More PV spikes, weaker inhibition
The study asks what happens when DNA methyltransferase 1 (DNMT1), an enzyme that maintains DNA methylation patterns, is removed from cortical PV interneurons. PV interneurons are fast-spiking inhibitory cells that gate excitatory input and help synchronize cortical rhythms. The authors used mice in which Dnmt1 was deleted conditionally and only in PV cells, then combined in vivo Neuropixels recordings with single-nucleus RNA-seq, snATAC-seq, histology and behavior.
The central result is a dissociation between firing and inhibition. In primary visual cortex (V1), spontaneous firing rates of identified fast-spiking PV interneurons were higher in knockout mice than in controls (54 control versus 37 knockout neurons; linear mixed-effects model, genotype p = 0.0174). Yet optogenetic activation of PV cells suppressed surrounding V1 neurons far less effectively in knockouts (332 control versus 238 knockout neurons; p = 5.66 × 10⁻⁸). More spikes did not mean more inhibitory control.
The authors interpret this as a homeostatic network adaptation: persistently elevated GABAergic drive down-weights inhibitory influence to stabilize excitation/inhibition balance. That interpretation is theirs, not a direct measurement of the adaptation process. What the recordings establish is the combination of increased PV firing and reduced inhibitory efficacy in the same animals.
Cortical responses lose timing and gamma synchrony
The weakened inhibition appeared across cortical areas and stimulus modalities. In V1, visually evoked activity in knockout mice was weaker, temporally delayed and less precise, with reduced stimulus specificity across layers; the inhibitory rebound normally associated with feedback inhibition was absent. Whisker stimulation in primary somatosensory cortex produced the same pattern: weaker, prolonged responses lacking the inhibitory rebound, and little suppression during PV optogenetic activation.
Visually evoked gamma oscillations, fast rhythms that depend on precise PV timing, were severely reduced in knockout mice. This matters because gamma synchrony is the network-level readout most tightly tied to PV interneuron output in the paper's framing. The reduction is a measured electrophysiological result; its link to mood symptoms is the authors' mechanistic argument, drawing on prior literature that the packet cites but does not reproduce.
Basic function was partly preserved. Knockout mice learned the Morris water maze and a visual evidence accumulation task as efficiently as controls, and optogenetic PV activation still impaired decision-making in both genotypes. The authors suggest compensatory mechanisms, possibly subcortical, can buffer impaired cortical precision. So the deficit is in timing and synchrony, not in every behavioral capacity tested.
Reduced activity without hypothalamic dysregulation
Knockout mice moved less in their home cages, built poorer nests, had deteriorated coat condition, ran less on wheels, showed diminished sexual motivation and took longer to approach food in the novelty-suppressed feeding test. The authors describe this as an apathy- and anhedonia-like phenotype. Motor performance on the rotarod and ladder rung tests was normal, and visual deficits did not explain the reduced movement.
Metabolic measurements argue against a primary hypothalamic cause. Food intake and circulating ghrelin, insulin and PYY were unchanged, whereas body weight and leptin were elevated, consistent with reduced energy expenditure rather than altered appetite signaling. PV cells in the arcuate nucleus showed no overlap with POMC and only partial overlap with AgRP populations. The authors therefore point to cortical rather than hypothalamic mechanisms for the dysphoric-like phenotype.
This behavioral evidence is correlative with the electrophysiology: the same animals show both. The paper's causal claim runs through the proposed cascade, not through a rescue experiment that restores behavior by correcting PNNs or gamma synchrony.
A PV-restricted deletion spreads to glia and matrix
Although Dnmt1 was deleted only in PV cells, single-nucleus RNA-seq revealed genotype-associated differences in the relative abundance of astrocytes and specific excitatory populations, most notably Tox+/Slc17a6+ corticospinal-like neurons, plus transcriptional changes in synaptic structure and transmission genes. Reduced GFAP signal in somatosensory cortex matched the lower astrocyte abundance. The authors flag the excitatory neuron abundance difference as needing validation and interpret the corticospinal-like enrichment as a possible activity-dependent transcriptional state change rather than a new subtype.
Cross-modal analysis linked chromatin and transcription selectively in layer 4 sensory neurons, the only cell type with FDR-significant directional concordance between knockout-open chromatin regions and upregulated RNA, and between knockout-closed regions and downregulated RNA. Layer 4 is densely innervated by PV interneurons, which the authors use to argue for a genuinely chromatin-anchored response in that population.
CellChat analysis of ligand-receptor co-expression predicted reorganized perisomatic signaling between PV interneurons and oligodendrocyte-lineage cells. Compartmental modeling of oligodendrocyte dynamics further predicted impaired formation of PNN-supporting perineuronal oligodendrocytes. These are model-based predictions, and the paper treats them as such; the histological finding below is the tissue-level check.
PNN integrity falls without oligodendrocyte loss
Perineuronal nets (PNNs) are extracellular matrix structures that ensheath PV interneurons, stabilize their synapses and support their excitability. Histology showed diminished net intensity of WFA-labeled PNNs in knockout mice, while total OLIG2+ cell numbers and neocortical myelination levels were preserved. The loss is therefore selective for the PNN scaffold rather than a general loss of oligodendroglia or myelin.
The authors propose a feed-forward model: DNMT1 loss increases presynaptic GABA release, triggering homeostatic desensitization of inhibitory signaling and glial remodeling, which culminates in extracellular matrix degradation and network desynchronization. Degraded PNNs may further destabilize PV function, amplifying the desynchronization. This cascade is the paper's synthesis, built from its own electrophysiology, transcriptomics, modeling and histology, and it reconciles GABAergic changes with mood-related behavior without requiring PV cell death.
What the mouse model can and cannot settle
All physiological and behavioral evidence comes from mice with a conditional PV-specific Dnmt1 deletion. The paper does not test whether restoring PNNs, blocking the glial changes or rescuing gamma synchrony reverses the behavioral phenotype, so the causal ordering in the feed-forward model remains a proposal. The oligodendrocyte and cell-cell communication results are model- and inference-based, and the authors themselves call the corticospinal-like neuron enrichment hypothesis-generating.
The behavioral phenotype is described as depression- and anxiety-like, a rodent readout rather than a clinical diagnosis. The link to human mood disorders rests on the paper's discussion of prior literature, not on new human data. The preserved performance on learning and evidence accumulation tasks also means the deficit is not a global cognitive or motor impairment, which narrows how the results should be generalized.
Inhibitory efficacy, not PV cell number, is the pivot
The most decision-relevant comparison in this paper is between PV firing rate and PV inhibitory efficacy. A reader tracking PV interneurons in psychiatric models might expect dysfunction to appear as fewer PV cells or less PV activity. Here the opposite pattern holds: more spontaneous PV spiking, weaker suppression of cortical activity during PV optogenetic activation, reduced gamma synchrony, and lower PNN intensity, with OLIG2+ cell numbers and myelination preserved.
That distinction sets the paper's priority. If DNMT1 loss acts through diminished inhibitory efficacy and PNN-supported timing rather than through PV cell loss, then measures of PV cell count or bulk PV expression could miss the relevant pathology, while assays of inhibitory strength, gamma coherence and PNN integrity would capture it. The paper's own untested step is whether intervening on the PNN or glial arm of this cascade changes the behavior.
Sources used in this answer
DNMT1 function in cortical PV interneurons modulates cortical network activity, mood disorder‑related behavior, and perineuronal nets
Conditional PV-specific Dnmt1 deletion in mice increased PV interneuron spiking but reduced inhibitory efficacy, desynchronized cortical networks, reduced visually evoked gamma oscillations and PNN intensity, and produced depression- and anxiety-like behavior without loss of OLIG2+ cells or myelination.