Biochemical studies of trophic dependencies in crayfish giant axons

Medial giant (MGA) and lateral giant (LGA) axons of crayfish were doubly cut in order to selectively isolate axonal segments from perikaryal and transsynaptic sources of trophic input. Isolated MGA segments remained morphologically intact for over 43 days, whereas isolated LGA segments usually degenerated within one week. The glial sheaths around isolated MGA segments had significantly increased in thickness within one week, but severed LGA segments showed no increase in sheath thickness at any time after lesioning. These data suggest that cells of the surrounding glial sheath can provide trophic support to isolated MGA segments but not to isolated LGA segments. Extent of glial hypertrophy seems dependent upon specific spatiotemporal parameters.The diameters of isolated MGA segments decreased more rapidly than the diameters of singly cut MGA segments. These data suggest that the MGA also receives some trophic support from pre- or postsynaptic sources. Conversely, some singly cut LGA segments completely degenerated within one week, whereas other singly cut LGA segments remained intact for at least 43 days after lesioning. Such results suggest that the LGA receives a significant trophic input from pre- or postsynaptic structures.     

Histological studies of trophic dependencies in crayfish giant axons

Medial giant (MGA) and lateral giant (LGA) axons of crayfish were doubly cut in order to selectively isolate axonal segments from perikaryal and transsynaptic sources of trophic input. Isolated MGA segments remained morphologically intact for over 43 days, whereas isolated LGA segments usually degenerated within one week. The glial sheaths around isolated MGA segments had significantly increased in thickness within one week, but severed LGA segments showed no increase in sheath thickness at any time after lesioning. These data suggest that cells of the surrounding glial sheath can provide trophic support to isolated MGA segments but not to isolated LGA segments. Extent of glial hypertrophy seems dependent upon specific spatiotemporal parameters.The diameters of isolated MGA segments decreased more rapidly than the diameters of singly cut MGA segments. These data suggest that the MGA also receives some trophic support from pre- or postsynaptic sources. Conversely, some singly cut LGA segments completely degenerated within one week, whereas other singly cut LGA segments remained intact for at least 43 days after lesioning. Such results suggest that the LGA receives a significant trophic input from pre- or postsynaptic structures.     

Lack of glia-axon transfer of proteins in the normal optic system of goldfish

An ultrastructural autoradiographic study of the goldfish optic tectum was carried out to determine whether proteins synthesized in glial cells are transferred into adjacent optic axons. Goldfish were injected intracranially over the optic tecta with [3H]leucine and fixed by perfusion 30 min, 4 and 24 h later. All unincorporated precursors were removed by repeated washings with fixatives, and tissue slices from the optic tecta were embedded and processed for electrom microscopy autoradiography (EMA). The densities of the silver grains were determined over optic axons and their myelin sheaths. The densities over the axons were lower than those over the myelin sheaths at all time intervals. The density of the intraaxonal grains, in absolute terms as well as relative to that of the myelin, was highest in the 4 h experiment. Analysis of the distribution of the grain densities over the myelin sheath and over concentric axonal compartments was carried out at this time interval to determine whether the grain density over the axon represented intraaxonal [3H]proteins or was only the result of grains 'scattered' from [3H]proteins located in the surrounding myelin sheaths. When the experimental distribution of the grain densities over the axon was compared with the theoretical distribution expected over the axon if only the myelin sheaths were labeled, no significant difference was found. This indicates that the silver grains present in the axons were scattered from the adjacent myelin sheath and did not represent intraaxonal radioactivity. It is therefore concluded that in our system there is not a quantitatively significant transfer of proteins between glial cells and adjacent axons.     

Glial polypeptides transferred into the squid giant axon

The proteins synthesized by the glial sheath of an isolated segment of squid giant axon and by the cell bodies of the giant axon in the isolated stellate ganglion were labeled by incubation in the presence of [3H]leucine. The axoplasm, which contained labeled proteins transferred from the glial sheath, was separated from the sheath by mechanical extrusion. The labeled proteins in the axoplasm, the empty sheath and the stellate ganglion were analyzed and compared by one- and two-dimensional polyacrylamide gel electrophoresis. Over 80 glial polypeptides were found to be selectively transferred into the axoplasm and many of these were distinct from stellate ganglion polypeptides which presumably could be supplied to the axon via axonal transport. Three of the more highly labeled transferred glial polypeptides (TGPs) were actin, a fodrin-like polypeptide and a polypeptide we have named traversin. Our observations, considered in the context of other reports, suggest that the squid axon receives a large number of polypeptides from its surrounding glia either by phagocytozing glial cell process that project into it or via cytoplasmic channels between adaxonal glia and the axon. These TGPs may help the axon survive unfavorable conditions.     

A glial-neuronal-glial communication system in the mammalian central nervous system

Previous studies have demonstrated that when tritiated proline [( 3H]Pro) is injected into the dorsal column nuclei (DCN) of cats, it labels macroglial cells, but fails to label neurons at the injection site. (Tritiated leucine [( 3H]Leu) in contrast, labels both neurons and some glial cells.) Despite the failure of [3H]Pro to label DCN neurons, labeling is still observed in DCN terminal targets. This result suggests that glial cells are involved in the translocation of [3H]Pro-labeled molecules from one part of the brain to another. The purpose of the present experiment was to use electron microscopic autoradiographic techniques to characterize the labeling produced in internal arcuate fiber tract axons arising from DCN neurons 24 h after injections of [3H]Pro (or [3H]Leu, for comparison) into DCN. It was reasoned that, if the translocation of [3H]Pro-labeled molecules from DCN to its targets is indeed carried out by glial cells, then only glial elements associated with the fibers should be labeled following [3H]Pro injections of DCN. If, on the other hand, the translocation involves an initial transfer of [3H]Pro-labeled molecules into neuronal perikarya followed by axonal transport, then only axoplasmic elements along the fiber pathway should be labeled. Injections of [3H]Pro into DCN labeled axoplasmic elements in samples of axons from the internal arcuate tract both 'near' (0.5-0.8 mm) and 'far' (2-4 mm) from the injection site at about an equal absolute density. However, glial elements associated with the axons were also labeled in both samples, but much more densely in the 'near' than in the 'far' axons. Injections of [3H]Leu labeled axoplasm more densely than did [3H]Pro (by a factor of 4 in the 'far' samples). Glial labeling by [3H]Leu near the injection site was much less than that of [3H]Pro, but, 'far' from the injection, the levels of [3H]Leu and [3H]Pro glial labeling were comparable. Taken together with the results of other studies, these data support the existence of a previously unrecognized system of communication between glial cells and neurons. In this putative system (Fig. 9), molecules containing both [3H]Leu and [3H]Pro are transferred from glial cells into adjacent neuronal soma and transported down the length of the axon where, all along the way, some of them are transferred from the axon into adjacent glial processes. The system is more readily apparent when [3H]Pro is used because of its avid and preferential uptake by glial cells. Potential functions of such a system are unknown, but could be trophic, protective and/or informative.     

Protein metabolism in transected peripheral nerves of the crayfish.

The significance of the protein metabolism in crayfish peripheral nerve was studied in relation the ability of crayfish motor axons to survive for over 200 days following axotomy. In contrast to frog peripheral nerves, the crayfish nerves appear to more closely resemble ganglia in their profiles of synthesis expressed on sodium dodecyl sulfate (SDS) gels, and have higher incorporation rates of [3H]leucine into protein than ganglia. Since anisomycin inhibits over 95% of protein synthesis in crayfish peripheral nerve, it was concluded that this local protein synthesis was dependent upon a eukaryotic ribosomal mechanism. Radioautography of isolated nerves reveals newly synthesized proteins in glial sheaths, and also within the axoplasm of large motor fibers. Based upon the data available at present, a hypothesis that the glia surrounding the axons are responsible for the local protein synthesis, and that some of these newly synthesized proteins are transported into the axon, is presented. Transection of crayfish peripheral nerves proximal to the neuron cell bodies produced a more than two-fold increase in [3H]leucine incorporation, but no significant changes in labeling profiles of the proteins on SDS gels. The data suggest that while an active local protein synthesis may be necessary for the maintenance of several crayfish motor axons, it is not a sufficient condition.     

Distribution of ultrastructural tracers in crustacean axons

Ruthenium red and horseradish peroxidase were used to compare the uptake of exogenous molecules into crayfish motor axons and their sheaths in severed and intact peripheral nerves. Both tracers penetrated the axonal sheath and were subsequently seen lining small vesicles and tubules in the axoplasm. Tracer appeared to enter the axon via pinocytotic vesicles. There were no perceptible quantitative or qualitative differences in ruthenium red uptake between intact and severed axons. However, counts of tracer-filled vesicles in axons exposed to peroxidase showed that at least three times as much tracer penetrated the severed as opposed to the intact axons.     

Heat-shock proteins in axoplasm: High constitutive levels and transfer of inducible isoforms from glia

To characterize heat-shock proteins (HSPs) of the 70-kDa family in the crayfish medial giant axon (MGA), we analyzed axoplasmic proteins separately from proteins of the glial sheath. Several different molecular weight isoforms of constitutive HSP 70s that were detected on immunoblots were approximately 1–3% of the total protein in the axoplasm of MGAs. To investigate inducible HSPs, MGAs were heat shocked in vitro or in vivo, then the axon was bathed in radiolabeled amino acid for 4 hours. After either heat-shock treatment, protein synthesis in the glial sheath was decreased compared with that of control axons, and newly synthesized proteins of 72 kDa, 84 kDa, and 87 kDa appeared in both the axoplasm and the sheath. Because these radiolabeled proteins were present in MGAs only after heat-shock treatments, we interpreted the newly synthesized proteins of 72 kDa, 84 kDa, and 87 kDa to be inducible HSPs. Furthermore, the 72-kDa radiolabeled band in heat-shocked axoplasm and glial sheath samples comigrated with a band possessing HSP 70 immunoreactivity. The amount of heat-induced proteins in axoplasm samples was greater after a 2-hour heat shock than after a 1-hour heat shock. These data indicate that MGA axoplasm contains relatively high levels of constitutive HSP 70s and that, after heat shock, MGA axoplasm obtains inducible HSPs of 72 kDa, 84 kDa, and 87 kDa from the glial sheath. These constitutive and inducible HSPs may help MGAs maintain essential structures and functions following acute heat shock. J. Comp. Neurol. 396:1–11, 1998. © 1998 Wiley-Liss, Inc.     

Maintenance and synthesis of proteins for an anucleate axon

The anucleate (distal) segment of a crayfish medial giant axon (MGA) remains intact for months in vivo after severing the axon from its cell body, a phenomenon referred to as long-term survival (LTS). We collected axoplasm from chronic anucleate MGAs by perfusing 2-cm lengths of axons with an intracellular saline. This axoperfusate was analyzed by SDS-PAGE and silver stained. Axoperfusate proteins from intact MGAs and from chronic anucleate MGAs exhibiting LTS for up to 6 months were the same. Furthermore, immunoreactive levels of actin and β-tubulin were similar in axoperfusates from intact and chronic anucleate MGAs. This maintenance of proteins in chronic anucleate MGAs must be due to a lack of protein degradation and/or to local protein synthesis by a source other than the cell body. To investigate local protein synthesis in vitro, we added [³⁵S]-methionine to the extracellular saline surrounding intact and chronic anucleate MGAs. After 4- to 6-h incubations, radiolabelled proteins were detected in axoperfusates analyzed by SDS-PAGE and fluorography. The similarity between radiolabelled proteins in axoperfusates and MGA glial sheaths indicated a glial origin for the radiolabelled axoperfusate proteins. Various observations and control experiments suggested that glial-axonal protein transfer occurred by a physiological process. Glial-axonal protein transfer may contribute to the maintenance of proteins during LTS of chronic anucleate MGAs.     

Characterization of the distinctive neurofilament subunits of the soma and axon initial segments in the squid stellate ganglion

The stellate ganglion, which gives rise to the giant axons of the squid, was dissected into two parts, one containing primarily cell bodies and the other axon initial segments. A neurofilament protein-enriched extract of each was prepared and compared biochemically and immunochemically with an axoplasmic neurofilament preparation and with the glial sheath that surrounds the axons. Both parts of the ganglion lacked the 220 kDa subunit of axoplasmic neurofilaments (NFs). However, they did contain a protein of about 190 kDa that reacted with the Pruss anti-intermediate filament antibody (aIFA; Pruss et al.: Cell 27:419-428, 1981), but not with a phosphorylation-dependent NF antibody (Cohen et al.: J Neurosci 7: 2056-2074, 1987). Dephosphorylation of the axoplasmic NF220 yielded a product that comigrated on two-dimensional (2D) gel electrophoresis with the 190 kDa ganglion protein, suggesting that the latter represented the incompletely phosphorylated precursor of NF220. The major low molecular weight aIFA-reactive species in the ganglion preparations was a polypeptide of about 65 kDa. A relatively small quantity of that polypeptide was also found in axoplasm and it comigrated in 2D gels with an aIFA-reactive polypeptide from the glial sheath. These results indicate that the site of modification of the 190 kDa NF precursor to the 220 kDa axonal form is probably at the point where the axon initial segments leave the ganglion, which is several mm distal to its site of synthesis in the cell body. Furthermore, the filament network of the axoplasm and possibly the cell bodies includes a glial-like intermediate filament protein in addition to the NF protein subunits.     

Heat shock-like protein is transferred from glia to axon

Glia-axon protein transfer was examined in the squid giant axon. Proteins synthesized by the glial sheath surrounding the axon were labeled with [³H]leucine. Raising the temperature of the incubation medium from 20 °C to 30 °C increased the synthesis of glial proteins that resembled heat-shock proteins. These proteins were among the group known to be transferred into the axon^12,18. Thus, glia provide the axon with proteins that may be involved in the reaction totrauma.     

Degeneration and regeneration of severed crayfish sensory fibers: an ultrastructural study

In crayfish peripheral nerve, terminal processes of most sensory fobers degenerated within two to three weeks of nerve severance. First, large clear vesicles appeared in the axoplasm which subsequently lost its substructure, became more dense and compact, and was finally reduced to dense inclusions within glial cytoplasm. Considerable traumatic degeneration was also observed in proximal segments of severed axons. The rate of fiber degeneration was generally more rapid in the smallest fibers and slower in the larger intermediate axons. Large numbers of regenerating axons were first seen crossing the lesion site after four weeks. Growing fibers, found in bundles wrapped by glial processes, were at first small and tightly packed with no glial processes separating them. Later fibers enlarged and acquired wrapping patterns typical of normal sensory fibers. Growth cones were observed during early stages of regeneration. Some secondary degeneration may occur.     

Glia-to-axon communication: Enrichment of glial proteins transferred to the squid giant axon

The transfer of newly synthesized proteins from the glial sheath into the axon is a well-documented process for the squid giant axon. In this study, we used a novel approach to separate the transferred glial proteins (TGPs) from the endogenous axoplasmic proteins of the squid giant axon. Axoplasm, containing radiolabelled TGPs, was extruded as a cylinder and immersed in an intracellular buffer. After 1–30 min, the TGPs were enriched in the intracellular buffer, because they were eluted from the axoplasm into the intracellular buffer much faster than the endogenous axoplasmic proteins. Most of the TGPs enriched in the intracellular buffer did not pellet when centrifuged at 24,000 g for 20 min and were susceptible to protease digestion without the addition of Triton X-100. Additionally, transmission electron microscopic autoradiography of intact axons, containing radiolabelled TGPs, suggested that most TGPs were not associated with vesicular organelles within the axon. We conclude that most of the TGPs are not contained within vesicles in the axoplasm of the squid giant axon, as would be expected if the mechanism of glia-to-axon transfer were conventional exocytosisendocytosis or microphagocytosis. © 1995 Wiley-Liss, Inc.     

Direction and specificity of the axonal and transcellular transport of nucleosides

The axoplasmic transport of nucleosides or their derivatives has been examined autoradiographically in avian and mammalian brains. Following injections of [3H]-adenosine or [3H]uridine intravitreally in chicks and rats or into the thalamus or neocortex of rats, three findings emerge: (1) axoplasmic transport of these materials occurs both in the anterograde and retrograde direction in birds and mammals; (2) anterograde transport in the axons of injected cells results in considerable cellular labeling in the region in which the axons terminate. This labeling is not exclusively transsynaptic, but probably results from some less specific, trancellular transport, since glial cells at the terminal site and along the path of the axons also become labeled; (3) injection of [3H]uridine in the chick optic and rat thalamocortical systems results in a pattern of labeling which differs considerably from that seen after injection of [3H]adenosine. After [3H]adenosine injections in the chick eye or rat thalamus, retrograde cell labeling is far more obvious than anterograde, transcellular labeling; after injections of [3H]uridine, anterograde, transcellular labeling is more intense than retrograde cell labeling.     

Analysis of neuritic outgrowth from severed giant axons in Lumbricus terrestris.

This study analyzes the detailed morphometric pattern at various postoperative times of neuritic outgrowths from the proximal and distal stumps of two uniquely identifiable axons. Morphological patterns of neuritic outgrowths from stumps of severed axons were compared for medial and lateral giant axons in the central nervous system of the earthworm Lumbricus terrestris. Outgrowths from proximal and distal stumps were labeled by injection of fluorescent dye into axonal stumps and assessed according to morphometric parameters. Outgrowths from axonal stumps of severed giant axons were statistically indistinguishable for most morphometric measures of neuritic quantity, shape, direction, and location. There were two exceptions to this general rule: 1) proximal stumps of medial giant axons produced significantly more neurites than distal stumps of medial giant axons, and 2) proximal stumps of lateral giant axons produced significantly longer neurites than proximal stumps of medial giant axons. No measure of neuritic outgrowth showed a significant change from the second through seventh postoperative week, suggesting that most outgrowth occurred in the first two postoperative weeks and that neuritic morphology remained stable through the seventh postoperative week. Neurites grew across the lesion site in relatively straight trajectories parallel to the longitudinal axis of the ventral nerve cord and often grew alongside the appropriate axonal stump across the lesion site. The length of neurites growing in close apposition to appropriate axonal stumps or giant axons was much greater than expected, had outgrowth been randomly directed. These data provide a basis for future investigations of the mechanisms that regulate neuritic outgrowth.     

The growth and myelination of central and peripheral segments of ventral motoneurone axons. A quantitative ultrastructural study.

This study compares the growth and myelination of those parts of cervical ventral motoneurone axons in the spinal cord (the intramedullary segments) and in the ventral roots of fetal and young rats (up to 21 days postnatal). The same fibre bundles are examined centrally and peripherally. Myelination begins centrally and peripherally at about birth. However, the peripheral segments of some fibres may begin to become myelinated before the central. Over the first 3 weeks after birth the minimum circumference of peripheral segments of myelinated axons remains relatively constant at 3 mum but that of central segments falls from 2.5 mum to just over 1 mum. Axons within the same fibre bundles tend to be thinner and less heavily myelinated centrally than peripherally. With ageing, axon circumference becomes more strongly correlated with sheath thickness. The thickness of the sheath surrounding an axon of a given circumference does not differ statistically from one age to another or between central and peripheral segments. Studies of myelin sheath growth rate show that in the early stages glial and Schwann cells vary independently of one another in the rates at which they add new turns to sheaths around central and peripheral segments of axons in the same bundles.     

Ultrastructural localization of slow retrograde axonal transport: an autoradiographic study

We used [3H]-N-succinimidylpropionate ([3H]-N-SP) to covalently label endogenous intra-axonal proteins within the nerve in order to study their bidirectional transport. At the time of injection virtually all of the labeled proteins are found at the injection site. At later times specific patterns of labeled proteins are found within the nerve both proximal to and distal from the injection site, as a result of retrograde and anterograde axonal transport, respectively. We undertook the current study to determine the ultrastructural distribution of the [3H]-N-SP-labeled transported proteins in the nerve. One microliter of [3H]-N-SP was injected subepineurially in sciatic nerve, and 5 days later the nerves were processed either for light and electron microscopic autoradiography or for gel electrophoresis and fluorography. At the injection site the labeled proteins are predominantly myelin proteins. Distally, a pattern similar to that described for slow anterograde transport is seen. Proximal to the injection site a constellation dominated by the 68-kilodalton protein is seen. Light microscopic autoradiography shows diffuse labeling both in axons and in myelin at the injection site, with predominant axonal labeling distant from the injection site. Electron microscopic autoradiography of segments distal to the injection site show silver grains which are distributed within the axoplasm without apparent relationship to organelles. In contrast, segments proximal to the injection site show silver grains which seem closely related to membrane-bound organelles, predominantly mitochondria. These results suggest that slow retrograde transport has a unique subcellular distribution that is distinct from that of slow anterograde transport.     

Organelle flux in intact and transected crayfish giant axons.

The flux of organelles moving by fast axonal transport in distal segments of severed crayfish medial giant axons (MGAs) and lateral giant axons (LGAs) was measured for survival times of up to 35 days (MGAs) or 60 days (LGAs). The response to transection occurred in 4 phases: (1) Organelle fluxes remained nearly normal for the first 24 h. (2) Fluxes then declined continuously until day 6 or 7. (3) A rebound toward normal levels lasted until day 21 (MGAs) or longer (LGAs). (4) During the final phase, fluxes declined either to zero (MGAs) or plateaued at a level which was a significant percentage of normal flux (LGAs). Changes in anterograde and retrograde flux were identical. The distribution of various size classes of translocating vesicles in distal segments of these axons was normal until day 4, with small and medium size, rapidly moving vesicles predominating. Afterwards, larger, slower vesicles predominated. During long-term survival, the axons remained physiologically intact, and cytoskeletons appeared to be normal, retaining intact microtubules which remained normally oriented with positive ends pointing distally. The evidence suggests that the two initial phases of the response to transection represent clearance from distal segments of organelle traffic which normally moves between axon and cell body. The rebound phase may be trauma induced, possibly a transient phase of cytoplasmic degeneration resulting from the loss of trophic support from the cell body. Differences between LGAs and MGAs with respect to organelle flux during prolonged survival, i.e. during the 4th phase of the response to transection, are consistent with different mechanisms of long-term survival which have been proposed for these axons.     

Distribution of [3H]RNA in goldfish optic tectum following intraocular or intracranial injection of [3H]uridine. Evidence of axonal migration of RNA in regenerating optic fibers.

The distribution of [³H]RNA in the goldfish optic tectum following eitherintra-ocular orintracranial injection of [³H]uridine during optic fiber regeneration has been studied by light (LMA) and electron (EMA) microscopic autoradiography.In one group of 4 fish both optic nerves were crushed, and 18 days later [³H]uridine was injected into the right eye. A second group of 5 fish, in which only one optic nerve had been crushed, received intracranial injections of [³H]uridine 18 or 22 days after the crush. All fish were sacrificed 24 days after crushing the optic nerves, a time when regenerating optic fibers have entered the tectum and are establishing functional reconnections. Tecta were fixed in situ with glutaraldehyde, dissected out, and samples were processed for LMA and EMA. Controls were carried out to ensure that [³H]RNA was the only radioactive component present in the tissue after fixation.The distribution of silver grains related to [³H]RNA in intraocularly injected goldfish was different from that following intracranial injection. Following intraocular injection virtually all the [³H]RNA was located in the layers of the left optic tectum (contralateral to the side of intraocular injection) where the regenerating optic fibers course and terminate, whereas virtually no radioactivity was present in the right optic tectum. EMA quantitative analysis of the labeled layers of the left optic tectum revealed that perikarya of cells, most of which are glial cells, had a density of grains related to [³H]RNA of 20–28 g/100 sq.μm; axonal growth cones had a density of 14–24 g/100 sq.μm. Grain densities over non-axonal cell structures were markedly lower, ranging between 3 and 6 g/sq.μm. Grains located over axons and growth cones accounted for 50–60% of all counted grains.Inintracranially injected goldfish, either 2 or 6 days after injection, silver grains were clustered over leptomeninges as well as vessels and parenchymal cells of the tectal strata containing the regenerating optic fibers. In the stratum opticum a high grain density was seen over glial cells, whereas virtually no grains were present over the fascicles of regenerating axons. EMA quantitative analysis revealed a grain density over glial and other parenchymal cells of the stratum opticum of 67 g/100 sq.μm, whereas densities over growth cones and regenerating axons were 1.3 g/100 sq.μm and 1.8 g/100 sq.μm respectively. Grains located over axons and growth cones accounted for 3.3% of all counted grains.On the basis of the present and previous findings it is suggested that followingintraocular injection of [³H]uridine the [³H]RNA present inside regenerating optic axons is transported from the ganglion cells of the retina; on the other hand, the [³H]RNA present in surrounding glial cells is the result of local utilization of [³H]RNA precursors which also migrate from the retina along with the [³H]RNA.It is also concluded that 2 and 6 days followingintracranial injection of [³H]uridine no substantial tranfer of [³H]RNA from glial cells to regenerating optic fibers occurs in the goldfish optic tectum.     

Local synthesis of axonal and presynaptic RNA in squid model systems

The presence of active systems of protein synthesis in axons and nerve endings raises the question of the cellular origin of the corresponding RNAs. Our present experiments demonstrate that, besides a possible derivation from neuronal cell bodies, axoplasmic RNAs originate in periaxonal glial cells and presynaptic RNAs derive from nearby cells, presumably glial cells. Indeed, in perfused squid giant axons, delivery of newly synthesized RNA to the axon perfusate is strongly stimulated by axonal depolarization or agonists of glial glutamate and acetylcholine receptors. Likewise, incubation of squid optic lobe slices with [3H]uridine leads to a marked accumulation of [3H]RNA in the large synaptosomes derived from the nerve terminals of retinal photoreceptor neurons. As the cell bodies of these neurons lie outside the optic lobe, the data demonstrate that presynaptic RNA is locally synthesized, presumably by perisynaptic glial cells. Overall, our results support the view that axons and presynaptic regions are endowed with local systems of gene expression which may prove essential for the maintenance and plasticity of these extrasomatic neuronal domains.