Since the highest expression of mOR was found in the IPL (Figure 1e, f), we initially probed the effect of a selective mOR agonist (DAMGO, 500 nM) on AMPA receptor-mediated transmission mediated by SP neurons that prominently innervate this subregion (Figure 1c). DAMGO application results in a significant decrease in light-driven (470 nM; typically, 10–50% arbitrary LED power equating to approximately 0.4–3.4 mW/mm²; CoolLED illumination system) AMPAR-mediated EPSCs in Tac1Cre:Ai32 mice that is partially reversed upon washout of the agonist (Figure 2a, b). This is accompanied by an increase in S2/S1 paired pulse ratio (PPR) suggesting the modulation is driven by presynaptic changes in release probability (Figure 2c). Thus, neurotransmitter release by this specific afferent input to the IPL is directly, negatively modulated like that seen in many other neural circuits within the brain (Reeves et al., 2022).

We next examined how mORs modulate mHb-IPN neurotransmission that occurs via the other major input mediated by cholinergic neurons. To this end, we employed ChatCre:Ai32 mice and focused on the IPR (Figure 3a) due to the relatively high levels of combined mOR expression and cholinergic neuronal innervation (Figure 1c, e, f) in this subregion. Under our basal experimental conditions and in agreement with previous studies (Souter et al., 2022; Singhal et al., 2025; Ren et al., 2022), light-evoked postsynaptic responses in the IPN mediated by mHb cholinergic neurons in response to either brief single or paired pulse stimulation are solely mediated by AMPARs as evidenced by complete pharmacological block by DNQX (refer to Figure 5—figure supplement 1a and Figure 7—figure supplement 1a–c). Remarkably, DAMGO application elicited a significant robust and reversible potentiation of light-evoked (470 nM; typically, 10–100% arbitrary LED power equating to approximately 0.4–6.9 mW/mm²; CoolLED illumination system) AMPAR EPSC amplitude (Figure 3a, b) in stark contrast to that seen with mHb Tac1-IPL glutamatergic signaling described previously (Figure 2a, b). Notably, the accompanying PPR changes were not consistent with an increase in release probability (Figure 3c), raising uncertainty as to the synaptic locus of the mOR effect.

In these experiments, we were ‘blind’ to the molecular identity of the postsynaptic IPR neuron. Since a major population of neurons in this IPN subregion is of the somatostatin (SST) GABAergic subtype (Ables et al., 2017; Hsu et al., 2013), it is likely that most of our recordings are from this subtype. Nevertheless, we also employed Chat-ChR2 transgenic mice containing fluorescently reported SST neurons (i.e. Chat-ChR2:SSTCre:Ai9; Figure 3d) to allow targeted recordings specifically from these neurons. Our data clearly show that DAMGO robustly increases AMPAR EPSCs impinging on IPR RFP+ SST neurons (Figure 3d, e) accompanied by a trending increase in PPR (Figure 3f). We also performed a series of additional experiments agnostic to the postsynaptic neuronal identity in IPC subdivision. Despite the relatively low expression levels of mOR in this subdivision of IPN (Figure 1e, f), we observed a robust increase in AMPAR EPSC amplitude at a similar extent to that seen in IPR (Figure 3—figure supplement 1), thus extending this remarkable potentiation of glutamatergic synaptic transmission to all IPN subfields where cholinergic neuronal terminals reside.

We considered non-specific actions of DAMGO such as those that could be mediated by a possible direct modulation of ChR2 itself to explain this non-canonical observation. However, our previous data demonstrating the reduction of glutamatergic neurotransmission mediated by SP neurons by DAMGO described (Figure 2) renders this possibility unlikely. Furthermore, a similar potentiation of cholinergic neuronal-mediated glutamatergic transmission in the IPN upon activation of GABA_B receptors has demonstrated a propensity for this circuit to undergo such modulation in response to a Gi-linked receptor (Koppensteiner et al., 2024; Zhang et al., 2016; Bhandari et al., 2021). Nevertheless, despite this precedent, we employed a pharmacological approach to further validate this unexpected result. We show that lower concentrations of DAMGO (100 nM) and alternative mOR agonists, Met-enkephalin (Met-enk, 3 µM) and morphine (10 µM), all induce potentiation of ChAT neuronal glutamatergic transmission to similar extents (Figure 3g). In addition, pre-treatment with the mOR selective antagonist (CTAP, 1 µM) completely prevents the DAMGO-induced response (Figure 3g). Together, the pharmacological battery of tests employing an experimental, an endogenous, and clinical/recreational mOR agonists and selective mOR antagonism clearly implicate this opioid receptor in mediating a non-canonical potentiation of AMPAR EPSC amplitude elicited by mHb cholinergic neurons in IPN.

In several recordings from both ‘blind’ patching of IPR neurons and in directly identified SST neurons, no measurable postsynaptic AMPA receptor EPSC could be elicited following light-evoked stimulation of cholinergic axons even with the maximal possible light intensity available (i.e. 100% arbitrary LED power; approximately 6.9 mW/mm²), and thus further interrogation was typically not performed. However, in a few instances, we nevertheless proceeded with DAMGO or Met-enk application, and surprisingly, in a subset of these recordings (9/15 cells tested), a robust and reversible emergence of a significant AMPAR-mediated EPSC was observed (Figure 3h, i). Thus, these data clearly reveal that mOR activation invokes a synaptic mechanism that, at its most extreme, results in an ‘un-silencing’ of ChAT neuronal glutamatergic transmission in IPN including in directly identified SST GABAergic neurons. Since PPR measurements are only valid if the same population of release sites are assayed before and after experimental manipulation, the additional engagement of a putative reluctant vesicular pool (Koppensteiner et al., 2024) by mOR activation could explain the discrepancy between the augmentation of EPSC amplitude and observed changes in PPR (Figure 3b, c, e, f). Future interrogation is warranted for definitive identification of the relative contributions of pre- and postsynaptic loci to the potentiation observed (but see Singhal et al., 2025).

To date, we have examined the effect of mOR on synaptic transmission during paired pulse light activation delivered at 20 Hz. Although mHb cholinergic neurons are capable of burst firing at much higher frequencies in response to afferent stimulation in vivo (Otsu et al., 2018), they typically exhibit low frequency (i.e. <10 Hz) intrinsic (i.e. independent of synaptic input) firing in vitro (Chung et al., 2023; Arvin et al., 2019; Görlich et al., 2013; Cho et al., 2020). In agreement with these previous studies, cholinergic neurons in ventral mHb (identified in ChatCre:Ai9 mice; Figure 1c) spontaneously elicit action potentials measured in cell-attached recordings at an average of ~5 Hz (range 2–10 Hz; Figure 4a). Therefore, we assessed the role of mOR activation on glutamatergic transmission mediated by cholinergic neurons elicited by light stimulation trains delivered at this frequency. Interestingly, even with this relatively low frequency paradigm, glutamatergic transmission is extremely labile in nature as evidenced by the large and rapid depression of AMPAR EPSC during the stimulus train (Figure 4b, c). Remarkably, activation of mORs essentially eliminates activity-dependent depression at this synapse (Figure 4b, c). This switch in transmission dynamics greatly facilitates the probability of excitatory postsynaptic potential-spike coupling (ES coupling) in response to each stimulus within the train (Figure 4d, e). Thus, these data clearly demonstrate that at physiologically relevant stimulus patterns, mORs serve to dramatically increase the fidelity of glutamatergic transmission in the IPR mediated by cholinergic neurons culminating in an enhanced recruitment of postsynaptic IPR neurons.

An intriguing inconsistency was observed during the conduction of our experiments regarding the DAMGO effect on light-evoked and spontaneous EPSCs (sEPSCs) impinging on IPR neurons. Under the experimental conditions employed, sEPSC events were exclusively mediated by AMPARs as evidenced by their virtually complete cessation upon DNQX application (Figure 5—figure supplement 1a). Remarkably, in contrast to the previously described robust potentiation of light-evoked AMPAR EPSCs, activation of mORs significantly attenuates sEPSC frequency in agreement with a recent study (Singhal et al., 2025), with no effect on amplitude (Figure 5—figure supplement 1b–d). Thus, in a single IPR postsynaptic neuron, opposing effects on light-evoked and spontaneous AMPAR EPSCs are evident (Figure 5—figure supplement 1e). Although these data do not directly identify the origin and neuronal subtype(s) responsible for the sEPSCs measured, the incongruent effect of mOR activation on evoked versus spontaneous events led us to consider the possible existence of an additional afferent system impinging on IPR neurons. Prevailing circuit schemas based on axonal arborization patterns portray a mutually exclusive, non-overlapping afferent input to IPN mediated by cholinergic and SP neurons to IPR/IPC versus IPL, respectively. However, in the current study by employing high-resolution imaging in Tac1Cre:Ai32 mice, a clear presence of synaptic bouton-like structures in the IPR is noted (Figure 5a, b). These structures do not co-localize with endogenous ChAT, but do express the glutamate vesicular transporter, VGluT1 (Figure 5a, b). This presence of putative presynaptic anatomical substrates points to the existence of a secondary input to IPR mediated by SP neurons distinct to the well-established one originating from mHb cholinergic neurons. Indeed, significant light-evoked AMPAR EPSCs in IPR neurons can be elicited in Tac1Cre:Ai32 mice with essentially similar amplitudes across all LED powers employed to that seen in ChatCre:Ai32 mice (Figure 5c). Thus, in contrast to recent ultrastructural EM analyses (Koppensteiner et al., 2024), these data indicate a previously unidentified functional glutamatergic input to IPR mediated by SP neurons. Interestingly, DAMGO application significantly depresses the SP neuronal glutamatergic output in IPR and increases PPR (Figure 5d–f) to a similar extent to that observed in IPL (Figure 2). This is in direct contrast to the role of mORs in positively modulating the ChAT neuronal glutamatergic transmission in the IPR (Figure 3a, b, d, e). Thus, for the first time, we demonstrate both SP and cholinergic input to the same subregion of IPN and that mOR activation elicits diametrically opposing effects on transmission mediated by these respective presynaptic neuronal populations.

Our functional studies thus far have been restricted to sexually mature, adult mice (>p40). mOR expression emerges in many brain structures during prenatal development largely overlapping with their final adult distribution profile by mid/late gestation (Zhu et al., 1998). In agreement, we reveal that mOR protein expression in IPR is present at appreciable levels from early postnatal stages (i.e. from P10 onwards; Figure 6a). This begs the question of whether modulation of habenulo-interpeduncular synaptic transmission described during these critical early epochs mirrors that seen in the adult. To examine this, we extended our analyses to earlier time points (P15–P40) and found that mOR modulation of glutamatergic transmission mediated by cholinergic neurons in the IPR undergoes a remarkable developmental regulation. This is characterized by an initial inhibitory effect on light-evoked AMPAR EPSCs transitioning to the previously described augmentation (Figure 6b, c, f). In contrast, the glutamatergic input to the IPL and the newly discovered input to the IPR elicited by SP neurons is not developmentally regulated with synaptic transmission being consistently inhibited by DAMGO application at all ages tested (Figure 6d–f). Thus, despite essentially similar levels of IPR mOR expression at the ages assayed (Figure 6a), at earlier development stages, opioids result in overall inhibition of afferent input to the IPR agnostic to afferent input with the emergence of the previously highlighted differential regulation of mHb cholinergic versus substance P neuronal output emerging during the late adolescent stages (Figure 6f).

In addition to mORs, this circuitry contains dense expression of nicotinic receptors located at pre- and postsynaptic sites (Chung et al., 2023; McGehee et al., 1995; Shih et al., 2014). However, the ability to reliably evoke postsynaptic nicotinic receptor (nAChR) EPSCs in response to physiologically appropriate stimuli has been challenging. Indeed, many studies have resorted to high frequency and prolonged stimulation (Ren et al., 2011) or bath/puff applications of nicotine to investigate nAChR function in this circuit (Zhao-Shea et al., 2013; Hsu et al., 2013; Arvin et al., 2019; Wei et al., 2022; Elayouby et al., 2021; Mondoloni et al., 2023). Despite the faithful co-expression of endogenous ChAT in ChatCre:Ai32 expressing terminals located in the IPR (Figure 7a), evoked transmission by brief light pulses (1–5 ms) from cholinergic neurons predominantly elicits pure AMPAR EPSC in postsynaptic IPR neurons under our basal conditions as described above and in agreement with previous observations (Souter et al., 2022; Ren et al., 2011; Singhal et al., 2025; Hsu et al., 2013) (see Figure 5—figure supplement 1a and Figure 7—figure supplement 1a-e). Additionally, DAMGO, at concentrations that strongly potentiate glutamatergic transmission, does not result in a measurable postsynaptic nAChR EPSC after AMPARs are pharmacologically silenced (Figure 7—figure supplement 1a–c). Furthermore, repeated light stimulation at either 25 or 50 Hz fails to elicit the emergence of any nAChR-mediated EPSCs in the presence of DNQX (Figure 7—figure supplement 1d, e). Finally, the potentiated ESPC following mOR activation is purely AMPAR mediated, as evidenced by complete block by DNQX (Figure 7—figure supplement 1f–h). Together, our functional analyses demonstrate that brief pulses of light (1–5 ms) in ChatCre:Ai32 mice do not result in sufficient ACh release, if any, to elicit measurable postsynaptic nAChR responses in response to high frequency stimulus regimens and even under conditions where mOR activation strongly enhances glutamatergic transmission.

Interestingly, cholinergic transmission at the neuromuscular junction can be boosted via potassium channel blockade with fampridine (4-aminopyridine) and has been clinically indicated in disorders such as myasthenia gravis and multiple sclerosis (Kostadinova and Danchev, 2019). Furthermore, low micromolar 4-AP in slices amplifies acetylcholine release, as assessed by Grab-ACh-mediated fluorescence in the IPN (Jing et al., 2018; Jing et al., 2020). Experimentally, these concentrations are relatively selective in inhibiting K⁺-channels of the Kv1 family. Probing the publicly available 10X single-cell RNAseq database provided by the Allen Brain Institute reveals that Kcna2 (the gene encoding Kv1.2) is the most prevalently expressed in the designated Chat Supertypes (Figure 1d and Figure 7b; mean expression levels including zero values = 9.1, 6.8, and 4.1 in Supertypes 0632, 0634, and 0635, respectively). Furthermore, spatial interrogation via in situ hybridization (Allen Brain; https://mouse.brain-map.org/gene/show/16263) demonstrates the presence of appreciable Kcna2 transcript levels in ventral habenula (Figure 7c) a region that is enriched in cholinergic neurons (Figure 1b, c). We therefore speculated that Kv1.2 block could result in enhancement of ACh release to perhaps reveal postsynaptic nAChR EPSCs in IPR neurons. Indeed, following the block of glutamatergic transmission (DNQX + APV), subsequent application of 4-AP (50–100 µM) or the more selective Kv1 channel antagonist, dendrotoxin-α (100 nM), results in the emergence of an EPSC elicited by brief single light pulses (Figure 7d) and in response to trains of 5 Hz stimulation (Figure 7e). This emergent EPSC is mediated by nAChRs as evidenced by ~80–90% block by the nicotinic receptor antagonist, DHβE (Mulle et al., 1991; Morton et al., 2018) (2 µM; Figure 7f).

Interestingly, this experimental approach allows for the calculation of nAChR;AMPAR peak amplitude ratio enabling an assessment of fast excitatory cholinergic transmission normalized to stimulus intensity required to produce a given AMPA EPSC response in an individual postsynaptic IPR neuron. Thus, the size of the nAChR response can be directly compared across multiple recordings from differing slices, mice, and experimental conditions. We first assessed the extent to which both 4-AP and DTX-α reveal light-evoked nAChR EPSCs, finding no significant difference in the nAChR:AMPAR ratio, demonstrating that the specific Kv1.1/Kv1.2 channel antagonist unmasks nAChR-mediated synaptic transmission in a similar manner to low concentration 4-AP (Figure 7g). This, in combination with the predominant expression of Kcna2 (Figure 7b and c) and relative lack of other Kcna transcripts (Figure 7b) in mHb ChAT neurons, confirms our hypothesis that Kv1.2 plays a central role in unmasking nAChR-mediated synaptic transmission. In addition to ChatCre:Ai32 mice, we also employed the Chat-ChR2 mouse. Not surprisingly, nAChR EPSCs could also be elicited in this mouse line following 4-AP application (Figure 7g). However, the measured nAChR:AMPAR ratio is significantly higher than that seen in ChatCre:Ai32 mice (Figure 7g), indicating a skew toward more prominent cholinergic transmission when compared to that mediated by glutamate. It is unclear as to the exact reason for this divergence, but it may be related to increased levels of the vesicular Ach transporter (vAChT) expression observed in Chat:ChR2 mice (Kolisnyk et al., 2013). Although Tac1 mHb neurons also express Kcna2 transcript, albeit lower than in the ChAT neurons (data not shown), 4-AP application following block of glutamatergic transmission does not result in emergence of any additional EPSC in response to single stimuli in Tac1Cre:Ai32 mice (Figure 7g). Finally, application of a cholinesterase inhibitor (ambenonium, 50 nM) markedly prolongs the EPSC waveform with minimal effect on amplitude (Figure 7h). Thus, together, these results corroborate the role of Kv1.2 as a molecular brake, removal of which results in the emergence of habenulo-interpeduncular synaptic transmission via postsynaptic nAChR signaling on IPR GABAergic neurons.

Having established the ability to reliably assess an otherwise reluctant cholinergic transmission, we next investigated how nAChR-mediated signaling in the IPR is impacted by mOR activation. As with AMPAR-mediated EPSCs, DAMGO produces a marked potentiation of light-evoked nAChR EPSC amplitude in IPR, with the augmented EPSC being completely blocked by DHβE (Figure 8a, b). The mOR-mediated potentiation of both the glutamatergic and cholinergic output is essentially similar (Figure 8i). Till now, we have utilized optogenetic approaches to assay synaptic transmission in the IPN. Interestingly, following acute block of Kv1, nAChR EPSCs could also be elicited via electrical stimulation (Figure 8c). Further, DAMGO reliably potentiated both electrical and light-evoked nAChR EPSCs in a single IPR neuron to similar extents, a response that is effectively blocked by DHβE (Figure 8c–e). Finally, we tested the molecular specificity of the DAMGO response employing Chat-ChR2:global Oprm1 knockout mice. In these mice, DAMGO application is ineffective in augmenting glutamatergic or cholinergic-mediated neurotransmission in IPR, while the potentiation upon GABA_B receptor activation previously described (Koppensteiner et al., 2024; Bhandari et al., 2021; Melani et al., 2019) remains intact (Figure 8f–h; note that in these experiments, CGP55845A, which is routinely included in all other experiments, was omitted). This genetic approach complements our previous pharmacological data (Figure 3g) to directly implicate mORs in the potentiation of mHb-IPN transmission. The overall effects of mOR activation on synaptic transmission mediated by the distinct mHb afferent systems in the various subdivisions of IPN tested are summarized in Figure 8i.

While conducting our current study, we became aware of one from the Hnasko laboratory also investigating the role of opioids on habenulo-interpeduncular synaptic transmission (Singhal et al., 2025). Their observations regarding a similar mOR-mediated potentiation of glutamatergic signaling by mHb cholinergic neurons point to the reproducibility of this unexpected, novel, and potentially important circuit mechanism.

All procedures in the current study involving experimental mice were conducted in strict accordance with an active animal protocol (ASP#23-045) approved by the Animal Care Use Committee of the National Institute of Child Health and Human Development. All transgenic mice were attained from The Jackson Laboratory (ME, USA) and were as follows: ChatCre (strain #031661), Tac1Cre (strain #021877), SSTCre (strain #018973); Ai9 (strain #007909), Ai32 (strain #024109), Chat-ChR2-EYFP (strain #014546); Oprm1 KO (strain #007599). For conditional expression of td-Tomato (Ai9) or ChR2 (Ai32), both male and female breeders were homozygous, and all immunocytochemical or electrophysiological experiments were performed in their progeny that were heterozygous for both the Cre and floxed alleles. Heterozygous Chat-ChR2 mice were used throughout the study either alone or crossed with the Oprm1 KO mouse (homozygous) or SSTCre:Ai9 (heterozygous for both alleles). Both male and female mice were investigated, and the data were pooled.

Pharmacological reagents used in this study are as follows: DNQX (Cat. No. 2312/10), DL-APV (Cat. No. 0105/10), CGP55845A (Cat. No. 1248/10), Picrotoxin (Cat. No. 1128), bicuculline (Cat. No. 0109/10), DAMGO (Cat. No. 1171/1), CTAP (Cat. No. 1560/1), and DHβE (Cat. No. 2349) were all purchased from Biotechne/Tocris (MN, USA); Dendrotoxin-α (Cat No. D-350) was purchased from Alomone Labs (Jerusalem, Israel). Met-enkephalin (Cat. No. M6638) was purchased from MilliporeSigma (MA, USA). Ambenonium (Cat. No. sc-203507) was purchased from Santa Cruz Biotechnology (TX, USA). Morphine Sulfate (Cat. No. NDC: 0641-6125) was from Hikma (NJ, USA) and attained via the NIH Division of Veterinary Resources (DVR) Pharmacy.

Primary and corresponding secondary antibodies used in this study for immunocytochemistry and the working dilutions employed are as follows: Guinea pig anti-RFP (Synaptic Systems, Cat. No. 390005, 1:500) with CF555-conjugated Donkey anti-Guinea Pig IgG (Biotium Cat. No. 20276, 1:1000). Chicken anti-GFP (AvesLabs, Cat. No. GFP-1010, 1:1000) with CF488-conjugated Donkey anti-Chicken IgY (Biotium Cat. No. 20166, 1:1000 working dilution). Rabbit anti-MOR (ABCAM, Cat. No. ab134054, 1:500) with CF555-conjugated Donkey anti-Rabbit IgG (Biotium cat#20038, 1:1000). Goat anti-ChAT, Millipore (Cat. No. AB144P, 1:1000) with CF555-conjugated Donkey anti-Goat IgG (Biotium, Cat. No. 20039, 1:1000 working dilution).

Mice (P20 and older) were perfused trans-cardinally using 4% PFA and dissected brain tissues were post-fixed in 4% PFA for 24 hr at 4°C. For P10 mice, brains were removed and drop-fixed in 4% PFA for 24 hr at 4°C. P10 mouse brains were not perfused but brains were removed and drop-fixed as stated above. Fixed brain tissues were thoroughly washed in 1x phosphate buffer (PB) followed by cryopreservation using 30% sucrose. 50 μm coronal sections were made on a frozen microtome. To perform floating section IHC, brain slices were washed with 1x PB at room temperature for 1 hr with two to three changes of 1x PB followed by blocking and permeabilizing in Blocking Solution (1x PB +10% donkey serum + 0.5% Triton X-100) at room temperature for at least 2 hr. Blocked brain slices were incubated in primary antibodies, which were diluted to working concentration using Antibody Solution (1x PB + 1% donkey serum + 0.1% Triton X-100), at 4°C for 24–48 hr. After washing with 1x PB at room temperature for 15 min with three repeats, brain slices were incubated in secondary antibodies diluted with Antibody Solution at room temperature for 1 hr, followed by DAPI staining at 1 µg/ml for additional 15 min. After being washed with 1x PB at room temperature for 15 min with three repeats, brain slices were mounted on gelatin-coated slides followed by air drying, covered with No. 1.5 cover slides ProLong Glass Antifade Mountant (Therm Fisher Scientific, Cat#P36984), cured in darkness overnight before imaging. Confocal images were acquired on a Zeiss LSM900 using ZenBlue software. 10x and 20x confocal images of whole habenula and IPN were acquired as tiles with z-steps of 3 µm × 5 and 0.9 µm × 20, respectively. Airy scan images were acquired at 63x using multiplex 2Y mode to achieve super resolution. 2 µm depth of images were captured at z-steps of 0.2 µm × 10, with or without tiling. All comparable images of tissues with different genotypes or developmental stages were acquired using the same light path configuration, channel input, gain, and offset. Images were imported into Fiji and Adobe Photoshop for processing and densitometry measurements where applicable.

P15–P91 mice (for details refer to the results section) were anesthetized with isoflurane, and the brain removed in ice-cold partial sucrose substituted ASCF (ssACSF) solution containing (in mM): 90 sucrose, 80 NaCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 3.5 KCl, 4.5 MgCl₂, 0.5 CaCl₂, 10 glucose, saturated with 95% O₂ and 5% CO₂ (pH 7.4; osmolarity 300–310 mOsm). Coronal sections containing either the mHb or IPN (270–300 µm) were cut using a VT-1000S vibratome (Leica Microsystems, Bannockburn, IL) in ice-cold ssACSF. The slices were allowed to recover in the ssACSF at 31–33°C for 20–30 min followed by an additional 20 min at room temperature. Slices were then transferred to our standard extracellular solution (ACSF) of the following composition (in mM): 130 NaCl, 24 NaHCO₃, 3.5 KCl, 1.25 NaH₂PO₄, 2.5 CaCl₂, 1.5 MgCl₂, and 10 glucose, saturated with 95% O₂ and 5% CO₂ (pH 7.4; osmolarity 300–310 mOsm) for storage until electrophysiological recording. Slices were transferred to an upright Olympus BX51WI microscope and visualized with infrared differential interference contrast microscopy and perfused (2–4 ml/min) with the above extracellular solution. Unless otherwise indicated, the recording extracellular solution was routinely supplemented with 2 µM CGP55845A hydrochloride (Biotechne/Tocris Cat#1248), 50 µM picrotoxin (Biotechne/Tocris Cat#1128), and 10 µM bicuculline methobromide (Biotechne/Tocris Cat#0131).

All recordings were performed at 31–33°C with electrodes (3–5 MΏ) pulled from borosilicate glass (World Precision Instruments, Sarasota, FL) filled with one of two intracellular solutions (in mM): (1) 135 CsMeSO₄, 8 KCl, 4 MgATP, 0.3 NaGTP, 5 QX-314, 0.5 EGTA, and 3 mg/ml biocytin. (2) 150 K-gluconate, 3 MgCl₂, 0.5 EGTA, 2 MgATP, 0.3 Na₂GTP, 10 HEPES, and 3 mg/ml biocytin. pH was adjusted to 7.3 with KOH or CsOH and osmolarity adjusted to 285–295 mOsm. Whole-cell patch-clamp recordings were made using a Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Signals were filtered at 4–10 kHz and digitized at 10–20 kHz (Digidata 1322A and pClamp 10 Software; Molecular Devices, Sunnyvale, CA). For cell-attached recordings, pipettes were filled with our standard extracellular solution.

For optogenetic stimulation of ChR2 light stimuli at a wavelength of 470 nm and duration of 1–5 ms were delivered through the 40X water immersion objective using a CoolLED pE-4000 Illumination system (Andover, UK). Arbitrary LED power typically ranged between 10% and 100% corresponding to ~0.4–6.9 mW/mm² as measured with a digital power meter (Thor Labs, NJ, USA). Where applicable, electrical stimulation (0.1ms; typically, 100–700 µA) was performed by placement of a tungsten bipolar electrode in the IPR (World Precision Instruments, FL, USA; Catalog # TST33A05KT) and a constant current stimulus isolator (World Precision Instruments; Catalog # A385 or A365). sEPSC analyses were performed on post hoc filtered episodic traces (Clampfit; Bessel 8-pole, –3 db cutoff 1000 Hz). Frequency of sEPSCs and average sEPSC amplitude were determined over five consecutive sweeps totaling a period of 4 s (i.e. 0.8 s per sweep). sEPSCs were detected via a thresholding procedure set at four times the RMS baseline noise. Note that only neurons with a baseline sEPSC frequency of >8 Hz were selected for further interrogation. Analyses of electrophysiological data were performed in Clampfit (Molecular Devices, Sunnyvale, CA) and curated in Microsoft Excel spreadsheets.

We utilized the Allen Institute dataset of the single-cell transcriptomes of ~4 million cells across the entire mouse brain (Yao et al., 2023). To probe the region and genes of interest, we adapted Python notebooks provided for accessing the Allen Brain Cell (ABC) Atlas (https://alleninstitute.github.io/abc_atlas_access/intro.html). The log₂-normalized expression matrices were downloaded and filtered to reach the cell types of interest within the habenula class ‘17 MH-LH Glut’ (~10.8 k cells). For the current study, we focused our analyses on the mHb Subclass ‘145 MH Tac2 Glut’ (~8k cells). Within this subclass, we probed the five supertypes ‘0632–0636 MH Tac2 Glut’, querying the genes of interest: Chat, Tac1, Slc17a7, Kcna1-6, and Oprm1. Chat supertypes were considered those with mean Chat expression including zero values >3 (0632, 0634, and 0635; 6803 cells), while Tac1 supertypes were defined as those with mean Tac1 expression including zero values >3 (0633; 878 cells). Violin and dot plots depicting gene expression levels were created in GraphPad Prism.

Tests for normality (Shapiro–Wilk test) were performed on all datasets to be compared prior to statistical analyses and appropriate parametric (unpaired or paired t-tests) or non-parametric (Mann–Whitney U or Wilcoxon signed rank tests) were accordingly performed. Exact p-values are stated within the figures and values of p < 0.05, p < 0.01, or p < 0.001 were deemed statistically significant and designated with 1, 2, or 3 asterisks, respectively. Error bars in the figures represent standard error of the means unless otherwise indicated. Box whisker plots were constructed as follows: symbol denotes mean value; line represents the median value; lower and upper box limits represent 25th and 75th percentiles, respectively. Lower and upper whiskers represent minimal and maximal data points, respectively.