Perinatal serotonin signalling dynamically influences the development of cortical GABAergic circuits with consequences for lifelong sensory encoding

The study investigates how perinatal serotonin signalling, which fluctuates with behavioural state, shapes developing neocortical circuits and sensory processing. The authors hypothesised that neuromodulatory signals varying with the sleep–wake cycle impinge on local cortical circuit elements to structure information transfer during perinatal life. Specifically, they examined serotonin’s role because it bridges behavioural state and sensory experience in adults and is tightly regulated in neonates, including transient expression of the serotonin transporter (SERT) in non-serotonergic forebrain neurons. Given clinical links between SERT polymorphisms or prenatal SSRI exposure and neurodevelopmental risk, the work aims to clarify how dysfunctional serotonin modulation alters systems-level activity and the maturation of cortical GABAergic microcircuits, ultimately affecting sensory encoding across development and into adulthood.

Background work shows that serotonin levels oscillate with sleep–wake states in adults, being lowest during active sleep, and fluctuate with sensory, reward, and punishment experiences. In early life, serotonergic innervation and dorsal raphe physiology evolve, and SERT is transiently expressed in thalamocortical and cortical projection neurons, disappearing from first-order thalamic nuclei around P10. Serotonin modulates thalamocortical transmission (attenuation via 5-HT1 receptors) and can promote cortico-cortical signalling via 5-HT3R-expressing interneurons (including VIP cells), which mediate top-down control. Clinically, SERT promoter polymorphisms interact with early trauma to increase risk for psychiatric disorders, and SSRIs cross the placenta with reports of neonatal serotonergic effects; meta-analyses link in utero SSRI exposure to increased risk of neurodevelopmental disorders, though confounding by maternal diagnosis is a concern. Prior animal work indicates excessive 5-HT1B activation disrupts sensory map formation and that transient SERT expression refines cortical development. These studies motivate testing how perinatal serotonin dynamics and SERT function influence early network activity and inhibitory circuit maturation.

• Animal models and timeline: Mouse pups were studied across two developmental windows: Th-SERT (P7–P10; transient thalamic SERT expression) and immediate Adult SERT pattern (P11–P16; raphe-restricted SERT). Adult assessments were performed at 8–12 weeks. Genotypes included SERT wild-type, heterozygous (SERT-Het), and knockout (SERT-KO). Maternal genotype effects were tested by breeding SERT-Het males with WT females to generate wt and het offspring.
• In vivo imaging of serotonin and calcium: AAV9.hSyn-g5-HT3.0 (serotonin sensor) and AAV1.hSyn.GCaMP6s were injected intracranially at P0–P1 into S1 barrel field (S1BF). Two-photon imaging (16x/0.8 NA objective, 30 Hz, ~150 μm depth; L2/3) recorded g5-HT3.0 and GCaMP6s signals daily from P7–P16. VIP or Nkx2-1 interneurons were labelled with tdTomato (VIP-Cre;Ai9 and Nkx2-1-Cre;Ai9) for interneuron-focused analyses.
• Sensory stimulation and behavioural state: Stimuli (pseudo-random) included single- and multi-whisker deflection (20 Hz, 40 ms), airpuff (aversive), smooth/rough textures, and complex tone (5 kHz carrier). Behavioural state (awake, quiet sleep, active sleep) was scored via infrared video and DeepLabCut tracking of forelimb movements.
• Pharmacology: WT pups received daily oral fluoxetine (SSRI; 10 mg/kg in 10% sucrose) or sucrose vehicle from P2–P14. Serotonin dynamics under dosing were validated by increased g5-HT3.0 signal during Th-SERT and Adult SERT windows.
• Longitudinal calcium dynamics: Network activity was quantified by detecting high- (H-) and low- (L-) synchrony events during 20 min baselines; ΔF/F was normalised to the 1st percentile of fluorescence across recording. Pairwise correlations over distance and whisker-evoked response amplitudes were quantified. VIP and Nkx2-1 interneuron event properties and sensory responses were analysed.
• Ex vivo electrophysiology and circuit mapping: Acute S1BF slices were used for whole-cell recordings. Optogenetics: Minimal LED stimulation of ChR2-expressing interneuron classes (VIP-, Nkx2-1-, or SST-Cre;Ai32) probed unitary IPSCs onto L2/3 pyramidal neurons. Thalamocortical ChR2 stimulation (Olig3-Cre;Ai32) probed unitary thalamic EPSCs and feed-forward IPSCs in L4 spiny stellate neurons. LSPS of caged glutamate mapped translaminar GABAergic inputs onto L4 SSNs and L2/3 PYRs across genotypes and maternal backgrounds.
• Adult encoding and decoding: Adult S1BF responses to stimulus battery were imaged, and logistic regression classifiers (scikit-learn, L2 penalty) assessed stimulus discriminability and whether genotype/treatment/maternal genotype could be decoded from post-stimulus population responses, with bootstrapped significance thresholds.
• Histology: Quantified layer-wise densities of VIP(tdTomato), Nkx2-1(tdTomato), PV (immunolabelled), and SST (immunolabelled) interneurons relative to DAPI using QuPath classifiers; imaging under consistent settings; analysis blinded to condition.
• Statistics: Normality by Shapiro–Wilk; appropriate parametric/nonparametric tests (t-tests, ANOVA, Kruskal–Wallis, Mann–Whitney; two-way ANOVA/permutation ANOVA for multi-factor); multiple-comparisons corrections as specified; Kolmogorov–Smirnov for distributions; significance at p<0.05.

• State-dependent serotonin dynamics in neonates: g5-HT3.0 signals in S1BF were highest during wakefulness and lowest during active sleep during Th-SERT (consistent with adult patterns). SSRI dosing increased g5-HT3.0 during active sleep and reduced time spent in active sleep (Fig. 1j–l; two-way ANOVA sleep state p<0.001; treatment p=0.017; AS time t-test p=0.015).
• Sensory-evoked serotonin is buffered by developmental SERT: During Th-SERT (P7–P10), whisker or sound did not elicit detectable 5-HT signals, while aversive airpuff did (Fig. 1b; airpuff p=0.031). After Adult SERT onset (P11–P13), whisker, airpuff, and sound evoked 5-HT signals (p=0.046; 0.003; 0.010). Spatially, responses transitioned from global to localized and were larger near VIP interneuron-rich regions in Adult SERT.
• Disrupting SERT elevates sensory-evoked 5-HT early: In SERT-Het and SERT-KO pups, whisker stimulation evoked prolonged g5-HT3.0 signals during Th-SERT (Fig. 1f; KO/Het p<0.05 across ~14–15 s; WT ns). Acute fluoxetine dosing replicated prolonged 5-HT responses to whisker and airpuff (Fig. 1g–h; whisker SSRI p<0.05 for 3 s; airpuff p<0.05 for 20 s), but not to sound at this age (Fig. 1i).
• Early cortical hypoactivity followed by hyperexcitability in SERT-KO: During Th-SERT, SERT-KO showed decreased H-event amplitude and duration with increased frequency versus WT (Two-way ANOVA: amplitude genotype p=0.003; duration p=0.007; frequency p=0.001). Baseline activity decorrelated across distance and ΔF/F distributions shifted left (reduced activity; KS p≤0.0007). After P11, both SERT-Het and SERT-KO exhibited right-shifted ΔF/F distributions and increased whisker-evoked amplitudes (Two-way ANOVA genotype p=0.006; WT–KO p=0.038; stimulus-triggered average WT–KO p=0.010), indicating hyperexcitability prior to active sensory awareness (Fig. 2).
• Interneuron-specific activity changes (in vivo): VIP interneurons in SERT-KO had higher H-event amplitude, frequency, and duration and larger whisker responses during Th-SERT and Adult SERT (Kruskal–Wallis p<0.001; multiple pairwise significant). Nkx2-1+ interneurons (MGE lineage) in SERT-Het had decreased H-event amplitude and duration but increased frequency; whisker-evoked responses were reduced during Th-SERT (Mann–Whitney p=0.001) and near-normal by Adult SERT (p=0.053) (Fig. 3).
• Synaptic underpinnings (ex vivo): Minimal VIP→L2/3 PYR IPSCs were unchanged across genotypes, indicating no compensatory strengthening (ANOVA p=0.50). Minimal Nkx2-1→PYR IPSCs were smaller in SERT-KO (WT–KO p=0.047), consistent with reduced PV-mediated input. SST→PYR minimal input was increased in SERT-KO (WT–KO p=0.019). Thalamic unitary EPSCs to L4 SSNs were unchanged (ANOVA p=0.514), but feed-forward inhibition was reduced in SERT-KO (ANOVA p<0.001). LSPS maps showed SERT-KO had predominantly translaminar (L5b) GABAergic input onto L4 SSNs, with sparse local PV-like input; WT pups from SERT-Het dams lacked expected transient L5b→L4 input (Fig. 4).
• Maternal genotype shapes transient SST circuits: Breeding SERT-Het males to WT dams restored the transient L5b→L4 SST connection in both wt and het offspring and revealed early L5b→L2/3 input in wt but not het pups, indicating maternal SERT-het background and postnatal serotonin dynamics affect supragranular SST connectivity trajectories (Fig. 5; significant differences at P4–6 and P7–9).
• Postnatal SSRI reproduces hypo→hyper trajectory: P2–P14 fluoxetine caused reduced H-event amplitude, increased frequency, unchanged duration (p=0.040; 0.016; 0.310), decreased pairwise correlations, left-shifted ΔF/F distributions (KS p=0.022), and reduced whisker responses during Th-SERT (p=0.018). After P11, cortical responses showed features of hyperexcitability (elevated mean signal; Fig. 6), though milder than SERT-KO.
• Adult sensory encoding: SSRI-treated adults (but not SERT-KO) exhibited increased amplitude and cell recruitment to single-whisker and airpuff stimuli (whisker amplitude p=0.004; recruitment p=0.032; airpuff amplitude p=0.046; recruitment p=0.007). Stimulus-type classifiers achieved high accuracy across groups, indicating preserved discriminability. Condition decoding: SERT genotype was not classifiable; maternal genotype was decodable from baseline and sound; postnatal SSRI treatment was classifiable from most stimuli, strongest for airpuff (Fig. 7n).
• Adult interneuron composition: In SERT-KO adults, PV% (per DAPI) decreased across layers (two-way ANOVA genotype p=0.004), with SST and VIP unchanged. In SSRI-treated adults, Nkx2-1(tdTomato) interneurons decreased (treatment p=0.032), VIP increased (treatment p=0.032), while PV and SST counts were not significantly altered, indicating divergent long-term cellular outcomes of genetic vs postnatal pharmacological SERT disruption.

The findings demonstrate that perinatal serotonin signalling is dynamically regulated with behavioural state and that transient developmental SERT expression buffers sensory-evoked serotonin, restricting detectable responses mainly to aversive stimuli during Th-SERT. Disrupting SERT genetically or pharmacologically elevates serotonin during this window, producing early cortical hypoactivity through attenuated thalamocortical drive (via 5-HT1 receptors), enhanced recruitment of 5-HT3R/VIP interneurons, and altered activity of MGE-derived interneurons. These changes impede activity-dependent maturation of PV-mediated feed-forward inhibition while enhancing transient SST translaminar inhibition. As networks desynchronise (~P11), insufficient PV inhibition yields cortical hyperexcitability. Notably, postnatal SSRI exposure induces lasting hyper-responsivity and interneuron composition shifts into adulthood, whereas lifelong SERT-KO, despite early hyperexcitability, does not show adult sensory hyper-responsivity, potentially due to continued serotonin dysregulation and compensations. Maternal SERT genotype further modulated transient SST circuits and adult baseline dynamics, highlighting fetal and early postnatal serotonergic milieu as determinants of circuit assembly. Together, the work links perinatal serotonergic state to the developmental integration of bottom-up and top-down pathways and excitation–inhibition balance, providing a mechanistic framework relevant to sensory hypersensitivity and neurodevelopmental risk associated with SERT polymorphisms and SSRI exposure.

This study longitudinally charts how perinatal serotonin dynamics sculpt cortical GABAergic microcircuit development and lifelong sensory encoding. Transient SERT expression buffers neonatal sensory-evoked serotonin; genetic or pharmacological disruption elevates serotonin, producing early hypoactivity, impaired maturation of PV-mediated feed-forward inhibition, and hyperexcitability after desynchronisation. Postnatal SSRI dosing leads to persistent adult sensory hyper-responsivity and shifts in interneuron composition, whereas SERT-KO does not show equivalent adult encoding changes. Maternal SERT status independently modulates transient SST circuits and adult baseline dynamics. These results identify a critical perinatal window wherein serotonin perturbations have enduring consequences, informing interpretation of genetic and pharmacological risks for neurodevelopmental disorders. Future work should dissect cell-type-specific receptor mechanisms, directly track PV maturation in vivo, test combined genetic–environmental interactions (e.g., SERT-Het with SSRI exposure), and relate altered developmental dynamics to behaviour and cognition.

• Direct in vivo interrogation of PV interneuron function in early postnatal stages was not feasible due to delayed PV marker expression; PV contributions were inferred from Nkx2-1 optogenetics and SST-subtractive approaches ex vivo.
• Nkx2-1Cre;SERT-KO animals did not survive to enable longitudinal in vivo interneuron imaging, limiting direct comparisons across development for this genotype.
• Early g5-HT3.0 expression was sufficient to detect large aversive responses, but low-level serotonin changes to non-aversive stimuli during Th-SERT could remain under detection thresholds in controls, potentially underestimating baseline 5-HT dynamics.
• Some adult analyses had modest sample sizes (e.g., SSRI group n=4 in certain assays), which may limit power for detecting subtler effects.
• Findings are in mice and generalisation to human development must consider species differences in timing and circuit maturation; auditory system immaturity at early ages also constrained interpretation of early sound-evoked responses.
• Dosing and genetic manipulations broadly alter serotonin signalling; receptor- and cell-type-specific contributions were not fully isolated in vivo.