The role of the adenosine system in epilepsy and its comorbidities

electrical activity in the brain that lead to seizures. Despite over 50 million people being affected worldwide, only (cid:1) 70% of people with epilepsy have their seizures successfully controlled with current pharma-cotherapy, and many experience significant psychiatric and physical comorbidities. Adenosine, a ubiquitous purine metabolite, is a potent endogenous anti-epileptic substance that can abolish seizure activity via the adenosine A 1 G protein-coupled receptor. Activation of A 1 receptors decreases seizure activity in animal models, including models of drug-resistant epilepsy. Recent advances have increased our understanding of epilepsy comorbidities, highlighting the potential for adenosine receptors to modulate epilepsy-associated comorbidities, including cardiovascular dysfunction, sleep and cognition. This review provides an accessible resource of the current advances in understanding the adenosine system as a therapeutic target for epilepsy and epilepsy-associated comorbidities. LINKED ARTICLES: This article is part of a themed issue Therapeutic Targeting of G Protein-Coupled Receptors: hot topics


| EPILEPSY: A BRIEF OVERVIEW
Epilepsy is a neurological disease affecting approximately 70 million people worldwide (Thijs et al., 2019).It is characterised by recurrent excessive synchronous electrical activity in the brain, which manifest as clinical seizures (Stafstrom & Carmant, 2015).Epileptic seizures are broadly categorised into three main subtypes: focal onset (seizures with discrete areas of electrical hyperactivity), generalised onset (seizures with widespread electrical hyperactivity) or seizures with unknown onset (Fisher et al., 2017).Although genetic causes are increasingly recognised, the aetiology of epilepsy is complex and multifactorial in many patients with epilepsy (Stafstrom & Carmant, 2015).The mainstay of treatment for epilepsy is pharmacotherapy, with more than 20 anti-seizure medications (ASMs) available on the market.Although the mechanisms of action for many ASMs are not clearly defined, most are thought to alter the threshold for neuronal excitability through modifying the activity of GABA A receptors (with benzodiazepines, tiagabine and phenobarbital), sodium channels (with phenytoin and carbamazepine) or calcium channels (with pregabalin and gabapentin) (Kwan et al., 2001).Currently used ASMs are associated with significant adverse effects and do not modify disease progression, including neuronal loss, inflammation and gliosis (Thijs et al., 2019).Up to 30% of patients with epilepsy continue to have seizures despite ASM treatment.The lack of effective treatment severely impairs patient quality of life (Kwan & Brodie, 2006).Moreover, epilepsy frequently presents with comorbidities contributing to increased mortality (Thijs et al., 2019).Accordingly, novel therapeutic agents are urgently needed to modify disease progression and to exhibit efficacy against drug-resistant epilepsy.
GPCRs play an important role in orchestrating neural signalling and maintaining homeostasis and thus are promising therapeutic targets for neurological diseases, including epilepsy (Hauser et al., 2017).
Here, we review the evidence for the adenosine GPCRs as potential drug targets to treat epilepsy and associated comorbidities.

| ADENOSINE: AN ENDOGENOUS ANTI-SEIZURE SUBSTANCE
2.1 | The role of adenosine receptors and seizure activity Adenosine is a key endogenous regulator of neuronal homeostasis, playing vital roles in cognition, sleep and learning (Ribeiro et al., 2002).
It is ubiquitously present throughout the brain as a breakdown product of the energy source ATP.Adenosine signals via four GPCRs, with the A 1 and A 2A receptor representing the subtypes most abundant in the brain, with dense distribution of the A 1 receptor in the cortex and hippocampus, and A 2A receptor in the striatum and nucleus accumbens (Ribeiro et al., 2002).The A 2B and A 3 receptor are also expressed in the brain, albeit at lower levels (Ribeiro et al., 2002).The roles of the A 2B and A 3 R receptor in epilepsy and seizures are less well defined (Masino et al., 2014;Tescarollo et al., 2020).The A 1 receptor preferentially signals via G i/o proteins in the brain, whereas the A 2A receptor initiates cellular communication preferentially through G s and G olf proteins (Fredholm et al., 2001).In addition to signalling via GPCRs, adenosine can engage in receptor-independent regulation of DNA methylation, thereby playing a key role in brain epigenetic programming and maintenance (Williams-Karnesky et al., 2013).
Often referred to as the brain's 'endogenous anti-convulsant', adenosine dampens neuronal excitability by acting at both pre-and post-synaptic A 1 receptors (Cunha, 2016;Dunwiddie & Masino, 2001;Spanoghe et al., 2020) (Figure 1).Activation of post-synaptic A 1 receptors promotes cell membrane hyperpolarisation, primarily driven F I G U R E 1 Neuronal A 1 and A 2A receptor signalling.Adenosine is present both extracellularly and intracellularly in neurons and glia.Enzymatic dephosphorylation of ATP and AMP increases levels of adenosine (ADO).Adenosine can be transported across the plasma membrane by equilibrative nucleoside transporters (ENT1 & 2).Clearance of adenosine occurs by its phosphorylation to AMP by adenosine kinase (ADK) and deamination to inosine by adenosine deaminase (ADA).The A 1 receptor (A 1 R) is present in pre-and post-synaptic neurons and astrocytes.The A 2A receptor (A 2A R) is found on post-synaptic neurons, astrocytes and microglia.The A 1 receptor reduces seizure activity by inhibiting neuronal excitability and reducing excitatory neurotransmitter release.Specifically, coupling of the A 1 receptor to G i/o proteins can inhibit calcium and sodium channels and increase potassium efflux.In contrast, the A 2A receptor can increase seizure activity, neuronal depolarisation and neuronal firing by activation of sodium/calcium channels via G s and G olf proteins.Adenosine also engages in receptor-independent modulation of DNA methylation, with ADK playing an important role.
Seizure activity significantly increases adenosine release, and this increased adenosine tone has been demonstrated to mitigate hyperexcitability and dampen seizure propagation in both animal models and patients (During & Spencer, 1992;Van Gompel et al., 2014).
Adenosine-mediated effects on seizures are temporally and spatially regulated.For instance, although activation of A 1 receptors reduces seizure potential, activation of A 2A receptors can have the opposite effect, initiating seizure activity through promoting membrane depolarisation and subsequent neuronal firing (Sebastiao & Ribeiro, 1996).
Thus, the balance of adenosine receptor subtype expression and adenosine clearance in different brain regions is vital in the overall modulation of seizure risk.
Such modulation of seizure activity by adenosine receptor signalling has been demonstrated in animal models and humans.For instance, A 1 receptor-KO mice have severe convulsions and significant widespread neuronal cell death, leading to lethal status epilepticus after treatment with intrahippocampal kainic acid (Fedele et al., 2006).In the same study, wildtype littermates experienced less severe seizures, minimal neuronal loss and did not die from these events.Similarly, A 1 receptor KO in a mouse model of experimental traumatic brain injury resulted in lethal status epilepticus (Kochanek et al., 2006).A 1 receptor antagonists increase seizure liability in animal models, an adverse effect also noted in a Phase 3 clinical trial of an A 1 receptor antagonist that was being developed to treat heart failure (Fukuda et al., 2010;Teerlink et al., 2012).In contrast, activation of A 1 receptors by exogenous adenosine, or a selective A 1 receptor agonist, suppresses seizure activity in animal models of epilepsy and human hippocampal epileptiform brain slices (Gouder et al., 2003;Klaft et al., 2016); also see (Beamer, Kuchukulla, et al., 2021) for a comprehensive summary.A 2A receptor KO raised the seizure threshold in mice treated with kainate and mitigated seizure-induced excitotoxicity and neurodegeneration (Canas et al., 2018;El Yacoubi et al., 2009).Similarly, pharmacological antagonism of the A 2A receptor also limited seizure activity and promoted neuroprotection, whereas A 2A receptor agonism increased seizure activity (Canas et al., 2018;D'Alimonte et al., 2009;Li et al., 2012;Zeraati et al., 2006).These findings are consistent with clinical studies of the A 2A receptor agonist, regadenoson, which is used for myocardial perfusion imaging, but can also lower seizure threshold (Agarwal & DePuey, 2014).Of note, caffeine, a well-known, non-selective, adenosine receptor antagonist, is regularly consumed by patients with epilepsy and caffeine intake may increase seizure susceptibility (van Koert et al., 2018).Inhibitors of adenosine kinase (ADK), which block the actions of a key enzyme that phosphorylates adenosine to AMP, can also mitigate seizures by decreasing adenosine clearance (Gouder et al., 2004;Sandau et al., 2019;Ugarkar et al., 2000).Encouragingly, seizure reduction mediated by the A 1 receptor has been shown in a range of models, including pharmacoresistant epilepsy models, highlighting the potential of specifically targeting the A 1 receptor for those patients refractory to current ASMs (Beamer, Kuchukulla, et al., 2021;Gouder et al., 2003;Klaft et al., 2016).Collectively, these studies highlight the potential for adenosine to play a crucial role in arresting seizure activity and promoting a state of post-seizure refractoriness, depending on the adenosine target and its mode of modulation.

| Evidence for disrupted adenosine homeostasis in epileptogenesis
Epileptogenesis defines the process by which the brain undergoes biological changes, rendering it susceptible to generate recurrent spontaneous seizures, that is, epilepsy (Goldberg & Coulter, 2013).
The adenosine system is implicated in epileptogenesis, as upregulation of ADK expression causes dysregulation of adenosine metabolism (Boison, 2012).Astrogliosis is a common hallmark of the epileptic brain, whereby disease progression can cause aberrant astrocyte proliferation (Li et al., 2019).Astrocytes are the predominant site of ADK expression in the adult human brain and regulate extracellular adenosine concentrations (Boison, 2012;Boison et al., 2010;Studer et al., 2006).Indeed, astrogliosis associated with ADK overexpression is a hallmark of epilepsy, independent of the animal model used (Boison, 2012;Fedele et al., 2005;Gouder et al., 2004;Li et al., 2007;Li et al., 2008).ADK overexpression precedes seizure activity, and therefore, ADK upregulation appears to be a precipitating factor, rather than a consequence of seizure activity (Li et al., 2007).Thus, disruption of astrocyte homeostasis, as seen with astrogliosis, coincides with disequilibrium in adenosine metabolism.
As indicated above, ADK catalyses the phosphorylation of adenosine to AMP.As such, increased ADK activity may limit the activation of A 1 receptors and, in turn, enhance neuronal excitability.Additionally, nuclear ADK indirectly modulates DNA methylation, and thus increased ADK activity could alter brain epigenetics and disease progression (Boison & Jarvis, 2021;Murugan et al., 2021).Indeed, transgenic mice overexpressing ADK develop recurrent spontaneous hippocampal seizures and, similarly, adenoviral overexpression of ADK in hippocampal astrocytes causes seizure activity (Li et al., 2007;Theofilas et al., 2011).Interestingly, early evidence suggests ADK inhibition during the developing phase of epileptogenesis may have disease-modifying properties.
In a kainic acid model of temporal lobe epilepsy (TLE), mice treated with an ADK inhibitor showed a significant reduction in seizure activity, and treatment with the inhibitor prevented ADK overexpression, thus limiting disease progression (Sandau et al., 2019).In resected hippocampal tissue from patients with epilepsy, significant overexpression of ADK was found in areas of astrogliosis (Aronica et al., 2011;Masino et al., 2011).Additionally, ADK is overexpressed in human astrocytic tumours, and associated with the incidence of tumour-related epilepsy (de Groot et al., 2012).Interestingly, patients with genetic ADK deficiency experience recurrent seizures and cognitive deficits, with these findings replicated in a mouse model of brain ADK deletion (Sandau et al., 2016).The A 2A receptor is implicated in mediating these effects, where its activation by excess adenosine caused excitotoxicity, seizure activity and aberrant synaptic plasticity, thus affecting learning and memory (Sandau et al., 2016).Inhibiting activity of the A 2A receptor by pharmacological or genetic means could mitigate these adverse effects, highlighting the fine balance between the expression and activity of ADK and adenosine receptors, and that this balance can modulate seizure risk.Indeed, in addition to ADK, there is an altered expression of A 1 and A 2A receptors in patients with epilepsy (Barros-Barbosa et al., 2016;Glass et al., 1996;Patodia et al., 2020).Therefore, altered adenosine signalling, as a result of changes to metabolism and/or adenosine receptor expression, may reflect a common feature in human epilepsy.As such, modulating the adenosine system during the early stages of epileptogenesis may represent a disease-modifying therapeutic intervention.

| Targeting the adenosine receptor system in patients with epilepsy
Despite promising pre-clinical data, there are no available therapeutic agents specifically targeting the adenosine system to treat epilepsy.
The development of ADK inhibitors to augment brain adenosine levels previously gained significant traction, with ABT-702 (preclinical) and GP-3269 (Phase I) being investigated in epilepsy (Boison & Jarvis, 2021;Jarvis, 2019).However, the progression of these compounds, and ADK inhibitors in general, was suspended because of severe toxicity (Jarvis, 2019).Over the last decade, adenosine-releasing brain implants, alongside RNA interference methods, have shown some promise (Ren & Boison, 2017;Theofilas et al., 2011;Young et al., 2014).However, A 1 receptor agonists or A 2A receptor antagonists have not yet progressed into clinical trials as treatments for epilepsy.This stagnation in development, particularly for A 1 receptor agonists, can largely be attributed to on-target adverse effects, which remain a significant problem for A 1 receptor-targeted drug discovery.Direct activation of A 1 receptors can mediate unwanted central and peripheral effects, including sedation, hypothermia, bradycardia and hypotension (Borea et al., 2018;Headrick et al., 2013).Novel medicinal chemistry efforts and alternative pharmacological paradigms will be needed to target these receptors successfully and these concepts have been extensively reviewed (McNeill et al., 2021;Vecchio et al., 2018).
Although pharmacotherapies targeting the adenosine system are not available, interesting links between the A 1 receptor and non-pharmacological interventions have been identified.Deep brain stimulation (DBS) of the anterior thalamic nucleus, an intervention available for patients resistant to traditional ASMs, reduces seizure activity (Miranda et al., 2014;Thijs et al., 2019).In rats treated with the muscarinic receptor agonist, pilocarpine (a common pharmacological precipitant of seizure activity), DBS caused a significant increase in adenosine release that reduced cellular excitability.This effect was abolished upon administration of an A 1 receptor antagonist, suggesting a role for this receptor in mediating the anti-seizure effects of DBS (Miranda et al., 2014).Moreover, in a pentylenetetrazole kindling model, DBS blunted the kindling process and was associated with reduced ADK expression, further supporting the hypothesis that DBS modifies epileptogenesis via the adenosine system (Gimenes et al., 2022).Similarly, a ketogenic diet is a treatment option for patients with drug-resistant epilepsy, with the A 1 receptor implicated in the anti-seizure effects of this diet (D'Andrea Meira et al., 2019;Masino et al., 2011;Masino et al., 2012).Indeed, a ketogenic diet suppresses spontaneous seizures in transgenic mice overexpressing ADK, an effect abolished in the presence of a selective A 1 receptor antagonist (Masino et al., 2011).Moreover, the anti-seizure effect of exogenous ketone administration in rats was inhibited by an A 1 receptor antagonist (Kovács et al., 2017).Taken together, these studies on DBS and the ketogenic diet suggest that the A 1 receptor can mediate anti-seizure effects associated with non-pharmacological interventions in humans.Confirmation would further support the A 1 receptor as a promising therapeutic target for epilepsy, particularly for those patients who are refractory to treatment.

| EPILEPSY AND ITS COMORBIDITIES: A ROLE FOR THE ADENOSINE SYSTEM
Epilepsy is a multifaceted disease associated with a range of comorbidities.Neuropsychiatric conditions include depression, anxiety, altered cognition and sleep.Physical comorbidities include cardiovascular disease, hypertension and diabetes (Keezer et al., 2016) (Figure 2).Approximately 50% of adults diagnosed with epilepsy develop at least one comorbidity (Keezer et al., 2016).Such comorbidities can significantly decrease patient quality of life and increase mortality risk.Although there is no clear mechanism defining how these comorbidities arise, the molecular changes associated with epilepsy, alongside ASM treatment (or lack thereof), can increase the likelihood of developing associated comorbidities (Keezer et al., 2016).Adenosine receptors have a well-established role in modulating the pathophysiology of many epilepsy-associated comorbidities.Indeed, the relevance of the adenosine system is two-fold, where it may represent a common link between epilepsy and comorbidities but could also represent a promising therapeutic target.

| Cardiovascular disease
Cardiac comorbidities have been reported in up to 80% of patients with epilepsy (Selassie et al., 2014).These patients display altered electrocardiogram traces, including arrhythmia as well as conductance and repolarisation perturbations, which are maintained during periods of seizure freedom (van der Lende et al., 2016;Verrier et al., 2020).
Population-based studies show chronic epilepsy carries a significantly increased risk for cardiac arrest, compared with that of the general population (Bardai et al., 2012;Janszky et al., 2009).A genetic link may underlie these cardiac changes, with the common expression of potentially pathogenic ion channels in the brain and heart (Johnson et al., 2009;Ravindran et al., 2016).However, a growing body of evidence suggests that damage caused by repeated seizures in uncontrolled epilepsy leads to incremental electrical and mechanical cardiac dysfunction (Tigaran et al., 2003;Verrier et al., 2020).Specifically, long-term monitoring studies of patients with chronic epilepsy revealed a significant proportion with clinically relevant cardiac arrhythmia and T-wave alternans, but this was not observed in newly diagnosed cases (Pang et al., 2019;Sivathamboo et al., 2022).Myocardial infarction (MI) is another serious cardiac comorbidity associated with epilepsy.A retrospective study found that patients without prior MI at the onset of epilepsy showed a 48% higher risk of MI, compared with patients experiencing a neurological disease (i.e., migraine) without seizures (Wilson et al., 2018).Patients with epilepsy also have an increased incidence of heart failure (Doege et al., 2021).Thus, such patients are more likely to experience cardiac electrical and mechanical dysfunction of genetic and or seizure-related origin.
In addition to disease processes, epilepsy treatments may drive undesirable cardiac effects.Chronic dosing with ASMs, including carbamazepine and gabapentin, has been associated with a significantly increased risk of sudden cardiac death (Hookana et al., 2016).Moreover, ASMs that induce CYP450 enzymes are also associated with increased dyslipidaemia and adverse cardiac effects (Renoux et al., 2015).In addition to ASMs, the high-fat ketogenic diet for epilepsy has potential to increase cardiac comorbidity risk, although the findings are unclear.Treatment with the ketogenic diet for more than 6 months has been associated with increased arterial stiffness and potential right ventricular diastolic dysfunction, although assessment of echocardiography parameters has not identified any alterations (Coppola et al., 2014;Ozdemir et al., 2016).As such, the relative safety of the ketogenic diet on cardiovascular physiology is still unknown.

| Adenosine and the cardiovascular system
Adenosine receptors are expressed in all cardiovascular tissues.Adenosine stimulates negative inotropy (cardiac contractility), chronotropy (heart rate) and dromotropy (AV node conduction), predominantly through activating A 1 receptors on cardiomyocytes (Headrick et al., 2011).Adenosine stimulates powerful cytoprotective effects in ischaemic tissues, including the heart, opposing the injury produced by oxidative stress, calcium overload and mitochondrial permeability transition pore opening (Hausenloy & Yellon, 2013;Kaminski et al., 2002).Activation of A 1 receptors as a consequence of elevated adenosine tone is a key driver of cardioprotection.There is a wellestablished role of A 1 receptor activation in preconditioning, i.e., the phenomenon by which hearts exposed to a prior ischaemic insult show reduced infarct size and preserved function (see Cohen et al., 2000, for review).Similarly, A 1 receptor activation post-MI stimulates beneficial cardioprotection (Lasley et al., 1990;Norton et al., 1992;Urmaliya et al., 2010).To date, no adenosine receptor ligand has been approved for use as a cardioprotective agent due to F I G U R E 2 Comorbidities associated with epilepsy.(a) Epilepsy is associated with numerous comorbidities, many of which can be segregated into conditions of the CNS and the periphery.(b) Current literature suggests targeting adenosine receptors may be beneficial for epilepsy and its comorbidities.In treatment-resistant epilepsy, A 1 receptor (A 1 R) agonists and A 2A receptor (A 2A R) antagonists may be useful.Non-pharmacological modalities, including the ketogenic diet and deep brain stimulation (DBS), involve the adenosine system.Activation of A 1 receptors may benefit cognition and cardiovascular comorbidity, whereas inhibition of A 2A receptors could improve cognition and mitigate SUDEP risk.
dose-limiting on-target bradycardia, negative dromotropy and hypotension.However, this also highlights the potential for exploiting novel paradigms of GPCR drug action, for example, on-target biased agonism, to overcome this limitation (McNeill et al., 2021;Vecchio et al., 2018).
A range of in vitro and in vivo models suggest that adenosine receptor activation prevents progression to heart failure in post-MI patients through decreasing pathological cardiac remodelling, particularly hypertrophy and fibrosis.An in vivo rat model of post-MI remodelling and heart failure found the administration of a stable adenosine analogue decreased pathological cardiac remodelling (Wakeno et al., 2006).Mechanistic in vitro studies demonstrated that activation of A 1 receptors attenuated hypertrophy in isolated cardiac myocytes (Chuo et al., 2016;Rueda et al., 2021), whereas activation of A 2B receptors prevented cardiac fibrosis (Wakeno et al., 2006).
Local adenosine concentrations also rise in response to increased neuronal activity during a seizure.In patients with epilepsy, peripheral blood concentrations of purine metabolites (including adenosine) are increased (Beamer, Lacey, et al., 2021).Similarly, purine metabolite concentrations were increased in the blood of mice following status epilepticus and correlated with the seizure burden and degree of brain damage (Beamer, Lacey, et al., 2021).These findings suggest that adenosine concentrations are elevated in both the brain and in peripheral blood of patients with epilepsy.It is currently unclear, however, whether seizure activity and its associated release of adenosine can influence cardiovascular physiology and pathophysiology.Notably, the A 1 receptor represents a potential therapeutic target for several cardiac conditions, such as MI and heart failure.As such, therapeutic approaches stimulating A 1 receptors to decrease seizure activity may also be beneficial for epilepsy-associated cardiovascular comorbidities.

| Sudden unexpected death in epilepsy
Patients with epilepsy have increased prevalence of premature mortality (Fazel et al., 2013).Sudden unexpected death in epilepsy (SUDEP) is defined as a sudden, unexpected, witnessed or unwitnessed, non-traumatic, and non-drowning death that occurs in benign circumstances in patients, with or without evidence for a seizure, in which post-mortem examination does not reveal a cause of death (Devinsky et al., 2016).SUDEP can affect individuals of any age, but it is most common in younger adults (20-45 years) and occurs more commonly during nocturnal hours (4 AM-8 AM) (Holst et al., 2013).SUDEP incidence is variable, but a recent meta-analysis estimated one to two cases per 1000 patients with epilepsy (Thurman et al., 2014).However, SUDEP incidence is increased in drug-resistant epilepsy and in patients after unsuccessful epilepsy surgery (Tomson et al., 2008).Other risk factors associated with SUDEP include male sex, epilepsy onset before age 16 years, chronic epilepsy >15 years, and the number of generalised tonic-clonic seizures per year (Hesdorffer et al., 2011).Of these, the most common risk factor for SUDEP is poor control of convulsive tonic-clonic seizures, particularly when they are nocturnal (Hesdorffer et al., 2011).Generalised tonic-clonic seizures associated with SUDEP may lead to persistent impairment of brain activity and increased sympathetic activation (Devinsky et al., 2016).This, in turn, can lead to respiratory dysfunction, arousal failure, non-tachyarrhythmia cardiac dysfunction and postictal suppression in an electroencephalogram (EEG), potentially culminating in postictal apnoea and bradycardia progressing to asystole and death (Sarkis et al., 2015).Indeed, continuous EEG monitoring of patients with epilepsy found generalised tonicclonic seizures precede SUDEP, with postictal respiratory failure, arousal impairment and bradycardia causing death (Ryvlin et al., 2013).
Neuropathological alterations have been described in SUDEP cases, however, these findings are variable (Devinsky et al., 2016).Similarly, although abnormalities in cardiac and pulmonary structures have been identified in SUDEP cases, there is not enough substantial evidence to indicate that structural abnormalities are a primary pathophysiological mechanism (Devinsky et al., 2016).Interestingly, assessment of genes related to SUDEP risk suggest mutations of ion channels expressed in the heart and brain may be involved (Bagnall et al., 2016;Devinsky et al., 2016;Klassen et al., 2014).A crucial element of SUDEP is brainstem dysfunction, for which postictal generalised EEG suppression may be a biomarker (Devinsky et al., 2016).Moreover, it has been suggested that aberrant brainstem neurotransmission, including glutamatergic, GABAergic, serotoninergic and adenosine signalling, may underlie SUDEP pathology (Devinsky et al., 2016).

| Adenosine system and SUDEP
Emerging evidence proposes a key mechanistic role for adenosine and adenosine receptors in SUDEP.Significant release of adenosine during seizure activity has the capacity to not only mitigate excessive excitatory activity, but also deleteriously suppress vital central respiratory function in key brain regions, including the brainstem and amygdala (Purnell et al., 2021); primarily via the A 1 and A 2A receptor (Barraco et al., 1990;Phillis et al., 1997;Reklow et al., 2019).In addition to centrally mediated mechanisms, adenosine receptors, particularly the A 2A and A 2B receptor, are expressed in the carotid body, and play a role in modulating the ventilation response, which has been proposed to be involved in SUDEP (Biggs et al., 2022;Conde et al., 2017).Moreover, due to its circadian rhythmicity, adenosine brain levels increase during the evening, which correlates with the nocturnal nature of SUDEP and thus suggests that increased adenosine concentrations may increase SUDEP risk (Purnell et al., 2018).Animal studies have linked the adenosine system with SUDEP.In a kainic acid model in mice, inhibition of adenosine metabolism initially protected mice from developing seizure activity, likely to be due to stimulation of A 1 receptors (Shen et al., 2010).However, following the initial protection, mice progressed to significant seizures and eventual death, which was not seen in kainic acid-only treated mice.Although not directly assessed, it was also suggested that impaired adenosine clearance facilitated the activation of pro-seizure A 2A receptors, and caused central cardiorespiratory depression.Co-treatment with caffeine could significantly increase survival time of mice, suggesting a role for adenosine receptors (Shen et al., 2010).Similarly, in a kainic acid model of epilepsy, treatment with an A 2A receptor antagonist in ADK knockdown mice significantly reduced the incidence of SUDEP (Shen et al., 2022).
Moreover, impairing adenosine clearance or administration of adenosine in mice enhanced seizure-induced respiratory arrest incidence, induced a decrease in oxygen saturation and/or suppressed phrenic nerve activity; all which play an essential role in breathing (Ashraf et al., 2021;Faingold et al., 2016;Kommajosyula et al., 2016).
Prior to SUDEP in humans, maladaptive changes, including changes in adenosine receptor expression, have been proposed to occur in brain regions essential for central cardiorespiratory control, including the brainstem (Patodia et al., 2021).In patients with TLE, those stratified into high risk for SUDEP had decreased expression of A 2A receptors in the temporal cortex but increased expression of A 1 receptors in the amygdala (Patodia et al., 2020).Although a preliminary finding, it was suggested that during seizure events, increased A 1 receptor expression could adversely modulate amygdala-driven apnoea and increase the risk of brainstem spreading depolarisation, of which the A 1 receptor may play a role (Patodia et al., 2020;Patodia et al., 2021;Purnell et al., 2021).Overall, the adenosine hypothesis of SUDEP suggests enhanced postictal adenosine, coupled with alterations in adenosine clearance and receptor expression, could facilitate cardiorespiratory depression and subsequent death.The protective effect of caffeine in SUDEP studies and models of traumatic brain injury-induced apnoea suggests adenosine receptor antagonism may represent a protective strategy against respiratory depression (Lusardi et al., 2012;Lusardi et al., 2020;Richerson et al., 2016).No epidemiological links have been identified between caffeine consumption and SUDEP risk; however, there is a reported relationship between caffeine intake and increased seizure risk (van Koert et al., 2018).There remains an inherent need to balance mitigating SUDEP risk whilst maintaining protective antiseizure adenosine receptor signalling.Moreover, the anti-somnogenic effects of caffeine limit its utility during the evening hours in which SUDEP occurs.Additionally, the relative roles of A 1 and A 2A receptors in mediating cardiorespiratory depression in the brainstem in SUDEP have not been fully delineated, although a protective effect of A 2A receptor, but not A 1 receptor, antagonism was identified in a SUDEP mouse model (Faingold et al., 2016;Reklow et al., 2019).Based on these insights, it is tempting to speculate whether an A 2A receptor antagonist could provide dual protection against seizure-induced excitotoxicity and prevention of central respiratory depression.

| Cognition
Cognitive impairment is the most common comorbidity in epilepsy and is multifactorial, as it depends on disease onset, seizure severity, interictal epileptic activity, ASM treatment, individual reserve capacities and the presence of other comorbidities (Elger et al., 2004).
Seizures and interictal spikes can generate transient cognitive effects; frequent or sustained multifocal or generalised epileptiform discharges have greater effects on cognition (Holmes & Lenck-Santini, 2006).Longstanding and uncontrolled epilepsy is correlated with worsening of cognitive status (Helmstaedter & Witt, 2017).
ASM treatments such as valproate, carbamazepine, lamotrigine, and phenytoin are associated with a reduction in cognitive function (Eddy et al., 2011).For the majority of patients with epilepsy, cognitive decline is slow (Jokeit & Ebner, 2002); however, in some cases, cognitive deficits (attention, executive function and memory) are already present at seizure onset (Pulliainen et al., 2000).Studies following patients after initial epilepsy treatment showed improvement in certain areas of cognition, such as attention, psychomotor speed and executive functions with appropriate seizure control (Witt et al., 2015;Witt & Helmstaedter, 2013).Despite this, cognitive dysfunction can persist and be resistant to treatment even with successful seizure control, highlighting a need to develop new cognition-enhancing therapies for patients with epilepsy.

| Adenosine receptors and cognition
The adenosine system is well recognised for modulating cognition, with caffeine, an adenosine receptor antagonist, being one of the most widely used psychostimulants (Camandola et al., 2019).Broadly, it is appreciated that inhibiting A 1 and A 2A receptors can have beneficial effects on cognitive domains, particularly learning, memory and attention.In naïve control animals, A 2A receptor activation impairs memory, whilst antagonism promotes cognitive improvement (Pagnussat et al., 2015;Pereira et al., 2005).Similarly, mice with global or conditional A 2A receptor knockdown show enhanced performance across several memory domains (Wang et al., 2006;Wei et al., 2011;Zhou et al., 2009).Although not studied extensively, the pro-cognitive benefits of A 2A receptor inhibition also extend to animal models of epilepsy or seizure events.The induction of a single early life seizure with kainic acid in young rats adversely affected memory later in adult life, treating with an A 2A receptor antagonist after the insult prevented the associated memory deficits (Cognato et al., 2010).Moreover, adult rats in this study showed evidence of synaptotoxicity, alongside increased A 2A receptor and decreased A 1 receptor expression, with an A 2A receptor antagonist mitigating the synaptotoxic effects.Similarly, treatment with the non-selective adenosine receptor antagonist, 8-phenyltheophylline, reduced memory deficits associated with exposure to pentylenetetrazole, although additional adenosine receptor subtypes cannot be excluded from facilitating this effect (Homayoun et al., 2001).The A 2A receptors are overexpressed in the hippocampi of patients with TLE and animal models of seizures, with this overexpression associated with increased neurodegeneration (Augusto et al., 2021;Barros-Barbosa et al., 2016;Canas et al., 2018;D'Alimonte et al., 2009).Unrelated preclinical neurodegeneration models have implicated A 2A receptors in mediating memory deficits, including dementia, Huntington's and Parkinson's diseases (Jenner et al., 2020).Interestingly, a common feature in these diseases and epilepsy models is overexpression of brain A 2A receptors.Given that the A 2A receptor plays a critical role in mediating synaptic plasticity, particularly in hippocampal regions, aberrant A 2A receptor signalling is likely to adversely modulate the balance of long-term potentiation and depression, thus affecting memory (Rebola et al., 2008).
Disruption to A 2A receptor homeostasis may be a hallmark of diseases associated with cognitive decline, irrespective of aetiology, offering broad therapeutic potential.The A 2A receptor antagonist, istradefylline, is approved in some countries as an adjunctive treatment for Parkinson's disease (Navarro et al., 2016).As such, clinical evaluation of A 2A receptor antagonists for mitigating seizureassociated A 2A receptor excitotoxicity and improving memory impairments in epilepsy may be a promising therapeutic avenue.
In contrast to the A 2A receptor, the effects of the A 1 receptor on cognitive processes are less characterised and context-dependent.
Memory and learning processes in global A 1 receptor-KO mice are largely unaffected under control conditions (Gimenez-Llort et al., 2002;Gimenez-Llort et al., 2005;Zhou et al., 2018).However, when kindled with pentylenetetrazole, A 1 receptor-KO mice demonstrate greater memory impairments compared to wildtype kindled controls, suggesting the A 1 receptor may protect against the development of memory deficits during the epileptogenic process (Zhou et al., 2018).Similarly, mice exposed to chronic intermittent hypoxia experience spatial learning and memory deficits; impairments significantly exacerbated in mice with A 1 receptor KO or treated with an A 1 receptor antagonist (Zhang et al., 2020).Likewise, chronic treatment with an A 1 receptor antagonist induced memory impairment in mice, whereas an A 1 receptor agonist improved learning and memory (Vollert et al., 2013;Von Lubitz et al., 1993).In addition to the beneficial effects on learning and memory mediated by A 1 receptors, some studies have reported that A 1 receptors may also induce cognitive deficits.
Stimulation of the A 1 receptor enhanced the pentylenetetrazoleinduced cognitive impairment in mice (Homayoun et al., 2001).Moreover, A 1 receptor antagonism improved cognitive deficits associated with animal models of sleep deprivation, suggesting A 1 receptor signalling facilitates impaired cognition in these settings (Chauhan et al., 2016;Florian et al., 2011;Oliveira et al., 2019).In several other models, including models of tauopathy and drug-induced memory impairment, the A 1 receptor negatively modulated cognition (Boison et al., 2012;Dennissen et al., 2016;Lu et al., 2010;Pagnussat et al., 2015;Sousa et al., 2011).Therefore, the effects mediated by the A 1 receptor could depend on the (patho)physiological context in which it is studied, and the type of memory measured.Indeed, in the majority of the studies mentioned above, spatial memory was assessed with this memory type correlating with hippocampal function and integrity (Murphy, 2013).Moreover, the varied effects of the A 1 receptor on memory and learning have been suggested to be due to either enhancement or disruption of synaptic plasticity, particularly in the hippocampus (Chasse et al., 2021;Florian et al., 2011;Zhou et al., 2018).Given that patients with TLE often have hippocampal sclerosis and impaired cognition, it is interesting to speculate whether the A 1 receptor modulates these cognitive deficits (Murphy, 2013).This is particularly pertinent given that resected hippocampal tissue from TLE patients showed decreased expression of A 1 receptors and increased ADK expression, suggesting lowered extracellular adenosine concentrations and thus reduced A 1 receptor signalling capacity (Aronica et al., 2011;Glass et al., 1996;Masino et al., 2011).Indeed, the effects of altered ADK expression on cognition is also demonstrated in animal models, with transgenic mice overexpressing ADK showing significant impairments in memory and learning (Shen et al., 2012;Yee et al., 2007).Collectively, these findings suggest that epilepsy-induced modifications to the adenosine system in the human hippocampus could drive cognitive deficits.More pre-clinical data are required to delineate whether agonism (biased or otherwise), antagonism, or allosteric modulation of the A 1 receptor would prove advantageous for cognition in epilepsy models; however, any benefits of A 1 receptor antagonism are likely to be outweighed by the risk of worsening seizures.

| CONCLUDING REMARKS
In summary, the adenosine system plays an essential role in maintaining the homeostasis of neuronal excitability.A summation of the published results highlights a key role for adenosine in modulating seizure activity and the epileptogenic process, with animal and human studies demonstrating that dysregulation of the adenosine system may underlie the pathology of epilepsy.Given the dearth of new and effective ASMs, adenosine receptors, particularly the A 1 receptor, represent promising drug targets and provide scope to develop new and effective medicines.This is especially relevant in the setting of drug-resistant epilepsy, where there is a critical need for new and efficacious medications.Harnessing the ability of the A 1 receptor to mitigate seizure activity in settings where there is an unmet clinical need is thus of critical importance and may represent a novel therapeutic target for epilepsy.
The relationship between adenosine receptors and many of the associated comorbidities of epilepsy indicates that these GPCRs represent a hitherto unappreciated common link between epilepsy and its comorbidities (Boison & Aronica, 2015).Indeed, targeting adenosine receptors may represent a novel mechanism to control seizures and additional symptoms of the disease.Despite previous limitations and hurdles in the adenosine receptor drug discovery pipeline, we propose these receptors remain of high importance, particularly in the context of pursuing 'non-canonical' methods of GPCR targeting at the level of the receptor subtype, as well as its pharmacological mode of modulation.
Ben Rollo: Writing-original draft (supporting); writing-review and