A quantitative comparison of the formation of synapses in the rat superior cervical sympathetic ganglion by its own and by foreign nerve fibres

Brain Research, 107 (1976) 445--470

445

© Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

Research Reports

A Q U A N T I T A T I V E C O M P A R I S O N OF T H E F O R M A T I O N OF SYNAPSES I N T H E R A T S U P E R I O R C E R V I C A L S Y M P A T H E T I C G A N G L I O N BY ITS O W N A N D BY F O R E I G N N E R V E FIBRES

A.-J. C t)STBERG, G. RAISMAN, P. M. FIELD*, L. L. 1VERSEN AND R. E. ZIGMOND** Laboratory of Neurobiology, National Institute for Medical Research, Mill Hill, London, N W 7 1.4A and MRC Neurochemical Pharmacology Unit, Cambridge, CB2 2QD (Great Britain)

(Accepted October 7th, 1975)

SUMMARY

The rat superior cervical sympathetic ganglion (SCG) has about 36,000 neurones in a volume of about 1 cu.mm. There are about 8.8 x 106 synapses, and 6000-9000 preganglionic axons. Section of the preganglionic chain causes a loss of 93 ~ of the synapses. In the denervated SCG there are 0.6 x 106 remaining ('intrinsic') synapses, and a proportion of the synaptic sites are identifiable as vacated synaptic thickenings (3 x 106 per SCG, as compared with 0.5 x 106 in the normal intact SCG). After deducting the intrinsic synapses, this indicates that each preganglionic axon forms about 1100 (900-1400) synapses. After freezing the preganglionic chain, subsequent axonal regeneration restores synapse numbers to 85 ~ of normal (7.5 x 106 synapses per SCG). After anastomotic repair by suture of the cut ends of the preganglionic chain (a necessary control for the foreign nerve anastomoses), the SCG contains only 60 ~ of the normal complement of synapses (5.2 x 106 synapses per SCG). The results of this anastomosis are very variable. However, in individual ganglia the numbers of synapses are directly correlated with the numbers of axons which reach the SCG. After deducting the intrinsic synapses it can be calculated that each axon forms about 700 synapses. This is probably an underestimation of the numbers which would be achieved at longer survival times. After anastomosis of the vagal nerve into the denervated SCG there are about * Research Training Fellow of the Foundations Fund for Research in Psychiatry. Alfred P. Sloan Foundation Research Fellow.

**

446 4.4 ~ 106 synapses per SCG. Morphologically the majority have anon terminals w~th large dense cored vesicles, and it is likely that these belong to the axons of the parasympathetic preganglionic neurones in the dorsal motor nucleus of the vagu~. A smaller population of axon terminals are devoid of large dense cored vesicles; their origin is unknown. The dorsal motor nucleus of the vagus has between 1000 and 2000 neurones. After deducting the intrinsic synapses, this indicates that each axon may form up to 1900-3800 synapses. To the extent that other, unldent~fied vagal fibres also contribute to the synapses found after this anastomosis, this figure Is an overestimate. After anastomosis of the hypoglossal nerve into the denervated SCG, there are 1.5 × 106 synapses per SCG. A morphologically distractive type of axon terminal is found, and it is argued that this may belong to a special category of skeletomotor neurones located in the caudoventral part of the hypoglossal nucleus and distinguished by pseudocholinesterase staining. There are about 600 of these neurones, which would indicate that they form about 1500 synapses per axon (after deducting the numbers of intrinsic synapses). The majority of the hypoglossal neurones do not form intragangliomc synapses; this suggests that although the possession of a cholinergic mechanism may be necessary for axons to be able to form ganglionic synapses, it is not in ~tself sufficient. For each of the types of anastomosis, the numbers of vacated thickenings are reversely proportional to the numbers of synapses. This indicates that regardless of whether the terminals belong to the normal sympathetic axons or any of the foreign axons, the synapses are always formed at the original postsynaptlc sites m the SCG. For each category of fibres, the number of synapses that each axon can form ~s limited, and in the normal SCG the preganglionic axons probably form close to their maximum number of synapses. The results of the different anastomoses suggest that only compatible axon types will form synapses (absolute specificity). Between different types of compatible axons, there ~s no indication of any relative differences m synaptogenic effectweness : foreign axons form at least as many synapses per axon as do the normal preganglionic sympathetic axons. The function of the foreign synapses has been examined by measuring various biochemical parameters - - choline acetyltransferase activity in the SCG and associated nerves, the reserpine-induced increase in ganglionic tyrosme hydroxylase, and the dark-induced rise in pineal serotonin N-acetyltransferase.

I NTRODUCTION

In a previous ultrastructural study of the superior cervical sympathetic ganglion (SCG) in the rat 40 it was found that estimates of the numbers of synapses m the ganglion can be used to follow the course of the processes of denervation and reinnervation. After interrupting the preganglionic axons, the terminals undergo rapid degeneration and removal, so that by two days after operation there are only about 7 ~ of the original number of synapses left in the ganglion. A proportion of the denervated synaptic sites are recognizable by virtue of the vacated synaptic thickenings. The cut axons

447 regenerate back to the ganglion, and within about 3 weeks, synapses begin to reappear; vacated thickenings become less numerous. Reinnervation is complete by about 3 months after operation, at which time the numbers of synapses reach similar levels to those in unoperated ganglia, and there is no further increase. The vacated synaptic thickenings disappear almost completely. These observations suggest that, over the periods studied, the postsynaptic sites in the SCG persist after deafferentation, and this was confirmed by the observation that even if the preganglionic chain was prevented from regenerating back to the ganglion, so that there was no reappearance of synapses, there was no diminution in the number of vacated thickenings. During the course of the reinnervation, the vacated synaptic thickenings progressively disappear as the new synapses reappear, indicating that the newly formed nerve terminals are reoccupying the previously denervated synaptic sites. Furthermore, the formation of new synapses ceases as the last vacated thickenings disappear, and there is consequently no hyperinnervation beyond normal levels. This suggests that the final density of innervation is determined by the numbers of available synaptic sites, although in this experiment it cannot be excluded that the numbers of regenerating axons may act as a limiting factor. In order to explore further the mechanism of the reinnervation process, we have compared the effects of innervating the denervated SCG with two foreign nerves - the vagus and the hypoglossalas. This sort of crossed reinnervation of the SCG has been studied by several authors (for review, see ref. 16). The cut central end of the cervical vagus will grow into the denervated SCG and establish functional contacts in cats and rats6,S,10Av,2a,a0,al. De Castro s claimed that the hypoglossal nerve also innervated the SCG (in the cat), but Hillarp 23 considered that this nerve could only form very occasional terminal specializations in the SCG in the rat. Our present experiments were designed: (i) to estimate the density of synaptic innervation achieved by these foreign nerves in the rat SCG, and (ii) to test whether the terminals of the vagal or hypoglossal axons can occupy the precise postsynaptic sites left vacant by degeneration of the normal sympathetic preganglionic fibres. Any differences in the mode of reinnervation of the SCG by the vagus or the hypoglossal nerves as compared with each other or with the regenerating preganglionic fibres may throw light on the specificity of matching of pre- and postsynaptic elements required for synapse formation. Some aspects of the hypoglossal anastomoses are described in a separate publication zs. MATERIALS AND METHODS

A total of 162 young adult Wistar rats of both sexes were used (78 for morphological studies, 84 for biochemical studies). The SCG of one side was used to study reinnervation (no systematic morphological or biochemical differences have been found between left and right ganglia) and the other side was left unoperated, denervated, or excised completely. Operations were carried out under tribromoethanol anaesthesia (Avertin, Winthrop, Surrey). For studies of reinnervation the preganglionic fibres were destroyed by direct application of a fragment of dry ice to the cervical

448 sympathetic trunk about 2 mm below the caudal pole of the SCG; this totally destroys all axons but preserves the continuity of the nerve sheath (for details see ref. 40). The foreign nerves were anastomosed into the SCG by cutting the cervical sympathetIc trunk 2 mm below the caudal pole of the ganglion and suturing the cut central end of either the vagus or the hypoglossal nerves with 10/0 polyamide monofilament thread (Ethilon, Eth~con, Scotland) in a loose end to end contact w~th the distal stump of the pregangliomc chain. The anastomosis was sheathed in a sleeve which was prepared by removing from the saphenous vein of the same animal a 10-15 mm segment which was soaked in 1 ,o~ aqueous sodium nimte and threaded over the cut nerve prior to anastomosis. To prevent spontaneous reinnervatlon the cut central end of the cerwcal sympathetic trunk was mobilized, turned down into the thorax and sutured to the external intercostal muscles. The nerves anastomosed into the SCG were (a) the central end of the vagus nerve cut at mid-cervical levels (23 ganglia), (b) the central end of the hypoglossal nerve cut at the point where it crosses the digastric muscle, i.e., including both its medial and lateral branches to the intrinsic tongue muscles, but excluding the contribution to the ansa hypoglossi, which has already left the nerve (15 ganglia), and (c) the cut central end of the cervical sympathetic chain (18 ganglia; as a control for possible effects on reinnervation caused by the surgical procedure necessary for anastomosing the vagus or hypoglossal nerves). Six ganglia with the preganglionic chain t~ed down into the thorax were left uninnervated. After survival periods varying from 4 to 436 days, the ganglia were removed under anaesthesia and those for electron microscopy were fixed by immersion either m 2 To osmium tetroxide in 0. t M phosphate buffer at pH 7.4 at room temperature for 2 h or in a mixture of 1 ° o glutaraldehyde and 1 o formaldehyde m 0.1 M phosphate buffer at pH 7.4 at 4 °C overnight, followed by osmication as above. The ganglia were dehydrated m a graded series of alcohols. Prior to embedding in resin (Taab, Reading, England) the ganglia were divided by transverse cuts into three blocks - - the caudal pole, including a length of the pregangliomc fibres, a middle section including the postganglionlc branches running with the branches of the external carotid artery, and the rostral pole including the postganglionlc fibres running in the carotid nerve, For the study of ganglionic synapses ultrathin sections taken from the middle block were mounted on uncoated 400 mesh grids (squares of 42 ,um side, Micron, EMscope, London). Sections were stained on the grid with 3 ,°i~uranyl acetate in methanol followed by Reynold's lead citrate. Under the electron microscope the section was outlined on squared paper. Every fifth square was scanned systematically and the total numbers of the different types of synapses, vacated synaptic thlckenlngs, desmosomes and other relevant structures were recorded. This method of sampling yields the number of synapses (n) in a sampled area (a). In order to correlate the numbers of synapses in the various normal and experimental ganglia with the numbers of preganglionic axons, it is necessary to arrive at some estimate of the total number of synapses in an entire ganglion. In a total ganglionic volume (V), the synapses are recogmzed by the apposed membrane specializations (presynaptic dense projections, vesicle clusters, postsynaptic density, etc.). For the purposes of estimation, the synaptic appositions are regarded as approximating to

449 uniform, two-dimensional, circular discs of diameter 2r, (estimated as 0.4 #m), in random orientation throughout the entire ganglion. When sections are cut, the synaptic appositions appear on the surface of each section as a number (n) of lines of varying length and orientation. Lines of length less than 2k (estimated as 0.15/~m) cannot be recognized as synaptic appositions. From these data it is possible to calculate the total number (N) of synaptic appositions in the entire ganglion, and small deviations from any of the above assumptions will cause only slight (i.e. proportional) changes in the calculation. When serial sections are cut through a synapse, the fraction (F) of sections in which the synaptic appositions can be recognized is given by the formula: r--d

F--

r

where d is the length of the part of the radius external to a chord of length 2k.

>t

By the chord theorem: (2r - - d)d --= kL Given r ---- 0.2/~m and k = 0.075 #m, then d = 0.0146, and F

--

a/r2--k ~ - -- 0.927. £

Each synaptic apposition will appear in a number (S) of adjacent sections and this number will depend on the section thickness (t), and the angle that the plane of the synlptic apposition makes to the section plane (e.g., when the synapse is at right angles to the section plane it will appear in 2r/t sections). Integrating over all possible angles, each synapse will appear in an average number of sections (S), where Sz~/2 (-~)cos0. dO 0 S

~

(-~) sin z~/2 __

S ~/2 dO

~/2

0.255

4r __

- -

~t

t

o

when the synapse makes an angle of 0 to a line vertical to the section plane. The total number of synapses (N) in the total ganglionic volume (V) is related to the number of synapses (n) counted in an area (a) of a section of thickness (t) by the formula: N

n

n

atFS

a

--

V

X

4 ~ r 2m k 2

450 For a ganghomc volume (V) of I cu.mm (109 cu.#m) N =- n/a .. 4.23

10°, where a ~s in sq.#m.

For estimation of the numbers of preganglionic axons, transverse sections were taken across the upper end of the preganglionic chain at its entry into the caudal pole of the SCG. Ultrathin sections were mounted on coated grids, photographed, and the number of axons deduced by counting the number of axons in a sample area and multiplying up to the total cross-sectional area of the chain (estimated from camera lucida drawings of semlthin sections from the block face). For light microscopy, gangha were fixed by immersion in 70 ° o ethanol w~th 1 ° o acetic acid for 24 h, dehydrated and embedded in paraffin wax. Serial sections of 20 #m were cut, mounted and stained by Bodlan's protargol silver method. The total number of ganghonic neurones and the volume of the SCG were estimated by counts of neuronal nuclei and area measurements from every fifth section 27.

Biochemicalprocedures A series of biochemical assays were carried ou~ on ganglia and pineals from rats which had been subjected to each of the various anastomoses and operative procedures in an attempt to establish whether the connections formed by the vagal and hypoglossal nerves in the ganglion were able to mediate either the reserpine-induced rise in ganglionic tyrosine hydroxylase (TOH) activity 22 or the dark-induced rise in pineal N-acetyltransferase (NAT) activity11, 2~. TOH was assayed by the method of Hendry and Iversen zl as described previously 4°. N A T was assayed in the pineal glands by the procedure of Deguchi and Axelrod 11. In addition, choline acetyltransferase (ChAc) activity (an enzyme contained in the preganglionic axons 4,2°,22) was assayed in the ganglia and nerve trunks 40. Some data on the normal ganglia, denervated ganglia and gangha reinnervated after freeze lesions of the cervical sympathetic trunk were given in this previous publication 40. The ganglia used in the present assays were taken at long survival times after operation (120-160 days), when the anatomical studies show that the process of reinnervatlon has reached completion. Twelve~anlmals were used in each group. Four animals were killed towards the end of the light period (at about 18:00 h; lights on 06:00 h, lights off 20:00 h); the ganglia were taken for assays of TOH and ChAc, and the pineals for NAT. Four animals were killed at 24:00 h (4 h after lights off) and the pineals taken for NAT; the ganglia from these animals were used in the anatomical analysis. Four animals, pretreated 72 h previously with reserpine (5 mg/kg s.c.), were killed at 18:00 h and the ganglia taken for assays of T O H and ChAc. In some animals, small segments of the preganglionic chain and of the postgangllomc axons in the internal carotid nerve were also taken for assays of ChAc. RESULTS

Light microscopy Longitudinal sections through the normal SCG stained with protargol silver

451 TABLE I VOLUMES OF S C G AND TOTAL NUMBERS OF NEURONES IN EACH OF THE EXPERIMENTAL GROUPS

Experiment

Unoperated Unoperated Unoperated Unoperated Freeze lesion Sympathetic anastomosis* Vagal anastomosis Hypoglossal anastomosis Denervated

Survival (days)

115 104 117 82 60

4

UNOPERATED S C G

AND ONE S C G

Volumeof SCG (cu.mm)

No. of neurones × 103

0.87 1.12 0.78 1.20 0.81 0.44 1.02 1.11 1.01

35.5 38.0 29.6 43.9 31.0 20.9 38.2 32.6 38.9

FROM

* This ganglion was found to have suffered d~rect trauma during the operation. showed the preganglionic axons breaking up into progressively smaller bundles and ramifying within the ganglion. Our findings with the vagal and hypoglossal anastomoses are in agreement with De Castro 8 and Hillarp 2a who showed that the axons of these nerves readily cross the anastomoses in large numbers and also ramify extensively within the SCG. In the case of the hypoglossal anastomoses the fibres were markedly thicker than the normal preganglionic axons and large numbers of them could be seen to traverse both the subcapsular space and the interior of the SCG before leaving the ganglion along the course of the postganglionic branches. Estimates of the volume of the SCG and the total numbers of ganglionic neurones were made in 4 normal ganglia, one denervated SCG, one SCG reinnervated after a freeze lesion, and one SCG with each type of anastomosis (Table I). With the exception of the single SCG with the sympathetic anastomosis, where the histology showed that some inadvertent direct trauma had occurred to the lower pole of the SCG, there was no significant change in ganglionic volume after any of these experimental procedures, and no loss of ganglionic neurones. Excluding the damaged ganglion, the mean volume of the other ganglia was 0.99 -4- 0.05 (S.E.M.) cu. mm, and the mean number of neurones was 36.0 ~ 1.7 (S.E.M.) × 103. Electron microscopy Normal ganglia

The ultrastructural analysis of the ganglia concentrated on features associated with the synaptic contacts. In the normal ganglia, the preganglionic axons terminated in a series of interconnecting varicosities, which contained 50 nm synaptic vesicles and a proportion of 100 nm diameter dense core vesicles. Synaptic contacts marked by clusters of the 50 nm vesicles, were established between the varicosities and the postsynaptic elements, which were most commonly small irregular protrusions from the dendritic shafts of the principal ganglionic neurones; comparable postsynaptic pro-

452

3

Fig. I A normal preganghomc anon terminal (with 50 nm synaptlc vesicles and 100 nm dense cored vesicles) making synaptlc contact wJth a dendritic shaft (H). Arrow shows synaptlc thlckemng. Scale bar, 0.5/era Unoperated SCG Fig 2 A small spme-hke profile containing the typically large, somewhat heterogeneous postsynapt~c vesicles and bearing a marked 'vacated' synaptic thickening (arrow) p, process of non-neuronal cell w~th basement membrane abutting on a collagen-containing space (c). Scale bar~ 0 5/~m. Denervated SCG, 200 days survwal Fig 3 A dendritic shaft (H) containing several clusters of vesicular material (asterisks) and a marked 'vacated' postsynaptic thickening (arrow) with an underlying row of about 6 subjunct~onal bodies The vacated thickening Js apposed by a process of non-neuronal cell (p) Scale bar, 0.5/~m. Denervated SCG, 83 days survival. Fig 4. A synapse of the supposedly intrinsic type with the axon terminal containing 50 nm synaptic vesicles of which a proportion show a dense central dot. The synaptlc contact has a vesicle cluster directed towards a postsynaptic thickening (arrow) on a dendritic shaft (H). Scale bar, 0.5/~m. Hypoglossal anastomosis, 436 days surwval Osmium fixation.

453 T A B L E II CALCULATED TOTAL NUMBERS (MEANS -~- S.E.M.) OF SYNAPSES AND VACATED SYNAPTIC THICKENINGS PER GANGLIONIN EACH OF THE GROUPS OF UNOPERATED AND LONG-TERM EXPERIMENTAL S C G Based on the counts f r o m animals with survival times of at least 75 days (59 days m the case of the denervated ganglia), i.e. at a time when the formation of synapses was judged to have reached completion, n, n u m b e r of animals in each group.

Experiment

n

Synapses per ganglion x 106

Vacatedthickenings per ganglion × 106

Unoperated Freeze lesion Sympathetic anastomosis Vagal anastomosis Hypoglossal anastomosis Denervated

6 6 14 14 12 6

8.78 d- 0.17 7.46 ± 0.58 5.21 4- 0.57 4.37 ± 0.25 1.53 :k 0.11 0.63 4- 0.23

0.51 ± 0.14 0.78 4- 0.12 1.48 ± 0.28 1.33 ± 0.08 2.51 i 0 30 2.95 ~ 0.12

trusions also arose directly from the surface of the perikaryon, or in some cases the synapses were directly upon dendritic shafts (Fig. 1). Direct axosomatic synapses were practically absent. The postsynaptic element frequently had a marked synaptic thickening. Postsynaptic elements with synaptic thickenings unapposed by axon terminals ('vacated synaptic thickenings', Figs. 2 and 3) were uncommon in normal ganglia. Symmetrical, desmosome-like contacts were frequently found between neural profiles, and tended to occur especially in the general vicinity of groups of synapses. Occasional synapses occurred (e.g. Fig. 4) in which the axon terminals contained a variable proportion of 50 nm vesicles with a dense central dot ('small dense core vesicles') of the type found in peripheral noradrenergic axons, and these possibly represent intrinsic intraganglionic terminals formed by axons of the principal ganglionic neuroneslL The characteristic small granule-containing cells with their afferent and efferent synapses 36 formed only a very small proportion of the synapses encountered, and they have been excluded from the present analysis. The quantitative study indicated that there were a total of 333 synapses and 13 vacated synaptic thickenings in a total area of 162 x 103 sq./~m from 6 normal ganglia. This figure is slightly higher than in the previously published series 4° and this difference may be due to the present random sampling method as opposed to the previous procedure which selected for synapses closer to the cell bodies. Use of the conversion formula described in Materials and Methods shows that the normal SCG has a total of 8.78 zk 0.17 (S.E.M.) × 10n synapses, and 0.51 dz 0.14 × l0 s vacated thickenings (Table II). The estimates of the total numbers of preganglionic axons in the normal cervical sympathetic trunk in 4 animals were 5980, 5910, 8180 and 9450 (mean = 7380). Denervated ganglia

Section o f the preganglionic trunk caused a loss of 93 ~ of the ganglionic synapses. A proportion of the remaining synapses had small dense cored vesicles (? intrinsic

454 synapses). Vacated synaptlc thlckenings were more common in the denervated gangha than in any of the other experimental groups. In 6 animals with survival periods from 59 to 143 days after operation, 16 synapses and 75 vacated thickemngs were counted m a total area of 108 ~ 103 sq./~m (i.e. 0.63 ± 0.23 ,, 106 synapses per ganghon, and 2.95 ± 0.12 ,< 106 vacated thickenmgs per ganglion; Table l l). No significant differences were seen between the animals w~th the different surwval tzmes.

Reinnervation of the SCG after a freeze lesion of the preganglionic axons Four animals were killed at 4 days after a freeze lesion of the preganglionic chain. As found in the previous study a0, the results were qualitatively and quantitatively comparable to those found in ganglia denervated by surgical section. In the present material there were totals of I0 synapses and 198 vacated thickenings in a total area of 234 × 10z sq.#m from the 4 ganglia. A further 6 animals had survivals of 100-140 days after a freeze lesion. Qualitatively the synapses m these 6 ganglia resembled those of the unoperated SCG. Quantitatively, synapses were almost as common as in the unoperated SCG, and vacated thickenings were infrequent. A total of 346 synapses and 44 vacated thickenings were counted in a total area of 216 × 103 sq.#m, i.e., 7.46 ± 0.58 ",~ 10n synapses, and 0.78 ± 0.12 × 106 vacated thickenings per ganglion (Table II). Reinnervation of the SCG after section and anastomotic union of the preganglionic axons Qualitatively, these ganglia and those with either vagal or hypoglossal anastomoses were distinguished from the previous groups of ganglia by a marked increase in the content of collagen throughout the ganglion. The types of synapses found after anastomosis of the cervical sympathetic trunk were similar to those in normal ganglia. The counts are based on a total of 860 synapses and 353 vacated thickenings in a total area of 812 × 10a sq.#m from 18 ganglia. Quantitatively, the results of this anastomosis were very variable (Fig. 5), ranging from almost complete absence of synapses (as in the denervated ganglia) to a maximum of around 60 ~/oof the level in the normal unoperated SCG. This variation tended to obscure the time course of the reinnervation process (Fig. 10). The numbers of vacated thickenings were higher than in the ganglia from normal animals or in the ganglia reinnervated after freeze lesions (Table II). Within the group of ganglia reinnervated by anastomosis of the preganglionic trunk, vacated thickenings were more numerous in the ganglia with fewer synapses (Fig. 11). Seven ganglia (points enclosed in diamonds in the top panel of Fig. 5) with a wide range of numbers of synapses (from 1.53 × 10n to 6.95 × 106 synapses per ganglion) but a fairly narrow range of survival periods (80-127 days), were selected for estimation of the numbers of preganglionic axons. It was found that the numbers of axons also ranged widely (from 0.77 × 10z to 8.98 × 103). Neither the numbers of synapses, nor the numbers of axons were correlated with the length of the surwval period within this group, but there was a significant correspondence between numbers of synapses and numbers of axons (Fig. 6).

455 ~10. -<

SYMPATHETIC

N6-

Z 4r'-

•

~

.

"-o

VAGUS

(~)-

O"

(~)--

D

8" HYPOGLOSSAL 6"

4,

2! o

N 5'0

16o

•

150

ee

260

•

•

--D 250 360 360 460 450 DAYS POSTOPERATIVE SURVIVAL

Fig. 5. The numbers of synapses in each of the individual experiments. Synapses are plotted as millions per ganglion (ordinate), against the survival time in days (abscissa). The three types of anastomoses are: in the top panel (sympathetic), middle panel (vagal), and lower panel (hypoglossal). To facilitate comparison, the numbers of synapses in the normal ganglia (N) and the denervated ganglia (D) are represented on each panel as broken horizontal lines. The 7 points enclosed in diamonds in the top panel indicate the ganglia which were also used for estimations of the numbers of axons in the preganglionic chain. The two circled points in the middle panel are ganglia in which the anastomosed vagus nerve was cut 6 days before killing.

Reinnervation of the SCG by the vagus nerve A f t e r r e i n n e r v a t i o n o f the S C G b y the vagus nerve there was an increased collagen content, a n d an increase in the n u m b e r s o f m y e l i n a t e d axons f o u n d in the ganglion ( m y e l i n a t e d axons were p r a c t i c a l l y a b s e n t f r o m a n y o f the previous g r o u p s o f ganglia). T h e synapses c o u l d be d i v i d e d into two general categories. T h e c o m m o n e r t y p e (Fig. 7) were fairly similar to the n o r m a l synapses f o u n d in the previous g r o u p s o f g a n g l i a ; they h a d a high p r o p o r t i o n o f large dense c o r e d vesicles, b u t the 50 n m

456

3>

/

O3

/

/

/~

•

•

/I

~1 1

~ 2•

0

1

" " " "" " '=' ' 0' r 799 2

3

4

5 6 -7--8 9 NUMBER OF AXONS ×10{

F~g 6. A graphical representation of the relationship between the numbers of synapses (m mdlions per ganghon, ordinate) and the total numbers of preganglionic axons (in thousands, abscissa) in each of 7 ganglia with sympathetic anastomoses and m the denervated ganglia. The numbers of synapses m the 6 denervated gangha are represented by a single point (since the values for mdiwdual denervated ganglia were based on so few synapses that they were too varmble to be plotted Individually). The straight line is fitted by the method of least squares for the regression of y on x. The correlation coeffioent, r -- 0.799 indicates a probabdlty, P < 0.02 ~ 0.01 (t 3.27, w~th 6 degrees of freedom) and broken hnes indicate 95 % confidence hmlts. The slopes of the curves for the regression o f y on x, and x on y are 550 and 863 respectively, gwmg a mean of 707. Thzs is a measure of the number of synapses per axon. Th~s may also be calculated by summing the numbers of synapses (m excess of the denervated level of 0.64 < 106) m all the 7 individual ganglia with sympathetic anastomoses (22.8 × 106) and diwding by the sum of all the indwldual numbers of axons (32.16 - 103): thts gives a value of 709 synapses per axon

s y n a p t i c vesicles w e r e m o r e s c a t t e r e d , f o r m i n g less d i s t i n c t clusters, and a n a p p r e c i a b l e p r o p o r t i o n o f the a x o n t e r m i n a l s w e r e l a r g e r t h a n t h o s e o f t h e n o r m a l S C G . It seems p r o b a b l e t h a t t h e s e t e r m i n a l s b e l o n g to v a g a l p r e g a n g t i o n i c a x o n s (see D i s cussion). A less n u m e r o u s c a t e g o r y o f s y n a p s e s (Fig. 8) h a d u n u s u a l l y l a r g e a x o n t e r m i n a l s w h i c h w e r e d i s t i n g u i s h e d by a p o p u l a t i o n c o n s i s t i n g a l m o s t e x c l u s i v e l y o f

Fig. 7. A synapse of the commoner type (presumed to belong to the vagal pregangliomc axons, see text) found after vagal anastomosis. The large axon terminal (A) contains 50 nm synaptic vesicles and a high proportion of 100 nm dense cored vesicles, and establishes synapt~c contact by means of a vesicle cluster towards a synaptic thickening (arrow) on a small, pedunculated spine-like postsynaptic profile. Scale bar, 0.5/zm. Vagal anastomosis. Survival, 117 days. Fig. 8. A synapse of the less common type found after vagal anastomos~s The large axon terminal (A) has almost entirely clear 50 nm synapt]c vesicles and a core of tubules (asterisk). The synaptlc contact has a vesicle cluster directed towards a postsynaptic thickening (arrow) on an irregular protrusion (P) borne on a dendritic shaft (H). The postsynaptic protrusion contains the typical irregular vesicular materml and has a slightly denser matrix than that of the parent shaft. Scale bar, 0.5 #m. Vagal anastomosis. 112 days survival. Fig. 9. A synapse of the type characteristic of hypoglossal axons. The large, highly irregular axon terminal (A) has a central cluster of mitochondria (M) and makes several synaptic contacts. Arrows indicate two contacts with vesicle clusters directed towards postsynaptic thiekenin~ on two small spree-like profiles. The terminal has practically no large dense cored vesicles. Scale bar, 0.5 /~m. Hypoglossal anastomosis, 87 days survival.

4~

458 50 nm synaptlc vesicles with no large dense cored vesicles. Such terminals are characteristic of the ganglia with vagal anastomoses. The quantitative study ~s based on counts of 1112 synapses and 506 vacated thlckenings in a total area of 1339 / 103 sq.#m from 23 ganglia. There is a progressive increase in synapse numbers over the 50 to 100 days after anastomosis (Fig. 5, middle panel) to a level of 4.37 ~ 0.25 < 106 synapses per ganglion (Table II), i.e., somewhat fewer than after sympathetic anastomoses. Vacated thickenings (1.33 ± 0.08 y 106 per ganglion) were slightly less frequent than after sympathetic anastomoses. In two animals (circled points in the middle panel of Fig. 5) with long survival periods (118 and 228 days) the vagus nerve was re-exposed and cut, and after a further 6 days the ganglia were removed and the synapses counted. This procedure resulted in a reduction of synapse numbers to 0.59 and 0.71 ~, 106 synapses per ganglion, i.e., the same level as in the denervated ganglia (0.63 J 106 synapses per ganglion), thus confirming that the synapses m the vagal anastomoses were derived from axons in the vagus nerve.

Reinnervation of the SCG by the hypoglossal nerve A detailed account of the hypoglossal anastomoses is given m Ostberg et aL as, but the main features are summarized here to facilitate comparison with the other types of anastomosis. After reinnervation of the SCG by the hypoglossal nerve there was an increased collagen content, and the ganglion was traversed by large numbers of myelinated and unmyelinated axons of unusually wide diameter. These axons had a filamentous axoplasm (in distmction to the neurotubules usually found in preganglionic axons) and tended to have a slightly ruffled surface membrane with a peripheral subsurface density somewhat resembhng the undercoating described in the region of axon hillocks m the central nervous system 39. Although there were only few synapses in these ganglia, a proportion were clearly different morphologically (Fig. 9) from those found in any of the previous groups. These synapses were formed by large, irregular axon terminals arising from wide pretermmal segments. They contained almost excluswely small clear 50 nm synaptlc vesicles (i.e. with only rare 100 nm dense cored vesicles), sacs of smooth membranes, and glycogen particles; the centre of the terminal usually contained a dense cluster of mitochondria. Synaptic contacts, which were usually multiple, were generally estabhshed with several small dendritic protrusions. In addition to these synapses, synapses with axon terminals containing small dense cored vesicles (as in denervated SCG; Fig. 4) formed a high proportion of the total population; this would be expected if these synapses represent an intrinsic group, since the number of extrinsic synapses is low. The quantitative study as based on counts of 214 synapses and 428 vacated thickenings from a total area of 616 × 103 sq.~m from 15 ganglia. There is a suggestion of a slight increase in synapse numbers over the first 50-100 days after anastomosis (Fig. 5, bottom panel, see also Fig. 10) to a level of 1.53 :k 0. l I × l06 synapses per ganglion (Table II), t.e., less than in any of the previous anastomoses, but more than the denervated level (0.63 ± 0.23 "< 106). Vacated thickenings were more frequent (2.51 ± 0.30 ~ l06 per ganglion) than in any of the previous anastomoses, but shghtly less common than in the totally denervated SCG (2.95 ~- 0.12 y 106).

459 TABLE III EFFECT OF RESERPINE PRETREATMENT ON TYROSINE HYDROXYLASE ACTIVITY IN

SCG

FROM UNOPERATED

AND SURGICALLY TREATED RATS

Tyrosine hydroxylase was assayed m control animals and 72 h after s.c. injection of reserpine (5 mg/kg). Values are means 4- S.E.M. for groups of 4 ammals, P values refer to comparison of control and reserpine treated groups. All operative procedures were performed at long time intervals (120 to 160 days) before the experiment. Group

Tyrosine hydroxylase --pmole DOPA/ganglion/h Control

Unoperated Unilateral ganglionectomy Denervated Unilateral ganglionectomy ÷ denervation *Freeze lesion Sympathetic anastomosis Vagal anastomosm Hypoglossal anastomosis

Reserpine-treated

P value

95.7 4- 11.2 77.4 4- 7.0 88.4 -4- 7.8

181.7 zk 29.3 149.9 ± 4.7 78.9 i 4.5

< 0.05 < 0.001 N.S.

82.9 -4- 5.4 99.4 4- 10.3 84.1 4- 10.6 110.5 4- 10.0 72.1 4- 4.2

79.9 i 12.5 148.0 zk 16.3 116.0 4- 12.5 139.3 4- 19.0 106.8 4- 27.0

N.S. < 0.05 < 0.1 N.S. N.S.

* Data from ref. 40. Biochemical data The tyrosine hydroxylase (TOH) assays (Table III) revealed that in none of the experimental groups was the basal ganglionic T O H activity significantly different from the value found in the unoperated control SCG. However, the basal T O H activity was higher (P < 0.001) after vagal than after hypoglossal anastomoses. Reserpine treatment induced a two-fold increase in T O H activity in the unoperated SCG 22. This response was unaffected by removing the SCG on the other side, but was abolished by denervating the SCG. Reserpine treatment caused a rise in T O H activity in ganglia reinnervated by the sympathetic preganglionic axons after a freeze lesion 40; it possibly induced a slight rise (P < 0.1) in the animals with anastomotic repair of the cervical sympathetic trunk, but was without detectable effect in any of the other groups of ganglia. The assays of pineal N-acetyltransferase (NAT) activity (Table IV) confirmed that in normal intact ganglia there is a 20-fold increase in N A T activity, from a level of 7 pmole/pineal/10 min in the late afternoon to 139 pmole/pineal/10 min at 4 h after the beginning of the dark period. These values are similar to those reported by Deguchi and Axelrod 11 (see also ref. 25). A comparable pineal N A T rhythm could be maintained by a single SCG with intact innervation (after preganglionic nerve section or ganglionectomy on the other side) but was abolished by denervating the remaining SCG (as in ref. 26). An initial series of experiments (not included in Table IV) showed that the loss of the nocturnal rise in pineal N A T activity was complete by 3 days after operation. A nocturnal rise in pineal N A T activity was not detected after reinnervation by surgical anastomosis of the sympathetic trunk or the vagus.

460 TABLE IV PINEAL SEROTONIN N-ACETYLTRANSFERASE ACTIVITY DURING LIGHT AND DARK PERIODS IN UNOPERAIED AND SURGICALLY TREATED RATS

Rats were killed 2 h before the end of the light period (Light) or 4 h after the beginning of the dark period (Dark) and pineal N-acetyltransferase activity assayed. Values are means ~ S.E.M for groups of 4 animals, P values refer to comparison of light and dark values

Group

N-acetyltransfera~e pmole/plneal/ lO rain Light

Unoperated Unilateral denervatlon Unilateral gangllonectomy Unilateral ganglionectomy + denervation Sympatheticanastomos~s Vagal anastomosis

Dark

P value

7.0 ± 2 3 6.6 JL: 1 5 6 8 ± 17

139.4 i 66.7 97 1 i 36.7 286.6 ± 119.6

< 0.05 < 0.05 < 0.05

7.5 ± 3 2 12.0 ± 3.0 15 7 i 3 4

46 ± 1.6 30.0 4- 15.7 25.2 ± 11 3

N.S. N.S. N.S.

TABLE V CHOLINE ACETYLTRANSFERASE ACTIVITY IN S C G AND NERVE TRUNKS IN UNOPERATED AND SURGICALLY TREATED RATS

Choline acetyltransferase was assayed m SCG and pre- and postganghonic nerve trunks from unoperated animals and animals operated at long time intervals previously. Results are mean values :t: S.E.M. Values for nerve trunks are expressed per mg protein because of variations in size of tissue samples dissected, n, number of SCG in group.

Group

Ch,4c activity SCG ( nmole/ganghon/h)

Unoperated Denervated *Freeze lesion Sympathetic anastomosis Vagal anastomosis Hypoglossal anastomosis

37.9 d: 1.7 0.3 i 0.2 21.5 ± 1 3 17.9 ± 2 6 24.4 £- 1.3 41.9 i 2 3

Nerve trunks (nmole/mg protem/h) Pregangliomc

Postgangliomc

41.8 0.24 0 15.8 64.7 96 9

2.9 ± 0.6 (n 0 (n 0 (n 0 (n 1.6 Z- 1.56 (n 49.0 ± 14 3 (n

i 35 ± 0.09 i 52 ± 47 ± 6.5

-- 8) - 8) -- 7) 4) 4) -- 4)

* Data from ref. 40. Th e ganglionic choline acetyltransferase ( C h A c ) activity (Table V) is n o r m a l l y a function o f the n u m b e r s o f preganglionic a x o n t er m i n al s present in the S C G 4,2°,22. C h A c activity disappeared after d e n e r v a t i o n an d returned t o w ar d s n o r m a l values after r e i n n e r v a t i o n f o l lo w i n g freeze lesions 4o or any o f the anastomoses. Assays o f C h A c w e r e also m a d e in preganglionic nerve trunks an d in the p o s t g a n g l i o n i c axons by s a m p l i n g a small segment o f the internal c a r o t i d nerve j u s t b e y o n d the p o i n t w h er e it emerges f r o m the rostral pole o f the S C G . As expected, n o r m a l p r e g a n g l io n i c nerve trunks c o n t a i n e d relatively high C h A c activity, a n d this was virtually abolished after section o f the p r eg an g l i o n i c t r u n k

461 (Table V). The ChAc activity of normal postganglionic nerve was very much lower, but nevertheless it was easily measurable by the sensitive radiochemical enzyme assay procedure used (values more than twice assay blank). After preganglionic nerve section the activity present in postganglionic axons disappeared completely, suggesting that it may be due to the presence of a small number of preganglionic fibres that pass through the ganglion. Both vagal and hypoglossal anastomoses restored preganglionic nerve trunk ChAc activity to values greater than those seen in normal material, but sympathetic anastomoses lead to only a partial restoration of preganglionic ChAc activity. Only very low ChAc activity was present in postganglionic nerve trunks from vagal or sympathetic anastomoses, but abnormally high values were seen after hypoglossal anastomosis, suggesting that a large proportion of the cholinergic hypoglossal fibres pass right through the SCG. DISCUSSION

Critical differences between the various anastomoses can be expressed in quantitative terms (Table II; Fig. 10). After a freeze lesion of the preganglionic chain, axonal regeneration restores a density of synaptic innervation (7.46 × 10e synapses per ganglion) approaching that in normal animals (8.78 × 106), and our previous study 4° with a larger series of ganglia showed that regeneration after this operation can com-

UNOPERATED

9"

rn Cn

,°,I

(6)t FREEZE LESION

C'3( C'3 r- 5 . O z

(4)

SYMPATHETIC

t(10 )

(7)

I(4)

VAGUS

% (6) (4)

(4)

(2) 5'0

1(~0

]-(5)

HYPOGLOSSAL

(8) DENERVATED 1~0

a6o

2~o

360

3~o

DAYS POSTOPERATIVE SURVIVAL

Fig. 10. A schematic representation of the course of the reinnervation process after anastomosis of the various types of nerves. Ordinate, synapses expressed as millions per ganglion; abscissa, postoperative survival in days. For the purposes of display, the ganglia from each 50 day survival period (0-50, 51-100, 101-150, 1 5 0 + ) were grouped (number of ganglia in each group shown in parentheses), and mean numbers of synapses plotted at a survival time which was the mean of the survival times in the group. The standard errors (represented by bars) are included only as a way of demonstrating the variation, and do not imply that the synapse numbers can be taken as normally distributed variables. The numbers of synapses in the unoperated ganglia are indicated by the continuous horizontal line at the top of the figure.

462 pletely restore a normal complement of synapses. Cutting and suturing the preganglionic chain (a necessary control for the surgical procedures mvolved in the remnervation by foreign nerves) results in a lower and more variable synaptic density (5.2 I . 10s). The ganglia with vagal anastomoses have an even lower synaptic density (4.37 :# 106), and those with hypoglossal anastomoses have still fewer synapses (1.53 :* IOs), although their density of innervation is still significantly greater than that of ganglia left denervated for comparable periods of time (0.63 K 10s). In agreement with De Castro8 and Hillarpss our silver stained material indicates that after vagal or hypoglossal anastomoses, there is no apparent failure of the regenerating fibres to traverse the anastomosis, enter the ganglion, and arborise profusely among the ganglionic neurones. The numbers of vacated synaptic thickenmgs show an inverse relation to the numbers of synapses (Table II). There are most vacated thickenings in the denervated ganglia (2.95 x lOs), less in the hypoglossal anastomoses (2.51 x IOs), and still fewer after vagal (1.33 Y 106) or preganglionic sympathetic anastomoses (1.48 \c 10s); after recovery from freeze lesions of the preganghonic chain there are only a very small number of vacated thickenings (0.78 >( 10s; cJ 0.51 x 106 in unoperated ganglia). The extent to which the vacated thickenings disappear is correlated with the extent of formation of new synapses (Fig. 11). This favours the suggestion that, regardless of whether they belong to sympathetic axons or to fibres of either of the foreign

Fig. 11. A graphical representation of the numbers of synapses (ordinate, in millions per ganglion) and the numbers of vacated thickenings (abscissa, in millions per ganglion) for each of the groups of ganglia shown in Fig. 10. The negative regression line fitted by the method of least squares has a correlation coefficient of r = -0.84, P < 0.001 (t = 5.7, with 14 deg#ea offreedom). Brokenfines show 95 ‘A confidence limits.

463 nerves, the newly formed axon terminals establish synapses only at the original synaptic sites of the ganglion. There is evidence from several studies that both vagal and hypoglossal anastomoses can restore ganglionic transmission, as judged by the occurrence of pupiUary and other sympathetic responses to electrical or reflex stimulation of the anastomosed nerve trunks6,S,10,17,23,30,al. Our present biochemical data confirm previous findings 26,4o that the reserpine-induced rise in ganglionic tyrosine hydroxylase (TOH) and the dark-induced rise in pineal N-acetyltransferase (NAT) both depend on an intact preganglionic innervation. The TOH response was restored after regeneration following freeze lesions of the sympathetic preganglionic trunk 4°, but neither this response, nor the pineal NAT rhythm were detected in ganglia with any of the various anastomoses (except possibly for a slight TOH response in the sympathetic anastomoses). However, it is impossible to say whether this failure of response is due to an intrinsic inability of the new synapses to mediate these transsynaptic enzyme regulatory effects, or whether the failure is merely a reflection of the reduced numbers of synapses, or of the abnormal pattern of afferent impulses in the anastomotic situations. Similarly, the observation that the basal TOH activity is higher after vagal than hypoglossal anastomoses (Table III) may be due to the larger number of synapses formed by vagal axons, but it may also be due to the higher level of 'tonic' electrical activity in the vagus nerve. The restoration of ChAc activity following the various procedures (Table V) is in general agreement with our other findings. In normal SCG the very low activity of ChAc in the postganglionic nerve bundles suggests that the great majority of preganglionic fibres terminate in the SCG. However, after hypoglossal anastomoses, relatively high ChAc activity was detected in the postganglionic trunk. This suggests that a considerable proportion of the ingrowing hypoglossal fibres pass through the SCG into the postganglionic trunk. In the subsequent discussion, it is convenient to consider the results under four headings: preganglionic anastomoses, vagal anastomoses, hypoglossal anastomoses, and denervated ganglia.

Preganglionic anastomoses In the present experiments, the recovery after freeze lesions of the preganglionic chain resulted in restoration of 85 ~ of the original numbers of synapses in the SCG. By contrast, the anastomotic repair of the chain by suture after section resulted in a much more variable number of synapses and failed to restore more than about 60 ~o of the normal complement of synapses. Possible reasons for this difference must be sought in the local conditions at the site of union, and must take at least three factors into account: disorganization of the regenerating axons, delay in reinnervation, and failure of axons to reach the target tissuet6,2a,42,4a. The present method of counting synapses does not yield satisfactory anatomical information regarding the pattern of organization within the ganglionic connections (cf. ref. 29), and analysis of the possible effects of delay at the site of union is hindered by lack of information on the precise time course of events during regeneration of individual connections. These

464 two factors cannot, therefore, be dismissed as negligible. Nonetheless, the counts of the numbers of axons (Fig. 6) in the 7 anastomoses indicate that the most likely explanation of the relative synaptogenic inefficiency of the anastomotic repair as compared with the freeze lesion is that a proporhon of the axons actually fail to reach the ganglion (either because they become diverted, or because they degenerate). The variable success rate after this anastomosis (as compared with the much more consistent results of vagal and hypoglossal unions) is largely due to the mechanical difficulty in suturing the much finer preganglionic chain in a situation where the cut nerve ends are under longitudinal tension. However, we can make use of this varmbility m order to calculate the numbers of synapses per axon (see below). The unoperated SCG has a total of 8.78 x 106 synapses. After pregangliomc denervation 0.63 y 106 synapses remain, indicatmg that about 8.18 × 106 of the synapses are formed by axons running in the preganglionic chain. In our material the normal rat preganglionic chain contains about 6000-9000 axons (mean of 7380), a figure somewhat higher than those of Bray and Aguayo 5 and Dyck and Hopkins 12. Thus each axon would form on the average about 1100 (range 900-1400) synapses. Quite simdar figures have been obtained for the cat SCG (see ref. 3) - - 10 x 106 synapses, and 7000 axons - - t.e., about 1400 synapses per axon (not allowing for any possible intrinslc synapses). After a crush lesion of the preganglionic chain low in the neck, Bray and Aguayo 5 counted the numbers of regenerating sprouts distal to the crush and found that at one month after operation there were about four times the normal number of axons; this had subsided to normal by 6 months after operation. It has been suggested that during peripheral nerve regeneration, loss of redundant axon sprouts is associated with their failure to estabhsh functional connections 43. In the present experiments we have counted the numbers of axons in the regenerated preganglionic chain at 80127 days after surgical section and anastomosis. The results show a positive correlation (Fig. 6) between the numbers of axons in the chain and the numbers of synapses m the SCG. Allowing for the 0.63 y 106 synapses remaining in the denervated SCG, it can be calculated that at this survival time each ingrowing axon forms on the average about 700 synapses. The shortfall between this figure and the number of 1100 synapses per axon in the normal unoperated S C G , may be due to some deficit inherent in the reinnervation mechanism, but it seems more likely that if the survival period is not sufficiently long, the axons may not have had sufficient time to form their maximum numbers of synapses, and further, that the transient increase in axon sprouts described by Bray and Aguayo 5 has not had time to subside to its final level. In calculating the ratio of synapses to axons, these factors would interact (reduction of the numerator and increase of the denominator) to cause a large fall in the number of synapses per axon. Whatever the actual number of synapses per axon, however, the total numbers of synapses achieved even after longer survival periods are always less than in the unoperated SCG, and there are still vacant sites remaining, indicating that the axons have a limit beyond which they are not able to reinnervate any further sites. This suggests that the 'synaptogenic' ability of the axons is limited; a given number of axons cannot innervate an infinitely large postsynaptic territory. This explanation

465 is important since it suggests that under appropriate circumstances the total number of synapses may be limited not by the number of available sites, but by the number of available axons.

Vagal anastomoses Successful reinnervation of the SCG by the cervical vagus has been described by several authors on the basis of recovery of sympathetic functions6,S,23, 8°, regeneration of pericellular axonal networks or boutonsS, 23, or reappearance of synapses in electron micrographs 6. In the present study we have made a quantitative estimation of the density of synaptic innervation in the SCG after anastomotic innervation by the vagus nerve, and find that there are about 4.4 × 106 synapses, i.e., less than the number found after a comparable anastomosis of the sympathetic preganglionic fibres (5.2 × 106). Since the vagus nerve is considerably larger than the preganglionic chain, and Bodian stained material shows that the regenerating vagal axons cross the anastomosis to reach the SCG in large numbers, there is no evidence that any lesser degree of synaptic innervation by the vagus could be due to an overall numerical deficiency in axons. The vagus contains several different fibre components, and, as suggested by Langley 8° and Hillarp 23, it seems most probable that the SCG is selectively reinnervated by the vagal preganglionic axons, whose cells of origin are located in the dorsal 'motor' vagal nucleus in the medulla. These axons normally innervate visceral autonomic (parasympathetic) ganglia. The evidence that it is the vagal preganglionic axons which selectively reinnervate the SCG is somewhat indirect (e.g. Hillarp's observation ~3 that the pericellular plexuses arose from the fine rather than the coarse diameter vagal fibres), and is based more on the relative unlikelihood of other vagal components being able to innervate the SCG. Thus, it seems unlikely that a large number of the synaptic contacts in the SCG are formed by the somatic motor fibres of the vagus (e.g., in the inferior laryngeal nerve) which normally form neuromuscular contacts with striated muscle. Although De Castro s reported reinnervation of the SCG by somatic motor fibres of the hypoglossal nerve, our own studies and those of Hillarp ~ on the hypoglossal nerve, as well as findings on other somatic motor nerves such as the phrenic31, 41 or the nerve to sternohyoid 37 on the whole indicate that only few synapses form in these situations. With regard to the sensory contingent of vagal axons, Langley and Anderson al were unable to obtain functional innervation of the SCG, suggesting that the sensory fibres of the peripheral vagus are not the source of the synapses in the vagally innervated SCG, although later authors 9,a5 were able to achieve functional reinnervation of the SCG by implanting the centrally directed branches of the vagal nodose (i.e., sensory) ganglion. By a process of elimination, therefore, the vagal preganglionic fibres become the most likely source of the majority of the vagal synapses in the SCG. The present observations agree with those of other authors 6 that the majority of the heterotypic synapses formed by vagal fibres in the SCG have axon terminals in which the usual 50 nm synaptic vesicles are associated with a number of larger (100 nm) dense core vesicles. This is comparable to the normal sympathetic preganglionic

466 axons, but Js unlike the motor endplates of skeletal muscle nerves or the synapses formed by the hypoglossal axons in the SCG. On the other hand, a similar vesicle population is found m the heterotypic axon terminals formed when the vagus nerve is grown into the rabbit diaphragm 2 or the frog sartorius muscle zs. Their morphological differences from motor endplate terminals favour the interpretation that m all these anastomoses the terminals belong to the pregangliomc rather than the skeletomotor fibres of the vagus. Evans and Murray 13 have made estimates of the numbers of fibres in the cervical vagus of the rabbit, and concluded that out of a total number of 23,000 axons, only about 10 ~o are preganglionic. Assuming the rat to be comparable, the total number of vagal preganglionic fibres is probably lower than that of the sympathetic preganglionic chain. This is supported by a study of Lewis et al. za whose Fig. 2 indicates that in the rat the number of cells in the dorsal motor nucleus of the vagus ~s between 1000 and 2000. The present results indicate that the vagal nerve forms about 3.7 × 106 synapses in the SCG (after allowing for the numbers of synapses in the denervated SCG). If the axons of cells m the dorsal motor nucleus formed all the synapses, they would be forming between 1900 and 3800 synapses per axon, considerably more than the figure of 1100 found for the normal sympathetic preganglionic axons. However, the morphological evidence suggests that a proportion of the vagal synapses (those without large dense cored vesicles in the axon terminals) are of a different type, and it seems likely that these (at least) are derived from a different category of vagal axons. To this extent, therefore, the figure of 1900-3800 synapses per axon for the vagal nerve should be taken as a probable overestimation of the true figure. A further difficulty in the interpretation of the results of vagal anastomoses is the absence of choline acetyltransferase in the postgangliomc trunks (Table V). In the case of the hypoglossal anastomoses, high ChAc activity in the postganglionic fibres can be correlated with a large number of cholinerglc axons passing through the SCG, but in the case of the vagal nerve, where it seems hkely that a similarly large proportion of skeletomotor cholinerglc axons would also pass through the SCG, there is no detectable ChAc in the postganghonic nerve bundle. Hypoglossal anastomoses

Despite the fact that the hypoglossal nerve forms only few synapses in the SCG (1.5 × 106), there are still more than twice as many as in the denervated SCG (0.6 × 106), and the occurrence of a morphologically distinctive type of axon terminal confirms the conclusion that some of these synapses do belong to axons in the hypoglossal nerve. Our findings indicate that, after deducting the number of synapses in the denervated SCG, the hypoglossal nerve forms considerably fewer synapses (0.9 × 106) than the vagus nerve (3.7 × 106). This is in distinction to De Castro s, who considered (using silver stained material) that the hypoglossal nerve was more effective than the vagus, but we are in agreement with Hillarp 28 who found that the hypoglossal only formed very occasional pericellular arborisations in the SCG. The hypoglossal nerve is an almost pure skeletomotor nerve, with only a small sensory (mechanoreceptor) component1,19. The presynaptic terminals formed by the

467 hypoglossal axons in the SCG as are often unusually large, with large numbers of 50nm synaptic vesicles and very few dense cored vesicles (in contrast to the terminals of the normal preganglionic axons). The occurrence of unusually large terminals in the SCG was also described after hypoglossal anastomoses by Hillarp ~z and De Castro s, who noted their resemblance to motor endplates. Assuming that the hypoglossal nerve innervates the ganglion by means of skeletomotor fibres, there remains the question of why such a large number of these fibres passing through the SCG (see also ref. 23) can only innervate a small proportion of the available ganglionic sites. Possibly the fibres are each relatively inefficient in this respect, but that as such a large number of axons are available, some synapses are formed. Alternatively, it may be that a small proportion of the sites in the SCG are specifically different from the remainder (e.g. the muscarinic sites TM, or those on cholinergic neuronesT, 24) and are thus able to accept hypoglossal axons. Thirdly, there is the possibility that specific differences exist between hypoglossal axons, such that a small group of axons have some specific property enabling them to form synapses in the SCG. Lewis et al. a3 found that whereas the majority of the neurones in the rat hypoglossal nucleus contained only acetylcholinesterase, a small group of cells in the caudoventral part of the nucleus contained pseudocholinesterase. The axons of these cells also contain pseudocholinesterase, and run in the medial branch of the hypoglossal nerve (probably innervating the intrinsic protrusor muscles of the tongue). These cells have a number of histochemical similarities to those found in the adjacent dorsal motor nucleus of the vagus a4, viz. both contain pseudocholinesterase as well as true acetylcholinesterase, and after axotomy both show a more rapid and severe fall in cholinesterase and a far less complete recovery than do the neurones of the main part of the hypoglossal nucleus14, a2. Thus the hypoglossal nerve contains a small contingent of fibres which have some affinities with vagal preganglionic neurones, and it is therefore tempting to suggest that it is this component of the hypoglossal nerve which forms the synapses we observe in the SCG. Lewis et al. aa estimate that there are about 600 of these pseudocholinesterase-containing cells in the hypoglossal nucleus, and roughly the same number of pseudocholinesterase-containing axons in the hypoglossal nerve. After allowing for the intrinsic synapses, these axons would therefore be responsible for about 0.9 × 106 ganglionic synapses - - i.e. 1500 synapses per axon - a figure not too far removed from the figure of 1100 (900-1400) found for the sympathetic preganglionic axons in the unoperated SCG. The suggestion that a large number of hypoglossal cholinergic axons pass through the SCG without terminating is supported by our observations of high choline acetyltransferase activity in the postganglionic nerves after hypoglossal anastomoses (Table V), and indicates that the presence of a cholinergic mechanism although possibly a necessary feature for axons to be able to form ganglionic synapses is not in itself sufficient. Denervated ganglia

An analysis of the residual synaptic composition of the denervated ganglia is outside the scope of this study, and it should also be noted that the synapses of the small granule-containing cells have not been included. The 7 ~o of synapses which

468 remain after cutting the pregangliomc chain may belong partly to pregangliomc axons which take a course enabhng them to escape destruction during the operation for anastomosis. On the other hand, the presence of small (50 nm) dense cored vesicles in many of the residual terminals suggests that they may contain catecholamines and may therefore belong to ganglionic neurones 15. The presence of such intrinsic synapses raises the question of why the axons of the ganglionic neurones (which are 5-10 t~mes as numerous as the preganghonic axons) do not form intragangliomc sprouts to innervate the denervated synaptic sates. Two general hypotheses arise from the foregoing discussion. Firstly, there appears to be a limit to the number of synapses which an axon can form, and under normal c~rcumstances, axons may be quite close to th~s limit. This may have a bearing on the question of the extent to which an undamaged axon may innervate adjacent denervated territory by 'collateral sprouting'. Secondly, the present results give indications of an absolute type of specificity of reinnervatlon. The numerical data can best be explained by assuming that the ganglionic postsynaptic sites will accept mnervation only from axons of an appropriate type, t.e. sympathetic pregangliomc, vagal preganghonic, or a specml category of hypoglossal axons. They do not recewe innervation from most other skeletomotor axons (even though cholinerglc), sensory axons or sympathetic postgangliomc axons. Thus the present numerical data can be explained without recourse to the more comphcated assumption that certain classes of nerve fibre are relatively inefficient m forming gangliomc synapses. The 'foreign' axons appear to be able to form at least as many ganglionic synapses per axon as the normal preganglionic axons. The present observations provide no suggestion that foreign nerve fibres are either slower or in some way more 'reluctant" to form ganglionic synapses than the proper preganglionic axons. ACKNOWLEDGEMEN'IS We wish to thank Dr. D. L. Misell for assistance with the mathematics.

REFERENCES 1 ADATIA, A. K., AND GEHRING, E. N., Proprloceptive irmervation of the tongue, J. Anat. (Lond.), 110 (1971) 215-220. 2 BENNETT,M. R., McLACHLAN,E. M., AND TAYLOR,R. S., The formation of synapses in mammahart striated muscle reinnervated with autonomic preganglionic nerves, J. PhysioL (Lond.), 233 (1973) 501-517 3 BmKs, R. I., The relationship of transmitter release and storage to fine structure in a sympathetic ganglion, J. Neurocytol., 3 (1974) 133-160. 4 BLACK, L B., HENDRY, I. A., AND IVERSEN, L. L., The role of post-synapt~c neurones in the b~ochemical maturation of presynaptic chotinergic nerve terminals in a mouse sympathetic ganglion, J Physiol. (Lond.), 221 (1972) 149-159. 5 BRAY, G. M, AND AGUAYO, A. J., Regeneration of peripheral unmyehnated nerves. Fate of the axonal sprouts which develop after injury, J. Anat. (Lond.), 117 (1974) 517-530. 6 CECCARELU,B., CLEMEWn,F., AND MANTEGAZZA,P., Synaptic transmission in the superior cervical ganglion of the cat after reinnervation by vagus fibres, J. PhysioL (Lond.), 216 (197I) 87-98.

469 7 DALE, H. H., AND FELDBERG,W., The chemical transmission of secretory impulses to the sweat glands of the cat, J. Physiol. (Lond.), 82 (1934) 121-128. 8 DE CASTRO, F., Sur la r6g6n6ratlon fonctionelle dans le sympathique (anastomoses crois6es avec des hefts de type iso- et h6t6romorphes). Une r6f6rence spt~ciale sur la constitution des synapses, Trav. Lab. Rech. Biol. Univ. Madrid, 31 (1937) 271-345. 9 DE CASTRO,F., Modelacl6n de un arco reflejo en el simp~ttico, unl6ndolo con la raiz aferente central del vago. Nuevas ideas sobre la sinapsis, Trab. Lab. Invest. Biol. Madrid, 34 (1942) 217-301. 10 DE CASTRO,F., Aspects anatomiques de la transmission synaptlque ganglionnaire chez les mammifibres, Arch. int. Physiol., 59 (1951) 479-525. 11 DEGUCm, T., AND AXELROD,J., Sensitive assay for serotonln N-acetyltransferase activity in rat pineal, Analyt. Biochem., 50 (1972) 174-179. 12 DYCK, P. J., AND HOPKINS, A. P., Electron microscopic observations on degeneration and regeneration of unmyelinated fibres, Brain, 95 (1972) 223-234. 13 EVANS,D. H. L., ANDMURRAY,J. G., Histological and functional studies on the fibre composition of the vagus nerve of the rabbit, J. Anat. (Lond.), 88 0954) 320-337. 14 FLUMERFELT,B. A., AND LEWIS, P. R., Chohnesterase activity in the hypoglossal nucleus of the rat and the changes produced by axotomy: a light and electron microscopic study, J. Anat. (Lond.), 119 (1975) 309-332. 15 GRILLO,M. A., Electron microscopy of sympathetic tissues, Pharmacol. Rev., 18 (1966) 387-399. 16 GUTH, L., Regeneration in the mammalian peripheral nervous system, Physiol. Rev., 36 (1956) 441-478. 17 GUTH, L., Functional recovery following vagosympathetic anastomosis in the cat, Amer. J. Physiol., 185 (1956) 205-208. 18 HANCOCK,J. C., AND VOLLE, R. L., Cholinoceptlve sites in rat sympathetic ganglia, Arch. int. Pharmacodyn., 184 (1970) 111-120. 19 HANSON,J., AND WIDI~N,L., Afferent fibres in the hypoglossal nerve of cat, Actaphysiol. scand., 79 (1970) 24-36. 20 HEBB,C. O., AND WA1TES,G. M. H., Choline acetylase in antero- and retro-grade degeneration of a cholinergic nerve, J. Physiol. (Lond.), 132 (1956) 667-671. 21 HENDRY, I. A., AND IVERSEN, L. L., Effect of nerve growth factor and its antiserum on tyrosine hydroxylase activity in mouse superior cervical sympathetic ganglion, Brain Research, 29 (1971) 159-162. 22 HENDRY,I. A., IVERSEN,L. L., ANDBLACK,I. B., A comparison of the neural regulation of tyrosine hydroxylase activity in sympathetic ganglia of adult mice and rats, J. Neurochem., 20 (1973) 1683-1689. 23 HILLARP,N. A., Structure of the synapse and the peripheral innervation apparatus of the autonomic nervous system, Acta anat. (Basel), Suppl. 4 (1946) 1-153. 24 HOLMSTEDT,B., LUNDGREN, G., AND SJ6QVIST,F., Determination of acetylcholinesterase activity in normal and denervated sympathetic ganglia of the cat. A biochemical and histochemical comparison, Acta physiol, scand., 57 (1963) 235-247. 25 KLEIN,O. C., AND WELLER, J. L., Indole metabolism in the pineal gland: a circadian rhythm in N-acetyltransferase, Science, 169 (1970) 1093-1095. 26 KLEIN, D. C., WELLER,J. L., AND MOORE, R. Y., Melatonin metabolism: neural regulation of pineal serotonin acetyl coenzyme A N-acetyltransferase activity, Proc. nat. Acad. Sci. (Wash.), 68 (1971) 3107-3110. 27 KONIGSMARK,B. W., Methods for the counting of neurons. In W. J. H. NAUTA AND S. O. E. EBBESSON(Eds.), Contemporary Research Methods in Neuroanatomy, Springer, Heidelberg, 1970, pp. 315-340. 28 LANDMESSER, L., Pharmacological properties, cholinesterase activity and anatomy of nervemuscle junctions in vagus-innervated frog sartorlus, J. Physiol. (Lond.), 220 (1972) 243-256. 29 LANGLEY,J. N., On the regeneration of pre-ganglionic and of post-ganglionic visceral nerve fibres, J. Physiol. (Lond.), 22 (1897) 215-230. 30 LANGLEY,J. N., On the union of cranial autonomic (visceral) fibres with the nerve cells of the superior cervical ganglion, J. Physiol. (Lond.), 23 0898) 240-270. 31 LANGLEY,J. N., AND ANDERSON,H. K., The union of different kinds of nerve fibres, J. Physiol. (Lond.), 31 (1904) 365-391. 32 LEWIS, P. R., BLUNDELLJONES, P., BREATHNACH,S. M., AND NAVARATNAM,V., Regenerative capacity of visceral preganghonic neurones, Nature New Biol., 236 (1972) 181-182.

470 33 LEwis, P. R., FLUMERFELT,B A., AND SHUTE,C. C. D., The use of chohnesterase techniques It) study topographical localization in the hypoglossal nucleus of the rat, J. Anat. (Lond.), l l0 (197l) 203-213. 34 LEwis, P. R., SCOTT, R. A , AND NAVARATNAM, V , Localization in the dorsal motor nucleus of the vagus in the rat, J. Anat. (Lond), 107 (1970) 197-208. 35 MATSUMURA,M., AND KOELLE, G. B., The nature of synaptic transmission in the superior cervical ganglion following reinnervation by afferent vagus, J. PharmacoL exp. Ther, 134 (1961) 28-46 36 MATTHEWS, i . R., AND RAISMAN, G., The ultrastructure and somatic efferent synapses of small granule-contaimng cells in the superior cervical ganglion, J. Anat. (Lond.), 105 (1969) 255-282 37 McLACHLAN, E. M., The formation of synapses m mammalian sympathetic ganglia reinnervated with pregangllonic or somatic nerves, J. Physiol. (Lond.), 237 (1974) 217-242 38 (~STBERG,A.-J C., MATTHEWS, M. R., AND RAISMAN, G., Some morphological observations on the superior cervical ganglion of the rat after relnnervation by the hypoglossal nerve, in preparation. 39 PALAY, S. L , SOTELO, C., PETERS, A., AND ORKAND, P. M., The axon hillock and the lmtial segment, J. Cell Biol., 38 (1968) 193-201. 40 RAISMAN, G., FIELD, P. M., (~STBERG, A.-J. C., [VERSEN, L. L., AND ZIGMOND, R. E , A quantitative ultrastructural and biochemical analysis of the process of remnervation of the superior cervical ganglion in the adult rat, Brain Research, 71 (1974) 1-16 41 TAXI, J., AND BABMINDRA, V. P., Light and electron microscopic studies of normal and heterogeneously regenerated ganglionic synapses of the dog, 3. Neurol. Trans., 33 (1972) 257-274. 42 WEISS, P., The technology of nerve regeneration: a review. Sutureless tubulation and related methods of nerve repair, J Neurosurg., 1 (1944) 400-450. 43 YOUNG, J. Z., Factors influencing the regeneration of nerves, Advanc. Surg., 1 (1949) 165-220.