Transmission of afferent information from urinary bladder, urethra and perineum to periaqueductal gray of cat

Brain Research 819 Ž1999. 108–119

Research report

Transmission of afferent information from urinary bladder, urethra and perineum to periaqueductal gray of cat Myto Duong a , John W. Downie a

a,b,)

, Huan-Ji Du

a

Department of Pharmacology, Dalhousie UniÕersity, Halifax, NoÕa Scotia, Canada, B3H 4H7 b Department of Urology, Dalhousie UniÕersity, Halifax, NoÕa Scotia, Canada, B3H 4H7 Accepted 24 November 1998

Abstract The micturition reflex pathway is a supraspinal pathway. Anatomical tracing evidence is compatible with an involvement of the periaqueductal gray ŽPAG. in the ascending limb of this reflex. We tested the involvement of the PAG in receiving urinary tract- or perineum-related information and attempted to characterize this ascending path in terms of what type of information is being conveyed. Electrical stimulation of the pelvic nerves, which carry afferent information from the urinary bladder, evoked maximum field potentials in the caudal third of the PAG, primarily in the dorsal part of the lateral PAG and in the ventrolateral PAG. Since the regions activated by pelvic nerve stimulation differed from those activated by stimulation of the sensory pudendal or superficial perineal nerves, it is possible that specific pathways for different nerve inputs to the PAG exist. Sacral spinal cord neurons ascending to the PAG were identified by antidromic activation and then tested for inputs from pelvic, sensory pudendal or superficial perineal nerves. Of 18 units identified, only five received inputs from any of the peripheral nerves tested and only two projecting neurons received a pelvic nerve input. Thus the PAG may receive inputs from bladder and perineum, but the small proportion of cells with direct projections to the PAG receiving inputs from our test nerves implies that the major part of this pathway is not directly related to lower urinary tract function. q 1999 Elsevier Science B.V. All rights reserved. Keywords: Pelvic nerve; Spinal cord; Ascending tract; Mesencephalon; Electrical stimulation; Extracellular unit recording; Micturition

1. Introduction Micturition Žurination. is mediated by activation of the sacral parasympathetic efferent pathway to the bladder and reciprocal inhibition of the somatic pathway to the urethral sphincter. This coordinated pattern can be evoked by stimulation at a particular site in the dorsolateral pontine tegmentum w18,20,37x termed Barrington’s nucleus, M-region w18x or the pontine micturition centre. Bilateral lesions in this area abolish micturition w4,5,17,21x. Intercollicular decerebration facilitates micturition apparently by eliminating inhibitory inputs from higher centers w38x. Anatomical studies suggest that Barrington’s nucleus in the cat does not receive a significant direct projection from the sacral spinal cord w6x. However, the periaqueductal gray ŽPAG. appears to be a major target of afferent input from the sacral spinal cord w6,8,41x and there is a rich projection from the lateral PAG to Barrington’s nucleus )

Corresponding [email protected]

author.

Fax:

q 1-902-494-1388;

E-mail:

w7,40x. It has been hypothesized that this spinal cord– PAG–Barrington’s nucleus pathway represents the afferent limb of the basic micturition reflex in the cat w6,7x. Some support for PAG involvement in micturition can be found in various studies. Stimulation of the PAG produces bladder contraction in cats w20x and rats w32x. Electrical stimulation of bladder afferents in pelvic ŽPL. nerve produced short latency field potentials in the rat PAG w32x. Blockade of synaptic transmission by microinjection of CoCl 2 into a restricted area of caudal ventrolateral PAG reversibly blocked micturition in urethane-anesthetized rats w26x. The spinal pathways that transmit sensory information from the visceral afferents to more rostral structures can be found in the dorsal, lateral and ventral columns. Primary afferent collaterals carrying touch and pressure sensation in the urethra and innocuous sensations from the pelvic floor muscle are conveyed via the dorsal column w30x. Projections to the dorsal columns have been traced from the PL w28x and pudendal ŽPud. w39x nerves in the cat. The dorsal column also has a synaptic path, the postsynaptic dorsal column pathway, which has been shown to convey

0006-8993r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 6 - 8 9 9 3 Ž 9 8 . 0 1 2 9 4 - 3

M. Duong et al.r Brain Research 819 (1999) 108–119

noxious colorectal and cutaneous inputs to the nucleus gracilus w1x. The lateral columns transmit information concerning temperature sensation in the urethra, the sensation of bladder fullness and desire to micturate, sexual sensations, and pain sensations from the bladder, urethra, lower ureter, and skin. This pathway is thought to be the spinothalamic tract in man w30,31x. Studies in the cat have identified a tract in the dorsal half of the lateral column near the surface of the spinal cord and a tract in the ventral and ventrolateral column which convey information from the bladder w13–15,19,21,27x. Fibers in the dorsolateral tract in the thoracic spinal cord form the primary pathway for activating the supraspinal micturition reflex as stimulation of this tract evokes coordinated micturition w14x. Although it has been demonstrated that there are spinal cord projections to the PAG, it is not known whether these projections are involved in bladder or bladder-related functions. Therefore we determined whether bladder-related information is carried on this pathway. Our objectives in this study were to identify areas of the PAG responsive to stimulation of bladder efferents and to characterize the inputs to sacral spinal neurons projecting to the PAG.

2. Materials and methods 2.1. Surgical preparation In 26 adult cats Ž2.9–5.7 kg. of either sex, anesthesia was induced with ketamine Ž25 mgrkg, i.m.. then maintained with 2–3% halothane in 1:1 nitrous oxide:O 2 Žtotal flow rate of 2 lrmin.. The trachea was cannulated for artificial respiration and the right femoral vein and left carotid artery were cannulated for the administration of 5% dextrose in Ringer’s solution and drug and for blood pressure monitoring, respectively. A catheter for recording bladder pressure was inserted into the urinary bladder through an incision in the urethra, 1–2 cm below the bladder neck. Bladder and arterial blood pressures were measured using pressure transducers connected to a chart recorder. Bladder-directed branches of the right and left PL nerves were exposed close to the bladder neck through a midline abdominal incision and freed of connective tissue. The leads of insulated silver foil electrodes, 2 mm = 10 mm with a 2 mm = 2 mm area of the electrode bared, were sutured to the urethra. The electrodes were placed under separately dissected areas of the intact nerve, with the bared portion in contact with the nerve. The electrodes were then folded over the nerves and were insulated from the surrounding tissue with Plastibase ŽSquibb Canada., or Kwik-Cast ŽWorld Precision Instruments.. The abdomen was then closed in two layers. With the cat prone, the sensory ŽsPud. and motor branches ŽmPud. of the pudendal nerve, the superficial perineal ŽSPeri. nerve, and in some experiments the caudal

109

cutaneous femoral ŽCCF. nerve were dissected on the left side. When possible, the mPud nerve was separated into anal and urethral branches. Laminectomy was carried out from L4–S2 to expose the lumbosacral spinal cord and the spinal roots. After the surgical preparation, halothane anesthesia was replaced by chloralose Ž60 mgrkg, i.v., supplemented Ž10–20 mgrkg, i.v.. as required based on a withdrawal response to pinching of the interdigital skin or toe pads. Gallamine triethiodide Ž20 mg initially and supplemented as required, i.v.. was used to immobilize the animal during the recording session. Assessment of the anesthetic level under these conditions was based on cardiovascular response to pinching interdigital skin or toe pads. Body temperature was maintained at 37–388C with a heating pad and infrared lamp. With the completion of the laminectomy, the cat was transferred to a stereotaxic frame and a spinal unit on a vibration isolation table. Bilateral pneumothoraxes were performed to decrease respiratory movements. sPud, mPud, SPeri, and CCF nerves were mounted on bipolar platinum iridium hook electrodes submerged in a pool of heated mineral oil formed with skin flaps. The dura was opened under mineral oil and the right S1 and S2 dorsal roots were identified and traced to their spinal segments. A silver ball recording electrode was placed near the S1–S2 junction to record the cord dorsum potential. A craniotomy was carried out and part of the cortex and the bony tentorium were removed so that brain surface landmarks could be seen and the entire midbrain was accessible. Dexamethasone Ž4 mg, i.v.. was given to reduce brain edema. 2.2. Stimulation of peripheral nerÕes The PL nerves were stimulated by a single monophasic rectangular pulse of 0.3–0.5 ms width or a train of 2–3 pulses at 333–1000 Hz, with an intensity 2–5 times the threshold current required to elicit a cord dorsum potential or a reflex discharge on the motor branch of the pudendal nerve. In the absence of a reflex or cord dorsum recording, a maximum intensity of approximately 100–200 mA was used. The central cut ends of the other peripheral nerves ŽsPud, SPeri, and CCF. were stimulated with single pulses of 0.2 ms width at an intensity of 5 times the threshold current required to elicit a cord dorsum potential. The interstimulus interval was 2 s. 2.3. Recording in PAG Monopolar tungsten microelectrodes with 50–60 mm tip diameters were used to search the PAG from A6.0 to P2.5 in planes 0.5–1 mm apart rostrocaudally w36x. Electrode tracks in each plane were placed 1–1.5 mm apart mediolaterally. Field potential recordings were made at

110

M. Duong et al.r Brain Research 819 (1999) 108–119

depths from 2–8 mm below the surface of the midbrain in steps of 1 mm. In three experiments, rostrocaudal searching was restricted to AP0–P1.5 to permit more data to be obtained in searching in the mediolateral plane. Here, mediolateral distances between tracks ranged from 0.5 to 1.0 mm. The signals were amplified, filtered Žbandpasss 1 Hz–3 kHz., and displayed on an oscilloscope. Cord dor-

Table 1 Central latencies of the earliest component of the PAG field potentials evoked by electrical stimulation of various peripheral nerves Nerve

Central latency wmsx Žnumber of sites.

ipsiPL contraPL sPud SPeri

8.2"0.6 Ž13. 8.3"0.6 Ž10. 10.5"0.9 Ž8.) 11.9"1.6 Ž6.)

Central latency was determined by subtracting the time of onset of the cord dorsum potential from the time of onset of the PAG field potential. Latencies for ipsilateral Žipsi. and contralateral Žcontra. stimulation sites are separated for PL nerve but combined for sPud and SPeri nerves. Data are presented as means"S.E.M. Žnumber of animals.. ANOVA detected between-group differences. Asterisks indicate significance Ž p- 0.05. in post-hoc between-group testing with Fisher’s LSD test between ipsiPL and other nerves.

sum and PAG field potentials were digitized Žsampling rate 2–10 kHz. and averaged over 16 sweeps ŽMacLab Scope, AD Instruments.. 2.4. PAG field potential data handling Locations where the field potential elicited by each nerve was maximum were determined and mapped onto standard outlines of the cat midbrain w36x. For the PL nerve stimulation, a maximum response farther than 1 mm from the border of the PAG was disregarded and the next largest field potential located in the PAG or within 1 mm of the PAG border was considered to be the maximum. 2.5. Stimulation of PAG

Fig. 1. Example of PAG field potentials evoked by PL nerve stimulation. The electrode track lay 2.2 mm lateral to midline at P1.0 w36x ŽA.. The numbered dots represent the sites at which recordings were made in 1 mm intervals down the track. The hatched area represents the site of an electrolytic lesion that was used to reconstruct the electrode path. Each trace shown in ŽB. is an averaged focal field response to 16 consecutive single stimuli applied to the PL nerve. The numbers correspond to the recording sites plotted in panel ŽA.. The maximum potential was found in the dorsal PAG at site 2. The field potentials were multiphasic. Identifiable components are indicated by arrows ŽB.. Vertical bar represents 20 mV. Horizontal bar represents 20 ms.

Prior to PAG stimulation, the bladder was distended with 5–10 ml of saline. For antidromic studies, two strategies were used to determine the appropriate placement of the PAG stimulating electrode. One strategy involved eliciting bladder contractions. Sites in the caudal PAG ŽA0.5– P1.5. were stimulated with monophasic 0.5 ms wide constant current pulses of 100–200 mA at 50 Hz, in search of an optimum site for evoking PAG-evoked bladder contraction. The other strategy used PL nerve stimulation-evoked field potential recordings in the PAG to determine the optimum PAG site. A stimulating tungsten microelectrode Ž60 mm tip diameter, bared for 1 mm. was placed in the midbrain with the indifferent electrode placed on muscle, in the cerebellum, or more rostrally in the PAG. Once a location in the PAG was found that could elicit a large bladder response, the stimulating electrode was kept at that PAG location. For single unit searches in the spinal cord, the PAG was stimulated with monophasic 0.5 ms wide constant current pulses of 100–200 mA in trains of 4–5 pulses at 333 Hz with an intertrain interval of 2 s.

M. Duong et al.r Brain Research 819 (1999) 108–119

111

filled with 4 M NaCl. A microelectrode manipulator mounted on a custom-built Lundberg arc was used to move the recording electrode to search the spinal cord for activated cells.

Fig. 2. Examples of PAG field potentials evoked by the stimulation of different peripheral nerves. At the site of the maximum field potential evoked by stimulation of the contralateral PL nerve Žat P1, 2.2 mm right of midline, q3 mm from horizontal zero w36x., there was virtually no field potential demonstrable in response to stimulation of SPeri or sPud nerves ŽA.. On the other hand, at the site where SPeri and sPud nerve stimulation evoked maximal PAG field potentials Žat AP0, 1.8 mm right of midline, q0.3 mm from horizontal zero w36x., contralateral PL nerve stimulation produced a submaximal response ŽB.. Traces taken from the same experiment as in Fig. 1. Vertical bar represents 20 mV and horizontal bar represents 20 ms for both panels.

2.6. Single unit recordings in the spinal cord Glass microelectrodes with resistances 2–5 M V were drawn from 1.2 mm ŽOD. filamented glass capillaries and

Fig. 3. Distribution of maximum PAG field potentials evoked by stimulation of different peripheral nerves. Positions of the maxima were determined by measurement from lesion sites and mapped on outlines redrawn from the atlas of Snider and Niemer w36x. PAG field potential maxima for sites ipsilateral Žfilled symbols. and contralateral Žopen symbols. to the peripheral nerve being stimulated were determined and plotted separately. When the maximum PAG field potential was at the same site for ipsilateral and contralateral peripheral nerve stimulation, a half-filled symbol was used. PL nerve stimulation produced identifiable maxima in histologically verifiable locations in 11 experiments Žcircles.. sPud stimulation evoked maximum field potentials within the PAG in eight experiments Žsquares.. SPeri stimulation evoked maximum field potentials within the PAG in six experiments Žtriangles..

112

M. Duong et al.r Brain Research 819 (1999) 108–119

The spinal cord was searched in the caudal to rostral direction from the S2rS3 junction to S1rS2 junction. Separation between microelectrode tracks was 150–200 mm in the mediolateral plane between the midline and the dorsal root entry zone. Tracks reached depths of 2000–2700 mm. Extracellular signals were amplified, filtered Žband pass s 300 Hz–3 kHz. and displayed on an oscilloscope. The signals were also digitized at a sampling rate of 21 kHz and displayed and stored on a Macintosh computer ŽSpike 2.2; 1401 plus, Cambridge Electronic Design.. Spinal units were tentatively considered to be antidromically activated from the PAG if they discharged at constant latency from the PAG stimulus and remained stimuluslocked during a brief train of PAG stimulation at a rate exceeding 333 Hz Žusually 1000 Hz.. Confirmation that discharges from PAG and peripheral nerve stimulation represented the same cell was provided by collision of orthodromic and antidromic action potentials during a critical interval. Spinal neurons were considered to be first-order interneurons if the latency difference between the onset of the cord dorsum potential and the unit dis-

charge evoked from stimulation of the peripheral nerve was ( 2 ms. 2.7. Histology Recording and stimulation sites in the PAG were marked electrolytically Ž50–100 mA, 2–4.5 min. at the end of the experiment. In the PAG field potential experiments, the animals were perfused transcardially with phosphate buffered saline containing 0.1% NaNO 2 , followed by 4% paraformaldehyde in phosphate buffer Ž0.1 M, pH 7.4.. The fixed brainstem was removed and post-fixed in 4% paraformaldehyde and transferred into 20% sucrose for cryoprotection. Fifty micrometer sections were cut with a cryotome and stained with thionin. The location of the electrolytic lesion was determined according to a cat brain atlas w36x. In spinal cord experiments the brainstem was perfused through the carotid artery with 350 ml of 0.9% saline followed by 350 ml of 10% formalin. Subsequent processing was as described above. Spinal cord recording locations were marked by cutting off and leaving the tip of the

Fig. 4. Sacral spinal neuron projecting to PAG and conveying information from PL nerve. Electrical stimulation Ž50 Hz. of a site in lateral PAG ŽP1.0, 1.8 mm right of midline, 4.5 mm from the surface of the midbrain.ŽA. evoked bladder contraction which was much smaller than that evoked by stimulation of the PL nerve Ž10 Hz. ŽC, top trace.. PAG stimulation produced a pressor response which was much larger than that produced by PL nerve stimulation ŽC, bottom trace.. PAG stimulation Ž200 mA, 3 pulses at 333 Hz. antidromically activated a neuron in an intermediate location in the dorsal horn of anterior S3 ŽaS3. ŽB. with a latency of 41 ms ŽD, top trace.. The neuron was also activated, at a latency of 22 ms, by stimulation of PL nerve Ž2.5 T, 3 pulses at 333 Hz. ŽD, bottom trace.. Vertical scale in the bladder pressure trace ŽC, top trace. is 10 cm H 2 O for the first peak and 20 cm H 2 O for the second peak. Vertical scale in the blood pressure trace ŽC, bottom trace. is in mmHg. Horizontal bars in panel C indicate duration of stimulus applied to the PAG and PL nerve and apply to both traces. Downward arrows in panel ŽD. indicate application of electrical stimulation to the PAG and PL nerve. In ŽD., vertical bar represents 100 mV and horizontal bar represents 50 ms.

M. Duong et al.r Brain Research 819 (1999) 108–119

glass electrode in one track of each plane. A block of the spinal cord containing the electrode tips was removed and placed in 10% formalin. Fifty micrometer sections were cut with a cryotome, stained with thionin and the microelectrode tracks were identified.

3. Results

113

was a cluster of maxima for sPud in the P0.5–P1.0 level at an intermediate depth in the PAG. 3.2. Spinal cord units The placement of the stimulating electrode in PAG was determined either by evoking a bladder contraction response to stimulation of the PAG or by recording a field potential in the PAG following stimulation of the PL

3.1. PAG extracellular field potentials Multiphasic field potentials could be elicited in the PAG by stimulation of the PL nerve in 13 experiments. Data from a representative experiment are shown in Fig. 1. The fastest component of the field potential illustrated in Fig. 1B was not consistently present in PL nerve stimulation-evoked field potentials in other experiments. Central latencies for the first consistent components of field potentials were measured as the differences in latency measured at the onset of the PAG field potentials and the onset of the corresponding cord dorsum potentials evoked by stimulation of different peripheral nerves. Ipsilateral and contralateral PL nerve stimulation evoked PAG field potentials at similar latencies, but these were significantly shorter than those for sPud and SPeri ŽTable 1.. Assuming no synaptic delays, this would yield a conduction velocity for the ascending pathway for PL afferents of about 43 mrs. The PL nerve stimulation-evoked field potentials diminished with distance from the maximum field potential ŽFig. 1.. The field potential recordings for sPud and SPeri displayed similar characteristics to those for PL nerve stimulation. They were multiphasic and the amplitude diminished with distance from the maximum Ždata not shown.. At the location where the maximum field potential was evoked following the stimulation of the left PL, SPeri and sPud stimulation did not evoke a maximum field potential ŽFig. 2A. Žless than 25% of maximum.. In this cat, at the sites where maximum field potentials were recorded for sPud and SPeri, left PL nerve stimulation elicited a smaller response Ž50% of maximum. ŽFig. 2B.. In 10 experiments, the PAG was searched for field potentials at antero-posterior locations rostral to and caudal to the intra-aural line. In three searching was confined to locations between AP0 and P2.0. The distributions of maximum field potentials evoked from PL, sPud, and SPeri nerve stimulation were not the same ŽFig. 3.. Most of the maximum field potentials elicited by PL nerve stimulation, were found at more caudal levels in the PAG ŽP0.5–P1.5.. Although most of these field maxima are concentrated in the dorsal region, there were also maxima in the ventrolateral region of the PAG in some experiments. Field potentials were recorded in response to stimulation of the PL nerve both ipsilateral and contralateral to the PAG electrode. The distribution of SPeri and sPud field potential maxima in the PAG ranged from A2.5 to P2.0 for sPud and A3.5 to P2.0 for SPeri ŽFig. 3.. There

Fig. 5. Distribution of lumbosacral neurons activated antidromically from stimulation sites in caudal PAG. Stimulation of PAG at sites Žfilled triangles. in P0.5–P1.5 ŽA. antidromically activated 18 units in S2–S3 ŽB–D.. Thirteen units did not respond to PL, sPud or SPeri nerve stimulation Žopen circles.. Three spinal units received input from sPud andror SPeri nerves Žfilled circles.. Two units received an input from PL nerve Žfilled squares.. Maps of anterior Ža. and posterior Žp. parts of spinal segments are presented separately.

114

M. Duong et al.r Brain Research 819 (1999) 108–119

nerves. PAG stimulation produced a rise in intravesical pressure ranging from 2 to 18 cm H 2 0 in nine experiments. Bladder pressure responses elicited by PAG stimulation were always small compared to those elicited by stimulation of the PL nerve at 10 Hz ŽFig. 4.. PAG stimulation in the region where bladder responses were produced often produced an increase in blood pressure of 25 to 75 mmHg. In five experiments, the PAG stimulating sites were determined by searching for the largest field potential responses in the PAG that could be produced with PL nerve stimulation. In three experiments, this response was confirmed by stimulating at this PAG site and examining the size of the bladder response. It was found that at PAG sites which showed a field potential in response to PL nerve stimulation, PAG stimulation produced bladder contractions. In 11 cats, 181 neurons were activated by PAG stimulation but either had a varying latency or did not follow a short train of stimuli at 333 Hz. They were thus considered to be orthodromically activated by the stimulus and were not examined further. In 3r11 cats, no units that followed

a short train of PAG stimulation at 333 Hz could be found. In eight cats, 18 units were found in the lumbosacral spinal cord which were tentatively considered to be antidromically activated from the PAG on the basis of their high frequency following and constant latency responses ŽFig. 5.. Two-thirds Ž12. of these neurons were located in the dorsal horn and intermediate area. Of the 18 neurons antidromically activated from the PAG in this study, two were spontaneously active and collision between the spontaneous activity and the PAG stimulation-evoked response could be demonstrated. Thirteen of the 18 neurons were held long enough to test all the peripheral nerves that had been prepared for stimulation. Five Ž38%. received inputs from at least one of these sources ŽFig. 5.. Three of them were located in S2 and responded to sPud only Žin lamina VII., sPudrSPerirCCF Žin lamina V. and SPerirCCF Žin lamina V., respectively ŽFig. 5.. Two neurons received an input from PL nerve. One, found in lamina VII of rostral S3, received an input from PL nerve but other inputs were not assessed since it was lost before they could be tested ŽFigs. 4 and 5.. Another, found ipsilateral to the PAG stimulation site, responded to both PL and sPud ŽFig. 5..

Fig. 6. Sacral spinal neuron projecting to PAG and receiving convergent input from sPud, SPeri and CCF nerves. PAG stimulation Žsame as in Fig. 4. antidromically activated a neuron in the dorsal horn of S2 ŽA. with a latency of 5.6 ms ŽB. and the neuron could follow a short train of stimuli at 1 kHz ŽC.. The neuron was also activated by stimulation of sPud Žnot shown., CCF ŽD., and SPeri ŽE. nerves. Collision between PAG stimulation-evoked response and the SPeri nerve stimulation-evoked response occurred at an interstimulus interval of 8 ms ŽF. but not 10 ms ŽE.. Vertical bar represents 1 mV. Horizontal bar represents 20 ms.

M. Duong et al.r Brain Research 819 (1999) 108–119

In four of five cases confirmation that the peripheral input reached the antidromically-activated neuron was obtained by using a collision test ŽFig. 6.. An S2 dorsal horn projecting neuron was identified by activation at a constant latency of 5.6 ms from the contralateral caudal PAG ŽP1.0. and followed a train of three stimuli in the PAG at 1000 Hz ŽFig. 6.. This neuron also responded to SPeri, and the SPeri stimulation-evoked discharge collided with PAG stimulation-evoked antidromic discharge at an interstimulus delay of 8 ms ŽFig. 6.. sPud and CCF ŽFig. 6. stimulation also evoked discharges which collided with PAG stimulation-evoked discharge at an interstimulus interval of 8 ms Žnot shown.. These results indicate that the spinal neuron projecting directly to the PAG received afferent input from three peripheral nerves. However, an input from PL nerve was not seen in that neuron. The mean latency for neurons that were antidromically activated by PAG stimulation was 21 " 5 ms Ž n s 17 evaluable cells., with a range of 5–60 ms, corresponding to a mean conduction velocity of 17 mrs. The projecting neuron in lamina VII which responded to left PL nerve stimulation, had a conduction velocity of 8.8 mrs. The conduction velocity for the neuron in lamina VII responding to only sPud was 12.4 mrs. The antidromically activated neurons receiving excitatory convergent peripheral input from sPudrSPerirCCF Žlamina V. and SPerirCCF Žlamina VI. had conduction velocities of 72.2 mrs and 40.1 mrs, respectively. In experiments in which antidromically activated neurons did not respond to PL inputs, there were indications that the PL nerves were functioning, e.g., responses to PL nerve stimulation in other tracks, bladder contraction following PL nerve stimulation, and the presence of PL nerve stimulation-evoked cord dorsum potentials or reflex responses. The 13 spinal neurons which were antidromically activated from the PAG but did not respond to PL, sPud, SPeri or CCF nerve stimulation could be receiving input from other visceral or somatic tissues which were not studied. 4. Discussion Previous models of bladder control postulated that bladder information was carried via the spinal cord to Barrington’s nucleus in the dorsolateral pontine tegmentum. This was recognized as the center for bladder control as lesions of this area abolished micturition w5,17x. Recently, it has been proposed that the periaqueductal gray ŽPAG. region of the brainstem may be involved in micturition. Neuroanatomical studies by Blok et al. w6x have shown that there are very few neurons in the cat spinal cord with direct projections to Barrington’s nucleus. Retrograde labelling revealed direct spinal cord projections to the PAG w41x. Our aim in this study was to determine whether this pathway indeed carries information from the bladder, urethra or perineal skin.

115

4.1. PAG extracellular field potentials By recording field potentials evoked in the PAG by electrical stimulation of peripheral nerves we showed that the caudal PAG receives information from peripheral afferents which innervate the bladder, urethra, and perineal skin. The maxima for PL nerve simulation-evoked field potentials lay on the dorsal margin of the lateral PAG and the ventrolateral margin of the PAG and were in different locations than the maxima for sPud, and SPeri nerve stimulation-evoked potentials. The distribution of maximum field potentials indicates that the region which is most responsive to PL nerve stimulation is around P0.5– P1.5 w36x. This is consistent with neuroanatomical data showing that the PAG at the intercollicular level is the major target of lumbosacral afferents w6,8,41,43x. Our neurophysiological results revealed maximum field potentials for PL nerve stimulation in both lateral and ventrolateral regions of the PAG. Precise correlation with the longitudinal columnar concept of PAG neuronal organization could not be made in these studies because the characteristic responses to stimulation associated with the different columns were not assessed and the tissue was not processed for characteristic neurochemical features w3x. Neural tracing studies have described projections from the lumbosacral cord that terminate in caudal lateral PAG in cats w6,29,41x. Several anterograde tracer studies have described a single concentration of terminal labelling in lateral PAG from lumbar or lumbosacral injections that become two concentrations of terminal labelling which are dorsally and ventrally placed at more caudal levels of the lateral PAG w8,42,43x. In the rat, retrograde tracer injections into the ventrolateral PAG resulted in denser neuronal labelling at the lumbosacral region, compared to dorsolateral PAG injection w24x. A HRP tracing study in cats revealed projections to Barrington’s nucleus from lateral regions of the PAG w6x. However in rat, there is greater Phaseolus Õulgaris-leucoagglutinin labelling in Barrington’s nucleus following tracer injection into lateral PAG than into ventrolateral PAG w9x. Based on these controversial findings, it is evident that the circuitry within the PAG is very complex and unclear with respect to the micturition pathway. There also may be species differences. The neuroanatomical studies mentioned above do not elucidate the functional role of axonal projections from the sacral cord to the PAG and from the different regions of the PAG to Barrington’s nucleus. Studies in rat have demonstrated an optimum site dorsally placed in the lateral PAG at which short latency PL nerve stimulation-evoked field potentials could be recorded w32x. However, electrical stimulation at this site did not evoke a bladder contraction w32x. Furthermore, studies from our laboratory showed that in lateral PAG, L-glutamate microinjection did not evoke a bladder contraction ŽS. Matsuura, J.W. Downie and G.V. Allen, unpublished., and interruption of synaptic transmis-

116

M. Duong et al.r Brain Research 819 (1999) 108–119

sion in this area with CoCl 2 did not interrupt ongoing micturition contractions w26x. Noto et al. w32x showed that electrical stimulation of a site in the ventral PAG in rat elicited PL nerve discharges and bladder contraction. CoCl 2 injection into a more restricted area in ventrolateral PAG interrupts voiding in the anesthetized rat w26x. These data imply that there are two regions in the PAG involved in transmitting bladder information, a finding that corresponds with our field potential data. The role in the micturition pathway of the neurons dorsally located in the lateral PAG is unclear, but at least in rat it has been shown that the ventrolateral PAG is important in supporting reflex micturition. In a study on spinal neurons projecting to unknown supraspinal targets, ascending axons responsive to PL nerve stimulation also responded to Pud and hypogastric nerve stimulation w27x. Based on our extracellular field potential findings, there is a possibility that the ascending pathway to the PAG is specific for bladder input since the stimulation of other nerves ŽsPud and SPeri. did not elicit maximum field potentials in the same location in the PAG. Resolution of this point required determining the response characteristics of individual neurons identified as projecting to the PAG. 4.2. Spinal cord units After obtaining evidence that the PAG receives input from the bladder branches of the PL nerve, we proceeded to determine whether this input was conveyed on directly projecting neurons and to determine whether these neurons carried other information as well. The sacral spinal cord was searched for neurons that could be antidromically activated by PAG stimulation, which would identify a direct Žnon-synaptic. projection from the spinal cord to the PAG. These projecting neurons were then tested for their responsiveness to electrical stimulation of PL, sPud, SPeri and CCF nerves to gain some insight into the information they might transmit. The PAG stimulation site was chosen by optimizing the bladder contraction produced by this stimulation. PAG stimulation has been shown previously to induce bladder contraction w16,20,32x. However, Skultety w35x found that stimulation of the rostral PAG elicited bladder contractions in only 3 out of 20 cats. Stimulation of lateral regions of rostral PAG did not elicit bladder contraction and ablation of the same area did not affect bladder activity w35x. Bilateral electrolytic lesion of the ventrolateral regions of the PAG at P1.5 also did not abolish micturition in cats w38x. These studies imply that the PAG is not involved in the micturition reflex pathway. However, the PAG covers a large area, so it is possible that lesioned areas did not eliminate the neurons involved in micturition. In this study, stimulation of more caudal regions of the PAG, a choice based on our field potential studies, not only elicited bladder contraction, but also antidromically

activated 18 sacral spinal cord neurons. Although this yield seems low, it corresponds with the results from other laboratories. Vanderhorst et al. w41x found a total of 115 retrogradely labelled neurons in S2 from an injection of WGA-HRP into the ventrolateral PAG ŽP1.0.. Furthermore, Yezierski and Schwartz w44x found only 13 antidromically activated neurons in L7–S1 following stimulation of 13 sites within the caudal PAG using an array of 2–4 stimulating electrodes. There are a number of possibilities for the paucity of directly projecting neurons found in our experiments. First, the PAG sites which elicited a large bladder response on electrical stimulation may not correspond to the site of termination of spinal projections. This could arise if there were an interneuron path within the PAG or if the descending axons of the bladder-activating pathway were stimulated. Even the location of the maximum PL nerve stimulation-evoked field potentials in the PAG may not correspond to the termination site of the ascending pathway in the PAG if the terminals of the projecting neurons spread diffusely in the PAG. Secondly, the stimulating electrode configuration and stimulation parameters may not have been optimum for antidromic activation of the projecting cells. Last, the spinal ascending neurons carrying afferent information from the PL nerve may not project directly to the area of maximum evoked field potential in the PAG. For example, the largest field potential could be produced postsynaptically. The mean conduction velocity for the projecting neurons found in this study Ž17 mrs. was similar to the conduction velocity Ž34.9 " 21.1 mrs. estimated for spinomesencephalic tract cells in lamina I and V–VIII of the lumbosacral spinal cord which were antidromically activated from the PAG ŽP1.5. w44x. Spinomesencephalic projecting cells originating in the superficial spinal laminae ŽI and II. have been reported to have slow conduction velocities Žmean: 14.1 " 5.7 mrs. while cells in lamina III–IV and VII–VIII conduct faster Žmean: 56.3 " 20.8 mrs.. The range of conduction velocity for cells found throughout lamina I–VIII and activated by stimulation of the PAG at AP0 was 8.8–102 mrs w44x. In our study, the only antidromically activated cell that received PLN input had a conduction velocity of 8.8 mrs, which would fall within the large range reported by Yezierski and Schwartz w44x, but this neuron was located in lamina VII. McMahon and Morrison w27x estimated that the conduction velocity for fibers conveying PL input between L4 and the brainstem to be about 30 mrs, which corresponded with the conduction velocities determined for other autonomic spino-bulbar pathways Ž20–30 mrs for cardiac and renal nerves w11x.. Their calculation supports the mean conduction velocity estimated in the present study. De Groat w12x reported a conduction velocity of 10–11 mrs for the ascending fibers to Barrington’s nucleus Žusing a latency of 30–40 ms from PL nerve stimulation-evoked field potentials in the rostral pontine areas ŽBarrington’s nucleus. and a conduction distance of 400 mm.. The latency

M. Duong et al.r Brain Research 819 (1999) 108–119

of PL nerve stimulation-evoked potentials recorded in Barrington’s nucleus w12,23x is longer Ž30–50 ms. than the latency recorded in the PAG in our study Ž11.5 ms.. This finding is compatible with the afferent information from the bladder being relayed through the PAG, or elsewhere, before it reaches Barrington’s nucleus. We sought to determine whether the population of sacral spinal neurons projecting to the PAG contained a subgroup that responded exclusively to PL input. It was reasoned that such a population would be most likely to represent the afferent limb of the micturition reflex model proposed by Blok and Holstege w7x. However, it was recognized that an earlier study failed to find ascending neurons with purely PL input w27x. This led to a hypothesis in which the specificity in the micturition reflex was determined by PL afferent gating of the spinal cord output rather than by transmission of bladder-specific afferent information to the brainstem w27x. In the present study, information from three peripheral nerves ŽsPud, SPeri, CCF. converged onto a sacral spinal neuron located in the dorsal horn and traveled in the same axon to a ventrolateral region of the PAG ŽP1.0.. Although it is not known how early along the transmission path the convergence of input occurs, the central latency of the response to the peripheral nerve stimulation indicates that this projecting neuron was not a first-order interneuron for any of the three nerve inputs. Different peripheral nerve inputs can travel and terminate onto specific neurons in the sacral cord where they may interact with other interneurons in the cord before the information is conveyed to the PAG. Of the five antidromically activated sacral neurons, only one was a first-order interneuron Žfor SPeri input.. This neuron also served as a higher-order interneuron for sPud and CCF transmission. Although our data demonstrate a convergence of different peripheral nerve inputs onto spinal-PAG projecting cells, the inputs that converged onto most of these projecting neurons did not include the PL nerve. Therefore, these results do not negate the possibility of a bladder-specific input to the PAG. However, the data do provide evidence for a non-specific pathway for sPud, SPeri and CCF inputs to the PAG. The fact that there was an antidromically activated neuron that responded only to sPud input suggests that there may be both specific and non-specific pathways for the transmission of sPud inputs to the PAG. These data would fit with the data from the field potential study in which maximum field potentials for sPud were found in a wide range of AP levels ŽA2.5–P2.0., and there was a cluster of maxima in the lateral Žintermediate. region of the PAG at P0.5–P1.0. The pudendal nerve conveys afferent information from other structures in addition to the urethra. It conveys information from somatic structures such as the skin of penis, clitoris and perineum w41x, striated muscles of the pelvic floor w34x and the urethral and anal mucosa w22x. Input from visceral structures such

117

as the vagina, and part of the uterine cervix are also conveyed by these nerves w41x. It is possible that inputs, conveyed in the Pud nerve from these different regions, separate at the level of the spinal cord and travel in different paths to terminate at different regions of the PAG responsible for different functions, while other Pud inputs converge with other peripheral nerve inputs onto spinal projecting neurons. The PAG has been shown to be involved in a variety of functions such as lordosis w33x, cardiovascular regulation w25x, vocalization w45x and pain modulation w2,10x. Vanderhorst et al. w41x observed that in cats, neurons retrogradely labelled by HRP injection to the lateral PAG were located in regions that overlapped with pelvic and pudendal afferent terminals in the sacral cord. Also in cats, Blok and Holstege w7x demonstrated that the lateral part of the PAG contained neurons projecting to Barrington’s nucleus. This is the region that, in our work, contains a clustering of maximum field potentials for sPud stimulation. Therefore, neurons in the lateral PAG which receive projections from the sacral cord Žin an area of Pud afferent terminations. could convey Pud information to the L-region, where neurons projecting to Onuf’s nucleus are located w18x. Pudendal afferent information related to micturition may be conveyed in a private pathway to the lateral region of the PAG. The other inputs carried by the pudendal nerve may terminate in other regions of the PAG related to lordosis, or pain modulation. This is supported by our finding that only 5 of the 18 cells projecting directly to the PAG received inputs from our test nerves, suggesting that the major part of this pathway is not directly related to lower urinary tract function. The finding that only 2 out of 18 neurons antidromically activated from the PAG received a PL input implies that there are very few direct projections to the PAG conveying bladder-specific information. Even so, this is also evidence for direct projections to the PAG from the sacral cord conveying PL input.

4.3. Conclusions By recording field potentials evoked in the PAG by electrical stimulation of peripheral nerves we have shown that the caudal PAG receives information from the bladder, urethra, and perineal skin. The maxima for PL nerve simulation-evoked field potentials lay on the dorsolateral and ventrolateral margins of the PAG and were in locations different than the maxima for sPud, and SPeri nerve stimulation-evoked potentials. Sacral spinal neurons antidromically activated from PAG were uncommon and did not often show response to PL nerve stimulation. Thus it seems that although direct projection from the sacral spinal cord to the periaqueductal gray occurs in the cat, the majority of information carried by this pathway concerns functions other than control of the urinary bladder.

118

M. Duong et al.r Brain Research 819 (1999) 108–119

Acknowledgements This work was supported by the Medical Research Council of Canada ŽMT-13486.. References w1x E.D. Al-Chaer, N.B. Lawand, K.N. Westlund, W.D. Willis, Pelvic visceral input into the nucleus gracilis is largely mediated by the polysynaptic dorsal column pathway, J. Neurophysiol. 76 Ž1996. 2675–2690. w2x R. Bandler, P. Carrive, S.P. Zhang, Integration of somatic and autonomic reactions within the midbrain periaqueductal grey: viscerotopic, somatotopic and functional organization, Prog. Brain Res. 87 Ž1991. 269–305. w3x R. Bandler, M.T. Shipley, Columnar organization in the midbrain periaqueductal gray: modules for emotional expression, Trends Neurosci. 17 Ž1994. 379–389. w4x F.J.F. Barrington, The relation of the hind-brain to micturition, Brain 4 Ž1921. 23–53. w5x F.J.F. Barrington, The effect of lesion of the hind- and mid-brain on micturition in the cat, Q. J. Exp. Physiol. 15 Ž1925. 81–102. w6x B.F.M. Blok, H. De Weerd, G. Holstege, Ultrastructural evidence for a paucity of projections from the lumbosacral cord to the pontine micturition center or M-region in the cat: a new concept for the organization of the micturition reflex with the periaqueductal gray as central relay, J. Comp. Neurol. 359 Ž1995. 300–309. w7x B.F.M. Blok, G. Holstege, Direct projections from the periaqueductal gray to the pontine micturition center ŽM-region.. An anterograde and retrograde tracing study in the cat, Neurosci. Lett. 166 Ž1994. 93–96. w8x A. Blomqvist, A.D. Craig, Organization of spinal and trigeminal input to the PAG, in: A. Depaulis, R. Bandler ŽEds.., The Midbrain Periaqueductal Gray Matter, Plenum, New York, 1991, pp. 345–363. w9x A.A. Cameron, I.A. Khan, K.N. Westlund, W.D. Willis, The efferent projections of the periaqueductal gray in the rat: a Phaseolus Õulgaris-Leucoagglutinin study: II. Descending projections, J. Comp. Neurol. 351 Ž1995. 585–601. w10x P. Carrive, R. Bandler, Viscerotopic organization of neurons subserving hypotensive reactions within the midbrain periaqueductal grey: a correlative functional and anatomical study, Brain Res. 541 Ž1991. 206–215. w11x H.J. Coote, C.B.B. Downman, Central pathways of some autonomic reflex discharges, J. Physiol. 183 Ž1966. 714–729. w12x W.C. de Groat, Nervous control of the urinary bladder of the cat, Brain Res. 87 Ž1975. 201–211. w13x M.J. Espey, H.-J. Du, J.W. Downie, Serotonergic modulation of spinal ascending activity and sacral reflex activity evoked by pelvic nerve stimulation in cats, Brain Res. 798 Ž1998. 101–108. w14x B. Fedirchuk, S.J. Shefchyk, Effects of electrical stimulation of the thoracic spinal cord on bladder and external urethral sphincter activity in the decerebrate cat, Exp. Brain Res. 84 Ž1991. 635–642. w15x H.L. Fields, L.D. Partridge, D.L. Winter, Somatic and visceral receptive field properties of fibers in ventral quadrant white matter of the cat spinal cord, J. Neurophysiol. 33 Ž1970. 827–837. w16x R. Gjone, Excitatory and inhibitory bladder responses of ‘limbic’, diencephalic and mesencephalic structures in the cat, Acta Physiol. Scand. 66 Ž1966. 91–102. w17x D. Griffiths, G. Holstege, E. Dalm, H. de Wall, Controls and coordination of bladder and urethral function in the brainstem of the cat, Neurourol. Urodyn. 9 Ž1990. 63–82. w18x G. Holstege, D. Griffiths, H. De Wall, E. Dalm, Anatomical and physiological observations on supraspinal control of bladder and urethral sphincter muscles in the cat, J. Comp. Neurol. 250 Ž1986. 449–461.

w19x C.-H. Jiang, S. Lindstrom, ¨ L. Mazieres, ` Segmental inhibitory control of ascending sensory information from bladder mechanoreceptors in cat, Neurourol. Urodyn. 10 Ž1991. 286–288. w20x H. Kabat, H.W. Magoun, S.W. Ranson, Reaction of the bladder to stimulation of points in the forebrain and mid-brain, J. Comp. Neurol. 63 Ž1936. 211–239. w21x M. Kuru, The nervous control of micturition, Physiol. Rev. 45 Ž1965. 425–494. w22x S. Kuzuhara, I. Kanazawa, T. Nakanishi, Topographical localization of the Onuf’s nuclear neurones innervating the rectal and vesical striated sphincter muscle: a retrograde fluorescent double labeling in cat and dog, Neurosci. Lett. 16 Ž1980. 125–130. w23x P.M. Lalley, W.C. de Groat, P.L. McLain, Activation of the sacral parasympathetic pathway to the urinary bladder by brainstem stimulation, Fed. Proc. 31 Ž1972. 386. w24x R.P.C. Liu, Laminar origins of spinal projection neurons to the periaqueductal gray of the rat, Brain Res. 264 Ž1983. 118–122. w25x T. Lovick, Integrated activity of cardiovascular and pain regulatory systems: role in adaptive behavioural responses, Prog. Neurobiol. 40 Ž1993. 631–644. w26x S. Matsuura, G.V. Allen, J.W. Downie, Volume-evoked micturition reflex is mediated by the ventrolateral periaqueductal gray in the anesthetized rats, Am. J. Physiol. 275 Ž1998. R2049–R2055. w27x S.B. McMahon, J.F.B. Morrison, Spinal neurones with long projections activated from the abdominal viscera of the cat, J. Physiol. 322 Ž1982. 1–20. w28x C. Morgan, I. Nadelhaft, W.C. de Groat, The distribution of visceral primary afferents from the pelvic nerve within Lissauer’s tract and the spinal gray matter and its relationship to the sacral parasympathetic nucleus, J. Comp. Neurol. 201 Ž1981. 415–440. w29x L.J. Mouton, V.G.J.M. VanderHorst, G. Holstege, Large segmental differences in the spinal projections to the periaqueductal gray in the cat, Neurosci. Lett. 238 Ž1997. 1–4. w30x P.W. Nathan, Awareness of bladder filling with divided sensory tract, J. Neurol. Neurosurg. Psychiatry 19 Ž1956. 101–105. w31x P.W. Nathan, M.C. Smith, The centripetal pathway from the bladder and urethra within the spinal cord, J. Neurol. Neurosurg. Psychiatry 14 Ž1951. 262–280. w32x H. Noto, J.R. Roppolo, W.D. Steers, W.C. de Groat, Electrophysiological analysis of the ascending and descending components of the micturition reflex pathway in the rat, Brain Res. 549 Ž1991. 95–105. w33x Y. Sakuma, D. Pfaff, Mesencephalic mechanisms for integration of female reproductive behaviour in the rat, Am. J. Physiol. 237 Ž1979. R278–R284. w34x M. Sato, N. Mizuno, A. Konishi, Localization of motoneurons innervating perineal muscles: an HRP study in cat, Brain Res. 140 Ž1978. 149–154. w35x F.M. Skultety, Relation of periaqueductal gray matter to stomach and bladder motility, Neurology 9 Ž1959. 190–198. w36x R.S. Snider, W.T. Niemer, A Stereotaxic Atlas of the Cat Brain, The Univ. of Chicago Press, 1961. w37x K. Sugaya, K. Matsuyama, K. Takakusaki, S. Mori, Electrical and chemical stimulations of the pontine micturition center, Neurosci. Lett. 80 Ž1987. 197–201. w38x P.C. Tang, T.C. Ruch, Localisation of brainstem and diencephalic areas controlling the micturition reflex, J. Comp. Neurol. 106 Ž1956. 213–245. w39x T. Ueyama, N. Mizuno, S. Nomura, A. Konishi, K. Itoh, H. Arakawa, Central distribution of afferent and efferent components of the pudendal nerve in the cat, J. Comp. Neurol. 222 Ž1984. 38–46. w40x R.J. Valentino, M.E. Page, P.-H. Luppi, Y. Zhu, E. Van Bockstaele, G. Aston-Jones, Evidence for widespread afferents to Barrington’s nucleus, a brainstem region rich in corticotropin-releasing hormone neurons, Neuroscience 62 Ž1994. 125–143. w41x V.G.J.M. Vanderhorst, L.J. Mouton, B.F.M. Blok, G. Holstege, Distinct cell groups in the lumbosacral cord of the cat project to

M. Duong et al.r Brain Research 819 (1999) 108–119 different areas in the periaqueductal gray, J. Comp. Neurol. 376 Ž1996. 361–385. w42x M. Wiberg, A. Blomqvist, The spinomesencephalic tract in the cat: its cells of origin and termination pattern as demonstrated by the intraaxonal transport method, Brain Res. 291 Ž1984. 1–18. w43x R.P. Yezierski, Spinomesencephalic tract: projections from the lumbosacral spinal cord of the rat, cat, and monkey, J. Comp. Neurol. 267 Ž1988. 131–146.

119

w44x R.P. Yezierski, R.H. Schwartz, Response and receptive-field properties of spinomesencephalic tract cells in the cat, J. Neurophysiol. 55 Ž1986. 76–96. w45x S.P. Zhang, P.J. Davis, R. Bandler, P. Carrive, Brainstem integration of vocalization: role of the midbrain periaqueductal gray, J. Neurophysiol. 72 Ž1994. 1337–1356.