Journal of Applied Physiology

Selected Contribution: Phrenic long-term facilitation requires 5-HT receptor activation during but not following episodic hypoxia

D. D. Fuller, A. G. Zabka, T. L. Baker, G. S. Mitchell


Episodic hypoxia evokes a sustained augmentation of respiratory motor output known as long-term facilitation (LTF). Phrenic LTF is prevented by pretreatment with the 5-hydroxytryptamine (5-HT) receptor antagonist ketanserin. We tested the hypothesis that 5-HT receptor activation is necessary for the induction but not maintenance of phrenic LTF. Peak integrated phrenic nerve activity (∫Phr) was monitored for 1 h after three 5-min episodes of isocapnic hypoxia (arterial Po 2 = 40 ± 2 Torr; 5-min hyperoxic intervals) in four groups of anesthetized, vagotomized, paralyzed, and ventilated Sprague-Dawley rats [1) control (n = 11), 2) ketanserin pretreatment (2 mg/kg iv; n = 7), and ketanserin treatment 0 and 45 min after episodic hypoxia (n = 7 each)]. Ketanserin transiently decreased ∫Phr, but it returned to baseline levels within 10 min. One hour after episodic hypoxia, ∫Phr was significantly elevated from baseline in control and in the 0- and 45-min posthypoxia ketanserin groups. Conversely, ketanserin pretreatment abolished phrenic LTF. We conclude that 5-HT receptor activation is necessary to initiate (during hypoxia) but not maintain (following hypoxia) phrenic LTF.

  • respiratory control
  • plasticity
  • modulation
  • serotonin
  • long- term facilitation
  • 5-hydroxytryptamine

episodic hypoxia or electrical stimulation of carotid chemoafferent neurons elicits a sustained, serotonin-dependent augmentation of respiratory motor output known as long-term facilitation (LTF; Refs. 15,23). LTF is an example of central neural plasticity in respiratory motor control and provides an experimental opportunity to examine serotonin-dependent plasticity in the mammalian central nervous system. LTF was first reported by Millhorn and colleagues (16, 17) who demonstrated that episodic electrical stimulation of the carotid sinus nerve caused a persistent facilitation of phrenic inspiratory activity for up to 90 min after stimulation in anesthetized cats. This form of LTF could be blocked by pretreatment with a serotonergic neurotoxin (5,7-dihydroxytryptamine), a serotonin receptor antagonist (methysergide), or serotonin depletion via p-chlorophenylalanine (17). Subsequently, it was demonstrated that phrenic LTF elicited by episodic hypoxia could be blocked by pretreatment with either a broad spectrum serotonin (5-hydroxytryptamine; 5-HT) receptor antagonist (methysergide; Ref.1) or an antagonist selective for the 5-HT2receptor subtype (ketanserin; Ref. 13). Although these studies demonstrate that 5-HT receptor activation is necessary for LTF, they do not clarify whether 5-HT receptor activation is necessary for LTF induction (during hypoxia), maintenance (after hypoxia), or both.

Morris and colleagues (19) report that neuronal discharge in some raphe neurons remains elevated after chemoafferent neuron activation in anesthetized cats. Accordingly, a persistent elevation of serotonin release from caudal raphe neurons posthypoxia and consequent 5-HT receptor activation may be necessary for the maintenance of LTF. However, Morris et al. did not monitor raphe neuron activity for more than 10 min after carotid chemoreceptor stimulation; therefore, it remains unclear if persistent augmentation of raphe activity and serotonin release contribute to LTF at later times. Millhorn et al. (17) demonstrated that methysergide administration after LTF induction reduces phrenic activity to baseline levels (or below) in cats. However, methysergide depresses phrenic activity in anesthetized cats, even without a background of LTF (9). Thus the apparent reversal of LTF by methysergide (17) is inconclusive because it could represent suppression of phrenic activity unrelated to phrenic LTF. An alternate hypothesis is that the relevant serotonin release (and 5-HT receptor activation) occurs only during the hypoxic episodes (and a relatively brief period posthypoxia). Raphe neuron activation and serotonin release during episodic hypoxia may trigger intracellular mechanisms that maintain LTF without further dependence on increased serotonin release and 5-HT receptor activation (10).

Our hypothesis was that 5-HT receptor activation during (but not after) episodic hypoxia is necessary for expression of phrenic LTF. This hypothesis was tested by administration of ketanserin at selected times before and after episodic hypoxia. Because ketanserin has no long-lasting effect on baseline phrenic activity in anesthetized rats, it is a suitable drug to determine whether 5-HT receptor activation is necessary for LTF maintenance after episodic hypoxia.


Experiments were performed on 300- to 400-g male Sprague-Dawley rats obtained from Charles River Laboratories (Wilmington, MA; rat colony K-62, Kingston, NY). The Animal Care and Use Committee at the University of Wisconsin-Madison approved all experimental procedures.

Experimental preparation.

Isoflurane anesthesia was induced (2.5–3.5%; inspired O2 fraction = 0.50, balance N2) via a nose cone. The trachea was cannulated to permit mechanical ventilation while continuing isoflurane anesthesia. A femoral venous catheter was inserted, and the rats were gradually converted to urethane anesthesia (1.6 g/kg in distilled water); anesthetic depth was assessed by monitoring blood pressure responses to toe pinch. A femoral arterial catheter enabled blood pressure measurements (Gould P23ID). Bilateral vagotomy was performed in the midcervical region, and paralysis was induced with pancuronium bromide (2.5 mg/kg iv). The end-tidal CO2 pressure (Pet CO2) was measured using a rapidly responding CO2 analyzer (model 1265, Novametrix, Wallingford, CT). Arterial partial pressures of O2 (PaO2) and CO2(PaCO2) and pH were determined from 0.2-ml blood samples (ABL-500, Radiometer, Copenhagen, Denmark). Rectal temperature was maintained (37–38°C) with a heated table. At the conclusion of all experiments, rats were euthanized via urethane overdose.

Nerve recordings.

The phrenic and XII nerves were isolated using a dorsal approach, cut distally, desheathed, bathed in mineral oil, and placed on bipolar silver wire electrodes. Nerve electrical activities were amplified (10,000×) and filtered (100–10,000 Hz) (model 1800, A-M Systems, Carlsborg, WA). The amplified signal was full-wave rectified and integrated (time constant = 50 ms; model MA-821RSP, CWE, Ardmore, PA) and digitized for recording on a computer using WINDAQ software (Akron, OH).

Experimental protocol.

The LTF protocol has been described previously (1, 13). After urethane anesthesia, a minimum of 1 h was allowed for stabilization. The CO2 apneic threshold for phrenic activity was determined by increasing the ventilator rate until respiratory-related activity ceased and then slowly decreasing ventilator rate and/or increasing inspired CO2 until phrenic activity resumed. Pet CO2 was maintained 3 Torr above this CO2 apneic threshold, thereby standardizing baseline conditions in each animal. After baseline conditions were established, at least 30 min were allowed to ensure steady state.

A baseline arterial blood sample was taken a few minutes before episodic hypoxia. Rats were then exposed to three 5-min hypoxic episodes (Fi O2 ∼ 0.11–0.12) separated by 5 min of hyperoxia (Fi O2= 0.50). An arterial sample was drawn during the fourth minute of the first hypoxic episode, and, if PaO2 was not between 35 and 45 Torr, Fi O2 was adjusted during the next episode. Phrenic and XII activities were monitored for 60 min after episodic hypoxia under isocapnic conditions (PaCO2 within 1 Torr of baseline; Table1). Arterial blood samples were drawn 15, 30, and 60 min after episodic hypoxia. At the conclusion of the experiment, rats were exposed to an additional 5-min hypoxic episode followed by a 5-min hypercapnic challenge (Pet CO2 of ∼80–90 Torr).

View this table:
Table 1.

PaCO2 and mean arterial pressure during baseline conditions, the first of 3 hypoxic episodes, and 15, 30, and 60 min after episodic hypoxia*

Experimental groups.

Three groups (each n = 7) were administered ketanserin tartrate (2 mg/kg in 1 ml of saline iv; RBI, Natick, MA) at different times. A pretreatment group (pre-ket) received ketanserin 20–25 min before episodic hypoxia. Other groups were given ketanserin immediately after episodic hypoxia (0-ket) or 45 min after episodic hypoxia (45-ket). A control group (n = 11) did not receive ketanserin. None of the following variables differed between groups: body weight, body temperature, CO2 apneic threshold, PaO2 at any time point, or PaCO2 regulation relative to baseline (Table 1).

We wished to differentiate between acute, nonspecific effects of ketanserin and those arising from the reversal of LTF. To help differentiate between these actions, the acute influences of ketanserin administration before episodic hypoxia were determined at 5-min intervals in the group pretreated with ketanserin, thus enabling us to describe the time course of acute changes in nerve activity unrelated to LTF.

Data analyses.

Peak integrated phrenic and XII nerve activities and arterial blood pressure were quantified over 30-s periods before the first hypoxic episode (baseline), at the end of the first hypoxic episode (i.e., the short-term hypoxic response), 15, 30, and 60 min after episodic hypoxia, and during the hypoxic and hypercapnic responses at the end of a protocol. Peak integrated phrenic and XII amplitudes (∫Phr and ∫XII, respectively) were quantified by measuring the peak height of the integrated neurogram and expressing this value relative to baseline and the hypercapnic response (i.e., “max”). Expressing changes in nerve activity relative to both the baseline and the maximum minimizes potential normalization artifacts that may occur when comparing neurograms within and between experimental preparations. We elected to present the data only as a percent change from baseline because all results were qualitatively similar when expressed as percent maximum.

Statistical comparisons between groups were performed using two-way ANOVA with a repeated measures design followed by the Student-Newman-Keuls post hoc test (Sigma-Stat version 1.0; Jandel Scientific, St. Louis, MO). One-way ANOVA was used to compare the following variables: body weight, blood gases, CO2 apneic threshold, and short-term hypoxic responses. Differences were considered statistically significant when the P value was <0.05.


Ketanserin exerted differential effects on baseline phrenic and XII activity (P = 0.003; Fig.1). Phrenic activity transiently decreased in most preparations (although the decrease did not reach statistical significance; P = 0.06) and returned to baseline within 10 min (Fig. 1). Therefore, ketanserin-induced shifts in baseline would not invalidate conclusions regarding the need for persistent 5-HT receptor activation in maintaining phrenic LTF. On the other hand, XII nerve activity decreased after ketanserin administration and did not return to baseline levels within 20 min (Fig. 1; P < 0.05). Thus it is not possible to make conclusions regarding the role of persistent 5-HT receptor activation in the maintenance of XII LTF using our experimental approach. Accordingly, XII data are not presented for the 0-ket and 45-ket groups. Respiratory burst frequency was transiently (although not significantly, P = 0.08) attenuated by pretreatment with ketanserin and returned to baseline values within 10 min (Fig. 1). Mean arterial pressure (MAP) was significantly decreased by ketanserin (34 ± 6 mmHg) and did not return to baseline levels during the study (Fig. 1; see Critique of methods below).

Fig. 1.

Time course of changes in peak integrated phrenic (●; n = 6) and peak integrated hypoglossal (■, n = 6) amplitude (∫Amp), respiratory burst frequency, and mean arterial pressure (MAP) following pretreatment with ketanserin (i.e., before intermittent hypoxia). Changes in integrated phrenic nerve activity (∫Phr) following ketanserin treatment approached but did not reach statistical significance (P = 0.06). Phrenic and XII responses were significantly different (P = 0.003). *Different from preketanserin values, P < 0.05

Phrenic amplitude and burst frequency responses during hypoxia were not significantly different between experimental groups (for ∫Phr: control was 101 ± 15, pre-ket was 130 ± 15, 0-ket was 118 ± 9, and 45-ket was 119 ± 24% baseline;P > 0.05). In contrast, ∫XII increases during hypoxia were significantly lower in the control (143 ± 26%), 0-ket (174 ± 28%), and 45-ket (134 ± 17%) groups relative to the pre-ket group (264 ± 52%; P < 0.05).

Control rats had augmented ∫Phr and ∫XII amplitudes after episodic hypoxia, indicating LTF (57 ± 11 and 69 ± 12% baseline at 60 min, respectively; both P < 0.05; Figs.2 and 3). Conversely, episodic hypoxia did not evoke LTF in either neurogram in rats pretreated with ketanserin (∫Phr: 9 ± 16% and ∫XII: 13 ± 12% baseline at 60 min after episodic hypoxia; bothP > 0.05; Figs. 2 and 3). Ketanserin had no significant effect on the development of phrenic LTF when administered 0 or 45 min after episodic hypoxia (0-ket: 48 ± 16% baseline, 45-ket: 46 ± 8% baseline; both P < 0.05 relative to baseline; Figs. 2 and 3). There were no significant differences in ∫Phr activity between the control, 0-ket, and 45-ket groups at any time point. Respiratory burst frequency was not different from baseline at any time following episodic hypoxia in any experimental group (Fig. 3), indicating a lack of frequency LTF in these experiments.

Fig. 2.

Changes in phrenic motor output following episodic hypoxia in 1 control rat and rats given intravenous ketanserin before (pre-ket) and immediately after (0-ket) episodic hypoxia. The integrated phrenic neurogram is shown during baseline conditions, the first hypoxic episode, and 30 and 60 min posthypoxia (see text for protocol). Control and 0-ket rats both demonstrate phrenic long-term facilitation (LTF), manifested as a progressive increase in ∫Phr after episodic hypoxia. In contrast, the animal pretreated with ketanserin did not develop phrenic LTF. Although ketanserin administered posthypoxia transiently decreases phrenic motor output, it does not impair phrenic LTF.

Fig. 3.

Changes in ∫Phr, ∫XII, and burst frequency from baseline following episodic hypoxia. Ketanserin was administered before episodic hypoxia (pre-ket; ■), immediately after episodic hypoxia (0-ket; ▴), or 45 min after episodic hypoxia (45-ket; ▾). Control rats (●) were treated the same but were not administered ketanserin. *Different from baseline, P < 0.05; †different from pre-ket value at same time, P < 0.05.

MAP decreased during hypoxia in all rats (Table 1). After hypoxia, MAP was successfully maintained relative to baseline in control and pre-ket rats (Table 1). However, ketanserin resulted in a decrease in MAP; therefore, MAP was significantly below baseline at all postepisodic hypoxia time points in the 0-ket group and at the 60-min time point in the 45-ket group (Table 1; see Critique of methods below).


Pretreatment with ketanserin (before episodic hypoxia) prevented phrenic LTF, confirming prior results (13). In contrast, when ketanserin was administered 0 or 45 min after episodic hypoxia, the magnitude of LTF was unchanged relative to control. Thus it appears that 5-HT receptor activation is necessary to induce but not maintain LTF of phrenic motor output. 5-HT receptor activation does not appear to contribute significantly to LTF once episodic hypoxia has ended. We suggest that activation of 5-HT receptors during episodic hypoxia initiates cellular or synaptic events that maintain phrenic LTF.

Pretreatment with ketanserin blocked LTF.

Pretreatment with methysergide before episodic chemoafferent stimulation prevents LTF of phrenic and XII motor output (1,17). Similarly, the more selective 5-HT2 receptor antagonist ketanserin blocks LTF of phrenic motor output when administered before episodic hypoxia (13). The present data confirm this finding. However, the experiments of Kinkead and Mitchell (13) were conducted on a substrain of Sprague-Dawley rats (Harlan) that do not normally exhibit XII LTF (11). Therefore, the influence of 5-HT receptor blockade on XII LTF was not determined in that study. The current experiments were done on a Sprague-Dawley substrain that exhibits robust XII LTF (Charles River; see Fig. 3). Our data demonstrate for the first time that ketanserin also prevents XII LTF in anesthetized rats.

Ketanserin treatment following episodic hypoxia does not block LTF.

When ketanserin was given either immediately or 45 min after episodic hypoxia, phrenic LTF was unaffected. These data strongly suggest that 5-HT receptor activation during episodic hypoxia is necessary to induce phrenic LTF but continued 5-HT receptor activation following episodic hypoxia is not necessary to maintain phrenic LTF.

These experiments were designed to investigate the relevant timing of 5-HT receptor activation in LTF but not the location of these 5-HT receptors. Although 5-HT receptors located in the brain stem, spinal cord, or other locations (4, 6, 15) could, in principle, participate in LTF, recent findings in our laboratory (2) indicate that the relevant 5-HT receptors for phrenic LTF are located in the spinal cord. By inference, we suspect that the relevant 5-HT receptors for XII LTF are within the hypoglossal motor nucleus.

Although the detailed cellular or synaptic mechanisms elicited by 5-HT receptor activation during episodic hypoxia are unknown, we recently proposed a working model of phrenic LTF (10). In this model, we hypothesize that 5-HT2A receptors located on phrenic motoneurons play a critical role in induction of phrenic LTF. In brief, we proposed that hypoxia activates raphe serotonergic neurons, which in turn release serotonin in close proximity to phrenic motoneuron dendrites and/or cell bodies. Activation of postsynaptic 5-HT2A receptors on phrenic motoneurons leads to an intracellular cascade, resulting in the persistent phosphorylation and potentiation of inward currents mediated by glutamate receptors. Consistent with this hypothesis, 5-HT receptor activation has been shown to potentiate glutamatergic currents in several different preparations (5, 24, 29). Potentiation of glutamate receptor currents will amplify descending respiratory drive, leading to greater phrenic output for the same presynaptic glutamate release (and thus LTF). Our model focuses on postsynaptic effects at spinal motoneurons; however, we do not preclude presynaptic effects or additional effects of 5-HT receptor activation at supraspinal levels (4, 6, 15).

Phrenic vs. XII responses to ketanserin.

Previous studies have quantified respiratory neural activity in phrenic and XII nerves before and after intravenous administration of serotonin receptor antagonists (1, 13). The broad-spectrum serotonin receptor antagonist methysergide has excitatory effects on phrenic and XII activity in anesthetized rats (1). In contrast, the more selective antagonist ketanserin was reported to transiently inhibit both phrenic and XII activity in anesthetized rats (13). The present data differ from our earlier report (13) in that ketanserin had a long-lasting inhibitory effect on XII (but not phrenic) activity (Fig. 1). Thus the present data suggest that 5-HT2 receptor activation provides a tonic facilitory influence on eupneic XII motor output. The differential effects of ketanserin on respiratory-related XII discharge may relate to the particular substrain of Sprague-Dawley rats being used. Kinkead and Mitchell (13) studied Sprague-Dawley rats obtained from Harlan (colony 205), whereas we used rats obtained from Charles River (colony K-62). Profound differences in the capacity for XII LTF exist between these Sprague-Dawley substrains (11). Specifically, the results of a blinded experiment show that Charles River Sprague-Dawley rats exhibit robust serotonin-dependent XII LTF, whereas Harlan Sprague-Dawley rats do not (11). The observation that respiratory neural control differs between rats supplied by different vendors is not unique; others have reported that genetic variations contribute to ventilatory phenotype differences among rodent strains (12, 26) and substrains (20, 21, 22, 27, 28). Thus serotonergic modulation of XII motor output may differ considerably between Harlan and Charles River Sprague-Dawley rats.

Acute hypoxic responses.

Serotonin increases phrenic and XII motoneuron excitability via 5-HT2 receptor activation in neonatal rat in vitro preparations (3, 14, 18). Accordingly, Kinkead and Mitchell (13) hypothesized that blockade of 5-HT2 receptors with ketanserin would diminish the magnitude of the short-term hypoxic phrenic response. However, ketanserin augmented phrenic (but not XII) burst amplitude during hypoxia, a response that could not be easily explained. In the present study, ketanserin enhanced the XII (but not phrenic) short-term hypoxic response. Thus the influence of ketanserin on short-term hypoxic phrenic and XII responses reported here are inconsistent with our earlier findings (13). Although we have no clear explanation for this apparent discrepancy, we speculate that the differential XII response may be an artifact of diminished baseline XII activity in the present study (i.e., a normalization artifact) or altered serotonergic function in the XII motor nucleus of Harlan vs. Charles River Sprague-Dawley rats (see above).

Critique of methods.

Previous publications from our laboratory have discussed potential limitations of our experimental model (1, 13). Ketanserin was administered intravenously, and, accordingly, we make no conclusions regarding the location of the relevant 5-HT receptors. Another concern is that, in addition to its antagonistic effects at 5-HT2 receptors, ketanserin could have directly or indirectly influenced other neurotransmitter systems. For example, ketanserin appears to bind to the monoamine transporter in bovine chromaffin granule membranes and rabbit cerebral cortex (7,30). Ketanserin can also have α-adrenoreceptor antagonistic effects in humans (8). Although we cannot be certain, the physiological effects of ketanserin in the present study were most likely due to 5-HT2 receptor antagonism because there is considerable literature indicating that phrenic LTF requires serotonin receptor activation in its underlying mechanism (1, 13, 17; reviewed in Ref. 10). Moreover, even if ketanserin influenced respiratory motor output by mechanisms not associated with the serotonergic nervous system, our fundamental conclusion regarding the timing of receptor activation remains intact. That is, phrenic LTF requires activation of (putatively serotonin) receptors during but not following episodic hypoxia.

Another concern relates to the acute effects of ketanserin on baseline respiratory motor output. The long-lasting inhibition of XII motor output after ketanserin administration under baseline conditions (Fig.1) precludes interpretation of the XII response to ketanserin administered after episodic hypoxia. However, because ketanserin effects on phrenic neural output were small and transient, drug-induced shifts in phrenic motor output unrelated to LTF did not influence the experimental outcome or our fundamental conclusions. Finally, we acknowledge the possibility that baroreceptor reflexes or other blood pressure-related effects could have influenced our results. In all cases, MAP declined ∼20 mmHg following ketanserin administration. However, the decrement in MAP after ketanserin administration is unlikely to have caused LTF in the 0-ket or 45-ket groups. Sixty minutes after episodic hypoxia, MAP was 66 mmHg in rats pretreated with ketanserin (no phrenic LTF), whereas MAP was 74 mmHg in rats that received ketanserin immediately following episodic hypoxia (significant phrenic LTF). If hypotension were a major contributing factor to LTF, the pre-ket group should have demonstrated phrenic LTF. Moreover, MAP decreases in the range reported here do not significantly affect respiratory motor output in urethane-anesthetized and vagotomized rats (1, 13, 25).


These experiments were supported by National Heart, Lung, and Blood Institute Grants HL-53319 and HL-36780 and Training Grant HL-07654.


  • Address for reprint requests and other correspondence: D. D. Fuller Dept. of Comparative Biosciences, School of Veterinary Medicine, Univ. of Wisconsin, 2015 Linden Drive West, Madison, Wisconsin, 53706 (E-mail: fullerd{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract