BY WILLIAM PROCTOR, STEPHEN ROPER AND BARBARA

J. Phyil. (1982), 326, pp. 173-188 173With 2 plate and 6 text-figures8Printed in Great Britain

SOMATIC MOTOR AXONS CAN INNERVATE AUTONOMIC NEURONESIN THE FROG HEART

BY WILLIAM PROCTOR, STEPHEN ROPER AND BARBARA TAYLORFrom the Departments of Anatomy and Physiology, University of Colorado School of

Medicine, Denver, CO 80262, U.S.A.

(Received 10 March 1981)

SUMMARY

1. The effects of sympathetic and parasympathetic stimulation on the heart ratein frogs were tested after hearts were reinnervated with a somatic motor nerve. Whenfrogs were vagotomized and hypoglossal axons were redirected to the heart for 8 ormore weeks, stimulating the redirected hypoglossus nerve produced a parasympath-etic-like inhibition of the heart. Stimulating sympathetic rami of the anastomosedhypoglossus nerve produced cardiac acceleration.

2. Individual parasympathetic neurones received synaptic input from hypoglossalterminals. The excitatory post-synaptic potentials evoked by hypoglossal stimulationwere much smaller than those evoked by vagal stimulation in control or vagal-reinnervated ganglia. However, hypoglossal axons innervated most (71 %) of theganglion cells and this level of innervation persisted for at least 60 weeks.

3. Hypoglossal axons formed networks ofvaricose terminals within cardiac gangliaand established axo-axonic synapses with parasympathetic neurones. Hypoglossalterminals did not reinnervate the neuronal perikarya, in contrast to vagal axons incontrol or vagal-reinnervated ganglia.

4. Axo-axonic synapses from redirected hypoglossal axons were identified incardiac ganglia by bathing isolated hearts in horseradish peroxidase (HRP) andstimulating the redirected nerve. Electron micrographs showed that axo-axonicsynapses contained HRP-labelled presynaptic vesicles.

5. The source offoreign innervation in experimental cardiac ganglia was confirmedto be hypoglossal motoneurones (a), by comparing the conduction velocity of theredirected presynaptic axons (1-32 m/sec) with regenerating vagal preganglionicfibres (< 0 3 m/sec), and (b), by retrograde HRP-labelling of large motoneurones inthe hypoglossal nucleus after applying peroxidase to the axons which had grown intothe heart.

INTRODUCTION

In some regions ofthe nervous system damaged neural tissue is capable ofextensiveregeneration which restores neuronal connexions in irrupted by the injury. Althoughregeneration of severed axons has been the focus of at ention for several decades, onlyrecently has it been possible to investigate the re-establishment of functionalconnexions in detail at the synaptic level. However, despite these advances, we still

W. PROCTOR, S. ROPER AND B. TA YLOR

do not fully understand how regenerating axons restore their former connexions withappropriate target cells in some cases, such as reinnervation of sympathetic ganglioncells (Langley, 1897; Nja & Purves, 1977), but not in other cases, such as reinnervationof skeletal musculature by regenerating motor nerves (Bernstein & Guth, 1961 ; Miledi& Stefani, 1969).To provide a greater understanding ofnerve tissue repair, particularly the accuracy

with which neuronal connexions are restored, we have studied an experimental modelwhere regenerating axons were surgically diverted to inappropriate target neurones.This investigation examines the extent to which spinal motoneurones can innervatedenervated autonomic neurones in the heart of the frog and describes the propertiesof the resulting synaptic connexions. Although there is ample precedence forfunctional connexions being formed between motor axons and autonomic neurones(e.g. Calugareanu & Henri, 1900; Langley & Anderson, 1904; Wolff, Hare & Cattell,1938; Hillarp, 1946; McLachlan, 1974; Ostberg, Raisman, Field, Iversen & Zigmond,1976) our objective was to study connexions that are formed between such disparateparts of the nervous system at the synaptic level, correlating intracellular recordingswith ultrastructural analyses. Some of these data have been reported previously inabbreviated form (Proctor, Frenk & Roper, 1978; Proctor, Taylor & Roper, 1979;Proctor, Frenk, Taylor & Roper, 1979).

METHODS

Animals. Adult Northern grass frogs (Rana pipiens) weighing 15-40 g were used for theseexperiments. Investigations were conducted throughout the year to reduce systematic errors causedby possible seasonal variations. For reliable and convenient identification of experimental frogsafter surgical operations, each frog was photocopied to reproduce the pattern of spots on its back.

Surgical procedures. Frogs were anaesthetized with tricaine methanesulphonate (100 ,tg/g bodyweight; Ethyl-m-aminobenzoate, Sigma), and were kept on ice during the operation. Thehypoglossal nerve on one side was redirected to the heart as follows (see Fig. 1). An incision inthe skin was made just caudal to the tympanic membrane on both sides of the animal. On the leftside underlying muscles were separated to expose the vagus and hypoglossal nerves. Both nerveswere sectioned and the central end of the hypoglossal nerve was sutured to the distal end of thecut vagus nerve using a 50 Aim curved needle with 10-0 monofilament nylon suture (Ethicon). Onthe other side, only the vagus nerve was exposed and resected. The central ends ofboth vagus nerveswere sewn into the overlying skin to redirect them away from the heart.Animals were then returned to tanks which had running water, and were force-fed three times

per week with a cocktail of powdered liver, multivitamins, and cod liver oil. In experiments wherethe effects of long-term hypoglossal innervation were studied, we re-anaesthetized the frogs every10-15 weeks after the initial surgery and again resected the central stumps of both vagi to excludevagal regeneration to the heart. Approximately half of the experimental animals recovered fromthese operations and were tested for hypoglossal nerve innervation of the heart 8-60 weeks later.

Measuring autonomic control of the heart in vivo. Animals were pithed and records of the heartrate were obtained in situ by inserting a 26 gauge stainless steel wire under the sternum. A referencelead was attached to the front legs, and the electrocardiogram was monitored with a high gain a.c.amplifier and pen chart recorder. Trains of stimuli (2 msec pulses, 10 Hz) were applied to vagaland hypoglossal roots via fine bipolar silver wire hook electrodes to examine the effect of nervestimulation upon heart rate.

Intracellular micro-electrode recording. Methods of recording intracellular activity and measuringinput resistances in single parasympathetic neurones in the hea: c are described in Dennis, Harris& Kuffler (1971); Roper (1976); and Roper & Ko (1978).Neuromuscular transmissmion at hypoglossal terminals. Neuromuscular junctions from normal

hypoglossal nerve targets, geniohyoideus and hyoglossus muscles, were examined in control

174

FOREIGN MOTOR INNERVATION OF GANGLION CELLS

animals. Single muscle fibres from isolated preparations were impaled with micro-electrodes to recordend-plate potentials evoked by hypoglossal nerve stimulation. D-Tubocurarine chloride (1-5-660 4uMfinal concn.) was added to prevent muscle contraction produced by nerve stimulation. Conductionvelocities ofmotoraxonsweredetermined by stimulus-response latencies. Muscleswere subsequentlyfixed for light and electron microscopy.

Unoperated Hypoglossal redirected

Brain

Hypogi. yp *

pinal cord

LHeart

Fig. 1. Schematic drawing of the surgical operation for redirecting the hypoglossal nerveto the heart. The left vagus nerve and the left hypoglossal nerve were cut as illustratedby the dark lines. The central end ofthe hypoglossal nerve was sutured to the distal stumpof the vagal nerve and the central end of the left vagal nerve was sutured into the skin.The right vagus nerve also was sectioned and its central end sutured into the skin (notshown).

Light and electron microscopic analyses. Tissue was fixed and stained with zinc iodide-osmium andwhole-mount embedded to facilitate light microscopic observations of nerve terminals on cardiacganglion cells and on muscle fibres, as described in Proctor et al. (1979). Methods for preparationof tissue for electron microscopy were similar to those described in McMahan & Kuffler (1971).

Labelling presynaptic terminals with horseradish peroxidase. To visualize functional synapses inthe cardiac ganglion we utilized the ability of nerve terminals to take up horseradish peroxidase(HRP) into presynaptic vesicles during nerve stimulation. The labelled vesicles can then beobserved with electron microscopy (Heuser & Reese, 1973). The redirected hypoglossus nerve wasstimulated central to the anastomosis with 1 see trains of stimuli (at 10 Hz) applied every 2 seefor a total of 30 min while the tissue was bathed in frog Ringer containing HRP (10 mg/ml., SigmaType VI). In addition, 2 /SM-dihydroxy-fl-erythroidine (Merck) was added to block post-synapticresponses. HRP-uptake into synaptic vesicles was not observed if the concentration of magnesiumwas increased to 15 mm or if calcium was eliminated from the external solution.

Labeling spinal neurones with retrograde transport of horseradish peroxidase. Cell bodies ofhypoglossal axons which innervated the heart were identified by retrograde transport ofHRP. Twoanimals in which the hypoglossus nerve had been redirected to the heart 10 weeks earlier werere-anaesthetized and the nerve supply to the heart was exposed. Small pieces of Gelfoam (Upjohn)

175


soaked in HRP (approximately 30 mg/ml. in 01% polyornithine, Sigma) were applied to freshlycut central ends of the redirected nerves where they entered the heart. Ten days were allowed forretrograde transport ofHRP into the central nervous system. The animals were then anaesthetizedand perfused with buffered 4% glutaraldehyde. Frozen sections of the brain and spinal cord wereprocessed as described by Mesulam (1976). For comparison, HRP was applied as above on cardiacbranches of the vagus nerve in two control frogs to show the central origin of vagal cardiacpreganglionic neurones.

A BI. .-.. a

C Dua>30-

E)~20-

0 1 2 3 4 0 1 2 3 4Time (min.)

Fig. 2. Effect of vagal nerve root and sympathetic chain stimulation on the heart ratein an unoperated control animal. A, B are electrocardiograms during nerve stimulation.A, stimulating the vagal nerve root blocked the heart. B. stimulating the sympatheticchain produced long lasting cardio-acceleration. Electrocardiograms from A, B are plottedgraphically. C, response to vagal stimulation. D, response to sympathetic stimulation.Horizontal bars indicate the duration of applied stimulation (5 V).

RESULTS

Autonomic innervation of the heart

Parasympathetic preganglionic axons in the vagus nerve reach the heart via thecardiac branches of the right and left vagosympathetic nerves. The predominanteffect of stimulating the vagus nerve was cardio-inhibition caused by the overridinginfluence of the parasympathetic supply (Fig. 2A). However, sympathetic cardio-acceleration was clearly seen when the sympathetic chain was stimulated (Fig. 2B).The data shown in Fig. 2A and B are plotted graphically in Fig. 2C and D, and resultsof all subsequent electrocardiograms will be presented graphically.

Motor innervation of skeletal muscles

Hypoglossal motor terminals in geniohyoideus and hyoglossus muscles of controlanimals resemble motor terminals in other muscles of the frog, such as the cutaneous

176

FOREIGN MOTOR INNERVATION OF GANGLION CELLS 177

pectoris (McMahan, Spitzer & Peper, 1972; Peper, Dreyer, Sandri, Akert & Moor,1974). Plate 1 A shows an electronmicrograph of a cross section through a hypoglossalnerve terminal in the geniohyoideus, and Plate 1 B shows a light microscopicphotograph of a zinc iodide-osmium stained hypoglossal terminal.The average conduction velocity for hypoglossal motor axons was 6-3 + 1P8 m/sec

A Vagus B Hypoglossus4 5 5 10 20

30-

20-

10-

0

t C D0 10 20 30 7 1030_

20'

10-

0 1 2 3 4 0 1 2 3 4Time (min)

Fig. 3. Effect of vagal and hypoglossal nerve root stimulation on the heart rate. A, B,results obtained from an unoperated frog. A, stimulating the vagal nerve blocked theheart. B, stimulating the hypoglossal nerve had no effect on the heart rate. C, D, resultsobtained from a frog in which the hypoglossus nerve had been redirected to the heart 18weeks earlier. C, vagal stimulation was ineffective since vagal regeneration had beenprevented. D, stimulating the redirected hypoglossal nerve produced a parasympathetic-like inhibition of the heart. Horizontal bars indicate duration of stimulation at thedesignated intensities (V).

at room temperature (mean+s.D.; seventeen fibres, two animals), consistent withvalues for myelinated motor axons in the frog (for review, see Stampfli & Hille, 1976).

Hypoglosal reinnervation of the geniohyoideusIn three animals the hypoglossus nerve was severed and the cut ends rejoined to

examine self-reinnervation of the geniohyoideus. The conduction velocity of regene-

178 W. PROCTOR, S. ROPER AND B. TA YLOR

rating hypoglossal axons 8-26 weeks after self-reinnervation was only 2-9 + 0 9 m/sec(112 fibres). Inspection of these muscles after zinc iodide-osmium staining revealedthat regenerating hypoglossal axons were smaller in diameter than those of controlanimals and that the unmyelinated terminals had distinctive and irregular varicositiesalong their length (Plate 1 C). At later times regenerated axons returned to theirnormal morphology.

A B C

40 . 4 5 4

. 30X

-20-

100

0 1 2 3 40 1 2 3 4 0 1 2 3

Time (min)Fig. 4. Effect ofsympathetic chain stimulation on the heart rate in an experimental animal.The records were taken 20 weeks after the hypoglossal nerve had been redirected to theheart. A, stimulating the redirected foreign nerve stopped the heart. B, stimulating thesympathetic chain caudal to the root of the hypoglossus nerve accelerated the heart rate.Between trials illustrated in B and C the sympathetic chain was resected just rostral tothe hypoglossal nerve root. C, stimulating the sympathetic chain as before still acceleratedthe heart rate, indicating that cardio-acceleration was not caused by regeneration ofvagosympathetic fibres.

Innervation by motor axons redirected into the heartTo test whether motor nerve axons could establish functional contact with

denervated parasympathetic neurones despite the marked structural and functionaldifferences between cholinergic innervation in cardiac ganglia and in skeletal muscle,we severed the vagosympathetic nerves and redirected the hypoglossal nerve fromone side into the heart. We first examined whether stimulating the hypoglossal nerveroot effected the heart rate in situ. Fig. 3 compares the heart rates during stimulationof the intact vagus and hypoglossus nerves in an unoperated control animal (Fig. 3Aand B) and in an experimental animal (Fig. 3C and D). The parasympathetic-likeinhibition of the heart produced by stimulating the hypoglossal nerve seen in Fig.3D indicates that hypoglossal axons had indeed grown into and innervated thevagotomized heart. Also, these observations could be demonstrated in vitro, andcardio-inhibition in isolated hearts produced by stimulating either the vagus (control)or hypoglossus (experimental) nerves was eliminated by the ganglionic blockerdihydroxy-f1-erythroidine (2 ,CM).We could elicit cardio-acceleration in experimental animals by stimulating the

sympathetic chain caudal to the root ofthe anastomosed hypoglossus nerve (Fig. 4B).


This cardio-acceleration was mediated by sympathetic axons reaching the heart viathe redirected hypoglossus nerve since the increased heart rate could be producedafter the sympathetic chain was severed rostral to the hypoglossus nerve (Fig. 4C).In another group of experimental animals, in which the sympathetic chain was leftintact but the redirected hypoglossus nerve was severed, stimulating the chain nolonger produced cardio-acceleration. Also, when isolated hearts from experimentalanimals were tested in vitro in the presence of dihydroxy-fl-erythroidine, stimulatingthe anastomosed hypoglossal nerve increased, rather than decreased, the heart rate,revealing the presence of sympathetic axons in the redirected hypoglossus nerve.

Innervation of ganglion cells by hypoglossal motor axonsTo determine whether redirected motor axons made synaptic contact with

parasympathetic neurones in the heart, we impaled cardiac ganglion cells withmicro-electrodes. Passive membrane characteristics of parasympathetic neurones inexperimental ganglia were indistinguishable from those in ganglia from unoperatedanimals. The average resting membrane potential in ganglion cells in experimentalanimals (mean+ S.D. = 50-2 + 3-9 mV; 982 cells, forty-three animals) was indistin-guishable from that in control ganglia (49-5 + 8-6 mV, eighty-six neurones, sixanimals), and there was no significant difference between the input resistances ofganglion cells in experimental animals (mean +S.D. = 159+88 Mf; fifty-five cellsfrom nine animals) and in control animals (188 + 97 MCI; thirty-three cells, fouranimals).

Hypoglossal stimulation produced excitatory post-synaptic potentials in 71 % ofthe impaled ganglion cells (87/123 cells; six animals) in experimental animalsexamined 8-20 weeks after redirecting the motor nerve into the heart, and remainedat this level for at least 60 weeks. In only 9% of the neurones (11/123 cells; sixanimals) were the responses large enough to generate action potentials. This contrastswith normal ganglionic innervation; in control animals 97 % of the impaled cells(83/86) received at least one and usually two or more vagal inputs, and in 92 % ofthese neurones (79/86) the responses were suprathreshold. Fig. 5A illustrates anexceptional case in an experimental animal where both sub- and suprathresholdresponses were evoked by varying the intensity of hypoglossal stimulation. Supra-threshold hypoglossal responses lacked the pronounced residual depolarization foundin vagal responses in vagal-reinnervated ganglia (Fig. 5B) or in control ganglia(Fig. 5C), indicating that the underlying post-synaptic potential was just sufficient toinitiate an action potential. Repetitive stimulation (0-5-2-0 Hz for 10-60 sec) fatiguedsynaptic transmission at hypoglossal inputs in the ganglion and eliminated actionpotentials, thereby revealing the underlying small synaptic responses. Subthresholdhypoglossal responses were weak (usually below 2 mV) and had slow rise timescompared with vagal responses in unoperated animals. Foreign innervation of theparasympathetic neurones was distributed uniformly throughout the ganglion andthere was no tendency for ganglion cells near the entry of the foreign nerve to receivemore, or stronger, hypoglossal innervation than cells in other regions.Two or more hypoglossal responses sometimes could be recruited in a single neurone

by increasing the stimulus intensity to the redirected motor nerve (e.g. Fig. 5A). Onthe average, however, hypoglossal axons contributed only 1 13± 0-19 (mean+S.E.M.)

179

18() W. PROCTOR, S. ROPER AND B. TA YLOR

A \

10mV

20 msec

|Sm. . J10 mV

20 msecFig. 5. Comparison of intracellular responses produced by redirected hypoglossal axonswith those produced by control and regenerating vagal axons. A, stimulating theredirected hypoglossus nerve in an animal 16 weeks after the operation produced small,subthreshold responses (arrow) followed by a suprathreshold post-synaptic potential. Thetwo responses could be evoked separately by adjusting the stimulus intensity and polarity,indicating that at least two hypoglossal axons had innervated this neurone. Notice thatthe suprathreshold post-synaptic potential does not outlast the action potential as in thecase for vagal responses. B, vagal responses in another animal in which both vagal nerveshad been crushed 16 weeks earlier to compare self-reinnervation with foreign innervation,above. Note the large residual depolarization after the impulse. C, vagal responses in aganglion cell from an unoperated animal. The two superimposed post-synaptic potentialswere evoked by stimulating the vagal nerve at two different intensities. The suprathresholdresponse includes a large residual component which outlasts the action potential. Thepresynaptic nerve lengths were similar in A, B and C, thereby illustrating the much fasterconduction velocity for redirected hypoglossal axons (A) than for regenerating (B) orunoperated (C) vagal axons.


inputs per innervated neurone. This is significantly less (P < 0-05) than the numberof inputs/cell found either in control ganglia (I-73 ± 007) or in vagally reinnervatedganglia (2-27+0-11; Ko & Roper, 1978).The mean conduction velocity of anastomosed hypoglossal axons innervating

experimental ganglia was 1-32 +0-70 m/sec (±S.D.; 412 inputs from twenty-twoanimals). This was significantly higher than the mean conduction velocity ofpreganglionic axons in unoperated animals or in vagally reinnervated ganglia(0 34 m/sec and 0-28 m/sec, respectively; Roper & Taylor, 1982), and was closer tothat of regenerating hypoglossal motor axons in the tongue (2-9 m/sec).

Morphology of hypoglossal terminals in the heartGanglia from experimental animals were stained with zinc iodide-osmium to reveal

the light microscopic structure of motor axons innervating the heart. Largeunmyelinated varicose nerve fibres in nerve bundles in the interatrial septum wereobserved. In some cases, varicose fibres formed a network near ganglion cells, asshown in Plate 1 D, and this appearance of varicose axons in experimental gangliadid not alter up to 60 weeks after cross-innervation. Varicose terminals weresometimes observed in close proximity to ganglion cells and their axons, but theynever appeared to spiral around the axon hillock or deposit boutons on the cell bodyas in the case of vagal terminals (McMahan & Kuffler, 1971; Roper & Ko, 1978).The absence offoreign motor terminals on neuronal perikarya ofzinc iodide-osmium

stained parasympathetic neurones, coupled with the finding that hypoglossal post-synaptic potentials usually were small and had slow rise times, suggested that foreignsynaptic innervation occurred remote from the cell body region. If the synaptic inputwere remote from the cell body, the apparent reversal potential (Er) measured at thecell body would be higher than if the synaptic input occurred on the soma, asdescribed in Roper & Taylor (1982). To test whether this was the case, cells werehyperpolarized during hypoglossal stimulation and the amplitudes of the evokedpost-synaptic potentials were plotted versus the membrane potential. Er wasextrapolated from the graph. To determine Er for synaptic responses on the soma,acetylcholine (ACh) was applied ionophoretically onto the cell body. Hypoglossalresponses in experimental animals usually had an apparent Er which was positiveto that obtained from ACh responses, consistent with the hypothesis that foreigninputs were remote from the perikaryon. From a total of forty-five animals, the meanEr for hypoglossal inputs was + 25-0 mV and the mean for ACh was - 13-9 mV. Fig.6 shows results from two cells where Er for ACh and for hypoglossal responses wasmeasured in each neurone, and Table 1 summarizes data from a subset of the aboveseries in which it was possible to obtain paired ACh potentials and hypoglossalresponses for each cell. The differences between Er for hypoglossal and ACh responsesin Table 1 were small, but reversal potentials for hypoglossal inputs were usually morepositive than for ACh potentials (9/11 cells), and the difference was significant at the0-005 level.The physiological evidence for hypoglossal innervation remote from the soma led

us to examine this further using electron microscopy and HRP labelling to identifyforeign synapses morphologically. Without a specific label, the hypoglossal origin ofremote synapses could not be distinguished from possible intrinsic connexions

181


(discussed below) using morphological criteria. Therefore, in three experimental frogsthe redirected hypoglossus nerve was stimulated repetitively while the ganglia werebathed in HRP-Ringer solution. The bathing medium also contained 2 /LM-dihydroxy-fl-erythroidine to block post-synaptic responses and thus eliminate polysynapticactivation of potential intrinsic pathways. After the development of the HRP

A -40 B

-35

-30

-25

20

-15

-10

- 5

-30

-25

-20 >Ea}m

-15 EX0CAc00)am

-10

-5

I I I I I I I_ _

-160 -120 -80 -40 0 -160 -120 -80 -40 0

Membrane potential (mV)

Fig. 6. Reversal potential measurements for ACh responses and for hypoglossal responses.A single current/recording micro-electrode in the ganglion cell body was used to alter themembrane potential and simultaneously to record responses evoked by stimulating theredirected hypoglossal nerve (open circles) or by applying acetylcholine at the soma (filledcircles). The amplitudes of nerve-evoked and ACh responses were plotted as a functionof the membrane potential for two cells (A, 11 weeks post-operative; B, 18 weekspost-operative). Least squares linear regression lines were calculated for the points andextrapolated to determine the reversal potentials for the hypoglossal and ACh responses.For A, these values were Er,(nerve) = - 12,6 mV and Er,(ACh) =-18-9 mV. For B,Er,(nerve) = + 11-8 mV and Er,(ACh) = - 199 mV.

reaction product, preparations were scanned for labelled and unlabelled synapses inthe electron microscope. Synapses which contained HRP-labelled vesicles contactedaxons of ganglion cells (Plate 2A), and not perikarya. Control ganglia which werebathed in HRP without nerve stimulation did not reveal any labelled synapses (Plate2B), nor did stimulation in the presence of 15 mMMg2+ plus HRP reveal nerveterminals containing peroxidase-labelled vesicles.

182

\

*0

* i

\*0 o

0

.

o o o o~~~~-1


Evidence that spinal motoneurone8 innervate ganglion cells in hypoglossally innervatedheartsAlthough the hypoglossus nerve in the frog is a mixed nerve containing sensory,

sympathetic and motor fibres, it is predominantly motor (Ecker, 1889; Nieuwenhuys& Opdam, 1976) and thus it was most likely that the foreign innervation of cardiacganglia was derived from spinal motoneurones in experimental animals. That thisindeed was the case was demonstrated by the following.

TABLE 1. Reversal potentials for hypoglossal responses and for ACh responses

Hypoglossal ACh Difference:Resting Er (hypoglossal)Potential Er r2 Er r2 -Er (ACh)

Experiment (mV) (mV) (mV) (mV)1 -42 +11-8 0-64 -19.9 097 +31-72 -42 -21V8 0 99 -25-1 0 99 +3-33 -40 +2-6 0-98 0.0 094 +2-64 -40 +8-8 091 +5-3 093 +3-55 -45 -19-8 0-64 -18-0 091 -1-86 -47 -9-2 0-77 -18.9 0-96 +9 77 -44 -95 0-83 -9-6 091 +0 18 -35 -13-9 0-97 -12-2 094 -1-79 -47 -12-6 0-90 -18&9 0-96 +6-310 -50 -26-9 0.95 -29-6 0-97 +2-711 -38 +14-6 035 -35 0-80 +1841

Mean= -43 -6-9 -13-7 +6-8

S.E.M. +1 +4-3 +3-3 +3 0ACh was applied ionophoretically onto the cell body and the soma was hyperpolarized during

hypoglossal and ACh responses. Reversal potentials were extrapolated by calculating the least-squares fit for the line drawn through the points (see Fig. 6). The correlation coefficients (r2) forthe linear regression lines are included in columns 4 and 6. The mean difference (+ 6-8 mV) betweenthe reversal potentials for hypoglossal inputs and for ACh responses is significant at the 04005 level(Wilcoxon signed rank test).

First, as previously stated, the conduction velocity for presynaptic axons inexperimental animals was much faster than regenerating autonomic fibres and closerto that of regenerating hypoglossal motor axons.

Secondly, in three experimental animals we dissected free the cardiac ganglion,along with the entire length of the redirected hypoglossal nerve, including dorsal andventral roots and the ramus connecting the sympathetic chain. Ganglion cells wereimpaled as before and the nerve roots and sympathetic chain were stimulatedseparately. In each case, stimulating ventral roots, but not dorsal roots, evokedpost-synaptic responses similar to those reported above. Stimulating the sympatheticchain did not produce intracellular potentials in ganglion cells; however, it causedthe isolated cardiac tissue surrounding the ganglion to begin vigorous contractionssimilar to the effects of sympathetic stimulation in experimental animals in situ (seeFig. 4).

183


Lastly, we identified the neurones which had innervated the heart in experimentalanimals by applying HRP to the regenerated axons where they entered the heart.HRP retrogradely labelled only large somatic motoneurones of the hypoglossal motornucleus (Plate 2C). On the average, 66+ 8 (mean + S.E.M.) hypoglossal spinalmotoneurones per animal were labelled by HRP in experimental frogs (n = 2). Thiscorresponds to 25 % of the entire population of hypoglossal spinal motoneurones on

one side, as determined by retrograde HRP labelling in three control frogs(mean+ S.E.M. = 262 +17 labelled motoneurones). For the latter data, HRP was

applied to the hypoglossal nerve where it exited the spinal cord and before itbranched. No peroxidase-labelled perikarya in experimental animals were observedin the vagal motor nucleus, in the autonomic motor columns, or elsewhere in the brainor spinal cord. For comparison, in two unoperated control animals, a total of 58+10preganglionic neurones (mean+ s.E.M.), located in the vagal motor nucleus, innervatedthe cardiac ganglion.

Intrinsic ganglionic synapses and foreign innervationThe above experiments do not rule out the possibility that post-synaptic potentials

produced by polysynaptic intrinsic pathways (Sargent & Dennis, 1977) may havecontributed to our data. That is, ganglion cells which were innervated by supra-

threshold hypoglossal inputs may have produced post-synaptic responses in neigh-bouring cells via intrinsic synapses and thus the extent of hypoglossal innervationmay have been over-estimated in the above values. Since few of the neurones (9 %)received suprathreshold input from the redirected hypoglossus nerve, it is unlikelythat intrinsic polysynaptic pathways were extensively activated in experimentalganglia. Nevertheless, to assess the incidence of intrinsic innervation directly, we

stimulated post-ganglionic branches of the interatrial septum after all extrinsic nerve

inputs had been severed and allowed to degenerate, as described in Roper (1976) andSargent & Dennis (1977). In animals in which the heart had been kept denervatedfor 12-13 weeks, we found that 35 % of the ganglion cells were innervated by intrinsicaxon collateral synapses (four animals, 42/120 neurones). These data are comparableto those found by Sargent & Dennis (1981). In a group of three experimental animals,the hypoglossus nerve was redirected to the heart and was allowed to establish foreigninnervation in the denervated cardiac ganglion (12-20 weeks). The hypoglossus nerve

was then severed in a second operation: 7-10 days later, when all extrinsic foreigninnervation had degenerated, we found that 23 % of the ganglion cells (23/99) receivedsynaptic input from post-ganglionic axon collaterals. The difference between theincidences of intrinsic innervation for these two series of animals was not significant(P> 0 05).These latter data can be used to estimate the incidence of intracellular responses

evoked via polysynaptic pathways. (The synaptic delays of polysynaptic responses

in denervated ganglia were too variable to use as a reliable characteristic of intrinsicsynapses.) Ifwe make the assumption that the formation of hypoglossal and intrinsicsynapses occurs independently, the expected incidence of polysynaptic responses isthe combined probability that a neurone received a suprathreshold hypoglossal input(9 %) and also formed intrinsic synapses with another ganglion cell (23%), i.e.0-09 x0-23 = 0-021. That is, only a small fraction, about 2 %, of the responses inhypoglossally innervated ganglia might be produced via polysynaptic pathways.

184


DISCUSSION

We have examined the capability of somatic motor axons to establish synapseson denervated parasympathetic ganglion cells in the frog. Although functionalhypoglossal innervation in the cardiac ganglion was widespread, transmission atforeign synapses was weak. Nevertheless, even though hypoglossal stimulation rarelyexcited ganglion cells, the small proportion which was excited (9%) appeared to havebeen sufficient to inhibit the heartbeat. It has been shown that excitation of only afew parasympathetic axons significantly inhibits the heart rate in the cat (McAllen& Spyer, 1978) and in the frog (N. McKenna & W. Proctor, unpublished observations).Furthermore, in the present experiment, a ganglionic blocker, erythroidine, was ableto block the inhibitory hypoglossal influence. Alternatively, redirected hypoglossalaxons may also have established synapses directly with muscarinic sites on cardiacmuscle fibres as well as synapsing with ganglion cells. In fact, branches of thehypoglossal axons could sometimes be traced to bundles of cardiac muscle fibressurrounding the cardiac ganglion in our histological preparations. This cholinergicinput may have augmented that from the hypoglossal-innervated ganglion cells.

Morphology offoreign synapses on cardiac ganglion cellsRegenerating hypoglossal axons in self-reinnervated geniohyoid muscles initially

formed an extensive network of unmyelinated varicose terminals like those of otherregenerating motor terminals in the frog (Letinsky, Fischbeck & McMahan, 1976).At later stages, regenerated terminals resembled nerve terminals at intact normalneuromuscular junctions. In contrast, hypoglossal motor axons in the heart remainedvaricose even up to a year after growing into the heart. The histological appearanceofforeign motor nerve terminals in the cardiac ganglion gave the impression that theywere in an arrested stage of maturation compared with motor axon regeneration inmuscles. However, well-defined synapses, as revealed by the electron microscope,were formed between hypoglossal fibres and ganglion cell axons.

Specificity of ganglion reinnervationThe finding that hypoglossal synapses occurred only on axons of parasympathetic

neurones was unexpected. Although regenerating vagal terminals first contactparasympathetic ganglion cell axons (as do foreign fibres), they rapidly restoresynaptic connexions on ganglion cell bodies (Roper & Taylor, 1982). This was notthe case for hypoglossal innervation, even at intervals of over one year after theoperation.We do not know why redirected hypoglossal axons failed to innervate neuronal

perikarya since vacant post-synaptic sites remained as potential targets on theperikaryon. Self-reinnervation by vagal axons does restore synapses on the soma(Roper & Taylor, 1982). Conceivably, the failure of foreign axons to reinnervate thecell body may merely reflect the fact that the surface area of available axonalmembrane and the numbers of vacant synaptic sites there are much greater than thaton the perikaryon. Thus, the difference would be a quantitative one whereby theperipheral arborizations of spinal motoneurones were over-extended and could onlysustain a limited number of peripheral synaptic contacts. According to this inter-pretation, hypoglossal synapses in the ganglion would be limited to the most readily

185


available sites, namely, post-ganglionic axons. If this were the case one might haveexpected a gradient of hypoglossal innervation in the ganglion, whereby parasym-

pathetic neurones at the proximal (central) end of the ganglion would receive more

extensive innervation from redirected motor axons than would more distal portionsof the ganglion. This, however, was not the case. Neurones at the distal portion ofthe ganglion were innervated to the same extent as were those in the proximal end;furthermore, in zinc iodide-osmium stained preparations, foreign axons could be seen

to bypass denervated perikarya which appeared directly in their pathway.Another explanation may be that foreign motor axons were prevented by the

surrounding satellite glial cells from reaching the neuronal perikarya. After vagotomy,glial lamellae surrounding the axon hillock region of the ganglion cells retract, leavingbehind large spaces filled with whorls of basal lamina and strands of collagen(Uchizono, 1964; Roper & Taylor, 1982). If vagal axons are allowed to regenerate,vagal terminals penetrate these spaces before depositing boutons on the axon hillockand cell body. In contrast, we did not observe hypoglossal axons within the satelliteglial spaces at the axon hillock. Thus, the glial lamellae and/or basal lamina may

represent an unfavourable environment for the hypoglossal terminals, therebyrestricting foreign synaptic contact to axonal regions of the ganglion cells.A third possibility for the distinction between axo-axonic and axo-somatic

synaptic sites is that the post-synaptic surface membrane itself differs from regionto region. There may be subtle distinctions such as differing 'chemoaffinities' (Sperry,1963) or surface adhesion properties between the axon and the perikaryon. As yet,little is known about any specific differences that may exist between the perikaryonand the axon which may be important in the formation of synaptic connexions.

We would like to express our gratitude to Drs A. R. Martin, N. McKenna, D. Purves and P.Sargent for advice and comments on the manuscript. We are indebted to Betty Aguilar, SharonFerdinandsen and Judy Paden for their expert secretarial assistance. The investigations were

supported by grants from the Colorado Heart Association, the American Heart Association andUSPHS NS1 1505 and NS00257.

REFERENCES

BERNSTEIN, J. J. & GUTH, L. (1961). Nonselectivity in establishment of neuromuscular connectionsfollowing nerve regeneration in the rat. Expl Neurol. 4, 262-275.

CALUGAREANU, M. M. & HENRI, V. (1900). IV. Suture croisee des nerfs pneumogastrique ethypoglosse. J. Physiol. Path.gen. 2, 709-711.

DENNIS, M. J., HARRIS, A. J. & KUFFLER, S. W. (1971). Synaptic transmission and its duplicationby focally applied acetylcholine in parasympathetic neurones in the heart of the frog. Proc. R.Soc. B 177, 509-539.

ECKER, A. (1889). The Anatomy of the Frog, translated by HASLAM, G. Oxford: Clarendon Press.Reprint edition, 1971. Amsterdam: Asher and Co.

HEUSER, J. E. & REESE, T.S. (1973). Evidence for recycling of synaptic vesicle membrane duringtransmitter release at the frog neuromuscular junction. J. Cell Biol. 57, 315-344.

HILLARP, N.-A. (1946). Structure of the synapse and the peripheral innervation apparatus of theautonomic nervous system. Acta anat. 2, suppl. 4, 1-153.

Ko, C.-P. & ROPER,S.(1978). Disorganized and 'excessive' reinnervation of frog cardiac ganglia.Nature, Lond. 274, 286-288.

LANGLEY, J. N. (1897). On the regeneration of pre-ganglionic and of post-ganglionic visceral nerve

fibres. J. Physiol. 22, 215-230.

186


LANGLEY, J. N. & ANDERSON, H. K. (1904). The union of different kinds of nerve fibres. J. Physiol.31, 365-391.

LETINSKY, M. S., FISCHBECK, K. H. & MCMAHAN, U. J. (1976). Precision ofreinnervation of originalpostsynaptic sites in frog muscle after a nerve crush. J. Neurocytol. 5, 691-718.

MCALLEN, R. M. & SPYER, K. M. (1978). Two types ofvagal preganglionic motoneurones projectingto the heart and lungs. J. Physiol. 282, 353-364.

MCLACHLAN, E. M. (1974). The formation of synapses in mammalian sympathetic ganglia re-innervated with preganglionit or somatic nerves. J. Physiol. 237, 217-242.

MCMAHAN, U. J. & KUFFLER, S. W. (1971). Visual identification of synaptic boutons on livingganglion cells and of varicosities in postganglionic axons in the heart of the frog. Proc. R. Soc.B 177, 485-508.

MCMAHAN, U. J., SPITZER, N. C. & PEPER, K. (1972). Visual identification of nerve terminals inliving isolated skeletal muscle. Proc. R. Soc. B 181, 421-430.

MESULAM, M.-M. (1976). The blue reaction-product in horseradish peroxidase neurohistochemistry:incubation parameters and visibility. J. Hi8tochem. Cytochem. 24, 1273-1280.

MILEDI, R. & STEFANI, E. (1969). Non-selective re-innervation of slow and fast muscle fibres inthe rat. Nature, Lond. 222, 569-571.

NIEUWENHUYS, R. & OPDAM, P. (1976). Structure ofthe brain stem. In Frog Neurobiology, ed. LLINAS,R. & PRECHT, W. Berlin, Heidelberg, New York: Springer-Verlag.

NJA, A. & PURVES, D. (1977). Re-innervation of guinea-pig superior cervical ganglion cells bypreganglionic fibres arising from different levels of the spinal cord. J. Physiol. 272, 633-651.

OSTBERG, A.-J. C., RAISMAN, G., FIELD, P. M., IVERSEN, L. L. & ZIGMOND, R. E. (1976). Aquantitative comparison of the formation of synapses in the rat superior cervical sympatheticganglion by its own and by foreign nerve fibres. Brain Res. 107, 445-470.

PEPER, K., DREYER, F., SANDRI, C., AKERT, K. & MOOR, H. (1974). Structure and ultrastructureof the frog motor endplate. Cell Tiss. Res. 149, 437-455.

PROCTOR, W. R., FRENK, S. & ROPER, S. (1978). 'Hybrid synapses produced between somatic axonterminals and autonomic neurons. Neurosci. Abs. 4, 533.

PROCTOR, W., FRENK, S., TAYLOR, B. & ROPER, S. (1979). 'Hybrid' synapses formed by foreigninnervation ofparasympathetic neurons: A model for selectivity during competitive reinnervation.Proc. natn. Acad. Sci. U.S.A. 76, 4695-4699.

PROCTOR, W., TAYLOR, B. & ROPER, S. (1979). Competition between foreign and native reinnervationof autonomic neurons. Neurosci. Abs. 5, 681.

ROPER, S. (1976). An electrophysiological study of chemical and electrical synapses on neuronesin the parasympathetic cardiac ganglion of the mudpuppy, Necturus maculosus: evidence forintrinsic ganglionic innervation. J. Physiol. 254, 427-454.

ROPER, S. & Ko, C.-P. (1978). Synaptic remodeling in the partially denervated parasympatheticganglion in the heart of the frog. In Neuronal Plasticity, ed. COTMAN, C. W. New York: RavenPress.

ROPER, S. & TAYLOR, B.'(1982). Reinnervation of denervated parasympathetic neurones in cardiacganglia from Rana pipiens. J. Physiol. 326, 155-171.

SARGENT, P. B. & DENNIS, M. J. (1977). Formation of synapses between parasympathetic neuronesdeprived of preganglionic innervation. Nature, Lond. 268, 456-458.

SARGENT, P. B. & DENNIS, M. J. (1981). The influence of normal innervation upon abnormalsynaptic connections between frog parasympathetic neurons. Devl Biol. 81, 65-73.

SPERRY, R. W. (1963). Chemoaffinity in the orderly growth of nerve fiber patterns and connections.Proc. natn. Acad. Sci. U.S.A. 50, 703-710.

STAMPFLI, R. & HILLE, B. (1976). Electrophysiology of the peripheral myelinated nerve. In FrogNeurobiology, ed. LLINAS, R. & PRECHT, W. Berlin, Heidelberg, New York: Springer-Verlag.

UCHIZONO, K. (1964). On different types of synaptic vesicles in the sympathetic ganglion ofamphibia. Jap. J. Physiol. 14, 210-219.

WOLFF, H. G., HARE, K. & CATTELL, M. (1938). Functional connections established by spinal nervesregenerating into preganglionic sympathetic nerve trunks. Am. J. Physiol. 123, 218-219.

187


EXPLANATION OF PLATES

PLATE 1A, electron micrograph of a hypoglossal nerve terminal in the geniohyoideus muscle from an

unoperated animal. This cross-section at the neuromuscular junction shows synaptic vesicles anda presynaptic thickening that lies opposite a subsynaptic cleft in the muscle fibre. A Schwann cellinvests the hypoglossal terminal and a distinct basal lamina, which surrounds the Schwann cell,extends into the synaptic cleft. G, Schwann cell; H, hypoglossal nerve terminal; M, muscle fibre;arrow, presynaptic thickening; curved arrows, basal lamina. Calibration bar = 0 5 jam.

B, photomicrograph of a zinc iodide-osmium stained geniohyoideus nerve terminal from anunoperated frog. A darkly stained unmyelinated hypoglossal nerve terminal branches and extendsalong the muscle fibre surface. Although the terminal branches are irregular, they are not markedlyvaricose. ax, hypoglossal axon; mf, muscle fibre. Calibration bar = 50 jam.

C, photomicrograph of a regenerating hypoglossal nerve terminal on the geniohyoideus musclestained with zinc iodide-osmium, 8 weeks after self-reinnervation. Arrows point to varicosities alongthe nerve ending. Abbreviations are the same as in B. Calibration bar = 50 #sm.

D, zinc iodide-osmium stained hypoglossal nerve terminals in a whole mount preparation of acardiac ganglion. Darkly stained varicose terminals form a diffuse network over a ganglion cellwhich is located below the plane of focus. Such networks of varicosities were not observed in controlganglia or in vagal-reinnervated ganglia. Calibration bar = 50 ,um.

PLATE 2A, electron micrograph of an axo-axonic hypoglossal synapse in a cardiac ganglion showing

vesicular uptake of HRP in a stimulated nerve ending. The hypoglossal nerve had been redirectedto the heart 16 weeks before. The cardiac ganglion was isolated and bathed in Ringer containingHRP, 10 mg/ml., and the redirected hypoglossus nerve was stimulated central to the anastomosis.H, hypoglossal nerve terminal; P, post-ganglionic cell axon; arrow, synaptic thickenings.Calibration bar = 0-5 ,m.

B, hypoglossal axo-axonic synapse from a ganglion bathed in HRP, as above, but without nervestimulation. Abbreviations are the same as in A. Calibration bar = 0 5 jm.

C, retrograde transport of HRP which was applied to the cut central end of a redirectedhypoglossal nerve 20 weeks after anastomosing the hypoglossus nerve to the vagosympatheticnerve supply to the heart. After 10 days following the HRP application, the animal was preparedas described in Methods. The photomicrograph (left) shows a cross-section through the medulla;two HRP-labelled spinal motoneurones are located in the hypoglossal motor nucleus (XII). Thedrawing (right) marks the labelled neurones as filled circles. Dark-staining structures elsewhere areerythrocytes and pigment cells and are readily distinguishable from retrogradely labelled neurones.fs, fasciculus solitarius; Xm, vagal motor nucleus; ri, nucleus reticularis inferior. Calibrationbar = 500,um.

188

The Journal of Physiology, Vol. 326

W. PROCTOR, S. ROPER AND B. TAYLOR

Plate I

(Facing p. 188)

The Journal of Physiology, Vol. 326 Plate 2

c

W. PROCTOR, S. ROPER AND B. TAYLOR

Documents

BY WILLIAM PROCTOR, STEPHEN ROPER AND BARBARA