Denervation-induced formation of adrenergic synapses in the superior cervical sympathetic ganglion of the rat and the enhancement of this effect by postganglionic axotomy

Neuroscience Vol. 16. No. 4, pp. 997-1026, Printed in Great Britain

1985

0306-4522/85 $3.00 + 0.00 Pergamon Press Ltd ‘C, 1985 IBRO

DENERVATION-INDUCED FORMATION OF ADRENERGIC SYNAPSES IN THE SUPERIOR CERVICAL SYMPATHETIC GANGLION OF THE RAT AND THE ENHANCEMENT OF THIS EFFECT BY POSTGANGLIONIC AXOTOMY D. A.

RAMSAY*

and M. R.

MATTHEWSt

Department of Human Anatomy, University of Oxford, South Parks Road, Oxford OX1 3QX, U.K. Abstract-A study has been made at the ultrastructural level of the effects of denervation and axotomy on the synapse population of the rat superior cervical ganglion. Superior cervical ganglia were subjected unilaterally to acute (survival, 48 h) or chronic preganglionic denervation (survival, 41-189 days) by cutting the cervical sympathetic trunk; in chronic denervation experiments regeneration of preganglionic nerve fibres into the ganglion was prevented by suturing the proximal (caudal) stump of the trunk into the sternomastoid muscle. In some chronic experiments the preganglionic denervation was combined with simultaneous crush axotomy of the major postganglionic branches of the ganglion, the internal and external carotid nerves (axotomized-denervated ganglia). Control observations were made in contralateral ganglia and in ganglia from normal rats. After excision and before fixation, ganglia were incubated briefly in the presence of 5hydroxydopamine to label adrenergic vesicles. Chronic denervation caused a statistically significant 12% decrease from control values in the cytoplasmic minor axes of the principal ganglionic neurones; axotomy combined with chronic denervation led to a 6% increase in this dimension, which was not statistically significant. The minor axes of the neuronal nuclei did not differ significantly from control values in either type of experiment. Axotomy combined with denervation led however to a 36% decrease in the incidence of nucleated neuronal profiles per unit area of ganglion. Counts of synapses were made in the various classes of ganglia and their incidence was expressed per nucleated neuronal profile, to permit comparison within and between experiments. Normal and control ganglia showed a high incidence of synapses of preganglionic cholinergic type. Nerve terminal profiles and synapses containing small dense-cored vesicles, as distinct from the efferent synapses of small granule-containing cells, were not found to be present on the principal neurones or their dendrites in these ganglia, despite strong 5-hydroxydopamine labelling of small dense-cored vesicles within cell bodies and dendrites. After acute denervation extremely few residual synapses were found in the ganglion, in areas remote from small granule-containing cells, and these residual synapses were of the cholinergic type. Acute denervation led to the appearance of vacated or isolated postsynaptic densities; such densities were also found, but were fewer in number, in chronically denervated and axotomizeddenervated ganglia. Chronic denervation was associated with the appearance of new synapses, in the proven absence of regeneration of the preganglionic nerve fibres, amounting overall to 12% of the control incidence of synapses per nucleated neuronal profile. The majority of these synapses arose from nerve endings containing a variable population of small dense-cored vesicles which became specifically labelled by 5hydroxydopamine, indicating that they were of the adrenergic type. The numerical incidence of these synapses showed a significant correlation with the post-operative survival interval. When chronic denervation was combined with post-ganglionic axotomy, the incidence of synapses per nucleated neuronal profile was 32.6% of the control incidence of synapses. The incidence of synapses from nerve terminals containing small dense-cored vesicles in these ganglia was increased three-fold when compared with that in the chronically denervated ganglia. In comparison with acutely denervated ganglia, both chronically denervated and axotomized-denervated ganglia showed also a statistically significant increase of synapses from nerve terminals containing regular clear vesicles but there was no significant difference in the incidence of these synapses between the two long-term experimental groups, Possible post-operative origins of the two classes of synaptic terminals, the one containing small dense-cored vesicles and the other, regular clear vesicles, are discussed. The most likely source of both is considered to be nerve sprouts arising from the intraganglionic portions of the post-ganglionic axons, of adrenergic and cholinergic neurones, respectively. The possible sources of stimuli for sprouting in the denervated and axotomized-denervated ganglia are discussed. Several sources are likely, including the products of neuronal and nerve degeneration and “sprouting factors” arising from the ganglionic neurones, their satellite cells and, or, their targets; these might interact with or be superimposed upon intrinsic changes in the metabolic machinery of the denervated, or axotomized-denervated, neurones. The source of postsynaptic sites for the novel adrenergic synapses is also discussed. -_ *Current address: Department of Neurology, 5-1 I9 Meyer Bldg, Johns Hopkins University Medical School, Baltimore, MA 21205, U.S.A. tTo whom all correspondence should be addressed. Abbreviarion.7: NGF, nerve growth factor: NNP, nucleated neuronal profile; 5-OHDA. 5-hydroxydopamine; PSD. postsynaptic density; RER, rough endoplasmic reticulum; SCG, superior cervical sympathetic ganglion; SC cell, small granule-containing cell; VCP, vesicle-containing profile. 997

998

D.

A.

RAMSAY

and M. R.

Studies on intact sympathetic ganglia have demonstrated unequivocally the normal occurrence of adrenergic intraganglionic synapses in some species. These include the rabbit superior cervical ganglion2s~‘0’ and the human thoracic and lumbar paravertebral sympathetic ganglia.44 The distinguishing feature of these synapses is the presence in the presynaptic profile, after appropriate fixation, of a varying proportion of small synaptic vesicles which contain an electron-dense core. They thus resemble the vesicle-containing varicosities of peripheral adrenergic nerve terminal networks, e.g. in smooth muscle.35,37,” Studies of synapses in the normal rat superior cervical sympathetic ganglion (SCG) have led to conflicting reports on the incidence of synapses containing small dense-cored vesicles.65,9’,‘04The interpretation of these studies must, however, be qualified by the fact that specific labels for catecholamine terminals were not used. A further complication arises from the presence of clusters of small densecored vesicles in the dendrites and cell bodies of normal sympathetic neurones, including those of the rat SCG,‘06 which are not in general demonstrably synapse-related, but some of which may occasionally be seen in an arguably synapse-like configuration in respect of a second apposed neuronal profile.5” A number of reports of adrenergic synapses in the normal rat SCG have relied on evidence from denervation experiments. For example, Raisman ef al.” and &tberg et al. ” found no change in the low incidence of such synapses from 1 to 191 days and 59 to 143 days, respectively, after preganglionic denervation of the rat SCG, and concluded that such synapses represented a persisting and stable population of “intrinsic” or residual synapses (which amounted to some 9”/, of the normal synapse population of the intact SCG). Similar conclusions have been reached by other authors, including Grillo,39.40 Quilliam and Tamarind,‘” Lakes” and Joci et al.‘” There is, however, a possibility that the synapses noted by these authors in the preganglionically denervated rat SCG arise subsequent to denervation, from collateral sprouts of post-ganglionic axons (or, indeed, from any other local source of postganglionic adrenergic sympathetic fibres, such as the vascular nerve plexuses), developing in consequence of the preganglionic denervation. This phenomenon would be analogous to the sprouting of intact axons in the partially or completely denervated frog parasympathetic cardiac ganglion,z2.9h in partially denervated striated muscle’(’ and in several partially denervated regions of the central nervous system.” That sprouting of intraganglionic axons can occur in the denervated mammalian SCG has been inferred from physiological experiments on both partially denervated4’,63.74and completely denervated preparations.89 Moreover, the capacity for these sympathetic neurones to form connexions with one another has

MATTHEWS

been amply demonstrated in tissue-culture ments on explanted ganglia (e.g. Ref. 94). On

the

basis

of the

foregoing

expert-

observation5

it

seemed desirable to investigate further. at an ul!rastructural level, the synapses which are found UI .rirrc following preganglionic denervation of the rat SC‘G. Preliminary experiments in this laboratory” have, moreover, indicated that the combination of decentralization with post-ganglionic axotomy of the SCG, which induces regenerative axon sprouting within the ganglion” and alters the connectivity of the injured neurons,‘O may significantly modify the synapse population found following decentralization and therefore this situation also has been further explored.

EXPERIMENTAL

PROCEDURES

Young male Wistar rats (43-64 days old. 150-210 g) were used for the chronic nerve-lesion experiments. In addition, four older male rats of the same strain (92-120 days old, 30943Og) were used for acute denervation experiments. The ages of these rats were selected to be comparable with the final ages of the animals used for the long-term experiments. For all surgical procedures intraperitoneal chloral hydrate (3.5g/lOOml distilled water) was used as an anaesthetic. Surgical procedures

In 8 animals the left cervical sympathetic trunk was divided under anaesthesia approximately 5 mm from the lower pole of the SCG and its proximal end was sutured into the left sternomastoid muscle to minimize (and effectively eliminate) the chance of regenerating fibres growing back into the ganglion. These animals were allowed to survive for the following periods: 41, 53, 72, 105, 112, 131, 185 and 189 days. At a second terminal operation both the left denervated and the right undisturbed ganglia were removed under anaesthesia, fixed and prepared for microscopy as described below. These ganglia will be termed chronically denervated ganglia. In a further 6 animals a similar denervation operation was carried out. In addition the left post-ganglionic external and internal carotid nerves, which together contain the axons of approximately 73P;, of the ganglionic neurones,‘” were exposed and crushed twice at the same site for 10 s between accurately apposed tips of a pair of watch-maker’s forceps. Sites of nerve crush were, for the internal carotid nerve, approximately 2-3 mm above the upper pole of the ganglion and for the external carotid nerve, immediately before it embraces the origin of the superior thyroid artery. At a second terminal operation, following survival periods of 43, 56, 78,97, 152 and 178 days, the left and right ganglia were removed under anaesthesia and treated as outlined below. These ganglia will be termed axotomized-denervated ganglia. In all the above experiments it was confirmed at the second operation that the cervical sympathetic trunk had remained embedded in the sternomastoid muscle. In the 4 older animals, the left SCG was preganglionically denervated as described above and the rats were then allowed to survive for 48 h, after which time the left and right ganglia were removed under terminal anaesthesia and prepared for microscopy. These ganglia will be referred to as the acutely denervated ganglia. Tissue-processing procedure

The excised ganglia were rapidly divided into three blocks (cranial, middle and caudal) with a sharp razor blade. The

999

Synapse formation in denervated rat sympathetic ganglion cut separating off the cranial block was made just cranial to

Vesicle-containing

the origin of the external carotid nerve and the second cut was placed mid-way between this level and the lower pole of the ganglion. The cranial and caudal blocks were fixed immediately (see below). The middle blocks were incubated for 30 minutes at 37°C in a freshly prepared Krebs-Henseleit solution containing I x 10-j g/ml of 5hydroxydopamine (S-OHDA, Sigma Chemical Co., St Louis, USA) and a similar quantity of ascorbic acid (as an anti-oxidant), through which was bubbled a gas mixture containing 95% 0, and 5% CO,. These blocks were then in three changes of oxygenated rinsed briefly Krebs-Henseleit solution and immediately transferred to fixative. All fixation was carried out by immersion for l-2 h at 4°C in I % osmium tetroxide, buffered with 0.1 M sodium phosphate, pH 7.3. After fixation the tissue was briefly washed in distilled water. dehydrated in graded concentrations of ethanol, transferred to propylene oxide and embedded in Araldite. Samples for electron microscopy of experimental material were taken from all left-sided ganglia and samples for control observations were taken from right-sided ganglia of three animals in the chronically denervated group and three animals in the axotomized-denervated group. The right and left SCGs from one normal animal (aged 241 days) were also sampled. Semi-thin longitudinal sections (l-2 pm) from the middle block of each ganglion were taken from two levels separated by at least 100 pm and stained with a mixture of methylene blue and Azur II. These were used for an initial light microscope survey and for identification of suitable neurone-containing regions from which were cut thin sections, thereby giving two levels, separated by IOOpm or more, for electron microscopy from each ganglion sampled. The use of the middle block for quantitative sampling avoided undue contamination of the samples from denervated ganglia by the small population (0.1-l%)” of ganglionic neurones projecting caudally within the cervical sympathetic trunk, which would have been axotomized by the trunk lesion: most of these lie within the caudal onethird of the ganglion. Thin sections (silver to gold interference colour) were collected on squared copper grids, each grid square having a side of IOOpm and thus an area of 10,000~m2. The sections were stained with alcoholic uranyl acetate and aqueous lead citrate for viewing on a Philips E.M. 200 electron microscope.

As a substantial part of the ultrastructural observations is concerned with the detection, description and quantification of various categories of “vesicle-containing profiles”, it is appropriate at this stage to define what is intended by the tkrm~ Vesicle-containing profiles (VCPs) were defined as nerve profiles whose internal organelles are predominantly vesicles of synaptic type. A proportion of them exhibits a-synapse, or svnaoses. in the mane of section (“svnaptic VCPs”); the remainder (“non-synaptic VCPs”), while lacking direct evidence of synapse, resemble the synaptic VCPs in their content of characteristic vesicles and in the limited range and distribution of other organelles which may be present (e.g. microtubules, mitochondria, profiles of smooth endoplasmic reticulum and sometimes lysosomal elements). The two classes of profiles (synaptic and “non-synaptic”) also resemble one another in size, shape and the characteristics of their limiting membranes; the “non-synaptic VCPs” are therefore considered to represent sections through nerve terminals or immediately pre-terminal regions. While these profiles are likely to be synaptic endings for which the plane of section has missed their synaptic zones,31-33this cannot be established definitively without the examination of consecutive serial sections. Nevertheless, a statistically significant correlation is observable between the incidences of synaptic and non-synaptic VCPs (see Results) and this is strong suggestive evidence that these two types of profiles are drawn from the same population of synaptic endings. For this reason the unqualified use of the term VCP refers, in this paper, to a vesicle-containing profile as defined above irrespective of whether it exhibits a synapse in the plane of section. Non-synaptic VCPs which contained fewer than 10 vesicles were excluded from the analysis because there were usually insufficient additional data to permit confident classification of these profiles. The features of VCPs varied, indicating the occurrence of distinctively different kinds of synaptic endings, classifiable according to the type and proportion of synaptic vesicles and other organelles and these were analysed separately. Non-synaptic VCPs were quantified in addition to synaptic VCPs because some categories of synapses have such a small synaptic region, and (partly for this reason) are so infrequently found, that an additional index of putative or potential synapse formation was considered desirable.

electron

microscopy

A map was drawn at low magnification of each ganglionic section as it lay on the grid in the electron microscope. Five grid squares (in the case of the operated ganglia), and two to three squares (in the case of the control and normal ganglia) from each of the two levels were selected for quantitative survey. Each square selected was separated from any other quantified square of the same level by at least one grid square (i.e. 100 pm plus the thickness of two 30 pm grid bars) and was, as far as possible, free from any major defects such as stain deposits, holes and knife marks. Squares were chosen which contained evenly distributed neuronal profiles, and those with large axon bundles or major blood vessels were excluded, in order to maximize and to equalize approximately the yield of neuropil per square. Each selected square was then scanned at a magnification of approximately x 21,000 for counting and examination of various features such as synapses, “vesicle-containing profiles” (see below for definition) and specialized contact regions. Any doubtful features were photographed for later discussion. Finally, the number of nucleated neuronal profiles per square was noted; a profile whose nucleus was partially obscured by a grid bar was counted as one half.

.

L

Nucleated

Quantitative

projles

neuronal

-

.

profles

It became apparent during microscopy of axotomized ganglia that the incidence of neurones per unit area was reduced compared with values from control, normal and denervated ganglia. This is consistent with the observations of Acheson and Schwarzacher,’ who noted a 50% reduction in the packing density of neurones per unit area between 14 and 28 days after surgical section of the post-ganglionic hypogastric nerve of the cat inferior mesenteric ganglion, which they attributed to neuronal loss. In agreement with this conclusion, Purves”’ described a 55% loss of neurones in the SCG of guinea-pigs at 30-90 days after crushing the major postganghonic nerve trunks and Smolen”’ found a comparable loss of neurones in the rat SCG at 2 months after cutting the postganglionic nerve trunks. Any reduction in the relative incidence of neurones may be expected to lead to a concurrent reduction per unit volume in the availability of membrane upon which synapses could form. For this reason all quantitative analysis of structures such as synapses, VCPs and non-synaptic specialized contact sites has been expressed in terms of their incidence per nucleated neuronal profile (NNP). Statistical

procedures

In general the statistical distribution and variance of the populations from which the present experimental data were

tOO0

D. A. RAMSAY

observations were available to establish that the data were drawn from populations of approximately distribution and similar variance. Otherwise the U-test9’ has been used to assess the of differences between paired sets of data. This test has been applied in the two-tailed form unless the alternative hypothesis (H,) predicted the direction of the difference. All unqualified P values given in the Results are for a two-tailed test and are based on the U-value. Where more than two groups of data were compared an initial screening for differences between the groups was performed using the Kruskal-Wallis one-way analysis of variance;97 where the H value obtained from this test indicated a significance level for accepting H, of P < 0.1, the Mann-Whitney U-test was used for further comparisons between paired sets of the data.

The median has been used as the measurement of location for non-parametrically treated data;14 its intervai estimate is indicated by the approximate, non-parametric 95”+, confidence limits.‘4,‘9 The arithmetic mean and its parametric 95”/, confidence limits have been used in describing the dimensions of neuronai nuclear and somatic profiles.

Apparent between two sets of data have been assessed by calculating the Spearman rank correlation coefficient (r,)97; P values relating to the statisticai significance of the rs are identified by the name “Spearman” In the graphic presentation of such data, linear regression analysis has sometimes been used to draw a “best-fit” straight line through the points in question;” this procedure is for illustrative purposes only and does not imply that the populations from which the data are drawn are considered to be normally distributed. The conventional level for accepting significance m a biological system (i.e. P 6 0.05) is used. Conventional symbols have been used to denote the results of statistical analyses in the tables and text (N.S., not s~gni~cant~ *~ P 6 0.05; **. P d 0.01: ***. P < 0.001. RESULTS

Examination of semi-thin sections stained with Azur II and methylene blue revealed no obvious difference between control and chronically denervated ganglia, or between these and normal ganglia. However, ganglia which had been subjected to postgangiionic axotomy 6-27 weeks previously showed

Figs I and 2. See p. 1001. Examples showing nucleated neuronai profiles from a control ganglion (Fig. I) and a chronically denervated ganglion (Fig. 2). Scale bars represent 5 pm. In the neuropil of Fig. I are indicated examples of dendrites (d) and a synapse (arrow); n, nucleus of satellite cell. Figure 2 is at a slightly lower magnification and the neuronal nuclei are sectioned closer to their poles, as shown by the obliquity of the nuclear membranes. In the neurones of the chronicaily denervated SCG (Fig. 2) the uitrast~cturai appearance is similar to that of normal or control neurones (Fig. 1) but collagen fibres and cells of fibrocytic type (f) appear to be more numerous in the surrounding ganghonic neuropil. Fig. 3. See p. 1002. This electron micrograph, from a chronically denervated SCG treated with S-OHDA, shows a dendrite dividing into lwo branches shortly after arising from the soma of an SCG neurone. Several clusters of small vesicles, many of which contain dense cores, are present in the dendrites, notably in their bases (arrow-heads); two “small dense-cored VCPs” are indicated by large arrows. A few solitary small dense-cored vesicles are visible in the neuronal soma (e.g. small arrow), some lying near to Golgi zones, and in the dendrites; the soma also contains a normal range of organelles (see Fig. l), including mnIt~vesicular bodies (e.g. m). These large dendrites show many longitudinally oriented microtubules and filaments and also contain an abundance of other organelles, including clumps of rough endoplasmic reticulum. B indicates the wall of an adjacent blood vessel. Scale bar represents 1pm. Figs 47. See p. 1003. Examples of synapses in control ganglia, with (Figs 6 and 7) and without 5-OHDA treatment (Figs 4 and 5). Scale bars represent 0.2 pm (Fig. 6, 0.3 pm). Fig. 4. A synapse on the shaft of a longitudinally sectioned dendrite. The presynaptic profile contains numerous small regular agranular vesicles (“regufar clear vesicles”) and a few Larger dense-cored vesicles with cores of varying electron density; the small electron-dense granules in the presynaptic profile are identifiable as glycogen particles. The postsynaptic profile contains longitudinally oriented microtubutes and other organelles typical of dendrites (see Fig. 3). This synapse shows a sub-synaptic apparatus: beneath the postsynaptic density. which is asymmetric, is seen a row of electron dense sub-junctional bodies which are linked with the deep aspect of the sub-synaptic web and have a row of associated vesicles in the subjacent cytoplasm. A specialized attachment plaque is also evident between the vesicle-containing profile and the dendrite (between arrowheads); the section may additionally have glanced through a second synapse (arrow). Fig. 5. This shows a synapse between a typical preganglionic “regular dear VCP” and a transversely sectioned dendrite. This synapse also is strongly asymmetric, and like that in Fig. 4 it too shows a sub-synaptic apparatus. Note that the sub-junctional bodies still appear discrete and circular in transverse as well as in longitudinal sections of the dendrites (Fig. 4), which indicates that these structures are spherical. Here too there are irregular vesicles associated with their deep aspect. Dendritic shaft synapses may show a sub-synaptic apparatus; spine synapses do not (Figs 6 and 7). Fig. 6. A non-synaptic regular clear vesicle-conta;nin~ profile (labelled cvp), and a synaptic profile of the same type (above, right) which forms a short synapse upon a spine-like profile containing a number of glycogen particles. Both VCPs contain larger dense-cored vesicles of similar type; in the non-synaptic VCP these are more numerous and accompanied by a cluster of mitochondria and by microtubules. Fig. 7. A typical regular clear vesicle-containing profile is seen forming a synapse upon a short spinous excrescence arising from a neuronal soma (sp) and a specialized attachment plaque with the soma membrane (between arrowheads), Spine synapses tend to be less strongly asymmetric than shaft synapses.

1001

1002

1003

Figs 9-14. See p. 1005. Scale bars represent 0.2pm. Figure 9, taken from a control SCG, illustrates a specialized attachment plaque between two neuronai profiles, comprising aggregations of cytopiasmic dense material along a region of parallel apposition of membranes: no vesicles or other organeiies approach close to the membranes in this region and the immediately subjacent cytoplasm flanking the attachment in each profile is occupied by a finely filamentous web. The upper profile is evidently dendritic: the lower, larger one is probably a neuronal soma. It contains a cluster of small vesicles (lower left) which do not show electron-dense cores. The tissue was not treated with 5-OHDA. Figs i&i4 illustrate the degeneration of synapses in acutely denervated ganglia (Figs 10, ii and 13) and the appearance of isolated postsynaptic densities in chronically denervated ganglia (Figs I2 and 14). Ail are from material treated with 5OHDA. Fig. 10. A surviving regular clear VCP of pregangiionic type contacts a dendrite forming an asymmetrical synapse, in the upper half of this figure. A vesicle-containing profile which appears to be undergoing dense degeneration is seen below and to the right. It is possible that the synaptic and degenerating profiles are related and that the synapse is in a very early stage of degeneration. Neither profile has become iabelled with 5-OHDA. Fig. il. A small, uniabelied vesicle-containing profile is seen in a “iucent” stage of degeneration. Fig. 12. The surface membrane of a ceil soma displays a vacant postsynaptic membrane density (between arrow-heads). There is a faint electron-dense fuzz on the extracellular surface of the PSD which may be the remains of dense material of the synaptic cleft. The PSD is covered by a satellite ceil process. Fig. 13. Profiles reminiscent of degenerating nerve endings are seen enclosed within a band of highly reactive-looking satellite cytoplasm. Fig. 14. A vacant postsynaptic membrane density upon a dendritic shaft (between arrows). The PSD lies in direct apposition to a region of neuronal somatic surface membrane. This relationship is quite distinct from the attachment plaque shown above in Fig. 9. Figs 1619. See p. 1006. These illustrate synaptic small dense-cored VCPs in ganglia with (Figs 16 and 18) and without 5-OHDA treatment (Figs 17 and 19); Figs 16 and 17 are from axotomized-denervated ganglia and Figs 18 and 19, from chronically denervated ganglia. Scale bars represent 0.2 pm (Fig. 18, 0.1 pm). Fig. 16. A profile containing small vesicles, some of which are characterized by a faint electron-dense core, is seen forming an asymmetrical synapse upon a dendrite. The dendrite contains a cluster of small vesicles (indicated by an asterisk), none of which contains a distinct dense core (cf. Fig. 3). Fig. 17. A profile containing numerous small vesicles heavily laden with a dense precipitate (caused by 5-OHDA treatment) forms two dendritic synapses in the regions indicated by the arrows. The postsynaptic thickenings of these synapses are dissimilar: one is symmetrical, on an obvious dendritic shaft (on the right) and the other (above) is asymmetrical, upon a profile which is equivocal, possibly dendritic, containing a number of scattered, small, uniabelled clear vesicles. This might belong to a cholinergic neurone. Small vesicles in the right-hand postsynaptic profile and in an adjacent dendrite (small arrow) are labeiled with 5-OHDA indicating that these belong to adrenergic neurones. Fig. 18. A small dense-cored VCP forms a synapse with a spine-like process arising from a dendrite. The vesicles in the VCP include a proportion with clearly distinguishable small electron-dense cores, although these are not as pronounced as those in tissue treated with 5-OHDA (cf. Figs 17 and 19). Note also in the adjacent neuropii several small transversely sectioned, axon-like, microtubule-containing profiles, one of which contains small dense-cored vesicles (arrow). Fig. 19. A synaptic varicosity containing small-dense-cored vesicles, in continuity below with a longitudinally-sectioned, narrow axon-like segment (arrow) which itself contains some small dense-cored vesicles (cf. arrow in Fig. 18). In this case, almost all the vesicles in the vesicle-containing profile are heavily iabelled by S-OHDA (cf. untreated examples illustrated in Figs 16 and 18). The synapse is asymmetric, upon a dendritic shaft, and has a filamentous type of sub-synaptic web. A short symmetric attachment links the dendrite with the point of origin of the VCP from its narrow segment, and a second postsynaptic-like specialization is seen in the dendrite opposite to the vesicle cluster in this segment. Figs 2&22. See p, 1007. These are ail from axotomized-denervated ganglia treated with 5-OHDA. Scale bars indicate 0.1 pm (Fig. 20) 0.2 pm (Fig. 21) and 0.25 pm (Fig. 22). Fig. 20. This high magnification electron micrograph of a somatic vesicle cluster illustrates some of the appearances which characterize the electron-dense deposit in small dense-cored vesicles from 5-OHDA-treated tissue. The deposit forms either a series of peripherally placed clumps, a central dense core or an eccentrically-positioned dense core. Pig. 21. This shows two consecutive varicosities containing small dense-cored vesicles (small dense-cored VCPs), which are linked by a narrow microtubule-containing segment of terminal axon-like character. The upper varicosity forms a synapse with a dendritic shaft. This synapse has a form of sub-synaptic specialization: postsynaptically it shows an indistinct array of sub-junctional bodies, associated with a row of vesicles of smooth endoplasmic reticulum. Fig. 22. A large profile which contains small dense-cored vesicles is seen to form two synapses, upon a dendritic shaft and a short bulbous-ended projection from the same shaft. Above and to the left the presynaptic vesicle-containing profile is continuous with a narrow axon-like microtubuie-containing segment similar to those shown in Figs 19 and 21. Both the postsynaptic dendrite and a second dendrite (above) contain small dense-cored vesicles and therefore presumably belong to adrenergic neurones. IOtl4

1005

IO06

1007

Figs 23 and 24. See p. 1009. These micrographs are both from chronically denervated ganglia into which a somatic nerve has been implanted. The nerve endings illustrated are, however, typical of the innervation of the intraganglionic arterioles. Scale bars indicate 0.2 pm. Fig. 23. The vesicle-containing profiles shown in this electron micrograph of tissue not treated with 5-OHDA form part of the innervation of a smooth muscle cell (sm) of an intraganghonic blood vessel. Some of the vesicles in these profiles contain small electron-dense cores but these are much less pronounced than those found in tissue treated with 5-OHDA (cf. Fig. 24). Fig. 24. Innervation of a vascular smooth muscle cell (sm) from a ganglion piece treated with 5-OHDA. Many of the vesicles in the vesicle-containing profiles contain a heavy electron-dense deposit indicating an adrenergic nature. Figs 25-27. See p. 1010. These are all from chronically denervated SCGs treated with 5-OHDA. Scale bars indicate 0.2 pm. Fig. 25. A small excrescence is seen arising from a neuronal soma; this structure contains a cluster of small vesicles, some of which contain dense cores and a rosette of ribosomes (arrow) which identifies it as dendrite- or spine-like. This type of profile might have been classified as a small dense-cored VCP if the section had passed through the centre of the vesicle cluster avoiding the polysome. Profiles of this type are, however, extremely infrequent and so are unlikely to have introduced significant bias into the counts of small dense-cored VCPs. Such excrescences have not been seen to give synapses. Fig. 26. Two vesicle-containing profiles are shown, of which that on the left contains small dense-cored vesicles and is a typical “non-synaptic small dense-cored VW”, whereas the right-hand profile (labelled cvs) contains small regular clear vesicles similar to those found in the presynaptic nerve terminal profiles of normal ganglia. This profile forms an asymmetric synapse with a neuronal profile which contains a multivesicular body and is probably a small dendrite. Fig. 27. A small profile containing small, regular clear vesicles (arrow) forms an asymmetric synapse with a substantial dendritic shaft, containing numerous mitochondria and small cytoplasmic dense bodies and a considerable sub-synaptic accumulation of smooth endoplasmic reticulum. An adjacent profile containing similar small regular clear vesicles (above right) exhibits transversely sectioned microtubules and forms a short specialized contact with a small spine-like profile; this junction is obliquely cut but is probably synaptic.

1008

1009

lOI0

Synapse formation in denervated rat sympathetic ganglion

1011

branches being very small. The variable spaces between the larger dendrites and the neuronal somata were frequently filled with a complex network of “neuropil” elements comprising clusters of dendrites and axons enclosed in satellite cell processes, together with vesicle-containing nerve terminal profiles (VCPs), some of which formed synapses with the dendritic arborizations; the incidence and predominant features of the VCPs depended on the group of animals from which the ganglion had been drawn (see below). In some areas, particularly in regions adjacent to the ganglionic capsule, the neuropil was relatively scanty with large tracts of tissue space containing loosely dispersed collagen fibres and associated connective tissue cells. In the axotomized-denervated ganglia the packing of the neuronal profiles was considerably reduced, the inter-neuronal neuropil clusters were diminished and there appeared to be an increase in the collagen content of the ganglia (noted also by other authors such as tftstberg et af.). 8’ The rough endoplasmic reticulum (RER) in the neuronal cytoplasm was often made up of short segments, rather than the more typical extensive, stacked, flattened cisternae; its intra-neuronal distribution was, however, similar to that of normal and denervated ganglia. The fragmented appearance of the RER was not exclusive to Ultrustructure axotomized tissue as a similar pattern was found General observations. A survey of the tissue at low from time to time in the control and denervated magnification did not reveal any major differences in tissue, possibly as a consequence of the plane of neuronal cytoarchitecture between the normal or sectioning. Conversely, RER of normal appearance control and denervated ganglia. The arrangement of was also encountered in some neurones of the axotomized-dene~ated ganglia, representing either the organelles in the neuronal somata was uniform, more fully recovered neurones or neurones which had with stacks of rough endoplasmic reticulum arranged not been injured, since the cervical and caudalward around a centrally placed nucleus, and with the Golgi branches of the ganglion were left uncrushed. complex occupying a perinuclear position, MitoThe median incidence of nucleated neuronal chondria, short segments of smooth endoplasmic profiles (NNPs) per 100 x lOO+m grid square in the reticulum, multivesicular bodies and cytoplasmic axotomized-denervated ganglia was significantly less, dense bodies lay scattered through the cytoplasm; autophagic vacuoles were infrequently seen (Figs 1 by 30--35x, than that of the right-sided controls (P = 0.002) and of the denervated ganglia and 2). Where dendrites were seen arising from the neu- (P < 0.002), though there was no significant ronal profiles they were often found to begin brandifference in the incidence of neuronal nuclear ching close to the soma (Fig. 3), some of these early profiles between the latter two groups (Table 1). an apparent loosening of the neuropil and a reduction in the packing density of the neurones in comparison with control and denervated ganglia, which suggested a degree of neuronal loss or dispersal. Despite this, the appearance of most of the neurones in the axotomized-denervated ganglia at these long post-operative intervals was remarkably close to normal except that they appeared to contain more dense bodies than usual, which were sometimes clustered in one peripheral region of the cytoplasm. Nuclear morphology was similar between the three groups of ganglia; nuclear profiles were usually circular or slightly elliptical, often showing one or more nucleoli in l-2 pm sections, and nuclei in the axotomizeddenervated ganglia rarely exhibited the eccentricity, indentation and crenation characteristic of recently axotomized neurones.“‘.” The distribution of Nissl substance in the neurones was also similar between the three groups. In the axotomized-denervated ganglia a few neuronal profiles were seen which were greatly shrunken, with small, e~entrically located and, or, distorted nuclei and with pale-staining cytoplasm which appeared virtually devoid of Nissl substance. These are considered likely to be neurones which had not re-established a sustaining contact with targets and which were severely atrophic.

Table 1. Median ineidences of nucleated neuronal profiles (NNP) per 100 x lOC+m grid square, with approximate 95% non-parametric confidence limits in parenthesis, and mean values for minor axes of neuronal nuclei and nucleated cytoplasmic profiles, with 95% parametric confidence limits. Significances of differences are indicated by conventional symbols (see Experimental Procedures). The Mann-Whitney U-test was used for the non-parametric data and the paired r-test for the parametric values Number of ganglia Control ganglia

6

Chronically denervated ganglia Axotomizeddenervated ganglia

8 6

NNP per 104pm2 5.15 (5.1-6.5) NS 5.40 (5.2-7.6) ** 3.75 (2.7-4.2)

Number of neurones measured

Mean minor axis of nucleus (pm)

297

9.33 + 0.4

300

NS 8.37 & 1.09

3.18

NS 9.63 i 0.76

**

Mean minor axis of NNP (pm)

NS

1011

ID. A.

KAMSAY

and M. R.

To check that the reduced incidence of nucleated neuronal profiles in axotomized-denervated ganglia was not an effect attributable simply to changes in nuclear size, the minor axes of the nuclear profiles of neurones from control ganglia, chronically denervated and axotomized-denervated ganglia were measured (Table 1); the lengths of these axes in the three groups were found not to differ significantly from one another. To determine whether there was any gross change in neuronal size between the control, chronically denervated and axotomized-denervated groups of ganglia, the minor axis of the cytoplasm of each nucleated neuronal profile used for the determination of the nuclear minor axes was measured. This procedure revealed a statistically significant reduction in the length of the cytoplasmic minor axes of the chronically denervated neurones of about 12% in comparison with the corresponding values for axotomized-denervated and control neurones; the cytoplasmic minor axes of the axotomizeddenervated and control neurones did not differ significantiy from one another (Table I). These findings indicate a shrinkage of the chronicalIy denervated SCG neurones. which would imply a reduction of the order of 239’, in the cross-sectional area of the soma, and this is in broad agreement with the observations of Hamlyn,“2 who reported a 17-l8”/a decrease in the cross-sectional area of rabbit SCG neurones 100 and I60 days after preganglionic denervation.

MATTHEWS

tudinal and transverse sections of the dendrites (cf. Figs 4 and 5) which indicates that they are spherical. Postsynaptic densities without an overlying synaptic bouton were rarely seen in the control ganglia (Table 3). The regular clear vesicle synapses in these adult rats contact either a dendrite (Figs. 4 and 5) or a small, spinous excrescence arising from a dendrite (Fig. 6); only two synapses (out of a total of 650) were seen directly on the neuronal somatic membrane. and one on a small spinous excrescence arising from the

soma (Fig. 7). The dendritic spines contain finely filamentous cytoplasm and occasionally some clear of membranous vesicles, strands material or glycogen-like particles (Fig. 6): dendritic shafts are characterized by the presence of microtubules.

1:

Observations concerning vesicle-containing profiles, attachment plaques und vesicle clusters. ~ormul und controt gang&

The neuropil of these ganglia showed numerous VCPs of which close to one half were seen to form synapses in the plane of section (Figs 4-7; Fig. 8; Table 2). These synapse-forming profiles were all of a similar type, characteristic of sympathetic ganin which the majority of vesicles are glia, 3’.32~6’~‘04~105 “clear” or electron-lucent (agranular), circular and of relatively uniform size (diameter. 30-50 nm); for the purpose of this analysis they are termed “regular clear VCPs”. Such profiles also contain a varying but generally small number of larger, dense-cored vesicles, SO-IO0 nm in diameter, which tend to be situated around the periphery of the profile and are less often seen in the immediately presynaptic zone. The synapses formed by the regular clear VCPs are usually asymmetrical, i.e. there is a well-marked postsynaptic membrane-associated cytoplasmic density (PSD), greater than that on the presynaptic side. The

PSD is more strongly marked at dendritic than at spinous sites, and in the former case, particularly on the large dendrites or on the soma, there is occasionally an underlying series of electron-dense bodies (the “‘subsynaptic apparatus”); these appear as rounded profiles, 50-60nm in diameter. in both longi-

A

B

C

D

A

8

C

D

Fig. 8. These sets of histograms (I, II) show the incidences of VCPs in the four categories of ganglia which were analysed quantitatively. In-each set &uhn A = control, B = acute denervation. C = chronic denervation and D = axotomy with chronic denervation. Set 1 presents regular clear VCPs; set II, small dense-cored VCPk In each &t, total height of histoefam = median incidence of all VCR Blank area = VCPs which do not display synapses in the plane of section. Obliquely hatched area = VCPs forming synapses upon spine-like processes. Dotted area = VCPs forming synapses upon dendritic shafts. Note the interruption in the vertical axis of both sets of histograms. The results of statistical comparisons between these data are presented in Table 2.

~__.__

(o-0.62)

8.09 (5.81-9.43) *** 0.10 (O-0.24) *** 0.16 (0.0330.8s) NS 0.18

Regular clear (rc) VCPs

0.10 (0.03-0.73) NS 0.16 (a-0.43) (25.8zlO) NS (a”;‘oO)

($0)

(@-Oloo)

57.1 (50.5-72.3)

47.2 (44.2-57. I)

3.65 (2.80-5.38) *** 0.007 (2.2)

-.-~--.

% of rc synapses on dendritic shafts

% of rc VCPs which were synaptic

Regular clear vesicle (rc) synapses ______~___..

_.._ ~.____.__.__~.__

(1.2-7.78)

0.88 (0.21-1.81) ** 3.15

0.09 (O-O.16) NS 0.06 (o;p;*o9)

_~._~ Small dense-cored (sd) VCPS

0.33 (0.09-0.44) * I .03 (0.38-2.09)

0

(lZ9)

34.6 (22-55.7) NS

0

Small dense-cored VCPs _-._________ .-... Small % of sd VCPs dense-cored which were vesicle (sd) synaptic synapses __-_.~________~0 0

68.3 (52.3-74.1)

,,clz;.,

0

0

% of sd synapses on dendritic shafts

-

1.22

Axotomizeddenervated (6)

(0.40-2.52)

0.43 (0.12-1.12) *

(&.02)

3.63 (2.80-5.38)

Total

Chronically denervated (8)

Acutely denervated (4)

Control (6)

Class and number of ganglia

11.8 (3.3-30.8) * 33.6 (1I X3-69.4)

(A.6)

100

% of control

Synapses

0.92 (0.36-1.68) NS 0.34 (0.16-1.13) NS 0.31 (0.22-0.61)

0.015 (Vi1 3)

Total

c&00,

@ES) NS 12.9 (60.;;00)

0

% of PSDs on dendritic shafts

Isolated PSDs

Total

(0.6Yz.74,

(2.&?.38) +* 0.93 (0.36-l .7) NS 0.76 (0.32-2.04) *

-_-

44.0

20.9

25.5

100

% of control

(0.3E.34) * I .08 (0.50-2.08)

(15-3.55) t 0.80 (0.33-1.44) NS

Dendritic

Pooled synaptic elements

:; (0.12-0.91) * 0.42 (0. M-0.66)

(1.04-1.88) * 0.13 (0.03-0.26)

1.47

Spinous

Table 3. Median incidences, per NNP. and approximate 95% non-parametric confidence limits (in parenthesis) of total synapses, isolated postsynaptic densities (PSDs) and pooled synaptic elements, with the corresponding proportions of PSDs which were on dendritic shafts and the incidences of pooled synaptic elements which were situated on dendritic shafts and on spines

denervated (6)

Axotomized-

denervated (8)

Chronically

Acutely denervated (4)

Control (6)

Class and number of gangiia

._.., ______

Regular clear VCPs

Table 2. Median incidences. per nucleated neuronal profile, and approximate 95% non-parametric confidence limits (in parenthesis) of regular clear VCPs and small dense-cored VCPs and of their associated synapses, with the corresponding proportions of VCPs which were seen to form synapses and of synapses which were situated on dendritic shafts: levels of significance are for comparisons between adjacent groups by the Mann-Whitney U-test

1014

D. A.

RAMSAY and

neurofilarnents, ribosomes (some arranged in polysomes), RER and mitochondria, as well as occasional multivesicular bodies or lysosomes (Figs, 4 and 5). Some dendritic branchlets are very small and may readily be confused with spines if a careful inventory of their contents is not taken. The ratio of dendritic to spinous synapses, on a strict analysis of postsynaptic profiles, was approximately 1.33: I (Fig. 8; Table 2). Non-synaptic. specialized contact sites of macula adhaerens type (“attachment plaques”“) are also present between neural profiles (median incidence in this series = 1.57/NNP; confidence limits 1.10 to 1.95); they are distinguishable from synapses by the parallel apposition of two neuronal membranes with symmetrical membrane-associated cytoplasmic densities, without vesicle clustering, and are most often placed between dendrites (approximately 70:/,, Fig. 9) which is in agreement with Elfvin’s”,” findings. Small vesicles, some of which contain an electrondense deposit, and most of which become labelled after 5-OHDA, are frequently found in clusters underlying the surface membrane of the neuronal somata (“somatic vesicle clusters”; median incidence 2.37 per NNP: confidence limits, 0.143.22) or dendrites (“dendritic vesicle clusters”, Figs 3 and 25; median incidence, 4.48 per NNP; confidence limits, 2.10-7.53)” The incidence of somatic vesicle clusters varies greatly between neuronal profiles: some of the large somatic profiles contain up to 10 such clusters whilst others of similar size have none. Some of the dendritic vesicle clusters are closely associated with regions of surface membrane which are “exposed” to the intraganglionic tissue space (i.e. are covered by basal lamina in place of the normal satellite cell process); the median incidence of these specialized associations was 0.19 per NNP. An additional, but infrequent, finding was the occurrence of nonspecialized apposition of two neuronal profiles, one of which locally contained a vesicle cluster; similar observations were made by Taxi ct o/.“‘~ in rat superior cervical ganglia which had been denervated 5d to 7d previously. In the present study, solitary small dense-cored vesicles were also frequently cncountered in the cytoplasm of the neuronal somata, particularly in the vicinity of the Golgi apparatus. from which they may have arisen (Fig. 3). (Cuello and Ivcrsen’” make a similar suggestion concerning the origin of 5-OHDA labelled somatic vesicles in neurones of the substantia nigra). In no instance were synapses formed by VCPs containing small, dense-cored vesicles found in control tissue. despite the pronounced uptake of 5-OHDA by adjacent clusters of dendritic and somatic vesicles and by vesicles of the nerve profiles innervating the intraganglionic arterioles. A few nonsynaptic VCPs containing small dense-cored vesicles were noted which could not be identified as either dendritic or somatic (Table 2).

M. R.

MATTHEWS

Acutely denervated ganglia Acute section of the cervical sympathetic trunk results in the almost complete disappearance ofintraganglionic synapses. 8’.9o9’ In the present series of experiments, the ages of the 4 animals which were subjected to preganglionic axotomy and were allowed to survive for 48 h corresponded to the ages at which the formation of small dense-cored VCPs was well under way in chronically denervated animals (see below), namely 91 and 122 days of age. In the 4 acutely denervated ganglia the ultrastructural survey covered an area containing 216 nucleated neuronal profiles, which yielded only IX regular clear VCPs and IO VCPs which contained various proportions of small vesicles with an electron-dense content (“small dense-cored VCPs”; Table 2). Synapses formed by regular clear VCPs were almost completely absent (Table 2): only two such synapses were found in 40 100 x 100 pm grid squares surveyed, representing II?, of the regular clear VCPs found, in contrast with the 47% which were seen to form synapses in normal ganglia. Of these two synapses, one was adjacent to a VCP in the early stages of degeneration (Fig. IO), suggesting that these may be synapses in which degeneration has been relatively delayed for some reason (cf. Ref. 43). The loss of synapses was accompanied by the appearance of numerous unoccupied, or isolated, postsynaptic densities (Figs I2 and 14; Table 3) as well as of nerve terminal profiles in different stages of degeneration (Figs 10, 11 and 13), although the latter were by no means frequent, a finding which underlines the rapidity with which terminal degeneration and its resolution occur in the SCG.4’.W.92Occasional degenerating myelinated nerve profiles were noted, in addition to normal myelinated fibres. Small dense-cored VCPs were no more frequently found in the acutely denervated ganglia than in normal ganglia and were not observed to give any synapses (Table 2). For the isolated postsynaptic densities (PSDs) the ratio of detected dendritic to spinous PSDs was 5.5: 1; thus they were more persistent. or more readily detected, upon dendritic shafts than upon dendritic spines after the detachment of the degenerating presynaptic nerve terminal. If the synaptic elements (i.e. the regular clear vesicle synapses and PSDs) in each of the control and acutely denervated ganglia are summed (Fig. 15, Table 3) the 2-day preganglionic lesions may be shown to have provoked a 74’3; loss of detected postsynaptic specializations. Chronically denervated ganglia In these ganglia 80 100 x 100 pm grid squares were surveyed, which contained a total of 433 nucleated neuronal profiles; this yielded 391 small dense-cored VCPs, 124 (32% overall) of which were seen to be synaptic and 85 regular clear VCPs, of which 53 (6296 overall, a greater proportion than in normal ganglia) formed synapses in the plane of section.

Synapse formation in denervated rat sympathetic ganglion (i)

Small dense-cored vesicle-containing profiles

The most notable difference between the chronically denervated ganglia and the acutely denervated and control ganglia was an almost IO-fold increase in incidence of small dense-cored VCPs (Figs 8 and 26, Table 2) which were not obviously either dendritic or somatic in nature. These profiles contained a mixed population of small (30-50 nm) regular spherical vesicles, some clear-centred and some containing an electron-dense deposit, the characteristics of which depended on whether or not the tissue had been treated with 5-OHDA (see below). A median proportion of approximately one in three of these VCPs was seen to form synapses in the plane of section (Figs 18 and 19; Table 2). The median incidence of these synapses was 9% of the median incidence of synapses in control ganglia. 63% of these synapses were upon dendrites and the remainder on spines, thus giving a dendritic to spinous ratio of 1.7: 1. This ratio did not differ significantly from that found for the normal

A

intraganglionic synapses (1.33 : 1). None of these synapses were found on neuronal somata. The synaptic membrane specializations of most of these synapses did not differ markedly from those found in normal ganglia: they were of similar extent and were usually asymmetric, more strongly so in the case of the dendritic shaft than of spine synapses. The symmetry of some of the synapses, both on shafts and spines, was, however, equivocal (e.g. Figs 17, I8 and 22). Occasionally a form of subsynaptic apparatus was present (e.g. Fig. 21). The small dense-cored VCPs and the synapses formed by them showed a slight tendency to clustering, but the clusters were evenly distributed in the tissue. In tissue which had been treated with 5-OHDA the morphology of the intra-vesicular deposits varied considerably (Figs 17, 19 and 20). Sometimes it formed a centrally placed core; sometimes it appeared as discrete lumps around the periphery of the vesicle, or as an even, dense halo around the inside of the vesicular membrane, or as a single but eccentrically placed clump; all forms of deposits could be found in the vesicles of the same VCP. Occasionally the vesicles labelled by 5-OHDA appeared to have swollen slightly; these were distinct from large dense-cored vesicles, which were only occasionally found in the small dense-cored VCPs. Such large dense-cored vesicles, which were reminiscent of but less numerous than the large dense-cored vesicles of normal preganglionic endings, usually did not seem to be distorted by 5-OHDA, although their cores could be very dense (Figs 17, 19, 21 and 22). In these chronically denervated ganglia, small dense-cored vesicles also continued to be found in the profiles which innervated arterioles (e.g. Figs 23 and 24) and amongst the clusters of small vesicles in the neuronal dendritic and somatic cytoplasm; the morphology of the cores, and of the deposits after 5-OHDA treatment in these vesicles, was identical to that seen in the small dense-cored VCPs (Fig. 20). The labelling of vesicles with 5-OHDA greatly facilitated the identification of small dense-cored VCPs and of somatic and dendritic vesicle clusters during the ultrastructural survey of the tissue (cf. Figs 1619). For this reason, only tissue treated with this label was used for quantitative observations. The small dense-cored VCPs were judged to be axonal varicosities for the following reasons: they were frequently seen in continuity with thin axon-like profiles containing microtubules (Figs 19 and 21); they were never seen to contain ribosomes or RER, which characterize dendrites, although they fre-

l_l_L B

C

D

Fig. 15. This histogram demonstrates the contribution made by the median incidences of various synaptic elements to the total incidence of all synaptic sites in the four groups of ganglia investigated, i.e. in control (A), acutely denervated (B), chronically denervated (C) and chronically denervated and axotomized ganglia (D). Obliquely hatched regions, regular clear vesicle synapses; dense cross-hatching, small dense-cored vesicle synapses; blank regions, isolated PSDs.

1015

quently contained mitochondria, smooth endoplasmic reticulum and strands of tubulovesicular material; and only rarely was a branchlet of a dendrite found to contain small dense-cored vesicles in such a topographical arrangement that, if sectioned through the vesicular region, it would have been likely to be classified as a small dense-cored VCP

(Fig. 25).

1016

D. A. RhMsAvand M. R. MATTHEWS

There was a statistically significant association between the incidence of small dense-cored VCPs and the post-operative survival interval, both fi,r data from the chronically denervated ganglia alone (P < 0.05, Spearman) and for data from acutely and chronically denervated ganglia. taken together (P < 0.01, Spearman; Fig. 28). This is consistent with the conclusion that small dense-cored VCPs and their attendant synapses appear in consequence of long-term preganglionic denervation and with the observation that in the rat such structures arc very infrequent or absent in the normally innervated KG. (ii)

Regular clear ~esjcie-co~tajn~ng profiies

Fifteen percent of all the VCPs in the chronically denervated, 5-OHDA-treated tissue did not exhibit small dense-cored vesicles, but contained vesicles typical of the regular clear VCPs in control ganglia; their median incidence was barely 2% of that in normal ganglia but, as noted earlier, a high proportion (median, 73%) of these profiles showed synapses in the plane of section (Figs 26 and 2’7; Table 2). These VCPs and their associated synapses were patchily and unevenly distributed, typically occurring in small clusters of four or five in a section face, IocaIly in a grid square. Isolated postsynaptic densities

(iii)

Isolated membrane densities of postsynaptic type were also found in these chronically denervated ganglia, though less frequently than in acutely

9

5

2 R AcutelyIa) and chronically (rt denervated ganglia . I1.

I

denervatcd ganglia (Table 3): the ditference bctwccu these incidences was however not found to bc stats+ ticaily significant (P = 0.1 1. two-tailed).

A limited quantitative analysis of the effect of preganglionic denervation on the innervation of ganghonic arterioles was carried out. For this, three tissue levels were cut at IOO-pm intervals and each arteriole encountered in the sections was inspected in the electron microscope for the presence of adjacent small dense-cored VCPs (located comparably with those illustrated in Figs 23 and 24) which were counted. The outer circumference of the smooth muscle coat of the blood vessel was measured, using a graticule engraved on the fluorescent screen of the microscope, to allow the incidence of small densecored VCPs to be expressed per micrometre of bloodvessel circumference. This procedure yielded 0.034 small dense-cored VCPs/gm in control (10 vessel profiles in one animal), 0.0067~~m in acutely denervated ganglia (22 vessel profiles in 2 animals) and 0.063/pm in chronically denervated ganglia ( I6 vessel profiles in 2 animals, 1X5 and 189 days’ postoperative survival). The only arteriolar profiles to show an innervation in acutely denervated ganglia were restricted to regions adjacent to or within the ganglionic capsule. These preliminary findings suggest that the innervation of intra-ganglionic arterioles arises from sympathetic neurones situated jn ganglia caudal to the superior cervical ganglion and enters the ganglion via the cervical sympathetic trunk. The exceptions to this are arterioles lying in or close to the ganglionic capsule, which have been found in the cat, dog and rabbit” to be supplied by post-ganglion~c sympathetic fibres accompanying those branches of the external carotid artery which supply the ganglion; such fibres presumably arise from the neurones of the SCG itself. These findings further suggest that the denervated arterioles are then reinnervated, probably by the same process as that involved in the innervation of ganglionic neurones by small dense-cored VCPS. Axotomized-denervated ganglia

I

0

50 loo 150 200 POST-OPERATIVE tNTERVAL (daysl

Fig. 28. This graph shows the relationship between postoperative survival interval (abscissa) and the incidence of small dense-cored vesicle-containing profiles (ordinate) in acutely and chronically denervated SC!Gs. Only two points are shown for the 4 acutely denervated ganglia because of the overlap of data from these animals. The data from ail 4 acutely denervated ganglia have, however, been used in the statistical calculations. The dashed line is a regression line which illustrates that there is a signifkant association (P < 0.01, Spearman) between the survival interval and the combined data from acutely and chronically denervated ganglia.

The characteristics of VCPs and of synapses in this group of ganglia were qualitatively similar to those of the chronically denervated ganglia described above. Synapses involving small dense-cored vesicles (Figs 16, 17,21,22) were found which were similar in location and in the extent and nature of their postsynaptic specializations to those found in chronically denervated ganglia. However, when the axotomizeddenervated ganglia were compared quantitatively with the latter, they were found to show a 3.6fold increase in the median incidence of small dense-cored VCPs per nucleated neuronal profile (NNP) and a 3-f&d increase in the median incidence per NNP of small dense-cored vesicle synapses (Table 2; Figs g

Synapse formation in denervated rat sympathetic ganglion

and 15). which brought the incidence of synapses per NNP in these ganglia to 34% of the control incidence. These data were obtained from a survey of 60 IO0 x IOOpm grid squares, which contained a total of 225 nucleated neuronal profiles. The neuropil of this tissue yielded 746 small dense-cored VCPs, of which 201 (27% overall) formed synapses in the plane of section and 44 regular clear VCPs, 33 of which (75% overall) were seen to be synaptic. As in the chronically denervated ganglia, the small dense-cored VCPs and the synapses formed by them showed some clustering but the clusters were evenly distributed and the regular clear VCPs and their synapses were patchily and unevenly distributed, in small clusters. The incidences and features of regular clear VCPs and their associated synapses (Table 2, Fig. 8) and of isolated postsynaptic densities (Table 3, Fig 15) did not differ significantly between the chronically denervated and the axotomized-denervated ganglia, neither did the proportions of PSDs and of small dense-cored vesicle synapses which were found on dendrites and on spine-like profiles (Tables 2 and 3, Fig. 8). When relevant data from the axotomizeddenervated ganglia and the chronically denervated ganglia were combined and compared, a statistically significant association was found between the incidences of non-synaptic and of synaptic small densecored VCPs (P < 0.01, Spea~an; Fig. 29). This is consistent with the assumption that these two classes of profiles were drawn from the same population of synapse-forming varicosities, the difference between the two being in the position of the plane of section with respect to the synaptic site. In both chronically denervated and axotomized-

p

6.0-

j5.0. S

4

x

,/ 9’

,/’ ,,*’

t*

,/

.

4.0

,/’

OC 0

I’

.

denervated ganglia, the overall proportions of small dense-cored VCPs and of regular clear VCPs which showed synapses in the plane of section differed considerably according to the type of VCP and differed from the normal in the case of the regular clear VCPs (see earlier) but did not differ greatly between the two sets of ganglia (Tables 2 and 3).

DISCUSSION

These ultrastructural observations indicate that, in the rat superior cervical ganglion, virtually all ganglionic synapses (apart from those arising from the small granule-containing cells) disappear within 48 h of preganglionic denervation but that in the longer term, in the proven absence of regeneration of the cervical sympathetic trunk, synaptic terminals containing small dense-cored vesicles appear and occupy synaptic sites amounting to approximately 9”/, of the control incidence of synapses per nucleated neuronal profile (on the assumption of approximate equality of areas of synaptic contact). If chronic preganglionic denervation is combined with axotomy of the external and internal carotid nerves (accounting for 7080% of the neurones,ba.70.7’there is a 3-fold increase in the median incidence per nucleated neuronal profile of small dense-cored vesicle synapses in comparison with the chronically denervated ganglia. In addition to the small dense-cored vesicle synapses, a small population of clear vesicle synapses appears in both the chronically denervated and chronically axotomized-denervated ganglia, with approximately the same incidence in each. Chronically denervated ganglia show a slight decline in the incidence of isolated postsynaptic membrane densities, compared with the acutely denervated ganglion, but there is no difference between the chronically denervated and the chronically axotomized-denervated ganglia in the incidence of these structures. The effects of mate de~ervut~off

,/

Y e

0

1017

,

1.5 2.0 2.5 R5 1.0 DENSE-CORED YESICLE SYNAPSES PER NUCLEATED NEPAL PRff ILE

Fig. 29. In this graph the incidence of “non-synaptic” small dense-cored VCPs (ordinate) has been plotted against the incidence of small dense-cored vesicle synapses in chronically denervated and ~otomi~d-denervat~ ganglia. The dashed line has been fitted from regression analysis of the data to illustrate that there is a significant association between the two sets of data (P < 0.01, Spearman).

In most reports of the effects of preganglionic denervation on the superior cervical ganglion there has been a tendency to assume that all synapses found shortly after decentralization are a normal population of synapses which has been unmasked by the loss of the major input from the cervical sympathetic trunk,39.40.54.8’.90.9’.‘~ irrespective of the species in question. The origin of such synapses has been variously attributed to wandering preganglionic fibres which have escaped damage, perhaps entering via the the ganglion postganglionic nerve trunks,sq,ni,~.9’.‘~ to specific interneurones” and to recurrent collaterals from cholinergic and adrenergic post_ganglionic

fibres."".40.54."l."~.~n.Yl

The present findings suggest that very few intact synapses remain in the rat SCG at 48 h after preganglionic denervation and that new synapses appear in the absence of regeneration of preganglionic fibres.

1018

0. A. RAMSAYand

Thus, with the exception of synapses arising from the Small granule-containing cells, which have a restricted distribution,‘“,” the normal rat SCG does not seem to contain a significant or widespread POpUlatiOn Of synapses arising from sources other than the cervical sympathetic trunk. A careful study of the literature lends support to this conclusion. Although Raisman et ~1.~’ pooled their data from rat SCGs denervated 5-191 days previously to yield a mean incidence of three “intrinsic” synapses per 400 mesh grid square (i.e. 10% of the incidence of synapses in control ganglia), inspection of these authors’ Fig. 9 shows that samples from three out of four ganglia denervated for 5-10 days contained no synapses. Moreover, in experiments where the regenerated preganglionic nerve trunk was re-cut all synapses disappeared indicating, in their words, that “the synapses in these selected areas (i.e. remote from small granule-containing cells) of ganglionic neuropil are effectively all of extrinsic origin”. Similarly no synapses were found at 2 days after denervation of the neonatal rat SCG, in which synaptogenesis has already begunlo although Smoleniw detected the presence of a few synapses at 4 days after decentralization in the same class of animals. Quilliam and Tamarind” also reported the complete disappearance of synapses from samples of rat SCG denervated 2 days previously, with the appearance of 0.11 and 0.46 synapses per unit area at 3 and 5 days, respectively, after decentralization. 6stberg et ~1.” found the equivalent of 2.4% of the normal incidence of synapses in samples of rat SCGs denervated 4 days previously; and Purves” estimated, using pooled data from guinea-pig SCGs denervated for 2-8 days, that “the small number of ganglionic synapses which remain after denervation” formed approximately 2-3% of the normal incidence of ganglionic synapses. Perri et ~1.‘~ found that acute preganglionic denervation of the guinea-pig SCG abolished the synaptic potentials normally elicitable in most SCG neurones by stimulation of the postganghonic trunks of ganglia with an intact preganglionic input. They concluded that no intraganglionic synapses arise collaterally from post-ganglionic axons in the guinea-pig but that there are preganglionic fibres projecting into the internal carotid nerve which give off collateral terminal branches as they pass through the SCG. The final destination of these fibres is unknown but they may innervate extraganglionic neurones in the post-ganglionic trunks. The apparent inconsistencies in the rat described above may be attributable to differences between strains of rats. A second possibility is that the normal rat SCG does not contain intrinsic synapses, i.e. synapses of intraganglionic origin, apart from those arising from the small, granule-containing cells but that denervation of the ganglion is followed rapidly by reactive synaptogenesis (the possible sources and mechanisms of which will be discussed below). There-

M. R. MATTHEWS

fore it seems that studies which group the postoperative incidences of synapses into broad survival interval groups may fail to reveal a gradual and ultimately incomplete reinnervation of the denervated postganglionic neurones by endings arising from another, presumably intraganglionic source. in the absence of regeneration of the cervical sympathetic trunk. Synapses

involving srnaN dense-cored

vesicles

A further inconsistency between the observations recorded in this study and those of other authors concerns the presence of small dense-cored vesicle synapses in control and normal rat SCG. At no time were such synapses found in the control and normal ganglia surveyed during the present experiments. despite the use of a highly selective label for monoaminergic vesicles (5-OHDA)‘# which was demonstrably effective in the same tissue, as evidenced by the labelling of vesicles in the neuronal somatic and dendritic vesicle clusters and in the vascular adrenergic endings. Neither of the two surviving synapses which were seen in the acutely denervated ganglia contained small dense-cored vesicles or showed tabelling with 5-OHDA. Only in the chronically denervated and axotomized-denervated ganglia were small dense-cored vesicle synapses found and these did become labelled with S-OHDA, which indicated that they were of monoaminergic type. Synapses containing small dense-cored vesicles have however been described in the intact rat SCG,‘“.4”,X’.‘“” “” though in none of these reports was there an illustration of such synapses taken from control or normal ganglia. In addition to the ever-present possibility of variation between different strains of rats. an cxplanation for this discrepancy might be that the occasional close relationship between superficially placed dendritic or somatic vesicle clusters and an adjacent dendrite without a clear-cut postsynaptic specialization has at times been interpreted as a small dense-cored vesicle synapse (e.g. see Ref. 56). Small dense-cored and sites

vesicles in synapses in other speri2s

In some other species there is clear-cut evidence for intrinsic synapses of this type in the SCG. Approximately 25% of the synapses in the normal rabbit SCG are characterized by small dense-cored vesicles2f,‘0J and these synapses remain after preganglionic denervation,2790 which suggests that they are endings either of collateral branches from post-ganglionic neurones or of adrenergic principal neurones which do not project outside the SCG, or of intraganglionic small granule~ontaining cells. Such synapses may also be found in paravertebral sympathetic ganglia of other species. For example, Helen and Hereonen* have described the occurrence of small dense-cored vesicle synapses amounting to some 22% of the ganglionic synapse poputation in human thoracic and lumbar sympathetic ganglia removed during surgical

Synapse formation in denervated rat sympathetic ganglion sympathectomy. Synapses involving small densecored vesicles have also been found in rat uterine paracervical ganglion75 and in guinea-pig pelvic ganglia. “*J” In the normally innervated rat SCG the formation of such synapses appears to be severely restricted or inhibited. Chronically

denervated ganglia

In ganglia denervated for between 41 and 189 days the median incidences of small dense-cored VCPs and of small dense-cored vesicle synapses were 11% and 9% respectively of the median incidences of regular clear VCPs and regular clear vesicle synapses in control ganglia. These figures are similar to the incidences of “intrinsic” synapses reported earlier in chronically denervated rat SCGS.*‘,~’ Although previously, as discussed above, these were interpreted as an unmasked, residual population of normal synapses the present experiments suggest that these small dense-cored vesicle synapses in the rat SCG are newly formed as a consequence of preganglionic denervation. This conclusion is supported by the statistically significant correlation observed between their incidence and the post-operative survival interval. The evidence from previous studies, cited earlier, indicates that these synapses begin to appear soon after denervation, probably within 4-7 days.65 This is consistent with the interval which elapses between neuronal injury and the appearance of sprouts in the SCG following post-ganglionic axotomy64.72 and in skeletal muscle following partial denervation of the muscle’ or tetrodotoxin nerve block.’ Although the normal preganglionic innervation is predominantly cholinergic,60 the small dense-cored VCPs and their related synapses are probably monoaminergic because their morphology resembles that of adrenergic endings found in a wide variety of tissues,‘3,35,95.‘08 including the smooth muscle of the intraganglionic blood vessels and because of their capacity to be labelled by 5-OHDA.“’ Thus denervation of the rat SCG, in the absence of reinnervation by the original fibres, results in the new formation of monoaminergic endings which form novel synaptic contacts with the principal neurones. This seeming paradox is not unique; there have been various other reports of apparent “neurotransmitter mis-match” as the result of denervation or crossreinnervation experiments. For example, Vera et al “I Koslow et aL5’ and Inestrosa et al5 have shown thit the predominantly adrenergic innervation of the nictitating membrane can be functionally replaced by cholinergic fibres of the hypoglossal nerve. Evans et a1.34 found that sympathectomy induced by chronic guanethidine administration in the rat led to permanent expansion for over 1 year of the innervation of the vas deferens by non-adrenergic fibres, “cholinergic” in ultrastructural type, which normally account for only lOaL of its innervation. Koslow et al.” have reported that the cat lumbar sympathetic ganglion may be reinnervated by the predominantly

1019

adrenergic splenic nerve (although contributions from possible cholinergic fibres in the splenic nerve or possible regeneration of the preganglionic trunk were not rigorously excluded). However, in each case of the apparent induction of synapses with novel neurotransmitter properties by a foreign nerve, there is evidence that the targets have a small complement of receptor sites which correspond to those normally activated by the foreign nerve. The nictitating membrane contains a small population of synapses containing small clear vesiclesI and the vas deferens, as already noted, a few non-adrenergic synapses.‘4 In the rat superior cervical ganglion the principal neurones have postsynaptic adrenergic receptors7.7”.28.99 and a certain proportion of themI receive synapses from small, granulecontaining cells7’.“4.“5 which are rich in monoamines, a further proportion probably being exposed to monoamines released non-synaptically from these cells.‘5,66.67 The available evidence suggests that targets can accept supernumerary synaptic contacts from an unaccustomed, “inappropriate” source, despite the fact that the predominant normal innervation of the target employs a neurotransmitter different from that of the foreign axons, provided that the target under normal circumstances expresses the receptor types of the incoming foreign nerve fibres. This may apply irrespectively of whether or not the receptors are normally related to a postsynaptic specialization. In addition, in the absence of reinnervation by normal fibres, foreign fibres may apparently be capable of inducing an increase in the relevant number of receptor sites expressed, beyond what would be normal for the tissue in question. This applies both in the case of the nictitating membrane reinnervated by cholinergic nerve fibres16 and in the vas deferens after guanethidine treatment.34 Neither the vas deferens nor the nictitating membrane exhibits conventional postsynaptic specializations; in the rat SCG a slightly more complex mechanism may be in operation because the novel synapses show postsynaptic specializations and these might have arisen either from previous synapses or from previously extra-synaptic sites. Their relative diversity, in terms of the postsynaptic membrane densities, suggests that both sources might contribute. It is not surprising that these sympathetic neurones can accept supernumerary monoaminergic synaptic connexions because similar synapses can form in tissue culture between explanted adult SCG neur0nes.r’ In the guinea-pig pelvic ganglia in sifu, where there is a small normal population (lo/,) of adrenergic synapses, “‘Yokota and Burnstock”” have found that these synapses increase 8-l O-fold following various extrinsic denervations. The outcome of tissue-culture experiments involving SCG neurones is also relevant to a consideration of the origin of the few synapses with cholinergic features which appear in the chronically denervated SCG. These latter synapses, arising in the absence of regeneration of preganglionic nerve

1020

D. A. RAMSAY and M. R. MATTHEWS

fibres, could result from either of two mechanisms: first. they might develop through collateral sprouting from the postganglionic axons of cholinergic principal neurones. which make up an estimated 57, of the population of SCG neurones;“” second, they might represent endings from SCG neurones which had become converted to a cholinergic state, by analogy with the cholinergic synapses which are formed between dissociated SCG neurones maintained in tissue culture under appropriate conditions.52.51ii.‘“‘q The adrenergic to cholinergic transformation of cultured, dissociated neonatal rat SCG neurones (most of which are already adrenergic at the time of birth) is promoted by co-culture with non-neuronal cells)‘* or by media previously conditioned by various cultured mesodermal cell lines.*’ Despite the well-documented possibility of this transformation in dissociated sympathetic neurones in tissue culture, however, it is unlikely that such a phenomenon could account for the few regular clear vesicle endings which appear in decentralized adult ganglia because SCG neurones progressively lose their capacity for cholinergic expression with increasing ages’ and the rats used in the present study were over the critical age beyond which such transformation occurs much less readily. Moreover, ganglia decentralized in oiuo still have access to their normal targets. Hill and Hendry were unable to demonstrate any in ciuo change in intrinsic choline acetyltransferase activity in SC@ of neonatal and young rats decentralized 14-21 days previously. This would not preclude a slight increase in the activity of the enzyme with longer survival periods, in accordance with the appearance here recorded of a few cholinergic-like synapses in chronically denervated ganglia but this is not inconsistent with the normal presence of a small proportion of pre-existing cholinergic neurones. Even in axotomizeddenervated ganglia, in which an adrenergic to cholinergic shift might be more likely on account of target deprivation, the incidence per neurone of regular clear vesicle synapses was not significantly different from that in chronically denervated ganglia. The most likely source of the nerve endings and synapses containing regular clear vesicles in these two sets of ganglia is therefore collateral sprouts arising from the post-ganglionic axons of the small population of cholinergic principal neurones. Their scanty, patchy distribution would be consistent with such a hypothesis. The high overall incidence of synaptic VCPs, as a proportion of total VCPs (62-75x), suggests that there is generally no shortage of suitable postsynaptic sites. An alternative possibility might be the expansion or new formation of a system of synaptic collateral branches from any sensory nerve fibres traversing the ganglia.68.“9

The source of small dense-cored vesicle-containing profiles in chronically denervated and axotomizeddenervated ganglia Three sources could theoretically contribute to the pool of small dense-cored VCPs. These are (1) the sympathetic nerve plexuses innervating the intra- and periganglionic vasculature, (2) the post-ganglionic sympathetic axons and (3) the small, granulecontaining (small, intensely fluorescent) cells. The vascular, sympathetic nerve plexuses as a source of “intrinsic” synapses. It is unlikely that the adrenergic plexuses around intraganglionic arterioles contribute substantially to the population of small densecored VCPs because it was found that most of these vessels are denervated by acute preganglionic nerve section. The innervation of these vessels represents a second difference between the rat and the rabbit SCG, since the latter, as well as containing a norm?\ population of “intrinsic” monoaminergic synapses,

also has an intrinsically innervated vasculature.” It is. however, possible that some small dense-cored VCPs in the denervated rat ganglion might arise from the surviving sympathetic terminal plexuses upon the blood vessels in the ganglionic capsule or even from those of the adjacent carotid artery and its main branches, provided that the intact capsule does not form an effective barrier. This would constitute collateral terminal branching of postganglionic sympathetic fibres. Such a phenomenon has been described in the rat central nervous system, where partial cholinergic denervation of the hippocampus by a medial septal lesion is followed by invasion of the denervated regions by catecholamine-fluorescent fibres arising from the perivascular nerve plexuses of local pial blood vessels, originating from the superior cervical ganglion;2’,6’ whether these sprouts from sympathetic fibres form synapses has yet to be shown on ultrastructural and physiological criteria.“,’ Collateral sprouting ,from post -ganglionic, axons By analogy with the neuromuscular system,“’ and in line with what has been postulated above for the regular, clear VCPs and their associated synapses, denervation of the SCG may cause collateral sprouting of the intraganglionic portions of the adrenergic post-ganglionic axons, which may develop endings containing small dense-cored vesicles and then establish contact with principal neurones, constituting a system of recurrent collaterals upon themselves or upon other members of the ganglionic neurone population. If this is the predominant source of such endings it is in some sense surprising that so few synapses are formed, since the rat SCC has in the region of 5 times as many neurones (approximately 31,000) and therefore post-ganglionic fibres as there are preganglionic fibres.*‘,‘(” It may be, however, that the axons near the cell body are relatively less responsive than the terminal arborizations to factors which encourage collateral and terminal sprouting. In favour of the possibility of such collateral synapse formation, Purvesx9 demonstrated a 3-fold increase in the number of guinea-pig ganglionic neurones which could be trans-synaptically driven by stimulating the post-ganglionic nerves at 4-5 months after preganglionic denervation, when compared with the number of neurones thus drivable after 2-i I days. The ultrastructural incidence of synapses in the chronically denervated ganglia in his study was 9 times greater than in the acutely denervated ganglia. Purves suggested that these synapses arose from sprouts formed by the “residual” intraganglionic terminals found at two to seven days after denervation. If the synapses observed by Purve? after chronic denervation of the guinea-pig ganglia correspond to those found in the rat under similar circumstances, this implies t&t monoamintrgic syna’pses might have an excitatory action on the ganglionic neurones. This is not entirely as catecholamines operating through various unlikely.

Synapse formation in denervated rat sympathetic ganglion

1021

may have both excitatory7”.‘z.‘x and effects on neuronal activity in the SCG. The fast excitatory postsynaptic potentials observed by PurvesH” were, however, of a nature which suggested a cholinergic excitatory transmitter. This raises the possibility that the small dense-cored VCPs may contain a dual transmitter system, as is characteristic of some of the functional synapses formed by SCG neurones in cholinergicallyconditioned tissue culture.v’.5” This, i.e. a degree of adrenergic-cholinergic dual function or shift on the part of some of the SCG neurones, rather than the loss of vesicular contents or variable degrees of amine filling,‘“scould explain why some of the small dense-cored VCPs in ganglion pieces not incubated in S-OHDA contain a variable population of clear vesicles. The fast excitatory postsynaptic potentials evoked by Purvess’ on stimulation of the post-ganglionic trunks of denervated ganglia might also be accounted for if

atypical synapses on the principal neurones; their scanty and focal distribution in the ganglion makes it improbable that they could have accounted for a widely dispersed synapse population equivalent to approximately 9% of the normal ganglionic synapse population. They normally provide somewhere in the region of 4 x 10’ synapses,15 i.e. not more than about 0.1% of the estimated 3 or 5 x lo6 intraganglionic synapses.‘0’,‘02When, in addition, the relatively much greater incidence of small dense-cored vesicle synapses in denervated-axotomized ganglia is taken into account, then the postganglionic adrenergic axons appear a much more likely source than the SG cells for these novel synapses.

the guinea-pig SCG contains a larger population of cholinergic neurones than does the rat, with the consequence that more cholinergic collateral sprouts might arise after preganglionic denervation.

Factors which may be involved in sprouting in the superior cervical sympathetic ganglion

mechanisms

inhjbitoryl7.?0.?X

Small, granule-containing cells as a source of small dense-cored vesicle-containing projles. These cells

with their high content of catecholamines are unlikely to contribute significantly to the pool of small densecored VCPs in chronically denervated ganglia for several reasons: at no time were small dense-cored VCPs seen to be in continuity with small, granulecontaining cells, or to contain the arrays of peripherally located, large dense-cored vesicles typical of these cells and their processes in the rat SCG (e.g. Ref. 71); and areas with small granule-containing cells were designedly avoided during the present experiments, no observations being made within 100 pm of identified small, granule-containing (SG) cells. These normally have only short processes7’s”s although their response to long-term deafferentation is not clearly known. The synaptic VCPs differed in morphology from the efferent synapses of SG cells and were frequently seen in continuity with narrow, axon-like structures, which are not typical of SG cells. Moreover, SG cells do not exhibit a highaffinity uptake for catecholamines’07 but the small dense-cored VCPs regularly became labelled by 50-OHDA. Dail and Evan26 reported an apparent increase in the extent of the catecholaminefluorescent varicose nerve fibre plexuses in the denervated rat inferior mesenteric ganglion but were unable to define their origin and could not establish whether the increment was due to accumulation of catecholamines in pre-existing nerve processes or to sprouting of nerve fibres in response to denervation. There is some evidence that SG cells in the rat SCG atrophy on denervation (Case and Matthews, submitted for publication). After post-ganglionic axotomy these cells initially lose their efferent synapses to the principal neurones, as do the preganglionic nerve fibres, but both subsequently tend to regain them”.” (Case and Matthews, submitted for publication). It is on balance unlikely that SG cells may have reacted to denervation in an unexpected way by producing long, axon-like processes which formed small dense-cored VCPs and established additional, morphologically

The higher incidence of nerve endings containing small dense-cored vesicles in the axotomizeddenervated than in the chronically denervated ganglia may be due to the heightened operation of a sprouting mechanism common to both experimental groups or to the appearance of an additional influence in the axotomized ganglia. It is therefore relevant, in assessing the potential sources of stimuli for ganglionic sprouting, to consider separately the factors which are common to the two experimental groups and those which are peculiar to the axotomized ganglia. Denervationand decentralization.In both sets of experimental ganglia the removal of the normal preganglionic input to the neurones may lead, through a change in their metabolic machinery, to the elaboration of regenerative factors or “retrophins”“’ whose function is to elicit growth of, and connexions from, any responsive nerve fibres to which they gain access. There is considerable evidence for the increased production of sprouting factors from denervated target tissues, for example from the denervated irisso and they may be, though they are not necessarily always, transmitter-specific. The iris, which is normally innervated by both adrenergic and cholinergic fibres, produces both a nerve growth factor (NGF)-like adrenotrophic factorM and a cholinotrophic factor.2.45Any regenerative or sprouting factor produced by the denervated SCG may be cholinotrophic and the ganglion may also be capable of responding to this, primarily through its small population (perhaps 5%) of cholinergic neurones, but perhaps also to a lesser extent non-selectively, Bjorklund and Stenevi” have shown that the catecholamine content of, and the outgrowth of fibres from, SCG implants in the rat hippocampus are greatly increased by cholinergic denervation of the hippocampus via lesions of the septohippocampal pathway. In this respect it may be significant that a relatively high proportion (21.7%) of all VCPs found in the 8 chronically denervated ganglia contained regular clear vesicles and so were probably cholinergic. Denervation is moreover accompanied by degeneration of the severed nerve endings, the products of which may be sufficient in themselves to elicit nerve sprouting. In the neuromuscular system, however, it has been shown that to induce muscle inactivity in the absence of tissue degeneration is sufficient to provoke terminal sprouting from its motor axons: this has been achieved by various means, including presynaptic paralysis,29 nerve conduction block3.* postsynaptic blockade@ and the silencing of motoneurones by spinalization. ” In the chronically denervated SCG, however, any new nerve terminals which originate

I022

D. A.

RAMSAY and

from postganglionic axons must arise as collaterals from their proximal parts, since their terminations lie overwhelmingly or exclusively outside the ganglion. This would be closer to collateral or nodal, as opposed to terminal. sprouting of the motor axons innervating skeletal muscle. Nodal sprouts in striated muscle do not appear when innervated muscle is paralysed and their formation seems to depend on the presence of nerve degeneration products.’ The relative infrequency of small dense-cored VCPs in the chronically denervated ganglion, as compared with the axotomized-denervated ganglion, may be attributable to a relative insensitivity of the proximal parts of intact postganglionic axons to such sprouting factors, probably cholinotrophic, as are elaborated by the denervated. silent postganglionic neurones, or their supporting cells, and to the relatively short-lived presence of degenerating preganglionic terminals. In the axotomized-denervated ganglion the axons of surviving neurones, including the axon stumps of axotomized neurones, may be more strongly stimulated by the superimposition of the greater extent and duration of degeneration associated with axotomy-induced neuronal death. In these axotomized-denervated ganglia. moreover, adrenotrophic nerve growth factors may appear, in addition, from the denervated target tissues. Sprout-inducing efSects of axotomy on neurones of the superior cervical sympathetic ganglion. Axotomy has two major and related effects on the neurones destined to survive: these are isolation of the neurone from its periphery and the induction of a series of metabolic changes in the nerve cell body, including chromatolysis, concerned ultimately with repair and re-growth of the damaged axon.72.“3 Separation from the periphery may act by interrupting the retrograde flow, from the periphery to the nerve cell bodies, of a trophic factor or factor elaborated by the target organs. This view has gained support from observations that axotomy-like changes can be induced by applying colchicine-containing cuffs to the post-ganglionic trunks, this effect being attributed to the interruption of axoplasmic transport,S5~84~86~88 although recent experiments suggest the need for caution in the interpretation of these results9* In the adrenergic neurone what is transferred from the periphery may be nerve growth factor or an NGF-like substance: NGF is essential for the in oitro survival of adrenergic neurones; exogenous NGF is selectively taken up by adrenergic nerve endings and retrogradely transported to the cell body46.‘03and the effects on the soma and dendrites of post-ganglionic axotomy can be partly prevented by local application of NGFSs,” and mimicked by treatment with anti-NGF antiserum” or by rendering an animal autoimmune to NGF.‘* The regeneration of severed postganghonic sympathetic axons is accentuated by exogenous NGF4 and reduced by anti-NGF antiserum.5 Adrenergic neurones have receptors for NGF on their somata and somatic uptake has been demonstrated after the destruction of their terminals by 6-hydroxydopamine.6’ In tissue culture, low concentrations of NGF (I ng/ml) suffice for the survival and maintenance of adrenergic neurones and slightly higher concentrations (IO ng/ml) elicit maximal sprout formation from the neurones; at greater concentrations ( > 100 ng/ml) fewer, though longer, outgrowths are formed and still higher concentrations permit the formation of transmitter-specific enzymes.“’ In the normal adult state in vivo, NGF may reach the neurones by retrograde axonal transport at quite high concentrations, sufficient for maintenance of an equilibrium level of transmitter formation; if the target is denervated it may release increased quantities of NGF which may reach the neurones by diffusion to the severed axons (if the distance is not great) or reach their somata by way of the bloodstream. in

M. R.

MATTHEWS

concentrations appropriate to induce or reinforce sproutmg Additional sources of sprouting factors could be the Schwann cells of the degenerating nerves and the ganglionrc non-neuronal cells, which proliferate in the vicinity of the chromatolytic neurones.” These cells m tissue culture t’xert a cholinergic conditioning effect on the neuroncy. In the axotomized-denervated ganglia. unlike the chronically denervated ganglia, only 5.9”,; of all VCPs found contained regular clear vesicles and the remainder contained small dense-cored vesicles. These figures are close to the relative proportions of presumed cholinergic and adrenergic neurones in the rat superior cervical ganglion”’ and they suggest that adrenergic and cholinergic sprout-inducing mechanisms may be equally active throughout the neurone population in the axotomized-denervated ganglia (if it is assumed that the axotomy, and the consequential loss of neurones, affect the cholinergic and adrenergic neurones in equal proportions). Formation of new svnapses But vesicle-containing profiles are more than merely sprouts, they are nerve terminal profiles and are closely correlated with synapse formation. The relative incidence of synapses in the two groups of ganglia differed according to the nature of the synapse. A few, presumably cholinergic synapses containing regular clear vesicles were formed in each group, amounting in either case to approximately 2.5% of the control incidence of synapses per nucleated neuronal profile. The remainder of the synapses which formed were of the adrenergic type, containing small dense-cored vesicles pre-loadable with 5-OHDA, equivalent in the chronically denervated ganglia to 9.5% and in the axotomized-denervated ganglia to 307: of the control incidence of synapses per nucleated neuronal profile. While the major limiting factors in sprout formation are likely to be the levels and types of sprout-inducing factors and the responsiveness of available neurones, the limiting factors in synapse formation must include also the availability of suitable postsynaptic sites. Since denervated ganglia contain large numbers of potential cholinergic postsynaptic sites, and can become fully”’ and remarkably accurately” re-innervated if the preganglionic axons regenerate, it is likely that the formation of cholinergic-like synapses in both groups of ganglia is limited by the availability of a source of appropriate terminals, presumably of cholinergic neurones. The equivalent incidence but patchy distribution of such synapses in these two groups of ganglia, and the high median proportions of the corresponding VCPs which were seen to be synaptic (62% and 75%, respectively. as compared with 47% in control ganglia), would be consistent with a scanty source and an abundance of suitable targets. The 3-fold greater incidence per neurone of synapses containing small dense-cored vesicles in axotomizeddenervated than in chronically denervated ganglia suggests that any or all of the limiting factors (the

Synapse formation in denervated rat sympathetic ganglion

presence of appropriate sprout-inducing factors, the responsiveness of neurones and the availability of suitable postsynaptic sites) may be at a higher level in the axotomized-denervated ganglia. Altered, even optimal, levels of adrenergic sprouting factors might be expected as a consequence of target denervation. The surviving axotomized neurones, lacking targets, should be directly and selectively responsive to the levels of these, and this response would be superimposed upon any endogenous sprout-inducing consequence of axotomy. The incidence per neurone of small dense-cored VCPs was 3.6 times greater in the axotomized-denervated than in the chronically denervated ganglia, but the median proportions of these VCPs which were seen to be synaptic in each group did not differ by so much (27%, 35%): this suggests that, while suitable postsynaptic sites may not have been unrestrictedly available, there was approximately the same balance between terminal profiles and receptive sites in the two classes of ganglia. This apparently near-parallel matching of synaptic termi-

1023

nal profiles to postsynaptic sites might be a consequence of inductive interaction between the two elements such that the one does not mature or develop, in the absence of the other. The present findings do not, however, exclude the possibility that axotomy, in addition to enhancing intraganglionic sprout formation, may also heighten the capacity of the post-ganglionic neuronal membrane to form new adrenoreceptive postsynaptic sites as the neurone recovers, either de novo or by organizing a postsynaptic specialization in association with adrenergic receptors already present extra-synaptically on the neuronal membrane; there are reasons for supposing that such an effect, mutatis mutandis, could apply also to receptivity for the “foreign” axons of a somatic motor nerve.93 Acknowledgements-The

authors wish to thank Mr. P. J. Belk for technical assistance, Miss J. Lloyd and Mrs J. Kilcoyne for photographic work and Miss J. Ballinger for secretarial help.

REFERENCES 1.

Acheson G. H. and Schwarzacher H. G. (1956) Correlations between physiological changes and morphological changes resulting from axotomy in the inferior mesenteric ganglion of the cat. J. camp. Neurol. 106, 247-265. 2. Adler R., Landak B., Manthorpe M. and Varon S. (1979) Cholinergic neurotrophic factors: intraocular distribution of trophic activity for ciliary neurons. Science, N. Y. 204, 14341436. 3. Betz W. J., Caldwell J. H. and Ribchester R. R. (1980) Sprouting of active nerve terminals in partially inactive muscles of the rat. J. Physiol., Lond. 303, 281-299. 4. Bjerre B., Bjiirklund A. and Mobley W. (1973) A stimulatory effect by nerve growth factor on the regrowth of adrenergic nerve fibres in the mouse peripheral tissues after chemical sympathectomy with 6-hydroxydopamine. Z. Zelljorsch. mikrosk. Anat. 146, 15-43. 5. Bjerre B., Bjorklund A. and Edwards D. C. (1974) Axonal regeneration of peripheral adrenergic neurones. Effects of antiserum to nerve growth factor in mouse. Cell Tissue Res. 148, 441-476. 6. Bjorklund A. and Stenevi U. (1979) Regeneration of monoaminergic and cholinergic neurones in the mammalian nervous system. Physiol. Rev. 59, 62-100. 6a. Bowers C. W. and Zigmond R. E. (1979) Localization of neurons in the rat superior cervical ganglion that project into different postganglionic trunks, J. camp. Neurol. 185, 381-392. 7. Brown D. A. and Caulfield M. P. (1979) Hyperpolarizing “ar”- adrenoceptors in rat sympathetic ganglia. Br. J. Pharrnac. 65, 435-445.

7a. Brown D. A. and Dunn P. M. (1983) Depolarization of rat isolated superior cervical ganglia mediated by /I-adrenoceptors. Br. J. Pharmac. 79, 429-439. 8. Brown M. C. and Ironton R. (1977) Motor neurone sprouting induced by prolonged tetrodotoxin block of nerve action potentials. Nafure 265, 459-461. 9. Brown M. C and Ironton R. (1978) Sprouting and regression of neuromuscular synapses in partially denervated mammalian muscles. J. Physiol., Lond. 278, 325-348. IO. Brown M. C., Holland R. L. and Hopkins W. G. (1981) Motor nerve sprouting. A. Rev. Neurosci. 4, 17-42. 11. Brown M. C., Holland R. L. and Ironton R. (1980) Nodal and terminal sprouting from motor nerves in fast and slow muscles of the mouse. J. Physiol., Lond. 306, 493-510. 12. Biilbring E. (1944) The action of adrenaline on transmission in the superior cervical ganglion. J. Physiol., Lond. 103, 55-67. 13. Burnstock G. and Costa M. (1975) Adrenergic Neurones: Their Organization, Function and Development in the Peripheral Nervous System. Chapman and Hall, London. 14. Campbell R. C. (1974) Statistics for Biologists (2nd edn). University Press, Cambridge.

15. Case C. P. and Matthews M. R. (1975) A quantitative study of structural features, synapses and nearest-neighbour relationships of small, granule-containing cells in the rat superior cervical ganglion at various adult stages. Neuroscience 15, 237-282. 16. Ceccarelli B., Clementi F. and Mantegazza P. (1971) Synaptic transmission in the superior cervical ganglion of the cat after reinnervation by vagus fibres. J. Physiol., Lond. 216, 87-98. 17. Christ D. D. and Nishi S. (1971) Site of adrenaline blockade in the superior cervical ganglion of the rabbit. J. Physiol., Lond. 213, 107-l 17. 18. Chungcharoen D., de Burgh Daly M. and Schweitzer A. (1952) The blood supply of the superior cervical sympathetic and the nodose ganglia in cats, dogs and rabbits. J. Physiol., Lond. 118, 528-536. 19. Colquhoun E. (1972) Lectures in Biostatistics. University Press, Oxford.

20. Costa E., Revzin A. M., Kuntzman R., Spector S. and Brodie B. B. (1961) Role for ganglionic norepinephrine sympathetic synaptic transmission. Science, N. Y. 133, 1822-l 823.

in

21. Cotman C. W., Nieto-Sampedro. M, and Harris E. W. (X981) Synapse replacement in the nervous system of adult vertebrates. Physiol, Rec. 61, 684-784. 22. Courtney K. and Roper S. (1976) Sprouting of synapses after partial denervation of frog cardiac ganglion. Xu!urr

259, 317-319. 23. Crutcher K. A., Brothers L. and Davis J. N. (1981) Sympathetic noradrenergic sprouting in response to central cholinergic denervation: a h~sto~hemi~al study of neuronal sprouting in the rat hipp~ampal formation. B&n &s. 210, IJS-#28. 23a. Crutcher K. A. and Chandler J. P. (1984) Association of basal Iamina with peripheral axons elongating within the rat central nervous system. Bruin Res. 308, 177 181. 24. Cuello A. C. and lvcrsen L. L. (1978) Interactions of dopamine with other neurotransmitters in the rat substantia nigra: a possible functional role of dendritic dopamine. In Interactions Between Pufutiw Neurorransnzitrers in thP Brain (eds Garattini S., Pujol J. F. and Samanin R.). pp. 127-149. Raven Press, New York. 25. Dail W. G. and Evan A. P. (1978) Ultrastructure of adrenergic terminals and SIF celis in the superior cervical ganghon of the rabbit. &n&r Res. 148, 469-477. 26* Dail W. G. and Evan A. P. (t978) Effects of chronic deafferentation on adrenergic ganglion cells and small intensely fluorescent cells. J. Neurocytol. 7, 25-37. 27. Dail W. G., Khoudary S., Barraza C., Murray H. M. and Bradley C. (1980) The fate of adrenergic fibres which enter the superior cervical ganglion. Adc. Biochem. Psychopharmucol. 25, 281-297. 28. de Groat W. C. and Voile R. L. (1966) The actions of catecholamines on transmission in the superior cervical ganglion of the cat. J. Pharmw, exp. Ther. 154, I- 13. 29. Duchen t. W. and Strich S. J. (1968) The effects of botulinum toxin on the pattern of innervation of’ skefetal muscle of the mouse. Q, 3. e.wp. P&r.riol. 53, 84-89. 286, 30. Ebendai T.. Olson L., Seiger A. and Hedlund K.-O. (1980) Nerve growth faclors in the rat iris. Na~iarure 25-28. 31. Elfvin L.-G. (1963) The ultrastructure uf the superior cervical sympathetic ganglion of the cat. I. ‘The structure of the ganglion cell processes as studied by serial sections. J. ultrastrucf. Res. S, 403-440. 32. Elfvin L.-G. (1963) The ultrastructure of the superior cervical sympathetic ganglion of the cat. II. The structure of the pregangiionic end fibers and the synapses as studied by serial sections. 3. ~~&r~sir~~~.Res. ST441-476. 33. Elfvin L.-G. ft97t) Ultrast~~tura~ studies on the synaptology of the inferior mesenteric ganglion of the cat. Hf. The structure and distribution of the axodendritic and dendrodendr~tj~ contacts. +r. ult~~~t$~~i. Res.37, 432-448. 34. Evans 8. K., Heath J. W. and Burnstock G. (1979) Reinnervation foliowing guanethidine-induced sympathectomy of adult rats. .I. Neurocytol. 8, 381-400. 35 Furness J. B. and lwayama T. (1972) The arrangement and identification of axons innervating the vas deferens of the guinea-pig. J. .&tar. 113, 179-196. 36. Furshpan E. J., MacLeish P. R., O’Lague P. H. and Potter D. D. (1976) ChemicaI transmission between rat sympathetic neurons and cardiac rnyocytes developing in microcultures. Evidence for cholinergic, adrenergic and dual function neurons. Prur. ~atn. Acad. Sci. U.S.A. 73, 42254229. 37. Gabeila G. (1976) The Structure of the Autonomic Nervous System. Chapman and Ha% London. 38, Gorin P. D. and Johnson E. M. (1980) Effects of long-term NGF deprivation on the nervous system of the adult rat: an experimental auto-immune approach. Brain Res. 198, 27-42. 39. Grille M. A. (1965) Synaptic morphology in the superior cervical ganglion of the rat before and after preganglionic denervation. j. Cell &1.q27, 136A. __ Rec. if$ 387-413. 40. Griilo M. A. (1966) Electron microscow of sympathetic tissues. Phmmeoi. 41. Guth L. and Bernstein J. J. (t961) Sele&ity in the re~~ablisbment of synapses in the superior cervical sympathetic ganglion of the cat. &ql NeuroL 4, 59-69. 42. Hamlyn L. H. (1954) The effect of preganglionic section on the neurons of the superior cervical ganglion in rabbits. J. Amt. 88, 184..19I. 43. Hamori J., Lang E. and Simon L. (1968) Experimental degeneration of the preganglionic fibers in the superior cervical ganglion of the cat. Z. ZeNforsch. mikrosk. Amt. 90, 37-52. 44. Helen P. and Hervonen A. (1981) Nerve endings in human sympathetic ganglia. Am. J. Anar. 162, 1f9-130. 45. Hendry I. A. and Hill C. E. ft980) Retrograde axonal transport of target-t~ssue~erived macromolecules. Narure tm, 647-649. 46. Hendry I. A., Stack R. and Herrup K, (1974) Characteristics of the retrograde axonal transport system for nerve growth factor in the sympathetic nervous system. Brain Res. $2, 117-128. 41, Hill C. E. and Hendry I. A. (1976) Differences in sensitivity to nerve growth factor of axon formation and tyrosine hydroxylase induction in cultured sympathetic neurones. Neuroscience 1, 489-496. 48. Hill C. E. and Hendry I. A. (1979) The influence of preganglionic nerves on the superior cervical ganglion of the rat. Netfrosfi Letr. 13, 133-139. 49. Holland R. L. and Brown M. C. ($980) Postsynaptic transmission block can cause terminal sprouting of a motor nerve, Science, Wmh. 20‘7. 649-651. neurotransmittersynthesis in cultured SO. Iacovitti L.. Jolt T. H., Park D. H. and Bunge R. P. (1981) Dual expression autanomic neurons. J. Neurosci. 1, 685-696 51. lnestrosa N. C.. Mendez B. and Luco J. V. (1979) Acetvlcholinesterase like that of skeletal muscle in smooth muscle reinnervated by’ a motor nerve. Nature Zso; 504506. _ 52. Johnson M. I., Ross D., Meyers M., Rees R., Bunge R., Wakshull E. and Burton H. (1976) Synaptic vesicle cytochemistry changes when cultured sympathetic neurons develop cholinergic interactions. Nnrure 262, 308310.

of

53, Johnson M. f., Ross C. D. and Bunge R. P. (19800)Morpboiog~~~ and biochemical studies on the d~ve~opment of cholinergic properties in cultured sympathetic neurons. 11. Dependence on post-natal age. .I. Cell &al, %%,692-704 53a. Johnson M. I., Ross C. D., Meyers M., Spitmagel E. L, and Bunge R. P. (1980) Morphological and biochemical studies on the development of choline&e properties in cultured sympathetic neurons. I. Correlative changes in choline acetyltransferase and synaptic vesicle cytochemistry. J. Cell Viol. 84, 680-691. 54. Job F., Lever J. D.. fvens C., Mottram D. R. and Presley R. (1971) A fine structural and electron histochemiea! study

Synapse formation in denervated rat sympathetic ganglion

1025

of axon terminals in the rat superior cervical ganglion after acute and chronic preganglionic denervation. f. Anat. 110, 181-189. 55. Kessler J. A. and Black I. B. (1979) The role of axonal transport in the regulation of enzyme activity in sympathetic ganglia of adult rats. Brain Res. 171, 415424. 56. Kondo H., Dun N. J. and Pappas G. D. (1980) A light and electron microscopic study of the rat superior cervical ganglion cells by intracellular HRP-labelling. Bruin Res. 1Q7, 193-199. 57. Koslow S. H., Stepito-Klauco M., Olson L. and Giacobini E. (1971) Functional reinnervation of cat sympathetic ganglia with splenic nerve homografts. Experientia 27, 799-801. 58. Koslow S. H., Giacobini E., Kerpel-Fronius S. and Olson L. (1972) Chohnergic transmission in the hypoglossal reinnervated nictitating membrane of the cat: an enzymatic, hist~hemi~l and physioto~cal study. f. ~~ur~uc. exp. Ther. 180, 664-67 1. 59. Lakos I. (1970) Ultrastructure of chronically denervated superior cervical ganglion in the cat and rat. Acta biol. hung. 21, 425-427. 60. Langley J. N. and Dickinson W. L. (1899) On the local paralysis of peripheral ganglia, and on the connexion of different classes of nerve fibres with them. Proc. R. SW. Land. Ser. B 46, 423-431. 61. Levi-Montalcini R. (1976) The nerve growth factor: its widening role and place in neurobiology. Ado. Biuchem. Psychopharmac. 15, 237-250. 62. Madison R., Crutcher K. A. and Davis J. N. (1981) Sympathohippocampal neurones are inside the blood-brain barrier. Bruin Res. 213, 183-189. 63. Maehlen J. and Nj& A. (1981) Selective synapse fo~ation during sprouting after partial deneffation of the guinea-pig superior cervical ganglion. J: Physiol., Lond. 319, 555-567. 64. Matthews M. R. (1973) An ultrastructural study of axonal changes following constriction of postganglionic branches of the superior cervical ganglion in the rat. Phil. Trans. R. Sot. Ser. B 264, 479-508. 65. Matthews M. R. (1974) The ultrastructure of ganglionic junctions. In The Peripherul Nevous System (ed. Hubbard J. I.), pp. 11I-150. Plenum, New York. 66. Matthews M. R. (I 976) Synaptic and other relationships of small granule-containing cells (SIF cells) in sympathetic ganglia. In Chromafin, Enterochromafin and Related Cells (eds Coupland R. E. and Fujita T.), pp. 13l-146. Elsevier, Amsterdam. 67. Matthews M. R. (1978) Ultrastructural evidence for discharge of granules by exocytosis from small granule-containing cells of the superior cervical ganglion in the rat. In Peripheral Neuroendocr~e In~erue~ion (eds Coupland R. E. and Forssmann W. G.), pp. 80-85 Springer, Berlin. 68. Matthews M. R. and Cuello A. C. (1982) Substance P-immunoreactive peripheral branches of sensory neurones innervate guinea-pig sympathetic neurones. Proc. nntn. Acad. Sci. U.S.A. 79, 1668-1672. 69. Matthews M. R. and Cuello A. C. (1984) The origin and possible significance of substance P immunoreactive networks in the prevertebral ganglia and related structures in the guinea-pig. Phil. Trans. R. Sot. Ser. B 304, 247-276. 70. Matthews M. R. and Nelson V. H. (197.5) Detachment of structurally intact nerve endings from chromatolytic neurones of rat superior cervical ganglion during the depression of synaptic transmission induced by post-ganglionic axotomy. J. Physiol., Lond. 245, 91-135. 71. Matthews M. R. and Raisman G. (1969) The ultrastructure and somatic efferent synapses of small granule-containing cells in the superior cervical ganglion. J. Anat. 105, 255-282. 72. Matthews M. R. and Raisman G. (1972) A light and electron microscopic study of the cellular response to axonal injury in the superior cervical ganglion of the rat. Proc. R. Sot. Ser. B 181, 43-79. 73. Matthews M. R. and Ramsay D. A. (1980) Evidence that chromatolysis enhances heterotypic innervation of sympathetic neurones. J. Physiol., Lond. 307, 15-16P. 74. Murray J. G. and Thompson J. W. (1957) The occurrence and function of collateral sprouting in the sympathetic nervous system of the cat. J. Ph@t., Land. 135, 133-162. 75. Mustonen T. and Terlvlinen H. (1971) Synaptic connections of the paracervical (Frankenhguser) ganglion of the rat uterus examined with the electron microscope after division of the sympathetic and sacral parasympathetic nerves. Acta physiol. stand. 82, 264-267. 76. NjH A. and Pitrves D. (1977)Reinnervation of guinea-pig superior cervical ganglion cells by preganglioni~ fibres arising from different levels of the spinal cord. J. Physiol., Lond. 272, 633-651. 77. Nja A. and Purves D. (1978) The effects of nerve growth factor and its anti-serum on synapses in the superior cervical ganglion of the guinea-pig. J. Physiol., Lond. 277, 53-75’. 78. O’Lague P. H., Obata K., Claude P., Furshpan E. J. and Potter D. D. (1974) Evidence for cholinergic synapses between dissociated rat sympathetic neurons in cell culture. Proc. nuns. Acad. Sei. U.S.A. 71, 3602-3606. 79. O’Lague P. H., Potter D. D. and Furshpan I.%J. (1978) Studies on rat sympathetic neurons developing in cell culture. III. Cholinergic transmission. Deuf Biol. 67, 424-443. 80. Olson L. and Malmfors T. (1970) Growth characteristics of adrenergic nerves in the adult rat. Acra physiol. wand. Suppl. 348, l-112. 81 t)stberg A.-J. C., Raisman G., Field P. M., Iversen L. L. and Zigmond R. E. (1976) A quantitative comparison of the formation of synapses in the rat superior cervical sympathetic ganglion by its own and by foreign nerve libres. Brain Res. 107, 445470.

82. Patterson P. H. and Chun L. L. Y. (1974) The influence of non-neuronal cells on catecholamine and acetylcholine synthesis and accumulation of cultures of dissociated sympathetic neurones. Proc. nnfn. Acad. Sci. U.S.A. 71, 3607-3610. 83. Patterson P. H. and Chun L. L. Y. (1977) The induction of acetylcholine synthesis in primary cultures of dissociated rat sympathetic neurons. I. Effects of conditioned medium. Devl Biol. 56, 263-280.

84. Perisic M. and Cuenod M. (1972)Synaptic transmission depressed by colchicine blockade of axoplasmic flow. Science, N. Y. 175, 1140-l 142. 85. Perri V., Sacchi 0. and Casella C. (1970) Synaptically mediated potential elicited by stimulation of post-ganglionic trunks in the guinea-pig superior cervical ganglion. PflzZgers Arch. ges Physiof. 314, 55-67. 86. Pilar G. and Landmesser L. (1972) Axotomy mimicked by localized colchicine application. Science, N.Y. 177, 1116-1118.

1026

D. A. RAMSAYand M. R. MATTHEWS

87. Purves D. (1975) Functional and structural changes in mammalian sympathetic neurones following interruption of their axons. J. Plzysic~/..Lond. 252, 429-463. 88. Purves D. (1976) Functional and structural changes in mammalian sympathetic neurones following colchicine application to postganglionic nerves. J. Physiol.. Lend. 259, 159-175. 89. Purves D. (1976) Competitive and non-competitive reinnervation of mammalian sympathetic neurones by native and foreign fibres. J. Physiol.. Land. 261, 453-475. 90. Quilliam J. P. and Tamarind D. L. (1972)Electron microscopy of degenerative changes in decentralised rat superior cervical ganglion. ~jr~o~ 3, 454472. 91. Raisman G.. Field P. M., C)stberg A.-J. C., lversen L. L. and Zigmond R. E. (1974) A quantitative ultrastructural and biochemical analysis of the process of reinnervation of the superior cervical ganglion in the adult rat. Brain Res. 71, l-16. 92. Raisman G. and Matthews M. R. (1972) Degeneration and regeneration of synapses. In The Structure nnd ~un~tjun of Nervous Tissue, Vol. IV (ed. Bourne G. H.), pp. 61-104. Academic Press, New York. 93. Ramsay D. A. and Matthews M. R. (1984) Clusters of small dense-cored vesicles in sympathetic neurons and the influence on these clusters of denervation and axotomy. J. Anat. 138, 562-563. 94. Rees R. and Bunge R. P. (1974) Morphological and cytochemical studies of synapses formed in culture between isolated rat superior cervical ganglion neurones. J. camp. h’euroi. 157, l-12. 95. Richardson K. C. (1966) Electron microscopic identification of autonomic nerve endings. Nature, Lo&. 210, 756. 96. Sargent P. B. and Dennis M. 3. (1977) Formation of synapses between parasympathetic neurones deprived of preganglionic innervation. Nature 268, 456-458. 97. Siegel S. (1956) Nonparametric Statistics fi>r fhe Behuvioural Sciences (International Student Edition). McGraw-Hill. New York. 98. Singer P. A., Mehler S. and Fernandez H. L. (1982) Blockade of retrograde axonal transport delays the onset of metabolic and morphological changes induced by axotomy. J. Neurosci. 2, 1299-1306. 99. Skok V. I. (1980) Ganglionic transmission: morphology and physiology. In Pharmacology ofGanglionicTransmik~ion. Handbuch der Experimentallen Pharmacologic, Vol. 53 (ed. Kharkevich D. A.), pp. 9-39. Springer, Berlin. 100. Smolen A. J. (198 1) Postnatal development of ganglionic neurons in the absence of preganglionic input: mo~hologica~ observations on synapse formation, &ir!t Brain Res. 1, 49-58. 101. Smolen A. J. (1983) Retrograde transneuronal regulation of the afi’erent innervation to the rat superior cervical sympathetic ganglion. J. Neurocytol. 12, 27-45. 102. Smolen A. and Raisman G. (1980) Synapse formation in the rat superior cervical ganglion during normal development and after neonatal deafferentation. Brain Ref. 181, 31.5-323. 103. St(ickel K., Paravicini U. and Thoenen H. (I 974) Specificity of the retrograde axonal transport of nerve growth factor. Brain Res. 16, 413422. 104. Tamarind D. L. and Quilliam J. P. (1971) Synaptic organization and other ultrastructural features of the superior cervical ganglion of the rat, kitten and rabbit. Micron 2, 204-234. 105. Taxi J. (1965) Contribution a 1’Ctudedes connexions des neurones moteurs du systkme nerveux autonome. Ann/s. SU. nar. (zoot.) 7, 413474.

106. Taxi J., Gautron J. and L’Hermite P. (1969) Don&es ultrastructurales sur une &entuelle modulation adrbnergique de l’activitt du ganglion cervical supkrieur du Rat. C. r. Acad. Sci. Paris, Ser. D 269, 1281-1284. 107. Taxi J. and Mikulajova M. (1976) Some cytochemical and cytological features of the so-called SIF cells of the superior cervical ganglion of the rat. J: Neurocytol. 5, 283-295. 108. Tranzer J. P. and Thoenen H. (1967) Significance of ‘empty vesicles’ in post-ganglionic sympathetic nerve terminals. Experientia 23, 123-124.

localization of 5-hydroxydopamine J. P. and Thoenen H. (1967) Electronmicroscopic 109. Tranzer (3,4,5-trihydroxy-phenyl-ethylamine), a new “false” sympathetic transmitter. Experientiu 23, 743-745. 110. Tranzer J. P. and Thoenen H. (1968) Various types of amine-storing vesicles in peripheral adrenergic nerve terminals. ~~perjentja 24, 484486. 111. Vera C. L., Vial J. D. and Luco J. V. (1957) Reinnervation of nictitating membrane of cat by cholinergic fibres. J. Neurophysiol. 20, 365-373.

Watanabe H. (1971) Adrenergic elements in the hypogastric ganglion of the guinea-pig. Am. J. Anut. 130, 305-330. Watson W. E. (1974) Cellular responses to axotomy and to related procedures. Br. med. Bull. 30, 112-l IS. WiiIiams T. H. (1967) Electron microscopic evidence for an autonomic interneuron. Nature, Land. 214, 309-310. Williams T. H. and Palay S. L. (1969) Ultrastructure of the small neurons in the superior cervical ganglion. Brain Res. 15, 17-34. 116. Yamauchi A., Lever J. D. and Kemp K. W. (1973) Catecholamine loading and depletion in the rat superior cervical ganglion: a formal fluorescence and enzyme histochemical study with numerical assessments. .f. Annt. 114, 271-282. 117. Yokota R. and Burnstock G. (1983) Synaptic organization of the pelvic ganglion in the guinea-pig. Cell Tissue Res.

112. 113. 114. 115.

232, 37&397.

118. Yokota R. and Burnstock G. (1983) Decentralisation of neurones in the pelvic ganglion of the guinea-pig: reinnervation by adrenergic nerves. CeN Tissue Res. 232, 399-411. (Accepted

IO June 198.5)