The urinary bladder operates in two modes: storage and voiding. During the storage phase whilst the bladder is filling, the detrusor muscle relaxes to accommodate the increase in fluid volume, whilst sustained contraction of the external urethral sphincter (EUS) induced by tonic activity in the pudendal nerves maintains urinary continence (Fowler et al., 2008; De Groat et al., 2015). The act of micturition is dependent on the functional integrity of central control circuitry, which permits voiding to occur only when it is safe and socially acceptable for the individual to do so. When the central micturition circuitry switches from storage to voiding mode, activity in parasympathetic pelvic nerve efferents initiates contraction of the detrusor muscle and pudendal nerve activity is inhibited so that the EUS relaxes to allow urine to exit through the urethra (Fowler et al., 2008; De Groat et al., 2015).

Disorders of bladder control may arise from malfunction at any stage of the control machinery. Urge urinary incontinence (UUI), “a sudden and uncontrollable desire to void which is impossible to defer” (Abrams et al., 2003) is extremely common (prevalence 13.3% for men; up to 30.0% for women, depending on age; Milsom et al., 2014) and is rated as the most bothersome of lower urinary tract symptoms (Agarwal et al., 2014). The condition may arise due to hyperexcitability of the detrusor muscle (overactive bladder). Alternatively, the control circuitry may become hyperexcitable and/or the facility to inhibit voiding in inappropriate social situations may fail. Current drug treatments for UUI are aimed at reducing the excitability of the bladder. These can be effective although undesirable side effects are not uncommon (Reynolds et al., 2015; Olivera et al., 2016). In patients refractory to pharmacological treatment intravesical injection of botulinum toxin may be offered, although this requires frequent re-injection for continued benefit. There is also a risk of urinary tract infection and high residual volume, which may require patients to perform self-catheterization (Gupta et al., 2015; Tubaro et al., 2015; Olivera et al., 2016; Truzzi et al., 2016).

Neuromodulation is another approach used to decrease the excitability of the overactive bladder. Percutaneous tibial nerve stimulation (PTNS) and sacral nerve stimulation (SNS) have both been adopted as clinical procedures for urge incontinence (Yamanishi et al., 2015). However, PTNS requires a frequent schedule of re-application whereas high levels of re-intervention are required with SNS and pain is a frequent side effect (Gupta et al., 2015; Tubaro et al., 2015; Olivera et al., 2016; Truzzi et al., 2016).

An alternative approach would be to suppress voids only when required i.e., at the onset of urge. Charge-balanced kilohertz frequency alternating current (KHFAC) has been shown to produce rapid onset, reversible conduction block in both myelinated and unmyelinated peripheral nerves (Joseph and Butera, 2009, 2011; Kilgore and Bhadra, 2014; Patel and Butera, 2015). We therefore considered whether KHFAC stimulation of the pelvic nerve might be employed to modulate urinary voiding. In a rat model we investigated whether KHFAC, initiated at the onset of an imminent involuntary void in rats, when humans would be expected to experience extreme urge sensation, could abort the void, and maintain urinary continence.

The study conforms to the national guidelines for the care and use of animals and was carried out under the authority of UK Home Office Project License PPL30/3200 and approved by the Local Ethical Committee of the University of Bristol. Every effort was made to minimize the risk of animal's pain or suffering. At the end of the experiment the animals were killed by an overdose of anesthetic.

Female Wistar rats (n = 30, 196–254 g) were obtained from Charles River UK Ltd and housed at the University of Bristol Animal Services Unit. They were anesthetized with urethane (1.4 g kg⁻¹ i.p.) and the right femoral artery and right femoral vein were cannulated to record, respectively, arterial blood pressure and heart rate and for infusion of fluids. The trachea was cannulated to maintain a patent airway and monitor respiratory airflow. Rectal temperature was maintained at 37°C by a homeothermic blanket system (Harvard Apparatus, Holliston, Massachusetts, USA). The depth of anesthesia was assessed throughout the experiment by monitoring the pedal reflex, blood pressure, and heart rate. If required, supplementary anesthesia and fluid replacement were administered via the cannula in the femoral vein.

Animals were positioned supine, and a midline laparotomy was performed. The pelvic nerve was located and gently separated from the uterine wall. The distal portion of the preganglionic pelvic nerve bundle was fitted with a custom made bipolar cuff-electrode (platinum iridium wire with cobalt core in silicone epoxy; Figure 1). Stimulus trains of varying waveform, pulse duration, intensity and frequency were delivered using a stimulus isolation device (STMISOLA, Biopac Systems Inc., Santa Barbara, USA), driven by an alternating voltage waveform generator.

The bladder dome was cannulated by piercing it with a 25 G needle tip attached to a length of saline-filled polythene tubing. The tubing was attached to a T-piece, enabling recording of intravesical pressure during infusion of saline into the bladder. In 17 experiments two insulated platinum wire electrodes were inserted into the space between the pubic symphysis and the EUS muscle to record electromyogram (EMG) activity (Figure 1).

In five experiments two insulated stainless steel needle electrodes were inserted into the abdominal wall to record EMG of the abdominal muscles. EMG activity was amplified (5000x) using a Neurolog system (Digitimer Ltd, Welwyn Garden City, Hertfordshire, UK) and digitally bandpass filtered (0.1–500 Hz) offline, using Matlab R2014a. Stimulus artifacts present in the EMG signal resulting from low frequency pulse trains (e.g., Figures 2D–F), were excluded off-line by subtraction of the averaged stimulus-triggered waveform. Uterine and rectal pressures were monitored either by inserting a balloon catheter (2F Embolectomy catheter, Intra special catheters GmbH, Rehlingen-Siersburg, Germany) into the vagina and advancing it until the tip reached the uterus (n = 9), or into the rectum via the anus (n = 5). The timing of each drop of urine exiting the urethra was counted by visual observation and logged electronically. All data were captured and displayed using a PowerLab 8SP data acquisition system running Chart v5 software. Statistical tests were carried out using Graphpad Prism v7. Group data are reported as mean ± standard error of the mean (SEM) unless otherwise specified.

Our experimental design followed the following sequence: verification of functional responsiveness of the pelvic nerve to bladder connection; establishing repeated reflex voiding in response to continuous infusion of saline into the bladder; determination of optimal parameters for stimulation of pelvic nerve to inhibit voids whilst producing minimal “off-target” side effects; investigation into the mechanism of action.

In urethane-anesthetized rats continuous infusion of saline into the bladder evoked repeated voiding, in agreement with previous reports (Kruse et al., 1990; Matsuura et al., 2000; Stone et al., 2011; Crook and Lovick, 2016). High frequency stimulation of the pelvic nerve initiated at the onset of an imminent void suppressed voiding. The effect was rapid in onset, reversible and reproducible and when stimulation parameters had been optimized to inhibit voiding, there were only minimal “off target” side effects. This powerful inhibitory effect on voiding was evoked by unilateral pelvic nerve stimulation. Indeed, there was no advantage to be gained by stimulating bilaterally.

In the present study spontaneous voiding in response to infusion of saline into the bladder persisted following unilateral nerve section. Although the bladder is innervated bilaterally, in rats the pelvic nerve trunks on either side each innervate the whole of the bladder (Carpenter and Rubin, 1967). Gap junction coupling between cells (Fry et al., 2004) also enables the detrusor to act as a functional syncytium. Thus, activity in only one pelvic nerve appears to be sufficient to produce a co-ordinated contraction of the detrusor and to initiate reflex bursting activity in the EUS.

Low frequency stimulation of the pelvic nerve evoked a rise in bladder pressure accompanied by contraction of the EUS, in agreement with previous reports (Danziger and Grill, 2016). In contrast, high frequency stimulation of the pelvic nerve, started within 1–2 s of the sharp rise in bladder pressure that signaled an imminent void, aborted the void and no urine escaped. The mechanism underlying this effect is intriguing. In other unmyelinated nerves high frequency stimulation has been shown to produce conduction block (Joseph and Butera, 2009, 2011; Patel and Butera, 2015). However, in the present study nerve block seems unlikely to have been a major underlying factor in inhibiting micturition. Firstly, if a nerve block had occurred under the electrode, the functional integrity of the contralateral nerve should have been sufficient to complete the void. In fact, voiding was suppressed during ipsilateral stimulation even when the contralateral nerve had been sectioned. Secondly, when the ipsilateral nerve was ligated distally, so that the repeated voiding prior to stimulation must have been mediated by activation of the contralateral nerve, high frequency stimulation ipsilaterally was still able to suppress voiding. These factors indicate that the effect of high frequency stimulation was mediated by a signal relayed centrally, which in some way blocked the generation of a motor command signal to the bladder.

Previous studies have shown that low frequency pelvic nerve stimulation (10–20 Hz) can evoke a long-lasting inhibition of spontaneous isovolumetric bladder contractions (De Groat and Ryall, 1968; De Groat, 1976). This effect was due to recurrent inhibition, via crossed and uncrossed pathways, of parasympathetic preganglionic neurons supplying the bladder whose axons were activated antidromically by stimulation of the nerve. However, unlike in the present study, the inhibition could be evoked only when bladder volume and resting pressure were low. This lead the authors to conclude that the effect was unlikely to be operative during the micturition reflex, which is triggered when bladder volume and pressure are relatively high, and more likely functioned during storage as an adjunct to the guarding reflex (De Groat and Ryall, 1968).

High frequency stimulation has been reported to produce axonal conduction block in invertebrate and mammalian nerves (Hulsebosch and Coggeshall, 1982; Bhadra and Kilgore, 2005; Joseph et al., 2007; Joseph and Butera, 2009, 2011; Patel and Butera, 2015). However, effective stimulation frequencies for blocking conduction in unmyelinated fibers (20–30 kHz; Joseph and Butera, 2009, 2011; Patel and Butera, 2015) were much higher than the optimal frequencies (1–3 kHz) that inhibited voiding in our study. Indeed we found stimulus frequencies above 10 kHz to be ineffective. Moreover, since inhibition of voiding persisted after ligation of the pelvic nerve distal to the stimulation site, which blocked conduction to the bladder, it is unlikely that a stimulation-induced blockade of pelvic nerve efferent transmission made a significant contribution to the effects seen in the present study. In support of these findings a recent modeling study (Pelot et al., 2017) predicts that stimulation of small diameter fibers within the range of effective parameters used by us, is likely to excite the nerve and impose on it a firing pattern that is physiologically meaningless with respect to voiding. When transmitted to the spinal cord and supraspinal micturition ccontrol, this would effectively block the circuitry setting up the motor command pattern to generate a void.

High frequency pelvic nerve stimulation that supressed voids, always evoked sustained contraction of the EUS, an effect that undoubtedly contributed toward maintaining continence. The pelvic nerve is a mixed nerve containing small myelinated afferents as well as motor fibers (Hulsebosch and Coggeshall, 1982; Park et al., 1997; Shea et al., 2000; D'Amico et al., 2011). Activation of stretch-sensitive Aδ bladder afferents during bladder filling (Moss et al., 1979) evokes reflex tonic contraction of the EUS: the guarding reflex (Park et al., 1997; D'Amico et al., 2011; Danziger and Grill, 2016). The sustained increase in tonic activity in the EUS during high frequency pelvic nerve stimulation is likely due to activation of these afferent fibers.

Stimulation of pelvic nerve afferents would also be expected to excite the spino-midbrain-spinal micturition control circuit that relays in the midbrain periaqueductal gray and pontine micturition center, which is essential to initiate co-ordinated voiding (De Groat et al., 2015). The mechanosensitive bladder afferents that respond to physiological levels of activation rarely fire in excess of 15 Hz (De Groat and Ryall, 1969; Shea et al., 2000). It is unlikely they would follow faithfully the stimulation in the low kHz frequency range used in the present study. Nevertheless, the stimulation would impose an unphysiological pattern of afferent activity, which would be transmitted to the central micturition control circuitry. This might make it impossible for the circuitry to initiate the co-ordinated activity in the spinal outflows to the detrusor and EUS that are required to produce a void. Whether such an effect would occur within midbrain or spinal levels is not clear. However, electrical stimulation of the midbrain part of this loop has also been shown to be able to block co-ordinated voids (Stone et al., 2015). The authors proposed that by imposing an unphysiological pattern of firing, the stimulus “jammed” the micturition circuitry in the manner that electrical signals are used to jam radio transmission (Stone et al., 2015). A similar mechanism, operating at spinal and/or midbrain level, may have contributed to the functional inhibition of voiding evoked by high frequency pelvic nerve stimulation.

Notwithstanding uncertainties about the precise underlying mechanism, the present study has demonstrated the ability of high frequency pelvic nerve stimulation to suppress imminent urinary voids. The rapid onset of the effect, its ready reversibility and the absence of significant effects on other organ systems suggests that pelvic nerve stimulation merits consideration as an alternative approach to manage urinary urge incontinence (UUI) in humans. Clinically, UUI may have a diversity of underlying causes. The current experiments were carried out in a rat preparation with no bladder pathology in which we were able to suppress uncontrollable, but essentially normal voiding. It will be important to replicate the findings in other models of UUI which involve bladder pathology.

TL conceived and designed the experiments with input from JC. JC performed the experiments and analyzed the data with input from TL. TL wrote the first draft of the manuscript, which was subsequently edited by JC and TL. All authors approved the final manuscript.