Episodic but not continuous hypoxia elicits long-term facilitation of phrenic motor output in rats

Abstract

• 1

  Intermittent hypoxia elicits long-term facilitation (LTF) of phrenic motor output in anaesthetized rats. We tested the hypothesis that an equal cumulative duration of continuous hypoxia would not elicit phrenic LTF.

• 2

  Integrated phrenic nerve activity was recorded in urethane-anaesthetized, vagotomized, paralysed and ventilated rats exposed to: (1) 3 × 3 min hypoxic episodes (inspired O₂ fraction (F_I,O2) = 0.11) separated by 5 min hyperoxia (F_I,O2= 0.5; n= 6), (2) 9 min continuous hypoxia (n= 6), or (3) 20 min continuous hypoxia (n= 7). Isocapnia was maintained throughout the protocol.

• 3

  Consistent with previous studies, phrenic amplitude was significantly elevated for at least 1 h following intermittent hypoxia (78 ± 15% 60 min post-hypoxia; P < 0.05) with an associated increase in burst frequency (11 ± 2.1 bursts min⁻¹; P < 0.05). In contrast, 9 or 20 min continuous hypoxia did not elicit LTF of either phrenic amplitude (4.7 ± 5.1 and 10.1 ± 10.2% 60 min post-hypoxia, respectively; P > 0.05) or frequency (4.6 ± 1.3 and 5.1 ± 2 bursts min⁻¹ 60 min post-hypoxia, respectively; P > 0.05).

• 4

  The results indicate that hypoxia-induced long-term facilitation of phrenic motor output is sensitive to the pattern of hypoxic exposure in anaesthetized rats.

The pattern of training has a profound effect on the formation and retention of memories. For example, episodic or spaced training trials are generally more effective in long-lasting memory formation than sustained or massed training trials (Carew et al. 1972; Tully et al. 1994; Yin et al. 1994; Mauelshagen et al. 1998). The cellular mechanisms that impart pattern specificity to memory formation are not well understood.

Long-term facilitation (LTF) of respiratory motor output is a time-dependent memory of the respiratory network following repetitive episodes of carotid chemoafferent activation by electrical stimulation of the carotid sinus nerve (Millhorn et al. 1980a,b; Hayashi et al. 1993) or hypoxia (Hayashi et al. 1993; Bach & Mitchell, 1996; Kinkead & Mitchell, 1999). LTF is serotonin dependent (Millhorn et al. 1980b) and requires the activation of 5-HT2 receptors during (Kinkead & Mitchell, 1999), but not following episodic hypoxia (Fuller et al. 2000b). LTF is primarily a central neural mechanism since LTF is elicited by carotid sinus nerve stimulation (Millhorn et al. 1980a,b; Hayashi et al. 1993), thus bypassing mechanisms of carotid chemosensory transduction.

There is evidence suggesting that LTF is a unique property of intermittent hypoxia. In awake goats, episodic but not sustained hypoxia elicits a long-lasting increase in breathing frequency and tidal volume (viz. Dwinell et al. 1997; Turner & Mitchell, 1997). However, the magnitude of facilitation in awake goats is modest (approximately 20-30% at 40 min) and degrades to baseline relatively quickly (between 40 and 60 min). Thus, long-term facilitation following sustained hypoxia might not be revealed given the small capacity for facilitation in awake goats. On the other hand, anaesthetized, vagotomized rats exhibit considerable facilitation of phrenic motor output 60 min post-episodic hypoxia (approximately 75%), an effect that appears to increase in a time-dependent manner (Bach & Mitchell, 1996). The objective of this study was to investigate the dependence of phrenic LTF on the pattern of hypoxia in anaesthetized, vagotomized rats, a preparation that exhibits robust LTF of phrenic motor output.

METHODS

Experiments were conducted on 19 adult male Harlan Sprague-Dawley rats (colony 236), weighing between 280 and 430 g. The Animal Care and Use Committee at the University of Wisconsin-Madison approved all experimental procedures.

Experimental procedures have been described in detail previously (Bach & Mitchell, 1996; Kinkead et al. 1998; Kinkead & Mitchell, 1999). Briefly, rats were anaesthetized initially with isoflurane (2.5%) in 50% O₂ (balance N₂), then slowly converted to urethane anaesthesia (1.6 g kg⁻¹, i.v.). The adequacy of anaesthesia was assessed periodically by testing blood pressure responses to toe pinch; supplemental doses of urethane were given as necessary. One hour after the beginning of surgery, an intravenous infusion of a 1:4 solution of sodium bicarbonate and lactated Ringer solution was initiated to maintain acid-base balance (5 ml kg⁻¹ h⁻¹).

Rats were vagotomized, paralysed with pancuronium bromide (2.5 mg kg⁻¹) and pump ventilated (V_T= 2-2.5 ml, Rodent Respirator, model 683; Harvard Apparatus, South Natick, MA, USA). Phrenic and hypoglossal nerves were dissected, cut distally and desheathed unilaterally using a left dorsal approach. The nerves were submerged in mineral oil and placed on bipolar silver recording electrodes. Nerve activity was amplified (gain = 10K, A-M systems, Everett, WA, USA), band-pass filtered (100 Hz to 10 kHz) and integrated (Paynter filter, CWE 821, Ardmore, PA, USA; time constant, 100 ms). The signal was then digitized, recorded and analysed using the WINDAQ data acquisition system (DATAQ Instruments, Akron, OH, USA).

Following surgery, at least 1.5 h were allowed for electroneurograms and blood pressure to stabilize. Baseline nerve activity was established in hyperoxic and normocapnic conditions (F_I,O2= 0.5; arterial P_O2 (P_a,O2) > 120 mmHg; arterial P_CO2 (P_a,CO2) ≈2-3 mmHg above the CO₂ apnoeic threshold). End-tidal CO₂ was monitored throughout the protocol with a flow-through capnograph (Novametrix; Wallingford, CT, USA), with sufficient response time to measure end-tidal CO₂ in rats. Inspired CO₂ or ventilator frequency was adjusted as necessary to maintain isocapnia. Blood gases and arterial pressure were monitored and corrected as necessary to ensure that post-hypoxia values were maintained near baseline levels. Experiments were included in the analysis only if: (1) P_a,O2 during hypoxia was between 35 and 45 mmHg, (2) P_a,O2 during the hyperoxic baseline and recovery periods was > 120 mmHg, (3) P_a,CO2 remained within 1.5 mmHg of baseline throughout the protocol, and (4) blood pressure remained within 20 mmHg of its baseline value.

Peak integrated phrenic and hypoglossal amplitudes (Phr and XII) were measured following 3 x 3 min episodes of hypoxia (F_I,O2= 0.11) separated by 5 min isocapnic hyperoxia (F_I,O2= 0.5; n= 6), and 9 (n= 6) and 20 min (n= 7) of continuous hypoxia. The durations of sustained hypoxia were chosen to correspond with the cumulative duration of hypoxic exposure (3 x 3 min = 9 min) or total stimulation protocol time (3 x 3 min episodes plus 2 x 5 min interval = 19 min) in episodic hypoxia. At the end of the final hypoxic exposure, rats were returned to baseline hyperoxic levels. Phrenic and hypoglossal nerve activity were monitored for 1 h post-hypoxia. At the end of the protocol, maximal phrenic and hypoglossal discharges were measured by elevating inspired CO₂ concentration until nerve activity failed to increase with further increases in CO₂ (approximately 12% CO₂). Blood samples (0.3 ml in a heparinized syringe) were drawn before, during, and 15, 30 and 60 min following hypoxia to ensure blood gases met the criteria outlined above; phrenic and hypoglossal measurements corresponding to these time points were used in the analysis. At the conclusion of all experiments, rats were killed with an overdose of urethane (i.v.).

Peak amplitudes and frequency (bursts min⁻¹) of phrenic and hypoglossal nerve activity were averaged in 1 min bins at each recorded data point (baseline and 15, 30 and 60 min post-hypoxia). Changes in amplitude data were normalized in two ways: (1) as a percentage of the baseline value, and (2) as a percentage of the CO₂-stimulated maximal nerve activity. All data are reported as percentage change from baseline, but comparable results were obtained when analysed as a percentage of maximal activity (data not shown). Burst frequency (bursts min⁻¹) is reported as an absolute change from baseline. Statistical analyses were conducted using a two-way ANOVA with a repeated measures design, and individual comparisons were made using the Student-Neuman-Keuls post hoc test. Differences were considered significant if P < 0.05. All values are expressed as means ±s.e.m.

RESULTS

Table 1 lists P_a,CO2 values obtained before and after hypoxia. There were no significant changes in any group at any time point. Thus, attempts to maintain arterial isocapnia were successful.

Table 1. P_a,Co2 values following episodic and continuous hypoxia

	Baseline	15 min	30 min	60 min
3 × 3 min	43.4 ± 1.7	43.6 ± 2.0	43.4 ± 2.1	43.8 ± 1.8
9 min	45.1 ± 1.0	44.9 ± 1.2	45.0 ± 0.9	44.8 ± 1.1
20 min	43.5 ± 1.1	44.9 ± 0.9	43.9 ± 1.5	43.5 ± 1.1

• Arterial P_Co2 values before, and 15, 30 and 60 min after 3 × 3 min episodic and 9 and 20 min continuous hypoxia. There were no significant differences between any groups.

Episodic hypoxia

As previously reported, episodic hypoxia elicited a sustained increase in phrenic burst amplitude (i.e. LTF; Hayashi et al. 1993; Bach & Mitchell, 1996; Kinkead et al. 1998; Kinkead & Mitchell, 1999). Phrenic burst amplitude was significantly above baseline levels at all post-hypoxia time points tested (P < 0.05; 1, 2). When expressed as a percentage increase from baseline, mean phrenic burst amplitude was increased by 78 ± 15% 60 min post-hypoxia. Phrenic LTF was significantly greater at 60 min than 15 min post-episodic hypoxia (78 ± 15 vs. 34.6 ± 6.6%, respectively; P < 0.05), indicating that hypoxia-induced LTF progressively increases for at least 1 h post-episodic hypoxia. Peak integrated hypoglossal amplitude showed no significant LTF following episodic hypoxia (17.8 ± 11.7% 60 min post-hypoxia; Fig. 2). Previous data from our laboratory suggest that hypoglossal LTF is rat sub-strain dependent (Fuller et al. 2000a). In a subsequent blind study (D. F. Fuller, T. L. Baker, M. Behan & G. S. Mitchell, unpublished observations), we found that Harlan Sprague-Dawley rats (colony 236) do not show significant hypoglossal LTF (< 20%), whereas Sasco Sprague-Dawley rats (colony K62) exhibit significant hypoglossal LTF (approximately 50-70%). The mechanism underlying this difference in the capacity to elicit hypoglossal LTF between rat sub-strains is not understood.

Figure 1

Integrated phrenic discharge in one rat exposed to episodic hypoxia and one rat exposed to 20 min continuous hypoxia

Integrated phrenic discharge during baseline, hypoxia, and 15, 30 and 60 min post-hypoxia. Only episodic hypoxia elicits LTF.

Figure 2

Changes in integrated phrenic and hypoglossal amplitude following episodic and continuous hypoxia

Mean integrated phrenic amplitude (±s.e.m.) at 15, 30 and 60 min following episodic (▪), and 9 min (•) and 20 min (▴) continuous hypoxia expressed as a percentage increase from pre-stimulation (baseline) levels. Episodic hypoxia elicited significant phrenic LTF at all time points. In contrast, 9 and 20 min continuous hypoxia did not elicit phrenic LTF. Neither episodic nor continuous hypoxia elicited significant hypoglossal LTF. *Significantly different from pre-stimulus baseline (P < 0.05); significantly different from 3 x 3 min episodic hypoxia.

Episodic hypoxia also elicited a sustained increase in nerve burst frequency at all post-hypoxia time points (P < 0.05, Fig. 3). At 60 min post-hypoxia, mean frequency had increased by 11 ± 2.1 bursts min⁻¹ above baseline levels.

Figure 3

Long-term facilitation of phrenic burst frequency following episodic, but not continuous hypoxia

Mean change in phrenic frequency from baseline (±s.e.m.) at 15, 30 and 60 min following episodic (▪) and 9 min (•) and 20 min (▴) continuous hypoxia. Episodic hypoxia elicited significant LTF of burst frequency at all time points. No LTF was observed following 9 or 20 min continuous hypoxia. *Significantly different from pre-stimulus baseline (P < 0.05); significantly different from 3 x 3 min episodic hypoxia (P < 0.05).

Continuous hypoxia

In contrast to episodic hypoxia, 9 or 20 min of continuous hypoxia failed to elicit LTF of phrenic burst amplitude (4.7 ± 5.1 and 10.1 ± 10.2% 60 min following 9 and 20 min hypoxia, respectively; P > 0.05; 1, 2). Continuous hypoxia also failed to elicit hypoglossal LTF (-11.1 ± 5.6 and 19.3 ± 12.2% 60 min following 9 and 20 min hypoxia, respectively; P > 0.05; Fig. 2). Phrenic burst amplitude following episodic hypoxia was significantly greater than values following 9 or 20 min continuous hypoxia at all time points (P < 0.05). Significant LTF of burst frequency was not observed following 9 or 20 min continuous hypoxia (+4.6 ± 1.3 and +5.1 ± 2 bursts min⁻¹ 60 min following 9 and 20 min hypoxia, respectively; P > 0.05; Fig. 3). The increase in burst frequency following episodic hypoxia was significantly greater than that following 9 min of continuous hypoxia at all time points (P < 0.05), or that following 20 min continuous hypoxia at 30 and 60 min post-hypoxia (P < 0.05). Thus, LTF of phrenic burst amplitude and frequency is a unique property of episodic hypoxia.

DISCUSSION

Long-term facilitation of phrenic motor output has a profound dependence on the pattern of hypoxic exposure in anaesthetized rats. Episodic hypoxia elicits LTF of phrenic amplitude and frequency, whereas continuous hypoxia of an equal cumulative duration does not. These results are consistent with previous reports in awake goats from different investigators, in which episodic, but not continuous hypoxia elicited LTF (Dwinell et al. 1997; Turner & Mitchell, 1997). Other lines of evidence suggest that continuous and episodic hypoxia have differential impact on ventilatory control. Chronic sustained (continuous) hypoxia elicits ventilatory acclimatization, a process thought to be primarily mediated by effects on the carotid body (Bisgard & Neubauer, 1995; Powell et al. 1998). Similarly, continuous, but not episodic hypoxia increases phosphorylation of the cyclic AMP response element binding protein in the carotid body (Wang et al. 2000). Thus, continuous hypoxia may initiate a cascade of cell signalling and gene expression events in the carotid body that enhances carotid body sensitivity, thus giving rise to ventilatory acclimatization. In contrast, episodic hypoxia elicits LTF, which is believed to be a central neural mechanism that arises from serotonin-dependent effects on the respiratory network. Our working hypothesis is that episodic hypoxia elicits repetitive serotonin release and subsequent 5-HT_2A receptor activation on respiratory motoneurons, initiating long-lasting effects on phrenic burst amplitude. However, given that episodic hypoxia also elicits LTF in respiratory frequency (albeit smaller than amplitude effects), it is possible that a component of LTF involves facilitation at the level of medullary respiratory neurons associated with rhythm generation. Collectively, available evidence suggests that the pattern of carotid body stimulation is important in determining the type of plasticity elicited, although the detailed cellular mechanisms distinguishing episodic and sustained hypoxia in the CNS and carotid body chemoreceptors remain to be elucidated.

We previously reported that one 5 min hypoxic episode is not sufficient to elicit LTF in anaesthetized rats (Bach et al. 1999). On the other hand, an earlier report suggested that one episode (10 min) of hypoxia is sufficient to elicit LTF in anaesthetized cats (Millhorn et al. 1980a). This earlier report is based on observations from only one cat and may not represent the response of a population of cats. Furthermore, the authors did not measure blood gases or pH before, during or after hypoxia, leaving open the possibility that inadequate control of blood gases or pH could account for the apparent LTF in this case. Alternatively, cats may be different from rats or goats and exhibit LTF following continuous hypoxia.

There are numerous accounts in the neurobiology literature concerning the importance of stimulation pattern in eliciting plasticity. For example, in Aplysia, both episodic and prolonged continuous (> 1 h) serotonin application elicit LTF (Montarolo et al. 1986; Clark & Kandel, 1993; Emptage & Carew, 1993; Ghirardi et al. 1995; Zhang et al. 1997). However, brief (25 min) continuous serotonin applications are less likely to elicit long-term facilitation (Mauelshagen et al. 1998). Intracellular signalling pathways thought to mediate synaptic facilitation in Aplysia are also sensitive to the pattern of serotonin application (Müller & Carew, 1998; Yanow et al. 1998). Episodic serotonin (4-5 pulses, 5 min duration) elicits sustained activation of protein kinase A at intermediate (1 h) and long (20 h) time intervals, whereas continuous (90 min) serotonin application induces persistent protein kinase A activation only at long time intervals (Müller & Carew, 1998). Furthermore, episodic (5 x 5 min episodes) but not continuous serotonin (90 min) increases translational protein synthesis 2 h following stimulation in the isolated pleural ganglia of Aplysia (Yanow et al. 1998). Another example of pattern sensitive neuroplasticity is odour avoidance learning in Drosophila (Yin et al. 1994, 1995; Tully, 1994); spaced training trials give rise to a protein synthesis-dependent long-term memory, whereas continuous training does not (Tully et al. 1994). Thus, the pattern dependence of hypoxia-induced respiratory LTF in rats provides another example of neuroplasticity elicited by episodic, but not continuous stimuli. Respiratory LTF elicited by episodic hypoxia may represent a highly useful and generalizable model for in vivo studies of central neural plasticity. Given the clear physiological significance of long-lasting changes in respiratory motor output, this model of plasticity may have inherent advantages not often found in other frequently investigated models of plasticity.

Acknowledgements

These experiments were supported by NIH grants HL 53319 and HL 36780. Tracy Baker was supported by NIH Training Grant HL 07654.