Dorsal Respiratory Group

The DRG contains almost entirely inspiratory neurons. One kind of inspiratory neuron, the Ia neuron, like the phrenic motor neurons, demonstrates an augmenting pattern of firing that peaks at end-inspiration. Axons from these cells decussate in the medulla at or immediately rostral to the obex and connect with phrenic and inspiratory intercostal motor neurons in the spinal cord. Collaterals are sent to the ipsilateral inspiratory neurons in the VRG, but a few also are distributed to the expiratory neurons in the VRG.
The firing of Ib neurons, as of Ia neurons, occurs primarily in inspiration but, in contrast to the discharge of Ia neurons, is augmented by inputs from pulmonary stretch receptors. In animals, the discharge peak of Ib is not as sharp and the decline in activity during expiration is slower than for Ia. In the absence of excitatory input from vagal stretch receptors, however, the discharge patterns of Ia and Ib neurons are similar.
Like Ia neurons, some of the Ib neurons project to the spinal cord. Those Ib neurons that do not project to the cord appear to undergo extensive axonal arborization in the NTS. Ib neurons seem to be responsible for the shortening of inspiratory time induced by lung inflation. Their responsiveness to stretch-receptor input is less during expiration than during inspiration.
The DRG also contains late-onset inspiratory neurons that reach their peak firing rate in the transition from inspiration to expiration. Their activity, like that of Ib neurons, is facilitated by stretch-receptor activity. These neurons may participate in the short phase of graded inhibition of inspiratory activity seen with volume changes occurring toward the terminal portion of inspiration.
Recently, a small number of early-expiratory neurons have been observed in the DRG (and possibly also in the VRG) intermingled with inspiratory cells. They begin their firing shortly before the end of inspiration and reach peak discharge rates quickly, and then their activity slowly diminishes during expiration, disappearing before inspiration begins. Increases in lung volume slow the rate of decline in activity of these cells, whereas prevention of lung inflation does the opposite. The activity of these neurons is related to postinspiration inspiratory activity (PIIA), which occurs in the diaphragm and intercostal and laryngeal muscles and retards expiratory flow and the rate of lung deflation.

Central Respiratory Neuron

The precise organization of the central respiratory neurons is still a matter of contention. Although there may be respiratory pacemaker cells in which spontaneous changes in transmembrane potential occur, in the intact system, the respiratory rhythm depends on interconnections between different respiratory neurons.
Because breathing is preserved in anesthetized animals even after removal of the brain rostral to the pons, it is believed that the neurons on which respiratory rhythm critically depends are located in the bulbopontine region. Many investigators believe that the essential features of the respiratory rhythm remain even after separation of the pons from the medulla, and that the central pattern generator must be anatomically located within the confines of the medulla. There is evidence, however, that pontine neurons, particularly the complex composed of the nucleus parabrachialis medialis (NPBM) and the Kölliker-Fuse nucleus (KFN), as well as nuclei in the tegmentum (magnocellular and gigantocellular nuclei), significantly modify breathing.
In addition to these pontine and medullary respiratory neuronal aggregates, neurons with activity that is modulated by respiration can be found all through the brain stem intermixed with nonrespiratory neurons. It also has been shown that when breathing is stimulated, respiratory modulation of the activity of these neurons decreases according to level of anesthesia and sleep state.
A number of neurons whose firing patterns demonstrate a respiratory modulation but whose phase relationships with phrenic motor activity and with one another differ have been identified in the brain stem. Some of these neurons project to the spinal cord (bulbospinal) and are therefore true premotor cells. The remainder have axons that project to other parts of the brain (propriobulbar). Only the function of the bulbospinal neurons has been determined with any degree of certainty. It is generally believed, however, that the propriobulbar cells actively inhibit or excite other neurons involved in the respiratory cycle. The precise function of these propriobulbar neurons remains under investigation, although it is generally agreed that they are organized into networks whose complicated interactions determine the level of excitation of the bulbospinal neurons and produce respiratory phase switching.

CLINICAL IMPLICATIONS

The most important cause of respiratory failure is derangement of lung mechanics. However, respiratory failure does not develop in all patients, even those with severe impairment of pulmonary function. It has long been suspected that those patients who have the poorest chemosensitivity are the ones in whom CO2 retention is most likely to develop when the performance of the chest bellows is reduced. The evidence for this is indirect. For example, normal offspring of hypercapnic subjects with COPD demonstrate significantly lower ventilatory and occlusion pressure responses to hypoxia and hypercapnia (~60% lower) than do normal offspring of eucapnic subjects with COPD.
The CO2 sensitivity of children who have retained CO2 because of upper airway obstruction resulting from hypertrophy of the adenoids and tonsils is depressed, even after the tonsils and adenoids have been removed. Asthmatic patients who have retained CO2 during an asthmatic attack also show persistently low ventilatory responses to CO2, even after recovery from the asthmatic episode. Moreover, subjects who have had asphyxial, near-fatal episodes of asthma display lower ventilatory and occlusion pressure responses to hypoxia than do either age-matched normal subjects or asthmatic subjects with no history of near-fatal episodes.
There is also a small group of subjects who retain CO2 even though lung function is normal. In some of these patients, the cause of the depressed CO2 sensitivity is not known, but in others it seems to be associated with specific diseases, certain metabolic abnormalities, such as alkalosis, or the long-term administration of respiratory depressant drugs, such as methadone. These conditions are summarized in Table 1. In a few conditions listed in Table 2, only the ventilatory response to hypoxia is depressed. Individuals with these conditions are able to maintain blood gas tensions within usual limits because of their normal CO2 drive. However, when CO2 sensitivity is reduced by the administration of drugs (e.g., premedication before surgery), significant hypoxemia can develop. Depressed ventilatory responses to hypoxia may also increase the tendency for CO2 retention to develop in COPD and may be a risk factor for acute mountain sickness.

TABLE 1. Conditions sometimes associated with depressed responses to hypercapnia and hypoxia

TABLE 2. Conditions associated mainly witha decreased response to hypoxia

Certain conditions seem to predispose to heightened responses to CO2 or hypoxia, even when the lungs are normal. These conditions are listed in Table 3.

TABLE 3. Conditions associated with increased responses to carbon dioxide and/or hypoxia

Abnormalities in mechanoreceptor function also can influence gas exchange. Patients with chronic airway obstruction who breathe with small tidal volumes tend to retain CO2, whereas those who breathe with larger tidal volumes do not. The small tidal volumes are caused by abbreviated inspiratory time and perhaps by heightened pulmonary or chest-wall receptor activity. Heightened mechanoreceptor activity may also be responsible for dyspnea in some patients with interstitial lung disease, as vagal blockade at times alleviates this sensation.

TESTS OF PERIPHERAL CHEMORECEPTORS

The peripheral chemoreceptors also respond to CO2 and contribute about 20%–30% of the total ventilatory increase observed when CO2 is inhaled. Because peripheral chemoreceptors react rapidly to changes in CO2, peripheral chemoreceptor responses have been evaluated by measuring the immediate increase in ventilation caused by a few breaths of inspired CO2 or by measuring the immediate decrease in ventilation observed when CO2 is abruptly removed.
The response to hypoxia, like the response to hypercapnia, can be measured by either rebreathing or steady-state techniques. Because of the prominent effects of CO2 on breathing, it is important to keep the CO2 constant while the response to hypoxia is measured. Because O2 stores are small, the peripheral chemoreceptor response to O2 also can be evaluated by measuring the effect on ventilation of a few breaths of N2 or 100% O2.
No matter how it is measured, the ventilatory response to hypoxia is curvilinear, making quantitation difficult. The response can be made linear, however, by relating ventilation to the reciprocal of PO2 or to the arterial O2 saturation.
There are insufficient data to establish the range of normal values of the ventilatory response to hypoxia; however, available information indicates that it is closely related to the metabolic rate and is at least as variable as the CO2 response.
Prolonged periods of hypoxia, particularly early in life, are associated with depression of the chemoreceptor response to hypoxia. The ventilatory response to hypoxia is reduced in native residents of regions at high altitudes and in children with congenital cyanotic heart disease. The carotid body appears to be larger in native residents of high altitudes. The change in size may be caused by increased carotid body vascularity, which raises PO2 and decreases responsiveness.
In the newborn, hypoxia causes only a transient increase in ventilation, which then subsides to nearly prehypoxic levels. It has been recently appreciated that in adult humans, hypoxia lasting for as short a time as 5 minutes produces a gradual reduction in ventilation from its initial peak level.
The initial increase in ventilation is of course mediated by the carotid body. The subsequent decrease seems to represent a depressant effect of hypoxia on central respiratory neurons by hypoxia-induced increases in cerebral blood flow and probably the release of inhibitory neuromodulators, such as adenosine.
Lung disease or respiratory muscle weakness can depress ventilatory responses to chemical stimuli. The depressant effect seems to be greater for the response to CO2 than for the response to hypoxia. As a result of studies in which airway resistance was increased by requiring subjects to breathe through external resistance, either during inspiration or expiration, it was suggested that the inspiratory work of breathing at a given level of PCO2 is fixed, so that when the ratio of inspiratory muscle work to ventilation is increased by disease, ventilation decreases. The hypercapnia observed in severe obstructive lung disease was explained by the increase in flow-resistive work associated with chronic airway obstruction.
More recent studies have indicated that small increases in airway resistance have little effect on resting ventilation or CO2 response in normal subjects and may even heighten ventilation. The mechanisms responsible for the preservation of ventilation under these circumstances could include intrinsic properties of the respiratory muscles, increased inspiratory augmenting output from lung and chest-wall mechanoreceptors, readjustment in the sequence of contractions of the respiratory muscles so that mechanical advantage of the muscles and their coordination is improved, and increased inspiratory drive originating from the motor cortex. This last mechanism may depend on the conscious perception of changes in airway resistance. It is interesting that the ability to detect changes in airway resistance varies, decreasing with increasing airway resistance, and that it decreases further when airway obstruction is chronic than when it is acute.
Because mechanical changes may limit the ventilatory response to chemical stimuli, other methods of assessing the output of respiratory motor neurons have been devised. Two methods have been employed: measurement of occlusion pressure and EMG of the diaphragm. Neither is perfect, but both are useful under certain circumstances.
In the measurement of occlusion pressure, the force of contraction of the inspiratory muscles under quasi-isometric conditions is determined as follows: The airway is momentarily blocked at the beginning of inspiration, and the negative pressure developed during inspiration is measured. In conscious subjects, the reproducibility of the response is greater when the airways are occluded for just a fraction of a second. The occlusion pressure increases with hypercapnia and hypoxia and can be related to change in PCO2 and PO2 to estimate chemosensitivity.
Airway occlusion at FRC produces a no-flow state at the relaxed position of the respiratory system. The absence of air flow and prevention of significant volume change during inspiration prevent increases in airway resistance or decreases in compliance from affecting this index of respiratory output. In patients with mechanical abnormalities of the ventilatory pump caused by diseases of the lung or chest wall, occlusion pressure therefore more accurately reflects the neuromuscular drive to breathe than does ventilation.
Because the tensions developed by the inspiratory muscles theoretically depend on their initial length, the occlusion pressure in patients with lungs hyperinflated by disease may not reflect respiratory drive accurately. Increased FRC in animals reduces occlusion pressure responses. However, studies in conscious humans in whom FRC has been changed by altering body position show little effect on occlusion pressure, even when changes in FRC are fairly large (1000 mL). A conscious person apparently maintains constant muscle tension successfully, despite changes in initial muscle length, by altering neural output.
Measurement of the electrical activity of the diaphragm is probably the most direct way of evaluating respiratory neuronal output. This can be accomplished by passing a catheter containing electrodes down the esophagus and positioning the electrodes so that they straddle both surfaces of the diaphragm.
Various ways have been devised to quantitate diaphragmatic electrical activity measured this way. In the method most used currently, diaphragmatic activity is integrated over small intervals of time (100 to 200 ms), and the average activity per time limit is recorded (the so-called moving average). Electrical activity measured in this way depends on the exact positions of the electrodes in relation to the diaphragm during breathing, so that it is difficult to compare one individual with another. It is possible, however, to use this method to determine the effect of different therapeutic interventions in the same person.

TESTS OF CHEMORECEPTOR SENSITIVITY

When lung function is normal, the sensitivity of the peripheral and central chemoreceptors to CO2 can be evaluated by measuring the ventilatory response to inspired CO2. In the conventional steady-state test, the inspired CO2 is increased in steps, and ventilation at each step is related to the change in PaCO2. Sensitivity to CO2 is determined from the slope of the line relating ventilation to CO2. Although the central chemoreceptors are readily accessible to CO2, the size of cerebral CO2 stores increases the time required for ventilation to reach a steady state when PCO2 changes. Usually, the inspired CO2 concentrations at each step must be maintained constant for 10 to 20 minutes to ensure equilibration. Relative rates of equilibration of PCO2 in arterial and brain venous blood (Fig. 9) indicate that PaCO2 reaches its steady-state level long before the venous PCO2. It is apparent from Fig. 9 that if ventilation, which closely tracks cerebral venous PCO2, is measured too soon, chemosensitivity will be underestimated.

FIG. 9. Changes of ventilation, PaCO2, and cerebral venous PaCO2 when inspired CO2 is changed.

When CO2 is rebreathed from a bag containing CO2 at the mixed venous level together with O2, arterial and venous blood equilibrate more rapidly. After a brief transition period, PCO2 at all sites in arterial, cerebral, and mixed venous blood and alveolar air rises at the same rate. Consequently, the rate of change of PCO2 in alveolar air can be used as an index of the rate of change in PCO2 in the central chemoreceptors. The exact length of the transition period depends on the size of the rebreathing bag. When the volume of the rebreathing bag is about the same as the vital capacity, chemosensitivity can be estimated by continuous recording of ventilation and PCO2 after 45 to 60 seconds of rebreathing. Measurements for wide variations in PCO2 can be obtained in a few minutes. Estimates obtained by this rebreathing method agree with those obtained by the more prolonged steady-state technique. However, the rebreathing tests measure CO2 sensitivity at much higher levels of PCO2 than are usually encountered. Moreover, differences have been noted between rebreathing and steady-state ventilatory response to CO2 when metabolic acidosis or alkalosis is present. With the steady-state technique, moderate alkalosis and acidosis produce larger changes in the position of the ventilatory response line but relatively small changes in its slopes, whereas the reverse is true in the rebreathing tests. The explanation for this difference is obscure, but it may be related to the different levels of PCO2 at which steady-state and rebreathing tests are performed.

The average ventilatory response to CO2 is about 2.5 L/min/mmHg in normal adult men. It is somewhat less in women than in men and tends to decline with advanced age. It varies greatly between individuals but is much more constant in repeated measurements from a single subject. Some of this variability is caused by differences in personality, genetic makeup, and body size, and it is reduced when the CO2 response is corrected for differences in vital capacity.

Cortical activity is known to affect the response to CO2. Ventilatory responses to CO2 measured with the subject's eyes open are greater than ventilatory responses to CO2 measured with the subject's eyes shut.

CONTROLLED SYSTEM EFFECTS ON REGULATION OF BREATHING

The translation of the output of the inspiratory neurons to ventilation involves, as shown in Fig. 8, the successive transformation of nerve impulses to muscle electrical activity, muscle shortening, force, and then ventilation. Usually, moderate changes in the mechanical properties of the muscles or chest bellows have little or no effect on the resting blood gas tensions. Compensating effects by the chemoreceptor and mechanoreceptor reflexes, conscious adjustments, and the intrinsic force-velocity relationships of the muscles themselves allow the force of contraction to increase whenever the rate of contraction is slowed. In the presence of sufficiently severe chest disease, however, gas exchange is inadequate despite all efforts to compensate.

FIG. 8. Steps by which respiratory neural activity is translated into ventilation.

Even when the compensatory responses prove ultimately to be adequate, changes in mechanical conditions (or metabolic rate) cause a transient period in which gas exchange is disturbed and gas tensions are abnormal. The degree to which blood gas tensions deviate from normal in such situations depends on the volume and arrangement of the body stores of O2 and CO2.

CO2 is contained in the body in large amounts as gas in the lungs, but mainly in the form of bicarbonate and carbonate solutions in blood and tissues. O2, on the other hand, is stored in much smaller amounts in alveolar gas, in solution, and in combination with hemoglobin and myoglobin. Disturbances in gas exchange cause small changes in PCO2 because of the large size of the CO2 stores, but large changes in PO2.

Rates of change of PO2 and PCO2 depend not only on the volume of gas stores, but also on organization—that is, the way O2 and CO2 contained in different body tissue compartments are linked by the circulation, rates of perfusion, and metabolic rates in the various body compartments—and the ability of the tissues in each compartment to bind CO2 and O2.

The rate at which peripheral and central chemoreceptors respond to changes in inspired CO2 and O2 depends on the arrangement of the body gas stores. The small size of the arterial compartment and the high rate of carotid body blood flow allow peripheral chemoreceptors to respond quickly to changes in both O2 and CO2. The larger CO2 stores of the brain cause the central chemoreceptors to respond more slowly to changes in inspired CO2. This difference in response time of central and peripheral chemoreceptors has been used to distinguish the contribution of each receptor to the CO2 response.

RESPIRATORY SENSORY RECEPTORS

The receptors of the lungs and airways are innervated through the vagi and superior laryngeal and trigeminal nerves, and respond, as in other hollow visceral structures, to irritation of the lining layers and changes in distending forces. The mechanoreceptors associated with the respiratory muscles are innervated by spinal nerves and, like those in other skeletal muscles, monitor changes in joint movement and in the length and tension of the muscle itself.
Pulmonary Receptors
There are basically three types of pulmonary receptors: stretch receptors in the smooth muscles of the airway, irritant receptors in the airway epithelium, and J (juxtacapillary) receptors situated in the lung interstitium.
Stretch Receptors
Stretch receptors are innervated by large myelinated fibers. As the lung is inflated, these receptors inhibit inspiration, promote expiration, and initiate the Hering-Breuer reflex. In animals, lung inflation cuts short inspiration and produces expiratory apnea; the duration of apnea is proportional to the degree of inflation.
Direct measurements of stretch-receptor activity indicate that stretch receptors in humans are excited by even small changes in lung volume during quiet breathing. In humans, however, unlike what occurs in animals, vagal blockade to abolish stretch-receptor input does not affect breathing frequency or tidal volume at rest. Vagal blockade in both humans and animals does, however, prevent the increase in breathing frequency that occurs when ventilation is stimulated by hypercapnia or hypoxia and tidal volume is larger.
In animals, stretch-receptor activity helps to preserve tidal volume whenever the usual movements of the lung are hindered by changes in airways resistance or respiratory system compliance. Anything that retards lung inflation diminishes inspiratory inhibitory stretch-receptor activity. Therefore, inspiration is prolonged and tidal volume tends to approach its usual level when the airway is obstructed or respiratory compliance is reduced, despite mechanical interference. When expiration is hindered and lung deflation slowed, increased stretch-receptor activity heightens the force of contraction of the expiratory muscles and also prolongs expiratory time. Both these stretch-receptor actions tend to prevent mechanical impediments to expiration from increasing end-expiratory volume and, as a consequence, decreasing the resting length of the inspiratory muscle. Stretch-receptor activity, by promoting full expiration, helps preserve inspiratory muscle function.
Although stretch receptors are not important in humans in shaping resting breathing patterns, they may help maintain tidal volume when breathing is stimulated or lung or chest-wall mechanical performance is impaired. The increase in breathing frequency caused by stretch-receptor activity in animals and during stimulated breathing in humans decreases the work of breathing of the respiratory muscles, conserving the energy that has to be expended to produce gas exchange. Although it is well-known that peripheral inputs from lung mechanoreceptors strongly affect the timing of respiratory motor activity, at the present time it is difficult to separate clearly the ventilatory effects of the pulmonary stretch-receptor afferents from those of other vagal sensory components (e.g., irritant and C-fiber afferents). However, changes in vagal afferent activity elicited by phasic lung volume changes seem to control predominantly the duration of inspiration, whereas tonic inputs predominantly affect the duration of expiration.
Irritant Receptors
Irritant receptors, like the stretch receptors, are innervated by myelinated fibers, whereas unmyelinated fibers supply the J receptors. Unlike the stretch receptors, both irritant and J receptors are rapid-adapting (within seconds). Neither irritant nor J receptors have a pattern of firing that is related to the phases of inspiration and expiration. Consequently, it is believed that neither receptor has an important influence in determining the pattern of breathing at rest.
Mechanical stimulation of the airways or the inhalation of potentially noxious agents (e.g., particulate matter, nitrogen dioxide, sulfur dioxide, ammonia, and antigens) seems to excite irritant receptors and produce airway constriction. Stimulation of irritant receptors augments the activity of the inspiratory neurons and, by interaction with the stretch receptors, promotes rapid, shallow breathing. This pattern of breathing, in combination with airway constriction, may limit penetration of dangerous agents into the lung and prevent them from reacting with the gas-exchanging surfaces.
The inspiratory augmenting effect of irritant-receptor excitation and the increase in breathing frequency it produces may help maintain ventilation in asthmatic patients, even when the work of breathing is massively increased.
Irritant receptors can be excited by traction on the airways and are stimulated if atelectasis reduces lung compliance. These receptors seem to cause augmented breathing and the large sighs that occur sporadically during normal breathing, and help to open collapsed areas of the lung. As a consequence, irritant receptors help maintain adequate gas exchange.
J Receptors
J receptors are stimulated by pulmonary interstitial edema, but they also can be activated by various chemical agents, such as histamine, halothane, and phenyldiguanide. Activation of the J receptors causes laryngeal closure and apnea, followed by rapid, shallow breathing. When pulmonary edema develops as a result of exercise, J receptors seem to depress the activity of the exercising limbs by a somatic reflex involving cingulate gyrus. J receptors, together with irritant receptors, may be responsible for the tachypnea seen in patients with pulmonary embolus, pulmonary edema, and pneumonia.
Laryngeal Receptors
Mechanoreceptors and chemoreceptors in the upper airway reflexively affect the level and pattern of breathing, motor outflow to the upper airway and chest-wall muscles, and airway tone. The best-studied of the upper airway receptors are the laryngeal receptors. In fact, all areas of the laryngeal mucosa and deeper structures contain sensory nerve endings. Several types of laryngeal receptors have been described: (1) pressure receptors, (2) “drive” receptors, and (3) cold receptors.
Pressure receptors, the most numerous of the laryngeal receptors, are activated by increases in negative (intraluminal less than extraluminal pressure) or positive transmural pressure. Pressure receptors fire in response to both dynamic and static pressure changes, and are slow-adapting. Approximately, two thirds of the pressure receptors respond to negative pressure; the remaining third respond to positive pressure. Approximately one half of laryngeal pressure receptors demonstrate a respiratory modulation in the absence of air flow in the isolated, bypassed upper airway, suggesting that they respond to laryngeal muscle shortening in response to descending motor drive. These so-called drive receptors fire primarily during inspiration. Their firing pattern is diminished by paralysis of the intrinsic muscles of the larynx.
Reflexes elicited by laryngeal pressure receptors tend to stabilize the upper airway, retard its tendency to collapse in response to subatmospheric pressure, and re-establish its patency following occlusion. Laryngeal pressure receptors reflexively activate upper airway muscles while inhibiting inspiratory muscles of the chest wall. Negative transmural airway pressure reflexes increase the activity of inspiratory upper airway muscles (e.g., genioglossus, sternohyoid, cricothyroid, levator alae nasi, posterior arytenoids), advance the onset of the upper airway-muscle EMG relative to that of the diaphragm, increase the duration of inspiration and expiration, and decrease the average rate of rise of diaphragmatic and inspiratory intercostal EMG activity. (Normally, activation of upper airway muscles occurs 50 to 100 ms before the outset of diaphragmatic activation.) Reflex responses to negative pressure in the upper airway mediated by pressure receptors may explain the greater tidal volume, expiratory time, and ventilation during nasal than in tracheostomy breathing in conscious animals and humans.
In contrast to pressure receptors, laryngeal cold receptors are silent near body temperature but are activated by decreases in laryngeal temperature to 34°C or below. When active, cold receptors demonstrate a phasic, inspiratory firing pattern and, in contrast to pressure receptors, appear to adapt rapidly. Cold receptors appear to be located superficially in the mucosa on the edge of the vocal cords near the arytenoid process. Increases in lower airway resistance elicited by laryngeal cooling may be mediated by these receptors.
Finally, mechanical or chemical irritation of the larynx (e.g., probe contact or application of acid) elicits cough, laryngeal closure, bronchoconstriction, an increase in tracheal production of mucus, and a decrease in heart rate and blood pressure. These reflex responses to laryngeal irritation suggest that laryngeal chemoreceptors and mechanoreceptors function to protect the lower airway from aspiration or inhalation of toxic fumes.
Of interest, reflex responses to laryngeal stimulation appear to be state-dependent and are qualitatively different during wakefulness and sleep. For example, in the dog, application of distilled water to the larynx during wakefulness consistently elicits cough and bronchoconstriction. In contrast, the same maneuver performed during REM sleep does not stimulate cough, but rather elicits apnea and bradycardia.
Chest-Wall Receptors
Three types of receptors in the chest wall—joint, tendon, and spindle receptors—signal changes in the force exerted by the respiratory muscles and movement of the chest wall. Specialized Ruffini receptors, as well as pacinian and Golgi organs, are present in joints. Joint-receptor activity, which can be consciously perceived, varies with the degree and rate of change of rib movement.
Inputs arising from muscular receptors, both proprioceptive (particularly muscle spindle) and nociceptor afferent (types III and IV) endings, influence the level and timing of respiratory activity. Proprioceptor afferents (chiefly from the intercostal and abdominal muscles) project to the phrenic motor neurons, where their effect is on firing rate only, and to medullary respiratory neurons in the DRG and NRA, where their predominant effect is on respiratory timing.
Tendon organs in the intercostal muscles and diaphragm monitor the force of muscle contraction and produce an inspiratory inhibitory effect. It was once thought that tendon organ activity was provoked only by unusual levels of muscle force, but it is now believed that tendon organs are stimulated by even small changes in force. Tendon organ input may be important in regulating both intercostal muscle and diaphragmatic contraction during breathing at rest.
Muscle spindles, which are abundant in the intercostal muscles but scarce in the diaphragm, are involved in several kinds of intercostal respiratory reflexes and also help coordinate breathing during changes in posture and speech.
Figure 7 shows schematically the operation of the spindle and its neural connections. Spindles are located on intrafusal muscle fibers aligned in parallel with extrafusal fibers, which move the ribs. Motor innervation of the extrafusal fibers originates in alpha motor neurons, whereas the intrafusal fibers receive motor innervation from gamma (fusimotor) motor neurons. Passive stretch of an intercostal spindle by lateral flexion of the trunk, for example, increases spindle afferent activity and activates a monosynaptic segmental reflex that causes contraction of the parent extrafusal fiber and restores the upright position. The spindles also can be stretched by an efferent fusimotor discharge, which causes contraction and shortening of the intrafusal fiber itself. Some fusimotor fibers fire phasically, so that their rate of discharge rises during inspiration and falls during expiration; other fusimotor fibers are tonically active. The cerebellum determines the balance between tonically and phasically active fusimotor fibers. Without phasic fusimotor activity, spindle discharge would decrease when the extrafusal fibers contract during inspiration. Simultaneous activation of fusimotor and alpha motor neurons causes the spindles to be under continuous stretch during inspiration and enhances the contribution made by the intercostal muscles to respiration. If inspiratory movements are impeded, afferent activity from a spindle innervated by a phasically active fusimotor fiber is enhanced, thus increasing inspiratory muscle force and helping to preserve tidal volume. Activity from lower intercostal muscle spindles, through an intersegmental spinal reflex, also enhances diaphragmatic contraction, allowing the diaphragm to contribute to the compensatory increase in muscle force that occurs when respiratory movements are hindered. In contrast, stretch of the intercostal spindles in the midthoracic region of the chest decreases the duration of inspiration and diminishes the force of inspiratory muscle contraction. This reflex may cut short ineffective inspirations. Ineffective inspirations sometimes are seen in the newborn when the negative intrathoracic pressure produced by powerful diaphragmatic contraction causes paradoxical inward movement of the flexible infant rib cage.

FIG. 7. Intercostal muscle spindle.

Of considerable importance, spindle afferents reach the highest level of the central nervous system, the sensorimotor cortex. Projection of spindle afferent activity to the cerebral cortex allows respiratory muscle length and tension to be sensed consciously and modulated with great precision, thereby allowing complex volitional acts to be performed (e.g., speaking, playing a wind instrument). Spindle afferent activity also likely contributes to the sense of breathlessness. It has been suggested that dyspnea occurs when spindle afferent activity is “high” relative to the intensity of central motor activity to the inspiratory muscles. This concept, which has been termed length-tension inappropriateness, explains the dyspnea that arises in the setting of lung diseases that increase inspiratory muscle load and impede muscle shortening. Of interest, the sense of breathlessness can be affected in patients with chronic obstructive pulmonary disease (COPD) by application of vibratory stimuli to the intercostal muscles, which changes spindle afferent activity. Dyspnea is ameliorated by vibratory stimuli applied in phase with muscle contraction and worsened when the vibratory stimulus is applied out of phase with muscle contraction.

Integration of Afferent Input

Although it is clear that afferent input to the medullary respiratory neurons from mechanoreceptors in the lungs, respiratory muscles, and cardiovascular and thermal regulatory systems (and even the exercising limbs) have significant effects on breathing, the precise manner in which these inputs are integrated is poorly understood. However, the changes in respiratory motor activity elicited by changes in these inputs are not stereotyped. The reflex responses to these inputs may affect the motor output to some respiratory muscles more than others. Pulmonary stretch-receptor input inhibits chest-wall muscle activity (i.e., diaphragm and external intercostal muscles) but increases the activity of the upper airway-dilating muscles (i.e., posterior cricoarytenoid) and the chest-wall expiratory muscles. Even more interesting, some receptors seem to have opposing effects on muscles that normally act as agonists. For example, stimulation of esophageal mechanoreceptors by balloon distension of the distal esophagus reflexively inhibits diaphragmatic activity, both costal and vertebral, but enhances external intercostal activity.

PERIPHERAL CHEMORECEPTORS

Sensors in both the carotid body (innervated by the ninth cranial nerve) and the aortic body (innervated by the tenth cranial nerve) respond to hypoxia by increasing ventilation. If the carotid and aortic bodies are removed, hypoxia depresses breathing. In most species, the increase in ventilation with hypoxia is more a consequence of carotid than of aortic body activity. The carotid body also responds, to a limited extent, to changes in PCO2 and hydrogen ion concentration, and it appears to be particularly important in the immediate increase in ventilation seen in metabolic acidosis. However, the increase in peripheral chemoreceptor activity caused by CO2 appears to be inconsequential under hyperoxic conditions.
With decreases in PaO2, afferent fibers from the carotid body increase their discharge hyperbolically. Reduction in PO2 rather than in O2 content in the arterial blood is mainly responsible for the increasing activity. The biochemical and physiologic mechanism that allows the carotid body to respond to even relatively mild hypoxia has not been completely elucidated, but some details are known. Although blood flow in the carotid body is unusually high, so is the metabolic rate. Vascular shunts through the carotid body as well as its high metabolic rate may produce areas of hypoxia within the carotid body, even when the arterial blood is fully saturated with O2. Measurements of carotid body PO2 have shown some extremely low tensions, but the range of tensions is wide. Cytochrome enzymes within the carotid body may have an especially low affinity for O2, thus accounting for the sensitivity of the carotid body to changes in PO2. Although the primary function of the peripheral arterial chemoreceptors is to transduce changes in arterial PO2, PCO2, and/or hydrogen ion levels into nerve signals, there is no general agreement as to how this is accomplished, nor is it known whether all stimuli act through a common mechanism.
Ultrastructural studies of the carotid and aortic bodies demonstrate the presence of two distinct types of cells. Afferent nerve terminals from the carotid sinus nerve appose type I glomus cells, which contain abundant, dense, clear-cored synaptic vesicles, mitochondria, and conspicuous rough endoplasmic reticulum. The cytology of type II (sustentacular) cells resembles that of Schwann cells. They envelop the afferent terminal-glomus cell complex.
Whereas it was originally proposed that the afferent terminals are chemosensitive and that the type I cells function as modulatory interneurons, subsequent studies have suggested that the integrity of the glomus cells (type I and perhaps type II cells) is essential for the process of chemoreception. After the glomus cells are destroyed, nerve endings alone seem unable to respond to physiologic stimuli.
Glomus cells (type I cells) contain a variety of agents, including acetylcholine, norepinephrine, dopamine, and 5-hydroxytryptamine. Recent immunocytochemical studies also have shown the presence of at least three polypeptides in the carotid body of cats and rats (i.e., substance P, vasoactive intestinal polypeptide [VIP], and enkephalins), suggesting that neuropeptides may play important roles in the transmission of nerve signals. Substance P, a member of the tachykinin group of polypeptides, has been proposed as a general transmitter/modulator for primary afferent fibers sensing nociceptive stimuli. In addition, substance P enhances the discharge of carotid body preparations in vivo and in vitro. These excitatory effects of substance P are dose-dependent, seem to be slow in onset, and last several seconds after intracarotid administration. Hypoxic excitation of the carotid body is markedly attenuated by substance P antagonists. The mechanism(s) for sensing O2 in the carotid body remains unclear. However, hypoxia depolarizes type I cells and increases cytosolic calcium, perhaps through effects on O2-sensitive, voltage-gated potassium channels and cytochrome protein(s) with a low affinity for O2. Depolarization of glomus cells in turn causes neurotransmitter release and activation of sinus nerve afferent terminals.
Efferent discharge to the carotid body from the CNS depresses afferent activity provoked by hypoxia. This efferent inhibition may prevent saturation of the carotid body response, allowing the carotid body to respond to a wider range of PO2 than it could otherwise. In part, efferent control depends on sympathetic nervous regulation of carotid body blood flow. However, other inhibitory efferent fibers that have no effect on the carotid body vasculature are also present.
With hypoxia, ventilation, like carotid body activity, increases hyperbolically (Fig. 6). Also, changes in PCO2 seem to enhance the ventilatory response to hypoxia, and vice versa (i.e., CO2 and hypoxia interact multiplicatively). Single carotid body fibers respond to both CO2 and hypoxia, so that some of the interaction of hypoxia and hypercapnia occurs at the cellular level in the sensor itself. However, other evidence suggests that convergence of input from central and peripheral chemoreceptors at the level of the CNS helps enhance the interaction of hypoxia and hypercapnia as ventilatory stimulants. Experimental studies on the effect of carotid nerve stimulation in different phases of breathing show that carotid body discharge is more effective in stimulating breathing during inspiration than during expiration. Carotid body discharge varies spontaneously during the breathing cycle as a result of variations in PaO2. The relationship between oscillations in carotid body activity and phase of breathing depends on the circulation time between the lungs and the carotid body. Thus, changes in cardiac output theoretically might affect both the level and pattern of breathing. When central chemoreceptor activity and the response to CO2 have been eliminated by destruction of the ventrolateral medullary surfaces, input from the carotid body alone is sufficient to maintain rhythmic breathing. Both central and peripheral chemoreceptors respond proportionally as the level of PCO2 is altered. Some studies suggest that increases in the rate of change of CO2 but not in the rate of change of PO2 also stimulate the carotid body.

FIG. 6. Effect of changing PaCO2 on (A) the ventilatory response to hypoxia and (B) the ventilatory response to hypercapnia.

CENTRAL CHEMORECEPTORS

When CO2-enriched gas is inspired, ventilation increases. The increase in ventilation tends to minimize the rise in PaCO2. Because the amount of CO2 delivered to the chemoreceptors depends on the CO2 carried to them by the arterial blood, the PaCO2 determines the PCO2 in the immediate environment of the chemoreceptors.
The effect of increases in ventilation on PaCO2 can be determined by the following equation:

where 

CO2 is the metabolic production of CO2 each minute, 

A is the alveolar ventilation, PICO2 is the partial pressure of inspired CO2, and K is a proportionality constant.

It can be seen that the greater the increase in ventilation caused by a change in PICO2, the lower is the PaCO2. In conscious humans, central chemoreceptors located within the medulla account for 70%–80% of the increase in ventilation. The peripheral chemoreceptors account for the remainder of the increase in ventilation when CO2-enriched gas is inspired and for all the increase in ventilation produced by hypoxia.

The exact location of the central chemoreceptors is still disputed, although most experimental data indicate that they (1) are distinct from the inspiratory motor neurons themselves, (2) are not located in the dorsal and ventral groups described earlier, and (3) respond to changes in hydrogen ion concentration of brain interstitial fluid but also may respond directly to changes in PCO2 (perhaps through a change in intracellular pH).

Studies in which drugs and temperature probes have been applied to the ventrolateral surface of the medulla have demonstrated abrupt and striking ventilatory effects, suggesting that many of the neurons comprising the central chemoreceptors or their associated axons may be located near the surface. Chemoreceptor activity can be influenced from three different superficial areas (Fig. 4). Recent studies suggest that respiratory cells near the ventral surface are intermingled with cells that also have significant vasomotor effects. Many agents that increase ventilation when applied superficially to the ventral medullary surface (e.g., nicotine, acetylcholine, kainic acid) also raise blood pressure. On the other hand, agents that decrease respiration when similarly applied (e.g., t-amino butyric acid, taurine, enkephalins) also decrease blood pressure. Nonetheless, discrete areas have been described from which either respiratory or vasomotor effects predominate (e.g., the nucleus paragigantocellularis, a collection of cells close to the ventral medulla).

FIG. 4. Ventrolateral medulla and its rostral (R), intermediate (I), and caudal (C) chemosensitive areas.

The crucial experiments in which activity from the central chemoreceptors would be directly recorded have never been performed. Hence, it has been possible to evaluate central chemoreceptor activity only indirectly. This is usually accomplished by measuring the increase in ventilation or phrenic nerve activity produced when the PICO2 is changed. Changes in ventilation theoretically should be related to changes in hydrogen ion concentration in the brain, but this concentration cannot be measured easily, either in humans or in animals. Instead, changes in ventilation are conventionally related to the measured levels of PaCO2. This kind of indirect estimation of central chemoreceptor activity is valid only under restricted circumstances and only if certain assumptions are made. It is assumed, for example, that in a steady state, after CO2 has been inspired for 10 to 20 minutes, changes in PaCO2 reflect changes in the hydrogen ion concentration of the brain. By the Henderson-Hasselbalch equation, hydrogen ion concentration in the brain, as in other tissues, varies according to the ratio PCO2/ HCO–3, where PCO2 is the partial pressure of CO2 at the chemoreceptor in the interstitial fluid and HCO–3 is the bicarbonate concentration of the medullary interstitial fluid. Hence, increases in bicarbonate concentration decrease hydrogen ion concentration, whereas decreases in bicarbonate concentration have the opposite effect.

The relationship between arterial PCO2 and PCO2 in the brain interstitial fluid depends on cerebral venous PCO2 and therefore on cerebral blood flow. The greater the cerebral blood flow, the smaller the difference between PCO2 in arterial blood and in interstitial fluid. As cerebral blood flow increases with PCO2, the change in ventilation produced by a change in PaCO2 depends on the CO2 responsiveness of cerebral blood vessels as well as on the sensitivity of the central chemoreceptors. This may be a significant factor in patients with cerebrovascular disease.

Changes in blood bicarbonate levels are not immediately mirrored in the brain interstitial fluid. In addition, evidence suggests that hydrogen ion concentration in interstitial fluid is actively regulated by cellular pumps at the blood-brain barrier or by the metabolism of brain cells. This means that in metabolic acidosis or alkalosis, neither PaCO2 nor hydrogen ion concentration in blood may reliably indicate the status of hydrogen or bicarbonate ion concentrations in interstitial fluid.

The stimulatory effect of acid injected into the blood on the peripheral chemoreceptors lowers PCO2. Because the transfer of PCO2 between blood and brain interstitial fluid is faster than the transfer of hydrogen or bicarbonate ions, the brain interstitial fluid may actually become alkaline when the blood PCO2 is acutely made acidic. Direct administration of acid into the cerebrospinal fluid to bypass the blood-brain barrier increases the hydrogen ion concentration in brain interstitial fluid and drives the PaCO2 down by stimulating central chemoreceptors.

With chronic acid-base disturbances, hydrogen ion changes in cerebrospinal fluid are usually qualitatively the same as those in the blood but are quantitatively less.

The effect of chronic metabolic acidosis and alkalosis on ventilatory responses to CO2 is shown in Fig. 5. It can be seen that in metabolic acidosis the level of ventilation is greater at any given level of PCO2, whereas in metabolic alkalosis ventilation decreases. These changes in ventilation reflect the altered level of bicarbonate in the brain interstitial fluid. If the same ventilation results are plotted as a function of hydrogen ion concentration in brain interstitial fluid, the response lines are identical.

FIG. 5. Effect of changes in blood bicarbonate on the relationship between ventilation and PCO2. The asterisk indicates the response line at the usual level of HCO–3.

In humans and animals, increases in PCO2 over a wide range cause a virtually linear increase in ventilation. At levels of PaCO2 >80 to 100 mmHg, the response to hypercapnia diminishes and may plateau. Decreases in PaCO2 below the usual level depress ventilation. In anesthetized and sleeping animals and humans, artificial hyperventilation with progressively reduced PCO2 eventually produces apnea. In a normal, awake human, however, active voluntary hyperventilation rarely causes apnea. In most cases, when voluntary hyperventilation is suspended, the increase in ventilation persists for perhaps 30 to 50 seconds. The persistence of ventilation in awake subjects at low levels of PCO2 has been attributed to a “wakefulness drive” caused by the continued impingement of other stimuli (e.g., noise, mechanoreceptor input, and light input) on the respiratory neurons. However, continuation of phrenic nerve activity at low levels of PCO2 has been described even in anesthetized animals made to hyperventilate actively by electrical stimulation of the carotid body nerves. This effect, which has been attributed to persisting reverberations in medullary respiratory neuron circuits, probably contributes to the wakefulness drive and helps stabilize breathing.

PATTERN OF MOTOR OUTFLOW TO THE INSPIRATORY MUSCLES

The firing of the bulbospinal inspiratory neurons projecting to the diaphragm and intercostal muscles increases progressively throughout inspiration and is terminated abruptly (off-switching). The ramplike increase in activity of these bulbospinal neurons (the central inspiratory activity) causes a progressive increase in excitation of the inspiratory muscles and hence their force of contraction (Fig. 2). The electrical and mechanical analogues of central inspiratory activity are, respectively, the integrated activity of the phrenic neurogram and diaphragmatic electromyogram (EMG) and the pleural pressure waveform. The progressively augmenting shape of central inspiratory activity allows the inspiratory musculature to overcome the progressive increase in elastic recoil of the lung during inspiration despite progressive shortening and a decrease in the intrinsic ability of the inspiratory muscles to generate force (i.e., the length-tension relationship).

FIG. 2. Effect of hypercapnia on the duration of the phrenic nerve electrical activity integrated as a moving time average and its rate of increase and the pleural pressure waveform. Note the similarity in shape of the integrated phrenic neurogram and the pleural pressure tracing. Following bilateral vagotomy, the duration of inspiration remains relatively constant despite the progressive increase in PCO2.

Control of the rate of rise in central inspiratory activity and hence the rate of lung inflation differs from control of inspiratory off-switching. Both chemical (e.g., hypoxia and hypercapnia) and nonchemical (e.g., thermal and mechanoreceptor afferents) inputs affect the steepness of the ramp of central inspiratory activity. On the other hand, the timing of inspiratory off-switching depends largely on inputs from pulmonary stretch receptors and from higher CNS structures, such as the NPBM and the KFN.

In anesthetized animals, phasic increases in lung volume resulting from the ramp of central inspiratory activity progressively increase pulmonary stretch-receptor activity. Integration of inputs from pulmonary stretch receptors and projections reflecting the intensity of the central inspiratory activity by as yet incompletely described pools of neurons terminates inspiration. Vagotomy eliminates stretch-receptor input, prolonging inspiration and increasing tidal volume, but the rate of rise in central inspiratory activity and hence the rate of inspiratory air flow are virtually unchanged. On the other hand, hypoxia and hypercapnia increase the steepness of the ramp of inspiratory activity and hence increase the rate of inspiratory air flow and tidal volume, but they have little effect on the duration of inspiration and frequency of breathing.

When the vagus is intact, so that respiratory neurons receive input from the stretch receptors as well as inputs reflecting central inspiratory activity, the duration of inspiration is reduced, because the inspiratory off-switch is activated earlier. Because central inspiratory activity increases with time, more stretch-receptor input (i.e., a greater change in lung volume) is needed early in inspiration to terminate a breath. This accounts for the curvilinear relationship between tidal volume (Vt) and inspiratory time (tinsp) that has been noted in studies of anesthetized animals (Fig. 3).

FIG. 3. Effect of lung volume information in determining off-switch and, hence, tinsp. Vagal input allows the off-switch threshold to be reached earlier in inspiration. The numbers refer to the PCO2 with the vagi intact. Inspiratory time declines and tidal volume rises with increasing hypercapnia. Without lung volume information (vagotomy), tinsp is fixed.

Consistent with these observations is the idea that ventilatory responses to hypercapnia and hypoxia depend on the sensitivity of both stretch receptors and chemoreceptors. Chemoreceptor sensitivity, because it influences the rate of increase in central inspiratory activity, is more closely related to the average level of air flow during inspiration than to minute ventilation. That is, the change in the ratio of tidal volume to inspiratory time, rather than the change in ventilation itself, most closely reflects chemical drive. On the other hand, the change in inspiratory time as a fraction of total breath duration indicates the activity of stretch receptors.

Although ventilation is conventionally thought to be equal to tidal volume times frequency (f), the concept of central respiratory neuronal organization suggests that ventilation should more realistically be considered to be the product of the following:

where tinsp is inspiratory time and texp is expiratory time.

Some studies in humans have tried to separate neural and chemical responses to hypoxia and hypercapnia by analyzing ventilatory responses with this approach. In some cases, depressed ventilatory responses to CO2 seem to be caused by altered mechanoreceptor function rather than by depressed chemosensitivity.

It is important to remember that this concept originated from experiments carried out in anesthetized animals and accordingly does not include the effects on breathing of inputs eliminated by anesthesia. These additional inputs, occurring during both wakefulness and sleep, may greatly distort the basic relationships between the medullary respiratory neurons observed in animals during anesthesia. Thus, in awake humans, increases in breathing frequency produced by hypercapnia and hypoxia are associated mainly with a shortening of expiratory time, whereas inspiratory time remains relatively constant. Rapid-eye-movement (REM) sleep is associated with an irregular breathing pattern and seems to eliminate ventilatory increases to hypercapnia, but not to hypoxia. In non-REM sleep, breathing is more regular, but responses to changes in CO2 remain lower than during wakefulness.

Even in anesthetized animals, influences from thermal and circulatory receptors can affect breathing. For example, temperature increases accelerate the frequency of breathing without changing tidal volume.

Timing Of Respiratory Motor Activity

During inspiration, firing rates increase monotonically in both inspiratory propriobulbar and bulbospinal neurons. Early in expiration, inspiratory propriobulbar neurons are silenced, but the activity of inspiratory bulbospinal neurons stops only momentarily, reappearing after a brief period of silence and then gradually declining as expiration proceeds. This PIIA corresponds in time to the period of firing of early-expiratory neurons in the DRG and VRG. Expiratory bulbospinal neurons are silent during this early phase of expiration, whereas inspiratory propriobulbar neurons are actively inhibited. Furthermore, because the time course of inhibition of the respiratory propriobulbar neurons is similar to the time course of activity of the early-expiratory units, it has been suggested that the respiratory rhythm is caused by inhibition of an inspiratory ramp generator by these early-expiratory neurons.
Based on these observations, it has been proposed that expiration be divided into two phases, EI and EII. The EI phase corresponds to the period of PIIA, whereas the EII phase corresponds to the period in which PIIA is absent and expiratory neuronal activity may be present. In some situations, PIIA may extend throughout expiration, suggesting that the respiratory rhythm does not depend on the occurrence of activity in expiratory neurons.
PIIA appears to be associated with “braking” of expiratory air flow by contraction of the inspiratory muscles. Increases in PIIA and prolongations in EI occur, for example, when the larynx is bypassed so as to decrease upper airway resistance. PIIA (EI) is markedly reduced or eliminated by vagotomy, suggesting that mechanoreceptors that sense lung volume and/or tracheal air flow are important inputs. Hypercapnia decreases the duration of EI, whereas hypoxia appears to do the reverse. Increases in PIIA may contribute to the increase in functional residual capacity (FRC) observed during hypoxia.
Respiratory timing can be significantly affected by the rostral pontine pneumotaxic center, which comprises the NPBM and KFN. This structure contains a number of neurons that have different patterns of firing: inspiratory, expiratory, or phase-spanning. When the vagi are intact, discharge patterns in the NPBM are mainly tonic, but they become more clearly phasic after vagotomy. Both the VRG and the DRG send projections to the pneumotaxic center, so that the respiratory activity seen in this center appears to be of medullary origin. Depending on the region involved, stimulation of the pneumotaxic center can either terminate or prolong inspiration. Stimulation of the dorsolateral region terminates inspiration. The earlier in inspiration the stimulation is applied, the stronger is the stimulus needed. If the pneumotaxic center is lesioned and the vagi are cut, an apneustic breathing pattern develops in anesthetized animals that is characterized by prolonged inspiratory time. If time is allowed for recovery, however, and the animal regains consciousness, breathing loses its apneustic quality. If the animal is then given anesthesia or allowed to go to sleep, the apneustic pattern returns. These observations suggest the lack of importance of the pneumotaxic center in generating the respiratory pattern, and indicate that an interaction between states of alertness and the activity of higher brain centers and the brain stem bulbopontine respiratory neurons can significantly affect respiratory rhythm.
It is not clear whether some or any of the different respiratory neurons described in fact make up the central pattern generator. Three different ways in which the central respiratory pattern may be produced in the brain have been proposed. In one, the pattern generator is composed only of inspiratory neurons; an inspiratory ramp continues until it is terminated by the activity of off-switch neurons. The off-switch neurons are triggered after some predetermined time or after the inspiratory ramp reaches some threshold level of activity. Both trigger and ramp neurons could be stimulated by hypoxia and hypercapnia. In this scheme, inspiration is a self-terminating process carried out by cells whose activity is confined to inspiration.
In a second hypothesis, the central pattern generator may include both inspiratory and expiratory cells affected by chemical drives causing tonic increases in the activity of each. The increasing ramplike discharge seen in inspiratory intercostal and phrenic nerves may result from a gradual decline in inhibition rather than a gradual increase in excitation. This hypothesis is based on the observation that during apnea induced by hypocapnia, decreases in PO2 elicit inspiratory tonic activity. Progressive decreases in PO2 during apnea elicit progressive increases in tonic inspiratory activity until at a critical level of hypoxia the respiratory rhythm reappears. On the other hand, hypocapnia under hyperoxic conditions produces continuous firing of expiratory neurons, which increases as PCO2 rises until rhythmic breathing resumes. This suggests hypoxia exerts an excitatory effect predominantly on inspiratory activity, and that hypercapnia affects expiratory motor activity.
The third idea is that respiratory rhythmogenesis arises in the antagonistic activity of inspiratory and early-expiratory neurons and does not depend on the activity of conventional expiratory neurons that peaks late in the expiratory phase.

Interrelationship Between Dorsal and Ventral Respiratory Groups

The precise interactions between the DRG and VRG remain unclear. Earlier studies indicated that inspiratory neurons of the DRG projected to the VRG, but a reciprocal connection was not apparent. These studies suggested that the central pattern generator was composed only of inspiratory cells and was located in the DRG. In this view, the DRG was the prime mover in the genesis of the respiratory rhythm, dominating the cells in the VRG and governing their activity. More recent studies indicate that cells from the VRG (Bötzinger complex) may inhibit inspiratory neurons in the DRG. Ablation experiments eliminating either the entire DRG or the Bötzinger complex in the VRG do not eliminate rhythmogenesis, indicating that substantial redundancy is present in the system.
The functional significance of the interconnections between groups of inspiratory cells in the DRG and VRG is also unclear. These interconnections may serve to synchronize the timing of neuronal firing in anatomically separate locations. For example, midline incisions through the medulla are associated with asynchronous firing of the two phrenic nerves.

Ventral Respiratory Group

The VRG in the medulla comprises several anatomically and probably functionally distinct populations (Fig. 1). One classification divides the neurons of the VRG into three aggregates: the nucleus retroambigualis (NRA), the nucleus para-ambigualis (NPA), and the nucleus retrofacialis (NRF).
The nucleus ambigualis (NA) is composed primarily of subnuclei of motor neurons innervating the laryngeal, pharyngeal, and facial muscles. This nucleus also contains the vagal motor neurons innervating the bronchial smooth muscles and the smooth muscles of the thoracic and abdominal viscera. These neurons are almost completely inactive during deep anesthesia, suggesting that they are not essential to respiratory rhythmogenesis.
The NPA is located in the region medial to the NA and 1 mm caudal to 3.5 mm rostral to the obex. The NPA is composed primarily of premotor inspiratory neurons, but some expiratory cells are present.
Most of the inspiratory neurons of the NPA (like Ia cells in the DRG) fire in a ramplike fashion, with peak activity occurring at the conclusion of the inspiratory phase. The NPA also contains a few inspiratory propriobulbar, so-called early-burst neurons. These cells begin to discharge slightly before the onset of the phrenic discharge, peak rapidly, and then demonstrate a decline and disappearance of activity in the latter half of inspiration. They send no projections to spinal motor neurons, but they have a rich pattern of arborization with expiratory neurons in the contralateral NRA, whose activity they appear to inhibit.
The activity of other neurons in the VRG (located in the NRA and NRF) is mainly directed to expiration. The expiratory neurons in the NRA demonstrate a slowly augmenting pattern of activity, with peak discharge late in expiration. Input from pulmonary stretch receptors prolongs the time of firing of these neurons. Hypercapnia causes these neurons to discharge earlier in inspiration and increases the steepness with which their rate of discharge rises. Lesioning experiments indicate that neurons in the NRA are the sole source of expiratory premotor neurons but are not of fundamental importance in generating the respiratory rhythm.
Respiratory neurons in the NRF (also called the Bötzinger complex) and in an area immediately rostral to it, called the pre-Bötzinger complex, have been described. Bötzinger neurons discharge mainly in expiration with a slowly augmenting firing pattern that peaks at end-expiration. They send projections to the DRG on the opposite side and seem to inhibit the inspiratory neurons located there. On the other hand, pre-Bötzinger neurons fire during inspiration, demonstrate pacemaker-like activity, and appear to be exclusively propriobulbar in type. In the neonatal rat, lesions in this pre-Bötzinger complex eliminate respiratory rhythmogenesis. Some pharyngeal motor neurons also can be found in the NRF.

Fig. 1

Medullary Respiratory Neurons

The respiratory neurons in the medulla seem to be aggregated into two groups (Fig. 1). One collection, the ventral respiratory group (VRG), forms a longitudinal column of neurons in the ventrolateral part of the medulla. It extends rostrally from the upper border of the spinal cord almost to the bulbopontine boundary. The other group, the dorsal respiratory group (DRG), is more circumscribed anatomically. It is located in a more medial and dorsal part of the medulla in the region of the ventrolateral nucleus of the tractus solitarius (NTS) and extends from the obex about 2.5 mm rostrally.

FIG. 1. Schematic depicting the organization of medullary respiratory neurons in the dorsal and ventral respiratory groups (DRG and VRG, respectively). Structures on one side only are shown. Axons from inspiratory bulbospinal neurons in the nucleus tractus solitarius (NTS) and the nucleus para-ambigualis (NPA) decussate rostral to the obex and extens caudally in the contralateral cord. Axons from expiratory bulbospinal neurons in the nucleus retroambigualis (NRA) decussate caudal to the obex. Bötz = Bötzinger complex in the nucleus retrofacialis; NA = nucleus ambigualis.

The nuclei of the solitary tract appear to function as relay stations for important respiratory and cardiovascular information. Afferents from pulmonary stretch receptors and carotid chemoreceptors and baroreceptors appear to synapse for the first time in the brain at this location.

Ventilation

The respiratory system, along with cardiovascular structures, operates as part of an intricate organization controlled by the central nervous system (CNS) to ensure optimal cell performance, providing sufficient oxygen to meet metabolic requirements and removing enough carbon dioxide so that cell function is not impaired by excessive changes in hydrogen ion concentration. The major function of the respiratory system is to maintain the arterial tension of oxygen (PaO2) and carbon dioxide (PaCO2) within acceptable limits in the face of changing metabolic needs and environmental conditions. To achieve this, the system is equipped with multiple sensors that monitor changes in blood chemistry (chemoreceptors) and changes in the mechanical properties of the lung and chest wall (mechanoreceptors). The chemoreceptors and mechanoreceptors allow ventilation to be continuously readjusted in accordance with metabolic needs, despite changes in body posture that alter the mechanical advantage or movement of the respiratory muscles. In addition, these receptors coordinate the contraction and relaxation of the respiratory muscles, so that adequate gas exchange is carried out with minimum expenditure of energy.
In addition, the respiratory chemoreceptors and mechanoreceptors participate in a protective network that adjusts the pattern of breathing and the mechanical conditions of the airways to minimize the deleterious effects on the lung of inhaled, noxious material.
Ventilation, unlike blood pressure and cardiac output, can be controlled consciously (voluntarily) as well as automatically (involuntarily). Indeed, the pathways for voluntary and automatic control of the respiratory muscles are anatomically separate. Voluntary as well as automatic control is essential for using the respiratory muscles in speech. In humans, afferent information continuously fed back to the CNS by mechanoreceptors in the airways, lungs, and chest wall allows the force of contraction of the respiratory muscles to be coordinated smoothly in volitional acts.
Besides inputs from respiratory system sensors, ventilation is influenced by projections from the vasomotor neurons to respiratory neurons and by signals received from thermoreceptors and vascular receptors. The multiplicity of inputs to the respiratory neurons ensures that ventilation is maintained when disease affects one or more afferent pathways or when the perception of some sensory cue is blunted by a depressed state of consciousness (e.g., sleep or anesthesia). However, conflicting demands and signals from different receptors may be responsible for dyspnea, a common symptom in respiratory disease.