversing the process. Normal root sheath. X 45,000.
8
Adaxonal glial cells possess large, highly anastamosed networks of tubules (arrows) -the tubular lattices. I n such cases, a regular pattern of construction of the lattice is suggested. Normal root sheath. X 114,000.
9
Tubular lattices c a n be seen in freeze-fracture replicas and exhibit a n extensive honeycomb appearance because of their complex branched and interconnected tubules (arrows). Compare this view of the lattice with that in figure 8. Four weeks post-operation. x 45,000.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W. Brightman
PLATE 4
15
PLATE 5 EXPLANATION
10
16
OF F I G U R E
Volcano-like projections (V) on the B-face (B2) of glial membranes are seen i n both cross-fracture and in lateral view thus revealing their continuity (arrows) with the round depressions (I)) on the corresponding A-face (A2) of the plasma membrane on the other side of the process. The main glial membrane face here is flanked on either side by an extracellular cleft (Ct) and the A-face of an adjacent glial membrane (Al) is seen i n the upper left of the figure. Junctional complexes (J) are present in some areas of the membrane surface. Three weeks postoperation. X 52,500.
TRANS-GLIAL CHANNELS Richard R . Shivers and Milton W . Brightman
PLATE 5
PLATE 6 EXPLANATION O F FIGURES
18
11
Trans-glial channels do not appear to be precisely aligned between adjacent glial lamellae however, they are in register to the extent that areas of membrane which possess them coincide across the sheath. No regular spacing of channels occurs (fig. 12). The extracellular clefts (Ct) and trans-glial channels (arrows) are filled with HRP reaction product. Three weeks post-operation. X 90,000.
12
Trans-glial channels in freeze-fracture replicas closely resemble those prepared for thin section (fig. 11). The A-face of the plasma membrane on one side of the cell is seen overlying the cytoplasm (G) and B-face of the plasma membrane of the opposite side of the cell. Note the transglial channels extend from one side of the cell to the other through the cytoplasm (brackets). Variation in form of the volcano-like projections on the B-face is appearent. Arrowheads identify fully-formed channel profiles, black arrows point to immature profiles and circled areas delimit forming channels. Normal root sheath. X 88,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 6
19
PLATE 7 EXPLANATION O F FIGURE
13
20
Fracture planes across the glial sheath show channel profiles i n corresponding areas of membrane of glial cells adjacent to one another. Aggregations of particles representing junctional complexes occur (dotted circles) i n close association with trans-dial channels: esuecially in regenerating glial sheaths. Three weeks post-operation. X 37,500.
TRANS-GLIAL C H A N N E L S Richard R. Shivers and Milton W. Brightman
PLATE 7
21
PLATE 8 EXPLANATION
14
22
O F FIGURE
This replica provides an extensive membrane surface for examination. The B-face is covered with channel profiles which exhibit some clustering -with adjacent areas of bare membrane surface. Interspersed among the protrusions (V) are junctional complexes (") which occur i n great abundance near the top of the figure. Fractures of trans-glial channels showing longitudinal profiles are identified with arrows. Three weeks post-operation. x 25,500.
TRANS-GLIAL C H A N N E L S Richard R. Shivers and Milton W. Brightman
PLATE 8
23
PLATE 9 EXPLANATION
15
24
O F FIGURE
Relationships between trans-glial channels and junctional complexes are seen in this high magnification replica which includes membrane surfaces from several glial processes. Both A- and B-faces are shown here and the B-faces contain volcano-like protuberances (V) and particle aggregates of junctions (circled), while the A-faces exhibit the complementary surfaces of these -round depressions (D) and clusters of small pits (circled). A spectrum of forms of the volcano-like projections is seen and identified as i n figure 12. Three weeks post-operation. X 40,000.
TRANS-GLIAL CHANNELS Richard R. Shivers and Milton W. Brightman
PLATE 9
25
TKANS-GLIAL C H A N N E L S R i c h a r d R . Shivers a n d Milton W . B r i g h t m a n
EXPLANATION O F FIGURES
26
16
Junctional particle aggregates (broken circle) are often closely associated with trans-glial channels. It is suggested that the junctional complexes are hemi-desmosomes which c a n be compared to those in figure 1 7 . Normal root sheath. X 90,000.
17
Trans-glial channels (C) are closely associated with hemi-desmosome type junctions ( H ) which are numerous in layers of the sheath. Compare this figure with figure 16. Normal root sheath. X 90,000.
P L A T E 10
