A fluorescent sensor for real-time measurement of extracellular oxytocin dynamics in the brain

Oxytocin is produced by neurons in the paraventricular and supraoptic nuclei of the hypothalamus and regulates diverse physiological processes in the periphery and brain, including sensory processing, feeding, social cognition and emotion. Dysfunction of OT signaling has been implicated in neurodevelopmental disorders and aging-related cognitive and emotional changes. Although exogenous OT has been explored as a therapy via peripheral routes (for example, intranasal), its efficacy and brain penetration remain controversial. A major barrier to progress is the lack of techniques for real-time measurement of OT dynamics in vivo across relevant timescales. Existing approaches such as microdialysis and reporter gene assays have limited temporal resolution. The study set out to develop and validate a sensitive, specific, minimally perturbative fluorescent sensor for extracellular OT that enables real-time recordings in living mouse brain, and to use it to uncover how endogenous OT dynamics relate to behavior and physiological state.

Prior methods to measure brain OT include microdialysis and reporter-based assays (iTango), which provide limited temporal resolution and sensitivity for fast neuromodulator dynamics. Recent advances in GPCR-based fluorescent sensors for neuromodulators (dopamine, acetylcholine, norepinephrine, adenosine, serotonin) incorporate circularly permuted GFP into the third intracellular loop to transduce ligand-binding-induced conformational changes into fluorescence signals. Structural information on OTR and maps of OTR expression guided target region selection (for example, AON). Reports questioning effective brain delivery of peripherally administered OT motivated direct in vivo monitoring of brain OT levels following peripheral versus central administration.

Sensor engineering: The authors screened oxytocin receptors (OTRs) from six vertebrates for plasma membrane trafficking in HEK293T cells and selected medaka OTR (meOTR) as scaffold. They inserted cpGFP into the third intracellular loop (IL3) and iteratively optimized (1) linkers flanking cpGFP, (2) transmembrane-to-intracellular loop regions, and (3) cpGFP residues through mutagenesis and screening. Three rounds yielded OT-1.0, OT-2.0, and OT-3.0, with OT-3.0 showing ~720% ΔF/F at 100 nM OT and antagonist sensitivity (L-368,899). OT-3.0 was named MTRIAOT (the fluorescent module termed MTRIA). A ligand-insensitive mutant (Y206A) served as control. Characterization in cells: In HEK293T cells, dose–response curves quantified Fmax and EC50, with minimal change in EC50 across variants. Ligand specificity testing showed high sensitivity to OT and isotocin, low sensitivity to vasopressin/vasotocin/inotocin, and insensitivity to nematocin. Coupling to downstream signaling was assessed by Ca2+ imaging (jRGECO1a) for Gα coupling and split NanoLuc complementation for β-arrestin recruitment; MTRIAOT showed no detectable coupling while wild-type OTR did. Chronic 100 nM OT exposure did not cause sensor internalization. Kinetics were measured using local OT and antagonist application, yielding on- and off-times of ~1.2 s and ~26 s in HEK293T. pH sensitivity was minimal from pH 6.6–8.2. Primary neurons: In rat hippocampal neurons, MTRIAOT localized to soma and neurites, responded robustly to OT and was blocked by L-368,899. Neuronal EC50 was ~20.2 nM; kinetics were faster (τon ~0.83 s; τoff ~9.5 s) than in HEK293T. In vivo expression and recording: MTRIAOT was delivered via AAV (Syn promoter) to the anterior olfactory nucleus (AON), a region with high OTR expression. Fiber photometry measured 470 nm excitation (signal) with 405 nm isosbestic reference in anesthetized and freely behaving mice. Intracerebroventricular (ICV) injections of OT (0–20 μg in 10 μl) assessed dose dependence; intranasal and intraperitoneal administration (20 μg) were compared to central delivery. Optogenetic validation used AAV-OT promoter–driven ChRmine-mScarlet in PVN OT neurons with red-light trains (647 nm, 10 ms, 20 Hz, 30 s; 0–25 mW) to evoke release while recording AON MTRIAOT. Kinetic rise/decay constants were extracted. Behavioral paradigms: In freely moving mice, long-duration recordings quantified spontaneous ultradian OT oscillations in AON. To test dependence on OT neuron release, TeLC (tetanus toxin light chain) was expressed under OT promoter in PVN. Social interaction responses were elicited by exposure to a conspecific mouse in a wire cage versus a toy; acute stress responses were elicited by a 10 s tail lift, with simultaneous 405 nm reference controls and no-stimulus trials. Physiological perturbations: Effects of anesthesia were tested using intraperitoneal dexmedetomidine/midazolam/butorphanol (mix-anes) with reversal by atipamezole, and graded isoflurane (1%, 4%, then 0%). To exclude direct optical effects on the sensor, HEK293T controls measured basal and OT-evoked fluorescence during anesthetic exposure. Food deprivation (~half day) and refeeding were used to probe metabolic state effects. Aging was assessed by comparing OT dynamics in mice aged ~2 months, ~6 months, ~1 year, and ~2.5 years; sensor expression and tissue viability were confirmed, and AON odor-evoked Ca2+ responses (jGCaMP8s) were recorded as control. Generalization to other GPCRs: The MTRIA module was fused into IL3 between TM5–TM6 positions (5.62–6.36) across 184 receptors for 46 ligands (human/mouse/zebrafish/medaka). Fluorescence responses to high ligand concentrations were screened in HEK293T; best performers per ligand were designated MTRIAligand sensors. Data analysis: Fluorescence signals were processed as ΔF/F0 in vitro and z-scores in vivo; area under the curve (AUC), peak counts, peak amplitudes, rise/decay constants, and time to peak were quantified. Appropriate statistical tests (t-tests, one-way ANOVA with post hoc corrections) were used; details are in Supplementary Notes.

- Sensor performance: Iterative engineering yielded MTRIAOT (OT-3.0) with ~720% ΔF/F at 100 nM OT, antagonist sensitivity (L-368,899), and an EC50 ~20.5 nM in HEK293T. On/off kinetics were ~1.2 s/~26 s in HEK293T and ~0.83 s/~9.5 s in primary neurons. Sensor showed high specificity for OT/isotocin, weak responses to vasopressin orthologs, and no detectable coupling to G protein or β-arrestin pathways. Fluorescence was stable over prolonged OT exposure and minimally pH-dependent (pH 6.6–8.2). - In vivo detection and pharmacology: ICV OT induced robust, dose-dependent AON signals with detectable increases at ≥0.2 μg. Intranasal or intraperitoneal 20 μg OT did not evoke detectable AON responses, implying <1% of peripherally administered OT reaches AON compared to central delivery. Optogenetic activation of PVN OT neurons evoked graded MTRIAOT signals in AON proportional to laser power (0–25 mW), with mean rise and decay times of ~7.8 s and ~14.6 s; controls without ChRmine showed no response. - Ultradian OT oscillations: In freely behaving mice, AON OT signals exhibited spontaneous transient increases approximately every ~2 hours (“OT oscillation”), absent in 405 nm reference and in control mutant sensor recordings. Expression of TeLC in PVN OT neurons significantly reduced oscillation frequency and peak amplitude (for example, peaks per hour and peak z-scores decreased; unpaired t-tests P=8.2×10−4 and P=1.6×10−4), demonstrating dependence on oxytocinergic release. - Behaviorally evoked responses: Social interaction with a conspecific (but not a toy) produced a gradual MTRIAOT increase in AON with a rise time constant ~1 minute (AUC increase, paired t-test P=0.03). Acute stress (10 s tail lift) near the central amygdala elicited rapid OT responses with time to peak ~6 s; robust 470 nm responses were absent in 405 nm reference and no-stimulus controls (ANOVA with Bonferroni, P≪0.001 for tail lift 470 vs 405; no difference without stimulation). - Modulation by physiological state: General anesthesia suppressed brain OT levels: mix-anes drove undershoot below baseline reversed by atipamezole; isoflurane produced dose-dependent reductions (1% to 4%) with recovery at 0%. Anesthetics did not directly alter sensor fluorescence in HEK293T (no significant changes in basal or OT-evoked ΔF/F0). Food deprivation induced disturbed oscillations with undershoot transients (“OT turbulence”), which normalized upon refeeding. Aging slowed the frequency of OT oscillations without significant amplitude change; control Ca2+ responses to odor in AON remained robust in 1-year-old mice. - Generalization: Of 184 GPCRs tested with MTRIA insertion, 54 (~30%) produced >50% ΔF/F0 upon ligand application in HEK293T; the top 24 per ligand were designated MTRIAligand, indicating broad applicability of the MTRIA system for sensor development.

The study demonstrates that a GPCR-based fluorescent sensor, MTRIAOT, enables real-time measurement of extracellular OT dynamics in vivo with high sensitivity, specificity, and fast kinetics while minimally perturbing downstream signaling. This capability addressed key gaps in the field by revealing endogenous OT patterns, including ultradian oscillations (~2-hour intervals) in AON that likely escaped detection with slower methods. MTRIAOT also resolved rapid, context-dependent OT responses to social interaction and acute stress across second-to-minute timescales. Pharmacological tests indicated that peripheral OT administration (20 μg intranasal or intraperitoneal) did not measurably elevate AON OT levels under the tested conditions, whereas central ICV OT did, informing debates on nose-to-brain OT delivery and blood–brain barrier transit. The modulation of OT dynamics by anesthesia, metabolic state (fasting/refeeding), and aging underscores the need to consider experimental context and subject state when interpreting OT-related measurements and behaviors. Finally, the MTRIA framework generalized to many other GPCRs, suggesting a scalable route to sensors for diverse neuromodulators.

MTRIAOT is a robust, fast, and specific fluorescent sensor that enables in vivo, real-time tracking of extracellular oxytocin dynamics. Using fiber photometry in mice, it uncovered ultradian OT oscillations, resolved rapid OT responses to social and stress stimuli, and revealed strong modulation of brain OT levels by anesthesia, fasting, and aging. The findings clarify that, under the tested conditions, peripherally administered OT does not measurably elevate OT in AON, while central administration and optogenetic release do. Beyond oxytocin, the MTRIA strategy generalizes to a broad set of GPCRs, offering a platform for building sensors for many ligands. Future directions include quantitative calibration of in vivo OT concentrations (for example, via fluorescence lifetime or FRET-based designs), investigation of mechanisms and functions of OT oscillations, exploration of regional OT dynamics across the brain, and application of MTRIA-based sensors to other neuromodulatory systems.

- Quantitation: Current readouts are relative (ΔF/F0 or z-scores); absolute in vivo OT concentrations are not directly measured. The authors propose future lifetime or FRET-based designs for quantitative sensing. - Regional generalizability: Peripheral OT delivery was assessed in AON; effects in other brain regions may differ and require validation. - Aging and surgery: Although controls support tissue viability, reduced oscillation frequency with age could be influenced by surgical vulnerability in older animals; this cannot be fully excluded. - Sensor kinetics and spatial sampling: Fiber photometry integrates signals over a volume and does not resolve cellular/subregional sources; sensor kinetics, though fast, may still integrate over release/clearance processes. - Context dependence: Anesthesia, metabolic status, and behavioral context significantly affect OT dynamics, complicating comparisons across conditions and studies.