I. INTRODUCTION

Top Previous Next References

The act of swallowing is a fundamental motor activity in mammals, since it serves two vital functions. By propulsing food from the oral cavity into the stomach via the pharynx and the esophagus, swallowing subserves an alimentary function and thus constitutes the first, irreversible step in feeding behavior. The swallowing reflex also protects the upper respiratory tract by cleaning the naso- and oropharynx and closing the nasopharynx and the larynx, to prevent the pulmonary aspiration of food particles (34, 87, 90, 110, 226, 248).

Swallowing is known to be a complex but stereotyped motor sequence, with the implication that it involves a fixed behavioral pattern. It constitutes, however, one of the most elaborate motor functions, even in humans, since it requires coordinating an extraordinary bilateral sequence of activation and inhibition among more than 25 pairs of muscles in the mouth, pharynx, larynx, and esophagus (70, 87, 226). One of the most striking characteristics of swallowing is that the motor sequence can be readily initiated by stimulating a nerve, namely, the internal branch of the superior laryngeal nerve (SLN) (61, 87, 149, 225, 279). Swallowing therefore probably constitutes, as Doty (87) has said, one of "the most complex stereotyped pattern of behavior that can be consistently evoked by electrical stimulation of a peripheral nerve." Because the motor event elicited by stimulating the SLN includes all the components of the physiological motor pattern, swallowing provides a suitable model for studying the neurophysiological mechanisms underlying motor pattern generation.

Since Magendie (208), the act of swallowing has generally been described as having three phases or stages (34, 51, 70, 87, 89, 226). As regards the motor pattern occurring during feeding behavior, swallowing has been subdivided into an oral, preparatory phase, followed by a pharyngeal phase, and then by an esophageal phase, corresponding to the primary peristalsis of the esophagus (78, 110, 142, 164, 280). In fact, the oral, preparatory phase during which the alimentary bolus is formed is almost entirely voluntary and can be interrupted at any time, whereas the pharyngeal and esophageal phases are involuntary. Moreover, once it has been initiated, the pharyngeal phase of swallowing constitutes an irreversible motor event. The same irreversible motor sequence is observed both during the swallowing, which is not associated with food consumption, but serves to clear the oropharynx of small amounts of saliva, and during the reflex swallowing elicited by stimulation of the SLN (87). Under these swallowing conditions, the pharyngeal phase of deglutition involves not only pharyngeal and laryngeal muscles, but also muscles in the oral cavity such as the tongue and suprahyoid muscles. Therefore, the actual motor event of swallowing can be best described as being composed of an oropharyngeal stage and a subsequent esophageal stage. It is this overall motor pattern, corresponding to the basic or fundamental swallow, which has been mainly studied in neurophysiological investigations (154-156).

Classically, swallowing has been taken to be a sequential motor event and not a rhythmic motor behavior like mastication, respiration, or locomotion. Deglutition can occur repeatedly during feeding, but this is generally regarded as simply involving the repetition at a low frequency of a single sequential motor performance (87, 164, 194). However, under the appropriate stimulus, some components of the swallowing act can take on the properties of a rhythmic motor behavior (51, 87, 149, 161, 279). A typical rhythmic pattern of swallowing is elicited during long-lasting repetitive stimulation of the SLN. Thus swallowing is a suitable model not only for the neurophysiological analysis of sequential motor events, but also for that of rhythmic motor behaviors. However, it should be stressed that the rhythmic motor pattern is mainly associated with the oropharyngeal phase of swallowing, and the uppermost part of the cervical esophagus when the rhythmic frequency is low. No esophageal peristalsis occurs during the successive oropharyngeal sequences, and only the last oropharyngeal event in the series is followed by the complete esophageal stage (79, 111, 279, 350, 353).

In mammals, all the muscles involved in the oropharyngeal stage are striated and are therefore driven by several pools of motoneurons located in various cranial motor nuclei in the brain stem and the uppermost levels of the cervical spinal cord (51, 87, 156, 226). The esophageal muscle is entirely composed of striated fibers in species such as rat, guinea pig, rabbit, dog, sheep, and cow and is therefore controlled by cranial motoneurons. In some species however, e.g., cats, opossums, and primates, a variable portion of the lower esophagus is composed of smooth muscle fibers, controlled by the autonomic nervous system (60, 79, 110, 111, 280).

It has by now been clearly established, as originally postulated in the pioneering work by Meltzer (218-220), that the sequential and rhythmic patterns of swallowing are formed and organized by a central pattern generator (CPG). The CPG was previously described as a swallowing center that can be subdivided into three systems: an afferent system corresponding to the central and peripheral inputs to the center; an efferent system corresponding to the outputs from the center, consisting of the various motoneuron pools involved in swallowing; and an organizing system corresponding to the interneuronal network that programs the motor pattern (51, 87, 156, 280). In fact, the concept of a swallowing center implies the idea of an anatomical localization, whereas that of a CPG is based on a more functional principle focusing on the activity of the various pools of neurons, i.e., motoneurons and interneurons, involved in the motor activity. The swallowing CPG is located within the medulla oblongata (156, 161). Swallowing is indeed a primitive reflex, which means that it is present in all animals with organs serving differentiated functions and that it emerges early during development. The normal human fetus can swallow by the 12th gestational week, before the cortical and subcortical structures have developed (139). It has also been reported that swallowing is still possible in human anencephalic fetuses (266, 269). Experimental studies have shown, moreover, that the destruction of several nervous structures such as the forebrain, cerebellum, and pons does not alter swallowing (34). In fact, the basic sequential motor pattern seems to remain nearly normal as long as the nervous structures located between the C₁ level and the trigeminal motor nuclei are intact (86, 89, 150, 152, 231).

Swallowing, which can be triggered by stimulating a peripheral nerve and involves limited regions of the central nervous system, is an attractive model for the neurophysiological analysis of sequential and rhythmic motor behaviors. However, it has received less attention than other fundamental motor activities such as locomotion, mastication, or respiration (65). This is probably due to the complexity of the motor pattern along with the great number of muscles and cranial nerves involved, which renders neurophysiological studies difficult. In addition, the whole swallowing sequence is difficult to initiate in anesthetized animals (149, 152).

Since the extensive review by Doty (87), several reviews on swallowing and its various aspects, such as those relating to esophageal motility, have been published (51, 70, 72, 78, 79, 84, 90, 110, 111, 129, 132, 154-156, 161, 164, 226-228, 278, 280, 353). However, the brain stem mechanisms contributing to the sequential and rhythmic motor events involved in swallowing have not been analyzed so far in detail. This review therefore focuses on the brain stem CPG responsible for swallowing. It deals with the results obtained over the last 25 years, first using classical electrophysiological techniques and then pharmacological approaches, as well as neuroanatomical methods, such as tract-tracing techniques. These results have been mainly obtained on anesthetized or decerebrate animals, and in some cases on in vitro brain stem slices. The brain stem CPG will be analyzed here in terms of its location, the firing behavior of the neurons, the patterns of connectivity and the neurotransmitters involved. The regulation of the central program by central or peripheral inputs and some tentative mechanisms possibly underlying CPG activity will also be examined. Although a detailed analysis of the swallowing motor pattern is beyond the scope of this review, the various motor patterns correlated with swallowing, which are relevant to the interpretation of central nervous mechanisms, will be briefly described before the neurophysiological analyses are discussed.

	II. MOTOR ACTIVITY

Top Previous Next References

The motor activity of the two stages of swallowing shows different features. The oropharyngeal stage of the basic or fundamental swallowing is a complex, stereotyped sequence of excitatory and inhibitory events (87). It involves a set of muscles that always participate in this fundamental motor pattern, and therefore, these muscles have been termed obligate muscles (87, 90). In addition to the obligate muscles, other muscles, such as extrinsic tongue muscles, facial muscles, lip muscles, and levator mandible muscles, may or may not participate in swallowing, depending on either the species or the swallowing conditions, and therefore constitute facultative deglutition muscles (90).

The onset of the oropharyngeal stage of swallowing starts in most species with the contraction of the mylohoid muscle, which can be taken to be the first muscle to become active in swallowing (87, 88). At the same time or after a very short delay of 30-40 ms, a contraction also begins in the anterior digastric and internal pterygoid muscles, concurrently with that occurring in the geniohyoid, stylohyoid, styloglossus, posterior tongue, superior constrictor, palatoglossus, and palatopharyngeus. All these muscles constitute, along with the mylohyoid, the "leading complex" that invariably initiates the act of swallowing (Fig. 1A) (88). The sequence continues with a pattern of excitation and inhibition in pharyngeal and laryngeal muscles. The middle and inferior pharyngeal constrictors fire in an overlapping sequence, and the inhibitory phenomena are particularly striking in laryngeal muscles such as the posterior crycoarytenoid, which becomes silent for several hundred of milliseconds (87). The oropharyngeal phase ends when the wave of contraction reaches the upper esophageal sphincter. At rest, the sphincter is closed by the tonic contraction of the cricopharyngeal muscle, which plays the main role in the sphincteric function (8, 9, 13, 49, 129, 301). Inhibition of the tonic contraction, resulting in the relaxation and opening of the sphincter, starts at the onset of swallowing and lasts until the cricopharyngeal muscle becomes active and propels the bolus into the esophageal body (Fig. 1, A and B) (110, 111, 164, 226).

View larger version (15K): [in this window] [in a new window]   Fig. 1. The swallowing motor pattern. A: diagram of the electromyographic activity (EMG) of muscles active during the swallowing sequence. Note that during the oropharyngeal phase several muscles are active synchronously, i.e., muscles of the leading complex, and that the sequentially activated muscles overlap considerably. Note the difference in the timing of the propagation of the peristaltic wave between the oropharyngeal and the esophageal phases. A downward deflection indicates inhibition of the tonic activity. The broken line indicates a possible inhibition that could not be definitely confirmed under the experimental conditions used (see text for comments). Whatever the case may be, note that when the inhibition is present, it begins at nearly the same time at each level of the swallowing tract. [Adapted from Doty and Bosma (88).] B: pressure profiles of the contraction wave in pharynx, superior esophageal sphincter, esophagus, and lower esophageal sphincter. Note that the contraction is stronger and faster during the oropharyngeal than during the esophageal phase.

The duration of the whole oropharyngeal sequence is in the range of 0.6-1 s and remarkably constant in all the mammals studied, including humans (87, 110). The electromyographic activity in each individual muscle consists of a phasic discharge, in the form of a burst of spikes, which ranges in general between 200 and 600 ms, depending on the muscle (88, 140, 171). The wave of contraction travels down the oro- and hypopharynx at a speed of ~10-20 cm/s before reaching the upper esophageal sphincter. This wave of contraction develops a force with a peak amplitude in the range of 100 mmHg in the oropharynx, reaching up to 200 mmHg in the human hypopharynx (Fig. 1B) (110, 353). When initiated by stimulation of the SLN, swallowing does not start immediately but occurs after a variable delay. Depending on the species, this delay ranges from 100 ms in sheep to 40-60 ms in rats (Fig. 4) (149, 152, 174). The presence of this delay suggests that the central nervous system comes into play during this period and participates in organizing the swallowing motor pattern.

In addition to the firing activity, when background activity is present in the muscles, it is abruptly inhibited with the onset of firing in the leading complex, and this inhibition is maintained until the actual swallowing contraction begins (Fig. 1A) (87, 88, 140, 171). Inhibition can also be detected in muscles of the leading complex just before they contract during swallowing (88). A strong inhibition of the background activity is also observed after the cessation of the firing activity. Therefore, the oropharyngeal stage comprises an extraordinary sequence of inhibition and activation in pairs of muscles, which have to be synchronized during the motor sequence (70, 87). It is particularly striking that this complex motor sequence constitutes an "all or none" motor activity, just like simpler reflexes. In other words, once it has been initiated, the sequence invariably reaches the pharyngo-esophageal junction (87, 231).

In comparison with the extraordinary complexity of the oropharyngeal phase, the esophageal phase of swallowing is quite simple. It consists of a peristaltic wave of contraction, which propagates down the esophagus, enabling the alimentary bolus to be transported from the pharynx to the stomach (111, 128, 129, 142, 164). At rest, the esophagus is electromyographically silent, i.e., there is no tonic or rhythmic activity. The peristaltic contraction moves from the proximal to the distal part of the esophagus at a speed that may show a fairly high degree of variability, depending on the species and the nature of the muscles, i.e., on whether an esophagus is composed of striated muscle alone or of both striated and smooth muscle. For example, the whole esophageal phase can last less than 2 s in anesthetized sheep and exceed 10 s in conscious humans (79, 87, 110, 149, 152, 353). Among all the species combined, the mean velocity of the peristaltic wave can be evaluated at between 2 and 4 cm/s, a value conspicuously lower than that obtained in the case of the oropharyngeal sequence. The amplitude of the peristaltic wave is more variable than in the oropharynx. It may be lower, ranging between 35 and 70 mmHg, but may reach upper values of 180-200 mmHg (Fig. 1B) (110, 111, 275, 353). The duration of the contraction at the segmental level is longer than that observed in the oropharyngeal muscles. It generally exceeds 1 s and can reach values of up to 4 s, depending on the level of the esophagus and the nature of the muscle (129, 336). The esophageal phase ends when the contraction wave reaches the lower esophageal sphincter and propels the bolus down into the stomach. At rest, the lower sphincter is the site of a high pressure zone that prevents the reflux of food from the stomach. The factors contributing to the basal sphincter tone are multiple (78, 79, 110, 111, 164). In most species, the tonic contraction is mediated by a tonic depolarization of the smooth muscle fibers, without any action potentials (111, 280). During swallowing, the sphincteric tone is inhibited at the beginning of a swallow or after a short delay. That is to say that this tone is inhibited during the whole swallowing sequence until a phasic contraction of the sphincter occurs, as at the upper sphincteric level (Fig. 1, A and B) (79, 110, 164).

Because there is no basal tone in the esophagus at rest, esophageal peristalsis has been thought of as just a sequence of excitatory events corresponding to the propagation of the contraction wave. However, several lines of evidence suggest that the esophageal peristaltic contraction, like the oropharyngeal sequence, not only consists of a single sequential contraction but that this contraction is preceded by an inhibitory input, which has been called deglutitive inhibition. The lower sphincter relaxation is a component of the deglutitive inhibition. At the other levels of the esophagus, the inhibition is not usually detectable in the absence of basal tone, but becomes clearly visible during repetitive swallowing (79, 110, 353). By artifically setting up a high pressure zone in the human esophagus by inserting an intraluminal balloon and inflating it to a critical level, Sifrim et al. (303, 304) have also observed the occurrence of a deglutitive-induced inhibitory influence before the esophageal contraction. This inhibition begins simultaneously at various levels of the esophagus but lasts progressively longer as it reaches the more distal segments. In addition, in vivo recordings on the opossum smooth muscle have shown that even without any pressure changes in the esophageal body, a membrane hyperpolarization of variable amplitude and duration precedes the muscular contraction (273, 311). Therefore, it seems very likely that, as in the oropharynx, the sequential pattern occurring in the esophagus is a sequence of inhibitory and excitatory events.

Upon comparing the duration of the contractions, the velocity of the contraction waves and the pressures developed, it becomes quite obvious that the esophageal phase has quite different characteristics from the oropharyngeal phase. In addition, unlike the oropharyngeal phase, esophageal peristalsis shows some degree of variability. The contraction wave can stop before reaching the sphincter, even in awake animals (9, 86, 149, 152, 218, 302, 353). Esophageal peristalsis is therefore not an all or none sequence, which suggests that some differences may exist between the central mechanisms involved in the esophageal versus the oropharyngeal sequence.

Rhythmic swallowing behavior can occur during physiological activities such as drinking, which involves a train of closely spaced swallows. It has been reported that depending on the amount of liquid being drunk, several successive swallows occur, numbering up to 100 in horses (87). Under experimental conditions, the best way of inducing rhythmic swallowing is to apply long-lasting repetitive stimulation to the SLN (Fig. 3) (86, 149, 225, 279). Under repetitive stimulation of the SLN, at 20-30 Hz, swallowing rates of 1-2/s can be readily induced in species such as rats and sheep during short periods of time (4-10 s) (149, 152, 174). Even during very long-lasting stimulation, the rhythmic pattern can continue, although obviously with a lower frequency. Under urethane anesthesia for example, a dog can swallow repetitively for 1 h at a rate of 0.4 Hz (87). In fact, as long as there is a bolus to swallow, or at least saliva, it is possible to observe sustained swallowing at a rapid rate with only slight signs of fatigue.

When swallows occur at a fast rate, the motor pattern consists of a series of complete oropharyngeal phases, whereas the esophagus remains quiescent and the lower esophageal sphincter is relaxed. A normal peristaltic contraction wave does not occur until after the last swallow in the series (Fig. 3A) (12, 221, 350). That is to say that during repetitive swallowing, the so-called deglutitive inhibition continuously blocks the esophageal activity. When the rate of the oropharyngeal swallowing is slow, the esophageal peristalsis is initiated but then interrupted or modified depending on when the subsequent swallow occurs during the peristalsis and on the nature of the muscle (striated or smooth) (350). When striated muscle only is involved, a second swallow initiated during the primary peristalsis results in the immediate and complete inhibition of the contractile activity induced by the first swallow, and the first wave progresses no further (Fig. 3A) (129, 152, 277, 279). In the species with smooth muscle distal esophagus, the pattern may be different, probably due to intricate interactions between central and peripheral effects. When the first swallowing wave has reached the smooth muscle segment of the esophagus, the peristaltic wave is not interrupted at this segment by the next swallow, but can proceed distally, although it decreases gradually in amplitude until it disappears (350). In these species, a previous swallow or the presence of a swallowing wave within the esophagus can dramatically alter the nature of the subsequent esophageal wave by decreasing its amplitude, modifying its velocity, and sometimes rendering it nonperistaltic (350). Whatever the case may be, depending on the duration of the whole swallowing sequence and the frequency of the rhythmic activities, the rhythmic swallowing motor pattern mainly involves the oropharyngeal phase, which suggests that an oscillatory activity may take place at the central level.

Part of the swallowing sequence, namely, an esophageal peristalsis, can also be induced in the absence of the oropharyngeal phase of swallowing and is called secondary peristalsis (219). Secondary peristalsis occurs in response to stimulation of sensory receptors in the esophagus. For example, the transient esophageal distension induced by rapidly inflating an intraluminal balloon can induce peristaltic contractions in the esophagus. This may correspond to the distensions produced when the esophageal content is not completely cleared by the first swallow, or when reflux of the gastric contents occurs into the esophagus. Secondary peristalsis may be initiated at the level of either the striated or the smooth muscle. The wave of contraction usually begins at the level of the distension or just above it (79, 110, 111, 280, 353).

Once initiated, the secondary peristaltic wave progresses distally through the esophagus, like the primary peristalsis (101, 129, 280). The amplitude and velocity of the peristaltic contractions induced by esophageal distension are similar to those observed during primary peristalsis. Secondary peristalsis can also be either complete or interrupted at a variable distance from the level at which it has been initiated. It should be noted, however, that whether or not a transient distension suffices to induce secondary peristalsis in the smooth muscle esophagus, the presence of a bolus may be required for the peristaltic wave to be propagated down the whole esophagus in species with a striated muscle esophagus (147, 149, 152, 276, 277). Secondary peristalsis also involves a sequence of inhibitory and excitatory events, as shown by the relaxation of the lower esophageal sphincter, which begins when the wave of contraction is initiated and lasts until the contraction reaches the sphincteric level (111, 263).

Finally, a peristaltic contraction can still be obtained in the smooth muscle segment of the esophagus in the absence of any extrinsic innervation. This peristaltic contraction is called tertiary or autonomic peristalsis because of the peripheral nature of the underlying mechanisms (60, 78, 79, 111, 280).

Electrical or natural stimulation of either the oro-pharyngo-laryngeal area or the esophagus can evoke in the same musculature a variety of simple reflexes. These responses can also be elicited by stimulating central end of nerves which innervate these regions such as the trigeminal, lingual, glossopharyngeal, vagal, superior laryngeal, and recurrent laryngeal nerves (Fig. 4A) (87, 88, 90, 212, 279). The reflex response evoked by nerve stimulation is generally ipsilateral, or at least it is much stronger ipsilaterally. The motor responses evoked consist of short electromyographic activities, lasting for ~10 ms. The latency of the responses is also short, since the minimal latency is in the range of 6-10 ms, depending on the muscle and the afferent fibers involved (87, 90). The reflex pathway may be monosynaptic, as in the case of the jaw closing muscles, but most of these elementary reflexes are mediated via oligosynaptic circuits (90).

With regard to the swallowing muscles, the elementary reflexes evoked by stimulating the SLN can also involve many nonlaryngeal muscles such as the geniohyoid, palatopharyngeus, and thyrohyoid muscles, the pharyngeal constrictors, and muscles of the tongue and upper cervical esophagus (87, 90, 279). All these muscles are obligate swallowing muscles, which means that the SLN strongly influences the motoneurons that innervate these muscles. The elementary reflex responses interact with the more complex swallowing behavior. The latter exerts the most powerful control on the motoneuronal pool and to a large degree overrides the effects of the elementary reflexes (Fig. 4A) (88). These findings indicate that the circuitry involved in the elementary reflexes, or at least part of this circuitry, also participates in swallowing.

Swallowing muscles in the oropharynx and larynx are also involved in several other integrated behaviors, such as the activities involved in the preparatory oral phases of ingestive behavior, i.e., lapping, licking, sucking, and mastication (90, 296, 326, 368). Muscles innervated by the trigeminal motor nucleus, such as the mylohyoid, anterior digastric, and lateral pterygoid, are involved in jaw opening and chewing, during which they have a typical rhythmic activity (90, 215, 322, 326, 368). Muscles innervated by the facial motor nucleus, such as the stylohyoid and posterior digastric muscles, intervene in jaw movements (90). Tongue muscles controlled by the hypoglossal nucleus, such as the genioglossus and styloglossus, which are involved in protrusion and retrusion of the tongue, respectively, are also among those mainly involved in all oral activities such as chewing, licking, and sucking (83, 200, 296, 326, 342).

Several swallowing muscles in the mouth, pharynx, and larynx exhibit either an inspiratory or an expiratory activity. This respiratory activity in oro-pharyngo-laryngeal muscles ensures the patency of the upper airway and regulates the airflow during the respiratory cycle (19, 81, 143, 352). Adductor muscles of the larynx are active during the expiratory phase of the respiratory cycle and regulate the rate of airflow during expiration, whereas abductor muscles become active during inspiration, thus ensuring airway patency (19, 87). In addition to participating in swallowing and respiration, laryngeal muscles obviously participate strongly in the various types of phonation (105, 239, 241, 369). Intrinsic and extrinsic tongue muscles, such as the genioglossus, also contribute to the respiratory activity and may be active during either inspiration or expiration (200, 201, 372). The tongue tonus, in particular, is important in the maintenance of the appropriate airway aperture. Pharyngeal muscles can also be active during one or the other phase of the respiratory cycle (112, 299, 352). With regard to the relationships between swallowing and respiration, in early asphyxia, most or all of the muscles active in deglutition are recruited and participate in the respiratory effort. In these cases, the muscles are active only in respiration (88). Under these conditions, the respiratory drive therefore overrides the swallowing drive.

Emesis is a complex sequential and coordinated motor activity, which involves some of the muscles which participate in swallowing (187, 188). During emesis, an opening of both the lower and upper esophageal sphincters occurs and antiperistaltic contractions of the esophagus are generated, particularly in its cervical portion (187-189). Several oropharyngeal muscles also contract, especially the laryngeal muscles acting as glottis closers, pharyngeal dilator muscles, geniohyoid, genioglossus, and digastric muscles (113, 347). Most of these muscles are also active during retching, which may precede the expulsion phase of the emetic reflex. Rumination is a reflex motor sequence whereby animals such as cow and sheep regurgitate food from the stomach back to the mouth, where it is chewed again and then reswallowed (52, 91, 110, 276, 277, 308). During rumination, the esophageal sphincters relax and an antiperistaltic wave occurs concomitantly, in this case involving the whole esophagus. Swallowing muscles are also activated during activities such as coughing, belching, eructation, regurgitation, and several other rather particular types of behavior (110).

This brief survey shows the extraordinary diversity of the behavior in which the muscles participating in swallowing are also involved. These muscles can participate in reverse motor activities, such as rumination or vomiting compared with deglutition, or even in such antagonistic pairs of activities as respiration and swallowing, speaking and swallowing, or jaw opening and swallowing. They constitute an example subscribing to the general rule whereby motoneurons serve as a common final pathway.

	III. THE NEURONAL NETWORK GENERATING SWALLOWING

Top Previous Next References

Primary lesion experiments as well as electrophysiological and more recent pharmacological and neuroanatomical experiments have generally yielded fairly concordant results as regards the swallowing CPG (24, 51, 87, 156, 226, 280). It was on the basis of microelectrode recordings, however, that the swallowing-related neurons were identified and the structures where they are located were mainly established, and a general picture of the organization and functional principles of the swallowing CPG was built up (154, 156, 161). Most of the data obtained in this connection were obtained in experiments performed on anesthetized or decerebrate animals. In addition, under these experimental conditions, swallowing was frequently reflexively elicited by applying electrical stimulation to the SLN, in many cases in paralyzed animals, thus resulting in a fictive swallowing pattern. Therefore, these results relate more to the stereotyped basic swallowing movement than to the physiological motor activity involved in alimentary and drinking behavior, which shows greater variability and a higher degree of complexity (87, 90, 164).

As regards the swallowing CPG, microelectrode data have been obtained on several species, such as the cat, dog, monkey, rat, and sheep (59, 149, 174, 246, 255, 314, 345), but the most extensive studies have been carried out on anesthetized sheep (5, 46, 47, 61, 62, 149, 151, 152, 159, 160). In addition, most of these results have been obtained by performing extracellular recordings. Only recently have intracellular recordings been successfully obtained on swallowing neurons, focusing mainly on the activity of motoneurons (337, 376, 377), whereas few intracellular recordings have been carried out so far on swallowing interneurons (106). When recorded extracellularly, the vast majority of the swallowing-related neurons, which will be referred to hereafter as swallowing neurons, are phasic neurons that are usually silent but produce a burst of spikes, which has been called "swallowing activity" (Fig. 2) (149), the timing of which is correlated with the swallowing motor activity. Spontaneously active neurons have also been found to participate in swallowing. Depending on the neuron, they can exhibit either a temporary increase in their discharge frequency or a phasic inhibition of their spontaneous discharge during swallowing (Fig. 2A) (149, 152, 174, 345). Depending on the temporal relationship between their activity and the onset of swallowing, these swallowing neurons have been classified into three categories: "early" or oropharyngeal neurons, firing before or during the oropharyngeal phase of swallowing, and "late" and "very late" esophageal neurons, discharging during the esophageal peristalsis, that is to say during the cervical or thoracic esophageal contractions, respectively (Fig. 2) (149, 152). In addition to firing during the motor swallowing sequence, the swallowing neurons can also be activated by a local distension of the region of the alimentary canal which contracts in phase with the neuronal firing (Fig. 6A1). The swallowing neurons can furthermore be subdivided into motoneurons and interneurons.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Sequential firing of dorsal medullary neurons during swallowing. A: firing patterns of oropharyngeal (early) neurons. MHm gives the EMG of the mylohyoid, indicative of the onset of swallowing, and SwN is the activity of various neurons recorded at the nucleus tractus solitarii (NTS) level during swallowing triggered by stimulating the superior laryngeal nerve. Recordings have been aligned on the onset of the mylohyoid EMG (t0, broken line). Only the burst firing ("swallowing activity") or the cessation of neuronal activity is shown; the initial activity induced by each laryngeal stimulation is not represented (see Fig. 4). Note the large overlap between the sequential burst firing. The neuronal activity shown in the first trace is that of a neuron with a preswallowing discharge; in this case, continuous firing occurred from the superior laryngeal nerve (SLN) stimulation up to the swallowing burst. [Adapted from Jean (149, 152).] B: activity of esophageal (late and very late) neurons. Note the differences between the discharge duration and burst firing frequency in esophageal versus oropharyngeal neurons. [Adapted from Jean (149).]

Motoneurons active in swallowing have been identified using criteria such as whether they are antidromically activated in response to stimulation of motor nerves, their discharge frequency, and the location of the recording electrode site. However, no very extensive electrophysiological studies have been carried out so far on swallowing motoneurons. These motoneurons are localized within the trigeminal (V), facial (VII), and hypoglossal (XII) motor nuclei, the nucleus ambiguus (IX, X), the dorsal motor nucleus of the vagus (X), and at the cervical spinal level between C₁ and C₃ (51, 87, 90, 156, 226). However, depending on which muscles are innervated by these motor nuclei and on the size of the population of motoneurons actually involved in swallowing, these motor nuclei do not all participate to an equal extent in swallowing, at least during the basic pattern.

V and VII motor nuclei do not deal mainly with swallowing (87, 90). They are most strongly involved in several other orofacial activities such as jaw reflexes, mastication, licking, and sucking (90, 326, 368). Lesion experiments have shown for example that abolishing the V motor nuclei does not affect the swallowing sequence (87, 89, 152). Trigeminal motoneurons mainly involved in swallowing innervate the mylohyoid, anterior digastric, lateral pterygoid, and tensor veli palatini (87). Within the VII motor nucleus, motoneurons greatly involved in swallowing control the posterior digastric and stylohyoid (87). Other muscles innervated by these two motor nuclei, such as the medial pterygoid, temporal, and masseter are more facultative swallowing muscles (70, 90). In fact, the main motor nuclei involved in swallowing are the XII motor nucleus and the nucleus ambiguus. Most, if not all, of the motoneurons within these nuclei, which innervate the intrinsic and extrinsic muscles of the tongue, such as the genioglossus, geniohyoid, styloglossus and hyoglossus, and the pharynx, larynx, and esophagus, participate in swallowing (87, 226).

Data have shown the existence of a myotopic pattern of organization within the motor nuclei. This pattern of organization is based on a series of dorsoventral and mediolateral subdivisions in V, VII, and XII motor nuclei (341), i.e., the mylohyoid and anterior digastric motoneurons are situated in the ventromedial region of V motor nucleus, and tongue protrudor motoneurons are located in XII ventrolateral nucleus, whereas the tongue retractors are to be found more dorsolaterally (102, 181, 182, 196, 207, 234, 256, 344). The myotopic organization of the nucleus ambiguus corresponds to the well-known rostrocaudal pattern of organization of the motoneurons innervating the esophagus, pharynx, and larynx (28, 191-193), where the esophageal motoneurons are localized in the rostral compact formation of the nucleus, the pharyngeal and soft palate motoneurons in the intermediate semi-compact formation, and most of the laryngeal motoneurons in the caudal loose formation of the nucleus. This scheme of organization results in the sequential firing of nucleus ambiguus motoneurons during swallowing. However, the various pools of motor nuclei are not involved in any ordered sequence during swallowing, since the contraction of muscles in the leading complex involves motoneurons in V, VII, and XII motor nuclei and the nucleus ambiguus (88).

With regard to the innervation of the smooth muscle esophagus, the majority of the preganglionic neurons are located within the X motor nucleus. It has been reported that experimental lesion of this nucleus in the cat impairs the motility of the smooth muscle esophagus (131). Moreover, after injecting tracer into the cat esophagus and lower esophageal sphincter, most of the labeled cells have been observed in the X motor nucleus, although some neurons have also been identified in the nucleus ambiguus (66, 247). Unlike the striated motoneurons, which are located compactly in a single region, the preganglionic esophageal neurons consist of two separate groups: the one located in the rostral part of the X motor nucleus and the other in its caudal portion (66, 284).

Extracellular recordings have shown that within V and XII motor nuclei and the nucleus ambiguus (46, 47, 149, 151, 152, 174), the oropharyngeal motoneurons, which are generally silent at rest, exhibit during swallowing a short burst of low-frequency (40-50 Hz) spikes with a duration in the 50- to 200-ms range and considerable overlapping between the discharge of the various neurons. The bursting discharge can either precede the beginning of swallowing by a few milliseconds or lag between 0 and 200 ms behind the onset of the sequence. The firing behavior of esophageal motoneurons is somewhat different (152, 183, 377). However, the results published to date on these motoneurons are mainly restricted to sheep. The bursting discharge has a longer duration (up to 800 ms) and a very low frequency (10-20 Hz) and lags from 200-300 ms to 2 s behind the onset of swallowing. Therefore, the swallowing activity has a longer duration and a lower frequency in the esophageal than in the oropharyngeal motoneurons; the later the neuron becomes active during swallowing, the longer it will fire and the lower its discharge frequency will be (see Table 1).

In addition to the burst firing that occurs in motoneurons during swallowing, they can exhibit a short-latency (7-12 ms) synaptic activation, which is also called initial activity. This initial activity consists of one spike elicited in oropharyngeal or esophageal neurons by stimulating the afferent fibers present in the SLN or the vagus nerve, respectively (46, 47, 149, 151, 152, 174, 317). This initial activity has been observed only in response to ipsilateral stimulation applied to afferent fibers. Several pulses are generally required to initiate the spike, the latency of which is variable and suggests the existence of a polysynaptic pathway.

No central recordings have been performed so far on the activity of the preganglionic vagal neurons that innervate the smooth muscle esophagus. The data available on this subject were obtained in cross-innervation experiments and by directly recording the activity of vagal fibers (109, 281). They show that these neurons operate an ordered sequence during swallowing, which suggests that a central control may be at work even in the case of the smooth muscle esophagus. The burst firing of the preganglionic vagal neurons driving the smooth muscle esophagus has a long duration, in the range of 1 s, and a very low frequency, i.e., 3-8 Hz (109, 281). Recordings of the vagal efferent fibers innervating the opossum esophagus and the dog lower esophageal sphincter, both of which are composed of smooth muscle, indicate that both the vagal excitatory and inhibitory pathways are involved in swallowing (109, 233). These results therefore show that the CPG exerts both an excitatory and an inhibitory drive on the smooth musculature.

Recent intracellular recordings on hypoglossal and nucleus ambiguus motoneurons (337, 376, 377) have shown that, during swallowing, the firing of motoneurons was superimposed on a bell-shaped membrane depolarization. This simple bell-shaped depolarization of the membrane potential is not the only event that has been found to accompany the firing of these motoneurons. In some cases, complex depolarizing-hyperpolarizing or hyperpolarizing-depolarizing waves of membrane potential are evoked in motoneurons during swallowing (Fig. 6D). This indicates that in addition to an excitatory drive, these motoneurons may also receive inhibitory inputs or have complex intrinsic properties that are activated by the swallowing drive.

Extensive microelectrode recordings, first performed on sheep (5, 61, 62, 149, 151, 152, 159) and subsequently on the rat, dog, cat, and monkey (59, 94, 174, 190, 202, 246, 345), have shown that the swallowing neurons are located in two main brain stem areas: 1) in the dorsal medulla within the nucleus tractus solitarii (NTS) and in the adjacent reticular formation, where they form the dorsal swallowing group (DSG), and 2) in the ventrolateral medulla, just above the nucleus ambiguus, where they form the ventral swallowing group (VSG). Swallowing neurons are therefore located in the same medullary sites where are situated neurons that belong to CPGs involved in respiration and cardiovascular regulation (21, 73, 99, 163). In addition, other swallowing-related neurons have been identified within or very close to the motor nuclei and in the pons, at the level of the sensory nucleus of the trigeminal nerve (46, 47, 152, 160, 174, 254).

The results of microelectrode recording studies have not corroborated the lesion and anatomical data, suggesting that the swallowing CPG might be located rostrally in the medial reticular formation. Doty et al. (89) postulated that the interneuronal network that organizes swallowing might be located within the medial reticular formation, between the posterior pole of the facial nucleus and the rostral pole of the inferior olive. This assumption was based on lesion experiments performed on various species (e.g., on cats, dogs, and monkeys), which severely affected the motor pattern. The anatomical results obtained on the cat by Holstege et al. (137) have suggested that part of the "swallowing center" might be present in the pontine tegmental area just dorsomedial to the superior olivary complex. In fact, the latter data were obtained with tracing techniques, which can be used to demonstrate connections between neurons but not to identify the function of any such connections. Moreover, there is no direct evidence that any swallowing-related neurons exist within these regions: their possible role in swallowing still remains to be determined.

A) DORSAL SWALLOWING GROUP: THE INTERNEURONAL POPULATION IN THE NTS. Within the NTS, there exist neurons that fire during either the oropharyngeal or the esophageal phase of swallowing (149, 152). These neurons exhibit a typical sequential firing pattern that parallels the sequential motor pattern typical of deglutition (Fig. 2). When rhythmic oropharyngeal phases of swallowing are elicited, NTS neurons involved in the oropharyngeal sequence produce rhythmic bursting discharges that are closely linked to the motor pattern (Fig. 3). Because these neurons are still active during fictive swallowing elicited in paralyzed animals, their bursting discharge cannot be due to peripheral afferent inputs generated by the muscular contraction and actually correspond to a central swallowing activity (149, 152, 174). These results have amply confirmed that swallowing is a centrally patterned motor activity, as suggested in the pioneer study by Meltzer (218), and demonstrate that NTS medullary neurons form part of the neuronal network that generates swallowing. In previous studies on the cat, Sumi (314) failed to detect any bursting activity in the NTS during fictive swallowing. However, recent studies have provided evidence that in the cat also, there exist NTS neurons involved in swallowing that have a centrally patterned activity (106, 345).

View larger version (21K): [in this window] [in a new window]   Fig. 3. Rhythmic patterns of swallowing. A: rhythmic swallowing elicited in sheep by repetitively stimulating the SLN at 30 Hz. Note that the rhythmic pattern consisted of a series of oropharyngeal stages (MHm: mylohyoid EMG), and in this case, rhythmic contractions of the uppermost part of the esophagus (cEs1: upper cervical esophagus EMG). Only the last swallow in the series is complete with primary peristalsis after the oropharyngeal stage, as shown by the EMG of the lower cervical esophagus (cEs2) and the pressure wave recorded at thoracic level (thEs). [From Jean (152).] B: rhythmic burst firing of an oropharyngeal NTS neuron with preswallowing discharge (SwN), during repetitive fictive swallowing elicited in paralyzed sheep by applying long-lasting stimulation to the SLN. XII shows the neuronal activity recorded in the hypoglossal nerve. (From Jean, unpublished data.) C: intracellular recording of a swallowing neuron within the cat NTS, during rhythmic swallowing elicited by applying repetitive stimulation to the SLN. Note that the burst firing was superimposed on a high-amplitude bell-shaped depolarization of the neuron. [Adapted from Gestreau et al. (106).]

Most of the oropharyngeal neurons are active either just a few milliseconds before or during the oropharyngeal phase of swallowing. The firing pattern of these neurons is characterized by a short burst of spikes, with a duration in the range of 100-300 ms, a steady or an increasing-decreasing frequency with a mean value in the range of 100 Hz, and an instantaneous frequency that can be as high as 400 Hz (Figs. 2A and 4B) (149, 152, 174). Recent intracellular studies on cats have shown that the bursting activity of oropharyngeal NTS neurons is superimposed on a high-amplitude depolarizing wave, between 15 and 20 mV, which indicates that a strong central drive is exerted during swallowing (Fig. 3C) (106). As in the case of the muscular pattern, considerable overlaps occur between the sequential burst firing of the various neurons. The esophageal neurons obviously fire later on during swallowing, with a phase lag of 200-300 ms to 2 s after the onset of the motor activity, at least in sheep (149, 151, 152). As we saw above with the motoneurons, the firing behavior of the esophageal neurons differs from that of the oropharyngeal neurons. The duration of the bursts of spikes is longer (between 200 ms and 1 s), and the firing frequency is much lower, since it does not exceed 40 Hz in the case of the late neurons and is as low as 10-20 Hz in that of the very late neurons (Fig. 2B). At the level of the interneurons, one can also observe an increase in the duration of the discharge and a decrease in the firing frequency when the neuron is active later during swallowing (Table 1). No central recordings have been performed so far on esophageal neurons in species with a smooth muscle esophagus.

When stimulation is applied to afferent fibers, both types (oropharyngeal and esophageal) of interneurons produce a synaptic response (initial activity) in the form of a single spike. This initial activity can be elicited by stimulating the ipsilateral SLN in all the oropharyngeal NTS neurons and in the late esophageal neurons, the firing behavior of which is linked to the activity of the upper cervical esophagus (Figs. 4B and 6A). In the very late esophageal neurons, an initial activity can be elicited by stimulating the cervical vagus (149, 152). The synaptic response may occur with a very short, stable latency of 1-2 ms, indicating that, at least some, of these neurons are monosynaptically connected to afferent fibers (149, 152, 255, 294). In these types of neurons, the synaptic response and the burst firing activity are clearly separated by an interval of variable length, depending on the order at which the neuron is recruited in the sequence (Figs. 4B and 6A).

View larger version (25K): [in this window] [in a new window]   Fig. 4. Characteristics of motor and neuronal swallowing patterns. A: elementary reflexes and swallowing motor event. EMG activity recorded in rat suprahyoid muscles (SHm) or sheep cervical esophagus (cEs) during swallowing evoked by repetitive (1) or single-pulse (2) stimulation of the SLN. Note that in addition to the swallowing bursting activity, the stimulating pulses elicited short-duration EMG responses with a brief latency, i.e., the so-called elementary reflexes. Note also in A1 the inhibition of the elementary reflexes during swallowing and several milliseconds later. [1) Adapted from Kessler and Jean, unpublished data; 2) adapted from Roman and Car (279).] B: activity of two oropharyngeal neurons (SwN) recorded within the sheep NTS during swallowing (indicated by the EMG of the mylohyoid: MHm) evoked by applying repetitive (1) or single-pulse (2) stimulation to the ipsilateral SLN. As in the muscles, the laryngeal stimulation elicited an initial synaptic response in the form of one short latency spike in addition to the swallowing burst firing. Note that when no swallowing occurs (first stimulation in 2), only the initial response is evoked. [Adapted from Jean (149, 152).] C: patterns of an oropharyngeal neuron with preswallowing activity. When the stimulation is ineffective in inducing swallowing, the ipsilateral SLN stimulation triggers only preswallowing activity with a variable duration; the preswallowing discharge pattern becomes a swallowing burst firing pattern when swallowing occurs (MHm: mylohyoid EMG; middle traces). Note that when swallowing is triggered by applying stimulation to the contralateral SLN (bottom traces), the preswallowing discharge is lacking, and only the swallowing burst is initiated. [Adapted from Jean (152).] D: neuronal firing (SwN) elicited in an oropharyngeal neuron with preswallowing discharge, recorded from the sheep NTS, in response to stimulation applied to either the ipsilateral SLN or the glossopharyngeal nerve (IX). Note that the IX stimulation, which does not trigger swallowing in sheep, evokes only preswallowing activity, which lasts longer when the stimulation is repeated. [Adapted from Ciampini and Jean (61).]

Among NTS oropharyngeal neurons, some exhibit a particular pattern of bursting activity starting before the onset of the swallowing motor sequence. Unlike those neurons, where the swallowing burst occurs with a phase lag after stimulation and only when a swallow is initiated, these neurons exhibit a continuous discharge in response to the stimulus. This discharge, called "preswallowing activity," starts long before the beginning of the muscular contraction. It decreases and stops quite rapidly when no swallowing occurs, but continues and increases, turning into a swallowing activity, when swallowing is initiated (Figs. 2 and 4, C and D). This pattern of discharge suggests that these neurons are involved in the initiation of swallowing, and it has been postulated that they may constitute the trigger neurons in deglutition (149, 152).

As regards the location of NTS swallowing neurons, most of the relevant data were obtained before the cytoarchitectonic subdivisions of the nucleus were established on the basis of anatomical studies (1, 17, 157, 167). Therefore, their location has not been specified in relation to these subdivisions. However, in all the species studied, it has emerged that the oropharyngeal neurons are situated rostrocaudally at the level of the intermediate-subpostremal portion of the NTS (17), within the medial part of the lateral NTS which overlaps the interstitial, intermediate, ventral, and, to some extent, the ventrolateral subdivisions of the nucleus (149, 152, 161, 174). The esophageal neurons, which to date have been clearly identified only in sheep, are also situated at the level of the intermediate-subpostremal part of the nucleus, between the tractus solitarius and the X motor nucleus (149, 152), a region which may correspond to the centralis subdivision of the NTS in the rat (1, 202). Interestingly, anatomical results have shown that both laryngeal and pharyngeal afferent fibers project mainly to the interstitial subdivision of the NTS in all the species studied and that the esophageal afferent fibers end within the NTS subnucleus centralis, at least in the rat (1, 238). Therefore, within the NTS, there are these two subnuclei that are mainly concerned with swallowing (see Table 1).

B) VSG. In the ventrolateral medulla above the nucleus ambiguus, there also exists a large population of oropharyngeal swallowing neurons (59, 94, 149, 151, 152, 174, 252, 345). These neurons have been identified as interneurons on the basis of criteria, such as the lack of antidromic activation by vagus nerve stimulation and a high discharge frequency. In addition, several neurons in this region fire during the leading phase of swallowing, before most of the motoneurons in the nucleus ambiguus become active. The burst firing behavior of the VSG neurons is very similar to that of the DSG neurons in terms of the sequential firing pattern, in which there is considerable overlap, discharge duration, and frequency. This population has, however, a lower instantaneous discharge frequency, which never reaches the higher instantaneous values (400 Hz) observed in the DSG neurons (149, 152, 174). The main difference lies in the synaptic response to stimulation of the SLN. The initial activity is not readily triggered in these VSG neurons with a single pulse, and several pulses are generally required to initiate the response. Moreover, the fact that the latency is visibly longer (between 7 and 12 ms) and shows some degree of variability suggests the existence of a polysynaptic connection between the afferent fibers and the neurons (149, 174, 255; see Table 1).

The existence of a large population of esophageal interneurons in the ventrolateral medulla is less clear. Bursting discharges in phase with esophageal peristalsis have been recorded in the medullary region above the nucleus ambiguus (152). However, in the case of esophageal neurons, applying stimulation to the vagus nerve systematically induces an antidromic field potential. Without any intracellular evidence, it is therefore not possible in the case of esophageal neurons to clearly distinguish between motoneurons and actual interneurons (151, 152).

Whatever the case may be, swallowing neurons within the ventrolateral medulla are still active during fictive swallowing. Their bursting activity is therefore actually a central discharge, and these neurons, like those in the NTS, belong to the neuronal network that generates swallowing.

C) INTERNEURONS IN THE MOTONEURONAL POOLS. Apart from the VSG in the ventrolateral medulla, which probably plays an important role within the swallowing CPG (see sect. IIIB), swallowing interneurons have been identified within V and XII motor nuclei or in their close vicinity. These neurons produce a high-frequency discharge, and when tested by stimulating the motor nerves, they failed to respond antidromically (46, 47, 149, 174). These neurons may play the role of premotor neurons or be involved in the organization of the swallowing drive to the various motoneurons involved in swallowing within a single motor nucleus (254). Electrophysiological and neuroanatomical data also suggest that they might be involved in the bilateral coordination of the motoneuronal pools (5, 46, 47, 158, 235). The exact function of these neurons has not yet been elucidated, however, and further experiments will be required before this is possible.

D) PONTINE INTERNEURONS. In sheep, a population of neurons exhibiting burst firing in phase with the oropharyngeal stage of swallowing has been found to exist more rostrally in the pons, in a region extending rostrally to the exit of the facial nerve, above the V motor nucleus, at the level of the principal sensory trigeminal nucleus (160). These neurons, which are spontaneously active, present a swallowing activity in the form of a burst of spikes lasting some 40-380 ms, with a similar discharge frequency to that of the medullary interneurons, ranging between 60 and 340 Hz. In addition, they can be synaptically activated by applying stimulation to the ipsilateral SLN with a latency of 1.5-4 ms resembling that of the NTS neurons. However, there are two striking differences between these and the NTS neurons. In the pons, the swallowing burst always occurs after the beginning of swallowing and is abolished after motor paralysis of the animal. In addition, unlike NTS neurons, the pontine neurons are antidromically activated when stimulation is applied to the ventro-postero-medial nucleus of the thalamus (160). On the basis of these results, these pontine neurons have been classified as sensory relay neurons and are thought to be involved in providing information from the oropharyngeal receptors to the higher nervous centers. They therefore do not belong to the swallowing CPG.

The detailed connections between the various groups of neurons within the swallowing CPG still remain to be mapped. However, the results of electrophysiological and anatomical experiments have provided some information about the organization of the DSG, VSG, and motoneurons (2, 152, 154, 156).

The latency of the synaptic response (initial activity) is shorter among the neurons of the DSG than those of the VSG. In the case of SLN stimulation, the latency is 1-4 ms in the DSG and 7-12 ms in the VSG (149, 151, 152, 159, 174, 255). A synaptic response can also be initiated in swallowing neurons by stimulating a specific cortical area, which induces swallowing (see sect. IV). Here again, the latency of the response is shorter in the NTS swallowing neurons (5-8 ms) than in the neurons in the ventrolateral medulla (10-16 ms) (152, 159). These results suggest that the neurons of the VSG are probably activated via neurons of the DSG. Indeed, regardless of which afferent pathway is stimulated, the initial response of the VSG neurons is abolished after lesion of the DSG (152, 159). Although no direct evidence is available at the single cell level that a connection of this kind exists between the DSG and VSG, connections between the NTS region and the ventrolateral reticular formation surrounding the nucleus ambiguus, where swallowing neurons are located, have been found to exist in several anatomical experiments (57, 69, 236, 249, 257, 258, 283, 288). Other anatomical results have shown that the ventrolateral medulla also projects to the NTS region (20, 126, 198, 282, 349). However, whether or not these possible reciprocal connections are involved in swallowing still remains to be established.

Electrophysiological experiments on sheep have shown that only the swallowing neurons within the VSG could be antidromically activated by stimulating the swallowing region of V motor nucleus, i.e., that region where swallowing motoneurons are located, while none of the neurons within the DSG exhibited any antidromic activation under the same conditions (5). These results indicate that the V motor nucleus is connected to only one of the medullary regions involved in swallowing, namely, the VSG. Similar results have also been obtained on the XII motor nucleus, the stimulation of which evoked an antidromic potential in swallowing neurons of the VSG (4). In addition, more recent electrophysiological studies on sheep and cats have established that within the VSG, the same identified swallowing neuron can project to more than one motor nucleus. It has been shown, for example, that the same neuron can project either to V and XII motor nuclei, or to the XII motor nucleus and the nucleus ambiguus (6, 94). Some data also suggest that the same neuron might project to V and XII motor nuclei and the nucleus ambiguus (94), which are all motor nuclei involved in swallowing. These electrophysiological data fit in well with those coming from anatomical studies. It has been shown that retrograde transport of peroxidase, injected under electrophysiological control into the region of the V motor nucleus where swallowing motoneurons are situated, resulted in labeling of neurons in the ventrolateral medulla. Control experiments with tritiated leucine, injected in the VSG region, show labeling in the ventromedial region of the V motor nucleus. In addition, these experiments showed that the ventrolateral medulla is also connected to the homologous contralateral medullary region and to VII, X, and XII motor nuclei, all of which are also involved in swallowing (158). The ventrolateral medulla, which contains swallowing neurons, is therefore connected to all the various groups of motoneurons involved in swallowing: this conclusion is in complete agreement with other anatomical data (138, 195, 343).

These results suggest that within the swallowing network, VSG neurons are activated via DSG neurons and that motoneurons are driven by neurons of the VSG (Fig. 5 and Table 2). If we therefore consider the swallowing CPG, there exist simple circuits linking together the afferent fibers, the DSG neurons, the VSG neurons, and the motoneurons. At each level obviously, i.e., within the NTS, the ventrolateral medulla and the motoneuronal pools, circuits more complex than a simple monosynaptic connection may exist. However, the trisynaptic circuits are probably basic elements in the functioning of the CPG. Such reflex loops are also probably basic in the elementary reflexes.

View larger version (15K): [in this window] [in a new window]   Fig. 5. Diagram of the swallowing central pattern generator (CPG). The CPG includes two main groups of neurons located within the medulla oblongata: a dorsal swallowing group (DSG) located within the nucleus tractus solitarii (NTS) and the adjacent reticular formation and a ventral swallowing group (VSG) located in the ventrolateral medulla (VLM) adjacent to the nucleus ambiguus (nA). The DSG contains the generator neurons involved in triggering, shaping, and timing the sequential or rhythmic swallowing pattern. The VSG contains the switching neurons, which distribute the swallowing drive to the various pools of motoneurons involved in swallowing. It should be noted that the pathway including the peripheral afferent fibers, neurons in the DSG and VSG, and motoneurons forms an oligosynaptic loop involved in swallowing and the elementary reflexes (see text for additional information). [Adapted from Jean (153).]

Recent results obtained with retrograde tract tracing techniques or by performing transneuronal labeling with pseudorabies virus indicate that a direct connection may exist between swallowing regions in the NTS and motoneurons in the nucleus ambiguus, at least in the rat (16, 18, 39, 125, 126). These studies show that NTS neurons, which are supposed to be esophageal neurons, since they are located in the subnucleus centralis where esophageal afferent fibers project, send axon terminals in the rostral compact formation of the nucleus ambiguus where esophageal motoneurons are situated. A direct connection between NTS esophageal neurons and esophageal motoneurons may exist, since the existence of an interneuronal pool of esophageal neurons has not been clearly demonstrated in the ventrolateral medulla. However, it seems unlikely that a direct connection of this kind exists in the case of oropharyngeal neurons (see sect. IIIA2). All the results available to date on functionally identified swallowing neurons have shown that motoneurons are driven by oropharyngeal neurons within the VSG. Therefore, it is puzzling that the oropharyngeal population of interneurons within the ventrolateral medulla, i.e., the VSG, was not determined in the anatomical studies. Recordings will have to be made on functionally identified neurons to be able to ascertain whether there exists a monosynaptic link between neurons in the DSG and motoneurons.

It has been established in several networks involved in basic motor behavior, such as locomotion, that within a given CPG all the neurons are not equal since some of them play a preeminent role (10, 117). As regards swallowing, data already obtained suggest that neurons in the DSG are likely candidates to act as generator neurons in the initiation and organization of the sequential or rhythmic motor pattern (149, 152). The swallowing network in mammals therefore provides a unique example of neurons located within a primary sensory relay, i.e., the NTS, which nevertheless play the role of generator neurons. Several lines of evidence support the idea that NTS neurons play a leading role in swallowing. NTS neurons exhibit a sequential or rhythmic firing pattern that parallels the motor pattern (149, 152, 174). As this firing remains unaltered after complete motor paralysis, it is clear that it is actually a centrally generated premotor activity. Moreover, most of the neurons, if not all those which have a preswallowing activity, are located within the NTS (149, 152). In addition, systematic exploration of the brain stem with concentric bipolar electrodes to determine which central structures responded to stimulation by triggering deglutition have shown that the active points are situated only in the region of the solitary complex (50, 174, 225). It may be that deglutition results from the stimulation of afferent fibers belonging to the solitary tract. The wide dispersion of the active points suggests, however, that not only afferent fibers were stimulated but also neurons in or very close to the NTS. The assumption that NTS neurons may trigger swallowing was fully confirmed by more recent pharmacological experiments using fine microinjections of excitatory amino acids (EAA), which are thought to stimulate neuronal cell bodies and not passing fibers. It was established in these experiments that activating specific EAA receptors within the NTS elicit both the sequential and rhythmic motor swallowing patterns, whereas microinjections of the same drugs into the ventrolateral medulla failed to induce the swallowing motor pattern (124, 173, 178). Furthermore, electrolytic lesion of the NTS results in the abolition of not only the swallowing elicited by stimulating the SLN but also that elicited by stimulating the swallowing cortical area (152, 159). Finally, fine lesions performed on sheep in the NTS region, which contains neurons that control esophageal motility, abolished the esophageal phase of swallowing without affecting the oropharyngeal phase, which indicates that some of the neurons actually involved in the generation of esophageal motility had been destroyed within the NTS (150, 152).

As regards the swallowing neurons in the ventrolateral medulla, the results available are consistent with the view that during swallowing these neurons are driven by NTS neurons. As neurons of the VSG are connected to motoneurons, one of their functions probably consists of activating the motoneuronal pools during swallowing. The existence of neurons with collaterals to several pools of motoneurons also suggests that they may also participate in the coordination of the motoneuronal pools during swallowing (6, 94, 158). Within the swallowing CPG, the ventral swallowing neurons can therefore be said to act as switching neurons that distribute and coordinate the sequential or rhythmic drive generated in the dorsal group to the various pools of motoneurons involved in swallowing. It may well be that these neurons are also concerned in the coordination of the activity of motor nuclei (V, VII, IX, X, and XII) that are involved in several other types of motor behavior (see sect. II).

This scheme of organization of the diverse groups of swallowing neurons does not mean that the connections between the dorsal group and the ventral group and those between the ventral group and the motoneurons include only excitatory connections. Indeed, recent results obtained in intracellular studies on swallowing motoneurons in XII motor nucleus and the nucleus ambiguus (337, 376, 377) have suggested that during the functioning of the network, both excitatory and inhibitory drives can be exerted along the anatomical pathways.

In fact, the CPG for swallowing consists of two hemi-CPGs, each located on one side of the medulla (87). The existence of two hemi-CPGs was established by making longitudinal midline sections of the medulla. After this splitting, stimulation applied to the SLN on one side triggered a "unilateral swallowing," i.e., a swallowing sequence involving only the ipsilateral oropharyngeal muscles, except for the middle and inferior pharyngeal constrictors in some species (87, 89, 150, 152). These results indicate that under physiological conditions, the two hemi-CPGs are tightly synchronized and organize the coordinated contraction of the bilateral muscles of the oropharyngeal region.

The mechanisms underlying the synchronization of the two hemi-CPGs are not known, and this matter has not yet been well documented. It is likely that the peripheral afferent fibers do not play an important role in the coordination of the CPGs, since lesion experiments have shown that splitting the medulla caudal to the obex, which interrupts the vagal afferent fibers crossing the midline through the solitary tract, does not affect swallowing, i.e., a bilateral