Organization of the serotonergic innervation of spinal neurons in rats--I. Neuropeptide coexistence in varicosities innervating some spinothalamic tract neurons but not in those innervating postsynaptic dorsal column neurons

Anatomical changes in descending serotonergic projections from the rostral ventromedial medulla to the spinal dorsal horn following repetitive neonatal painful procedures

Excessive noxious stimulation during the critical neonatal period impacts the nociceptive network lasting into adulthood. As descending serotonergic projections from the rostral ventromedial medulla (RVM) to the spinal dorsal horn develop postnatally, this study aims to investigate the long-term effect of repetitive neonatal procedural pain on the descending serotonergic RVM-spinal dorsal horn network. A well-established rat model of repetitive noxious procedures is used in which neonatal rats received four noxious needle pricks or tactile stimulation with a cotton swab per day in the left hind paw from day of birth to postnatal day 7. Control animals were left undisturbed. When animals reached adulthood, tissue was collected for quantitative immunohistochemical analysis of serotonin (5-hydroxytryptamine, 5-HT) in the RVM and spinal dorsal horn. Both repetitive noxious and tactile procedures in the neonate decreased the 5-HT staining intensity in the adult ipsilateral, but not contralateral spinal dorsal horn. Repetitive neonatal noxious procedures resulted in an increased area covered with 5-HT staining in the adult RVM ipsilateral to the side of injury, whereas repetitive neonatal tactile stimulation resulted in increased 5-HT staining intensity in both the ipsi- and contralateral RVM. The number of 5-HT cells in adult RVM is unaffected by neonatal conditions. This detailed anatomical study shows that not only neonatal noxious procedures, but also repetitive tactile procedures result in long-lasting anatomical changes of the descending serotonergic system within the RVM and spinal dorsal horn. Future studies should investigate whether these anatomical changes translate to functional differences in descending serotonergic modulation after neonatal adverse experiences.

Projections from the nucleus reticularis magnocellularis to the rat cervical cord using electrical stimulation and iontophoretic injection methods

The aim of this study is to clarify the fiber distribution of the nucleus reticularis magnocellularis (NRMC) and adjacent areas in the rat spinal cord. Biotinylated dextran amine was injected iontophoretically through a glass capillary into the areas, in which a single cell responded to noxious electrical stimulation of the sciatic nerve and to a pinch of the thigh skin with multiple spikes. Labeled fibers descended bilaterally through the ventral funiculi of the medulla oblongata and then through the ventral and lateral funiculi of the cervical cord with an ipsilateral predominance, and terminated in the spinal gray (laminae I-X). A single fiber sometimes ran through several laminae while bifurcating many short branches with axon varicosities and terminal buttons in one transverse section, that is, through laminae V, VII and X, through laminae V, IIl-IV and I-II, and through laminae VII to I-II. The present study showed that the wide distribution of a single fiber and a mass of fibers descending from the NRMC and adjacent areas might modulate not only somatic sensory and motor functions but also autonomic functions in the spinal cord.

Selective innervation of lamina I projection neurones that possess the neurokinin 1 receptor by serotonin-containing axons in the rat spinal cord

Axons containing serotonin descend from brainstem to spinal cord and are thought to contribute to stimulation-produced and opioid analgesia, partly by a direct inhibitory action of serotonin on projection neurones. The density of serotoninergic innervation is highest in lamina I, which contains many nociceptive projection neurones. Two sets of anatomical criteria have been used to classify lamina I projection neurones: somatodendritic morphology and presence or absence of the neurokinin 1 receptor. To test whether the strength of serotoninergic innervation of lamina I projection neurones was related to morphology or neurokinin 1 receptor expression, we used confocal microscopy to determine the density of serotoninergic contacts on 60 cells retrogradely labelled from the caudal ventrolateral medulla. The contact density on neurones with the neurokinin 1 receptor was variable, with some cells receiving heavy input and others having few contacts. However, on average they received significantly more contacts (5.64 per 1000 microm(2) plasma membrane +/- 0.47, S.E.M.) than neurones which lacked the receptor (2.49 +/- .36). Among the neurokinin 1 neurones, serotoninergic innervation density was not related to morphology. Since the majority of serotoninergic boutons in lamina I of rat spinal cord do not appear to form synapses, we carried out electron microscopy on three heavily innervated neurokinin 1 receptor-immunoreactive projection neurones. Symmetrical synapses were found at 89% of serotoninergic contacts. These results indicate that serotoninergic innervation of lamina I projection neurones in the rat spinal cord is related to expression of neurokinin 1 receptors, but not to morphology, and that (at least on heavily innervated neurones) most serotonin-containing boutons which are in contact form synapses.

Descending control of pain

Upon receipt in the dorsal horn (DH) of the spinal cord, nociceptive (pain-signalling) information from the viscera, skin and other organs is subject to extensive processing by a diversity of mechanisms, certain of which enhance, and certain of which inhibit, its transfer to higher centres. In this regard, a network of descending pathways projecting from cerebral structures to the DH plays a complex and crucial role. Specific centrifugal pathways either suppress (descending inhibition) or potentiate (descending facilitation) passage of nociceptive messages to the brain. Engagement of descending inhibition by the opioid analgesic, morphine, fulfils an important role in its pain-relieving properties, while induction of analgesia by the adrenergic agonist, clonidine, reflects actions at alpha(2)-adrenoceptors (alpha(2)-ARs) in the DH normally recruited by descending pathways. However, opioids and adrenergic agents exploit but a tiny fraction of the vast panoply of mechanisms now known to be involved in the induction and/or expression of descending controls. For example, no drug interfering with descending facilitation is currently available for clinical use. The present review focuses on: (1) the organisation of descending pathways and their pathophysiological significance; (2) the role of individual transmitters and specific receptor types in the modulation and expression of mechanisms of descending inhibition and facilitation and (3) the advantages and limitations of established and innovative analgesic strategies which act by manipulation of descending controls. Knowledge of descending pathways has increased exponentially in recent years, so this is an opportune moment to survey their operation and therapeutic relevance to the improved management of pain.

Contributions of the medullary raphe and ventromedial reticular region to pain modulation and other homeostatic functions

The raphe magnus is part of an interrelated region of medullary raphe and ventromedial reticular nuclei that project to all areas of the spinal gray. Activation of raphe and reticular neurons evokes modulatory effects in sensory, autonomic, and motor spinal processes. Two physiological types of nonserotonergic cells are observed in the medullary raphe and are thought to modulate spinal pain processing in opposing directions. Recent evidence suggests that these cells may modulate stimulus-evoked arousal or alerting rather than pain-evoked withdrawals. Nonserotonergic cells are also likely to modulate spinal autonomic and motor circuits involved in thermoregulation and sexual function. Medullary serotonergic cells have state-dependent discharge and are likely to contribute to the modulation of pain processing, thermoregulation, and sexual function in the spinal cord. The medullary raphe and ventromedial reticular region may set sensory, autonomic, and motor spinal circuits into configurations that are appropriate to the current behavioral state.

Inhibition of pudendal reflexes in spinal rats. Reassessing the role of serotonin

The effects of serotonin (5-HT) and thyrotropin-releasing hormone (TRH) on penile reflexes were investigated in intact and spinally transected male rats. Doses of intrathecal 5-HT (0.0, 1.13, 2.26, 11.3, 22.6, and 113.0 nmol), in a range previously shown to inhibit pudendal reflexes in anesthetized spinal preparations, prolonged the latency to the first penile erection in awake intact rats. However, these doses also provoked hyperreactivity and vocalization. Doses of intrathecal TRH (100 and 500 pmol) that effectively inhibited penile erection in intact animals were less effective in spinalized animals. Finally, a combination of subthreshold doses of TRH (100 pmol) and 5-HT (4.0 nmol) at a ratio known to affect other TRH/5-HT-mediated circuits significantly extended erection latency in animals with spinal transections. These data suggest that 5-HT and TRH are both involved in the inhibitory circuits regulating penile erection, either through corelease onto the same population of cells or through independent release onto different populations of neurons.

Peptidergic input to immunohistochemically-identified Renshaw cells

Peptidergic influences on Renshaw cells were assessed in rat using gephyrin-immunoreactivity, as a Renshaw cell specific marker, in combination with substance P, calcitonin gene-related peptide- and nicotinic acetylcholine receptor-immunolabelling. An average of 3.9 substance P-, and 8.1 calcitonin gene-related peptide-, and 16.3 nicotinic acetylcholine receptor-immunoreactive close contacts or puncta were observed per Renshaw cell. Most appositions were somatic. These results provide neuroanatomical support for the peptidergic modulation of Renshaw cells.

The induction of pain: an integrative review

The highly disagreeable sensation of pain results from an extraordinarily complex and interactive series of mechanisms integrated at all levels of the neuroaxis, from the periphery, via the dorsal horn to higher cerebral structures. Pain is usually elicited by the activation of specific nociceptors ('nociceptive pain'). However, it may also result from injury to sensory fibres, or from damage to the CNS itself ('neuropathic pain'). Although acute and subchronic, nociceptive pain fulfils a warning role, chronic and/or severe nociceptive and neuropathic pain is maladaptive. Recent years have seen a progressive unravelling of the neuroanatomical circuits and cellular mechanisms underlying the induction of pain. In addition to familiar inflammatory mediators, such as prostaglandins and bradykinin, potentially-important, pronociceptive roles have been proposed for a variety of 'exotic' species, including protons, ATP, cytokines, neurotrophins (growth factors) and nitric oxide. Further, both in the periphery and in the CNS, non-neuronal glial and immunecompetent cells have been shown to play a modulatory role in the response to inflammation and injury, and in processes modifying nociception. In the dorsal horn of the spinal cord, wherein the primary processing of nociceptive information occurs, N-methyl-D-aspartate receptors are activated by glutamate released from nocisponsive afferent fibres. Their activation plays a key role in the induction of neuronal sensitization, a process underlying prolonged painful states. In addition, upon peripheral nerve injury, a reduction of inhibitory interneurone tone in the dorsal horn exacerbates sensitized states and further enhance nociception. As concerns the transfer of nociceptive information to the brain, several pathways other than the classical spinothalamic tract are of importance: for example, the postsynaptic dorsal column pathway. In discussing the roles of supraspinal structures in pain sensation, differences between its 'discriminative-sensory' and 'affective-cognitive' dimensions should be emphasized. The purpose of the present article is to provide a global account of mechanisms involved in the induction of pain. Particular attention is focused on cellular aspects and on the consequences of peripheral nerve injury. In the first part of the review, neuronal pathways for the transmission of nociceptive information from peripheral nerve terminals to the dorsal horn, and therefrom to higher centres, are outlined. This neuronal framework is then exploited for a consideration of peripheral, spinal and supraspinal mechanisms involved in the induction of pain by stimulation of peripheral nociceptors, by peripheral nerve injury and by damage to the CNS itself. Finally, a hypothesis is forwarded that neurotrophins may play an important role in central, adaptive mechanisms modulating nociception. An improved understanding of the origins of pain should facilitate the development of novel strategies for its more effective treatment.

Apposition of enkephalin- and neurotensin-immunoreactive neurons by serotonin-immunoreactive varicosities in the rat spinal cord

The descending serotonergic system provides a powerful inhibitory input to the dorsal horn of the spinal cord. Little is known about the chemical identity of the spinal neurons that the serotonergic system innervates, although spinal enkephalinergic neurons are likely candidates. This study investigated the apposition of serotonin-immunoreactive varicosities onto enkephalin- and neurotensin-immunoreactive neurons in the rat lumbosacral spinal cord. Using a double immunofluorescence technique, serotonin-immunoreactive varicosities were observed to abut the soma or proximal dendrites of [Met]enkephalin- and neurotensin-immunoreactive neurons. Nearly 75% of all [Met]enkephalin- and neurotensin-immunoreactive neurons were apposed by serotonin-immunoreactive varicosities in the marginal zone and dorsal gray commissure. In substantia gelatinosa, approximately half of the [Met]enkephalin- and neurotensin-immunoreactive neurons were juxtaposed by serotonin-immunoreactive varicosities. [Met]enkephalin-immunoreactive neurons also were bordered by serotonin-immunoreactive varicosities in the nucleus proprius (65%) and sacral parasympathetic nucleus (75%). The results of this study suggest that the descending serotonergic system mediates nociception via probable contacts with intrinsic enkephalin and neurotensin spinal systems. The mode of action of spinal serotonin on enkephalin and neurotensin neurons may be through "volume" transmission vs synaptic or "wiring" transmission.

Neuroanatomy of the pain system and of the pathways that modulate pain

We review many of the recent findings concerning mechanisms and pathways for pain and its modulation, emphasizing sensitization and the modulation of nociceptors and of dorsal horn nociceptive neurons. We describe the organization of several ascending nociceptive pathways, including the spinothalamic, spinomesencephalic, spinoreticular, spinolimbic, spinocervical, and postsynaptic dorsal column pathways in some detail and discuss nociceptive processing in the thalamus and cerebral cortex. Structures involved in the descending analgesia systems, including the periaqueductal gray, locus ceruleus, and parabrachial area, nucleus raphe magnus, reticular formation, anterior pretectal nucleus, thalamus and cerebral cortex, and several components of the limbic system are described and the pathways and neurotransmitters utilized are mentioned. Finally, we speculate on possible fruitful lines of research that might lead to improvements in therapy for pain.

A confocal microscopic survey of serotoninergic axons in the lumbar spinal cord of the rat: co-localization with glutamate decarboxylase and neuropeptides

Patterns of co-localization of serotonin with glutamate decarboxylase (the synthetic enzyme for GABA) or each one of eight neuropeptides (calcitonin gene-related peptide, dynorphin, enkephalin, galanin, neuropeptide Y, neurotensin, substance P and somatostatin) were investigated with dual-colour confocal laser scanning microscopy in the lumbar spinal cords of three adult rats. Four regions of the gray matter were studied (laminae I-II, V, IX and X). The extent of co-localization was estimated by direct assessment of merged pairs of optical sections and by automated image analysis. Co-localization of serotonin and glutamate decarboxylase was found only in a few axons of laminae I-II but was not detected in other laminae. Peptides were not co-localized with serotonin in the superficial dorsal horn but considerable co-localization was found in motor nuclei and sparse co-localization was found in laminae V and X. Galanin and substance P frequently co-existed with serotonin in lamina IX but some co-localization with dynorphin, somatostatin, [Met]enkephalin and neuropeptide Y was also detected. Galanin, substance P and dynorphin were also co-localized with serotonin in a few axons of the deep dorsal horn and in the gray matter around the central canal. Neurotensin and calcitonin gene-related compound did not co-exist with serotonin in any of the laminae investigated. This evidence suggests that different populations of serotoninergic axons project to different regions of the spinal gray matter. Those containing glutamate decarboxylase terminate in the superficial dorsal horn and are likely to be involved in antinociception, whereas those containing peptides terminate principally in motor nuclei and are likely to modulate motor activity.

Peptidergic transmission: from morphological correlates to functional implications

Like non-peptidergic transmitters, neuropeptides and their receptors display a wide distribution in specific cell types of the nervous system. The peptides are synthesized, typically as part of a larger precursor molecule, on the rough endoplasmic reticulum in the cell body. In the trans-Golgi network, they are sorted to the regulated secretory pathway, packaged into so-called large dense-core vesicles, and concentrated. Large dense-core vesicles are preferentially located at sites distant from active zones of synapses. Exocytosis may occur not only at synaptic specializations in axonal terminals but frequently also at nonsynaptic release sites throughout the neuron. Large dense-core vesicles are distinguished from small, clear synaptic vesicles, which contain "classical' transmitters, by their morphological appearance and, partially, their biochemical composition, the mode of stimulation required for release, the type of calcium channels involved in the exocytotic process, and the time course of recovery after stimulation. The frequently observed "diffuse' release of neuropeptides and their occurrence also in areas distant to release sites is paralleled by the existence of pronounced peptide-peptide receptor mismatches found at the light microscopic and ultrastructural level. Coexistence of neuropeptides with other peptidergic and non-peptidergic substances within the same neuron or even within the same vesicle has been established for numerous neuronal systems. In addition to exerting excitatory and inhibitory transmitter-like effects and modulating the release of other neuroactive substances in the nervous system, several neuropeptides are involved in the regulation of neuronal development.

Contacts between serotoninergic fibres and dorsal horn spinocerebellar tract neurons in the cat and rat: a confocal microscopic study

Contacts between serotoninergic nerve fibres and dorsal horn dorsal spinocerebellar tract neurons were analysed in order to investigate the morphological basis of actions of serotonin upon dorsal spinocerebellar tract neurons. In a series of experiments dorsal spinocerebellar tract neurons were labelled with intracellularly injected rhodamine-dextran in the cat. The neurons were monosynaptically excited by group II muscle afferents and cutaneous afferents and were identified by antidromic activation following stimuli applied in the cerebellum. In the second series of experiments dorsal spinocerebellar tract neurons were labelled by retrograde transport of Fluorogold injected into the cerebellum in the rat. In both series, serotoninergic fibres were labelled by using a specific anti-serotonin antiserum and were revealed by immunofluorescence. Appositions between the serotoninergic fibres and the cells were inspected with a dual channel confocal microscope. The merged images obtained with the two channels of the microscope were viewed in single optical planes 2 microns apart and in rotated three-dimensional reconstructions. Serotoninergic nerve fibres were found in apposition to cell bodies of all feline dorsal spinocerebellar tract neurons (n = 7) and of 75% of rat dorsal spinocerebellar tract neurons (n = 90). The numbers of putative contacts on cell bodies varied between less than 100 and nearly 300 (mean 160) in the cat and between about five and 30 in the rat. Contacts with dendrites of feline neurons were seen on 96% of 72 dendrites within 300 microns from soma and on 91% of 23 dendrites at distances of 300-500 microns. The number of such contacts varied from less than five to 150 on a single dendrite within these ranges of distances. Their total number within 100 microns from the soma was comparable or exceeded the number of contacts on the soma.

Quantitative and qualitative aspects on the distribution of 5-HT and its coexistence with substance P and TRH in cat ventral medullary neurons

By use of the indirect immunofluorescence technique the distributions of 5-hydroxytryptamine (5-HT)-, substance P- and thyrotropin-releasing hormone (TRH)-immunoreactive (IR) neurons have been studied in the midline raphe nuclei and nucleus reticularis lateralis of the caudal brainstem (levels P18.3-P8.5; according to Berman (1968), in the cat, after treatment with colchicine. In addition, by use of the double-labelling technique, the coexistence between 5-HT-, substance P- and TRH-like immunoreactivity (LI) in these neurons was analysed. The results show that cell bodies in the midline raphe nuclei and nucleus reticularis lateralis contain 5-HT-, substance P- and TRH-LI. 5-HT-IR cells were more abundant than peptidergic neurons in all areas analysed. Quantitative estimations indicated that the total number of 5-HT-IR cells in the regions studied was about 17 x 10(3), while the corresponding numbers for substance P- and TRH-IR cells were 11 x 10(3) and 12 x 10(3), respectively. From double-labelled sections it was concluded that the vast majority of peptidergic cells also contained 5-HT-LI (87-100%). However, a subpopulation of 5-HT-IR neurons lacked peptide-LI (10-55%). The degree of coexistence varied with the brainstem level, in that neurons at more rostral locations showed a lower incidence of coexistence between 5-HT and peptide(s). The presence of all three compounds in one and the same cell body could also be demonstrated. In summary, 5-HT-, substance P- and TRH-IR cell bodies were encountered in medullary nuclei known to contain neurons with projection to the spinal cord. A high degree of coexistence between the compounds was demonstrated in these nuclei. The obtained results fit well with earlier studies on the patterns of distribution and peptide colocalization of 5-HT fibres in the spinal cord. The existence of biochemically distinct neuronal subpopulations within the 5-HT bulbospinal pathway is discussed.

Serotonin and pain: evidence that activation of 5-HT1A receptors does not elicit antinociception against noxious thermal, mechanical and chemical stimuli in mice

In this study, we examined whether activation of 5-HT1A receptors elicits antinociception in response to acute noxious chemical, thermal and mechanical stimuli in mice. In the writhing test, both agonists (e.g., 8-OH-DPAT, S 14671 and WY 50,324) and partial agonists (e.g., buspirone and gepirone) elicited a pronounced antinociception. However, antagonists (e.g., (-)-alprenolol and WAY 100,135) also induced antinociception and, at lower (inactive) doses, failed to modify the action of agonists. In addition, the separation between doses required for induction of antinociception as compared to those required for induction of ataxia (in the rotarod test) was variable and low for both agonists (median: 1.9) and partial agonists (median: 1.3), although it was somewhat greater for antagonists (> or = 3.3). In the hot-plate test, only certain agonists (e.g., 8-OH-DPAT) and partial agonists (e.g., gepirone) elicited antinociception and their actions were not attenuated by 5-HT1A antagonists which, themselves, were inactive in this paradigm. The 5-HT1C/2 antagonist, ritanserin, the 5-HT3 antagonist, ondansetron, the dopamine D2 receptor antagonist, raclopride, and the alpha 1-adrenoceptor antagonist, prazosin, were also ineffective in modifying the antinociception evoked by 5-HT1A agonists and partial agonists in the hot-plate test. In contrast, their actions were strongly attenuated by the alpha 2-adrenoceptor antagonist, idazoxan. In the tail-flick tests to noxious heat and noxious pressure, 5-HT1A receptor agonists, partial agonists and antagonists generally failed to induce antinociception. Moreover, modulation of stimulus intensity (from very weak to very intense) did not reveal any influence upon the latency to respond. In conclusion, in the writhing test, the data provide no evidence for a specific antinociceptive effect of the activation of 5-HT1A receptors. Further, in the hot-plate test, for those 5-HT1A agonists and partial agonists which induce antinociception, alpha 2-adrenoceptors rather than 5-HT1A receptors are implicated in their actions. Finally, in reflexive tests, irrespective of stimulus quality or intensity, 5-HT1A agonists and partial agonists do not mediate antinociception. These data suggest that the activation of 5-HT1A receptors does not, under these conditions of acute noxious stimulation, elicit antinociception.

Neurokinin A coexists with substance P and serotonin in ventral medullary spinally projecting neurons of the rat

The coexistence of neurokinin A (NKA) with substance P (SP) and serotonin (5-HT) in ventral medullary neurons of the parapyramidal region and nucleus raphe pallidus of the rat was studied using multiple immunofluorescence labeling. Nearly all of the NKA-immunoreactive (IR) cells in the parapyramidal region and raphe pallidus were SP-IR nd 5-HT-IR, whereas about 70% of the SP-IR neurons and about 60% of the 5-HT-IR neurons contained NKA-IR. There were no apparent differences in the patterns of coexistence between parapyramidal and raphe pallidus neurons. NKA-IR neurons, which colocalized SP-IR and 5-HT-IR, were studied for projections to the lumbar and thoracic spinal cord by use of retrograde transport of fluorescent tracer. Whereas about 50% of the retrogradely labeled neurons of the parapyramidal region and raphe pallidus contained NKA-IR, nearly all of the NKA-IR neurons projected to the thoracic and lumbar spinal cord. In addition, some NKA-IR neurons in the ventral medulla were retrogradely labeled with tracer from localized injections into the thoracic intermediolateral cell column. In summary, this study demonstrated that NKA-IR is colocalized with SP-IR in bulbospinal serotonergic neurons of the parapyramidal region and raphe pallidus, which are known to regulate sensory, motor, and autonomic activities of the spinal cord.

Organization of the serotonergic innervation of spinal neurons in rats--III. Differential serotonergic innervation of somatic and parasympathetic preganglionic motoneurons as determined by patterns of co-existing peptides

The spinal cord is innervated by brainstem serotonergic neurons, some of which contain substance P and/or thyrotropin-releasing hormone in addition to serotonin. These neurons project at least three types of axons to the spinal cord: those containing both substance P and thyrotropin-releasing hormone, those containing thyrotropin-releasing hormone but not substance P, and those containing neither substance P nor thyrotropin-releasing hormone. However, the organization of the different types of serotonergic processes is unclear. In the present studies, the types of serotonergic axons projecting to two kinds of spinal neurons were examined. Somatic and parasympathetic preganglionic motoneurons were labeled retrogradely from the pelvic or sciatic nerve, respectively. Sections containing these neurons were stained either for serotonin and substance P, or for serotonin and thyrotropin-releasing hormone. Of a total of 428 profiles examined that were retrogradely labeled from the sciatic nerve, 425 (99%) were apposed by serotonin-immunoreactive varicosities; similarly, of a total of 382 profiles examined that were retrogradely labeled from pelvic nerve, 353 (92%) were apposed by serotonin-immunoreactive varicosities. However, differences appeared to exist between the types of serotonergic varicosities innervating these two groups of neurons. Among the profiles labeled from the sciatic nerve, it was estimated that over 97% were apposed by serotonin-immunoreactive varicosities in which serotonin co-existed with substance P and thyrotropin-releasing hormone. In contrast, among the profiles labeled from pelvic nerve that were apposed by serotonin-immunoreactive varicosities, it was estimated that less than 1% were apposed by serotonin-immunoreactive varicosities containing both thyrotropin-releasing hormone and substance P. We estimate that most of the remainder (about 80%) were apposed by serotonin-immunoreactive varicosities containing thyrotropin-releasing hormone but not substance P. We conclude that both the cell bodies of neurons retrogradely labeled from the pelvic nerve and those labeled from the sciatic nerve were apposed by serotonin varicosities. However, these two systems of neurons appear to be innervated largely by two different populations of serotonergic cells. This suggests that the raphe-spinal serotonergic system may independently modulate the activities of somatic motoneurons and parasympathetic preganglionic motoneurons.