The idea of a goal-directed arousal system in the brain (Chapter 2) implies the existence of some mechanism for selecting appropriate goals, for initiating the behaviours required to achieve them, and for signalling when they have been attained. If a goal proves favourable for survival in the prevailing circumstances, it is advantageous to reinforce behaviour leading to it; if the goal proves to be unfavourable, behaviour leading to it must be suppressed and avoidance action taken in future. Such a signalling system may be provided by certain ‘reward’ and ‘punishment’ pathways in the brain discovered by Olds and Milner (1954). These are closely integrated with arousal systems and with learning and memory, and appear to be fundamental for motivation, and for goal-seeking and avoidance behaviour. They are thought to form the basis for instinctive drives such as hunger, thirst, and sex, and are probably the substrate of more complex emotional/cognitive states such as hope and disappointment.

Olds and Milner (1954), and Olds (1956, 1977) reported a series of experiments showing that rats will work to obtain electrical stimulation through electrodes implanted at specific sites in the brain. When allowed to stimulate themselves by pressing a lever, they would sometimes do so at the rate of over 100 times a minute for hours on end. The animals appeared to like the stimulation and it seemed that activity in certain parts of the brain generated a pleasurable or rewarding sensation. Furthermore, motivation for self-stimulation was so strong that the rats would learn to perform various tasks such as traversing complex mazes in order to obtain it. When food reward and the opportunity for intracranial self-stimulation were both restricted, animals preferred the stimulation, even at the cost of starvation (Routtenberg 1978). Such findings suggested that the effects of intracranial self-stimulation and food reward were similar and that rewarding stimulation sites were located in neural pathways subserving reinforcement of goal-directed behaviour in the natural state (Redgrave and Dean 1981).

The above type of self-stimulation behaviour was elicited most strikingly when the electrodes were implanted in the lateral hypothalamus, an area known to be related to feeding mechanisms. However, with some electrode placements responses are only obtained if the rats are motivated by sexual arousal rather than by hunger. If multiple sites for stimulation are provided, there is a tendency to alternate the self-stimulation from electrode to electrode and preferred sites may change depending on whether the animals are hungry or thirsty (Olds 1977; Redgrave and Dean 1981; Routtenberg 1978).

Sites which support intracranial self-stimulation have been found in all vertebrate species studied, including man, and patients describe different sensations according to electrode locations (Heath 1964; Redgrave and Dean 1981). These results suggest that there may be a complex of rewarding pathways in the brain subserving different types of reinforcement behaviour.

Anatomical locations from which various types of self-stimulation behaviour can be obtained have been mapped in many investigations reviewed by Redgrave and Dean (1981) and Routtenberg (1978). Table 5.1 and Fig. 5.1 show many of the large number of sites which have been implicated. It is clear that self-stimulation sites are distributed widely in the brain, from the frontal lobes to the medulla, and that they include areas of very different function, from sensory processing to motor activity. Self-stimulation is also supported from sites in the fibre tracts connecting many of these areas, notably the median forebrain bundle. This runs in the lateral hypothalamus and carries ascending projections from the brainstem nuclei, including the locus coeruleus and raphe nuclei, to the diencephalon and telencephalon, as well as descending projections from the median forebrain. Routtenberg and Santos-Anderson (1977) suggest that the prefrontal cortex is vital to the intracranial self-stimulation system, and point out that it is the origin of fibre tracts which run through many self-stimulation loci throughout the neuraxis and that these pathways intermingle with the median forebrain bundle at the level of the hypothalamus. At least five well-established areas for self-stimulation lie in the path of the frontal cortex descending fibre system (Fig. 5.1).

Forebrain

frontal cortex; olfactory nucleus; nucleus accumbens; septal area; amygdaloid nucleus; hypothalamus

entorhinal cortex; caudate nucleus; entopeduncular nucleus; hippocampus; ventral and medial thalamus; median forebrain bundle; dorsal noradrenergic bundle

Midbrain and brainstem

ventral tegmental area; raphe nuclei; superior cerebellar peduncle; mesencephalic nucleus of trigeminal nerve

substantia nigra; nucleus coeruleus; periaqueductal grey matter

Cerebellum Medulla

deep cerebellar nuclei motor nucleus of trigeminal nerve; nucleus of tractus solitarius

other cerebellar areas

Forebrain

frontal cortex; olfactory nucleus; nucleus accumbens; septal area; amygdaloid nucleus; hypothalamus

entorhinal cortex; caudate nucleus; entopeduncular nucleus; hippocampus; ventral and medial thalamus; median forebrain bundle; dorsal noradrenergic bundle

Midbrain and brainstem

ventral tegmental area; raphe nuclei; superior cerebellar peduncle; mesencephalic nucleus of trigeminal nerve

substantia nigra; nucleus coeruleus; periaqueductal grey matter

Cerebellum Medulla

deep cerebellar nuclei motor nucleus of trigeminal nerve; nucleus of tractus solitarius

other cerebellar areas

Pathways of reward in the rat brain. Circles: location of cell bodies; rectangles: regions where reliable self-stimulation is obtained. (From Routtenberg 1978, by kind permission of W.H. Freeman and Co.)

Most, if not all, of the sites which support self-stimulation have anatomical connections with limbic structures, where the emotional, autonomic, and motor responses appropriate to reward may be generated. Many of these same pathways are also involved in arousal and in learning and memory systems. The discovery of brain rewarding areas seems to hold the key for understanding the normal processes of motivation, reinforcement, and learning. Nevertheless, intracranial self-stimulation itself remains an enigma: the exact neural circuitry involved, and the relation of this strange activity to physiological behaviour is far from clear.

At first it seemed likely that noradrenaline was crucially involved in brain stimulation reward. Sites which elicit self-stimulation coincide with histological maps of noradrenergic nerve distribution (Stein et al. 1977; Stein 1978; Fig 2.7a). In particular, a dorsal noradrenergic pathway originates in the locus coeruleus and innervates the neocortex, cerebellum, hippocampus, and thalamus. A ventral pathway originates from noradrenaline-containing cells in the medulla and pons and innervates the hypothalamus and ventral parts of the limbic system. A periventricular pathway originates from various noradrenergic cell bodies and innervates the median regions of the thalamus and hypothalamus. All these pathways, and the locus coeruleus itself, support self-stimulation, and electrical stimulation of rewarding areas in the median forebrain bundle results in an increased liberation of noradrenaline and its metabolites from the lateral hypothalamus, while stimulation of neutral areas does not (Stein and Wise 1969). However, Routtenberg (1978) and Routtenberg and Santos-Anderson (1977) showed that complete destruction of the locus coeruleus had little effect on the rate of self-stimulation from an electrode in the dorsal brainstem in rats. At present the status of noradrenaline is uncertain: it seems to play a part in self-stimulation reward but it does not appear to be the only transmitter involved.

Although there are self-stimulation sites in the brain which are not near a dopaminergic system (Wise 1980), it seems likely that dopamine is an important transmitter in some reward pathways. There are dopaminergic fibres in the median forebrain bundle. These include the nigrostriatal pathway from substantia nigra to caudate nucleus and the mesolimbic pathway from the ventral tegmental area to the nucleus accumbens, olfactory tubercle, septal area, and frontal cortex (Redgrave and Dean 1981; Fig 2.7b). Neuronal mapping and stimulation studies of ventral tegmental and substantia nigra areas seem to indicate that self-stimulation is uniquely associated with dopamine containing cells (Wise 1980).

Pharmacological evidence supports the involvement of dopamine in reward systems. Drugs which inhibit dopamine synthesis or block dopamine receptors disrupt intracranial self-stimulation from a number of sites, even at the locus coeruleus, a predominantly noradrenergic structure. When dopamine blockade is limited to one hemisphere, self-stimulation responses are suppressed for that hemisphere but not for the other. Dopamine blockade also raises the electrical threshold for self-stimulation in a dose-dependent manner. Drugs which increase dopaminergic activity (amphetamines, cocaine) can increase rates of electrical self-stimulation, are avidly self-administered by animals, and are drugs of abuse in man (Chapters 6, 7). Stein and Belluzi (1987, 1988, 1989) demonstrated that the firing rate of individual neurones in the CA, area of isolated rat hippocampal brain slices could be modified by microinjection of dopamine or cocaine. The preparation was arranged so that increased rates of firing could trigger a self-microinjection of dopamine or cocaine into individual neurones. When delivered non-contingently these drugs had little effect on the firing rate of such cells, but after suitable priming, cells responded by increasing their firing rates contingently to trigger microinjections of cocaine or dopamine, but not placebo or opioid injections. Thus the firing of individual, isolated, cells could apparently be reinforced by dopaminergic drugs. It seems that even at the cellular level animals are programmed to seek reward. Other individual neurones in the CA3 hippocampal fields responded similarly to the endogenous opioid dynorphin A (see below) but not to dopaminergic drugs. In parallel experiments in intact rats, dopamine and cocaine reinforced self-administration when injected into the same CA1hippocampal field but not into other hippocampal areas, while dynorphin A reinforced self-administration specifically by microinjection into the CA3 hippocampal field.

These results led Stein (1989) to suggest that the functional unit of reward is a population of individual neurones (‘hedonistic neurones’) scattered around reward areas of the brain, which are responsive specifically to dopamine or opioids, and which are presumably connected to pathways controlling motivated behaviour. Phillips et al. (1989) and Phillips and Fibiger (1989) showed that there is an increase in dopamine metabolism, synthesis and release in the ventral tegmental area and nucleus accumbens during brain self-stimulation behaviour in rats, and that this increase is proportional to the stimulation rate and intensity. Dopaminergic pathways also appear to be involved in food rewards, both anticipation and ingestion (Blackburn et al. 1989; Royall and Klemm 1981) and to precopulatory and copulatory behaviour in male rats. Dopamine thus appears to be closely linked both with intracranial self-stimulation and with natural rewards, and dopamine release itself may be rewarding (Wise 1980). However, the increase in self-stimulation behaviour induced by dopaminergic drugs (amphetamine, cocaine) is blocked by naloxone (Schaefer and Michael 1990), suggesting that opioid mechanisms are also involved in dopaminergic reward functions.

Both noradrenaline and dopamine have general effects in heightening arousal and increasing goal-seeking behaviour. However, reward can also be identified with reduction of arousal when the goal is achieved and satisfaction results. Stein (1978) suggested that the latter aspect of reward may be mediated by enkephalin or a related opioid peptide. In line with this suggestion is the observation that pharmacological agents which are apparently rewarding include not only stimulants such as amphetamine and cocaine, which release catecholamines, but also depressants such as morphine which act on endogenous opioid receptors.

In many brain areas the distribution of cell bodies containing endogenous opioids and of opioid binding sites overlaps very closely with that of catecholamine-containing cell areas, and rewarding sites (including the amygdala, locus coeruleus, pontine central grey, zona compacta of the substantia nigra, bed nucleus of the stria terminalis, and nucleus accumbens) contain beta-endorphin and other polypeptides as well as catecholamines (Elde et al. 1976; German and Bowden 1974). Thus, it seems likely that stimulation of the same rewarding areas releases both classes of neurotransmitters or modulators, and that both play an essential part in reward mechanisms. Stein (1978) reports experiments in which rats were found to work for injections of various opiates and opioids, including morphine and enkephalin, directly into the cerebral ventricles. The response was blocked by the specific opioid antagonist naloxone and also by noradrenaline depletion. Cooper (1984) notes that rats will also work to self-inject morphine specifically into the ventral tegmental area, and that this behaviour can be blocked by haloperidol, a dopamine receptor antagonist, or by 6-hydroxydopamine lesions of ascending dopaminergic pathways. Opioid-based reinforcement may therefore operate through catecholaminergic links, a possibility supported by Wise and Bozarth (1987). However, as described above, Stein (1989) found that the endogenous opioid dynorphin A (Table 5.2) could act directly as a reinforcer to isolated cells in the CA3 hippocampal area, and it seems likely that opioid release may itself be reinforcing.

Precursor

pro-enkephalin

pro-dynorphin

pro-opiomelanocortin

Peptides

(met) enkephalin (leu) enkephalin peptide E

beta-neoendorphin dynorphin A dynorphin B

beta-endorphin

Precursor

pro-enkephalin

pro-dynorphin

pro-opiomelanocortin

Peptides

(met) enkephalin (leu) enkephalin peptide E

beta-neoendorphin dynorphin A dynorphin B

beta-endorphin

Endogenous agonists*

beta-endorphin dynorphin A enkephalins

Enkephalins beta-endorphin dynorphin A

dynorphin A beta-endorphin enkephalins

?

Agonist drugs

Narcotic analgesics

DPDPE

Pentazocine ketacozine

Phencyclidine cyclazocine

Antagonist

naloxone

Naloxone (less potent than at mu receptors)

Naloxone (less potent than at mu receptors)

Naxolone (less potent than at mu receptors)

Effector pathways

cAMP ↓ K+ channel activation

cAMP ↓ K+ channel activation

Ca2+ channel inhibition

?

Some biological effects

analgesia, euphoria, respiratory depression, cough suppression, miosis, endocrine effects

analgesia, euphoria, motor functions, endocrine effects

analgesia (spinal), sedation, miosis

dysphoria, hallucinations, respiratory and motor stimulation, mydriasis

Endogenous agonists*

beta-endorphin dynorphin A enkephalins

Enkephalins beta-endorphin dynorphin A

dynorphin A beta-endorphin enkephalins

?

Agonist drugs

Narcotic analgesics

DPDPE

Pentazocine ketacozine

Phencyclidine cyclazocine

Antagonist

naloxone

Naloxone (less potent than at mu receptors)

Naloxone (less potent than at mu receptors)

Naxolone (less potent than at mu receptors)

Effector pathways

cAMP ↓ K+ channel activation

cAMP ↓ K+ channel activation

Ca2+ channel inhibition

?

Some biological effects

analgesia, euphoria, respiratory depression, cough suppression, miosis, endocrine effects

analgesia, euphoria, motor functions, endocrine effects

analgesia (spinal), sedation, miosis

dysphoria, hallucinations, respiratory and motor stimulation, mydriasis

listed in order of potency

DPDPE: D-Penicillamine, D-Penicillamine enkephalin (experimental drug)

References: Hughes and Kosterlitz (1983); Morley (1983); Atweh and Kuhar (1983); Jaffe and Martin (1980); Trends in Pharmacol. Sci. (1990).

It seems reasonable to conclude that reward processes are regulated by the closely related joint actions of dopamine, possibly noradrenaline, and endogenous opioids, and Stein (1978) suggests that a normal action of endogenous opioids is to bring successful reward-seeking behaviour to a satisfying termination. A particular role for endogenous opioids in mediating social reward has been postulated by Panksepp (1981) and is discussed in Chapter 11.

Activity at certain sites in the brain appears to generate sensations that are strongly aversive; animals will work as avidly to avoid stimulation at these sites as they will to obtain stimulation at rewarding points (Olds and Olds 1963; Delgado et al. 1954). A major anatomical pathway subserving aversive effects appears to be the periventricular system, a group of fibres running between the midbrain and thalamus with extensions into the hypothalamus, basal ganglia, limbic system, and cerebral cortex (Stein 1968). The median forebrain bundle and periventricular system probably interact, since they distribute fibres to various common sites along their paths (Criswell and Levitt 1975). A further pathway subserving aversion may originate in the dorsal raphe nuclei and distribute to periventricular regions of the brain, but some fibres run in the median forebrain bundle and terminate in various parts of the limbic system (Stein and Wise 1974). Destruction of either of these pathways results in a generalized defect in passive avoidance, so that an animal will no longer suppress behaviour that precipitates an aversive stimulus, such as an electric shock to the feet. It has been suggested that these pathways act as a ‘punishment’ system favouring avoidance behaviour and also selecting the appropriate reward behaviour that will terminate a particular aversive state, for example, feeding in hunger, drinking in thirst, etc. (Stein 1971). The interaction between reward and punishment systems allows for many dimensions of reward and punishment. Activity in reward systems not only engenders active reward, but also inhibits activity in punishment systems. Conversely, activity in punishment pathways is positively aversive and also inhibits activity in reward pathways. Presumably, there can be innumerable degrees of partial inhibition or excitation in the different pathways, resulting in finely graded shades of reward/punishment activity. The arrangement appears to be similar to that described for the reciprocally connected arousal and sleep mechanisms in the brain (Chapter 2). Reward and punishment systems are also closely integrated with learning and memory systems (Chapter 8). Thus, lack of expected reward, as well as active punishment, is unpleasant; similarly, lack of expected punishment, as well as active pleasure, is rewarding. The hippocampal comparator system described in relation to anxiety neurosis (Chapter 3) is thought to be involved in forming the expectation of reward or punishment as a result of learning.

The periventricular system appears to be at least partly cholinergic (Stein 1968). The deficit of passive avoidance resulting from destruction of this system can be reversed by local instillation of cholinergic drugs, such as carbachol or acetylcholine, while the local application of anticholinergic drugs, such as atropine, produces similar effects to surgical destruction.

Cholinergic systems are also closely involved in learning and memory (Chapter 8), and it is likely that certain aspects of these functions which require suppression of behaviour are mediated through the periventricular system (Criswell and Levitt 1975). As already mentioned, the periventricular system communicates with the limbic circuit, and many limbic structures receive innervation from both the periventricular system and the median forebrain bundle. Thus, together the two systems can be envisaged to promote the seeking of reward and the suppression of punished behaviour.

Other neurotransmitters may also mediate aversive effects in periventricular structures. From experiments involving the local injection of various agonists and antagonists into the central grey and medial hypothalamus, Schmitt et al. (1984) conclude that GABA, excitatory amino acids, and opioids may all interact to modulate aversive reactions in the rat.

The fibres from the raphe nuclei which run in the median forebrain bundle are serotonergic (Fig 2.7c). Stein and Wise (1974) suggest that this system acts antagonistically to the median forebrain bundle reward system and that goal-directed behaviour is reciprocally regulated through noradrenergic and serotonergic pathways here. Thus, rewarding intracranial self-stimulation in the lateral hypothalamus is facilitated by the intraventricular injection of noradrenaline, but suppressed by serotonin. It is thought that the anxiolytic effects of benzodiazepines and other anxiolytic drugs are exerted at least partly by decreasing activity in serotonergic punishment pathways, and that activity in these pathways is increased in anxiety.

Reward and punishment systems are thought to be involved not only with certain types of behaviour, but also with their subjective accompaniments or mood. Thus, rewarding events may presumably elicit a range of pleasurable feelings (joy, contentment, hope, repletion), and aversive events a range of unpleasant feelings (pain, fear, disgust, guilt, depression). Omission of expected rewards or punishments may result in other emotions (disappointment, relief) (Stein et al. 1977). The various neurotransmitters presumably interact in complex ways and through different pathways to involve particular emotions, and individuals may differ in their sensitivity to reward and punishment (Gray 1981b).

A large component of the punishment mechanisms must be provided by the systems responsible for signalling pain and nociception. Indeed, pain and fear of pain are the strongest of punishing stimuli, and the same limbic structures involved in the reward and punishment systems described above also provide the neural basis for the aversive drive and affect that comprise the motivational dimensions of pain (Melzack 1986). Pain systems are reviewed by Yaksh and Hammond (1982), Fields (1987), and Melzack and Wall (1988).

Pain is a complex experience resulting from variable interactions between physical, emotional, and rational components. The physical component (sensory-discriminative) is supplied by the nociceptive system described below; the emotional component (motivational-affective) involves the limbic system, including the punishment pathways; the rational component (cognitive-evaluative) is derived from the cerebral cortex. Excitatory and inhibitory feedback systems link all components.

Two types of physical pain sensation are recognized: (a) first pain is sharp or pricking, rapid in onset and brief in duration, well localized, and can only be elicited from the skin, and (b) second pain is burning in character, delayed in onset (up to 1 s after the stimulus) but prolonged, poorly localized, and may be elicited from both skin and deep structures. The reflex response to first pain is a phasic withdrawal reaction, whilst that to second pain is a slowly developing tonic muscular contraction (guarding or rigidity). First pain is responsible for withdrawal reflexes and protects the body from injury; second pain probably only occurs in the presence of tissue damage and may serve to splint or rest the affected part. It is interesting that potent analgesic drugs such as morphine have little effect on first pain in subanaesthetic doses.

In clinical and experimental settings, it is important to distinguish between pain threshold and pain tolerance. The intensity of a stimulus required to be perceived as painful (threshold) is variable, but the intensity of a painful stimulus that a subject will tolerate is even more variable and depends on personality and on social, educational, cultural, and environmental factors. Analgesic drugs and procedures have differing effects on pain tolerance and pain threshold. A further important clinical distinction is that between acute pain and chronic pain (Chapter 6).

The basic anatomical organization of pain pathways is shown diagrammatically in Fig. 5.2.

Diagram of the organization of pain pathways. For explanation see text. (References: Thompson 1984a, b; Zimmerman 1981.)

Peripheral nociceptors are situated at the terminals of afferent neurones whose cell bodies lie in the dorsal root ganglia. These nociceptors are of two types: (a) high threshold mechanoreceptors connected to myelinated A delta axons which are relatively large diameter (6–30 μm) and fast-conducting (5–10 m/s), and (b) polymodal nociceptors, consisting of the bare terminals of C fibres which are unmyelinated, small diameter (0.25–1.5μm) and slow conducting (1–2.5 m/s) (Melzack and Wall 1988). These different nociceptors largely underlie first and second pain respectively, but they are not specific for pain. They respond to other stimuli (pressure, chemicals, and heat) at stimulus intensities well below those that evoke pain, and central processes determine the threshold, intensity and time course of pain (Wall and McMahon 1986).

The central axons of these sensory neurones enter the dorsal horns of the spinal cord where they give off branches which run up and down the cord for one or two segments. Many collaterals penetrate the substantia gelatinosa and terminate there on Golgi Type II cells. These cells also receive connections from adjacent spinal regions and from descending pathways from the brain. The inter neurones of the substantia gelatinosa are probably important sites for the gating mechanism of pain described below.

The axons of the substantia gelatinosa interneurones terminate in the chief nucleus of the dorsal horn and synapse with the second order afferent neurones. These cross to the opposite side of the spinal cord and ascend in the lateral spinothalamic tract. In the brainstem, the spinothalamic tracts from each side merge and pass through the brainstem as the spinal lemniscus before terminating in the ventral posterior nucleus of the thalamus. From here, third order sensory axons pass to discrete localized projections on the sensory cerebral cortex. This system appears to be primarly responsible for well-localized, sharp first pain.

Impulses in nociceptive fibres are also projected centrally through diffuse connections. Small spinal nerve root fibres connect, partly directly and partly through interneurones in the substantia gelatinosa, with cells in the dorsal horn of the spinal cord. These give rise to spinoreticular axons which ascend in crossed and uncrossed multisynaptic pathways, making numerous connections with the medullary reticular formation, to the intralaminar thalamic nuclei, the hypothalamus, and limbic areas including frontal cortex. From the thalamus, diffuse, non-specific projections pass to widespread areas of the cortex of both hemispheres, probably mediating the poorly localized second pain. Recent work suggests that the anterior cingulate cortex is critically involved in pain perception in man (Roland 1991). The intralaminar thalamic nuclei also give off more localized projections to the corpus striatum, which may be concerned with the reflex motor responses to painful stimuli. The hypothalamic projections connect with autonomic centres which supply the autonomic concomitants of pain, while the limbic connections are probably responsible for the emotional components. These spinoreticular pathways, with connections to reticular formation, limbic system and cortex, are also closely concerned in the arousal systems of the body (Chapter 2).

Inhibitory pathways descend from the periventricular and periaqueductal grey matter, and from the nucleus raphe magnus, the giant cell nucleus and other structures in the reticular formation, to the spinal cord. In the dorsal horn, these axons terminate on the endings of the first sensory neurones or on the connected interneurones in the substantia gelatinosa. These descending inhibitory pathways contribute to the spinal pain gate mechanisms.

The original gate-control theory of Melzack and Wall (1965) has been modified in the light of later experimental findings (Melzack and Wall 1988). However, the basic concept that, due to local inhibitory control mechanisms, not all nociceptive impulses which reach the spinal cord are transmitted to the brain, remains unchallenged. It appears that the propagation of nociceptive impulses can be inhibited at spinal cord level by at least three systems, all of which probably interact with short interneurones in the substantia gelatinosa (Fig. 5.3).

Gate-control mechanisms in pain regulation. SP, substance P; ENK, enkephalin; END, endorphin; NA, noradrenaline; 5-HT, serotonin. For explanation see text. (Reference: Thompson 1984a.)

The responses of dorsal horn neurones to peripheral nociceptive stimuli can be inhibited by simultaneous stimulation of large A fibre afferents. One mechanism for this inhibition is activation by the large A fibres of short substantia gelatinosa interneurones, which then presynaptically inhibit the dorsal horn spinothalamic cells. The effect of the presynaptic inhibition is to block the onward transmission of spinothalamic impulses resulting from nociceptor stimulation of dorsal horn cells, particularly those from small C fibre primary afferents. The degree to which potentially painful stimuli are propagated to higher centres thus depends on the relative proportions of small and large fibre activity. Stimulation of A fibres is thought to be an important mechanism of the analgesic action of transcutaneous electrical nerve stimulation, acupuncture, massage, and other counter-irritation procedures. As discussed below, substance P may be the excitatory transmitter from C fibre afferents to dorsal horn cells, and enkephalin may be the inhibitory transmitter liberated by the short interneurones in this system. There are in addition other mechanisms, including post-synaptic inhibition, for gating of pain by afferent stimuli, and for the excitatory and inhibitory effects of different spinal cord neurones (Melzack and Wall 1988).

Firing of dorsal horn cells in response to noxious stimuli can also be inhibited, and profound analgesia produced, by electrical stimulation of the periaqueductal grey matter in the brainstem or of the raphe nuclei in the medulla, to which it projects. Such stimulation is mediated by descending impulses from the medullary region impinging on dorsal horn cells or interneurones through pathways which are probably serotonergic but may involve an enkephalinergic link (Bowsher 1978a).

A further descending inhibitory system appears to originate from structures in the lateral reticular formation. Electrical stimulation of this area also produces deep analgesia and decreases the firing rate of dorsal horn cells in response to noxious stimuli. This system is believed to involve a catecholamine neurotransmitter, probably noradrenaline.

These pain gate control systems in the spinal cord provide another example of the way in which the nervous system controls its own sensory input, as discussed in relation to arousal (Chapter 2). In fact, the system is undoubtedly closely related to arousal systems, and pain sensation and responses are largely modulated by the degree and type of activity in arousal systems. For example, potentially painful stimuli may pass unnoticed during the excitement of sporting activities or may be greatly aggravated by fear of its consequences or by depression. Learning and memory are equally closely involved in pain, and fear of a previously experienced pain can generate similar responses to those evoked by the pain itself. Thus, limbic and cortical gate control mechanisms can add to those operating at spinal cord level.

Nociceptors are stimulated by mechanical and thermal stimuli and are also sensitive to a number of chemical agents. Certain types of clinical pain, notably that of inflammation, probably derive from chemical activation by endogenous algesic agents. These include serotonin, histamine, bradykinin, adenophosphate, potassium, prostaglandins, leukotrienes, acetylcholine, and substance P. They are released peripherally after tissue injury and either activate or sensitize primary afferents. Many of these substances are also present in the central nervous system where they may act as transmitters or modulators, but their central functions are not understood. Chemical mediators of pain at sensory nerve endings are reviewed by Bond (1979), Zimmermann (1981), and Terenius (1981).

Substance P is a undecapeptide which acts both as a neurotransmitter and a neuromodulator in various body systems; Oehme and Krivoy 1983; Otsuka and Yanagisawa 1987). Its major role in pain appears to be as an excitatory neurotransmitter at primary nociceptive nerve endings in the dorsal horn of the spinal cord. Substance P is present in small peripheral nerve fibres, their dorsal root ganglion cells, and in synaptic vesicles at their terminals in the substantia gelatinosa. It is released in a calcium-dependent manner in response to stimulation of nociceptive neurones but not after stimulation of large diameter afferents alone. When iontophoretically applied to the spinal cord or trigeminal nucleus caudalis, it excites only those neurones which respond to noxious stimuli. Intrathecal injection of substance P produces hyperalgesia, while depletion of spinal cord substance P (with capsaicin) or the application of substance P antagonists produces analgesia. Substance P produces slow excitatory postsynaptic potentials in dorsal horn neurones and it may be co-released from primary afferent nerve terminals with a fast-acting transmitter such as glutamate.

Under physiological conditions, the pre-or post-synaptic release of substance P may be controlled, at least partly, by the actions of short interneurones containing the endogenous opioid enkephalin. The distribution of enkephalin in the spinal cord and brain is similar to that of substance P, and enkephalinergic interneurones project onto the terminals of substance P-containing primary afferent fibres. Release of enkephalin from the interneurones is believed to reduce the release of substance P and the interaction between substance P-containing and enkephalinergic neurones is probably one of the mechanisms for the gate control of pain in the spinal cord.

In addition to substance P, other non-opioid polypeptides, including somatostatin, neurotensin, angiotensin II, vasoactive intestinal peptide and cholecystokinin, may be involved in pain modulation in the spinal cord and brain. The possible role of cholecystokinin is reviewed by Baber et al. (1989). This octapeptide is present in central areas associated with pain modulation, including cortex, periaqueductal grey, thalamus and spinal dorsal horn neurones. Small doses of cholecystokinin antagonize the analgesic effects of opioids while antagonists of cholecystokinin enhance opioid analgesia.

The supraspinal descending inhibitory systems which modulate nociceptive transmission in the spinal cord appear to be mediated by monoamine neurotransmitters. The analgesia produced by electrical stimulation of the periaqueductal grey and the medullary raphe nuclei is accompanied by release of serotonin in the spinal cord (Wilson and Yaksh 1980). This analgesia is blocked by agents which decrease serotonin synthesis and enhanced by agents which increase synaptic concentrations of serotonin.

A serotonergic link in the analgesia produced by opiates is suggested by the observation that the intrathecal administration of methysergide blocks the analgesic effect of morphine injected into the periaqueductal grey. The analgesia produced by periaqueductal stimulation is naloxone-reversible; tolerance and cross-tolerance to morphine develop after repeated stimulation (Lewis and Liebeskind 1983). Beta-endorphin may also contribute to this pain suppression system (Bolles and Fanselow 1982). Similar opioid and serotonergic pathways have been implicated in the anaesthesia produced by acupuncture (Han and Terenius 1982).

Electrical stimulation of the lateral reticular formation, on the other hand, produces an analgesic effect which appears to be mediated by noradrenaline. This analgesia is accompanied by release of noradrenaline in the spinal cord and is not affected by manipulation of serotonin concentrations (Zimmerman 1981). Direct application of noradrenaline and alpha-adrenergic stimulants to the spinal cord produces analgesia which is not affected by opioid or serotonin antagonists. Noradrenaline also appears to be involved in the analgesia produced by opiates (Yaksh 1982) and acupuncture (Han and Terenius 1982). Noradrenergic and serotonergic systems probably interact to produce optimum analgesia and may also control the activity of a third inhibitory pathway which is dopaminergic, descending from the substantia nigra to the spinal cord (Fitzgerald 1986).

There is increasing evidence (reviewed by Hartvig et al. 1989) that cholinergic (muscarinic) mechanisms are involved in pain pathways. Autoradiographic studies have revealed the existence of muscarinic receptors in the substantia gelatinosa of the spinal cord and of cholinergic interneurones on the dorsal horn. Cholinergic neurones are thought to interact with enkephalinergic, noradrenergic and serotonergic neurones, which all have terminals in the same areas. Muscarinic agonists have been shown to have antinociceptive effects when administered intrathecally in the rat. Nicotine receptor agonists have little antinociceptive activity. Cholinergic antagonists attenuate the analgesic effects of intrathecally administered morphine, suggesting that in the spinal cord opioids act through a cholinergic link. In preliminary clinical studies, cholinomimetic drugs such as physostigmine and THA, administered parenterally or orally, appear to have analgesic effects and to potentiate morphine analgesia.

There appears to be a complex interaction between GABA and excitatory amino acids (and other transmitters) controlling nociceptive transmission in the spinal cord and brain. GABA activity in the dorsal horn of the spinal cord inhibits onward transmission of afferent nociceptive impulses (Duggan and Foong 1985) while in the medullary reticular formation and periaqueductal grey GABA modulates activity in descending monoaminergic pain suppressing pathways (Lovick 1987; Behdehami et al  1990). These effects appear to be mediated by GABAA receptors since they are blocked by bicuculline.

Glutamate and other excitatory amino acids are present in C fibres and their terminals in the dorsal horn and co-exist in most substance P-containing fibres. Stimulation of peripheral nociceptive afferents probably results in the co-release of both peptides and excitatory amino acids in the spinal cord. Glutamate appears to be involved in plastic neuronal changes following repetitive nociceptive stimulation (Dickenson 1990; Fitzgerald 1990). In particular, activation of NMDA receptors (Table 8.1) produces long-lasting increase in the firing rate of dorsal horn cells, amplifying, enhancing and prolonging the initial nociceptive discharge. This phenomenon is akin to hippocampal long-term potentiation (Chapter 8; Fig. 8.3) and may be important in prolonged nociceptive hypersensitivity states and in chronic pain syndromes (Chapter 6).

Endogenous opioids are reviewed in the British Medical Bulletin (1983), and much of the information is encapsulated briefly by Thompson (1984b). The term opioid refers to directly acting compounds whose actions are specifically antagonized by naloxone. Opiates are products derived from opium and the term is generally applied to morphine derivatives. Narcotic analgesics (Chapter 7) are agents which act on opioid receptors to produce naloxone-reversible analgesia.

Of the endogenous opioid polypeptides, three classes appear to be of major physiological importance: enkephalins derived from the precursor pro-enkephalin; dynorphins derived from prodynorphin; and endorphins’, derived from pro-opiomelanocortin (Table 5.2). These opioids are closely related structurally but differ in the length of their peptide chains. They function as short-acting neurotransmitters (shorter-length: enkephalins) or long-acting neuronal or hormonal modulators (longer-length: beta-endorphin, dynorphin) at their respective opioid receptors. Many of them interact with more than one receptor. There are several other endogenous opioids, including peptide E, alpha- and beta-neoendorphins, and others, whose function is less clearly understood.

There appear to be several distinct opioid receptor subtypes. These have different pharmacological profiles, tissue distributions, and binding properties, and probably mediate different though overlapping actions (Table 5.2). Mid-receptors are the main sites of action of the narcotic analgesics; naloxone and nalorphine are antagonists. Delta-receptors are activated by enkephalins, which have a greater affinity for them than for mu-receptors, and by beta-endorphin which has equal agonist activity at mu- and delta-receptors. Naloxone has less antagonist activity at delta than at mu-receptors. It is possible that mu-and delta-receptors represent high and low affinity states of a unitary receptor (Atweh and Kuhar 1983) and Pasternak (1987) suggests that there are at least two subtypes of mu receptors. Kappa-receptors respond to dynorphin and also to the synthetic analgesics ketazocine and pentazocine. These receptors appear to be involved in spinal analgesia. It has been suggested that mu-, delta- and kappa-receptors are interchangeable forms of a single opioid receptor complex (Barnard and Demoliou-Mason 1983). Sigma-receptors are probably not involved in analgesia, but may mediate some adverse effects of opioids. Agonists for these receptors include cyclazocine and phencyclidine (Chapter 7), and that activation of these receptors may account for the hallucinogenic properties of these drugs.

Like other receptors, the opioid receptors consist of a recognition site, to which drugs bind, and a translating mechanism which ultimately produces the biological response. The response is generally one of cellular inhibition which is achieved by membrane hyperpolarization due to opening of potassium channels, and by depression of transmitter release (Henderson 1983; North and Williams 1983). The endogenous opioid system as a whole appears to operate in the body as a widespread and complex inhibitory signalling mechanism in which selectivity is achieved by particular combinations of opioid peptides and receptors (Thompson l984a, b).

The distribution of opioid peptides and receptors in the brain is shown in Table 5.3. The distribution of the peptides differs somewhat from that of the receptors, and the distribution of each different opioid peptide is distinct. Within the central nervous system, opioid peptides and their associated receptors are found most often in association with sensory, limbic and neuroendocrine systems. Enkephalinergic systems consist mainly of short neurones diffusely distributed; dynorphin systems have longer neurones, also widespread; endorphins are largely found in endocrine cells, but some neurones project as far down as the spinal cord. Enkephalins are co-stored with catecholamines in chromaffin tissue, while dynorphins are co-stored with vasopressin and endorphins are co-synthesized with corticotrophin in the hypothalamus and pituitary.

Methionine enkephalin

corpus striatum, caudate putamen, globus pallidus limbic system: olfactory bulb, tubercle, septum, nucleus accumbens, hippocampus hypothalamus medulla, pons, spinal cord: periaqueductal grey, substantia gelatinosa.

Beta-endorphin

hypothalamus, pituitary

Dynorphin

hypothalamus

Distribution of opioid receptors (mu, delta and kappa)

Methionine enkephalin

corpus striatum, caudate putamen, globus pallidus limbic system: olfactory bulb, tubercle, septum, nucleus accumbens, hippocampus hypothalamus medulla, pons, spinal cord: periaqueductal grey, substantia gelatinosa.

Beta-endorphin

hypothalamus, pituitary

Dynorphin

hypothalamus

Distribution of opioid receptors (mu, delta and kappa)

Thalamic nuclei Periaqueductal grey Spinal cord-substantia gelatinosa, spinal nucleus trigeminal nerve, dorsal horns

Analgesia supraspinal spinal

Cortex Limbic structures—hippocampus, thalamic nuclei, nucleus accumbens, amygdala

Behavioural and mood effects Affective and cognitive components of pain

Anterior and posterior pituitary Hypothalamus

Endocrine effects stimulation (growth hormone, ACTH, prolactin) inhibition (leuteinising hormone vasopressin, oxytocin) Thermoregulation

Brainstem nuclei

Autonomic effects cough suppression, hypotension, respiratory depression, vomiting, miosis

Striatum

Locomotor behaviour

Ventral tegmentum

Appetite modulation, feeding behaviour

Thalamic nuclei Periaqueductal grey Spinal cord-substantia gelatinosa, spinal nucleus trigeminal nerve, dorsal horns

Analgesia supraspinal spinal

Cortex Limbic structures—hippocampus, thalamic nuclei, nucleus accumbens, amygdala

Behavioural and mood effects Affective and cognitive components of pain

Anterior and posterior pituitary Hypothalamus

Endocrine effects stimulation (growth hormone, ACTH, prolactin) inhibition (leuteinising hormone vasopressin, oxytocin) Thermoregulation

Brainstem nuclei

Autonomic effects cough suppression, hypotension, respiratory depression, vomiting, miosis

Striatum

Locomotor behaviour

Ventral tegmentum

Appetite modulation, feeding behaviour

References: Cuello (1983); Atweh and Kuhar (1983); Khachaturian et al. (1985); Mansour et al. (1988)

Nociceptive systems. Opioid receptors and peptides are closely associated with systems subserving pain sensation at several levels and are of physiological importance in pain modulation. Enkephalins appear to control the responses of dorsal horn neurones; they may also modulate pain at higher sites in the central nervous system, but they are rapidly destroyed in vivo by enkephalinases and their analgesic effects are shortlived. Beta-endorphin is less rapidly degraded in the body and has more enduring analgesic effects. The role of dynorphin in pain modulation is still not clear.

While enkephalinergic systems in the spinal cord and elsewhere may be tonically active in pain modulation, the beta-endorphin system and further enkephalinergic activity appear to be triggered into action by noxious stimuli and other stresses. Thus, naloxone does not usually cause hyperalgesia unless the subject already has pain or has been subjected to prolonged pain (Buchsbaum et al. 1983). A variety of stresses appear to induce the release of endogenous opioids. These include electric foot-shock in rats (Bowsher 1978a), pregnancy (Sicuteri 1981), various types of severe pain (Sicuteri 1981), depressive disorders (Terenius 1982), and endotoxic, haemorrhagic, and spinal shock (Holaday and Faden 1982). In these conditions beta-endorphin may be co-released with ACTH, and enkephalins with adrenaline from the adrenal medulla and noradrenaline from peripheral nerves (Hughes 1983), as part of the general reaction to stress. Endorphins may also be involved in placebo analgesia (Fields and Levine 1984). The responsivity of an individual’s endogenous opioid systems may affect his vulnerability or sensitivity to pain and stress.

Limbic system. Very high concentrations of opioid receptors are found in the amygdala and in the corpus striatum, especially the globus pallidus. Delta-receptors are located on presynaptic dopaminergic terminals in the corpus striatum and their action is to inhibit dopamine release. Mu-receptors are found in the same areas, but are not associated with dopaminergic terminals. Opioid receptors and peptides are also present in the hippocampus and cortex although in relatively low concentrations. In the cortex they tend to be distributed in polysensory association areas rather than in primary sensory cortex.

The opioid systems in limbic areas may play a role in mood and behaviour (Koob and Bloom 1983). Opioid receptors are found in rewarding areas (Chapter 5), and opioid agonists, such as morphine, are potent reinforcers in animals and support intracranial self-stimulation. The role of opioids in reward systems has been discussed above; their possible role in affective disorders is considered in Chapter 11. Limbic opioid systems are probably also involved in modulating the emotional components of pain, especially during arousal and stress, and they may also be involved in memory (Chapter 8). Thus, they appear to be intimately concerned in ‘the whole pleasure-pain modality’ (Bolles and Fanselow 1982, p. 26).

Endocrine and autonomic areas. Endogenous opioids also have important modulating actions in the endocrine system, especially on pituitary, hypothalamic, and associated autonomic functions. These actions are closely integrated with pain modulation and limbic activity, especially under stress, and may be involved in the endocrine abnormalities found in depression (Chapter 11).

It is clear that, like other complex functional systems, the nociceptive system utilizes a multitude of chemical mediators which together modulate nociceptive information at all levels from peripheral nociceptor to cerebral cortex to produce the final sensation of pain. The interactions of this apparent plethora of transmitters and modulators seem particularly intricate. However, of all the sensations pain is perhaps the one most immediately important for survival and is of fundamental importance in arousal, reward and punishment, and learning and memory. It is possible that several overlapping back-up systems have developed during the course of evolution. Of particular interest is the existence of multiple neurochemically and anatomically discrete pain suppressive systems. It appears that the physiological trigger for some of these systems is stress and that they represent an adaptation to certain emergency conditions in which pain suppression favours optimal coping behaviour. The particular pain suppression system activated appears to depend on the type of stress. (Lewis and Liebeskind 1983; Frenk et al. 1988).

Despite the elaborate organization of pain suppression systems, malfunction appears to occur in some chronic pain syndromes described in Chapter 6 and pain perception is altered in a number of psychiatric conditions, such as depression, anxiety, and schizophrenia.