Role of Adenosine in Epilepsy and Seizures

Abstract

Adenosine is an endogenous anticonvulsant and neuroprotectant of the brain. Seizure activity produces large quantities of adenosine, and it is this seizure-induced adenosine surge that normally stops a seizure. However, within the context of epilepsy, adenosine plays a wide spectrum of different roles. It not only controls seizures (ictogenesis), but also plays a major role in processes that turn a normal brain into an epileptic brain (epileptogenesis). It is involved in the control of abnormal synaptic plasticity and neurodegeneration and plays a major role in the expression of comorbid symptoms and complications of epilepsy, such as sudden unexpected death in epilepsy (SUDEP). Given the important role of adenosine in epilepsy, therapeutic strategies are in development with the goal to utilize adenosine augmentation not only for the suppression of seizures but also for disease modification and epilepsy prevention, as well as strategies to block adenosine A_2A receptor overfunction associated with neurodegeneration. This review provides a comprehensive overview of the role of adenosine in epilepsy.

Introduction

Because epilepsy has traditionally been considered a condition based on an imbalance of neuronal excitation and inhibition, conventional antiseizure drugs (ASDs) act by blunting neuronal transmission.¹ Unfortunately, those treatments are exclusively symptomatic, they do not work in more than one third of all persons with epilepsy, fail to affect underlying pathogenic processes, and also fail to affect the comorbidities of epilepsy.^2,3

In addition to the well-characterized impairment of neurotransmission and a shift in the excitatory/inhibitory balance toward more excitation, there is now growing consensus that epilepsy is a complex network disorder with a profound derailment of metabolic functions, as well as maladaptive responses of glial function and inflammatory processes which play a major role in the pathogenesis and pathophysiology of epilepsy.^4–10 Therefore, novel therapeutic approaches are needed. Network pharmacology offers new options to treat epilepsy more holistically, with the prospect of affecting underlying pathogenic processes (i.e., epileptogenesis)¹¹ and the potential to prevent the development of epilepsy or its progression.¹²

Because adenosine has multiple functions globally controlling allostasis,¹³ it is ideally suited to act as a prototypic network modulator. Through a combination of adenosine receptor (AR)-dependent and -independent effects, adenosine has emerged not only as a potent seizure suppressor¹⁴ but also as an antiepileptogenic, disease modifying agent.^15,16

Epilepsy and Seizures

Epilepsy is the fourth most common neurological disorder, affecting more than 70 million persons worldwide according to the International League Against Epilepsy (ILAE). Epilepsy is a complex group of cerebral disorders characterized by a lasting predisposition for the occurrence of unpredictable spontaneous and recurrent epileptic seizures that are a consequence of anomalous brain activity.^17–19 Accordingly, epilepsy exists when a person has an epileptic seizure and their brain “demonstrates a pathologic and enduring tendency to have recurrent seizures,”¹⁹ and an epileptic seizure is defined as “a transient occurrence of signs and/or symptoms due to abnormal excessive or synchronous neuronal activity in the brain”¹⁹ that lasts from 10 seconds to several minutes.²⁰ Ictogenesis is the generation of a seizure by an epileptic brain.²⁰ It is a progressive process that leads to synchronous high-frequency electroencephalographic activity, which triggers an epileptic seizure.²¹ Therefore, ictogenesis can occur spontaneously or be objectively related to a stimulus, as in cases of reflex seizures.^22–24

Although a large number of ASDs are available for controlling epileptic seizures, they are symptomatic treatments and fail to prevent epilepsy developments.²⁵ In addition, over 35% of persons with epilepsy have intractable epileptic seizures, in which ASDs have no effect on treating the symptoms.^2,3 A major risk factor for those with epilepsy is sudden unexpected death in epilepsy (SUDEP).^26–29

Epileptogenesis

Epileptogenesis is the process of progressive neurobiological alterations that converts a nonepileptic brain into a brain capable of generating spontaneous and recurrent seizures.^30,31

Approximately 20–60% of all epilepsies are caused by acute insults to the brain, such as traumatic brain injury, status epilepticus (SE), cerebrovascular insult, or infections, which trigger a cascade of events, including inflammatory processes, microglial, and astroglial activation, as well as epigenetic changes, which can trigger early seizures within 1 week after the injury.^10,32 However, after a primary acute insult to the brain, a “clinically silent” latent period of epileptogenesis is often observed that is characterized by a lack of a clinical phenotype but ongoing remodeling processes that can last from days to years between the insult and the onset of clinically defined epilepsy.¹⁰ During this seizure-free period, evidence of pathophysiological alterations in the brain are apparent in both patients and animal models of posttraumatic epilepsy.^10,32 These alterations include metabolic impairments, increased inflammatory processes, molecular reorganization, changes in gene and voltage-gated channels and receptor expression, astrogliosis, neurogenesis, aberrant synaptic plasticity, remodeling of neuronal circuits, axonal sprouting, neuronal loss, and epigenetic reprogramming that includes DNA hypermethylation.¹⁰

In addition, mesial temporal lobe epilepsy (MTLE), the most common form of acquired epilepsy in adults³³ with a high incidence of pharmacoresistence,³⁴ is frequently, but not always, associated with hippocampal sclerosis (HS), which includes hippocampal tissue atrophy and stiffening as a result of intense neuronal loss in CA1 and CA3 subfields and further aberrant astrocyte proliferation, known as reactive astrogliosis^35,36 as well as neuronal circuit rebuilding, which includes mossy fiber sprouting and granule cell dispersion.^34,37,38 These changes during epileptogenesis result in persistently increased neuronal excitability, which is a determinant for seizures and for epilepsy progression, including interictal discharges, high-frequency oscillations, and for augmented responses to stimuli.^30,39

Although for a long time, it was thought that epileptogenesis is a restricted process over a limited time, evidence is recently emerging that epileptogenesis is an extended and continuous process that continues beyond the expression of spontaneous seizures.^32,40 Therefore, it is becoming crucial to develop therapies addressing the epileptogenic process rather than only the symptoms of epilepsy.

Role of ARs in Synaptic Function, Synaptic Plasticity, and Synaptotoxicity in Epilepsy

The antiictogenic potential of adenosine has been recognized for the first time 45 years ago by identifying an inhibitory action of adenosine on neuronal activity.⁴¹ Since then, experimental studies have shown that maladaptative changes in G-protein coupled ARs (A₁ receptor [A₁R], A_2AR, A_2BR and A₃R) expression and signaling contribute to the pathophysiology of epilepsy.⁴²

The actions of adenosine in the brain are mostly mediated by A₁R and A_2AR, whereas the possible central roles of A_2BR and A₃R are still poorly studied.⁴²

ARs were first classified according to their differential coupling to adenylyl cyclase to regulate cyclic adenosine monophosphate (cAMP) levels. A₁Rs and A_3ARs are coupled to Gi/o proteins, leading to the inhibition of cAMP production and leading to the inhibition of presynaptic glutamate release. A₁R activation may also increase phospholipase C (PLC) activity and directly couple to and activate K⁺ channels and Q, N, and P type voltage sensitive Ca²⁺ channels.^42,43 In contrast, A_2ARs and A_2BRs are pleiotropic receptors, mostly coupled to Gs/olf proteins, increasing the production of cAMP, leading to the activation of protein kinase A (PKA), which in turn leads to the recruitement and opening of Ca²⁺ channels, increasing cell excitability.⁴²

In the brain, A₁Rs are widely distributed on neurons, mostly located in excitatory synapses,^44–46 astrocytes,⁴⁷ oligodendrocytes,⁴⁸ and microglia⁴⁹ in the cortex, hippocampus, and cerebellum.⁴² A_2ARs have a more restrictive and heterogenous distribution in the brain with higher levels in the nucleus accumbens shell, striatum, and tuberculum olfatorium,⁵⁰ whereas lower levels were found in hippocampus, cerebral cortex, amygdala, and cerebellum.⁵¹ A_2ARs are not only most densely located in synapses of neurons⁵² but also present in microglia^53,54 and astrocytes.^43,53,55

Detailed in vitro studies of circuits of one of the most afflicted regions in MTLE—the hippocampus—then followed⁵⁶ and shortly after, Dunwiddie and Worth in 1982⁵⁷ associated the anticonvulsant action of adenosine with A₁R activation. Since then, it has been repeatedly reported that endogenous adenosine, as well as systemically administered selective A₁R agonists, such as 2-chloro-N⁶-cyclopentyl-adenosine (CCPA) or N⁶-cyclohexyladenosine (CHA), possess anticonvulsant actions in a variety of acute and chronic models of epilepsy, which are prevented by A₁R antagonists such as 8-cyclopentyl-1,3-dimethylxanthine (DPX) or 8-cyclopentyl-1,3-dipropylxanthine (DPCPX).⁵⁸

Importantly, so far available information shows that A₁R activation is not only capable of preventing seizure occurrence when administered before the trigger but also to abort an already emerging seizure event, thus working as a true anticonvulsant substance.

The hyperpolarizing effect of A₁R activation is still considered one of the main mechanisms that contribute to the A₁R-dependent suppression of seizures, but other synaptic mechanisms likely coexist. In fact, presynaptic actions of A₁Rs control epileptiform activity by altering glutamatergic transmission, in accordance with the predominant presynaptic localization of the A₁R,^44,59 mainly in glutamatergic terminals.⁴⁶ This includes reduction of the probability of glutamate release through inhibition of voltage-dependent Ca²⁺ channels^60–62 or reduction of Ca²⁺-independent spontaneous release.^63,64

Postsynaptic actions are mostly related to a control of the recruitment of AMPA and NMDA receptors,⁶⁵ since both are implicated in the initiation and establishment of epileptic seizures.^66–69 In this sense, suppression of A₁R tonus in vitro causes sustained synaptically mediated epileptiform activity that has an AMPAR-dependent^70,71 and NMDAR-dependent component.⁷² A₁R also directly influence GABA function in several brain areas.⁷³ It is, however, important to stress that A₁Rs supress tonic extrasynaptic GABA_AR currents in pyramidal cells and in a specific population of postsynaptic GABAergic inhibitory neurons, the CB1R-expressing interneurons,⁷⁴ not affecting phasic GABAergic inputs to glutamatergic neurons.^75–77 These effects may be of particular importance during hyperexcitable conditions to reduce disinhibitoy actions of interneurons to pyramidal neurons.

When evaluating alterations of the adenosine modulation system that may prompt likely therapeutic targets, one concludes that both adenosine metabolism and ARs are modified. Thus, there is an increased activity of adenosine kinase (ADK), the major enzyme for the metabolic clearance of adenosine, which removes adenosine by phosphorylating it to AMP,⁷⁸ and an expected decreased formation of astrocyte-derived adenosine⁷⁹ associated with a lower tonic activation of A₁R in animal models of epilepsy reviewed in.⁸⁰ In parallel, there is a decreased density of A₁R in excitatory synapses in different animal models of epilepsy,^{45,46,81–83} as well as in patients with epilepsy.⁸⁴ This decrease of A₁R density results from the repeated activation of A₁R, which tend to develop tolerance in the control of excitatory transmission⁸⁵ and a gradual loss of their anticonvulsant capacity.⁸⁶

Furthermore, peripheral actions of systemically administered A₁R agonists, mostly related to cardiovascular adverse effects such as decrease in arterial blood pressure and heart rate,^87,88 have prevented the use of A₁R agonists for the treatment of central nervous system diseases.

With regard to the role of A_2AR and epilepsy, there is a false impression of inconsistency of results. In fact, major evidences indicate that the tonic activation of A_2AR by endogenous adenosine results in proepileptic effects, although the activation with exogenous agonists, such as 2-p-(2-carboxyethyl)phenethylamino-5′-N-ethylcarboxamidoadenosine hydrochloride (CGS21680), of different populations of A_2AR^89,90 can cause a paradoxical attenuation of seizure activity.

Some animal studies claimed that activation of A_2AR with exogenously added agonists 2-hexynyl-5′-N-ethylcarboxamidoadenosine (2HE-NECA)⁸⁶ or CGS21680^91,92 contributed to seizure suppression in acute models of seizure and that its blockade with the antagonist 3,7-dimethyl-1-propargylxanthine (DMPX) had proconvulsant actions⁹³; however, these findings probably result from the activation of A₁R rather than A_2AR since DMPX has an A_2A/A₁ selectivity ratio of 3 in rats and 5 in cloned human A₁R⁹⁴ and the anticonvulsant effect of CGS21680, as well as other central effects,⁹⁵ has been shown to be abrogated by the selective A₁R antagonist, DPCPX.⁹⁶

Most studies using both selective antagonists and genetic manipulations of A_2AR have consistently demonstrated proconvulsant effects of endogenously activated A_2AR.^97–101 Notably, the genetic ablation or pharmacological inhibition of A_2AR in animal models of epilepsy mostly prevents proconvulsant epileptogenic changes during evolution of kindling.^102–104 These results are indicative of a main role of A_2ARs in the cumulative aggravation of convulsions rather than in the onset of seizure activity.

The neuroprotective role of the A_2AR antagonist istradefylline (KW-6002) was also documented to dampen long-term deleterious consequences of early-life convulsions¹⁰⁵ and tinkering with the ability of A_2AR to control the migration of cortical interneurons during brain development in mice results in increased susceptibility to seizures.¹⁰⁶ Moreover, it has also been reported that genetic variations in the human adenosine A_2AR gene (ADORA2A), associated with increased expression of A_2AR and higher levels of cAMP production, is a predisposing factor for childhood encephalopathy following severe febrile seizures.¹⁰⁷ Overall, these evidences implicate a deleterious role of endogenous A_2AR activation in epilepsy.

A putative gain of function of A_2AR during epileptogenesis is best heralded by the reported increased density of A_2AR in different animal models of epilepsy,^{46,104,108,109} as well as in patients with epilepsy,¹¹⁰ which is most evident in excitatory synapses.^46,104 Clearly, multiple evidences suggest that both decreased A₁R and increased A_2AR function contribute to the establishment of epileptic conditions.¹¹¹ Thus, experimental data support a key role of A₁R as a hurdle to the spreading of abnormal neuronal function in animal models of epilepsy.^112–114 This prompts the suggestion that a combination of controlling A₁R desensitization¹¹⁵ and attenuating A_2AR function should be an effective strategy to control epilepsy.

Much less is known about the role of A₃R in epilepsy, in line with the current view that the central effects of adenosine mostly involve A₁R and A_2AR.¹¹⁶ Due to their lower affinity for adenosine, A₃R are activated only when relatively high concentrations of extracellular adenosine are reached, as occurs during high frequency neuronal discharges. Knowledge about the endogenous role of A₃R to epilepsy is still limited. While there is still conflicting information about whether the activation of A₃R shows a pro- or anticonvulsant effect,^93,117–121 it is suggested that A₃R alter the stability of GABA_AR, thereby fine-tuning neuronal excitability.¹¹⁸ A heterologous desensitization of A₁R has been described after activation of A₃R by high levels of adenosine, which could explain a possible proconvulsant effect of A₃R activation.¹²²

The scarce understanding of the role of A₃Rs in epilepsy, together with some evidence for their dual role in seizure control, makes it even more difficult to understand their mechanism of action. Not surprisingly then, very few studies addressed the synaptic mechanisms that might be linked to A₃R in epilepsy. These effects may be direct or indirect, mediated through A₁Rs, influencing glutamatergic transmission presynaptically or postsynaptic K⁺ channel function,¹²² or even causing desensitization of GABA_ARs.¹²³

AR-Independent Effects of Adenosine

Epigenetic changes such as histone modification, DNA methylation, and noncoding RNAs have been recently identified as novel AR-independent roles of adenosine. These epigenetic changes play important roles in controlling gene expression without changing the DNA sequence.¹⁶ Thereafter, alterations due to these factors have been linked to several neurological disorders.¹²⁴ In particular, DNA hypermethylation has been associated with the development of epilepsy giving rise to the methylation hypothesis of epileptogenesis.^125–131

In addition to the receptor-mediated effects of adenosine mentioned above, adenosine can act as an endogenous regulator of DNA methyltransferase activity.¹⁶ Adenosine is a product of the transmethylation pathway, which ultimately converts S-adenosylmethionine (SAM) into adenosine and homocysteine via methyltransferases and S-adenosylhomocysteine (SAH) hydrolase.¹³² As a product of the transmethylation pathway, changes in adenosine levels can have a major effect on DNA methylation, which proceeds only if adenosine and homocysteine are effectively metabolized by ADK.¹³² Therefore, dysregulation of the intracellular metabolism of adenosine during epileptogenesis is directly linked to epigenetic changes thought to drive the development of epilepsy.

Disruption of Adenosine Metabolism in Epilepsy

Astroglial dysfunction in epilepsy

Astrocytes are the most abundant cells in the brain of higher mammalian species.¹³³ Among other functions, astrocytes provide energetic and homeostatic support for neurons¹³⁴; guide the formation and function of synapses^5,135–137; and also perform a crucial role in the blood–brain barrer creating a tight interface in the brain that regulates the exchange of molecules in and out of the brain to maintain homeostasis.¹³⁸

Astrogliosis, also known as reactive astrogliosis, is a prevalent pathological hallmark of many forms of epilepsy, including MTLE,¹³⁹ and refers to a broad spectrum of functional and structural alterations in glial cells, particularly astrocytes and microglia.¹⁴⁰ In MTLE, an important feature of astrogiosis is HS that is characterized by loss of neurons in the hippocampus, induction of hypertrophy of astroglial cell bodies and processes, upregulation of the expression of several proteins, including intermediate filament proteins such as glial fibrillary acidic protein (GFAP) and vimentin. In addition to cellular proliferation, the formation of a glial scar rich in chondroitin sulfate and proteoglycans is also observed in HS.^141,142

Functionally, astrogliosis may be implicated in seizure generation and also, most importantly, in epileptogenesis.^5,136 For example, dysregulation of astrocytic activity may increase epileptiform activity through the release of gliotransmitters, including glutamate, D-serine, and ATP, which increases the excitability of neurons.¹⁴³ Therefore, impairment of the homeostatic control of network excitability can be a consequence of astrocyte activation. This might involve impairment of K⁺-buffering,¹³⁶ dysfunctional gap junctions,¹⁴⁴ dysfunctional glutamate homeostasis,¹⁴⁵ changes in calcium buffering and gliotransmitter release,¹⁴⁶ and dysfunctional adenosine homeostasis^5,8,147 and modulation.¹⁴⁸ On the contrary, astrogliosis may occur as a compensatory mechanism.

Astrocytes can reduce neuronal excitability by facilitating the reuptake of excessive glutamate released during a seizure.¹⁴³ Moreover, astrogliosis alone seems to be capable of inducing spontaneous recurrent seizures.¹⁴⁹

The ADK hypothesis of epileptogenesis

As mentioned before, ADK is the major enzyme for the metabolic clearance of adenosine by phosphorylating it to AMP.⁷⁸ In the adult brain, ADK is predominantly expressed in astrocytes.¹⁵⁰ After acute insults, such as a SE or a stroke, ADK is rapidly downregulated, contributing to a large astrocytic-derived surge in extracellular adenosine, which serves to terminate seizures and to provide neuroprotection.^78,151 On the contrary, high amounts of ATP-derived adenosine released into the synaptic cleft can contribute to synaptotoxicity where adenosine activates A_2AR on neurons and leads to an increase of intracellular calcium concentrations.^{71,104,152–154}

During the clinically silent latent period of epileptogenesis, the inflammatory responses triggered after the acute insult lead to astrogliosis.^6,155 Moreover, experimental data show an important association of astrogliosis with overexpression of ADK, which results in a chronic deficiency in the availability of adenosine.¹⁵⁶ Accordingly, results obtained from studies in both animal models of epileptogenesis¹⁴⁷ and specimens from human patients with MTLE associated with HS, who underwent surgical resection of their hippocampus as a treatment of epilepsy, showed that ADK is robustly overexpressed in those brain areas involved in epileptogenesis.^157–160 Indeed, ADK overexpression by itself is sufficient to trigger seizures.^151,160

Evidence described by Li et al. attest that transgenic Adk-tg mice, that intrinsically overexpress ADK, experience spontaneous recurrent seizures limited to astrogliotic hippocampal CA3 that were similar in rate, frequency, and duration to those in a mouse model of intra-amygdaloid kainic acid (KA)-induced focal epilepsy.¹⁶¹ Taken together, these observations lead to the hypothesis that overexpression of ADK is a crucial mechanistic contributor to the development of epilepsy.

Altered purinergic modulation of synaptic function in epilepsy

Despite the evidence supporting a role for ADK modifications during epileptogenesis, the link between the altered ADK activity during epileptogenesis and the alteration of neuronal function to trigger the hyperexcitability and hypersynchronicity characteristics of circuits engaged in epileptic seizures is still poorly understood. A possible involvement of the purinergic system in this astrocyte-to-neuron adaptive process is proposed, based on the role of astrocytic ATP-derived adenosine to activate A₁R to control heterosynaptic depression,^{69,146,162,163} the alteration of the function of A_2AR in excitatory synapses upon tinkering with astrocytes,¹⁴⁸ with astrocytic ADK¹⁶⁴ or ARs.¹⁴⁸ However, the exact mechanisms underlying a possible role of purines in the astrocyte-to-neuron adaptive processes underlying epileptogenesis still remain to be experimentally scrutinized.

Irrespective of the trigger(s) of the alterations engaging epileptogenesis, it is clear that epileptic phenotypes are cumulative in the sense that seizures aggravate subsequent seizures in a process designated as seizures beget seizures; conversely, the control of seizure activity is a sine qua non condition to successfully manage epilepsy and hopefully treat these disorders.^165,166 This implies that critical factors governing epileptogenesis should be identified upon triggering a convulsant episode in naive animals and these factors might be targeted to develop antiepileptic strategies since epilepsy can be pharmacologically controlled.

As mentioned before, the main brake of excitatory transmission and neuronal excitability, A₁Rs, have their density decreased in the epileptic brain. In contrast, increased neuronal activity and seizure-like firing patterns trigger a sustained augmentation of the release of synaptic ATP^45,167–170 and of the density and activity of ecto-5′-nucleotidase (ecto-5′-NT).^{45,110,171,172} Thereafter, ATP and ecto-5′-NT are responsible for ATP-derived adenosine formation and for the selective activation of synaptic A_2AR.^154,173,174 The A_2ARs are the main bolsterer of excitatory transmission and neuronal excitability and have the expression increased in epilepsy, being a critical player in linking increased excitability with neurodegeneration.¹⁰⁴ However, it is not yet clear if any of these changes is causative of epilepsy or if they are an adaptive consequence of epilepsy.

Aging—Altered Adenosine Modulation with Increased Incidence of Epilepsy

Epidemiological evidences show that the incidence of seizures and epilepsy is dramatically increased in persons older than 60 years. This may be due to a higher risk of older people to suffer a stroke or a head trauma as a result of a fall, leading to the development of cryptogenic or poststroke, and posttraumatic epilepsy.¹⁷⁵

Another hypothesis argues that the prevalence of higher rates of epilepsy in elderly individuals may also be linked to alterations of the adenosine modulation system.^176–178 In addition to the alterations of excitatory and inhibitory transmission systems¹⁷⁹ as well as numerous neuromodulation systems,¹⁸⁰ there is a modification of the adenosine modulation system upon aging that matches the alterations found in epileptic animal models and patients. In fact, there is a decrease of the extracellular levels of adenosine with aging.^168,181 which is paralleled by a decreased density and function of inhibitory A₁R in aged rodents^{180,182–186} and humans.^187,188 In addition, in aged brain preparations, there is an increased release of ATP from recruited synapses^168,189 with an increased density and activity of ecto-5′-NT^168,190,191 and an increased density and function of facilitatory A_2AR in animal models of aging^{180,192–195} and in humans.¹⁵³

In particular, the overfunction of A_2AR is deleterious for brain function^196–198 and was recently proposed to be a causative factor of an imbalance of excitatory transmission by generating a hippocampal long-term depression to long-term potentiation (LTD-to-LTP) shift accompanied by increased NMDA receptor gating and enhanced Ca²⁺ influx dependent of mGluR5¹⁵³ precipitating synaptic alterations that increase frailty of neuronal networks¹⁹⁹ and accelerating neuronal dysfunction in animal models of brain diseases.²⁰⁰ This striking parallel supports the contention that a synaptic dysfunction of the adenosine modulation system might be a triggering factor for the increased susceptibility to epilepsy in the elderly.

Sudden Unexpected Death in Epilepsy

SUDEP is defined as “sudden, unexpected, witnessed or unwitnessed, non-traumatic and non-drowning death”²⁰¹ and is a significant comorbidity for persons with epilepsy. In most witnessed or recorded cases, sudden deaths were immediately preceded by a seizure.²⁰² SUDEP can affect individuals of any age, but is most common in young adults (aged 20–45 years) with an incidence of 1.2 per 1000 person-years.²⁰³

The biggest risk factor for SUDEP are generalized tonic-clonic seizures,²⁰³ although other risk factors have been discussed, such as cardiac problems and respiratory arrest. Accordingly, respiratory problems were observed in most cases of SUDEP.^27,204 Indeed, epileptic seizures can induce significant hypoventilation.^205,206 However, apnea and O₂ desaturation are more common than previously realized after generalized partial or convulsant seizures,²⁶ and recent evidence supports a significant role of apnea and hypoventilation in SUDEP.²⁰⁷

Respiration is controlled mainly by two neuronal groups located at the brainstem: (1) the parafacial respiratory group (pFRG) and (2) the pre-Bötzinger complex (inspiratory pacemaker population).²⁰⁸ The brainstem also contains nuclei responsible by other vital functions, such as the vagal nuclei, which significantly participates in the central control of cardiac activity.²⁰⁹ Molecularly, the control of respiration depends on the availability of a number of neuroactive substances, such as serotonin (5-HT), adenosine, and endogenous opioid peptides, that are released during seizures and whose concentrations are altered during the phase of postictal depression.²⁶

Experimental evidences show that mice that had 5-HT_2c receptor genetically deleted presented spontaneous generalized seizures that developed to a tonic extension phase and rapidly died.²¹⁰ In addition, according to the “adenosine hypothesis of SUDEP,” seizure-induced adenosine release with impaired metabolic clearance by astrocytes, specifically in the brainstem may trigger lethal apnea or cardiac arrest. Accordingly, AR antagonists, such as caffeine or theophylline might prevent SUDEP.²¹¹ Therefore, serotonin and ARs, or other components of those pathways, may be promising targets for pharmacological SUDEP prevention therapy.^26,27,212

Adenosine Augmentation Therapies for Epilepsy

Pharmacological approaches

Although seizure activity causes a robust elevation of adenosine levels in afflicted brain regions,^213,214 the global augmentation of extracellular adenosine still emerges as a promising strategy to limit the spreading of seizure activity to still naive brain areas reviewed in.¹¹¹ ADK inhibition is an ideal strategy for adenosine augmentation for several reasons. First, ADK is the major metabolizing enzyme of adenosine.⁷⁸ In tissue studies, the K_m for adenosine is 1–2 orders of magnitude lower than that of adenosine deaminase (ADA). Thereby, a minor change in ADK activity can result in a major change in adenosine concentration.^78,215 In agreement with this, ADK inhibition, but not ADA inhibition, increased endogenous adenosine and neuronal inhibition in hippocampal slices²¹⁶ and supressed bicuculine-induced seizures in rats.²¹⁷

Furthermore, there is a pathological upregulation of ADK expression at an astrocyte enriched seizure focus in the epileptic brain in animal models and in tissue resected from humans with temporal lobe epilepsy.¹⁵⁷ Consequently, pharmacological inhibition of ADK with 5-iodotubercidin was able to suppress seizures in the KA mouse model of pharmacoresistant epilepsy.²¹⁸ Past drug discovery efforts have focused on finding a clinically useful ADK inhibitor^219,220 and both effective nonnucleoside and nucleoside inhibitors have been identified.^221–224 Although highly efficient in suppressing seizures, even those resistant to common ASDs,¹⁴ the clinical use of ADK inhibitors has been hampered by cardiovascular and sedative side effects of systemic adenosine augmentation.^78,224

Cell-based adenosine delivery

To avoid side effects of systemic adenosine augmentation, local adenosine delivery approaches have been developed with the goal to increase adenosine concentrations locally in the brain in the vicinity of a seizure producing focus.²²⁵

In a pioneering proof of principle study, it was shown that encapsulated cell grafts of baby hamster kidney fibroblasts engineered to release adenosine based on the disruption of ADK expression, effectively suppressed seizures after implantation into the brain ventricles of rats submitted to electrical kindling.²²⁶ This strategy of local adenosine augmentation demonstrated that the focal paracrine release of adenosine was not only sufficient to suppress seizures²²⁷ but also to avoid sedative side effects associated with systemic ADK inhibition or A₁R activation.²²⁸

Cell-based adenosine release for seizure suppression in epilepsy was further refined by engineering mouse embryonic stem cells to lack both alleles of the Adk gene. Those Adk^−/− embryonic stem cells were differentiated into adenosine-releasing neural precursor (NP) cells and transplanted into the infrahippocampal cleft of rats 1 week before the onset of hippocampal electrical kindling.²²⁹ The Adk^−/− NP cell recipient rats showed reduction of epileptogenesis induced by electrical kindling, and furthermore, 48 hours after kindling stimulations, these rats did not develop generalized seizures, suggesting for the first time a novel antiepileptogenic effect of adenosine.²²⁹

In a different study, the same Adk^−/− NP cells were transplanted into the infrahippocampal cleft of mice 24 hours after receiving an intra-amygdaloid injection of KA. Mice, which received the implant of Adk^−/− NP cells did not develop any spontaneous seizures, and when analyzed after 3 weeks of the KA injection, it was seen that astrogliosis was significantly reduced and ADK expression levels were close to normal when compared to control or sham animals.²³⁰ These latter two studies for the first time suggested a hitherto unknown antiepileptogenic potential of adenosine releasing cellular brain implants.

In an attempt to generate a cell line compatible for future clinical applications, Ren et al. used a model of ADK ablation in human mesenchymal stem cells based on the lentiviral expression of a microRNA (miRNA) directed against ADK, achieving around 80% reduction of ADK expression in these cells.^231,232 Transplantation of these cells into mice before the injection of KA significantly reduced the acute brain injury and seizures in these mice.²³²

Gene therapy

Gene therapy focuses on the attenuation of the endogenous ADK gene. The logic is as follows: by reducing the activity of ADK, which is overexpressed in an epileptic brain, adenosine levels will increase and ultimately exert their anticonvulsant effects. Gene therapy is a logical avenue of treatment as it corrects the pathology, being the overexpression of ADK, at the genetic level rather than just suppressing symptomology, which is typically the case in pharmacological treatments.

Promising results were seen in a study that utilized an adeno-associated virus (AAV) vector that expressed ADK-complementary DNA (cDNA) in antisense orientation under the control of gfaABC1D promotor was injected unilaterally into the CA3 region of the hippocampus to knock down ADK in astrocytes in adult male Adk-tg mice.²³³ The mice showed a significant suppression of local seizure activity ipsilateral to the injection side with 0.6 ± 0.6 seizures/h compared to 5.8 ± 0.5 seizures/h in the noninjection contralateral side.²³³ These results show that antisense-mediated knockdown of ADK in astrocytes is an effective strategy for the suppression of seizure activity.²³³

Dietary interventions

An additional strategy to augment adenosine levels is through a ketogenic diet (KD). A KD, which consists of a high-fat, low-carbohydrate intake, produces an increase in ATP levels within the brain, ultimately increasing adenosine concentration as one of its beneficial mechanisms.²³⁴

When carbohydrate intake is limited, as in a KD, blood glucose levels decline causing the liver to convert fatty acids to ketone bodies, which enable a more efficient ATP production. Since a small increase in ATP translates into a disproportional larger increase in adenosine concentration reviewed in,²³⁵ the KD becomes an attractive method of augmenting adenosine as a therapy for epilepsy. Accordingly, an in vitro study showed that a mimic of KD afforded a hyperpolarizing effect in hippocampal CA3 pyramidal cells (a cell type and location involved in seizures), through a combined increased activation of A₁R and opening of ATP-sensitive potassium channels in response to reduced glucose levels.¹⁵⁸

Through the augmentation of adenosine, KD offers not only anticonvulsant effects, but antiepileptogenic effects as well. In fact, mice fed a KD were significantly less susceptible to chemical kindling epileptogenesis than mice fed a control diet, indicating that the KD delayed the development of epilepsy.²³⁶

To determine the longevity of the antiepileptogenic effects of the KD, the KD-fed mice and PTZ-kindled mice were administered a single injection of glucose, which is known to terminate acute impacts of the KD and model a diet reversal. After 4 days, the mean Racine's scale seizure score of the two KD groups were alike and significantly lower than the control group, demonstrating a lasting antiepileptogenic impact.²³⁶

In a long-term study performed in the pilocarpine model of epilepsy in rats, it was shown that in KD-treated rats a lasting reduction of spontaneous recurrent seizure activity was achieved even 6–8 weeks after diet reversal, whereas control diet exposed rats experienced robust prolonged seizures with varying patterns and intensity.²³⁶ The KD-fed rats had reduced levels of DNA methylation compared with the control diet fed rats even after diet reversal, which supports that the reduction in seizures is linked to an antiepileptogenic disease-modifying mechanism induced by epigenetic changes.²³⁶ This study warrants further research of the antiepileptogenic effects of a KD, specifically in controlled clinical studies. Indeed, the few available clinical studies indicate that a KD has significant disease-modifying impacts, leaving some patients seizure-free even after discontinuation of the KD.^237,238

Notably, the KD has been most successful in mediating seizures in pediatric epilepsy. In a randomized controlled clinical trial, the impacts of a 3-month long trial of a KD was studied in children between the ages of 2 and 16 years, diagnosed with either generalized or focal epilepsy who had at least a seizure per day.²³⁹ The results showed that 38% of patients displayed a >50% reduction, with some reaching as high as >90% reduction.²³⁹

Another pediatric study, which focused on children with pharmacoresistant epilepsy, showed after introducing a modified Atkins diet (a less restrictive variation of the KD), 30% of children experienced 90% seizure reduction and 52% of children experienced 50% seizure reduction after 3 months.^238,240 In an adult model, 32% of adult patients with pharmacoresistant epilepsy showed >50% seizure reduction using a KD.^237,238

While the therapeutic impact of adenosine augmentation is the focus of this section, it is important to note that the KD offers seizure control through other mechanisms as well. There is evidence that the presence of ketone bodies, free fatty acids, and overall glucose restriction has anticonvulsant properties and can lead to seizure protection.²⁴¹ In addition, the KD offers a variety of other physiological benefits, especially for adult patients, including reduction of comorbidities, enhancing adenosine's role as a neuroprotective molecule, aiding in sleep, and a significant role in pain reduction.²³⁴

Strategies for Epilepsy Prevention

Epilepsy prevention is the Holy Grail of epilepsy research and therapy development. Aproximately 15% of all diagnosed epilepsies are caused by an acute acquired brain insult, such as traumatic brain injury, stroke, or encephalitis.^10,242 The latent period following a potentially epilepsy triggering event presents a unique opportunity for preventive interventions. Several lines of evidence show that the transient therapeutic augmentation of adenosine has lasting antiepileptogenic and disease-modifying effects.

The first indication that adenosine has antiepileptogenic effects was provided in the cell transplantation studies discussed above. Adenosine-releasing hippocampal cell grafts suppressed hippocampal electrical kindling development in rats²²⁹ and epilepsy development in the mouse intraamygdaloid KA model.²²⁵

To understand the mechanisms behind those antiepileptogenic effects, the biomaterial silk was engineered to release a defined dose of adenosine for a restricted time span of 10 days.²⁴³ As expected, adenosine releasing silk implanted into the lateral brain ventricles of rats suppressed electrically induced seizures in rats submitted to electrical kindling.²⁴³

The same implants were introduced into the brain ventricles of rats 9 weeks after the systemic injection of KA. At this timepoint, all rats had experienced at least four to six spontaneous convulsant seizures, had overexpressed ADK, and increased DNA methylation levels in the hippocampus. The acute release of adenosine (during 10 days) suppressed DNA methyltransferase activity and reduced DNA methylation levels to baseline levels seen in control animals. Importantly, normalization of DNA methylation status was maintained even after cessation of the adenosine release. While seizure rate (seizures per week) and intensity (duration) continuously increased in all sham or control silk-treated animals, the development of epilepsy was completely stalled for at least 3 months in animals that received the adenosine-releasing brain implants.

In parallel to the suppression of seziure development further, sprouting of mossy fibers was blocked. In line with suppression of epilepsy progression, DNA methylation levels were found to be normalized. This study identified a novel adenosine-dependent epigenetic mechanism responsible for the prevention of epilepsy progression.¹⁶

Adenosine-releasing silk might not be the most practical approach for clinical translation. Therefore, ADK inhibitors might provide a better alternative for epilepsy prevention. A recent study demonstrated that a transient dose of the ADK inhibitor 5-iodotubercidin given systemically from 3 to 8 days after the intrahippocampal injection of KA prevented epilepsy development in 50% of all of tested mice.¹⁵ This is an important finding, because the transient use of an ADK inhibitor would be acceptable in light of side effects associated with chronic inhibition of ADK.

Apart from these strategies based on the augmentation of adenosine, there is also recent evidence showing that the pharmacological blockade of A_2AR can also interupt the casacde of events converting abnormal brain activity into neurodegeneration,¹¹¹ a key process in the development of circuit remodeling and establishment of epilepsy.¹⁶⁵ In fact, a recent study identified a key role of A_2AR linking increased excitability with neurodegeneration in a kainate model of temporal lobe epilepsy,¹⁰⁴ in accordance with the ability of A_2AR blockade to dampen the proconvulsant epileptogenic changes during evolution of kindling.^102–104

This neuroprotective role of A_2AR antagonists was also documented in animal models for long-term deleterious consequences of early-life convulsions¹⁰⁵ and also in clinical findings with the report that genetic variation in human adenosine A_2AR gene (ADORA2A) associated with increased expression of A_2AR and higher levels of cAMP production is a predisposing factor for childhood encephalopathy following severe febrile seizures.¹⁰⁷ It is worth mentioning that the interest of considering the use of A_2AR antagonist as coadjuvants to manage the evolution of epilepsy is particularly attractive in view of the ability of these A_2AR antagonists to control the most common comorbibities with epilepsy, namely memory impairment,^244–248 mood dysfunction,^249–251 migrane,^252,253 and sleep disorders.^254,255

Conclusions and Outlook

There is now solid evidence that the adenosine system is dysregulated on several different levels in epilepsy. In particular, maladaptive changes in adenosine metabolism can explain the generation of seizures (ictogenesis), as well as comorbidities of epilepsy including SUDEP, whereas A_2AR overfunction seems paramount for the emergence of neurodegenerative consequences of seizure activity and circuit remodeling to implement an epileptic condition. Consequently, adenosine-based therapies are based on a solid rationale to directly affect multiple mechanisms involved in the expression of epilepsy.

Novel findings document the therapeutic potential of A_2AR antagonists as well as AR-independent roles of adenosine as an epigenetic regulator, which form the basis for the development of antiepileptogenic therapies. The development of novel compounds acting on the nuclear form of ADK might capitalize on the epigenetic effects of adenosine without causing AR-mediated effects that still remain to be tested in conjunction with A_2AR antagonists to maximize the therapeutic possibilities of targeting the adenosione system to manage epilepsy. Future work needs to determine whether those strategies are safe and also effective in clinically realistic models of posttraumatic epilepsy.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The authors acknowledge grant support from the National Institutes of Health to D.B. (NINDS: NS103740, NS065957); support to A.M.S. by LISBOA-01-0145-FEDER-007391, project cofunded by Fundo Europeu Para o Desenvolvimento Regional through POR Lisboa 2020 (Programa Operacional Regional de Lisboa) from PORTUGAL 2020 and Fundação para a Ciência e Tecnologia (FCT), by an FCT project (PTDC/MED-FAR/30933/2017) and by Twinning action (SynaNet) from the EU H2020 programme (project number: 692340); and support to R.A.C. and A.R.T. by Fundacion LaCaixa (LCF/PR/HP17/52190001), Centro 2020 (CENTRO-01-0145-FEDER-000008: BrainHealth 2020 and CENTRO-01-0246-FEDER-000010) and FCT (POCI-01-0145-FEDER-03127).