All experimental manipulations were performed in accordance with the Animal Welfare Guidelines of Third Military Medical University (TMMU) and the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and were approved by the TMMU Animal Care and Use Committee. Wild-type C57BL/6J mice and CRH-IRES-Cre (Cat# 12704) transgenic mice originated from the Laboratory Animal Center of TMMU or Jackson Laboratory were utilized in this study. All animals were group-housed (four to five individuals per cage) at 20–25°C under 12-h light/dark cycle conditions with food and water readily available. Adult CRH-Cre mice that were 8–20 weeks of age were utilized in RV-based retrograde tracing experiments. The animals used in the other experiments were adult C57BL/6J mice. Animals implanted with optical fibers for fiber photometry recordings were housed individually. Animal numbers of each experiment are indicated in each figure legend.

For PRV retrograde tracing experiments, adult C57BL6/J mice were anesthetized with 10 ml/kg sodium pentobarbital (1%). The bladder was accessed through a midline incision in the lower abdomen. Using a glass micropipette (20 μm diameter of the tip) connected with injection pump, 1.5 μl PRV531-EGFP (2.00 × 10⁹ PFU/ml, BrainVTA Technology Co., Ltd., Wuhan, China) was injected into the dome of the bladder at a speed of 300 nl/min. After being held in the injection region for 5–10 min, the micropipette was slowly pulled out to avoid the virus leaked from the bladder detrusor. The injection site was disinfected with alcohol cotton balls and the surgical wound was sutured. Postoperative mice were sent back to their cages and were perfused before showing signs of death.

Animals received isoflurane (4% induction, 1%–1.5% maintenance throughout the surgery) for anesthesia and were then stabilized in a stereotaxic frame (RWD Technology Corp., Ltd.; Shenzhen, China) with a heating pad for temperature support. Animals were considered to be kept at a sufficient depth of anesthesia, characterized by no response to tail pinch and the loss of pedal withdrawal reflex. After local application of 100 μl lidocaine hydrochloride (2%) under the skin, a midline scalp incision was made by using eye scissors to expose the skull. The craniotomy (1 mm × 1 mm) above each target brain region was drilled using a dental drill (0.5 mm in diameter). Different types of viruses were delivered into the PMC or the VLPAG through pressure injection using glass micropipette attached to a syringe (1 ml). The injection speed was about 20 nl/min. After each injection, the micropipette was left at the injection point for 10 min before withdrawal. The scalp wound was sutured. Then, animals were put back to their home cages and allowed to recover for 3 weeks prior to optic fiber implantation. The coordinates for target injection areas included the PMC (AP, −5.45 mm; ML, −0.7 mm; DV, −3.14 mm from dura) and the VLPAG (AP, −4.72 mm; ML, −0.5 mm; DV, −2.10 mm from dura), as determined by the mouse brain atlas (Paxinos and Franklin Mouse Brain Atlas, second edition).

For fiber photometry experiments, 100 nl volume of AAVRetro-hSyn-Cre (AAV2/2, titer: 10.0 × 10¹² vg/ml, vector genome per ml, TaiTool Bioscience Co., Ltd., Shanghai, China) mixed with CTB555 (invitrogen, C34776) was unilaterally injected into the PMC. Then, 100 nl of AAV-Dio-GCaMP6f (AAV2/9, titer: 0.5 × 10¹² vg/ml, Obio Biotechnology Co., Ltd. Shanghai, China) was injected into the ipsilateral VLPAG of the experimental group wild-type C57BL/6J mice, whereas 100 nl of AAV-Dio-EGFP (AAV2/9, titer: 4.20 × 10¹² vg/ml, BrainVTA Technology Co., Ltd., Wuhan, China) was unilaterally injected into the ipsilateral VLPAG of the control group wild-type C57BL/6J mice. A 200 μm diameter optic fiber (NA 0.48, Doric lenses, Quebec City, QC, Canada) that was fastened with a small metal cannula (ID. 0.51 mm, OD. 0.82 mm) was implanted 50 μm above viral injection region of the VLPAG three weeks after AAVs injection. Then, the optic fiber was attached to each animal’s skull using dental cement as an adhesive. Finally, these animals rested for 3–5 days prior to GCaMP6f detection during urination events.

To label the upstream PAG neurons of PMC^CRH neurons, we first injected about 50 nl of a mixture containing 1:1 volumes of AAV-Dio-EGFP-TVA (AAV2/5, titer: 2.00 × 10¹² vg/ml, BrainVTA Technology Co., Ltd., Wuhan, China) and AAV-Dio-RVG (AAV2/5, titer: 2.00 × 10¹² vg/ml, BrainVTA Technology Co., Ltd., Wuhan, China) into the unilateral PMC of CRH-Cre mice. After waiting three weeks for these AAVs expression, 100 nl of RV-EnVA-ΔG-DsRed (titer: 2.00 × 10⁸ IFU/ml, BrainVTA Technology Co., Ltd., Wuhan, China) was stereotaxically injected into the same area of these CRH-Cre mice. To explore the upstream brain areas of PMC-projecting VLPAG neurons, 100 nl volume of AAVRetro-hSyn-Cre (mixed with CTB555) was injected into the right PMC and 50 nl of the 1:1 volume mixture of AAV-Dio-EGFP-TVA and AAV-Dio-RVG was administered into the right VLPAG of wild-type mice of a C57BL/6J background at the same time. Three weeks later, 100 nl of RV (RV-EnvA-ΔG-DsRed) was injected into the same site of the VLPAG of these wild-type C57BL/6J mice. These mice were perfused on day 7 after RV injection.

Fiber photometry recordings were performed as described previously (Zhang et al., 2017; Yao et al., 2018, 2019). The fiber photometry system (FOM-02M FiberOptoMeter, Suzhou Institute of Biomedical Engineering and Technology, China) was used for measuring population neuronal activity. In order to decrease photo bleaching, light power at the tip of the optic fiber was adjusted to 0.22 mW/mm². Before recordings of PMC-projecting VLPAG neurons activities during urination, each animal was injected intraperitoneally (i.p.) with diuretic furosemide (40 mg/kg) to increase the frequency of voiding and the implanted fiber was directly connected to the fiber photometry system. Then, mice were moved into the experimental chamber that had a glass bottom and habituated for 20 min. A video recorder (Sony, Japan) was placed under the chamber to record videos (30 Hz with a spatial resolution of 1,280 × 720 pixels) of the urination events. The activities of the PMC-projecting VLPAG neurons and mouse behavior videos were detected simultaneously. Each recording session lasted approximately 40 min to 1 h. Fiber photometry recording data were sampled via acquisition software (LabVIEW platform, National Instruments, USA). According to the video data, we defined the onset of urine appearance from the mouse urinary meatus as the start of urination events, and defined no urine appearance at the urinary meatus as the termination of urination events (the urination offset).

Animals were anesthetized with isoflurane (4% induction, 1%–1.5% maintenance throughout the surgery) and positioned on top of a heating pad. After local application of 100 μl lidocaine hydrochloride (2%) under the skin and lower abdominal muscle, we accessed the mouse bladder dome and exteriorized it via a 1-cm midline incision in the lower abdomen. A PE10 catheter (7 cm) was inserted and implanted into the bladder. Then, the other side of the PE10 catheter was removed through a 1 cm cut at the back of the mouse’s neck after local application of lidocaine hydrochloride (2%, 100 μl) under the skin. The surgical wound was sutured after verifying the seal and normal flow of the bladder catheter. Animals were rested for 3–5 days before cystometric measurement incorporating fiber photometry recording experiments were initiated. We used a high sensitivity pressure transducer (YPJ01H; Chengdu Instrument Factory, China) connected with the multi-channel physiological information recording device (RM6240; Chengdu Instrument Factory, China) to measure the bladder pressure. The injection pump (RWD404; RWD Technology Corp., Ltd., Shenzhen, China), pressure sensor, and adapter were connected to each other through a three-way tab. A PE50 tube was connected to the end of the adapter. Before cystometry recordings, the PE10 bladder catheter was moved into the other end of the PE50 tube. Freely moving animals were simultaneously measured for VLPAG neuronal activity and cystometry with bladders receiving an infusion of normal saline via the PE10 catheter at a constant speed of 30–50 μl/min. The cystometry data were collected by the commercially available acquisition software (Chengdu Instrument Factory, China).

Immunohistochemistry staining was carried out as we previously described (Yao et al., 2018, 2019). After each experiment, mice were anesthetized with 10 ml/kg sodium pentobarbital (1%) and initially transcardially perfused using phosphate-buffered saline (PBS) or saline, followed by transcardial perfusion using 4% paraformaldehyde (PFA). Murine brains underwent dissected and were then fixed by placing samples in 4% PFA at 4°C for at least 24 h until sectioning. These brain samples were sliced into coronal slices or sagittal slices (40 μm) by use of a vibratome (NX50, Thermo). For tyrosine-hydroxylase (TH) immunohistochemistry staining, the primary antibody was rabbit anti-TH (1:200, Millipore, AB152) and a donkey anti-rabbit antibody conjugated with Alexa Fluor 647 was used as the secondary antibody (1:200, Invitrogen, A31573). Lastly, all sections were stained with DAPI (Beyotime, Shanghai, China, C1006). These brain slices were mounted on glass slides, cover-slipped, and imaged using a fluorescence microscope (BX53, Olympus) or a scanning confocal microscope (TCS SP5, Leica). Viral injection sites and fiber tip sites for each mouse was confirmed via post hoc images. Animals with improper locations of viral injection or fiber implantation were excluded from final data analyses.

Fiber photometry recording data were recorded at a 2,000 Hz sample rate and bladder pressure data were obtained at an 800 Hz sample rate. The optic-fiber recording data underwent low-pass filtering by use of a Savitzky-Golay filter with 3 as the polynomial order and 50 side points. We calculated the relative fluorescence change values, Δf/f = (f − f_baseline)/f_baseline), as Ca²⁺ transients, and f_baseline was defined as the minimum fluorescent signal detected when recording urination events. We considered the Ca²⁺ transient amplitude, which was three times higher than the standard deviation of the noise band, as a positive signal. To shuffle the GCaMP6f fluorescence data, we divided the GCamp6f fluorescence data into 10 segments that were then randomly assigned these segments with detected urination events. We counted the cell number of PRV531-labeled neurons and RV-labeled neurons manually.

All data in this article were represented as mean ± s.e.m‥ The dot plots in each figure are overlaid with the corresponding bar graphs. We used MATLAB programs (MathWorks, USA) to conduct our cross-correlation analyses and non-parametric statistical analyses. We used the Wilcoxon signed-rank test for the nonparametric paired data and the Wilcoxon rank-sum test for the two independent data. Statistical significance was determined as follows: ^***P < 0.001, ^**P < 0.01, ^*P < 0.05; ns indicates no significant difference was detected.

To confirm the distribution of VLPAG neurons that may participate in control of bladder and urination, we directly injected pseudorabies virus 531 (PRV531) expressing robust green fluorescent protein, a valuable retrograde trans-multisynaptic tracer (Jia et al., 2019) into the bladder detrusor of adult mice to label supraspinal neural circuits (Figure 1A). These mice were perfused when showing serious symptoms of infection and abnormal behaviors. Their average survival time after PRV531 injection was 103.9 h ± 3.0 h (mean ± s.e.m., n = 7 mice). Depending on the average survival time, PRV531 transported and infected transneuronally from one-order neurons to fourth-order neurons. In addition to the PAG, the major PRV531-labeled brain structures were found in the gigantocellular reticular nucleus (Gi), raphe nuclei, A5 noradrenaline cells, PMC, locus coeruleus (LC), red nucleus (Red N), lateral hypothalamic area (LH), paraventricular nucleus (PVN), medial preoptic area (MPOA), and cerebral cortex (data not shown). The above PRV531-labeled brain areas were similar to those reported in our previous study using PRV152 (Yao et al., 2018).

In the present experiment, we aimed to demonstrate the precise anatomical location of PRV531-labeled VLPAG neurons. According to the mouse brain atlas, morphological data of sagittal brain section (Figure 1B, n = 1 mouse) and serial coronal brain sections (Figures 1C,D) showed that PRV531-infected VLPAG neurons were bilaterally present in the whole VLPAG region (Bregma: −4.12–5.20 mm). A closer overlay analysis revealed that the majority of PRV531-labeled VLPAG neurons were located between Bregma −4.48 mm and Bregma −4.96 mm (Figure 1E, n = 6 mice). Quantification of retrogradely labeled cell numbers of the VLPAG in serial coronal brain sections showed no significant difference between the ipsilateral and contralateral sides (Bregma −4.48 to −4.96 mm, left side 53 ± 9–96 ± 7, right side 53 ± 8–93 ± 8, P > 0.05, Figure 1F, n = 6 mice). In addition, we also found the PRV531-infected LPAG neurons in these brain sections (Figures 1B–E). Quantification of PRV531-labeled neuron numbers of the LPAG in serial coronal brain sections showed no significant difference between the ipsilateral and contralateral sides (Bregma −4.48 to −4.96 mm, left side 34 ± 11–67 ± 12, right side 37 ± 10–67 ± 15, P > 0.05, Figure 1G, n = 6 mice). PRV-infected neurons in the VLPAG/LPAG should be fourth-order neurons (Figure 1H), according to our previous study (Yao et al., 2018). Altogether, these results indicate that both VLPAG and LPAG neurons innervate the bladder detrusor.

PMC^CRH neurons were previously reported to be an important controller in bladder contractions (Hou et al., 2016; Ito et al., 2020). PMC^CRH neurons received direct input projections from PAG (Hou et al., 2016; Verstegen et al., 2019), but the cell body distribution of these PAG neurons has not been well-characterized. To further investigate the anatomical distribution of VLPAG neurons innervating the bladder detrusor that project to PMC^CRH neurons, we used cell-type-specific rabies virus (RV) system in CRH-Cre mice (Figure 2A). Firstly, we injected two Cre-dependent helper adeno-associated viruses (AAV-Dio-RVG and AAV-Dio-EGFP-TVA) into the right PMC (Figure 2A). Then, we injected RV (RV-EnvA-ΔG-DsRed) into the same area 3 weeks after the first injection (Figure 2A). One week later, we perfused these mice and obtained their brains (Figure 2A).

The starter neurons (yellow cells) were restricted to the right PMC and co-labeled by helper AAV and RV (Figures 2B,C). Retrogradely labeled monosynaptic input neurons from the VLPAG only expressed RV-DsRed (Figures 2B,D,E). The representative serial coronal brain sections of the VLPAG (Figures 2D,E) revealed that RV-labeled PAG neurons were widely distributed in the whole ipsilateral VLPAG (Bregma: −4.12 to −5.20 mm). The quantification of RV-labeled neuronal number in the ipsilateral VLPAG showed that most of RV-labeled VLPAG neurons were also located in the medial and caudal regions of the VLPAG (Bregma −4.48 to −4.96 mm, Figure 2F, n = 6 mice). The average number of cells in the VLPAG ranged from 15 ± 1 (Bregma −5.20 mm) to 121 ± 7 (Bregma −4.72 mm). We also found the RV-labeled LPAG neurons in these serial coronal brain sections (Figures 2D,E). Quantification of RV-labeled cell numbers in the ipsilateral LPAG (Figure 2G) showed that the average number of neurons ranged from 2 ± 0 (Bregma −5.20 mm) to 44 ± 7 (Bregma −4.72 mm). These findings confirm the accurate distribution of VLPAG neurons that project to PMC^CRH neurons. Moreover, our results suggest that LPAG neurons project to PMC^CRH neurons directly. We used the above location of the VLPAG for the following optical-fiber recording experiments.

The VLPAG is considered to be the major PAG column for bladder control via its direct projection to the PMC (Zare et al., 2019). The role of LPAG neurons in controlling bladder function is rarely reported. Therefore, we only explored the real-time Ca²⁺ activity of PMC-projecting VLPAG neurons in response to natural urination events through monitoring activities of these VLPAG neurons and urination events in freely behaving mice simultaneously in the following experiment. We expressed genetically encoded Ca²⁺ indicator GCaMP6f in PMC-projecting VLPAG neurons by injecting AAVRetro-hSyn-Cre mixed with CTB555 into the right PMC and AAV encoding Cre-dependent GCaMP6f (AAV-Dio-GCaMP6f) into the right VLPAG (Figure 3A). After three weeks, we implanted a 200 μm diameter optical fiber above the VLPAG and applied an optical fiber-based Ca²⁺ signal recording system to specifically monitor the Ca²⁺ activity of GCaMP6f-labeled PMC-projecting VLPAG neurons during urination (Figures 3B,D). Representative images revealed that the injection site of AAVRetro-hSyn-Cre mixed with CTB555 and axon terminals of these GCaMP6f-labeled VLPAG neurons were only restricted to the PMC (Figure 3C). We observed large increases in calcium activity of PMC-projecting VLPAG neurons at each time of urination (Figure 3E).

We observed that Ca²⁺ signals of GCaMP6f-labeled PMC-projecting VLPAG neurons were sustained throughout the entire duration of urination events (Figures 4A,B). We also found that the onsets of Ca²⁺ signals of GCaMP6f-labeled PMC-projecting VLPAG neurons were earlier than the urination event onsets in 181/199 trials (91% of trials), suggesting that the activities of these VLPAG neurons may contribute to the initiation of the urination events (Figures 4A,B, Supplementary Figures 1A,B). A robust increase in GCaMP6f fluorescence of PMC-projecting VLPAG neurons started 500 ms (505 ± 32 ms) before the urination event onsets (Supplementary Figure 1A). However, no change in fluorescence of EGFP-labeled PMC-projecting VLPAG neurons during urination was detected (Figures 4A–D, GCaMP6f-labeled group, max Δf/f = 8.25% ± 0.20%; EGFP-labeled control group, max Δf/f = 0.74% ± 0.04%; P < 0.001), suggesting that the urination-related GCamp6f signals were not a result of movement artifacts. Across all the recording trials, each urination event corresponded to the Ca²⁺ signal of GCaMP6f-labeled PMC-projecting VLPAG neurons (Figures 4B,C). Further analysis revealed that the urination-correlated Ca²⁺ activity was absent in randomly chosen intervals (shuffled data, Figure 4E). Thus, our data indicates that the population Ca²⁺ activity of PMC-projecting VLPAG neurons highly correlates with urination.

To further investigate whether bladder contraction coincides with the calcium signals of PMC-projecting VLPAG neurons, we simultaneously measured bladder pressure by use of cystometry and GCaMP6f fluorescence changes in freely moving animals (Figure 5A). We determined that each instance of increased activity of GCaMP6f-labeled PMC-projecting VLPAG neurons was associated with a spike in bladder pressure (Figures 5B–D, n = 7 mice). We did not identify an increase in fluorescence in EGFP-labeled PMC-projecting VLPAG neurons during bladder contraction (Figures 5C,D, n = 7 mice). Averaging Ca²⁺ signals synced to the onset of bladder contractions showed a sharp increase in GCaMP6f fluorescence when the initial surge of bladder pressure (Figure 5D). Time-locked Ca²⁺ signals of GCaMP6f-labeled PMC-projecting VLPAG neurons started 68 ms on average before the bladder contraction onsets (Figure 5D). The onsets of Ca²⁺ signals of these PMC-projecting VLPAG neurons were earlier than the bladder contraction onsets in 34/63 trials (54% of trials). Each bladder contraction was 100% associated with Ca²⁺ activity of VLPAG neuronal populations projecting to the PMC. Cross-correlation analysis between GCaMP6f signals and the measured bladder pressure indicated that the mean peak’s correlation coefficient for GCaMP6f data was found to be significantly higher compared to the shuffled data (Figures 5E,F, GCaMP6f data, 0.61 ± 0.04; shuffled data, 0.14 ± 0.01; P < 0.05). Our findings suggest that these Ca²⁺ signals from PMC-projecting VLPAG neurons have a significant association with bladder contraction.

The above data indicate that PMC-projecting VLPAG neurons are activated during urination and bladder contractions. We wondered which brain regions could direct input to PMC-projecting VLPAG neurons. To unravel monosynaptic inputs to VLPAG neurons that ultimately project to the PMC, a RV-based retrograde trans-synaptic tracing was used. We injected AAVRetro-hSyn-Cre mixed with CTB555 into the right PMC and two Cre-dependent helper AAV (presented above) into the right VLPAG (Figure 6A). At 2–3 weeks after the first injection, we injected RV-EnvA-ΔG-DsRed into the right VLPAG (Figure 6A). Representative expanded images were obtained 1 week later and showed that starter neurons (yellow cells) were restricted to the VLPAG (Figure 6B).

Coronal brain images from a representative mouse revealed that major input neurons to PMC-projecting VLPAG neurons were observed in many upstream brain areas (on the ipsilateral side), including prefrontal areas, primary motor cortex (M1), primary somatosensory cortex (S1), bed nucleus of the stria terminalis (BNST), MPOA, PVN, LH, zona incerta (Zi), central amygdalar nucleus (CeA), substantia nigra pars compacta (SNc), substantia nigra pars reticulata (SNr), parabrachial nucleus (PBN), PMC, and lateral reticular nucleus (LRt; Figure 6C). Then we performed a statistical analysis on the input cell numbers obtained from the above major brain regions on the ipsilateral side of the injection site (Figure 6D, n = 6 mice). The average number of RV-labeled neurons throughout the above brain regions (Figure 6D) ranged from 57 ± 5 (PVN) to 931 ± 16 (LH). These results show a clear distribution of input across the whole brain to PMC-projecting VLPAG neurons.