The Anatomy and Physiology of Pain

This web-page assumes a basic knowledge of medicine. We assume that the reader is familiar with such basic concepts as interneurones, synapses, oncogenes, G proteins and prostaglandins, and has had previous sketchy exposure to neuroanatomy. We will try to briefly describe the neuroanatomy of pain. Once we have described the main structures transmitting pain, we will explore the connections between them, neurotransmitters mediating pain, and what happens within nerve cells carrying painful stimuli. We conclude with a short note on pain pharmacology.

What is pain?

We attach different attributes to pain. If a child injures herself, she will cry, and say "It's sore". Her mother might enquire "Where does it hurt, darling?" Think of these as two different approaches to pain - the "motivational-affective" component which is phylogenetically primitive, and concerned with pain as something nasty to be avoided, and the evolutionarily more recently acquired "sensory/discriminative" ability to perceive exactly where the pain is, and respond appropriately. As we explore the complex structures that mediate our appreciation of and response to pain, we will find that we can conveniently categorize them into two groups, those that deal with the response to pain as an unpleasant sensation, and those that are more concerned with sensory/discriminative aspects of pain. We will consistently colour code the pathways concerned with these two 'types' of pain.

Pain in the Cortex

It used to be said that cortical structures are only tangentially involved in the perception of pain, if at all. This is clearly extremely silly, as a host of connections link higher cortical structures with pain-centred nuclei in the thalamus and brainstem. Major cortical players are:

The next three pictures clearly show their location - S I is concerned with localisation of pain, while the other three structures are thought to be concerned with the motivational-affective aspects which from now on for the sake of brevity we will call "affective" pain, although "sore" pain might be a better term!

The human brain seen from the side, showing the SI and SII sensory cortex. SI is a thin strip made up of Brodman areas 3,1 and 2 posterior to the central sulcus, while SII lurks just above the lateral sulcus.

The insula

The temporal lobe has been retracted to reveal the insula in all its splendour
- the anterior portion is probably concerned with pain perception.

Medial view of hemisphere, showing cingulate gyrus

Next we look at the medial aspect of a hemisphere, showing the cingulate gyrus.
The anterior part of the cingulate gyrus is important in the perception of 'affective' pain.

The Thalamus

The thalamus is the 'central switching station' of the brain. Several of its multiple nuclei are concerned with pain. The lateral nuclei deal mainly with sensory/discriminative aspects, the medial ones with 'affective' pain. Because the thalamus is such a complex three-dimensional structure it's extremely difficult to visualise it. We therefore adopt the approach of "building it" from medial to lateral (successively adding nuclei) in the following diagram (follow the arrows):

Medially we add the poorly-defined midline nuclei. Lateral to these we plonk on the dorsomedial nucleus (dm) and, more anteriorly the anterior nuclei (ant).

The internal medullary lamina bounds the dorsomedial nucleus laterally, and separates it from the anterior nuclei. Within the internal medullary lamina are the intralaminar nuclei, including the centromedian (cm) and centrolateral nuclei.

Lateral to the internal medullary lamina (iml) lies the ventral posteromedial nucleus (vpm) and anterior to this is the ventral anterior nucleus (va).

Even further laterally we have the lateral dorsal(ld), lateral posterior(lp), ventral lateral(vl) and ventral posterolateral nuclei. The pulvinar (grey) is situated posteriorly.

Diagram of the thalamic nuclei, building from medial to lateral

A Diagram of the Thalamic nuclei. The lateral thalamus is thought to be mainly concerned with 'discriminative' pain, the medial with "affect/motivation".

Although difficult to identify, and glossed over in most neuroanatomy texts, these nuclei, particularly the submedius, may turn out to be very important in pain perception!

Midbrain

There is a host of pain-related structures in the midbrain. Most of this circuitry is involved in 'affective' pain, with extensive connections to the reticular system of the brainstem. Important components are:

Rostral midbrain, with the anterior pretectal nucleus, interstitial nucleus of Cajal, and nucleus of Darkschewitsch in red.

Picture of Caudal midbrain

A more caudal section through the midbrain, showing the superior colliculus (SC), periaqueductal grey matter (PAG), and red nucleus (RN).

The Pons

The most important pain-related nucleus in the pons is probably the locus coeruleus. This is jam-packed with noradrenaline-containing neurones, and projects to a variety of brainstem structures that modulate pain through pathways that descend to the spinal cord. Another notable group is the parabrachial nuclei, which receive a vast number of ascending spinoreticular fibres. Both are shown in the following cross-section through the pons.

The Medulla

This too is involved in the motivational/affective aspects of pain. Important cell groups are the nucleus gigantocellularis and related nuclei, the lateral reticular nucleus, and a variety of other nuclei. Some of these are shown in the following diagram.

Picture of a section through the rostral medulla

The rostral medulla, with the nucleus gigantocellularis and lateral reticular nucleus clearly shown.
Take note of the raphe nuclei near the midline, which are important in the descending pathways that suppress pain.

How do these nuclei tie together?

Connections between the above structures are complex and confusing, but can be divided into several groups. We give you four:

Before we look at the connections in more detail, let's look at pain connections in the spinal cord:

The Spinal Cord

Traditionally it was thought that most pain fibres entered the dorsal root of the spinal cord (the "sensory" root) and then synapsed in the dorsal part of the spinal grey matter, before passing the message up through the spinothalamic tract. It is now known that this is a gross oversimplification. In fact, anything up to 40% of sensory fibres enter in the ventral root!

Histologically the grey matter of the spinal cord is divided into ten 'laminae'. The dorsal part is divided into five laminae (I to V), components of which deal with most incoming pain fibres. VII is in between these laminae and the more ventral laminae VIII and IX, and X refers to the grey matter around the central canal of the spinal cord. And what about lamina VI? This is only discernible in the bulges in the cord related to where the innervation of the limbs originates.

Much enthusiasm was generated when the "gate control" theory of pain was first described. Although this mechanism has now been well documented, and has some clinical utility, it is known to be a gross over-simplification. The basic idea is that incoming pain stimuli can be "gated" (shut off) by other stimuli, because many nerve cells talk to one another in the dorsal horn. Important fibres coming from the periphery into the dorsal horn include:

Unpleasant stimuli entering via the C fibres can be suppressed by concurrent stimulation of A delta fibres (high amplitude low frequency stimulation, for example by acupuncture) or by impulses passing through A beta fibres. Examples of the latter include TENS (transcutaneous electrical nerve stimulation) and the simple expedient of rubbing the skin, which is well known by mothers to decrease perception of pain!

Ascending Pain Connections

We now have sufficient resources to examine the various ascending pain pathways. First, the primitive spino-reticulo-diencephalic connections:

Pathways from the spinal cord to the brainstem, and from there to the thalamus (diencephalon).
Some fibres pass directly to the medial thalamus, while others end in (or send collaterals to) a variety of nuclei in the brainstem. Connections between the reticular system and thalamus are not shown for reasons of clarity.

Next, we examine the phylogenetically more modern pathways from cord to lateral thalamus and thence to the S I cortex. These pathways are discriminative pain pathways, and have little to do with perception of pain as a 'sore' stimulus! These pathways have few or no opioid receptors - morphine (for example) will have no effect on such pathways!

Descending Pain Connections

As important as the ascending pathways are fibres that descend from brainstem to spinal cord to modulate the incoming signals. Notable neurotransmitters mediating this anti-nociceptive effect include noradrenaline (norepinephrine), especially in the locus coeruleus, and serotonin in the raphe nuclei. Opioid receptors are prevalent here. Some descending connections are:

Descending connections that modulate incoming pain impulses.

Incoming painful stimuli are transmitted (A) to the dorsal horn, and from there (B) to the periaqueductal grey (PAG). Descending impulses pass (C) to the raphe nuclei, especially the nucleus raphe magnus, in the upper medulla, and thence back to the dorsal horn via reticulospinal fibres (D).

The above shows only the serotonergic descending fibres. Other pain-suppressing impulses pass from the PAG to the locus coeruleus, and from there to the dorsal horn.

Pain in the periphery - the nociceptor

The above connections are awfully complex. One might think that once we moved out to the periphery, things might become more simple. Not so! Most tissues are well provided with specific pain receptors called nociceptors. Formerly it was thought that painful stimuli were detected through 'overstimulation' of receptors for other modalities. This is incorrect. The quality of the pain perceived on stimulation of nociceptors seems to depend on the site of stimulation, and the nature of the fibres transmitting the sensation. Even in the periphery, there is a distinction between the sharp immediate pain ("first pain") transmitted by A delta fibres, and the prolonged unpleasant burning pain mediated through the smaller unmyelinated C fibres.

Nociceptors have numerous different receptors on their surfaces that modulate their sensitivity to stimulation. These include GABA, opiate, bradykinin, histamine, serotonin and capsaicin receptors, but the various roles of these receptors are poorly characterised.

The most fascinating aspect of pain perception in the periphery is that normally most nociceptors lie dormant. Inflammation sensitizes this vast population of nociceptors, making them far more sensitive to stimulation (hyperalgesia). Hyperalgesia may be primary (felt at the site of stimulation, related to sensitization of the neurones innervating that area) or secondary (felt at a site remote from the original injury, and probably related to NMDA-mediated "wind-up").

& Neurotransmitters

A plethora of neurotransmitters mediates transmission of the sensations of pain in both brain and spinal cord. The list is intimidating, and grows daily. We can try and 'lump' these neurotransmitters into various groups:

Specific neurotransmitters

Glutamate
A brief Medline search for articles using the abbreviation "NMDA" in the past ten years will garner about twelve thousand references. This attests the fanatical interest researchers have in this, the hottest of the glutamate receptors, but one mustn't forget that there are at least two others, the "AMPA" receptor and the obscure and devious metabotropic receptor. The NMDA receptor mediates a host of spinal responses to severe painful stimulation, but there are several catches to understanding how it works. Normally, the receptor is inactive as it is physiologically choked by a magnesium ion sitting in its ion channel. In order for this ion to be removed, adjacent peptide receptors have to be stimulated - the Mg⁺⁺ then pops off, and an emphatic painful response occurs. Neurophysiologists have known about this phenomenon for ages, gracing it with the label "wind-up" - as the frequency of C-fibre stimulation increases there is a dramatic and long-lasting central response, with some populations of spinal neurones becoming more and more sensitive to stimulation.

Consequences of glutamate receptor activation include production of c-fos (discussed below) and spinal production of prostanoids and the ubiquitous Dr NO, nitric oxide. Unfortunately all this knowledge benefits clinicians surprisingly little, as drugs that antagonise the effect of glutamate at the NMDA receptor tend to induce psychosis in humans, but the combination of low dose NMDA antagonists with opioids may be supra-additive with fewer side effects.

GABA
GABA is widespread in the brain and spinal cord. Together with its partner glycine, it has major inhibitory effects, dramatically evident in poisoning with strychnine, which antagonises glycine. Interneurones in laminae I, II and III are GABA-rich, and mediate gate control in the dorsal horn by synapsing on neurones that contain substance P.

There are several distinct GABA receptors that work quite differently - the GABA_A receptor is a "ligand-gated ion channel" that allows chloride ions to leak into the cell, while the GABA_B receptor is a "seven-spanning" transmembrane structure that activates G proteins.

Again, the clinical utility of this knowledge is small, as GABA_B receptor agonists such as baclofen, which are analgesic at the spinal level in rats, have little effect in man, although they may potentiate the analgesia of morphine. Benzodiazepines modulate the GABA_A receptor allosterically - but GABA_A seems more important at supraspinal than spinal sites.

Tachykinins
It will probably be several years before newer agents such as neurokinin antagonists have been tested sufficiently for widespread clinical use, although (for example) NK-1 antagonists such as CP-96 345 have been shown to moderately decrease isoflurane MAC in tail-clamped rats (shudder!) when given intrathecally. Neurokinin receptors probably do mediate pain in the spinal cord - substance P binds to the NK-1 receptor while neurokinins A and B bind respectively to the NK-2 and NK-3 receptors. Collectively these substances are known as 'tachykinins'. The tachykinin receptors are G-protein coupled, and increase intracellular calcium levels, triggering gene transcription.

at the Cellular Level

Probably the most significant discovery ever in the field of pain has been the gene c-fos. The cellular analogue of a viral oncogene, this rather special gene and its cellular product, the protein called Fos seem crucial to the profound central nervous system changes that occur when an animal (or man) feels pain. Central nervous system c-fos expression correlates extremely well with painful stimulation. Generically, Fos is one of the inducible transcription factors (ITFs) that controls mammalian gene expression.

We now have a molecular marker for pain! Even more important, we know that because c-fos is a proto-oncogene - that is, it can promote vast intracellular changes including cellular restructuring and proliferation - it is almost certainly involved in the long-term neurological consequences of noxious stimulation.

Noxious peripheral stimulation not only causes Fos to appear in the spinal cord, but also the ITFs called Jun and Krox. Certain "constitutive transcription factors" also change their activity (for example, CREB becomes phosphorylated). Brief stimulation (10 min) causes ITFs to appear within 30 min, peak at one to two hours, and disappear within about eight hours. Prolonged stimulation causes a many-fold increase in ITF expression, and substantially prolongs expression. Nociceptive C-fibre stimulation seems to be the main stimulus for ITF production in the spinal cord, as is the case with painful visceral stimulation (for example, introduction of cyclophosphamide into the bladder). In the latter c-fos appears within an hour in laminae I and II, the parasympathetic column, the dorsal grey commissure, as well as in the hindbrain (periaqueductal grey matter, locus coeruleus, the parabrachial area, and several other areas including the dorsal vagal complex and ventrocaudal bulbar reticular area).

Centrally, the reuniens, rhomboid and submedius nuclei all express c-fos, as do a number of areas in the cortex, thalamus, hypothalamus and amygdala. More caudally, the parabrachial nucleus also lights up!

A remarkable finding is that with prolonged stimulation, c-fos disappears from spinal neurones after two to seven days! This disappearance is despite increased neuronal excitability and a marked increase in expression of neurokinin and glutamate receptors, and may be simply because the neuronal changes are fixed, so the ITF is no longer needed! Conversely, chronic lesions of sensory nerves (for example, the neurophysiologist's favourite, partial sciatic nerve ligation) can induce chronic c-fos expression even in nerves which don't normally express the ITF!

Consequences of ITF production include neuropeptide production and synthesis of a variety of receptors. Fos is involved in cell replication and differentiation, but we are still horribly confused about its precise effects on neurones.

Fos may be important in expression of genes for dynorphin, enkephalin and tyrosine hydroxylase, for example, as these all have binding sites for AP1, the transcription complex that actually binds to DNA and does the job! (AP1 is the term used to describe the combination of Fos and the ITF Jun, or the dimer Jun:Jun. AP1 seems to bind the DNA consensus sequence TGACTCA).

It is now well known that "anaesthesia" (for example the combination of halothane and N2O) does not suppress production of c-fos within the spinal cord, despite an apparent "adequate" level of anaesthesia. Fentanyl reduces c-fos production by about 50%, and appropriate neuraxial block with local anaesthetic agents can almost totally ablate the c-fos response.

Opiates

Opioids act by stimulating mu, delta and kappa receptors. Other receptors (such as sigma and epsilon) have been postulated, but probably don't exist. The status of opioid receptor subtypes (mu1, mu2, delta 1, delta 2 and so on) is also controversial. If one looks at the genes for these receptors, there do not appear to be subtypes, but this doesn't mean that post-translational modification may not have created distinct subclasses of, say, the mu receptor. Of more interest is the identification of a so-called "orphan" receptor called ORL-1 (for "opiate-like receptor 1"), with about 60% sequence homology with the other opiate receptors. An endogenous ligand that acts at ORL-1 has recently been identified and called nociceptin, otherwise known as orphanin F2.

Even more recently its precursor, prenociceptin, has been found to give rise to other opioid-like substances called Noc II, Noc III and nocistatin, but the function (if any) of all these fancy new substances is unclear, to say the least!

A variety of G proteins are associated with opioid receptors, including G_i and G_o. Generally, mu and delta agonists seem to decrease cyclic AMP production, and hyperpolarise membranes by stimulating an inward rectifying potassium channel. This hyperpolarisation decreases the release of neurotransmitters from the nerve cell, as such release depends on opening of voltage-sensitive calcium channels.

Many researchers have tried to find out where opiates work in the central nervous system. The best-characterised site is undoubtedly the periaqueductal grey, where morphine micro-injection is profoundly analgesic. Other probable sites of action include the mesencephalic reticular formation, the substantia nigra, and the nucleus gigantocellularis & areas near the nucleus raphe in the medulla. Mu receptors also seem to be present in the amygdala, and the ventral forebrain.

Descending control mechanisms: Adrenergic and serotonergic pathways from the brainstem to the spinal cord inhibit incoming painful stimuli. Opiates release these pathways from GABAergic inhibition, increasing activity in the descending pathways and thus suppressing pain. Stimuli for opiate release in for example the periaqueductal grey include pain stimuli ascending from the spinal cord, and stimuli passing along the multiple connections that the brainstem pain structures have with other local nuclei, the midbrain, and higher centres.

Spinal opioid receptors: These are prevalent and are very effective in modulating pain - in lamina I the predominant receptors seem to be mu and delta, and in laminae II to V, kappa.

Peripheral opioid receptors: Although some studies have shown an inhibitory effect of opiates on peripheral nociceptors, the mechanisms and relevance of such observations remain uncertain. If these peripheral opioid receptors (predominantly mu and kappa) have a function, it is probably mainly in the terminals of C fibres already sensitized by inflammation. Practically, intra-articular morphine has a powerful analgesic-sparing effect after knee surgery! Another unusual revelation is that immune cells appear to be profoundly influenced by opiates, following stimulation of their delta and kappa receptors.

Non-steroidal anti-inflammatories (NSAIDs)

Classically the effectiveness of NSAIDs against inflammation has been attributed to their effects in inhibiting prostaglanding biosynthesis. They do this by several different effects on cyclo-oxygenase (COX), the crucial enzyme in the initial synthesis of prostaglandins:

Aspirin, the prototype NSAID, acts through the first mechanism, irreversibly acetylating COX. Many drugs such as ibuprofen, mefenamic acid and sulindac are reversible competitive inhibitors, and (surprisingly) paracetamol may scavenge hydroperoxides generated during arachidonic metabolism, and thus quench prostaglandin formation (The hydroperoxides normally stimulate COX, causing positive feedback)!

Much in the news has been the identification of two subtypes of COX. COX-1 is the constitutive variant that makes prostaglandins vital for protecting the stomach through mucus production, and maintenance of renal blood flow. COX-2 is the inducible form that mediates the pain of inflammation by sensitising peripheral nociceptors. Most conventional NSAIDs inhibit COX-1 more than COX-2, consequently pharmaceutical companies have invested fortunes in making selective COX-2 inhibitors.

NSAIDs have many other effects far removed from their COX inhibition. These include cell membrane effects (inhibition of neutrophil aggregation, inhibition of H₂O₂ generation (piroxicam), and even phosphodiesterase inhibition (indomethacin, diclofenac). Some of these effects may be mediated through interference with G protein function. In addition, NSAIDs may have central antinociceptive effects! This has been seen both experimentally (NSAIDs prevent the usual rise in cerebrospinal fluid prostaglandins after NMDA receptor activation) and clinically. Descending serotonergic pathways seem to be activated by NSAIDS, and part of the central action of NSAIDs in animal models appears to be prevented by naloxone! In addition, NSAIDs may interfere with spinal production of nitric oxide in response to NMDA receptor stimulation. They may even reduce c-fos expression!

Summary

In confronting the bewildering complexity of pain pathways, we tend to forget that fundamentally, above all else, pain is sore! There is another aspect of pain, that is, the localisation of a painful stimulus, but long before our primitive ancestors acquired the high-tech pathways necessary to perceive the sensory-discriminative aspects of pain, they were still sore. Throughout the nervous system, we can identify these two forms of pain transmission. In the periphery we have fast pricking pain transmitted by myelinated A-delta fibres, and 'slow' burning pain mediated by the primitive C fibres. In the spinal cord, the affective component of pain is transferred by spinoreticular fibres, and fibres to the medial thalamus, and thence to those parts of the cortex where 'sore' pain is perceived. The phylogenetically newer 'neospinothalamic tract' transmits discriminative components of pain to the SI cortex via the lateral thalamus.

As important as the ascending pathways, are descending pathways that modulate the incoming signal. These are predominantly noradrenergic and serotonergic, and can be modulated by stimulation of opiate receptors. Opiates, still the mainstay of therapy for moderate to severe pain, have a variety of receptors that are widespread in the central nervous system. Only now are we beginning to understand the subcellular mechanisms of pain, particularly how c-fos works. We hope that such understanding will eventually help us to manage more effectively man's second greatest scourge - pain.

References

From the vast pain literature, we've pulled out a few good articles and books for your delectation. We used these extensively in constructing the above review.

Brilliant Reviews

Worthwhile Reading

A few Web References on Brain Anatomy!

Pain Physiology