Recent studies have drawn apparently contrasting conclusions about the role and importance of Barr^CRH neurons in the regulation of volitional voiding (Hou et al., 2016; Keller et al., 2018). To further define their role, the light-activated cation channel ChR2 (channelrhodopsin-2) was selectively expressed in Barr^CRH neurons of CRH^CRE mice (Taniguchi et al., 2011) using a Cre-dependent adeno-associated viral vector (AAV-EF1α-DIO-ChR2-mcherry) (Figure 1A–C, Figure 1—figure supplement 1). Saline infusion to the bladder in urethane anaesthetised mice produced a regular cycle of autonomous micturition (voids at ~5 min intervals, average infusion rate 23 ± 4 µl/min). Tonic unilateral activation of Barr^CRH neurons (5–20 Hz x 20 ms, 465 nm light pulses, applied for three completed voiding cycles) produced an increase of micturition frequency (to 153.2 ± 15.6% of basal after 10 Hz stimulation, Figure 1D) manifesting as a significant shortening in inter-void interval associated with a reduction of the threshold pressure for voiding (84.0 ± 4.7% at 10 Hz, Figure 1E). Similar illumination in control mice (CRH^CRE mice injected with AAV-DIO-hm4Di-mCherry instead of ChR2) had no effect on voiding frequency.

Chemogenetic-inhibition of Barr^CRH neurons was achieved by expressing an inhibitory DREADD (Figure 2A, bilateral injection of AAV-DIO-hM4Di to CRH^CRE mice). Administration of Clozapine N-oxide (CNO, 5 mg/kg i.p) produced a prolonged reduction in the frequency of voids seen during a continuous infusion protocol (reduced to 66.8 ± 6.3% of control at 20 mins, Figure 2B). This was associated with increases of threshold pressure, fill volume and micturition pressure (Figure 2C).

The chemogenetic inhibition of Barr^CRH during this continuous filling protocol caused the bladder to become increasingly distended leading to incomplete voids. To control for this effect a ‘fill and hold’ protocol was employed to fill the bladder to a maximum pressure of 15 mmHg (close to the threshold pressure for voiding). This volume was held for up to 10 min until either the mouse voided spontaneously, or the bladder was manually emptied, and the filling cycle restarted (Figure 2D). Using this protocol, chemogenetic inhibition caused a prolongation of the latency to void (from 207 ± 17 s at baseline to 738 ± 60 s at 20mins after CNO). This effectively produced a period of urinary retention, that persisted for 2 hr (Figure 2E).

These findings indicate that Barr^CRH neurons have a potent ability to modulate the autonomous micturition cycle and that their basal level of activity is of functional importance.

To assess whether Barr^CRH neurons act as a tightly-coupled, pre-motor drive to bladder parasympathetic neurons (Fowler et al., 2008; de Groat and Wickens, 2013) bladder pressure was recorded while parametrically opto-activating Barrington’s nucleus unilaterally (9.5 ± 0.3 mW, 465 nm). In initial experiments, unilateral opto-activation of Barr^CRH (20 ms x 20 Hz for 5 s) evoked non-voiding contractions of the bladder (eNVC, Figure 3A and B, with the bladder filled to half of its threshold capacity, n = 7 mice). These eNVC were similar to the transient bladder contractions triggered by optoactivation of Barr^CRH neurons previously noted by Hou et al. (2016). Varying stimulus frequencies and pulse durations produced modestly graded changes in eNVC with a 20 ms x 20 Hz protocol producing near maximal responses (3.9 ± 0.8 mmHg, Figure 3C and D). The eNVC had a consistent latency to onset of 1.3 ± 0.1 s and a time to peak of 6.0 ± 0.3 s following stimulus onset and an average duration of 8.2 ± 0.6 s. With each of the stimulus parameters there were ‘failures’ where there was no detectable bladder response (Figure 3 and E). The probability of eNVC increased with stimulation frequency (71.4 ± 7.6% at 2.5 Hz and 97.1 ± 2.5% at 20 Hz) with 20 Hz being the most reliable. Single light pulses of longer duration (1–3 s) were also able to reliably generate eNVC. In contrast, illumination in control mice (CRH^CRE mice injected with AAV-DIO-hm4Di-mCherry instead of ChR2) had no effect on the bladder pressure (101.4 ± 0.6% compared to the pressure immediately before optoactivation at 20 Hz x 20 ms, n = 3).

Previous anatomical studies with retrograde tracing using pseudorabies virus have suggested that there is a route for communication between afferent neurons innervating the distal colon and Barrington’s nucleus and further that Barrington’s neurons were activated by distention of the distal colon (Rouzade-Dominguez et al., 2003). This has led to the suggestion that Barrington’s nucleus may control the lower gastrointestinal tract as well as the lower urinary tract. Indeed, defaecation was suggested to occur on occasion (but not quantitated) following opto-activation of Barr^CRH neurons in mice implying a role in motor control (Hou et al., 2016). To investigate a possible relationship between Barr^CRH and activity in the distal colon, a balloon catheter was inserted to monitor pressure in pilot experiments (n = 2 mice). Distal colonic pressure was not synchronised with bladder pressure during the normal micturition cycle (Figure 3—figure supplement 1B). Furthermore, optogenetic activation of Barr^CRH neurons did not alter distal colonic pressure (Figure 3—figure supplement 1C and D), despite the generation of bladder eNVC. This preliminary evidence suggests that this Barr^CRH population of Barrington’s neurons is not involved in motor control of the colon.

It was postulated that recruitment of a larger population of Barr^CRH neurons in synchrony might be more effective in triggering larger or more reliable bladder contractions. This was tested with bilateral expression of ChR2 and a dual-fibre optical cannula allowing independent activation of one, the other or both Barrington’s nuclei (Figure 4A–C). Bilateral activation of Barr^CRH produced larger eNVC (7.1 ± 2.0 mmHg at 20 ms and 20 Hz, n = 7 mice) than optoactivation of either side alone (2.9 ± 0.5 and 3.1 ± 0.8 mmHg for right or left side, respectively), particularly at higher frequencies of stimulation (Figure 4D and E). The effect of bilateral stimulation on eNVC amplitude was additive rather than synergistic. The probability of generating eNVC was increased by bilateral stimulation (evident at lower stimulus frequencies that is increased by 154 ± 18% for bilateral vs right alone or by 158 ± 17% for bilateral vs left alone at 2.5 Hz, Figure 4E). It was notable that, with the bladder filled to half of its threshold capacity, bilateral Barr^CRH stimulation never triggered voids.

These findings support the proposal that Barr^CRH neurons can selectively generate bladder contractions and indicate that this is a probabilistic process, with failures, rather than being a simple high-fidelity pre-motor drive to the bladder.

This raised the question of whether the stage of the micturition cycle influences the bladder pressure response to Barr^CRH opto-activation as the cycle phase may modulate Barr^CRH neuronal excitability. During continuous bladder filling, it was noted that the amplitude of Barr^CRH eNVC increased progressively through the micturition cycle (increase of 17.0 ± 3.9 fold, comparing eNVC obtained during the 2^nd versus 5^th quintile of micturition cycle, Figure 5A–D). This phenomenon was also observable, albeit not quantitated or commented upon, in the recordings of Hou et al. (2016), see Figure 5B). Similarly, the probability of obtaining a bladder contraction with optoactivation also increased with progressive filling, with most ‘failures’ being seen when the bladder was <40% filled (Figure 5E). The same phase dependence of eNVC was also apparent with bilateral stimulation of Barr^CRH (Figure 5—figure supplement 1A and B).

The phase-dependence of eNVC amplitude may, in part, be a consequence of bladder distension, leading to raised passive detrusor tension and an increase of length-dependent contractions. To test this proposition, the effect of pelvic nerve stimulation was assessed in the pithed decerebrate, arterially-perfused mouse preparation (Ito et al., 2018; Ito et al., 2019). The amplitude of bladder contractions induced by pelvic nerve stimulation (4–20 Hz, 10V, 3 s) increased with bladder filling (Figure 5—figure supplement 2) with a doubling (2.2 ± 0.34 fold at 20 Hz) of the pressure generated between empty bladder and 70 µl fill (close to voiding threshold in an intact mouse). However, this amplitude increase plateaued at a volume of ~50 µl – and showed a much less steep relationship than that observed for Barr^CRH eNVC in vivo which increased by 17-fold over the same range of bladder distension. Additionally, this relationship did not account for the observed probabilistic nature of eNVC, as failures were never observed with pelvic nerve stimulation.

Although tonic stimulation of Barr^CRH increased voiding frequency, it was not possible to trigger full voiding contractions with phasic Barr^CRH stimulation with the bladder up to 50% filled, even with bilateral stimulation. However, by applying stimuli systematically at points through the micturition cycle it was possible to trigger fully co-ordinated voids by activating Barr^CRH neurons later in the cycle (>50% filled, Figure 5B–D). The pattern and amplitude of the evoked bladder contraction was similar to that seen with spontaneous voids and they occurred at a similar latency to eNVC. In addition, voiding was complete and the empty bladder relaxed to the basal pressure level after each void.

The mouse external urethral sphincter (EUS) shows bursting activity during spontaneous voids which facilitates urine expulsion (Ito et al., 2018; Keller et al., 2018). Injections of pseudorabies virus into either the bladder or EUS has shown labelling in the vicinity of Barrington’s nucleus, suggesting it is part of the EUS control circuit (Nadelhaft et al., 1992; Nadelhaft and Vera, 1996; Marson, 1997). However, recent evidence suggests that it is the Barr^ESR-1, rather than Barr^CRH, neurons which project to local circuit interneurons in L4-5 that may regulate EUS motoneurons (Keller et al., 2018) analogous to the lumbar spinal coordinating centre (LSCC; Chang et al., 2007). Therefore, recordings were made from the EUS to investigate the relationship of the voiding-associated bursting to Barr^CRH activation. The Barr^CRH eNVC (irrespective of their magnitude) were never associated with EUS activity (Figure 5B and D).

However, when Barr^CRH activation evoked a voiding contraction then bursting EUS activity was always found (Figure 5D and F). These Barr^CRH induced voids had EUS activity that was indistinguishable from spontaneous voids in terms of burst duration (spontaneous 4.4 ± 0.9 vs opto-induced 4.4 ± 1.0 s, n = 7, paired t-test, ns) and frequency (spontaneous 22.7 ± 2.7 vs opto-induced 21.5 ± 3.0 Hz, n = 7, paired t-test, ns)). These results indicate that the Barr^CRH neurons can trigger voids that are in all aspects similar to those seen spontaneously but that can be triggered to occur earlier in the normal micturition cycle.

The axons from Barr^CRH were noted to provide a specific innervation of the sacral parasympathetic neurons but not to the ventral horn at the level of Onuf’s nucleus (Figure 6A, Figure 6—figure supplement 1). To investigate whether optogenetic stimulation of spinal axons of Barr^CRH is sufficient to directly generate eNVC, bladder pressure was recorded while light was applied from an optic fibre located above the spinal cord. Optogenetic stimuli (either 20 ms x 20 Hz for 5 s or single 1 s pulse) applied to the spinal cord reliably induced bladder contractions (Figure 6B–D, p=0.025, bladder half filled). These eNVCs tended to occur with a shorter latency than those evoked directly from pontine stimulation (1.0 ± 0.2 s vs 1.26 ± 0.1 s, n = 5). Similarly, during continuous bladder filling, spinal activation could trigger full voids (Figure 6E). These data support the principle that the Barr^CRH neurons can evoke both voiding and eNVC through their spinal projections.

It has been proposed that CRH released from Barrington’s neurons at a spinal level augments bladder pressure responses (Klausner and Steers, 2004; Klausner et al., 2005) although others have reported the opposite action (Pavcovich and Valentino, 1995; Kiddoo et al., 2006; Wood et al., 2013) and genetic knock out of CRH expression in Barrington’s neurons was without phenotype (Verstegen et al., 2019). If the release of CRH does increase during the micturition cycle, then this might be predicted to act as a positive feedforward mechanism to augment the parasympathetic and hence bladder pressure responses to Barr^CRH drive. To test this hypothesis, the effect of intrathecal Astressin (a broad-spectrum CRH antagonist, 5 µg in 5 µl) on Barr^CRH eNVC was assessed through the micturition cycle. Counter to the prediction, Astressin significantly and reversibly increased the amplitude of eNVC, an action that was more pronounced as the bladder filled (333 ± 75%, p=0.008 (n = 7), 20 mins after Astressin, Figure 7). Intrathecal Astressin also decreased the infused volume required to trigger a void (Figure 7D).

This indicates that CRH is providing a negative feedback signal to limit the extent of the spinal parasympathetic response to Barr^CRH neuronal activity (in agreement with Pavcovich and Valentino, 1995; Kiddoo et al., 2006; Wood et al., 2013). Therefore, increased release of CRH cannot account for the augmented responses to Barr^CRH activation with progression through the micturition cycle.

Neural recordings from cats (Sasaki, 2005a) and rats (Manohar et al., 2017) indicates that some putative Barrington’s neurons fire intermittently during the storage phase with an increase of firing that occurs around voiding, consistent with a role in mediating the drive to bladder parasympathetic neurons. Recent fibre photometric recordings of Barr^CRH neurons, using the genetically encoded calcium indicator GCaMP6, indicate that the activity of these neurons is ‘in phase’ with the micturition cycle (Hou et al., 2016; Keller et al., 2018). However, fibre photometry is unable to resolve the action potential discharge patterns from Barr^CRH neurons in vivo. As such it has not previously been possible to directly assess the functional relationship between Barr^CRH firing and bladder pressure.

Neuronal activity was recorded in the vicinity of Barrington’s nucleus using a 32-channel silicon probe to test whether changes in the excitability of Barr^CRH neurons during the micturition cycle accounts for the observed variation in the evoked pressure responses of the bladder. An optic fibre was placed above Barrington’s nucleus enabling optogenetic identification (Figure 8A). Recordings were made of cell activity during the normal micturition cycle (with simultaneous bladder pressure and EUS EMG activity) and in response to the application of light stimuli. A total of 113 individual neurons were identified by clustering from recordings made in the vicinity of Barrington’s nucleus (n = 3 mice, Figure 8BD). Definitive opto-identification of Barr^CRH neurons (n = 12) was indicated by reliable short latency spike entrainment to light (20 ms pulses, Figure 8C) with time-locked, maintained firing in response to longer light pulses (≥1 s, Figure 8C).

These Barr^CRH neurons showed a characteristic pattern of activity during the micturition cycle with bursting at the time of voiding (Figure 9A, 20.5 ± 4.1 Hz peak firing frequency). A second population of neurons was recorded with a similar pattern of activity (but were not activated by light) that are henceforth termed Barr^CRH-like (n = 32, Figures 9B and 10A,B) in distinction to the remainder of non-identified neurons (n = 69). These Barr^CRH-like neurons had a short-latency synchrony with the Barr^CRH neurons that was evident in cross-correlograms (Figure 9—figure supplement 1). Both Barr^CRH and Barr^CRH-like neurons showed a clear temporal relationship to bladder pressure (Figure 9C) with their firing preceding and ramping up with the pressure during voiding.

For both the Barr^CRH and Barr^CRH-like neurons there was a strong sigmoid relationship between bladder pressure and neuronal firing (Figure 10C), which wasnot seen in the non-identified group of neurons. The directionality of this influence was investigated by examining the cross-correlation between firing rate and bladder pressure – this indicated that the increases in firing frequency (for both Barr^CRH and Barr^CRH-like neurons) preceded increases in bladder pressure by ~3 s for both sets of neurons (Figure 10D). These data indicated that the pattern of firing of both Barr^CRH and Barr^CRH-like neurons anticipated changes in bladder pressure as would be expected for a pre-motor population upstream of bladder parasympathetic neurons.

Our recording probe trajectory passes close to the locus coeruleus (LC) on its lateral edge raising the possibility that some of our recorded (non-optoidentified) neurons could be LC neurons, which have been shown in rats to increase their firing in anticipation of a void (Manohar et al., 2017) and as such could fall into the Barr^CRH-like group. To examine this possibility, we made recordings from LC neurons under identical recording conditions in mice which had received injections of CAV2-PRS-ChR2-mCherry directly into LC, causing selective expression of ChR2 in LC cells only (Li et al., 2016). A total of 29 opto-identified LC neurons were recorded (n = 3 mice). They showed a characteristic pattern of spontaneous firing and a phasic burst of activity with a paw pinch. LC cells increased their firing around the void – a pattern that was evident in individual firing rate plots and in the z-scored firing heatmap (Figure 10—figure supplement 1A and B). However this peri-void activation was significantly less pronounced than the increase in firing seen in the Barr^CRH-like (and Barr^CRH) neurons (Figure 10—figure supplement 1C).This suggests that the Barr^CRH-like neurons are most likely to be part of the population of Barr^CRH neurons, of which only a subset recorded by the probe are exposed to enough light to be formally opto-identified.

Analysis of spontaneous Barr^CRH firing rates over the micturition cycle shows a pattern of activity that is consistent with what would be expected for a high-fidelity controller of bladder pressure. However, this is at odds with our optogenetic activation findings. To resolve this discrepancy the relationship between cycle phase and the light-evoked Barr^CRH activity and voiding was examined in more detail.

During all phases of the voiding cycle it was possible to opto-excite Barr^CRH neurons (Figure 11A) and the increase in firing frequency in both absolute and relative terms was independent of the phase of the micturition cycle (Figure 11B, ranging from 22.4 ± 7.7 to 24.0 ± 6.5 Hz across micturition phases). These data indicate that the intrinsic excitability of the Barr^CRH neurons does not vary across the micturition cycle and that augmentation of the bladder pressure responses (by 17.0 ± 3.9 fold) occurs downstream of the firing output from Barrington’s nucleus.

To further explore this proposition, the relationship between spontaneous non-voiding contractions (sNVC) and Barr^CRH neuronal firing was mapped. sNVC are defined as phasic increases of intravesical pressure seen during filling cystometry, not associated with passage of urine and have been seen in many studies of murine urodynamics (Pavcovich and Valentino, 1995; Hou et al., 2016; Ito et al., 2017; Keller et al., 2018; Verstegen et al., 2019). A burst of firing in the Barr^CRH neurons preceded the sNVC by 1.5–3.0 s – suggesting that they were triggered by a signal from the pons (Figure 11C and D). However, there was only a weak relationship between the magnitude of each Barr^CRH burst and the amplitude of the associated sNVC (see Figure 11E). A linear fit of these data indicates that an increase in burst size of 20 spikes (close to the maximum observed range) would only account for 0.5 mmHg difference in sNVC size (less than 20% of the observed range of amplitudes). Even this modest relationship was noted to be dependent upon a single outlier value of a large NVC occurring close to a void (circled). Again, this finding is consistent with Barr^CRH providing a trigger signal rather than a pre-motor drive which determines the amplitude of the bladder contraction.

This indicates a model of autonomous micturition where a spinal circuit gates the output to the bladder (shown schematically in Figure 12A). The Barr^CRH – parasympathetic - bladder afferent component of this circuit was modelled in NEURON using an existing preganglionic neuronal model (Briant et al., 2014) and a combination of a fast, excitatory synaptic drive descending from Barrington’s nucleus plus a bladder afferent synaptic drive (based on recordings of pelvic nerve afferents from Ito et al., 2019. The incrementing frequency of afferent drive, as the bladder fills, leads to summation and a maintained membrane depolarisation that increases parasympathetic excitability (Figure 12B).

The resulting output from the parasympathetic neuron when driven by Barr^CRH (with a mimicked 20 Hz optogenetic drive) was strongly dependent upon the phase of the micturition cycle with a ~ 10 fold increase over the voiding cycle which closely parallels the experimental data. Note also that in the early phase of the voiding cycle the Barr^CRH input is unable to evoke action potentials – thus producing ‘failures’.

The observations described above provided the basis for a new integrated model of the autonomous micturition cycle which incorporates the observed drive from Barr^CRH neurons and the known afferent feedback from the bladder (Figure 13A and methods including a summary of the evidence supporting the model). This afferent feedback governs both the excitability of the spinal parasympathetic neurons (demonstrated in the NEURON model above) and the output of a synaptic generator driving Barr^CRH activity. The resulting feedback loop (depicted schematically in Figure 13B) closely reproduces characteristic of the observed micturition cycle with graded NVCs, periodic voids and patterns of Barr^CRH firing.

The varying excitability of spinal parasympathetic neurons is represented in the integrated model by a pressure-modulated logistic relationship which determines the change in bladder pressure generated from a given level of Barr^CRH firing. The afferent drive also determines the probability of a high frequency Barr^CRH discharge in a given epoch. The bladder pressure-dependent synaptic drive for Barr^CRH is represented as a logistic relationship. This synaptic generator is commonly believed to be relayed via the PAG (Drake et al., 2010; de Groat and Wickens, 2013; de Groat et al., 2015) however, there is also evidence for direct spinal inputs to Barrington’s nucleus (in the rat) that could also act as a generator (Ding et al., 1997; Blok and Holstege, 2000). In addition, elegant recent studies indicate there are also direct functional inputs from the cortex and hypothalamus (Hou et al., 2016; Yao et al., 2018; Verstegen et al., 2019).

This circuit organisation generates NVCs: dynamic perturbations whose magnitude and frequency increase with progress through the micturition cycle. As pressure increases these contractions become more frequent and higher in amplitude – eventually summating to cause sustained increases in bladder pressure. This in turn increases the rate of firing of Barr^CRH, making further contractions more likely, and shifting the system into a positive feedback loop in which pressure rapidly increases. A void occurs when the pressure reaches 15 mmHg which is presumed to be effected via a spinal mechanism and the micturition cycle restarts.

In line with experimental data, attenuation of the variance in Barr^CRH firing (underpinning the NVCs) delays the time to void – indicating their importance in the process (Figure 13—figure supplement 1A). Similarly, augmenting the spinal parasympathetic sensitivity to the Barr^CRH drive (as seen experimentally with intrathecal Astressin) increases the amplitude of the NVCs and shortens the inter-void interval (Figure 13—figure supplement 1C). We note that additional drive into the Barr^CRH neurons (as is proposed to come from higher centres with voluntary voiding) would increase the variance and could trigger voiding earlier. This effect is demonstrated with the simulated optogenetic drive of Barr^CRH neurons (20 Hz x 1 s, Figure 13—figure supplement 1B) which produces both failures, eNVCs and triggers voids earlier in the cycle than would otherwise have happened.

The serotype two recombinant AAV-EF1α-DIO-hChR2(H134R)-mCherry (Nagel et al., 2005; Gradinaru et al., 2007) (1.6 × 10¹² viral genomes/ml) used for optogenetic activation experiments was obtained from University of North Carolina vector core facilities (a gift from Karl Deisseroth). The serotype 2 AAV-hSyn-DIO-hM4Di-mCherry (Krashes et al., 2011) (7 × 10¹² vg/ml) used for chemogenetic inhibition experiments was obtained from Addgene (a gift from Brian Roth). To target the locus coeruleus selectively for optogenetic activation we used CAV-PRS-ChR2-mCherry (2 × 10¹¹ pp/ml; Li et al., 2016).

To target Barr^CRH neurons, homozygous CRH^cre mice were anaesthetised with ketamine (70 µg/g) and medetomidine (0.5 µg/g) and placed in a small animal stereotaxic frame (Kopf, USA) with a drill-injection robot attachment (Neurostar, Germany). After exposing the skull under aseptic conditions, a small burr hole was drilled and AAVs were injected (200 nl x three injections per side) unilaterally or bilaterally through a pulled glass pipette at a rate of 100 nl/min. Injection coordinates for Barrington’s nucleus were 5.3 mm posterior to bregma, 0.70 mm lateral and 3.25, 3.5 and 3.75 mm below brain surface. Injections for Locus coeruleus were identical but targeted 0.8 mm lateral. After surgical procedures, all mice were returned to their home cage for at least 21 days for recovery to maximise protein expression.

To target Barrington’s nucleus, adult CRH^Cre mice had injections of AAV-DIO-ChR2-mCherry (1.6 × 10¹² vg/ml) or AAV-DIO-hM4Di-mCherry (as control). To target Locus coeruleus, adult C57Bl/6 mice had injections of CAV-PRS-ChR2-mCherry (2 × 10¹¹ pp/ml). Mice were used in experiments at least 3 weeks after vector injections. They were anaesthetised with urethane and prepared for cystometry as described below. Light from a 465 nm LED (Plexon, Dallas USA) was delivered in pulses with a maximum duty cycle of 50%. The light train was delivered once every 60 s for fixed-interval stimulation, or at randomised intervals between 30 s and 90 s. The light power exiting the fibre tip was set at approximately 10 mW and was measured before and after each experiment. For unilateral opto-activation light was delivered via a tapered optical fibre (Lambda-B, 0.39NA, 17 mm long, 1.2 mm emitting length, Optogenix, Italy) with the fibre lowered down the original vector injection track. For bilateral simultaneous opto-activation a dual fibre implant was used (DFC 200/250–0.66 15 mm DF1.4 C60, Doric, Canada) and coupled via a dual fibre optic cable to two separate LEDs.

For light delivery to the spinal cord, soft tissue was removed between T11 and T12 vertebral spines after skin incision. The exposed spinal cord was illuminated using a 473 nM laser (PhoxX, Omicron, Germany) via a bare ended fibre (Thorlabs, 400 µm) positioned above the cord and delivered in 20 ms pulses at 20 Hz for 5 s or a prolonged pulse of 1000 ms. The light power at the fibre tip was 29 ± 0.3 mW.

To inhibit Barr^CRH neurons, CRH^cre mice were bilaterally injected with AAV-DIO-hM4Di-mCherry (7.0 × 10¹⁰ vg/ml) into Barrington’s nucleus (as described above) and allowed at least 3 weeks of recovery (control mice had AAV-DIO-ChR2-mCherry injected). They were anaesthetised with urethane and prepared for cystometry as described below. Intraperitoneal CNO (5 mg/kg, 1 mg/ml stock) or saline (as control) was applied after obtaining >five baseline micturition cycles. In an initial set of experiments, saline was continuously infused to the bladder around the time of CNO injection to investigate the effects on mice micturition. Subsequently, to determine the CNO effect on threshold for micturition, a cyclical infuse and hold protocol was adopted whereby saline infusion was stopped at the threshold for voiding and then held at that volume for 10 min or until a void occurred before emptying the bladder and restarting the infusion phase.

Mice were anaesthetised with urethane (0.8–1.2 mg/kg) and the bladder was exposed via a 2 cm midline abdominal incision. A flanged catheter (PE50) was secured with a purse-string suture into the bladder and connected to a syringe pump and pressure transducer. The infusion rate was adjusted on an individual mouse basis (10–40 µl/min) to produce an equivalent proportionate speed of fill to threshold for voiding (typically 600 s) taking account of differing bladder volumes of the mice. External urethral sphincter (EUS) was recorded with insulated stainless steel wires, bared at the tip (0.075 mm, AISI316 Advent) inserted through a 30 G needle bilaterally into the EUS just proximal to the pubic symphysis. A balloon catheter (2.5 mm diameter x 12 mm when fully distended, Medtronic Sprinter) was inserted into distal colon and the tip of balloon was placed 40 mm from the anus. To monitor colonic pressure the balloon catheter was filled with distilled water.

Once a regular rhythm of micturition cycles was established (typically ~1 hr after starting saline infusion into the bladder) then the following variables were measured (and averaged over at least three voiding cycles (see Figure 1—figure supplement 2):

The pithed DAPM preparation was used to examine the influence of bladder filling on pelvic nerve stimulation-evoked bladder contractions. The methods were as previously described (Ito et al., 2018; Ito et al., 2019) but in brief, mice were terminally anaesthetised with isoflurane, disembowelled through a laparotomy and the bladder was cannulated. The mouse was then cooled, exsanguinated, decerebrated and its spinal column was pithed to remove all central neuronal control. It was then moved to a recording chamber, perfused through the heart with warm (32 °C) Ringer’s solution (composition (mM): NaCl (125), NaHCO₃ (24), KCl (3.0), CaCl₂ (2.5), MgSO₄ (1.25), KH₂PO₄ (1.25); glucose (10); pH 7.35–7.4 with 95% O2/5% CO2). Ficoll-70 (1.25%) was added as an oncotic agent to the perfusate. The flow rate was adjusted (from 15 to 20 ml/min) to achieve a perfusion pressure of 50–60 mmHg. The pelvic nerve was identified, traced proximally and cut allowing the distal end to be aspirated into a bipolar stimulating electrode. Stimuli (10V, 1 ms, 4–10 Hz for 3 s) were applied to the nerve. The bladder was filled with saline to perform cystometry as above with filling limited to a ceiling pressure of 15 mmHg. The effect of pelvic stimulation on bladder pressure was examined with different degrees of bladder filling (0–70 µl).

Recordings were made from Barrington’s nucleus and the locus coeruleus in urethane anaesthetised mice using a 15 μm thick silicone probe with 32 channels (NeuroNexus, Model: A1 × 32-Poly3-10mm-25 s-177-A32). For recordings in Barrington’s nucleus, the recording probe was lowered down a track using the same co-ordinates as the vector injection. An optical fibre was lowered on an intersecting track to target the nucleus from a caudal vector (bregma –8.8, ML 0.7 or 0.8 and 4.9 mm deep on an angle of 45° to the vertical). For recordings in LC, an identical configuration was used, but with ML coordinate 0.8 mm. Each channel (177 µm²) was spaced from the neighbouring channels by 50 μm. A reference electrode (Ag/AgCl) was inserted into the scalp. The probes were connected to an amplifier-digitising headstage (INTAN, RHD2132). The signals were amplified and filtered (100 Hz-3 kHz) and digitized at 30 kHz before being processed and visualised online within the Open Ephys system.

To investigate the Barr^CRH projection to the spinal cord, Cre-dependent AAV (AAV- EF1α-DIO-ChR2-mcherry) was unilaterally injected to Barrington’s nucleus in CRH^CRE mice. After a minimum of four weeks the mice were killed, and perfusion fixed for immunohistochemistry. To examine the Barr^CRH projection into spinal cord, 40 µm transverse sections were taken from T11 to S2, processed for mCherry and Choline acetyltransferase immuno- (to demarcate motoneurons) followed by confocal imaging (detailed below).

Mice were killed with an overdose of pentobarbital (20 mg per mouse, i.p; Euthetal, Merial Animal Health) and perfused trans-cardially with 4% formaldehyde (Sigma) in phosphate buffer (PB; pH 7.4, 1 ml/g). The brain and spinal cord were removed and post-fixed overnight before cryoprotection in 30% sucrose in phosphate buffer. Coronal tissue sections were cut at 40 µm intervals using a freezing microtome and left free floating for fluorescence immunohistochemistry. Tissue sections were blocked and incubated in phosphate buffer containing 0.3% Triton X-100 (Sigma) and 5% normal donkey serum (Sigma). Incubated on a shaking platform with primary antibodies for 14–18 hr at room temperature. After washing, sections were then incubated for 3 hr with appropriate Alexa Fluor secondary antibodies.

A Leica DMI6000 inverted epifluorescence microscope equipped with Leica DFC365FX monochrome digital camera and Leica LAS-X acquisition software was used for widefield microscopy. For confocal and tile scan confocal imaging, a Leica SP5-II confocal laser-scanning microscope with multi-position scanning stage (Märzhäuser, Germany) was utilized. Primary antibodies used were rabbit anti-mCherry (1:4,000; Biovision), sheep anti-tyrosine-hydroxylase (1:1,000; AB1542, Millipore) and goat anti-ChAT (1:250; AB144P, Millipore). Alexa Fluor 488-conjugated donkey secondary antibodies were used against goat IgG (1:500; Jackson ImmunoResearch) and sheep IgG (1:400; Jackson ImmunoResearch). Alexa Fluor 594-conjugated donkey secondary antibody was used against rabbit IgG (1:1000; Invitrogen).

A model of the integration of the synaptic drive to the bladder parasympathetic preganglionic neurons in the spinal cord was constructed using NEURON (Carnevale and Hines, 2006). The preganglionic neuron was based on using an existing preganglionic neuronal model (Briant et al., 2014) that was modified slightly to have a resting potential of ~−65 mV by altering the leak conductance reversal potential. A synaptic drive from Barr^CRH neurons was modelled by adding an EXPSYN to the soma which was driven with trains of action potentials (20 Hz x 1 s) generated from a NETSTIM to mimic an optogenetic stimulus of Barr^CRH. The synapse was subthreshold if triggered alone as an input despite a modest degree of summation (~20%) when driven at 20 Hz. A second fast excitatory synaptic drive to model a bladder afferent was added as an EXPSYN to a proximal medial dendrite. This was driven with an incrementing frequency of action potentials (from a NETSTIM based on recordings of pelvic nerve afferents from Ito et al., 2019) to model bladder distension-evoked increase in afferent firing over a period of 5 min to represent a typical micturition cycle. The model files are available from data.bris.ac.uk (DOI:10.5523/bris.20l920gl27ufi204brn8ilonsf).

The model of autonomous micturition employs an iterative algorithm in which bladder pressure and Barr^CRH firing rate are updated in each cycle (assumed to last 1 s). At each time point, bladder pressure is incremented by 0.015 mmHg (modelling a continuous infusion). The level of pressure in the bladder is used to determine the maximum level of Barr^CRH firing via a logistic function (modelling an afferent input):smax(t)=fmin+( fmax-fmin⁡)1+e-ks(p(t)-ms)

Where smax is the maximum spiking rate, and fmax and fmin represent the maximum and minimum levels of firing – here set to 25 and 4.3Hz respectively, based on spiking rates seen in unit data recordings. The gradient and mean were set to ks=4 and ms=4.5. Actual spiking rates were then generated probabilistically by sampling from a random uniform distribution with a maximum level set by smax.

The change in bladder pressure produced by the firing of Barr^CRH neurons was determined by a logistic function modulated by bladder pressure (representing the level of parasympathetic neuron excitability at the level of the spinal cord):ΔP(t)=ΔPmin+(ΔPmax-ΔPmin)1+e-kp(s(t)-mp)

Where s is the Barr^CRH firing rate and ΔPmax,ΔPmin⁡ are the maximum and minimum amplitude of bladder contractions, and kp=0.5 and mp=6. The maximum change in pressure depends on the current bladder pressure such that:ΔPmax=6*log⁡pt

The output of these equations is shown in Figure 13 and Figure 13—figure supplement 1. Bladder contractions are modelled with a Gaussian bump function, of amplitude ΔPmax and variance 2s. These had a duration of 6 timepoints (approximating the characteristics of non-voiding contractions observed experimentally). At the beginning of each cycle, the current pressure is calculated by summing the baseline pressure (including incrementing due to constant filling) with increases in pressure caused by Barr^CRH triggered bladder contractions generated in the current and past cycles.

Voiding occurs at a pressure of 15 mmHg, at which point bladder pressure is decremented at 2 mmHg per cycle (as the bladder empties) until a baseline pressure of <0.3 mmHg is reached and the cycle restarts. The model is coded in MATLAB.

In cycles where optogenetic activation of Barr^CRH firing was simulated, the firing rate s was set to 20Hz. To simulate intrathecal Astressin, the mean of the logistic curve for ΔP(t) was reduced to mp=5 To simulate attenuation of NVCs, the variance in the level of Barr^CRH firing was reduced by 80% but without any change in the mean level of firing.

Custom MATLAB scripts used to analyse the data are available along with example data are at DOI:10.5523/bris.20l920gl27ufi204brn8ilonsf along with the MATLAB code for the model of autonomous micturition and the Parasympathetic Preganglionic NEURON model.