Sensory Nerves in Man and Their Role in Primary Headaches

Abstract

The sensory innervation of intracranial vessels originate in the trigeminal ganglion and comprise the following signal substances; calcitonin gene-related peptide (CGRP), substance P, neurokinin A, pituitary adenylate cyclase activating peptide (PACAP) and nitric oxide (NO). Studies in patients have revealed a clear association between head pain and the release of CGRP. In cluster headache and in a case of chronic paroxysmal headache there is in addition release of vasoactive intestinal peptide (VIP), which was associated with the facial symptoms (nasal congestion, rhinorrhea). In parallel with triptan administration, acting via 5-HT_1B/1D receptors, head pain subside and neuropeptide release normalise. These data show the involvement of sensory and parasympathetic mechanisms in the pathophysiology of primary headaches.

Introduction

With the advances in the understanding of cranial pain processing that have been made over the last few years it has become clear that cranial vessels play some role in either the pathogenesis or the expression of vascular pain syndromes such as migraine and cluster headache. Recent data also show that specific populations of cerebrovascular sensory nerves may be involved in the pathophysiology of these pain syndromes. In the present review, the origin, immunocytochemical distribution and ultrastructural features of perivascular nerves supplying human and laboratory animal cerebral blood vessels are described and their activation in primary headaches commented upon (due to space a selected list of references is given at the end where all the original papers can be found).

The sensory innervation of the cerebrovascular bed

The trigeminal ganglion

Most sensory fibres to cranial structures derive from the trigeminal ganglion. Several neuropeptides have been found in sensory nerves, of which those containing substance P (SP), neurokinin A (NKA) and CGRP dominate. Immunohistochemical studies have shown that CGRP is a major constituent in the sensory nervous system. The function of CGRP in sensory neurones is poorly understood; it has been suggested that CGRP acts as a vasodilator or ‘antivasoconstrictor’ in the pathogenesis of migraine headache. Stimulation of the trigeminal ganglion or the superior sagittal sinus has been regarded as the experimental equivalent of the alterations observed during migraine attacks. Electrical stimulation of the trigeminal ganglion induces structural alterations of CGRP-immunoreactive perivascular nerve terminals and releases SP and CGRP.

In the human trigeminal ganglia, CGRP-immunoreactive neurones occur in high numbers (40% of all neuronal cells), whereas SP-immunoreactive neurones occur in lower numbers (18%). This agrees well with observations from the cat and the rat in which the relation of CGRP to SP is 3:1. In situ hybridization has revealed that 40% of all nerve cell bodies contain CGRP immunoreactivity and CGRP mRNA. A smaller number of the neurones contain substance P (18%), NOS (15%) and PACAP (20%). Co-localization of CGRP-and SP-immunoreactivity has also been seen in human trigeminal cell bodies. Both CGRP and SP have been described as potent vasodilators in vivo and in vitro, the former being 10–1000 times more potent. Several studies have suggested that SP is involved in plasma extravasation from post-capillary venules in the dura mater during primary headache attacks. While neurokinin receptor antagonists are potent inhibitors of neurogenic inflammation, recent studies have shown that such blockers do not have any effect in acute migraine attacks. Furthermore, while CGRP is released during the headache phase of a migraine attack, SP is not. In addition, there are indications that SP does not take part in vascular nociception in humans.

The expression of NOS immunoreactivity in trigeminal nerve cell bodies has led to the suggestion that NO is a key molecule for inducing migraine attacks. NO release, either from the endothelium or from the perivascular nerves, may activate the guanylate cyclase system in the smooth muscle cells. This leads to a decrease in the intracellular Ca⁺⁺ level, giving rise to vasodilatation that hypothetically may activate the pain sensitive structures around the cranial vessels. Few trigeminal neurones express NOS in the rat and the cat. In human trigeminal ganglia about 15% of the cell bodies contain NOS immunoreactivity. Double immunostaining has revealed that only few CGRP-immunoreactive neurones (less than 5%) are NOS positive.

In situ hybridization and immunocytochemistry have revealed the expression of PACAP in both sensory and parasympathetic ganglia, which has led to the suggestion that PACAP may play an important role as a neuromodulator in the sensory and the autonomic nervous systems. PACAP dilates cerebral arteries and can increase cerebral blood flow. Co-existence with CGRP immunoreactivity can be seen in some cells. It is possible that this peptide may participate both in sensory mechanisms and in antidromic vasodilatation following activation of the trigemino-vascular reflex.

Sensory nerves

For many years, morphological studies of the sensory innervation of the cerebral vascular system were based on methylene blue staining and silver impregnation techniques. However, the results obtained were limited due to the fact that staining was inconsistent and did not allow the discrimination of nerve sub-populations. Significant advances have been made in the morphological analysis of presumptive vascular nerve endings by the introductions of electron-microscopic methods. The presence of an unusual abundance of mitochondria within nerve varicosities has been repeatedly shown to be one of the most distinguishing features of the sensory nerve terminals. In rat cerebral arteries, mitochondriarich nerve varicosities were interpreted as sensory in nature because they could be traced back to myelinated fibres and did not degenerate after sympathetic denervation by removal of the superior cervical ganglion. Presumptive vascular sensory nerve terminals may contain, besides a large number of mitochondria, small clear and/or dense vesicles (30–60 ran in diameter) and large dense-cored vesicles (80–150 ran in diameter), autophagic vacuoles, pleomorphic dense bodies, lamellated or myelin figures, and large amounts of glycogen-like granules. These nerve terminal varicosities have been postulated to represent nerve specializations for pressure or tension reception based on the structural analogy they share with sensory or baroreceptor nerve terminals. However, the use of immunocytochemical techniques has increased our knowledge of the distribution of sensory nerve fibres supplying the vasculature. To date, the most widely distributed neuropeptides in afferent cerebrovascular nerve fibres are the neurokinins and CGRP(59).

Neurokinins

Immunohistochemical studies have shown that SP, which is the best known neurokinin, is present in nerve fibres supplying cerebral vessels of a variety of species including man. Although interspecies differences have not been systematically studied, cerebral arteries from guinea pig and cat receive a relatively dense supply of SP-immunoreactive nerve fibres. Porcine cerebral arteries are innervated by a moderate nerve supply and cerebral vessels from humans, rabbits and rats appear to have a relatively sparse distribution of SP-containing nerve fibres. In general, SP-immunoreactive nerve fibres are more abundant in anterior vessels of the circle of Willis, such as the anterior cerebral artery. Nerve fibres containing SP-immunoreactivity are also associated with cerebral veins.

Retrograde tracing studies have demonstrated that the major source of cerebrovascular SP-containing nerve fibres is the trigeminal ganglion; however, some of the perivascular nerve fibres may also originate in the upper cervical dorsal root ganglia. Unilateral excision of the trigeminal ganglion in the cat decreases the cerebrovascular SP levels and induces an almost complete loss of SP-immunoreactive nerve fibres in the wall of cerebral arteries. The sensory origin of SP-immunoreactive nerve fibres has been substantiated by using the sensory neurotoxin capsaicin. Systemic capsaicin treatment of guinea pigs leads to a marked loss of SP-immunoreactivity in the cerebral vasculature. However, capsaicin does not deplete SP immunoreactivity from rat brain microvessels. This observation, together with the presence of substantial amounts of SP even after division of the trigeminal nerve, has led to the conclusion that while the trigeminal ganglion is the major source of SP in large cerebral arteries, it may not be the exclusive source of this neuropeptide for brain microvessels. For instance, in the dorsal raphe and interpeduncular nucleus it has been shown that dendrite processes of central perikarya containing SP immunoreactivity are intimately associated with small cerebral blood vessels.

NKA-immunoreactive nerve fibres have a similar distribution to that of SP-containing nerves and are also depleted after capsaicin treatment. Furthermore, NKA is colocalized with SP in cerebrovascular nerve fibres and in cell bodies in the trigeminal ganglion.

Calcitonin gene-related peptide (CGRP)

CGRP immunoreactivity is present in perivascular nerve fibres supplying the major cerebral arteries and pial arterioles of the cortical surface of all species examined including man. Marked species and regional variations are observed in the density of CGRP-immunoreactive cerebrovascular nerve fibres. While cerebral arteries of laboratory animals receive a relatively dense supply of CGRP-immunoreactive nerve fibres, human cerebral vessels are only innervated with a sparse nerve fibre network. From surgical denervation studies it appears that the trigeminal ganglion is the only source for the CGRP-containing cerebrovascular innervation. CGRP frequently coexists with SP and NKA in a population of sensory ganglion cells, although the number of CGRP-containing cells exceeds the number of SP/NKA-containing neurones. CGRP immunoreactivity is often colocalized with SP/NKA immunoreactivity in perivascular nerve fibres supplying rodent and feline cerebral arteries and the human superficial temporal artery. These results are further supported at the ultrastructural level, where the use of double immunogold staining demonstrated that CGRP and SP immunoreactivities are consistently found colocalized in the same large granular vesicles (70–150 nm in diameter) in both sensory neurones of the trigeminal ganglion and in varicosities of perivascular nerve fibres of the guinea pig.

PACAP and NOS

PACAP displays marked sequence homology with VIP and related peptides. Immunochemistry has revealed the occurence of PACAP in various parts of the sensory system, e.g. dorsal horn of the spinal cord, and in cell bodies of spinal and trigeminal ganglia. It coexists with SP and CGRP in fibres and ganglia. With in situ hybridization mRNA for PACAP has been detected in these ganglia. With the use of specific antibodies we early noted the presence of PACAP immunoreactive nerve fibres in the cat cerebral circulation. In a subsequent study the origins of the PACAP fibres were in the rat traced to the cranial ganglia; however, the majority of these reside in the parasympathetic ganglia. In the human trigeminal ganglion PACAP-containing cell bodies were more numerous than in the laboratory animals, amounting to about 15%. Colocalization was seen with CGRP. It is thus likely that PACAP may have a role also in the sensory system.

In the sensory system there is also expression of NOS in a sub-population of CGRP-containing cell bodies. The origin and distribution of cerebrovascular nerve fibres containing NOS have been evaluated previously in several species. The perivascular innervation is denser in the rostral part of the circle of Willis as compared with vessels posteriorly located. In the human circle of Willis the network of NOS-containing fibres is relatively sparse. Although it has been proposed that the trigeminal ganglion in the rat harbours a significant population of NOS-ir nerve cell bodies we could only detect occasional NOS-positive cell bodies. In the rat up to 40% of the trigeminal cell bodies store CGRP. Thus, NOS-ir perivascular nerve fibres in the rostral part of the cerebral circulation originate mainly in the ipsilateral sphenopalatine and otic ganglia with a near complete colocalization of NOS-ir with VIP- and PACAP-ir. The role of NO in the sensory system is not clear, although there is a suggestion that NO may participate in the pathogenesis of primary headaches.

Release of neuropeptides in primary headaches

Migraine attacks

The results of experimental trigeminal ganglion stimulation in man led us to examine the levels of various neuropeptides in patients during migraine attacks. Blood samples were drawn from the external jugular vein during headache attacks. The concentrations of NPY (a marker for sympathetic activity), VIP (parasympathetic activity) and CGRP or SP (markers for trigeminal activity) were measured. There were no changes in the levels of NPY, VIP or SP in the jugular vein. However, a marked increase in jugular vein CGRP was observed during migraine headache. Two individuals with symptoms similar to those seen in cluster headache (e.g. nasal congestion and rhinorrhea), displayed also increased VIP, but this was not the case in the whole group. The changes in VIP in these two individuals suggested the involvement of a parasym-pathetically mediated event. There was no difference between migraine with aura or without aura; both resulted in substantial increases in venous CGRP levels.

We have proposed that the release of CGRP rather than of SP, VIP or NPY might be due to the fact that the cerebral circulation is preferentially innervated by CGRP-containing fibres from the trigeminal ganglion. The observations in patients have been confirmed in subsequent studies. In addition, following sumatriptan administration the CGRP levels returned to control with successful amelioration of the headache.

Cluster headache

Cluster headache is a well-described clear-cut clinical syndrome. Patients with episodic cluster headache, fulfilling the criteria of the International Headache Society (IHS), were examined during acute spontaneous attacks of headache to determine the local cranial release of neuropeptides. During the attacks, the blood levels of CGRP and VIP were raised, while there were no changes in NPY or SP. Treatment with oxygen or subcutaneous sumatriptan promptly normalized the CGRP levels. Opiate administration on the other hand did not alter the peptide levels. This agrees well with the results of others, demonstrating release of CGRP in nitroglycerine-elicited attacks of cluster headache. CGRP in the jugular vein on the pain side in cluster headache patients was found to be elevated during the attack period and was elevated further at the peak of the provoked attack. There were no alterations in SP levels.

Thus, CGRP, a marker of the trigeminovascular system, and VIP, a marker of the parasympathetic nerve activity, are both elevated in the cranial venous blood of patients suffering acute spontaneous attacks of cluster headache. The termination of the attack with either sumatriptan or oxygen normalized the CGRP levels, probably reflecting cessation of the activity in the trigeminovascular system. In contrast, administration of an opiate agonist induced pain relief but did not end the trigeminovascular activity. The finding of elevated levels of both CGRP and VIP in the cranial venous blood during attacks suggests that there is activation of a brainstem reflex, the afferent arc of which is the trigeminal nerve and the efferent the cranial parasympathetic outflow from the VIIth nerve.