Optopharmacological tools for precise spatiotemporal control of oxytocin signaling in the central nervous system and periphery

Oxytocin is a neuropeptide critical for maternal physiology and social behavior, and is thought to be dysregulated in several neuropsychiatric disorders. Despite the biological and neurocognitive importance of oxytocin signaling, methods are lacking to activate oxytocin receptors with high spatiotemporal precision in the brain and peripheral mammalian tissues. Here we developed and validated caged analogs of oxytocin which are functionally inert until cage release is triggered by ultraviolet light. We examined how focal versus global oxytocin application affected oxytocin-driven Ca2+ wave propagation in mouse mammary tissue. We also validated the application of caged oxytocin in the hippocampus and auditory cortex with electrophysiological recordings in vitro, and demonstrated that oxytocin uncaging can accelerate the onset of mouse maternal behavior in vivo. Together, these results demonstrate that optopharmacological control of caged peptides is a robust tool with spatiotemporal precision for modulating neuropeptide signaling throughout the brain and body.


Main
Oxytocin is an evolutionarily conserved neuropeptide with essential roles in many tissues across mammalian and many non-mammalian species [1][2][3][4] . In mammals, oxytocin is primarily synthesized in the paraventricular (PVN) and supraoptic nuclei (SON) of the hypothalamus 5,6 . The release of oxytocin into the bloodstream is critical for a number of peripheral physiological processes, including inducing uterine contractions during parturition to facilitate childbirth, and milk ejection during nursing [7][8][9] . Oxytocin is also released throughout the central nervous system and has been shown to facilitate social behaviors such as maternal care 10,11 and pair bonding 12,13 . Moreover, a deficit in oxytocin signaling is believed to be associated with several disorders, such that oxytocin supplements are being explored as a potential clinical treatment for obesity 14 , pain 15,16 , addiction 17 , and psychiatric disorders such as autism spectrum disorders, social anxiety, posttraumatic stress disorder, and post-partum depression [18][19][20] . However, the mechanisms of how oxytocin release and oxytocin receptor (OXTR) signaling modulate various cell types, tissues, and brain areas to regulate internal states and social interactions remains unclear. Thus far, the inability to control oxytocin peptide levels with high spatiotemporal precision in vivo, have hampered these efforts.
For example, stimulating endogenous release or providing systemic OXTR agonists/antagonists to regulate signaling at specific sites or tissues without off-target effects has been challenging.
Furthermore, both peptide and synthetic OXTR ligands can effectively bind and activate multiple receptor types present on the same cell or nearby cells. Direct peptide application in vivo and in brain slices by perfusion, pressure injection 21 or iontophoresis 22 produces slow, prolonged and spatially imprecise presentation of the peptide offering poor control over the concentration and gradients of peptide delivered. Therefore, it is imperative to design new tools to enable spatiotemporally precise oxytocin delivery for regulating physiology and behavior.
In the periphery, circulating oxytocin is primarily released from the brain into the blood via the posterior pituitary, but there are also other peripheral sources of oxytocin 23 . These sources converge on their peripheral targets via diffusion 24 . OXTR expression is found in both neuronal and non-neuronal cell types in various organs including uterus, ovaries, testis, prostate gland, mammary, kidneys, heart, thymus, adipocytes, pancreas, adrenal glands, and others 2,25 .
Consequently, oxytocin regulates a variety of critical physiological functions ranging from reproduction and lactation to control of body fluid and heart homeostasis 2,25 . Despite how fundamental oxytocin is in the periphery, spatiotemporal optogenetic studies of these different systems and processes have been inaccessible. Unlike in the brain, where the optogenetic release of oxytocin is, for the most part, practical (with some limitations and caveats), it is infeasible in the periphery. This is partly due at least to the lack of known oxytocin projections into peripheral tissues beyond the spinal cord and a lack of means to activate them 26 .
In the brain, previous studies of endogenous oxytocin release in vivo (e.g., through chemogenetic or optogenetic activation of hypothalamic oxytocin neurons) have provided essential information on the modulation of social behavior, mainly in rodents 10,27,28 . However, many hypothalamic cell types, including oxytocin neurons, are believed to co-express and release other neurotransmitters and peptides. Oxytocin-releasing neurons have been found to synthesize corticotropin-releasing hormone, cholecystokinin, and enkephalin [29][30][31] , each of which has various functions for regulating physiological states and behavior (and could be released under different physiological conditions).
Co-transmission of neuropeptides and small-molecules even by a single neuron can affect microcircuit activity and behavior through multiple contributing mechanisms and over a range of timescales 32 . Also, the spatial extent at which oxytocin signaling can affect a target brain region remains unclear. Oxytocin fibers into different brain areas do not seem to make conventional synapses. Fiber density may be sparse in many regions, and oxytocin can be released from the dendrites of hypothalamic neurons, suggesting that volume transmission must occur at least across some distances and in some regions 24,33,34 . Additionally, there are many examples of projectionreceptor mismatch across many brain regions which are mainly modulated by volume transmission and it may be difficult to activate with optogenetic approaches. Thus, compared to classical neurotransmitters, the spatial and functional specificity of oxytocin and other neuropeptides may be limited. Therefore, while genetics-based approaches have been useful for examining peptidergic modulation of neural circuit function, it can be difficult to parse out the contributions of specific co-transmitters from a particular cell population or the action of oxytocin in specific spatial locations or time points. Moreover, these approaches might be difficult to adopt for other species in which genetic manipulation is more challenging.
Here we developed optopharmacological variants of caged oxytocin, to precisely establish how oxytocin release drives OXTR signaling in peripheral (in mouse mammary gland) and brain tissue (in mouse hippocampus and cortex) and to ultimately control maternal behavioral and physiological responses. Optopharmacology enables a light-meditated selective activation of oxytocin signaling within a defined target region [35][36][37] . This method complements optogenetics and pharmacology by endowing light sensitivity to specific biomolecular targets. It thus provides molecular control of a given system, in this case OXTR activation and regulation of mammary gland function in the periphery and neuronal activity in the central nervous system. This tool does not require genetic manipulations or viral expression and thus can be used in any organism that expresses OXTRs. Furthermore, our approach exemplifies that this technique can be generalized to small signaling peptides that target G-protein-coupled receptors in the brain and body and would be applicable to both peptide agonists and antagonists for spatiotemporal specificity.

Design and photophysical characterization of caged oxytocin analogs
Oxytocin is an ideal candidate for caging because of its small size (nine amino acids) and established role in various biological functions. Photosensitive caging has previously been successfully achieved for other peptides of similar size [38][39][40] . To generate oxytocin that is functionally inert until light exposure, we produced analogs where individual amino acids were replaced with photo-caged unnatural amino acid versions of the corresponding residue using solidphase peptide synthesis. Previous functional studies of oxytocin have shown that the Cys and Tyr residues are important for receptor activation 41,42 . Additionally, the disulfide between the two Cys residues is responsible for the cyclic structure of oxytocin, which is important for the potency and metabolic half-life. Therefore, we hypothesized using photocage moieties on these residues would render oxytocin reversibly inert. Based on these considerations, we designed three caged oxytocin peptide analogs (Fig. 1a). For the first caged compound ('cOT1'), we replaced Tyr with orthonitrobenzyl-tyrosine. The second analog ('cOT2') incorporated a 4,5-dimethoxy-2-nitrobenzyl caging group at the N-terminus of Cys1. For the third caged oxytocin compound ('cOT3'), we replaced both cysteines with 4,5-dimethoxy-2-nitrobenzyl cysteine to impede cyclization through disulfide formation.
To predict if each photocaged oxytocin analog (cOT1, cOT2, and cOT3) would be inert, we performed molecular simulations of each compound in comparison to oxytocin bound to the OXTR, Protein Database (PDB) Code: 7RYC 43 (Fig. 1b-e). Computational modeling indicated that each docked compound would have significant steric clashes residues within the binding pocket of the OXTR, and the weakest binding analog was cOT1, and then cOT3, and cOT2 respectively (Fig. 1f, Extended Data Fig. 1). These simulations suggested that photocaging oxytocin would modify the binding affinity of oxytocin due to alterations in the peptide/protein interaction. This is consistent with structural studies of oxytocin bound to the OXTR. Cryo-EM structure (PDB: 7RYC) shows that the tyrosine residue of oxytocin penetrates deeply into the binding pocket of the helical bundle, forming strong hydrophobic and hydrogen bond interactions with OXTR internal binding pocket residues 43 . Adding a bulky caging group (ortho-nitrobenzyl group) with starkly different chemical properties at this position (as in the case of cOT1) should therefore diminish its affinity for OXTR.
Photolysis of the selected caging groups can produce bioactive uncaged products on a millisecond time scale 44 . Under optimal conditions, photorelease can proceed with >95% yield 44,45 . We designed these caged oxytocin analogs for photorelease with 365 nm light and confirmed that the ultraviolet (UV) and visible absorption spectra of each analog corresponded to the uncaging wavelength of 365 nm (Fig. 1a,g). The spectrum of each caged oxytocin analog irradiated with UV light over 30 minutes (365 nm LED, 3 mW) showed a significant change in absorbance corresponding to photo-uncaging and release of oxytocin with UV light while oxytocin remained unchanged ( Fig. 1h-k). Thus, oxytocin itself is stable towards 365 nm irradiation for durations of at least 30 minutes which is substantially higher than durations of irradiation that would be used in experimental conditions. (a) Amino acid sequences and chemical structures of oxytocin (OT) and the photocage-modified oxytocin analogs cOT1, cOT2, and cOT3. For the OT structure the functional groups that are modified with cages are color-coded: the Tyr residue is labeled in red, the N-terminus in purple, and the disulfide in blue.
(b) Structure of OT bound to the oxytocin receptor, adapted from PDB: 7RYC.
(f) Heat map analysis of OT, cOT1, cOT2, and cOT3 screened against the oxytocin receptor (PDB: 7RYC) via molecular docking modeling. In the gradient ruler, green color indicated strong binding, while red color indicates weak binding. Results listed from weakest binder: cOT1, cOT3, and cOT2.

In vitro validation of caged oxytocin analogs
We next determined if the caged oxytocin analogs cOT1, cOT2, and/or cOT3 are inactive at OXTRs prior to photolysis and then become active after photolysis. As OXTRs are believed to primarily couple to Gq/11 and mobilize intracellular calcium 2 , we compared the effects of uncaging to oxytocin wash-on using an in vitro functional calcium flux fluorescence assay of OXTR activation (Fig. 2a). To demonstrate the utility of uncaging of cOT1 in cell culture, we nm UV light. We did not observe any change in fluorescence after UV illumination. We then treated these cells with 1 µM of cOT1 and again used full-field illumination with 365 nm UV light, observing a substantial increase in FLUO-8 fluorescence after photolysis (Fig. 2c).
We used calcium imaging of CHO-K1 cells stably expressing OXTRs and loaded with the calcium-sensitive dye FLUO-8 to examine oxytocin wash-in (1 µM) and uncaging of each of the three candidate caged oxytocin compounds (cOT1, cOT2, cOT3; 1 µM). Before photolysis, cOT1 and cOT2 demonstrated minimal OXTR activation, comparable to that measured in HBBS buffer.
However, cOT3 showed significant baseline activation before uncaging ( Fig. 2d LED off, fluorescence: 40.4±1.9, n=5 cell culture wells, p=0.0008; Student's two-tailed paired ttest). After photolysis with an LED (365 nm, 3 mW), both cOT1 and cOT2 produced robust calcium signals, similar to that produced by oxytocin itself when compared to HBBS ( Fig. 2d Fig. 2). These experiments confirm that the caged oxytocin analogs we synthesized can effectively photorelease oxytocin peptide. For the remainder of this study, we focused on the effects of cOT1, as this reagent was the most inert of the caged-oxytocin analogs before photolysis and showed the most robust activation of OXTRs after photolysis.
To demonstrate that the effects of cOT1 uncaging are specific for OXTRs, we performed similar UV photolysis experiments with cOT1 onto CHO-K1 cells expressing OXTRs but also included the specific OXTR antagonist OTA in the bath. When inhibiting OXTR with OTA, both oxytocin and cOT1 failed to activate OXTR+ cells after photolysis, indicating that uncaging of oxytocin from cOT1 is specific for OXTR binding ( Fig. 2e We were concerned that cOT1 might potentially act as an OXTR antagonist. We performed calcium imaging and applied varying concentrations of cOT1 (0 µM, 0.25 µM, 0.5 µM, 1 µM, 10 µM) to OXTR+ CHO-K1 cells loaded with FLUO-8, and then subsequently treated with 1 µM of oxytocin and no UV light (LED off). Over the entire concentration range, there was no significant reduction in the calcium signal when compared to cells treated solely with 1 µM oxytocin ( Fig.   2g). As expected, when the LED was then turned on, there was an increase in calcium signal for cells treated with both 1 µM oxytocin and 10 µM cOT1 compared to oxytocin alone ( Fig. 2g; 1 µM oxytocin LED off, fluorescence: 143.3.1±4.9, n=5 cell culture wells; 1 µM oxytocin 10 µM cOT1 LED on, fluorescence: 171.2±2.9, n=5 cell culture wells, p=0.0043, Student's two-tailed paired t-test). These results demonstrate that cOT1 is not an OXTR antagonist, and thus this compound might enable uncaging experiments in which balance between activation and inhibition in a receptor is important, such as in ex vivo tissue and in freely behaving animals.
To quantify the difference in binding affinity of cOT1 compared to oxytocin, we obtained dosedependent competitive binding curves of cOT1 and oxytocin, either caged or uncaged (Fig. 2h).

Photorelease of oxytocin in mammary tissue
Oxytocin is critically important for the milk ejection reflex; knockout mice lacking either oxytocin peptide or oxytocin receptors are unable to expel milk to support their young 46,47 . With other methods such as systemic release or optogenetic stimulation of the hypothalamus, precise regulation of mammary function can be especially challenging because oxytocin is released into the bloodstream from the posterior lobe of the pituitary gland. We therefore next asked about the utility of caged oxytocin in mammary tissue. During lactation, mammary glands express high levels of oxytocin receptors (Fig. 3a,b). We performed three-dimensional time-lapse imaging of ex vivo mammary tissue from lactating GCaMP6f::K5CreERT2 mice treated with cOT1 and subsequently bathed with oxytocin. With a 405 nm laser, we uncaged a region of interest (ROI) at 100% laser power for 3 seconds at three different time points. We observed that uncaging on one alveolar cluster typically led to local activation just of that cluster, and did not seem to induce global synchrony of calcium release and contractions throughout adjacent mammary tissue.
Increases in cytosolic calcium and alveolar movement occurred only in the uncaging ROI (Fig.   3c,d and Supplementary Movie 3). In contrast, in response to whole-tissue wash on of oxytocin, the star-shaped basal epithelial cells robustly responded with an increase in intracellular calcium, followed by cell and alveolar contraction across all cells throughout the sample in a diffusive manner after bath application (Fig. 3c,d), as expected from previous results 48 . This implies that global alveolar contraction and milk ejection requires a minimum concentration and more diffuse spatial distribution of oxytocin in the tissue.

Photorelease of oxytocin in neural tissue
Next we examined cOT1 uncaging in tissue from the mouse brain. To validate the ability to uncage cOT1) to activate OXTR + neurons in the brain, we first sought to carry out a functional assay on the depolarizing effect of oxytocin previously identified in OXTR + neurons in the hippocampus 49,50 . OXTRs are highly expressed in CA2 pyramidal neurons in the hippocampus 33 and their activation drives neuronal depolarization 49,50 . We visually targeted OXTR-expressing (OXTR + ) neurons in the dorsal CA2 region (dCA2) (Fig. 4a), using OXTR-Cre mice crossed with an Ai9 tdTomato reporter line [51][52][53] .
However, additional 1-2 minutes of 365 nm UV light induced changes in firing pattern as well as resting Vm (from -67.6±2.5 mV to -55.9±2.9 mV, n=10 neurons, p<0.0001, unpaired t-test), to a similar degree as 1 µM oxytocin (two-way ANOVA, p=0.59) (Fig. 4b,c,e-g). We confirmed that the effects of cOT1 uncaging were due to OXTR activation, as both the firing pattern and the depolarization driven by UV uncaging were blocked by pre-application of 10 µM OTA, a specific OXTR antagonist (Fig. 4d,h-i).
We also asked if oxytocin uncaging would have similar effects on neurons in another brain region, in this case the mouse auditory cortex. We have previously reported lateralized OXTR expression to the left auditory cortex (AC), and the increased OXTR expression has been shown to be important for maternal behaviors such as retrieving isolated pups back to the nest 10,11 . We targeted our recordings to OXTR-expressing (OXTR+) neurons in the auditory cortex (Fig. 4j) using OXTRcre::Ai9 mice [51][52][53] . We performed whole-cell recordings in acute brain slices of the mouse auditory cortex. We found that photorelease of the caged oxytocin caused significant membrane depolarization of OXTR+ neurons (n=9 neurons, p =0.0003, paired t-test) (Fig. 4k,l). The effect of photolyzed cOT1 was similar to the depolarization produced by pharmacological application of oxytocin (Fig. 4k,m; n=10 neurons, p =0.0013, paired t-test). Together, these in vitro experiments in CA2 and auditory cortex demonstrate that cOT1 is a robust optopharmacological tool to activate OXTRs in neuronal tissue. Additionally, this compound can be applied to slices for hours without apparent toxic effects on neurons. We suggest that this tool will enable spatiotemporal delivery of

Photorelease of oxytocin enables control of maternal behavior
Finally, we asked if caged oxytocin could be useful in vivo for regulation of behavior; in this case, the emergence of mouse alloparenting behavior. One important mouse maternal behavior is pup retrieval; when pups are separated from the nest, they make ultrasonic distress calls, which experienced mothers (dams) then use as a cue to retrieve the isolated pups. Initially, most inexperienced (pup-naïve) virgin female mice do not retrieve pups but can become alloparents during co-housing with dams. This process requires native oxytocin and is accelerated by oxytocin supplements 10,11,55 .
To deliver cOT1 into the mouse auditory cortex in vivo, we implanted an optofluidic cannula positioned in the left auditory cortex of pup-naïve female virgin C57Bl/6 mice (Fig. 5a), and after each experiment we verified the position of the cannula (Fig. 5b). Virgin females were co-housed with experienced dams and their litters over several days, and we initially verified that dams retrieved pups (Fig. 5c-e). Two cohorts of virgin females were infused with cOT1 and were tested for pup retrieval (Fig. 5d); one experimental 'LED on' group for uncaging cOT1, and one control 'LED off' to retain cOT1 as functionally inert (Fig. 5e-f and Supplementary Movies 4 -5). Within 12 hours of co-housing with a dam and her pups, virgin females with the LED on after cOT1 infusion began retrieving more than the LED off cohort (Fig. 5e; off 'LED on': 7 / 9 animals retrieved; LED off: 3 / 12 animals retrieved, p=0.03, two-tailed Fisher's exact test). This meant that the emergence of this alloparental behavior was slower for the LED off cohort compared to the LED on cohort, although eventually most animals in both groups began retrieving (Fig. 5f), as expected from past work on co-housing 10,11,55 . (c) System for monitoring pup retrieval behavior compatible with an optofluidic cannula.
(d) Schematic of pup retrieval behavior paradigm of virgin female mice co-housed with experienced dams comparing caged (LED Off) and uncaged (LED On) oxytocin.

Discussion
In this study, we designed several photoactivatable oxytocin analogs by replacing key residues (Cys and Tyr) with their photocaged unnatural amino acid complement. We characterized the photophysical and photochemical properties of each caged oxytocin peptide to ensure successful uncaging with UV light. We compared each caged oxytocin analog using calcium sensitive assays in cell culture as a proxy for the activation of OXTR in the absence and presence of UV light. cOT1 was caged at the Tyr residue hydroxyl group, and we found this compound to be the best performing of the caged oxytocin derivatives, without antagonistic effects on the oxytocin receptor under our experimental conditions. The binding affinity of cOT1 to the oxytocin receptor in the caged form is ~280X less than in the uncaged form, where after photolysis with UV light, its binding affinity for the oxytocin receptor is approximately the same as oxytocin (because the reaction yields native oxytocin). These results point to the Tyr residue of oxytocin as the most effective caging site to attenuate the potency of. In this study, we specifically used the orthonitrobenzyl photo labile group to cage cOT1, which has been successfully used in various biological applications, including the photocontrol of proteins such as nicotinic acetylcholine receptors 45 , protease activity 56 , and has been established as a general tool for turning off and on protein activity (applied to GTPases, kinases, RNA demethylases, caspases and bacterial effector proteins) through caging and decaging actions 57 . Other variations of caged tyrosine unnatural amino acids analogs have also been successfully used 56 . The design of cOT1 serves as a scaffold for designing and tuning the photoactivation properties of caged oxytocin. For example, the caging group on the Tyr can be replaced with other caging functional groups with distinct chemical and physical properties such as to optimize for faster photolysis or multiphoton uncaging. Together, this study serves as a model for designing photoactivatable derivatives of similar neuropeptides of similar sizes, such as the closely related neuropeptide, vasopressin. Furthermore, this strategy will enable photocontrol of specific peptidergic agonists and antagonists for neuropeptide receptors such as OXTR and AVRP for spatiotemporally controlled pharmacology in effort to minimize potential off-target interactions.
Oxytocin signaling is fundamentally important in non-neuronal tissues, most notably the mammary gland where oxytocin receptor activation triggers milk ejection in response to infant suckling 7 . Basal epithelial cells respond to oxytocin with increased intracellular calcium, followed by cell and alveolar contraction 48 . Both stochastic and coordinated events have been observed in mammary tissue, and mechanisms for cell entrainment and tissue-level synchronization are the topic of ongoing investigations. Caged compounds may be particularly advantageous for deciphering cellular connectivity in the mammary gland by optical spatiotemporal activation, particularly since genetic methods present challenges in this epithelial system. We show that caged oxytocin can be released in thick, lactating mammary tissue with a brief 405 nm pulse, causing local release of oxytocin and alveolar unit contraction. These findings point to the utility of this synthetic compound, both with conventional microscopes and in non-neuronal systems.
As the toolbox of optical tools in biology continues to grow, combining these tools becomes an increasingly powerful approach for dissecting information within brain circuits and other organs.
Combining these tools will allow for the control and monitoring of more signals simultaneously.
For example, a useful feature of using caged oxytocin is that it allows for three-channel (potentially more) multiplexing for imaging studies, as we have shown with the imaging of contracting mammary tissue activated by uncaging oxytocin (Fig. 3). Specifically, uncaging oxytocin with 405 nm light (or 365 nm), and then online imaging of calcium dynamics using GCaMP6f and cell structural changes using CellTracker Red in the green and red channels correspondingly. This strategy has the potential to be useful in cases where oxytocin can be uncaged and have two different readouts in the green and red channels with a combination of sensors and/or structural markers. Additionally, uncaging oxytocin can be used in parallel with optogenetic stimulation and inhibition because of minimal spectral overlap in the excitation wavelength between the tools. In

Synthesis
All caged oxytocin analogs (cOT1, cOT2, and cOT3) were synthesized using solid-phase peptide synthesis. In short, resin was prepared using rink amide MBHA resin (1.5 mmol, 1.00 eq, Sub 0.55 mmol/g) in DMF agitated with N2 for 2 hours at 20°C. Then deprotection was carried out by adding 20% piperidine in DMF (30 mL) and agitated the resin with N2 for another 30 minutes. The resin was then washed with DMF (30 mL, 4 times) and filtered to obtain the resin. To couple the amino acids, a solution of HBTU (2.85 eq) and Fmoc-protected amino acid (3.00 eq) in DMF (20 mL) and DIEA (6.00 eq) was added to the resin and agitated with N2 for 30 minutes at 20°C. The resin was then washed with DMF (30 mL, 4 times). The deprotection and coupling steps were repeated to add each amino acid (including caged amino acids). After completion of peptide synthesis. The

Molecular Modeling
To understand how caging of oxytocin at different residues alter the peptide/protein interaction, molecular docking was performed for each caged oxytocin analog (cOT1, cOT2, cOT3) in comparison to oxytocin. The structure of oxytocin bound to OXTR, PDB: 7RYC was used as a template for the molecular docking experiments 43 . For each designed peptide, the geometry and structure were optimized by using a fast, Dreiding-like force field via Discovery Studio (Dassault System). The generation of the receptor grid was prepared by using oxytocin bound to OXTR structure (PDB: 7RYC) and was cleaned for docking. The defined dimensions of X, Y, and Z coordinates for site-specific docking of OXTR were -140.667, 134.000, and 113.361, respectively and the box size was 86 X 126 X 118 for all three dimensions was used. Each compound was docked using Autodock Vina (Schrodinger Maestro yielded similar results). Lastly, the van-der Waals (vdW) repulsion forces between the caged oxytocin analogs and OXTR were calculated using the "Show bumps" plugin, implemented in the PyMOL interface. Heatmaps of compound binding was analyzed and plotted using Prism 8 (GraphPad, San Diego, CA).

Photophysical and photochemical characterization
All measurements were performed in the dark or under red light. Samples were stored at -20 °C.
UV-Vis spectroscopy was performed using a Cary 60 UV-Vis Spectrophotometer equipped with a PCB 1500 High-Performance Peltier thermostat (Agilent Technologies, Santa Clara, CA). It is worth noting the uncaging in this experiment was done with a 405 nm laser at the highest power, which is not optimized (less than 10% efficiency) for uncaging the oNBZ cage group of cOT1 (Klán et al., 2012). We are confident that using a UV laser with a wavelength between 340 -385 nm would significantly improve uncaging efficiency. However, it is noteworthy that succeeding in uncaging cOT1 with 405 nm light broadens the access and utility of this tool because most standard microscope setups are equipped with 405 nm lasers. Additionally, this experiment demonstrates the potential for experiments where three wavelengths multiplexing could be useful. Here we uncaged oxytocin with UV light and tracked the cells in the red channel and calcium imaged in the green channel. To assess effects of oxytocin, a stable baseline was established for 5 minutes and 1 μM oxytocin (Tocris) in ACSF was washed on for 30 minutes. To assess effects of cOT1 uncaging, after baseline stability was achieved in ACSF, 1 μM cOT1 was washed in for another 5-10 minutes to maintain the stable baseline. Then a CoolLED pe-4000 illumination system was used to generate 365nm of UV light pulsed at 20Hz for 1 to 2 minutes over the slices through the objective. For further control experiment using OXTR antagonist OTA, OTA was continuously presented before the wash-in of cOT1 to the end of the recording. The whole set of experiments was conducted in dark as much as possible to avoid pre-emptive uncaging due to ambient UV light. Peak depolarization was measured as the mean membrane potential over a 2-minute period of highest membrane depolarization at least 10 minutes after uncaging of cOT1, subtracted from the mean membrane potential before the light stimulation. Input resistance and series resistance were monitored throughout the experiments, and recordings were rejected if series resistance increased to above 25MΩ or more than 20%.

In vitro electrophysiology
Experimental design and statistical analysis: Experiments were not performed blindly. In all cases, four or more animals with both sexes were used for each parameter collected and were pooled for analysis. Each recorded neuron came from one brain slice of one experimental animal. There was no repeated use of any brain slice. Individual sample sizes for slice patch clamp recording (n=number of neurons, included in each figure legend) are reported separately for each experiment.
All statistical analysis was performed using GraphPad Prism 9. Statistical comparisons before and after the uncaging of cOT1 were made using paired two-tailed Student's t-test. Statistical comparisons for different groups were made using one-way or two-way ANOVA and post hoc Turkey's test. Each statistical method is clearly stated in the result section or the figure legends.
All statistical tests were two sided. Data distribution was assumed to be normal, but this was not formally tested. Data are presented as mean ± sem. Individual data points are plotted in figures.
All raw data sets are openly accessible upon request. To assess the effects of oxytocin, a stable baseline was established for 3-10 minutes, and 1 μM oxytocin (Tocris, United Kingdom) in ACSF was washed on for 10-15 minutes, followed by a washout period. To assess the effects of cOT1, after baseline stability was achieved and following 5-10 minutes of cOT1 wash-in, 365 nm light was pulsed at 20-Hz for 1s over the recording chamber using a CoolLED pe-4000 illumination system. The peptide (cOT1) was pre-emptively kept in the dark as much as possible to avoid uncaging due to ambient UV light. Peak depolarization was measured as the mean membrane potential over the 2minute period of highest membrane depolarization at least 10 minutes after wash-in of cOT1, subtracted from the mean membrane potential during the 2 minutes of the baseline period immediately before initiation of drug wash-in.
Statistical analysis: Current clamp recordings were analyzed offline using Clampfit 10.7 (Molecular Devices, San Diego, CA). Recordings were excluded from analysis if the access resistance (Ra) changed >30% compared to baseline. Spontaneous PSCs were detected offline using pClamp software for event detection. Student's t-test was used to compare two groups using Prism 9.0 GraphPad software. Data is displayed as the mean +/-the standard error of the mean (s.e.m.).

Behavior
Pup-naïve female C57BL/6 wildtype virgin mice were bred and raised at NYU Grossman School of Medicine and kept isolated from both dams and pups until approximately five weeks old. Before implantation with an optofluidic cannula (details below), naïve virgins were prescreened for pup retrieval (detailed below) to exclude spontaneous retrievers or pup mauling. Typically, <30% of naïve virgins retrieve at least one pup or maul pups during pre-screening 10 . Dams were also prescreened to ensure they retrieved pups before using them in this study.
Optofluidic cannula implantation: In brief, a total of 30 female C57BL/6 wildtype mice at 5 weeks of age were anesthetized and implanted with 1 mm Optical Fiber Multiple Fluid Injection Cannula system (Doric Lenses, Canada) (part: OmFC_SM3_400/430-0.66_1.0mm_FLT_0.5mm) targeting the auditory cortical region-coordinates: AP: -2.8 mm and ML: -4.35 mm. The implant was attached to the skull with dental cement. Post-surgery the mice were group-housed (cannula was protected using a cap) for 4 -5 weeks to allow recovery from the surgery. Daily food intake and body weights were monitored. Then each animal was screened again for spontaneous pup retrieval for the baseline experiment. 7 animals were removed during recovery due to bodyweight drop (1 animal) or spontaneous pup retrieval post-surgery (6 animals).
Post hoc verification of implantation was conducted after the completion of behavior tests.
Animals were sacrificed and fixed with 4% paraformaldehyde via transcardial perfusion. Then, 100 μm brain slices were cut using a cryostat, and cannula and optic fiber tracks were viewed under a stereo microscope. Confocal images were taken using Zeiss LSM700 or LSM800 (Zeiss, White Plains, NY). Data from mice with incorrect cannula placement were excluded from the data analysis (2 animals).
Pup retrieval: A single test session consists of 10 individual trials of pup retrieval for an individual mouse. The pup retrieval test was performed as pre-screening before surgery, baseline, and postinfusion with caged oxytocin (cOT1). The mice were placed in a behavioral arena (38 × 30 × 15 cm) for prescreening or a customized behavioral arena setup of identical size for the baseline and post-infusion sessions. The mice were first given 20 minutes to acclimate to the behavioral arena before each testing session. Subsequently, at least 4 pups ranging from postnatal day 1-4 was placed in a corner of the arena under nesting material. A single pup was removed from the nest and placed in an opposite corner of the arena for each trial. The experimental mouse was given ten two-minute trials to retrieve the displaced pup back to the nest. If the displaced pup was not retrieved within two minutes, the trial was scored as a failure, and the pup was returned to the nest.
If the pup was retrieved back to the nest, the time of retrieval was recorded. This is repeated until ten trials are complete but each time removing a different pup from the nest. We consider reliable retrieval when the mouse successfully retrieves pups in at least two out of ten trials. After ten trials, the pups were returned to their home cage with their dam. An ultrasonic microphone (Avisoft, Germany) was used during all pup retrieval experiments to confirm that the isolated pups vocalized distress calls.
After the baseline test for pup retrieval was performed, each virgin mouse was cohoused with a dam and her pups. After 1 hour, the virgin mouse is placed back into the customized behavioral setup equipped for delivering fluid containing cOT1 and 365 nm light for uncaging using both a fluid microinjector (World Precision Instruments, United Kingdom) and an LED (Thor Labs 365 nm LED controlled by High-Power 1-Channel LED Driver with Pulse Modulation, Newton, NJ) respectively. All the female virgin mice were infused with cOT1 (50 μM in saline, 1.5 μl at 1 μl/min). Immediately after the infusion was complete, one group (n = 9) of the mice had the 365 nm LED light turned on (385 nm, 10 mW, 500 ms, 2 Hz) for the photorelease of native oxytocin and the other group (n= 12) had it off. The mice were then untethered from the LED and microinjector and allowed to complete the 20-minute acclimation period to the behavior arena before pup retrieval testing. Then the 10 trials of pup retrieval were carried out as described previously 10 . Pup retrieval was tested after infusion of cOT1 with or without light for the following time points: 1, 3, 6, 12, 18, 24, 36, 48, and 72 hours. Fisher's two-tailed exact test was used to compare the LED on group (uncaged) versus the LED off group (caged).