ABSTRACT
Connectives of the ventral nerve cord of Manduca sexta consist of glia-ensheathed axons surrounded by a perineurium and an acellular neural lamella, which is greatly expanded on the dorsal surface. The glial cells are linked to one another by desmosomes and tight junctions; the latter also occur between adjacent perineurial cells. There no continuous circum-neural fat-body sheath.
A ten-fold change in the external potassium concentration results in a 43 mV change in the resting potential of de-sheathed connectives. Action potentials of such exposed axons are rapidly blocked in low-sodium or sodium-free saline and under these conditions neither calcium nor magnesium is able to maintain conduction. Spikes from de-sheathed preparations are rapidly abolished on exposure to 10−6 M tetrodotoxin. These findings indicate a conventional ionic basis of excitation for the axonal membrane of this insect.
Analyses of the haemolymph reveal a mean sodium concentration of 25·4(s.E. ± 0·98) mm/1 and a mean potassium concentration of 25-1(s.E. ± 1·74) HIM/1.
Action potentials recorded from sheathed connectives are maintained for extended periods in sodium-free saline. 5. Exposure of most sheathed connectives to elevated potassium concentrations results in a two-stage depolarization. A rapid, single-stage, apparently extraneuronal potential change is, however, observed in some preparations.
These results on sheathed connectives indicate the presence of some peripheral barrier to the movements of sodium and potassium; the tight junctions between adjacent perineurial cells are considered to be possible sites of this restriction.
INTRODUCTION
Current knowledge of the ionic basis of excitation in insect nerve derives largely from studies on the cockroach, Periplaneta americana (cf. Narahashi, 1963; Pichón, 1969).The axonal membrane of this insect resembles that of the squid axon in the possession of a predominantly potassium-dependent resting potential and a sodiumdependent action potential (Boistel & Coraboeuf, 1958; Yamasaki & Narahashi, 1959; Narahashi, 1963, 1965; Pichon, 1967; Pichon & Boistel, 1967a, b\Pichon, 1969). Analyses of the haemolymph of this omnivorous insect have revealed a relatively high mean concentration of sodium ions in relation to potassium ions (Tobias, 1948; Clark & Craig, 1953; van Asperen & van Esch, 1956; Treherne, 1961; Pichon, 1963, 1970). The work of Boné (1944, 1946) clearly demonstrated that this situation is not found in all insects. For example, sodium: potassium ratios of approximately unity have been observed in a number of phytophagous insects including the rather primitive exopterygote Carausius morosus (Cheleutoptera) and several species of the more advanced endopterygotes (Lepidotera, Hymenoptera and Coleoptera) (cf. Florkin, 1966). The question of the ionic basis of conduction processes in the neurones of these insects has, therefore, been raised (Hodgkin, 1951; Hoyle, 1952, 1953).
Potassium-dependence of the resting potential and sodium-dependence of the action potential have been demonstrated by Treherne and Maddrell (1967b) for axons in de-sheathed nerve cords of Carausius. These findings appear anomalous in the fight of the haemolymph ionic composition observed in this insect (15±3mm-Na; 18 ± 3 mm-U; Wood, 1957) and led to the proposal of two different mechanisms for local regulation of the ionic composition of the extra-axonal fluid. Treherne & Maddrell (1967 a, b) considered that the possible sites for this regulation could be situated at different levels in the central nervous tissue, i.e. in the neural lamella (the outer acellular connective tissue sheath), in the perineurium (the peripheral layer of modified glial cells) or in the more deeply situated glial cells. Alternatively, Weidler & Dieke (1969, 1970) proposed that the fat-body sheath which encompasses the ventral nerve cord of this insect (Maddrell & Treherne, 1966) actively maintains an elevated sodium concentration in the fluid space around the nerve cord. Recent experiment militate against this second hypothesis for the following reasons.
The peripheral nervous system of Carausius is not surrounded by a fat-body sheath (Lane & Treherne, 1971). The mechanism proposed by Weidler & Diecke (1969, 1970) would only operate over part of the nervous system and would therefore necessitate two physiologically distinct processes of ionic regulation.
Quantitative considerations of ionic regulation reveal that inappropriately high rates of sodium pumping would be required to overcome the back diffusion of this ion through the open intercellular channels separating adjacent fat-body cells (Lane & Treherne, 1971).
Ionic regulation cannot be restored by re-investing a de-sheathed connective in intact lengths of fat-body sheath (Treherne, 1972).
The aim of the present investigation is to study the ionic aspects of nervous function in another phytophagous insect. The moth Manduca sexta has been chosen as a representative of the Lepidoptera, a phylogenetically advanced insect order, in contrast with the Cheleutoptera (e.g. Carausius), but in which the sodium level of the haemolymph is also often very low. For example, in adults of Sphinx ligustri (Lepidoptera) a value of 3·5 mm/1 sodium has been obtained (Pichon & Sattelle, unpublished observations). Following a study of the fine structure of the nervous tissue and an analysis of the blood ionic content, electrophysiological experiments have been carried out on isolated abdominal nerve cords in order to evaluate the role of major blood cations in excitability and to test the ability of this nervous tissue to regulate its ionic microenvironment. These results will be discussed in terms of structure-function relationships in the central nervous system of Manduca and the findings will be compared to those obtained for Periplaneta and Carausius.
MATERIAL AND METHODS
For fine-structural studies abdominal nerve cords from adult specimens of Manduca sexta were dissected out and fixed at room temperature in either Karnovsky’s fixative (1965), or 3% glutaraldehyde in 01 M cacodylate buffer, pH 7·4, plus 0·2 M sucrose (Sabatini, Bensch & Barmett, 1963). The tissues were subsequently washed in several rinses of cacodylate buffer with added sucrose, post-fixed in 1 % osmium tetroxide in MM cacodylate buffer, dehydrated through an ascending series of ethanols and embedded in Araldite. Sections 1−2 /zm thick were cut on an LKB ultrotome III and treated with a solution of 1 % toluidine blue in 1 % borax. Ultrathin sections were stained with uranyl acetate and lead citrate and examined in a Philips EM 300.
Analyses of the ionic composition of the haemolymph were performed on samples obtained from unanaesthetized adults of Manduca by puncturing the dorsal surface of the integument in the region of the second or third abdominal segment. The extruded haemolymph was drawn into a calibrated glass micropipette of 5−10 µA capacity. Samples thus obtained were discharged into 5 ml of distilled water. Determinations of the sodium and potassium concentrations in the diluted haemolymph were carried out using a Pye Unicam SP 90 Series 2 Atomic Absorption Spectrophotometer. The standard solutions consisted of either sodium chloride or potassium chloride at conentrerions close to those normally occurring in the haemolymph of adult Lepidoptera (cf. Florkin, 1966).
For electrophysiological experiments, the abdominal portion of the ventral nerve cord was dissected from unanaesthetized adults of Manduca and transferred into a drop of physiological saline on a glass slide. This preparation comprised the last four abdominal ganglia and their connectives. As originally described in the wax moth Galleria melonella (Ashhurst & Richards, 1964a), the nerve sheath of Manduca includes an extensive dorsal mass of connective tissue (cf. Results, Section A). This portion was removed by microdissection under a binocular microscope using fine stainless-steel needles, care being taken first not to stretch the preparation and secondly not to puncture the neural lamella surrounding the connectives. This procedure was necessary to provide satisfactory insulation between the recording electrodes (cf. recording technique). Such preparations are hereafter referred to as sheathed preparations.
De-sheathing was performed, when necessary, using the technique employed for the cockroach nerve cord (Boistel & Coraboeuf, 1954). Electron-microscopical studies on Manduca have revealed that in such de-sheathed preparations not only was the neural lamella completely removed, but also damage was suffered by the underlying perineurial cells (Lane, 1971). Peripheral glial and axonal elements were also likely to be damaged by the de-sheathing procedure but the bulk of the axons in de-sheathed preparations conducted impulses for several hours. The normal physiological solution employed consisted of NaCl 150 mm/1; KC1 3 mm/1; CaCl2 3 mm/1; phosphate buffer to pH 6·6−6·8. This was based on an optimum saline devised by Narahashi (1963) capable of maintaining spontaneous activity of the ventral nerve cord of the silkworm for long periods.
Electrical potential changes were recorded using the sucrose-gap technique (Stamplfi, 1954; Pichon & Treherne, 1970) since the small diameters of axons precluded an investigation employing conventional intracellular micro-electrode methods. The preparation was stimulated by depolarizing pulses of short duration delivered by a Famell pulse-generating system via an RF stimulus-isolation unit. Membrane potentials were led through a high-impedance, unity-gain amplifier to a Tektronix 561 oscilloscope. Action potentials were filmed using a Nihon-Kohden oscilloscope camera. Slow potential changes were monitored continuously on a Smith Servoscribe potentiometric recorder.
RESULTS
A. Ultrastructure
The abdominal nerve cord of the adult form of Manduca is encased by an acellular connective tissue sheath. This sheath or neural lamella measures about 2−8 µm. in width on the ventral and lateral surfaces, but on the dorsal surface it is extended into a large mass of tissue to which are connected the muscular bundles from the ventral diaphragm (Pl. 1). This enlarged dorsal neural lamella projects about 350 µm beyond the nervous tissue; its development in pupal and adult Lepidoptera has been described previously (Ashhurst & Richards, 1964a). The dorsal mass of connective tissue consists of fibroblast cells and trecheoles embedded in a matrix of collagen-like fibres and amorphous material. In the wax moth Galleria, the matrix has been shown by histochemical methods to consist of neutral and acid mucopolysaccharides, as well as some lipid (Ashhurst and Richards, 1964b). In Manduca, clumps of fat-body cells are found lying near the neural lamella (Pl. 1) but these are not arranged as a continuous sheath. The lamellar layer on the ventral and lateral surfaces consists of many collagen-like fibrils embedded in an amorphous matrix (Pl. 2).
Beneath the neural lamella lies a cellular layer called the perineurium; as in other insects its constituent cells appear to be modified glial cells. They contain a number of mitochondria, lipid globules (Pl. 2) and large aggregations of microtubules, particularly in the region of the lateral cell borders (Pl. 2 and Pl. 3, fig. 1). These microtubules appear to be oriented parallel to the longitudinal axis of the nerve cord, since they are usually seen in cross-section when transverse sections of the nerve cord are examined. At the neural lamellar surface, adjacent perineurial cells are separated laterally by large extracellular spaces of varying size and shape (Pl. 2). These spaces contain an electron-dense substance similar in appearance to the ground substance of the neural lamella; in some cases fibrils of collagen are also present (Pl. 2). This apparent intrusion of the neural lamella into the perineurium terminates at the lower lateral surface of the perineurial cells, at which point the cells are linked by tight junctions (Pl. 3, fig. 1). Hemi-desmosomes occur at intervals on those surfaces of the lateral perineurial plasma membrane which bound the enlarged extracellular spaces containing connective tissue material (Pl. 2); microtubules tend to be concentrated near the junctions. Herni-desmosomes also occur on the peripheral surface of the perineurial cells which lies adjacent to the neural lamella (Pl. 3, fig. 1). These are fairly numerous and presumably maintain the structural integrity of the cells with respect to the connective tissue matrix.
Beneath the perineurium lie the glial cells ensheathing the nerve-cell bodies and axons which comprise the ganglia and connectives. The glia are attenuated cells with elongated nuclei; they contain massive arrays of microtubules (Pl. 3, fig. 1) aligned, as in the perineurium, along the longitudinal axis of the nerve cord. The microtubules are often associated with desmosomes (Pl. 3, fig. 2) which are present in large numbers, and form junctions between adjacent glial cells (Pl. 2 and Pl. 3, fig. 2). It seems probable that these desmosomes ensure that no changes occur in the spatial relations between cells during the sinuous flexions of the nerve cord; such movements have been observed in all preparations prior to dissecting away the muscle attachments to the lateral body wall.
The precise relationship of the microtubules to these desmosomes has been examined in the wax moth Galleria (Ashhurst, 1970), and in Manduca the tubules also appear to lie parallel to the plaque of dense material which forms the desmosome. In addition, tight junctions occur between adjacent glial cells (Pl. 3, fig. 2). One to three folds of a glial cell surround each axon, but the glial plasma membrane need not necessarily lie in intimate association with the axolemma; sometimes gaps occur (Pl. 3, fig. 2). In some preparations there is considerable extracellular space between adjacent glial cells in the connective (as in Pl. 3, fig. 2). This space is not, as in the cockroach and stick insect (Smith & Treherne, 1963), full of an electron-opaque material. In other preparations, however, the intercellular spaces between glia are negligible. It is not clear whether these differences represent different physiological states or whether the spaces are a fixation artifact. The axons themselves contain mitochondria and neurotubules which are usually oriented along the longitudinal axis of the connective (Pl. 3, fig. 2). Most axons in the nerve cord of Manduca are small or of medium size (Pl. 2 and Pl. 3, fig. 2) in contrast with the giant axons of the cockroach.
B. Ionic composition of haemolymph
The results of determinations of sodium and potassium concentrations together with calculated sodium: potassium ratios for samples of haemolymph from six adults of Manduca sexta are summarized in Table 1. The mean concentration of sodium was 25·4 (S.E. ±0-98) mm/1 and that of potassium 25·1 (S.E. ± 1·74) MM/1. In both cases little variation in ionic concentration was observed either between individual animals or between successive samples from the same animal. The mean sodium: potassium ratio of haemolymph was 1·01 (S.E. ± 0·23).
C. Electrophysiological study of the isolated abdominal nerve cord
It is clear that in Manduca the same paradoxical situation occurs as is observed for Carausius in that the classical ionic theory for excitation in nerve (Hodgkin, 1951) is difficult to reconcile with the ionic composition of the haemolymph. The same tentative explanations proposed for Periplaneta (Pichon, 1969) and Carausius (Treherne & Maddrell, 1967b) are considered for Manduca:
The mechanism of excitation is unconventional.
The ionic micro-environment of the nervous elements differs from the blood and is maintained by passive or active mechanisms.
Some haemolymph ions may be bound to blood proteins or sequestered within the haemocytes, thus reducing the effective concentration.
It is technically very difficult to test this last possibility; furthermore, in insects where it has been investigated, its relative importance has been found to be limited (cf. Carrington & Tenney, 1959; Pichon, 1969). Our study has therefore been confined to the two other mechanisms. The purpose of the first set of experiments has been to investigate the ionic basis of membrane-potential production. They have been carried out on de-sheathed connectives using the sucrose-gap technique.
(a) Ionic basis of membrane-potential production
Resting potential*
(1) Effects of potassium ions
Elevating the potassium concentration of the bathing medium surrounding the de-sheathed connective from 3 to 100 mm/1 (potassium replaced sodium to maintain isotonicity) resulted in a rapid depolarization accompanied by a conduction block (Text-fig. 1). Five desheathed connectives were exposed to various external potassium concentrations over the range 3-300 mm/1. The sodium concentration of the saline was maintained constant at 3 mm/1 throughout, and for low potassium concentrations tris chloride was used to maintain isotonicity. For potassium concentrations above 150 mm/1 the potassium chloride concentration was simply increased. The results have been summarized in Text-fig. 2. Over the straight-line section of the graph a ten-fold change in the potassium concentration corresponded to a 43 mV change in the recorded potential. This figure differed appreciably from the predicted 58 mV potassium slope for a single membrane behaving as an ideal potassium electrode.
(2) Effects of sodium ions
To test whether or not deviations from the ideal situation were in any way attributable to a sodium contribution to the overall resting potential, the external concentration of this ion was varied between 3 and 300 mm/1 and changes in the resting potential were observed. The results which are summarized in Textfig. 3 indicate that the resting sodium permeability is effectively relatively large. Over the linear section of the graph, a tenfold change in sodium concentration corresponded to a 15 mV change in the resting potential. The possible contribution of other ions to the resting potential was not assessed.
Action potential
(1) Effects of sodium ions
The exposure of de-sheathed connectives to either sodium-free (tris) saline or saline containing only 3 mm/1 sodium resulted in a rapid block of the action potential associated with hyperpolarization of 8–20 mV (Text-fig. 4). Both effects were completely reversible upon re-exposure of the preparation to normal saline. Sodium ions are therefore important for normal action-potential production.
(2) Effects of calcium and magnesium ions
Another set of experiments was performed to test whether or not these two divalent cations were also involved in action-potential production. These revealed that both ions were unable to carry a detectable amount of the action current. Text-fig. 6 shows that removal of most of the sodium (147 mm/1) from the original saline (150 mm/1) in the presence of 50 mm/1 magnesium was followed by a hyperpolarization and a conduction block. These effects could not be recovered when magnesium ions were replaced by calcium ions. Return to the original saline (which included 150 mm/1 sodium and 50 mm/1 magnesium) restored the original resting potential and the action potential.
(3) Effects of tetrodotoxin
The puffer fish poison tetrodotoxin (TTX) has been shown to block selectively the ‘sodium channels’ in the giant axon of the squid (Narahashi, Moore & Scott, 1964; Nakamura, Nakajima & Grundfest, 1965 and others), in the node of Ranvier of the frog (Hille, 1966) and in the giant axon of the cockroach (Pichon, 1969). Text-fig. 6 shows that this poison is also effective in blocking the conduction of the compound action potential in de-sheathed connectives of Manduca sexta.
(b) Regulation in sheathed preparations
The ionic basis for resting potential and action potential does appear to be conventional as in the two other insect species previously studied. This second set of experiments has therefore been performed to check the function of the nerve sheath and associated tissues in the regulation of ionic movements between the blood and the extra-axonal fluid. In these experiments, the effects of changes in concentration of both potassium and sodium on the polarization of the connective and on axonal conduction have been tested simultaneously on sheathed preparations.
(1) Effects of a low-sodium, high-potassium saline
The effects obtained were somewhat variable between different preparations. In most cases, however, replacement of the normal low-potassium, high-sodium saline by a solution containing 150 mm/1 potassium and 3 mm/1 sodium was followed by a two-stage depolarization, the initial rapid phase being followed by a much slower phase(Text-fig. 7). Long exposures were characterized by a slow reduction in the size of the action potential. Re-exposure to the normal saline was followed by a slow recovery of both the resting and action potential. In a limited number of preparations the slow potential change did not appear and recovery took place much more rapidly following re-exposure to the normal saline. In these preparations the depolarization was not associated with the reduction of the size of the conducted action potential(Text-fig. 8).
(2) Effects of variations in the external potassium concentration.*
The potassium-dependence of the D.C. potential changes in the two categories of preparations, distinguished by their response to a high-potassium saline, is illustrated in Text-figs. 9 and 10. A 35·5 mV slope for a ten-fold change of the potassium concentration was obtained for the preparations of the first type described in the preceding paragraph after a 15 min exposure to the test solution. This slope was only 11 mV for the other preparations.
(3) Effects of low-sodium saline
Exposure of sheathed connectives to low sodium(3 mm/1) saline, in which tris replaced sodium, resulted in a rapid and reversible hyperpolarization(10·20 mV) without any associated change in the action potential(Text-fig. 11).
(4) Effects of sodium-free saline
As in the experiments with low-sodium saline, replacement of all the sodium ions by tris ions led to a hyperpolarization. A 30 minexposure to this solution did not result in a decline of the recorded action potential(Text-fig. 12).
(5) Effects of tetrodotoxin
Tetrodotoxin applied at 10−6 M and 3 × 10−6 M in normal saline was ineffective in blocking the conducted spike even after 15 min(Text-fig. 13).
These preceding experiments on the sheathed nerve cords of M. sexta showed that The nerve elements are protected by a highly effective barrier from changes in the concentration of ions and the levels of pharmacological agents in the bathing medium.
DISCUSSION
The ionic basis of the resting potential recorded from the connectives of M. sexta resembles in some respects that of the giant axons of Periplaneta americana(Yamasaki & Narahashi, 1959) and of cell bodies of Carausius morosus(Treherne & Maddrell, 1967b). For example, when the potassium concentration of the bathing medium is varied, independently of sodium, the exponential slope obtained on plotting potassium concentration against resting potential shows a 43 mV change for a ten-fold change in concentration. This value closely resembles the 42 mV slope obtained for the giant axons of the cockroach(Yamasaki & Narahashi, 1959) and is not dissimilar to the 37 mV slope recorded from cell bodies of de-sheathed mesothoracic ganglia of the stick insect(Treherne & Maddrell, 1967b). Deviations from the theoretical 58 mV slope for a membrane behaving as an ideal potassium electrode can be attributed to a short-circuiting effect in the sucrose-gap arrangement(cf. footnote p. 722). It seems more likely, however, that, as in other insect species, this deviation is due to the fact that the resting potential is not solely determined by the potassium equilibrium potential, but that conductances to other ions contribute to its generation. In Manduca, the resting membrane conductance to sodium ions is relatively large(15 mV slope for a ten-fold change in external sodium) and can be considered responsible for the deviation.
De-sheathed nerve cords of Manduca develop a rapid and reversible conduction block in solutions of very low sodium concentration or in the absence of external sodium ions. This suggests that sodium is essential for the production of normal action potentials. The fact that neither calcium nor magnesium ions is capable of maintaining nerve conduction in the absence of sodium shows that the so-called ‘sodium channel’ is highly selective to sodium. Experiments with 10−6 M tetrodotoxin demonstrate that these channels are pharmacologically similar to those of the squid and cockroach axons. There appears, therefore, to exist a strong similarity in the ionic mechanism for the production of resting and action potentials in nerve between M. sexta and other species including Periplaneta americana and Carausius morosus.
This similarity of the axonal membrane properties provides a striking contrast with the differences in ionic composition of the haemolymph of these insects. In Manduca, for example, a sodium concentration of 25·4(s.E. ± 0·98) mm/1 and a potassium concentration of 25·1(s.E. ± 1·74) HIM/1 has been found. The sodium: potassium ratio of 1·01(s.E. ± 0·23) resembles the ratio of 0·83 for Carausius(calculated from the data of Wood, 1957) and contrasts strongly with the ratios of between 6·2(Tobias, 1948) and 19·6(van Asperen & van Esch, 1953) obtained for Periplaneta. Clearly some form of control of the ionic micro-environment within the central nervous tissues must be proposed for Manduca as has been done for Carausius and Periplaneta(cf. Treherne & Pichon, 1972).
Possible structural barriers to the free movements of ions through central nervous tissues of insects can be listed as follows: extra-neural structures, such as the fat-body sheath; the neural lamella; the perineurium; the glial cells; and the extracellular system(cf. Treherne & Moreton, 1970). In Manduca the ventral nerve cord and peripheral nerves are not surrounded by a fat-body sheath, although isolated patches of fat-body tissue occasionally occur in close proximity to the ventral nerve cord(cf. Text-fig. 1). The neural lamella, although 2−8 µm in width on the ventral and lateral surfaces, is expanded dorsally into a large mass of tissue. However, the neural lamella is not generally regarded as a barrier to the diffusion of ions(cf. Smith & Treherne, 1963) and although exogenous peroxidase penetration is greatly retarded by the expanded dorsal mass, it does penetrate the neural lamella on the ventral and lateral surfaces fairly rapidly(Lane, 1971).
The perineurium in Manduca does not appear specialized for active transport, but the tight junctions observed between lateral walls near the inner margins of this cell layer may constitute a barrier to the intercellular exchange of ions. Experiments using the exogenous tracer substance peroxidase show that in sheathed preparations restriction to penetration of this molecule occurs at the perineurial junctions(Lane, 1971). These tight junctions at the base of the perineurium are also found in both Periplaneta and Carausius(Maddrell & Treherne, 1967), where they also proved to be the sites of restriction to entry of tracer substances such as peroxidase(M.W. 40000)(Lane & Treherne, 1970) and microperoxidase(M.W. 2000)(Lane, unpublished observations). By comparison with other systems(Brightman & Reese, 1969; Bennett & Trinkaus, 1970) these results suggest that the junctions here are true tight junctions occluding the interspaces between the perineurium and the underlying extra-neuronal spaces. In the channels between adjacent perineurial cells, peripheral to the tight junctions, electron-dense material containing collagen-like fibres occurs. This material is very similar to the mucopolysaccharide occurring in the neural lamella and hence is unlikely to represent a barrier to ion movements but may constitute a potential cation reservoir.
The extensive investment of neuronal elements in the nerve cord of Manduca by glia represents a further possible source of impedance to the movements of ions by increasing the length of the extracellular pathway. Junctional connections between glial cells, however, may represent a short-circuit transport route from the periphery of the central nervous tissue to the neuronal surface. The tight junctions between adjacent glial cells may in fact be ‘gap’ junctions, which could represent sites of low resistance pathways between the cytoplasm of adjoining cells, possibly permitting exchanges of ions and molecules between the interiors of such coupled cells. If this is so, then the glial cells, as in the cockroach(Lane & Treherne, 1970), may form an additional faster diffusion pathway for movement of ions and molecules from the bathing medium to the axonal surfaces. This pathway would only be functional under certain conditions, as, for example, after the disruption of the neural lamella and perineurium by de-sheathing. Preliminary experiments with peroxidase support this speculation as in de-sheathed preparations of Manduca this exogenous protein is taken up by the glial cytoplasm(Lane, 1971), as it is in cockroach connectives.
Finally, the extracellular system, though tortuous due to the extensive glial investment of neurones, does not contain an electron-dense material such as that observed in the nervous system of Periplaneta and Carausius and considered to be an acid mucopolysaccharide(Ashhurst, 1968). Hence, whatever the nature of the non-opaque substance present in these spaces, it is unlikely to be similar to the dense mucopolysaccharide material found in those of the cockroach and stick insect, which may re-pesent cation reservoirs. From these findings it is clear that ultrastructural studies alone e unlikely to lead to the location of the control mechanism regulating the ionic micro-environment. Such studies have, however, eliminated, in this species, the fatbody and the neural lamella as major sites of regulation and concentrated attention on the region of the perineurium.
Electrophysiological experiments show that the movements of both potassium and sodium ions to and from the central nervous tissues are considerably slower in sheathed than in de-sheathed preparations. Exposure of sheathed preparations of Manduca to high-potassium and/or sodium-free solutions gives rise to potential changes which by analogy with the results obtained for Periplaneta(Treherne et al. 1970) are very likely to be extraneuronal in nature. It has been suggested that such extraneuronal potential changes could reflect ionic effects on the outwardly directed perineurial membranes, access to the inner surfaces being reduced by the presence of tight junctions between adjacent perineurial cells(Pichon, Moretón & Treherne, 1971). This would account for the observation, in a limited number of preparations of Manduca, that rapid extraneuronal potential changes in response to elevation of the external potassium concentration are not associated with changes in the action potential. An 11 mV potential change of this presumed extraneuronal component accompanies a ten-fold change in the potassium concentration of the bathing medium surrounding sheathed connectives. This figure is similar to the 17 mV slope observed in intact Periplaneta connectives(Pichon et al. 1971). The 35-5 mV potassium slope most frequently observed in sheathed preparations of Manduca presumably results from a combination of neuronal and perineurial depolarizations. As demonstrated in Periplaneta(Treherne, et al. 1970; Pichon & Treherne, 1970) even slight stretching or drying of the preparation during isolation of sheathed connectives could account for such an increase in the accessibility of the axonal surface to ions in the bathing medium.
The presence in Manduca of an apparent peripheral restriction to intercellular diffusion of small water-soluble ions by the perineurial tight junctions would obviously facilitate extra-axonal regulation. Two mechanisms have been proposed which could lead to elevated extra-axonal sodium levels. It has been suggested, for example, that anion groups associated with extracellular acid mucopolysaccharides could maintain a sodium-reservoir in close proximity to the axon surfaces(Treherne, 1962). However, in the absence of active processes, the thermodynamic activity of cations associated with such an anion matrix would be equivalent to that of the external medium and would not produce an effective elevation of the extra-axonal sodium level(Treherne, 1967). Furthermore, in Manduca, the absence of electron-dense material from the extracellular spaces around the axons has been demonstrated. The alternative hypothesis attempts to relate extra-axonal sodium regulation to the activity of glial elements(Treherne, 1967; Treherne & Maddrell, 1967b; Pichon, 1969; Pichon & Boistel, 1967 a). The results presented here for Manduca sexta showing prolonged axonal function in sodium-deficient media apparently accord with the latter hypothesis.
ACKNOWLEDGEMENTS
The authors thank Dr J. E. Treherne, for helpful conversations during the course of this work and for comments on the manuscript. The assistance with photography provided by Miss Y. Carter is gratefully acknowledged.
REFERENCES
EXPLANATION OF PLATES
PLATE 1
Light micrograph of a cross-section through the abdominal nerve cord of Manduca sexta showing the paired connectives(C) surrounded by the neural lamella(NL) which is extended dorsally into a mass of connective tissue(DM). Fibroblasts occur within this mass, and muscle bundles(arrows) connect it to the body wall. Clumps of fat body cells(FB) lie nearby. The rectangle indicates the area shown in Pl. 2. X250.
PLATE 2
Electron micrograph through the periphery of a transversely sectioned connective from the ventral nerve cord of Manduca as indicated by the rectangle drawn on Pl. 1. This region is near the latero-dorsal edge where the neural lamella(NL) begins to become extended into the dorsal mass. The perineurium(PN) possesses many microtubules and lipid globules(L); between adjacent cells lie extracellular spaces(ES) containing dense material. Desmosomes(arrows) attach these perineurial cells to the underlying glia(G), whose microtubule-laden(MT) processes ensheath the axons(A). T, trecheoles, x 13300.
PLATE 3
Fig. 1. Section through the perineurium showing the tight junctions(arrows) formed between the interdigitating processes near the bases of two component cells. Note the hemi-desmosomes(HD) between perineurium(PN) and neural lamella(NL). MT, microtubules. Insert shows part of some perineurial tight junctions at higher magnification, x 45 100; insert, x 89800.
Fig. 2. Cross-section through axons, containing mitochondria and neurotubules(NT), which are ensheathed by glial cells. Note the desmosomes(D) between adjacent glial processes as well as the tight junctions(arrows) adjoining them at intervals. MT, microtubules of glial cells; *, extracellular space. Insert shows tight junctions between adjacent glial cells at higher magnification, x45 100; insert, x 96450.
Note added in proof
Since this paper went to press more specific cytochemical reports have occurred in the literature to indicate that the dorsal mass of the moth Galleria(see p. 719) contains chondroitin, dermatan sulphates and neutral glycoproteins and that the electron-dense material in the extracellular spaces of Periplaneta(see page 730) represents hyaluronic acid(Ashhurst, D. E. and Costin, N. M. 1971, Histochem. J. 3, 297−310; 379−387).
It must be pointed out that the sucrose-gap technique as applied to insect nerve is not so accurate as for the de-sheathed frog nerve for which it was originally devised (cf. Stamplfi, 1954; Straub, 1956),due to differences in structure of the two preparations. In the present experiments a relatively large proportion of the ‘nerve’ is composed of sheath, glia and extracellular spaces so that the potential changes can be considered neither as absolute values nor as purely axonal.
Sodium chloride was used to maintain isotonicity.