A brainstem area mediating cerebrovascular and EEG responses to hypoxic excitation of rostral ventrolateral medulla in rat
Eugene V. Golanov,
Division of Neurobiology, Department of Neurology and Neuroscience, Weill Medical College of Cornell University, 411 East 69th Street, New York, NY 10021, USA
Corresponding author E. V. Golanov: Division of Neurobiology, Department of Neurology and Neuroscience, Weill Medical College of Cornell University, 411 East 69th Street, New York, NY 10021, USA. Email: email@example.com
Author's present address D. A. Ruggiero: Departments of Anatomy & Cell Biology and Neuroscience in Psychiatry, Columbia University College of Physicians and Surgeons, 1051 Riverside Drive, Box 42, New York, NY 10032, USA.
1We sought to identify the medullary relay area mediating the elevations of regional cerebral blood flow (rCBF) and synchronization of the electroencephalogram (EEG) in the rat cerebral cortex elicited by hypoxic excitation of reticulospinal sympathoexcitatory neurons of the rostral ventrolateral medulla (RVLM).
2In anaesthetized spinalized rats electrical stimulation of RVLM elevated rCBF (laser-Döppler flowmetry) by 31 ± 6 %, reduced cerebrovascular resistance (CVR) by 26 ± 8 %, and synchronized the EEG, increasing the power of the 5–6 Hz band by 98 ± 25 %. Stimulation of a contiguous caudal region, the medullary cerebral vasodilator area (MCVA), had comparable effects which, like responses of RVLM, were replicated by microinjection of L-glutamate (5 nmol, 20 nl).
3Microinjection of NaCN (300 pmol in 20 nl) elevated rCBF (17 ± 5 %) and synchronized the EEG from RVLM, but not MCVA, while nicotine (1.2 nmol in 40 nl) increased rCBF by 13 ± 5 % and synchronized the EEG from MCVA. In intact rats nicotine lowered arterial pressure only from MCVA (101 ± 3 to 52 ± 9 mmHg).
4Bilateral electrolytic lesions of MCVA significantly reduced, by over 59 %, elevations in rCBF and, by 78 %, changes in EEG evoked from RVLM. Bilateral electrolytic lesions of RVLM did not affect responses from MCVA.
5Anterograde tracing with BDA demonstrated that RVLM and MCVA are interconnected.
6The MCVA is a nicotine-sensitive region of the medulla that relays signals elicited by excitation of oxygen-sensitive reticulospinal neurons in RVLM to reflexively elevate rCBF and slow the EEG as part of the oxygen-conserving (diving) reflex initiated in these neurons by hypoxia or ischaemia.
The autonomic responses have been attributed to direct hypoxic excitation of tonically active sympathoexcitatory reticulospinal vasomotor neurons of the rostral ventrolateral reticular nucleus of the medulla (RVLM) (Sun & Reis, 1994a). In vivo these neurons are selectively, rapidly and reversibly excited by blockade of the respiratory chain by microinjection or microiontophoresis of NaCN (histotoxic hypoxia) (Sun & Reis, 1993). Moreover, restricted lesions of RVLM block the systemic and cerebrovascular responses to ischaemia and/or hypoxaemia (Dampney & Moon, 1980; Underwood et al. 1994; Golanov & Reis, 1996a). Indeed, to protect the brain from lack of oxygen may be an important function of the RVLM (Reis et al. 1994).
Reticulospinal neurons of the RVLM function as premotor neurons, acting to initiate the integrated responses to hypoxia by activation of effector (motor) neurons distributed elsewhere in the CNS. Thus, the sympathetic responses can be attributed to activation of the RVL-spinal sympathoexcitatory pathway (Sun & Reis, 1996a), apnoea from activation through short intramedullary projections of the adjacent ventral respiratory cell group (Kita et al. 1994; Sun & Reis, 1996b), and bradycardia – from excitation of medullary cardiovagal neurons (Ross et al. 1984; Benarroch et al. 1986). However, the pathway(s) through which rCBF and the EEG are modified is unknown. Since the RVLM does not innervate the cerebral cortex (Ruggiero et al. 1989), the projection must be indirect.
In this study we investigate whether a small area of the ventrolateral medulla serves to relay the signals generated by hypoxic excitation of the RVLM to elevate rCBF and synchronize the EEG in the cerebral cortex of rats. We report that an heretofore unrecognized paraambigual region of the bulbar reticular formation, which we term the medullary cerebral vasodilator area (MCVA), subserves this role.
Male Sprague-Dawley rats were maintained in a thermally controlled (27°C), light-cycled (07.00 h on, 19.00 h off) environment with free access to water and lab chow. Anaesthesia was induced by 5 % isoflurane in a gas mixture (nitrogen 80 %, oxygen 19.5 %, carbon dioxide 0.5 %) and maintained during surgery by 2.5 % isoflurane. For the remainder of the experiment, isoflurane was maintained between 2 and 1.2 %. The depth of anaesthesia was assured by an absence of a desynchronized EEG, variation in arterial pressure (AP), corneal reflexes and hind leg flexing in response to pinch throughout the procedure. The experimental protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Weill Medical College of Cornell University.
Both femoral arteries and veins were cannulated with polyethylene catheters (o.d. = 0.97 mm). One arterial cannula was used to continuously measure AP, while the second was used to sample blood for measurement of blood gases. One venous catheter was used for continuous infusion of phenylephrine to maintain AP after spinal cord transection or placement of bilateral lesions of RVLM; another was used for injection of drugs.
A tracheal cannula was inserted, wounds were closed, and animals with the isoflurane-gas mixture. The stroke volume was adjusted to be equal (in ml) to 1/100 the animal's weight in grams. The concentrations of O2, N2 and CO2 were adjusted using calibrated flowmeters. Blood gases were measured several times during the experiment in 0.1 ml samples of arterial blood by a blood gas analyser. Blood gases were sampled and adjusted if necessary to keep them in the normal range for rat (pH, 7.46 ± 0.023; Pa,O2, 95.3 ± 1.1; Pa,CO2, 34.2 ± 0.8) (Loeb & Quimby, 1989). Body temperature was monitored with a rectal probe connected to an electric thermometer-controlled heating pad, maintaining temperature at 37 ± 0.6°C.
After instrumentation, animals were placed in a stereotaxic frame with the bite bar adjusted at −11 mm below the interaural line. The calvarium was exposed through a mid-line incision from the frontal bone to the atlanto-occipital junction, and the inferior half of the occipital bone was removed to expose the dorsal medulla. The calvarium was carefully thinned by a dental drill irrigated with saline at room temperature over an area of 3 mm × 4 mm and centred 3 mm caudal to the frontal suture and 2.5 mm lateral to the sagittal suture. Bone was removed until only the internal layer (lamina vitrea) remained. A stainless steel screw was inserted ipsilaterally extradurally (0.5 mm rostral to the frontal and 1 mm lateral to the sagittal suture) to record EEG monopolarly. The indifferent lead was an electrode placed on the exposed neck muscles.
The spinal cord was transected at the C1-C2 segment. To avert the acute elevation in AP associated with the procedure, 0.1 ml of 2 % procaine was injected locally just prior to sharp transection. Gelfoam was inserted into the cut and an infusion of phenylephrine started immediately (1.3–6.4 μg min−1) to maintain AP (90–100 mmHg). Completeness of the transection was established histologically post mortem. Animals were allowed to stabilize for 30–40 min before initiating the experiment.
Pulsatile AP was recorded using a strain gauge pressure transducer connected to an amplifier. EEG was recorded monopolarly, and amplified and filtered (1–100 Hz). The respective signals were simultaneously displayed on channels of a chart recorder and also digitized and stored in a computer. rCBF was recorded with a laser-Döppler flowmeter (LDF). The probe (0.8 mm in diameter) was mounted on a micromanipulator and placed, under magnification, just over the exposed lamina vitrea, avoiding visible vessels (Golanov & Reis, 1994) that can generate false signals. The opening was filled with paraffin oil and the probe left in place for the remainder of the experiment. Flow values were expressed in arbitrary units (perfusion units, PU). Cerebrovascular reactivity was assessed immediately after positioning of the probe and randomly during the experiment by increasing the concentration of CO2 in the inhaled gas mixture to 5–7 % for 2 min. The procedure, which increased Pa,CO2 to 51–75 mmHg, but did not change Pa,O2 (95–118 mmHg), rapidly elevated rCBF by 60–90 %. If, during the course of an experiment, CO2 reactivity was lost, the experiment was discontinued and the animal killed.
Electrical and chemical stimulation
To stimulate the medulla oblongata electrically or chemically, electrodes or micropipettes, respectively, were attached to the stereotaxic electrode holder and positioned over the calamus scriptorius at a 10 deg posterior inclination. The coordinates of the calamus were taken as stereotaxic zero. An area of medulla 0 to 3 mm rostral to calamus scriptorius, 0 to 2.5 mm lateral, and from −0.5 to −2.5 mm ventral to stereotaxic zero, was explored systematically for evoked changes in rCBF and/or AP.
The medulla was stimulated through monopolar electrodes consisting of Teflon-insulated stainless steel wire (150 μm) exposed only at the cut tip. An indifferent electrode was attached to the neck muscles. Cathodal square-wave pulses (5–150 μA, 0.5 ms, 50 Hz) were generated by a square-wave stimulator and delivered to the electrodes through a constant current stimulus isolation unit. To determine the threshold of the reaction the current was increased from 5 μA in steps of 5 μA until the maximum amplitude of reaction exceeded 2 standard deviations of baseline values. Electrolytic lesions were made through the same electrodes by a constant anodal current (500 μA, 30 s).
Intracerebral microinjections were made through glass capillary micropipettes with an outer diameter of 0.9 mm. The pipette tip was broken back to 40–60 μm in diameter. Drugs were injected in a volume of 20 nl manually by pressure over 15–20 s. L-Glu (5 nmol) was injected mixed with rhodamine beads to mark the site of injection. NaCN (300 pmol) was microinjected as a fresh solution in saline (pH 7.3 after adjustment with phosphate buffer). Nicotine (62 nmol) was injected as a freshly prepared solution in saline (pH 7.3 adjusted with phosphate buffer). Control microinjections were 20 nl of phosphate buffered saline at pH 7.3. The micropipette was advanced in 0.5 mm steps and, in each animal, four to five injections were made on either side during one experiment (i.e. a total of 8–10). At least 10–15 min were allowed between injections.
At the end of each experiment, animals were killed by a bolus of 1.0 ml (i.v.) of saturated KCl. Brains were removed, frozen and sectioned in 20 μm slices in a cryostat microtome without fixation to prevent shrinking. Alternate slices were stained with thionine. Flourescent images of unstained sections were superimposed on the images of corresponding stained sections using a computerized imaging system. Structures were identified microscopically and their distribution in the brain plotted.
Five rats were anaesthetized and instrumented. A double-barrelled micropipette filled with L-Glu and a 10 % solution of the anterograde tracer biotinylated dextran amine (BDA) in 0.1 M PBS (pH 7.4) was inserted stereotaxically into the physiologically defined regions of medulla. BDA was iontophoresed (positive current of 5 μA, 7 s pulses, 50 % duty cycle, for 10 min) through an adjacent barrel of the micropipette into an active site of RVLM or MCVA, identified by microinjections of 5 nmol (in 20 nl) of L-Glu. The active site was identified as a site eliciting a rise in mean AP of >15 mmHg and elevation of rCBF. Control deposits were made into adjacent inactive loci in the lateral tegmental area. Animals were killed after 10 days and perfused with 50 ml of heparinized normal saline, followed by 400 ml of a solution of 4 % paraformaldehyde in 0.1 M PBS. The brainstem was blocked from the spinomedullary junction to the cerebellopontine angle, postfixed, cryoprotected and sectioned transversely at 35 μm on a sledge microtome. Every second and third serial section was saved, and either stained with thionine to distinguish nuclear boundaries or processed histochemically by incubating tissue in a solution of peroxidase-avidin complex (diluted at 1:100 with 0.1 % BSA), to which 0.035 % Triton-X 100 and 100 ml of 0.1 M TBS containing 22 mg of 3,3-diaminobenzidine and 10 ml of 30 % H2O2 were added. BDA-positive fibre trajectories, axon shafts and punctate processes with morphological characteristics of terminals were mapped with a camera lucida attached to a microscope.
After amplification and filtering AP, EEG and rCBF signals were digitized by a computer-based data acquisition system and stored on hard disk for further processing. The principal parameters of the responses (latency, amplitude and duration) were extracted and averaged. Deviation of any parameter by over 2 standard deviations from the baseline was established as the threshold in a reaction. rCBF measurements made by LDF were expressed as percentage changes over baseline.
To estimate changes in cerebrovascular resistance (CVR), we compared the ratio between AP and rCBF before and after each experimental procedure, calculated as a percentage of baseline values.
Fast Fourier transform analysis (FFT) of EEG was performed over a 10 s epoch before, and a 10 s epoch immediately after the stimulation or microinjection of drugs. Resulting data were averaged across experimental animals. Statistical analysis of differences in the means before and after intervention was performed using Student's paired t test, or, for different groups of animals, Student's t test for independent samples. Results were assessed as significant at P < 0.05. All data are expressed as means ±s.e.m.
Distribution of medullary sites elevating rCBF
To localize medullary sites elevating rCBF, we systematically mapped the ventral medulla in 14 rats while recording rCBF, EEG and AP (Fig. 1). The rats were artificially ventilated and spinalized with AP and blood gases maintained so as to eliminate evoked changes in sympathetic or respiratory activity. Microelectrodes were lowered through a region bordered caudally at the level of the calamus scriptorius (stereotaxic ‘0′), rostrally (+3.0 mm) at the caudal end of the facial (VIIth) nucleus and laterally by an area extending 3 mm from the mid-line. Figure 1 depicts the electrode tracks. During mapping the brain was electrically stimulated every 200 μm with a 5 s train of pulses at 50 Hz and with a stimulus current of 60 μA and rCBF changes were observed. At the end of the last track a small electrolytic lesion was placed for subsequent reconstruction. In Fig. 1, only sites that produced an increase in rCBF are depicted.
Maximal elevations of rCBF were obtained from two principal areas as depicted by black circles in Fig. 1. One was located +2.2–2.5 mm rostral to the calamus scriptorius and corresponded to the RVLM, confirming previous reports that excitation of the nucleus elevates rCBF (Saeki et al. 1989; Underwood et al. 1992; Golanov & Reis, 1994). A second active area was located +1.1 mm rostral to the calamus and caudal to the RVLM (Fig. 1). This region was bordered rostromedially by the RVLM and, dorsally, was contiguous with the parvicellular reticular nucleus (PCRt) (Fig. 1). It lies in an anatomically undefined subsection of the lateral tegmental field within a ventrolateral extension of the PCRt bordering the semicompact division of nucleus ambiguus (Swanson, 1992; Paxinos & Watson, 1997). We refer to this zone as the medullary cerebral vasodilator area (MCVA) to emphasize function rather than topography.
Effects of stimulation of MCVA on rCBF, EEG and AP in spinalized rats
We characterized the effects on rCBF, CVR, EEG and AP of electrical stimulation of MCVA in 21 spinalized rats. Electrodes were placed into the MCVA, identified functionally by lowering a stimulating electrode through the appropriate medullary level to locate the site along the track from which the largest elevations of rCBF were evoked by a 10 s train (50 Hz, 40 μA). The electrode was fixed in place and its location verified post mortem.
Electrical stimulation of the MCVA elevated rCBF as depicted in Fig. 2A (left panel). (Data were obtained by averaging of 21 trials in 6 animals, 3–4 trials in each animal. This mode of analysis is used throughout this paper.) The optimal stimulus frequency, established while stimulating with a constant stimulus current, was ∼50 Hz (data not shown). At 50 Hz the threshold for elevating rCBF was 12.5 ± 4.7 μA (n= 18, P < 0.01) and the response was graded with increasing stimulus currents (data not shown). In all subsequent studies the MCVA was stimulated for 10 s at 50 Hz with stimulus currents 5 × the threshold current.
Stimulation of the MCVA almost immediately elevated rCBF, reduced CVR and elicited a small and delayed elevation of AP. rCBF began to rise within 3 s of the stimulus onset and reached a maximum of 54 ± 12 % (n= 21, P < 0.01) at 36 ± 5 s. It gradually returned to baseline within 3–5 min. The fall of CVR inversely paralleled the elevation of rCBF and, at maximum, was reduced by 21 ± 9 % (P < 0.01).
With stimulation, AP initially fell for 15–25 s by 13 ± 4 % (P < 0.05), and then reversed to rise to 25 ± 9 % (from 92 ± 2 to 113 ± 5 mmHg, P < 0.05). It recovered 3–4 min later (Fig. 2A, left and right panels). The initial fall and delayed elevation in AP could be abolished by simultaneously administering the selective V1 vasopressin receptor antagonist [deamino-Pen1,Val4,D-Arg8]-vasopressin (10 mg kg−1, i.v.) and the ganglionic blocker hexamethonium (10 mg kg−1) without altering the evoked changes in rCBF or EEG (see below) (Fig. 2A, right panel). The pressor component of the response is identical to the ‘delayed pressor response’ elicited by electrical stimulation of RVLM (Golanov & Reis, 1994) or cerebellar fastigial nucleus (Del Bo et al. 1983) and which, in spinalized rat, results from release of AVP sufficient to elevate AP when the baroreflex arc is interrupted by spinal cord transection (Del Bo et al. 1983; Golanov & Reis, 1994). The depressor response can probably be attributed to cholinergic vasodilatation of extracerebral cranial vessels (Nakai et al. 1993).
Stimulation of the MCVA also synchronized the EEG within 5 s (Fig. 2C). Quantitative analysis of group data by FFT (Fig. 2D) indicated a 52 ± 12 % increase in the power of the 5–6 Hz band (P < 0.05) and a 28 ± 13 % decrease in the power of the 1–3 Hz band, without a change in amplitude (683 ± 117 μV before, and 725 ± 154 μV after, stimulation, n= 21; P > 0.1) (Fig. 2C). EEG activity recovered in parallel with rCBF.
Microinjection of L-Glu (5 nmol, 20 nl, n= 3) into MCVA reversibly increased rCBF by 23 ± 10 % (P < 0.05) over ∼2 min and, in parallel, reduced CVR by 10 ± 6 % (P < 0.01) (Fig. 3). It also increased AP by 11 ± 5 % (P < 0.05) and synchronized the EEG, increasing the power of the 4–6 Hz components by 50 ± 19 %, without changing amplitude (682 ± 152 μV before and 731 ± 127 μV after, P > 0.01).
Stimulation of RVLM
Stimulation of the RVLM in five spinalized rats elicited qualitatively similar changes in rCBF to those evoked from MCVA, but of smaller magnitude (Fig. 4A). rCBF increased to 30 ± 6 % (P < 0.001) within 29 s from the onset of the stimulus, with recovery in 3–5 min. CVR was reduced in parallel by 26 ± 8 % (P < 0.05). The threshold current for elevating rCBF in RVLM was 23.2 ± 6.2 μA, significantly higher than in MCVA (n= 5, P < 0.05).
Stimulation of RVLM elicited, within 3 s, a limited drop of AP from 93 to 78 mmHg, followed, after 31 s, by a small delayed pressor response (to 105 mmHg). RVLM stimulation also synchronized the EEG (Fig. 4C and D); within 5 s there was a significant increase (to 38 ± 25 %, P < 0.05) in the prevalence of the 4–6 Hz rhythm, without a change in amplitude (703 ± 131 μV before and 742 ± 128 μV after; P > 0.1).
L-Glu microinjected into RVLM increased rCBF by 9 ± 3 % (P < 0.05, n= 5) with synchronization of the EEG about 50 % smaller than that elicited from MCVA (data not shown).
Stimulation of MCVA, RVLM or CVLM in non-spinalized (intact) rats
The average response to stimulation of MCVA, RVLM and the caudal ventrolateral medulla (CVLM) on rCBF, CVR and AP in rats with intact spinal cords is depicted in Fig. 5. Stimulation of the MCVA (10 s train, 50 Hz, 5 × threshold of 8.5 ± 2.7 μA) in five intact rats elevated AP and rCBF and synchronized the EEG. However, the responses differed from spinalized rats (see Fig. 2) in that the elevation of AP appeared and resolved very rapidly, appearing at 2 ± 1 s, peaking at 24 ± 6 % (from 95 ± 2 to 119 ± 2 mmHg, P < 0.05; n= 5) at 5 ± 1 s, and terminating ∼6 s later.
The increase in AP was accompanied by an elevation of rCBF, which reached a maximum of 18 ± 5 % (P < 0.01) at 5 ± 2 s, and gradually recovered over 3–5 min. As AP rose, CVR initially increased by 6 ± 2 % (P < 0.05), but then fell by 14 ± 4 % at 17 ± 4 s to recover in parallel with rCBF (Fig. 5A), indicating vasodilatation. The results are consistent with the fact that an abrupt and rapid elevation of AP is immediately counteracted by a brief reactive cerebrovascular vasoconstriction (Heistad & Kontos, 1983), which is then overridden by the long-lasting vasodilatation. As in spinalized animals, stimulation synchronized the EEG (not shown).
Stimulation of RVLM in intact rats (Fig. 5B) also rapidly and reversibly increased AP, reaching 28 ± 4 % (from 93 ± 3 to 126 ± 6 mmHg, n= 5, P < 0.01) in 7 ± 2 s and returning to the baseline in 6 ± 2 s. In parallel, rCBF increased by 19 ± 4 % in 8 ± 3 s. Changes in CVR were biphasic; after an initial increase of 12 ± 4 % at 9 ± 3 s, it dropped by 11 ± 3 % at 16 ± 4 s (Fig. 5B). EEG changes were comparable to those observed in spinalized animals (data not shown).
We also sought to determine whether the MCVA was anatomically and functionally distinct from the CVLM, a sympathoinhibitory region which lies just ventral and caudal to the MCVA (Cravo et al. 1991). In three intact rats, a site corresponding to the CVLM was stimulated for 10 s (50 Hz, 35 μA). Stimulation elicited an almost immediate (Blessing & Reis, 1982; Blessing, 1997) acute stimulus-locked decrease in AP by 28 % (from 102 ± 5 to 73 ± 7 mmHg, P < 0.05, n= 3), with a parallel reduction in rCBF (16 ± 9 %). There was a modest reduction in CVR (17 ± 6 %) (Fig. 5C). These results indicate that the MCVA and CVLM are topographically and functionally distinct.
In summary, the studies indicate that excitation of neurons of the MCVA elevates rCBF and synchronizes the EEG. The response is comparable to responses evoked from neurons of the RVLM, but is larger. The MCVA is anatomically and functionally distinct from the CVLM.
Effects of NaCN or nicotine in MCVA and RVLM
The cerebrovascular and electrocortical responses elicited from neurons of MCVA and RVLM are qualitatively similar. However, it is not certain whether the two regions represent an extended neuronal pool or are functionally distinct. That they may be independent is suggested by two prior observations. First, microinjection of NaCN into the RVLM elevates AP and rCBF, while injections of the agent into a region corresponding to MCVA are without effect (Sun et al. 1992; Golanov & Reis, 1996a). Second, the MCVA appears to correspond topographically to that region of the underlying medullary surface from which nicotine lowers AP. This response is not evoked by nicotine when injected into a region overlying RVLM (Guertzenstein & Lopes, 1984). We, therefore, compared the effects of microinjecting NaCN or nicotine into MCVA or RVLM.
In five intact and five spinalized rats, the distribution of sites (Fig. 6) from which microinjection of NaCN (300 pmol in 20 nl) increased rCBF was analysed. The dose and volume were selected since they reproducibly and reversibly elicit maximal elevations of AP and rCBF from the region (Sun et al. 1992; Golanov & Reis, 1996a).
In confirmation of previous studies (Sun et al. 1992; Golanov & Reis, 1996a), NaCN microinjected into RVLM of five intact rats increased rCBF by 17 ± 5 % in 32 ± 2 s, to recover 213 ± 63 s later (Fig. 7A, left panel). In parallel, CVR fell by 6 ± 2 % (Fig. 7A). NaCN also elicited a small, immediate and delayed elevation in AP (Fig. 7A), which rose 12 ± 6 mmHg from 98 ± 11 mmHg by 37 ± 12 s, to recover 129 ± 34 s later. In contrast, NaCN injected into the MCVA was without effect. In five spinalized rats, NaCN also increased rCBF (Fig. 7A, right panel) to a maximum of 18 % by 94 ± 29 s (P < 0.05), with recovery in 186 ± 48 s.
Within 8 s after injection of NaCN in intact and spinalized rats, the EEG became synchronized with a significant increase (64 ± 15 %; P < 0.05) in the power of the 6 Hz rhythm, and a reduction in power of the 1–3 Hz rhythm (by 32 ± 10 %) without change in total power and overall amplitude (692 ± 169 μV before and 625 ± 187 μV after treatment, P > 0.1). EEG activity returned to pre-injection levels in parallel with recovery of rCBF. The distribution of positive sites reflected onto the ventral surface of the medulla indicated that they lay over the RVLM (Fig. 6), confirming previous observations (Benarroch et al. 1986).
In five intact rats nicotine (1.2 nmol in 40 nl) microinjected into MCVA (Fig. 8) profoundly reduced AP from 101 ± 3 to 52 ± 9 mmHg (P < 0.001) after 15 ± 7 s, recovering in 7–8 min. Its effects on rCBF were more complex. It initially reduced rCBF by 24 ± 14 % (P < 0.001) in 8 ± 7 s, with a transient recovery at 36 ± 13 s (Fig. 8A, left panel), followed by a secondary depression. It elicited, in parallel, a marked reduction in CVR by 62 ± 21 % (P < 0.001) at 34 ± 15 s. Nicotine significantly increased the power of the 6 Hz rhythm of EEG activity (by 121 ± 19 %, P < 0.05), without affecting total power and overall amplitude (723 ± 134 μV before and 675 ± 148 μV after treatment, P > 0.1).
The hypotension elicited by nicotine may lower AP into the range expected to elicit hypoxic activation of neurons of RVLM. Thus, we analysed the actions of the drug in spinalized rats in which AP was controlled. In five spinalized rats, injection of nicotine into MCVA did not change AP. However, it increased rCBF maximally by 13 ± 5 % (P < 0.01) at 27 ± 12 s, returning to the basal levels in 132 ± 32 s (Fig. 8A, right panel). In parallel, CVR decreased by 15 ± 9 % (P < 0.01). The finding suggests that in intact rats the profound fall of AP initiates ischaemic excitation of RVLM neurons acting on rCBF to produce the biphasic response.
In all rats, nicotine increased the power of the 6 Hz EEG rhythm (by 89 ± 27 %, P < 0.05) without altering amplitude (731 ± 125 and 672 ± 193 μV, P > 0.1) (Fig. 8B and C). In contrast, nicotine microinjected into the RVLM of intact or spinalized rats was without effect.
The results thereby directly indicate that RVLM and MCVA differ with respect to their sensitivity to NaCN and nicotine, and further verify that it is neurons of the RVLM that can be excited by the hypoxic stimulus. The failure of NaCN to elicit responses from the MCVA endorses the view that the excitability of RVLM neurons to histotoxic hypoxia is not shared by all neurons of the ventral medulla (Sun & Reis, 1994b).
Effects of lesions of RVLM and MCVA
The findings that NaCN elevates rCBF and synchronizes the EEG only within RVLM, but that L-Glu elevates rCBF from both RVLM and MCVA, could indicate, first, that neurons of the MCVA are trans-synaptically excited by oxygen-sensitive neurons of the RVLM, and second, that it is neurons of the MCVA that, over ascending projections, elicit the cerebrovascular vasodilatation and synchronization of the EEG in the cerebral cortex. If correct the hypothesis predicts that: (a) in spinalized rats bilateral electrolytic lesions of the MCVA should attenuate or abolish the cerebrovascular and EEG responses elicited by stimulation of RVLM, (b) bilateral lesions of RVLM will not affect the cerebral responses elicited by electrical stimulation of MCVA, and (c) the RVLM and MCVA are anatomically interconnected. The following experiments tested the hypothesis.
Lesions of RVLM.
In five spinalized rats a stimulating electrode was placed unilaterally into a functionally active site of the MCVA and into functionally defined regions of RVLM. After measuring the effects of MCVA stimulation (10 s train, 50 Hz, 5 × threshold current) on rCBF, CVR and EEG, small bilateral electrolytic lesions were made in RVLM. After a recovery period of 20–30 min the MCVA was restimulated.
Before lesions, electrical stimulation of MCVA elevated rCBF by 24 ± 12 % (P < 0.001), reduced CVR by 17 ± 7 % (P < 0.05), and increased the power of the 6 Hz band of the EEG by 109 ± 24 % (Fig. 9A, continuous line). Acute bilateral electrolytic lesions of RVLM (Fig. 9C) did not affect resting rCBF or EEG, nor did they affect responses elicited by stimulating MCVA (Fig. 9A, dotted line, and Fig. 9B). Thus, rCBF was elevated by 29 ± 9 % (P < 0.01), CVR was reduced by 22 ± 9 % (P < 0.05), and the power of the 6 Hz band was increased 187 ± 39 %. These values did not differ from prelesion responses (P > 0.1). As depicted in Fig. 9C, the common area of destruction in all five animals was confined to the RVLM.
Lesions of MCVA.
In five spinalized rats we reversed the procedure by examining the effects of bilateral lesions of the MCVA (Fig. 10) on responses evoked from RVLM. Before lesions, electrical stimulation of RVLM elevated rCBF (Fig. 10A, continuous line) by 17 ± 6 % (P < 0.05), decreased CVR by 18 ± 8 % (P < 0.05), and increased the power of the 6 Hz EEG band by 94 ± 31 %. Lesions themselves did not affect baseline rCBF, CVR or EEG. However, the lesions significantly attenuated the elevation of rCBF (to 7 ± 5 %), the reductions in CVR (to 10 ± 6 %) (P < 0.05 for both), and the synchronization of the EEG (to 22 ± 19 % P > 0.05, compared to baseline EEG), without changing EEG amplitude (723 ± 134 μV before, and 675 ± 148 μV after treatment, P > 0.1).
Anterograde transport from RVLM to MCVA
Injections into RVLM.
In four rats, the anterograde tracer BDA was iontophoresed into an active site in RVLM from which microinjection of 5 nmol of L-Glu maximally increased rCBF and synchronized the EEG (Fig. 11A and C). Fibre trajectories were traced within the medullary reticular formation in tissue processed histochemically for BDA. BDA injection deposits were centred on the RVLM and abutted but did not encompass the caudal pole of the facial nucleus, and were concentrated within the physiologically active area. In each case, the dense central core of the injection site was subjacent to, without involving, the nucleus ambiguous compactus area and encompassed the rostral sympathetic-premotor region neighbouring the ventral medullary surface and coincided with the area containing perikarya of reticulospinal sympathoexcitatory neurons of the C1 area (Benarroch et al. 1986; Ruggiero et al. 1994).
Efferent projections anterogradely labelled for BDA formed bundles of axons traversing the lateral tegmentum and running ipsilaterally to the injection sites. In the reticular formation, axon terminals were heavily labelled at levels of ventrolateral medulla caudal to the injection site. Punctate processes with morphological characteristics of terminals were concentrated in the paraambigual region of the parvicellular reticular formation coinciding with the MCVA (Fig. 11B and D). While the projection arising from RVLM coincided with MCVA, terminals were distributed to contiguous regions of ventral reticular formation, which is consistent with the functional complexity of the reticular regions. Commissural projections were especially pronounced to loci in the ventral tegmentum on the contralateral side, their terminal distribution mirroring the injection sites (not illustrated).
Anterograde transport from MCVA to the RVLM.
In three rats the iontophoretic deposits of BDA were confined to the MCVA (Fig. 12A). The dense central core of each deposit was well circumscribed and limited to a paraambigual zone, located immediately posterior to the rostral vasopressor region of RVLM. Labelled axons were arranged in fascicles concentrated in the ventral tegmentum, from which branches could be traced to areal subunits distributed diagonally across the lateral tegmental field. A rostral fibre trajectory heavily terminated in the RVLM, from which branches issued to a subambigual region coinciding with the C1 adrenergic cell group and surround (Figs 12B and C). The rostral pole of the terminal field abutted the medial border of the facial nucleus. Control data were available from anterograde tracer deposits that were centred on medial, gigantocellular reticular zone (Aicher et al. 1994) or ventral portions of the trigeminal nucleus oralis and adjacent loci in lateral parvicellular reticular formation (E. V. Golanov, D. A. Ruggiero and D. J. Reis, unpublished data). In these cases, terminals were skewed ventromedially or laterally to the cardiovascular active paraambigual and subambigual regions, respectively.
RVLM and MCVA
We confirmed that excitation of the RVLM, electrically or chemically with L-Glu, increases AP and rCBF, and synchronizes the EEG (Golanov & Reis, 1996a). The fact that cortical responses persisted after spinal cord transection indicates that they cannot be attributed to evoked changes in AP or blood gases, both of which were stabilized and normal. The active region was not restricted to the C1 area of RVLM, the site of perikarya of the sympathoexcitatory reticulospinal neurons (Ruggiero et al. 1989), but extended caudally into a paraambigual area of the lateral tegmental field. While the changes in AP, rCBF and EEG responses elicited from RVLM and MCVA were qualitatively similar, those evoked from MCVA were significantly larger.
In addition to topography and magnitude, the RVLM and MCVA also differed with respect to the responses elicited by microinjection of NaCN or nicotine into each region. Nicotine was selected because its topical application to an area of the ventral medullary surface underlying the MCVA, but not the RVLM, lowers AP (Feldberg & Guertzenstein, 1976; Guertzenstein & Lopes, 1984; Benarroch et al. 1986). In intact rats NaCN, as previously reported (Sun et al. 1992; Golanov & Reis, 1996a), elevated AP and rCBF, and synchronized the EEG from RVLM, but not from MCVA. After spinalization the changes in rCBF and EEG were preserved indicating that these responses were independent of sympathetic activation.
In contrast, in intact rats nicotine, when injected into MCVA, but not RVLM, lowered AP, as expected. It also elevated rCBF, reduced CVR, and synchronized the EEG. When systemic effects of nicotine were eliminated by spinalization the drug still elevated rCBF and reduced CVR from MCVA, but not RVLM. The elevation in rCBF elicited from RVLM or MCVA could not be attributed to excitation of cranial parasympathetic neurons (Goadsby, 1990), since it was not blocked by ganglionic blockade with hexamethonium (Goadsby & Shelley, 1990).
These observations indicate that the RVLM and MCVA not only differ topographically, but also pharmacologically. The differences between RVLM and MCVA with respect to actions of nicotine probably do not reflect differences in a cholinergic innervation (Ruggiero et al. 1990), nor in responsiveness to acetylcholine, since cholinergic agonists excite RVLM neurons through stimulation of cholinergic muscarinic receptors (Giuliano et al. 1989; Huangfu et al. 1997). Rather, it might reflect a restricted expression in MCVA of nicotinic cholinergic receptors, as receptor immunocytochemistry suggests (Swanson et al. 1987).
The MCVA appears to be topographically and functionally distinct from nearby cardiovascular control areas of the medulla, notably a major sympathoinhibitory region of the lateral tegmentum, the caudal ventrolateral medulla (CVLM) (for review, see Blessing, 1997). Not only does the functionally defined CVLM lie ventrocaudally to MCVA, as shown here, but unlike MCVA, its activation reduces AP (Blessing & Reis, 1982; Jeske et al. 1993; Blessing, 1997) and rCBF (Maeda et al. 1991), while interrupting its function elevates AP (Cravo et al. 1991). The MCVA is also anatomically and functionally distinct from the nearby gigantocellular depressor area (GiDA) (Aicher et al. 1994; Aicher & Reis, 1997), which, like CVLM, lowers AP when stimulated and raises it when inactivated. It is far removed from other cardiovascular control areas of the medulla (e.g. Dampney, 1994).
Interaction between RVLM and MCVA
The fact that excitation of the MCVA replicates the cerebrovascular and electrocortical responses evoked from RVLM, but that only the RVLM is sensitive to NaCN, is consistent with the proposition that the MCVA relays responses initiated from the oxygen-sensing neurons of RVLM to the cerebral cortex. Two sets of observations support the hypothesis.
First, anatomically anterograde tracing with BDA indicates that some neurons of the RVLM innervate the MCVA region. Demonstration of this local network extends previous studies from this and other laboratories of abundant intrareticular connections emanating from other physiologically distinct areas of the ventral medulla (e.g. Ruggiero et al. 1989; Ellenberger et al. 1990; Zagon, 1995; Lipski et al. 1995), but which appear to have some specificity of projections (Aicher et al. 1994). It also supports evidence of caudal projections from intracellularly labelled reticulospinal neurons of RVLM into areas corresponding to MCVA (Lipski et al. 1995).
Second, in spinalized rats bilateral electrolytic lesions of MCVA, while not affecting basal activity, reduced, by over 50 %, the changes in rCBF and EEG elicited by stimulating the RVLM, a finding confirming previous results (Underwood et al. 1992; Golanov & Reis, 1994). In contrast, lesions of the RVLM had no effect on responses elicited from MCVA. The preservation of cerebrocortical responses from MCVA after RVLM lesions indicates that their abolition cannot be non-selective.
Together the results, therefore, strongly suggest that the MCVA serves to relay the cerebrovascular and EEG responses elicited by stimulation of the RVLM to some as yet undefined rostral site.
Relationship of rCBF and EEG
Stimulation of RVLM or MCVA, elevating rCBF and decreasing CVR, invariably synchronized the EEG. Moreover, the coupling failed when the RVLM was stimulated after lesions of the MCVA. Synchronization was reflected by a shift in the power spectrum to the 4–6 Hz bandwidth. The response was not secondary to selective changes in rCBF, since synchronization persisted when nicotine lowered rCBF.
The observation that electrical or chemical stimulation of RVLM or MCVA always changed these two functions in parallel suggests that the evoked changes in rCBF and EEG are in some manner linked and share, at least in part, common projections. In particular changes in EEG and rCBF were unrelated to the direction of change in AP since they were identical whether AP was elevated (from stimulating RVLM or MCVA) or lowered in response to injection of nicotine into the MCVA. An association between synchronization of electrocortical activity and cerebrovascular vasodilatation is consistent with observations that stimulation of RVLM with single electrical pulses evokes a ‘burst-cerebrovascular wave’ complex (burst-CW complex) consisting of bilateral short bursts of synchronized EEG activity, followed (within ∼1.4 s) by a single wave of vasodilatation (Golanov & Reis, 1995a, 1996b).
The present study underscores the essential role of the rostral medulla in organizing and integrating the cerebrovascular adjustments to hypoxia and/or cerebral ischaemia. While modest levels of hypoxaemia stimulate arterial chemoreceptors, these do not have actions on rCBF (Heistad et al. 1976). Rather, it appears that much of the cerebrovascular response to hypoxaemia is reflex and generated by activation of the RVLM neurons. This is supported by the facts that stimulation of RVLM, like hypoxaemia, elevates rCBF without changing cerebral metabolism (Underwood et al. 1992; Golanov & Reis, 1994), and that bilateral lesions of the RVLM reduce, by up to 50 %, the elevations of cortical blood flow elicited by hypoxaemia without affecting cerebrovascular responses to hypercapnia or impairing cerebrovascular autoregulation (Underwood et al. 1994; Golanov & Reis, 1996a). On the other hand, the fact that after MCVA lesions almost 50 % of the cerebrovascular response is preserved indicates that other pathways also contributed to the response.
Reticulospinal neurons of the RVLM are ‘premotor’ neurons and, as such, are critical for detecting and initiating the vascular, cardiac and respiratory responses to brainstem hypoxia and ischaemia. The systemic response to excitation of RVLM neurons, however, results from activation of a network of effector neurons distributed elsewhere in the CNS. Thus, sympathetic excitation is mediated by an excitatory projection to spinal preganglionic sympathetic neurons and the bradycardia via projections to cardiovagal motor medullary neurons. The integrated response functions to redistribute blood from viscera to brain in response to a challenge to cerebral metabolism.
This study suggests that MCVA, a topographically and functionally, but as yet not cytologically, distinctive region of the bulbar reticular formation, functions as the interposed relay mediating the cerebral effects of hypoxic excitation of RVLM. However, the identity of the next link in the ascending pathway is unknown. While the trajectory of the projection is uncertain, it must involve at least one other subcortical relay, since the ventral medulla does not project to the cerebral cortex (Ruggiero et al. 1994). One potential subcortical candidate may be a region of a caudal subthalamic area which, when stimulated chemically or electrically, increases cortical rCBF and synchronizes the EEG globally and, when bilaterally lesioned, blocks the cerebrovascular and electrocortical responses elicited from MCVA (Golanov & Reis, 1998). The ascending pathways are presently under investigation.
The study also raises the question of whether the MCVA may also be a common relay mediating comparable elevations of rCBF, but not rCGU, and EEG synchronization elicited from other regions of brain, including the cerebellar fastigial nucleus (FN) (Golanov & Reis, 1995a) or nucleus tractus solitarii (NTS) (Golanov & Reis, 1995b). That this might be the case is suggested by observations that lesions of MCVA abolish the primary elevations of rCBF and EEG synchronization elicited from NTS (Golanov & Reis, 1995b), as well as the vasodilatation from the cerebellar fastigial nucleus (Chida et al. 1990).
This work was supported by NIH grants NS36154 (E.V.G.), NS36363 (D.A.R.) and HL18974 (D.J.R.), and a gift from the Irving Harris Foundation.