Corresponding author M. Sasaki: Department of Physiology, Tokyo Medical University, 6-1-1 Shinjuku, Shinjuku-ku, Tokyo 160, Japan. Email: email@example.com
The purpose of the present study was to clarify how Barrington's nucleus regulates bladder contractility. Single neurones that discharge at higher rates during micturition contraction were recorded from Barrington's nucleus. Spinal-projecting neurones were identified by antidromic stimulation of the spinal cord. Seventy-six spinal-projecting neurones were classified into four types based on the firing patterns displayed during the relaxation phase of the micturition contraction–relaxation rhythm: (1) ramp-tonic neurones displayed a ramp increase in firing throughout the relaxation phase, (2) ramp-silent neurones were silent initially during the relaxation phase and displayed a ramp increase later, (3) flat-tonic neurones fired constantly, and (4) flat-silent neurones displayed little firing, being virtually silent throughout relaxation. During the relaxation phase, discharge volleys from Barrington's nucleus to sacral neurones were estimated to increase exponentially as micturition contraction approached. Twenty-two neurones increased firing even further within 3 s of micturition contraction, suggesting that they are involved in the final stages of initiation of micturition contraction. During micturition contraction, 18 neurones (of which 14 belonged to the ramp-silent class) displayed maximal firing rates before maximal bladder pressures were reached; firing gradually decreased during micturition contraction. Thirty-nine neurones (of which 25 belonged to the ramp-tonic class) displayed constant firing during micturition contraction. This suggests that ramp-silent neurones might be involved in increasing bladder pressure rapidly and strongly via feed-forward regulation, while ramp-tonic neurones might be involved in maintaining high bladder pressure via positive feedback from the bladder afferents. Sixty neurones continued to fire for 1–8 s after the onset of bladder relaxation, suggesting that Barrington's nucleus does not trigger bladder relaxation.
Previous anatomical, electrophysiological and pharmacological studies have revealed that the micturition reflex centre is localized in a limited area of the dorsolateral pontine tegmentum in several species (for review, see Yoshimura & de Groat, 1997). In cats, a distinct population of neurones located ventral and adjacent to the mesencephalic tract of the trigeminal nerve (5MET) projects directly to the sacral parasympathetic nucleus (Holstege et al. 1979, 1986; Blok & Holstege, 1994, 1997, 1999). The same area has also been labelled transneuronally in a retrograde direction after inoculation with the pseudorabies virus (PRV) into the bladder wall (de Groat et al. 1998). Electrical stimulation of this area in particular causes bladder contraction, inhibition of the external urethral sphincter, and relaxation of the urethra (Holstege et al. 1986; Griffiths et al. 1990), supporting the idea that this restricted part of the pontine tegmentum is a central component of the micturition circuit. The site has been described as the pontine micturition centre, the M-region, or Barrington's nucleus.
In rats (Willette et al. 1988; Tanaka et al. 2003) and cats (de Groat et al. 1998), there are two types of neurones in Barrington's nucleus that increase or decrease their discharge during micturition contraction. Considering that electrical or chemical stimulation evokes bladder contraction, neurones that increase firing during micturition contraction are expected to be closely associated with bladder contraction. It has been assumed that neurones that are completely silent in the absence of a voiding reflex, but that increase their firing just prior to a bladder contraction, play an essential role in the initiation of micturition, and may be the source of bulbospinal axons carrying excitatory signals to the sacral parasympathetic nucleus (de Groat et al. 1998; Tanaka et al. 2003). There exist neurones that send axons to the sacral cord, fire tonically in the absence of a voiding reflex, and exhibit high frequency firing during bladder contraction (Sasaki, 2002). Despite the importance of Barrington's nucleus in the micturition circuit, the mechanism controlling bladder contractility associated with this nucleus remains unclear.
The present study aimed to elucidate the functional role of Barrington's nucleus in controlling micturition contraction. For this purpose, we sought to record neurones, both within and around Barrington's nucleus, that increased their firing during micturition contraction. Identification of the spinal projection of Barrington's neurones is essential, since neurones that send descending axons to the spinal cord (spinal-projecting neurones) are expected to be directly associated with bladder contractility. Projections to the spinal cord are commonly examined by antidromic stimulation of the spinal cord, and thus we employed this method. In the present study different classes of spinal-projecting neurones were encountered that displayed several firing patterns, not only during absence of micturition contraction but also during micturition contraction. Two different types of regulating mechanisms for evoking and maintaining micturition contraction will be presented. Based on our findings, and those of others, the idea is discussed that tonic firing in the absence of the contraction has a potential role in the initiation of micturition contraction, and that the neuronal circuit within Barrington's nucleus lacks a neural mechanism to cause bladder relaxation. Preliminary results have been presented in abstract form (Sasaki, 2000a,b).
Experiments were performed on 32 adult cats of both sexes weighing between 2.8 and 5.9 kg. All experimental procedures were approved by the Tokyo Medical University Institutional Animal Care and Use Committee in accordance with NIH Guidelines for the Care and Use of Laboratory Animals. Cats were initially injected with ketamine (100–150 mg; i.m.). After a tracheotomy was performed, animals were prepared under halothane–N2O anaesthesia (1.0–1.5% halothane in a 33% N2O and 67% O2 inhalation mixture) before recording. A femoral artery and vein were cannulated for blood pressure recording and drug administration. The urethra was cannulated for bladder-pressure recording. Animals were then mounted in a spinal cord recording frame and a stereotaxic frame. Preparation included craniotomy of the interparietal bone, partial aspiration of the anterior cerebellum for insertion of a recording electrode, and laminectomy for electrical stimulation of the spinal cord. Throughout surgery, cats were artificially ventilated without immobilizing drugs. The effectiveness of the anaesthesia was assessed by corneal and pedal withdrawal reflexes and adjusted accordingly. Halothane was then discontinued, and the animals were anaesthetized throughout the remainder of the experiment with α-chloralose (initial dose: 35–40 mg kg−1; i.v.). After this anaesthetic change, there was no sign of animal movement for 1–2 h. Then, additional doses of 3–5 mg kg−1 of α-chloralose (i.v.) were injected every 1 or 2 h; the total dosage over the course of the experiment was 60–100 mg kg−1. Anaesthesia level was monitored continually.
To prepare for recording and stimulation, the dura was opened, and the exposed spinal cord was covered with mineral oil to prevent desiccation. Experimental equipment for recording units and for stimulating the spinal cord was set up and adjusted. During recording, the animals were subject to neuromuscular block with pancuronium bromide (Mioblock, Organon, Holland; 0.06–0.12 mg kg−1 h−1, i.v.) and artificially ventilated, and as stated above, anaesthesia level was maintained with α-chloralose. Rectal temperature was maintained around 38°C with a thermo-regulated heating blanket. Blood pressure was recorded with a chart recorder (RTA1200, Nihon Kohden, Japan) throughout the experiments. Blood pressure was stable in all animals, and their pupils, which were checked several times an hour, were constricted, showing that the anaesthesia was maintained at a level at which no indication of pain could be detected in intact animals. Also, to ensure that the neuromuscular blocking agent was not masking signs of animal discomfort and to confirm that the anaesthetic level was appropriate, we discontinued the pancuronium bromide once or twice for 1–2 h during the experiments. At the end of the experiment, animals were deeply anaesthetized and were perfused with 20% formalin, and the brainstem was removed for histological identification of recording and stimulating sites. The brainstem was sectioned serially (100 μm) in the transverse plane, and sections were stained with cresyl violet.
Preparation of the bladder
A catheter (size, Fr 5) was inserted through the urethral orifice in 19 male cats. In 13 female cats, the urethra was exposed via an abdominal incision, and a catheter was inserted through a small puncture made in the urethra about 1–2 cm distal to the bladder neck. The catheter was connected via a three-way stopcock to a low-pressure transducer (LPU-0.1, Nihon Kohden) for measurement of the intravesical pressure, or to a syringe for infusion or withdrawal of the intravesical fluid. The micturition contraction–relaxation rhythm under this isovolumetric condition was displayed on the chart recorder.
Microstimulation of the spinal cord was done using a carbon monopolar electrode insulated with a glass pipette, except for the tip (diameter of the carbon fibre, 8 μm; uninsulated tip length, 10–20 μm). Prior to each experiment, the optimal position for cord stimulation for evoking bladder contraction was systematically examined by stimulating the cord with 30–50 train pulses (200 Hz, 50–200 μA, 150 μs cathodal square pulses). Bladder contraction with a 0.25–0.5 s latency was effectively evoked from the surface of the dorsolateral funiculus (DLF) of the spinal cord. For eliciting antidromic spikes in the pontine tegmentum, one or two stimulating electrodes were placed on the surface of the DLF at the upper lumbar cord (the border between the thoracic (T13) and lumbar cord (L1) (indicated as T/L hereafter) or between L1 and L2 (L1/2); 17 animals), at the middle lumbar cord (L3/4 or L4/5; 15 animals), and/or at the lower lumbar cord (L6/7 or L7/S1; 21 animals). In one animal, a carbon electrode was inserted into the lateral funiculus caudal (one segment) to that placed on the surface of the DLF; the most effective sites were explored by moving the electrode up and down through the funiculus.
Recording of single units
Glass micropipettes filled with 2 m NaCl saturated with fast green FCF dye (resistance 2–5 MΩ) were used for extracellular recording of single units. Recordings were made with a differential amplifier (AB-651J, Nihon Kohden; low cut: 50 Hz). Spikes of single units were counted with pulse-counting units (QC-111J and ET-612J, Nihon Kohden). Recording electrodes were angled 30–45 deg in the rostral direction from the Horsley-Clarke frontal plane. Exploration for units was restricted to the region around the dorsolateral pontine tegmentum 1.5–3.5 mm lateral to the midline, 0–4 mm caudal to the level of the trochlear decussation, and up to 3 mm deep from the surface of the midline. When bladder-contractility-related units were found, they were further examined for antidromic activation from the spinal cord. Antidromic spikes evoked from the stimulating electrode were tested for positive collisions with spontaneous spikes generated in the single unit to verify that they were being emitted from the same neurone. Neuronal firing, firing rate and bladder contractility were displayed on the chart recorder, and selected units were also recorded with a DAT recorder (PC-108M, Sony, Japan) for off-line analysis. Digitized waveforms of antidromic spikes were stored on disk with a conventional computer system. In each experiment, the location of a recorded single unit was marked with iontophoretically deposited fast green FCF dye by passing negative current (30–40 μA, 10–30 min) before changing recording electrodes. The recording sites of single units were reconstructed with reference to the dye marks and the depths of the microelectrodes noted during the experiment.
In the present study 185 neurones were recorded from the dorsolateral pontine tegmentum that increased their firing during micturition contraction. Neurones displayed a variety of firing patterns during the relaxation phase of the micturition contraction–relaxation rhythm, as well as during micturition contraction. By evaluating three to four cycles of the rhythm, neurones were classified into two main groups and four subgroups based on their firing properties during the relaxation phase. They were also examined for spinal projections by antidromic stimulation in the upper lumbar, middle lumbar, and/or lower lumbar segments (Table 1). One hundred and four single units were verified to be spinal-projecting neurones; the stimulus-current threshold for evoking antidromic spikes was 16–500 μA. The remaining 81 neurones were not activated antidromically from the spinal cord, even with stimulus intensities equal to or exceeding 1 mA. The data of this group are described separately in later sections.
Table 1. Spinal levels of antidromic stimulation in 185 dorsolateral pontine neurones
Spinal level of antidromic stimulation
‘+’: antidromically activated neurones, ‘−’: antidromically not-activated neurones. The sum of neurones (‘T/L’+‘L3–5’+‘L7-L/S’) is larger than the number of neurones in the third raw (no.), because one or two different level(s) were stimulated in each neurone.
Antidromically activated neurones
Neurones firing with a ramp increase. Figure 1 shows a typical example of a dorsolateral pontine neurone that increased its firing when the bladder contracted. This neurone displayed spontaneous low-frequency firing, which increased gradually throughout the relaxation phase of the micturition contraction–relaxation rhythm (Fig. 1A). During micturition contraction, the neurone displayed high-frequency firing. This increase occurred almost simultaneously with the onset of micturition contraction (Fig. 1C; fast sweep). Antidromic spikes with a fixed latency of 29.6 ms were elicited in this single unit after stimulation of the DLF at the L/S level (Fig. 1D; filled circle). Antidromic spikes collided with the preceding spontaneous spikes when the interval between spontaneous spikes and antidromic stimulation was shortened to less than the latency of antidromic spikes plus the refractory period (cf. Fig. 1D and E), indicating that antidromic and spontaneous spikes were evoked in the same neurone and that this neurone was a spinal-projecting neurone. Such collision tests were performed in all antidromically activated neurones.
Another neurone, exemplified in Fig. 2A, was silent in the early stage of the relaxation phase. It started to fire when the bladder was still quiescent, and discharge gradually increased. As the firing increased further, bladder contractions eventually began (Fig. 2B; fast sweep). Discharge reached a maximum during the rising phase of contraction.
While the above two types of neurones displayed a ramp increase during the relaxation phase, one type fired tonically and the other type was silent during the early stage of the bladder relaxation. To clarify whether such firing properties were inherent to individual neurones or were a class property, the influence of the base intravesical pressure on firing patterns was examined. When lowering the base pressure to slightly above the micturition threshold (from 108.1 mmH2O to 88.3 mmH2O in the mean) by withdrawal of intravesical fluid (4–6 ml), 3 of 8 neurones ceased firing at the early stage of bladder relaxation (cf. Fig. 3A and B). The remaining five neurones displayed tonic firing throughout the relaxation phase: they still displayed tonic firing when the base pressure was lowered to below the micturition threshold, but firing of them ceased within 10 min when the bladder was emptied. On the other hand, when the base pressure was increased by infusing fluid into the bladder (mean increase of the base pressure: 14.4 mmH2O), 3 of 4 neurones still were silent during the early stage (cf. Fig. 3C and D). The remaining one neurone that was previously silent during the early stage began to display a ramp increase throughout the relaxation phase in response to increasing base pressure. These observations indicate that there are two general types of neurones that have different firing patterns in the relaxation phase. However, some of these could not clearly be assigned to different groups. Therefore, the neurones displaying a ramp increase in the relaxation phase were classified as ‘ramp neurones’en bloc. The neurones displaying a ramp increase throughout the relaxation phase (e.g. Fig. 1) were expediently subclassified as ‘ramp-tonic neurones’, and neurones that were silent in the early stage (e.g. Figs 2 and 3C–D) were subclassified as ‘ramp-silent neurones.’ The kind of neurone presented in Fig. 3A and B, which displayed both characteristics, was subclassified into a ramp-tonic type, since it recovered quickly, exhibiting tonic firing early in the relaxation phase (after the 2nd contraction in Fig. 3B and following contractions).
Thirty ramp-tonic and 18 ramp-silent neurones were recorded from the dorsolateral pontine tegmentum. Among ramp-silent neurones, the onset of the ramp increase varied from 3 to 26 s before the onset of bladder contraction (Fig. 4A). Figure 4B shows the cumulative histogram derived from the histogram of Fig. 4A. This shows that the number of active ramp-silent neurones increases exponentially as initiation of micturition contraction approaches.
Silent neurones and neurones firing constantly in the relaxation phase. The neurone presented in Fig. 5A displayed relatively constant firing during the relaxation phase. The firing patterns of this class of neurones were clearly different from those of the ramp-tonic neurones, which displayed a ramp increase throughout the relaxation phase. The firing rate of this neurone started to increase prior to micturition contraction, and continued to increase around the onset of micturition contraction without a clear relationship to the onset (Fig. 5B; fast sweep). Other neurones (Fig. 6) were silent during the relaxation phase, and firing increased after (Fig. 6A) or before (Fig. 6B) the onset of micturition contractions. The neurone shown in Fig. 6B displayed burst-like firing preceding small amplitude bladder contractions (upward arrows with a dashed line). This neurone, however, displayed continuous firing during relatively large bladder contractions (from upward arrow with a straight line). The firing rate was maximum at the rising phase of contraction and began to decrease before peak bladder pressure was reached.
When lowering the base pressure to a level slightly above the micturition threshold by withdrawing intravesical fluid (4–8 ml; mean reduction of the base pressure: 18.6 mmH2O), 2 of 9 neurones that previously fired tonically in the relaxation phase became silent. The remaining seven neurones displayed tonic firing throughout the relaxation phase. The influence of the base pressure was further examined in four neurones: three neurones still displayed tonic firing when the base pressure was lowered to below the micturition threshold, and firing of these neurones ceased within 2 min when the bladder was emptied. In contrast, 1 of 2 neurones that were previously silent in the relaxation phase began to discharge tonically in response to increasing base pressure caused by infusing fluid into the bladder. With the remaining one neurone, the firing pattern remained unaffected by this procedure. Because these neurones could not be classified simply into two groups, as in the case of ramp neurones, they were classified en bloc as ‘flat neurones.’ Neurones with persistent tonic activity and those that were silent in the relaxation phase were expediently subclassified as ‘flat-tonic neurones’ and ‘flat-silent neurones’, respectively. Nineteen flat-tonic and nine flat-silent neurones were recorded from the dorsolateral pontine tegmentum.
Firing rates were measured in the early and late stages of the relaxation phase. For the early stage, rates were measured when micturition contraction was complete and bladder pressure fell to the base pressure (see, for example, upward arrow in Fig. 1A), while for the late stages, rates were measured just before the neurones displayed an abrupt increase associated with the onset of micturition contraction (see, for example, ▲ in Fig. 1A). Firing rates in the early stage ranged from 1 to 10 Hz (mean value ±s.d. 4.7 ± 3.7 Hz; n= 30) in ramp-tonic neurones and 1–18 Hz (5.7 ± 4.9 Hz; n= 19) in flat-tonic neurones. All ramp-silent and flat-silent neurones did not discharge during this stage. Firing rates in the late stage ranged from 7 to 24 Hz (12.7 ± 6.2 Hz) in ramp-tonic neurones, 1–21 Hz (5.8 ± 3.4 Hz) in flat-tonic neurones, 5–22 Hz (11.3 ± 5.9 Hz; n= 18) in ramp-silent neurones and 0–3 Hz (1.0 ± 1.4 Hz; n= 9) in flat-silent neurones.
Analysis of firing properties during the micturition phase
As discussed above, spinal-projecting neurones were classified into four subgroups based on firing patterns recorded during the relaxation phase. These neurones also displayed differences in firing patterns during the micturition phase. Two different types of neurones discharged at or around the onset of micturition contraction; examples are shown in Fig. 6. One neurone (Fig. 6B) started firing prior to the contractions, while the other neurone (Fig. 6A) fired after the onset of micturition contraction. Tonically firing neurones also increased their firing even further just prior to (Figs 2B, 5B and 7A) or after (Fig. 3B) micturition contraction. Of 76 ramp and flat neurones analysed, 21 increased their firing rates 0.2–3 s before (Fig. 7B, filled bars) the onset of micturition contraction, and 25 neurones increased their rates 0.1–2 s after (Fig. 7B, open bars) contraction. These two firing patterns were observed in all four subtypes of neurones. A firing increase occurring before contraction was most frequently observed in ramp-silent neurones. In contrast, a firing increase occurring after the onset of contraction was most frequently observed in ramp-tonic neurones. The patterns of the remaining 30 neurones (stippled bar) could not be determined because the firing increase occurred too close to the somewhat unclear onset of micturition contraction (typically less than 0.1 s; for example, see Fig. 1C).
During micturition contraction, neurones displayed three different firing patterns: constant, decrement, and rise. Neurones with constant-type firing (Figs 3A–B and 5) displayed relatively constant firing rates, even when bladder pressure fluctuated by small amounts. Neurones with decrement-type firing (Figs 2 and 7A) displayed maximal firing rates at the rising phase of micturition contraction. Firing began to decrease before bladder pressure reached a maximum and continued to decrease, even when the bladder was maintained at the highest pressure. Neurones with rise-type firing (Figs 1, 3C–D and 6B) displayed firing increases that coincided with small rises in bladder pressure; firing rate began to decrease before peak bladder pressure was achieved (see Fig. 3E; fast sweep). Although the firing pattern of this type generally resembled that of the decrement-type, neurones with rise-type firing did not gradually decrease firing during micturition contraction.
The firing patterns of 76 neurones were examined during micturition contraction. Neurones were classified as the decrement-type if their firing rate at the last stage of micturition contraction was less than 90% of the maximal firing rate (see Fig. 7A). Of the 18 neurones with decrement-type firing, most (n= 14) were ramp-silent neurones, and the remainder were either ramp-tonic (n= 2) or flat-tonic (n= 2) neurones (Fig. 7C, filled bar). The firing rate at the last stage of micturition contraction ranged between 56 and 89% (mean = 71.5%) of the maximal rates (range: 13–51 Hz; mean value ±s.d. 29.1 ± 12.1 Hz) achieved for the 18 neurones showing decrement-type firing. In contrast, of the 39 neurones with constant-type firing, most (n= 25) were ramp-tonic neurones, and the remainder were either ramp-silent (n= 2), flat-tonic (n= 7), or flat-silent neurones (n= 4) (Fig. 7C, open bar). Of the 20 neurones with rise-type firing, most (n= 15) were flat neurones, and the remainder were ramp neurones (n= 5) (Fig. 7C, hatched bar). The average firing rates during micturition contraction for the constant type was 10–44 Hz (24.7 ± 11.8 Hz), and 11–42 Hz (20.0 ± 10.1 Hz) for the rise-type.
After the onset of bladder relaxation, high firing rates persisted for a while in some neurones (see Figs 1, 2, 3 and 7A; between arrowhead and downward arrow); these rates were comparable to those recorded during micturition contraction. In other neurones (e.g. Fig. 6B), firing rates began to decrease before the onset of relaxation. The histogram of Fig. 7D shows the frequency distribution of neurones showing persistent high-frequency firing at the indicated times after the onset of the bladder relaxation (plus values of abscissa), or showing firing decreases at the indicated times before relaxation (minus values). For most ramp neurones (45/48), high-frequency firing was maintained for 1–8 s after the onset of bladder relaxation (Fig. 7D, open columns). The remaining 3 ramp neurones started to decrease their firing at approximately the onset of bladder relaxation (see Fig. 5A). The mean duration of firing after relaxation onset for ramp-tonic and ramp-silent neurones was 3.9 s (n= 30) and 3.2 s (n= 18), respectively. In contrast to ramp neurones, flat neurones tended to show persistent firing of a shorter duration after relaxation onset, or a firing decrease before relaxation onset, as is clear in the distinct leftward bias in the frequency distribution (Fig. 7D, symbols). About half (n= 15) of this neurone type fired continuously after bladder relaxation, while the remaining ones started to decrease their firing 2–4 s prior to (n= 7), or around (n= 6) the onset of, the bladder relaxation. All seven neurones that decreased firing before the onset of the micturition contraction were the rise type. Mean values were 1.2 s for flat-tonic neurones and –0.2 s for flat-silent neurones.
Unclassified neurones. The remaining 28 out of 104 spinal-projecting neurones also displayed elevated firing during bladder contraction. Sixteen neurones were tonically active during the relaxation phase, and 12 were silent during the initial stage of relaxation. Several uncontrollable, intermittent factors precluded us from grouping these neurones into our firing pattern classification scheme. For example, sometimes the relaxation phase was too short (less than 10 s), amplitudes of micturition contractions were too small (typically less than 200 mmH2O), and/or contractions were too brief (typically less than 5 s) to analyse firing patterns in the relaxation phase as well as during micturition contractions. Other times, action potentials of individual spinal-projecting neurones could not be discriminated confidently from those arising from nearby single units. Therefore, they were not classified into groups. These neurones did not display firing patterns that were any different from those observed for ramp and flat neurones.
Latency and conduction velocity of antidromic spikes
Antidromic spikes were evoked from the lower lumbar segments (L7 or L/S) in 70 neurones (Table 1). The latency of antidromic spikes ranged between 13.3 and 40.8 ms. The mean latencies for classified neurones were as follows: (1) ramp-tonic (n= 23), 26.1 ± 8.4 ms (s.d.); (2) ramp-silent (n= 11), 27.2 ± 6.5 ms; (3) flat-tonic (n= 12), 24.8 ± 9.8 ms; and (4) flat-silent (n= 5), 28.5 ± 5.7 ms. The conduction velocity of the descending axon between the cell body and the upper lumbar segments (T/L or L1/2) was estimated using the distance between the recording and stimulating sites and the measured antidromic latency; 0.2 ms was subtracted to account for the latent period of spike initiation at the stimulating site (Jankowska & Roberts, 1972). The conduction velocity ranged between 8.0 and 27.7 m s−1(n= 53). The mean velocities for classified neurones were as follows: (1) ramp-tonic (n= 18), 15.5 ± 5.1 m s−1; (2) ramp-silent (n= 12), 15.5 ± 5.7 m s−1; (3) flat-tonic (n= 6), 16.9 ± 7.2 m s−1; and (4) flat-silent (n= 3), 13.2 m s−1. No significant group differences were found for the latency and the conduction velocity parameters (P > 0.1, Student's two-tailed t test).
Recording sites of 98 spinal-projecting neurones were successfully reconstructed from histological sections. Recording sites were located in the rostral part of the dorsolateral pontine tegmentum. They formed a rostrocaudally orientated column, extending from the level of the inferior colliculus, rostrally, to the rostral edge of the motor trigeminal nucleus, caudally (Fig. 8A–C; 73 subclassified neurones). Neurones were recorded at sites ventral to the 5MET at rostral levels, and at sites medial and ventral to the 5MET at middle and caudal levels. Both ramp neurones and flat neurones were distributed throughout these areas. Ramp-tonic neurones were mainly distributed in the medial and the ventral parts of the 5MET (Fig. 8B and C). Ramp-silent neurones were predominantly distributed ventral to the 5MET. Flat neurones were preferentially distributed ventral to the 5MET, although some flat neurones were recorded in the medial part to the 5MET. Unclassified neurones were distributed similarly to the classified neurones (not shown).
Neurones not antidromically activated from the spinal cord
There were 81 neurones that increased their firing when the bladder contracted but were not activated antidromically from the spinal cord; we call these ANTI (−) neurones. Fifty-seven of these were subclassified as ramp-tonic, ramp-silent, flat-tonic and flat-silent neurones (Table 1). Firing patterns of ANTI (−) neurones were qualitatively similar to those observed for the spinal-projecting neurones. The remaining 24 neurones could not be confidently subclassified into any of the four groups for the same reasons mentioned above for the unclassified spinal-projecting neurones.
Ramp and flat ANTI (−) neurones were located in the dorsolateral pontine tegmentum (Fig. 8D–F). There were some differences in the anatomical distributions of ANTI (−) neurones and the spinal-projecting neurones. In ventral parts of the 5MET, where spinal-projecting neurones were densely packed, the distribution of ANTI (−) neurones was somewhat sparse at three rostrocaudal levels (cf. Fig. 8D–F with A–C). The distribution of ANTI (−) neurones in the medial part of the 5MET was also sparse at mid levels of the rostrocaudal axis (cf. Fig. 8E with B). Although there was some overlap with the distribution of spinal-projecting neurones, ANTI (−) neurones tended to surround spinal-projecting neurones. ANTI (−) neurones were also distributed more ventral and ventrolateral to the 5MET. Unclassified ANTI (−) neurones were distributed similarly to the ramp and flat ANTI (−) neurones (not shown).
The present study identified 104 spinal-projecting neurones in Barrington's nucleus by verifying that antidromic spikes could be elicited; the upper, middle, and/or lower lumbar cord were stimulated. We previously showed that antidromically activated neurones in Barrington's nucleus project to the lumbosacral segments: most of them to the sacral cord, very few to the caudal part of the lumbar cord, and no neurone to the coccygeal cord (Sasaki, 2002). It is likely therefore that most of the antidromically activated neurones in the present study project to the sacral cord.
These spinal-projecting neurones were located in the rostral part of the dorsolateral pontine tegmentum, medial to the ventral part of the 5MET (Fig. 8A–C). They formed a rostrocaudally orientated column, extending from the level of the inferior colliculus to the rostral edge of the motor trigeminal nucleus. The location of these neurones corresponds well to those revealed by a neuronal tracing study of Barrington's nucleus (Blok & Holstege, 1999).
The present study also demonstrated that there exist dorsolateral pontine neurones that are not activated antidromically from the spinal cord. Eighty-one neurones we studied failed to produce antidromic spikes using appropriate stimulation. Anatomical studies show that descending axons from Barrington's nucleus are packed tightly as they pass through the DLF (e.g. Holstege et al. 1986). Electrical stimulation of this site (30–50 train pulses, 50–200 μA) evokes consistent bladder contractions with latencies between 0.25 and 0.5 s (Sasaki, 2002). We stimulated the surface of the DLF with strong stimuli, varying in intensity between 1 and 5 mA (typically, 1 mA). Failure to elicit antidromic spikes in the 81 neurones was not overcome even when stimulating two different levels of the spinal cord. We also attempted repositioning the stimulating electrode. In one experiment, the lateral funiculus was also stimulated by inserting the electrode into the white matter, but this was not as effective as stimulating the surface of the DLF. Thirty-nine dorsolateral pontine neurones could not be activated from the upper lumbar segments (Table 1), implying that at least some of these neurones do not project to the lumbar cord. These neurones were distributed in a somewhat broader region than the spinal projecting neurones (Fig. 8D and E). But medially and just ventral to the 5MET they were somewhat sparse; spinal-projecting neurones are densely packed here. These differences further suggest that there are neurones in Barrington's nucleus that do not project to the spinal cord, but because their firing is correlated with phases of bladder contractility, they are likely part of the micturition circuit (see discussion below).
Neurones examined in the present study overlap to some extent with the locus coeruleus/subcoeruleus complex, where there are noradrenergic neurones that also send fibres through the dorsolateral funiculus to the sacral cord; these axons terminate on many spinal neurones, including sacral parasympathetic bladder motoneurones (Westlund & Coulter, 1980). There are several reasons why we believe that these neurones were not sampled in our present study. Axons of noradrenergic neurones are unmyelinated (Brown et al. 1972), and the conduction velocity of unmyelinated axons is estimated to be less than 2.5 m s−1 (Häbler et al. 1990). The conduction velocity of axons of the Barrington's spinal-projecting neurones we studied was 8–28 m s−1, indicating that they are most likely not noradrenergic neurones. It has also been suggested that noradrenergic neurones of the locus coeruleus do not show activity related to micturition contraction in rats (Tanaka et al. 2003).
Barrington's nucleus has been labelled transneuronally in the retrograde direction after inoculation with pseudorabies virus into various pelvic organs: bladder, urethra, external urethral sphincter, penis, clitoris, colon, uterus and ovary (see Papka et al. 1998; Gerendai et al. 1998; Vizzard et al. 2000). Barrington's nucleus neurones project directly to sacral parasympathetic bladder motoneurones (Blok & Holstege, 1997). It is also proposed that the descending connection is intercalated by the sacral parasympathetic interneurones (Matsumoto et al. 1995). Spinal-projecting neurones that were recorded in the present study may activate sacral parasympathetic bladder motoneurones directly or indirectly, as has been suggested (Sasaki, 2002). Barrington's nucleus has also been suggested to inhibit external urethral sphincter motoneurones via GABA-ergic interneurones located in the dorsal grey commissure of the sacral spinal cord (Blok et al. 1998). The other organ that has a strong link with Barrington's nucleus might be the urethra; this smooth muscle also relaxes during micturition contraction. The functional significance of links between Barrington's nucleus and the remaining pelvic organs will be clarified by combining anatomical, physiological, pharmacological, and lesion studies. Consideration of the firing properties of the spinal-projecting neurones identified in the present study may serve to advance our understanding of a possible contribution of this nucleus to the control of other organs.
The present study was the first to demonstrate that spinal-projecting neurones in Barrington's nucleus display different kinds of firing patterns. Single units were classified into two categories, ramp and flat neurones, based on their firing patterns displayed during the relaxation phase of the micturition contraction–relaxation rhythm. They were further subclassified into ramp-tonic, ramp-silent, flat-tonic, and flat-silent neurones. Ramp-tonic neurones increased their firing throughout the relaxation phase of bladder contraction. Ramp-silent neurones were silent in the early period of the relaxation phase, displaying ramp increases thereafter. Flat-tonic neurones displayed relatively constant firing. Flat-silent neurones were silent throughout most of the relaxation phase. It should be noted, however, that ramp-tonic versus ramp-silent neurones, and flat-tonic versus flat-silent neurones are expedient subclassifications. The tonic neurones recorded in the present study are expected to be silent when the bladder is empty. In fact, all ramp-tonic (n= 5) and flat-tonic (n= 4) neurones that were examined became silent when the bladder was emptied by withdrawal of the fluid.
During micturition contraction, firing patterns were also found to vary. Three different classes of firing patterns were identified: constant, decrement and rise. Neurones with constant-type firing display relatively constant firing rates during contraction. Neurones with decrement-type firing display maximal rates at the rising phase of micturition contraction: this elevated rate gradually decreases over the duration of the contraction phase. Neurones with rise-type firing display firing increases and decreases that covary with small rises and falls in bladder pressure during the micturition phase. Although Barrington's neurones display varieties of firing patterns, as has been demonstrated, the firing types observed during micturition contraction are closely associated with four subclassified neurones: ramp-tonic, ramp-silent, flat-tonic and flat-silent (Fig. 7). That is, most ramp-tonic neurones displayed an additional, delayed firing increase after micturition contraction, and showed constant-type firing during contraction. In contrast, most ramp-silent neurones increased their firing before contraction, displaying decrement-type firing during micturition contraction. Flat-tonic and flat-silent neurones typically showed rise-type firing during the micturition phase.
What initiates micturition contraction under the isovolumic condition?
One of the striking findings of the present study is that the majority (65/104) of neurones fired tonically throughout the relaxation phase. Nevertheless, the bladder continued to relax, implying that tonic bombardment from Barrington's neurones to sacral neurones does not necessarily evoke bladder contraction.
The present study also demonstrated that the majority (48/76) of classified neurones were ramp neurones. Eighteen of 48 ramp neurones were ramp-silent neurones. Although the start of the ramp firing differed among ramp-silent neurones, the number of recruited neurones increased exponentially as the onset of micturition contraction approached (Fig. 4B). Together with ramp-tonic neurones, the summed population firing of all ramp neurones would be estimated to be much higher at the last stage of the relaxation phase compared with the early stage. This implies that the potential for evoking micturition contraction becomes more and more robust as the onset of micturition contraction approaches. There is evidence that supports the idea that the potential for evoking micturition contraction increases with time. Kruse et al. (1992) showed that bladder contractions evoked directly after electrical stimulation of Barrington's nucleus were small immediately after completion of micturition contraction, and the directly evoked contractions progressively increased in amplitude at longer intervals. In rats, a similar depression occurs and is thought to be due to recurrent inhibition from the parasympathetic bladder motoneurones (Noto et al. 1989). The identification of ramp neurones and the demonstration of progressive recruitment of neurones during the relaxation phase (Fig. 4B) in the present study is consistent with the findings of these two previous studies. Ramp neurones may participate in either activating the spinal excitatory pathway or disinhibiting a strong inhibitory neural mechanism, such as recurrent inhibition, that prevents parasympathetic bladder motoneurones from discharging easily. Related to this scenario is the observation that the base pressure of the bladder often displays small fluctuations during the later part of the relaxation phase (e.g. Fig. 3D; indicated by dot), and that the base pressure gradually, rather than suddenly, increases around the time micturition contraction is initiated (see fast sweeps in Figs 1, 2 and 5). These pressure profiles further support the increased potential for evoking micturition contraction. During the later stage of the relaxation phase, a majority of ramp-silent neurones and a portion of flat-tonic neurones displayed additional firing increases (Fig. 7B). Some flat-silent neurones also started to fire just prior to micturition contraction. These discharge profiles may further increase the probability of micturition contraction.
After a long period of relaxation, very high-pressure micturition contraction starts as if it were ‘switched on’ (de Groat & Kawatani, 1985). This ‘switching’ mechanism does not appear to be a simple ‘on-switch’ (also see Sasaki, 1998), in that at least two steps seem to be necessary. The first one involves the gradual excitation and/or disinhibition of spinal neural substrates mediated by spinal-projecting ramp neurones, and the other involves concentrated synaptic bombardments, just prior to bladder contraction, from Barrington's nucleus onto spinal premotor and/or parasympathetic bladder motoneurones, which ultimately leads to initiation of micturition.
Regulation of the contraction
Findings from the present study suggest that Barrington's nucleus regulates micturition contraction by means of three types of firing patterns of its spinal-projecting neurones: constant, decrement and rise. The constant type was observed predominantly in ramp-tonic neurones. Firing rate was relatively constant during micturition contraction, even when pressure of the contraction displayed small rise-and-fall fluctuations. This indicates that firing rate does not necessarily reflect precisely bladder pressure. The existence of such neurones was predicted previously. That is, when carried out beyond certain bladder pressures, isotonic stimulation of the bladder did not additionally increase the firing rate of sacral efferents to the bladder (Sasaki, 1998); the firing rate of sacral efferents, however, still increases (Häbler et al. 1993). This observation led to the hypothesis that the CNS basically operates as an amplifier of afferent inputs (positive feedback loop), but with an upper limit built into the loop.
The observation that some neurones in Barrington's nucleus fire constantly during micturition contraction is consistent with this hypothesis, and this type of neurone may be the predicted neural substrate. Together with the fact that a firing increase tended to be delayed in relation to onset of micturition contraction, neurones with constant-type firing may have a role in maintaining high bladder pressure via a positive feedback loop (with an upper limit), rather than a role in triggering of micturition contraction.
Other neurones, on the other hand, displayed a decrement type of firing pattern during micturition contraction. Most of these neurones were ramp-silent neurones. Neurones with decrement-type firing displayed an additional firing increase prior to micturition contraction and maximal discharge during the rising phase of micturition contraction. This discharge began to decrease even though bladder pressure had not yet reached a maximum. The decrement pattern appears to strongly contribute to raising bladder pressure rapidly and forcefully from the base pressure to high levels, but its role in maintaining high pressure levels is a minor one. This firing pattern fits well with the concept of feed-forward control.
The majority of flat-tonic and flat-silent neurones displayed a rise-type firing pattern. Firing increases coincided with small elevations in bladder pressure, and firing decreases began before reaching the peak pressure of each pressure elevation. This indicates that discharge is not pressure dependent, but rather may be caused by an active mechanism, such as feed-forward control. The rise type of firing pattern is similar to the decrement type in that both are characterized by increased firing during the rising phase of bladder pressure and by decreased firing before peak pressures are reached. The main difference between these two firing patterns is that the rise type is phasically active with each small contraction during the rising phase of pressure, while the decrement type is active vigorously over the whole range of the rising phase. This difference may reflect the firing properties of these neurones during the relaxation phase. The rise type was predominantly observed in flat neurones that started to increasingly discharge more just before the onset of micturition contraction. On the other hand, the decrement type was predominantly observed in ramp-silent neurones that started to discharge 3–26 s before the onset of micturition contraction and also displayed further increased firing just before the onset of micturition contraction.
Feed-forward control is a well established regulating mechanism implemented in the motor system. The present study is the first to demonstrate the existence of feed-forward control in the autonomic nervous system. It is interesting that this control exists in the ‘slowly acting’ autonomic system, which is in stark contrast to the rapid feed-forward control in the motor system. In the latter, feed-forward control is used for rapid control of effect organs, happening on the order of milliseconds without feedback from the peripheral receptors, which would take more time. In the bulbospinal reflex pathway of the micturition system, reflex firing in the sacral parasympathetic bladder motoneurones occurs with a long latency (60–75 ms) following stimulation of the bladder afferents (de Groat et al. 1981). In addition, it takes 200–300 ms for electrical stimulation of the sacral ventral root to the bladder to evoke bladder contraction (M. Sasaki, unpublished observation). Thus, feed-forward control seems indispensable for producing large micturition contractions ‘rapidly’.
Another striking finding of the present study was that most neurones displayed high firing rates for several seconds after the onset of bladder relaxation, maintaining rates that were comparable to those recorded during micturition contraction (Fig. 7D). This observation indicates that these neurones themselves do not trigger bladder relaxation. It is not plausible that neurones, which are active during the relaxation phase and decrease firing during micturition contraction, trigger bladder relaxation, since firing of these neurones starts after the onset of bladder relaxation (see specimen figures of Willette et al. (1988) and de Groat et al. (1998)), and there are very few of this type of spinal-projecting neurone in Barrington's nucleus (Sasaki, 1999). Considering that only a small number of neurones decreased their firing before the onset of bladder relaxation, it is conceivable that Barrington's nucleus does not have an active mechanism that triggers bladder relaxation. This also implies that the inhibitory mechanism that causes bladder relaxation most likely does not exist in the micturition pathway between the bladder afferents and Barrington's nucleus.
The periaqueductal grey (PAG) relays afferent information from the bladder to Barrington's nucleus (Blok et al. 1995). Although it has inhibitory as well as excitatory effects on bladder contractility (Kruse et al. 1990), this area may not be the source that triggers bladder relaxation. The inhibitory mechanism, or ‘off-switch’, seems to be downstream to Barrington's nucleus, perhaps at the sacral cord, where recurrent inhibition from parasympathetic bladder motoneurones is suggested to cause bladder relaxation (de Groat & Ryall, 1968; Noto et al. 1989).
Where are the firing patterns constructed?
The present study demonstrated that four types of neurones can be subclassified according to their firing activity in the relaxation phase. Each type displayed three different firing patterns during the micturition phase. The firing pattern exhibited by flat-tonic neurones resembled that of bladder afferents that faithfully transmit information regarding bladder pressure (Häbler et al. 1993). But the firing patterns exhibited by the other types of neurones were completely different from those of the bladder afferents. This implies that the micturition pathway is not a simple reflex arc, but that different firing patterns are formed in the afferent pathway from the bladder afferents to Barrington's nucleus.
Recent neurone tracing studies have identified direct projections from the lumbosacral spinal cord to the caudal part of the lateral PAG and Barrington's nucleus: projections to the former are much larger than those to the latter (Blok et al. 1995; Vanderhorst et al. 1996) (see Fig. 9). An electrophysiological study in cat identified sacral interneurones with long ascending axons projecting to the brainstem (McMahon & Morrison, 1982), presumably to the above two regions. The firing patterns are already differentiated in these long ascending spinal neurones: with increasing, stepwise bladder pressure, firing rates increase in one group and decrease in the other group. In addition, both of these are further subclassified into slowly adapting and rapidly adapting types (McMahon & Morrison, 1982). The PAG, a relay area between the sacral cord and Barrington's nucleus, may be a candidate for a component that further processes pattern formation. Firing properties of neurones in this area remain to be clarified.
The present study revealed that many neurones in the dorsolateral pontine tegmentum could not be activated antidromically from the spinal cord. These neurones displayed firing patterns similar to spinal-projecting neurones. Firing patterns also appear to be constructed within and around Barrington's nucleus.
In conclusion, the present study identified spinal-projecting neurones in Barrington's nucleus. These spinal-projecting neurones displayed different firing patterns. They were classified into four subgroups – ramp-tonic, ramp-silent, flat-tonic and flat-silent neurones – based on the relaxation phase of the micturition contraction–relaxation rhythm. The four subgroups of neurones further showed three different firing patterns during micturition contraction: constant, decrement and rise. Despite the variety of firing patterns, several tendencies emerged: most of ramp-tonic neurones displayed constant-type firing during the micturition phase, most of ramp-silent neurones displayed decrement-type firing, and flat-tonic and flat-silent neurones typically displayed rise-type firing. Considering all these findings, a model begins to emerge that places Barrington's nucleus in the micturition pathway that is involved in increasing bladder pressure rapidly and strongly via feed-forward regulation, while also maintaining high-bladder pressure via positive feedback from bladder afferents. Pudendal afferents from the urethra might also participate in the latter feedback loop: urine passing through the urethra also causes bladder contraction via pudendal afferents and via Barrington's nucleus (Barrington, 1921). Neurones that display decrement-type firing during the micturition phase may be the main source responsible for the former mechanism, while neurones displaying constant-type firing may be the source for the latter mechanism. In addition, ramp neurones may have another role, that of increasing the probability of micturition contraction by activating a spinal excitatory pathway or disinhibiting a spinal inhibitory mechanism. Finally, our observations suggest that Barrington's nucleus does not trigger bladder relaxation. The present study thus elucidated the basic function of Barrington's nucleus. It should be noted that the study was conducted under the isovolumic condition, where the base bladder pressure is maintained slightly above the micturition threshold. While this condition resembles the occasion when subjects endure urination, normal micturition is performed voluntarily, indicating that Barrington's neurones are also under the control of the cerebrum. It is of interest to know how firing patterns of Barrington's neurones are modulated, before, during and after micturition. An answer will be obtained by recording Barrington's neurones in the alert animal in a future study.
The author would like to thank Professor Y. Uchino for constant encouragement and valuable advice throughout these experiments. This work was supported by the Grant-in Aid for Scientific Research (C) from the Ministry of Education, Science, Sports and Culture of Japan (11680810).