An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Noebels JL, Avoli M, Rogawski MA, et al., editors. Jasper's Basic Mechanisms of the Epilepsies. 5th edition. New York: Oxford University Press; 2024. doi: 10.1093/med/9780197549469.003.0030
Neuroinflammation and epilepsy appear to exist in a self-reinforcing circular arrangement, whereby inflammatory pathways lower seizure threshold and the ensuing seizures trigger neuroinflammation. This forms the essential logic of anti-inflammatory therapies, which seek to interrupt the vicious cycle or to prevent it. Many if not all neuroinflammatory mediators are known to play physiological roles in the healthy brain, for example, regulation of astroglial potassium buffering by TGFβ, and the strength of excitatory synaptic transmission by PGE2 and IL-1β ανδ TNΦ. These neuromodulatory roles are usurped in epilepsy, when the neuroinflammatory mediators are transformed from their physiologic functions to agents of havoc as their local concentration rises or they persist too long in brain tissue. This chapter focuses on IL-1β, TNF, IL-6, PGE2, CXCL1, CCL2, and drivers of the complement pathway, which have all received substantial attention over the past decade. This chapter describes these inflammatory mediators and discusses how they may create a pathological milieu permissive for seizures, cell loss, and behavioral dysfunctions.
Classical inflammation outside the nervous system proceeds by recruitment of immune cells to the inflamed tissue, which leads to the well-known triad of increased local blood flow, swelling, and sensitized pain receptiveness. Neuroinflammation is a multifaceted brain state that engages many pathophysiological and homeostatic activities that are only partially understood. Neuroinflammation is common in epilepsy, often focal but sometimes widespread, characterized by activation of innate immune cells and brain resident cells (e.g., microglia, astrocytes, pericytes, and endothelial cells). These activated cells produce inflammatory cytokines and chemokines that both disrupt normal physiology and attract peripheral immune cells (monocytes/macrophages and neutrophils, occasionally T- and B-cells), which may exacerbate the inflammatory microenvironment. Neurons are not always just victims, but their excessive activity during seizures or after various sterile injuries can drive so-called neurogenic inflammation (Xanthos and Sandkuhler, 2014) by releasing molecules (e.g., prostaglandin E2, danger signals such as ATP and High Mobility Group Box 1 [HMGB1], glutamate, fractalkine) that activate autocrine receptors or receptors on glial cells and brain endothelium. The homeostatic purpose of neuroinflammation appears to be repair of brain injury; for example, microglia activated by tissue injury recognize and engulf damaged or dead neurons and other cells. However, persistent elevation of cytokines such as IL-1β and TNF can lower the seizure threshold by altering homeostasis of synaptic and intrinsic neuronal excitability. Similarly, overly zealous synaptic pruning or displacement by activated microglia can modify the normal function of cortical pathways (Chen et al., 2014b).
We focus here on the triggers of neuroinflammation, the pathways involved, and how the disruption of normal physiology by inflammatory pathways can influence established epilepsy and its development and progression (i.e., epileptogenesis). Related topics will be covered in more depth in other chapters, including control of blood–brain barrier (BBB) integrity by TGF-β signaling; inflammatory biomarkers in blood; epilepsy therapies targeted to inflammatory pathways; role of neuroinflammation in comorbidities of epilepsy (e.g., depression); autoimmune epilepsies; and the interface between inflammation and oxidative stress.
Evidence exists for a strong association between a peripheral inflammatory disease and the subsequent development of epilepsy (type unspecified) in a Taiwanese cohort of 32,000 rheumatoid arthritis patients compared with a similar number of age- and sex-matched controls (Chang et al., 2015). Overall, arthritis patients were at elevated risk of developing epilepsy during the 6–12 years after their diagnosis compared with age-, sex-, and comorbidity-matched controls (adjusted hazard ratio [AHR] = 1.52). However, when the rheumatoid arthritis cases were stratified in quartiles according to the duration of nonsteroidal anti-inflammatory drug (NSAID) use, those who had been using NSAID for less than 1.5 years had the highest risk of developing epilepsy (AHR = 2.57), and risk declined with more prolonged NSAID use (Fig. 30–1A). Similarly, a study of 843,000 adult patients hospitalized for sepsis revealed a 5.25-fold increased risk of experiencing seizures that required hospitalization or emergency department visits over the subsequent 8 years (Reznik et al., 2017; Fig. 30–1B). Peripheral inflammation often induces neuroinflammation, and the conclusions of these epidemiological studies are reinforced by clinical data that demonstrate activation of microglia, astrocytes, neurons, and perivascular macrophages in the brain of patients with temporal lobe epilepsy (TLE) and epileptogenic malformation of cortical development, and elevated cytokines and COX-2 in these conditions (Arena et al., 2019; Broekaart et al., 2018; Ravizza et a., 2006, 2008; Zattoni et al., 2011). Finally, as described below, a wealth of evidence documents pathogenic roles for neuroinflammatory pathways in animal models of epilepsy.
Neuroinflammation as a risk factor for epilepsy. Panel A, Increased incidence of epilepsy over a 6–12 year period after diagnosis of rheumatoid arthritis is offset by prolonged used of NSAIDS (from Rojas et al., 2019, modified from Chang (more...)
Upon cell activation, glia synthesize and release numerous neural mediators, including IL-1β, TNF, IL-6, CXCL8/IL-8, CXCL1, CCL2, and HMGB1, whereas hyperactive neurons synthesize COX-2 and release prostaglandins, HMGB1, ATP, and fractalkine. What is the normal function of these proinflammatory molecules? Many inflammatory mediators fine-tune synaptic transmission in nonpathological conditions, including roles for IL-1β, IL-6, and TNF in synaptic plasticity that underlies learning and memory (e.g., del Rey et al., 2013; Vezzani and Viviani, 2015). Neural networks in the healthy adult brain are being continuously modified by experience. Under physiological conditions, these pathways regulate the remodeling of neural circuits that form long-term potentiation and neurogenesis. At low concentrations, proinflammatory molecules, particularly IL-1β, IL-6, TNF, and prostaglandins, are important neuromodulators in this form of synaptic plasticity (Gruol, 2015; Vezzani and Viviani, 2015; Yirmiya and Goshen, 2011). A major unsolved question is whether elevated and sustained levels of these same mediators, as would occur after seizures or an injurious epileptogenic trigger, engage different pathways or drive their normal physiological pathways in a maladaptive manner.
Several cytokines that are main constituents of the inflammatory microenvironment in epileptogenic brain regions activate nonconventional intracellular signaling pathways in target brain cells, which may result in cellular dysfunction. These actions ultimately affect neuronal network excitability and neuronal cell survival. Below, we review the main molecular mechanisms underlying the neuromodulatory effects exerted by pro-inflammatory cytokines, focusing on those with a demonstrated role in the mechanisms of seizure generation in experimental models and in some clinical syndromes.
Acute and chronic seizures and epileptogenic brain injuries in animal models cause a rapid and long-lasting increase in the expression of IL-1β and its cognate receptor IL-1 receptor type 1 (IL-1R1) in ictogenic brain regions chiefly involving glia, neurons, and endothelial cells of brain vessels (van Vliet et al., 2018). These changes similarly occur in human drug-resistant epilepsy foci resected for therapeutic purposes in structural focal-onset epilepsies (Vezzani et al., 2019).
The activation of the IL-1β-IL-1R1 axis promotes seizure generation in animal models of acute seizures, status epilepticus, and in chronic epilepsy. The activation of this signaling contributes also to neurodegeneration and neurological comorbidities, as assessed with pharmacological interventions or (epi)genetic approaches aimed at either blocking or exacerbating this pathway activation (van Vliet et al., 2018; Vezzani et al., 2019).
A search for the molecular mechanisms underlying these pathologic effects revealed a specific functional interaction of IL-1β with glutamatergic and GABAergic neurotransmission (Vezzani and Viviani, 2015). Moreover, the cytokine can modulate voltage-gated and receptor-operated ion channels (Vezzani and Viviani, 2015). These interactions directly involve neurons bearing IL-1R1, or they are indirectly mediated by astrocytes.
Activation of the IL-1β-IL-1R1 axis in the mouse hippocampus was shown to enhance the tyrosine phosphorylation of the NR2B subunit of the NMDA receptor by inducing ceramide-mediated Src protein kinase activation (Balosso et al., 2008). This chain of events resulted in increased glutamate-mediated Ca2+ influx into hippocampal neurons, ultimately leading to excitotoxicity (Viviani et al., 2003). This pathway was subsequently shown to be involved in the in vivo ictogenic effects of both IL-1β and of the danger signal HMGB1 (Balosso et al., 2008, 2014; Maroso et al., 2011).
Another mechanism by which IL-1β may increase neuronal excitability is inhibition of NMDA-induced outward current in CA1 pyramidal cells, involving p38-MAPK and the subsequent phosphorylation of Ca2+-dependent K+ channels (Zhang et al., 2008). It has been shown that the cytokine also down-regulates the surface expression of the GluA1 subunit of the AMPA receptors, an indirect effect dependent on NMDA receptor activation (Lai et al., 2006).
Functional interactions between IL-1β and the GABAergic system were also reported. The application of IL-1β to rat hippocampal slices reduced synaptic inhibition onto CA3 pyramidal cells (Zeise et al., 1997) and decreased the peak magnitude of GABA-evoked current in cultured hippocampal neurons by activating protein kinases (PK) other than PKC (Wang et al., 2000). Notably, pathophysiological concentrations of IL-1β significantly reduced GABAA-mediated currents in oocytes transfected with hippocampal and cortical membranes from brain tissue surgically resected from TLE patients. This effect was mediated by PKC and was replicated in entorhinal cortex slices from chronic epileptic rats (Roseti et al., 2015). These data demonstrate that IL-1β reduces GABAA-mediated inhibitory neurotransmission, which likely contributes to seizures. In addition to its direct postsynaptic effects, IL-1β was shown to enhance neuronal excitability and promote excitotoxicity by increasing extracellular glutamate concentration (Casamenti et al., 1999). For example, in cortico-striatal mouse slices, IL-1β blocked the inhibitory function of cannabinoid receptor 1 on presynaptic glutamate release (De Chiara et al., 2013).
The inward rectifying K+ channel 4.1 (Kir4.1) in astrocytes is responsible for K+ buffering; thus, its reduction impairs K+ uptake and increases seizure susceptibility (Steinhauser et al., 2012; Vezzani et al., 2022). The exposure of human fetal astrocytes to IL-1β, but not to TNF or IL-6, downregulates the mRNA and protein levels of Kir4.1 (Zurolo et al., 2012), thus suggesting that IL-1β can increase neuronal excitability by enhancing extracellular K+ levels. Of note, Kir4.1 is reduced in experimental and human epileptogenic brain tissue, and this phenomenon contributes to seizures and epileptogenesis (Kinboshi et al., 2020).
Chronic exposure of human astrocytes to IL-1β reduces the expression of the glial glutamate transporter GLT-1, leading to a decrease in glutamate reuptake, thereby enhancing its extracellular concentrations (Hu et al., 2000). Moreover, IL-1β treatment of mixed cortical neuron-astrocyte cultures enhances glutamate export by increasing the activity of the cystine/glutamate antiporter (system xc−) that is induced in both neurons and astrocytes in epilepsy models (Fogal et al., 2007; Pauletti et al., 2019).
Chronic brain IL-1β overexpression induced by adenovirus-mediated gene delivery in naïve rats led to BBB breakdown, as assessed by the horseradish peroxidase extravasation in brain parenchyma after its intravenous injection (Ferrari et al., 2004). In accord, IL-1β promoted disassembling of tight junctions in endothelial cells, and induced nitric oxide production and the activation of matrix metalloproteinases that contribute to BBB leakage (Rempe et al., 2016). Interestingly, the downregulation of the tight junction protein ZO-1 associated with epileptiform activity in organotypic hippocampal slices was mediated by activation of Src kinases (Morin-Brureau et al., 2011), similar to the pathway involved in IL-1β ictogenic effects (see above). BBB dysfunction and the subsequent extravasation of serum albumin in brain parenchyma was shown to activate TGF-β signaling in perivascular astrocytes that became inflammatory and impaired in their capacity to reuptake extracellular K+ and buffer water, thus promoting changes in the extracellular milieu that are permissive for seizure generation (see Chapter 36, this volume; Vezzani et al., 2022). In support of a role of IL-1β in astrocytic dysfunctions, primary human fetal astrocytes exposed to IL-1β showed reduced gap junction coupling (Kielian, 2008). Astrocyte uncoupling has been described in human and experimental epileptic tissues (Bedner et al., 2015), resulting in impaired extracellular K+ clearance (see Chapter 26, this volume). This set of evidence suggests that the activation of the IL-1β-IL-1R1 axis in endothelial cells and perivascular astrocytes plays a role in BBB-related dysfunctions occurring in epilepsy (Ravizza et al., 2008).
The available evidence shows that IL-1β affects different cell populations in the brain by activating either similar or distinct intracellular pathways. The IL-1β-induced activation of NFkB primarily occurs in hippocampal glia and endothelial cells, thus sustaining the inflammatory response by increasing the transcription of inflammatory mediators (Srinivasan et al., 2004). On the other hand, IL-1β activates p38MAPK and CREB in hippocampal neurons, which are molecules involved in synaptic plasticity (Coogan et al., 1999; Frank and Greenberg, 1994). Interestingly, IL-1β-mediated responses in neurons, such as potentiation of NMDA-induced Ca2+ influx and activation of p38 and CREB, appear to be mediated by a specific neuronal isoform of IL-1R accessory protein, named IL-1RAcPb, which is proximal to IL-1R1 and pivotal for activation of IL-1 intracellular pathway (Huang et al., 2011). These findings highlight the complexity of the neuromodulatory role of the IL-1β-IL-1R1 axis and suggest that the association of IL-1R1 with specific accessory proteins may govern the activation of cell-specific intracellular signaling leading to differing biological actions.
Experimental status epilepticus induces a rapid and transient increase of TNF in hippocampal glial cells, concomitantly with a progressive increase of TNFR1 expression in neurons and astrocytes, and a neuronal reduction of TNFR2 (Balosso et al., 2013; De Simoni et al., 2000). Similar molecular changes were observed in human TLE brain tissue (Balosso et al., 2013). TNF has a dual role in seizures depending on its concentration, the target cells, and the receptor subtype induced in the brain. In particular, TNF ictogenic effects are mediated by TNFR1, while TNF reduced seizures by acting on TNFR2 (Balosso et al., 2005; Weinberg et al., 2013). Transgenic mice overexpressing TNF in astrocytes spontaneously develop seizures, whereas increased neuronal expression of TNF did not induce apparent phenotypic abnormalities, suggesting that the pathologic outcomes depend on cellular origin of the cytokine.
Similar to IL-1β, TNF affects neuronal excitability and cell loss by modulating excitatory and inhibitory neurotransmission acting on neuronal receptors (Vezzani and Viviani, 2015).
Among glutamate receptors, TNF showed functional interactions mainly with AMPA receptors. TNF released by microglia enhanced the surface expression of AMPA receptors in cultured hippocampal neurons. This molecular change was associated with an increase in the excitatory synaptic efficacy, as shown by the enhancement in the mean frequency of AMPA-induced miniature excitatory postsynaptic currents (Beattie et al., 2002). Notably, the newly recruited surface receptors lacked the GluA2 subunit, thus resulting in receptors that are more permeable to Ca2+ (Leonoudakis et al., 2008; Stellwagen et al., 2005). This effect was mediated by activation of a TNFR1-PI3K-dependent pathway. Incubation of cortical neurons with TNF transiently potentiated NMDA-dependent Ca2+ influx by involving TNFR2 and ERK (Jara et al., 2007).
The intracellular signaling activated by TNF receptor subtypes was shown to control the subunit composition of glutamate receptors. Thus, constitutive knockout mice lacking TNFR1 or TNFR2 showed distinct modification in the hippocampal levels of NMDA, AMPA, and KA receptor and their subunit composition. The basal extracellular glutamate level was significantly reduced in mice lacking TNFR1 as compared to wild-type animals, possibly because of a reduced astrocytic glutamate release, whereas basal extracellular glutamate levels did not change in mice lacking TNFR2. These changes were consistent with the alterations in seizure susceptibility in these mice (Balosso et al., 2009).
In cultured hippocampal neurons, TNF induced a rapid and persistent endocytosis of GABAA receptors containing the α1, α2, β2/3, and γ2 subunits, resulting in reduction of the surface receptors and a consequent decrease of inhibitory synaptic strength (Pribiag and Stellwagen, 2013; Stellwagen et al., 2005). The alterations in GABAA receptor trafficking were mediated by neuronal TNFR1, and they required the activation of p38-MAPK, PI3K, and protein phosphatase 1 signaling (Stellwagen et al., 2005).
TNF also modulates Nav, Cav, and Kv functions: the cytokine was shown to increase Na+ and Ca2+ channel currents in cultured hippocampal neurons by enhancing the channel membrane expression (Chen et al., 2015), or reducing inwardly rectifying K+ currents in cortical astrocytes via PKC signaling (Köller et al., 1998).
TNF-TNFR1 axis activation reduced the expression of the cell adhesion molecule N-cadherin in cultured hippocampal neurons by STAT3 tyrosine phosphorylation, thus resulting in modifications of dendritic spine morphology (Kubota et al., 2009). Since N-cadherin is down-regulated in rat hippocampus and cortex during epileptogenesis (Avdic et al., 2018), TNF may contribute to this effect. The TNF-driven changes were detected primarily on excitatory synapses, possibly representing a compensatory mechanism to counteract seizure-induced hyperexcitability. Accordingly, conditional deletion of N-cadherin gene in mice resulted in reduced levels of postsynaptic proteins in glutamatergic excitatory synapses and decreased the severity of kainic acid-induced behavioral seizures (Nikitczuk et al., 2014). TNF-driven dendritic modifications may be implicated in cognitive deficits.
TNF increased extracellular glutamate concentration by promoting the release of glutamate from activated astrocytes (Bezzi et al., 2001) and microglia (Takeuchi et al., 2006), and by decreasing glutamate reuptake by astrocytes (Hu et al., 2000; Zou and Crews, 2005). The astrocytic effects are mediated by COX-2 activation and PGE2 synthesis (Bezzi et al., 2001). In microglia, it involves the upregulation of glutaminase that, by converting glutamine to glutamate, leads to accumulation of intracellular glutamate that is subsequently released via connexin 32 hemi-channels (Takeuchi et al., 2006). Similar to IL-1β, TNF also induces changes in BBB permeability via TNFR1. In particular, TNF overexpression in rats led to IgG extravasation in the brain parenchyma (Weinberg et al., 2013). This event includes the remodeling of tight junctions (Capaldo and Nusrat, 2009) via the activation of soluble guanylate cyclase, protein tyrosine kinase, and COX1/2 (Mayhan, 2002). Overall, these TNF-dependent alterations contributed to neuronal excitability and excitotoxicity.
Seizure-induced changes in IL-6 expression are similar to those described for TNF. Both mRNA and protein expression were rapidly enhanced in glial cells for several days after the onset of status epilepticus (De Simoni et al., 2000), together with upregulation of the signaling transducer protein Gp130 (Lehtimaki et al., 2003).
There are discordant data on the IL-6 effect on seizures. Mice overexpressing IL-6 in astrocytes showed a reduced threshold to kainate seizures, and also developed spontaneous seizures. These effects were likely mediated by an impairment of the GABAergic system, since transgenic mice displayed a loss of GABA- and parvalbumin-immunoreactive neurons in the hippocampus (Samland et al., 2003). Similarly, intranasal administration of human recombinant (hr)IL-6 in adult rats increases the severity of behavioral pentylentetrazole (PTZ)-induced seizures (Kalueff et al., 2004). However, intranasal delivery of similar amount of hrIL-6 in developing rats reduced hyperthermia-induced seizures (Fukuda et al., 2007). Finally, mice lacking IL-6 exhibited increased seizure susceptibility to various chemoconvulsants and enhanced seizure-induced hippocampal injury (Penkowa et al., 2001).
The IL-6-dependent molecular mechanisms identified so far in neuronal cells are associated with increased excitatory neurotransmission and decreased GABAergic inhibitory function. In accord, IL-6 treatment of cultured rat hippocampal neurons potentiated the NMDA-dependent Ca2+ fluxes by activating the JAKs/STATs pathway (Orellana et al., 2005) and reduced the levels of the AMPA-GluA2 and metabotropic mGluR2/3 receptor subunits (Vereyken et al., 2007), thus increasing Ca2+ permeability and impairing inhibitory action on glutamate synaptic release. These effects were confirmed in transgenic mice overexpressing IL-6 in astrocytes (Vereyken et al., 2007), suggesting these cells are major sources of IL-6, which upon release activates IL-6 neuronal receptors. Accordingly, IL-6 is enhanced in astrocytes in epilepsy models (De Simoni et al., 2000).
In cortical rat slice preparations, IL-6 selectively and reversibly reduces the amplitude of GABA inhibitory postsynaptic currents, and this effect was abolished by drugs targeting GABAA receptor trafficking and/or internalization, and inhibitors of PI3K/AKT signaling (Garcia-Oscos et al., 2012). Finally, there is evidence of an inhibitory role of IL-6 on cortical Nav and hippocampal Cav (Vereyken et al., 2007; Xia et al., 2015), possibly representing a compensatory neuroprotective mechanism.
Notably, therapeutic effects on drug-resistant seizures have been attained in new-onset refractory status epilepticus clinical cases using anti-inflammatory drugs such as anakinra, adalimumab, and tocilizumab targeting the signaling activated by IL-1β, TNF, and IL-6, respectively (see Chapter 74, this volume).
There is emerging evidence indicating that neuroinflammation per se affects synaptic network activity. This aspect has been investigated by exposing rodents of different ages to systemic or intracerebral inflammatory triggers mimicking bacterial or viral infections, and by analyzing changes in expression and properties of neuronal glutamate and GABA receptors.
Lipopolysaccharide (LPS) is the most abundant component of the outer membrane of Gram-negative bacteria. Its peripheral or intracerebral administration to rodents boosts an innate immunity-driven inflammatory response in the brain via the activation of Toll-like receptor 4 (TLR4), which results in the synthesis and release of various pro-inflammatory cytokines and their downstream mediators (Lehnardt et al., 2003). One consequence of neuroinflammation produced by systemic LPS is long-lasting memory loss, which can be prevented by transient blockade of the EP2 receptor (Jiang et al., 2020).
LPS injection in adult rats induces alterations in the mRNA levels of several NMDA receptor subunits. The subunits whose expression is altered, and the direction of subunit changes depend on the brain area (hippocampus or cortex), the time elapsed from LPS injection, the systemic or intracerebral administration, and the LPS acute or chronic exposure (Harré et al., 2008; Ma et al., 2014; Rosi et al., 2004). This evidence highlights the complexity of LPS-dependent mechanisms in determining the functional readout on glutamatergic system.
Notably, single or multiple systemic LPS injections between postnatal day 5 and 30 induce long-lasting modifications in specific NMDA and AMPA receptor subunits in the rat hippocampus and cortex (Harré et al., 2008; Zubareva et al., 2020), with the direction of subunit changes depending of the age at which LPS was administered (Harré et al., 2008). These receptor changes were associated with cognitive deficits, a reduction in long-term potentiation, and with increased seizure susceptibility in adulthood (Galic et al., 2008; Harré et al., 2008; Zubareva et al., 2020). These findings indicate that the exposure to inflammatory episode(s) early in life may permanently alter brain function.
LPS also modulates GABAergic neurotransmission. Mice treated systemically with LPS during postnatal development showed decreased GABAA receptor level in the adult hippocampus, associated with impaired neurogenesis and depressive-like behavior (Liang et al., 2019). In rat spinal cord slices, LPS reduced mIPSC amplitude and frequencies and impaired GABA synthesis, thereby decreasing GABAergic synaptic activity. These effects were mediated by IL-1β release and the activation of PKC (Yan et al., 2015).
LPS was shown to promote a reduction in HCN1 channel protein in the hippocampus, due to dysregulation of the trafficking of its anchoring protein TRIP8. The associated Ih current was reduced in the distal dendrites of hippocampal CA1 pyramidal cells (Frigerio et al., 2018). This effect was likely mediated by IL-1β, which was induced by LPS in microglia, and implies compromised dendritic information processing. Of note, dysregulation of HCN1 is implicated in hyperexcitability and cognitive deficits in human epilepsy and experimental models (Noam et al., 2011).
Poly I:C is a double-stranded RNA molecule used to evoke an immune response by activation of TLR3, which mimics viral infection. The impact of poly I:C on glutamatergic and GABAergic functions has been investigated by exposing dams to the viral trigger during the early or late stages of gestation, and then analyzing the brain neuronal receptors in offspring from weaning to adulthood.
Maternal immune activation induced by poly I:C administration during gestation resulted in age-related alterations in the subunit composition of NMDA receptors in the offspring’s forebrain (Forrest et al., 2012). Using a similar experimental design, a reduction in the GABAergic transmission has been described in the prefrontal cortex of mouse offspring, as assessed by a decrease in the mRNA levels of enzymes regulating the biosynthesis of GABA, vesicular GABA transporter, and the α-subunits of the GABAA receptors (Richetto et al., 2014). These alterations were associated with increased anxiety, cognitive impairment, and pre-pulse inhibition deficits throughout life (Hao et al., 2019; Richetto et al., 2014).
After intracerebroventricular poly I:C injection during brain development in rats, Galic et al. (2009) found that the transient inflammatory response induced in forebrain was followed by increased mRNA levels of selected NMDA and AMPA receptor subunits in the hippocampus of adult animals. These receptor changes were associated with enhanced seizure susceptibility to chemoconvulsive drugs and deficits in the contextual fear conditioning test.
Overall, this evidence indicates that the inflammatory cascade evoked by a systemic inflammatory challenge is sufficient to affect glutamatergic and GABAergic neurotransmission in the absence of overt neuropathology. These changes are associated with reduction in seizure threshold and negatively impact neurobehavioral performances.
COX-2, the cyclooxygenase isoform that is induced by injury or neuronal activity, is the rate-limiting enzyme that converts arachidonic acid to five prostanoids. Its substrate is cleaved from membrane phospholipids by phospholipase A2 (PLA2) or from the endocannabinoid, 2-arachidonoylglycerol, by the enzyme monoacylglycerol lipase (MAGL). COX-2 sits in the middle of a major lipid signaling cascade that includes both inflammatory and inflammation-resolving pathways (see Fig. 74–2 in Chapter 74, this volume; Rojas et al., 2019b). The animal literature is inconsistent regarding the potential benefits of pan-specific NSAIDs or COX-2 selective coxibs in epilepsy models, as expected when interfering with a multidimensional signaling pathway (see Chapter 74, this volume). Moreover, the cardiotoxicity associated with chronic coxib use, and the expectation that COX-2 inhibition can influence the levels of endocannabinoids, leukotrienes, and lipoxins as well as the prostaglandins and their endocannabinoid metabolite analogs, is shifting attention toward downstream synthases and receptors that mediate inflammation in the brain. Of particular relevance for epilepsy are the PGE2 receptors, EP1 and EP2.
Antiseizure drug resistance in approximately 30% of epilepsy patients can in some patients be attributed to seizure-induced upregulation of the multidrug transporter, P-glycoprotein (Pgp), in the brain endothelial cells and perivascular astrocytes comprising the BBB (Loscher and Friedman, 2020), the net effect of which is to pump drugs out of the brain. COX-2 upregulation by glutamate acting on capillary endothelial cells appears essential in this process. Thus, glutamate-induced Pgp synthesis in isolated rat brain capillaries, which was associated with enhanced transporter activity, could be blocked by the selective COX-2 inhibitor, celecoxib (Bauer et al., 2008), and celecoxib also abolished Pgp upregulation in brain capillaries in a rat model of status epilepticus (Zibell et al., 2009). Importantly, systemic injection of an EP1 antagonist, SC-51089, both prevented upregulation of Pgp after status epilepticus and restored the antiseizure effect of phenobarbital in a drug-resistant kindling model (Pekcec et al., 2009).
These results, taken together, suggest a picture in which autocrine or paracrine activation of EP1 receptors by PGE2 formed in brain capillary endothelial cells induces Pgp, thereby contributing to drug-resistant seizures. Further tests of this hypothesis would employ more selective EP1 antagonists (e.g., ONO-8711) and endothelial-selective conditional ablation of EP1 and COX-2.
A common response of the brain to epileptogenic triggers such as acute injury or status epilepticus is the rapid activation of microglia, achieved in part by release of early danger signals from injured or rapidly firing neurons (e.g., ATP, HMGB1, and PGE2) (Klein et al., 2018). Activated microglia synthesize and release numerous cytokines and chemokines that reinforce neuroinflammation. Accumulating evidence over the past decade indicates that activation of EP2 receptors by PGE2 exacerbates neuroinflammation caused by status epilepticus and thereby contributes to the pathophysiological consequences. These data comprise both in vitro studies of microglia in cell culture and in vivo studies after status epilepticus.
In vitro cultured rat microglia express EP2 receptors that, upon activation by the EP2-preferring agonist butaprost, result in a rapid rise of cytoplasmic cyclic AMP (Fu et al., 2015; Quan et al., 2013) followed by induction of the inflammatory mediators IL-6, COX-2, and iNOS. EP2 activation also potentiated the increased levels of these inflammatory mediators after treatment with LPS. Based on the effects of a selective protein kinase A (PKA) inhibitor, and a selective Epac (Exchange protein activated by cyclic AMP) activator, the inflammatory effects of cAMP produced by EP2 activation appear to proceed both through the Epac and PKA pathways (Quan et al., 2013; see Fig. 30–2B). Similar findings have been reported for a BV2 microglial cell line engineered to express human EP2 (Rojas et al., 2019a).
COX-2 and EP2 related mechanisms. Panel A (from Vezzani et al., Neuropharmacology, 2013) depicts the main intersections among the COX-2 pathway and other major inflammatory pathways described in the text. Panel B depicts the downstream molecular (more...)
The development of selective brain-permeant EP2 antagonists beginning a decade ago allowed exploration of the role of EP2 receptor activation in the neuroinflammation produced by status epilepticus, as judged by blunting of microgliosis and reduced induction of a panel of cytokines and other inflammatory mediators. Systemic administration of five EP2 antagonists in three rodent models of status epilepticus consistently reduced the neuroinflammatory reaction in the hippocampus (Jiang et al., 2013, 2015, 2019; Rojas et al., 2015, 2016, 2021; Varvel et al., 2021, 2023). The therapeutic window for EP2 antagonism matched that of COX-2 induction after status epilepticus as expected if PGE2 derived from neuronal COX-2 was responsible for EP2 activation (Jiang et al., 2015). EP2 antagonist administration had to be delayed 2–4 hours after status epilepticus onset to be effective, suggesting opportunities for adjunctive therapy with conventional status epilepticus therapies (see Chapter 74, this volume). The anti-inflammatory effect was accompanied by accelerated physiological recovery after status epilepticus, reduced mortality, protection from delayed working and spatial memory loss, and, for two compounds, neuroprotection. Importantly, and in contrast to IL-1β, neither global pharmacologic antagonism of EP2 nor conditional ablation in myeloid cells modified the intensity or time-course of status epilepticus itself.
A strong expectation based on the in vitro culture work was that microglial EP2 receptors mediate neuroinflammation after status epilepticus. However, recent work indicates that circulating monocytes are another major target for the beneficial effects of EP2 antagonists. First, blocking monocyte entry into the brain after status epilepticus, either with a CCR2 knockout (KO) (Varvel et al., 2016) or a selective CCR2 antagonist (Aleman-Ruiz et al., 2023), phenocopied the beneficial effects of systemically delivered EP2 antagonists. Second, systemic administration of an EP2 antagonist also prevented monocyte infiltration. Finally, although both microglia and monocytes express EP2 receptors, only the monocyte EP2 pool was ablated in myeloid conditional EP2 KO mice created by crossing CD11b-cre mice with floxed EP2 mice (Varvel et al., 2021). The EP2 myeloid conditional KO mice replicated some but not all of the effects of EP2 antagonists after status epilepticus. The induction of hippocampal IL-6 after pilocarpine was nearly abolished, and the breakdown of the BBB was attenuated. The recovery from sickness behaviors following status epilepticus was accelerated. Surprisingly, however, neurodegeneration was not alleviated in myeloid conditional KOs. These findings drive the unexpected conclusion that monocytic EP2 contributes strongly to the neuroinflammatory consequences of status epilepticus. The data to date are consistent with neuron-derived PGE2 activating EP2 receptors on nearby microglia or perivascular macrophages, causing the formation and release of CCL2. This chemokine then attracts EP2-bearing monocytes into the brain that greatly intensify the incipient neuroinflammatory response.
Among the questions that remain to be addressed are the following: (a) Which cell types express EP2 that produces neuroinflammation? (b) What is the mechanism underlying EP2-mediated neurodegeneration, and (c) Which cells express the relevant EP2 receptors? Neuron-conditional EP2 ablation is being used to address the latter (Varvel et al., 2021), but pathway-specific adjustments are needed to explore the former. (d) How do the roles of invading monocytes and microglia differ? (e) Is salt-dependent hypertension seen in the global EP2 KO mouse relevant for advancing EP2 antagonists into the clinic? Recent data address this and find no adverse cardiovascular effects after acute or chronic administration of high-dose EP2 antagonists (Rawat et al., 2022).
CXCL1 or Gro KC is the murine ortholog of the human chemokine CXCL8 (IL-8). It activates G protein–coupled CXCR1 and CXCR2 receptors that are expressed by leukocytes and endothelial cells. In the brain, these receptors are expressed by neurons, glial cells, and endothelial cells of the BBB. After soman-induced status epilepticus in rats, CXCL1 levels were increased in the hippocampus and thalamus, remaining significantly elevated up to 24 hours (Johnson et al., 2011). Immunohistochemical analysis identified neurons and endothelial cells as the main cellular source of CXCL1 in the hippocampus, thalamus, and piriform cortex (Johnson et al., 2011; Xu et al., 2017). In drug-resistant human epilepsy tissue, IL-8 and CXCR2 are increased in glial cells and neurons (Choi et al., 2009; Morin-Brureau et al., 2018; Pernhorst et al., 2013; Strauss and Elisevich, 2016; Xu et al., 2017), while CXCR1 receptors are mainly expressed by neurons (Di Sapia et al., 2021).
Emerging preclinical evidence indicates that the CXCL1-CXCR2 axis plays a role in seizure mechanisms. Thus, CXCL1-neutralizing antibody delayed the onset and reduced the duration of acute PTZ seizures (Liu et al., 2020). This CXCL1-related effect involved a decrease in astrocytic glutamate reuptake due to dysregulation of GLT-1 that was driven by CXCL1 released from brain endothelium and activating CXCR2 receptor in astrocytes. Moreover, administration of a CXCR2 selective antagonist for 2 weeks after pilocarpine-induced status epilepticus in mice delayed the onset of spontaneous seizures and reduced their frequency after treatment discontinuation, thus indicating a disease-modifying effect (Xu et al., 2017).
We recently found that CXCL1 is induced in the brain but not in blood during status epilepticus evoked by intra-amygdala KA in mice. Inhibition of CXCL1 receptors using the selective antagonist reparixin reduced the duration of status epilepticus and afforded neuroprotection. Both the acute symptomatic and chronic seizures arising post status epilepticus in mice were reduced during drug administration, but the antiseizure effects elapsed after drug discontinuation, suggesting that the chemokine has ictogenic properties (Di Sapia et al., 2021).
The neuromodulatory effects of CXCL1 were investigated in models of neuropathic pain, highlighting a functional interaction between CXCL1 and the glutamatergic system. For example, CXCL1 potentiated NMDA-induced currents in spinal neurons by increasing the expression and phosphorylation of GluN2B-containing NMDA receptors, and involving both CXCR2 and COX-2 (Cao et al., 2014; Yang et al., 2016). In accord, the exposure of spinal cord slices to CXCL1 increased the frequency of spontaneous excitatory postsynaptic currents via activation of CXCR2 (Chen et al., 2014a). Moreover, the application of CXCL1 on primary cultured dorsal root ganglia neurons increased Ca2+ influx by activating CXCR2 and TRPV1 receptors (Deftu et al., 2018; Qin et al., 2005), and enhanced Na+ current amplitude and excitability (Wang et al., 2008). Overall, these mechanistic insights suggest a functional link between CXCL1-CXCR1/2, hyperexcitability, seizures, and neuronal cell loss.
The increase of IL-8 in human epilepsy foci (Choi et al., 2009; Morin-Brureau et al., 2018; Pernhorst et al., 2013; Strauss and Elisevich, 2016) and in cerebrospinal fluid (CSF) of patients with status epilepticus or drug-resistant seizures (Sakuma et al., 2015), and the reduction of IL-8 CSF levels in drug-resistant patients responding to anti-inflammatory therapy (De Herdt et al., 2009; Kenney-Jung et al., 2016) prompt further investigations into the role of CXCL1 in epilepsy, and the potential therapeutic effects attained by blocking its receptors.
Monocyte chemoattractant protein-1 (CCL2), the primary ligand for CCR2, is elevated in brain tissues from epileptic patients (Choi et al., 2009; Wu et al., 2008), and brain CCL2 levels are increased in rodent models of status epilepticus (Arisi et al., 2015; Foresti et al., 2009; Manley et al., 2007). CCL2 expression is correlated with seizure frequency in epileptic rats (Broekaart et al., 2018). Transcriptional upregulation of hippocampal CCL2 is rapid, increasing more than 10-fold within 30 min of pilocarpine-induced status epilepticus onset and almost 100-fold 4 days after status epilepticus (Jiang et al., 2015), indicating a rapid and robust involvement of CCL2 engagement in the immediate status epilepticus sequela. Infiltration of CCR2-expressing monocytes into the brain has been reported in several different rodent models of status epilepticus generated by intracerebroventricular (Feng et al., 2019; Tian et al., 2017) or intrahippocampal kainate administration (Cerri et al., 2016), systemic kainate and pilocarpine injection (Varvel et al., 2016), and in the Theiler’s virus model of encephalitis (Kaufer et al., 2018). This indicates that peripheral monocytes involvement in the neuroimmune response to status epilepticus is model-independent.
In the pilocarpine mouse model, blocking inflammatory monocyte brain recruitment after status epilepticus, via Ccr2 KO, is neuroprotective, reduces hippocampal neuronal damage, attenuates the expression of hippocampal IL-1β, prevents deterioration of the BBB, and enhances weight regain (Varvel et al., 2016), indicating that monocyte brain infiltration is reliant on CCR2 and that CCR2+ inflammatory monocytes are pathologic. This conclusion was supported by Tian et al. (2017) using the intracerebroventricular kainate mouse model, where they showed that blocking monocyte brain entry relieved hippocampal neuronal damage, suppressed the induction of hippocampal IL-1β, and quenched inflammatory STAT3 signaling. The authors further demonstrated that preventing monocyte brain entry rescued anxiety-like behaviors in the open field test and improved recognition of a novel object 14 days post status epilepticus.
Enhanced weight regain and neuroprotection were recently observed in rats subject to a CCR2 antagonist after pilocarpine-induced status epilepticus. However, the CCR2 antagonist had no effect on the incidence of spontaneous recurrent seizures or seizure severity occurring between 11 and 30 days after status epilepticus. CCR2-treated rats were equally susceptible to PTZ-induced seizures as controls. Notably, the efficacy of the CCR2 antagonist to block monocyte brain recruitment was unclear as the numbers of Iba1+ macrophages (sum of microglia and monocytes) was not reduced by the antagonist (Foresti et al., 2020).
Collectively, these data strongly support a role for peripheral monocytes in the neuroimmune response and the pathologic sequela following status epilepticus. Future studies designed to interrogate the signaling mechanisms involved in monocyte-mediated toxicity and the communication between brain-invading monocytes and brain-resident cells are warranted.
The complement system is an evolutionarily conserved arm of the innate immune system comprising 40 known proteins critical for the recognition and clearing of pathogens, apoptotic cells, and cellular debris. Whereas the canonical effects of complement activation include opsonization of cells and debris for clearance, immune cell chemotaxis, and cellular lysis through the formation of the membrane attack complex (MAC), recent work demonstrates complement proteins and glia are involved in synapse pruning within the brain during development (reviewed in Stephan et al., 2012). Elevated brain levels of complement proteins are observed in several neurologic diseases, including epilepsy (Schartz and Tenner, 2020).
Complement transcripts and proteins are encountered in brain material harvested from epilepsy patients. In particular, C1q, C3, and MAC (C5b-C9) are induced in neurons, astrocytes, and microglia in hippocampal tissue from epileptic rats (Aronica et al., 2007; Boer et al., 2008; Wyatt et al., 2017) and in resected specimens from TLE patients (Aronica et al., 2007; Wyatt et al., 2017). Notably, CD59, a complement-related protein preventing the formation of MAC was expressed to a lower extent than MAC itself, suggesting inefficient mechanisms of control of the complement cascade in epilepsy (Aronica et al., 2007). Cortical brain material from patients with refractory epilepsy has higher levels of C1q and iC3b with C1q located near points of contact between microglia and dendrites (Wyatt et al., 2017). Thus, there is a robust activation of the complement cascade in the epileptic brain.
The proinflammatory C5a receptor (C5ar1) is upregulated in the hippocampus in both the pilocarpine and intrahippocampal kainate models. Treatment with a brain-permeant C5ar1 inhibitor (PMX53) is anticonvulsant in the 6 Hz and corneal kindling models and in the chronic epilepsy phase 6 weeks after intrahippocampal kainate. PMX53 administration prior to systemic pilocarpine quenched seizure power measured by EEG, reduced the numbers of mice entering status epilepticus, rescued mortality, and reduced hippocampal neuronal damage (Benson et al., 2015). Taken together, these findings demonstrate an anticonvulsive effect of C5ar1 inhibition in acute and chronic seizure models.
Complement proteins C1q, C3, and iC3b remain elevated in the rat hippocampus for up to 5 weeks following pilocarpine-induced status epilepticus. Interestingly, iC3b levels positively correlated with spontaneous seizure frequency after status epilepticus, but not with cognitive performance in the novel object recognition test and Barnes maze, suggesting a role for iC3b in the generation of spontaneous seizures (Schartz et al., 2018). Pilocarpine-treated rats administered with a C1 esterase inhibitor (C1-INH) showed enhanced weight regain and reduced anxiety-like behaviors in the open field test, but no rescue of performance in the novel object recognition test and Barnes maze after status epilepticus. In addition to the contribution to seizures, specific components of the complement system (i.e., C1, C3, C5, C6, and MAC) are involved also in neuronal cell loss (Benson et al., 2015; Libbey et al., 2010; Schartz et al., 2019; Xiong et al., 2003). This set of data therefore indicates a potential therapeutic value to targeting complement activation after status epilepticus (Schartz et al., 2019).
In summary, the complement system is activated in epilepsy and engaged in the inflammatory aftermath of status epilepticus in rodent models. Inhibiting complement signaling has shown some beneficial effects in seizure models, but broad recovery of status epilepticus–induced cognitive comorbidities and inhibition of epileptogenesis has yet to be demonstrated.
Similar to cytokines, the complement system may increase neuronal excitability by modulating the glutamatergic system. In fact, complement proteins increased glutamate release from rodent forebrain synaptosomes, and from human cortex and hippocampus (Merega et al., 2014). This mechanism was prevented by DL-TBOA, a blocker of glutamate transporters, implying a carrier-mediated process. Moreover, the complement system, in cooperation with pentraxins, induced synaptic AMPA receptors clustering (reviewed in Perry and O’Connor, 2008), thus potentiating excitatory neurotransmission.
Finally, in adult rodent the activation of C1q-C3 signaling was associated with microglia-dependent excessive phagocytosis of hippocampal synapses in a mouse model of Alzheimer disease (Hong et al., 2016) and West Nile Virus infection (Vasek et al., 2016), impacting negatively on neurocognitive performances. Notably, in the hippocampus of patients with drug refractory epilepsy, phagocytic microglia signaling molecules (Trem2, Pros1) are up-regulated alongside the C1q-C3 complement components (Wyatt et al., 2017), suggesting that aberrant synaptic pruning may occur in epilepsy, leading to pathological remodeling of the neuronal circuitries that possibly contribute to hyperexcitability and cognitive deficits.
The IL-1β-511T SNP, located in the promoter region, is associated with susceptibility to TLE with hippocampal sclerosis in a Japanese population (Kanemoto et al., 2003) and pediatric febrile seizures (FS) in Finland (Virta et al., 2002). However, the IL-1β-511T SNP was not found to be a risk factor for TLE with hippocampal sclerosis in a population of European decent (Buono et al., 2001) or FS in Taiwanese children (Chou et al., 2003). The reasons for these discordant results are unclear but might be attributed to small genetic effects, insufficient sample sizes, or population differences (Colhoun et al., 2003).
A recent meta-analysis of 13 case-control studies revealed an association between general FS risk and the IL-1β-511 SNP (rs16944) in Caucasian and Asian populations (Yu et al., 2018). A separate case-control study found a significant association between the IL-1β -511 SNP and susceptibility to simple sporadic FS (Kira et al., 2010). Also, the IL-1β -511 SNP showed a modest association with TLE with hippocampal sclerosis (Kauffman et al., 2008), further supporting an association between genetic variants in IL-1β, particularly the -511 SNP, with seizure risk. Notably, the IL-1β-511 SNP is in a haplotype that influences LPS-induced expression of IL-1β in leukocytes (Wen et al., 2006). Thus, genetic variants in the promoter region of IL-1β can influence gene expression under an immune challenge, and this might provide a mechanistic explanation for the genetic association of IL-1β variants with seizure risk.
A recent genome-wide association study comparing children with measles, mumps, and rubella (MMR)-related FS have identified two loci, interferon-stimulated gene IFI44L and the measles virus receptor CD46 (Feenstra et al., 2014). A missense SNP rs208294 in P2X7R was recently reported in a case-control study of Caucasian childhood-onset FS (Emsley et al., 2014). In contrast to genetic variants that confer increased susceptibility with FS, the IL-10-592C allele and -1082A/-819C/-592C haplotype confer resistance to FS (Ishizaki et al., 2009).
Recently, functional assessment of endogenous IL1RA activity in FIRES patients revealed attenuated inhibition of IL1R signaling, and sequencing of IL1RN revealed multiple variants. This was accompanied by reduced expression of intracellular but not secreted isoforms of IL1RA in the patient’s blood mononuclear cells (Clarkson et al., 2019).
Since neuroinflammation was identified as a common hallmark of tissue pathology in patients with structural epilepsies and following status epilepticus, preclinical research has been addressing the pathophysiological changes mediated by the array of inflammatory molecules detected in epileptogenic tissue. The discovery that several inflammatory mediators affect seizures, cell loss, and neurological comorbidities in animal models (see Chapter 74, this volume), begged the question about the mechanisms underlying these effects.
Experimental research highlighted previously overlooked neuromodulatory actions of inflammatory molecules that were found to significantly affect neuronal excitability either directly or indirectly. Of note, besides the cytokines’ well-known induction of inflammatory genes in immune cells mediated by NFkB and AP-1 transcriptional factors, various cytokines and chemokines were shown to induce acquired channelopathies by causing rapid posttranslational changes in ion channels themselves (Figs. 30–2A and 30–3). This occurs by activation of their constitutive or up-regulated receptors on neuronal cells thus challenging the notion that immune receptors were mainly operative in leukocytes and microglia. Indirect effects of brain inflammatory mediators on neuronal network excitability mostly involve glial cells and the BBB: dysfunctional astroglia and increased BBB permeability due to neuroinflammation lead to an extracellular milieu permissive for seizures. Overall, these brain tissue modifications contribute in concert to reduce the excitability threshold in neurons, thereby promoting seizure generation.
Schematic representation of the cellular mechanisms underlying the effects of neuroinflammatory molecules on epilepsy outcomes.
New evidence also showed that peripheral monocytes recruited into the brain by chemokines play a significant role, in concert with PGE2, cytokines, and complement activation, in neuronal cell loss and comorbid behaviors ensuing after status epilepticus
The elucidation of the mechanisms of action of neuroinflammatory mediators has revealed the complexity of the neuroinflammatory network activated in epilepsy (Fig. 30–2A). A deep understanding of this network is instrumental for designing specific interventions apt to preclude the activation of specific pathologic mechanisms while preserving the homeostatic arm that the same mediators may elicit. A good example of this dual role is represented by the COX2-PGE2 axis in seizure-induced neurodegeneration. Moreover, knowing the timing of activation and persistence of each pathologic neuroinflammatory mechanism is required to develop targeted, safe, and effective therapies.
Finally, a challenging task that still remains is the identification of nodal points of intersection where the individual pathogenic neuroinflammatory mechanisms might converge to enhance excitability or promote cell loss and comorbidities. Gene expression modifications induced by mechanism-driven anti-inflammatory treatments in animals undergoing disease development may provide insights for the discovery of checkpoints to attain broad anti-inflammatory effects on the relevant pathways. This may lead to the identification of new or repurposed drugs for effective therapeutic intervention.
This review was supported in part by the National Institutes of Health, Office of the Director, Neurological Disorders and Stroke grants R01 NS097776 and R01 NS112308 (to RD), R01 NS112350 (to NHV), the Brightfocus Foundation (to NHV), and Citizens United for Research in Epilepsy (CURE) (to NHV), and by FIRE-AICE, Fondazione Monzino, Era-Net Neuron 2019 (Ebio2), and AES/NORSE Institute Seed Grant (to AV).
RD is a founder of, and has equity in, Pyrefin Inc, which has licensed technology from Emory University in which RD is an inventor. Otherwise, the authors declare no relevant conflicts.
This is an open access publication, available online and distributed under the terms of a Creative Commons Attribution-Non Commercial-No Derivatives 4.0 International licence (CC BY-NC-ND 4.0), a copy of which is available at https://creativecommons.org/licenses/by-nc-nd/4.0/. Subject to this license, all rights are reserved.
Your browsing activity is empty.
Activity recording is turned off.
See more...
