Abstracts

Rationale: During an absence seizure, abnormal synchronization between cortex and thalamus temporarily generates a high-amplitude spike-wave EEG and LFP signature. It has been proposed that strong, synchronized output from thalamocortical (TC) neurons is required for full seizure expression. Multimodal TC cells whose axons innervate superficial layers may play a privileged role in generating EEG signature. However, single cell recordings suggest that overall, thalamic output may be sparse. To directly observe the extent of multimodal TC activation during absence seizures, we recorded calcium activity from TC axons projecting to superficial cortex in awake, head-fixed epileptic mice. We compared seizure-related activity to that evoked by sensory input or running to identify seizure-specific activity patterns. Methods: To image thalamocortical axon activity during seizures, the calcium indicator GCaMP6s was expressed in somatosensory thalamus of a mouse model that has spontaneous absence seizures due to a sodium channel mutation (Scn8a-med). A 3mm cranial window was implanted over somatosensory barrel cortex, 4 silver wires were implanted under the skull to record electrocorticogram (ECoG), and a head post was secured to the skull. Animals were adapted to head fixation on a treadmill over 5 sessions before recording. TC axons <50µm from the brain surface were imaged at 8.4Hz using a custom 2-photon microscope. ECoG and animal running were recorded simultaneously with OpenEphys. In some recordings, air puffs were delivered to the contralateral whiskers. Results: Up to 50% of recorded TC axons exhibited seizure-related activity. Most imaging fields (11/12) contained a mix of axon types with and without seizure-related activity. The most common activity pattern was a gradual increase and decrease in activity corresponding to the seizure envelope (3/3 animals, at least on axon in 10/12 fields). These seizure-activated axons reached peak GCaMP dF/F at a reproducible time across seizures. Seizure-related activation was preceded in many axons by a gradual decrease in fluorescence over 5 seconds before seizure onset. We also recorded axons with the opposite seizure-related response. These axons were active in the pre-seizure period but turned off just before or at seizure onset (2 animals, 3 fields). Seizure-activated axons also encoded sensory signals. In recording sessions with whisker stimulation (3 animals, 4 fields), a subset of whisker-activated axons were also seizure-activated. Whisker stimulation drove a sharp increase in calcium flux, unlike the gradual rise and fall recorded during seizures. Seizure-activated axons were also active during activity transitions, displaying a strong dF/F volley either just before running onset or offset, depending on the axon. Conclusions: For the first time, we directly show that up to 50% of superficial TC axons are reliably active during absence seizures. These fibers that are well-positioned to be recorded by EEG, supporting the idea that the EEG readout of an absence seizure is driven in substantial part by TC output. We find that seizure-activated axons are multimodal, responding to sensory inputs and signaling activity transitions, but only half of sensory- or running-activated TC axons participate in seizures. These results suggest a specialized absence seizure circuit within the broader thalamocortical network that could be selectively targeted for seizure intervention. We also define two types of pre-seizure activity in TC axons which could be used to define seizure predictors for real-time interruption.  Funding: NIH T32NS007280-32NIH R01 NS034774-21A1