Complexity of migraine

Migraine is a chronic neurological condition that manifests with recurrent attacks of severe cranial pain associated to sensorial (photo, phono, and osmophobia) and gastrointestinal (nausea, repeated vomiting) symptoms. These recurrent attacks reflect the involvement of a complex set of brain structures: cortical, subcortical, and brainstem areas that regulate autonomic, affective, cognitive, and sensory functions. Clinical and neurophysiological studies point to an important role for abnormal nociceptive processing, altered habituation of somatosensory stimuli, central sensitization, and changes in cerebrovascular and immune functions [1].

Migraine can be viewed as the more or less frequent recurrence of a brain network abnormality, where several nodes of which—cortical and subcortical (hypothalamus, thalamus, trigeminocervical complex, hypothalamus, and thalamus) are central to migraine pathophysiology.

Genetics plays an important role in the preparation of migraine terrain; however, until now no convincing evidence has suggested specific genetic alterations with large effect sizes, even though a multitude of genomic loci have been associated to migraine, to configure a condition of predisposition to a generalized neuronal excitability [2, 3].

Although migraine subjects function normally interictally, migraine brain differs from non migraine brain in many aspects also interictally. This is supported by several functional neuroimaging studies as well as by neurophysiological investigations. Functional neuroimaging studies have shown that interictally the migraine brain has stronger activation in pain-facilitating regions and hypoactivation in pain-inhibiting regions [4]. It seems that the migraine brain is also morphologically different from the normal brain, as suggested by the reduced gray matter volume in pain processing areas, such as the anterior cingulate cortex, amygdala, insula, operculum, and the frontal, temporal, and precentral gyri, and the increased gray matter volume in the caudate nuclei bilaterally in migraineurs with high-frequency attacks as compared with patients with low frequency [5, 6]. Neurophysiological studies have suggested that the migraine brain is poorly wired in terms of the protective mechanism represented by the habituation to repeated stimuli [7].

The extreme complexity of migraine pathophysiology—and the diffuse involvement of brain and neruvascular systems—has been partly unveiled through the adoption of experimental models. These are not capable of mimicking the entire array of structures involved anatomically and functionally, but still have allowed to identify and understand several components of migraine, thus leading to tremendous advances in the knowledge of migraine neurobiology. One for all is represented by the recent availability of the first target therapy for the prevention of migraine: the monoclonal antibodies targeting calcitonin gene-related peptide (CGRP) [8].

Several experimental models of migraine pain have been proposed and applied to investigate the mechanisms underlying headache pathophysiology. Most of them are focused on trigeminal sensory processing. Human models, associated to advanced imaging techniques, are useful to identify potential biomarkers and to improve the treatment. Advantages and drawbacks of each model should be considered to evaluate the predictive feature of antimigraine activity in the clinical setting, information about tolerability profile of drugs, neuro-anatomy of the structures (neuronal and vascular) involved in migraine, and non-pain associated symptoms. Here, we briefly discuss the preclinical and clinical research impact in the migraine field via a description of the relevant experimental models used, their utility in clinical setting, and therapeutic targets individuation.

Animal models of migraine

Plain vascular models: in vitro and in vivo

Within in vitro studies on blood vessels, vascular segments are mounted in organ baths and contraction or relaxation is measured isometrically to determine the potency (pEC50) and efficacy (Emax) of a potential antimigraine agent. In vitro models [9,10,11,12] (Table 1) provide the possibility of studying drug–receptor interactions and detailed pharmacological analysis that can be performed on multiple vessels segments in parallel.

Table 1 In vitro animal vascular models
Full size table

In vivo vascular models are based on the involvement of vascular and neuronal components involved in disease and, although not including all migraine pathophysiological features, they contributed to the development of new antimigraine drugs [13,14,15,16,17,18,19,20,21,22] (Table 2).

Table 2 In vivo animal vascular models
Full size table

Neurovascular models: in vitro and vivo

Neurogenic inflammation (vasodilatation and plasma protein extravasation) within cephalic tissues has been suggested as a potential mechanism of migraine pathogenesis. Numerous clinically effective abortive antimigraine drugs, such as ergots and triptans, have been shown to reduce the release of neuropeptides, such as CGRP and Substance P, and thus to inhibit neurogenic plasma protein extravasation [23].

In vitro neurovascular models

A trigeminal ganglion-skull cavity preparation, hemi-skull Preparation, has been developed to preserve some degree of trigeminal/meningeal interface and used as neurovascular model in vitro. In the hemi-skull preparation the nervus spinosus (branch of the V3 branch of the trigeminal nerve) is cleaned to remove the dura mater and prepared for placement into the recording system. This nerve innervates a region of the medial meningeal artery, a key structure for migraine pain [24]. This model overcomes several limitations of the in vivo models and allows to directly record the activity in nociceptive terminals under conditions of adequate drug concentration control and without application of concomitant anesthesia.

In vivo neurovascular models

Stimulation of the trigeminal ganglion and plasma protein extravasation

The trigeminal ganglion of anesthetized animals can be electrically stimulated using bipolar electrodes inserted stereotactically. Trigeminal ganglion neurons are activated with low-frequency ( ̴ 5-Hz) stimulation [25, 26], which induces plasma protein extravasation in the dura mater detected by radiolabeled albumin given intravenously prior to stimulation [27]. Protein leakage also occurs in extracranial tissues [28]. The model induces tissue-specific changes in the meninges and cerebral activation by c-Fos protein expression, a phenomenon that is inhibited by antimigraine drugs [29]. Stimulation of the trigeminal ganglion, however, failed to induce uniform activation of the brainstem nuclei related to migraine, although it activated the descending pain modulatory system [30]. Other studies have used the prolonged stimulation of the trigeminal ganglion (approximately 30 min) to induce morphological changes, whereas a shorter stimulation (3–5 min) causes the peripheral release of CGRP, a phenomenon that was inhibited by triptans and dihydroergotamine [31]. The repetitive electrical stimulation of the trigeminal ganglion, an experimental paradigm simulating migraine chronification [32], causes allodynia. Additionally, with this model it was possible to detect a direct link between pituitary adenylate cyclase-activating polypeptide (PACAP, another vasoactive neuropeptide) and the kynurenine system during trigeminal vascular system activation [33].

It is from the late 1980s the demonstration that a specific 5-HT receptor subtype mediates sumatriptan effects in the trigeminal ganglion stimulation model [34] and results obtained with the electrical stimulation of the trigeminal ganglion using diverse 5-HT agonists and antagonists showed that sumatriptan effect were most evident by 5-HT1B/D activation [31]. It must be noted that the drawbacks of this model are the generation of inflammatory responses both locally and in the dura mater due to the insertion of electrodes into the brain parenchyma.

Stimulation of the superior sagittal sinus

Electrical stimulation of meningeal nerve terminals, innervating the superior sagittal sinus (SSS) [35], the transverse sinus [36], or middle meningeal arteries [37], have also been used as preclinical model of migraine. The SSS is one of the sources of cephalic pain; indeed, its stimulation in humans causes pain referred to the head. SSS stimulation has been performed in rats, cats, and nonhuman primates to evaluate mechanisms underlying cephalic pain [38]. Data gathered with this model showed central actions of antimigraine drugs such as ergotamine and sumatriptan, but also of acetylsalicylic acid on transmission of trigeminal nociceptive input in the brainstem [39]. SSS stimulation has been used to investigate the effect of a variety of drugs on c-Fos protein expression [40], a neuronal marker of activation. For instance, a significant reduction of SSS stimulation-evoked c-Fos expression after MK-801 administration was reported in rat, indicating a role for glutamate in neurotransmission within the trigeminocervical complex [41].

Electrical and chemical stimulation of the meningeal nerve terminals

Electrical stimulation of the dura mater with stimuli that are able to activate thin myelinated and unmyelinated nerve fibers increases the meningeal blood flow in the rat. This change is attenuated by 5-HT1 receptor agonists and abolished by a CGRP receptor antagonist [42, 43].

Direct stimulation of primary sensory neurons supplying meninges has been obtained through infusion of irritant substances (capsaicin, carrageenan) on the meninges via microcannulas located into the cisterna magna of anesthetized rats, mice, or guinea pigs [44, 45]. These substances induce c-Fos protein expression within both sides of the trigeminal nucleus caudalis (TNC) 2 h after the administration. The limitation of this model is the lack of intra-animal control, since lateralization of c-Fos protein expression within the TNC cannot be explored and may be associated with possible damage caused by capsaicin [46].

Local application or infusion of inflammatory or algesic substances onto the dura mater or chemical stimulation of the dural receptive fields in rat causes hypersensitivity to mechanical and thermal stimulation together with direct activation of the trigeminal ganglion [47, 48]. The application of an inflammatory soup to the dura mater alters additional neural networks indirectly related to the primary nociceptive pathways via the spinal cord to the thalamus and cortex [49]. This inflammatory solution elicits both facial and plantar allodynia that can be reversed by sumatriptan and CGRP8–37 [50].

In mice, dural administration of allyl-isothiocyanate (a TRPA1 agonist), low pH, interleukin-6, or inflammatory soup lead to cephalic/extracephalic allodynia, that can be reversed by systemic injection of sumatriptan [51]. Repetitive infusion of inflammatory soup (histamine, bradykinin, serotonin, prostaglandin E2) induces a chronic periorbital hypersensitivity in rat that continues for up to 3 weeks. This condition is reversed by propranolol treatment via blocking the chronic sensitization of descending pain controls and preventing the central sensitization within the trigeminocervical complex [52]. Moreover, the repeated dura mater inflammatory soup infusion causes depression and anxiety like behaviors [53]. Migraine-like pathological changes and abnormal behaviors in conscious rhesus monkeys have been also reported after chronic inflammatory soup infusion. Higher expression levels of c-Fos, nNOS, and CGRP were found in various brain structures, including the trigeminal nucleus caudalis (TNC), thalamus, hypothalamus, midbrain, pons, and other areas involved in pain perception after infusion [54]. In selecting the present model, the researcher must consider that upstream events leading to trigeminal activation are bypassed and the chemical cocktail utilized requires cautious control to prevent supramaximal stimulation.

Nitric oxide donors

Nitric oxide (NO) donors, including nitroglycerin (NTG), have appeared as the most noticeable exogenous algogenic substances to date in the field of migraine research [55]. The animal and human models based on NTG administration have proved over the years to be reliable tools for investigating migraine mechanisms and have provided a wealth of data [56].

Systemic NTG via intraperitoneal administration (10 mg/kg) or intravenous infusion (4 μg/kg/min, for 20 min) in rodents induces neuronal activation in several brain nuclei, such as TNC, that are relevant for migraine pain [57], as well as behavioral nocifensive response. Such effects can be abolished by anti-migraine drugs [58, 59]. NTG-induced changes in central and/or peripheral neurotransmission are associated with trigeminal and spinal hyperalgesic state [60], which have been reported also in migraine subjects [61]. The abovementioned model, with a single NTG administration, is used to mimic the episodic migraine condition; whereas a repeated and intermittent treatment with NTG (10 mg/kg or 5 mg/Kg i.p.) experimentally reproduces the condition of chronic migraine. Within this experimental setting of chronic migraine, NTG induces sustained mechanical hyperalgesia, light hypersensitivity, and hypoactivity [62].

The ability of the NTG model to act at both the neuronal and vascular compartments has provided a relevant amount of evidences for better understanding of the pathophysiology of migraine attacks, in which NO and other mediators play a pivotal role (for extensive information see for review Demartini et al. [56]). Although highly attractive, several variables need to be carefully controlled when adopting the NTG based animal model for the study of the trigeminovascular system: (1) NTG dose and route of administration; (2) NTG vehicle; (3) choice of the time of observation, and (4) differences in rodent species.

Another NO donor worth mention is sodium nitroprusside (SNP). When administered intraperitoneally (4 mg/Kg) it induces in rat neuronal activation limited to the areas involved in the neuroendocrine control [63]. By contrast, when infused slowly over 2 h it causes a persistent increase in the neuronal activity of the TNC, which can be inhibited by a high dose of the CGRP antagonist olcegepant (900 μg/kg, i.v) [64].

Others algogenic substances

CGRP has been shown to trigger photophobia, periorbital hypersensitivity, and spontaneous pain behaviors in rodents [65,66,67]. Intravenous CGRP facilitated vibrissal responses, which seemed to suggest that CGRP-induced vasodilation activating the primary afferent meningeal nociceptors [68]. However, this observation was not supported by electrophysiological studies [69] when CGRP was administered via dural infusion. Whereas CGRP failed to increase c-Fos and Zif268 (neuronal pain markers) expression in the TNC of rat [70]. It must be noted that within this model, different results related to pain sensitization (hyperalgesia and allodynia) are obtained depending on the route of CGRP administration.

Pituitary adenylate cyclase-activating peptide (PACAP) is a neuropeptide implicated in a wide range of functions, such as nociception and primary headaches. PACAP infusion in rodents causes a delayed sensitization of trigeminovascular nociceptive processing [71]; subcutaneous PACAP injection in the periorbital area (10 μl/site) elicits periorbital mechanical sensitivity, which was attenuated by PACAP receptor antagonist, PACAP6-38. Moreover, PAPAP induces vasodilation in vivo suggesting that this effect might be mediated through degranulation of mast cells. This latter issue is confirmed by in vitro observations [72].

Models of cortical spreading depression

Cortical spreading depression (CSD) is supposed to mimic the underlying mechanism of migraine aura. CSD can be induced in animals by either chemical (superfusion of potassium chloride solution), pinprick, or electrical stimulation over cortex surface. CSD results in an increase in the extracellular (K+) and intracellular (Na + and Ca2+) ions and neurotransmitter (glutamate). CSD provokes the expression of c-fos protein-like immunoreactivity within neurons of the TNC via trigemino-vascular mechanisms, and it is a useful tool for testing current and novel prophylactic drugs (see Costa et al. [73] for review).

CSD induction showed similar transcriptomic profiles with and without drug treatment in cortex, involving genes related to hormone stimulus, apoptosis, synaptic transmission, and interleukin signaling. Glutamate NMDA receptors, voltage-dependent sodium channels, and CGRP play roles in CSD agents [74].

Human models

In vitro human models

Besides the animal in vitro models, also specimens of human origin have been used in the attempt to improve the understanding of antimigraine drug-induced effects. For instance, CGRP has been widely studied by using in vitro human models; the development and characterization of CGRP receptor antagonists, such as olcegepant, were performed on human and also other species arteries [75,76,77,78]. Table 3 summarizes in vitro human models.

Table 3 In vitro human models
Full size table

In vivo human models of migraine

The animal models of migraine are critical to understanding the pathobiology of migraine and the factors involved such as the trigeminovascular system, CSD, central and peripheral nociception, localization of specific receptors, and key mediators. On the other hand, the human models can be used to test whether endogenous signaling molecules or other putative trigger factors may induce migraine-like attacks and are therefore suitable for investigating the mechanisms underlying spontaneous migraine attacks.

The first human migraine model was developed using the oral administration [79, 80] or i.v. infusion of NTG [81] to provoke migraine attacks without aura in migraine patients. To date, other human models are currently used, e.g., intravenous infusion of CGRP [82] and PACAP38 [83]; however, differently from the NTG model, for these models the comparative analysis with their pre-clinical counterparts is still limited due to the scarce use in animal experimental models.

NTG human model

Infusion or sublingual administration of NTG causes a biphasic response in migraineurs with an immediate, non-specific headache followed, hours later, by a migraine-like attack [79,80,81], with intensity and characteristics that are reminiscent of the patient’s own spontaneous attacks without aura. The reliability of the NTG model resides in its ability to reproduce headache attacks with features that are reminiscent of the spontaneous migraine attack. For instance, NTG in migraine patients induces premonitory symptoms together with the activation of specific brain areas (see for review Demartini et al. [56]).

CGRP human model

Intravenous CGRP infusion triggers delayed migraine-like attacks in patients with migraine without aura. Like NTG, CGRP is able to induce both an immediate and a delayed headache, although bearing generally milder and less frequently specific migraine features (see Ashina et al. [82] for review). Infusion of CGRP in healthy volunteers showed that the CGRP-induced headache was prevented by pretreatment with the olcegepant, but not abolished by sumatriptan [84].

Other vasoactive peptides

Besides CGRP, also, the vasoactive peptides VIP and PACAP may trigger migraine attacks in humans. Systemic administration of VIP induced mild headaches for a short time in healthy volunteers. Although VIP is able to induce a marked dilation of cranial arteries, it does not trigger a migraine-like attack in migraine sufferers [85], while PACAP38 infusion does [86].

Informative differences and similarities of the mostly used human models, NTG and CGRP, with pre-clinical models are illustrated in Table 4.

Table 4 Comparison of NTG and CGRP models in human and animal settings. The features reported in the human setting rely, unless otherwise stated, on the delayed migraine-like attack experimentally provoked in migraine patients
Full size table

Other ongoing stories

The wealth of scientific output gathered with the experimental models described above has greatly contributed to identify potential therapeutic targets for migraine, paving the way, for instance, to the development of monoclonal antibodies targeting CGRP. In this frame, other potential therapeutic molecules are currently under evaluation and will hopefully lead to additional advances in migraine science. In Table 5 we show some potential targets/pathways for both acute and prophylactic treatments of migraine.

Table 5 Other potential therapeutic leads for both acute and prophylactic treatments of migraine
Full size table

Another interesting story that is going on regard the role of acute medication overuse in the transformation of episodic migraine into chronic migraine. The long-lasting ongoing debate has not yet clarified how acute medications can intercept the neurobiology of migraine to negatively affect the disease outcome. A series of recent data provided by different groups using animal models of medication overuse clearly points to the capability of some acute medications to induce a condition of hyperalgesia and will therefore contribute to disentangle this critical issue [116, 117].

Conclusions

Experimental models of migraine have yielded significant insights into brain structures that mediate migraine symptoms and the recurrence of attacks. These models have contributed to elucidate the role of small peptides as neurotransmitters within this network, thus fostering the generation of novel therapeutic approaches that have been validated by randomized placebo-controlled trials and will hopefully provide a substantial improvement to the lives of a large multitude of migraine sufferers.