Epilepsy, a multifaceted disease characterized by the development of recurrent, unprovoked seizures, is a neurological disorder that affects over 65 million people worldwide. Its high-incidence and high-prevalence (6.4 cases per 1,000 persons) is nevertheless minored by its high mortality owing to accidents, status epilepticus or sudden unexpected death in epilepsy, often occurring during seizures (Devinsky et al., 2018). Being the most frequent, chronic, severe neurological disease it is a huge burden for public health systems (Moshé et al., 2015; Devinsky et al., 2018; Vezzani et al., 2019; Cunha-Reis et al., 2021) and since over one third of epilepsy cases are refractory to treatment with multiple antiseizure drugs (ASDs), it is imperative to identify new therapeutic targets.

Epileptogenesis, the process by which healthy brain networks develop enhanced susceptibility to generate spontaneous recurrent seizures leading to the development of epilepsy (Pitkänen et al., 2015; Vezzani et al., 2019), is currently perceived as a continuous and progressive process extending far beyond the latent period. Yet, the early pathophysiological mechanisms involved in epileptogenesis are still for the most part unknown and tackling epileptogenesis is an unmet clinical need (Pitkänen et al., 2015; Cunha-Reis et al., 2021). One key element is that different triggers likely lead to different seizure-onset, which will ultimately determine its first line of treatment (Sloviter, 2017). As such, establishing therapeutic targets in this early phase of epileptogenesis requires a better knowledge of these mechanisms.

Mesial temporal lobe epilepsy (MTLE) involves seizures typically arising in the hippocampus or other mesial temporal lobe structures. Only about 11–26% of patients with MTLE with hippocampal sclerosis achieve complete seizure control under pharmacological treatment with existing ASDs. As such, patients often require neurosurgical resection of the epileptic focus as a last resource to ameliorate seizure recurrence and to prevent fast worsening of the disease (Liu et al., 2012; Kuang et al., 2014; Thom, 2014). The etiology of MTLE epileptogenesis in still unknown, yet putative precipitating events such as trauma, complex febrile seizures, status epilepticus, inflammatory insults, or ischemia have been implicated. Such events may trigger epileptogenesis by generating aberrant synaptic plasticity/neuronal excitability, excitotoxicity, secondary non-convulsive status epilepticus, neuroinflammation and generation of reactive oxygen species that develop during a silent period, when spontaneous seizures do not occur (Thom, 2014; Gambardella et al., 2016; Sloviter, 2017), but their exact chronology it is not often easy to establish partly due to the multitude of putative epileptogenic events.

Within the first days of disease progression, enhanced neurogenesis and axonal sprouting accompany changes in neuronal excitability and synaptic plasticity driving the formation of novel synaptic connections and aberrant neural networks (Beck and Yaari, 2008). Synaptic plasticity, the ability of synapses to undergo long-lasting, activity-dependent bidirectional changes in the strength of synaptic communication, can be mediated by altering the gain of stimulus-release coupling at the presynaptic component or by changes in the type, number, or properties of the neurotransmitter receptors and their coupling to the intracellular signaling machinery at the postsynaptic level (Bliss and Collingridge, 2019; Cunha-Reis and Caulino-Rocha, 2020). It has been hypothesized that such changes may occur maladaptively immediately following non-epileptic seizures, thus triggering epileptogenesis (Avanzini et al., 2014), but this has not so far been thoroughly investigated.

Lipid rafts are membrane microdomains involved in synaptic receptor clustering, synaptic signaling, synaptic vesicle recycling and neurotransmitter release (Allen et al., 2007) and may have a crucial role in both synaptic and homeostatic plasticity under such pathological conditions (Sebastião et al., 2013). Synaptic lipid raft markers include both classic raft-associated proteins like caveolin-1 and flotillin-1, characteristic of caveolae and planar lipid rafts, respectively, and structural postsynaptic receptor anchoring proteins like PSD-95, present at glutamatergic synapses, and gephyrin, distinctive of GABAergic synapses (Tulodziecka et al., 2016; Papadopoulos et al., 2017; Bai et al., 2021). The latter two are also recognized as important NMDA, AMPA, and GABA_A membrane-anchoring proteins, respectively.

In vitro models of ictogenesis have provided important insights into the therapeutic potential of several candidate anti-seizure drugs (Gloveli et al., 1995; Debanne et al., 2006; Antonio et al., 2016; Dulla et al., 2018). In this paper we used such in vitro models of epileptiform activity to evaluate the time-course of changes in LTP occurring within 30 min to 1 h 30 m following seizures. LTP was evoked in vitro by theta burst stimulation (TBS), a sequence of electrical stimuli that mimic CA1 pyramidal cell burst firing as occurring during the hippocampal theta rhythm (4–10 Hz), an EEG pattern linked to hippocampal memory storage in rodents and involves the recruitment of GABAergic mechanisms (Vertes, 2005; Larson and Munkácsy, 2015; Artinian and Lacaille, 2018; Rodrigues et al., 2021). We characterized also the synaptic molecular alterations in AMPA receptors potentially underlying impaired LTP. In addition, its association with modified lipid raft dynamics, an important requirement for several molecular mechanisms involved in synaptic plasticity, was also investigated. A preliminary account of some of the results has been published as an abstract (Carvalho-Rosa and Cunha-Reis, 2019).

All animal procedures and protocols were performed according to ARRIVE guidelines for experimental design, analysis, and their reporting. Animal housing and handing was performed in accordance with the Portuguese law (DL 113/2013) and European Community guidelines (86/609/EEC and 63/2010/CE). The experiments were performed on hippocampal slices (400 μm thick), cut perpendicularly to the long axis of the hippocampus, obtained from young-adult (6–7 weeks old) male outbred Wistar rats (Harlan Iberica, Barcelona, Spain) as previously described (Aidil-Carvalho et al., 2017). Female rats were not used due to hormonal influences on LTP (Warren et al., 1995; Good et al., 1999). Rats were anesthetized with halothane, decapitated, and both hippocampi were extracted in ice-cold artificial cerebrospinal fluid (aCSF), composition in mM: NaCl 124, KCl 3, NaH₂PO₄ 1.25, NaHCO₃ 26, MgSO₄ 1, CaCl₂ 2, glucose 10, and oxygenated with a 95% O₂ - 5% CO₂ mixture.

This is known as paired-pulse facilitation (PPF) and is believed to result from the presence of residual calcium generated by the first stimulation in glutamatergic terminals, that results in an enhanced release of neurotransmitter in response to the second stimulation (Zucker, 2003) at CA3–CA1 synapses, where the release probability is low under the current stimulation conditions (Wu and Saggau, 1994; Debanne et al., 1996). PPF in the CA1 area of the hippocampus has also been attributed to paired-pulse depression in the fast inhibitory postsynaptic potential (IPSP) (Nathan and Lambert, 1991) believed to result from GABAB autoreceptor-mediated depression of neurotransmitter release at interneuron–pyramidal cell synapses (Davies et al., 1990; Lambert and Wilson, 1994).

To understand the impact of EA on the function of hippocampal synapses we studied its immediate influence on synaptic transmission in the CA1 area and early hour synaptic plasticity using electrophysiological recordings. We characterized spontaneous activity during seizure-inducing conditions, changes in electrically evoked synaptic transmission and LTP induced by mild TBS after EA.

In the present work we first describe that: (1) interictal-like EA induces a mild impairment in TBS-induced LTP observed 30 min after the insult that is fully recovered 1 h after; (2) ictal-like EA, in turn, induces a marked impairment in TBS-induced LTP within 30 min -1 h after the insult and this is reversed to a mild enhancement 1 h 30 m after; (3) Impairment of LTP induced by EA was globally associated to a decrease in AMPA GluA1 subunit levels and an increase in GluA2 subunit levels, resulting in a marked decrease of the GluA1/GluA2 ratio, together with a marked increase in GluA1 P831 phosphorylation and a mild decrease in GluA1 P845 phosphorylation; (4) Kv4.2 K⁺ channels levels were decreased following interictal-like EA but not following ictal-like EA; (5) EA also decreased the levels of classic lipid raft markers caveolin-1 and flotillin-1, while increasing synapse specific lipid raft markers PSD-95 and, most prominently, gephyrin.

In this work, the observed physiological alterations in LTP induction and expression, together with the alterations in the molecular composition of synapses, provide a good insight on the very early alterations of synaptic plasticity in adult rodents following two forms of epileptiform activity: (1) seizure-like activity (the one induced by bicuculline, 10 μM, Bic) and (2) neuronal interictal-like activity (induced by Mg²⁺ suppression, 0Mg²⁺), characteristic resting brain activity pattern of epilepsy patients, often observed in the latent period of epileptogenesis in animal models. The fact that LTP expression is influenced by previous neuronal and/or synaptic activity is not itself new (Abraham and Bear, 1996; Colgin et al., 2004; Zhang et al., 2005) and previous studies reported structural alterations in synapse architecture following seizures (Leite et al., 2005; Postnikova et al., 2017). Altered synaptic plasticity was also observed following numerous physiologic stress conditions, (Félix-Oliveira et al., 2014; Arias-Cavieres et al., 2021) and associated with the clustering of synaptic markers and altered synaptic morphology (Sebastian et al., 2013). Nonetheless, LTP changes occurring following in vitro events resembling seizure-like activity versus the ones caused by altered resting neuronal activity observed in epileptic patients was never investigated concomitantly in the same model. To this purpose, the use of the hippocampal slice in vitro system for seizure generation was preferred to the in vivo model of epilepsy also used by our research group (Serpa et al., 2022), since it allowed to assess LTP in the first hours following EA, thus avoiding the dissection and recovery time after slice preparation (usually 1 h– 1 h 30 m) that would expand the time window for LTP inspection. This is of particular interest to the development of early intervention therapies to prevent epileptogenesis.

Interictal-like EA, lasting for about 30 min, was characterized by spaced large amplitude burst of neuronal activity separated by more silent periods with occasional single spikes while ictal like activity, lasting for 20–25 min, was characterized by a continuous activity showing both population spikes and EPSP-SWs. As previously observed by many others (Gloveli et al., 1995; Debanne et al., 2006; Postnikova et al., 2020; Ergina et al., 2021), synaptic transmission was potentiated following both ictal-like and interictal-like EA (Figures 1E, 3D), an LTP-like effect of EA that is similarly, dependent on NMDA receptor activation (Ben-Ari, 2001; Debanne et al., 2006). However, in this paper, we focused on the ability of EA to condition further LTP induction by mild TBS (five trains of four pulses delivered at 100 Hz with 200 ms interburst interval) in a time window from 30 min to 1 h 30 min post EA. Decreased LTP observed when induced 30 min following interictal-like EA (Figure 2), was virtually gone when LTP was induced 1 h after EA. In contrast, LTP impairment was much stronger when LTP was induced 30 min following ictal-like activity (Figure 4), and a milder impairment was still observed when LTP was induced 1 h following ictal-like EA. LTP induced 1 h 30 min after ictal-like activity was, however, slightly enhanced, suggesting an overshoot in synaptic recovery mechanisms (Russo et al., 2013). Previous studies addressed the alterations in LTP following in vitro 4-AP induced ictal-like EA in hippocampal slices or 3 h after Li²⁺-pilocarpine-induced seizures in juvenile rats (Kryukov et al., 2016; Postnikova et al., 2020) and report a similar small reduction of LTP. Although this partially agrees with our observations, a slightly stronger TBS pattern was used, likely a requirement to induce a robust LTP in the juvenile rats (Rodrigues et al., 2021). LTP impairment after ictal-like EA was more pronounced in our work and was nearly recovered 1 h 30 m later, likely reflecting the shorter duration of the ictal-like EA in our work. The differences in the age, the LTP and EA induction methods and/or in the time frame of LTP evaluation after EA preclude a more elaborated discussion of all these findings. In this work, ictal-like activity, although shorter in duration than interictal-like activity, had a much greater impact on hippocampal LTP. This is in line with previous theories arguing that interictal-like activity is protective in epileptogenesis (Avoli, 2001; Avoli et al., 2006) and suggests that altered LTP mechanisms are not only important for epileptogenesis but also a relevant therapeutic target in its prevention (Albensi et al., 2004, 2007; Avanzini et al., 2014).

As previously mentioned, altered LTP expression following EA could be due to changes in presynaptic regulation of glutamate release, that could be assessed studying by PPF a synaptic short-term synaptic plasticity property that is influenced by such changes. In this work, PPF was mildly increased following 0Mg²⁺ induced EA and was slightly decreased following Bic-induced EA, suggesting that each of the two types of EA have a different impact on presynaptic function. The reduction in PPF caused by Bic-induced EA, although not very pronounced, is consistent with an increased glutamate release probability and may reflect enhanced presynaptic Ca²⁺ post EA (Kovac et al., 2017). Conversely, the mild increase in PPF caused by 0Mg²⁺-induced EA is consistent with a decrease in glutamate release probability and with a putative neuroprotective role of this type of activity in epileptogenesis. Regarding its impacts on LTP expression following EA, it can be argued that enhanced glutamate release probability following ictal-like Bic-induced EA may somewhat contribute to enhanced LTP impairment by promoting glutamate depletion at CA3-CA1 synapses. However, it is likely that this is only a small contribution to this effect, given the small alteration in the PPF ratio.

Induction of CA1 LTP classically relies on the NMDA-mediated postsynaptic Ca²⁺ rise (Bliss et al., 2018), that follows primary AMPA receptor mediated postsynaptic depolarization. Yet, LTP outcomes are also dependent on AMPA receptor subunit composition, a variable tetrameric combination of GluA1-4 subunits influencing channel function (Wiltgen et al., 2010; Chater and Goda, 2022). In hippocampal synapses, GluA1/2 heterodimers are the most common, although GluA2/3 heterodimers are also present (Chater and Goda, 2022). Importantly, GluA2 containing AMPA Rs, being impermeable to Ca²⁺, limit the contribution of AMPA Rs to the postsynaptic Ca²⁺ rise. Diminished levels of GluA2 containing AMPA receptors were linked to aging and pathological conditions such as brain lesions (Chater and Goda, 2022). In this work we found that synaptic GluA1 levels were decreased while GluA2 levels were increased in hippocampal slices exposed to EA (Figures 5A, D), an effect more pronounced following ictal-like activity. A concomitant decrease in GluA1/GluA2 ratio was observed for both conditions (Figure 6E). These findings suggest that epileptiform activity transiently reduces the number of Ca²⁺-permeable AMPA receptors at hippocampal synapses in adult rats, a likely endogenous neuroprotective measure. Interestingly, decreased AMPA Ca²⁺ permeability following 0Mg²⁺ induced EA was previously reported in mixed neuroglial cultures (Gaidin et al., 2019), which is consistent with our results. A recent work also implicates synaptic GluA2-lacking Ca²⁺ permeable receptors as the main event in the LTP-like effects of 4-AP induced EA in CA1 pyramidal neurons in juvenile rats (Postnikova et al., 2020). Given the different role played by AMPA subunits in synaptic plasticity at these young ages (Liu et al., 2020; Chater and Goda, 2022), it is likely that the AMPA glutamatergic component of this LTP-like effect may be distinct in adult circuitry (Goodkin et al., 2008; Avoli et al., 2022; Lévesque et al., 2022). Since the initiation of EA is itself dependent on GABAergic transmission (Avoli et al., 2022), the role of GABAergic transmission in the LTP-like effects of EA should also be elucidated. This would require a distinct approach to induce ictal-like activity than the one used in this study since direct effects of bicuculline on GABAergic receptors (Tang et al., 2023) may alter also the long-lasting response to EA.

The experimental approach used in this paper to monitor changes in synaptic AMPA receptors does not provide sufficient detail on their subsynaptic distribution (Chater and Goda, 2022), and the distinction between synaptic and perisynaptic AMPA receptors, or between pre- or postsynaptic receptors. In the context of seizure induced epileptogenesis, activation of extrasynaptic receptors by glutamate spillover from active synapses may be crucial to the establishment of metaplastic events contributing to maladaptive homeostatic plasticity such as changes in AMPA/NMDA ratio at glutamatergic synapses (Postnikova et al., 2020; Ergina et al., 2021; Chater and Goda, 2022). Future works should aim the elucidation of the exact changes occurring at synaptic and perisynaptic to clarify the contribution of AMPA receptors to early alterations in synaptic plasticity leading to epileptogenesis.

Long term potentiation (LTP) expression involves the phosphorylation of AMPA GluA1 subunits and their synaptic recruitment (Appleby et al., 2011; Park et al., 2021), a mechanism dependent on CamKII and intracellular kinases like PKA and PKC (Lee et al., 2003). Epileptiform activity could limit hippocampal LTP by inducing an activity-dependent enhancement in GluA1 phosphorylation status, thus generating a ceiling effect for LTP induction. In this work, the phosphorylation of AMPA GluA1 subunits at Ser831, a prominent target of CaMKII and PKC, was enhanced after both ictal-like and interictal like EA, an effect slightly more pronounced following interictal-like EA (Figure 6B). This suggests enhanced GluA1 phosphorylation at Ser831 contributes to the LTP-like effects of EA, likely by increasing channel conductance and promoting the recruitment of AMPA receptors to the synapse (Derkach et al., 1999, 2007). This may in turn impair LTP expression following EA, by limiting additional TBS-induced phosphorylation at Ser831. Thus, although the overall number of GluA1 subunits is decreased, the remaining GluA1 containing AMPA receptors are more likely found at synapses. Our observations early following EA contrast with the previously observed decrease in GluA1 Ser831 phosphorylation observed 3 h after pilocarpine induced seizures (Russo et al., 2013).

Phosphorylation of GluA1 at Ser845, a main target of PKA, was contrariwise decreased following both types of EA, and this decrease was more pronounced following ictal-like activity (Figure 6C). This change opposes the one observed after common TBS protocols in vitro (Park et al., 2021) but is consistent with our findings of decreased GluA1 containing receptors at synapses, an effect that may also reflect and endogenous neuroprotective measure against hyperexcitability induced by EA, as phosphorylation of GluA1 subunits at Ser845 drives AMPA receptor synaptic insertion, which could contribute to maladaptive homeostatic plasticity following seizures. This is fundamentally different from what is observed in the chronic period in the Li²⁺-pilocarpine model of epilepsy, where both GluA1 P831 and P845 phosphorylation levels are increased (Serpa et al., 2022).

This work also found a decrease in synaptic Kv4.2 channels following interictal-like EA but not following ictal-like EA (Figure 6F). Expression of Kv4.2 channels is more prominent in dendritic spines vs. dendritic shafts, being largely responsible for the Ca²⁺-activated delayed rectifying A-current (I_A) in CA1 pyramidal neuron distal dendrites, where it acts to control signal propagation and compartmentalization (Kim et al., 2007; Beck and Yaari, 2008; Kim and Hoffman, 2008). LTP induction with strong TBS stimuli reduces Kv4.2 dendritic membrane levels, leading to a shift in the voltage-dependence of I_A, and Kv4.2 channel activity contributes to LTP expression (Frick et al., 2004; Kim and Hoffman, 2008). Kv4.2 channels are responsible for the precision of the time window of pre and postsynaptic activity allowed for LTP induction (Zhao et al., 2011) and I_A activation also influences synaptic NMDA receptor composition at CA1 pyramidal neurons (Jung et al., 2008) and is synaptic morphology and seizure susceptibility (Tiwari et al., 2020). Altogether, our findings suggest that LTP will be easier to induce following interictal-like EA, resulting from an enlarged time window provided by Kv4.2 withdrawal from synapses. This may explain the lower impact and faster recovery of LTP expression following interictal-like vs ictal-like activity in our study, as this effect opposes the overall observed putative neuroprotective molecular remodeling at synapses (e.g., in AMPA receptor levels and phosphorylation). Being observed only following interictal-like (and not ictal-like EA), this effect is either not contributing to the putative protective effects of interictal-like EA (Avoli, 2001; Avoli et al., 2006) in epileptogenesis or reflects the ability of interictal-like activity to promote a faster synaptic recovery.

Synaptic α-tubulin levels were also altered in hippocampal slices following ictal-like and interictal-like EA (Figure 7F). As such, we could not use this protein as loading control for western blot analysis. Until recently the role of the microtubule cytoskeleton at synapses was not fully acknowledged but recent evidence has shed light on its importance in synaptic function, neurotransmitter release and synaptic plasticity (Waites et al., 2021). Decreased α-tubulin levels found in this work following EA likely indicate structural changes in synaptic morphology, as previously reported after seizures (Leite et al., 2005), and suggest that α-tubulin may undergo incorporation into microtubules following EA. Interestingly, the control of synaptic Ca²⁺ overload by microtubule dynamics has recently been reported (Vajente et al., 2019). Whether this mechanism may play a role endogenous neuroprotection against hyperexcitable states it remains to be established.

Chemical synapses are specialized structures composed of microdomains such as the presynaptic active zone and the postsynaptic density. Synaptic membranes, in turn, comprise nanodomains such as lipid rafts, liquid-ordered or gel-like membrane nanodomains segregated from the bulk membrane and enriched in cholesterol, sphingolipids and GPI-anchored proteins, deemed essential for dynamic synaptic signaling in processes like neurotransmitter release and neurotransmitter receptor membrane cycling (Allen et al., 2007; Borroni et al., 2016). The membrane proteins caveolin-1 and flotillin-1 characterize the two main types of lipid rafts found at synapses, caveolae and planar lipid rafts (Allen et al., 2007; Borroni et al., 2016), and play an active role in promoting the phase separation that sustains these lipid nanodomains (Allen et al., 2007; Frick et al., 2007), yet, lipid domain stability is also regulated by its lipid composition and lipid leaflet distribution (Sodero et al., 2011). In this work we observed an extremely marked decrease in the synaptic levels of caveolin-1 and flotillin-1 (Figures 7A, B), suggesting that EA is a strong disruptor of normal lipid raft stability and of trafficking mechanisms regulated by these proteins. Previous studies showed that enhanced excitatory neurotransmission induces a massive cholesterol loss, and this was implicated in endogenous neuroprotection against oxidative stress (Iannilli et al., 2011; Sodero et al., 2011). Furthermore, the interaction of postsynaptic anchoring proteins like PSD-95 and gephyrin and neurotransmitter receptors, like AMPA receptor, with lipid raft domains is regulated by synaptic activity through post-translational modifications like palmitoylation or GPI anchoring (Papadopoulos et al., 2015; Tulodziecka et al., 2016; Itoh et al., 2018). This may in turn disrupt the normal synaptic signaling through AMPA, NMDA and GABA_A receptors and impact synaptic plasticity mechanisms. Caveolin-1 was shown to enhance AMPA receptor binding capacity through direct inhibition of PLA2 activity (Gaudreault et al., 2004), thus suggesting that reduced caveolin levels may impair glutamatergic transmission and LTP expression. Caveolin-1 was also deemed essential in post-injury remodeling of the neuronal membrane following different insults (Gaudreault et al., 2005), such as brain injury by cerebral ischemia (Jasmin et al., 2007). Flotillin-1 in turn was implicated in glutamatergic synapse maturation, dendritic pruning and promotion of glutamate release. Flotillin is also involved in clathrin-independent endocytosis, by inducing membrane curvature and formation of plasma-membrane invaginations morphologically similar to caveolae (Frick et al., 2007) a process linked to neurotransmitter transporter internalization (Cremona et al., 2011). As such, decreased levels of flotillin-1 following seizures may contribute to decreased synaptic stability, impaired glutamatergic transmission and the impaired synaptic plasticity observed in this work. Altogether, this suggests that caveolins and flotillins are promising therapeutic targets to prevent epileptogenesis. The exact mechanism to be targeted requires still further investigation.

One possible explanation for the massive loss of some synaptic lipid raft markers could be the occurrence of massive synaptic protein internalization, degradation and autophagy in response to synaptic activity during hyperexcitable states (Li et al., 2012; Soykan et al., 2021). This would also imply a reduction in nerve terminal synaptic vesicles and its characteristic markers. Our observations that the synaptic vesicle marker synaptophysin is not significantly altered following both ictal-like and interictal-like EA (Figure 7C), suggests that these mechanisms are not likely contributing to reduce levels of synaptic proteins.

Distinct from classic lipid raft markers, the synaptic levels of postsynaptic scaffolding protein PSD-95, present at glutamatergic synapses, were moderately enhanced while the levels of gephyrin, present at GABAergic synapses, were massively enhanced following ictal-like and interictal-like EA (Figures 7D, E). PSD-95 is a lipid raft nucleating protein and plasma membrane cholesterol depleting was shown to reduce PSD-95 targeting to neuronal dendrites (Hering et al., 2003; Tulodziecka et al., 2016). Our results are in conflict with previous observations reporting that enhanced excitatory synaptic transmission causes cholesterol depleting (Sodero et al., 2011), expected to lower PSD-95 levels, and are also not in line with evidence in this work that synaptic lipid rafts are destabilized by EA and with reports that synaptic AMPA receptor levels, stabilized at synapses by palmitoylated PSD-95 (Han et al., 2015; Matt et al., 2019), are reduced following EA. However, this may reflect a specific response of the postsynaptic active zone to EA, since maintaining postsynaptic lipid rafts may be crucial to prevent epileptogenesis and abnormal spine growth by keeping together the molecular machinery allowing for AMPA receptor depalmitoylation and internalization (Han et al., 2015; Itoh et al., 2018).

GABAergic transmission and interneuron networks are key players in seizure initiation and in epileptogenesis (Magloire et al., 2019; Avoli et al., 2022). GABA_A receptor postsynaptic recruitment to the inhibitory GABAergic synapses requires the scaffold protein gephyrin, that establishes a dynamic lattice-like structure with nanodomains of different size and density (Pizzarelli et al., 2020), and the action of the guanine nucleotide exchange factor collybistin through a PI3P dependent-process (Papadopoulos et al., 2017). Collybistin mutants with reduced lipid binding affinity disrupt GABAergic synapses by preventing gephyrin synaptic clustering and GABA_A receptor targeting and this was in turn reported to trigger epilepsy (Papadopoulos et al., 2015). Gephyrin palmitoylation drives its membrane insertion and macromolecular clustering thus facilitating GABAergic transmission (Dejanovic et al., 2014; Matt et al., 2019). As such, the enhanced synaptic levels of gephyrin found in this work following EA may reflect its synaptic recruitment to reinforce GABAergic transmission, through GABA_A recruitment, as an endogenous neuroprotective effort against hyperexcitability during ictal-like and interictal-like EA (Barberis, 2020). A similar response of GABAergic synapses was observed following anoxia (Lushnikova et al., 2011). This contrasts with the decreased levels of PSD-95 and gephyrin found in animal models and human MTLE (Sun et al., 2009; Fang et al., 2011; Bento-Oliveira, 2018), reinforcing the idea that acute and chronic synaptic responses to seizures are distinct.

Long term potentiation (LTP) expression is impaired following EA, and this impairment is stronger following ictal-like than after interictal-like activity. Altered LTP expression is accompanied by altered PPF, evidencing opposing presynaptic modifications of glutamate release, following ictal-like and interictal-like EA. In addition, we observed marked modifications in several synaptic proteins involved in synaptic plasticity mechanisms such as AMPA receptor subunits and its phosphorylation, and Kv4.2 channels. Furthermore, we detected altered synaptic membrane structure and domain regulating proteins like caveolin-1, flotillin-1, PSD-95 and gephyrin. Altogether, these results suggest that early changes in synaptic plasticity are a promising target for antiepileptogenic therapies and strategies focusing on modulation of synaptic lipid raft domain dynamics may prove relevant to prevent epileptogenesis.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This animal study was reviewed and approved by the Ethics Committee of the Faculty of Medicine, University of Lisbon (Comissão de Ética para a Saúde do CHLN/FMUL). Approval was given in writing based on a detailed procedure report by the authors.

JC-R, NR, and AS-C: formal analysis, methodology. SV: methodology. DC-R: formal analysis, methodology, resources, supervision, funding acquisition, project administration, writing—original draft, review, and editing. All authors contributed to the article and approved the submitted version.