To investigate to what extent antagonists of spinal neurotransmitters interact with the effects of sacral neuromodulation in a rat model of a chronically hyperactive urinary bladder.
To investigate to what extent antagonists of spinal neurotransmitters interact with the effects of sacral neuromodulation in a rat model of a chronically hyperactive urinary bladder.
In female rats the urinary bladder was instilled with turpentine oil 2.5% to induce cystitis. After surviving for 10 days the rats were anaesthetized with urethane, the bladder catheterized and connected to a pressure transducer. Stimulating electrodes were placed in the sacral foramina bilaterally. The spinal cord was exposed by a laminectomy, and a small pool was placed on the cord for intrathecal administration of neurotransmitter antagonists. Sacral neuromodulation was applied before and after administering the antagonists. The antagonists used were: memantine, an antagonist for N-methyl- d-aspartate (NMDA) receptors; CNQX, an antagonist for non-NMDA receptors, and L-NAPNA, a blocker of nitric oxide synthase.
With no electrical neuromodulation, memantine and L-NAPNA abolished the cystitis-induced bladder contractions for ≈ 4 and ≈ 37 min, respectively. The effect of CNQX was similar to that of artificial cerebrospinal fluid. Electrical sacral modulation with no antagonists also transiently abolished the bladder contractions; at the highest intensity used, the pause was 2–3 min. Superfusion of the spinal cord with CNQX reduced this effect of neuromodulation significantly, whereas memantine had no influence, and L-NAPNA increased the neuromodulation-induced pause.
The results suggest that non-NMDA receptors are involved in the effects of sacral neuromodulation, whereas NMDA receptors appear to have no role. Nitric oxide is essential for maintaining the chronic hyperactive state of the urinary bladder.
neuronal NO synthase
motor threshold of the tail muscles
L-N-nitro arginine p-nitroanilide
6-cyano-7-nitroquinoxaline-2,3-dione disodium salt
Electrical stimulation of the sacral nerves with electrodes in the sacral foramina (sacral neuromodulation, SN) has become an established method for treating lower urinary tract dysfunction in patients [1–4], for dysfunction of other pelvic organs , and for chronic pain conditions  (for a review of the indications, see ). In most cases, the desired effect of SN is inhibition of neural activity, either in sacral reflex pathways or in nociceptive tracts. The effects of SN differ from those of other types of electrical stimulation, e.g. microstimulation of the ventral segment S2, which causes bladder contractions instead of inhibition .
Even though SN is used internationally the mechanisms underlying its effects remain largely unclear [9,10]. There have been numerous animal experiments investigating the mechanisms of SN [11–13], and recently down-regulation of the proton- and heat-sensitive vanilloid receptor VR1 (now called TRPV1) was described as a possible mechanism for at least some aspects of chronic SN . Indeed, there is strong evidence for a contribution of TRPV1 to the hyper-reflexia of an experimentally inflamed urinary bladder .
To our knowledge, systematic studies of the interaction between neurotransmitters involved in bladder hyperactivity and the effects of SN are lacking, particularly in chronic pathological situations. The neurotransmitters or neuromodulators reported as possible mediators of bladder hyperactivity are numerous; practically all of the known neurotransmitters, neuropeptides and inflammatory cytokines have been assumed to be involved. Of particular importance for maintaining an unstable bladder appears to be activation of N-methyl- d-aspartate (NMDA) receptors [16–18]. Nitric oxide (NO) has been addressed as a further factor that may be involved in chronic bladder disorders .
Previous work from our group showed that in the spinal cord and ganglia of rats with chronic cystitis, the expression of neurotransmitters/neuromodulators, e.g. substance P, calcitonin gene-related peptide, neuronal nitric oxide synthase (nNOS) and galanin, was either increased or decreased, with the degree of change varying markedly between spinal levels . Interference with one of these modulators may underlie the effects of electrical neuromodulation. However, as neuropeptides are present in thin fibres, which have a high electrical threshold, and as these fibres are unlikely to be influenced by the relatively low intensities of electrical neuromodulation used in patients, their role in SN is questionable. Therefore, the release of neuropeptides from fibres activated by electrical neuromodulation is similarly unlikely. Obviously an indirect involvement of neuropeptides in neuromodulatory effects is possible, if it is assumed that electrical neuromodulation stimulates thick fibres which then influence the action or release of neuropeptides.
The most common neurotransmitter released in the spinal cord by both thick and thin fibres is glutamate. The postsynaptic neurones in the spinal dorsal horn express a multitude of ionotropic and metabotropic receptors for glutamate . As the involvement of ionotropic glutamate receptors in chronic pathological conditions is well known , we concentrated on antagonists of the main ionotropic receptors, i.e. NMDA and α-amino-3-hydroxy-5-methyl-4-isoxazolepropanoic acid/kainate (AMPA/KA) receptors [22–24]; the latter are called non-NMDA receptors.
NO has been addressed as another factor possibly maintaining chronic bladder disorders and hyper-reflexia , but there are data indicating that under special conditions, NO relaxes bladder muscles . Therefore, NO could also be involved in the inhibitory effects of SN. Interestingly, ‘knock-out’ mice that lack the gene encoding for the nNOS void normally . Collectively, the available data indicate that NO plays a role in lower urinary tract function, but the exact nature of this role is unclear.
The aims of the present study were: (i) to study the interaction between some of these neurotransmitters and the effects of SN; and (ii) to identify a neurotransmitter that is involved in neuromodulatory effects. To this end, SN was applied without and under the influence of glutamate antagonists and a blocker of NO synthesis. The experiments were based on the hypothesis that SN releases neurotransmitters in the spinal cord which act on spinal neurones, and thus reduce the bladder contractions. If one of these neurotransmitters was involved in the neuromodulatory effects, intrathecal administration of an antagonist should reduce the effects of neuromodulation. The animal model used was the rat with chronic cystitis induced by turpentine oil.
The study was carried out in accordance with the German law on the protection of animals and approved by the state ethics authority for animal experimentation. The experiments were conducted using 39 adult female Sprague-Dawley rats. To induce cystitis, 24 rats were anaesthetized with xylazine (7.5 mg/kg, i.p.) and ketamine (100 mg/kg, i.p.). The urinary bladder was catheterized through the urethra and inflamed by instillation with turpentine oil (2.5% in olive oil). The turpentine oil was left in the bladder for 30 min, then the bladder was rinsed with Tyrode solution. All rats survived for 10 days. If the relatively short life-expectancy of a rat (≈ 2 years under laboratory conditions) is considered, such a duration of cystitis is chronic. Fifteen rats served as a control group, in which the bladder was instilled with physiological saline.
For the final experiment the rats were anaesthetized with urethane 1.1 g/kg i.p. (Fluka Chemika, Switzerland). The anaesthesia was deep enough to abolish flexion reflexes and marked blood pressure reactions (>10 mmHg) to noxious mechanical stimuli. Arterial blood pressure, body core temperature, and an electrocardiogram were continuously monitored and kept at physiological levels. Most rats breathed room air spontaneously, but some that showed irregular respiration were artificially ventilated with a gas mixture of 47.5% O2, 2.5% CO2, and 50% N2.
To record bladder contractions a catheter was introduced into the bladder through the urethra and connected to a pressure transducer. A perfusion pump was also connected to the catheter to fill the bladder, if necessary, to start the contractions.
The rats were mounted in a spinal frame and the sacral foramina S1–S3 surgically exposed. Two stimulating cathodes were placed in the foramina of S1 bilaterally. The indifferent anode was placed under the skin of the back (Fig. 1). The frequency of stimulation was 20 Hz throughout. The intensity of stimulation is given in multiples of the motor threshold of the tail muscles (Tmot). The threshold was determined by increasing the voltage (square stimuli, width 0.1 ms) until vibrating movements of the tail were just discernible. This intensity was Tmot (stimulus amplitude 200–300 mV). In those rats which were not stimulated, the stimulating electrodes were fixed in the exposed sacral foramina as in the test group, but no stimulating current was applied.
For topical spinal (intrathecal) administration of the antagonist or blocker, respectively, a laminectomy from vertebrae L6 to Th12 was performed and the dura opened. On the dorsal surface of the exposed spinal segment S1, a small pool was formed with a plastic ring, and its contact sites with the spinal cord were sealed with silicone grease. This technique of intrathecal administration is termed ‘spinal superfusion’. As the electrodes for SN had been placed in the S1 foramina, the main neuromodulatory effects were expected to occur in this segment. The method of spinal superfusion of substances has the advantages that: (i) the neurones in the dorsal horn are easily reached, because the blood-brain barrier is circumvented; (ii) drug administration is restricted to one spinal segment and therefore systemic effects are unlikely; (iii) drug concentrations in the dorsal horn can be kept constant for long periods.
The pool was filled with 75 µL of either artificial cerebrospinal fluid (CSF) or antagonist/blocker solutions. In most cases where administration of the test substances abolished the contractions, they returned within 1 h (the exception was L-N-nitro arginine p-nitroanilide, L-NAPNA, see Results). Only after the return of bladder contractions could the effectiveness of SN be tested. The test substances were: memantine, an antagonist of NMDA receptors (Sigma Co., Munich, Germany, 100 µm in CSF); 6-cyano-7-nitroquinoxaline-2,3-dione disodium salt (CNQX), an antagonist of AMPA/KA receptors (Tocris, Bristol, UK, 100 µm in CSF); and L-NAPNA, a blocker of nNOS (Sigma, 100 µm in CSF). Memantine was preferred over other well-known NMDA antagonists because it is relatively short acting, has few side-effects and is assumed to effectively block transmission of chronic low-frequency input to the spinal cord [28,29]. In all experiments, CSF (the solvent of the antagonists) was administered as a control.
The contractions of the urinary bladder were continuously measured using the interface CED 1401 and Spike 2 software (Cambridge Electronic Design Ltd, UK). To obtain the ‘increase of interval’ between contractions as a measure of the effect of SN, the mean of the three contraction intervals before the start of SN was subtracted from the entire interval between contractions (the ‘pause’ in bladder activity) during and after SN.
In each experiment in which SN was used the electrical stimulation was at 3 × and 12 × Tmot. In rats with both inflamed and intact bladder, the pool was first filled with CSF (in many rats the filling procedure caused a small pause in the contractions, see below), and SN applied after the contractions had reappeared. After this test, the contents of the pool were exchanged against the glutamate antagonists or NOS blocker, and SN repeated. SN was applied for 2 min throughout. Only one antagonist was tested in each rat, because the duration of the effect of the antagonists on the neurones in the dorsal horn is unknown.
Results were compared statistically using the Mann–Whitney U-test, with P < 0.05 (two-tailed) considered to indicate significance.
In a previous study chronic cystitis was induced with 0.4% mustard oil . This model proved suitable for studying stimulation parameters of SN in anaesthetized rats, but the mustard oil model could not be used for the present study, because the bladder contractions were irreversibly abolished after a laminectomy, and the pool was placed on the spinal cord. Therefore, a model of chronic cystitis was developed that had more robust bladder contractions but still showed clear effects of SN. Rats with cystitis induced by 2.5% turpentine oil fulfilled this requirement. The turpentine model had larger bladder contractions at more regular intervals even after laminectomy, but the effects of SN were smaller (the pause in the bladder activity shorter) than in the mustard-oil model. In contrast to the cystitis induced by 0.4% mustard oil, the micturition rate after 2.5% turpentine oil showed no decline at the end of the survival period of 10 days.
Filling the pool with CSF or the test substances caused a short pause in the contractions, which occurred almost instantaneously on administering the solutions, and therefore this effect cannot be due to diffusion of the agents into the cord. A temperature effect can likewise be excluded, because the test solutions were warmed to 37°C. A possible explanation is stimulation of sensitive mechanoreceptors of the arachnoidea or pia mater by filling the pool.
In rats with cystitis that were not treated with SN, as shown in Figs 2 and 3, superfusion of the spinal segment S1 with memantine was followed by an insignificant increase in the interval between bladder contractions. In the rather extreme case shown in Fig. 2, the duration of the pause was ≈ 12 min (mean pause after memantine 200 s; Fig. 3).
CNQX did not influence the interval between bladder contractions, i.e. after the pool had been filled with CNQX, the mean interval was similar to that after administering CSF (Fig. 3). L-NAPNA was extremely effective in suppressing the bladder contractions. After administering the NOS blocker, the bladder was silent for a mean of ≈ 37 min (Fig. 3). This pause was significantly longer than that after CSF (P < 0.01).
In rats with cystitis and treated with SN, memantine caused no significant change (compared with CSF) of the effects of SN. Superfusion with CNQX abolished the (small) effect of SN at 3 × Tmot completely, and reduced that at 12 × Tmot. Original recordings from such a test are shown in Fig. 4. These data suggest that SN stimulates fibres in the dorsal root S1 which release glutamate and thus activate non-NMDA glutamate receptors. Activating the non-NMDA receptors apparently inhibits the spinal reflex arc that drives the bladder contractions.
In the recording shown in Fig. 5, changing the content of the spinal pool from CSF to L-NAPNA caused a contraction pause of ≈ 33 min. After the contractions had reappeared despite the continuous presence of L-NAPNA in the pool, SN was applied at 3 × Tmot. This caused a short pause of 12 min, while after SN at 12 × Tmot the pause was 20 min.
The comparison of SN during superfusion with CSF or with the test substances is shown in Fig. 6. Compared with the effects of CSF, CNQX reduced significantly the neuromodulatory effect at 12 × Tmot (P < 0.01). Memantine had no significant influence on the effects of SN, and during superfusion with L-NAPNA the pause between contractions was significantly longer. However, if this neuromodulatory effect under L-NAPNA is compared with the effect of the NOS blocker alone (Fig. 3), the combination of SN with L-NAPNA had a weaker inhibitory effect on the activity of the bladder than L-NAPNA alone. This indicates that SN reduced the inhibitory action of the NOS blocker on bladder contractions.
The extremely prolonged inhibitory effects of L-NAPNA in rats with cystitis raised the question as to the action of nNOS blockage in rats with an intact bladder. In these experiments the bladder was instilled with 0.9% saline instead of turpentine oil. If no contractions were present at the start of the experiments they were elicited by slow infusion of 0.9% saline into the bladder (maximum 2 mL).
As shown in Table 1, in rats with an intact bladder and no SN the pause in bladder activity after spinal superfusion with L-NAPNA was even longer than in rats with cystitis. No mean value for the duration of the L-NAPNA-induced pause is given, because the contractions did not return within the following 2 h. Therefore, the experiment was terminated at that time (7200 s), and no SN was applied. However, in intact rats with no SN, spinal superfusion with memantine or CNQX had smaller effects on bladder contractions than in rats with cystitis, i.e. the pause was shorter. The values for CSF and those of CNQX and memantine, respectively, were not quite significant on the U-test (P < 0.054 for both CSF against CNQX, and CSF against memantine), but the effect sizes indicated that there was a true difference (Cohen's d-value for CSF against CNQX was 1.52; size-effect correlation 0.60; for CSF against memantine, 1.50; size-effect correlation 0.60). The size-effect data are within the range of a large effect.
|Spinal superfusion||Mean (sem) increase in interval, s, between contractions with:|
|No SN||0||19.60 (8.26)||30.20 (12.56)||7200*|
|3 × Tmot||21.32 (15.70)||0||1.00 (0.52)||0†|
|12 × Tmot||29.27 (12.41)‡||32.02 (8.38)||74.70 (55.72)||0†|
In intact rats in which SN was applied the neuromodulatory effects were likewise smaller than in rats with cystitis. Qualitatively, the effects had the same direction; CNQX reduced the effects of SN at 3 × Tmot (not significant) and had no clear effect at 12 × Tmot. Memantine again had no significant influence on the effects of SN. As noted above, the influence of L-NAPNA on neuromodulatory effects could not be tested.
The stronger and more regular contractions in rats with turpentine-induced cystitis than in the mustard-oil model could be a result of bladder inflammation after turpentine causing less damage to the sensory endings in the bladder wall. Histologically, the bladder was clearly inflamed in both models, but the damage to the sensory endings in the bladder wall may have differed. A strong inflammatory lesion is likely to desensitize many of the receptive endings in the bladder . Our experience that not all rats had bladder contractions and therefore did not yield scientific data, could be caused by oestrous influences on the contractile state of the bladder . As stated in the methods, only female rats were used for the study, but the ovulatory cycle was not determined.
The results show that two of the substances tested (CNQX and L-NAPNA) influenced the effects of SN. In the present experiments we did not use the stimulus intensity that was found to be most effective in a previous study (24 × Tmot) , because we had to assume that at this stimulus strength unmyelinated fibres are also excited, many of which are nociceptive and therefore cause pain. In patients with no spinal cord injury such a stimulus intensity cannot be used for SN.
To mimic the clinical application of SN as closely as possible  the stimulating electrodes were placed in the sacral foramina of S1 bilaterally. Using the dorsal root L6 for neuromodulation (which is known to constitute the other major pathway for bladder afferent fibres) would have required a technique (hook electrodes) that is different from that used in hospital for SN. Bilateral stimulation was used, because in patients it had greater effects than unilateral neuromodulation .
At the highest stimulus strength used in the present study (12 × Tmot) all thick myelinated and part of the thin myelinated fibres in the dorsal roots were probably excited. The most likely neurotransmitters released from the terminals of these fibres in the spinal cord are glutamate and possibly aspartate. Therefore, we concentrated on antagonists of ionotropic glutamate receptors (NMDA and non-NMDA channels).
The most important findings of the present study are that: (i) in rats with cystitis the neuromodulatory effects could be attenuated or abolished by superfusion of the spinal cord with an antagonist of non-NMDA receptors; (ii) NMDA-receptor antagonists had no significant effect on the bladder contractions; and (iii) a spinal block of NOS was extremely effective in inhibiting the contractions of the urinary bladder.
Of the substances tested only the non-NMDA antagonist CNQX was able to reduce the effects of SN. There is little published information about the relation between non-NMDA receptors and bladder contractions, but these receptors are known to transmit the effects of thick fibre activity to dorsal-horn neurones [21,23]. As the relatively low-intensity SN used in the present experiments excites mainly thick fibres, the finding that the AMPA antagonist blocked or reduced the effects of SN is in accordance with these data. Activation of this fibre system is not associated with a long-term change in the excitability of dorsal-horn neurones; they fire a short train of action potentials, and thereafter resume their previous state of excitability. In contrast, activation of the NMDA pathway is assumed to require a pathological (strong or long-lasting) input via nociceptive afferent fibres and probably changes the excitability of the neurones.
As glutamate is primarily an excitatory transmitter the additional assumption must be made that under the conditions of the present study, glutamate excited AMPA receptors on dorsal-horn neurones that contacted parasympathetic preganglionic cells through inhibitory interneurones. Alternatively, the AMPA receptors may have been located on inhibitory interneurones. As the effects of stimulation at 12 × Tmot, which probably included activation of some of the Aγ-fibres, were also reduced by CNQX, it is likely that these fibres too released glutamate acting on AMPA receptors. However, the neuromodulatory effects of stimulation at 12 × Tmot were not completely abolished. This could be explained by assuming that by releasing glutamate together with neuropeptides such as substance P, thin fibres are capable of activating both the AMPA and NMDA channel. The latter channel is not influenced by CNQX. In the present experiments a stimulus intensity of 12 × Tmot corresponded to 2.4–3.6 V; this voltage is similar to that applied for neuromodulation in patients . However, in patients the effective stimulus strength that reaches the nerve fibres is probably less, because the stimulation conditions cannot be optimized, as in our experiments.
That the NMDA antagonist memantine had only a marginal influence on the activity of the inflamed bladder was surprising, because there are published data indicating that NMDA receptors are essential for bladder contractions [16,17], particularly if the bladder is hyperactive . For instance, Maggi et al. reported that a block of the NMDA receptors with MK-801 suppressed voiding of the intact bladder in anaesthetized rats, and MK-801 has also been shown to inhibit the micturition reflex of a chronically inflamed bladder . However, the role of NMDA receptors for bladder contractions is unclear because in awake rats, MK-801 appears to have the opposite action . Another explanation for the lack of an effect of memantine would be that the concentration administered in our experiments (100 µm in the pool) was out of the dose range. However, this explanation appears unlikely, because other groups who recorded the activity of dorsal-horn neurones in vivo[36,37] used similar concentrations of memantine (calculated tissue concentrations 40 and 24 µm, respectively) and reported a reduction of neuronal firing rate.
The present in vivo experiments are unsuitable for dose–response measurements, because this would require many more rats, even if cumulative dose-effect relations are studied. Therefore, we applied one working concentration of each drug that could be assumed to be effective, based on published evidence.
The lack of an effect of memantine also raises the question of whether in the present experiments the urethane anaesthesia interfered with the NMDA mechanism. Such an interaction was reported for many neural systems, including cerebral centres, and can result in an enhancement or attenuation of glutamatergic transmission . If there had been an accentuation of NMDA mechanisms under urethane memantine would be expected to have an effect on SN (provided that NMDA mechanisms are involved in bladder hyperactivity and a sufficiently high dose of memantine was administered). However, if NMDA-mediated transmission was attenuated by urethane, memantine could lack any recognisable effect. If this was the case, urethane inhibited NMDA- but not AMPA-mediated transmission, because CNQX had an effect on SN. At present we cannot offer a convincing explanation for the lack of effect of memantine on bladder contractions. The conditions of the present rat model (turpentine oil-induced cystitis, laminectomy, spinal superfusion) may have played an additional role.
As noted, the function of NO in micturition is largely obscure. Apparently NO can promote bladder hyper-reflexia during inflammation, or conversely inhibition of NO synthesis can prevent bladder hyperactivity . This latter effect is what occurred in the present experiments, but the effects of SN do not appear to be mediated by NO, because during inhibition of its synthesis by L-NAPNA, the effectiveness of SN was greater, not less.
The NO effect may be restricted to special situations: results obtained from the isolated bladder trigone in an in-vitro system showed that a pharmacological block of NO synthesis abolished the smooth muscle-relaxing effects of low-frequency stimulation (1 Hz), but not those of stimulation at 10 Hz . However, in mice in which the gene encoding for NOS was neutralised, there was unimpaired bladder voiding . Taken together, the NO data indicate that this neurotransmitter may have a function in micturition only under special circumstances.
Neuropeptides such as substance P and vasoactive intestinal polypeptide have also been discussed as mediators of bladder hyperactivity  and neuromodulatory effects. Neuropeptides are stored in and released from thin myelinated and unmyelinated primary afferent fibres, and are known to be involved in the induction of CNS hyperexcitability. Substance P is present in bladder afferent fibres [39,40]. Acting together with glutamate, substance P is known to activate NMDA receptors, which are essential for the induction of central hyperexcitability in many animal models of a hyperactive bladder. However, it is unlikely that with the low-intensity stimulation used in the present study, sufficient unmyelinated fibres were stimulated to release effective amounts of neuropeptides during SN. An increased expression of nerve growth factor (which has been shown recently to induce bladder hyperactivity ) likewise is unlikely to be involved in the neuromodulatory effects in the present study, because the sacral stimulation periods were too short to increase its expression.
The study was supported by a research grant from the EU, Project REBEC, QLG5-CT-2001-00822. The competent assistance by B. Quenzer is gratefully acknowledged.
None declared. Source of funding: EU Project.