Long-term facilitation of ventilation following acute continuous hypoxia in awake humans during sustained hypercapnia

Abstract

In awake humans, long-term facilitation of ventilation (vLTF) following acute intermittent hypoxia (AIH) is only expressed if CO₂ is maintained above normocapnic levels. vLTF has not been reported following acute continuous hypoxia (ACH) and it is not known whether this might be unmasked by elevated CO₂. Twelve healthy participants completed three trials. In all trials end-tidal pressure of CO₂ was elevated 4–5 mmHg above normocapnic levels. During Trial 1 (AIH) participants were exposed to eight 4 min episodes of hypoxia. During Trial 2 (ACH) participants were exposed to continuous hypoxia for 32 min. In Trial 3 (Control) participants were exposed to euoxia throughout. To assess the contribution of the carotid body (CB) in observed ventilatory responses, CB afferent discharge before and after each trial was transiently inhibited with hyperoxia. Minute ventilation () increased following all trials, but was significantly greater in Trials 1 and 2 when compared with Trial 3 (Trial 1: 4.96 ± 0.87, Trial 2: 5.07 ± 0.7, Trial 3: 2.55 ± 0.98 l min⁻¹, P < 0.05). Hyperoxia attenuated to a similar extent in baseline and recovery in all trials (Trial 1: 3.0 ± 0.57 vs. 3.27 ± 0.68, Trial 2: 1.97 ± 0.62 vs. 2.56 ± 0.62, Trial 3: 2.23 ± 0.49 vs. 2.15 ± 0.55 l min⁻¹, P > 0.05). Data are means ± SEM. In awake humans with elevated CO₂, ACH evokes a sustained increase in ventilation that is comparable to that evoked by AIH. However, a gradual positive drift in ventilation in response to elevated CO₂ accounts for approximately half of this apparent vLTF. Additionally, our data support the view that the CB is not directly involved in maintaining vLTF.

Key points

• In awake humans, when CO₂ is maintained above normal levels, exposure to acute intermittent hypoxia causes a sustained elevation in ventilation that persists when normoxic breathing is resumed.

• In this study we have demonstrated that when CO₂ is maintained above normal levels, exposure to acute continuous hypoxia also causes a sustained elevation in ventilation when normoxic breathing is resumed.

• This sustained elevation in ventilation following both acute intermittent hypoxia and acute continuous hypoxia is maintained by mechanisms other than increased activity of the carotid body.

• These results help develop our understanding of respiratory control in humans and may aid future development of treatments for respiratory control disorders, such as obstructive sleep apnoea.

Introduction

Long-term facilitation of ventilation (vLTF) is a sustained elevation in minute ventilation () that is initiated upon resumption of normal inspired O₂ levels following episodic chemical stimulation with acute intermittent hypoxia (AIH). In addition to vLTF, respiratory LTF has been expressed in animals as an increase in phrenic (pLTF) and hypoglossal nerve activity (hLTF), and muscle activity of their respective innervations, diaphragm (dLTF) and genioglossus (gLTF) (Mateika & Sandhu, 2011). Interestingly, although pLTF is unaffected, hLTF and gLTF can also be initiated by episodic mechanical stimulation mediated via the vagus nerve (Tadjalli et al. 2010). However, vLTF and gLTF, the two forms of respiratory LTF that can be safely measured in humans, are not expressed during wakefulness (McEvoy et al. 1996; Jordan et al. 2002; Mateika et al. 2004; Khodadadeh et al. 2006; Diep et al. 2007) unless CO₂ levels are elevated and maintained above normocapnic levels throughout AIH and during recovery (Harris et al. 2006; Wadhwa et al. 2008; Lee et al. 2009; Gerst et al. 2011).

The carotid body (CB) is the primary O₂ sensor in mammals (Kumar & Prabhakar, 2012) and increased afferent discharge during hypoxia is considered as a logical initiating step in the development of respiratory LTF. Indeed, episodic stimulation of the cut carotid sinus nerve of rodents can initiate respiratory LTF (Millhorn et al. 1980a; Hayashi et al. 1993). However, respiratory LTF is sustained long after cessation of carotid sinus nerve stimulation suggesting that it is a central mechanism(s) that maintains vLTF rather than CB afferent discharge (Millhorn et al. 1980a; Hayashi et al. 1993). Furthermore, direct recordings of rodent (in and ex vivo) CB discharge show a return to baseline levels immediately upon each re-oxygenation following AIH (Peng et al. 2003) and respiratory LTF is evident in rodents following AIH despite CB inhibition with hyperoxia (Bach & Mitchell, 1996; Baker & Mitchell, 2000; Xing & Pilowsky, 2010). Thus, although strong evidence exists to suggest CB afferent discharge plays a primary role in initiating respiratory LTF during AIH in rodents, it is generally accepted that it does not have an active role in maintaining respiratory LTF beyond hypoxic cessation.

In humans, elevated sympathetic nervous activity during AIH or acute continuous hypoxia (ACH) is sustained for up to 3 h following resumption of normal inspired O₂ levels (Morgan et al. 1995; Xie et al. 2000, 2001; Cutler et al. 2004a,b; Leuenberger et al. 2005; Tamisier et al. 2005; Querido et al. 2010). It has recently been demonstrated that this sympathetic LTF following ACH is transiently attenuated during brief inhibition of CB afferent discharge with hyperoxia (Querido et al. 2010). Because the pattern of sympathetic activity in humans (Seals et al. 1993) and rodents (Dick et al. 2004) is tightly coupled to that of the central respiratory drive we hypothesised that in awake humans, the CB discharge necessary for maintaining the full expression of sympathetic LTF may also maintain respiratory LTF following ACH and AIH. Currently, there are no available data to specifically link CB activity in the maintenance of vLTF in humans and this remains to be determined. The absence of vLTF following ACH may simply be because ACH is a stimulus that does not elicit respiratory LTF in awake humans. This would be in agreement with previous animal research that has failed to link ACH with respiratory LTF (Dwinell et al. 1997; Turner & Mitchell, 1997; Baker & Mitchell, 2000; Mitchell et al. 2001b; McKay et al. 2004). Alternatively, vLTF following ACH may not have been evident in previous human studies (Morgan et al. 1995; McEvoy et al. 1996; Xie et al. 2001; Tamisier et al. 2005; Querido et al. 2010) because CO₂ levels were not maintained above normocapnic levels as previously shown to be a requirement during AIH (Harris et al. 2006).

We aimed to examine whether ACH of the same total duration of hypoxia as AIH was able to elicit vLTF when CO₂ levels were maintained above normocapnic levels. In addition, by transiently inhibiting CB discharge with brief hyperoxic exposures we aimed to investigate whether CB discharge was involved in the maintenance of vLTF following AIH and ACH.

Methods

Ethical approval

After receiving detailed information on the procedures and risks all participants gave written consent to take part. The study was performed according to the latest revision of the Declaration of Helsinki and was approved by the local ethics committee (University of Birmingham ethical review committee).

Participants

Twelve healthy male subjects participated in the study. All subjects were non-smokers, had no history of cardiovascular, respiratory and metabolic disease and they were not taking any medication.

Protocol

Participants visited the laboratory for three experimental trials >24 h apart: Trial 1, AIH; Trial 2, ACH; Trial 3, Control. Prior to experimental visits participants also undertook a preliminary visit to familiarise themselves with the instrumentation and exposure to the various gas mixtures. For all experimental trials participants were asked to refrain from alcohol consumption and moderate to vigorous exercise for 24 h, and caffeine intake for 12 h, prior to arriving at the laboratory. In addition all experimental trials were performed at least 3 h after food consumption and the actual duration was matched between experimental trials for each individual. All trials were performed at the same time of day and were performed in a randomised order.

Preliminary visit

On arrival at the laboratory participants positioned themselves comfortably on a couch in the supine position where they were exposed to two 4 min episodes of hypoxia during elevations in CO₂ as would be experienced in experimental trials. Instrumentation for this visit was identical to all other experimental visits as described below (‘Instrumentation’).

Trial 1

Once positioned comfortably on the couch participants breathed room air for 15 min in order to establish normocapnic baseline, followed by 1 min exposure to hyperoxia. Subsequently, end-tidal partial pressure of CO₂ () was elevated by 4–5 mmHg above normocapnic levels and it was maintained at this level until the start of normocapnic recovery as described below. During the elevation in , partial end-tidal pressure of O₂ () was maintained at a euoxic level (100 mmHg), except during hypoxic and hyperoxic exposures. After 20 min of acclimatisation to the newly elevated (hypercapnic baseline), participants were exposed to 1 min of hyperoxia followed by an additional 2 min recovery to restore euoxia. Subsequently, participants experienced eight 4 min episodes of hypoxia each separated by 4 min periods of euoxia. Following the final hypoxic episode, hypercapnic recovery consisted of four 5 min periods (R1–R4), each followed by 1 min of hyperoxia. After an additional 2 min recovery to restore euoxia following the final hyperoxia, inspired CO₂ was removed to enable 15 min of normocapnic recovery, followed by a final 1 min exposure to hyperoxia. See Fig. 1 for an illustration of the protocol.

Figure 1.

Schematic diagram demonstrating protocols performed in Trials 1, 2 and 3. Arrows denote 1 min exposures to 100% inspired O₂.

Trial 2

Trial 2 was identical to Trial 1, except that the eight 4 min hypoxic episodes were replaced by one continuous hypoxic episode of the same total duration (32 min). See Fig. 1 for an illustration of the protocol.

Trial 3

In Trial 3 euoxia was maintained throughout the entire protocol and was otherwise identical to Trial 1, thus serving as a control for the sustained elevation in CO₂ experienced in Trial 1 and Trial 2. See Fig. 1 for an illustration of the protocol.

Instrumentation

To measure ventilation, participants wore a comfortable face mask that allowed breathing through either the mouth or the nose (Hans Rudolph 7450 Series V2 oro-nasal mask, total deadspace 100–125 ml depending on facemask size). The facemask was connected to the gas supply with a short flexible tube that allowed small head movements without lifting of the facemask seal. A good seal was ensured at all times by careful observation of gas waveforms on the computer monitor. End-tidal gas was sampled continuously from a catheter within the facemask and it was analysed using a mass spectrometer (AirSpec 2000, Case Scientific, London, UK). End-tidal profiles were generated using a computerised dynamic end-tidal forcing system, with end-tidal gas composition recorded each breath and compared with desired values by a computer-controlled, fast gas-mixing system. Using an integral-proportional feedback control system, deviations of actual values from the desired values were used to modify inspired gas mixtures on a breath by breath basis. The control scheme has been described in detail previously (Robbins et al. 1982). Inspired gases were heated and humidified. Respiratory volumes were measured by use of a turbine and ventilation is reported in BTPS.

During hypoxia in Trial 1 and Trial 3, was reduced to 50 mmHg within 2–4 breaths. Upon completion of each of the eight hypoxic episodes in Trial 1 and the 32 min of continuous hypoxia in Trial 2, was returned to 100 mmHg within 1–2 breaths. The gas inspired during the exposures to hyperoxia consisted of 100% O₂ with additional CO₂ to maintain at the desired hypercapnic level. Hyperoxic exposures during normocapnic baseline and normocapnic recovery involved 100% inspired O₂ without any additional CO₂.

Throughout all trials arterial O₂ saturation () was continuously measured using a pulse oximeter, worn on the participant's ear lobe (Datex-Ohmeda 3900).

Ventilatory analyses

In all experimental trials, , breathing frequency (BF), tidal volume (TV), and were recorded continuously. In all trials, baseline ventilatory measurements were made during both the normocapnic and hypercapnic periods by averaging for the last 3 min of each period. Likewise, in all trials, ventilatory measurements were made every 5 min during both the hypercapnic and normocapnic recovery periods by averaging for the last 3 min of each 5 min segment. Ventilatory measurements during the intervention period in Trial 1 (AIH) were made for each hypoxic and euoxic interval by averaging for the last 2 min of each interval. In Trials 2 and 3, ventilatory measurements during the intervention periods (ACH and euoxia, respectively) were made every 4 min to match the measurement points in Trial 1.

To assess the effect of inhibition of CB afferent discharge, ventilatory measurements were averaged over the entire minute of each hyperoxic exposure and compared with the ventilatory measurements made during the immediately preceding period of euoxia. The initial two breaths of 100% inspired O₂ were excluded from the analysis to account for lung-to-CB circulation delay.

Statistics

A two-way analysis of variance (ANOVA) with repeated measures in conjunction with a post hoc Fisher's least significant difference test was used to assess whether ventilatory measurements (, BF, TV, , and ) during normocapnic and hypercapnic recovery were significantly different from their respective baselines within each trial and whether there were significant differences between trials. In addition, the same statistical approach was used to assess whether the hyperoxia-induced reduction in during hypercapnic recovery was significantly different from that during hypercapnic baseline within and between each trial. Finally, in Trial 1 a one-way ANOVA with repeated measures was used to assess whether there were significant differences between ventilatory measurements during the eight hypoxic episodes and also during the eight immediately preceding euoxic episodes. Differences were considered significant if P≤ 0.05.

Results

Participants

Twelve healthy male participants completed the experiment with age (mean ± SD) 23.5 ± 0.42 years, weight 77.3 ± 2.4 kg and height 180.8 ± 1.8 cm. The mean body mass index was 23.6 ± 0.56.

Gas control

Figure 2 shows the and that were achieved during all trials. The elevation in above normocapnic levels during the periods of hypercapnia in all trials was nearly identical (Trial 1: 4.6 ± 0.2, Trial 2: 4.6 ± 0.2, Trial 3: 4.4 ± 0.2 mmHg). was controlled with considerable precision during all hypercapnic periods including when ventilation was altered due to superimposed hypoxia or hyperoxia. was maintained at 100 mmHg during all periods of euoxia and at 50 mmHg during all periods of hypoxia. Inspiration of 100% O₂ was successful in elevating well above the threshold that is required to inhibit CB afferent discharge. Tables 1–4 list the average end-tidal partial pressures achieved in all trials.

Figure 2.

Minute averages for and for each trial. Black circles indicate hyperoxic exposures.

Table 1.

Averaged ventilatory and measurements during normocapnic baseline (NB) and normocapnic recovery (NR)

Time	NB	NR 5 min	NR 10 min	NR 15 min
Trial 1 (AIH)
TV	665 ± 29	671 ± 44	627 ± 30	598 ± 26*
	38.8 ± 0.6	37.1 ± 0.5*	37.4 ± 0.5*	37.4 ± 0.5*
	104 ± 0.8	109 ± 0.7*	105 ± 0.6	105 ± 0.6
	97.5 ± 0.2	97.5 ± 0.2	97.3 ± 0.2	97.3 ± 0.2
BF	16.9 ± 0.6	16.6 ± 0.9	17 ± 0.7	17.7 ± 0.7
Trial 2 (ACH)
TV	684 ± 27	683 ± 41	663 ± 30	647 ± 23
	38.9 ± 0.5	36.6 ± 0.5*	37.0 ± 0.5*	37.1 ± 0.6*
	103 ± 0.9	106 ± 0.7*	104 ± 0.6	104 ± 0.8
	97.3 ± 0.2	97.4 ± 0.1	97.1 ± 0.8	97.3 ± 0.2
BF	16.6 ± 0.8	16.8 ± 0.8	17.0 ± 0.7	17.1 ± 0.6
Trial 3 (Control)
TV	635 ± 29	626 ± 26	605 ± 27*	612 ± 22
	39.4 ± 0.5	37.6 ± 0.4*	38.1 ± 0.4*	38.3 ± 0.4*
	103 ± 1.0	107 ± 0.7*	104 ± 0.8	104 ± 0.8
	97.8 ± 0.2	97.6 ± 0.2	97.4 ± 0.2	97.3 ± 0.2
BF	17.0 ± 0.7	16.5 ± 0.5	17.1 ± 0.7	16.9 ± 0.7

^*Significantly different from normocapnic baseline.

Mean ± SEM; Tidal volume, TV; Arterial saturation in %, ; breathing frequency (per min), BF.

Table 4.

Averaged ventilatory and measurements for minute hyperoxic exposures during hypercapnic baseline (HB) and hypercapnic recovery (HR)

Time	HB	HR 5 min	HR 10 min	HR 15 min	HR 20 min
Trial 1 (AIH)
TV	1004 ± 75	1105 ± 74*	1111 ± 63*	1104 ± 73*	1149 ± 85*
	43.5 ± 0.5	43.5 ± 0.5	43.4 ± 0.5	43.4 ± 0.5	43.5 ± 0.5
	526 ± 14	544 ± 10	544 ± 11	545 ± 12	547 ± 11
	99.2 ± 0.2	98.9 ± 0.1	98.9 ± 0.1	98.8 ± 0.2	98.7 ± 0.1
BF	17.3 ± 0.7	18.4 ± 0.8	19.0 ± 0.7†	19.2 ± 0.8*	19.3 ± 0.8*†
Trial 2 (ACH)
TV	1057 ± 83	1147 ± 88*	1169 ± 81*	1193 ± 78*	1185 ± 80*
	43.6 ± 0.5	43.6 ± 0.4	43.5 ± 0.4	43.4 ± 0.4	43.6 ± 0.4
	504 ± 11	507 ± 12	524 ± 12	529 ± 9	532 ± 11
	98.8 ± 0.2	98.9 ± 0.2	99.1 ± 0.1	99.0 ± 0.2	99.0 ± 0.2
BF	17.3 ± 0.8	18.0 ± 0.8	18.9 ± 0.9*	18.4 ± 0.8*	19.2 ± 0.8*†
Trial 3 (Control)
TV	1065 ± 78	1067 ± 88	1091 ± 88	1120 ± 99	1102 ± 93
	43.8 ± 0.5	43.8 ± 0.5	43.9 ± 0.5	43.8 ± 0.4	43.7 ± 0.4
	509 ± 13	515 ± 15	522 ± 17	521 ± 16	521 ± 12
	99.1 ± 0.2	98.9 ± 0.2	99.1 ± 0.1	99.1 ± 0.2	99.2 ± 0.2
BF	16.7 ± 0.7	18.1 ± 0.8*	18.1 ± 0.9*	18.3 ± 0.9*	18.4 ± 0.9*

^*Significantly greater than baseline. †Significantly greater than hypercapnic recovery 5 min.

Mean ± SEM; Tidal volume, TV; Arterial saturation in %, ; breathing frequency (per min), BF.

Table 2.

Averaged ventilatory and measurements for hyperoxic exposures during normocapnic baseline (NB) and normocapnic recovery (NR)

Time	NB	NR
Trial 1 (AIH)
TV	650 ± 34	555 ± 34*
	39.7 ± 0.6	38.0 ± 0.5*
	433 ± 10	420 ± 11.9
	99.0 ± 0.2	98.8 ± 0.2
BF	15.1 ± 0.8	16.3 ± 0.7
Trial 2 (ACH)
TV	644 ± 33	610 ± 34
	39.4 ± 0.6	37.3 ± 0.6*
	418 ± 11	428 ± 10
	98.8 ± 0.2	98.9 ± 0.2
BF	15.7 ± 0.5	16.2 ± 0.8
Trial 3 (Control)
TV	614 ± 42	629 ± 46
	39.8 ± 0.7	38.4 ± 0.5*
	434 ± 14	420 ± 18
	99.1 ± 0.1	98.9 ± 0.2
BF	16.0 ± 1.2	14.8 ± 0.9

^*Significantly lower than normocapnic baseline.

Mean SEM; Tidal volume, TV; Arterial saturation in %, ; breathing frequency (per min), BF.

Ventilation

Euoxia

In Trial 1, was significantly elevated above hypercapnic baseline during the entire hypercapnic recovery period (Fig. 3). Increased was a product of a significant increase in both BF and TV (Table 3). Intriguingly, in Trial 2, during the entire hypercapnic recovery period was also significantly elevated above baseline to an extent equivalent to that of Trial 1 (Fig. 4) and was likewise a product of increased BF and TV (Table 3). Furthermore, in both trials, continued to increase during hypercapnic recovery and was significantly greater during the final quarter than in the initial quarter (Figs 3 and 4).

Figure 3. Trial 1 (AIH).

Black circles indicate average values for recorded during hypercapnic baseline, each hypoxic and euoxic interval and during hypercapnic recovery. Grey circles indicate average values for recorded during normocapnic baseline and during normocapnic recovery. White circles indicate average during hyperoxic exposures which immediately followed that of the euoxic time point positioned above (as indicated by a black circle). *Significantly greater than the respective baseline (i.e. euoxic or hyperoxic). †Significantly greater than during the corresponding hyperoxic exposure. ‡Significantly greater than at the start of hypercapnic recovery. §Significantly greater than the first hypoxic episode. Inset (left): tidal volume during each of the eight hypoxic episodes. Inset (right): breathing frequency during each of the eight hypoxic episodes. *Significantly greater than the initial hypoxic episode. The three plots show mean values. All error bars are SEM.

Table 3.

Averaged ventilatory and measurements during hypercapnic baseline (HB) and hypercapnic recovery (HR)

Time	HB	HR 5 min	HR 10 min	HR 15 min	HR 20 min
Trial 1 (AIH)
TV	1127 ± 78	1202 ± 84*	1229 ± 69*	1249 ± 74*	1282 ± 82*
	43.4 ± 0.5	43.4 ± 0.5	43.4 ± 0.5	43.4 ± 0.5	43.4 ± 0.5
	100 + 0.1	100 ± 0.18	100 ± 0.04	100 ± 0.1	100 ± 0.05
	97.5 ± 0.1	97.5 ± 0.2	97.3 ± 0.2	97.2 ± 0.2	97.3 ± 0.2
BF	18.2 ± 0.7	19.2 ± 0.7*	19.5 ± 0.8*	19.7 ± 0.8*	19.8 ± 0.8*
Trial 2 (ACH)
TV	1144 ± 65	1163 ± 61	1238 ± 72*	1262 ± 69*	1291 ± 67*†
	43.4 ± 0.4	43.3 ± 0.4	43.4 ± 0.4	43.4 ± 0.4	43.3 ± 0.4
	100 ± 0.1	100 ± 0.1	100 ± 0.04	100 ± 0.05	100 ± 0.03
	97.3 ± 0.2	97.5 ± 0.2	97.6 ± 0.2	97.5 ± 0.2	94.5 ± 0.2
BF	17.7 ± 0.7	19.7 ± 0.8*	19.7 ± 0.9*	19.4 ± 0.9*	19.6 ± 0.8*
Trial 3 (Control)
TV	1104 ± 65	1141 ± 80	1188 ± 83	1180 ± 79	1171 ± 86
	43.7 ± 0.4	43.7 ± 0.4	43.7 ± 0.5	43.7 ± 0.4	43.7 ± 0.4
	100 ± 0.1	100 ± 0.05	100 ± 0.1	100 ± 0.1	100 ± 0.1
	97.3 ± 0.1	97.6 ± 0.1	97.4 ± 0.1	97.6 ± 0.2	97.6 ± 0.2
BF	18.2 ± 0.9	19.0 ± 0.8*	19.2 ± 0.8*	19.4 ± 0.9*	19.3 ± 0.8*

^*Significantly greater than baseline. †Significantly greater than hypercapnic recovery 5 min.

Mean ± SEM; Tidal volume, TV; Arterial saturation in %, ; breathing frequency (per min), BF.

Figure 4. Trial 2 (ACH).

Black circles indicate average values for during hypercapnic baseline, continuous hypoxia and during hypercapnic recovery. Grey circles indicate average values for recorded during normocapnic baseline and during normocapnic recovery. White circles indicate average during hyperoxic exposures which immediately followed that of the euoxic time point positioned above (as indicated by a black circle). *Significantly greater than respective baseline (i.e. euoxic or hyperoxic). †Significantly greater than during the corresponding hyperoxic exposure. ‡Significantly greater than at the start of hypercapnic recovery.

However, in Trial 3, where participants were exposed to hypercapnia without hypoxia, during the hypercapnic recovery period was also significantly greater than hypercapnic baseline (Fig. 5). This increase in was predominantly mediated by a significant increase in BF without a noticeable change in TV (Table 3). Unlike Trial 1 and Trial 2, did not increase further during the hypercapnic recovery period but remained constant in Trial 3 (Fig. 5). The magnitude of the increase in from baseline to the final quarter of the recovery period was significantly greater in Trial 1 and Trial 2 (Fig. 6).

Figure 5. Trial 3 (Control).

Black circles indicate average values for during hypercapnic baseline, continuous euoxia and during hypercapnic recovery. Grey circles indicate average values for recorded during normocapnic baseline and during normocapnic recovery. White circles indicate average during hyperoxic exposures which immediately followed that of the euoxic time point positioned above (as indicated by a black circle). *Significantly greater than respective baseline (i.e. euoxic or hyperoxic). †Significantly greater than during the corresponding hyperoxic exposure.

Figure 6.

Change in from hypercapnic baseline to the final quarter of recovery, during euoxia (filled columns) and hyperoxia (open columns). *Significantly greater than control trial during euoxia. †Significantly greater than control trial during hyperoxia.

In Trial 1, during the later hypoxic episodes began to increase and as such the final hypoxic episode was significantly greater than that of the first (Fig. 3). A significant increase in BF was solely responsible for this augmentation in , as TV did not change between episodes (Fig. 3 insets). during each euoxic episode that immediately preceded the eight hypoxic episodes did not change over time.

In all trials during normocapnic recovery had returned to normocapnic baseline levels within the first or second minute and remained at this level throughout the entire 15 min period (Figs 3–5).

Hyperoxia

In all trials each exposure to hyperoxia significantly reduced (Figs 3–5). The hyperoxia-induced reduction in was greater during concurrent hypercapnia, when CB activity would be expected to be elevated above that of room air breathing (Figs 3–5).

In contrast to our hypothesis that CB afferent discharge would be actively involved in maintaining vLTF, inhibition of the CB with hyperoxia did not abolish or even attenuate the apparent vLTF. Hyperoxia reduced ventilation to the same extent before and following the intervention in each trial (i.e. AIH, ACH and Control) (Fig. 7) meaning that the increase in observed in all trials in the period following AIH, ACH or Control was independent of CB discharge (Fig. 6).

Figure 7.

Hyperoxia-induced reduction in during hypercapnic baseline (filled columns) and during the final quarter of hypercapnic recovery (open columns).

Discussion

The primary findings of this study are threefold: (1) in awake humans, exposure to ACH during hypercapnia induces a similar sustained elevation in ventilation to that of exposure to AIH; (2) approximately half of this elevation in ventilation may be attributed to vLTF as exposure to sustained hypercapnia causes a gradual increase in ventilation that is independent of hypoxic exposure; and (3) hyperoxia reduced ventilation to the same extent before and following AIH and ACH suggesting the CB is not actively involved in maintaining vLTF in awake humans.

Hypercapnia and respiratory LTF

Although respiratory LTF has been repeatedly demonstrated in a range of animal species using various preparations since the first publication in 1980 (Millhorn et al. 1980a), evidence for its existence in awake humans remained elusive until relatively recently despite the completion of numerous studies designed to investigate this phenomenon (Mateika & Sandhu, 2011). It was postulated that previous unsuccessful attempts to induce vLTF in humans may be due to CO₂ levels being below the central and peripheral chemoreflex thresholds and thus ventilation may be predominantly driven by arousal and/or behavioural stimuli (Harris et al. 2006). Under these conditions, mechanisms responsible for adaptations to peripheral or central chemoreceptor processes that may mediate vLTF could have been active during AIH, but the expression of vLTF would have been restrained by the low CO₂ levels. By raising 5 mmHg above normocapnic levels and maintaining this throughout AIH and recovery, vLTF was expressed in awake humans (Harris et al. 2006). A number of subsequent studies by the same group have since replicated these results and also demonstrated the existence of vLTF following AIH in women and in patients suffering from obstructive sleep apnoea (OSA) (Wadhwa et al. 2008; Lee et al. 2009; Gerst et al. 2011).

Using an almost identical AIH protocol and a similar level of elevated CO₂ as used by Harris et al. (2006) we demonstrated the appearance of vLTF of a similar magnitude in a population of young males. Furthermore, to the best of our knowledge we are the first to investigate whether ACH also evokes vLTF during elevated CO₂ levels in awake humans. In contrast to a substantial body of evidence in animal research showing that only AIH produces respiratory LTF we demonstrated an equivalent magnitude of vLTF with ACH to that following AIH. However, our results from Trial 3 (Control) also demonstrated a gradual but significant rise in ventilation during sustained elevations in CO₂ that was independent of the exposure to hypoxia. Although the increase in was considerably less, at approximately half of the increase demonstrated in Trial 1 and Trial 2 following AIH and ACH, we believe that the increase in ventilation due to sustained hypercapnia must account for part of the apparent vLTF following both AIH and ACH.

The possibility that the apparent vLTF following AIH is not exclusively dependent on the exposure to intermittent hypoxia but also on the gradual increase in ventilation to sustained hypercapnia was also considered by Harris et al. (2006). In their study, a subset of their population was exposed to sustained hypercapnia without exposure to hypoxia, but in contrast to our results, ventilation remained constant after baseline measurements suggesting that it was indeed AIH that mediated vLTF. However, a more recent paper has demonstrated that ventilation can also be significantly elevated in OSA patients following a slightly longer but lower intensity of sustained elevation in CO₂ without hypoxic exposure (Gerst et al. 2011). The significant increase in due to hypercapnia alone in our study occurred even though we extended the duration of the hypercapnic acclimatisation phase by 7 min from that used by Gerst et al. (2011) in order to allow longer for ventilation to plateau following the step increase in CO₂. The exact mechanism(s) responsible for this gradual rise in ventilation due to mild hypercapnia remains to be determined, although our data showing that hyperoxia reduced ventilation to the same extent at baseline and recovery suggest a central mechanism is responsible.

Respiratory LTF following AIH and ACH

Despite our results demonstrating that there is a significant gradual increase in ventilation to sustained hypercapnia that is independent of hypoxic exposure (Trial 3), this increase is significantly less than that occurring following the addition of AIH or ACH (Trial 1 and Trial 2). We believe these results demonstrate that in awake humans vLTF does manifest following exposure to both AIH and ACH during sustained hypercapnia albeit by a smaller magnitude than previously documented.

Our findings are in contrast to animal research that suggests that intermittent but not continuous exposure to hypoxia can evoke respiratory LTF. Respiratory LTF following AIH has been demonstrated in various anaesthetised and conscious animal species (Mateika & Sandhu, 2011) but not following ACH (Dwinell et al. 1997; Turner & Mitchell, 1997; Baker & Mitchell, 2000; Mitchell et al. 2001b; McKay et al. 2004; Tadjalli et al. 2007).

The magnitude of vLTF following AIH and ACH in our study was equivalent and comparable to that of previous studies demonstrating vLTF following AIH in healthy awake humans (Harris et al. 2006; Wadhwa et al. 2008; Lee et al. 2009). Small differences in the magnitude of vLTF between studies are inevitable due to differences in protocols such as the total number of hypoxic episodes, level of sustained CO₂, duration of recovery and differences in the participant population. However, one noticeable difference worth discussing is that the time course of vLTF development following AIH and ACH in our study is different to that previously demonstrated in studies following AIH (Harris et al. 2006; Wadhwa et al. 2008; Lee et al. 2009; Gerst et al. 2011). In these previous studies, does not change during hypercapnic recovery but remains stable and elevated above baseline having progressively increased during each euoxic episode that follows hypoxic exposures. In contrast the increase in during euoxic episodes was more modest in our study, with only a noticeable increase following the first hypoxic exposure and then again during the final two hypoxic exposures. Furthermore, there was a gradual rise in during hypercapnic recovery following both AIH and ACH and as such was significantly greater during the final quarter of hypercapnic recovery compared with the first. No change in during hypercapnic recovery in Trial 3 suggests the increase in during hypercapnic recovery following AIH and ACH is due to hypoxic exposure rather than a continued increase in to the sustained hypercapnia. Interestingly, OSA patients have been shown to both demonstrate (Lee et al. 2009) and not demonstrate this increase in (Gerst et al. 2011) during hypercapnic recovery. Although the pattern of increasing vLTF during hypercapnic recovery shown in our study is different to that of previous studies demonstrating vLTF development in healthy humans, this pattern is more common in animals that express respiratory LTF following AIH. In anaesthetised rats phrenic and hypoglossal nerve and diaphragmatic muscle LTF as well as vLTF demonstrate a progressively augmenting pattern during recovery from AIH (Bach & Mitchell, 1996; Baker & Mitchell, 2000; Mitchell et al. 2001a; Olson et al. 2001; Ryan & Nolan, 2009).

In addition to the expression of vLTF following AIH the significant increase in during the last hypoxic episode compared with the initial exposure is indicative of an additional form of respiratory plasticity termed progressive augmentation of ventilation. A progressive increase in BF during hypoxic episodes was responsible for this elevation in as TV did not change. Progressive augmentation is only observed in humans when CO₂ levels are maintained above normocapnic levels (Harris et al. 2006) and in a range of animal species when hypoxic-induced hypocapnia is prevented (Mateika & Narwani, 2008).

The role of the CB in maintaining vLTF

We hypothesised that CB afferent discharge during recovery from AIH and ACH is actively involved in maintaining vLTF in awake humans. In contrast to this hypothesis, hyperoxia reduced ventilation to a similar extent during hypercapnic baseline and recovery in both trials suggesting that the CB does not contribute to vLTF as shown by the increase in ventilation following AIH and ACH.

Our results suggest that mechanisms independent of the CB maintain vLTF in awake humans and, accordingly, agree with a substantial body of evidence from animals. The demonstration of respiratory LTF following AIH during CB inhibition with hyperoxia as shown in our study has previously been demonstrated in rats (Bach & Mitchell, 1996; Baker & Mitchell, 2000; Xing & Pilowsky, 2010). Although these studies were not designed specifically to investigate the involvement of the CB in maintaining respiratory LTF, rats were exposed to 50% (Bach & Mitchell, 1996; Baker & Mitchell, 2000) or 100% (Xing & Pilowsky, 2010) inspired O₂ throughout baseline and recovery, which would be expected to inhibit CB afferent discharge. However, despite this, phrenic discharge was shown to increase progressively during recovery and remained elevated for >1 h. This does not preclude the possibility of a stimulus to the carotid body that arises during AIH, but is not inhibited by hyperoxia. To date, however, there is no evidence of any such residual, oxygen-independent component arising during AIH. Of course, if such a stimulus did exist then a role for the CB cannot be excluded. In addition, the possibility of a necessary input from the CB in LTF development is further eroded by the finding that episodic stimulation of the cut carotid sinus nerve to mimic episodic activation of the CB during AIH initiates pLTF which is maintained for >30–90 min, following the total absence of any CB input to the central nervous system (CNS) (Millhorn et al. 1980a; Hayashi et al. 1993).

Direct recordings of in and ex vivo CB afferent discharge in rats not previously exposed to AIH demonstrated that the elevated afferent discharge during hypoxic episodes promptly returned to baseline levels upon re-oxygenation and remained constant throughout recovery (Peng et al. 2003). However, rodents previously exposed to chronic intermittent hypoxia (CIH), aimed to mimic nightly IH experienced by OSA patients, evoked a progressive increase in CB afferent discharge during successive hypoxic episodes which stayed elevated for >1 h in recovery. This sustained elevation in CB afferent discharge following AIH that is only evident in CIH rats has been termed sensory LTF and is reactive oxygen species dependent as antioxidant treatment inhibits its manifestation (Peng et al. 2003). All participants in our study reported no current or previous history of sleep apnoea and thus can be considered naive to previous IH exposure. As such, our healthy participants would not be expected to develop sensory LTF following AIH. Using an almost identical AIH protocol as used in our study, Lee et al. (2009) demonstrated that OSA patients produced greater vLTF than healthy controls. Interestingly, antioxidant treatment attenuated this augmentation in vLTF back to the same level of vLTF as the healthy controls where antioxidant treatment had no effect. Presumably this was consequent to the abolishment of sensory LTF by antioxidant treatment in OSA patients that did not manifest in healthy controls who had previously not been exposed to CIH. However, the contribution that sensory LTF makes to respiratory LTF remains unknown and requires future studies utilising techniques such as hyperoxic inhibition of the CB.

CB afferent discharge may reach the presympathetic neurons without intermediate communication in the respiratory network (Guyenet, 2000). Thus, it is possible that augmented CB afferent discharge (sensory LTF) or enhanced integration of CB afferent discharge at the CNS could mediate sympathetic LTF as previously shown by Querido et al. (2010) but not mediate a parallel increase in ventilation (respiratory LTF). We did not record sympathetic activity in our study and therefore we cannot conclude whether the hyperoxic exposures that failed to attenuate vLTF also attenuated sympathetic LTF.

CB-independent mechanisms maintaining vLTF in AIH and ACH

Our findings that the CB does not actively maintain respiratory LTF suggest that a central mechanism may be involved. Whilst there is no evidence to suggest that an increase in cerebral blood flow due to relatively short, acute exposures to hypoxia and/or hypercapnia adapts towards control levels (Poulin et al. 1996) causing a relative accumulation of CO₂ in the cerebral circulation, a possible stimulus could be a lower pH in the cerebrospinal fluid caused by impaired regulation of [H⁺] during hypoxia (Duffin, 2005). Thus, vLTF may reflect the time course of central pH recovery following exposure to systemic hypoxia. Whilst we have no data to support or refute this possibility, this would be an important consideration for any mechanistic explanation of vLTF. Similarly, a contribution from altered mechanical feedback during a period of AIH or ACH might induce a vagal-mediated form of LTF, although the evidence to date suggests that this may be confined to an effect delimited to the tongue via hypoglossal LTF and genioglossus LTF rather than to an effect upon ventilation mediated through phrenic LTF (Tadjalli et al. 2010).

Animal studies provide evidence for a serotonin-dependent mechanism initiating and maintaining respiratory LTF (Mahamed & Mitchell, 2007; MacFarlane et al. 2008). pLTF requires spinal serotonin release (Baker-Herman & Mitchell, 2002) and activation of serotonin receptors (5-HT₂) on respiratory motoneurons in the spinal cord (Kinkead & Mitchell, 1999; Fuller et al. 2001) which initiate the synthesis of spinal proteins such as brain-derived neurotrophic factor that subsequently maintains pLTF (Baker-Herman et al. 2004). As stimulation of the cut carotid sinus nerve initiates the same serotoninergic-dependent pathway of pLTF as AIH (Millhorn et al. 1980b; Bach & Mitchell, 1996), the above model for serotonin-dependent respiratory LTF does not imply that the CB is not necessary for initiating respiratory LTF, rather that spinal protein synthesis maintains respiratory LTF as opposed to any continued input from the CB. Additionally, inhibition of 5HT₂ receptors prior to AIH exposure but not immediately following AIH abolished pLTF (Fuller et al. 2001). Although this serotonin-dependent mechanism for AIH-induced respiratory LTF has been well described in animal studies (Mahamed & Mitchell, 2007; MacFarlane et al. 2008), ACH does not elicit the same pattern of respiratory LTF (Dwinell et al. 1997; Turner & Mitchell, 1997; Baker & Mitchell, 2000; Mitchell et al. 2001b; McKay et al. 2004) and only intermittent but not continuous application of serotonin or 5-HT₂ receptor agonists evokes pLTF (Lovett-Barr et al. 2006; MacFarlane & Mitchell, 2009) and hLTF (Bocchiaro & Feldman, 2004). Therefore, animal studies can only provide us with limited clues as to the possible mechanism(s) for vLTF following ACH in awake humans as shown in our study. However, a recent study reported an equivalent magnitude and pattern of pLTF following ACH to that of AIH, when serine/threonine protein phosphatase in the spinal cord of rats exposed to ACH was inhibited with okadaic acid (Wilkerson et al. 2008). Furthermore, in the same study, intravenous infusion of the broad spectrum serotonin inhibitor, methysergide, blocked pLTF in rats exposed to AIH or ACH with okadaic acid. Okadaic acid administration alone did not evoke pLTF and these results thus suggest that the equivalent pLTF following AIH and ACH with okadaic acid was mediated by the same serotonin-dependent mechanism. Reactive oxygen species (ROS) are known to inhibit many protein phosphatases and observations from various studies would suggest that AIH-induced ROS may inhibit these okadaic acid-sensitive serine/threonine phosphatases and thus remove their inhibitory restraint on pLTF which does not occur during ACH, a stimulus not known to induce ROS generation (Wilkerson et al. 2007). Indeed, MacFarlane & Mitchell (2009) showed that intermittent serotonin application to the spinal cord induces pLTF but was abolished with NADPH oxidase inhibition. An attractive explanation for the appearance of vLTF following ACH in our study is that the existence or regulation of these specific okadaic acid-sensitive serine/threonine phosphatases during hypoxic exposure is different in rats and thus the serotonin-dependent mechanism for respiratory LTF during ACH is not constrained in humans.

Although serotonin-induced pLTF (Lovett-Barr et al. 2006; MacFarlane & Mitchell, 2009) and hLTF (Bocchiaro & Feldman, 2004) require intermittent activation of serotonin receptors, different respiratory pools may not require an intermittent pattern. Indeed, continuous spinal serotonin application is sufficient to induce LTF in thoracic (intercostals) inspiratory motor output in the neonatal rat brainstem/spinal cord preparation (Lovett-Barr et al. 2006). As ventilation is a product of combined respiratory output from various respiratory motor pools it is plausible that significant thoracic LTF following ACH could in part mediate the vLTF demonstrated in our study. Furthermore, humans may demonstrate a variation from the situation in the rat in which respiratory motor pools express serotonin-induced LTF to continuous activation of serotonin receptors and thus pLTF or hLTF may be evoked by ACH.

If hypoxic-induced CB activity was the only mechanism capable of initiating respiratory LTF then CB denervation should completely abolish respiratory LTF when in fact it has been shown in rats to only attenuate AIH-induced pLTF (Bavis & Mitchell, 2003; Sibigtroth & Mitchell, 2011). Therefore, although pLTF following AIH and episodic stimulation of the carotid sinus nerve is completely abolished with methysergide (Millhorn et al. 1980b; Bach & Mitchell, 1996) suggesting a serotonin-dependent mechanism is capable of mediating the full expression of pLTF in rats with intact CBs, it would appear that it is a redundant system as an additional central mechanism(s) must exist for pLTF to develop in CB-denervated rats. The exact central mechanism(s) remains unknown but in the absence of chemoreceptor feedback from the CB during hypoxia the initiating stimulus must be central hypoxia. It is therefore possible that central hypoxia may have been responsible for part or all of the vLTF following AIH and more likely following ACH where the 32 min of sustained hypoxia may have allowed greater development of central hypoxia than during the 4 min episodes in AIH.

A further intriguing possibility is that the vLTF following AIH and ACH in our study represents two different forms of respiratory LTF. The majority of animal studies investigating respiratory LTF use anaesthetised rats and demonstrate a progressive increase in pLTF or vLTF following AIH that does not peak until >30 min and is sustained for >1 h (Mitchell et al. 2001a). vLTF following AIH in awake unrestrained rats follows an equivalent pattern to anaesthetised rats (Olson et al. 2001). However, under these awake and unrestrained conditions that most closely mimic the experimental conditions of our study, ACH evoked a significant sustained elevation in ventilation that peaked immediately following ACH and then declined back to baseline levels within 40 min. It was suggested that the differences in the pattern of the elevated ventilation following AIH and ACH might arise as they represent different forms of LTF (Olson et al. 2001). In contrast, there did not appear to be noticeable differences in the pattern of the elevated ventilation following AIH and ACH in our study. It is possible that differences would have been displayed if our relatively short recovery period was extended.

Implications

Inducing respiratory LTF in OSA patients prior to sleep has been considered a possible treatment option as it may help to maintain upper airway patency and thus prevent repetitive nocturnal collapse, although there remains a concern that respiratory LTF may in fact promote upper airway instability (Mateika & Narwani, 2008). Exposing patients to hypoxia prior to sleep is not a realistic treatment option but a clearer understanding of the mechanisms involved may guide the development of feasible treatments that can mimic hypoxia-induced respiratory LTF. ACH has long been considered to be a stimulus that cannot evoke respiratory LTF in animals or humans, but our data show that the opposite might be the case in awake humans and thus highlight the need for further studies of respiratory LTF in humans.

Conclusions

We have demonstrated that, in awake humans, ACH evokes a magnitude and pattern of sustained elevation in ventilation equivalent to that of AIH and that part of this apparent vLTF is due to a gradual ventilatory ‘drift’ induced by the concomitant hypercapnia. Furthermore, by demonstrating that CB inhibition with hyperoxic exposure does not attenuate vLTF we provide further corroborative evidence against a role for the CB in actively maintaining vLTF.

Glossary

ACH

acute continuous hypoxia

AIH

acute intermittent hypoxia

BF

breathing frequency

CB

carotid body

CIH

chronic intermittent hypoxia

gLTF

genioglossal long-term facilitation

hLTF

hypoglossal long-term facilitation

OSA

obstructive sleep apnoea

pLTF

phrenic long-term facilitation

TV

tidal volume

vLTF

ventilatory long-term facilitation

Author contributions

All experiments were performed in the laboratory of G.M.B. Conception and design of research: H.S.G., P.K. and G.M.B. Data were collected by H.S.G., K.P. and G.M.B., and analysed by H.S.G. and G.M.B. The manuscript was written by H.S.G and G.M.B with assistance from P.K. All authors have read and approved the final version of the manuscript.