Abstract
Liou JY, Ma H, Wenzel M, Zhao M, Baird-Daniel E, Smith EH, Daniel A,
Emerson R, Yuste R, Schwartz TH, Schevon CA Brain
2018;141:2083–2097. Focal seizure propagation is classically thought to be spatially
contiguous. However, distribution of seizures through a large-scale
epileptic network has been theorized. Here, we used a multielectrode
array, wide field calcium imaging, and two-photon calcium imaging to
study focal seizure propagation pathways in an acute rodent neocortical
4-aminopyridine model. Although ictal neuronal bursts did not propagate
beyond a 2–3-mm region, they were associated with hemisphere-wide
field potential fluctuations and parvalbumin-positive interneuron
activity outside the seizure focus. While bicuculline surface
application enhanced contiguous seizure propagation, focal bicuculline
microinjection at sites distant to the 4-aminopyridine focus resulted in
epileptic network formation with maximal activity at the two foci. Our
study suggests that both classical and epileptic network propagation can
arise from localized inhibition defects, and that the network appearance
can arise in the context of normal brain structure without requirement
for pathological connectivity changes between sites.
Commentary
Understanding how focal seizures spread throughout the brain is a longstanding question in epilepsy research that has both clinical and basic science implications. Over 50 years ago, it was reported that cortical seizure foci were characterized by large depolarization shifts at the area of insult, but were surrounded by strongly inhibited areas, which were hypothesized to limit the spread of seizure activity (1, 2), and later that blocking inhibition greatly enhanced the speed and extent of epileptiform activity spread in neocortical brain slices (3). More recent work has refined these ideas, showing that principle neurons demonstrate intense, synchronized burst firing in the ‘ictal core’, but that surrounding regions, termed the “penumbra,” are characterized by relatively sparse, heterogeneous firing. This sparse firing is due to strong feedforward inhibitory synaptic activity onto principal neurons preventing their firing and recruitment into the ictal core (4, 5). Seizures are proposed to spread if and when this inhibition in the penumbra is overwhelmed. Interestingly, epileptiform local field potentials are observed in both of these areas, meaning that EEG recordings do not easily distinguish between the two types of fundamentally different underlying neuronal activity. Single-unit recordings in animals and humans, and patch clamp recordings in vitro, have provided evidence for the difference.
In a recent paper, Liou et al. combined multielectrode array recordings with both widefield and two-photon Ca2+ imaging of cortical activity through acute cranial windows in vivo to investigate how this inhibitory synaptic activity prevents the spread of epileptiform activity. To model focal seizures, they injected the potassium channel blocker 4-aminopyridine (4-AP) in to the cerebral cortex of anesthetized rats and mice. First, they demonstrated that the focal nature of the epileptiform activity induced by 4-AP is apparent in multiunit activity (MUA) and in the local field potential, but that the local field potential is altered in a larger spatial area. In areas proximal to the 4-AP injection site there was intense MUA, but these signals declined sharply at distances farther from the injection site. Increased local field potenital line length, however, was detectable beyond the region defined by MUA firing greater than baseline. In agreement with the MUA data, widefield Ca2+ imaging showed a large, highly correlated signal near the site of injection that declined sharply beyond 2 mm. This pattern was observed with 4-AP injection into both somatosensory and visual cortical areas.
The inhibitory restraint model of seizure propagation predicts that inhibitory neurons in the penumbra should be activated by excitatory projections from the ictal core. To test this, the authors performed two-photon imaging of parvalbumin-expressing (PV) neurons specifically labeled with the calcium indicator GCaMP6 in mice. They injected 4-AP and recorded the LFP at a posterior site while imaging the activity of the PV neurons and recording the LFP at a site approximately 4 mm away, a distance that should be well away from the ictal core and not display intense activation of principal neurons. They found that the PV neurons were reliably activated during epileptiform events, and that the short temporal delay at which they were activated relative to the LFP change was consistent with synaptic feedforward recruitment.
If the activation of these inhibitory neurons is providing synaptic inhibition that is restraining the spread of the seizure from the focus, then compromising this inhibition should alter the spread of the seizure. To test this, the authors first bathed the entire cranial window in bicuculline, a GABA-A receptor antagonist, and examined the spread of ictal activity from the 4-AP injection site. They found that this resulted in ictal activity propagating outward contiguously from the 4-AP site, eventually invading the entire sampled area. In support of the idea that the ictal activity represented spread from the 4-AP injection site, and was not directly due to the GABA antagonism, they noted that the character of the LFP was more similar to the ictal events generated by 4-AP alone, as opposed to events triggered by focal bicuculline, which produces high amplitude paroxysmal epileptiform discharges.
Second, instead of bathing the entire area in bicuculline, they injected it focally in an area 4 to 5 mm away from the 4-AP focus in the same cortical hemisphere. This resulted in ictal activity at both sites. However, instead of observing intense MUA over the whole imaging area with short latency, the strongest MUA was seen at the two microinjected sites with nearly simultaneous onset, while the in between areas showed relatively slower activation that proceeded from the bicuculline-injected area back toward the 4-AP injected area. This indicates that the seizure focus “jumped” from the 4-AP focus to the disinhibited area, providing evidence that synaptic inhibition at distant sites protects these areas from being recruited into the ictal core. When they injected the bicuculline in a homotypic region in the contralateral hemisphere, however, the same phenomenon was not observed, indicating that cross-callosal projections are not strong enough to propagate the seizure in their model, even when inhibition is compromised.
These experiments provide evidence that a seizure focus engages inhibitory neurons at a considerable distance, and that the activity of these neurons provides synaptic inhibition that protects these brain areas from being recruited into the seizure, even though the LFP may not distinguish between these two states. When this synaptic inhibition is compromised, however, epileptiform activity in these distant regions rapidly appears and shows very similar dynamics to the primary focus. Synaptic inhibition is known to be compromised in many epilepsies, either by genetic or epileptogenic mechanisms (6, 7). It is unknown how well bicuculline application used here mimics compromised inhibition in an epileptic brain, or other mechanisms that may make brain areas vulnerable to recruitment into a seizure. More elegant methods for focally impairing inhibition, such as optogenetics, chemogenetics, or genetic reduction of inhibitory synapses, would be a nice extension to this work. Also, how analogous the underlying networks in these nonepileptic brains are to an epileptic brain is also unknown, meaning that repeating these experiments in an epileptic animal may show different results. Nevertheless, the evidence here suggests that in epileptic brains, the appearance of ictal activity in an originating seizure focus and other foci in distant brain areas may appear simultaneously upon seizure onset. If this is the case, the location and extent of the originating focus would be obscured, which would in turn make it difficult to pinpoint the critical seizure focus for resection surgery.