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/9780197549469.003.0022
The thalamus, a deep brain structure with broad connectivity, plays key roles in local and global rhythmic activity in sleep, arousal, and cognition. Disruptions in thalamic and thalamocortical circuits—and the ensuing hypersynchrony and hyperexcitability—have been widely studied in the context of genetic epilepsies and have garnered increasing interest in the context of acquired epilepsies. In this chapter, key structural, synaptic, cellular, and biophysical elements underlying the rhythmogenic properties of the thalamus are described, and then how diverse perturbations of these properties in rodent genetic models ultimately generate or facilitate seizures. The chapter briefly highlights recent studies which have identified mechanisms of thalamic involvement in the development or modulation of epilepsies acquired following incidents such as traumatic brain injury and ischemic stroke. Understanding how diverse etiologies converge upon thalamic hyperexcitability can pinpoint elements of vulnerability, resilience, redundancy, necessity, and sufficiency in the thalamocortical circuit. Understanding how the thalamus generates and modulates aberrant activity—even when it is not the primary or sole site of genetic perturbation—will be key to identifying therapeutic targets and paradigms to treat epilepsies. Such efforts will benefit from continued advancements in our knowledge of cell-type heterogeneity, meso- and macro-scale connectivity, and interspecies differences in the thalamus.
The thalamus, a subcortical diencephalic structure involved in sleep, arousal, consciousness, and cognition, has been a site of investigation in seizures since the 1940s, starting with experimental studies in cats and monkeys by Dempsey, Morison, Jasper, and Droogleever-Fortuyn. These studies showed that intralaminar stimulation evoked bilateral 3-Hz spike-and-wave discharges and behavioral arrest (reviewed by Jasper, 1948, 1991). In 1950, the first depth recordings in the thalamus during petit mal seizures (now known as “absence seizures”) were obtained from patients with absence type epilepsy (Spiegel and Wycis, 1950), and subsequent studies of thalamic activity during seizures supported the hypothesis that “the clinical case of petit mal epilepsy is due to a disturbance in the thalamus which causes a rhythmic discharge throughout the cortex” (Williams, 1953). While there has been a debate on whether the seizures initiate in cortex or thalamus, thalamic firing has indeed been found necessary to maintain absence seizures in rodents (Sorokin et al., 2017). The study of epileptic discharges, in both patients and animal models, has yielded basic insights about thalamic organization and function. Reciprocally, advances in our understanding of thalamic structure and function have elucidated mechanisms of the generation and propagation of epileptic activity such as spike-wave discharges underlying absence seizures, and it has opened avenues for the therapeutic modulation of seizures.
Endowed with extensive, reciprocal connections with the entire cerebral cortex and other subcortical structures (Jones, 1985), and possessing rhythmogenic properties (reviewed by Fogerson and Huguenard, 2016), the thalamus plays a critical role in the generation of normal and pathological synchronized oscillatory activity (Steriade, McCormick, and Sejnowski, 1993). Decades of research have revealed disruptions in thalamocortical excitability and synchrony as a convergence point for epileptic activity related to multiple genetic etiologies and pharmacological manipulations. More recently, cell-type-specific optogenetics have enabled causal manipulations of rodent thalamocortical synapses and circuits, demonstrating a role for the thalamus as a “choke-point” in certain forms of seizures and epilepsies (Paz et al., 2013; reviewed by Paz and Huguenard, 2015). Indeed, in 2018 the U.S. Food and Drug Administration (FDA) approved deep brain stimulation of the anterior thalamus as a treatment for refractory focal epilepsy (reviewed by Fisher and Velasco, 2014).
By understanding exactly which aspects of thalamic function are vulnerable—or resilient—to disruptions, we can clarify strategies to effectively interfere with seizure expression while minimizing side effects. In this chapter, we describe the key anatomical, structural, cellular, and biophysical elements of the thalamus which confer its rhythmogenic properties. We then provide a brief overview of the extensive literature on genetic rodent models of seizures, with particular emphasis on the diverse role of calcium channels in absence seizures and compensatory changes in the thalamocortical network. Finally, we briefly discuss recent work highlighting thalamic involvement in the development or modulation of epilepsies acquired after brain injuries.
The thalamus is composed of various nuclei embedded in diverse, multiregional neural circuits, whose complexity and detailed connectivity are not yet fully understood (reviewed by Halassa and Sherman, 2019; Shepherd and Yamawaki, 2021). Given the widespread connectivity between thalamus and the entire cerebral cortex, the thalamus is well-positioned to control the generation and maintenance of cortical rhythms. Furthermore, the thalamus is notable for its generation and maintenance of synchronous oscillatory activity (reviewed by Steriade, McCormick, and Sejnowski, 1993; Huguenard and McCormick, 2007; Beenhakker and Huguenard, 2009; Crunelli and Hughes, 2010; Fogerson and Huguenard, 2016). Such oscillations are considered critical substrates of states of arousal, sleep, and consciousness (reviewed by Steriade and Llinas, 1988; Steriade, McCormick, and Sejnowski, 1993; McCormick and Bal, 1997; Crunelli et al., 2018).
Three cardinal rhythms of the thalamus include delta, spindles, and slow-wave oscillations, all of which occur during non–rapid eye movement (NREM) sleep (reviewed by Crunelli and Hughes, 2010; Fogerson and Huguenard, 2016; Fernandez and Lüthi, 2020). We will describe them in the sixth section. Key features of the thalamus which enable the generation of these synchronous rhythms include cell-intrinsic firing properties, microcircuit connectivity, long-range connections with the cerebral cortex and basal ganglia (reviewed by Deschênes, Veinante, and Zhang, 1998; Jones, 2007; Paz et al., 2007; Charpier, Beurrier, and Paz, 2010), and the precise timing of synaptic inputs (reviewed by Huguenard and McCormick, 2007). The cellular and circuit properties underlying the generation and regulation of these oscillations can subserve aberrant oscillations underlying epileptic seizures, and they will therefore be reviewed in the following sections. Thalamic connectivity with basal ganglia will not be reviewed in this chapter.
The multiple nuclei of the thalamus are each embedded in reciprocal circuits that can be local (i.e., between thalamic relay nuclei and the reticular thalamic nucleus) or long range (i.e., between thalamic relay nuclei and cortex) (Fig. 22–1A).
Basic structural elements of thalamic and thalamocortical circuits. A. Coronal section of mouse brain labeled with the neuronal nuclear protein NeuN. Colored regions highlight the reticular thalamic nucleus (known as nRT, for nucleus reticularis thalami (more...)
The reticular thalamic nucleus (abbreviated nRT for nucleus reticularis thalami) is a thin shell of GABAergic neurons which surrounds the lateral boundary of dorsal thalamocortical (TC) relay nuclei (Houser et al., 1980; Pinault, 2004). nRT neurons form reciprocal connections with glutamatergic neurons in thalamic relay nuclei (Jones, 1975; Sherman and Guillery, 1996; Guillery and Sherman, 2002) and receive corticothalamic (CT) collateral afferents; thus, the nRT provides both feedback and feedforward inhibition in the thalamus and critically shapes net thalamic output (reviewed by Pinault, 2004; Paz and Huguenard, 2015; Halassa and Acsády, 2016; Crabtree, 2018). Approximately 10%–25% of synapses in the nRT are GABAergic (Williamson, Ohara, and Ralston, 1993; Liu and Jones, 1999). There is histological and ultrastructural evidence that the nRT receives GABAergic input from the external segment of the globus pallidus in squirrel monkeys (Asanuma, 1994), the substantia nigra pars reticulata in cats (Paré et al., 1990), the basal forebrain and mesopontine tegmentum in rats (Jourdain, Semba, and Fibiger, 1989), and notably, from the nRT itself in cats, rats, and mice (Houser et al., 1980; Pinault, Bourassa, and Deschênes, 1995; Liu and Jones, 1999; Clemente-Perez et al., 2017). However, the existence of functional, GABAergic synapses between nRT neurons is debated (discussed below).
The majority of thalamic territory is occupied by various thalamic relay nuclei, which earned the term “relay” due to their well-studied role in transmitting afferent sensory information to other brain regions. Other than the nRT, sources of GABAergic inhibition in thalamic relay nuclei include the zona incerta, anterior pretectal nucleus, basal ganglia (including substantia nigra pars compacta, internal globus pallidus, ventral pallidum), and the pontine reticular formation, which are well described elsewhere (reviewed by Halassa and Acsády, 2016). Based on their primary afferent source, thalamic relay nuclei are broadly organized into first-order and higher-order nuclei. First-order nuclei receive sensor-specific driving input from subcortical sources, while higher-order nuclei receive driving input from the cortex (reviewed by Halassa and Sherman, 2019). Based on their input-output connectivity, thalamic relay nuclei can also be broadly organized into sensory, limbic, motor, and executive/cognitive domains, which are preserved along the thalamo-cortico-reticular pathway (Sherman and Guillery, 1996; Jones, 2007; also see recent studies, e.g. Harris et al., 2019; Phillips et al., 2019). The vast majority of neurons in thalamic relay nuclei are glutamatergic and are called “thalamocortical” (TC) neurons because they form reciprocal, long-range, excitatory, and topographic connections with cortical neurons. For example, key aspects of visual information processing involve TC neurons in the dorsal lateral geniculate nucleus (dLGN), which receive retinal input, project to primary visual cortex (V1), and receive corticothalamic input from V1; in addition, dLGN neurons form reciprocal connections with the GABAergic neurons of the nRT, which also receive corticothalamic input from V1 (Sherman and Guillery, 1996). Similarly, key aspects of somatosensory information processing involve TC neurons from the ventrobasal nucleus (VB), which receive input from the spinothalamic tract and medial lemniscus (Mo et al., 2017), project to primary somatosensory cortex (S1), and receive corticothalamic input from S1; in addition, VB neurons form reciprocal connections with the GABAergic neurons of the nRT, which also receive corticothalamic input from S1 (Deschênes, Veinante, and Zhang, 1998; Jones, 2009). The somatosensory VB-S1-nRT circuit is one of the most well-characterized circuits in the context of basic intrinsic and synaptic properties, and in rodent models of seizures and epilepsy.
To understand the cellular basis of thalamocortical rhythmic activity, it is important to understand the relevant cell types and their connectivity (i.e., microcircuit motifs). Thalamocortical circuits consist of multiple key cell types, including glutamatergic TC relay neurons in thalamic relay nuclei, glutamatergic corticothalamic (CT) neurons, primarily in cortical layers 5 and 6, and GABAergic neurons in the nRT (Fig. 22–1B) (Deschênes, Veinante, and Zhang, 1998; Shepherd and Yamawaki, 2021). Although there is increasing appreciation of cell-type heterogeneity among these populations (Clemente-Perez et al., 2017; Phillips et al., 2019; Li et al., 2020; Martinez-Garcia et al., 2020), the basic organization of these microcircuit motifs in the thalamus is generally accepted (reviewed by Paz and Huguenard, 2015).
Thalamocortical:
Recurrent excitation (formed by reciprocal glutamatergic projections; prominent in cortical networks)
Feedforward inhibition (disynaptic component, whereby excitatory inputs recruit local inhibition en route to its recipient; can effectively modulate the strength of the efferent signal)
Intra-thalamic:
Feedback inhibition (reciprocally connected glutamatergic and GABAergic neurons; an inhibitory mechanism to control local excitatory activity)
Counter-inhibition (reciprocally connected GABAergic neurons)
In addition to cell-type-specific connectivity patterns, these distinct microcircuits include layer-specific connectivity patterns (Halassa and Sherman, 2019; Shepherd and Yamawaki, 2021). For example, in first-order thalamic relay nuclei such as the dLGN and VB, TC neurons send long-range glutamatergic projections to related cortical regions in a topographic manner, primarily in layer 4 (Deschênes, Veinante, and Zhang, 1998; Jones, 2007). Layer 6 CT neurons send long-range glutamatergic projections to TC neurons, and also form glutamatergic collaterals onto nRT, which enable feedforward inhibition of TC neurons (Jones, 2007; Sherman, 2007).
Functional connections between nRT neurons, which form a reciprocal counter-inhibition microcircuit, introduce a distinct set of principles into the framework of thalamic function in health and disease (Huntsman et al., 1999; Sohal and Huguenard, 2003; Paz and Huguenard, 2015; Makinson et al., 2017; Crabtree, 2018). Gap junction-coupled intra-nRT electrical synapses have been consistently demonstrated by multiple groups (Landisman et al., 2002; Long, Landisman, and Connors, 2004; Deleuze and Huguenard, 2006). However, the existence of intra-nRT chemical GABAergic synapses has been a subject of controversy (reviewed by Paz and Huguenard, 2015; Crabtree, 2018), with multiple electrophysiological studies, suggesting their existence and functionality in ferrets, rats, and mice (Shu and McCormick, 2002; Deleuze and Huguenard, 2006; Makinson et al., 2017), and others failing to find them (Parker, Cruikshank, and Connors, 2009; Cruikshank et al., 2010; Hou, Smith, and Zhang, 2016). Nevertheless, in this chapter we refer to several studies which highlight the relevance of intra-nRT connections for our understanding of thalamic circuit synchrony.
Disruption of any microcircuit motif may lead to thalamic hyperexcitability and seizures. For example, in the Gria4–/– mouse model of absence epilepsy, there is a specific and selective deficit of the cortico-nRT projection (Paz et al., 2011). The consequent loss of feedforward inhibition (despite normal feedback inhibition) underlies the generation of pathological oscillation. In mice lacking the GABAA receptor β3 subunit, a selective loss of GABAAR-mediated inhibition in reciprocally connected nRT neurons leads to dramatic thalamic hypersynchrony in slices. The latter finding highlights the desynchronizing role of counter-inhibition within the nRT (Huntsman et al., 1999; Sohal, Huntsman, and Huguenard, 2000; Sohal and Huguenard, 2003). This role is further supported by the observation that selective loss of the voltage-gated sodium channel Scn8a in nRT neurons, which leads to loss of intra-nRT, but not nRT-TC, inhibition, causes mice to develop spontaneous absence seizures (Makinson et al., 2017).
TC and nRT neurons are generally thought to exhibit two distinct firing modes: tonic firing in response to depolarization (Jahnsen and Llinás, 1984), and burst firing (in response to hyperpolarization, or depolarization in nRT (Steriade and Llinas, 1988; Huguenard, 1996; Sherman, 2001; however, see Wolfart et al., 2005). During states of arousal (e.g., wake), thalamic neurons are relatively depolarized and exhibit “tonic” firing of action potentials at low frequencies; such tonic firing is proposed to be involved in sensory information processing (Swadlow, Gusev, and Bezdudnaya, 2002; Sherman, 2007). During states of low arousal (e.g., sleep), thalamic neurons are relatively hyperpolarized and exhibit high-frequency bursts of action potentials. Burst firing can occur periodically through both cell-intrinsic and circuit mechanisms (as described in the sixth section), and it is a critical building block of thalamic oscillations (Llinás and Steriade, 2006; Beenhakker and Huguenard, 2009; Fogerson and Huguenard, 2016). T-type voltage-gated calcium channels (subtypes CaV3.1, CaV3.2, and CaV3.3) are crucial elements of burst firing in thalamic neurons and are described in detail in the fifth section.
Oscillatory burst firing is a critical substrate of sleep. It is abolished in the nRT neurons of CaV3.3–/– mice, whereas tonic firing is not, according to ex vivo whole-cell, patch-clamp recordings (Astori et al., 2011). Lack of CaV3.3-mediated oscillatory bursting results in a selective reduction of the sleep-associated sigma frequency band power in the EEG at transitions from NREM to REM (where sleep spindles occur). Thus, CaV3.3-mediated burst firing has been proposed to be the predominant pacemaker underlying sleep spindles in the thalamus (Astori et al., 2011).
Burst firing is also a critical substrate of thalamic seizures, because of its involvement in spike-and-wave discharges (SWDs). Generalized SWDs are hallmarks of absence seizures (reviewed by Crunelli and Leresche, 2002; Blumenfeld, 2005; Tenney and Glauser, 2013; Fogerson and Huguenard, 2016). The electroencephalographic (EEG) features of SWDs are bilateral, high-amplitude events composed of stereotyped spike-and-wave signatures which typically occur at 3–4.5 Hz in humans and 6–10 Hz in rodents (reviewed by Crunelli and Leresche, 2002; Tenney and Glauser, 2013; Fogerson and Huguenard, 2016). Multi-unit recordings have shown that TC, CT, and cortical neurons exhibit synchronous burst firing, time-locked to the “spike” component of the SWD in vivo in rats and mice (Inoue et al., 1993; Sorokin et al., 2017); at the intracellular level, TC and nRT neurons exhibit burst firing correlated with the “spike” component of the SWD in vivo in rats (Slaght et al., 2002; Paz et al., 2007).
Recent studies have provided further compelling evidence for the role of thalamic burst firing in seizures (reviewed by Huguenard, 2019). Sorokin and colleagues expressed a stable step-function opsin (Yizhar et al., 2011) in VB TC neurons of Wistar Albino Glaxo from Rijswijk (WAG/Rij) rats and stargazer mice—rodent models of absence epilepsy (Sorokin et al., 2017)—which enabled them to switch the firing mode of TC cells between tonic and burst mode. They found that toggling the firing from bursting to tonic mode stopped ongoing absence seizures. Stable step-function opsins have also been used to demonstrate the role of TC bursting in atypical, nonconvulsive seizures observed in a mouse model of Dravet syndrome—seizures that have been found in human patients (Ritter-Makinson et al., 2019). Thus, there is strong evidence for the role of thalamic neuronal population bursting in the maintenance of nonconvulsive seizures (Sorokin et al., 2017). Consistent with these findings, in a different rodent model of absence epilepsy—the Genetic Absence Epilepsy Rat from Strasbourg (GAERS), McCafferty and colleagues have found that the output of TC neurons during absence seizures is strong and rhythmic, even though burst firing in each TC cell is infrequent, suggesting that bursting in each TC cell is not required for maintaining the seizure (McCafferty et al., 2018). In contrast, burst firing in nRT neurons is very robust during absence seizures and likely serves to synchronize the TC output to the cortex (Slaght et al., 2002; McCafferty et al., 2018).
Ultimately, the thalamocortical network dynamics underlying seizures is likely an emergent property of both cell-intrinsic mechanisms (e.g., T-type mediated burst firing) and circuit properties (e.g., cortical feedforward inhibition of the thalamus). Nevertheless, given that disruptions to firing mode can alter thalamic synchrony in ways that can lead to seizures, and that thalamic synchrony depends in part on T-type calcium channels, we next provide an overview of the key enabler of thalamic switching in firing mode: the low-threshold T-type calcium channel.
Voltage-gated calcium channels (VGCCs) are widely expressed throughout the mammalian body and brain (reviewed by Catterall, 2000), and support numerous critical functions, including cellular excitability, synaptic transmission, and synaptic plasticity (reviewed by Catterall and Few, 2008) (Fig. 22–2). VGCCs are organized into two main classes, distinguished by their biophysical properties: high-voltage activated (HVA) VGCCs, which include L-type, N-type, and P/Q-type channels (Nowycky, Fox, and Tsien, 1985; Rajakulendran, Kaski, and Hanna, 2012), and low-voltage activated (LVA) VGCCs, which include T-type calcium channels (Perez-Reyes, 2003); R-type channels are considered to have intermediate properties (Soong et al., 1993; Randall and Tsien, 1997) and are discussed further in the seventh section. VGCCs are composed of four or five distinct subunits, including a pore-forming α1 subunit (of which there are three families, CaV1, CaV2, and CaV3), a transmembrane complex of α2 and δ subunits, an intracellular β subunit, and in some cases a transmembrane γ subunit (Catterall, 2000). Because the α1 subunit contains the conduction pore, voltage sensors, gating mechanisms, and most of the known sites of channel regulation, its family and subtype typically correlate with whether the calcium channel is of the HVA or LVA type (Catterall and Few, 2008). Here we provide an overview of LVA T-type calcium channels, given their prominent role in thalamic bursting rhythmogenesis (which we will describe in the sixth section). The role of HVA R and P/Q-type calcium channels in seizures will be discussed in the seventh section.
CaV3.1 and CaV3.3 are the dominant T-type calcium channel subunits expressed in TC relay nuclei and in the nRT. A. Sagittal section of mouse brain indicating location of the nRT and TC relay nuclei (labels are not drawn to scale). All images obtained (more...)
T-type calcium channels are critical determinants of thalamic neuron electrophysiology (Huguenard, 1996). In addition to their activation at low voltages, T-type calcium channels have unique inactivation properties (time-dependent inactivation, steady-state inactivation, and recovery from inactivation), as well as lower conductance (thus resulting in transient and “tiny” currents), compared to HVA VGCCs. Thus, currents generated by T-type calcium channels (referred to as IT, or T-type current) are uniquely positioned to regulate cellular excitability and firing pattern, particularly near the resting membrane potential in neurons with robust burst firing behaviors (discussed further in the sixth section).
In TC neurons at resting membrane potential (between –60 to –65 mV), the majority of T-type calcium channels are in an inactivated state. Hyperpolarization (to at least –70 mV) enables de-inactivation, and subsequent depolarization leads to activation which allows a transient influx of Ca2+ and generates the low-threshold spike (LTS), also known as the rebound plateau potential. The LTS further depolarizes the membrane potential such that voltage-gated sodium channels generate action potentials, culminating in a barrage of rebound high-frequency burst firing. At this point, T-type calcium channels are inactivated (reviewed by Cheong and Shin, 2013; Fogerson and Huguenard, 2016). The sequence of events that underlies oscillatory activity is summarized in Figure 22–3, and discussed later in various pathophysiological contexts.
Cell-intrinsic and synaptic properties enabling thalamic burst firing and oscillations. (A) Schematic of the interplay between nRT and TC neurons, which gives rise to rhythmic activity in the intra-thalamic circuit such as spindle oscillations. Cortical (more...)
The α1 subunit of T-type calcium channels are encoded by the CaV3 family, which consists of three subtypes which are differentially expressed in the brain. Within the rodent thalamus, channels containing CaV3.1 (encoded by Cacna1g) are the primary source of T-type calcium channels in TC neurons. nRT neurons primarily express the CaV3.3-containing T-type calcium channel (encoded by Cacna1i) and to a lesser extent, CaV3.2-containing T-type channel calcium (encoded by Cacna1h) (Talley et al., 1999; Joksovic et al., 2006) (Fig. 22–2). Differences among the three subtypes of T-type calcium channels include inactivation kinetics and recovery from steady-state inactivation. Because inactivation influences the ability of T-type calcium channels to trigger an LTS, the expression pattern and subcellular localization of subtypes are important considerations for thalamic electrophysiological properties. T-type channels are prominently expressed in the soma and dendrites. In rodent TC neurons, electrophysiological, immunohistochemical, and computational models have found that CaV3.1 is prominently expressed in the soma and dendrites as well. While the distribution along the dendritic axis remains debated (e.g., Destexhe et al., 1998; Parajuli et al., 2010), it is a particularly interesting question, given that the spatial distribution should shape the interaction of synaptic inputs, burst generation, and downstream calcium-mediated signaling pathways (e.g., Cueni et al., 2008; Crandall, Govindaiah, and Cox, 2010).
TC neurons mainly express Cav3.1-containing channels, which recover from inactivation the fastest out of the three subtypes (reviewed by Cheong and Shin, 2013). nRT neurons mainly express Cav3.3-containing channels, which display relatively slow inactivation and are nearly voltage-independent (Huguenard and Prince, 1992). This results in longer calcium-dependent spike bursts generated in nRT, which translates into longer bouts of GABA release onto TC neurons. Thus, the difference in inactivation kinetics critically determines the precise timing and rhythmicity of the intra-thalamic oscillations.
The rich literature in this field is well-described in other reviews (reviewed by Crunelli, Cope, and Hughes, 2006; Cheong and Shin, 2013; Fogerson and Huguenard, 2016). In the future, increased specificity of genetic manipulations and increased sampling capacity of recording techniques may lead to a more definitive understanding of the combinations of calcium channel subtypes in distinct cell types that are necessary and sufficient for the expression of various rhythmic oscillations.
Thalamic circuits play key roles in local and global oscillations (reviewed by Huguenard and McCormick, 2007; Fogerson and Huguenard, 2016; Crunelli et al., 2018). Three cardinal rhythms of the thalamus during NREM sleep include delta (1–4 Hz), spindles (7–15 Hz), and slow-wave oscillations (<1 Hz). Recent work has also unveiled the thalamic involvement in alpha waves (Hughes et al., 2011) and gamma oscillations (Hoseini et al., 2021), which will not be addressed in this chapter. We have already highlighted how burst firing is mediated by T-type calcium channels (fifth section). Below we discuss how these channels, along with cellular and circuit properties of the thalamus, enable the generation of delta, spindles, and slow-wave rhythms, and what disruptions in these circuits can hijack these rhythms and contribute to seizures.
TC relay neurons display intrinsic pacemaker activity in the 1–4 Hz frequency range; this frequency band, referred to as delta, is associated with deep sleep (i.e., stage 3 NREM) and anesthesia (Brown, Lydic, and Schiff, 2010).
Cell-intrinsic mechanisms: This pacemaker activity arises from an intrinsic interplay between two currents generated by the low-threshold T-type calcium channel (IT) and the hyperpolarization-activated cyclic nucleotide-gated (HCN) (Ih) (McCormick and Pape, 1990; Soltesz et al., 1991; Nuñez, Amzica, and Steriade, 1992; Destexhe et al., 1996) (Fig. 22–3). Upon sufficient membrane hyperpolarization, IT is de-inactivated and Ih is activated. Ih is a slowly developing, non-inactivating inward current which has a depolarizing effect. Ih depolarization of the membrane activates IT and triggers a LTS, which further depolarizes the membrane and triggers Na+-dependent action potentials. The depolarization associated with the LTS deactivates Ih; in the absence of Ih activation, the membrane hyperpolarizes, which then reactivates Ih and de-inactivates IT, and the cycle continues.
Synaptic mechanisms: Synchronization of this autonomously generated delta-frequency membrane oscillation is thought to be mediated by cortico-TC and cortico-nRT-TC feedforward inhibition during cortical UP states (Steriade, Dossi and Nunez, 1991; Neske, 2016). Evidence suggests that the local thalamic delta-frequency oscillations indeed contribute to global delta waves observed in the EEG during NREM sleep. The extent to which (and the mechanisms whereby) this local thalamic delta-frequency oscillation contributes to global delta waves observed in the EEG during NREM sleep has been an active area of research. Global CaV3.1–/– mice lack delta waves at the EEG level (Lee, Kim, and Shin, 2004), at least in part due to the loss of burst firing in TC neurons (Kim et al., 2001). Recent studies employing optogenetic and chemogenetic techniques have suggested that the nRT is capable of modulating delta waves in spatially restricted regions of the cortex (Lewis et al., 2015; Fernandez et al., 2018).
Pro-epileptic mechanisms: Disruption of calcium channels appears to be a common node of dysfunction across various genetic models of absence seizures. T-type calcium channels have been the most widely studied calcium channels in absence epilepsy, and the diverse ways in which IT has been linked to absence seizures are discussed further in the seventh section. The disruption of Ih in the thalamus can also promote thalamic hyperexcitability and seizures. For instance, deletion of HCN2, an Ih subunit prominently expressed in both the thalamic relay nuclei and the nRT (Notomi and Shigemoto, 2004), leads to spontaneous absence seizures in mice. Although Ih plays a key role in pacemaker bursting activity, the loss of Ih in TC neurons in these models can still lead to seizures because it can lead to hyperpolarization of TC neurons, which in turn facilitates the de-inactivation of T-type calcium channels and burst firing (Ludwig et al., 2003; Chung et al., 2009). Interestingly, deletion of the tetratricopeptide-repeat-containing Rab8b-interaction protein (TRIP8b), an auxiliary subunit of HCN channels expressed in TC and CT neurons but not in nRT neurons, results in a milder absence seizure phenotype (Heuermann et al., 2016). Thus, preservation of Ih in the nRT of TRIP8b–/– mice appears to partially constrain the heightened excitability of the thalamocortical circuit due to the loss of Ih in TC neurons. This property is reminiscent of the apparent ability of nRT Ih to constrain cortico-nRT synaptic integration in normal contexts (Ying et al., 2007), and it suggests that Ih in the nRT confers partial resilience to hyperexcitability in the circuit.
Thalamic Ih can also be disrupted in the context of epileptogenesis. In a rodent model of post-stroke epilepsy, whereby stroke in the S1 cortex leads to secondary injury in the VB thalamus, TC neurons exhibited alterations in the biophysical properties of Ih, such as depolarized half-activation voltage (Paz et al., 2013). Biophysical modeling demonstrated that these changes in Ih were sufficient to enhance thalamic network oscillations, and optogenetic inhibition of these hyperexcitable TC neurons was sufficient to abort epileptic seizures (Paz et al., 2013).
Finally, Ih in TC neurons has also been reported to be increased, rather than decreased, in two different rat models of absence epilepsy (Kanyshkova et al., 2012; Cain et al., 2015). Perhaps a noteworthy commonality across several of these models featuring Ih disruption (in which Ih is abolished in both TC and nRT, Ih is abolished in TC, or Ih exhibits depolarized half-activation) is that they result in seizures which appear to have slower internal frequencies than typical absence seizures observed in rodents. While typical rodent SWDs are 6–10 Hz (Fogerson and Huguenard, 2016) and characteristically faster than typical human SWDs which are 3–4.5 Hz (Tenney and Glauser, 2013), the mice described above exhibited SWDs in the range of 5 Hz (HCN2–/– mice), 5.3 ± 1.7 Hz (Trip8b–/– mice), and 4–5 Hz (post-stroke epilepsy rats) (Ludwig et al., 2003; Chung et al., 2009; Paz et al., 2013; Heuermann et al., 2016).
Spindle oscillations (7–15 Hz), a global rhythmic signature of stage 2 NREM sleep, emerge in the intrathalamic circuit as a result of recurrently connected TC and nRT neurons, which exhibit alternating bouts of rebound burst firing (Llinás and Steriade, 2006; Fernandez and Lüthi, 2020).
Cell-intrinsic mechanisms: nRT neurons exhibit intrinsic rhythmicity in the spindle frequency range (7–12 Hz), which can be maintained for up to 1 second following release from hyperpolarization (von Krosigk, Bal, and Mccormick, 1993). This intrinsic oscillation in nRT arises from the interplay of IT and Ca2+-dependent small-conductance type 2 potassium channels (SK2 channels) (Köhler et al., 1996; Cueni et al., 2008; Wimmer et al., 2012), similar to the interplay of IT and Ih which gives rise to the intrinsic delta oscillation in TC neurons described above. Ca2+ influx through the Cav3.3 subtype (the main T-type calcium channel in nRT) activates the SK2 channel-mediated K+ current (ISK), which generates a slow afterhyperpolarization (AHP) following the LTS burst. The slow AHP allows the de-inactivation of IT, and thus the cycle of repetitive burst firing continues (Cueni et al., 2008) (Fig. 22–3). Another key contributor is the Cav3.2 R-type calcium channel, which has been proposed to enable prolonged elevation of Ca2+ required for induction of the slow AHP, which in turn is critical for allowing the continued oscillatory bursting in nRT (Zaman et al., 2011; Paz and Huguenard, 2012).
Synaptic mechanisms: During a spindle oscillation, rhythmic firing of nRT neurons triggers a transient burst of IPSPs in reciprocally connected TC neurons. The barrage of IPSPs, mediated by both GABAA and GABAB receptors, transiently hyperpolarizes TC neurons such that IT is de-inactivated. Upon release of inhibition from nRT, TC neurons fire LTS-mediated rebound bursts, which in turn re-excite nRT neurons and initiate another nRT burst (Huguenard and Prince, 1994a; Fogerson and Huguenard, 2016) (Fig. 22–3). The precise timing of this interplay is influenced by the duration of the IPSPs (governed by the relative expression of GABAA versus GABAB receptors in the recipient TC neurons), as well as the kinetics of the LTS (governed by the T-type calcium channel subtype expressed in each cell type). Recent studies have shown that nRT neurons can play a causal role in initiating spindles (e.g., Halassa et al., 2011; Barthó et al., 2014; Fernandez et al., 2018).
Pro-epileptic mechanisms: The relationship between spindle oscillations and spike-and-wave discharges is an unresolved debate (reviewed by Beenhakker and Huguenard, 2009; Leresche et al., 2012). Recent studies have provided support for the hypothesis that spindles can be hijacked into epileptiform activity. Children with childhood epilepsy with centrotemporal spikes (the most common focal epilepsy syndrome) exhibit transient, sleep-activated focal deficits in spindles (Kramer et al., 2021). Using high-density EEGs during sleep and cognitive testing, the authors found an inverse relationship between spike and spindle rate during NREM sleep. In addition, they found that spindle rate predicted performance on cognitive tasks. These findings provide compelling support for the hypothesis that the same underlying thalamocortical circuit can give rise to spindles and epileptic activity (Beenhakker and Huguenard, 2009), as well as underlie seizures and cognitive dysfunction (Kramer et al., 2021).
Further support for the shared circuit hypothesis comes from a recent study of a rodent model of thalamocortical dysfunction following mild traumatic brain injury (TBI) (Holden et al., 2021). Unilateral injury in the primary somatosensory cortex led to chronic secondary damage of the thalamus, including loss of nRT neurons and a selective deficit in nRT synaptic function, but not in CT and TC neuronal functions. These mice develop epileptic spikes and exhibit a loss in sleep spindles; both changes were focal to the S1 cortex ipsilateral to injury (Holden et al., 2021).
Finally, thalamic circuits can also generate slow oscillations (<1 Hz) (reviewed by Crunelli and Hughes, 2010; Neske, 2016; Crunelli et al., 2018), although these were traditionally considered to be exclusively generated in cortical networks (Steriade, McCormick, and Sejnowski, 1993). The slow rhythm, present in almost all NREM stages, has been proposed to group together periods of sleep spindles and delta waves (Crunelli and Hughes, 2010). Both TC and nRT neurons have been shown to play an active role in shaping the slow rhythm during sleep. In TC neurons, the slow oscillation arises from the interplay of IT and the leak K+ current, and is influenced by Ca2+- activated nonselective cation (CAN) current and mGluR1a activation (Hughes et al., 2002). In nRT neurons, the slow oscillation arises from IT and Ih, as well as being influenced by the CAN current, a Ca2+-activated K+ current, and a Na+-activated K+ current (Blethyn et al., 2006).
Pro-epileptic mechanisms: Pathological slow oscillations such as the slow spike-and-wave discharges (1.5–2.5 Hz) are observed in certain epilepsies such as Lennox-Gastaut Syndrome (Steriade and Amzica, 2003), but the mechanisms that lead to these remain largely unknown.
There is a growing appreciation of the genetic determinants underlying a number of epilepsies linked to thalamic dysfunction. For instance, childhood absence epilepsy (CAE) has been linked with multiple susceptibility loci and polymorphisms in genes encoding voltage-gated calcium channels and GABA receptors, which will be discussed below.
Another epileptic condition also associated with aberrant thalamocortical network activity is Lennox-Gastaut syndrome (LGS), a severe childhood epileptic encephalopathy characterized by multiple seizure types (Scheffer et al., 2017). Although LGS was not traditionally recognized as having a genetic etiology, recent advances in whole-exome sequencing have identified genetic variations in LGS patients, including loci encoding voltage-gated calcium channels (Yang et al., 2021). In LGS, epileptic activity is detected both in the cortex and in the centromedian thalamic nucleus, during generalized paroxysmal fast activity and slow spike-and-wave (1–2 Hz) epileptic discharges (Velasco et al., 1991; Dalic et al., 2020; Warren et al., 2020). Notably, high-frequency deep brain stimulation of the centromedian thalamus has been shown to alleviate seizure frequency in some LGS patients (Velasco et al., 2006; Fisher and Velasco, 2014; Warren et al., 2020), confirming the importance of the thalamus in this disorder.
In the case of Dravet syndrome, a severe myoclonic epilepsy of infancy and epileptic encephalopathy with a known monogenic mutation in SCN1A (Claes et al., 2001; Dravet, 2011) (further described later), a small but notable clinical finding has implicated the potential modulatory role of the thalamus: deep brain stimulation of the centromedian thalamus resulted in seizure reduction in one of two patients with Dravet syndrome (Andrade et al., 2010).
In this section, we will discuss findings from rodent models of genetic epilepsies and highlight the diversity of etiologies which feature some extent of disruption to thalamic firing and connectivity (reviewed by Beenhakker and Huguenard, 2009; Cheong and Shin, 2013; Maheshwari and Noebels, 2014; Paz and Huguenard, 2015; Fogerson and Huguenard, 2016; Gobbo, Scheller, and Kirchhoff, 2021). In particular, we will highlight disruptions of the rhythmogenic machinery in thalamic neurons in two epilepsies: CAE, for which there is an abundance of evidence for thalamic involvement, and Dravet syndrome, for which there exists limited but convincing evidence for thalamic involvement. Understanding how various genetic etiologies converge upon a similar phenotype can pinpoint elements of vulnerability, resilience, redundancy, necessity, and sufficiency in the thalamocortical circuit, and help enumerate strategies to intervene, prevent, or compensate for epileptic activity.
The role of the thalamus in epilepsy has been best characterized in the context of typical absence seizures (formerly known as petit mal seizures), which occur in several idiopathic generalized epilepsies such as CAE and juvenile epilepsies (Scheffer et al., 2017). Absence seizures are characterized by generalized SWDs which occur alongside abrupt behavioral arrest and loss of consciousness. SWDs consist of stereotyped spike-and-wave signatures which occur at 3–10 Hz, depending on the species (generally, 6–10 Hz in rodents, and 3–4.5 Hz in humans; reviewed by Crunelli and Leresche, 2002; Blumenfeld, 2005; Tenney and Glauser, 2013; Fogerson and Huguenard, 2016), typically lasts 2–30 seconds, and occur frequently (few to several hundreds of times a day; Panayiotopoulos, 2008). SWDs are bilateral, high-amplitude waveforms, easily detected by cortical EEG, and are well-understood to reflect an aberrant, hypersynchronous thalamocortical network.
CAE includes a group of genetically determined epilepsies—some of which show strong familial inheritance and incomplete penetrance—with SWDs being inherited as an autosomal-dominant feature (Crunelli and Leresche, 2002). Numerous susceptibility loci and polymorphisms with various extents of linkage to CAE have been identified (reviewed by Crunelli and Leresche, 2002; Maheshwari and Noebels, 2014), and the evidence indicates that absence epilepsies are complex polygenic disorders. CAE mutations have been identified in voltage-gated calcium channel subunits CACNA1A, CACNA1G, CACNA1H, and CACNG3, some of which are described further later (Jouvenceau et al., 2001; Chen et al., 2003; Liang et al., 2006; Everett et al., 2007). Mutations in GABA receptor subunit genes have also been identified in human genetic studies of CAE, including GABRG2 (Baulac et al., 2001; Wallace et al., 2001), GABRA1, and GABRB3 (reviewed by Cheong and Shin, 2013).
While genetic mutations identified in human patients typically inform the design of preclinical disease models, in some cases, the identification of genetic mutations in humans has provided powerful support for mechanisms of seizure generation previously identified from genetic and pharmacological models in rodents in vivo and ex vivo. For example, the pro-epileptic effects of the GABRB2 mutation identified in human patients, which abolished benzodiazepine-induced potentiation of GABA currents when expressed in a heterologous cell system (Wallace et al., 2001) was consistent with the previously identified anti-oscillatory mechanism of clonazepam—a benzodiazepine anti-epileptic drug used to treat absence seizures (Farrello, 1986)—in rodent ex vivo thalamic slices wherein inadequate endogenous GABAAR-mediated inhibition of nRT neurons leads to strengthening of the synchronizing nRT-TC output (Huguenard and Prince, 1994b; Huguenard, 1999).
Absence epilepsies are rarely monogenic disorders, and variants in a given gene may not all lead to aberrant function. Thus, it is critical to understand how single-nucleotide polymorphisms (SNPs) identified within the same locus give rise to epileptic cells and circuits. For example, Vitko and colleagues carried out a functional characterization of 12 SNPs found in the CACANA1 (CaV3.2) locus of CAE patients in a heterologous cell system, and then used computational modeling to understand the impact of various SNPs on T-type calcium channel function; interestingly, some, but not all, SNPs were predicted to increase burst firing propensity (Vitko et al., 2005). Such efforts will only be facilitated by the increased access to whole-exome sequencing and patient-derived induced pluripotent stem cell platforms, and they will aid in the development of precision therapies for patients with epilepsy (Klassen et al., 2011; Noebels, 2017).
Basic science and epilepsy research alike have been advanced by rodent models carrying monogenic mutations in genes encoding voltage-gated calcium channels, GABA receptors, and glutamate receptors subunits associated with absence epilepsy (reviewed by Maheshwari and Noebels, 2014). Such rodent models have provided a fertile foundation for the investigation of the thalamus’s contribution to aberrant oscillations, in some cases at the level of specific synapses (e.g., Paz et al., 2011).
For thalamic neurons, IT critically determines the coordinated rhythmic bursting between TC and nRT neurons (described in the sixth section). Unsurprisingly, aberrant calcium channel function—and subsequent changes to cellular and circuit excitability and/or firing—often leads to thalamic hyperexcitability and seizures. Aberrant T-type calcium channel function can arise from a mutation in a channel subunit, a compensatory reaction, and/or the interaction of multiple susceptibility variants. Here, we review diverse ways in which T-type calcium channels have been linked to absence seizures.
Two of the three T-type calcium channel subtypes have been linked to absence epilepsy in human genetic studies: gain-of-function polymorphisms in the CACNA1H gene (encoding CaV3.2) to sporadic CAE, and functional polymorphisms in the CACNA1G gene (encoding CaV3.1) to juvenile absence syndromes (reviewed by Maheshwari and Noebels, 2014). Similarly, enhanced T-type calcium channel function—and subsequently, enhanced burst firing—in thalamic neurons seems to be a common feature across multiple rodent models of absence epilepsies. For instance, nRT neurons in GAERS rats display elevated amplitude of IT (Tsakiridou et al., 1995) and increased CaV3.2 mRNA (Talley et al., 2000), owing to a gain-of-function point mutation in Cacna1h (Powell et al., 2009). And CaV3.1–/– mice—in which burst firing, but not tonic firing, is abolished in TC neurons—are resistant to SWDs induced by baclofen (GABABR agonist) and gamma-hydroxybutyrate (Kim et al., 2001). Accordingly, overexpression of CaV3.1 results in increased T-type calcium currents in TC neurons (although rebound burst firing was not measured) and spontaneous, ethosuximide-sensitive absence seizures (Ernst et al., 2009). The latter two did not investigate the nRT, because it expressed very low levels of endogenous and exogenous CaV3.1. Nevertheless, we cannot rule out the possibility that nRT neurons made post-translational compensatory changes in T-type calcium channels in these conditions.
Even when T-type calcium channels are not directly affected by mutations, they appear to be critical components of spontaneous absence seizures in genetically determined rodent models. Across several inbred mouse strains exhibiting absence epilepsy and ataxia, spontaneous mutations have been identified in various subunits of P/Q-type calcium channels (e.g., CaV2.1/α1A subunit in tottering mice; β4 subunit in lethargic mice; and γ2 subunit in stargazer mice (reviewed by Maheshwari and Noebels, 2014); notably, the Cacna1a mutation was identified in tottering mice prior to the first discovery of a CACNA1A mutation in cases of CAE (Fletcher et al., 1996; Jouvenceau et al., 2001). Interestingly, in tottering, lethargic, and stargazer mice with mutations in P/Q-type calcium channels, T-type calcium channels exhibit compensatory changes in their electrophysiological properties despite no changes in mRNA expression of their α1 subunits (Zhang et al., 2002). In particular, whole-cell recordings of TC neurons in thalamic slice preparations demonstrated increased peak current density and depolarized shifts in the steady-state inactivation curve (i.e., increased availability of T-type calcium channel availability upon activation threshold) (Zhang et al., 2002). Although IT properties of nRT neurons were not assessed, the changes observed in TC neurons were hypothesized to favor the enhanced cellular excitability, rhythmic bursting, and overall propensity to generate SWDs observed in these mice. Indeed, a later study cross-bed these P/Q-channel-deficient mice with T-channel- deficient mice and provided compelling evidence that T-type CaV3.1/α1G is a crucial component of the absence seizure phenotype in these mice: when harboring double null mutations in both CaV2.1/α1A and CaV3.1/α1G subunits, TC neurons are depleted of IT and these mice displayed no SWDs (Song et al., 2004). Yet another example are the HCN2-deficient mice which exhibit spontaneous absence seizures (Ludwig et al., 2003). TC neurons in HCN2–/– mice exhibit a complete loss of Ih, which leads to TC neuron hyperpolarization, which facilitates T-type calcium channel de-inactivation and therefore facilitates burst firing; ex vivo thalamic slices exhibited hypersynchronous oscillatory activity which was blocked by a T-type calcium channel inhibitor (Ludwig et al., 2003). Finally, ethosuximide, a T-type calcium channel antagonist, reduces absence seizures in GAERS rats (Polack and Charpier, 2009). Altogether, these studies provide strong support for the critical role of T-type calcium channels in absence seizures.
Nonetheless, several studies have called into question the extent to which IT-mediated burst firing in various components of single TC versus nRT cells is necessary for the generation of absence seizures (reviewed by Crunelli and Leresche, 2002; Cheong and Shin, 2013; Maheshwari and Noebels, 2014). For example, in GAERS rats and in a pharmacological model of absence seizures, McCafferty and colleagues have reported infrequent burst firing in TC neurons during absence seizures but an increased burst firing in nRT neurons (McCafferty et al., 2018). They suggested that despite the rare burst firing of TC neurons, overall thalamic output (i.e., shaped by both TC and nRT) remained strong and rhythmic during seizures (McCafferty et al., 2018). An alternative interpretation is that it is IT-mediated burst firing in the nRT neurons, and not the TC neurons, that is critical to the expression of absence seizures in these rat models. On the other hand, Lee and colleagues have reported that burst firing in nRT neurons is not essential for pharmacologically induced absence seizures in mice (Lee et al., 2014). Mice harboring double mutations of CaV3.2 and CaV3.3 displayed enhanced pharmacologically induced SWDs despite the complete abolishment of nRT burst firing; the authors attributed the SWD susceptibility to an increase in nRT tonic firing (Lee et al., 2014).
In spite of these controversies, there is abundant evidence for the powerful role of T-type calcium channels—and their disruptions—in mediating thalamic hyperexcitability and SWDs.
Although T-type calcium channels have sustained the most attention in the study of thalamic rhythms and seizures in the past decades, the role of R-type and P/Q-type calcium channels in thalamic hypersynchrony and hyperexcitability cannot be discounted. Loss-of-function mutations in CACNA1A (encoding the CaV2.1 subunit of P/Q-type channels) have been identified in CAE (Jouvenceau et al., 2001). In addition, de novo pathogenic variants in CACNA1E (encoding the CaV2.3 subunit of R-type channels) have recently been identified in patients with developmental and epileptic encephalopathies (Helbig et al., 2018). Although these patients were not reported to exhibit absence seizures, the CaV2.3 subunit is highly expressed in nRT neurons (Soong et al., 1993; Weiergräber et al., 2006), and it has recently been shown to play an important role in oscillatory burst firing of the nRT and in pharmacologically induced SWDs (Zaman et al., 2011; Paz and Huguenard, 2012). R-type calcium channels are structurally similar to HVA calcium channels, yet exhibit electrophysiological properties closer to T-type calcium channels; a critical difference, however, is that R-type channels are activated at higher thresholds than T-type channels (Soong et al., 1993; Randall and Tsien, 1997). In nRT neurons of CaV2.3–/– mice, or nRT neurons of WT mice in the presence of SNX-482 (a specific blocker of R-type channels), reduced AHP rendered them unable to generate oscillatory bursts. Notably, genetic deletion or pharmacological blockade of R-type channels promoted resilience to gamma-hydroxybutyrate-induced SWDs in mice, suggesting that nRT oscillatory bursts, enabled by R-type mediated-AHP, are critical for generation of SWDs (Zaman et al., 2011).
P/Q-type channels are widely expressed throughout the central nervous system but especially abundant in presynaptic terminals in Purkinje and granule cells of the cerebellum. Unlike postsynaptic T-type calcium channels which play a major role in shaping firing properties, HVA P/Q-type channels are primarily involved in coupling calcium influx to neurotransmitter release in presynaptic terminals (reviewed by Rajakulendran, Kaski, and Hanna, 2012). Loss-of-function mutations in various subunits of P/Q-type channels lead to absence epilepsy, in some cases accompanied by ataxia (Noebels, 2012); the enhancement of T-type channel function in mice harboring these mutations was described earlier. Interestingly, investigations of rodent absence models harboring P/Q-type loss-of-function mutations have highlighted another source of vulnerability at the synaptic level: an incomplete overlap of P/Q-type subunit family members across the brain (Noebels, 2012). Different brain regions have distinct patterns of β1–4 subunit expression; thus, P/Q-type calcium channel composition, and thus function, is differentially affected. One of the earliest demonstrations of “subunit reshuffling” (a compensatory process) was in lethargic mice, which exhibit SWDs and ataxia (Burgess et al., 1999). In the cerebellum of lethargic mice, in the absence of the β4 subunit, there is increased steady-state association of the α1A subunit with β2 and β3 subunits instead; these complexes generate functional P/Q-type channel, with normal amplitude and voltage dependence of the P-type current in dissociated cerebellar Purkinje neurons (Burgess et al., 1999). Interestingly, thalamic nuclei appear to lack significant expression of β1 and β3 subunits, thus decreasing the chance that the α1A subunit can form a functional heteromer in the absence of a functional β4 subunit (Noebels, 2012).
Thus, selective deficits or compensations at specific synapses may arise from regulatory subunit “reshuffling.” In addition, cell-type-specific alternative splicing of various subunits may also explain resilience and vulnerability (Noebels, 2012).
Dravet syndrome is a childhood epileptic encephalopathy associated with severe nonconvulsive seizures and sudden unexplained death in epilepsy. It is caused by loss-of- function mutations in the SCN1A gene encoding the type I voltage-gated sodium channel (Claes et al., 2001; Dravet, 2011). In mouse models of Dravet syndrome, cortical inhibitory neurons exhibit reduced excitability, consistent with a loss of function in a channel critical for action potential generation and propagation (Han et al., 2020; Rubinstein et al., 2015). Specific inactivation of Scn1a in parvalbumin-positive interneurons in the hippocampus and cortex increases seizure susceptibility (Dutton et al., 2013). Interestingly, hypoexcitability of parvalbumin-positive cortical interneurons have been reported to normalize in adulthood in mouse models, suggesting that there may be other mechanisms which explain the ongoing, chronic epilepsy in Dravet syndrome (Favero et al., 2018). Recent work by our group has uncovered a surprising role for thalamic bursting in Dravet syndrome.
Given that Scn1a is highly expressed in parvalbumin-positive nRT cells, our group investigated thalamic firing in the Scn1a-deficient mouse model of Dravet syndrome (Ritter-Makinson et al., 2019). We found that GABAergic nRT neurons—but not glutamatergic TC neurons—exhibited heightened intrinsic excitability, most notably in the form of augmented rebound burst firing. Interestingly, enhanced nRT rebound burst firing was due to a compensatory reduction in the calcium-activated small potassium current (ISK) density rather than an enhancement of the low-threshold T-type calcium current. While nRT neurons also exhibited changes expected from Scn1a loss-of-function such as depolarized action potential threshold and increased action potential width, nRT neurons fired more action potentials in response to depolarization. This finding was in contrast to the decreased excitability of cortical inhibitory neurons described in previous studies. Moreover, reducing SK conductance, but not the sodium conductance, in a biophysical model was sufficient to generate enhanced rebound bursts as observed in nRT neurons. As a consequence of SK-deficit mediated augmentation of nRT burst firing, mice exhibited thalamic microcircuit hyperexcitability, an ex vivo hallmark of seizures involving the thalamus (reviewed by Cho et al., 2017). Finally, treatment with an SK channel agonist (1-ethyl-2-benzimidazolinone, or EBIO) was sufficient to reduce the frequency of non-convulsive seizures in Dravet syndrome mice. Altogether, we propose that Scn1a loss-of-function results in a compensatory reduction of ISK in nRT neurons, which contributes to enhanced input resistance and intrinsic excitability, and increased rebound burst firing. Augmented burst firing in nRT neurons recruits stronger oscillatory bursting in the intra-thalamic circuit, which maintains nonconvulsive seizures (Ritter-Makinson et al., 2019).
This study demonstrated a novel role for the thalamus in a type of nonconvulsive seizures observed in human patients. It also points to a novel role for SK2 channels (a normal component of nRT rhythmicity; Cueni et al., 2008)) and raises questions about how this compensatory mechanism arises from a Scn1a loss of function mutation. Importantly, by tapping into the rhythmogenic machinery of thalamic neurons, this study identified the nRT as a potential novel therapeutic target to treat nonconvulsive seizures in Dravet syndrome.
While the thalamus has been well-studied in the context of genetic epilepsies, there is also an increasing appreciation of thalamic involvement in acquired epilepsies such as those that develop after acute insults and in mesial temporal lobe epilepsies. Following cortical injuries such as traumatic brain injury and stroke, the thalamus sustains robust and chronic secondary damage including neurodegeneration and neuroinflammation in humans (Pappata et al., 2000; Scott et al., 2015; Grossman and Inglese, 2016) and in rodent models (Paz et al., 2010; Hazra et al., 2014; Cao et al., 2020; Holden et al., 2021; Necula et al., 2021). Thalamic damage has gained increasing recognition as a prognostic biomarker of posttraumatic epileptogenesis (PTE) in rodent models (Paz et al., 2010; Immonen et al., 2013; Pitkänen et al., 2020; Manninen et al., 2021), and it has also been suggested as a biomarker of cognitive impairment after TBI in patients (Ramlackhansingh et al., 2011; Grossman et al., 2012). Despite being remote from the initial epileptogenic insult, the thalamus can modulate seizure activity in rodent models of PTE (Paz et al., 2013; Paz and Huguenard, 2015), and it is the only FDA-approved brain region for deep brain stimulation in patients with refractory temporal lobe epilepsy (Fasano et al., 2021; Fisher and Velasco, 2014).
Neuroinflammation, commonly implicated in epileptogenesis, has been notoriously challenging to dissect and target effectively (reviewed by Aronica et al., 2017; Klein et al., 2018; Patel et al., 2019; Vezzani, Balosso, and Ravizza, 2019). The extent to which neuroinflammation—specifically in the thalamus—can drive epileptogenic changes following injuries, and whether it can be a disease-modifying target in the context of PTE, is an active field of investigation.
Our group recently conducted a systematic study of secondary thalamic inflammation across four distinct thalamocortical circuits in two rodent models of cortical lesions and found that secondary thalamic neuroinflammation mirrors functional connectivity of thalamocortical circuits (Necula et al., 2021). Such investigations, coupled with open access to large-scale anatomical and functional connectivity atlases—for example, the Allen Mouse Brain Connectivity Atlas (Harris et al., 2019), the Janelia MouseLight project (Winnubst et al., 2019), and the Allen Brain Observatory (Siegle et al., 2021)—may help start to elucidate potential “rules” by which secondary neuroinflammation evolves after cortical injuries.
Neuroinflammation following stroke or TBI colocalizes with robust changes in cellular and circuit properties of the thalamus. In rodent models, cortical stroke and TBI both lead to thalamic circuit hyperexcitability and thalamocortical epileptic discharges (Paz et al., 2010, 2013; Holden et al., 2021). Our group recently identified the C1q complement pathway as a mediator of the chronic emergence of epileptic activity and sleep disruption after a mild TBI in mice (Holden et al., 2021), highlighting the feasibility and promise of targeting thalamic neuroinflammation to prevent circuit dysfunction after injuries. Altogether, these findings support the role of thalamic neuroinflammation as a meaningful driver of chronic pathological changes implicated in epileptogenesis (Cho et al., 2022) and motivate further dissection of neuronal-glial interactions in the secondarily damaged thalamus.
In summary, we have reviewed evidence of pro-epileptic disruptions in thalamic and thalamocortical circuits, and propose that the thalamus as a key convergence point of dysfunction in genetic epilepsies. Understanding how the thalamus generates and modulates aberrant activity will aid in the identification of therapeutic targets and paradigms to treat epilepsies.
What aspects of thalamic organization and function predispose the thalamus to vulnerability? The thalamus generates vigorous rhythmic burst firing at the cellular and microcircuit levels, and the mechanics underlying such firing appears to be vulnerable to genetic perturbations. However, whether that means they can just as easily be “corrected,” without equally debilitating off-target effects, remains to be seen.
Cell-intrinsic vulnerabilities: We reviewed diverse ways in which T-type calcium channels, as well as R-type and P/Q-type channels, have been linked to absence seizures, including mutations which lead to direct or compensatory increases in the magnitude of IT, or to changes in channel subunit expression and composition. We particularly highlighted the abundant evidence that burst firing properties of thalamic neurons appear to be a common node of dysfunction across genetic models of epilepsies. One novel and notable example is the identification of a compensatory reduction of the SK current and subsequent increase of nRT burst firing in a Scn1a loss-of-function mouse model of Dravet syndrome (Ritter-Makinson et al., 2019). The finding that pharmacological augmentation of the SK current reduced the severity of nonconvulsive seizures suggests that restoring burst firing, even when it may not be the sole or primary electrophysiological disruption, can be potentially beneficial in modulating seizures.
Micro-circuit vulnerabilities: We also reviewed the diverse ways in which disruptions to feedforward inhibition and counter-inhibition increase the propensity of thalamic circuits to become hypersynchronous and hyperexcitable. Further investigations of what mediates selective vulnerability of specific synapses, and whether restoration of unperturbed microcircuit functions can compensate for pro-epileptic changes, will undoubtedly be fruitful in identifying therapeutic strategies to prevent and treat seizures.
Are there similar vulnerabilities in the thalamocortical circuit across genetic and acquired epilepsies and what are they? Thalamic damage after cortical injuries has recently emerged as a biomarker of PTE. Due to its broad connectivity, focal and even remote damage to the thalamus is well-positioned to mediate disruptions in “bystander” circuits which are not immediately involved in seizure dynamics. Thus, recent advances in our understanding of thalamic organization and function in various psychiatric and cognitive conditions (e.g., Krol et al., 2018)—which can often be comorbidities of seizures—may be harnessed to understand and modulate the potentially far-reaching role (beyond seizure activity) of the thalamus in acquired epilepsies. Ultimately, if the thalamus is indeed an Achilles’ heel—a critical node of dysfunction—in the context of epilepsy, then understanding how diverse genetic and acquired etiologies converge upon thalamic hyperexcitability will pinpoint elements that contribute to its vulnerability as well as those that contribute to its resilience.
J. T. P. is supported by NIH/NINDS grant R01NS096369, the Department of Defense (EP150038), the Gladstone Institutes, and the Kavli Institute for Fundamental Neuroscience. F. S. C. is supported by NINDS F31 NS111819-01A1. The authors would like to thank Drs. Francoise Chanut and Kathryn Claiborn for editorial assistance. F. S. C. would also like to thank Dr. Tilo Gschwind for constructive feedback and insightful discussions.
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...
