Affecting approximately 50 million people (World Health Organization, 2019), epilepsy is one of the most common neurological disorders in the world. Epilepsy is characterized by an individual suffering from repeated unprovoked seizures, which are a result of synchronous discharge of thousands of neurons which give rise to an abnormal EEG and are associated with a variety of behavioral abnormalities.

Temporal lobe epilepsy (TLE), which designates seizures originating in the temporal lobe, is the most common form of epilepsy observed in adults and adolescents (Blair, 2012). The majority of patients with TLE have seizures originating from internal structures of the region, which is further classified as mesial TLE (MTLE). It is well characterized by the pathological hallmark of mesial temporal lobe or hippocampal sclerosis (HS), which involves clearly demarcated regions of neuronal loss and reactivity of glial cells, or gliosis, throughout the subfields of the hippocampus and surrounding areas (Thom, 2014). Approximately one-third of patients with epilepsy are treatment-resistant, with MTLE-HS patients making up the majority of those, highlighting the need for therapeutic treatments that can specifically address the neuronal loss and gliosis that characterize HS. Although the roles of neurons and the consequences of their loss are crucial to understanding the progression of epilepsy, it has become quite evident that glial cells, particularly reactive astrocytes, can contribute to epileptogenesis; that is, the processes occurring in the brain that lead to seizures and subsequent epilepsy.

Astrocytes have long been acknowledged as essential for normal brain function as well as being major contributors to injury and diseases. These specialized glial cells tile the entire brain and contact vasculature, synapses, and each other, forming gap junctions between the individual cells. In the healthy brain, these interactions enable astrocytes to be engaged in energy metabolism, blood-brain-barrier maintenance, glutamate clearance, and neurotransmitter uptake and homeostasis.

At the synaptic level, astrocytes effectuate not only neurotransmitter regulation but also synaptic formation, maturation, pruning, and stability (Dityatev and Schachner, 2003; Dityatev and Rusakov, 2011; Chung et al., 2015; Hösli et al., 2022). They do so by extending branching processes with small terminal extensions, often called leaflets, to contact and stabilize pre and postsynaptic partners (Khakh and Sofroniew, 2015; Torres-Ceja and Olsen, 2022), resulting in the classic “tripartite synapse.” The presence of astrocytic leaflets, which contain a variety of membrane receptors, permits astrocytes to closely monitor and respond to molecular changes in their immediate domains. These are crucial in regulating and redistributing molecules associated with neuronal firing released into the extracellular space (ECS), particularly potassium and glutamate.

Astrocytic processes are highly enriched in potassium (K⁺) channels, enabling them to clear K⁺ from the synaptic cleft and surrounding area following neuronal activity. Under normal homeostatic conditions, their inwardly rectifying potassium channels (Kir) maintain a membrane potential that hovers around the equilibrium potential for K⁺, so that upon K⁺ concentration increase in the ECS, astrocytes are able to swiftly take up excess K⁺ ions. They are then conveyed via K⁺-permeable gap junctions to neighboring astrocytes, enabling them to redistribute ions from regions of high to low K⁺ concentration (Olsen and Sontheimer, 2008; Blutstein and Haydon, 2014; Ohno, 2018). A similarly astrocyte-driven mechanism controls the extracellular concentration of the excitatory neurotransmitter glutamate, which influences neuronal excitability and can become excitotoxic if allowed to remain in the ECS. Astrocyte-specific glutamate transporters EAAT1 (GLAST) and EAAT2 (GLT-1) transport glutamate into the astrocyte along with 3Na⁺ in exchange for 1K⁺, after which the glutamate is converted to glutamine by glutamine synthetase (GS) and shuttled back to the neurons.

At each step of these synaptic processes, astrocytes inherently alter and are altered by their interactions with not only neurons, but also immune cells, signaling molecules, and even non-cellular components of the brain such as extracellular matrix. In pathological states, astrocytes can easily become reactive and transition to an inflammatory state, altering their interactions with the other brain constituents and potentially creating neuroinflammatory feedback loops.

In a neuroinflammatory situation, astrocytes can very quickly become reactive. Also referred to as “astrogliosis,” “astrocytosis,” “gliosis,” or “reactive gliosis,” these astrocytes undergo molecular, chemical, morphological, proliferative, and functional changes following an immune challenge (Sofroniew and Vinters, 2010; Vezzani et al., 2011; Zamanian et al., 2012; Robel et al., 2015; Escartin et al., 2019). These changes vary in degree of reactivity depending on the intensity or nature of the initial instigator; in fact, the heterogeneity of astrocytic responses seems to be the one agreed-upon facet of this widespread immune reaction (Escartin et al., 2019). Some reactive astrocytes are considered more beneficial or neuroprotective as they release more anti-inflammatory and health-associated signaling molecules, and others are considered more harmful or neurodegenerative as they release more pro-inflammatory, disease-associated molecules like cytokines and chemokines. This has led to a classic “good vs. bad” taxonomy of reactive astrocytes that some consider too disparate. An excellent consensus paper covers this topic (Escartin et al., 2021). For this paper, it is sufficient to express that reactive astrocytes exist along a spectrum and a single astrocyte can express both beneficial and detrimental growth factors and signaling molecules.

Keeping in mind their clearly important roles in supporting normal brain function, astrocytic dysfunction is linked to many pathologies that involve neurodegeneration including Alzheimer disease (AD) (Nwaobi et al., 2016; Pajarillo et al., 2019), Huntington’s disease (HD) (Tong et al., 2014), and amyotrophic lateral sclerosis (ALS) (Rossi et al., 2008; Ferrer, 2017; Neal and Richardson, 2018). Astrocytes and astrogliosis are also heavily implicated in epilepsy and epileptogenesis, as evidenced in the brains of human epilepsy patients (Lee et al., 1995; Crespel et al., 2002; Wetherington et al., 2008; Das et al., 2012; Devinsky et al., 2013; Eid et al., 2013; Gibbons et al., 2013; Bedner et al., 2015; Coulter and Steinhaeuser, 2015; Hayatdavoudi et al., 2022), and recapitulated in a variety of animal models.

Neuroinflammation and its associated changes have been found in practically every neurodegenerative disorder (Escartin et al., 2019). Many studies have linked neuroinflammation with epilepsy in human patients (Ravizza et al., 2008; Aronica et al., 2012; Gibbons et al., 2013; Bedner et al., 2015; Ferrer, 2017; DeSena et al., 2018; Wenzel et al., 2019; Tan et al., 2021; Aulická et al., 2022), which has been replicated by a variety of animal epilepsy models including but not limited to: traumatic brain injury (TBI) associated epilepsy (Abdul-Muneer et al., 2016; Kim et al., 2016; Webster et al., 2017; Sharma et al., 2019; Zhou et al., 2020; Gao et al., 2022; Golub and Reddy, 2022), post-ischemic stroke epilepsy (Tröscher et al., 2021), glioma-associated epilepsy (Olsen and Sontheimer, 2008; Buckingham et al., 2011; Buckingham and Robel, 2013; MacKenzie et al., 2016; Tewari et al., 2018; Campbell et al., 2020; Komiyama, 2022), kainic acid (KA)-induced epilepsy (Canto et al., 2022; Han et al., 2019; Huang et al., 2022; Hubbard et al., 2016; McRae et al., 2010; Takahashi et al., 2010; Wolinski et al., 2022; Wu, Z et al., 2021), pilocarpine-induced epilepsy (Borges et al., 2003; Canto et al., 2022; Han et al., 2019; Kong et al., 2012; Mátyás, A et al., 2021; Ravizza et al., 2008; Schauwecker, 2012; Shapiro et al., 2008; Wyeth et al., 2012), kindling models of epilepsy (Kołosowska et al., 2016; Ueno et al., 2020), and a β1-integrin knockout astrogliosis mouse model (Robel et al., 2015). The models particularly analogous to human MTLE-HS include the pilocarpine model and the KA model, which exhibit varying degrees of HS in addition to upregulation of proteins associated with immune responses and inflammation (Canto et al., 2022).

Notably, both short-term and chronic exposure to inflammation can increase brain excitability and lead to lower seizure thresholds (Inyushin et al., 2010; Vezzani et al., 2013). In fact, application of lipopolysaccharide (LPS) to induce neuroinflammation in rat models of epilepsy has been shown to increase susceptibility to KA, pilocarpine, and pentylenetetrazol (PTZ)-induced seizures, as well as increased hippocampus neuronal degeneration (Galic et al., 2008; Huang et al., 2022).

The molecular components of perineuronal nets are both neuronal and glial in origin (Brückner et al., 1993; Giamanco and Matthews, 2012) and include hyaluronic acid, hyaluronan and proteoglycan link (Hapln) proteins, tenascins R and C, and a variety of chondroitin sulfate proteoglycans (CSPGs), mainly of the lectican family, including aggrecan, versican, brevican, and neurocan (Figure 1).

Aggrecan is the primary lectican component of PNNs (Giamanco and Matthews, 2012; Morawski et al., 2012; Wen T. H. et al., 2018) as well as the most well-studied; it is the loss of aggrecan that is most associated with critically impaired (Kwok et al., 2010) to practically ablated (Rowlands et al., 2018) PNN structures. Aggrecan and its fellow lecticans are anchored to neuronal cell membranes by hyaluronic acid (HA), which is produced by hyaluronic acid synthase (HAS) and stabilized by Hapln proteins, predominantly Hapln1 and 4 (Kwok et al., 2010; Mohamedi et al., 2020; Jakovljević et al., 2021). HA, HAS and Hapln proteins are also critical for PNN formation (Carulli et al., 2010; Giamanco et al., 2010; Kwok et al., 2010), as is tenascin-R (TnR). TnR, which links the lecticans of the structure, is a direct component of the PNNs, whereas tenascin-C (TnC) is affiliated with the structure but does not appear to physically contribute to it (Morawski et al., 2014). Instead, it interacts with cell surface receptors like integrins and cell adhesion molecules, and indirectly modulates the other constituents of the ECM (Jakovljević et al., 2021). Aggrecan is primarily produced by neurons (Giamanco et al., 2010) and CA2 pyramidal neurons (Carstens et al., 2016), whereas most of the other CSPGs appear to be expressed by astrocytes, which express transcripts for HAPLN1, TnR, and the other three lecticans (Giamanco and Matthews, 2012). Hyaluronic acid binding protein (HABP) is associated with both neurons and glia; glial removal results in diminished but not completely depleted HABP expression in vitro (Giamanco and Matthews, 2012).

Visualization of these structures is most often achieved using the plant lectin marker Wisteria floribunda agglutinin (WFA) (Figure 2), which binds to the glycosaminoglycan (GAG) side chains of the PNNs and is considered a fairly universal marker (Giamanco et al., 2010; Slaker et al., 2016). GAGs, which adhere to the CSPG/lectican backbone of the structure, express various sulfation patterns that contribute heavily to the negative charge of PNNs as well as influencing their overall heterogeneity, dividing CSPGs into primarily two groups with either 4-sulfated or 6-sulfated GAG chains (Bonneh-Barkay, 2009; Miyata et al., 2018). These sulfation patterns- much like the nets themselves- are dynamic and have been observed to change during development, adolescence, and through adulthood (Carulli et al., 2010; Yutsudo and Kitagawa, 2015).

Despite being first immortalized in published form by Camillo Golgi in 1898 (Celio and Blumcke, 1994), the purposes of PNNs are still not fully elucidated. In general, the ECM is important for organization, support, and maintenance of the neural and glial cells it surrounds and encapsulates. PNNs specifically are furthermore intimately involved with the formation, stability and remodeling of synapses and synaptic signaling (Dityatev and Schachner, 2003; Frischknecht and Gundelfinger, 2012; Bozzelli et al., 2018; Lipachev et al., 2019; Reichelt et al., 2019) and thus neuronal plasticity and learning and memory (Romberg et al., 2013; Tsien, 2013; Thompson et al., 2018; Bosiacki et al., 2019; Wei et al., 2019; Chelyshev et al., 2022; Fawcett et al., 2022).

Juvenile animals still in early development display experience-dependent neuronal plasticity. This capability, observed during what is referred to as the “critical period,” is fairly depleted by the time postnatal development ends, which also coincides with the formation of PNNs (Pizzorusso et al., 2002; Hensch, 2004; Gundelfinger et al., 2010; Miyata and Kitagawa, 2015; Cornez et al., 2018). PNNs appear to stabilize or “lock” synapses into place to reduce synaptic plasticity at this point, as their physical presence restricts the placement of astrocytic leaflets and presynaptic boutons. This is particularly relevant in the sensory system, where incoming sensory information competes for cortical representation. In the visual system, for example, the closure of the ocular dominance that allocates cortical territories to each of the eyes is marked by the deposition of PNNs. Experimental degradation of the visual system PNNs using the enzymatic drug chondroitinase ABC (ChABC) reverses this process, restoring a more adolescent-like plasticity in adult animal models (Pizzorusso et al., 2002; McRae et al., 2007; Carulli et al., 2010; Hou et al., 2017). Degradation of PNNs around CA1 and CA2 hippocampal regions in mice replicated these findings, shifting the excitatory/inhibitory balance and reinstating juvenile-like plasticity (Carstens et al., 2016; Khoo et al., 2019). This characteristic stabilization of synapses further suggests a role in the formation or retention of memory (Romberg et al., 2013; Howell et al., 2015; Yang et al., 2015; Rowlands et al., 2018; Thompson et al., 2018; Wei et al., 2019) [See (Wingert and Sorg, 2021; Fawcett et al., 2022) for thorough reviews on PNNs in plasticity and memory]. Application of ChABC for treatment of glial scars, often associated with areas of neuroinflammation, was found to promote axonal regeneration and a return to plasticity in the spinal cord after injury as well (Bradbury et al., 2002; Massey et al., 2006). It is therefore proposed that PNN structures form a “repulsive barrier” that inhibits axonal and dendritic growth, not only by physically blocking leaflet and bouton formation but also via their highly negative charges as well as their interactions with growth-suppressing signaling molecules (Bonneh-Barkay, 2009; Sharma et al., 2012) [For a thorough review on glial scar formation and its immunological interactions, see (Raposo, 2014)].

Other studies have alluded to further purposes of PNNs such as helping to regulate extracellular reactive oxygen species (ROS) or protect against oxidative stress (Morawski, 2004; Beurdeley et al., 2012; Cabungcal et al., 2013) and enabling the inhibitory, GABAergic PV⁺ fast-spiking neurons (FSNs) that they envelop to fire action potentials at extremely high rates (Balmer, TS, 2016; Tewari et al., 2018). They also appear to play a role in ion buffering, as indicated by their highly anionic structures mentioned above (Brückner et al., 1993; Brückner et al., 1998; Härtig et al., 1999).

Though expressed and active in normal healthy physiology due to the constant turnover of ECM components, MMPs and their regulators are important mediators in CNS inflammation (Klein and Bischoff, 2011; Gaudet and Popovich, 2014) and neuroinflammatory processes (Rosenberg, 2002; Berezin et al., 2014; Reinhard et al., 2015) and have been associated with pathological disorders and diseases such as TBI(Abdul-Muneer et al., 2016; Pijet et al., 2018), stroke (Pielecka-Fortuna et al., 2015; Akol et al., 2022), glioma (Markovic et al., 2005; Varol and Sagi, 2018), and of course epilepsy (Wilczynski et al., 2008; Rankin-Gee et al., 2015; Zybura-Broda et al., 2016; Dubey et al., 2017; Pijet et al., 2018).

Systemic inflammation is well associated with increases in remodeling enzymes, especially with regards to epileptic activity. MMP-2 and MMP-9 in particular have been found to be upregulated in glia and neurons in general seizure activity, TLE, and post-status epilepticus (Dubey et al., 2017; Dzwonek et al., 2004; Kim et al., 2017; Lukasiuk et al., 2011; Pijet et al., 2018; Reinhard et al., 2015; Szklarczyk et al., 2002; Ulbrich, P. et al., 2020; Wegrzyn, D. et al., 2020; Wilczynski et al., 2008; Zybura-Broda et al., 2016). Rankin-Gee et al. (2015) find that seizure activity increases MMP proteolysis of aggrecan, which suggests a mechanism by which PNNs are degraded in epilepsy and thus contribute to the progression of the disorder. Indeed, one study found that two different strains of MMP-9 overexpressing rats displayed higher seizure susceptibility to PTZ kindling than wild type rats (Wilczynski et al., 2008). Kim et al. (2017) found upregulation of genes encoding MMP9 and 14 and ADAMTS1 in multiple brain injury and BBB leakage models as well as in resected tissue from human TLE patients (Kim et al., 2017).

In glioma-associated epilepsy, for example, epileptic activity in peritumoral areas may be attributed to MMP-driven PNN degradation. Glioma cells and tumor-associated macrophages (TAMs) release MMPs, but the host’s inflammatory cells can also release MMPs in response to the tumor cells. Glioma has also been found to overexpress ADAMTS-5, which as mentioned previously targets CSPGs, specifically cleaving brevican (Nakada et al., 2005). Furthermore, the ECM itself is thought to actively promote cancer growth by altering collagen degradation and re-deposition via remodeling enzymes so as to clear space and allow progression and growth of the tumor (Shapiro, 1998; Markovic et al., 2005; Varol and Sagi, 2018). Peritumoral areas immediately surrounding resected low-grade epilepsy-associated tumors exhibit not only increased inflammatory markers, but also an increased ripple rate, possibly implicating an MMP-driven discrepancy in excitatory-inhibitory balance (Sun et al., 2022).

Treatment of cultured rat astrocytes and microglia with inflammatory mediators such as IL-1β, TNF-α, and LPS also stimulates the production of MMP-2 and MMP-9 (Gottschall and Deb, 1996; Shapiro, 1998; Dzwonek et al., 2004), and accelerates the epileptogenesis process and/or increases seizure susceptibility in rat models of kindling induced seizures (Wilczynski et al., 2008; Kołosowska et al., 2016). In concurrence, application of MMP inhibitors or knockout of MMP-9 seemingly protected mice and rats against KA-induced and kindling-induced seizures (Wilczynski et al., 2008; Pollock et al., 2014) as well as TBI-induced spontaneous seizures (Pijet et al., 2018). Notably, a critical amount of MMPs seems to be required for optimal function-inhibiting MMP-2 and MMP-9 can suppress plasticity in the visual cortex, but briefly inhibiting the same MMPs post-stroke can rescue plasticity (Akol et al., 2022), indicating that intervention timing and intensity are crucial.

Degraded segments of PNNs and ECM are known to act as alarmins or damage-associated molecular patterns (DAMPs) and thus amplify CNS inflammation (Gaudet and Popovich, 2014; Jang et al., 2020). Buildup of fragmented HA in particular, specifically the low molecular weight (LMW) HA (10–500 kDa) generated due to ECM damage, is known to serve as an injury and inflammatory signal, binding to CD-44 and TLR4 to induce pathways such as NFκB signaling and increasing IL-1β and TNF-α in vitro (Noble, 2002; Wang et al., 2006). Hyaluronidase treatment of cultured rat astrocytes induced more stellate-like, branching morphology, indicating cleavage of HA may be associated with astrocytic form and/or function (Konopka et al., 2016).

CSPGs and tenascins are released from activated astrocytes following CNS injury, with TnC specifically increasing after exposure to IL-1β, TNF-α and INF-γ (Laywell et al., 1992; Jang et al., 2020). TnC serves as an activator of TLR4, which is well associated with increased pro-inflammatory cytokines and neuroinflammation (Midwood et al., 2009; Eidson et al., 2017). Studies have also found increased CSPG expression in the pathological hallmarks of neurodegenerative diseases with chronic inflammation components such as AD plaques and tangles and MS lesions (Jang et al., 2020).

This is not to say that all ECM components and cleavage products are pro-inflammatory; in fact, the GAG sidechains of CSPGs may have different effects due to increased or decreased affinity for specific chemokines depending on their sulfation pattern [see (Monneau et al., 2016) for a thorough review on chemokine-GAG interactions]. 6-sulfated CSPGs, for example, appear to help suppress microglial activation and production of IL-6 and TNF-α (Tan and Tabata, 2014; Jang et al., 2020).

Many of the molecules released by reactive astrocytes can indirectly or directly lead to ECM-altering outcomes, including changes in the expression of HA, CSPGs, and tenascin proteins (Wiese et al., 2012; Bosiacki et al., 2019). Activated astrocytes are known to migrate to injury sites in the CNS and release inflammatory factors such as CCL2 and 3 (Fabene et al., 2010), as well as increasing secretion of CSPGs including neurocan, versican, and brevican, likely via TGF-β and subsequent signaling (Schiller et al., 2004). Genes associated with ECM and integrin signaling are also significantly upregulated in rat models of kainic acid and pilocarpine epilepsy after SE (Han et al., 2019) as well as being associated with genes upregulated in astrogliosis (Zamanian et al., 2012). PTZ-induced seizures were found to trigger astrogliosis in the targeted hippocampus and many cortical areas, as well as overall increases in the amount of extracellular matrix (Ueno et al., 2020) [For a thorough review of glial-ECM remodeling, see (Kim et al., 2016)].

Removal of multiple lecticans in vivo and in vitro is associated with not only abnormal PNN morphology, but also altered synaptic function, including reduced inhibitory synapses and increased excitatory presynaptic markers (Geissler et al., 2013) (Mueller-Buehl et al., 2022). Aggrecan knockout cells show a complete lack of PNNs in vitro (Kwok et al., 2010), but removal of other CSPGs does not appear to lead to such drastic changes. Brevican, for example, appears to be required for modulating synapses and excitatory contacts of inhibitory interneurons; lack of brevican at PV⁺ interneurons led to altered pruning of excitatory synapses and thus alterations in spike properties and miniature EPSCs (Favuzzi et al., 2017). Although a genetically deleted brevican mouse model did not show changes in the structure of the PNN itself, multiple studies reported seeing significant alterations of synaptic plasticity and transmission (Frischknecht and Gundelfinger, 2012; Blosa et al., 2016). A knockout mouse model of neurocan also correlated with notable decreases in brevican mRNA levels and visibly altered brevican ECM structures (Sonntag et al., 2018; Schmidt et al., 2020), though neurocan itself did not appear to be altered. Astrocytically released ECM molecules such as TnR and laminins interact directly with voltage-gated Ca² channels, AMPARs, and GABARs, and as such, influence synaptic organization and function (Dityatev and Schachner, 2006); thus, upon astrogliosis, upregulation or alteration of ECM components and signaling molecules can easily follow.

Replicating CSPG degradation along with removal of HA in vitro was found to increase synaptogenesis and decrease glutamate sensitivity (Pyka, 2011), both of which could readily lead to increases in excitability. Increases in excitability are not the only method by which the excitation/inhibition balance can be altered; in fact, a study in which PNNs were experimentally degraded using ChABC showed reduced excitability of PV⁺ neurons and inhibitory synaptic transmission in the visual cortex (Liu et al., 2019). Similarly, a recent study (Dzyubenko, 2021) that also experimentally degraded PNNs saw a decrease in the density of inhibitory synapses to both excitatory and inhibitory neurons, along with an increase in the strength of inhibitory synapses. However, the action potential threshold for excitatory neurons also decreased, and as such, the strengthened inhibitory neuron outputs were insufficient to balance the excitatory activity, leading to overall network changes (Dzyubenko, 2021).

TnC, which interacts with other cell surface receptors and helps to regulate cell growth, adhesion, and migration, is upregulated early in inflammation, either by the pro-inflammatory IL-1 pathway or possibly by IL-4, IL-13, or TGF-β, which are considered anti-inflammatory (Chiquet-Ehrismann and Chiquet, 2003). It is expressed at high concentrations in disorders characterized by chronic inflammation, and its production further induces inflammatory responses, as seen in astrocytes in vitro (Wiese et al., 2012) and a mouse model of AD, for example (Xie et al., 2013). TnC further appears to regulate astrocytic maturation during embryonic development in cortical cells and in the spinal cord (Karus et al., 2011; Wiese et al., 2012), implicating this ECM molecule in astrocyte development and providing a potential feedback loop effect of altered PNNs affecting astrocytes. HA also interacts with astrocytes at two specific molecular weights (low HA, 10–500 kDa and high HA, >500 kDa), both of which appear to modulate astrocytic responses to TLR agonists and upregulate IL-10 expression via TLR pathways (Chistyakov et al., 2019).

Taken together, it is clear that the presence of PNNs is well-associated with astrocytes at the synapse, and that neuroinflammation can not only alter these interactions but can be upregulated as a result of these interactions as well.

At this point, there appears general agreement that PNNs are crucial for synaptic function and retention of memory, and that removal of or otherwise diminished PNNs can be associated with a variety of brain disorders. Whether PNNs are altered in neurodegenerative disorders specifically has historically been controversial. A recent study (Crapser et al., 2020b) has provided strong evidence for microglial engulfment of PNNs in AD, further showing that induction of an inflammatory state using LPS injections induced similar PNN-degrading phenotypes in wild type mice. There have also been a number of studies looking at alterations of PNNs in psychiatric disorders in particular [see (Carceller et al., 2022) for a thorough review].

Due to the inflammation aspect of our hypothesis, although it was not discussed, other immune cells such as microglia certainly have their own interactions with astrocytes and PNNs. As mentioned above, Crapser et al. (2020b) found that activated microglia are heavily implicated in PNN degradation, whether directly or indirectly, and another 2020 study (Nguyen et al., 2020) found that cytokine IL-33 released by hippocampal neurons induces microglial ECM remodeling. Astrocytically-released IL-33 has also been found to drive synaptic engulfment by microglia (Vainchtein et al., 2018) and microglial activation has been linked to epilepsy in general (Shapiro et al., 2008; Hiragi et al., 2018) [see (Andoh and Koyama, 2021) for review of microglia and plasticity]. Another paper looking specifically at depletion of microglia in Huntington’s disease found that knockout of microglia resulted in decreased PNN degradation, with denser PNN expression in all brain regions as well as reduced astrogliosis (Crapser et al., 2020a).

The time course of all this PNN remodeling may still be up for debate as well; one study suggests PNN modification occurs during each sleep cycle, varying with circadian rhythms (Pantazopoulos et al., 2020). However, such a quick turnaround of PNN degradation and production would likely have larger implications in multiple disease states as well as in healthy brains, where again, PNNs appear to play critical roles in plasticity and stability and are required for the normal function of the enveloped PV⁺ FSNs. A more recent study observed that although PNN expression did not change diurnally, it does increase in the absence of microglia, which display changes in ramification during the circadian cycle in mice (Barahona et al., 2022). If PNN integrity does in fact alter every 24-h cycle, the question of how normal brain function is maintained-especially with regards to inhibitory neuronal activity from FSNs-comes to the forefront.

Lastly, although touched upon earlier, the physiology of reactive astrocytes and the classification of such has continued to be controversial. Formerly considered in more binary terms such as “reactive” versus “nonreactive,” or “neuroprotective” vs. “neurotoxic,” astrocytes are now more likely to be classified holistically and along a continuum, categorized by their morphological, functional, and molecular changes, as well as taking into consideration their immunoreactivity markers and the brain regions they are expressed in, amongst others factors (Escartin et al., 2021).

To fully explore this hypothesis, designing experiments to artificially alter the proposed PNN-astrocyte interactions is the crucial next step. Although there are enzymes that can be applied to degrade ECM and PNNs in vitro and in vivo and a viable aggrecan knockout mouse has been developed (Rowlands et al., 2018), there is currently no method of artificially inducing ECM growth or PNN formation. As discussed above, removal of or interfering with normal microglia function results in more highly condensed, intense, or concentrated PNNs (Liu et al., 2021; Barahona et al., 2022), but does not appear to result in de novo synthesis of the structures. To this end, the advent of a true PNN synthesis method would be a significant step towards truly confirming and/or revealing the roles of these structures in healthy and diseased brains alike.

One of the other stumbling blocks in determining the purposes and characteristics of PNNs and ECM in general is the fact that so many molecules comprise these complex structures that it becomes difficult to study. However, efforts to analyze and replicate its complexity have resulted in widely used biomaterials like basement membrane-like matrix (Matrigel) (Benton et al., 2011) and a variety of ECM-based polymers used for 3D modeling [see (Vigier, 2016) for a thorough review].

In addition to therapeutically targeting PNNs and the ECM to treat disorders such as epilepsy and AD, some suggest that manipulating these structures may be a potential anti-aging technique (Yang et al., 2021). As discussed previously, removal of PNNs using ChABC or hyase can restore the plasticity of the brain to critical period-like levels, implicating careful “editing” of the brain as a way to potentially mitigate or rewind the effects of age on memory formation and retention.