Professional Documents
Culture Documents
047597
© The Physiological Society 2003 www.jphysiol.org
Physiology in Press
As first demonstrated by Haldane & Priestley (1905), the mechanisms have been calibrated to ensure that ◊E rises in a
increase in metabolism that occurs as a result of muscular manner so precisely matched to ◊CJ.
exercise is matched exactly by an increase in pulmonary
In an attempt to overcome this problem, Somjen (1992)
ventilation (◊E). They proposed that during exercise, an
has hypothesised that feedforward control is a learned
increase in alveolar PCJ (PA,CJ), and hence arterial PCJ
response. He proposed that the brain ‘knows’ exactly how
(Pa,CJ), would stimulate the chemoreceptors, and that this
much O2 is demanded and CO2 produced, by the level of
would increase ◊E, thus providing the link between ◊E and
exercise being undertaken and has learned to anticipate
metabolic CO2 production (◊CJ). However, subsequent
Journal of Physiology
breathing rest. Outside the laboratory, subjects were requested to was able to maintain for a full minute. Heart rate was monitored
refrain from taking exercise other than the minimum required to throughout the test to check that the expected maximal value was
travel around (slow walking and cycling on level ground). Subjects reached, and plots of PET,CJ vs. work rate were inspected visually
were equipped with a pulse meter to warn them if they were to check that anaerobic threshold had been achieved several
exerting themselves to too great a degree. Subjects tended to only minutes prior to the end of the test (i.e. plots became alinear).
wear the monitor for the first couple of days of each protocol, until
During training, the inspired gas composition was controlled
they became accustomed to maintaining this minimal level of
manually using rotameters (Roxspur Measurement & Control,
exertion.
Bramley, UK). For protocol EX + CO2, values for inspiratory PCJ
In order to determine whether the subjects’ perception of and PJ were chosen that would raise PET,CJ by 8 Torr above the
breathlessness differed between the three training protocols they normal level during exercise and lower PET,J to 8 Torr below the
were asked to rate their sensations of breathlessness during the normal level during exercise. PET,J was lowered in addition to
final minute of each bout of training, according to a modified raising PET,CJ in order that the stimulus would mimic the effects of
Borg scale (Borg, 1982). This scale associates numerical ‘scores’ hypoventilation as closely as possible. With a background flow of
with verbal expressions; in this case, descriptions of the strength of air, the flows of CO2 and N2 were adjusted until the desired end-
sensations of breathlessness, which ranged from zero (‘none at tidal gas values were achieved. For protocol EX, subjects breathed
all’) to 10 (‘very, very heavy’). A card displaying the modified Borg air throughout the training periods. For protocol CO2, values for
scale was held in front of the subject, who was asked to point at the inspiratory PCJ and PJ were chosen that would raise PET,CJ by
score associated with the expression that best described their 8 Torr above the normal level at rest and lower PET,J to 8 Torr
sensations of breathlessness at that moment. below the normal level at rest.
The respiratory response to exercise before and after training was Measurements of the ventilatory response to exercise on days
assessed in two ways. In the first, subjects undertook an 8 min before and after the training periods were made using a turbine
period of exercise breathing air, which leaves the biological volume transducer (Cardiokinetics) to measure respiratory
feedback loop between CO2 and ventilation intact (closed-loop volumes (Howson et al. 1986). All data were logged to a personal
measurements). In the second, subjects undertook an identical computer. Two sets of measurements were made under closed-
period of exercise, but with their PET,CJ regulated at a fixed level, loop conditions followed by two sets of measurements under
which opens the biological feedback loop to allow measurement open-loop conditions. For each measurement, there was an initial
of ◊E at a fixed level of PCJ (open-loop measurements). Measure- period of 5 min at rest, followed by 8 min of exercise, and then a
ments were undertaken the day before training (day _1), on the final 5 min of rest, throughout which the subject was seated on the
1st and 2nd days after training (days +1 and +2) and 2 weeks after cycle ergometer. For the closed-loop measurements, subjects
training (day +14). breathed air throughout. For the open-loop measurements, the
end-tidal gases were regulated using an end-tidal forcing system
Experimental technique
(Robbins et al. 1982). PET,CJ was held constant at 2 Torr above the
Subjects sat upright in a chair for the procedures exclusively at
subject’s normal air-breathing value in exercise during the 8 min
rest, and on a cycle ergometer for those involving any exercise. For
of exercise and at 2 Torr above the subject’s normal air-breathing
the preliminary measurements of the end-tidal gases at rest, gas
value at rest during the 5 min periods of rest before and after
was sampled from a catheter taped just below the subject’s nostril
exercise. PET,J was held at the subject’s normal level during
and analysed by mass spectrometry (Airspec MGA300, UK). For
exercise for the 8 min of exercise, and at the subject’s normal level
all other experiments, subjects breathed through a mouthpiece
at rest for the 5 min periods of rest before and after exercise.
with the nose occluded by a clip. Gas was sampled from a port in
the mouthpiece. Data analysis
In order to assess visually the effects of the different protocols,
All exercise in this study was performed on the same cycle
breath-by-breath data from the pre- and post-training measure-
ergometer (Mijnhardt KEM-3, Cardiokinetics, Salford, UK). This
ments were first averaged over 20 s intervals. This enabled the
ergometer was electromagnetically braked and provided a
responses to be averaged across subjects for both the closed- and
constant work rate, independent of pedal frequency. A coloured
open-loop measurements. These average responses were then
Journal of Physiology
EX + CO2 were compared with pooled data from the day _1 RESULTS
measurements and the days +1 and +2 data from the other two
protocols. Subjects
The individual physical characteristics of the subjects are
Daily average ratings of breathlessness (Borg scores) were calculated
for each subject and these were averaged overall to give a mean given in Table 1, along with values for maximal O2 uptake
score for each protocol. Differences between these scores were (◊O2,max) and the work rate equivalent to 50 % of maximal.
assessed statistically using ANOVA. There may have been some bias away from regular
sportsmen in the selection of subjects due to the
All statistical analysis was performed using the SPSS statistical
software package. Significance was accepted at P < 0.05 and unless limitations that the protocol placed on sporting activities.
indicated otherwise, data are presented as the mean ± S.E.M. However, of the eight subjects selected, only one usually
took no regular exercise.
Closed-loop responses
The closed-loop responses to (air-breathing) exercise
before and after training with the three protocols are
shown in Figs 2 and 3. Average values for PET,CJ, ◊E,
respiratory frequency (fR) and tidal volume (VT) over the
final 3 min of both rest and exercise are given in Table 2.
Inspection of the figures suggests that there was no
training effect of the repeated inspiratory CO2 load at rest
(protocol CO2) in the sense that the regulation neither of
PET,CJ during exercise nor ◊E during exercise was affected
by the training. Following training with repeated exercise
(protocol EX), there was again no change in the level at
which PET,CJ was regulated during exercise compared with
Journal of Physiology
Following training with protocol EX + CO2, the increase No significant difference was detected between protocols
in PET,CJ with exercise was less on days +1 and +2 in the changes in VT from rest to exercise. ANOVA
compared with the two control protocols and with pre- conducted on the changes in fR from rest to exercise did not
training responses (reduction in PET,CJ = 1.3 ± 0.4 Torr, quite reach significance (P = 0.07).
ANOVA, P < 0.05). By day +14 the responses of both
Sensations of breathlessness
PET,CJ and ◊E to exercise had returned to normal.
The mean ± S.D. ratings of breathlessness for each of the
Examining the pattern of breathing during the closed-loop training protocols were 1.7 ± 1.1 for protocol EX + CO2,
response (see Table 2), there was no significant difference 1.6 ± 1.0 for protocol EX and 1.3 ± 1.1 for protocol CO2.
in changes in fR from rest to exercise between protocols. ANOVA revealed no significant difference between these
However, after training with protocol EX, VT increased less values.
with exercise on days +1 and +2 compared with the other
two protocols and with pre-training responses (P < 0.05). DISCUSSION
The response of VT had returned to normal by day +14.
The major finding of this study is that repeated
Open-loop responses disturbances in chemoreception during exercise in
The open-loop responses to exercise at a regulated PET,CJ humans modify the normal ventilatory response to
before and after training are shown in Fig. 4. Average exercise. In particular, repetitive occurrences of increased
values for PET,CJ, ◊E, fR and VT over the final 3 min of both chemoreceptor activity during exercise result in an
rest and exercise are given in Table 2. Consistent with the augmentation of subsequent ventilatory responses to
expectations from the closed-loop responses, ◊E during exercise, which in a naturally occurring situation would
exercise was higher after training with protocol EX + CO2 act to strengthen the proportionality between ◊E and ◊CJ
(increase in ◊E = 11.5 ± 3.8 l min_1), lower after training and tighten control over Pa,CJ.
with protocol EX (reduction in ◊E = 13.2 ± 4.5 l min_1)
and unchanged after training with protocol CO2. The Critique of the study
increases in ◊E from rest to exercise were significantly In this study, PET,CJ has been used as a proxy measurement
different between day _1 and days +1 and +2 between for Pa,CJ, whereas ideally, direct measurements of Pa,CJ
protocols (ANOVA, P < 0.05). By day +14 the responses would have been made. However, with a study design where
had returned to normal. the same subjects undertook all three protocols, this would
Journal of Physiology
Figure 3
◊E measured in normal closed-loop (air-breathing) conditions Figure 4
during exercise (0–8 min) before training (day _1) and after ◊E measured in open-loop conditions (at regulated PET,CJ) during
training (days +1, +2 and +14). Pre-training (day _1) exercise (0–8 min) before training (day _1) and after training (days
measurements (also shown in post-training panels for +1, +2 and +14). Pre-training (day _1) measurements (also shown
comparison), open symbols; post-training measurements, filled in post-training panels for comparison), open symbols; post-
symbols. Values are breath-by-breath data averaged over 20 s training measurements, filled symbols. Values are breath-by-
intervals, averaged across subjects. breath data averaged over 20 s intervals, averaged across subjects.
972 H. E. Wood, M. Fatemian and P. A. Robbins J Physiol 553.3
Physiology in Press
have required arterial catheterisation of our volunteers on at or a reduction in the respiratory dead space. The latter
least six occasions, which we felt was difficult to justify. would seem less likely given that there were no significant
Although PET,CJ is a fairly good approximation of Pa,CJ, changes in fR. Importantly, from our perspective the
there is nevertheless a difference between the two variables, change that occurred with training was a change in ◊E
and this difference changes with the level of exercise without a change in PET,CJ. Such a change would be
(Robbins et al. 1990). At rest, PET,CJ is lower than Pa,CJ, but entirely in keeping with the notion that the respiratory
during exercise PET,CJ rises to a value above Pa,CJ. Thus, if response to exercise slowly adapts to ensure isocapnia
Pa,CJ remains constant with exercise, PET,CJ will rise. when internal factors change.
Robbins et al. (1990) reported that the difference between
In contrast to protocol EX, protocol CO2 had no effects on
Journal of Physiology
have been demonstrated in goats following surgery to component to the ventilatory response to exercise in
denervate proprioceptors in the chest wall (Mitchell et al. humans relates to the training protocol employed. In the
1988). After surgery, the normal ventilatory response to present study there were 70 rest-to-exercise transitions in
exercise was abolished, with goats becoming severely the training period, whereas for previous studies the
hypercapnic during exercise. Over repeated trials of maximum number has been 16. In general, it is well-
exercise, a normal response was re-established. This can recognised that the learning of a motor skill often requires
not be explained by functional recovery, and hence the many repetitions over several training sessions, and in
authors concluded that it was due to adaptive control addition a period of consolidation may be required in
mechanisms that had modified components of the order for performance gains to be fully expressed (Karni et
respiratory control system (Mitchell et al. 1990). They al. 1998). There is little reason to suppose that learning the
proposed that this mechanism was elicited by repeated respiratory response to exercise should be any different in
associations of chemoreceptor feedback with exercise, and this respect. Indeed, given the length of time over which
indeed, subsequently demonstrated that repeated exercise learning has occurred, it may well be the case that longer
paired with hypercapnia (induced by increasing the protocols than that employed in the present study would
respiratory dead space) augmented the subsequent have greater effects on the ventilatory response to exercise.
ventilatory response to exercise alone (Martin et al. 1990). Clearly, much more experimental work is required before
the components that are important for learning to occur
In humans, results from similar studies pairing exercise
are properly identified.
with dead space have been far less convincing. It does
appear that pairing exercise with dead space may have A further difference between our study and those
some effect on the time course of the rapid transient in ◊E at conducted previously lies in the nature of the respiratory
the start of exercise (Helbling et al. 1997; Turner & stimulus that was paired with exercise. Previous studies
Sumners, 2002). Furthermore, there are two preliminary (Helbling et al. 1997; Cathcart et al. 2000; Moosavi et al.
reports (Adams et al. 1992; Turner et al. 1996) that 2002; Turner & Sumners, 2002) have all paired added dead
repeated pairings of dead space and exercise alter the space with exercise. However, it has been observed that
subsequent steady-state ventilatory response to exercise. dead space may have mechanical as well as chemical
However, neither of these findings has been substantiated influences on ◊E, such as changes in the endogenous
within the context of a full publication, and furthermore resistance that may affect the proprioceptors (Martin &
the preliminary findings of Adams et al. (1992) have been Mitchell, 1993). We have avoided any possible mechanical
attributed to habituation to the experience of performing influence on ◊E by inducing hypercapnia by adding CO2 to
exercise in the laboratory in a later full publication by the the inspired air. The dead space used by Turner & Sumners
same group of authors (Moosavi et al. 2002). (2002) induced hypercapnia of 6–8 Torr, which is similar
to the level used in the present study. We can assume that
Three full publications and one preliminary publication
the degree of hypercapnia would have been similar in the
have reported no effect of repeated pairings of dead space
study by Helbling et al. (1997) as they used a very similar
and exercise on the subsequent steady-state response to
dead space volume (1.5 l) to Turner & Sumners (2002).
exercise (Helbling et al. 1997; Cathcart et al. 2000; Moosavi
Moosavi et al. (2002) have stated that their protocol may
et al. 2002; Turner & Sumners, 2002). In all but one of
have been adversely affected by using a small dead space
these studies (Helbling et al. 1997), the period of
volume (0.6 l), which induced a lesser degree of
conditioning has been undertaken within a single day. In
hypercapnia.
the study by Helbling et al. (1997), the conditioning
Journal of Physiology
However, we do not consider that their data really support Martin PA & Mitchell GS (1993). Long-term modulation of the
this conclusion. In particular, from the individual data exercise ventilatory response in goats. J Physiol 470, 601–617.
plotted in their figures, we estimate that the variance of the Martin PA, Mitchell GS, Brown KL & Kaarakka P (1990). Paired
exercise and chemoreceptor stimulation alter subsequent
change in PET,CJ from rest to exercise is some five- to sixfold
ventilatory responses to exercise. FASEB J 4, A540.
greater in the children with CCHS than in their normal Mitchell GS, Douse MA & Foley KT (1990). Receptor interactions in
controls. Finally, although it is clear that patients with modulating ventilatory activity. Am J Physiol 259, R911–920.
CCHS lack a reflex stimulation of ◊E by CO2, it is not Mitchell GS, Foley K, McGuirk S, Selby BD, Schaefer SA & Lange KJ
necessarily clear that the defect is in the afferent limb of the (1988). Effects of thoracic dorsal rhizotomy (TDR) on ventilatory
reflex loop and that these patients also lack chemoreception control during mild exercise in goats. FASEB J 2, A1508.
through which the ventilatory response to exercise could be Moosavi SH, Guz A & Adams L (2002). Repeated exercise paired
with ‘imperceptible’ dead space loading does not alter the VE of
calibrated.
subsequent exercise in humans. J Appl Physiol 92, 1159–1168.
Concluding remarks Pandit JJ & Robbins PA (1992). Ventilation and gas exchange during
In conclusion, the results of the present study demonstrate sustained exercise at normal and raised CO2 in man. Respir Physiol
88, 101–112.
that humans, like goats, can learn the appropriate
Rasmussen B, Klausen K, Clausen JP & Trap-Jensen J (1975).
calibration of the feedforward mechanisms that regulate Pulmonary ventilation, blood gases, and blood pH after training of
breathing in exercise, with the chemoreceptors providing the arms or the legs. J Appl Physiol 38, 250–256.
the sensory feedback that is necessary for such learning to Robbins PA, Conway J, Cunningham DA, Khamnei S & Paterson DJ
occur. Such a mechanism provides a solution to the long- (1990). A comparison of indirect methods for continuous
standing problem of how, during exercise, feedforward estimation of arterial PCO2 in men. J Appl Physiol 68, 1727–1731.
stimuli can provide a respiratory response that is so tightly Robbins PA, Swanson GD & Howson MG (1982). A prediction-
correction scheme for forcing alveolar gases along certain time
coupled to metabolism.
courses. J Appl Physiol 52, 1353–1357.
Shea SA, Andres LP, Shannon DC & Banzett RB (1993). Ventilatory
REFERENCES responses to exercise in humans lacking ventilatory
chemosensitivity. J Physiol 468, 623–640.
Adams L, Moosavi S & Guz A (1992). Ventilatory response to
Somjen GG (1992). The missing error signal-regulation beyond
exercise in man increases by prior conditioning of breathing with
negative feedback. News Physiol Sci 7, 184–185.
added dead space. Am Rev Respir Dis 145, A882.
Turner DL, Greenway JR, Lawrence H, Lyons P, Taylor MR & Iqbal
Borg GAV (1982). Psychophysical bases of perceived exertion. Med
ZM (1996). Long-term modulation of ventilatory control in
Sci Sports Exerc 14, 377–381.
exercising humans (Abstract). Neurosci Abstr 22, 1602.
Casaburi R, Storer TW & Wasserman K (1987). Mediation of
Turner DL & Sumners DP (2002). Associative conditioning of the
reduced ventilatory response to exercise after endurance training. J
exercise ventilatory response in humans. Respir Physiol Neurobiol
Appl Physiol 63, 1533–1538.
132, 159–168.
Cathcart AJ, Herrold N, Turner AP, Wilson J & Ward SA (2000).
Wasserman K, Van Kessel AL & Burton GG (1967). Interaction of
Absence of long-term modulation in response to external dead
physiological mechanisms during exercise. J Appl Physiol 22,
space loading during moderate exercise in humans. J Physiol 44P,
71–85.
528P.
Zuntz N & Geppert J (1886). Ueber die Natur der normalen
Haldane JS & Priestley JG (1905). The regulation of the lung-
Atemreize und den Ort ihrer Wirkung. Pflügers Arch Gesamte
ventilation. J Physiol 32, 225–266.
Physiol Menschen Tiere 38, 337–338.
Helbling D, Boutellier U & Spengler CM (1997). Modulation of the
ventilatory increase at the onset of exercise in humans. Respir
Physiol 109, 219–229. Acknowledgements
We thank Mr David O’Connor for his skilled technical assistance.
Journal of Physiology