All experimental protocols and animal studies were performed in accordance with the approved guidelines of the Institutional Animal Care and Use Committees of the Research Institute of Environmental Medicine, Nagoya University, Japan (Approval numbers: R210096 and R210729). All efforts were made to decrease the usage of animals and to minimize the affliction and pain of animals. Animals were maintained on a 12-h light dark cycle (light period: 08:00–20:00; dark period: 20:00–08:00), with free access to food and water.

Adult male and female mice between 8 and 16 weeks of age were used in this study. To specifically label the Gad67-positive neural population in the VTA we used Gad67-Cre mice [Gad1tm2(cre)Tama]. To visualize DRN_5–HT, we first bred tryptophan hydroxylase 2 (Tph2)–tetracycline transactivator (tTA) mice [Tg(Tph2-tTA)1Ahky] with tetracycline operator sequence (TetO) Yellow-Cameleon Nano50 (YC) mice (Actb^{TM 2}.^1Kftnk) to generate Tph2-tTA;TetO-YC bigenic mice in which Tph2 [which encodes an enzyme that synthesizes 5-HT (Walther et al., 2003)] promoter drives tTA (tetracycline-controlled transactivator) expression in 5-HT neurons. tTA then binds to the TetO sequence and drives expression of YC in 5-HT neurons (Kanemaru et al., 2014; Ohmura et al., 2014; Chowdhury and Yamanaka, 2016; Mukai et al., 2020). These bigenic mice were then crossed with Gad67-Cre mice to finally generate Gad67-Cre; Tph2-tTA; TetO YC trigenic mice. To generate Gad67-Cre; Tph2-tTA bigenic mice, we bred Gad67-Cre mice with Tph2-tTA mice.

Adeno-associated virus (AAV) vectors were produced using the AAV Helper-Free system (Agilent Technologies, Inc., Santa Clara, CA, United States). The virus purification method was modified from a published protocol (Inutsuka et al., 2016). pAAV plasmid including a gene of interest, pHelper and pAAV-RC (serotype 9) were mixed with 0.3 M CaCl₂, followed by the addition of 2× HEPES buffered saline in a dropwise manner with this solution during continuous vortex mixing. For transfection, the mixture of calcium-phosphate with DNA was added to cultured HEK-293 cells in Dulbecco’s Modified Eagle’s Medium-high glucose medium (DMEM) (Sigma-Aldrich, Merck, Darmstadt, Germany). After 3 days, transfected cells were collected and centrifuged. The pellet was suspended with phosphate buffered saline (PBS). The cell suspension was then freeze-thawed four times with the following cycle: frozen at −80°C for 15 min, incubated at 37°C for 10 min and vortexed at room temperature (RT) for 1 min. Then, benzonase nuclease (0.025 U, Merck, Darmstadt, Germany) was added to the cell suspension solution and it was incubated at 37°C for 30 min. The solution was centrifuged twice at 16,000 rpm for 10 min. The supernatant was used as the AAV-containing solution. Quantitative PCR was performed to measure the number of particles of purified AAV. The AAV was aliquoted and stored at −80°C until use.

At 8–16 weeks of age, male and female mice (Gad67-Cre, Gad67-Cre;Tph2-tTA;TetO YC and Gad67-Cre;Tph2-tTA) received either unilateral (in DRN) or bilateral (in VTA) injection of AAV9-CMV-FLEX-hrGFP (200 nl, 7.45 × 10¹² copies/ml), AAV9-CMV-FLEX-ChR2(ET/TC)-EYFP (200 nl, 1.5 × 10¹³ copies/ml), AAV9-CMV-FLEX-tdTomato (200 nl, 2 × 10¹² copies/ml) or AAV9-TetO(3G) tdTomato (200 nl, 6.6 × 10¹¹ copies/ml) under 1.0–1.2% isoflurane (Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) anesthesia (MK-A100W, Muromachi Kikai Co., Fukoaka, Japan). Injection sites were as follows: from bregma -3.3 mm, lateral ±0.5 mm, ventral -4.2 mm for the VTA and from bregma -4.5 mm, lateral 0 mm, ventral -3.1 mm for the DRN.

Mice were deeply anesthetized with isoflurane (Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) and transcardially perfused with 20 ml of ice-cold saline, followed by 20 ml of ice-cold 10% formalin (Fujifilm Wako Pure Chemical Industries, Osaka, Japan). Brains were removed and post-fixed in a 10% formalin solution at 4°C overnight. Brains were then cryoprotected in 30% sucrose in PBS at 4°C for at least 2 days. After that, brains were frozen in embedding OCT compound (Sakura Finetek, Osaka, Japan) at −80°C, and coronal brain slices of 40 or 80 μm thickness were made using a cryostat (CM3050-S, Leica Microsystems K.K., Wetzlar, Germany) at −20°C. Slices were placed in PBS containing 0.02% NaN₃ at 4°C until staining. For staining, coronal brain slices were washed 3 times with blocking buffer (1% BSA and 0.25% Triton-X in PBS) for 10 min and then incubated with primary antibodies at 4°C overnight. For Gad67 staining, slices were incubated with the primary antibody at 4°C for 2 days. After washing 3 times with blocking buffer for 10 min, slices were then incubated with secondary antibodies for 1 h at RT. For Gad67 staining, slices were incubated with secondary antibody for 2 h at RT. After another 3 washes with blocking buffer, slices were stained with 4 μM DAPI (4′,6-diamidino-2-phenylindole, Wako Pure Chemical Industries, Osaka, Japan) (except for Figure 7B) for 10 min at room temperature. Finally, slices were washed 3 times with blocking buffer and then mounted and examined with a fluorescence microscope (BZ-X710, Keyence, Osaka, Japan) and/or confocal microscope (LSM 710, Zeiss, Jena, Germany). Every fourth coronal brain section of both hemispheres was used for cell counts.

Primary and secondary antibodies were diluted in the blocking buffer at the following concentrations: anti–tyrosine hydroxylase (TH) rabbit antibody (AB152, Millipore, Darmstadt, Germany) 1/1000, anti-Gad67 mouse antibody (MAB5406, Millipore, Darmstadt, Germany) 1/250, anti-Tph2 sheep antibody (AB1541, Millipore, Darmstadt, Germany) 1/250, anti-green fluorescent protein (GFP) chicken antibody (Aves labs, Inc., Davis, CA, United States) 1/1000, CF 350-conjugated anti-sheep antibody (20148, Biotium, Fremont, CA, United States) 1/1000, CF 488-conjugated anti-chicken antibody (20166, Biotium, Fremont, CA, United States) 1/1000, CF 594-conjugated anti-rabbit antibody (20152, Biotium, Fremont, CA, United States) 1/1000, CF 594-conjugated anti-sheep antibody (20156, Biotium, Fremont, CA, United States) 1/1000 and CF 647-conjugated anti-mouse antibody (20046, Biotium, Fremont, CA, United States) 1/1000.

Adult male and female mice between 10 and 24 weeks of age were used for electrophysiological recording experiments. After 3–4 weeks of AAV injection, mice were deeply anesthetized using isoflurane (Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) and decapitated. Brains were rapidly isolated and chilled in ice-cold bubbled (95% O₂ and 5% CO₂) cutting solution (in mM: 110 K-gluconate, 15 KCl, 0.05 EGTA, 5 HEPES, 26.2 NaHCO₃, 25 glucose, 3.3 MgCl₂, 0.0015 3-[(±)-2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP)]. Coronal brain sections of 300-μm thickness were made using a vibratome (VT-1200S, Leica, Tokyo, Japan). The slices were incubated in a bubbled (95% O₂ and 5% CO₂) bath solution (in mM: 124 NaCl, 3 KCl, 2 MgCl₂, 2 CaCl₂, 1.23 NaH₂PO₄.2H₂O, 26 NaHCO₃, 25 glucose) at 35°C for 30 min following another 30-min incubation at RT in the same solution. After the incubation period, the brain slice was placed in a recording chamber (RC-26G, Warner Instruments, Hamden, CT, United States) that was perfused with a bubbled (95% O₂ and 5% CO₂) bath solution at a rate of 1.5 ml/min using a peristaltic pump (Miniplus3, Gilson, United States). An infrared camera (C3077-78, Hamamatsu Photonics, Hamamatsu, Japan) was installed in an upright fluorescence microscope (BX51WI, Olympus, Tokyo, Japan), together with an electron multiplying charge-coupled device camera (Evolve 512 delta, Photometrics, Tucsaon, AZ, United States), and both images were visualized on separate monitors. Glass micropipettes were prepared from borosilicate glass capillaries (GC150-10, Harvard Apparatus, Cambridge, MA, United States) using a horizontal puller (P1000 Sutter Instrument, Novato, CA, United States), maintaining a pipette resistance of 4–8 MΩ. During recording from ChR2-eYFP–expressing VTA_Gad67+ and DRN_Gad67+ neurons, patch pipettes were filled with a KCl-based internal solution (in mM: 145 KCl, 1 MgCl₂, 10 HEPES, 1.1 EGTA, 2 Mg-ATP, 0.5 Na-GTP; brought to pH 7.3 with KOH), with an osmolality between 280 and 290 mOsm. A K-gluconate–based internal solution (in mM: 138 K-gluconate, 10 HEPES, 8 NaCl, 0.2 EGTA, 2 Mg-ATP, 0.5 Na₂-GTP; brought to pH 7.3 with KOH), with an osmolality between 285 and 290 mOsm was used while recording from CeA, DRN and LC non-specific neurons, as well as from DRN_5–HT, DRN_non–5–HT, and DR_Gad67+ neurons in current clamp mode and from a few DRN_non–5–HT neurons in loose cell-attached mode. After confirming that the cell expressed green (hrGFP/eYFP), red (tdTomato) or YC, the pipette was moved toward the cell. A positive pressure was applied manually in the patch pipette. This pressure was released when the tip of the pipette touched the cell membrane and a gigaseal was formed. Then, the patch membrane was ruptured by briefly applying strong suction to form a whole-cell configuration. During recording, electrophysiological properties were continuously observed using the Axopatch 200B amplifier (Axon Instrument, Molecular Devices, San Jose, CA, United States). Output signals were low-pass filtered at 5 kHz and digitized at a 10-kHz sampling rate. Patch clamp data were recorded through an analog-to-digital converter (Digidata 1550A, Molecular Devices, San Jose, CA, United States) using pClamp 10.7 software (Molecular Devices, San Jose, CA, United States). Blue and yellow light (with a wavelength of 475 ± 17.5 and 575 ± 12.5 nm, respectively) was generated by a light source that used a light-emitting diode (Spectra light engine, Lumencor, Beaverton, OR, United States) and was directed to the microscope stage with a 1.0 cm diameter optical fiber. Coronal brain slices were illuminated through the objective lens of the fluorescence microscope. In whole-cell voltage clamp recordings at a −60 mV holding potential, QX-314 (1 mM) was added to the KCl-based pipette solution. A six-channel perfusion valve controller (VC-6, Warner Instruments, Holliston, MA, United States) was used for local substance perfusion.

Immunostaining data were quantified, analyzed with ImageJ (United States National Institutes of Health) and BZ-X Analyzer (Keyence BZ-X710 microscope, Itasca, IL, United States) software. Electrophysiological data were analyzed by Clampfit10 (Molecular Devices, San Jose, CA, United States). Statistical analyses were performed using OriginPro 2020 software. Graphs were generated using Canvas X (Canvas GFX Inc., Boston, MA, United States). Simple comparisons of the means and standard errors of the mean (SEMs) were performed using a Student’s t-test. Multiple comparisons of means and SEMs were performed by a one-way ANOVA (with/without repeated measures) analysis followed by Tukey’s test. p values < 0.05 were considered significant.

To visualize VTA_Gad67+ neurons, Gad67-Cre mice were bilaterally injected with a Cre-dependent AAV expressing humanized renilla GFP (hrGFP) into the VTA (Figure 1A). Two weeks after injection, many hrGFP-expressing neurons were observed in the VTA (Figure 1B). We used every fourth section of 40 μm coronal brain slices where double immunostaining for Gad67 (a marker of GABAergic neurons) and TH (a marker of DA neurons) confirmed that most of the hrGFP-expressing neurons in the VTA colocalized with Gad67, but not with TH (Figure 1B). Of all the VTA neurons, 66.2 ± 2.3% were TH-positive, 33.6 ± 2.3% were hrGFP-positive and only 0.2 ± 0.1% were co-labeled with TH and hrGFP (Figure 1C). We counted 1140 ± 58 neurons per animal (n = 4 mice).

Next, we performed slice patch clamp recording to assess the activity of VTA_Gad67+ neurons upon application of various substances. Coronal brain sections of hrGFP-expressing VTA_Gad67+ neurons from Gad67-Cre mice were prepared (Figure 1D). By using a six-channel perfusion valve controller, substances were sequentially and locally applied for 4 s under a loose cell-attached mode recording. To minimize mechanical or any other indirect effects on the recording cell by local application of the substance, artificial cerebrospinal fluid (aCSF) was applied before and after substance application (Figures 1E,G). At first, glutamate (Glu, 100 μM) and GABA (100 μM) (Figures 1E,G) were applied to confirm responsiveness and positioning of the local application. After that, serotonin (5-HT, 50 μM), noradrenaline (NA, 100 μM), dopamine (DA, 100 μM), histamine (HA, 100 μM), carbachol (CCh, 100 μM), and orexin A (OX-A, 100 nM) were applied in random order. We found that 5-HT, DA and HA application significantly reduced, whereas CCh increased, the firing rate (Figures 1F,G). Surprisingly, the wake-promoting neurotransmitters, NA and OX-A had no significant effect on firing frequency (Figures 1F,G). We then counted the fraction of cells that responded to substance application. We considered cells to be responsive if they showed a ≥20% increase or decrease in firing frequency during the 4 s substance application compared with before application. We found responses in 10/11 (90.9%) cells with 5-HT, 4/24 (16.7%) cells with NA, 10/18 (55.6%) cells with DA, 8/21 (38.1%) cells with HA, 11/19 (57.9%) cells with CCh, 3/22 (13.6%) cells with OX-A, 23/27 (85.2%) cells with Glu, 44/44 (100%) cells with GABA and 3/42 (7.1%) cells with aCSF (Figure 1G, pie charts). These results indicated that VTA_Gad67+ neurons responsiveness to substances was not uniform.

Next, we sought to reveal the downstream target neurons of VTA_Gad67+ neurons. We bilaterally injected a Cre-dependent AAV carrying the blue light–gated cation channel channelrhodopsin-2 (E123T/T159C) (ChR2) (Berndt et al., 2011) into the VTA of Gad67-Cre mice (Figure 2A). Immunohistochemical study confirmed that ChR2 fused with eYFP was expressed exclusively in the VTA (Figure 2B). In vitro slice patch clamp recording from ChR2-expressing VTA_Gad67+ neurons confirmed that a 10-Hz blue light pulse depolarized and significantly increased the spontaneous firing frequency by 919 ± 277% compared with pre-stimulation values (Figures 2C–E). VTA_Gad67+ neurons densely projected to the distal part of the brain, including to the CeA, DRN, and LC, consistent with our previous study (Figure 2F; Supplementary Figures 1A,B; Chowdhury et al., 2019). We then recorded from neurons in the CeA, DRN, and LC by stimulating the VTA_Gad67+ nerve terminals. Current clamp recording showed that only the recorded neurons of the DRN were hyperpolarized and there was a complete cessation of spontaneous firing following a 20-Hz blue light stimulation, whereas activity in CeA or LC neurons was not affected (Supplementary Figures 1C–K). We then focused on the DRN—we randomly selected cells in the DRN and applied different light frequencies, i.e., 5-, 10-, and 20-Hz blue light stimulation in whole-cell current clamp mode (Figure 2G). We found that blue light decreased the firing rate and hyperpolarized the membrane potential in a frequency-dependent manner. However, stimulation with the 20-Hz yellow light showed no effect on firing frequency (Figures 2H–J).

To further reveal the mechanism of action, we recorded from neurons in the DRN in whole-cell voltage clamp mode at a -60 mV holding potential. Activation of VTA_Gad67+ nerve terminals in the DRN induced post-synaptic currents (PSCs) in response to a 5-ms blue light pulse (Figures 3A,B). These light-induced PSCs were blocked by a GABA_A receptor antagonist gabazine (10 μM), but not by AP5 (50 μM) and CNQX (20 μM), which are antagonists for the NMDA (N-methyl-D-aspartic acid) and AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) glutamate receptor types, respectively (Figures 3B,C). The mean response delay from light onset was 3.5 ± 0.2 ms (Figure 3D). To rule out the effect of indirect input, we applied tetrodotoxin (TTX, 1 μM), a voltage-gated sodium channel blocker, which blocked the light-induced PSCs (Figures 3E–G). The combination of TTX and 4AP (4-aminopyridine, 1 mM), a voltage-gated potassium channel blocker, rescued the light-induced PSCs. These were again blocked by gabazine but not by CNQX (Figures 3F,G). We found the mean response delay was 3.8 ± 0.2 ms (Figure 3H). We recorded 35 cells from 5 mice (included in Figure 2I and Supplementary Figure 1H), where all the cells (35/35, 100%) responded to blue light stimulation (Figure 3I). Together, these results confirmed that DRN neurons were directly innervated and inhibited by VTA_Gad67+ neurons through the neurotransmitter GABA.

It has been reported that there are different subtypes of neurons in the DRN (Huang et al., 2019). Among them, 5-HT neurons are a major population and known to be involved in many physiological functions. Thus, the DRN neurons we recorded in the earlier experiment might have been serotonergic. To confirm this, we generated trigenic mice—Gad67-Cre;Tph2-tTA;TetO YC—in which tTA was exclusively expressed in 5-HT neurons under the control of the Tph2 promoter (Gossen et al., 1995). tTA binds to the TetO sequence and induces the YC expression (Tanaka et al., 2012). We then bilaterally injected a Cre-dependent AAV carrying ChR2 into the VTA of trigenic mice (Figure 4A). Immunostaining results confirmed that ChR2 fused with eYFP was exclusively expressed in the VTA (Figure 4B). In vitro slice patch clamp recording from ChR2-expressing VTA_Gad67+ neurons confirmed that these neurons were depolarized by a 10-Hz blue light pulse, and this significantly increased the spontaneous firing frequency by 908 ± 289% compared with pre-stimulation values (Figures 4C–E). Since YC is composed of yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP), we could clearly isolate the expression of YFP versus CFP by confocal imaging. Although CFP co-labeled with YFP, the nerve terminals from VTA_Gad67+ neurons expressed only YFP and not CFP (Figure 4F, white arrowheads). We stained every fourth 40 μm section of DRN-containing brain slices with an anti-Tph2 antibody to label 5-HT neurons. Colocalization of YC and TPH2-positive cell bodies confirmed that 5-HT neurons expressed YC (Figure 4F, white arrows). Cell counting indicated that 58.4 ± 2.6% of cells were both TPH2- and YC-positive, 41.2 ± 2.7% were only positive for TPH2 and 0.4 ± 0.1% were only YC-positive (Figure 4G). We then recorded from the YC-expressing DRN5-HT neurons by stimulating VTA_Gad67+ nerve terminals (Figure 4H). In whole-cell current clamp recording, we found that a 5-, 10- and 20-Hz blue light pulse decreased the firing rate and hyperpolarized the membrane potential in a frequency-dependent manner. However, stimulation with a 20-Hz yellow light had no effect on firing frequency (Figures 4I–K).

Next, to further clarify the mechanism of action, we recorded from the 5-HT neurons in whole-cell voltage clamp mode at a -60 mV holding potential. Activation of nerve terminals from VTA_Gad67+ neurons in the DRN induced PSCs in response to a 5-ms blue light pulse (Figures 5A,B). These light-induced PSCs were blocked by gabazine (10 μM), but not by combined AP5 (50 μM) and CNQX (20 μM) (Figures 5B,C). The mean response delay from light onset was 4.0 ± 0.2 ms (Figure 5D). To rule out indirect connections, TTX (1 μM) was applied. TTX blocked the light-induced PSCs (Figures 5E–G). The combination of TTX and 4AP (1 mM) rescued the light-induced PSCs, which was again blocked by gabazine, but not by CNQX (Figures 5F,G). We found the mean response delay was 4.2 ± 0.2 ms (Figure 5H). We recorded 24 cells from 5 mice, and all cells (24/24, 100%) were inhibited by blue light stimulation (Figure 5I). Taken together, these results confirmed that DRN_5–HT neurons were directly innervated and inhibited by VTA_Gad67+ neurons through GABA neurotransmission.

During the experimental recordings from YC-positive 5-HT neurons, we also recorded from YC-negative non–5-HT neurons. Interestingly, 3/22 non–5-HT neurons showed an inhibitory response to a 20-Hz blue light pulse (Supplementary Figure 2). Therefore, we assumed that some non–5-HT neurons also received inhibitory input from VTA_Gad67+ neurons. In addition to 5-HT neurons, GABAergic neurons are also distributed throughout the DRN (Monti, 2010b). To elucidate the connection between VTA_Gad67+ neurons and DRN_Gad67+ neurons, we used Gad67-Cre mice and injected a Cre-dependent AAV expressing ChR2-eYFP into the VTA and another Cre-dependent AAV expressing tdTomato into the DRN to visualize GABAergic neurons in the DRN (Figure 6A). Immunohistochemical study confirmed that ChR2 was exclusively expressed in the VTA (Figure 6B). In vitro slice patch clamp recording from ChR2-expressing VTA_Gad67+ neurons revealed that these neurons were depolarized by a 10-Hz blue light pulse, which also significantly increased the spontaneous firing frequency by 989 ± 319% compared with pre-stimulation values (Figures 6C–E). We stained DRN-containing slices with an anti-Gad67 antibody to label GABA neurons. Co-labeling of tdTomato and Gad67-positive cell bodies confirmed that GABA neurons expressed tdTomato (Figure 6F, white arrow), which was innervated by nerve terminals from VTA_Gad67+ neurons (Figure 6F, white arrowheads). We then recorded from the tdTomato-expressing DRN_Gad67+ neurons by stimulating VTA_Gad67+ nerve terminals (Figure 6G). In whole-cell current clamp recordings from DRN_Gad67+ neurons, we found that a 20-Hz blue light pulse also silenced them. However, the 20-Hz yellow light stimulation had no effect (Figures 6H,I). Next, to explore the connection patterns, we recorded from DRN_Gad67+ neurons in whole-cell voltage clamp mode at a -60 mV holding potential. Activation of VTA_Gad67+ nerve terminals induced PSCs in response to a 5-ms blue light pulse (Figures 6J,K). These light-induced PSCs were blocked by the combination of TTX (1 μM), 4AP (1 mM), and gabazine (10 μM), but not by TTX and 4AP (Figures 6J,K), confirming a monosynaptic GABAergic connection. The mean response delay was 3.7 ± 0.4 ms (Figure 6L). We recorded 36 cells from 7 mice, where 15 cells (41.7%) responded to blue light stimulation (Figure 6M). From these results, we concluded that a fraction of DRN_Gad67+ neurons were also directly innervated and inhibited by VTA_Gad67+ neurons.

Through the experiments thus far, we found that both DRN_5–HT and DRN_Gad67+ neurons were inhibited by VTA_Gad67+ neurons. However, previous reports show that DRN_5–HT neurons are also inhibited by local GABAergic interneurons (Huang et al., 2017; Zhou et al., 2017). Thus, we aimed to reveal whether the inhibitory role of DRN_Gad67+ neurons on DRN_5–HT neurons was the same in Gad67-Cre mice. To individually visualize DRN_Gad67+ and DRN_5–HT neurons, we generated a bigenic mouse line, Gad67-Cre; Tph2-tTA. We then mixed and injected into the DRN two AAV viral vectors, Cre-dependent ChR2-eYFP and tTA-dependent TetO tdTomato (Figure 7A). We stained DRN-containing brain slices with an anti-Tph2 antibody to label 5-HT neurons. Co-labeling of tdTomato- and Tph2-positive cell bodies confirmed that 5-HT neurons expressed tdTomato (Figure 7B, white arrows), and they were also innervated by nerve terminals from local DRN_Gad67+ neurons (Figure 7B, white arrowheads). To confirm ChR2 expression, we recorded from DRN_Gad67+ neurons. A 10-Hz blue light pulse significantly increased the spontaneous firing frequency by 1161 ± 375% compared with pre-stimulation values (Figures 7C–E). We then recorded from the tdTomato-expressing DRN_5–HT neurons by stimulating the local DRN_Gad67+ nerve terminals (Figure 7F). In whole-cell current clamp, we found that a 20-Hz blue light stimulation significantly reduced the firing frequency. However, a 20-Hz yellow light stimulation had no effect (Figures 7G,H). Next, to explore the connection patterns, we recorded from DRN_5–HT neurons in whole-cell voltage clamp mode at a -60 mV holding potential. Activation of DRN_Gad67+ nerve terminals in the DRN induced PSCs in response to a 5-ms blue light pulse (Figure 7I). These light-induced PSCs were blocked by the combination of TTX (1 μM), 4AP (1 mM), and gabazine (10 μM), but not by TTX and 4AP (Figures 7I,J), confirming a monosynaptic GABAergic connection. The mean synaptic delay was 3.7 ± 0.4 ms (Figure 7K). We recorded 55 cells from 5 mice, where 14 cells (25.5%) responded to blue light stimulation (Figure 7L). Finally, we confirmed that a small fraction of DRN_5–HT neurons was also directly innervated and inhibited by local DRN_Gad67+ neurons.