Hippocampal hyperactivity throughout memory encoding tasks in individuals with MCI and presymptomatic people that carried the E280A presenilin-1 (PS1) mutation, which is the primary driver of early-onset familial AD, has been identified by fMRI (Hector and Brouillette, 2020). Hyperexcitability of the neurons causes hyperactivity and hypersynchrony of the neural networks, which causes epileptiform and seizure activity in the early stages of AD. Seizures and epileptiform activity have been recorded in AD and MCI patients, as well as in animal models of AD (Vossel et al., 2013). Electrophysiological studies in transgenic AD animal models have reported a disturbance in neuronal network excitability, which can be related to Aβ or tau. It has been shown that in the early and presymptomatic stages of the disease, soluble Aβ damages inhibitory neurons and inhibits glutamate reuptake, which disrupts the excitatory-inhibitory balance and leads to the hyperexcitability of cortical and hippocampal neurons (Palop and Mucke, 2016; Zott et al., 2019). Hyperactivity of the cortical and hippocampal areas may play an important role in the accumulation of Aβ plaques and the spread of NFTs throughout the brain. Therefore, it can be one of the key mechanisms of disease progression and cognitive disorders (Busche and Konnerth, 2015; Targa Dias Anastacio et al., 2022). In numerous transgenic AD mouse models, neuronal hyperactivity has been identified. It was found Around 21% of the neurons in the cortex of the APP23 × PS45 animal model exhibited an elevated influx of Ca2+, primarily close to amyloid plaques (Busche et al., 2008). Moreover, hyperactivity was observed in the CA1 area of the hippocampus in young transgenic mice with AD (aged 1–2 months) at a stage when the accumulation of Aβ oligomers (Aβo) starts but before the emergence of detectable plaques (Busche et al., 2012).

These findings indicate that hyperactivity is an initial pathogenic event that depends on the accumulation of Aβo, rather than the mere existence of plaques. The plaques may serve as a storage site for harmful Aβo, thus intensifying the elevated neuronal activity that is partly responsible for the synaptic and neuronal losses found near the plaques. Alongside this hyperactivity, approximately 29% of the cortical neurons exhibited hypoactivity in mice aged 6–10 months with established plaques. It is worth mentioning that hypoactive neurons were only detected after the development of plaques. This suggests that as AD advances, the initially hyperactive neural networks eventually transition to a state of hypoactivity (Hector and Brouillette, 2020). It is noteworthy that the modulation of hyperactivity using anticonvulsant drugs such as levetiracetam improves AD-related memory disorders (Bakker et al., 2012). Although most of the neuronal hyperactivity is attributed to Aβ accumulations, the effects of tau on neuronal excitability are conflicting in different studies. However, recent studies suggest that hyperphosphorylated tau is more related to reduced neuronal excitability in the hippocampal and cortical areas (Harris et al., 2020).

It is imperative to understand hippocampal and cortical excitability in AD, but understanding EC neuronal excitability also provides valuable insight into AD pathology as well. Considering this, various studies have investigated this issue. Using four-month-old Tg2576 mice carrying the Swedish mutation (APP695 with double mutations at KM670/671NL), a reduction in the firing rate was observed among LEC fan cells. The overproduction of different Aβ peptides in the MLT regions reported in this model may be involved in the hypoactivity of LEC fan cells (Marcantoni et al., 2014). In contrast, a study conducted in 3-and 6-month-old Tg2576 mice demonstrated neuronal hyperactivity in the LEC through the measurement of local field potentials and single-unit spontaneous activity in the LEC, which was correlated with amyloid precursor protein (APP) metabolites (Xu et al., 2015). Also in another study, which was conducted in Tg2576 mice, but at an earlier stage in 2 to 4-month-old mice, there was an increase in soluble Aβ expression in the LEC area. Hyperexcitability in the LEC slices was observed by increased repetitive field potentials in response to stimulus (Duffy et al., 2015). The results of these studies suggest that EC hyperexcitability may also occur before plaque deposition and thus may serve as an early indicator of AD. Other transgene models of AD have also demonstrated hyperexcitability of EC. Hyperexcitability, characterized by an elevated discharge of action potentials, increased frequency, and amplitudes of spontaneous excitatory postsynaptic potentials (sEPSPs), was also detected in the LEC of a mouse model with AD (AppNL-F/NL-F). This hyperexcitability was associated with the malfunctioning of parvalbumin (PV) interneurons and Wnt signaling (Petrache et al., 2019). Furthermore, mice expressing mutant human APP in the EC had hyperexcitable neurons. This was supported by the presence of more frequent and longer-lasting spontaneous extracellular field potentials (sEFPs) in the LEC and the occurrence of epileptiform-ictal-like discharges in the MEC. In contrast, expressing mutant human tau in the EC reduced excitability (Angulo et al., 2017). In an additional investigation, electrophysiological recordings using the whole cell patch clamp technique were conducted on stellate neurons in layer II of the MEC in three-month-old 3xTg transgenic mice, which exhibit both Aβ and tau pathologies. The results revealed that the excitability of these neurons remained unchanged in animals of this age. Additionally, no impairments in spatial memory were detected in mice at this stage. However, in mice that were 10 months old, there was a noticeable hyperexcitability in these neurons, which was accompanied by impairments in spatial memory. This suggests that spatial memory impairments may be associated with the hyperexcitability of these neurons (Chen et al., 2023). A further investigation conducted on the rTg4510 tauopathy model demonstrated a reduction in neuronal excitability in the dorsal region of the MEC, whereas the excitability of neurons in the ventral region remained unaffected (Booth et al., 2016). A reduction in grid cell firing rate was observed after the in vivo recording of the MEC in old EC-Tau mice with tau pathology in the MEC. Furthermore, this study found that tau pathology mostly caused neuronal death in excitatory neurons, which indicates that these neurons are more susceptible to tau than other types of neurons (Fu et al., 2017). Interestingly, the hyperexcitability of LEC pyramidal neurons was reported even in mice expressing the risk factor gene of AD (APOE4) (Nuriel et al., 2017). According to the above evidence, most evidence suggests that Aβ accumulation in the EC can cause hyperexcitability. The increase in hyperexcitability can also result in increased Aβ peptide secretion and greater plaque accumulation (Leal et al., 2017). It has been found that Aβ can facilitate tau spread in the brain, both locally and remotely (Busche and Hyman, 2020; Lee et al., 2022). Neuronal hyperactivity can also increase the release of tau from neurons which can be taken up by neighboring cells, thereby seeding even more tau to develop (Wu et al., 2016). It was found that optogenetic stimulation of the rTg4510 mice hippocampus and chemogenetic stimulation of the rTauEC mice EC increased tauopathy in mice (Schultz et al., 2018). Therefore, the increase in excitability, either directly or indirectly through the increase of Aβ facilitates the spread of tau in the hippocampal and cortical regions. As tau spreads in the brain, it can cause the disease to progress to its advanced stages and cause extensive dementia in the brain (Braak and Del Tredici, 2012). It can be therapeutically beneficial to control hyperexcitability in this area to prevent the progression of the disease.

Aberrant network excitability and dysfunction of inhibitory neurons can also impair the oscillatory or rhythmic activity of the brain (Palop and Mucke, 2016). Neuronal oscillations are the rhythmic fluctuation of the neuronal population’s electrical activity that can be measured by different methods such as electroencephalography (EEG), local field potentials (LFP), and magnetoencephalography (MEG). They are classified into six categories based on frequency, which includes delta (δ), theta (θ), alfa (α), beta (β), gamma (γ), and sharp-wave ripples (SWRs). Oscillatory activity in different regions of the brain enables various brain areas to connect more easily and efficiently through the synchronization of neuronal oscillations, either by coupling the phase of neuronal oscillations (phase coupling) or by coupling the amplitude of neuronal oscillations (power to power correlation) or by coupling the phase of a slower oscillation with the power of a faster one (phase-amplitude coupling), making communication faster and easier. As a result, neural oscillations are one of the basic mechanisms for processing various cognitive functions of a healthy brain, and their abnormal activity may indicate the presence of brain disorders such as AD (Ward, 2003; Buzsáki and Watson, 2012).

Dysfunction in neural networks oscillatory activities plays a critical role in the pathology and clinical manifestations of AD (Aron and Yankner, 2016). It occurs in the early phase of a pathogenic cascade that leads to exacerbation of AD progression and finally disruption of neural circuits that underlie higher cognitive functions in affected individuals (Canter et al., 2016). Therefore, several EEG investigations that encompassed individuals with both MCI and AD have consistently observed alterations in brain oscillations compared with a group of healthy individuals. These alterations include a reduction in alpha and beta activity, along with an increase in delta and theta activity. In addition, it has been observed that decreased complexity and coherence in EEG recordings could be used as biomarkers for diagnosing AD (Le Roc’h, 1994; Stam et al., 2003). Decreased EEG synchronization and loss of oscillatory activity, particularly of gamma-frequency oscillations, have been observed in MCI and AD patients (Koenig et al., 2005).

Due to the need for invasive methods for studying the neuronal oscillations of subcortical areas, mouse rodent animal models are used. In this regard, the recording of electrophysiological signals from hippocampal areas of rodent AD models shows that gamma and theta oscillations and the phase-amplitude coupling between them (theta–gamma coupling) are more disturbed during AD (Mehak et al., 2022). Theta, gamma, and SWRs are oscillatory activities that are involved in various memory processes, including encoding, consolidation, and retrieval (Düzel et al., 2010; Joo and Frank, 2018). A decrease in the power of the low gamma frequency band has been reported in the hippocampus of many AD animal models. Studies suggest that soluble Aβ peptides cause damage to PV neurons, Considering the important role of PV neurons in generating gamma oscillations, they disrupt the generation of these oscillations (Verret et al., 2012; Martinez-Losa et al., 2018). It is important to note that, impairment of the hippocampal gamma oscillatory activity band has been reported even before the accumulation of Aβ plaques (Iaccarino et al., 2016). For this reason, gamma oscillations have attracted a lot of attention in AD research. Therefore, restoring gamma oscillations is proposed as one of the solutions for the treatment of AD (Adaikkan and Tsai, 2020). In this regard, preclinical studies have shown that the entrainment of 40 Hz gamma oscillations by using different techniques, including optogenetic stimulation, sensory visual and auditory stimulation, and transcranial-focused ultrasound has shown effective results in improving AD-like pathologies (Adaikkan et al., 2019; Martorell et al., 2019; Park et al., 2021).

Overall, it can be deduced that the impairment of neuronal oscillations is an important component for examining the normal activity of the neural network in AD and is considered a biomarker for AD early diagnosis and a potential therapeutic target.

The proper functioning of the EC during spatial navigation relies on its oscillatory dynamics. The main neuronal oscillations in the EC are gamma and theta, which play critical roles in transferring spatial information to the hippocampus (Quilichini et al., 2010). Theta-gamma coupling is an essential phenomenon for accurate coordination of the hippocampus and EC during the process of encoding and retrieval of memory. It involves the emergence of gamma oscillations at specific phases of theta oscillations (Colgin, 2015). Considering the importance of EC oscillatory activity in memory, disturbance in EC oscillatory activity can result in the dysfunction of various memory processes. This has motivated various studies to investigate neuronal oscillations in EC using animal models of AD. One of the first studies conducted on this issue was conducted using amyloid precursor protein-knock-in mice (APP-KI). LFP recordings in the MEC of APP-KI mice demonstrated impaired theta-fast gamma coupling and spike phase locking of pyramidal neurons. However, the power of gamma oscillations and theta oscillations was not affected. The disruption of gamma temporal organization in the MEC is evident from these findings (Nakazono et al., 2017). In a subsequent study, MEC grid cells were found to have poor spatial tuning in aged APP-KI mice with memory disorders. Furthermore, fast gamma oscillations, which are crucial to transmitting information to the hippocampus, were also disrupted. Consequently, there was a noticeable reduction in gamma coherence between the MEC and the hippocampus. When MEC activity was recorded in young APP KI mice who did not yet have significant cognitive disorders, grid cells showed abnormal spatial tuning. In contrast, hippocampal place cells did not show abnormal tuning. This result indicated that MEC dysfunction occurs earlier in this disease than in other areas, and most spatial memory disorders originate here (Jun et al., 2020).

Researchers conducted a recent study on J20 male mice and found a dysfunctional pattern in the grid cells. The pattern was characterized by reduced spatial stability and lack of synchronization with head direction cells and associated with poor path integration (Ying et al., 2022). Video EEG recordings of MEC in 6-month-old APP KI mice showed impaired low gamma power. These mice were also recorded during SWRs in the MEC and CA1, revealing disrupted synchronization in the early stages of the disease. Coordination between the MEC and CA1 may support hippocampal long-duration SWRs. Disruption of this synchronization during SWRs could contribute to later memory consolidation dysfunction in AD, as these long-duration SWRs are essential for memory consolidation (Funane et al., 2022). Consequently, the main alterations observed in Aβ-based models include diminished power and temporal coherence of gamma oscillations, as well as disrupted connectivity between the MEC and hippocampus, which is primarily attributed to disturbances in gamma oscillations (Figure 2). Researchers have also investigated the oscillatory activity of tau-based models in EC-tau mice. This model showed an increase in theta activity in this region. This may be due to the loss of excitatory neurons, which is consistent with a higher level of theta rhythm in patients with mild cognitive impairment in the early stages of AD (Fu et al., 2017). Further research on the rTg4510 tauopathy transgenic mice showed a disruption in gamma oscillations, particularly in the dorsal region of the MEC, but no significant change in the oscillatory activity of the ventral part (Booth et al., 2016).

LEC is regarded as a critical node of AD (Mandino et al., 2022). However, little research has been conducted on the alterations in oscillatory activity within this area. A recent study discovered that injecting Aβ i.c.v. resulted in the aggregation of plaques in the LEC and disrupted synchronization between the LEC, olfactory bulb, and hippocampal area. This synchronization impairment has been linked to a decline in recognition memory during testing for novel object recognition (Salimi et al., 2022).