Frog sympathetic ganglion cells have local axon collaterals

THE JOURNAL OF COMPARATIVE NEUROLOGY 289~294-303 (1989)

Frog Sympathetic Ganglion Cells Have Local Axon Collaterals

CYNTHIA J. FOREHAND AND LUKASZ M. KONOPKA Department of Anatomy and Neurobiology, University of Vermont College of Medicine,

Burlington, Vermont 05405

ABSTRACT Amphibian autonomic ganglia have been used as simple models for stud-

ies involving the physiology of synaptic transmission. These models assume an anatomical simplicity where the ganglion is a simple relay for central nervous system output to peripheral autonomic targets. Cholinergic preganglionic fibers innervate the soma and proximal axon of the unipolar ganglion cells, which were thought to relay the information to the periphery with little ganglionic processing. However, several different types of' synaptic potentials occur in response to preganglionic stimulation. Also, a variety of neuropeptides are found in both preganglionic fibers and ganglion cells; at least one of the peptides found in preganglionic fibers is known to act as a neurotransmit- ter in the ganglion. Finally, there may be communication between ganglion cells.

In the present study, we have explored the morphology of lumbar sympathetic chain ganglion cells by intracellular injection with horseradish peroxidase to determine whether an anatomical substrate exists for processing information within these ganglia. We have shown that 39% of these cells have axons that branch within the ganglion. While both major classes of ganglion cells (B cells and C cells) had intraganglionic axon collaterals, there was a marked difference in the frequency: 65% of the C cell axons had collaterals while only 19% of the B cell axons collateralized within the ganglion. Ultra- structural examination of labeled axon collaterals indicated that these collaterals receive synaptic input; whether the collaterals also make synapses has not been definitively established.

Key words: autonomic nervous system, synapses, Anura, horseradish peroxidase

Autonomic ganglia in the frog are traditionally described as simple relay stations. This idea has been challenged by current investigations of the complexities of membrane and synaptic mechanisms in paravertebral sympathetic chain ganglion cells in the frog (see Horn and Dodd, '83; and Weight, '83 for reviews). Thus these cells exhibit fast and slow excitatory postsynaptic potentials (EPSPs) mediated by nicotinic and muscarinic acetylcholine receptors, respectively, an inhibitory postsynaptic potential (IPSP) also mediated by muscarinic acetylcholine receptors, and a late slow EPSP mediated by the peptide luteinizing hormone-releasing hormone (LHRH) (Jan and Jan, '83). In addition to LHRH, which has been shown to act as a transmitter in these ganglia, other neuropeptides are found in both ganglion cells and their preganglionic inputs. These peptides mark functionally specific pathways through the ganglia (Horn and Stofer, '87; Hornet al., '87, '88). Finally, evidence

exists that ganglion cells may interact with one another via catecholamine release within the ganglion (Minota and Ko- ketsu, '78; Nishimura and Akasu, '88). The anatomical substrate for this interaction is unknown, but the suggestion has been made that catecholamine released directly from the somata of active neurons may play a role (Suetake et al., '81; Nishimura and Akasu, '88).

Perhaps the only aspect of presumed simplicity within frog sympathetic ganglia that has not been challenged is the view of ganglion cells as simple unipolar neurons (see Taxi, '76, for review). This view is supported by both light and electron microscopic observations on ganglion cell morphology. These observations indicate a single process arising from individual ganglion cells. While this description is

Accepted May 26,1989.

0 1989 ALAN R. LISS, INC.

AXON COLLATERALS IN AUTONOMIC GANGLIA

valid to a certain extent, no one has actually performed electron microscopic reconstructions of individual ganglion cells to ensure only a single process exists or to determine whether the known process branches within the ganglion. Nor has anyone systematically explored the morphology of these cells with intracellular injection of visualizable sub- stances. In the present study, we have examined the morphology of neurons in the lumbar chain ganglia of the frog with the intracellular marker horseradish peroxidase (HRP). We have also taken advantage of the fact that there are two types of principal ganglion cells (B cells and C cells) that receive preganglionic input from anatomically separate preganglionic fibers (B fibers and C fibers, respectively) (Nishi et al., '65; Skok, '65; Libet e t al., '68; Francini and Urbani, '73) to ask whether there are differences in morphology between these two populations of cells, which are known to be functionally distinct (Horn et al., '87, '88). We found that these ganglion cells are indeed unipolar, but a significant number of them (39%) have axons that branch within the ganglion. While both B cells and C cells can have axon collaterals within the ganglion, collaterals were found three times as often on C cells compared with B cells. Ultra- structural examination of labeled axon collaterals indicates that these collaterals receive synaptic input; whether the collaterals also make synapses has not been definitively established. Some of these results have appeared in abstract form (Forehand and Konopka, '88).

295

MATERIALS AND METHODS Experiments were performed on eight bullfrogs and two

grass frogs. The animals were pithed and the fourth through tenth sympathetic chain ganglia and associated ventral roots were removed bilaterally and placed in a recording chamber filled with Ringer's solution. Cells in the ninth and tenth lumbar ganglia were impaled with glass microelec- trodes filled to the shoulder with 5% HRP (Sigma, Type VI) in 0.2 M potassium acetate buffered to pH 7.6 and back- filled with 3 M potassium chloride. Ganglion cells were identified by a resting membrane potential of a t least 30 mV and an action potential generated in response to depolarizing current injection or an anode break spike following hyper- polarizing current injection. In some experiments, the fourth through sixth or the seventh and eighth ventral roots were drawn into suction electrodes for synaptic activation of B cells or C cells, respectively. Ganglionic B cells are innervated by rapidly conducting preganglionic axons (-2.5 m/ second) from the sympathetic chain above ganglion 7 , whereas C cells are innervated by slowly conducting preganglionic axons (<1 mhecond) from spinal nerves 7 and 8 (see Dodd and Horn, '83, for review). Ganglion cells in these experiments were identified by the presence of an action potential in response to stimulation of their preganglionic inputs in the ventral roots. In each ganglion, HRP was ion- tophoresed into three to six cells for 5 minutes each with 50 msec depolarizing pulses of 1 nA at 5 Hz; 1 hour was allowed from the time of the last cell injection before fixation. Gan- glia were fixed overnight in 1.25 % glutaraldehyde and 0.5 % paraformaldehyde in HEPES buffer (pH 7.6 a t 4OC). They were then rinsed in the buffer and processed en bloc for the histochemical visualization of HRP with the method of Hanker e t al. ('77) for subsequent light microscopic examination or with diaminobenzidine (DAB) (Graham and Kar- novsky, '66) for electron microscopic examination.

Fig. 1. Photomicrographs of intraganglionic axonal branch sites on two HRP-labeled paravertebral ganglion cells. A Two branches (filled arrows) arise from the main shaft of a ganglion cell axon. The cell body is cut off at the top of the field. The main axon that eventually exited the ganglion is marked with open arrows. Calibration = 25 pm. B: A single intraganglionic branch (filled arrow) arises from this cell. The termination of this branch within the ganglion is marked with an arrowhead (part of the length of the branch is out of the focal plane). Open arrows indicate the portion of the axon that exited the ganglion. Calibration =

25 Fm.

For light microscopy, ganglia were dehydrated and cov- erslipped in DPX (BDH Chemicals, Ltd.). Ganglion cells were drawn with the aid of a camera lucida at 400x. Mor- phometric measurements were obtained from the camera lucida drawings with the aid of a digitizing tablet, computer, and Neurolucida software (Glaser and Glaser, '88). Data col- lected included the area of the cell bodies, the number of

296 C.J. FOREHAND AND L.M. KONOPKA

u 100 urn

Fig. 2. Camera lucida drawings of three HRP-labeled paravertebral ganglion cells, each with intraganglionic axonal branches. On each drawing, the portion of the axon that exited the ganglion is marked with an asterisk. The numbered arrows on the left (arrow 1) and middle (arrow 2) cells indicate the type of area sectioned for electron microscopy and shown in Figures 6 and 7, respectively. The middle cell corresponds to Figure 1A.

axon branches, the length of axon branches, and the distance from the cell body at which the branches arose.

Two ganglia containing HRP-labeled cells were processed for electron microscopy. After histochemical processing for the visualization of HRP, the ganglia were embedded in rapidly fixed gelatin-albumin (Bowers and Zigmond, '79) and sectioned serially at 40 bm on a vibratome. The sections were postfixed in 1 % osmium tetroxide in phosphate buffer for 30 minutes, rinsed, and dehydrated through graded alco- hols. The sections were wafer-embedded in Araldite with propylene oxide as the transition solvent. Embedded sections were examined a t 400x and two sections containing a cell with a branched axon were identified. These sections were serially thin sectioned and stained with lead citrate and uranyl acetate.

RESULTS A total of 135 HRP-labeled cells were analyzed. Only

those cells whose axons were labeled a t least as far as their exit from the ganglion were included in the analysis. (About 75', of the injected cells met this criterion.) Fifty-three of the cells (39'0 of the total number analyzed) had axon branches that arose within the ganglion (Fig. 1). Thirteen of the 53 cells with intraganglionic axon collaterals had more than one collateral; thus 81 collaterals were observed on the 53 cells; some cells had as many as six different branches arising from the axon within the ganglion. Camera lucida drawings of three typical cells with intraganglionic axon collaterals are shown in Figure 2.

The length of individual intraganglionic axon collaterals ranged from 5 to 331 pm with an average length of 66 pm

(Fig. 3A). Collaterals arose from the axon anywhere from immediately (i.e., at the junction of the axon and cell body) to 427 pm away from the soma, with the average distance being 76 Frn (Fig. 3B). There was no correlation between the length of the axonal branches and the distance from the cell body a t which they arose.

In an initial effort to determine whether the 39% of ganglion cells with local axon collaterals represented a func- tional class of cells, the size distribution of cells with and without collaterals was examined (Fig. 4A). The size distri- butions were not obviously different, but there was a tendency for the largest cells not to have axon collaterals. Given this tendency, and because B cells are generally larger than C cells (Dodd and Horn, '83), some cells were identified physiologically as B cells or C cells prior to filling with HRP. Twenty-seven identified B cells and 20 identified C cells were examined. While local axon collaterals were observed on both cell types, there was a marked difference in the rela- tive frequency of branches: 65% of the C cells had axon branches while only 19% of the B cells had branches (Fig. 4B,C).

In addition to the presence of intraganglionic axon collaterals on many of these frog sympathetic ganglion cells, some other unexpected morphologies observed with much less frequency should be mentioned. In general, these cells are, as they are usually described, unipolar. However, two cells were observed with processes other than the axon, presumably dendrites, arising directly from the soma. One of these cells had three such processes and the other two. These cells were not the small multipolar interneurons (SIF cells) found in many sympathetic ganglia since they had large cell bodies and an axon which exited the ganglion. Moreover, SIF cells

AXON COLLATERALS IN AUTONOMIC GANGLIA 297

tional significance of such doublets are unknown, we have also observed them (more frequently) in frog cardiac ganglion cells both by HRP injection and by electron microscopic observation; a t the ultrastructural level in the cardiac ganglion we have observed two separate cell bodies, with separate nuclei, joined by a cytoplasmic bridge (Forehand and Konopka, unpublished).

In order to determine whether the intraganglionic axon collaterals observed on many of these cells may be involved in local circuit interactions within the ganglion, a prelimi- nary ultrastructural analysis of two HRP-labeled cells with axon branches was made. In each case, the axon collateral received synaptic input from an unlabeled axon of unknown origin (Figs. 6, 7). The axon collateral of one of these cells made intimate contact with an unlabeled axon whose morphology resembled a ganglion cell axon; however, the den- sity of the HRP reaction product in the contact site made it impossible to determine whether a synapse was made (Fig. 6). Interestingly, the collateral branch of the other cell examined ultrastructurally ended in a growth-cone-like structure that received several synapses from unknown sources (Fig. 7).

0

B

m w I 0 Z Q K m L 0

100 200 300 400 500 LENGTH OF AXON BRANCH (urn)

0 100 200 300 400 500

DISTANCE OF BRANCH POINT FROM SOMA ( u r n )

Fig. 3. Histograms of the lengths of individual intraganglionic axon branches (A) and the distance from the soma at which they arose (B).

in bullfrog paravertebral ganglia do not have processes (Weight and Weitsen, '77). While ganglion cells with dendrites are not too surprising, four other cells exhibited a somewhat bizarre morphology. Each of these cells was actually two cells connected together, either with two separate axons or with a single axon between them (Fig. 5). These doublets were clearly two cell bodies separated by a connecting piece of cytoplasm that was about the diameter of the initial part of a normal ganglion cell axon. Two of these doublets had a short connecting segment between the somata and a single axon leaving one of the somata as shown in Fig- ure 5A. The other two doublets each had two axons; one of these doublets was similar to that shown in Figure 5A, while the other was quite different in that the two cells were joined by their axons at a considerable distance from the somata (Fig. 5B). Two separate axons exited the ganglion from the connecting point between the two cells shown in Figure 5B; other axonal branches that did not exit the ganglion also arose from this connecting point. While the func-

DISCUSSION This study using intracellular injection of HRP to delin-

eate the morphology of frog sympathetic ganglion cells indicates that these cells are more complex than previously imagined. While the cells are unipolar, a significant percentage of them (39%]) have axons that issue collaterals within the ganglion. These collaterals may be involved in local circuit interactions within the ganglion as evidenced by the presence of synapses on the collaterals and by intimate contacts made by the collaterals onto other processes within the ganglion. There may be some specificity in these interactions in that C cells were three times as likely to have intraganglionic axon collaterals as were B cells.

Intraganglionic axon coilaterals have not been widely reported in sympathetic ganglia. Ganglion cell morphology has been studied by intracellular HRP injection in a variety of mammalian sympathetic and parasympathetic ganglia with a very low reported incidence (tl R ) of axon collaterals (Purves and Hume, '81; Forehand and Purves, '84; Purves and Lichtman, '85; Snider, '86, '87). These reports do indicate, however, that ganglion cell axons often arise from a common trunk with a dendrite. Moreover, mammalian ganglion cell axons receive synapses close to their point of origin (Forehand and Purves, '84; Forehand, '87). In this sense, frog sympathetic neurons are similar to these mammalian cells; the axon of frog sympathetic cells close to the cell body is densely innervated (Taxi, '76; Jan et al., '80; Marshall, '81). The ultrastructural morphology of frog sympathetic axons where they arise from the cell body is not typical of an axon hillock; however, typical hillock morphology is observed some distance from the soma (Taxi, '76). This observation, coupled with the dense innervation of the axon near the soma, led Taxi ('76) to suggest that perhaps the actual axon initiation site was distal to the cell soma. One wonders whether the process leaving the cell body, and perhaps the collaterals reported here, may actually be dendritic in nature. The HRP reaction product prevents an ultrastructural determination of the axonal or dendritic nature of the intraganglionic collaterals. Our initial attempts to get at this question with antibodies specific to dendritic or axonal proteins have been inconclusive; however, recent evidence in


A UNBRANCHED CEUS 0 BRANCHED CEUS 0

3000

24s N-

E - lgoo

2

250

28001 2400

0 0 0 1600

1200

0 0 0

- 0 1.0 0 0 0

800 rn I 0 4 8 12 16

CELL'S I D P ( T I F I ~ T l 0 N NUYEER

Fig. 4. Size distribution of HRP-labeled paravertebral ganglion cells with (filled circles) and without (open circles) intraganglionic axon branches. A The distribution of all labeled cells is shown (n = 135 cells). B: The distribution of 27 physiologically identified HRP-labeled B cells

the cardiac ganglion of Necturus indicates that the proximal part of the single process arising from these amphibian parasympathetic neurons contains dendrite-like rather than axon-like cytoskeletal proteins (McKeon and Parsons, '89).

Intraganglionic axon collaterals (or cells with more than one axon) have been reported in both mammalian and amphibian autonomic ganglia during development and in response to experimental manipulation. Snider ('86) reported that 10% of neonatal rat lumbar sympathetic neurons possessed two axons, though all of these cells had only a single unbranched axon in the adult. Similarly, developing cardiac ganglion cells in X e n o p u s exhibit more than one axon, whereas the adult cells possess a single axon that branches only a t the target (Heathcote and Sargent, '85). These Xenopus neurons differ from their counterparts in Rana, where extensive axon collateralization is observed close to the soma as well as at the target (Forehand and Konopka, '88). If neonatal rats are treated with injections of nerve growth factor (NGF), approximately 50% of superior cervical ganglion cells exhibit intraganglionic axonal branches (Snider, '88). These cells also exhibit extremely

A

15001 nc

l::I 600

0

0

0 0

300 4 0 5 10 15

CELL'S IDENllFlC4TlON NUYBER

is shown. C: The distribution of 20 physiologically identified HRP- labeled C cells is shown. The insets in B and C show typical orthodrom- ically elicited action potentials for physiologically identified B and C cells, respectively. Calibration: x = 20 msec, y = 20 mV.

complex dendritic arbors compared with superior cervical ganglion cells a t the same age in untreated rats and are growing more rapidly than their normal counterparts. Intra- ganglionic axon collaterals are prevalent on frog sympathetic ganglion cells whose axons are regenerating in response to a crush or transection (Kelly et al., '89). These authors examined only B cells and noted that in normal frogs, 20'0 of the B cells in the lumbar ganglia had axon col-

Fig. 5. Photomicrographs of HRP-labeled paravertebral ganglion cells. In each case only a single cell was intracellularly filled with HRP, but two cell bodies are apparent with a physical connection between them. A: Two cell bodies connected by a thick process (single arrow) with a single axon (double arrows) that exited the ganglion. Calibra- tion = 50 pm. B: Two cell bodies (open arrows) connected by an anastomosis of their axons (large arrow). Several branches arise from the anastomosis, and two separate axons come off this site and exit the ganglion (small arrows marked 1 and 2 ) . Calibration = 50 Fm. C: Higher magnifi- cation of the anastomosis shown in B. The axons arising from the cell bodies are marked with open arrows. Two axonal branches (small arrows 1 and 2) exit the ganglion. Calibration = 50 Mm.


Figure 5


Fig. 6. Electron micrographs of HRP-labeled paravertebral ganglion cell axon. A. The main shaft of a labeled axon is marked with asterisks. A thin branch arose from the main axon (out of the plane of section) and contacted a nearby axon (arrow) that is presumably a ganglion cell axon based on similar morphology and glial ensheathment compared with the labeled axon. The branch site was approximately 100 ym from the soma and was similar a t the light microscopic level to the area in Figure 2

marked arrow 1. Calibration = 1 ym. B: Two unlabeled synapses (arrows) contact the main shaft of the labeled axon. This section is the same area shown in A, but is 0.5 pm deeper. While the postsynaptic ultrastructure is obscured by the HRP reaction product, a clustering of synaptic vesicles and a membrane thickening can be seen on the presynaptic side. Calibration = 0.5 Wm.


Fig. 7. Electron micrographs of the terminal end of an HRP-labeled intraganglionic axonal branch. A The termination of this axonal branch (dark process) resembled a growth cone in that several filopodia-like extensions (arrows) arose from it. A t the light microscopic level, this termination point was similar to the point in Figure 2 marked arrow 2. Cal-

ibration = 1 pm. B An unlabeled synapse (arrow) contacts the terminal area of this axon branch. Several synapses contacted this area; the one shown is from a section about 0.3 pm deep to the section shown in A and corresponds to the vesicle-filled profile contacting the lower left side of the process in A. Calibration = 1 pm.


laterals. This percentage agrees with our observation of 19 Yo of B cells exhibiting axon collaterals. The response to injury resulted in intraganglionic axonal sprouting to the extent that nearly 100% of the cells had collaterals. Interestingly, the percentage of B cells with axon collaterals fell back to the pre-injury figure of 20% if the axons were allowed to regenerate; however, if the axons were prevented from regenerating, the percentage of axon collaterals remained high (Kelly et al., '89). A common theme to these instances where intraganglionic axon collaterals have been observed appears to be that the cells are actively growing. In that regard, it is interesting that the tip of the axon collateral we examined ultrastructurally resembled a growth cone. That ganglion cells may be growing in the adult frog is not surprising given that the dendrites of mammalian autonomic ganglion cells are known to increase in length throughout the life of the animal in parallel with the increase in animal size (Voyvodic, '87). This neuronal growth is thought to be related to the size of the target innervated by a mechanism of trophic stimulation, and it occurs across and within species (see Purves et al., '88, for review).

An intriguing aspect of the presence of intraganglionic axon collaterals in normal frog sympathetic ganglia con- cerns their possible function. Frog sympathetic ganglion cells are adrenergic (Kojima et al., '78). The presence of axon collaterals implies that adrenaline may be released within the ganglion when these cells are active. Adrenaline has several direct effects on frog sympathetic ganglion cells: It produces a fast, a slow, and a mixed depolarizing response (Akasu, '88). Both the fast and slow adrenaline-induced de- polarizations in these neurons are blocked by the P-adreno- receptor antagonist propranolol (Akasu, '88). An investigation of the ionic basis for the slow adrenaline depolarization (Akasu, '88) showed that adrenaline causes this response by blocking the M-current, which is derived from voltage- dependent K' channels (Adams et al., '82a,b). The observed responses to adrenaline indicate that adrenaline released within the ganglion could exert a modulatory role on ganglion cell excitability (Akasu, '88). This idea was bolstered by the report of Suetake et al. ('81) that active ganglion cells could release adrenaline from their somata; apparently, the possibility of local axon collaterals was not considered. In addition to a direct effect on ganglion cells, adrenaline may act on preganglionic terminals within the ganglion. Direct intracellular stimulation of bullfrog sympathetic ganglion cells results in recurrent synaptic activation in 5% of the cells (referred to as type 2 cells) (Minota and Koketsu, '78). This response is blocked both by curare (Minota and Koketsu, '78) and by propranolol (Nishimura and Akasu, '88). Minota and Koketsu ('78), and Nishimura and Akasu ('88) suggest that the response which characterizes type 2 cells is a result of the ganglion cell releasing adrenaline from its soma; the adrenaline would then activate the preganglionic nerve terminals impinging on the ganglion cell. While this scenario is possible, we suggest another possibility, which is that ganglionic axon collaterals contact the preganglionic fibers. Short axon collaterals arose close to the cell body in many instances and were thus in the vicinity of the preganglionic inputs. Electrophysiological characteriza- tions are difficult with HRP-filled electrodes in these ganglia and the percentage of type 2 cells is quite small so we were not able to test this hypothesis directly. Whether type 2 cells are more prevalent in the C cell population, which has a high percentage of axon collaterals, is not known. We have not yet examined proximal axon collaterals at the ultra-

structural level; we are currently extending the electron microscopic investigation of intraganglionic axon collaterals to determine both the source of synapses to these collaterals and which ganglionic elements are contacted by the collaterals.

In summary, we have shown that in frog lumbar sympathetic ganglia 19% of the B cells and 65% of the C cells exhibit intraganglionic axon collaterals. Synapses are made on these collaterals, and the collaterals contact other ganglionic elements. The collaterals may reflect continued growth of the ganglion cells in the adult frog. These results provide an anatomical framework for the consideration of physiological interactions within the ganglia.

ACKNOWLEDGMENTS This study was supported by National Science Founda-

tion grant BNS-8646758 and by a Grant-in-Aid from the American Heart Association (C.J.F.), and by National Insti- tutes of Health grant NS-25973 to Dr. Rod Parsons (L.M.K.). We are indebted to Greg Hendricks, Phil Prather, and Brad Vietje for technical assistance.

LITERATURE CITED Adams, P.R., D.A. Brown, and A. Constanti (1982a) M-currents and other

potassium currents in bullfrog sympathetic neurones. J. Physiol. (Lond.) 330537-572.

Adams, P.R., D.A. Brown, and A. Constanti (1982b) Pharmacological inhibi- tion of the M-current. J. Physiol. (Lond.) 332.223-262.

Akasu, T. (1988) Adrenaline depolarization in paravertebral sympathetic neurones of bullfrogs. Pflugers Arch. 411.80-87.

Bowers, C.W., and R.E. Zigmond (1979) Localization of neurons in rat superior cervical ganglion that project into different postganglionic trunks. J. Comp. Neurol. 185~381-392.

Dodd, J., and J.P. Horn (1983) A reclassification of B and C neurones in the ninth and tenth paravertebral sympathetic ganglia of the bullfrog. J. Phy- siol. (Lond.) 334255-269.

Forehand, C.J. (1987) Ultrastructural analysis of the distribution of synaptic boutons from labeled preganglionic axons on rabbit ciliary neurons. J. Neurosci. 7:3274-3281.

Forehand, C.J., and L.M. Konopka (1988) Bullfrog autonomic ganglion cells have local axon collaterals. SOC. Neurosci. Abstr. 14:1314.

Forehand, C.J., and D. Purves (1984) Regional innervation of rabbit ciliary ganglion cells by the terminals of preganglionic axons. J. Neurosci. 4:l- 12.

Francini, F., and F. Urbani (1973) Organization of the efferent extraspinal pathways in Rana escuelenta. Arch. Fisiol. 70:149-165.

Glaser, J.R., and E.M. Glaser (1988) Neurolucida: A PC based 3-dimensional imaging system for visual and video neuron tracing. SOC. Neurosci. Abstr. 14550.

Graham, R.C., Jr., and M.J. Karnovsky (1966) The early stages of absorption of injected horseradish peroxidase in the proximal tubules of the mouse kidney: Ultrastructural cytochemistry by a new technique. J. Histochem. Cytochem. 14291-302.

Hanker, J.S., P.E. Yates, C.B. Metz, and A. Rustioni (1977) A new, specific, sensitive and non-carcinogenic reagent for the demonstration of horseradish peroxidase. Histochem. J. 9:789-792.

Heathcote, R.D., and P.B. Sargent (1985) Loss of supernumerary axons during neonatal morphogenesis. J. Neurosci. 5~1940-1946.

Horn, J.P., and J. Dodd (1983) Inhibitory cholinergic synapses in autonomic ganglia. Trends Neurosci. 6:180-184.

Horn, J.P., and W.D. Stofer (1987) Calcitonin gene-related peptide-like immunoreactivity appears to be present in preganglionic sympathetic B- type, but not C!-type, neurons in the bullfrog. SOC. Neurosci. Abstr. 13297.

Horn, J.P., S. Fatherazi, and W.D. Stofer (1988) Differential projections of B and C sympathetic axons in peripheral nerves of the bullfrog. J. Comp. Neurol. 278:570-580.

Horn, J.P., W.D. Stofer, and S. Fatherazi (1987) Neuropeptide Y-like immu-


noreactivity in bullfrog sympathetic ganglia is restrided to C cells. J. Neurosci. 7~1717-1727.

Jan, Y.N., and L.Y. Jan (1983) Coexistence and corelease of cholinergic and peptidergic transmitters in frog sympathetic ganglia. Fed. Proc. 42t2929- 2933.

Jan, L.Y., Y.N. Jan, and M.S. Brownfield (1980) Peptidergic transmitters in synaptic boutons of sympathetic ganglia. Nature 288:380-382.

Kelly, M.E.M., A.G.M. Bulloch, K. Lukowiak, and M.A. Bisby (1989) Regen- eration of frog sympathetic neurons is accompanied by sprouting and retraction of intraganglionic neurites. Brain Res. 477:363-368.

Kojima, H., S. Anraku, K. Onogi, and R. Ito (1978) Histochemical studies on two types of cells containing catecholamines in sympathetic ganglia of the bullfrog. Experientia 34:92-93.

Libet, B., S. Chichibu, and T. Tosaka (1968) Slow synaptic responses and excitability in sympathetic ganglia of the frog. J. Neurophysiol. 31~383- 395.

Marshall, L.M. (1981) Synaptic localization of bungarotoxin binding which blocks nicotinic transmission a t frog sympathetic neurons. Proc. Natl. Acad. Sci. USA 78:194%1952.

McKeon, T.W., and R.L. Parsons (1989) Cytoskeletal proteins in postganglionic neurons and SIF cells in the cardiac parasympathetic ganglion of Necturus. SOC. Neurosci. Abstr. 15:in press.

Minota, S., and K. Koketsu (1978) Recurrent synaptic activation of the bullfrog sympathetic ganglion cells by direct intracellular stimulation. Jpn. J. Physiol. 28r799-806.

Nishi, S., H. Soeda, and K. Koketsu (1965) Studies on sympathetic B and C neurons and patterns of preganglionic innervation. J. Cell. Comp. Phy- siol. 66:19-32.

Nishimura, T., and T. Akasu (1988) Propranolol blocks recurrent synaptic activation in paravertebral sympathetic ganglia of bullfrogs. Neurosci. Lett. 86~306-310.

Purves, D., and R.I. Hume (1981) The relation of postsynaptic geometry to

the number of presynaptic axons that innervate autonomic ganglion cells. J. Neurosci. 5441-452.

Purves, D., and J.W. Lichtman (1985) Geometrical differences among homol- ogous neurons in mammals. Science 228.298-302.

Purves, D., W.D. Snider, and J.T. Voyvodic (1988) Trophic regulation of nerve cell morphology and innervation in the autonomic nervous system. Nature 336:123-128.

Skok, V.I. (1965) Conduction in tenth ganglion of the frog sympathetic trunk. Fed. Proc. [Tranl. Suppl.] 24:T363-T367.

Snider, W.D. (1986) Rostrocaudal differences in dendritic growth and synap- togenesis in rat sympathetic chain ganglia. J. Comp. Neurol. 244.245- 253.

Snider, W.D. (1987) The dendritic complexity and innervation of subman- dibular neurons in 5 species of mammals. J. Neurosci. 7r1760-1768.

Snider, W.D. (1988) Nerve growth factor enhances dendritic arborization of sympathetic ganglion cells in developing mammals. J. Neurosci. 8~2628- 2634.

Suetake, K., H. Kojima, K. Inanaga, and K. Koketsu (1981) Catecholamine is released from non-synsptic cell-soma membrane: Histochemical evidence in bullfrog sympathetic ganglion cells. Brain Res. 205,436-440.

Taxi, J. (1976) Morphology of the autonomic nervous system. In R. Llinas and W. Precht (eds): Frog Neurobiology. Heidelberg: Springer-Verlag, pp. 93-150.

Voyvodic, J.T. (1987) Development and regulation of dendrites in the rat superior cervical ganglion: Influence of preganglionic innervation. J. Neu- rosci. 7~904-912.

Weight, F.F. (1983) Synaptic mechanisms in amphibian sympathetic ganglia. In L.-G. Elfvin (ed): Autonomic Ganglia. Chichester: John Wiley & Sons, pp. 309-344.

Weight, F.F., and H.A. Weitsen (1977) Identification of small intensely fluo- rescent (SIF) cells in bullfrog sympathetic ganglia. Brain Res. 128~213- 226.

Documents

Frog sympathetic ganglion cells have local axon collaterals