Déjà Vu: Same Pattern of Neuron Activation From Seizure to Seizure, Only the Timing Changes

Commentary

Reliable and Elastic Propagation of Cortical Seizures In Vivo.

Wenzel M, Hamm JP, Peterka DS, Yuste R. 2017;19:2681–2693.

Mapping the fine-scale neural activity that underlies epilepsy is key to identifying potential control targets of this frequently intractable disease. Yet, the detailed in vivo dynamics of seizure progression in cortical microcircuits remain poorly understood. We combine fast (30-Hz) two-photon calcium imaging with local field potential (LFP) recordings to map, cell by cell, the spread of locally induced (4-AP or picrotoxin) seizures in anesthetized and awake mice. Using single-layer and microprism-assisted multilayer imaging in different cortical areas, we uncover reliable recruitment of local neural populations within and across cortical layers, and we find layer-specific temporal delays, suggesting an initial supra-granular invasion followed by deep-layer recruitment during lateral seizure spread. Intriguingly, despite consistent progression pathways, successive seizures show pronounced temporal variability that critically depends on GABAergic inhibition. We propose an epilepsy circuit model resembling an elastic meshwork, wherein ictal progression faithfully follows preexistent pathways but varies flexibly in time, depending on the local inhibitory restraint.

Most recordings of seizures, both in human patients and in animal models of epilepsy, employ techniques that record neuronal activity at the macroscale. EEG, electrocorticography (ECog), and local field potential (LFP) recording techniques have varying levels of invasiveness and spatial resolution, but share the fundamental characteristic of recording voltages associated with the summed activity of many neurons. Importantly, these techniques offer ‘brain-wide’ views that have shaped our thinking about both seizure generation and propagation. In recent years, however, there has been a push to ‘zoom in’—to understand seizures at the microscale by observing the activity of individual neurons. Studying how seizures emerge in, and spread through, microcircuits may reveal specific neuron types that promote or restrain pathological activity. Thus far, the most common approach to recording individual neurons during seizures in the intact brain has been single-unit recording, in which action potentials from individual neurons are isolated using clustering algorithms. Although single-unit recording has precise temporal resolution, the exact identity of neurons remains unknown, and in dense networks the sampling of neurons is sparse. Furthermore, action potentials can change shape during seizures and electrodes can drift, making tracking the same neurons over long durations difficult.

In the last few decades, several groups have developed in vivo two-photon imaging of calcium activity to monitor large neuron ensembles in intact neural networks. Two-photon excitation of fluorescent indicators allows for unprecedented spatial resolution of neural activity. In their recent publication, Wenzel et al., were among the first to use this technique to record micro and macro scale dynamics of seizure propagation in vivo in both anesthetized and awake mice. Their study is exemplary in design, matching experimental questions to experimental techniques. In particular, they were interested in how seizures propagate through networks of excitatory neurons in the cortex and whether patterns of activation during seizure propagation were stereotyped or unique to each seizure. To induce seizures, the authors injected a small volume of either 4-AP or picrotoxin to cortical layer V, 1.5 to 3 mm from the area of cortex that they imaged. By inducing seizures pharmacologically, they were able to control when the seizures occurred, and also to image several seizures in the same neuron set.

Wenzel et al. find that as seizures spread laterally across layer II/III of cortex (or layer V), individual neurons are recruited in a pattern that resembles a continuous wave (i.e., neighboring neurons are recruited consecutively). This demonstrates the power of their technique because sparser sampling would not have been able to distinguish between continuous and patchy recruitment. This continuous wave likely reflects local recurrence in the cortex. Similarly, another recent in vivo two-photon study of seizure propagation also sees a continuous pattern of propagation and gives evidence that it occurs along defined anatomic connections in area V1 (1). Further questions thus arise. Would it be possible to stop the seizure from spreading if the recruitment of one cortical column was experimentally blocked? Or would the seizure find another way around—for example, through subcortical nuclei?

The authors also found that from seizure to seizure, the same neurons were recruited in roughly the same temporal order. This finding again supports the idea that the seizure is spreading via anatomic connections, travelling perhaps by way of previously strengthened connections. It corroborates findings in one human study, in which reproducible neuron activation during seizures was observed (2), but contrasts with another human study, in which neuron activation during seizures was not consistent between seizures (3). It is not immediately clear why different levels of stereotypy are observed, but this may point to differences between unique networks, disease states, or types of epilepsy. Indeed, as candidly noted by the authors, in their experiments the brain is not in a chronic disease state, which may account for differences seen in human studies. This need not detract from Wenzel and colleague's findings, however, for the following reasons. First, some argue that fundamental aspects of seizure generation may be invariant across the diverse conditions from which they emerge (4). Second, any epilepsy study should be considered in its most clinically relevant context. In this case, for example, the seizures in a patient experiencing uncontrolled status epilepticus for the first time may have dynamics similar to those studied here. Nevertheless, in chronic epilepsy, in which the brain undergoes cell death and rewiring, seizure propagation likely looks different from that in a healthy brain (5). Still, because seizures can arise in both cases, finding differences between chronically epileptic and naïve brains in regard to seizure spread at the microscale could be beneficial for therapeutic intervention.

Finally, and perhaps most surprising, Wenzel et al. report that while the order of neural recruitment remains constant from seizure to seizure, the timing of the evolution of the spread does not. They refer to this variable timing as the ‘elastic’ property of propagation. This elasticity appeared to depend on GABAergic interneurons because when GABA receptors in the imaged region were blocked, seizure propagation between neurons became faster and no longer variable. This finding is consistent with in vitro work showing that feedforward inhibition regulates the propagation speed of seizures (6). It is also consistent with the inhibitory restraint model of seizure propagation (7). One limitation of the technique employed by Wenzel et al. is that it can only resolve the moment a neuron is first recruited by the seizure, but not the later dynamics. It is thus difficult to determine what occurs when the activity eventually spreads to neighboring neurons. In some cases, several hundred milliseconds passed before the seizure spread to neighboring neurons. Another recent study, which employed single unit recordings in the cortex and hippocampus, gives insight to what might be happening during those hundreds of milliseconds. They found that small groups of neurons were activated during individual ictal spikes, and reactivated across subsequent ictal spikes (8). These recordings also showed fast-spiking interneurons to be strongly recruited in ictal sequences. We can therefore imagine that when a seizure front recruits a given group of neurons, that group gets sequentially recruited in space, and then feeding back onto itself thus repeats the cycle of activation. For each cycle, feedforward interneurons are also repeatedly activated. The interneurons would thus ‘protect’ the next group of neighboring neurons from being recruited until, they could not. The moment the breakdown in inhibition occurs, the seizure propagates to the next group of neurons. What exactly causes the breakdown in inhibition, and why the time-course of the breakdown differs from seizure to seizure (and even from group of neurons to group of neurons within one seizure) are open questions. Several groups propose that fast-spiking interneurons undergo depolarization block, and some propose that glial-neuron interactions are important for regulating when this happens (9). Wenzel and colleague's work will hopefully refine our thinking along these lines; variability in the timing of the inhibitory breakdown suggests that underlying mechanisms are dynamic in ways that are currently still a mystery.