Review

Cephalalgia 0(0) 1–26 ! International Society 2018 structure and function related Reprints and permissions: sagepub.co.uk/journalsPermissions.nav to headache: structure and DOI: 10.1177/0333102418784698 function in headache journals.sagepub.com/home/cep

Marta Vila-Pueyo1 , Jan Hoffmann2 , Marcela Romero-Reyes3 and Simon Akerman3

Abstract Objective: To review and discuss the literature relevant to the role of brainstem structure and function in headache. Background: Primary headache disorders, such as and , are considered disorders of the brain. As well as head-related , these headache disorders are also associated with other neurological symptoms, such as those related to sensory, homeostatic, autonomic, cognitive and affective processing that can all occur before, during or even after headache has ceased. Many imaging studies demonstrate activation in brainstem areas that appear specifically associated with headache disorders, especially migraine, which may be related to the mechanisms of many of these symptoms. This is further supported by preclinical studies, which demonstrate that modulation of specific brainstem nuclei alters relevant to these symptoms, including headache, cranial autonomic responses and homeostatic mechanisms. Review focus: This review will specifically focus on the role of brainstem structures relevant to primary , including medullary, pontine, and , and describe their functional role and how they relate to mechanisms of primary headaches, especially migraine.

Keywords Primary headaches, migraine, medulla, , midbrain, cluster headache

Date received: 8 February 2018; revised: 24 April 2018; accepted: 23 May 2018

Introduction relaying somatosensory signals from the head and . The brainstem consists of the , the A detailed description of all ascending/descending pons and the midbrain, moving rostrally through the circuits associated with all brain structures relevant to brain, and at its most caudal it is contiguous with primary headaches can be found elsewhere (2–6), and the (Figure 1(a)–(b)). Its major function is through other relevant reviews of this special issue. Ten the integration of information from the rest of the body of the 12 cranial (III–XII) arise in the brainstem, through ascending connections with higher centres of the brain, such as the , , , and cortex (1). These are related to physio- 1Headache Group, Department of Basic and Clinical , logical survival regulation and reflex functions, includ- Institute of Psychiatry, Psychology and Neuroscience, King’s College ing cardiovascular and respiratory control, alertness, London, London, UK 2Department of Systems Neuroscience, University Medical Center awareness, consciousness, and sensitivity to pain (som- Hamburg-Eppendorf, Hamburg, Germany atosensory function). These crucial connections are 3Department of Neural and Pain Sciences, University of Maryland mediated via the for motor signals, Baltimore, Baltimore, MD, USA the posterior column-medial pathway involved with fine touch, vibration sensation and Corresponding author: Simon Akerman, Department of Neural and Pain Sciences, University of , and the for pain, Maryland Baltimore, 650 W. Baltimore Street, Baltimore, MD 21201, temperature, itch, and crude touch. In addition, there USA. is the trigeminothalamic tract, specifically involved in Email: [email protected] 2 Cephalalgia 0(0)

Midbrain

Pons Brainstem

Medulla oblongata

Thalamus

Hypothalamus Midbrain

Trochlear (IV) (V) Pons (VI) Vestibular- (VIII) Glosspharyngeal (VII) nerve (IX) (X) (XII) Medulla oblongata

Ventral root of 1st cervical verve (XI)

Spinal cord

Figure 1. Anatomical organisation of the brainstem and in the . (a) The brainstem includes, moving rostrally, the medulla oblongata (emerging from the ), the pons (emerging from the ) and the midbrain (emerging from the mesencephalon). At its most caudal, it is contiguous with the spinal cord connecting the trigeminal of the medulla. At its most rostral, it connects with diencephalic nuclei from the hypothalamus. (b) Ten of the 12 pairs of cranial nerves (III–XII) target or emerge from the brainstem. Nuclei of the oculomotor (III) and trochlear nerves (IV) are located in the midbrain. Nuclei of the trigeminal nerve (V), abducens (VI), facial (VII) and vestibulocochlear nerves (VIII) are located in the pons. Nuclei of the glossopharyngeal (IX), vagus (X), accessory (XI) and hypoglossal nerves (XII) are located in the medulla. Figure 1(b) is adapted from (1). Vila-Pueyo et al. 3 including the trigeminal (V), facial (VII) and vagus (X) the head and face. It is also responsible for the motor nerves (Figure 1(b)). These nerves supply the head, face nerves of the (12–15). The tri- and viscera. In relation to primary headaches, it is sim- geminal sensory root arises from the trigeminal gan- plistic to suggest that the major function of the brain- glion, and is split into three major branches: The stem is in the processing of noxious somatosensory ophthalmic or first division (V1), the maxillary or information relevant to the head and face. The com- second division (V2), and the mandibular or third div- plexity of their associated symptoms, particularly in ision (V3). The V3 division is the only branch to have a migraine and trigeminal autonomic cephalalgias sensory and motor component. Their intra- and extra- (TACs), such as triggers, which involve homeostatic cranial innervation is summarised in Figure 2(a)–(b). stresses including , feeding and sleep disruption, and premonitory symptoms, including tiredness/alert- Trigeminal brainstem nuclear ness, irritability, as well as cranial autonomic symp- complex (TBNC) toms, suggest an overall role in primary headache symptomatology processing and pathophysiology. In relation to primary headaches, the severe and throb- While many primary headache disorders involved bing head pain particularly associated with migraine mechanisms related to the brainstem, particularly the and TACs, localized in the frontal, temporal, parietal, medullary trigeminal nucleus, migraine is specifically occipital and high cervical region, is thought to be the associated with the brainstem through many imaging consequence of activation of the ophthalmic trigeminal studies, which highlight selective brainstem activation, projection to the intracranial . In TACs, the compared to other headache types such as the TACs. pain also predominates in the peri-orbital region, in and As the most prevalent and most well studied, this around the eye. A rich plexus of nociceptive nerve fibres review will largely focus on the brainstem role in that originate in the trigeminal (Tg) innervate migraine, although where relevant we will reference the pial, arachnoid and dural blood vessels, including other primary headache disorders. the and middle meningeal (MMA), as well as large cerebral (16–18). Medulla oblongata Noxious manipulation of these structures, particularly the dura mater, results in severe headache pain very The medulla oblongata, more commonly referred to as the similar to the pain in migraine and TACs, as well as ‘‘medulla’’, is the most caudal portion of the brainstem other symptoms associated with migraine, including and it connects the spinal cord with the rest of the brain nausea and photophobia (16–18). Interestingly, stimu- via the pons (Figure 1(a)–(b)). It is here that there is decus- lation of sites away from these blood vessels is much sation of approximately 90% of to their contra- less nociceptive, with correspondingly milder symptoms lateral side, ascending to cortico-thalamic structures. of headache. The nociceptive innervation of the intra- Structures within the medulla are primarily responsible cranial vasculature and includes non-myeli- for autonomic (involuntary; sympathetic and parasympa- nated (C-fibres) and thinly myelinated (Ad-fibres) thetic nervous systems) functions including vomiting and axonal projections. This is mainly through the ophthal- sneezing, as well as cardiac and respiratory centers. It mic (V1) division of the trigeminal nerve, but also, to a therefore deals with the autonomic functions of breathing, lesser extent, through the maxillary (V2) and mandibu- heart rate and blood pressure. The medulla is also import- lar divisions (V3). There is also neuronal innervation of antly involved in pain processing. Here is situated the ros- the dura mater from the cervical dorsal root ganglia tral ventromedial medulla (RVM), including the nucleus (19). Central afferent projections from the Tg of these raphe magnus (NRM). This forms part of the ‘‘endogen- dural nociceptive primary afferent Ad- and C-fibres ous pain modulatory pathway’’, thought to provide des- enter the caudal medulla of the brainstem, terminating cending modulation of nociceptive processing to the spinal on second-order neurons of the trigeminal brainstem and medullary dorsal horns (7–11). At the most caudal nuclear complex (TBNC), via the trigeminal tract. portion of the medulla is the trigeminal nucleus (Figures The TBNC is a long column of cells in the dorsolateral 2–3). These neurons are thought to be functionally con- brainstem, which extends from the rostral pons to the tiguous with the upper cervical dorsal horn and relay all C2 spinal level. This is subdivided into four subnuclei somatosensory information from the face and head, (Figure 2(c)): The principal or main sensory trigeminal including intracranial structures such as the dura mater. nucleus (Pr5), and three spinal trigeminal nuclei (Sp5), oralis (Sp5O), interpolaris (Sp5I) and caudalis (Sp5C) The trigeminal nerve (20). Sp5 is topographically organized in long rostro- caudal columns of cells, with midline structures (nose) The trigeminal nerve is the fifth (V) and largest of the represented medially and more lateral represented cranial nerves, and serves as the main of laterally, although each trigeminal branch conveys 4 Cephalalgia 0(0)

(a) (c) SSS Cerebral vessels MMA

Meningeal vessels

Nasal glands Trigeminal nerve

V1/Ophthalmic Vm V2/Maxillary V V3/Mandibular

(b) Facial (VII) Glossopharyngeal (IX) Vagus (X)

Pons Mesencephalic Motor Principalis Oralis Spinal trigeminal Interpolaris complex C1 Caudalis Substantia gelatinosa Medulla Nucleus proprious

Figure 2. Craniofacial sensory innervation territories of branches of the trigeminal nerve and cytoarchitecture of the trigeminal brainstem nuclear complex (TBNC). The trigeminal nerve is split into three major branches, the ophthalmic or first division (V1), the maxillary or second division (V2), and the mandibular or third division (V3). Intracranially (a), the V1 branch innervates the ethmoid and sphenoid sinuses, as well as the dura mater of the tentorium cerebelli and , along with portions of the dural vasculature, including (MMA), superior, sigmoid and occipital sinuses. The V2 branch innervates part of the , the , maxillary and , the dura of the , and portions of the dural MMA. The V3 branch innervates the oral cavity, the gums of the , the mucous membrane of the anterior two thirds of the tongue, the muscles of mastication, the external surface of the ear drum, and a small portion of the dural MMA. In the extracranial region (b), the V1 branch innervates the superior region of the face, including skin, eye and nose, the V2 branch innervates the cutaneous structures of the middle portion of the face, including the lower and cheeks. The V3 branch innervates the cutaneous structures overlying the mandible, the auricular, the lower , and the external surface of the eardrum. (c) Central afferent projections of the trigeminal nerve terminate on second-order neurons of the trigeminal brainstem nuclear complex (TBNC), which extends from the rostral pons to the C2 upper spinal cord. It is subdivided into four subnuclei, the principal or main sensory trigeminal nucleus (Pr5), and three spinal trigeminal nuclei (Sp5), oralis (Sp5O), interpolaris (Sp5I) and caudalis (Sp5C; trigeminal nucleus caudalis). GON: greater occipital nerve; C: cervical. Here we present the somatotopical organization in humans, but it is relatively conserved across species, for the rat organization see (245). Figures (a) and (b) are adapted from (20). Figure (c) is adapted from (246). signals to all nuclei of Sp5 (Figure 3(a)–(b)). Indeed, Trigeminocervical complex (TCC) animal tracing and electrophysiological studies indicate a dorso-ventral manner of somatotopic organization The most caudal portion of the trigeminal nerve, the throughout the TBNC. Allowing for the natural trigeminal nucleus caudalis (TNC; Sp5C), extends from curvature of the trigeminal nucleus, in general, the the to the upper cervical spinal cord (C2). Indeed, dorsal-most portion is mainly the V3 division, the stimulation of the dural vascular structures in animal V2 division is mainly central, and the V1 division is models, including the superior sagittal and transverse mainly the ventral-most portion of the dorsal horn sinuses, and MMA, results in activation of neurons (Figure 3(b)). in the TNC, C1 and C2 regions of the cervical spinal Vila-Pueyo et al. 5

(a) (c) (d) Mouse dorsal horn Sp5C /dorsal horn Dorsal root Peripheral nerve Obex ganglion endings I

layer NS/WDR Marginal IIi NS/WDR

IIo gelatinosa Substantia Non-N

III Non-N

C1 IV (b) Non-N

Nucleus proprius WDR

V NS-WDR

VI

deep laminae NS/WDR

X C2 NS/WDR Circumcanular

V3 (Mandibular)

V2 (Maxillary)

V1 (Ophthalmic)

Figure 3. Somatotopical organization of the spinal trigeminal nucleus. The spinal trigeminal nucleus (Sp5) is topographically orga- nized in rostro-caudal columns of cells (a)–(b). Sp5O (oralis) receives information relating to the mouth, , tongue, teeth and front of jaws, Sp5I (interpolaris) receives cutaneous and visceral signals from the middle of the face including nose, and Sp5C (caudalis) receives somatosensory signals from the most lateral territory of the face. The trigeminal brainstem nuclear complex is also somatotopically organized (b). In general, the dorsal-most portion of the trigeminal nucleus reflects mainly innervations from the V3 (mandibular) region of the trigeminal nerve. The V2 (maxillary) innervation reflects mainly the middle portion of the face, including the nose and just below the eye. The V1 (ophthalmic) division of the trigeminal nerve predominantly innervates the area above the eye, including the intracranial dura mater, and is found in the most ventral region of the dorsal horn. These boundaries are not clearly separated, and there is overlap between the specific divisions of the trigeminal nerve and the regions of the face, head and intracranial structures that they do innervate. The trigeminal nucleus caudalis (Sp5C; (c)–(d)) has a similar laminae organization and properties to the spinal dorsal horn. These layers include a marginal zone (lamina I, 1Sp), a substantia gelatinosa layer (lamina II, 2Sp), and a magnocellular layer, equivalent to nucleus proprius (lamina III-IV, 3-4Sp; (Figure 3(c))). (d) summarizes the fibre types and their somatosensory properties that terminate in the specific laminae of the medullary dorsal horn (Sp5C–C2). (c) is a hand drawing of the proposed lamina delineation of the mouse medullary and spinal dorsal horn adapted from the Allen Mouse Brain Atlas (http://atlas.brain-map.org/atlas), Allen Mouse Spinal Cord Atlas (http://mousespinal.brain-map.org), and Paxinos and Watson mouse atlas (247). NTS: nucleus tractis solitarius; Spt5: spinal trigeminal tract. cord – together known as the trigeminocervical com- a similar to the spinal dorsal plex (TCC) (21–24). Furthermore, stimulation of the horn (lamina I-V) (20) (Figure 3(c)), which disappears greater occipital nerve also causes neuronal activation around the caudalis/interpolaris transition zone. This in the same regions and enhances convergent inputs has led it to also be known as the medullary dorsal from the dural vasculature (25,26). These data suggest horn. Laminas V–VI have been described by some that the trigeminal nucleus extends beyond its caudalis authors, although this appears to be less well defined. boundary in a functional continuum, which includes Nociceptive Ad- and C-fibres predominantly terminate the upper spinal dorsal horn. The TNC also has in the superficial laminae, I and IIo, as well as deeper 6 Cephalalgia 0(0) laminae V–VI (24,27–31) of the TNC and cervical but also with the significant co-morbidity of several extension, whereas Ab-fibres terminate predominantly headache disorders with orofacial pain disorders, such in central laminae III–IV. Trigeminal primary afferent as temporomandibular disorders (TMDs) (37–41). This nerve fibres converge on three types of second overlap of innervation impacts the likely pathophysi- order neurons in the medullary dorsal horn (32–34) ology and occurrence of co-morbid pain disorders. (Figure 3(d)): Rostral ventromedial medulla (RVM) . Low threshold (LT) neurons, which are activated by innocuous touch, vibration and pressure, receiving The rostral ventromedial medulla (RVM) is a cluster of inputs predominantly from Ab-fibres neurons located within the medullary region that . Nociceptive-specific neurons, which are silent at rest, stretches as far as the pons, close to the midline floor. become active in response to high intensity, noxious For anatomical localisation see Figures 1, 4, 6 and 8. stimuli, including pain, temperature and itch, receiv- Nuclei within the RVM make bidirectional connections ing inputs from Ad- and C-fibres with the ventral lateral PAG (vlPAG), and spinal and . Wide dynamic range (WDR) neurons, which are medullary dorsal horns. This pathway is thought to activated by a range of noxious and innocuous sti- provide descending modulation of pain processing to muli (mechanical, chemical, thermal), and receive the spinal and medullary dorsal horns. This was first convergent inputs from first order extracranial cuta- evidenced by the high concentration of endogenous neous and intracranial visceral (viscerosomatic con- signalling, via ON, OFF and Neutral cells, and vergence) structures onto the same second-order is described as the ‘‘endogenous pain modulatory path- neurons, receiving inputs from Ab-, Ad- and C-fibres way’’ (8,42). Within the RVM, non-serotonergic ON cells facilitate pain and are inhibited by , The major function of the trigeminal nucleus and its whereas OFF cells are tonically active, and they are cervical extension is to relay all somatosensory infor- activated by opioids (43,44). Neutral cells do not mation from craniovascular structures (face and head respond to peripheral noxious input, and whether relevant to many headache forms), including intracra- they play a role in the transmission and modulation nial and extracranial structures, to other areas of the of central pain syndromes is not entirely clear. brainstem and the , via ascending connec- Functionally, ON and OFF cells in the RVM (including tions involved in somatosensory processing. This the ) are also involved in the con- includes direct projections to thalamic neurons, via trol of external stimuli, motor activity, and homeostatic the trigeminothalamic tract, as well as projections to processes (10,45), discussed in more detail in the NRM hypothalamic nuclei via the trigeminohypothalamic and PAG sections. tract (2,4,6). There are also projections to brainstem nuclei including the (LC) and periaque- Vagus nerve ductal gray (PAG). Trigeminovascular neurons are also under the control of pain-modulatory circuits in the The vagus nerve is the Xth cranial nerve and arises in brainstem, including the PAG and rostral ventromedial the medulla (Figure 1(b)). It is composed of 80% affer- medulla (both discussed in detail below), and therefore ent fibres, projecting to the brain, and 20% efferent these neurons are considered as integrative relay neu- fibres, projecting to the rest of the body, each of rons between peripheral and central nociceptive mech- which carries signals in both directions. These nerve anisms. In the context of primary headaches, the fibres originate from cell bodies in the superior (jugular) convergence of neuronal inputs into the TCC (21–23) and the larger, inferior (nodose) vagal ganglion (Figure from intracranial and extracranial structures and cer- 4(a)), and are made up of A-, B-, and C-fibres (46). The vical regions of different nerve fibre types, probably vagus nerve represents the major parasympathetic accounts for the distribution of pain and innervation of the autonomic . Efferent cutaneous allodynia in migraine and TACs over the fibres project to the larynx, lungs, heart, , liver, frontal and temporal regions, plus the involvement of pancreas, and gut, and are involved in the primary parietal, occipital, and higher cervical regions (35). function of the vagus nerve; regulating involuntary Furthermore, while primary headaches such as function including the control of heart rate, respiration migraine and the TACs are predominantly mediated and digestion. Visceral efferents participate in the pre- via V1 activation, there is overlap between the specific ganglionic parasympathetic nervous system and arise divisions in the regions of the face, head and intracra- from the dorsal motor nucleus of the vagus (DMN) nial structures that they innervate. The result is consid- and nucleus ambiguous, innervating all thoracic and erable overlap of head pain into other trigeminal abdominal organs, and striate muscle. Fibres arising divisions beyond V1, especially in facial migraine (36), from DMN do not directly innervate peripheral Vila-Pueyo et al. 7

(a)

Midbrain

Amygdala Pons

Medulla Vagus nerve oblongata

Superior (jugular) ganglion of the vagal nerve

Inferior (nodose) ganglion of the vagal nerve

(b)

ACh, NO,VIP PACAP

Amygdala

TG

SCG

Sympathetic

Spinal cord

Figure 4. Summary of the neurophysiological connections of the medullary-pontine autonomic parasympathetic and sympathetic region. (a) The vagus nerve is the major parasympathetic innervation of the . Its efferents project to all areas of the body, including larynx, lungs, heart, stomach, liver and pancreas. Afferent nerve fibres originate in cell bodies in the superior (jugular) and inferior (nodose) vagal ganglion. From the jugular ganglion, they project to the nucleus tractus solitarius (NTS), (continued) 8 Cephalalgia 0(0) organs, but on adjacent neurons in the parasympathetic , bladder control, as well as facial move- ganglia close to these organs. Post-ganglionic parasym- ment, and posture. The pons contains cranial pathetic neurons travel to cardiovascular, respiratory, nerve nuclei, including the pontine and motor nuclei and GI systems. of the trigeminal nerve (V), and the facial nerve (VII) The majority of afferent vagus nerve fibres, bringing (Figure 1(b)). These are involved in sensory processes information from the rest of the body, project bilaterally related to touch and pain, facial sensation and expres- to the nucleus tractis solitarius (NTS) in the medulla. sion, and secretion of saliva and . Nuclei within the The remaining fibres project ipsilaterally to the SpV, pontine region are also involved in the control of touch , the DMN and (47). and pain sensation, specifically cranial sensation, Visceral afferent fibres from the , heart and abdo- through descending projections to the medullary men carry information whose cell bodies are located in dorsal horn (trigeminal), via the LC, nucleus raphe the nodose ganglion and transmitted to the caudal NTS. magnus (NRM), and parabrachial nucleus. Structures Somatic afferents transmit sensory information from the within the pons are also concerned with autonomic lower part of the pharynx, larynx, trachea, bronchi, function, specifically the parasympathetic nervous oesophagus, the ear and ear canal, and dura mater system, via the preganglionic superior salivatory lining the , via the jugular gan- nucleus (SuS) and its craniovascular projection, as glion, terminating in SpV, where they project to somato- well as the parabrachial nucleus. sensory thalamic neurons. From the NTS, vagal afferents project to the LC, , preganglionic Superior salivatory nucleus (SuS) parasympathetic neurons, the thalamus, the parabra- chial nucleus, the PAG, the amygdala and , The SuS is part of the trigeminal autonomic reflex and the , traversing many synapses in the (Figure 4(b)), receiving a reflex connection from the process (46,48,49). Through these many projections, TNC (50), connecting these two important somatosen- the vagus nerve directly influences visceral sensory path- sory and autonomic pathways. The SuS contains the ways, somatosensory, higher autonomic, extrapyram- cell bodies of neurons that are part of the cranial para- idal motor and limbic function, all of which are sympathetic autonomic vasodilator pathway (51). relevant to headache symptomatology and implicate it These neurons project predominantly through the in its pathophysiology and treatment mechanisms. greater petrosal branch of the facial (VIIth) nerve, via the sphenopalatine (pterygopalatine; SPG) ganglion Pons (52) to the cranial vasculature, including the dura mater, and to the nasal and oral mucosa, and lacrimal The pons (or bridge) sits between the medulla and mid- glands. The SuS also receives inputs from the vagus brain, and essentially forms an important bridge nerve afferents, via the NTS, from the heart, thorax between these two regions to allow these areas to com- and (Figure 4(a)). municate with each other. It provides an important Preclinical studies demonstrate that activation of connection between the and this SuS-parasympathetic projection to the cranial vas- related to motor function, and is also involved with culature mediates vasodilation of cerebral blood vessels sleep, control of motor function such as respiration, (53–55) and secretory responses from glands. Indeed,

Figure 4. Continued. where they directly or indirectly innervate many areas of the brainstem implicated in headache mechanisms, including locus coeruleus (LC), parabrachial nucleus (PB), preganglionic parasympathetic superior salivatory nucleus neurons (SuS), and raphe nuclei (DRN). From the nodose ganglion afferent fibres primarily project to the trigeminocervical complex (TCC), as well as the dorsal motor nucleus of the vagus (DMN) and nucleus ambiguus (NAm). (b) The SuS also receives a connection from the TCC and projects to the cranial vasculature via the (green), and its synapse with the sphenopalatine ganglion (SPG), as well as via the facial nerve (VIIth cranial nerve; blue). Together these form the cranial parasympathetic autonomic projection. There is also a cranial sympathetic autonomic projection (orange) to the intracranial vasculature, and facial region, predominantly via the superior cervical ganglion (SCG). Activation of these pathways contributes to cranial autonomic symptoms in primary headaches, especially trigeminal autonomic cephalalgias, but also migraine via the release of (, ACh; vasoactive intestinal peptide, VIP; nitric oxide, NO; and pituitary adenylate cyclase-activating peptide, PACAP). The SuS also receives descending projections from a host of hypothalamic nuclei, the PB, as well as limbic (amygdala, hippocampus, thalamus) and cortical regions, and indirectly from the (PAG). Together they control SuS activation. Many of the hypothalamic and brainstem (PAG, RVM, parabrachial nucleus) nuclei are also involved in trigeminovascular nociceptive processing, of which the SuS is thought to contribute. RVM: rostral ventromedial medulla; TG: trigeminal ganglion; PH: posterior; SON: supra-optic; VMH: ventromedial, PON: pre-optic; LH: lateral; DMH: dorsal medial; PVN: paraventricular hypothalamic nuclei. Vila-Pueyo et al. 9 direct stimulation of the SuS in rats produces blood but it is now thought that vasodilation during TACs flow changes in the lacrimal gland, indicative of cranial and migraine is an epiphenomenon, and less relevant to autonomic features (56,57). These changes are specific- the pathophysiology of the headache symptoms. ally mediated by activation of the cranial parasympa- The SuS also has a network of bidirectional connec- thetic pathway, as a specific SPG blocker attenuates tions with areas of the hypothalamus, including lateral these responses, as does the cluster headache treatment, (51,69), paraventricular, dorsomedial and pre-optic inhaled oxygen (56,57). Activation of the cranial hypothalamic nuclei (51,70,71), as well as the A5 (lat- SuS-parasympathetic pathway is therefore thought to eral tegmental area) and the parabrachial nucleus, and directly contribute to cranial autonomic symptoms limbic and cortical areas (51). These are regions found in primary headaches, including TACs such as involved in homeostatic functions including feeding, cluster headache and paroxysmal hemicranias, as well sleep and stress, which alongside the central autonomic as in migraine (57,58). The main neurotransmitters of system are thought to be ultimately under the direct this projection include acetylcholine (ACh), vasoactive control of the hypothalamus. Many of these regions intestinal peptide (VIP), pituitary adenylate cyclase- also send direct projections to the TCC, and therefore activating peptide (PACAP), and nitric oxide (NO). are thought to contribute to the triggering mechanisms ACh is likely involved in the secretory responses, involved in migraine and TACs, as well as to associated whereas VIP, NO and PACAP likely mediate vasodila- neurological symptoms (72). The SuS sits within a tory and, potentially, sensory responses. In primary brainstem network that is therefore ideally placed to headaches, it is likely that both VIP and PACAP con- integrate and relay nociceptive and autonomic informa- tribute to the cranial autonomic symptoms. In experi- tion to and from the trigeminovascular system, as well mental clinical studies in migraineurs, both VIP and as being under descending control of the hypothal- PACAP are known to cause severe cranial autonomic amus, therefore integral in the pathophysiology of symptoms (59–61), as well as extracranial vasodilation. headache and cranial autonomic symptoms in primary Also, VIP (62,63) and PACAP (64,65) levels in the headaches. extracranial and peripheral vasculature are increased during migraine and cluster headache. However, VIP Locus coeruleus (LC) levels are only increased during migraine when accom- panied with severe cranial autonomic symptoms, sug- The locus coeruleus (LC) (Latin for blue spot)isa gestive of a definitive role. Taken together, it is clear small cluster of noradrenaline (NA)-synthesizing neu- that activation of the cranial parasympathetic pathway, rons located in the upper dorsolateral pontine tegmen- via VIP and PACAP, is important in mediating cranial tum, in the lateral floor of the autonomic symptoms. (Figure 5(a)–(b)). It is one of several noradrenergic The cranial parasympathetic vasodilator pathway cell groups distributed throughout the brainstem, also influences the dural microenvironment, and as a referred to as the A6 cell group according to the clas- consequence central trigeminovascular neurons and sification of Dahlstrom and Fuxe (73). The LC is the head pain in migraine and TACs. The dural blood ves- largest of the noradrenergic groups and it sends select- sels are richly innervated by parasympathetic nerve ive widespread CNS projections providing one of the fibres (66) (Figure 4(b)), and activation of this pathway only sources of NA to the , spinal cord, hypo- causes the release of ACh, VIP, PACAP and NO from thalamus, hippocampus, cerebellum and parts of the dural vascular terminals of post-ganglionic sphenopa- thalamus (Figure 5(a)–(b)) (74–76). Furthermore, the latine neurons. It is believed that activation can lead to LC receives an extensive and varied number of inputs a sensory neuro-inflammatory response, with dilation which contribute to the high complexity of the LC-NA of intracranial vessels, plasma protein extravasation system (77,78), including from the hypothalamus. and local dural release of inflammatory mediators Through these varied connections, the LC acts as a (67). Studies in rodents also demonstrate that direct modulator of a wide diversity of neural networks, stimulation of the SuS activates two populations of including being a key component in , cognition, neurons in the TCC (57,68), one specifically via activa- autonomic function, and stress circuits (6). tion of the parasympathetic outflow to the cranial vas- Moreover, the LC has a complex and specific role in culature. It is thought that this in turn indirectly the control of behaviour. In animal studies, it has been activates trigeminal afferents from the dura mater to shown that changes in behavioural state are preceded the TCC. This neuronal response is attenuated by by changes in LC activity. LC neurons fire in either inhaled oxygen and an autonomic ganglion blocker tonic or phasic modes. During tonic activity, neurons (57), suggestive of direct activation of the cranial para- discharge in a sustained and highly regular pattern, sympathetic projection. Interestingly, only a modest with higher rates during , lowest during dural vasodilation accompanies this neuronal response, slow-wave sleep and with almost no firing during 10 Cephalalgia 0(0)

(a) Rat brain

(b) Human brain

Midbrain

Amygdala Pons

Lateral tegmental NA cell system Medulla oblongata

Figure 5. Noradrenergic projection neurons, including locus coeruleus, located in the pons and medulla. (a) Noradrenergic (A groups) and adrenergic (C groups) neurons are located in the pons and medulla. A1/C1 cell groups in the nucleus ambiguus, and A2/C2 groups in the nucleus tractus solitarius both project to the hypothalamus. The A5 (lateral tegmental area), A6 (locus coeruleus; LC) and A7 () cell groups in the pons project to the spinal cord and modulate autonomic and pain sensation (a)–(b). The LC is the major synthesiser of noradrenaline in the brain. Its project to many forebrain structures, including the cerebral cortex, hippocampus, amygdala, cerebellum and thalamus, and it has an important role in arousal and . (a) is a representation of projections in the rat brain, (b) is a representation of projections in the human brain.

REM sleep (79). Highest rates are also found in disruptions are reported to increase migraine attack fre- response to stimuli that induce arousal and exploratory quency and chronification (84), fatigue is prevalent behaviour (80). On the other hand, phasic activity is throughout the multiple attack phases (85,86), and driven by complex sensory input, task-relevant stimuli, sleep is frequently reported as a common abortive strat- and is critical for vigilance and for the efficient process- egy (84). Furthermore, migraine attack onset is most ing of salient information (75). frequent in the early morning when arousal networks, The LC has a key role in regulating the sleep-wake including the LC and the hypothalamic nuclei, are cycle and levels of arousal (81,82) through neural net- most active (87,88). The prominent role of the LC in works with hypothalamic nuclei. In connection with regulating the sleep-wake cycle positions it as one of the this, it is relevant to highlight the importance of arousal potential mechanistic links between migraine patho- dysregulation in migraine (83). Sleep and circadian physiology and its sleep-wake disruption (83). Vila-Pueyo et al. 11

In support of this hypothesis, several neuroimaging the caudal groups, which are separated by a narrow studies in migraine patients have shown an altered -free transversal segment located at the hypothalamic functional connectivity with the dorsal middle of the pons (113). The rostral group, located rostral pontine region, a region including the LC within the pontine-midbrain region, includes the among other nuclei, which could partially explain auto- dorsal raphe (see midbrain section), the median raphe nomic symptoms occurring during migraine attacks and the caudal linear nuclei, with major projections to (89–91). In preclinical studies, the LC is activated by the forebrain (Figure 6(a)–(b)). The caudal group, known clinical migraine triggers including nociceptive located within the medullary pontine region, includes trigeminal activation (92) and systemic nitroglycerin the raphe magnus, raphe obscurus and raphe pallidus injection (93), and electrical or pharmacological nuclei and parts of the adjacent lateral reticular forma- manipulation of the LC modulates dural-evoked noci- tion, and mainly projects to the brainstem and spinal ceptive trigeminovascular activation (94). Additionally, cord (Figure 6(a)–(b)). modulation of the LC influences the susceptibility to cortical spreading (CSD), the experimental Nucleus raphe magnus (NRM) correlate of migraine aura. Direct electrical stimulation of the LC induces cerebral hypoperfusion, maximally in The nucleus raphe magnus (NRM) is located at the the occipital cortex (95–97), and indirect stimulation junction between the pons and the medulla, at through vagus nerve activation (98,99) decreases CSD the level of the facial nucleus, and it is restricted to susceptibility (100). the midline (Figure 6(a)–(b)). It is the largest serotoner- Besides containing NA, LC neurons also co-express gic nucleus of the raphe nuclei caudal group (113). This other distinct in a specific anteroposter- caudal group sends major projections to the caudal ior and dorsoventral distribution, providing heterogen- brainstem and spinal cord (114). The NRM specifically eity to their function (101). Given the prominent role of sends projections to the dorsal horn of the spinal cord, calcitonin gene-related peptide (CGRP) in migraine with dense innervation of laminae I and II of the TNC pathophysiology (102), it is important to highlight (115), the , parafascicular nucleus, that the highest CGRP immunoreactivity in the brain- vlPAG, noradrenergic nuclei including LC, A5 and A7, stem is found in the LC with approximately 80% of LC and NTS (116). The NRM also receives afferent pro- cell bodies being positive for CGRP in humans jections from the TNC, several cortical areas, lateral (103,104). Other neuropeptides known to be involved habenular nucleus, dorsolateral PAG (dlPAG), caudal in migraine are also expressed in the LC, including pontine reticular nucleus and lateral paragigantocellu- PACAP, which is found in 40% of LC cells (103), sub- lar reticular nucleus (117). NRM neurons are largely stance P, found in LC fibres (103), and Y, serotonergic and differ in morphology depending on found in lower density and only shown in rodents their localization. Bipolar serotonergic neurons are (105,106). LC neurons also express a variety of recep- mainly found in the midline, and multipolar neurons tors that are modulated by other brain structures impli- are found in the paramedian and lateral reticular for- cated in headache pathophysiology. These include mation (114). Besides serotonin, NRM neurons can neurokinin-1 tachykinin (NK1) receptor (107); NPY also co-express (118) , neuropeptide Y receptor type 1, 2, 4 and 5 (108); 5-HT1F receptor (119), VIP (103), thyrotropin-releasing hormones (109); m- (110); and receptor (120) and other neuropeptides including encephalin, type-1 (111), which can be modulated through the somatostatin and (121). dense projections received from the lateral hypothal- Until recently, the main function of the RVM-NRM amus (112), and could be involved in the dysfunction nuclei was thought to primarily be in the modulation of of arousal in migraine. nociceptive inputs, forming part of the ‘endogenous pain modulatory pathway’. It provides descending con- Raphe nuclei trol of nociceptive processing to the spinal and medul- lary dorsal horns via non-serotonergic ON and OFF The raphe nuclei constitute a collection of cell groups cells. ON cells facilitate pain, and are most active distributed in the midline region of the during waking, whereas OFF cells are tonically active, (Figure 6). The term raphe refers to a seam or ridge their firing inhibits nociceptive inputs, and they are along the midsagittal plane of the body where both left most active during sleep (43,44). There also exists a and right-side structures fuse or join. The raphe nuclei reciprocity in the connectivity between the PAG and contain heterogeneous populations of neurons, with a the NRM that may be involved in the modulation of majority of serotonergic neurons that provide the the PAG-NRM-spinal cord pathway (116). In fact, the major source of serotonin synthesis in the brain. PAG (see PAG section) has excitatory connections with They are allocated into two groups: the rostral and the NRM, indicating that the anti-nociceptive effects of 12 Cephalalgia 0(0)

(a) Rat brain

Pons (b) Human brain Medulla

Striatum

Midbrain

Amygdala Pons

Medulla oblongata

Figure 6. Serotoninergic-raphe projection neurons located in the midbrain, pons and medulla. (a) The raphe nuclei are the major source of serotonin synthesis in the brain. The serotoninergic-raphe (B cell groups) neurons are located in the midbrain, pons and medulla of the brainstem, and project throughout the brain. Within the medulla B cell groups (B1, ; B2, nucleus raphe obscurus; B3 nucleus raphe magnus (NRM); B4, region of (MVe) and ) project throughout the medulla and spinal cord, and are involved in modulating afferent pain signals, , cardiovascular control, and breathing. (a)–(b) Within the pons and midbrain B cell groups (B5, nucleus raphe pontis; B6 forms the caudal portion of the nucleus raphe dorsalis (DRN), with B7 the rostral potion of the DRN adjacent to the peraqueductal gray (PAG); D8 (MRN); and B9 is the region of the ) are involved in arousal, mood and cognition. These neurons also project throughout forebrain stuctures, including the cerebral cortex, hippocampus, hypothalamus, thalamus, and cerebellum. (a) is representation of projections in the rat brain, (b) is representation of projections in the human brain.

PAG stimulation may be mediated by the NRM (122). (124–127), and ON and OFF cells respond to various Stimulation of the NRM in animals has been shown to craniofacial inputs that are modulated by antimigraine be anti-nociceptive in behavioural experiments (123), treatments; intravenous naratriptan was shown to which has been linked to the inhibition of nociceptive increase spontaneous activity of OFF cells and to dorsal horn neurons through the action of serotonin, decrease ON cell activity (125). Moreover, electrical endogenous and/or NA. Interestingly, several or chemical manipulation of the NRM modulates tri- reports also implicate the NRM in modulation of tri- geminovascular nociceptive activation in the TCC. geminovascular pain pathways. In animals studies, the Specifically, electrical stimulation of the NRM inhibits NRM is activated by dural nociceptive stimulation both dural and periorbital-evoked nociceptive neuronal Vila-Pueyo et al. 13 activation in the TCC of cats (126). Whereas in rats, described, as well as impacting somatosensory nocicep- microinjection of GABA or muscimol (GABAA recep- tive processing. tor agonist) in the NRM increases dural-evoked noci- ceptive neuronal activation in the TCC, while Parabrachial nucleus bicuculline (GABAA receptor antagonist) decreases it and neither baclofen (GABAB receptor agonist) nor The parabrachial nucleus is located in the pons 2-hydroxysaclofen (GABAB receptor antagonist) has (Figure 4) and plays a significant role in many physio- a significant effect on the neuronal activation (128). logical processes such as arousal (135,136), thermo- On the same line, microinjection of orexin A and B in regulation (137) and the regulation of blood sugar the NRM increases trigeminovascular nociceptive acti- (138). Beyond this, it is involved in the modulation of vation in the TCC, an effect that is mainly driven by pain, including trigeminal pain (139–141). The TCC, in OX2 receptors (129). Lastly, using experimental condi- particular laminae I and II, connect to the medial and tions to simulate potential migraine triggers in cats, lateral parts of the parabrachial nucleus (139,140,142– using CSD and bright light stress, both were shown 145). Rostrally, the parabrachial nucleus connects to to inhibit spontaneous neuronal discharge firing in the the RVM to access the descending pain modulatory NRM (126). In the same study, CSD antagonized the system (141), as well as to higher subcortical structures inhibitory effect of NRM stimulation on dural-evoked, including the thalamus and hypothalamus for pain pro- but not peri-orbital-evoked, trigeminovascular nocicep- cessing and other subcortical structures such as the tive activation in the TCC. amygdala for integration with the central autonomic Despite this plethora of data, the functional role of network (139,143,144,146–148). Some of these pro- the NRM is now thought to go beyond simply modu- cesses of the parabrachial nucleus and its interactions lating somatosensory mechanisms. Via a network of with other structures, such as the thalamus, are connections with other brain nuclei, including the mediated by CGRP (149,150) and other neuropeptides, PAG and hypothalamus, the RVM-NRM is also including PACAP (151), relevant to primary headaches thought to respond in external innocuous stimuli (152). Given the high importance of these neuropep- (130), motor activity (131,132), and is involved in mech- tides in the pathobiology of migraine and the described anisms related to (44,133,134). ON cells neuroanatomical connections within the relevant pain are state-dependently active during waking hours, but pathways, it may be hypothesized that the parabrachial not during feeding and micturition, or during sleeping nucleus plays a role in migraine. Following this hypoth- (44,130,133,134). OFF cells are active during sleep, and esis, results from preclinical in vivo studies indicate that only active during waking prior to micturition, as well noxious trigeminal stimulation induces an increase in as during feeding. Therefore, OFF cell firing during Fos expression, a marker of neuronal activation, in the important homeostatic functions such as sleep, feeding parabrachial nucleus (140). The same observation has and urination seems to suppress arousal, to prevent been made after administration of the NO-donor, against responding to innocuous and even acute nox- nitroglycerin (153,154). Using NO-donors as stimulat- ious stimuli. During waking, ON cell firing seems to ing agents is particularly interesting, as in humans the facilitate alertness to sensory stimuli. In combination substances can trigger migraine-like attacks in migrain- with the PAG, RVM-NRM neurons are now thought eurs (155). Tracing techniques revealed that the acti- of as brain nuclei involved in modulating sensory, auto- vated regions in the parabrachial nucleus that showed nomic and motor processes in the spinal cord across a significant Fos expression project to most of the normal conditions, and their network of connections higher brain regions described above (154). to forebrain structures suggest that they are primed in However, little is known so far on the role of the readiness to prioritize responses to nociceptive inputs. parabrachial nucleus in primary headache disorders. If then one takes the bigger picture of migraine patho- The neuronal activation observed, and the neuroana- physiology in particular, triggers such as sleep and food tomical and functional connections, could well be part deprivation are homeostatic processes that now seem, of a modulatory network that may regulate the incom- in part, to be regulated by RVM-NRM and PAG path- ing pain signal. On the other hand, the parabrachial ways. Likewise, symptoms of migraine that accompany nucleus is a major relay center that converges nocicep- pain, including excessive urination, and eating tive, visceral and thermoreceptive information to the (or altered feeding), the need for sleep and the tendency forebrain and the central autonomic network. to avoid physical activities, are all processes that are a Through the parabrachial nucleus, trigeminal nocicep- result of OFF cell discharge. It is therefore possible that tive inputs reach the subcortical areas of the autonomic if altered responses of RVM-NRM ON and OFF cells network including the amygdala, hypothalamus, thal- are involved in migraine pathophysiology, they would amus, the PAG and raphe nuclei (148). Activation of result in homeostatic changes, similar to those this network induces autonomic and behavioural 14 Cephalalgia 0(0) responses as well as pain modulatory effects. Some of The DRN is a complex and heterogeneous nucleus these effects such as alterations of arousal (tiredness), that has a marked topographical organization with dis- emotional responses, food intake (craving) and an tinct subdivisions receiving and sending specific projec- increase in blood pressure can be observed during the tions to several areas of the brain (160). Mainly, the premonitory, pain or postdromal phases of acute DRN receives afferents from the hypothalamus (arcu- migraine attacks (85,86). Interestingly, this network, ate nucleus, lateral, perifornical and preoptic areas), to some extent, uses endogenous CGRP as one of its amygdala, lateral , PAG, cerebral cortex and main neurotransmitters (156). In this system, CGRP (159). It sends a vast majority of has been shown to act directly through the neuronal serotonergic projections to the caudate and effects elicited after binding its receptor, as well as to substantia nigra, and to a lesser extent, to the subtha- interact with other receptor systems. In this context, lamic nucleus, , motor cortex CGRP acts on NMDA receptors and potentiates and, interestingly, to the TNC (161–164). The DRN NMDA receptor-mediated synaptic transmission also sends dopaminergic projections to the nucleus (157). Moreover, an interaction with GABAergic neu- accumbens, lateral septum and medial prefrontal rons is also described (158), highlighting the complexity cortex (165). While the DRN contains predominantly of this modulatory system. In addition, the central serotonergic neurons (166), accounting for 70% of all autonomic network involves parasympathetic neurons DRN cells in humans (167), it also contains glutama- that elicit the facial autonomic symptoms (e.g. lacrima- tergic, GABAergic and dopaminergic neuronal cell tion) during attacks of migraine or cluster headache. types (168). These neurons also co-express other neuro- Therefore, it remains to be elucidated to what extent peptides including nitric oxide (169), neuropeptide Y, the parabrachial nucleus and the neuronal activations substance P and VIP (170), among others, which have observed in animal experiments are linked to pain pro- been previously linked to migraine pathophysiology. cessing, or activation and transmission of neuronal sig- The DRN modulates pain networks through direct nals of the central autonomic network. and indirect descending projections to spinal dorsal horn neurons and through ascending projections that Midbrain regulate the activity of thalamic pain sensitive neurons (171). It is also involved in regulating waking and rapid The midbrain is the smallest brainstem component, sit- (REM) sleep (160) through hypothal- ting rostral to the pons. It is involved in linking compo- amic and LC connections. Serotonergic neurons display nents of the , such as the cerebellum, basal a regular firing pattern during waking that is reduced ganglia and , involved in the control during non-REM sleep and silent or almost silent of skeletal muscle. Indeed, the substantia nigra provides during REM sleep (172–174). This suggests a possible important inputs to the , which regulates mechanistic role between sleep-wake disruptions and voluntary movements; of particular interest in headache, as discussed above. The DRN may also be Parkinson’s disease through damage to dopaminergic involved in CSD, as DRN stimulation modulates cor- neurons. Several regions are involved in the direct control tical (175,176) and carotid blood flow (177,178), of eye movement, and a relay for auditory and visual and chronic degeneration of serotonergic neurons of information. Many structures, including the PAG and the DRN increases CSD velocity and amplitude in dorsal raphe also provide important connections between rats (179). pontine and medullary structures involved in modulating The DRN is described as having a role in mood somatosensory inputs, as well as ascending connections disorders, which are highly co-morbid with headache involved in pain processing, such as hypothalamic, thal- and migraine (180). This is especially the case with amic and cortical structures. These midbrain structures depression (181); the DRN of depressed patients pre- are believed to be important in the pathophysiology of sents a 31% decrease in overall number (182), head pain, as well as integrating inputs from other and hydroxylase (the enzyme involved in regions related to associated headache symptoms, such serotonin synthesis) immunoreactivity and mRNA as altered arousal, reward and feeding states. levels in the DRN are higher in depressed suicide vic- tims compared to controls (183). Furthermore, the dor- (DRN) somedial part of the DRN is innervated by forebrain structures involved in the regulation of states The DRN, located in the rostral pontine and caudal and is activated by a number of stress- and anxiety- , within the ventral part of the related stimuli (184). PAG, contains the largest group of serotonergic neu- Despite the low spatial resolution found in human rons within the raphe nuclei complex (Figure 6 (a)–(b)) imaging, several studies have hypothesised an import- (113,159). ant role for the DRN based on its general location Vila-Pueyo et al. 15 within activated regions. Early PET studies reported an increase in cerebral blood flow in a region that includes the DRN during spontaneous migraine attacks. This dlPAG activation persisted after effective treatment with the dlPAG 5-HT1B/1D receptor agonist, sumatriptan (185). More recently, a functional MRI study has reported an altered hypothalamic functional connectivity within the dorsal rostral pons, a region that also includes the DRN among other nuclei, which could partially explain autonomic symptoms occurring during migraine attacks, as reported above (90). A PET study has pro- Figure 7. Functional subdivisions of the central periaqueductal vided indirect support for 5-HT receptors expressed 1B gray (PAG). The PAG is thought to be divided into four functional in the DRN being temporarily downregulated during subdivisions, organised as longitudinal columns, running parallel the migraine attack, probably in response to higher with the central aqueduct (248). These four columns are the cerebral serotonin levels in the ictal phase (186). dorsomedial (dmPAG), dorsolateral (dlPAG), lateral (lPAG), and Preclinically, the DRN is activated by trigeminal ventrolateral (vlPAG). The dmPAG, lPAG and vlPAG each project nociception induced by intracranial capsaicin (92), an directly to the lower brainstem, the dlPAG does not. These dif- experimental model of trigeminal nociception and ferent columns have different functions; for instance, both the CGRP release, but it is not activated by electrical lPAG and vlPAG are involved in the control of heart rate and stimulation of the trigeminal ganglion (127). The effects blood pressure, but with opposing effects, hypertensive vs. of antimigraine drugs in the DRN have been analysed hypotensive, respectively. The PAG is also involved in the in several studies. In one, using autoradiographic tech- endogenous descending modulation of pain mechanisms, via rostral ventromedial medullary projections. Again the lPAG and niques in the cat, the highest density of specific binding 3 vlPAG appear to have opposing effects, with activation of lPAG sites of H-dihydroergotamine was found in the DRN producing an aversive response, whereas the vlPAG is thought to and the medial raphe nucleus. Both nuclei were also be . Further, within these separate subdivisions there are 3 labelled after intravenous administration of H-dihy- also differences in their response to opioids and development of droergotamine (187) and ex vivo analysis, suggesting tolerance to , and the dual role of ascending/descending that dihydroergotamine could be acting at these sites pathways, or descending only. This figure is adapted from (248). in . In fact, this was confirmed in a later study where dihydroergotamine and its main modulation of pain by cognitive influences, emotional metabolite inhibited the firing of serotoninergic neu- factors and stress. rons in the DRN within brainstem slices through its The descending modulatory system, and in particu- binding to 5-HT1A receptors (188). Furthermore, in lar the PAG-RVM pathway (Figure 8), is involved in the rat, acute, but not chronic, subcutaneous treatment all pain conditions, regardless whether the pain is of with sumatriptan or zolmitriptan induced a decrease in peripheral or central origin. As discussed in the the serotonin synthesis rates within the DRN (189), RVM-NRM section, this descending modulatory suggesting one of the possible mechanisms of action system is also involved in mechanisms beyond pain, of triptans as antimigraine drugs. Together, these to include motor function, responses to innocuous data suggest that the DRN may be involved in migraine stimuli, as well as related to homeostatic functions. pathophysiology through its descending and ascending The function of the endogenous ‘pain’ modulatory projections, but also may be involved in the therapeutic system was initially discovered in animal experiments actions of dihydroergotamine and triptans. in rats (190) and later confirmed in humans (191,192). Beyond that, these studies impressively demonstrate the Periaqueductal gray (PAG) potency of this system, allowing or endurance of severe lesions without experiencing pain. Given the The PAG (Figure 7) represents one of the most sig- importance of this mechanism, it could be expected that nificant elements of the endogenous descending mod- it would play a role in headache disorders. Raskin et al. ulatory system. Caudally, the PAG projects to the published a case series in which patients that underwent TCC and the spinal dorsal horn, predominantly electrode implantation in the PAG started to suffer through the RVM, rostrally it connects to supraspinal from headache with migrainous features, which in structures including the hypothalamus, thalamus, most cases was responsive to dihydroergotamine amygdala and (Figure 8). Given (193). The fact that PAG stimulation induced rather this neuroanatomical setting, the PAG is ideally than inhibited headache underlines the fact that the placed to modulate incoming pain and other sensory action of the PAG-RVM pathway is not unidirectional. and homeostatic signals, including the endogenous In addition, it can’t be excluded that the frequency 16 Cephalalgia 0(0)

of the electrical stimulation was decisive of the func- tional outcome of the PAG stimulation. A few years later, the hypothesis that the PAG is relevant in migraine was given significant support by the studies of Weiller et al. (185). Using PET imaging, they identi- fied specific brainstem activation during spontaneous migraine headache. They hypothesised that the mid- brain region that was activated during these spontan- eous migraine attacks was very likely the vlPAG, and these active areas remained so, even after headache resolution with abortive treatment. Based on these find- ings, a number of in vivo studies were conducted to further elucidate the role of the PAG during trigeminal activation. Electrical activation of the vlPAG inhibits Transmission stimulus-evoked nociceptive neuronal activity in the of nociceptive input TCC resulting from electrical stimulation of the dura mater (194). The same effect is observed after local injection of the GABAA antagonist bicuculline into the vlPAG (195,196). In the PAG endogenous modulation of pain signals is mediated to a large extent through opioid receptor-based mechanisms (9,197–200). In migraine, serotonin receptor agonists (5-HT1B/1D) are more effective than opioids, and for this and several other reasons the use of opioids is not recommended (201,202). These observations raised the possibility that in primary headache disorders other non-opioid receptor systems may be as important for Figure 8. Brainstem descending control of nociceptive inputs pain transmission, and that this may also apply for pain in the medullary dorsal horn. Nociceptive trigeminovascular transmission and modulation in the PAG. Given that neurons, including those that innervate the dural vasculature and triptans are highly effective in the treatment of migraine, project to the trigeminocervical complex (TCC) region, are Bartsch et al. investigated their effect in the PAG. The under the control of various endogenous descending pain mod- data showed that microinjection of the 5-HT recep- ulatory mechanisms. These impact the of the 1B/1D tor agonist naratriptan into the vlPAG inhibited nocicep- nociceptive input, and therefore the perception of pain, or not, during primary headaches, as their signal is relayed and processed tive neuronal activity in the TCC following electrical via ascending projections to hypothalamic (H), thalamic (Thal) stimulation of the dura mater (196). A similar effect was and cortical neurons. The periaqueductal gray (PAG) is indirectly observed with administration of sumatriptan into the involved in the control of pain transmission in the medullary RVM in two distinct models of visceral pain (203). dorsal horn via projections to the rostral ventromedial medulla Considering that triptans act, at least in part, through (RVM). This involves both inhibitory (green) and facilitatory (red) modulation of CGRP release, it would be conceivable projections. The PAG receives projections from the anterior that CGRP itself may also act, among other sites, cingulate, S1 somatosensory and insula cortices, the hypothal- within the PAG. This hypothesis was recently confirmed amus, and amygdala, and thus is the main output of the limbic in an animal model of migraine in which microinjection system and its involvement in the emotional and motivational of CGRP into the vlPAG facilitated nociceptive activity control of pain. The PAG also projects to the locus coeruleus within the TCC, an effect that could be reversed with the (LC), which has descending noradrenergic projections to provide control of TCC nociceptive neurons. Within the RVM there are CGRP receptor antagonists, CGRP8-37 and olcegepant also inhibitory and facilitatory projections to the TCC involved in (204). These results suggest that the pathobiology of the control of nociceptive transmission. In addition, there are migraine could potentially involve a dysfunction within separate serotoninergic projections (yellow) from the nucleus this descending modulatory system. This hypothesis is raphe magnus in the RVM involved in pain modulation. Opioid fuelled by the , which may also act at specific receptors present in the PAG and RVM further provide orexin receptors in the PAG. Orexins are synthetized descending mechanisms for endogenous pain modulation of exclusively in the hypothalamus and regulate arousal. dural-nociceptive inputs to the TCC. Finally, direct descending Beyond that, and particularly important in migraine, projection neurons from the cortex and hypothalamus also orexins modulate trigeminal nociceptive processing provide descending modulatory control nociceptive inputs to (205–207). The site of action of this modulation is not the TCC. The figure represents the rat brain. limited to the hypothalamus, as orexinergic neurons Vila-Pueyo et al. 17 project to the prefrontal cortex, thalamus and other headache as other triptans, and highly effective at subcortical areas, including the PAG (31,206,208– aborting cluster headache. So how can this be 211). In this context, Holland et al. demonstrated rationalised? that administration of orexin A into the vlPAG inhi- Either triptans and other therapeutics do not need to bits dural nociceptive responses in the TCC (206). The gain access to the brain to be effective, or they are access- exact mechanisms of this action are not entirely clar- ing the brain via another mechanism to mediate their ified, but the fact that the orexin A-induced antinoci- therapeutic effects. The only area of the brainstem not ceptive effect can be blocked by a 5-HT1B/1D receptor protected by a blood-brain barrier (BBB) is the area antagonist suggests an interaction with the triptan postrema. It sends projections to brainstem nuclei, and other receptor systems. A similar observation including the parabrachial nucleus, as well to several was made with a CB1 receptor antagonist, which hypothalamic nuclei. It is possible therapeutics are when microinjected into the vlPAG also had an accessing the brain this way, but at the moment there inhibiting effect on dural-evoked nociceptive activity is no evidence to support this. Another hypothesis is that in the TCC that could be modulated through an there is BBB disruption, particularly during migraine action at 5-HT1B/1D receptors (212). Therefore it with aura, which contributes to the spread of noxious remains unclear if these receptor systems have direct mediators in the dura mater, but also access of thera- influence on the PAG or if their effect is mediated via peutics to the brain (222). However, several recent stu- 5-HT1B/1D or perhaps even 5-HT1F receptors. dies now suggest that there is no disruption of the BBB during migraine with and without aura, and no differ- Brainstem actions of headache ence in brain permeability during migraine compared to therapies and the blood-brain their interictal state (223,224). Dihydroergotamine is barrier (BBB) also an effective migraine abortive and is able to inhibit dural-evoked central trigeminovascular neurons (225). Within this review we have highlighted how many Yet in a human binding study there was little or no headache therapeutics, such as triptans and targeting binding of 11 C-dihydroergotamine in the brain paren- CGRP and its receptors, can be used to modulate tri- chyma during nitroglycerin-induced migraine (226), geminovascular nociceptive mechanisms within the again suggesting an intact BBB. brainstem. Some of these are from preclinical studies Migraine and TAC preventive drugs, including that allow direct administration into specific nuclei. topiramate, sodium valproate, lamotrigine, and ami- While they are suggestive of a potential therapeutic triptyline, are widely accepted to cross the BBB, based mechanism of action, they are not definitive. At best on their primary use as anti-epileptics and anti- they demonstrate that certain pharmacologies are very depressants. Preclinical studies support this, as they likely relevant to the pathophysiological mechanisms are effective in models of cortical spreading depres- that underpin these headache disorders, with potential, sion (227–229) and dural-nociceptive trigeminovascu- rather than defined, mechanisms of action in thera- lar activation (230–232), which require access to peutics. The question of access to the brain for thera- discrete areas of the brain. The development of non- peutics is a long running issue, which has been vascular centrally acting novel therapeutics such as amplified since the general acceptance that vascular the -ditans, also suggest that access to the brain in mechanisms are unlikely to be as relevant to many pri- medullary regions is an effective mechanistic mary headache disorders, especially migraine and the approach (233,234). However, the development of TACs. It is relevant that probably the most commonly molecules that target CGRP and its receptors raises prescribed headache therapeutic, sumatriptan, is actu- a further issue. Olcegepant was proven to be highly ally the least brain penetrant of all the triptans (213). In effective in the abortive treatment of migraine (235), preclinical studies, sumatriptan is unable to inhibit and several preclinical studies suggest it is also able to dural-evoked activation of central trigeminovascular modulate central trigeminovascular neurons (57,236– neurons (214), or reverse dural-inflammatory mediated 238). The lipophilicity of olcegepant is not known but central sensitisation (215,216), unless the BBB is dis- it is thought to have poor brain penetrance, based on rupted (217). The more brain penetrant naratriptan a lack of effect on the cerebral vasculature (239). The (218), zolmitriptan (214) and rizatriptan (219), are recent development and reported success of CGRP- able to inhibit dural-nociceptive central trigeminovas- related antibodies (240–243) in migraine treatment, cular responses. Furthermore, zolmitriptan has been molecules considered too large to cross the BBB, sug- shown to bind to the medullary TNC and its C1 and gest they are acting in the periphery to be effective. C2 cervical extension (220). Yet, it is clear sumatriptan Despite the overwhelming data describing the import- can act via an exclusively neural mechanism (221), and ant role of brainstem nuclei in mediating symptoms of it is at least equally effective in aborting migraine primary headache, particularly the medullary 18 Cephalalgia 0(0) trigeminovascular relay, we are left in the situation arousal and sensitivity to pain, from the rest of the where we have to conclude that access to the brain body to higher centres in the brain. The role of brain- to reach brainstem areas is not obligate for either stem nuclei in headache mechanisms is to integrate acute or preventive treatment success in many pri- inputs from the periphery, such as from environmen- mary headaches. While in the main it is not clear tal stimuli and intracranial nociceptive inputs, relayed exactly where many drugs are acting to mediate through the medullary trigeminovascular system, with their therapeutic effects, what is very likely, and per- those mediated centrally from endogenous homeo- haps proven by the success of many preclinical models static stresses, such as altered arousal, feeding and in predicting therapeutic efficacy, is that modulating life stress. Due to the complex integration and pro- dural-responsive central trigeminovascular neurons cessing of information in brainstem nuclei with each and their ascending brainstem projections is likely to other, and their roles in normal function, the net lead to therapeutic success. Much of the trigemino- result of activation of this complex neural network vascular system sits outside the BBB, including the can be headache, due to an altered perception of cra- dura mater, trigeminal ganglion, the central axonal niovascular inputs. But also, there is a host of asso- trigeminal projections, as well as cranial autonomic ciated symptoms based on the nuclei involved, such as projections, modulating brainstem structures includ- a generalized increase in sensitivity to sensory inputs, ing the medullary trigeminovascular relay. Therefore, mood change, fatigue, yawning, stiffness, gastro- via these peripheral networks, it is very likely that intestinal disturbance, autonomic features, and cogni- some migraine drugs mediate their therapeutic tive phenomena. All of these can precede, accompany action to modulate brainstem networks, whereas or follow the headache (86,244). This activation might other drugs, which do pass the BBB, mediate their be mediated as a normal functional response to exter- effects more directly. nal stimuli and endogenously-mediated homeostatic stressors, or potentially through dysfunctional pro- Summary cessing within the brainstem to what are considered normal stimuli and stresses. Together, the normal Brainstem nuclei make up an important part of the function of brainstem nuclei, and their potential dys- central pain pathway related to primary headaches. function in processing neural information, is integral Their normal function is to integrate information to mechanisms related to primary headaches. related to physiological survival and reflex functions,

Article highlights . The brainstem consists of the medulla oblongata, pons, and midbrain, and includes the trigeminal nucleus, raphe nuclei, locus coeruleus, parabrachial nucleus and periaqueductal gray. . Normal brainstem function is to integrate information related to physiological survival (cardiovascular/ respiratory/autonomic control), arousal and sensitivity to pain, from the rest of the body to higher centres in the brain. . In a headache context, activation within brainstem nuclei can mediate cranial pain, and associated headache symptoms that reflect altered sensory, autonomic, affective, and cognitive processing. . Activation in the brainstem, and primary headache symptoms, might be mediated as a normal functional response to external stimuli and endogenously mediated homeostatic stressors. . Alternatively, primary headache symptoms are a consequence of dysfunctional processing within the brain- stem to what are considered normal stimuli and stresses, but which are perceived as noxious, and become a homeostatic threat.

Declaration of conflicting interests report an unrestricted grant, honoraria and travel reimburse- The authors declared the following potential conflicts ments from electroCore LLC, unrelated to the submitted of interest with respect to the research, authorship, and/ work. or publication of this article: MVP declares no conflict of interest. JH has received honoraria for consulting activities and/or serving on advisory boards as well as for speaking Funding from Allergan, Autonomic Technologies Inc. (ATI), The authors received no financial support for the research, Chordate Medical AB, Novartis and Teva. MRR and SA authorship, and/or publication of this article. Vila-Pueyo et al. 19

ORCID iD 20. Waite ME and Ashwell KWS. Trigeminal sensory Marta Vila-Pueyo http://orcid.org/0000-0003-0652-2988 system. In: Mai JK and Paxinos G (eds) The human ner- Jan Hoffmann http://orcid.org/0000-0002-2103-9081 vous system, 3rd edn. London, UK, Waltham, MA & San Diego, CA: Academic Press, 2012, pp.1110–1143. References 21. Kaube H, Keay KA, Hoskin KL, et al. Expression of 1. Kandel ER, Schwartz JH and Jessell TM. Principles of neural c-Fos-like immunoreactivity in the caudal medulla and science, 4th edn. McGraw-Hill Publishing Co. upper cervical cord following stimulation of the superior sagittal sinus in the cat. Brain Res 1993; 629: 95–102. 2. Akerman S, Holland PR and Goadsby PJ. Diencephalic and brainstem mechanisms in migraine. Nat Rev Neurosci 22. Goadsby PJ and Hoskin KL. The distribution of trigemi- 2011; 12: 570–584. novascular afferents in the nonhuman primate brain Macaca nemestrina: a c-fos immunocytochemical study. 3. May A. Cluster headache: Pathogenesis, diagnosis, and J Anat 1997; 190: 367–375. management. Lancet 2005; 366: 843–855. 23. Hoskin KL, Zagami A and Goadsby PJ. Stimulation of 4. Noseda R and Burstein R. Migraine pathophysiology: the middle meningeal artery leads to Fos expression in the Anatomy of the trigeminovascular pathway and asso- trigeminocervical nucleus: A comparative study of ciated neurological symptoms, cortical spreading depres- monkey and cat. J Anat 1999; 194: 579–588. sion, sensitization, and modulation of pain. Pain 2013; 154: S44–S53. 24. Burstein R, Yamamura H, Malick A, et al. Chemical stimulation of the intracranial dura induces enhanced 5. Akerman S and Goadsby PJ. A novel translational responses to facial stimulation in brain stem trigeminal animal model of trigeminal autonomic cephalalgias. neurons. J Neurophysiol 1998; 79: 964–982. Headache 2015; 55: 197–203. 25. Bartsch T and Goadsby PJ. Stimulation of the greater 6. Goadsby PJ, Holland PR, Martins-Oliveira M, et al. occipital nerve induces increased central excitability of Pathophysiology of migraine: A disorder of sensory pro- dural afferent input. Brain 2002; 125: 1496–1509. cessing. Physiol Rev 2017; 97: 553–622. 7. Fields HL, Basbaum AI, Clanton CH, et al. Nucleus 26. Bartsch T and Goadsby PJ. Increased responses in trige- raphe magnus inhibition of spinal cord dorsal horn neu- minocervical nociceptive neurones to cervical input after rons. Brain Res 1977; 126: 441–453. stimulation of the dura mater. Brain 2003; 126: 8. Fields HL, Bry J, Hentall I, et al. The activity of neurons 1801–1813. in the rostral medulla of the rat during withdrawal from 27. Millan MJ. Descending control of pain. Prog Neurobiol noxious heat. J Neurosci 1983; 3: 2545–2552. 2002; 66: 355–474. 9. Fields HL, Vanegas H, Hentall ID, et al. Evidence that 28. Akerman S and Goadsby PJ. and migraine: disinhibition of brain stem neurones contributes to mor- Biology and clinical implications. Cephalalgia 2007; 27: phine analgesia. 1983; 306: 684–686. 1308–1314. 10. Mason P. Deconstructing endogenous pain modulations. 29. Liu Y, Broman J and Edvinsson L. Central projections of J Neurophysiol 2005; 94: 1659–1663. the sensory innervation of the rat middle meningeal 11. Fields H. State-dependent opioid control of pain. Nat artery. Brain Res 2008; 1208: 103–110. Rev Neurosci 2004; 5: 565–575. 30. Liu Y, Broman J and Edvinsson L. Central projections of 12. Shankland WE 2nd. The trigeminal nerve. Part I: An sensory innervation of the rat superior sagittal sinus. over-view. Cranio 2000; 18: 238–248. Neuroscience 2004; 129: 431–437. 13. Shankland WE 2nd. The trigeminal nerve. Part IV: The 31. Holland PR, Akerman S and Goadsby PJ. Modulation of mandibular division. Cranio 2001; 19: 153–161. nociceptive dural input to the trigeminal nucleus caudalis 14. Shankland WE. The trigeminal nerve. Part II: The oph- via activation of the orexin 1 receptor in the rat. Eur J thalmic division. Cranio 2001; 19: 8–12. Neurosci 2006; 24: 2825–2833. 15. Shankland WE 2nd. The trigeminal nerve. Part III: The 32. Hu JW, Dostrovsky JO and Sessle BJ. Functional proper- maxillary division. Cranio 2001; 19: 78–83. ties of neurons in cat trigeminal subnucleus caudalis 16. Penfield W and McNaughton FL. Dural headache and (medullary dorsal horn). I. Responses to oral-facial nox- the innervation of the dura mater. Arch Neurol Psych ious and nonnoxious stimuli and projections to thalamus 1940; 44: 43–75. and subnucleus oralis. J Neurophysiol 1981; 45: 173–192. 17. Ray BS and Wolff HG. Experimental studies 33. Millan MJ. The induction of pain: An integrative review. on headache. Pain sensitive structures of the head and Prog Neurobiol 1999; 57: 1–164. their significance in headache. Arch Surg 1940; 41: 813–856. 34. Millan MJ. Descending control of pain. Prog Neurobiol 18. McNaughton FL and Feindel WH. Innervation of intra- 2002; 66: 355–474. cranial structures: A reappraisal. In: Rose FC (ed.) Phy- 35. Bartsch T and Goadsby PJ. Anatomy and physiology of siological aspects of clinical neurology. Oxford: Blackwell pain referral in primary and cervicogenic headache dis- Scientific Publications, 1977, pp.279–293. orders. Headache Currents 2005; 2: 42–48. 19. Marfurt CF. The central projections of trigeminal pri- 36. Headache Classification Committee of the International mary afferent neurons in the cat as determined by the Headache Society. The International Classification of tranganglionic transport of horseradish peroxidase. Headache Disorders, 3rd edition (beta version). J Comp Neurol 1981; 203: 785–798. Cephalalgia 2013; 33: 629–808. 20 Cephalalgia 0(0)

37. Fernandes G, Franco AL, Goncalves DAD, et al. Tem- 55. Nakai M, Tamaki K, Ogata J, et al. Parasympathetic poromandibular disorders, sleep bruxism, and primary cerebrovascular center of the facial nerve. Circ Res headaches are mutually associated. J Orofac Pain 2013; 1993; 72: 470–475. 27: 14–20. 56. Akerman S, Holland PR, Lasalandra MP, et al. Oxygen 38. Goncalves DAG, Speciali JG, Jales LCF, et al. Tempor- inhibits neuronal activation in the trigeminocervical com- omandibular symptoms, migraine, and chronic daily plex after stimulation of trigeminal autonomic reflex, but headaches in the population. Neurology 2009; 73: not during direct dural activation of trigeminal afferents. 645–646. Headache 2009; 49: 1131–1143. 39. Goncalves DA, Camparis CM, Speciali JG, et al. Tem- 57. Akerman S, Holland PR, Summ O, et al. A translational poromandibular disorders are differentially associated in vivo model of trigeminal autonomic cephalalgias: Ther- with headache diagnoses: A controlled study. Clin J apeutic characterization. Brain 2012; 135: 3664–3675. Pain 2011; 27: 611–615. 58. Lai T-H, Fuh J-L and Wang S-J. Cranial autonomic 40. Goncalves DA, Bigal ME, Jales LC, et al. Headache and symptoms in migraine: Characteristics and comparison symptoms of temporomandibular disorder: An epidemio- with cluster headache. J Neurol, Neurosurg Psychiatry logical study. Headache 2010; 50: 231–241. 2009; 80: 1116–1119. 41. Franco AL, Goncalves DAG, Castanharo SM, et al. 59. Amin FM, Hougaard A, Schytz HW, et al. Investigation Migraine is the most prevalent primary headache in indi- of the pathophysiological mechanisms of migraine viduals with temporomandibular disorders. J Orofac Pain attacks induced by pituitary adenylate cyclase-activating 2010; 24: 287–292. polypeptide-38. Brain 2014; 137: 779–794. 42. Fields HL, Heinricher MM and Mason P. Neurotrans- 60. Rahmann A, Wienecke T, Hansen JM, et al. Vasoactive mitters in nociceptive modulatory circuits. Ann Rev intestinal peptide causes marked cephalic vasodilation, Neurosci 1991; 14: 219–245. but does not induce migraine. Cephalalgia 2008; 28: 43. Fields HL and Heinricher MM. Anatomy and physiology 226–236. of a nociceptive modulatory system. Phil Trans Royal Soc 61. Schytz HW, Birk S, Wienecke T, et al. PACAP38 induces London B Biol Sci 1985; 308: 361–374. migraine-like attacks in patients with migraine without 44. Foo H and Mason P. Brainstem modulation of pain aura. Brain 2009; 132: 16–25. during sleep and waking. Sleep Med Rev 2003; 7: 62. Goadsby PJ and Edvinsson L. Human in vivo evidence 145–154. for trigeminovascular activation in cluster headache. 45. Mason P. Ventromedial medulla: Pain modulation and Neuropeptide changes and effects of acute attacks thera- beyond. J Comp Neurol 2005; 493: 2–8. pies. Brain 1994; 117: 427–434. 46. Ruffoli R, Giorgi FS, Pizzanelli C, et al. The chemical 63. Goadsby PJ, Edvinsson L and Ekman R. Vasoactive pep- of vagus nerve stimulation. J Chem tide release in the extracerebral circulation of humans Neuroanat 2011; 42: 288–296. during migraine headache. Ann Neurol 1990; 28: 183–187. 47. Kalia M and Sullivan JM. Brainstem projections of sen- 64. Tuka B, Szabo N, Toth E, et al. Release of PACAP-38 in sory and motor components of the vagus nerve in the rat. episodic cluster headache patients – an exploratory study. J Comp Neurol 1982; 211: 248–265. J Headache Pain 2016; 17: 69. 48. Cheyuo C, Jacob A, Wu R, et al. The parasympathetic 65. Zagami AS, Edvinsson L and Goadsby PJ. Pituitary nervous system in the quest for therapeutics. adenylate cyclase activating polypeptide and migraine. J Cereb Blood Flow Metab 2011; 31: 1187–1195. Ann Clin Transl Neurol 2014; 1: 1036–1040. 49. Henry TR. Therapeutic mechanisms of vagus nerve 66. Suzuki N, Hardebo JE and Owman C. Origins and path- stimulation. Neurology 2002; 59: S3–S14. ways of cerebrovascular vasoactive intestinal polypep- 50. May A and Goadsby PJ. The trigeminovascular system in tide-positive nerves in rat. J Cereb Blood Flow Metab humans: Pathophysiological implications for primary 1988; 8: 697–712. headache syndromes of the neural influences on the cere- 67. Burstein R, Noseda R and Borsook D. Migraine: Multi- bral circulation. J Cereb Blood Flow Metab 1999; 19: ple processes, complex pathophysiology. J Neurosci 2015; 115–127. 35: 6619–6629. 51. Spencer SE, Sawyer WB, Wada H, et al. CNS projec- 68. Akerman S, Holland PR, Lasalandra MP, et al. Oxygen tions to the pterygopalatine parasympathetic pregan- inhibits neuronal activation in the trigeminocervical com- glionic neurons in the rat: A retrograde transneuronal plex after stimulation of trigeminal autonomic reflex, but viralcellbodylabelingstudy.Brain Res 1990; 534: not via direct dural activation of trigeminal afferents. 149–169. Headache 2009; 49: 1131–1143. 52. Gray H. Anatomy of the human body. Philadelphia: Lea & 69. Hosoya Y, Matsushita M and Sugiura Y. A direct hypo- Febiger, 1918 (www.bartleby.com/107/). thalamic projection to the superior salivatory nucleus 53. Goadsby PJ. Effect of stimulation of facial nerve on neurons in the rat. A study using anterograde autoradio- regional cerebral blood flow and utilization in graphic and retrograde HRP methods. Brain Res 1983; cats. Am J Physiol 1989; 257: R517–R521. 266: 329–333. 54. Goadsby PJ. Sphenopalatine ganglion stimulation 70. Hosoya Y, Sugiura Y, Ito R, et al. Descending projec- increases regional cerebral blood flow independent of tions from the hypothalamic paraventricular nucleus to glucose utilization in the cat. Brain Res 1990; 506: the A5 area, including the superior salivatory nucleus, in 145–148. the rat. Exp Brain Res 1990; 82: 513–518. Vila-Pueyo et al. 21

71. Robert C, Bourgeais L, Arreto CD, et al. Paraventricular 89. Moulton EA, Becerra L, Johnson A, et al. Altered hypo- hypothalamic regulation of trigeminovascular mechan- thalamic functional connectivity with autonomic circuits isms involved in headaches. J Neurosci 2013; 33: and the locus coeruleus in migraine. PLoS One 2014; 9: 8827–8840. e95508. 72. Burstein R and Jakubowski M. Unitary hypothesis for 90. Schulte LH and May A. The migraine generator revis- multiple triggers of the pain and strain of migraine. ited: Continuous scanning of the migraine cycle over 30 J Comp Neurol 2005; 493: 9–14. days and three spontaneous attacks. Brain 2016; 139: 73. Dahlstrom A and Fuxe K. Localization of monoamines 1987–1993. in the lower brain stem. Experientia 1964; 20: 398–399. 91. Afridi SK, Matharu MS, Lee L, et al. A PET study 74. Aston-Jones G and Cohen JD. An integrative theory of exploring the laterality of brainstem activation in locus coeruleus- function: Adaptive gain migraine using glyceryl trinitrate. Brain 2005; 128: and optimal performance. Ann Rev Neurosci 2005; 28: 932–939. 403–450. 92. Ter Horst GJ, Meijler WJ, Korf J, et al. Trigeminal 75. Berridge CW and Waterhouse BD. The locus coeruleus- nociception-induced cerebral Fos expression in the con- noradrenergic system: Modulation of behavioral state scious rat. Cephalalgia 2001; 21: 963–975. and state-dependent cognitive processes. Brain Res 93. Tassorelli C and Joseph SA. Systemic nitroglycerin Brain Res Rev 2003; 42: 33–84. induces Fos immunoreactivity in brainstem and fore- brain structures of the rat. Brain Res 1995; 682: 167–181. 76. Bruinstroop E, Cano G, Vanderhorst VG, et al. Spinal 94. Vila-Pueyo M, Strother L, Goadsby PJ, et al. The projections of the A5, A6 (locus coeruleus), and A7 nor- role of the locus coeruleus in regulating trigeminovascu- adrenergic cell groups in rats. J Comp Neurol 2012; 520: lar nociception. Cephalalgia 2016; 36(Suppl 1): 152. 1985–2001. 95. Goadsby PJ and Duckworth JW. Low frequency stimu- 77. Samuels ER and Szabadi E. Functional neuroanatomy lation of the locus coeruleus reduces regional cerebral of the noradrenergic locus coeruleus: Its roles in the blood flow in the spinalized cat. Brain Res 1989; 476: regulation of arousal and autonomic function part II: 71–77. physiological and pharmacological manipulations and 96. Goadsby PJ, Lambert GA and Lance JW. Differential pathological alterations of locus coeruleus activity in effects on the internal and external carotid circulation of humans. Curr Neuropharmacol 2008; 6: 254–285. the monkey evoked by locus coeruleus stimulation. 78. Samuels ER and Szabadi E. Functional neuroanatomy of Brain Res 1982; 249: 247–254. the noradrenergic locus coeruleus: Its roles in the regula- 97. Goadsby PJ, Lambert GA and Lance JW. Effects of tion of arousal and autonomic function part I: principles locus coeruleus stimulation on carotid vascular resist- of functional organisation. Curr Neuropharmacol 2008; 6: ance in the cat. Brain Res 1983; 278: 175–183. 235–253. 98. Dorr AE and Debonnel G. Effect of vagus nerve stimu- 79. Takahashi K, Kayama Y, Lin JS, et al. Locus coeruleus lation on serotonergic and noradrenergic transmission. neuronal activity during the sleep-waking cycle in mice. J Pharmacol Exp Ther 2006; 318: 890–898. Neuroscience 2010; 169: 1115–1126. 99. Cunningham JT, Mifflin SW, Gould GG, et al. 80. Benarroch EE. The locus ceruleus norepinephrine system: Induction of c-Fos and DeltaFosB immunoreactivity Functional organization and potential clinical signifi- in rat brain by Vagal nerve stimulation. cance. Neurology 2009; 73: 1699–1704. 2008; 33: 1884–1895. 81. Berridge CW, Schmeichel BE and Espana RA. Noradre- 100. Chen SP, Ay I, de Morais AL, et al. Vagus nerve stimu- nergic modulation of wakefulness/arousal. Sleep Med lation inhibits cortical spreading depression. Pain 2016; Rev 2012; 16: 187–197. 157: 797–805. 82. Espana RA, Schmeichel BE and Berridge CW. Norepi- 101. Schwarz LA and Luo L. Organization of the locus coer- nephrine at the nexus of arousal, motivation and relapse. uleus-norepinephrine system. Curr Biol 2015; 25: Brain Res 2016; 1641: 207–216. R1051–R1056. 83. Holland PR. Headache and sleep: Shared pathophysio- 102. Cady RJ, Glenn JR, Smith KM, et al. Calcitonin gene- logical mechanisms. Cephalalgia 2014; 34: 725–744. related peptide promotes cellular changes in trigeminal 84. Kelman L. The triggers or precipitants of the acute neurons and implicated in peripheral and central migraine attack. Cephalalgia 2007; 27: 394–402. sensitization. Mol Pain 2011; 7: 94. 85. Giffin NJ, Lipton RB, Silberstein SD, et al. The migraine 103. Tajti J, Uddman R and Edvinsson L. Neuropeptide postdrome: An electronic diary study. Neurology 2016; localization in the ‘‘migraine generator’’ region of the 87: 309–313. human brainstem. Cephalalgia 2001; 21: 96–101. 86. Giffin NJ, Ruggiero L, Lipton RB, et al. Premonitory 104. Unger JW and Lange W. Immunohistochemical map- symptoms in migraine: An electronic diary study. ping of neurophysins and calcitonin gene-related peptide Neurology 2003; 60: 935–940. in the human brainstem and cervical spinal cord. J Chem 87. Alstadhaug K, Salvesen R and Bekkelund S. 24-hour Neuroanat 1991; 4: 299–309. distribution of migraine attacks. Headache 2008; 48: 105. Sutin EL and Jacobowitz DM. Immunocytochemical 95–100. localization of peptides and other neurochemicals in 88. Solomon GD. Circadian rhythms and migraine. Cleve the rat laterodorsal tegmental nucleus and adjacent Clin J Med 1992; 59: 326–329. area. J Comp Neurol 1988; 270: 243–270. 22 Cephalalgia 0(0)

106. Chronwall BM, DiMaggio DA, Massari VJ, et al. The 123. Oliveras JL, Redjemi F, Guilbaud G, et al. Analgesia anatomy of neuropeptide-Y-containing neurons in rat induced by electrical stimulation of the inferior centralis brain. Neuroscience 1985; 15: 1159–1181. nucleus of the raphe in the cat. Pain 1975; 1: 139–145. 107. Caberlotto L, Hurd YL, Murdock P, et al. Neurokinin 1 124. Wang X, Yu S, Dong Z, et al. The Fos expression in rat receptor and relative abundance of the short and long brain following electrical stimulation of dura mater sur- isoforms in the human brain. Eur J Neurosci 2003; 17: rounding the superior sagittal sinus changed with the 1736–1746. pre-treatment of rizatriptan benzoate. Brain Res 2011; 108. Kask A, Rago L and Harro J. -like effect of 1367: 340–346. neuropeptide Y (NPY) and NPY13-36 microinjected 125. Ellrich J, Messlinger K, Chiang CY, et al. Modulation into vicinity of locus coeruleus in rats. Brain Res 1998; of neuronal activity in the nucleus raphe magnus by the 788: 345–348. 5-HT(1)-receptor agonist naratriptan in rat. Pain 2001; 109. Vila-Pueyo M, Strother L, Page K, et al. Lasmiditan 90: 227–231. inhibits trigeminovascular nociceptive transmission. 126. Lambert GA, Hoskin KL and Zagami AS. Cortico- Cephalalgia 2016; 36(Suppl 1): 152–153. NRM influences on trigeminal neuronal sensation. 110. Reyes BA, Glaser JD and Van Bockstaele EJ. Ultra- Cephalalgia 2008; 28: 640–652. structural evidence for co-localization of corticotropin- 127. Bohar Z, Fejes-Szabo A, Tar L, et al. Evaluation of c- releasing factor receptor and mu-opioid receptor in the Fos immunoreactivity in the rat brainstem nuclei rele- rat nucleus locus coeruleus. Neurosci Lett 2007; 413: vant in migraine pathogenesis after electrical stimulation 216–221. of the trigeminal ganglion. Neurol Sci 2013; 34: 111. Zitnik GA. Control of arousal through neuropeptide 1597–1604. afferents of the locus coeruleus. Brain Res 2016; 1641: 128. Supronsinchai W, Storer RJ, Hoffmann J, et al. 338–350. GABAA receptors in the nucleus raphe magnus modu- 112. Espana RA, Reis KM, Valentino RJ, et al. Organization late trigeminal cell firing responsive to activation of cra- of hypocretin/orexin efferents to locus coeruleus and niovascular and dural afferents. J Headache Pain 2013; basal forebrain arousal-related structures. J Comp 14: S67. Neurol 2005; 481: 160–178. 129. Supronsinchai W, Hoffman J, Akerman S, et al. The 113. Hornung JP. The human raphe nuclei and the seroto- role of the orexin-2 receptor in the nucleus raphe nergic system. J Chem Neuroanat 2003; 26: 331–343. magnus on trigeminovascular nociceptive transmission. 114. Hornung JP. The neuroanatomy of the serotonergic Cephalalgia 2013; 33: 223–224. system. In: Muller C and Jacobs B (eds) Handbook of 130. Leung CG and Mason P. Physiological properties of the behavioral neurobiology of serotonin. 1st edn, Vol 21. raphe magnus neurons during sleep and waking. Academic Press, 2010, pp.51–64. J Neurophysiol 1999; 81: 584–595. 115. Lopez Costa JJ, Averill S, Saavedra JP, et al. Serotonin 131. Foo H and Mason P. Discharge of raphe magnus ON innervation of containing neurones in the rat and OFF cells is predictive of the motor facilitation spinal trigeminal nucleus. Neurosci Lett 1994; 168: evoked by repeated laser stimulation. J Neurosci 2003; 167–171. 23: 1933–1940. 116. Sim LJ and Joseph SA. Efferent projections of the 132. Foo H and Mason P. Movement-related discharge of nucleus raphe magnus. Brain Res Bull 1992; 28: 679–682. ventromedial medullary neurons. J Neurophysiol 2005; 117. Hermann DM, Luppi PH, Peyron C, et al. Afferent pro- 93: 873–883. jections to the rat nuclei raphe magnus, raphe pallidus 133. Baez MA, Brink TS and Mason P. Roles for pain mod- and reticularis gigantocellularis pars alpha demon- ulatory cells during micturition and continence. strated by iontophoretic application of choleratoxin J Neurosci 2005; 25: 384–394. (subunit b). J Chem Neuroanat 1997; 13: 1–21. 134. Foo H and Mason P. Sensory suppression during 118. Halliday GM, Li YW, Joh TH, et al. Distribution of feeding. Proc Natl Acad Sci USA 2005; 102: substance P-like immunoreactive neurons in the 16865–16869. human medulla oblongata: Co-localization with mono- 135. Kaur S, Pedersen NP, Yokota S, et al. Glutamatergic amine-synthesizing neurons. Synapse 1988; 2: 353–370. signaling from the parabrachial nucleus plays a critical 119. Halliday GM, Li YW, Oliver JR, et al. The distribution role in hypercapnic arousal. J Neurosci 2013; 33: of neuropeptide Y-like immunoreactive neurons in the 7627–7640. human medulla oblongata. Neuroscience 1988; 26: 136. Fuller P, Sherman D, Pedersen NP, et al. Reassessment 179–191. of the structural basis of the ascending arousal system. 120. Poulat P, Marlier L, Rajaofetra N, et al. 5-Hydroxytryp- J Compar Neurol 2011; 519: 933–956. tamine, substance P and thyrotropin-releasing hormone 137. Yahiro T, Kataoka N, Nakamura Y, et al. The lateral synapses in the intermediolateral cell column of the rat parabrachial nucleus, but not the thalamus, mediates thoracic spinal cord. Neurosci Lett 1992; 136: 19–22. thermosensory pathways for behavioural thermoregula- 121. Bjo¨rklund AHT and Kuhar MJ. Neuropeptide in the tion. Sci Rep 2017; 7: 5031. CNS, part II. Amsterdam: Elsevier. 138. Garfield AS, Shah Bhavik P, Madara JC, et al. A para- 122. Willis WD and Westlund KN. Neuroanatomy of the brachial-hypothalamic cholecystokinin neurocircuit pain system and of the pathways that modulate pain. controls counterregulatory responses to hypoglycemia. J Clin Neurophysiol 1997; 14: 2–31. Cell Metab 2014; 20: 1030–1037. Vila-Pueyo et al. 23

139. Bester H, Menendez L, Besson JM, et al. Spino (trige- 155. Olesen J, Iversen HK and Thomsen LL. Nitric oxide mino) parabrachiohypothalamic pathway: Electrophy- supersensitivity: A possible molecular mechanism of siological evidence for an involvement in pain migraine pain. Neuroreport 1993; 4: 1027–1030. processes. J Neurophysiol 1995; 73: 568–585. 156. Shinohara K, Watabe AM, Nagase M, et al. Essential 140. Ter Horst GJ, WJ Meijler WJ, Korf J, et al. Trigeminal role of endogenous calcitonin gene-related peptide in nociception-induced cerebral Fos expression in the con- pain-associated plasticity in the central amygdala. Eur scious rat. Cephalalgia 2001; 21: 963–975. J Neurosci 2017; 46: 2149–2160. 141. Roeder Z, Chen Q, Davis S, et al. The parabrachial 157. Okutsu Y, Takahashi Y, Nagase M, et al. Potentiation complex links pain transmission to descending pain of NMDA receptor-mediated synaptic transmission at modulation. Pain 2016; 157: 2697–2708. the parabrachial-central amygdala synapses by CGRP 142. Noseda R, Monconduit L, Constandil L, et al. Central in mice. Mol Pain 2017; 13: 1744806917709201. nervous system networks involved in the processing of 158. Lu Y-C, Chen Y-Z, Wei Y-Y, et al. Neurochemical meningeal and cutaneous inputs from the ophthalmic properties of the synapses between the parabrachial branch of the trigeminal nerve in the rat. Cephalalgia nucleus-derived CGRP-positive axonal terminals and 2008; 28: 813–824. the GABAergic neurons in the lateral capsular division 143. Edvinsson L. Tracing neural connections to pain path- of central nucleus of amygdala. Mol Neurobiol 2015; 51: ways with relevance to primary headaches. Cephalalgia 105–118. 2011; 31: 737–747. 159. Lowry CA, Evans AK, Gasser PJ, et al. Topographic 144. Liu Y, Broman J, Zhang M, et al. Brainstem and thal- organization and chemoarchitecture of the dorsal raphe amic projections from a craniovascular sensory nervous nucleus and the median raphe nucleus. In: Monti JM, centre in the rostral cervical spinal dorsal horn of rats. Pandi-Perumal SR, Jacobs BL, et al. (eds) Serotonin and Cephalalgia 2009; 29: 935–948. sleep: Molecular, Functional and Clinical Aspect. Basel, 145. Saito H, Katagiri A, Okada S, et al. Ascending projec- Switzerland: Birkshauser Verlag AG, pp.25–67. tions of nociceptive neurons from trigeminal subnucleus 160. Monti JM. The structure of the dorsal raphe nucleus caudalis: A population approach. Exper Neurol 2017; and its relevance to the regulation of sleep and wakeful- 293: 124–136. ness. Sleep Med Rev 2010; 14: 307–317. 146. Gauriau C and Bernard JF. Pain pathways and para- 161. Imai H, Steindler DA and Kitai ST. The organization brachial circuits in the rat. Exp Physiol 2002; 87: of divergent axonal projections from the midbrain 251–258. raphe nuclei in the rat. J Comp Neurol 1986; 243: 147. Moga MM, Herbert H, Hurley KM, et al. Organization 363–380. of cortical, basal forebrain, and hypothalamic afferents 162. Azmitia EC. Bilateral serotonergic projections to the to the parabrachial nucleus in the rat. J Compar Neurol dorsal hippocampus of the rat: Simultaneous localiza- 1990; 295: 624–661. tion of 3H-5HT and HRP after retrograde transport. 148. Benarroch EE. Pain-autonomic interactions. Neurol Sci J Comp Neurol 1981; 203: 737–743. 2006; 27: s130–s133. 163. Li YQ, Takada M, Matsuzaki S, et al. Identification 149. Yasui Y, Saper CB and Cechetto DF. Calcitonin gene- of periaqueductal gray and dorsal raphe nucleus neu- related peptide immunoreactivity in the visceral sensory rons projecting to both the trigeminal sensory complex cortex, thalamus, and related pathways in the rat. and forebrain structures: A fluorescent retrograde J Compar Neurol 1989; 290: 487–501. double-labeling study in the rat. Brain Res 1993; 623: 150. Yasui Y, Saper CB and Cechetto DF. Calcitonin gene- 267–277. related peptide (CGRP) immunoreactive projections 164. Kirifides ML, Simpson KL, Lin RC, et al. Topographic from the thalamus to the and amygdala in organization and neurochemical identity of dorsal the rat. J Compar Neurol 1991; 308: 293–310. raphe neurons that project to the trigeminal somatosen- 151. Missig G, Roman CW, Vizzard MA, et al. Parabrachial sory pathway in the rat. J Comp Neurol 2001; 435: nucleus (PBn) pituitary adenylate cyclase activating 325–340. polypeptide (PACAP) signaling in the amygdala: Impli- 165. Stratford TR and Wirtshafter D. Ascending dopamin- cation for the sensory and behavioral effects of pain. ergic projections from the dorsal raphe nucleus in the Neuropharmacology 2014; 86: 38–48. rat. Brain Res 1990; 511: 173–176. 152. Block CH and Hoffman GE. Neuropeptide and mono- 166. Bang SJ, Jensen P, Dymecki SM, et al. Projections and amine components of the parabrachial pontine complex. interconnections of genetically defined serotonin neu- Peptides 1987; 8: 267–283. rons in mice. Eur J Neurosci 2012; 35: 85–96. 153. Tassorelli C, Greco R, Cappelletti D, et al. Comparative 167. Baker KG, Halliday GM, Hornung JP, et al. analysis of the neuronal activation and cardiovascular Distribution, morphology and number of monoamine- effects of nitroglycerin, sodium nitroprusside and l-argi- synthesizing and substance P-containing neurons in the nine. Brain Res 2005; 1051: 17–24. human dorsal raphe nucleus. Neuroscience 1991; 42: 154. Tassorelli C, Joseph SA and Nappi G. Reciprocal cir- 757–775. cuits involved in nitroglycerin-induced neuronal activa- 168. Soiza-Reilly M and Commons KG. Unraveling the tion of autonomic regions and pain pathways: A double architecture of the dorsal raphe synaptic neuropil immunolabeling and tract-tracing study. Brain Res 1999; using high-resolution neuroanatomy. Front Neural 842: 294–310. Circuits 2014; 8: 105. 24 Cephalalgia 0(0)

169. Johnson MD and Ma PM. Localization of NADPH 186. Deen M, Hansen HD, Hougaard A, et al. Low 5-HT1B diaphorase activity in monoaminergic neurons of the receptor binding in the migraine brain: A PET study. rat brain. J Comp Neurol 1993; 332: 391–406. Cephalalgia 2018; 38: 519–527. 170. Smith GS, Savery D, Marden C, et al. Distribution of 187. Goadsby PJ and Gundlach AL. Localization of 3H-dihy- messenger RNAs encoding enkephalin, substance P, droergotamine-binding sites in the cat central nervous somatostatin, galanin, vasoactive intestinal polypeptide, system: Relevance to migraine. Ann Neurol 1991; 29: 91–94. neuropeptide Y, and calcitonin gene-related peptide in 188. Hanoun N, Saurini F, Lanfumey L, et al. Dihydroergo- the midbrain periaqueductal grey in the rat. J Comp tamine and its metabolite, 8’-hydroxy-dihydroergota- Neurol 1994; 350: 23–40. mine, as 5-HT1A receptor agonists in the rat brain. Br 171. Wang QP and Nakai Y. The dorsal raphe: An important J Pharmacol 2003; 139: 424–434. nucleus in pain modulation. Brain Res Bull 1994; 34: 189. Dobson CF, Tohyama Y, Diksic M, et al. Effects of 575–585. acute or chronic administration of anti-migraine drugs 172. Trulson ME and Jacobs BL. Raphe unit activity in sumatriptan and zolmitriptan on serotonin synthesis in freely moving cats: Correlation with level of behavioral the rat brain. Cephalalgia 2004; 24: 2–11. arousal. Brain Res 1979; 163: 135–150. 190. Reynolds DV. Surgery in the rat during electrical anal- 173. McGinty DJ and Harper RM. Dorsal raphe neurons: gesia induced by focal brain stimulation. Science 1969; Depression of firing during sleep in cats. Brain Res 164: 444–445. 1976; 101: 569–575. 191. Mayer DJ. Analgesia produced by electrical stimulation 174. Sakai K and Crochet S. Differentiation of presumed of the brain. Prog Neuropsychopharmacol Biol serotonergic dorsal raphe neurons in relation to behav- Psychiatry 1984; 8: 557–564. ior and wake-sleep states. Neuroscience 2001; 104: 192. Young RF and Brechner T. Electrical stimulation of the 1141–1155. brain for relief of intractable pain due to cancer. Cancer 175. Underwood MD, Bakalian MJ, Arango V, et al. Regu- 1986; 57: 1266–1272. lation of cortical blood flow by the dorsal raphe nucleus: 193. Raskin NH, Hosobuchi Y and Lamb S. Headache may Topographic organization of cerebrovascular regulatory arise from perturbation of brain. Headache 1987; 27: regions. J Cereb Blood Flow Metab 1992; 12: 664–673. 416–420. 176. Adams RW, Lambert GA and Lance JW. Stimulation 194. Knight YE and Goadsby PJ. The periaqueductal grey of brainstem nuclei in the cat: Effect on neuronal activ- matter modulates trigeminovascular input: A role in ity in the primary visual cortex of relevance to cerebral migraine? Neuroscience 2001; 106: 793–800. blood flow and migraine. Cephalalgia 1989; 9: 107–118. 195. Knight YE, Bartsch T and Goadsby PJ. Trigeminal 177. Goadsby PJ, Piper RD, Lambert GA, et al. Effect of antinociception induced by bicuculline in the periaque- stimulation of nucleus raphe dorsalis on carotid blood ductal gray (PAG) is not affected by PAG P/Q-type flow. I. The monkey. Am J Physiol 1985; 248: calcium channel blockade in rat. Neurosci Lett 2003; R257–R262. 336: 113–116. 178. Goadsby PJ, Piper RD, Lambert GA, et al. Effect of 196. Bartsch T, Knight YE and Goadsby PJ. Activation of 5- stimulation of nucleus raphe dorsalis on carotid blood HT(1B/1D) receptor in the periaqueductal gray inhibits flow. II. The cat. Am J Physiol 1985; 248: R263–R269. nociception. Ann Neurol 2004; 56: 371–381. 179. Cui Y, Li QH, Yamada H, et al. Chronic degeneration 197. Fields HL, Heinricher MM and Mason P. Neurotrans- of dorsal raphe serotonergic neurons modulates cortical mitters in nociceptive modulatory circuits. Ann Rev spreading depression: A possible pathophysiology of Neurosci 1991; 14: 219–245. migraine. J Neurosci Res 2013; 91: 737–744. 198. Fields H. State-dependent opioid control of pain. Nat 180. Amoozegar F. Depression comorbidity in migraine. Int Rev Neurosci 2004; 5: 565–575. Rev Psychiatry 2017: 1–12. 199. Fields HL and Heinricher MM. Anatomy and physi- 181. Michelsen KA, Schmitz C and Steinbusch HW. The ology of a nociceptive modulatory system. Phil Trans dorsal raphe nucleus – from silver stainings to a role Royal Soc London B, Biol Sci 1985; 308: 361–374. in depression. Brain Res Rev 2007; 55: 329–342. 200. Harasawa I, Johansen JP, Fields HL, et al. Alterations 182. Baumann B, Bielau H, Krell D, et al. Circumscribed in the rostral ventromedial medulla after the selective numerical deficit of dorsal raphe neurons in mood dis- ablation of -opioid receptor expressing neurons. Pain orders. Psychol Med 2002; 32: 93–103. 2016; 157: 166–173. 183. Boldrini M, Underwood MD, Mann JJ, et al. More 201. Tepper SJ. Opioids should not be used in migraine. tryptophan hydroxylase in the brainstem dorsal raphe Headache 2012; 52: 30–34. nucleus in depressed suicides. Brain Res 2005; 1041: 202. Holland S, Fanning KM, Serrano D, et al. Rates and 19–28. reasons for discontinuation of triptans and opioids in 184. Lowry CA, Hale MW, Evans AK, et al. Serotonergic episodic migraine: Results from the American Migraine systems, anxiety, and affective disorder: Focus on the Prevalence and Prevention (AMPP) study. J Neurol Sci dorsomedial part of the dorsal raphe nucleus. Ann N 2013; 326: 10–17. Y Acad Sci 2008; 1148: 86–94. 203. Vera-Portocarrero LP, Ossipov MH, King T, et al. 185. Weiller C, May A, Limmroth V, et al. Brain stem acti- Reversal of inflammatory and noninflammatory visceral vation in spontaneous human migraine attacks. Nat pain by central or peripheral actions of sumatriptan. Med 1995; 1: 658–660. Gastroenterology 2008; 135: 1369–1378. Vila-Pueyo et al. 25

204. Pozo-Rosich P, Storer R, Charbit A, et al. Periaqueduc- 219. Cumberbatch MJ, Hill RG and Hargreaves RJ. Riza- tal gray calcitonin gene-related peptide modulates triptan has central antinociceptive effects against durally trigeminovascular neurons. Cephalalgia 2015; 35: evoked responses. Eur J Pharmacol 1997; 328: 37–40. 1298–1307. 220. Goadsby PJ and Knight YE. Direct evidence for central 205. Bartsch T, Levy MJ, Knight YE, et al. Differential sites of action of zolmitriptan (311C90): An autoradio- modulation of nociceptive dural input to [hypocretin] graphic study in cat [see comments]. Cephalalgia 1997; orexin A and B receptor activation in the posterior 17: 153–158. hypothalamic area. Pain 2004; 109: 367–378. 221. Hoskin KL, Kaube H and Goadsby PJ. Sumatriptan 206. Holland PR, Akerman S, Lasalandra MP, et al. can inhibit trigeminal afferents by an exclusively Antinociceptive effects of orexin A in the vlPAG are neural mechanism. Brain 1996; 119: 1419–1428. blocked by 5-HT1B/1D receptor antagonism. 222. Pietrobon D and Moskowitz MA. Pathophysiology of Headache 2008; 48: S6. migraine. Annu Rev Physiol 2013; 75: 365–391. 207. Hoffmann J, Supronsinchai W, Akerman S, et al. Evi- 223. Amin FM, Hougaard A, Cramer SP, et al. Intact blood- dence for orexinergic mechanisms in migraine. Neurobiol brain barrier during spontaneous attacks of migraine Dis 2015; 74: 137–143. without aura: A 3T DCE-MRI study. Eur J Neurol 208. Smart D and Jerman JC. The physiology and 2017; 24: 1116–1124. pharmacology of the orexins. Pharm Ther 2002; 94: 224. Hougaard A, Amin FM, Christensen CE, et al. 51–61. Increased brainstem perfusion, but no blood-brain bar- 209. Holland PR, Akerman S and Goadsby PJ. Orexin 1 rier disruption, during attacks of migraine with aura. receptor activation attenuates neurogenic dural Brain 2017; 140: 1633–1642. vasodilation in an animal model of trigeminovascular 225. Hoskin KL, Kaube H and Goadsby PJ. Central activa- nociception. J Pharmacol Exp Ther 2005; 315: tion of the trigeminovascular pathway in the cat is inhib- 1380–1385. ited by dihydroergotamine. A c-Fos and 210. Holland P and Goadsby PJ. The hypothalamic orexiner- electrophysiological study. Brain 1996; 119: 249–256. gic system: Pain and primary headaches. Headache 2007; 226. Schankin CJ, Maniyar FH, Seo Y, et al. Ictal lack of 47: 951–962. binding to brain parenchyma suggests integrity of the 211. Peyron C, Tighe DK, van den Pol AN, et al. Neurons blood-brain barrier for 11C-dihydroergotamine during containing hypocretin (orexin) project to multiple neur- glyceryl trinitrate-induced migraine. Brain 2016; 139: onal systems. J Neurosci 1998; 18: 9996–10015. 1994–2001. 212. Akerman S, Holland PR, Lasalandra MP, et al. Activa- 227. Akerman S and Goadsby PJ. Topiramate inhibits cor- tion of the CB1 receptor in the vlPAG inhibits nocicep- tical spreading depression in rat and cat: Impact in tive dural inputs in the TCC via neurons that are migraine aura. Neuroreport 2005; 16: 1383–1387. modulated by the 5-HT1B/1D receptor. Headache 228. Ayata C, Jin H, Kudo C, et al. Suppression of cortical 2008; 48: S1–S67. spreading depression in migraine prophylaxis. Ann 213. Tfelt-Hansen P, De Vries P and Saxena PR. Triptans in Neurol 2006; 59: 652–661. migraine: A comparative review of pharmacology, 229. Bogdanov VB, Multon S, Chauvel V, et al. Migraine pharmacokinetics and efficacy. Drugs 2000; 60: preventive drugs differentially affect cortical spreading 1259–1287. depression in rat. Neurobiol Dis 2011; 41: 430–435. 214. Hoskin KL and Goadsby PJ. Comparison of more and 230. Akerman S and Romero-Reyes M. Targeting the central less lipophilic serotonin (5HT1B/1D) agonists in a projection of the dural trigeminovascular system for model of trigeminovascular nociception in cat. Exp migraine prophylaxis. J Cereb Blood Flow Metab 2017: Neurol 1998; 150: 45–51. 271678X17729280. 215. Levy D, Jakubowski M and Burstein R. Disruption of 231. Andreou AP and Goadsby PJ. Topiramate in the treat- communication between peripheral and central trigemi- ment of migraine: A kainate (glutamate) receptor antag- novascular neurons mediates the antimigraine action of onist within the trigeminothalamic pathway. 5HT 1B/1D receptor agonists. Proc Natl Acad Sci USA Cephalalgia 2011; 31: 1343–1358. 2004; 101: 4274–4279. 232. Storer RJ and Goadsby PJ. Topiramate inhibits trige- 216. Burstein R and Jakubowski M. Analgesic triptan action minovascular neurons in the cat. Cephalalgia 2004; 24: in an animal model of intracranial pain: A race against 1049–1056. the development of central sensitization. Ann Neurol 233. Farkkila M, Diener HC, Geraud G, et al. Efficacy and 2004; 55: 27–36. tolerability of lasmiditan, an oral 5-HT(1F) receptor 217. Kaube H, Hoskin KL and Goadsby PJ. Inhibition by agonist, for the acute treatment of migraine: A phase 2 sumatriptan of central trigeminal neurones only after randomised, -controlled, parallel-group, dose- blood-brain barrier disruption. Br J Pharmacol 1993; ranging study. Lancet Neurol 2012; 11: 405–413. 109: 788–792. 234. Vila-Pueyo M. Targeted 5-HT1F therapies for migraine. 218. Goadsby PJ and Knight YE. Inhibition of trigeminal Neurotherapeutics: the journal of the American neurones after intravenous administration of naratrip- Society for Experimental NeuroTherapeutics 2018; 15: tan through an action at 5-hydroxy-tryptamine 291–303. (5-HT(1B/1D)) receptors. Br J Pharmacol 1997; 122: 235. Olesen J, Diener HC, Husstedt IW, et al. Calcitonin 918–922. gene-related peptide receptor antagonist BIBN 4096 26 Cephalalgia 0(0)

BS for the acute treatment of migraine. N Engl J Med of migraine: A phase 2, randomised, double-blind, 2004; 350: 1104–1110. placebo-controlled study. Lancet Neurol 2014; 13: 236. Storer RJ, Akerman S and Goadsby PJ. Calcitonin 885–892. gene-related peptide (CGRP) modulates nociceptive tri- 242. Bigal ME, Dodick DW, Rapoport AM, et al. Safety, geminovascular transmission in the cat. Br J Pharmacol tolerability, and efficacy of TEV-48125 for preventive 2004; 142: 1171–1181. treatment of high-frequency episodic migraine: A multi- 237. Fischer MJ, Koulchitsky S and Messlinger K. The non- centre, randomised, double-blind, placebo-controlled, peptide calcitonin gene-related peptide receptor antag- phase 2b study. Lancet Neurol 2015; 14: 1081–1090. onist BIBN4096BS lowers the activity of neurons with 243. Sun H, Dodick DW, Silberstein S, et al. Safety and effi- meningeal input in the rat spinal trigeminal nucleus. cacy of AMG 334 for prevention of episodic migraine: A J Neurosci 2005; 25: 5877–5883. randomised, double-blind, placebo-controlled, phase 2 238. Koulchitsky S, Fischer MJ and Messlinger K. Calcito- trial. Lancet Neurol 2016; 15: 382–390. nin gene-related peptide receptor inhibition reduces 244. Schoonman GG, Evers DJ, Terwindt GM, et al. The neuronal activity induced by prolonged increase in prevalence of premonitory symptoms in migraine: A nitric oxide in the rat spinal trigeminal nucleus. questionnaire study in 461 patients. Cephalalgia 2006; Cephalalgia 2009; 29: 408–417. 26: 1209–1213. 239. Tvedskov JF, Tfelt-Hansen P, Petersen KA, et al. 245. Akerman S, Holland PR and Hoffmann J. Pearls and CGRP receptor antagonist olcegepant (BIBN4096BS) pitfalls in experimental in vivo models of migraine: does not prevent glyceryl trinitrate-induced migraine. Dural trigeminovascular nociception. Cephalalgia Cephalalgia 2010; 30: 1346–1353. 2013; 33: 577–592. 240. Dodick DW, Goadsby PJ, Silberstein SD, et al. Safety 246. Wilkinson J. Neuroanatomy for medical students, 3rd and efficacy of ALD403, an antibody to calcitonin gene- edn. London: Butterworth-Heinemann, 1998. related peptide, for the prevention of frequent episodic 247. Paxinos G and Franklin KBJ. The mouse brain in stereo- migraine: A randomised, double-blind, placebo-con- taxic coordinates, 2nd edn. London: Academic Press, trolled, exploratory phase 2 trial. Lancet Neurol 2014; 2003. 13: 1100–1107. 248. Carrive P and Morgan MM. Periaqueductal Gray. 241. Dodick DW, Goadsby PJ, Spierings EL, et al. Safety In: Mai JK and Paxinos G (eds) The human nervous and efficacy of LY2951742, a monoclonal antibody to system, 3rd edn. London, UK, Waltham, MA & San calcitonin gene-related peptide, for the prevention Diego, CA: Academic Press, 2012, pp.368–400.