THE JOURNAL OF COMPAR,A'I"E NEUROLOGY 29tk334-342 (1990)

Morphologyof Sympathetichgangjionic Neurons in the Neonatal Rat Spinal C o d AnInimdhlarHorserPeroxidme Study CYNTHIA J. FOREHAND Department of Anatomy and Neurobiology, University of Vermont College of Medicine, Burlington, Vermont 05405

ABSTRACT Understanding the central neural control of autonomic functions requires a knowledge of the morphology of the preganglionic neurons, for the location of the dendritic arborizations of these neurons will indicate which central pathways may have access to them. In the present study, individual sympathetic preganglionic neurons in the neonatal rat spinal cord have been examined by the intracellular injection of horseradish peroxidase (HRP) in an in vitro preparation. Seventeen HRP-labeled preganglionic neurons in thoracic segments Tl-T3 were examined in detail; of these, 12 somata were located in the intermediolateral cell column (IML), one in the lateral funiculus (LF), two in the intercalated nucleus (IC), and two at the border between IML and IC. All of the neurons had extensive dendritic arborizations arising from an average of six primary dendrites; the average total dendritic length for these cells was 2,343 pm. The morphology of preganglionic neurons differed depending on the location of their cell bodies. Preganglionic neurons located in the IML were essentially two-dimensional:the cells had some dendrites that coursed rostrocaudally for 300-500 pm within the IML and others that coursed mediolaterally,extending to the lateral surface of the cord and close to the central canal. Axons of these cells coursed ventrally from the cell body and exited from the spinal cord at the first ventral root caudal to the cell body. No intraspinal axon branches were observed. In contrast, preganglionic neurons with somata outside the IML had dendritic arborizations that radiated away from the cell body in three dimensions, extending into the IML, LF, and the dorsal and ventral horns of the spinal cord while traversing a rostrocaudal distance of about 400 pm. Two of these cells had axons that branched within the spinal gray. These results indicate that the dendrites of neonatal rat sympathetic preganglionic neurons extend outside the IML, where they may be contacted by inputs that do not terminate within IML. Moreover, the different dendritic arborizations of preganglionic neurons in different preganglionic nuclei may mean that the connectivity of preganglionic neurons varies according to cell body location. Finally, the intraspinal axon collaterals on some preganglionic neurons provide an anatomical substrate for local circuit interactions. Key words: autonomic,cardiovascularregulation

Over the last decade it has become apparent that the traditional view of autonomic regulation, which states that the major role of the sympathetic nervous system is to act en masse during fight or flight situations (Cannon, '151,is no longer tenable. Whereas this type of response clearly does occur, autonomic control is now recomized as a finely tuned system that plays an active role in minute-to-minute maintenance Of autonomic functions. A concept in this view is that there is differential Control Over Separate autonomic functions (Korner, '79; Schramm, '82). An O

1990 WILEY-LISS, INC.

implication of differential control is that the autonomic nervous system must be organized in a highly ordered and selective fashion. In fact, at the most peripheral level, specific populations of ganglion cells are known to innervate separate target organs (Sjoqvist, '63; Lichtman et al., Accepted May 23, 1990. Address reprint requests to Dr. C.J. Forehand, Department of Anatomy and Neurobiology, University of Vermont College of Medicine, ~i~~~ Building, Burlington, VT 05405.

MORPHOLOGY OF MAMMALIAN PREGANGLIONIC NEURONS '79; Meckler and Weaver, '84). Moreover, ganglion cells projecting to different organs can be discretely activated by complex reflex patterns (Weaver, '82). Because preganglionic neurons represent the final common pathway for central autonomic regulation, differential activation of targets requires discrete functional groupings of preganglionic neurons. Recent evidence indicates that there are targetspecific subnuclei within the intermediolateral cell column (Appel and Elde, '88). In addition to the intermediolateral cell column (IML),there are three other anatomical groupings of sympathetic preganglionic neurons: the lateral funiculus (LF) adjacent to the IML, the intercalated nucleus (IC) medial to the IML, and an area dorsal and lateral to the central canal (Petras and Cummings, '72; Hancock and Peveto, '79). These nuclei may represent functionally different groups of preganglionic neurons since there is a tendency for the more medial preganglionic neurons to preferentially innervate prevertebral ganglia involved in visceromotor structures (Petras and Cummings, '78; McLachlan et al., '85). Most of what is known about the organization of preganglionic neurons is derived from cytoarchitectural or retrograde tracing studies. These studies are limited to delineating the position of preganglionic somata and proximal dendrites. In order to understand the descending and propriospinal control of preganglionic neurons, the organization of the dendritic arborization of these neurons must also be known. In 1972, RBthelyi performed an extensive investigation on Golgi-impregnatedcat preganglionic neurons and concluded that the dendrites of these neurons were oriented primarily in the longitudinal axis and stayed within the IML, thus making IML a "closed nucleus"; afferent input to these cells would have to enter the IML directly. This interpretation has been called into question by retrograde tracing studies (e.g., Vera et al., '86) and immunohistochemical localization of choline acetyltransferase in the intermediate spinal gray (Barber et al., '84). These studies indicate that there may be both a longitudinal and a transverse orientation of preganglionic dendrites. In the present study, I have examined the morphology of preganglionic neurons in the neonatal rat spinal cord by intracellular injection of horseradish peroxidase (HRP). I utilized an in vitro preparation of the spinal cord originally described by Otsuka and Konishi ('74) and modified by McKenna and Schramm ('83) to obtain recordings from preganglionic neurons. The results demonstrate that most neonatal rat preganglionic neurons located in IML have extensive dendritic arborizations in both the rostrocaudal and mediolateral axes of the spinal cord, with very little dorsoventral arborization. Moreover, preganglionic neurons located in IC or LF were quite different from IML cells in their morphologies; these cells had dendrites radiating away from the cell body in three dimensions. Two of the HRP-labeled preganglionic neurons with somata outside IML had axon collaterals within the spinal gray. The morphology of neonatal rat preganglionic neurons is similar to that recently reported in the adult pigeon (Cabot and Bogan, '87),but different from that reported in the adult cat (Dembowsky et al., '85a). Some of these results have appeared in abstract form (Forehand, '85).

335

immersed in ice for 10-20 minutes, and sacrificed by decapitation. The vertebral column and surrounding tissue were removed and placed in a dissecting dish filled with ice-cold, oxygenated, buffered saline (composition in mM: Na' 149, K' 5, Mg+' 1, Ca++2, C1- 136, H,PO,- 1,HC0,24, glucose 11)maintained at pH 7.4 via aeration with 95% 0,/5% CO,. The spinal cord was exposed by a dorsal laminectomy and the spinal roots cut distal to the dorsal root ganglia. The spinal cord was then removed from the vertebral column by grasping the sacral end and gently lifting the cord while cutting the dentate ligaments and freeing the dorsal and ventral roots from the surrounding tissue. The dura was removed and the dorsal and ventral roots separated by cutting just proximal to the dorsal root ganglia. A longitudinal hemisection was accomplished by grasping each side of the sacral cord with forceps and gently pulling the cord apart along the midline. A section of the hemisected cord 5-8 segments (lower cervical-upper thoracic) in length was removed, placed in fresh chilled saline, and allowed to come to room temperature.

The isolated hemicord was placed in a recording chamber and continuously superfused with oxygenated saline at room temperature. The recording chamber was a small petri dish (60 x 15 mm) containing sylgard to a depth of 2 mm. The spinal cord was placed lateral surface up on a nylon gauze platform and pinned to the sylgard with minutien pins through its rostral and caudal ends. The dorsal and ventral root of a particular segment were drawn into suction electrodes for extracellular stimulation and recording. The viability of each hemicord was assessed by recording the ventral root response to dorsal root stimulation (DR-VR).(The DR-VR improves over the first 3 hours after isolation, presumably because the effects of anesthesia, hypothermia, and spinal shock wear off slowly. Thus the cords were maintained in the recording chamber for 3 hours at room temperature before recordings began.) If a particular segment lacked a DR-VR because of damage to either root or the cord, another segment was tried; usually the experiment was carried out on thoracic segments one to three (Tl-T3). If several segments had a poor DR-VR, the preparation was discarded (about 1in 5 preparations was unusable for this reason). Preganglionic neurons were located by inserting a glass microelectrode filled with HRP (see below) into the spinal cord through the lateral surface in the vicinity of the lateral horn. The ventral root was continuously stimulated at an arbitrarily chosen intensity of 0.5V greater than that required for a maximum DR-VR (usually about 20 V and 2 Hz). The electrode was advanced through the tissue with a micromanipulator; cell penetration was facilitated by tapping the air table on which the recording chamber sat and passing small hyperpolarizing current through the electrode (0.2 nA, 5 msec, 2 Hz). Preganglionic neurons were identified by the presence of an antidromic spike elicited by ventral root stimulation. As no attempt was made to separate the sympathetic ramus from the ventral root to antidromically stimulate only preganglionic neurons, it is possible that some of the cells identified as preganglionic neurons were actually somatic motor neurons. Since cells were identified histologically by the presence of HRP, any MATERIALSAND METHODS such mistaken identification would be readily apparent. In Neonatal (5-16 days) Holtzman rats were deeply anesthe- fact, the lateral horn containing the preganglionic neurons tized with sodium pentobarbitol (50 mg/kg body weight), is sufficiently far from the somatic motor column when

C.J. FOREHAND

336 approached from the lateral surface (see, e.g., McKenna and Schramm, '83) that in only one instance (the first preparation tried) did a filled cell turn out to be a somatic motor neuron. When a preganglionic neuron was successfully impaled, the neuron was filled with HRP in a manner similar to that used to fill preganglionic axons and ciliary ganglion cells with HRP in vitro (Forehand and Purves, '84). The microelectrodes were pulled from triangle glass and filled with a solution of 5% HRP in 0.2 M potassium acetate buffered to pH 7.6; electrode resistances ranged from 125-190 mohm. The HRP was iontophoresed into the cells using 3 nA,50 msec depolarizing pulses at 5 Hz for 10 minutes. In some instances more than one cell was filled with HRP. The number of cells per experiment was limited to three, both because of the several hours required to set up the preparation and locate cells and because the dendritic arborization of these cells was so extensive that only one cell could be filled and analyzed per segment. One hour after the last cell was filled, the hemicord was fixed overnight in 1.25% glutaraldehyde and 0.5% paraformaldehyde in HEPES buffer at pH 7.6 in the refrigerator.

Histology and cell analysis Fixed cords were rinsed in buffer and embedded in a gelatin albumin matrix (Bowers and Zigmond, '79) and sectioned with a vibratome; 100-F.m transverse sections were serially collected in phosphate buffer. The sections were rinsed briefly in alcohol (5 min each: 25%, 50%, 25%) and the HRP visualized using the Hanker-Yates histochemical reaction (Hanker et al., '77). The sections were then mounted on slides, coverslipped with DPX, and examined by light microscopy. The HRP-labeled preganglionic neurons were drawn with the aid of a camera lucida. Each cell extended rostrocaudally through several 100-pm transverse sections; a two-dimensional representation of each cell was made by tracing all of the sections through each cell onto a single sheet of paper. Measurements of total dendritic lengths and cell body circumference and area were made from these drawings using a digitizing tablet attached to a computer and a graphic reconstruction language (Voyvodic, '86).

RESULTS HRP was iontophoresed into 49 antidromically identified preganglionic neurons in 37 preparations of the spinal cord. Thirty-two of the 49 cells were recovered after histochemical processing to visualize the HRP; 25 of the cells had somata in the intermediolateral cell column (IML), 2 in the lateral funiculus (LF), and 5 in the intercalated nucleus (IC) or a t the border of the IML and IC. Because the spinal cord was hemisected at the midline, no cells in the CA area were obtained. Fifteen of the 32 recovered cells were deemed unacceptable for further analysis. Some of these cells were discarded because they were poorly filled, others because two cells were filled too close to each other to permit camera lucida tracings of individual cells. Still other cells were rejected because their axons could not be traced to their exit from the spinal cord. (Although all cells that were filled fired an action potential in response to ventral root stimulation, no conditions were employed to be certain the activation was not synaptic. The added requirement that cells analyzed further have axons entering the ventral root as verified histologically insures that the neurons examined were not activated by ventral root afferents or

recurrent collaterals of preganglionic or somatic motor neurons.) Characteristics of the 17 cells that were well labeled are described below.

Elechophysiology Stable intracellular recordings from sympathetic preganglionic neurons in the neonatal rat are difficult to maintain; because the intent of this study was to obtain morphological data for these neurons, the time required for intracellular filling with HRP precluded extensive electrophysiological analysis of the cells. However, the resting membrane potential was noted for all cells, as was the response to dorsal root stimulation. All of the cells were obtained from hemicord preparations demonstrating a ventral root response to dorsal root stimulation (Fig. 1A). Although monosynaptic responses were present in the DR-VR (presumably due to activation of somatic motor neurons), monosynaptic excitation or inhibition of identified preganglionic neurons by dorsal root stimulation was never observed. The average resting membrane potential for successfully filled preganglionic neurons was 43 mV (range = 2060 mV). The duration of the antidromically driven action potential was relatively long (Fig. 1B). This observation could be due in part to the immaturity or poor health of the neurons, though others (Fernandez de Molina et al., '65; Mclachlan and Hirst, '80; Dembowsky et al., '85b) have observed long-duration action potentials in sympathetic preganglionic neurons of the cat recorded in vivo. Coote and Westbury ('79) found that some sympathetic preganglionic neurons in the in vivo cat spinal cord had long-duration action potentials, but that most had short-duration action potentials. They suggested that the long-duration action potentials resulted from poor penetrations of the cells. Action potentials of short duration ( < 5 msec) have also been observed in intracellular recordings obtained from sympathetic preganglionic neurons in vitro in thick transverse sections of the spinal cord of the rat (Ma and Dun, '86) and cat (Yoshimura et al., '86). Transverse sections may provide a more stable recording situation for these neurons. In addition to preganglionic neurons, several other types of neurons were impaled in the immediate vicinity of the intermediolateral cell column. One commonly observed type was monosynaptically excited by dorsal root stimulation but had no response to ventral root stimulation. Another type of neuron impaled with some regularity exhibited excitatory postsynaptic potentials in response to ventral root stimulation but had no response to dorsal root stimulation. Spontaneous activity ( < 5 Hz) was recorded in some of these neurons. These neurons may be interneurons activated by axon collaterals from preganglionic neurons (Cabot and Bogan, '87; also see below), or they could be activated by afferents entering the cord via the ventral roots (Light and Metz, '78). These cells are probably not Renshaw cells activated by recurrent collaterals from antidromically driven somatic motor neurons because they did not respond when these motor neurons were activated by dorsal root stimulation. A final type of neuron commonly encountered in these experiments displayed a long-lasting inhibitory postsynaptic potential in response to dorsal root stimulation. Any or all of these cell types may be involved in local circuit processing of autonomic responses.

Morphology of preganglionic neurons Cell bodies. Twelve of the 17 HRP-labeled preganglionic neurons examined in detail had cell bodies located

MORPHOLOGY OF MAMMALIAN PREGANGLIONIC NEURONS

A

TABLE 1. MorphologicalCharacteristics of P~ganglionicNeurons by Intracellular Iontophoresis ofHRP ~

Cell number

10 msec I!

*

----

I

B

337

I\

'"3 10 msec

Fig. 1. Electrophysiological recordings from the isolated rat spinal cord maintained in vitro at room temperature. A. Extracellular recording of the response of the third thoracic ventral root to stimulation of the third thoracic dorsal root. B. Intracellular recording of the action potential in a preganglionic neuron in the third thoracic segment following stimulation of the third ventral root. Resting membrane potential of this cell was - 40 mV. C. Photomicrograph of a section through the cell body of a preganglionic neuron intracellularly filled with HRP. Calibration = 10 wm.

within the IML. One cell body was located in the LF, two in the IC, and two in the IML-IC border. The cell bodies varied from spindle-shape to round. An equivalent diameter for the cells was calculated using the cross-sectional area measured on the camera lucida drawings. Equivalent diameters ranged from 18-28 p.m with an average of 22 pm (Table 1).

1 2 3 4 5 6

7 8 9 10 11 12 13 14 15 16 17

Cellbody Cellbody diameteP loeation" (pm) IML IML IML IML IML IML IML

IML IML IML IML IML IMLIC IMLIC IC IC

LF

22 22 24 22 24 19 20 26 23 24 28 18 22 22 21 20 25

Numberof primary dendrites 4 6 5 5 5 6 6 7 5 6 4 8 5 9 4 8 7

~

Total Rostro-caudal dendritic extentof length (km) dendrites' (km) 2,107 1,193 1.674 1;895 2,219 1,785 1,673 3,159 2,032 2,824 2.811 3,028 4,241 872 1,243 4,335 2.747

500 400 500 600 500 500 300 600 400 700 300 700 500 400 300 700 500

"LF, lateral funidus; IML, intemediolateral cell column; IC, interealated area; IMLIC, border between IML and IC. bCal&ted from the cm-sedional area measured from cameralucidadrawinge of the cells. %rived from the number of 100-pmtransverse sections containing pi- of the labeled ceh.

Dendrites. All of the HRP-labeled preganglionic neurons had extensive dendritic arborizations arising from an average of 6 (range = P 9 ) primary dendrites (Table 1).The number of primary dendrites may actually be higher than the numbers shown in Table 1 because it is difficult to assess accurately the number of dendrites that arise perpendicular to the plane of the section. The average total dendritic length for these neurons was 2,343 p.m (range = 8724,335 pm) (Table 1). In contrast to the impression gained from Golgi or retrograde HRP labeling studies, individual dendrites extended 100's of microns from the cell bodies and well out of the area defined by the intermediolateral cell column (Fig. 2). Moreover, preganglionic neurons exhibited quite different patterns of dendritic arborization that correlated with the position of their somata (Fig. 2). The most commonly observed pattern of arborization is shown in Figure 2A,B. The somata of these neurons were in the intermediolateral cell column; they had prominent dendrites extending laterally nearly to the pia and medially to the vicinity of the central canal. These medially and laterally oriented dendrites remained in the plane of the cell body and were thus nearly orthogonal to the rostrocaudal axis of the spinal cord. This type of neuron also had an extensivedendritic arborization oriented longitudinallythat remained within the confines of the IML (Fig. 3). Typically, these dendrites traversed 300-500 pm in the rostrocaudal dimension. Ten of the 12 preganglionic neurons recovered that had somata in the IML had this type of dendritic arborization; i.e., their dendrites were oriented in two dimensions with extensive mediolateral and rostrocaudal projections. The other two cells with somata in the IML had dendrites that radiated away from the somata in stellate fashion. In all cases, spinelike appendages protruded from the surface of the dendrites throughout the extent of the arborization (Fig. 2). None of the HRP-labeled preganglionic neurons with somata located outside the IML had the type of twodimensional dendritic arborization common to IML cells. Transversely oriented dendrites spanning the mediolateral extent of the cord in the plane of the cell body were lacking in these cells. Rather, their dendrites radiated away from the cell body extending into the IML, LF, and dorsal and ventral horns of the spinal cord traversing a rostrocaudal distance of about 400 pm (Fig. 2C,D). As with the IML cells,

C.J. FOREHAND

338

4

!

i

,

Fig. 2. Camera lucida drawings of HRP-filled preganglionic neurons. Six 100-Km transverse sections through the cells have been compressed onto a single plane. The dendrites have a spiny appearance, whereas the axons (dotted lines) are relatively smooth. A,B. Preganglionic neurons with cell bodies located within the IML. These cells have medio-laterally directed dendrites that remain in the plane of the cell body and span an area from near the pial surface to the central canal. The convoluted dendritic arbor in the vicinity of the cell body is a two-dimensional representation in the transverse plane of an arborization that remains within the confines ofthe IML and extends rostrocaudally for 300-500 pm (see Fig. 3). The axons were unbranched and exited the spinal cord in the first ventral root caudal to the location of

the cell bodies. C,D. Preganglionic neurons with cell bodies located outside the IML. These cells, located in the LF (C) and at the border of IML and IC (D), have dendrites that radiate away from the cell body in three dimensions. The axonal trajectories for these neurons were more complicated than for cells with somata in the IML, but each exited the spinal cord in the first ventral root caudal to the location of the soma. Insets show the location and orientation of the cells in camera lucida drawings of transverse hemisections of the spinal cord. In each inset, dorsal is up, lateral is left, and the dashed line indicates the gray-white border in the cord. Calibration = 100 pm for the cells and 600 pm for the insets.

339

MORPHOLOGY OF MAMMALIAN PREGANGLIONIC NEURONS

A

50 un

Fig. 3. Computer assisted reconstruction and rotation of an HRPfilled preganglionicneuron with its soma in the IML. The neuron was reconstructed in three dimensions from five 100-pm transverse sections through the cell using Neurolucida software (Glaser and Glaser,

'88). The axon is indicated by asterisks. A. Two-dimensional representation of the cell in the transverse plane. B. Two dimensional representation of the cell in the horizontal plane. The rostrocaudal extent of dendrites within IML is represented by the vertical axis in this drawing.

340

C.J. FOREHAND

spinelike appendages were apparent throughout the dendritic arborization (Fig. 2). Axons. All but four of the axons of these HRP-labeled preganglionic neurons arose as a distinct process from the cell body. Four axons arose from the proximal portion of a dendrite (within 20 Fm of the soma). Axons exited the spinal cord in the first ventral root caudal to the position of the cell body. Preganglionic somata located within IML had axons that coursed almost directly ventrally from the cell body and turned into the ventrolateral funiculus at the edge ofthe ventral gray (Fig. ZA,B).These axons never branched within the gray matter. Axons from somata outside the IML extended dorsally, medially or laterally before projecting in a caudoventral direction (Fig. 2C,D). Interestingly, two of these axons branched within the gray matter of the spinal cord (Fig. 4).

DISCUSSION The morphology of neonatal rat sympathetic preganglionic neurons in thoracic segments Tl-T3 has been examined by intracellular injection of HRP in an in vitro preparation of the spinal cord. The results indicate that whereas these neurons averaged six primary dendrites and a total dendritic length of 2,343 pm, the morphology of their dendritic arborizations differed according to the location of their cell bodies. Most cells whose somata lay within IML were two dimensional, with some dendrites oriented rostrocaudally within IML and others oriented mediolaterally. In contrast, preganglionic neurons with somata outside IML had dendrites that radiated away from the cell body in three dimensions, extending into the IML, LF, and the dorsal and ventral horns. The dendrites of all of the HRP-labeled cells had spinelike protruberances. The axons of all cells exited the spinal cord in the first ventral root caudal to the position of the cell body. Two of the HRPlabeled preganglionic neurons had intraspinal axon branches; both of these neurons had somata outside IML.

The fact that most preganglionic neurons in IML had extensive mediolateral dendritic projections is at odds with the generally accepted notion that these cells have only rostrocaudally projecting dendrites that remain within the confines of the IML (Rethelyi, '72).This notion has recently been challengedby retrograde tracing studies and immunohistochemical detection of choline acetyltransferase in the spinal cord. Thus several studies have demonstrated transverse bundles of dendrites when preganglionic neurons in guinea pig, cat, rat, or rabbit are retrogradely labeled (Rubin and Purves, '80; Deuschl and Illert, '81; Oldfield and McLachlan, '81; Rando et al., '81; Vera et al., '86; Bacon and Smith, '88). Immunohistochemical localization of choline acetyltransferase in the adult rat spinal cord also indicates transverse bundles of dendrites from preganglionic neurons in IML (Barber et al., '84; Markham and Vaughn, '89). In these retrograde tracing and immunohistochemical studies, it is difficult to be sure that particular dendrites arise from cells in IML rather than from more medially located preganglionic neurons. The intracellular staining of individual preganglionic neurons provides a more complete representation of these cells. Dembowski et al. ('85a) have reported that cat preganglionic neurons intracellularly labeled with HRP have only rostrocaudally oriented dendrites. However, their sample size was small (7 cells) and the medially projecting dendrites suggested by retrograde labeling in the cat do not occur at all levels of the IML (Oldfield and McLachlan, '81). Preganglionic neurons with extensive mediolateral (and rostrocaudal) dendritic projections were described in an intracellular HRP labeling study in the pigeon (Cabot and Bogan, '87).Each of the nine preganglionic neurons these authors labeled in the principal preganglionic cell column (Column of Terni) of the pigeon had a dendritic morphology similar to that commonly observed here for rat preganglionic neurons located in the IML. Cabot and Bogan ('87) also described two HRP-labeled preganglionic neurons located within the intercalated nucleus of the pigeon; these neurons had dendritic

Fig. 4. Photomicrograph of an HRP-labeled preganglionic axon that branched within the spinal gray. The branch point is indicated by the thick arrow and the fine branch that did not exit the spinal gray by the thin arrows. The branch that continued on to exit the spinal gray is indicated by arrow heads. Calibration = 10 urn.

MORPHOLOGY OF MAMMALIAN PREGANGLIONIC NEURONS morphologies similar to those rat preganglionic neurons described in the present study whose somata lay outside the IML. In contrast to both the cat (Dembowski et al., '85a) and pigeon (Cabot and Bogan, '87) preganglionic neurons, the axons of HRP-labeledrat preganglionic neurons commonly arose from the cell body. In only four cases did the axon arise from a common process with a dendrite. As in cat and pigeon, axons always exited the spinal cord in the first ventral root caudal to the cell body. Two (of 17) HRPlabeled preganglionic neurons, both with somata outside IML, had axons that exhibited an intraspinal axon branch. The possibility of such preganglionic intraspinal axon collaterals has been suggested by physiological studies (Lebedev, '72; deGroat, '76; Gebber and Barman, '79). Dembowski et al. ('85a) did not observe axon collaterals on preganglionic neurons examined in the cat. However, Cabot and Bogan ('87) reported that two of 11HRP-labeled pigeon preganglionic neurons had intraspinal axon collaterals. One of these neurons was in the principal preganglionic nucleus; the other was in the intercalated nucleus. It would appear that although intraspinal axon collaterals are present in a subpopulationof both rat and pigeon sympathetic preganglionic neurons, they are relatively rare. In the rat, axon collaterals may be restricted to preganglionic neurons whose somata do not lie in IML. The description of the morphology of rat sympathetic preganglionic neurons in the present study was obtained in neonatal animals (5-16 days). The possibility exists that preganglionic neurons in the adult rat could differ considerably. The dendrites of sympathetic ganglion cells in the rat continue to grow throughout life (Voyvodic, '871, and there is no reason not to expect preganglionic neuronal dendrites also to continue to grow. More importantly, there may be a differential growth of dendrites such that the,overall pattern of dendritic arborization may change. In fact, Schramm et al. ('76) utilizing retrograde labeling, reported an extensive mediolateral dendritic projection from caudal thoracic preganglionic neurons projecting to the adrenal gland in neonatal rats that was lost as the animals matured. It should be noted, however, that adult rat sympathoadrenal preganglionic neurons labeled with a more sensitive retrograde tracer than that used previously exhibit medially projecting dendrites (Bacon and Smith, '88). Whether the mediolateral dendritic extensions observed on neonatal preganglionic neurons in the upper thoracic cord are less extensive in the adult awaits elucidation by intracellular labeling studies in the adult. However, the immunohistochemical data localizing choline acetyltransferase in the adult rat suggest the continued presence of mediolaterally projecting dendrites at all spinal levels (Barber et al., '84; Markham and Vaughn, '89). In addition to the possible change in the pattern of dendritic arborization with maturation, it is also possible that the axons of these neurons may exhibit changes. The intraspinal axon branches reported here may be confined to the neonate; however, the presence of such collaterals in adult pigeon preganglionic neurons (Cabot and Bogan, '87) and their implied existence from physiological studies (Lebedev, '72; deGroat, '76; Gebber and Barman, '79) make it likely that such collaterals are present in adult rat preganglionic neurons. In summary, intracellular staining with HRP indicates that the dendrites of neonatal rat sympathetic ganglion cells extend well outside the IML where they may be

341

accessed by central pathways that do not terminate within IML. Moreover, the different dendritic arborizations of preganglionic neurons in different preganglionic nuclei may mean that the connectivity of preganglionic neurons varies according to cell body location. Finally, the presence of intraspinal axon collaterals on some preganglionic neurons provides an anatomical substrate for local circuit interactions.

ACKNOWLEDGMENTS I thank Mrs. Pat Newton for excellent technical assistance and Dr. Dale Purves for advice and support. This work was supported by NSF Grant BNS 85-18844 and by NIH Grant 18629 (to Dr. D. Purves).

Appel, N.M., and R.P. Elde (1988) The intermediolateral cell column of the thoracic spinal cord is comprised of target-specific subnuclei: Evidence from retrograde transport studies and immunohistochemistry. J. Neurosci. 8:1767-1775. Bacon, S.J., and A.D. Smith (1988) Preganglionic sympathetic neurones innervating the rat adrenal medulla: immunocytochemical evidence of synaptic input from nerve terminals containing substance P, GABA or 5-hydroxytryptamine. J. Autonom. Nerv. Syst. 24:97-122. Barber, R.P., P.E. Phelps, C.R. Houser, G.D. Crawford, P.M. Salvaterra, and J.E. Vaughn (1984) The morphology and distribution of neurons containing choline acetyltransferase in the adult rat spinal cord: An immunocytochemical study. J. Comp. Neurol. 229:329-346. Bowers, C.W., and R.E. Zigmond (1979) Localization of neurons in the rat superior cervical ganglion that project into different postganglionic trunks. J. Comp. Neurol. 185t381-392. Cabot, J.B., and N. Bogan (1987) Light microscopic observations on the morphology of sympathetic preganglionic neurons in the pigeon, Columbia liua. Neurosci. 20r467-486. Cannon, W.B. (1915) Bodily Changes in Pain, Hunger, Fear and Rage.An Account of Recent Researches into the Function of Emotional Excitement. New York Appleton. Coote, J.H., and D.R. Westbury (1979) Intracellular recordings from sympathetic preganglionic neurones. Neurosci. Lett. 15:171-175. deGroat, W.F. (1976) Mechanisms underlying recurrent inhibition in the sacral parasympathetic outflow to the urinary bladder. J. Physiol. 257;503-5 13. Dembowsky, K.J., J. Czachurski, and H. Seller (1985a) Morphology of sympathetic preganglionic neurons in the thoracic spinal cord of the cat: An intracellular horseradish peroxidase study. J. Comp. Neurol. 238:453465. Dembowsky, K., J. Czachurski, and H. Seller (1985b) An intracellular study of the synaptic input to sympathetic preganglionic neurones of the third thoracic segment of the cat. J. Autonom. Nerv. Syst. 13:201-244. Deuschl, G., and M. Illert (1981) Cytoarchitectonic organization of lumbar preganglionic sympathetic neurons in the cat. J. Autonom. Nerv. Syst. 3:193-213. Fernandez de Molina, A,, M. Kuno, and E.R. Per1 (1965) Antidromically evoked responses from sympathetic preganglionic neurones. J. Physiol. 180:321-335. Forehand, C.J. (1985) Morphology of sympathetic preganglionic neurons in the rat spinal cord revealed by intracellular staining with horseradish Neurosci. Abstr. 11:34. peroxidase. SOC. Forehand, C.J., and D. Purves (1984) Regional innervation of rabbit ciliary ganglion cells by the terminals of preganglionic neurons. J. Neurosci. 4:l-12. Gebber, G.L., and S.M. Barman (1979) Inhibitory interaction between preganglionic sympathetic neurons. In P. Meyer and H. Schmitt (eds): Nervous System and Hypertension. New York: John Wiley & Sons, pp. 137-145. 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. 14:550. Hancock, M.B., and C.A. Peveto (1979) A preganglionic autonomic nucleus in the dorsal gray commissure of the lumbar spinal cord of the rat. J. Comp. Neurol. 183335-72.

342 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 peroxidme. Histochem. J. 9r789-792. Korner, P.I. (1979) Central nervous control of autonomic cardiovascular function. In R.M. Berne, N. Sperelakk and S. Geiger (eds): Hdbk. Physiol., Sec. 2, v. 1, The Cardiovascular System. New York American PhysiologicalSociety,pp 691-739. Lebedev, V.P. (1972) Properties of axons of sympathetic preganglionic neurons of the lower thoracic spinal cord. Neurosci. Behav. Physiol. 5t377-384. Lichtman, J.W., D. Purves, and J. Yip (1979) On the purpose of selective innervation of guinea-pig superior cervical ganglion cells. J. Physiol. 292t69-84. Light, A.R., and C.B. Metz (1978) The morphology of spinal cord efferent and afferent neurons contributing to the ventral roots of the cat. J. Comp. Neurol. 179501-516. Ma, R.C., and N.J. Dun (1986) Excitation of lateral horn neurons of the neonatal rat spinal cord by 5-hydroxytryptamine. Dev. Brain Res. 24:m-ga. McKenna, K.E., and L.P. Schramm (1983) Sympathetic preganglionic neurons in the isolated spinal cord of the neonatal rat. Brain Res. 269r201-210. McLachlan, E.M., and G.D.S. Hirst (1980) Some properties of preganglionic neurons in upper thoracic spinal cord of the cat. J. Neurophys. 43:12511265. McLachlan, E.M., B.J. Oldfield, and T. Sittiracha (1985) Localization of hindlimb vasomotor neurones in the lumbar spinal cord of the guinea pig. Neurosci. Lett. 54.269-275. Markham, J.A., and J.E. Vaughn (1989) Ultrastructural analysis of autonomic preganglionic neurons with a monoclonal antibody to choline acetyltransferase. SOC.Neurosci. Abstr. 15.263. Meckler, R.L., and L.C. Weaver (1984) Comparison of the distributions of renal and splenic neurons in sympathetic ganglia. J. Autonom. New. Syst. 11t189-200. Oldfield, B.J., and E.M. McLachlan (1981) An analysis of the sympathetic preganglionic neurons projecting from the upper thoracic spinal roots of the cat. J. Comp. Neurol. I96r329-345. Otsuka, M., and S. Konishi (1974) Electrophysiology of mammalian spinal cord in uitro. Nature 252:733-734. Petras, J.M., and J.F. Cummings (1972) Autonomic neurons in the spinal cord of the Rhesus monkey: A correlation of the findings of cytoarchitec-

C.J. FOREHAND tonics and sympathectomy with fiber degeneration following dorsal rhizotomy. J. Comp. Neurol. 146:189-218. Petras, J.M., and J.F. Cummings (1978) Sympathetic and parasympathetic innervation of the urinary bladder and urethra. Brain Res. 153:363-369. Uthelyi, M. (1972) Cell and neuropil architecture of the intermediolateral (sympathetic) nucleus of the cat spinal cord. Brain Res. 164:413425. Rubin, E., and D. Purves (1980) Segmental organization of sympathetic preganglionic neurons in the mammalian spinal cord. J. Comp. Neurol. 192r163-174. Schramm, L.P. (1982) Ganglionic, spinal and medullary substrates for functional specificity in circulatory regulation. In O.A. Smith, R.A. Galosy and M. Weiss, (eds): Circulation, Neurobiology, and Behavior (Developments in Neuroscience, v. 15). New York Elsevier Science Publishing, pp. 23-33. Schramm, L.P., J.M. Stribling, and J.R. Adair (1976) Developmental reorientation of sympathetic preganglionic neurons in the rat. Brain Res. 106:166-171. Sjoqvist, F. (1963) The correlation between the occurrence and localization of acetylcholinesterase-rich cell bodies in the stellate ganglion and the outflow of cholinergic sweat secretory fibres to the fore paw of the cat. Acta Physiol. Scand. 57:339-351. Vera, P.L., H.H. Ellenberger, J.R. Haselton, C. Haselton, and N. Schneiderman (1986) The intermediolateral nucleus: an “open” or “closed” nucleus? Brain Res. 386:84-92. Voyvodic, J.T. (1986)A general purpose image processing language (IMAGR) facilitates visualizing neuronal structures in fixed tissue and i n uiuo. SOC. Neurosci. Abstr. 12r390. Voyvodic, J.T. (1987) Development and regulation of dendrites in the rat superior cervical ganglion: Influence of preganglionic innervation. J. Neurosci. 7r904-912. Weaver, L.C. (1982) Neural basis for differential cardiovascular control. In O.A. Smith, R.A. Galosy, and M. Weiss (eds): Circulation, Neurobiology, and Behavior (Developments in Neuroscience, v. 15).New York: Elsevier Science Publishing, pp. 77-84. Yoshimura, M., C. Polosa, and S. Nishi (1986) Afterhyperpolarization mechanisms in cat sympathetic preganglionic neuron in uitro. J. Neurophys. 55:1234-1246.