How do clock neurons in the brain modulate daily seizure timing?

Flies show a daytime peak in spontaneous seizures

The study's entry point is a measurement problem: to ask whether seizures follow a daily rhythm, the authors needed to catch them as they happened, without a stimulus. They placed individual flies in wells containing picrotoxin, a GABAA-receptor antagonist, and recorded video continuously for 96 hours. A tracking pipeline converted each fly's position over time into movement measures, and a seizure was scored when hyperkinetic events clustered densely enough to match previously defined thresholds. The authors call these "spontaneous" seizures because no acute trigger is delivered.

Wild-type flies fed picrotoxin seized preferentially during the daytime, and a permutation-based Rayleigh test rejected a uniform distribution across the 24-hour day. The comparison group matters: timeless (tim01) mutants, which lack a functional molecular clock, still showed normal light-dark activity patterns, yet their seizures were spread randomly across the day. Total seizure number and seizure duration did not differ between the genotypes; what changed was when seizures occurred.

Two controls narrow the interpretation. Picrotoxin-treated wild-type and tim01 flies slept at similar levels, and shifting picrotoxin exposure by 12 hours to the nighttime still produced daytime-predominant seizures. The authors therefore argue the timing effect is not a sleep artifact and is not simply driven by when the drug is present. One exception is worth keeping: lethal seizures in tim01 mutants still clustered in the early evening, as in wild-type flies, which the authors read as evidence that this particular rhythm may not require a functional molecular clock.

Seizure timing and seizure severity are separate readouts

The paper moves between two seizure models, and they answer different questions. The picrotoxin video assay measures when spontaneous seizures occur and how many there are. The second model uses tko25t "bang-sensitive" mutants, in which a mechanical stimulus such as vortexing reliably triggers a stereotyped seizure with an atonic phase, a tonic-clonic convulsive phase, and a post-ictal recovery phase. Because each phase can be scored by eye from video, this model measures severity and duration rather than spontaneous timing.

In tko25t flies stimulated at six time points across the day, total seizure duration oscillated, peaking around ZT0, the light-on transition. That oscillation was carried mainly by the tonic-clonic phase, with a weaker cycle in the atonic phase. The probability that a stimulus would induce a seizure did not vary significantly across the day. Crossing tko25t to tim01 removed the 24-hour rhythm in seizure duration, with a possible ultradian rhythm appearing instead.

The authors also tested two clinically relevant knock-in models carrying mutations in paralytic, the fly ortholog of human SCN1A, which is implicated in Genetic Epilepsy with Febrile Seizures Plus and Dravet Syndrome. At 30 °C these flies showed a diurnal seizure rhythm peaking at the day-to-night transition. Across the pharmacological and genetic models, then, rhythms in seizure severity were widespread, but the phase of the peak was etiology-specific rather than fixed to one time of day.

Morning and evening cells set the daily seizure window

To find which cells drive the rhythm, the authors screened clock-neuron subsets. Thermogenetic activation with TrpA1, together with cell-specific clock perturbations, pointed to two clusters: the ventrolateral neurons, which include the small morning cells, and the dorsolateral neurons, the evening cells. In the fly brain, roughly 240 clock neurons are divided between hemispheres and organized into lateral and dorsal clusters; the small ventrolateral neurons increase activity in the early morning and are orthologous to the mammalian suprachiasmatic nucleus, while the dorsolateral neurons peak in the early evening.

Optogenetic manipulation made the division of labor visible. Activating morning cells throughout the day with csChrimson increased the number of daytime seizures, and it also increased seizure duration. Activating evening cells throughout the day qualitatively suppressed daytime seizures. Activating either population at times when those cells are not normally active suppressed daytime seizures instead of promoting them. The authors interpret this as a switch: the same cells can push seizure risk up or down depending on when they are engaged.

The screen also separates the cells from the clock itself. tim01 mutants lack a functional molecular clock globally, but the cell-specific perturbations show that particular clock-neuron populations, not the clock in every cell, carry the timing signal for seizures.

PDF and sNPF raise daytime risk; glycine lowers it at night

The molecular side of the switch involves three signals from morning cells. Disrupting pigment-dispersing factor signaling, either in pdfrhan5304 mutants or by pdf RNAi in morning cells, changed seizure severity. Disrupting sNPF with RNAi did the same. Both manipulations altered the severity readouts, consistent with the peptides promoting seizures during the day.

Glycine behaved differently. Knocking down Shmt, an enzyme needed for glycine synthesis, in morning cells did not change severity but did change seizure timing, producing a significant shift relative to genetic controls by Watson's two-sample test of homogeneity. The authors' model is that morning cells release glycine at night, likely acting on glycine receptors on evening cells, to reduce seizure risk. The word "likely" is doing real work here: the evening-cell receptor step is proposed rather than directly demonstrated in the excerpted evidence.

This is where the paper's central claim becomes concrete. The same morning cells use peptidergic signaling to raise risk in the day and a different transmitter, glycine, to lower it at night. The switch is a change in signaling mode, not merely a change in firing rate.

Morning-cell activity raises global brain activity only early in the day

If morning cells raise daytime seizure risk, one plausible route is broad excitation. The authors tested this with a calcium reporter, CaLexA, driven pan-neuronally and measured in dissected brains at six time points. The signal cycled through the day, peaking during the daytime, which establishes that global brain activity itself has a daily rhythm.

A second experiment asked whether morning cells can drive that activity. Flies expressed csChrimson in morning cells and GCaMP7b pan-neuronally, and were imaged in a fly-on-the-ball preparation at four times of day with 100 ms whole-field red-light stimulation. The calcium response increased after morning-cell stimulation between ZT0 and ZT4, but not at the other time points tested.

The timing restriction is the important part. Morning cells are not simply excitatory whenever they fire; their ability to raise brain-wide activity is itself gated to the early day. That fits the behavioral result that activating morning cells outside their normal active window suppressed rather than promoted seizures.

Giant fiber neurons carry the clock signal to motor output

The paper places the clock neurons upstream of the giant fiber neurons, which mediate the tonic-clonic movements that define these seizures. Activating giant fiber neurons around the clock with csChrimson increased seizure duration without a significant change in seizure timing or lethality, which the authors read as evidence that these neurons sit downstream of the clock input.

A narrower test was more striking. Inhibiting giant fiber neurons only from ZT0 to ZT4, the window when morning cells are active and daytime seizures worsen, reduced the daily number of seizures by about 50 percent. Survival improved from 25 percent to 75 percent at 96 hours. The authors conclude that descending motor neurons mediating seizures are downstream of clock inputs, and that inhibiting the downstream pathway during a narrow time window can reduce seizures and improve survival.

This result also reframes what the clock neurons are doing. They are not necessarily generating seizures themselves; they appear to modulate the gain of a motor circuit that produces the convulsive behavior.

What the fly switch does and does not establish for epilepsy

The authors are explicit about the translation gap. The fly brain contains roughly 140,000 neurons and lacks a cerebral cortex, the primary seat of seizures in humans, and the specific brain regions driving fly seizures beyond the giant fiber system remain unmapped. Fly seizures are defined behaviorally because electrical access in freely moving flies is technically challenging, so seizures that do not produce sustained tonic-clonic movements are likely missed. The conserved pieces the authors point to are molecular: voltage-gated ion channels, synaptic release machinery, and the molecular clock.

Within those limits, the paper's contribution is a mechanism for the phase of daily seizure risk rather than a claim about seizure frequency in people. The authors connect their results to the Molecular Oscillations and Rhythmicity of Epilepsy hypothesis, which proposes that diurnal molecular rhythms place the brain at oscillating levels of seizure risk, and they argue that accounting for clock-neuron contributions should improve prediction of when susceptibility is greatest.

The practical priority that follows is specific to this paper's variables. The phase of the seizure peak differed by etiology across the models studied, so a single "high-risk hour" is not the takeaway. The takeaway is that the signaling mode of morning cells, peptidergic by day and glycinergic at night, sets the direction of the effect, and that the downstream giant fiber window from ZT0 to ZT4 was where inhibition cut seizure counts and improved survival. For seizure forecasting, that argues for tracking clock-neuron state and its downstream motor gate, not just the time on the clock.

Sources used in this answer