The research team, based at Boston Children’s Hospital and Harvard Medical School, obtained acute brain slices from cortical tissue surgically resected from patients with pharmaco-resistant epilepsy, most of them children and young adults between one and twenty-one years of age. Working within hours of resection, the researchers kept the tissue alive in oxygenated artificial cerebrospinal fluid and delivered trains of high-frequency, high-intensity electrical pulses through a bipolar electrode placed roughly 250 micrometers from the cells under study. Meanwhile, patch-clamp recordings captured the electrical behavior of three distinct neuronal populations in layer 2/3 of the neocortex: pyramidal neurons, the principal excitatory cells of the cortex; fast-spiking interneurons, the inhibitory workhorses of cortical circuits; and non-fast-spiking interneurons, a more heterogeneous collection of inhibitory cells.

The results were strikingly asymmetric. When stimulation was delivered at 20 hertz, pyramidal neurons responded reliably, firing in step with the pulses. But as the frequency climbed to 50 and then 100 hertz, the pyramidal neurons hit a wall. Their firing rates collapsed to levels no higher than those seen at 20 hertz, because the stimulation outpaced their intrinsic maximum firing rates. The fast-spiking interneurons told a completely different story. These cells fired continuously and with remarkable fidelity throughout the entire 400-millisecond stimulation train, matching or even exceeding the stimulation frequency itself. The non-fast-spiking interneurons, despite being inhibitory like their fast-spiking cousins, were suppressed in much the same way as the pyramidal neurons.

This selectivity is not a trivial observation. Fast-spiking interneurons, which in molecular terms overlap heavily with parvalbumin-expressing interneurons, are the primary source of fast inhibitory feedback onto pyramidal neurons in the cortex. Their dysfunction has been implicated in epilepsy, schizophrenia, and autism spectrum disorders, and their loss or silencing is thought to be a central contributor to the excitation-inhibition imbalance that allows seizures to ignite and propagate. The finding that a blunt, non-selective tool like extracellular electrical stimulation can preferentially recruit precisely these cells suggests that the therapeutic benefit of neurostimulation may rest on a surprisingly elegant cellular foundation.

To test whether this selectivity is a general property of cortical circuitry rather than a quirk of human epileptic tissue, the team performed parallel experiments in mice, using viral labeling to identify parvalbumin-expressing and somatostatin-expressing interneurons in the prefrontal cortex. The pattern held. Mouse parvalbumin interneurons, which are electrophysiologically fast-spiking, fired with high fidelity throughout high-frequency stimulation trains, while pyramidal neurons and somatostatin interneurons were progressively suppressed. Mouse parvalbumin cells even fired doublets or more in response to individual pulses, making the contrast with pyramidal neurons even sharper than in the human tissue. This cross-species consistency strengthens the argument that the selective recruitment of fast-spiking interneurons reflects fundamental circuit architecture rather than pathology-specific adaptation.

Beyond the acute effects on firing, the researchers uncovered a second, longer-lasting mechanism. After the stimulation trains ended, they measured the strength of individual excitatory and inhibitory synapses onto both pyramidal neurons and fast-spiking interneurons using minimal stimulation protocols. At pyramidal neurons, both excitatory and inhibitory postsynaptic currents were depressed to roughly 70 percent of baseline, leaving the net balance between excitation and inhibition essentially unchanged. At fast-spiking interneurons, however, the picture was radically different. Excitatory inputs were untouched, while inhibitory inputs were depressed by half. The result was an approximately twofold increase in the ratio of excitatory to inhibitory synaptic conductance at these inhibitory cells, effectively amplifying their own excitatory drive and priming them to fire more readily against future network hyperactivity.

This rebalancing of synaptic weights represents a form of homeostatic plasticity that operates specifically at inhibitory interneurons. Long-term potentiation and depression have been studied extensively at excitatory synapses and at inhibitory synapses onto pyramidal neurons, but inhibitory plasticity at postsynaptic GABAergic interneurons has only recently begun to be explored. The observation that high-frequency stimulation, traditionally associated with potentiation rather than depression, can induce selective inhibitory long-term depression at fast-spiking interneurons adds a new dimension to our understanding of how stimulation protocols reshape cortical circuits over time. Notably, these synaptic changes occurred without meaningful alterations in intrinsic neuronal properties: firing rates outside the stimulation epoch, rheobase, action potential threshold, and resting membrane potential were all essentially unchanged, confirming that the lasting effects of stimulation are mediated primarily through synaptic mechanisms.

The implications for clinical practice are significant. Current neurostimulation protocols for epilepsy vary widely in their parameters, constrained by charge density limits at the electrode-tissue interface to prevent tissue damage from electrolysis, pH shifts, or electrode corrosion. Without a mechanistic framework linking specific stimulation parameters to specific circuit-level outcomes, optimization has been largely empirical. This study provides exactly that framework. By quantifying how different frequencies selectively modulate the firing rates and synaptic weights of defined neuronal populations, the researchers have laid the groundwork for predictive models of electrical neuromodulation. If the general effectiveness of a stimulation protocol can be characterized by its impact on fast-spiking interneuron recruitment, clinicians could in principle design stimulation parameters that maximize this recruitment while minimizing off-target effects.

Important caveats remain. The tissue used in this study came from patients with epilepsy, a condition known to produce synaptic and circuit-level reorganization, including alterations in inhibitory function. The researchers acknowledge that generalizing their findings to non-pathological cortex requires caution, although the consistency of the mouse results argues against a purely pathology-driven effect. Additionally, most of the human tissue was pediatric, and neuronal properties change substantially across developmental stages. Sample sizes were not large enough to quantify age-dependent effects within each cell type. The stimulation used in the slices, while designed to approximate clinical parameters, was delivered at a saturating intensity to maximally activate nearby afferents, and the relationship between this ex vivo condition and the distributed fields generated by implanted electrodes in living brain remains to be fully established.

Nevertheless, the study represents a rare and valuable window into the cellular mechanics of a therapy that has already changed lives. The idea that the human cortex is wired such that a simple, non-selective electrical pulse train can selectively amplify its own inhibitory control system is both unexpected and encouraging. As the researchers note, the human neocortex contains a higher proportion of inhibitory interneurons than rodent cortex, with particularly strong excitatory synapses onto basket cells and chandelier cells, the two major subclasses of parvalbumin interneurons. These features appear to favor the efficient recruitment of fast-spiking cells during synchronous high-frequency activation, exactly the condition that stimulation creates. Future work aimed at systematically exploring the stimulation parameter space, combined with direct measurements of synaptic dynamics between identified neurons, could transform neurostimulation from an art of trial and adjustment into a precision therapy guided by a deep understanding of the circuits it modulates.

Subject of Research: Selective activation of human fast-spiking interneurons by high-frequency electrical stimulation and its role in seizure suppression

Article Title: High-frequency, high-intensity electrical stimulation selectively activates human fast-spiking interneurons

Article References: Yoon, J., Silva, R., Stone, S., Lidov, H., Osterweil, E., & Ferguson, B. R. (2026). High-frequency, high-intensity electrical stimulation selectively activates human fast-spiking interneurons. Frontiers in Synaptic Neuroscience, 18, Article 1843277. https://doi.org/10.3389/fnsyn.2026.1843277

Image Credits: AI Generated

DOI: 10.3389/fnsyn.2026.1843277

Keywords: epilepsy, brain stimulation, fast-spiking interneurons, neurostimulation, excitation-inhibition balance, synaptic plasticity, human brain slices, parvalbumin interneurons, High-frequency, high-intensity, electrical, stimulation