67
Brain Research, 598 (1992) 67-75 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 18321
Rostral ventrolateral medullary neurons projecting to locus coeruleus have cardiorespiratory inputs Donghai Huangfu, Anthony J.M. Verberne
* a n d P a t r i c e G. G u y e n e t
University of Virginia School of Medicine, Department of Pharmacology, Charlottesville, VA 22908 (USA) (Accepted 7 July 1992)
Key words: Rostral ventrolateral medulla; Locus coeruleus; Respiratory and cardiovascular control
This study was designed to characterize some of the properties of the rostral ventrolateral medullary (RVLM) cells with axonal projection to the locus coeruleus (LC) in urethane anesthetized, vagotomized, paralyzed and artificially respirated rats. The vast majority of RVLM units antidromically (AD) activated from LC (RVLM-LC units) were silent and unresponsive to peripheral chemoreceptor stimulation or nociceptive stimulation. Twenty seven spontaneously active RVLM-LC neurons, AD activated from LC with currents below 30 p.A (17+2 tzA) were analyzed. AD mapping (n = 18) indicated that the lowest threshold for AD activation occurred within the LC itself. Axonal branching within or close to LC was suggested by the presence of sudden jumps in AD latency. Maximal AD latencies ranged from 7 to 37 ms. Most spontaneously active RVLM-LC neurons displayed marked central respiratory modulation characterized by either a post-inspiratory or an inspiratory pattern. The majority of the tested neurons were affected (excited or inhibited) by brief peripheral chemoreceptor stimulation (N 2 inhalation). Most cells were inhibited by raising arterial pressure but none exhibited any detectable pulse synchrony. Reticulospinal sympathetic premotor neurons of RVLM were not found to project to LC (sample of 9) and very few RVLM cells with on-off respiratory discharges appeared to project to LC (2 out of 110). This study suggests that much of the information conveyed by the RVLM to LC could be of a mixed cardiorespiratory nature.
INTRODUCTION The rostral ventrolateral medulla (RVLM) is a major source of afferent inputs to the somata and proximal dendrites of the noradrenergic cells of the locus coeruleus ( L C ) 1'2'7'13. This pathway includes phenotypically adrenergic cells of the C1 group e°'23. The physiological role of the RVLM-LC projection is largely unknown. It may somehow be involved in the excitation of LC cells during opiate withdrawal e5 and some evidence suggests that RVLM might be a relay in the poly-synaptic pathway responsible for LC activation by nociceptive somatic stimulation 2,s. The latter notion has been challenged by Rasmussen and Aghajanian's observation that massive high frequency bilateral thermal lesions of RVLM fail to alter the response of LC neurons to nociceptive stimuli 26. The prominent role of RVLM in vasomotor and respiratory control e7 (for reviews, see refs. 10,11) suggests an alternative role of the RVLM-LC projection: namely, it might convey information of a cardio-re-
spiratory (or enteroceptive) nature. This view derives some support from prior evidence that LC unit activity is influenced by arterial pressure, hypoxia and blood volume changes 5,6,21. In addition the LC also displays a marked central respiratory rhythmicity during hypoxia or CO 2 inhalation (Guyenet, Huangfu and Koshiya, unpublished observations). The purpose of the present study was to test whether the LC might receive cardio-respiratory information directly from its afferent inputs from RVLM. This goal was pursued by analyzing the properties of RVLM cells antidromically (AD) activated from LC with very low current thresholds. MATERIALS AND METHODS General procedures Experiments were carried out in 17 male Sprague-Dawley rats weighing from 320 to 380 g. The rats were initially anesthetized with ether. After tracheostomy, anesthesia was maintained by artificial ventilation with 1.3-1.5% halothane in 100% 0 2 for the remainder of the surgery. The right femoral artery and a femoral or jugular vein
Correspondence: P.G. Guyenet, Department of Pharmacology, University of Virginia, Charlottesville, VA 22908, USA. Fax: (1) (804) 982-3878. * Present address: Austin Hospital, Heidelberg 3084, Victoria, Australia.
68 B
A
v
O
•
I v
I I
J
Fig. 1. Antidromic (AD) identification of RVLM-LC neurons. A: an example of a collision test (4 superimposed sweeps per trace) used to identify a RVLM cell with projection to LC. Spontaneous spikes (©), AD spikes (*), stimuli delivered at arrow. Calibration bar, 100 ~V, 5 ms. B: example of sudden jumps in AD latency with increased stimulation currents (4 sweeps per trace; stimulation intensity from top to bottom: 30, 100 and 200 ~A, 0.5 ms duration). Calibration bar, 100/xA, 2 ms.
were cannulated for measurement of arterial pressure (AP) and injection of drugs. Bilateral vagotomy was performed to desynchronize the central respiratory rhythm from ventilation ~4. Body temperature was maintained at 37-39°C. The facial nerve (mandibular branch) was exposed for AD stimulation to help locate RVLM which lies just caudal to the facial motor nucleus 15-17. The left phrenic nerve was isolated, cut distally, placed over a bipolar silver electrode and immersed in Silgel (Wacker compound 907) for multi-unit recording t4. After completion of surgery, halothane was withdrawn, and the anesthesia was replaced by intravenous administration of urethane in the dose of 1 g / k g (the anesthetic was changed because halothane depresses the chemo-reflex). After half an hour stabilization and verification of the adequacy of anesthesia (lack of corneal reflex and absence of retraction of distal phalanges to strong pinch of the hindpaw), the rat was paralyzed with pancuronium bromide (1 m g / k g initial dose followed by maintenance dose of 0.3 m g / k g / h infused from the femoral vein with a syringe pump). Peripheral chemoreceptor stimulation was performed by respirating the rats with 100% N 2 for 10 s. This paradigm produces increases in AP, phrenic activity and a generalized increase in sympathetic
outflow 4 (Guyenet, Huangfu and Koshiya, unpublished observations). In the present preparation, the cardio-respiratory effects of brief N~ inhalation were completely eliminated by selective bilateral section of the carotid sinus nerve; therefore, in confirmation of the results of Biesold et al. 4, these effects were entirely due to carotid chemoreceptor stimulation (Guyenet, Huangfu and Koshiya, unpublished observations). Baroreceptor stimulation and unloading were done by intravenous administration of a bolus of phenylephrine (4 /.tg in 20 p.I) and sodium nitroprusside (10/xg in 20 ~zl)TM respectively. Nociceptive stimulation was done by squeezing the toe of the hindlimb with a pair of forceps. Combined central and peripheral chemoreceptor stimulation was done by adding 5 - 8 % CO 2 to the breathing mixture (balance 02).
Electrical stimulations and recordings The rat was placed in a stereotaxic apparatus with the incisor bar 3.5 mm below the interaural line and suspended by a clamp holding the spinal processes at the Ts_ 6 level. A parieto-occipital craniotomy and a laminectomy at the T 2 level were performed. A fine metal electrode with impedance from 2 to 5 M O was inserted into LC transcerebrally with the manipulator angled 35 ° rostrally for recording or stimulating. The approximate coordinates were 3 mm anterior to lambda, 1.1 mm lateral to midline, 7.5-9.0 mm below the brain surface. Extracellular recordings from the electrode during implantation were used for accurate placement in the LC. The nucleus was identified by its characteristic spontaneous and nociceptive evoked firing patterns and its location medial to the mesencephalic nucleus of the trigeminal nerve, as previously described m2. A second monopolar stimulating electrode was inserted into the spinal cord at the T 2 level with its tip 0.9 mm below the dorsolateral sulcus for AD activation of RVLM bulbospinal cells 15'16. RVLM neurons were recorded with glass electrodes filled with 2% Pontamine sky blue in 0.5 M sodium acetate (impedance 5-10 Mg2). The medullary area explored was between 1.5 and 2.2 mm lateral to the midline, within 1 mm of the ventral surface, and from 0 to 1.5 mm caudal to the facial motor nucleus, initially mapped by AD field potentials evoked by stimulation of the mandibular branch of the facial nerve 15-iv. For AD activation of RVLM neurons, LC was routinely stimulated with 200 /zA of 0.5 ms pulses at 1 Hz. Antidromic identification used the following criteria3J4: (1) constant latency of evoked spikes, (2) ability to follow paired pulse stimulation
~ -0.68
!'~
'
~ '~-
-
~ ~y
1
.
6
~-~
Fig. 2. Anatomy of recording and stimulation sites. Anatomical location of stimulation sites (locus coeruleus on the left) and RVLM unit recordings (on the right). Coronal sections redrawn after ref. 22. Numbers next to pontine sections (left side) refer to Paxinos's nomenclature of coronal planes 22 (ram interaural). Numbers next to medullary sections (right side) refer to mm rostral to obex. 7n, facial nerve; A, nucleus ambiguus; IO, inferior olive; LC, locus coeruleus; LPB, lateral parabrachial nucleus; Me5, mesencephalic trigeminal nucleus; Mo5, trigeminal motor nucleus; MPB, medial parabrachial nucleus; nTS, nucleus of solitary tract; Pr5, principal trigeminal sensory nucleus; PY, pyramidal tract; s5, sensory root of trigeminal nerve; SO, superior olive; sp5, spinal trigeminal tract; Sp5i, interpolar trigeminal nucleus; Sp50, oral spinal trieeminal nucleus.
69 at frequencies of 200 Hz or greater, and (3) collision of evoked spikes with spontaneous impulses (Fig. 1A). Neuronal responses to stimulation of the spinal cord was also performed in an attempt to identify spinal projections ( < 1 mA, 0.5 ms pulses). In a few cases, five-barrel iontophoretic electrodes were used, filled with the following drugs: L-glutamate (Aldrich, 0.2 M in 160 m M NaCI, p H 8), y-aminobutyric acid (GABA, Aldrich, 0.2 M in 160 m M NaCl, pH 6), and clonidine (Sigma, 0.1 M in 160 m M NaC1, pH 5). Current balancing was made through a channel filled with 2 M NaCl. Retaining currents of 5 nA were used between periods of drug ejection. The recording channel contained 2% Pontamine sky blue in 2 M NaCl. Arterial blood pressure (AP), heart rate, intratracheal pressure (ITP), phrenic nerve discharge (PND, band pass of 10-3,000 Hz, half-wave rectified and integrated), integrated rate histogram of neuron activity and percentage of oxygen in inspired air (FiO 2) were displayed on a chart recorder. AP, ITP, P N D and neuronal activity were also recorded on tape for later analysis. PND-triggered histograms of R V L M unit discharges (100 sweeps, 4 ms bins, 2,400 ms sweep duration), and A P pulse-triggered histograms (200-500 sweeps, 0.5 ms bins, 300 ms sweep duration) were obtained as described before 14d5.
RESULTS
RVLM cells activated from LC with a threshold of less than 30 IzA In 17 rats, 554 cells were encountered in 224 micropipette penetrations through the RVLM, and 150 units were putatively AD activated from the region of the LC, judging from the presence of evoked spikes with invariant latency and a 200 Hz following frequency. This section focuses on the cells with a threshold for AD activation of < 30 /.tA, based on the notion that the lower the threshold the more likely the cell was to project t o / t h r o u g h LC. This group numbered 112 units (mean latency 11 + 1 ms, range 6-46 ms, mean threshold 19 + 1 /xA), the vast majority of which were silent (n = 81). These silent units were activated neither by nociceptive stimuli (toe pinch) nor by peripheral chemoreceptor stimulation, and were not studied further. The rest of the RVLM-LC neurons were spontaneously active and all met the requirements of the collision test (Fig. 1A). Twenty-seven were recorded for a period sufficient to run a series of experimental tests. The remainder of this section focuses on the
Histology R V L M recording sites (1 per animal) were marked by ejection of Pontamine sky blue ( 1 0 / x A cathodal current for 5 min). Stimulation sites (LC only) were marked by iron deposits (10 ~ A anodal current for 10 s) revealed by the Prussian blue reaction. Recording and stimulation sites were identified in Nissl-stained coronal sections and mapped on standardized sections redrawn with reference to the atlas of Paxinos and Watson 22. All data reported in this paper are from rats in which LC stimulation sites were histologically verified and recording sites were confirmed to be in R V L M (Fig. 2). Results are presented as means_+ S.E.M. TABLE I
Properties o f R V L M neurons projecting to L C c / s , spikes per second; lat, latency; thre, threshold; OD, orthodromic (with reference to activation from SC, spinal cord); glu, glutamate; clo, clonidine; ion, iontophoretic application; / , cell not tested; - , inhibition; + , excitation; 0, no change; + - , excitation followed by inhibition.
No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Firing rate (c / s)
A D response to L C st lat (ms) thre (~ A)
OD ~ t to SC st (ms)
Responses to N2
5% CO 2
Toe pinch
PE L c.
glu ion
GABA ion
clo ion
clo i.v.
10 4 4 2 3 1.2 1.2 3 2 1.5 6 4.5 5 1 12 5 4 10 4 22 10 10 16 3 10 0.4 0.2
19 19 15 18 16 8 16 18 37 22 8 13 9 22 8 18 16 14 8 8 18 19 28 13 7 15 12
15 30 0 0 0 10 0 0 16 0 / / 0 / 0 0 0 18 12 12 25 12 0 0 0 0 /
0 0 + / + / + / + + + + + 0 + + + + 0
+ + + / / + / / / / + / + / / / / / / + + + / / / /
/ / / / / 0 + / + / / + 0 0 0 0 + ++ 0 0 0 / +
0 0 / 0 / / 0 0 0 / 0 /
+ + / / / / / / / / / / / / / / / / / / / / / / + + /
/ / / / / / / / / / / / / / / / / / / / / / /
/ / / / / / / / / / / / / / / / / / / / / / /
+ / + / / / + / / / 0 0 / 0 / / 0 /
20 15 10 15 15 6 3 20 25 10 15 10 25 28 20 20 15 15 20 20 8 20 25 20 25 10 20
Respiratory rhythm
Cardiac rhythm
post-I peak I peak 0 I peak I peak I nadir post-I peak I nadir I peak 0 post-I peak I peak post-I peak 0 post-I peak 0 0 I peak post-I peak 0 I peak 0 I peak post-I peak I peak post-I peak /
0 0 0 0 0 0 0 0 0 0 0 0 0 0 + 0 0 0 0 0 0 0 0 0 0 0 /
70 properties of these 27 spontaneously firing cells recorded in 11 animals (summarized in the Table I). Fig. 2 shows the distribution of the histologically verified stimulation sites in the LC and RVLM recording sites (these sites were either directly identified from Pontamine sky blue deposits, n = 11, or reconstructed from the coordinates of the cells in relation to those of another cell that was directly identified by the dye). The firing rate of the 27 R V L M - L C neurons varied from 0.2 to 22 spikes/s with a mean of 5.7_+ 1.1 spikes/s. A D latencies of these neurons to LC stimulation ranged from 7 to 37 ms (mean 16 + 1 ms, P < 0.05 compared with that of silent group, 10 _+ 1 ms, n = 81). This latency corresponds to an average conduction velocity of 0.34 _+ 0.02 m / s (range 0.15-0.79 m / s ) based on the mean straight line distance between stimulation and recording sites (estimated at 5 mm). The refractory period of these cells ranged from 0.8 to 4.0 ms. The threshold for A D activation ranged from 3 to 28 ~ A (17 _+ 1 ~A, current measured by ohmic voltage drop across a 1 kS2 resistor). A D mapping was performed in 18 cases by stepping the stimulation electrode through and on each side of the LC. The lowest threshold for A D activation was found at a mean depth of 8.3 _+ 0.2 mm below the brain surface and marked. All the marked spots were found within the LC. Fig. 3 shows an example of a series of 3 depth-threshold curves designed to characterize the projection of neuron 21 of Table I. Note that the lowest threshold (8 p.A) was found in the middle tract, while higher thresholds and
narrower curves were found in the two lateral tracts only 0.2 mm distant. This particular case also illustrates the occurrence of sudden shifts in antidromic latency as the stimulation current was gradually raised a n d / o r the electrode was moved through the LC. This phenomenon, suggesting terminal branching 3, is also illustrated in Fig. lB. These sudden shifts generally ranged from 1 to 5 ms, but could be as large as 14 ms in an occasional case. Depth-threshold curves were also constructed in the case of 6 silent cells. The sites with the lowest threshold for AD activation were also within LC. Nine out of 23 tested cells were excited orthodromically by stimulation of spinal cord with a mean onset latency of 17 + 2 ms. The effect of peripheral chemoreceptor activation on the activity of 24 RVLM-LC neurons was tested by respirating the animals with 100% N 2 for 10 s. Twelve cells were excited from a control firing rate of 7.4 + 2.2 to 16.0 + 3.5 spikes/s (Fig. 4A1); 8 cells were inhibited from 4.2 + 1.1 to 1.3 +_ 0.8 spikes/s (Fig. 4B1). The remaining 4 cells showed no detectable response. We include in this category one unit which was slightly inhibited by chemoreceptor activation but probably as a result of a secondary baroreceptor-mediated effect of the hypoxia-induced response. Indeed this inhibition did not occur when the peripheral chemoreceptor stimulation was combined with the i.v. injection of a vasorelaxant (sodium nitroprusside) to prevent the hypoxiainduced hypertension.
7.5 o 18 ms O 16ms A 15 ms • 13ms
LI.I
LI.I --8.0
.-8.s
,~ ,ca
- -9.0
-9.5
1 mm
-I0.0 O
I~
Threshold
2~
3~
(kLA)
Fig. 3. Antidromic mapping of the projection of a RVLM neuron to LC. Left panel: scheme of 3 parallel penetrations through LC (for abbreviations see legend to Fig. 2). Characters above the 3 tracks refer to mm lateral to the midline. Right panel: depth threshold curves corresponding to tracks shown in the left panel. The lowest threshold (8 ~A) was identified by a small iron deposit revealed by the Prussian blue reaction (star in the left panel). Symbols inset into the right panel indicate AD latencies of the RVLM neurons which are different along the stimulation tracks in LC.
71 AI 30
N~
PE
V
V
BI
N2 V
30
PE V
SNP N2 V V
10 0
,
25O 100
~-~
,
•
,
_
.
,
"~
2013.
~
~'~
,
~-~
loo.
~
50.
~:
-
5
o
.
0
~
3
6
i 150. I 0 0 . 1 ~ ~ ~ 4 ~ ~ 50 0
TIME (mini A2
100
3 T I M E (rain)
6
B2
600
ioo
8oo
50 z 0
0 0
Z
25
50
0
Z
25
BIN (48 m s )
A3
B3 160
20
0
10,
Z
25
5o
BIN (6 ms)
50
BIN (48 m s ) 30
180
~
130
~ z
0
~
80 0
25
50
Bin (6 ms)
Fig. 4. Properties of RVLM-LC neurons. A and B represent two different cells. A1,BI: integrated rate histogram of cell discharges (top traces), mean arterial pressure (MAP) and averaged phrenic nerve discharge (PND; 0 PND units represent the central apnea, 100 units equal activity at rest, in arbitrary units). N 2 (10 s of nitrogen inhalation), phenylephrine (PE, i.v.) and sodium nitroprusside (SNP, i.v.) administered at arrow heads. A2,B2: histograms of unit discharge (thick traces) triggered by PND (thin traces, 100 sweeps; total sweep duration, 2,400 ms). A3,B3: histograms of unit discharge (thick traces) triggered on femoral arterial pulse (thin traces, 200 sweeps; sweep duration, 300 ms).
The effect of combined central and peripheral chemoreceptor stimulation ( 5 - 8 % CO 2 in 0 2 inhaled for 3 min) was also tested on 10 cells. The firing rate of 9 cells was increased from 3 . 0 + 0 . 8 to 6.4_+ 1.8 spikes/s, while that of the remaining one was inhibited from 4 to 1 spike/s. The response of RVLM-LC cells to nociceptive stimulation (toe pinch) was variable (Table I). This stimulus was applied by pinching the rear paw with a hemostat clamp. Although crude, this stimulus was used because it produces powerful and extremely reliable excitation of LC cells in the same preparation. Only 4 of 18 tested cells responded with a brisk excitation-inhibition sequence reminiscent of the pattern observed in the LC. The remainder were unaffected (n = 8), mildly excited (n -- 3) or even inhibited (n = 3). As indicated before, not a single one of the silent units with putative projection to the LC was activated by this stimulus.
A majority of the RVLM-LC neurons (15 out of 22) were inhibited when AP was raised (from a mean of 108 + 2 to 184 + 3 mmHg by i.v. injections of 10 tzg/kg phenylephrine, Fig. 4A1,B1). Mean inhibition was 70 + 5% (from 7.7 + 1.7 to 2.3 + 0.4 spikes/s, range 50100%). No effect was observed on the remaining 7 neurons although they were subjected to similar elevations of mean AP (from 112 + 4 to 186 + 5 mmHg). Nine out of 16 tested cells were partially inhibited by intravenous injections of clonidine (15 ~ g / k g , from 7.3 + 1.2 to 3.1 + 0.9 spikes/s) but 3 cells were excited (from 3.3 + 1.5 to 5.0 + 2.1 spikes/s), and the other 4 showed no significant change, although mean AP was reduced in all cases from 109 + 3 to 97 + 3 mmHg, and PND was inhibited by 60 + 6% by the drug. Four RVLM-LC units were recorded with iontophoretic electrodes. These were excited by glutamate (3-20 nA), silenced by GABA (3-10 nA), and inhibited (30-100%) by clonidine (8-40 nA).
72
A ~
~o
N~
~r
sNv~
B
8o
800
4OO "~ Z c, 0
o
0
C
ooy L/ '°°I n
~ P-"
40 I 0
25
5O
160
30.
110 ~
_ . _
•
"
-- " 3
"
~ ~o. "
TIME (rain)
Z 6
o
6O o
~/s BIN (6 ms]
so
Fig. 5. Properties of RVLM sympathetic premotor neurons. A: integrated histogram of unit discharges, mean arterial pressure (MAP) and averaged phrenic nerve discharge (PND). N 2 (10 s of nitrogen inhalation), phenylephrine (PE, i.v.) and sodium nitroprusside (SNP, i.v.) administered at arrow heads. B: PND-triggered histogram of activity of the unit (thick trace) shown in A (100 sweeps; total sweep duration, 2400 ms; thin trace = PND). C: histogram of the unit activity of the same cell (thick trace) triggered by pulses of AP (thin trace, 200 sweeps; sweep duration, 300 ms).
PND-triggered histograms of unit discharge were constructed in 26 cases to determine the type of central respiratory pattern exhibited by RVLM-LC units. The mean duration of the central respiratory cycle and of the inspiratory phase were 1306 + 37 ms and 345 + 16 ms, respectively. The vast majority of RVLM-LC cells (19 out of 26 neurons) exhibited a pronounced central respiratory rhythmicity. Nine cells showed an inspiratory (I) pattern characterized by an increased probability of discharge during phrenic activation, i.e. a peak of unit activity coincided with that of the phrenic discharge (the peak of unit activity was 160 + 48 ms after PND onset, Fig. 4B2). Eight cells had a post-inspiratory pattern (p-I) characterized by a reduced probability of discharge during inspiration, and a peak following cessation of PND (post-I peak). This peak occurred 390 + 24 ms following PND onset (Fig. 4A2). The remaining two cells displayed simple inhibition during phrenic activation (I nadir). RVLM-LC cells usually showed no cardiac rhythmicity (Fig. 4A3,B3). One cell (out of 26) exhibited a very weak cardiac rhythm. R V L M units A D activated from L C with a threshold above 30 tz A As indicated above, an additional 38 cells were AD activated from LC with higher current intensities (ranging from 30 to 200 /~A). Again, the majority of the units was silent and not activated by nociceptive stimuli. Sixteen fired spontaneously and 9 of these were subjected to a few tests. The mean firing rate of this group of 9 was 4.1 + 1.0 spikes/s, and the AD latency
to LC stimulation was 20 + 4 ms. Three cells were excited, 2 were inhibited, and 4 showed no change after N 2 inhalation. Two cells were inhibited, 1 excited, and 6 showed no change after i.v. injection of phenylephrine. Two of the 9 cells showed a burst of activation followed by a pronounced inhibition to toe pinch, 2 of them were inhibited and the remainder had no response to the nociceptive stimuli. PND-triggered histograms of the unit discharge were done in 7 out of 9 cases. Only 2 cells had respiratory synchrony (1 with a post-I pattern and 1 with inspiratory inhibition only. None of the cells tested (n = 7) had cardiac rhythmicity. O n - o f f respiratory cells and sympathetic premotor neurons of R V L M As expected z7, a large fraction of the active units recorded during the exploration of the RVLM of these 17 rats had o n - o f f respiratory activity synchronized with various phases of the phrenic cycle (n = 110). These were tested for AD activation from LC. Only 2, both of the inspiratory type, were AD activated from LC with currents of less than 25/~A (#18 and #25 in Table I). Nine reticulospinal cells were recorded in the RVLM area (mean firing rate 10.6 _+ 2.0 spikes/s, latency to stimulation of spinal cord 21 +_ 5 ms). All were completely inhibited by raising the mean AP above 170 mmHg by means of i.v. injections of phenylephrine (Fig. 5A). Eight out of 9 showed prominent cardiac rhythmicity at a mean AP ranging from 90 to 120 mmHg (Fig. 5C): 1 did not, but the mean AP could
73 have been too low for detection (80 mmHg) since baroreceptor threshold for inhibition of these cells is in the range of 80-90 mmHg 1°. These 9 cells therefore qualified as putative sympathetic premotor neurons 1°'~7. Seven were activated orthodromically by LC stimulation with a onset latency of 12 ± 1 ms, and 3 out of 9 were AD activated by LC stimulation with threshold currents ranging from 50/.~A to 1.2 mA. AD mapping was done in one case (the cell which was AD activated with a current of 1.2 mA from LC). The lowest threshold for AD activation (40/~A) was found at a site 0.9 mm below the ventral border of LC. A further 9 sympathetic premotor neurons (AD latency from T3:6.3-10.5 ms, firing rate 4-30 spikes/s) were studied in vagally intact, halothane-anesthetized rats in the context of another, unpublished, set of experiments. Only 1 cell was AD activated from LC, with a threshold of 180/zA.
DISCUSSION The AD mapping 3'18 performed in the present study argues that the axonal processes of the RVLM units AD activated from LC with low current intensity were coursing through the cell body-rich portion of this catecholaminergic nuclear group. The probability that this is correct is enhanced by the fact that AD activation required very low current intensities, especially for unmyelinated axons (lack of myelination is suggested by the consistently long AD latencies of these units). An inherent limitation of the AD mapping technique is that it does not provide proof that these axonal processes were synaptically connected with LC neurons 3,18. Yet, the sudden jumps in AD latency, which frequently occurred with very limited increments of stimulation current, suggest that axonal branching occurred in or very close to LC. Axonal branching and the presence of processes with extremely low conduction velocity both suggest the presence of terminal fields in the vicinity of the stimulating electrode 3. The range of AD latencies found in the present study is in agreement with a prior study and so is the observation that most units activated from LC at constant latency are silent 7. No spontaneous activity could be elicited in these units by any of the stimuli that we applied, and we were equally unsuccessful in our attempt to record from these cells with iontophoretic electrodes. Thus there is still no indication that these units represent RVLM cell bodies, and they could equally well be axons of passage. In any event, it is clear that these units cannot be involved in relaying nociceptive somatic information to the LC.
The responses of the tonically active RVLM-LC neurons to the various tests applied in the present study were variable, a result which is not unexpected since this projection seems to consist of at least two classes of cells, i.e. phenotypically adrenergic and nonadrenergic cells 1'7'24. Yet there was a unifying characteristic to most of these cells, namely they generally responded to chemoreceptor and baroreceptor activation and had a discharge pattern modulated by the central respiratory generator. With the exception of two of these cells, their central respiratory pattern was probabilistic, i.e. their firing pattern did not belong to the on-off type commonly associated with medullary cells involved in respiratory pattern generation 27. In this respect, RVLM-LC cells resemble the bulbospinal cells of RVLM considered to represent vasomotor sympathetic premotor neurons in the rat 1°'15-17. However, according to the present results, sympathetic premotor neurons and the cells with projection to LC represent different populations intermingled in RVLM. Indeed, while a significant fraction of RVLM sympathetic premotor neurons send rostral projections through the dorsal pons (4 out of 18 in the present case, see also ref. 17), we could not trace these collaterals to the LC itself. Moreover, bulbospinal sympathetic premotor neurons, as previously identified 1°'17, exhibit marked pulse synchrony above the baroreceptor threshold and are completely silenced by raising the mean AP in the 150-170 mmHg range. In contrast, most RVLM-LC cells were only partially inhibited by raising the mean AP to more than 180 mmHg and they did not exhibit pulse synchronicity (with one possible exception). However, as with the bulbospinal premotor neurons in RVLM 14'15, RVLM-LC cells were excited by combined central and peripheral chemoreceptor stimulation. This might underlie the involvement of the LC cells in the hypercapnic response; the LC ceils were also found to be excited during CO 2 inhalation (Guyenet, Huangfu and Koshiya, unpublished observation). It is conceivable that several of the tonically active RVLM-LC cells recorded in the present study might have been adrenergic. Indeed a catecholamine-like oxidizable material (thought to be DOPAC) is released in vivo by cell bodies in an area of the RVLM where virtually all catecholaminergic cell bodies are adrenergic 9'19. DOPAC release in RVLM is partially inhibited by clonidine (ca. 55% at saturating concentrations) ~9, suggesting that a2-adrenergic agonists inhibit the firing rate of RVLM adrenergic ceils only partially. In the present study, effects of comparable magnitude were found on many RVLM-LC ceils during clonidine administration i.v. or iontophoretically. The phrenic nerve-related rhythm of RVLM-LC
74 neurons was of central origin since the animals were vagotomized 14. The two major patterns, inspiratory and post-inspiratory, were similar to the predominant central respiratory rhythm of RVLM sympathetic premotor neurons tS. The p-I pattern of RVLM-LC cells closely corresponded to pattern III of RVLM sympathetic premotor neurons and the I pattern of RVLMLC ceils was reminiscent of pattern II of the premotor neurons. These two patterns were also found in LC cells (Guyenet, Huangfu and Koshiya, unpublished observations). This rhythmicity is in all cases considerably enhanced by stimulation of peripheral and central chemoreceptors (Guyenet, Huangfu and Koshiya, unpublished observations). The similarity between the respiratory pattern of LC cells and that of RVLM-LC cells supports the notion that the information conveyed by the RVLM input to the noradrenergic cells of LC is of mixed cardio-respiratory nature. There were no consistent or dramatic differences between the properties of the RVLM cells AD activated with a very low threshold (30 /xA) and those activated with higher current intensities (30-200 /~A). This was somewhat expected since (i) the cut-off (30/xA) was arbitrarily selected and (ii) the physiological tests were few and would be expected to influence a large number of reticular formation cells. Two differences appeared, however. First, a smaller proportion of cells was affected by raising AP, and a small number of putative sympathetic premotor neurons could be backfired from LC with these higher currents. Finally, very few RVLM-LC cells were sensitive to nociceptive inputs, especially if one includes the large contingent of silent unresponsive cells. Since a few were found to be activated, the present results do not exclude the possibility that nociceptive inputs to LC might be transmitted via RVLM. However, the small number of responsive cells would tend to support Rasmussen and Aghajanian's contention that RVLM is not or is minimally involved in relaying nociceptive inputs to L C 26. Acknowledgments. The authors gratefully thank Ms. Tina Riley for technical assistance. This work was supported by grants from the National Institutes of Health to P.G.G. (HL 28785 and HL 39841).
REFERENCES 1 Aston-Jones, G., Astier, B. and Ennis, M., Inhibition of nora-
drenergic locus coeruleus neurons by C1 adrenergic cells in the rostral ventral medulla, J. Neurosci., 273 (1991). 2 Aston-Jones, G., Ennis, M., Pieribone, V.A., Nickell, W.T. and Shipley, M.T., The brain nucleus locus coeruleus: restricted afferent control of a broad efferent network, Science, 234 (1986) 734-737. 3 Barman, S.M. and Gebber, G.L., Lateral tegmental field neurons
of cat medulla: a source of basal activity of ventrolateral medulIospinal sympathoexcitatory neurons, J. NeurophysioL, 57 (1987) 1410-1424. 4 Biesold, D., Kurosawa, M., Sato, A. and Trzebski, A., Hypoxia and hypercapnia increase the sympathoadrenal medullary functions in anesthetized, artificially ventilated rats, Jap. J. Physiol., 39 (1989) 511-522. 5 Elam, M., Yao, T., Svensson, T.H. and Thoren, P., Regulation of locus coeruleus neurons and splanchnic sympathetic nerves by cardiovascular afferents, Brain Res., 290 (1984) 281-288. 6 Elam, M., Yao, T., Thoren, P. and Svensson, T.H., Hypercapnia and hypoxia: chemoreceptor-mediated control of locus coeruleus neurons and splanchnic sympathetic nerve, Brain Res., 222 (1981) 373-381. 7 Ennis, M. and Aston-Jones, G., Two physiologically distinct populations of neurons in the ventrolateral medulla innervate the locus coeruleus, Brain Res., 425 (1987) 275-282. 8 Ennis, M. and Aston-Jones, G., Activation of locus coeruleus from nucleus paragigantocellularis: a new excitatory amino acid pathway in brain, Z Neurosci., 8 (1988) 3644-3657. 9 Gillon, J.Y., Richard, F., Quintin, L., Pujol, J.F. and Renaud, B., Pharmacological and functional evidence for extracellular 3,4-dihydroxyphenylacetic acid as an index of metabolic activity of the adrenergic neurons: an in vivo voltammetry study in the rat rostrol ventrolateral medulla, Neuroscience, 37 (1990) 421-430. 10 Guyenet, P.G., Role of the ventral medulla oblongata in blood pressure regulation. In A.D. Loewy and K.M. Spyer (Eds.), Central Regulation of Autonomic Functions, Oxford University Press, New York, 1990, pp. 145-167. 11 Guyenet, P.G., Respiratory-sympathetic integration in the medulla oblongata. In G. Kunos and J. Ciriello (Eds.), Central Neural Mechanisms in Cardiocascular Regulation, Vol. H, Birkhauser, Boston, 1992 (in press). 12 Guyenet, P.G. and Aghajanian, G.K., ACh, substance P and met-enkephalin in the locus coeruleus: pharmacological evidence for independent sites of action, Eur. J. Pharmacol., 53 (1979) 319-328. 13 Guyenet, P.G. and Young, B.S., Projections of nucleus paragigantocellularis lateralis to locus coeruleus and other structures in rat, Brain Res., 406 (1987) 171-184. 14 Guyenet, P.G., Darnall, R.A. and Riley, T.A., Rostral ventrolateral medulla and sympathorespiratory integration in rats, Am. J. PhysioL, 259 (1990) R1063-R1074. 15 Haselton, J.R. and Guyenet, P.G., Central respiratory modulation of medullary sympathoexcitatory neurons in rat, Am. J. Physiol., 256 (1989) R739-R750. 16 Haselton, J.R. and Guyenet, P.G., Electrophysiological characterization of putative C1 adrenergic neurons in the rat, Neuroscience, 30 (1989) 199-214. 17 Haselton, J.R. and Guyenet, P.G., Ascending collaterals of medullary barosensitive neurons and C1 cells in rats, Am. J. PhysioL, 258 (1990) R1051-R1063. 18 Jankowska, E. and Roberts, W.J., An electrophysiological demonstration of the axonal projections of single spinal interneurones in the cat, J. Physiol., 222 (1972) 597-622. 19 Mermet, C. and Quintin, L., Effect of clonidine on catechol metabolism in the rostral ventrolateral medulla: an in vivo electrochemical study, Eur. J. Pharmacol., 204 (1991 ) 105-107. 20 Milner, T.A., Abate, C., Reis, D.J. and Pickel, V.M., Ultrastructural localization of phenylethanolamine N-methyltransferase-like immunoreactivity in the rat locus coeruleus, Brain Res., 478 (1989) 1-15. 21 Morilak, D.A., Fornal, C. and Jacobs, B.L., Single unit activity of noradrenergic neurons in locus coeruleus and serotonergic neurons in the nucleus raphe dorsalis of freely moving cats in relation to the cardiac cycle, Brain Res., 399 (1986) 262-270. 22 Paxinos, G. and C. Watson. The Rat Brain in Stereotaxic Coordinates', 2nd edn., Academic Press, Sydney, 1986. 23 Pieribone, V.A. and Aston-Jones, G., Adrenergic innervation of the rat nucleus locus coeruleus arises predominantly from the C1 adrenergic cell group in the rostral medulla, Neuroscience, 41 (199l) 525-542.
75 24 Pieribone, V.A., Aston-Jones, G. and Bohn, M.C., Adrenergic and non-adrenergic neurons in the C1 and C3 areas project to locus coeruleus: a fluorescent double labeling study, Neurosci. Lett., 85 (1988) 297-303. 25 Rasmussen, K. and Aghajanian, G,K., Withdrawal-induced activation of locus coeruleus neurons in opiate-dependent rats: attenuation by lesions of the nucleus paragigantocellularis, Brain Res., 505 (1989) 346-350.
26 Rasmussen, K. and Aghajanian, G.K., Failure to block responses of locus coeruleus neurons to somatosensory stimuli by destruction of two major afferent nuclei, Synapse, 4 (1989) 162-164. 27 Schwarzacher, S.W., Wilhelm, Z., Anders, K. and Richter, D.W., The medullary respiratory network in the rat, J. Physiol., 435 (1991) 631-644.
Brain Research, 598 (1992) 67-75 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 18321
Rostral ventrolateral medullary neurons projecting to locus coeruleus have cardiorespiratory inputs Donghai Huangfu, Anthony J.M. Verberne
* a n d P a t r i c e G. G u y e n e t
University of Virginia School of Medicine, Department of Pharmacology, Charlottesville, VA 22908 (USA) (Accepted 7 July 1992)
Key words: Rostral ventrolateral medulla; Locus coeruleus; Respiratory and cardiovascular control
This study was designed to characterize some of the properties of the rostral ventrolateral medullary (RVLM) cells with axonal projection to the locus coeruleus (LC) in urethane anesthetized, vagotomized, paralyzed and artificially respirated rats. The vast majority of RVLM units antidromically (AD) activated from LC (RVLM-LC units) were silent and unresponsive to peripheral chemoreceptor stimulation or nociceptive stimulation. Twenty seven spontaneously active RVLM-LC neurons, AD activated from LC with currents below 30 p.A (17+2 tzA) were analyzed. AD mapping (n = 18) indicated that the lowest threshold for AD activation occurred within the LC itself. Axonal branching within or close to LC was suggested by the presence of sudden jumps in AD latency. Maximal AD latencies ranged from 7 to 37 ms. Most spontaneously active RVLM-LC neurons displayed marked central respiratory modulation characterized by either a post-inspiratory or an inspiratory pattern. The majority of the tested neurons were affected (excited or inhibited) by brief peripheral chemoreceptor stimulation (N 2 inhalation). Most cells were inhibited by raising arterial pressure but none exhibited any detectable pulse synchrony. Reticulospinal sympathetic premotor neurons of RVLM were not found to project to LC (sample of 9) and very few RVLM cells with on-off respiratory discharges appeared to project to LC (2 out of 110). This study suggests that much of the information conveyed by the RVLM to LC could be of a mixed cardiorespiratory nature.
INTRODUCTION The rostral ventrolateral medulla (RVLM) is a major source of afferent inputs to the somata and proximal dendrites of the noradrenergic cells of the locus coeruleus ( L C ) 1'2'7'13. This pathway includes phenotypically adrenergic cells of the C1 group e°'23. The physiological role of the RVLM-LC projection is largely unknown. It may somehow be involved in the excitation of LC cells during opiate withdrawal e5 and some evidence suggests that RVLM might be a relay in the poly-synaptic pathway responsible for LC activation by nociceptive somatic stimulation 2,s. The latter notion has been challenged by Rasmussen and Aghajanian's observation that massive high frequency bilateral thermal lesions of RVLM fail to alter the response of LC neurons to nociceptive stimuli 26. The prominent role of RVLM in vasomotor and respiratory control e7 (for reviews, see refs. 10,11) suggests an alternative role of the RVLM-LC projection: namely, it might convey information of a cardio-re-
spiratory (or enteroceptive) nature. This view derives some support from prior evidence that LC unit activity is influenced by arterial pressure, hypoxia and blood volume changes 5,6,21. In addition the LC also displays a marked central respiratory rhythmicity during hypoxia or CO 2 inhalation (Guyenet, Huangfu and Koshiya, unpublished observations). The purpose of the present study was to test whether the LC might receive cardio-respiratory information directly from its afferent inputs from RVLM. This goal was pursued by analyzing the properties of RVLM cells antidromically (AD) activated from LC with very low current thresholds. MATERIALS AND METHODS General procedures Experiments were carried out in 17 male Sprague-Dawley rats weighing from 320 to 380 g. The rats were initially anesthetized with ether. After tracheostomy, anesthesia was maintained by artificial ventilation with 1.3-1.5% halothane in 100% 0 2 for the remainder of the surgery. The right femoral artery and a femoral or jugular vein
Correspondence: P.G. Guyenet, Department of Pharmacology, University of Virginia, Charlottesville, VA 22908, USA. Fax: (1) (804) 982-3878. * Present address: Austin Hospital, Heidelberg 3084, Victoria, Australia.
68 B
A
v
O
•
I v
I I
J
Fig. 1. Antidromic (AD) identification of RVLM-LC neurons. A: an example of a collision test (4 superimposed sweeps per trace) used to identify a RVLM cell with projection to LC. Spontaneous spikes (©), AD spikes (*), stimuli delivered at arrow. Calibration bar, 100 ~V, 5 ms. B: example of sudden jumps in AD latency with increased stimulation currents (4 sweeps per trace; stimulation intensity from top to bottom: 30, 100 and 200 ~A, 0.5 ms duration). Calibration bar, 100/xA, 2 ms.
were cannulated for measurement of arterial pressure (AP) and injection of drugs. Bilateral vagotomy was performed to desynchronize the central respiratory rhythm from ventilation ~4. Body temperature was maintained at 37-39°C. The facial nerve (mandibular branch) was exposed for AD stimulation to help locate RVLM which lies just caudal to the facial motor nucleus 15-17. The left phrenic nerve was isolated, cut distally, placed over a bipolar silver electrode and immersed in Silgel (Wacker compound 907) for multi-unit recording t4. After completion of surgery, halothane was withdrawn, and the anesthesia was replaced by intravenous administration of urethane in the dose of 1 g / k g (the anesthetic was changed because halothane depresses the chemo-reflex). After half an hour stabilization and verification of the adequacy of anesthesia (lack of corneal reflex and absence of retraction of distal phalanges to strong pinch of the hindpaw), the rat was paralyzed with pancuronium bromide (1 m g / k g initial dose followed by maintenance dose of 0.3 m g / k g / h infused from the femoral vein with a syringe pump). Peripheral chemoreceptor stimulation was performed by respirating the rats with 100% N 2 for 10 s. This paradigm produces increases in AP, phrenic activity and a generalized increase in sympathetic
outflow 4 (Guyenet, Huangfu and Koshiya, unpublished observations). In the present preparation, the cardio-respiratory effects of brief N~ inhalation were completely eliminated by selective bilateral section of the carotid sinus nerve; therefore, in confirmation of the results of Biesold et al. 4, these effects were entirely due to carotid chemoreceptor stimulation (Guyenet, Huangfu and Koshiya, unpublished observations). Baroreceptor stimulation and unloading were done by intravenous administration of a bolus of phenylephrine (4 /.tg in 20 p.I) and sodium nitroprusside (10/xg in 20 ~zl)TM respectively. Nociceptive stimulation was done by squeezing the toe of the hindlimb with a pair of forceps. Combined central and peripheral chemoreceptor stimulation was done by adding 5 - 8 % CO 2 to the breathing mixture (balance 02).
Electrical stimulations and recordings The rat was placed in a stereotaxic apparatus with the incisor bar 3.5 mm below the interaural line and suspended by a clamp holding the spinal processes at the Ts_ 6 level. A parieto-occipital craniotomy and a laminectomy at the T 2 level were performed. A fine metal electrode with impedance from 2 to 5 M O was inserted into LC transcerebrally with the manipulator angled 35 ° rostrally for recording or stimulating. The approximate coordinates were 3 mm anterior to lambda, 1.1 mm lateral to midline, 7.5-9.0 mm below the brain surface. Extracellular recordings from the electrode during implantation were used for accurate placement in the LC. The nucleus was identified by its characteristic spontaneous and nociceptive evoked firing patterns and its location medial to the mesencephalic nucleus of the trigeminal nerve, as previously described m2. A second monopolar stimulating electrode was inserted into the spinal cord at the T 2 level with its tip 0.9 mm below the dorsolateral sulcus for AD activation of RVLM bulbospinal cells 15'16. RVLM neurons were recorded with glass electrodes filled with 2% Pontamine sky blue in 0.5 M sodium acetate (impedance 5-10 Mg2). The medullary area explored was between 1.5 and 2.2 mm lateral to the midline, within 1 mm of the ventral surface, and from 0 to 1.5 mm caudal to the facial motor nucleus, initially mapped by AD field potentials evoked by stimulation of the mandibular branch of the facial nerve 15-iv. For AD activation of RVLM neurons, LC was routinely stimulated with 200 /zA of 0.5 ms pulses at 1 Hz. Antidromic identification used the following criteria3J4: (1) constant latency of evoked spikes, (2) ability to follow paired pulse stimulation
~ -0.68
!'~
'
~ '~-
-
~ ~y
1
.
6
~-~
Fig. 2. Anatomy of recording and stimulation sites. Anatomical location of stimulation sites (locus coeruleus on the left) and RVLM unit recordings (on the right). Coronal sections redrawn after ref. 22. Numbers next to pontine sections (left side) refer to Paxinos's nomenclature of coronal planes 22 (ram interaural). Numbers next to medullary sections (right side) refer to mm rostral to obex. 7n, facial nerve; A, nucleus ambiguus; IO, inferior olive; LC, locus coeruleus; LPB, lateral parabrachial nucleus; Me5, mesencephalic trigeminal nucleus; Mo5, trigeminal motor nucleus; MPB, medial parabrachial nucleus; nTS, nucleus of solitary tract; Pr5, principal trigeminal sensory nucleus; PY, pyramidal tract; s5, sensory root of trigeminal nerve; SO, superior olive; sp5, spinal trigeminal tract; Sp5i, interpolar trigeminal nucleus; Sp50, oral spinal trieeminal nucleus.
69 at frequencies of 200 Hz or greater, and (3) collision of evoked spikes with spontaneous impulses (Fig. 1A). Neuronal responses to stimulation of the spinal cord was also performed in an attempt to identify spinal projections ( < 1 mA, 0.5 ms pulses). In a few cases, five-barrel iontophoretic electrodes were used, filled with the following drugs: L-glutamate (Aldrich, 0.2 M in 160 m M NaCI, p H 8), y-aminobutyric acid (GABA, Aldrich, 0.2 M in 160 m M NaCl, pH 6), and clonidine (Sigma, 0.1 M in 160 m M NaC1, pH 5). Current balancing was made through a channel filled with 2 M NaCl. Retaining currents of 5 nA were used between periods of drug ejection. The recording channel contained 2% Pontamine sky blue in 2 M NaCl. Arterial blood pressure (AP), heart rate, intratracheal pressure (ITP), phrenic nerve discharge (PND, band pass of 10-3,000 Hz, half-wave rectified and integrated), integrated rate histogram of neuron activity and percentage of oxygen in inspired air (FiO 2) were displayed on a chart recorder. AP, ITP, P N D and neuronal activity were also recorded on tape for later analysis. PND-triggered histograms of R V L M unit discharges (100 sweeps, 4 ms bins, 2,400 ms sweep duration), and A P pulse-triggered histograms (200-500 sweeps, 0.5 ms bins, 300 ms sweep duration) were obtained as described before 14d5.
RESULTS
RVLM cells activated from LC with a threshold of less than 30 IzA In 17 rats, 554 cells were encountered in 224 micropipette penetrations through the RVLM, and 150 units were putatively AD activated from the region of the LC, judging from the presence of evoked spikes with invariant latency and a 200 Hz following frequency. This section focuses on the cells with a threshold for AD activation of < 30 /.tA, based on the notion that the lower the threshold the more likely the cell was to project t o / t h r o u g h LC. This group numbered 112 units (mean latency 11 + 1 ms, range 6-46 ms, mean threshold 19 + 1 /xA), the vast majority of which were silent (n = 81). These silent units were activated neither by nociceptive stimuli (toe pinch) nor by peripheral chemoreceptor stimulation, and were not studied further. The rest of the RVLM-LC neurons were spontaneously active and all met the requirements of the collision test (Fig. 1A). Twenty-seven were recorded for a period sufficient to run a series of experimental tests. The remainder of this section focuses on the
Histology R V L M recording sites (1 per animal) were marked by ejection of Pontamine sky blue ( 1 0 / x A cathodal current for 5 min). Stimulation sites (LC only) were marked by iron deposits (10 ~ A anodal current for 10 s) revealed by the Prussian blue reaction. Recording and stimulation sites were identified in Nissl-stained coronal sections and mapped on standardized sections redrawn with reference to the atlas of Paxinos and Watson 22. All data reported in this paper are from rats in which LC stimulation sites were histologically verified and recording sites were confirmed to be in R V L M (Fig. 2). Results are presented as means_+ S.E.M. TABLE I
Properties o f R V L M neurons projecting to L C c / s , spikes per second; lat, latency; thre, threshold; OD, orthodromic (with reference to activation from SC, spinal cord); glu, glutamate; clo, clonidine; ion, iontophoretic application; / , cell not tested; - , inhibition; + , excitation; 0, no change; + - , excitation followed by inhibition.
No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Firing rate (c / s)
A D response to L C st lat (ms) thre (~ A)
OD ~ t to SC st (ms)
Responses to N2
5% CO 2
Toe pinch
PE L c.
glu ion
GABA ion
clo ion
clo i.v.
10 4 4 2 3 1.2 1.2 3 2 1.5 6 4.5 5 1 12 5 4 10 4 22 10 10 16 3 10 0.4 0.2
19 19 15 18 16 8 16 18 37 22 8 13 9 22 8 18 16 14 8 8 18 19 28 13 7 15 12
15 30 0 0 0 10 0 0 16 0 / / 0 / 0 0 0 18 12 12 25 12 0 0 0 0 /
0 0 + / + / + / + + + + + 0 + + + + 0
+ + + / / + / / / / + / + / / / / / / + + + / / / /
/ / / / / 0 + / + / / + 0 0 0 0 + ++ 0 0 0 / +
0 0 / 0 / / 0 0 0 / 0 /
+ + / / / / / / / / / / / / / / / / / / / / / / + + /
/ / / / / / / / / / / / / / / / / / / / / / /
/ / / / / / / / / / / / / / / / / / / / / / /
+ / + / / / + / / / 0 0 / 0 / / 0 /
20 15 10 15 15 6 3 20 25 10 15 10 25 28 20 20 15 15 20 20 8 20 25 20 25 10 20
Respiratory rhythm
Cardiac rhythm
post-I peak I peak 0 I peak I peak I nadir post-I peak I nadir I peak 0 post-I peak I peak post-I peak 0 post-I peak 0 0 I peak post-I peak 0 I peak 0 I peak post-I peak I peak post-I peak /
0 0 0 0 0 0 0 0 0 0 0 0 0 0 + 0 0 0 0 0 0 0 0 0 0 0 /
70 properties of these 27 spontaneously firing cells recorded in 11 animals (summarized in the Table I). Fig. 2 shows the distribution of the histologically verified stimulation sites in the LC and RVLM recording sites (these sites were either directly identified from Pontamine sky blue deposits, n = 11, or reconstructed from the coordinates of the cells in relation to those of another cell that was directly identified by the dye). The firing rate of the 27 R V L M - L C neurons varied from 0.2 to 22 spikes/s with a mean of 5.7_+ 1.1 spikes/s. A D latencies of these neurons to LC stimulation ranged from 7 to 37 ms (mean 16 + 1 ms, P < 0.05 compared with that of silent group, 10 _+ 1 ms, n = 81). This latency corresponds to an average conduction velocity of 0.34 _+ 0.02 m / s (range 0.15-0.79 m / s ) based on the mean straight line distance between stimulation and recording sites (estimated at 5 mm). The refractory period of these cells ranged from 0.8 to 4.0 ms. The threshold for A D activation ranged from 3 to 28 ~ A (17 _+ 1 ~A, current measured by ohmic voltage drop across a 1 kS2 resistor). A D mapping was performed in 18 cases by stepping the stimulation electrode through and on each side of the LC. The lowest threshold for A D activation was found at a mean depth of 8.3 _+ 0.2 mm below the brain surface and marked. All the marked spots were found within the LC. Fig. 3 shows an example of a series of 3 depth-threshold curves designed to characterize the projection of neuron 21 of Table I. Note that the lowest threshold (8 p.A) was found in the middle tract, while higher thresholds and
narrower curves were found in the two lateral tracts only 0.2 mm distant. This particular case also illustrates the occurrence of sudden shifts in antidromic latency as the stimulation current was gradually raised a n d / o r the electrode was moved through the LC. This phenomenon, suggesting terminal branching 3, is also illustrated in Fig. lB. These sudden shifts generally ranged from 1 to 5 ms, but could be as large as 14 ms in an occasional case. Depth-threshold curves were also constructed in the case of 6 silent cells. The sites with the lowest threshold for AD activation were also within LC. Nine out of 23 tested cells were excited orthodromically by stimulation of spinal cord with a mean onset latency of 17 + 2 ms. The effect of peripheral chemoreceptor activation on the activity of 24 RVLM-LC neurons was tested by respirating the animals with 100% N 2 for 10 s. Twelve cells were excited from a control firing rate of 7.4 + 2.2 to 16.0 + 3.5 spikes/s (Fig. 4A1); 8 cells were inhibited from 4.2 + 1.1 to 1.3 +_ 0.8 spikes/s (Fig. 4B1). The remaining 4 cells showed no detectable response. We include in this category one unit which was slightly inhibited by chemoreceptor activation but probably as a result of a secondary baroreceptor-mediated effect of the hypoxia-induced response. Indeed this inhibition did not occur when the peripheral chemoreceptor stimulation was combined with the i.v. injection of a vasorelaxant (sodium nitroprusside) to prevent the hypoxiainduced hypertension.
7.5 o 18 ms O 16ms A 15 ms • 13ms
LI.I
LI.I --8.0
.-8.s
,~ ,ca
- -9.0
-9.5
1 mm
-I0.0 O
I~
Threshold
2~
3~
(kLA)
Fig. 3. Antidromic mapping of the projection of a RVLM neuron to LC. Left panel: scheme of 3 parallel penetrations through LC (for abbreviations see legend to Fig. 2). Characters above the 3 tracks refer to mm lateral to the midline. Right panel: depth threshold curves corresponding to tracks shown in the left panel. The lowest threshold (8 ~A) was identified by a small iron deposit revealed by the Prussian blue reaction (star in the left panel). Symbols inset into the right panel indicate AD latencies of the RVLM neurons which are different along the stimulation tracks in LC.
71 AI 30
N~
PE
V
V
BI
N2 V
30
PE V
SNP N2 V V
10 0
,
25O 100
~-~
,
•
,
_
.
,
"~
2013.
~
~'~
,
~-~
loo.
~
50.
~:
-
5
o
.
0
~
3
6
i 150. I 0 0 . 1 ~ ~ ~ 4 ~ ~ 50 0
TIME (mini A2
100
3 T I M E (rain)
6
B2
600
ioo
8oo
50 z 0
0 0
Z
25
50
0
Z
25
BIN (48 m s )
A3
B3 160
20
0
10,
Z
25
5o
BIN (6 ms)
50
BIN (48 m s ) 30
180
~
130
~ z
0
~
80 0
25
50
Bin (6 ms)
Fig. 4. Properties of RVLM-LC neurons. A and B represent two different cells. A1,BI: integrated rate histogram of cell discharges (top traces), mean arterial pressure (MAP) and averaged phrenic nerve discharge (PND; 0 PND units represent the central apnea, 100 units equal activity at rest, in arbitrary units). N 2 (10 s of nitrogen inhalation), phenylephrine (PE, i.v.) and sodium nitroprusside (SNP, i.v.) administered at arrow heads. A2,B2: histograms of unit discharge (thick traces) triggered by PND (thin traces, 100 sweeps; total sweep duration, 2,400 ms). A3,B3: histograms of unit discharge (thick traces) triggered on femoral arterial pulse (thin traces, 200 sweeps; sweep duration, 300 ms).
The effect of combined central and peripheral chemoreceptor stimulation ( 5 - 8 % CO 2 in 0 2 inhaled for 3 min) was also tested on 10 cells. The firing rate of 9 cells was increased from 3 . 0 + 0 . 8 to 6.4_+ 1.8 spikes/s, while that of the remaining one was inhibited from 4 to 1 spike/s. The response of RVLM-LC cells to nociceptive stimulation (toe pinch) was variable (Table I). This stimulus was applied by pinching the rear paw with a hemostat clamp. Although crude, this stimulus was used because it produces powerful and extremely reliable excitation of LC cells in the same preparation. Only 4 of 18 tested cells responded with a brisk excitation-inhibition sequence reminiscent of the pattern observed in the LC. The remainder were unaffected (n = 8), mildly excited (n -- 3) or even inhibited (n = 3). As indicated before, not a single one of the silent units with putative projection to the LC was activated by this stimulus.
A majority of the RVLM-LC neurons (15 out of 22) were inhibited when AP was raised (from a mean of 108 + 2 to 184 + 3 mmHg by i.v. injections of 10 tzg/kg phenylephrine, Fig. 4A1,B1). Mean inhibition was 70 + 5% (from 7.7 + 1.7 to 2.3 + 0.4 spikes/s, range 50100%). No effect was observed on the remaining 7 neurons although they were subjected to similar elevations of mean AP (from 112 + 4 to 186 + 5 mmHg). Nine out of 16 tested cells were partially inhibited by intravenous injections of clonidine (15 ~ g / k g , from 7.3 + 1.2 to 3.1 + 0.9 spikes/s) but 3 cells were excited (from 3.3 + 1.5 to 5.0 + 2.1 spikes/s), and the other 4 showed no significant change, although mean AP was reduced in all cases from 109 + 3 to 97 + 3 mmHg, and PND was inhibited by 60 + 6% by the drug. Four RVLM-LC units were recorded with iontophoretic electrodes. These were excited by glutamate (3-20 nA), silenced by GABA (3-10 nA), and inhibited (30-100%) by clonidine (8-40 nA).
72
A ~
~o
N~
~r
sNv~
B
8o
800
4OO "~ Z c, 0
o
0
C
ooy L/ '°°I n
~ P-"
40 I 0
25
5O
160
30.
110 ~
_ . _
•
"
-- " 3
"
~ ~o. "
TIME (rain)
Z 6
o
6O o
~/s BIN (6 ms]
so
Fig. 5. Properties of RVLM sympathetic premotor neurons. A: integrated histogram of unit discharges, mean arterial pressure (MAP) and averaged phrenic nerve discharge (PND). N 2 (10 s of nitrogen inhalation), phenylephrine (PE, i.v.) and sodium nitroprusside (SNP, i.v.) administered at arrow heads. B: PND-triggered histogram of activity of the unit (thick trace) shown in A (100 sweeps; total sweep duration, 2400 ms; thin trace = PND). C: histogram of the unit activity of the same cell (thick trace) triggered by pulses of AP (thin trace, 200 sweeps; sweep duration, 300 ms).
PND-triggered histograms of unit discharge were constructed in 26 cases to determine the type of central respiratory pattern exhibited by RVLM-LC units. The mean duration of the central respiratory cycle and of the inspiratory phase were 1306 + 37 ms and 345 + 16 ms, respectively. The vast majority of RVLM-LC cells (19 out of 26 neurons) exhibited a pronounced central respiratory rhythmicity. Nine cells showed an inspiratory (I) pattern characterized by an increased probability of discharge during phrenic activation, i.e. a peak of unit activity coincided with that of the phrenic discharge (the peak of unit activity was 160 + 48 ms after PND onset, Fig. 4B2). Eight cells had a post-inspiratory pattern (p-I) characterized by a reduced probability of discharge during inspiration, and a peak following cessation of PND (post-I peak). This peak occurred 390 + 24 ms following PND onset (Fig. 4A2). The remaining two cells displayed simple inhibition during phrenic activation (I nadir). RVLM-LC cells usually showed no cardiac rhythmicity (Fig. 4A3,B3). One cell (out of 26) exhibited a very weak cardiac rhythm. R V L M units A D activated from L C with a threshold above 30 tz A As indicated above, an additional 38 cells were AD activated from LC with higher current intensities (ranging from 30 to 200 /~A). Again, the majority of the units was silent and not activated by nociceptive stimuli. Sixteen fired spontaneously and 9 of these were subjected to a few tests. The mean firing rate of this group of 9 was 4.1 + 1.0 spikes/s, and the AD latency
to LC stimulation was 20 + 4 ms. Three cells were excited, 2 were inhibited, and 4 showed no change after N 2 inhalation. Two cells were inhibited, 1 excited, and 6 showed no change after i.v. injection of phenylephrine. Two of the 9 cells showed a burst of activation followed by a pronounced inhibition to toe pinch, 2 of them were inhibited and the remainder had no response to the nociceptive stimuli. PND-triggered histograms of the unit discharge were done in 7 out of 9 cases. Only 2 cells had respiratory synchrony (1 with a post-I pattern and 1 with inspiratory inhibition only. None of the cells tested (n = 7) had cardiac rhythmicity. O n - o f f respiratory cells and sympathetic premotor neurons of R V L M As expected z7, a large fraction of the active units recorded during the exploration of the RVLM of these 17 rats had o n - o f f respiratory activity synchronized with various phases of the phrenic cycle (n = 110). These were tested for AD activation from LC. Only 2, both of the inspiratory type, were AD activated from LC with currents of less than 25/~A (#18 and #25 in Table I). Nine reticulospinal cells were recorded in the RVLM area (mean firing rate 10.6 _+ 2.0 spikes/s, latency to stimulation of spinal cord 21 +_ 5 ms). All were completely inhibited by raising the mean AP above 170 mmHg by means of i.v. injections of phenylephrine (Fig. 5A). Eight out of 9 showed prominent cardiac rhythmicity at a mean AP ranging from 90 to 120 mmHg (Fig. 5C): 1 did not, but the mean AP could
73 have been too low for detection (80 mmHg) since baroreceptor threshold for inhibition of these cells is in the range of 80-90 mmHg 1°. These 9 cells therefore qualified as putative sympathetic premotor neurons 1°'~7. Seven were activated orthodromically by LC stimulation with a onset latency of 12 ± 1 ms, and 3 out of 9 were AD activated by LC stimulation with threshold currents ranging from 50/.~A to 1.2 mA. AD mapping was done in one case (the cell which was AD activated with a current of 1.2 mA from LC). The lowest threshold for AD activation (40/~A) was found at a site 0.9 mm below the ventral border of LC. A further 9 sympathetic premotor neurons (AD latency from T3:6.3-10.5 ms, firing rate 4-30 spikes/s) were studied in vagally intact, halothane-anesthetized rats in the context of another, unpublished, set of experiments. Only 1 cell was AD activated from LC, with a threshold of 180/zA.
DISCUSSION The AD mapping 3'18 performed in the present study argues that the axonal processes of the RVLM units AD activated from LC with low current intensity were coursing through the cell body-rich portion of this catecholaminergic nuclear group. The probability that this is correct is enhanced by the fact that AD activation required very low current intensities, especially for unmyelinated axons (lack of myelination is suggested by the consistently long AD latencies of these units). An inherent limitation of the AD mapping technique is that it does not provide proof that these axonal processes were synaptically connected with LC neurons 3,18. Yet, the sudden jumps in AD latency, which frequently occurred with very limited increments of stimulation current, suggest that axonal branching occurred in or very close to LC. Axonal branching and the presence of processes with extremely low conduction velocity both suggest the presence of terminal fields in the vicinity of the stimulating electrode 3. The range of AD latencies found in the present study is in agreement with a prior study and so is the observation that most units activated from LC at constant latency are silent 7. No spontaneous activity could be elicited in these units by any of the stimuli that we applied, and we were equally unsuccessful in our attempt to record from these cells with iontophoretic electrodes. Thus there is still no indication that these units represent RVLM cell bodies, and they could equally well be axons of passage. In any event, it is clear that these units cannot be involved in relaying nociceptive somatic information to the LC.
The responses of the tonically active RVLM-LC neurons to the various tests applied in the present study were variable, a result which is not unexpected since this projection seems to consist of at least two classes of cells, i.e. phenotypically adrenergic and nonadrenergic cells 1'7'24. Yet there was a unifying characteristic to most of these cells, namely they generally responded to chemoreceptor and baroreceptor activation and had a discharge pattern modulated by the central respiratory generator. With the exception of two of these cells, their central respiratory pattern was probabilistic, i.e. their firing pattern did not belong to the on-off type commonly associated with medullary cells involved in respiratory pattern generation 27. In this respect, RVLM-LC cells resemble the bulbospinal cells of RVLM considered to represent vasomotor sympathetic premotor neurons in the rat 1°'15-17. However, according to the present results, sympathetic premotor neurons and the cells with projection to LC represent different populations intermingled in RVLM. Indeed, while a significant fraction of RVLM sympathetic premotor neurons send rostral projections through the dorsal pons (4 out of 18 in the present case, see also ref. 17), we could not trace these collaterals to the LC itself. Moreover, bulbospinal sympathetic premotor neurons, as previously identified 1°'17, exhibit marked pulse synchrony above the baroreceptor threshold and are completely silenced by raising the mean AP in the 150-170 mmHg range. In contrast, most RVLM-LC cells were only partially inhibited by raising the mean AP to more than 180 mmHg and they did not exhibit pulse synchronicity (with one possible exception). However, as with the bulbospinal premotor neurons in RVLM 14'15, RVLM-LC cells were excited by combined central and peripheral chemoreceptor stimulation. This might underlie the involvement of the LC cells in the hypercapnic response; the LC ceils were also found to be excited during CO 2 inhalation (Guyenet, Huangfu and Koshiya, unpublished observation). It is conceivable that several of the tonically active RVLM-LC cells recorded in the present study might have been adrenergic. Indeed a catecholamine-like oxidizable material (thought to be DOPAC) is released in vivo by cell bodies in an area of the RVLM where virtually all catecholaminergic cell bodies are adrenergic 9'19. DOPAC release in RVLM is partially inhibited by clonidine (ca. 55% at saturating concentrations) ~9, suggesting that a2-adrenergic agonists inhibit the firing rate of RVLM adrenergic ceils only partially. In the present study, effects of comparable magnitude were found on many RVLM-LC ceils during clonidine administration i.v. or iontophoretically. The phrenic nerve-related rhythm of RVLM-LC
74 neurons was of central origin since the animals were vagotomized 14. The two major patterns, inspiratory and post-inspiratory, were similar to the predominant central respiratory rhythm of RVLM sympathetic premotor neurons tS. The p-I pattern of RVLM-LC cells closely corresponded to pattern III of RVLM sympathetic premotor neurons and the I pattern of RVLMLC ceils was reminiscent of pattern II of the premotor neurons. These two patterns were also found in LC cells (Guyenet, Huangfu and Koshiya, unpublished observations). This rhythmicity is in all cases considerably enhanced by stimulation of peripheral and central chemoreceptors (Guyenet, Huangfu and Koshiya, unpublished observations). The similarity between the respiratory pattern of LC cells and that of RVLM-LC cells supports the notion that the information conveyed by the RVLM input to the noradrenergic cells of LC is of mixed cardio-respiratory nature. There were no consistent or dramatic differences between the properties of the RVLM cells AD activated with a very low threshold (30 /xA) and those activated with higher current intensities (30-200 /~A). This was somewhat expected since (i) the cut-off (30/xA) was arbitrarily selected and (ii) the physiological tests were few and would be expected to influence a large number of reticular formation cells. Two differences appeared, however. First, a smaller proportion of cells was affected by raising AP, and a small number of putative sympathetic premotor neurons could be backfired from LC with these higher currents. Finally, very few RVLM-LC cells were sensitive to nociceptive inputs, especially if one includes the large contingent of silent unresponsive cells. Since a few were found to be activated, the present results do not exclude the possibility that nociceptive inputs to LC might be transmitted via RVLM. However, the small number of responsive cells would tend to support Rasmussen and Aghajanian's contention that RVLM is not or is minimally involved in relaying nociceptive inputs to L C 26. Acknowledgments. The authors gratefully thank Ms. Tina Riley for technical assistance. This work was supported by grants from the National Institutes of Health to P.G.G. (HL 28785 and HL 39841).
REFERENCES 1 Aston-Jones, G., Astier, B. and Ennis, M., Inhibition of nora-
drenergic locus coeruleus neurons by C1 adrenergic cells in the rostral ventral medulla, J. Neurosci., 273 (1991). 2 Aston-Jones, G., Ennis, M., Pieribone, V.A., Nickell, W.T. and Shipley, M.T., The brain nucleus locus coeruleus: restricted afferent control of a broad efferent network, Science, 234 (1986) 734-737. 3 Barman, S.M. and Gebber, G.L., Lateral tegmental field neurons
of cat medulla: a source of basal activity of ventrolateral medulIospinal sympathoexcitatory neurons, J. NeurophysioL, 57 (1987) 1410-1424. 4 Biesold, D., Kurosawa, M., Sato, A. and Trzebski, A., Hypoxia and hypercapnia increase the sympathoadrenal medullary functions in anesthetized, artificially ventilated rats, Jap. J. Physiol., 39 (1989) 511-522. 5 Elam, M., Yao, T., Svensson, T.H. and Thoren, P., Regulation of locus coeruleus neurons and splanchnic sympathetic nerves by cardiovascular afferents, Brain Res., 290 (1984) 281-288. 6 Elam, M., Yao, T., Thoren, P. and Svensson, T.H., Hypercapnia and hypoxia: chemoreceptor-mediated control of locus coeruleus neurons and splanchnic sympathetic nerve, Brain Res., 222 (1981) 373-381. 7 Ennis, M. and Aston-Jones, G., Two physiologically distinct populations of neurons in the ventrolateral medulla innervate the locus coeruleus, Brain Res., 425 (1987) 275-282. 8 Ennis, M. and Aston-Jones, G., Activation of locus coeruleus from nucleus paragigantocellularis: a new excitatory amino acid pathway in brain, Z Neurosci., 8 (1988) 3644-3657. 9 Gillon, J.Y., Richard, F., Quintin, L., Pujol, J.F. and Renaud, B., Pharmacological and functional evidence for extracellular 3,4-dihydroxyphenylacetic acid as an index of metabolic activity of the adrenergic neurons: an in vivo voltammetry study in the rat rostrol ventrolateral medulla, Neuroscience, 37 (1990) 421-430. 10 Guyenet, P.G., Role of the ventral medulla oblongata in blood pressure regulation. In A.D. Loewy and K.M. Spyer (Eds.), Central Regulation of Autonomic Functions, Oxford University Press, New York, 1990, pp. 145-167. 11 Guyenet, P.G., Respiratory-sympathetic integration in the medulla oblongata. In G. Kunos and J. Ciriello (Eds.), Central Neural Mechanisms in Cardiocascular Regulation, Vol. H, Birkhauser, Boston, 1992 (in press). 12 Guyenet, P.G. and Aghajanian, G.K., ACh, substance P and met-enkephalin in the locus coeruleus: pharmacological evidence for independent sites of action, Eur. J. Pharmacol., 53 (1979) 319-328. 13 Guyenet, P.G. and Young, B.S., Projections of nucleus paragigantocellularis lateralis to locus coeruleus and other structures in rat, Brain Res., 406 (1987) 171-184. 14 Guyenet, P.G., Darnall, R.A. and Riley, T.A., Rostral ventrolateral medulla and sympathorespiratory integration in rats, Am. J. PhysioL, 259 (1990) R1063-R1074. 15 Haselton, J.R. and Guyenet, P.G., Central respiratory modulation of medullary sympathoexcitatory neurons in rat, Am. J. Physiol., 256 (1989) R739-R750. 16 Haselton, J.R. and Guyenet, P.G., Electrophysiological characterization of putative C1 adrenergic neurons in the rat, Neuroscience, 30 (1989) 199-214. 17 Haselton, J.R. and Guyenet, P.G., Ascending collaterals of medullary barosensitive neurons and C1 cells in rats, Am. J. PhysioL, 258 (1990) R1051-R1063. 18 Jankowska, E. and Roberts, W.J., An electrophysiological demonstration of the axonal projections of single spinal interneurones in the cat, J. Physiol., 222 (1972) 597-622. 19 Mermet, C. and Quintin, L., Effect of clonidine on catechol metabolism in the rostral ventrolateral medulla: an in vivo electrochemical study, Eur. J. Pharmacol., 204 (1991 ) 105-107. 20 Milner, T.A., Abate, C., Reis, D.J. and Pickel, V.M., Ultrastructural localization of phenylethanolamine N-methyltransferase-like immunoreactivity in the rat locus coeruleus, Brain Res., 478 (1989) 1-15. 21 Morilak, D.A., Fornal, C. and Jacobs, B.L., Single unit activity of noradrenergic neurons in locus coeruleus and serotonergic neurons in the nucleus raphe dorsalis of freely moving cats in relation to the cardiac cycle, Brain Res., 399 (1986) 262-270. 22 Paxinos, G. and C. Watson. The Rat Brain in Stereotaxic Coordinates', 2nd edn., Academic Press, Sydney, 1986. 23 Pieribone, V.A. and Aston-Jones, G., Adrenergic innervation of the rat nucleus locus coeruleus arises predominantly from the C1 adrenergic cell group in the rostral medulla, Neuroscience, 41 (199l) 525-542.
75 24 Pieribone, V.A., Aston-Jones, G. and Bohn, M.C., Adrenergic and non-adrenergic neurons in the C1 and C3 areas project to locus coeruleus: a fluorescent double labeling study, Neurosci. Lett., 85 (1988) 297-303. 25 Rasmussen, K. and Aghajanian, G,K., Withdrawal-induced activation of locus coeruleus neurons in opiate-dependent rats: attenuation by lesions of the nucleus paragigantocellularis, Brain Res., 505 (1989) 346-350.
26 Rasmussen, K. and Aghajanian, G.K., Failure to block responses of locus coeruleus neurons to somatosensory stimuli by destruction of two major afferent nuclei, Synapse, 4 (1989) 162-164. 27 Schwarzacher, S.W., Wilhelm, Z., Anders, K. and Richter, D.W., The medullary respiratory network in the rat, J. Physiol., 435 (1991) 631-644.
