Introduction

Adenosine, a vasoactive amine produced by the hydrolysis of adenosine monophosphate (AMP) or S-adenosylhomocysteine (SAH) [43], is involved in numerous physiological processes such as metabolism, inflammation, respiration and pain [39]. Adenosine binds to four G-protein coupled receptors (GPCR), A1, A2A, A2B and A3, with a unique profile of tissue distribution, signaling pathways and function (Table 1 & Fig. 1) [20, 21]. Adenosine receptors activate the mitogen-activated protein kinase (MAPK), leading to survival, cell growth, and differentiation [20], and modulate the activity of adenylate cyclase [11, 20], the enzyme that regulates intracellular concentration of cyclic adenosine monophosphate (cAMP) [38]. Adenosine A1 and A3 receptors are coupled to a Gαi subunit that downregulates cAMP by inhibiting adenylate cyclase [20, 38], while A2A and A2B receptors are coupled to a Gαs subunit which stimulates adenylate cyclase to upregulate cAMP [20, 38]. A1 and A3 receptors are considered to have anti-nociceptive effects, whereas, activation of A2A and A2B receptors induces nociception [11]. Adenosine A1 receptors are expressed at the trigeminovascular system (TVS), including the trigeminal ganglion (TG) and trigeminal nucleus caudalis (TNC) [6, 32], which is considered to be the anatomical and physiological substrate of migraine pain [2]. Stimulation of this receptor causes inhibition of the TVS by reducing neuronal firing from the trigeminal nucleus and decreasing the release of calcitonin gene-related peptide (CGRP) [6, 13]. A2A and A2B receptors are located in vascular smooth muscle cells [20, 38] and in pre- and postsynaptic nerve terminals [38], and stimulation of these receptors causes dural vasodilation [16], leading to stimulation of the TVS. Collectively, adenosine signaling pathways are complex and might be involved in headache and migraine pathophysiology.

Table 1 Adenosine receptors and their functions
Full size table
Fig. 1
figure 1

Adenosine signaling pathway. Adenosine binds to G-protein coupled receptors (GPCR) resulting in either activation (adenosine A2A and A2B receptors) or inactivation (adenosine A1 and A3 receptors) of adenylyl cyclase (AC). Activation of AC increases the formation of cyclic adenosine monophosphate (cAMP) which binds to protein kinase A (PKA). Active PKA then phosphorylates and thereby modulates cellular responses. ATP: adenosine triphosphate

Full size image

Here, we systematically review preclinical studies on the involvement of adenosine in trigeminal pain pathway, to make the case that adenosine signaling pathways may play a role in migraine and discuss adenosine receptors as potential target for future treatment of migraine (Table 2).

Table 2 Molecules presented here that target adenosine receptors. None are approved for treatment
Full size table

Method

Data source

We conducted two searches on PubMed and Embase on September 5th, 2021. Firstly, we searched “(“adenosine”[MeSH Terms] OR “adenosine”[All Fields] OR “adenosin”[All Fields] OR “adenosine s”[All Fields] OR “adenosines”[All Fields]) AND (“migrain”[All Fields] OR “migraine disorders”[MeSH Terms] OR (“migraine”[All Fields] AND “disorders”[All Fields]) OR “migraine disorders”[All Fields] OR “migraine”[All Fields] OR “migraines”[All Fields] OR “migraine s”[All Fields] OR “migraineous”[All Fields] OR “migrainers”[All Fields] OR “migrainous”[All Fields])”. Secondly, we searched for “(“adenosine”[MeSH Terms] OR “adenosine”[All Fields] OR “adenosin”[All Fields] OR “adenosine s”[All Fields] OR “adenosines”[All Fields]) AND (“headache”[MeSH Terms] OR “headache”[All Fields] OR “headaches”[All Fields] OR “headache s”[All Fields])”. Both searches were restricted to English language.

Selection criteria and study inclusion

Studies were restricted to pre-clinical or clinical studies that investigated adenosine, adenosine agonist, adenosine antagonist, adenosine deaminase, adenosine deaminase inhibitor and adenosine reuptake inhibitors in headache and migraine pathophysiology. We excluded reviews, meta-analysis, conference proceedings and case reports.

Two investigators (J.T. and L.K.) screened all studies by title and abstract, followed by full text screening to confirm eligibility. References of the included studies were screened to find studies that were missed by the search. For each included study, both investigators (J.T. and L.K.) extracted hypothesis or purpose of the study, method, sample size, main outcome, conclusion, and limitations. Any disagreements were resolved through discussion by the two investigators (J.T. and L.K.). If the conflict remained, a third investigator (M.M.K) made the final decision.

Results

The database search identified 3209 citations of which 701 were duplicates (Fig. 2). An additional two studies were included through a manual search of identified primary articles. A total of 2510 studies were screened by title and abstract and 44 were full text screened. Of these, 20 studies were included – 13 preclinical (Table 3) and 7 clinical studies. Data for clinical studies has been published recently [45].

Fig. 2
figure 2

Flow chart of search strategy

Full size image
Table 3 Summary of preclinical studies
Full size table

Narrative summaries

Arulmani et al. [1]. In pigs, intravenous infusion of adenosine A1 receptor agonist, GR79236, was compared to vehicle prior to capsaicin infusion. Total carotid blood flow, conductance, and plasma CGRP concentrations in jugular vein were assessed at baseline and after the infusions. GR79236 dose-dependently attenuated the capsaicin-induced carotid hemodynamic changes but not the CGRP release, compared to vehicle infusion.

Carruthers et al. [6]. In rats, adenosine agonists and antagonist (A1 receptor agonist GR79236X, adenosine A1 receptor antagonist, DPCPX, adenosine A2A agonist, CGS21680, and A3 receptor agonist, 2-CI-IB-MECA) were applied to cultured trigeminal neurons in combination with forskolin or vehicle to induce release CGRP. Immunocytochemical studies and Western analysis assessed whether these pharmacological agents could modulate the forskolin induced CGRP release. GR79236X concentration-dependently inhibited forskolin-stimulated CGRP release, while DPCPX abolished GR79236X´s effect. CGS21680 and 2-CI-IB-MECA were unable to attenuate forskolin-induced CGRP secretion.

Faraci et al. [9]. Intravenous infusion of adenosine was administered to anesthetized dogs. Blood flow was measured with labelled, radioactive microspheres. Adenosine decreased aortic pressure along with blood flow and vascular resistance in the dura. Adenosine infusion did not alter cerebral blood flow.

Goadsby et al. [13]. Intravenous infusion of adenosine A1 receptor agonists were administered in anesthetized dogs, following electrical stimulation of the superior sagittal sinus (SSS). Jugular vein blood samples were taken at baseline, immediately after the SSS stimulation and following the A1 agonist infusion, for detection of CGRP levels. Both A1 receptor agonists, GR79236 and GR190178, inhibited SSS-induced activation in TNC and CGRP release in cranial circulation, in a dose-dependent manner. Moreover, adenosine A1 receptor antagonist, DPCPX, was able to reverse GR79236's inhibitory effect on TNC activation.

Haanes et al. [15]. Adenosine was applied to pre-contracted middle meningeal artery (MMA) segments isolated from rats. RT-PCR was used to characterize the expression of purinergic receptor and myography to access the vascular effects. Notably, all purinergic receptor mRNAs were detected in the trigeminal ganglion and MMA. Adenosine caused dilation of MMA, which was reversed by SCH58261 (A2A receptor antagonist) and caffeine (adenosine receptor antagonist).

Haanes et al. [16]. Adenosine and caffeine were administered intravenously to six rats. Adenosine resulted in dural vasodilation and decrease in blood pressure. However, pre-treatment with caffeine inhibited adenosine´s effect. Caffeine caused an increase in blood pressure and a non-significant dilation of dural arteries. Secondly, intravenous infusions of different adenosine A2A receptor antagonists (JNJ compounds) were given following intravenous administration of adenosine A2A receptor agonist, CGS21680, or periarterial electrical stimulation (mode of CGRP-release), in rats. The closed cranial window was used to evaluate the antagonists´ effect on the CGS21680 and CGRP -induced dural dilation. CGS21680 caused vasodilation and decrease in arterial blood pressure. All A2A receptor antagonists blocked CGS21680-induced dural vasodilation with a more potent respond with A2A over A1 selectivity, while they did not affect electrical stimulated neurogenic vasodilation.

Hardebo et al. [17]. Adenosine, cAMP, ADP and ATP were applied in segments of middle cerebral artery and extracranial arteries of feline and humans. The dissected vessels were pre-constricted by prostaglandin F2a (PGF2a) or 5-hydroxytryptamine (5-HT). The tension was measured with force displacement transducers and recorded on a Grass polygraph. All adenine compounds dilated feline pial arteries, however the dilatory response was less pronounced when extracellular K+ concentration increased. Adenine compounds did not influence the diameter of human and feline extracranial arteries.

Honey et al. [19]. In rats, intravenous infusion of adenosine A1-receptor agonist, GR79236, was compared to saline infusion in models of neurogenic dural vasodilation. Vasodilation was induced by either electrical stimulation of perivascular trigeminal nerves or intravenous CGRP. Cranial window was used to evaluate the vascular responses. GR79236 inhibited electrically induced neurogenic vasodilation in a dose-dependent manner but had no effect on vasodilation caused by CGRP. Selective A1 receptor antagonist, DPCPX, inhibited the effect of GR79236 on electrically evoked vasodilation, compared to vehicle.

Jenkins et al. [22]. Adenosine deaminase was applied in cultured rat trigeminal neurons. Application of prostaglandin E2 (PGE2) led to CGRP release from the cultured cells. Adenosine deaminase did not alter baseline or PGE2-evoked CGRP levels.

Lindquist et al. [31]. Accumulation of adenosine following spreading depolarization (SD) was investigated in brain slices of mice and in vivo. Amperometric recordings from adenosine-sensitive enzyme-linked electrochemical were made in brain slices and applied in vivo. SD generated transient adenosine accumulation in vivo which could reliably report underlying metabolic status in brain slices.

Lu et al. [32]. Εlectrical stimulation of the trigeminal ganglion (ESTG) or sham operation was performed in rats to investigate its effect on CGRP, adenosine A1 receptor and adenosine A2A receptor expression. RT-qPCR and Western analysis was used for detection and quantification of the proteins. In the trigeminal nucleus caudalis (TNC) and ipsilateral trigeminal ganglion (TG), CGRP and A2A expression increased following ESTG, while A1 decreased. Interestingly, pretreatment with hinese medicine Tianshu capsule (TSC) decreased CGRP and A2A expression and increased A1 receptor expression.

Wei et al. [46]. In cats, the effect of topical application of adenosine and adenosine diphosphate was investigated before and following application of CGRP receptor antagonist, CGRP8-37. Cranial window was used to evaluate the vascular responses. CGRP8-37 was not able to reverse vasodilating effect of adenosine and adenosine diphosphate. Moreover, guanylate cyclase inhibitor, LY83583, had no effect on adenosine-induced vasodilation.

Yegutkin et al. [48]. The effects of adenine nucleotides were assessed in meninges of rats and cultured trigeminal cells following application of CGRP or placebo. Bioluminescent and fluorometric techniques were used to measure purine levels in trigeminal ganglion cells, and electrophysiology to record the nociceptive spikes in the meningeal trigeminal nerves, before and after pre-treatment with CGRP. CGRP decreased adenosine levels in cultured cells but not in the meninges, while adenosine was not able to activate nociceptive firing in the meningeal nerves. Moreover, basal levels of adenosine and AMP where higher compared to ATP and ADP in trigeminal cells.

Discussion

The main findings of the present systematic review are that adenosine receptors modulate pain transmission through the TVS. While A1 receptor has an inhibitory effect, stimulation of A2A receptor causes vasodilation and activation of trigeminal pain pathway.

In rats, electrical stimulation of the TG decreased adenosine A1 receptor expression and increased adenosine A2A receptor expression in ipsilateral TG and TNC [32]. The former is suggested to be involved in migraine attack initiation, while upregulation of adenosine A1 receptors or activation of this receptor might block migraine attacks [32]. In support, one study found that treatment with adenosine A1 receptor agonists, GR79236 and GR190178, inhibited TVS activation after electrical stimulation of the superior sagittal sinus in cats [13]. Upregulation of both adenosine A2A and CGRP receptors following electrical stimulation implies that when combined, the two receptors activate trigeminal pain transmission and cause migraine [32]. Overall, these findings implicate both adenosine A1 and A2A receptors in pain regulation and transmission during migraine [13, 32].

Three studies showed that adenosine caused prominent dilation of pre-contracted middle meningeal artery, dural and pial arteries in vitro [15,16,17]. Another study showed that adenosine caused dural vasodilation in dogs [9]. Pretreatment with caffeine or adenosine A2A receptor antagonist, SCH58261, was able to block adenosine-induced dilation in vitro [15, 16], suggesting that adenosine-induced vasodilation might be mainly dependent on adenosine A2A receptor [15].

Adenosine A1 receptor agonist, GR79236, inhibited electrically-induced vasodilation and capsaicin-induced hemodynamic changes in carotid artery [1, 19]. Pretreatment with DPCPX prevented the inhibition following GR79236, indicating that its inhibitory or vasoconstricting effect is mediated through adenosine A1 receptor [19]. The same agonist, GR79236, inhibited CGRP release induced by adenylate cyclase activator, forskolin [6] without any effect on CGRP-induced vasodilation in rats [19]. These data indicate that GR79236 inhibits CGRP release via a pre-junctional inhibition, and that adenosine A1 receptors are present on CGRP-positive neurons [6, 13, 19]. Together with its vasoconstricting ability, it is suggested that GR79236 and adenosine A1 receptors hold anti-migraine potential [1, 6, 19].

In contrast to GR79236, adenosine A2A receptor agonist, CGS21680, had no effect on forskolin-induced CGRP release [6]. However, CGS21680 caused dural vasodilation that was blocked by adenosine A2A receptor antagonists (JNJ-compounds) [16]. The study showed that the lower the selectivity for A2A receptor over A1 receptors, the higher the potential to attenuate the CGS21680 induced vasodilation. It was suggested that blocking both A1 and A2A receptors might be necessary to completely attenuate dural vasodilation [16].

Of note, 2-CI-IB-MECA, an adenosine A3 receptor agonist, had no effect on forskolin-induced CGRP secretion in rats [6]. The involvement of adenosine A3 receptor in migraine has not been further investigated, however, adenosine A3 receptor agonist exhibited anti-nociceptive properties in models of chronic pain in rats and mice [24].

While CGRP antagonist, CGRP(8–37), and guanylate cyclase inhibitor, LY83583, inhibited CGRP and nitroglycerine induced vasodilation, both compounds did not alter adenosine-induced vasodilation [46]. This finding demonstrates that adenosine induced dilation is not dependent on activation of CGRP receptors or an increase in cyclic guanosine monophosphate. Another study showed that pretreatment of trigeminal ganglion cells with CGRP is followed by decreased adenosine levels compared to baseline [48]. It is suggested that the finding might be a part of migraine sensitization but due to other contradicting findings (i.e., no change in nociceptive firing), further investigation on adenosine’s mechanisms was recommended [48].

Collectively, studies showed adenosine receptors expression in the trigeminal pain pathway and indicated that adenosine-induced pronociceptive effect is mediated through A2A receptor activation, whilst A1 receptor mediates antinociception. A1 receptor agonist, GR79236, inhibits activation of the TVS and vasodilation, while A2A receptor antagonist, SCH58261, attenuated adenosine-induced vasodilation [13, 15, 19], designating adenosine A1 and A2A receptors as possible targets in the treatment of migraine.

Limitations and future perspective

The major limitations of the studies included, were differences in methodological approaches including designs, subjects, substances, and sampling sources. Additionally, concentrations and types of adenosine A1 receptor agonists applied, differed across the studies [6, 13, 19]. Different CGRP releasing mechanisms were applied throughout the studies, potentially affecting the potency of adenosine receptor agonists and antagonists in modulating the CGRP release [1, 6, 13, 22].

Human studies are needed to elucidate the headache inducing effect of adenosine in patients with migraine. A specific focus on adenosine A1 receptor agonists and A2A receptors antagonists would be of great interest because of their potentially opposite effects based on current knowledge. Several adenosine receptor agonists and antagonist are currently available for research purpose only, while only one adenosine receptor antagonist, istradefylline, is currently U.S. Food and Drug Administration (FDA) approved as treatment for Parkinson’s disease (JF and RA 2020). To our best knowledge, no studies have been conducted on the adenosine A2B receptor in migraine and adenosine A3 has only once been investigated in migraine [6]. This leaves a huge gap in our knowledge that needs to be explored in both clinical and pre-clinical setting.

Conclusion

Preclinical data demonstrated that adenosine caused vasodilation and modulated CGRP release. We suggest that the adenosine A1 receptor and adenosine A2A receptor could be potential targets for migraine treatment.