12923 Journal of Physiology (2001), 536.3, pp.

905–915 905

In vivo reduction in ATP cost of contraction is not related

to fatigue level in stimulated rat gastrocnemius muscle

Benoît Giannesini, Marguerite Izquierdo, Yann Le Fur, Patrick J. Cozzone

and David Bendahan
Centre de Résonance Magnétique Biologique et Médicale (CRMBM), UMR CNRS 6612,
Faculté de Médecine de Marseille, 27 Boulevard Jean Moulin, 13005 Marseille, France
(Resubmitted 14 June 2001; accepted 3 July 2001)
1. We tested whether the reduction in ATP cost of contraction during in vivo stimulation of rat
gastrocnemius muscle was related to fatigue level.
2. Muscles (n = 44) were electrically stimulated to perform 6 min repeated isometric contractions
at different frequencies; one non-fatiguing protocol (stimulation at 0.8 Hz) and five fatiguing
protocols (2, 3.2, 4, 5.2 and 7.6 Hz) were used. Anaerobic and oxidative ATP turnover rates
were measured non-invasively using 31P-magnetic resonance spectroscopy.
3. At the onset of the stimulation period, no signs of fatigue were measured in the six protocols
and ATP cost of contraction did not differ significantly (P = 0.45) among protocols (mean value
of 1.76 ± 0.11 mM (N s)_1).
4. For the six protocols, ATP cost of contraction was significantly reduced (P < 0.05) at the end of
the stimulation period when compared with the initial value. This reduction did not differ
significantly (P = 0.61) among the five fatiguing protocols (averaging 35 ± 3 % of initial
value), whereas isometric force decreased significantly as stimulation frequency increased. No
significant correlation (P = 0.87, r 2 = 0.01) was observed between isometric force and ATP cost
of contraction at the end of the stimulation period. In addition, this reduction was significantly
lower (P < 0.05) for the non-fatiguing protocol (67 ± 9 % of initial value) when compared with
the fatiguing protocols.
5. These results demonstrate that (i) the reduction in ATP cost of contraction during in vivo
stimulation of rat gastrocnemius muscle is not related to the fatigue level; (ii) surprisingly, this
reduction was significantly larger during the fatiguing protocols compared with the non-
fatiguing protocol.

The ATP demand in working muscle is tightly coupled to
mechanical output (Hochachka & McClelland, 1997;
Hogan et al. 1998; Stary & Hogan, 2000) and depends on
numerous variables such as the type of contractile
activity, i.e. dynamic or isometric (Ryschon et al. 1997),
the intensity and duration of contractions (Crow &
Kushmerick, 1982; Bergstrom & Hultman, 1988; Hogan
et al. 1998) and the fibre type composition of the muscle
(Crow & Kushmerick, 1982; Newham et al. 1995;
Ratkevicius et al. 1998). For these reasons, the rate of
ATP utilization shows considerable variability among the
different forms of exercise (Fitts, 1994). The assessment
of the ATP cost of contraction provides a good tool to
compare the bioenergetics of these different forms of
exercise, by normalizing the ATP utilization to force
output.
It is well established that ATP cost of contraction is
reduced throughout fatiguing exercise. Chemical analysis
of isolated mouse muscle (Crow & Kushmerick, 1982) and
of rat muscle stimulated in situ (de Haan et al. 1986) has
shown that the ATP cost of a single tetanic contraction
inducing fatigue decreases progressively with the increased
duration of the contraction. Similar observations have
been reported in human muscle; 31P-magnetic resonance
spectroscopy (31P-MRS) studies have shown that the ATP
cost of contraction declines during a single 90 s maximal
voluntary contraction (MVC) (Boska, 1994; Newcomer et
al. 1999) and also throughout a 15 min protocol consisting
of brief repeated MVCs (Newcomer & Boska, 1997).
Likewise, the ATP cost of a brief MVC (5 s duration) has
been demonstrated to be lower during recovery from a
4–5 min exhaustive dynamic exercise than during the
pre-exercise period (Smith et al. 1999). It is of interest to
stress that all these previous studies have investigated
the changes in ATP cost of contraction throughout a
given fatiguing protocol. However, the relationship
between the level of fatigue and the reduction in the ATP
cost of contraction has to our knowledge never been
addressed.
906 B. Giannesini and others
The purpose of the present study was to test in vivo
whether the reduction in ATP cost of contraction during
rat gastrocnemius muscle stimulation was related to fatigue
level. Muscles were electrically stimulated to perform
6 min isometric contractions at different frequencies; one
non-fatiguing protocol (0.8 Hz) and five fatiguing protocols
(2, 3.2, 4, 5.2 or 7.6 Hz) were used. We have quantified
non-invasively using 31P-MRS the rates of ATP turnover
from oxidative and anaerobic pathways (Kemp & Radda,
1994; Conley et al. 1998; Newcomer et al. 1999; Walter et
al. 1999) in order to calculate the ATP cost of contraction.
A preliminary account of this work has been presented
previously (Giannesini et al. 2000).
METHODS
Animal care and feeding
Forty-four male Wistar rats (CERJ, Le Genest St Isle, France)
weighing 350–375 g were used for these experiments, following the
guidelines of the National Research Council Guide for the care and
use of laboratory animals, and the French Law on the Protection of
Animals. Rats were housed in an environmentally controlled facility
(12 h light–12 h dark cycle, 22 °C), and received water and standard
rat food ad libitum until the time of experiment. At the end of
experiments, animals were immediately killed by an intracardiac
injection of sodium pentobarbitone.
Hindlimb surgical preparation
General anaesthesia was induced with an intraperitoneal injection
of sodium pentobarbitone (50 mg (kg body weight)_1), and was
maintained during the experiment by repeated administrations of
anaesthetic (10 mg (kg body weight)_1, every 30 min) through an
intraperitoneal catheter. The left hindlimb was surgically prepared
for in situ sciatic nerve stimulation of the gastrocnemius muscle. The
Achilles’ tendon was exposed and removed from the foot at the site
of calcaneus bone attachment, leaving intact the neurovascular
supply of the muscle. The distal part of the gastrocnemius tendon
was attached to a home-built force transducer via a silk thread.
Through a small incision made at the hip level, the left sciatic nerve
was exposed in its gluteal course between the quadratus femoris and
the caudofemoralis muscles, and was carefully cleared of connective
tissue. A home-built bipolar electrode connected to an electrical
stimulator (SI-10, NARCO, USA) was placed around the sciatic
nerve, which was cut proximally to the electrode attachment. The
part of the sciatic nerve connected to the bipolar electrode was put
back in place at the hip level and the incision was surgically closed.
The rat was placed backwards in a custom-made Perspex cradle
integrating a warm-water heating pad. Body temperature (typically
35–36 °C) was monitored throughout the experiment using a rectal
probe. The left leg was firmly immobilized by securing the foot with
straps and by inserting a non-magnetic brass pin into the tibia head.
In that position, the belly of the gastrocnemius muscle was located
above the 31P-MRS surface coil. At rest, muscle was passively loaded
(typically 1.3–1.5 N) by adjusting the position of the force transducer
in order to give maximum isometric twitch tension in response to
supramaximal square wave pulses (1–10 V, 1 ms duration) delivered
to the sciatic nerve. At this stage, the cradle was inserted into the
magnet.
Stimulation protocol
The stimulation protocol consisted of 6 min of isometric contractions,
which were electrically induced via the sciatic nerve with supra-
maximal square-wave pulses (1–10 V, 0.2 ms duration). Animals were
J. Physiol. 536.3
randomly divided to perform six different stimulation protocols: one
non-fatiguing protocol at 0.8 Hz (n = 7) and five fatiguing protocols
at 2 Hz (n = 7), 3.2 Hz (n = 9), 4 Hz (n = 9), 5.2 Hz (n = 6) and 7.6 Hz
(n = 6). Isometric contractions were synchronized to MRS data
acquisition.
Force measurements
Force measurements were conducted with a home-built force
transducer, which was linear from 0 to 10 N. During muscle
stimulation, the electrical signal coming from the force transducer
was amplified (reference: 13-4515-50, Gould, USA), converted to a
digital signal and processed on a personal computer using ATS
software (SYSMA, France). Isometric force (in N s) was calculated
every 15 s of stimulation by integrating isometric tension (in N)
relative to time (in s) and was expressed as tension–time integral.
Isometric force per twitch (N s twitch_1) was calculated by scaling
isometric force to the stimulation frequency. Muscle relaxation was
characterized as the extent of the return to rest tension between
consecutive twitches.
Magnetic resonance spectroscopy and data processing
31
P-MRS investigations were performed in a horizontal super-
conducting magnet (Brüker 47/30 Biospec system, Karlsruhe,
Germany) operating at 4.7 T. Magnetic resonance (MR) data were
collected with a home-built 31P-MRS surface coil (10 w 14 mm).
Magnetic field homogeneity was optimized by monitoring the water
signal until the width of the water resonance at half height was less
than 0.25 p.p.m. 31P-MR signal was acquired at 81 MHz following
20 µs radiofrequency pulses applied with a repetition time of 2.4 s.
MR data acquisition was gated to stimulation in order to record
signals between consecutive contractions and then reduce motion
artifacts due to contraction. Free-induction decays (FIDs; 12 scans,
4000 Hz sweep width, 512 data points collected) were continuously
recorded in 30 s blocks throughout the experimental protocol: during
the 6 min before stimulation (rest), during stimulation (6 min) and
during the 30 min after stimulation (recovery). FIDs were
transferred to an IBM RISC 6000 workstation and processed using
NMR1 spectroscopy processing software (New Methods Research,
USA). After deconvolution to a line broadening of 15 Hz and
application of zero filling (2 K), FIDs were Fourier-transformed into
spectra and baseline correction was performed as previously
described (Mazzeo & Levy, 1991). Signal areas corresponding to PCr,
Pi and b-ATP were measured by curve fitting of the spectrum signals
to a Lorentzian shape function (Mazzeo & Levy, 1991), and were
corrected for magnetic saturation effects using fully relaxed spectra
collected at rest with a TR of 20 s. Absolute concentrations of PCr
and Pi were expressed relative to a resting b-ATP concentration of
5.8 mM reported from fluorimetrical measurements in the same
muscle from the same strain of rat (Kemp et al. 1996). Intracellular
pH (pHi) was calculated from the chemical shift of Pi relative to PCr
(_2.45 p.p.m.) as previously described (Arnold et al. 1984). Time
points for the time course of pHi and phosphorylated metabolite
concentrations were assigned to the midpoint of the acquisition
interval as previously described (Kemp et al. 1994, 1996).
Calculations
The ATP turnover rates (mM s_1) by PCr hydrolysis (D), oxidative
phosphorylation (Q) and glycolysis (L) were calculated as previously
described (Kemp & Radda, 1994; Kemp et al. 1996; Ratkevicius et al.
1998; Walter et al. 1999) and are detailed in the Appendix. Total
ATP turnover rate was calculated: (i) at the onset of the stimulation
period from the sum of anaerobic ATP contributions (D + L),
considering oxidative contribution (Q) as negligible at this stage of
stimulation (Kemp et al. 1994); (ii) at the end of the stimulation
period from the sum of anaerobic and oxidative ATP contributions
J. Physiol. 536.3 ATP cost of contraction and fatigue 907

(D + L + Q). Total ATP turnover per twitch (mM twitch_1) was
calculated by scaling the total ATP turnover rate to the stimulation
frequency. ATP cost of contraction (mM (N s)_1) was calculated as the
ratio between total ATP turnover per twitch and isometric force per
twitch.
Statistics
All results are presented as means ± S.E.M. Statistical difference was
tested within each protocol using Student’s two-tailed t test for
paired observations, and among protocols using one-way analysis of
variance (ANOVA) followed by Student-Newman-Keuls test for
multiple comparisons. The level of significance was set at P < 0.05.
RESULTS
Time-dependent changes in [PCr] and pHi
Typical 31P-MRS spectra from a single rat gastrocnemius
muscle are presented in Fig. 1. [PCr] and pHi were not
significantly different among resting muscles, averaging
20.7 ± 0.3 mM and 7.04 ± 0.02, respectively, over the six
groups (Table 1). Throughout the stimulation period, [PCr]
and pHi decreased to reach an end-of-stimulation value,
which was lower at higher stimulation frequencies (Table 1).
Time-dependent changes in [PCr] and pHi are illustrated
in Fig. 2A and B.
Time-dependent changes in isometric force
Samples of isometric tension traces are presented in Fig. 3.
At the onset of muscle stimulation (t = 0.25 min), no sign
of fatigue was observed within each protocol (Fig. 4A). At
this time, isometric force per twitch did not differ
significantly (P = 0.45) among the six protocols (Fig. 4A),
averaging 0.124 ± 0.004 N s twitch_1 over the six protocols.
Throughout muscle stimulation, isometric force per
twitch was not significantly altered (104 ± 7 % of initial
value) for the non-fatiguing protocol (at 0.8 Hz) but
decreased significantly for the fatiguing protocols
(Fig. 4A). Time-dependent changes in isometric force are
illustrated in Fig. 2C. At the end of the fatiguing
protocols (t = 5.75 min), the level of fatigue increased
proportionally with stimulation frequency (Fig. 4A):
isometric force per twitch reached 0.105 ± 0.006 N s
twitch_1 at 2 Hz (82 ± 4 % of initial value), 0.067 ±
0.006 N s twitch_1 at 3.2 Hz (60 ± 7 % of initial value),
0.054 ± 0.006 N s twitch_1 at 4 Hz (51 ± 8 % of initial
value), 0.044 ± 0.006 N s twitch_1 at 5.2 Hz (35 ± 4 % of
initial value) and 0.024 ± 0.004 N s twitch_1 at 7.6 Hz
(19 ± 4 % of initial value). Differences were statistically
significant (P < 0.01) among protocols at 0.8, 2 and 3.2 Hz

Figure 1. Typical 31P-MRS spectra from a single rat gastrocnemius muscle at rest (A) and at the
end of 6 min repeated isometric contractions produced at 2 Hz (B)
MR data acquisition was synchronized to stimulation in order to record 31P signals between consecutive
twitches. Abbreviations for peak assignment (in p.p.m.) are: PME (phosphomonoester), Pi (inorganic
phosphate), PCr (phosphocreatine), and y-, a-, and b-resonances of ATP.
908 B. Giannesini and others J. Physiol. 536.3

Table 1. PCr concentration and intracellular pH (pHi) determined by 31P-MRS in rat

gastrocnemius muscle at rest and at the end of 6 min repeated isometric contractions produced
during non-fatiguing (0.8 Hz) and fatiguing (2, 3.2, 4, 5.2 and 7.6 Hz) stimulation protocols

Stimulation protocol
———————————————————————————————————————
0.8 Hz 2 Hz 3.2 Hz 4 Hz 5.2 Hz 7.6 Hz
n 8 7 9 9 6 6
[PCr] (mM) Rest 20.9 ± 0.6 20.5 ± 0.5 20.8 ± 0.7 20.9 ± 0.8 20.8 ± 0.8 20.5 ± 0.4
End 11.9 ± 0.6 * 6.5 ± 0.5 * 6.9 ± 1.0 * 5.8 ± 0.8 * 6.0 ± 1.4 * 2.9 ± 0.6 *
pHi Rest 7.04 ± 0.03 7.03 ± 0.02 7.04 ± 0.02 7.04 ± 0.03 7.04 ± 0.02 7.02 ± 0.02
End 6.89 ± 0.03 * 6.42 ± 0.08 * 6.37 ± 0.05 * 6.37 ± 0.06 * 6.31 ± 0.05 * 6.22 ± 0.06 *
Values are means ± S.E.M. n, number of rats. * Significantly different from rest value (P < 0.05).

Table 2. Rat gastrocnemius muscle relaxation measured at the onset and at the end of 6 min
repeated isometric contractions produced during non-fatiguing (0.8 Hz) and fatiguing (2, 3.2, 4,
5.2 and 7.6 Hz) stimulation protocols

Stimulation protocol
————————————————————————————————————
0.8 Hz 2 Hz 3.2 Hz 4 Hz 5.2 Hz 7.6 Hz
Onset 100 % 100 % 100 % 100 % 100 % 100 %
End 100 % 100 % 100 % 100 % 91 ± 6 % * 75 ± 11 % *
Values are means ± S.E.M. Muscle relaxation was determined from the extent of the return to rest tension,
expressed as a percentage, between consecutive twitches. * Significantly different from onset stimulation
value (P < 0.05).

Figure 2. Typical time-dependent changes in [PCr] (A), pHi (B) and isometric force (C) during
6 min repeated isometric contractions produced at 3.2 Hz
Values are means ± S.E.M. (n = 9). For the time-dependent changes in [PCr] and pHi, the first point
indicates the resting value.
J. Physiol. 536.3 ATP cost of contraction and fatigue 909

and between protocols at 4 and 7.6 Hz, but were not
statistically (P = 0.10) significant between protocols at
5.2 and 7.6 Hz (Fig. 4A).
Muscle relaxation
For the six protocols, muscle relaxation was fully
achieved (100 %) at the onset of the stimulation period
when isometric tension returned to rest values between
consecutive twitches (Table 2). At the end of the
stimulation period, muscle relaxation was also fully
achieved for protocols ranging from 0.8 to 4 Hz, but not
for protocols at 5.2 and 7.6 Hz when muscle tension did
not return to rest values between consecutive twitches
(91 ± 5 and 75 ± 10 % relaxation at 5.2 and 7.6 Hz,
respectively) (Table 2). This observation is illustrated in
Fig. 3.
ATP turnover rate
Total ATP turnover per twitch did not differ significantly
(P = 0.29) among the six protocols at the onset of the
stimulation period (Fig. 4B), averaging 0.208 ± 0.012 mM
twitch_1 over the six protocols. At this time, relative
contributions of PCr hydrolysis (D) and glycolysis (L) to
the total ATP turnover per twitch were affected by
increasing stimulation frequencies: D declined with
increasing stimulation frequency, from 0.167 ± 0.012 mM
twitch_1 at 0.8 Hz (73 ± 7 % of the total ATP turnover) to
0.107 ± 0.009 mM twitch_1 at 7.6 Hz (43 ± 3 % of the
total ATP turnover) (Table 3); on the other hand, L
increased by a factor of 2 between 0.8 Hz (0.066 ±
0.019 mM twitch_1) and 7.6 Hz (0.150 ± 0.023 mM twitch_1)
(Table 3).
At the end of the stimulation period, the total ATP
turnover per twitch was significantly reduced in the six
protocols (Fig. 4B). The extent of this reduction increased
proportionally as the stimulation frequency increased
(Fig. 4B): total ATP turnover per twitch reached 0.140 ±
0.021 mM twitch_1 (68 ± 7 % of initial value) at 0.8 Hz,
0.066 ± 0.009 mM twitch_1 (38 ± 4 % of initial value) at
2 Hz, 0.042 ± 0.007 mM twitch_1 (25 ± 4 % of initial value)
at 3.2 Hz, 0.029 ± 0.004 mM twitch_1 (14 ± 2 % of initial

Figure 3. Typical isometric tension traces from three rat gastrocnemius muscles electrically
stimulated at 0.8, 4 or 7.6 Hz
For each trace, isometric tension (in N) is shown at the onset (left panels) and at the end (right panels) of
6 min repeated isometric contractions. The dashed horizontal lines indicate the rest tension. Muscle
stimulation produced no fatigue at 0.8 Hz (A), an intermediate level of fatigue at 4 Hz (B) and a high level
of fatigue at 7.6 Hz (C). Note that muscle relaxation was not fully achieved (65 % relaxation) at the end
of muscle stimulation at 7.6 Hz when muscle tension did not revert to rest values between consecutive
twitches.
910 B. Giannesini and others J. Physiol. 536.3

value) at 4 Hz, 0.017 ± 0.003 mM twitch_1 (11 ± 3 % of ATP cost of contraction

initial value) at 5.2 Hz and 0.011 ± 0.002 mM twitch_1
(4 ± 1 % of initial value) at 7.6 Hz. Differences were
statistically significant (P < 0.04) among protocols at 0.8,
2 and 3.2 Hz, but not among protocols at 4, 5.2 and 7.6 Hz
(Fig. 4B).
ATP cost of contraction did not differ significantly
(P = 0.45) among the six protocols at the onset of the
stimulation period (Fig. 4C), with a mean value of
1.76 ± 0.11 mM (N s)_1 over the six protocols. At the end
of the stimulation period, ATP cost of contraction was

Figure 4. Isometric force (A), total ATP turnover (B) and ATP cost of contraction (C) at the onset
and at the end of 6 min repeated isometric contractions produced during non-fatiguing (0.8 Hz)
and fatiguing (2, 3.2, 4, 5.2 and 7.6 Hz) stimulation protocols
Values are means ± S.E.M. * Significant difference (P < 0.05) between the onset (t = 0.25 min) and the
end (t = 5.75 min) of each protocol. At the end of the stimulation period, statistical difference
(P < 0.05) was tested among stimulation protocols using one-way analysis of variance (ANOVA)
followed by a Student-Newman-Keuls test for multiple comparisons: a significantly different from
protocol at 0.8 Hz; b significantly different from protocol at 2 Hz; c significantly different from protocol
at 3.2 Hz; d significantly different from protocol at 4 Hz; e significantly different from protocol at 5.2 Hz;
f
significantly different from protocol at 7.6 Hz.
J. Physiol. 536.3 ATP cost of contraction and fatigue 911

Table 3. ATP turnover per twitch by PCr hydrolysis (D) and glycolysis (L) at the onset of 6 min
repeated isometric contractions produced during stimulation protocols at 0.8, 2, 3.2, 4, 5.2 and
7.6 Hz

Stimulation protocol
———————————————————————————————————————————
0.8 Hz 2 Hz 3.2 Hz 4 Hz 5.2 Hz 7.6 Hz
D (mM twitch_1) 0.167 ± 0.012 0.143 ± 0.026 0.106 ± 0.011 0.115 ± 0.016 0.101 ± 0.026 0.107 ± 0.009
L (mM twitch_1) 0.066 ± 0.019 0.038 ± 0.030 0.066 ± 0.012 0.116 ± 0.019 0.108 ± 0.037 0.150 ± 0.023
Values are means ± S.E.M.

significantly reduced in the six protocols (Fig. 4C), DISCUSSION

reaching 1.08 ± 0.13 mM (N s)_1 (67 ± 9 % of initial value)
at 0.8 Hz, 0.63 ± 0.07 mM (N s)_1 (48 ± 7 % of initial
value) at 2 Hz, 0.64 ± 0.15 mM (N s)_1 (41 ± 8 % of initial
value) at 3.2 Hz, 0.56 ± 0.07 mM (N s)_1 (29 ± 4 % of
initial value) at 4 Hz, 0.41 ± 0.07 mM (N s)_1 (31 ± 6 % of
initial value) at 5.2 Hz and 0.54 ± 0.17 mM (N s)_1
(26 ± 6 % of initial value) at 7.6 Hz; differences were
statistically significant (P < 0.006) between the non-
fatiguing protocol and the five fatiguing protocols
(Fig. 4C). However, ATP cost of contraction did not differ
significantly among the five fatiguing protocols and
averaged 0.57 ± 0.05 mM (N s)_1 (35 ± 3 % of initial
value).
Relationship between isometric force and ATP cost of
contraction
Regression analysis performed on data pooled across the
five fatiguing protocols did not show any significant
correlation (P = 0.87, r 2 = 0.01) between the isometric
force per twitch and the ATP cost of contraction at the
end of the stimulation period (Fig. 5).
This work represents the first attempt to investigate the
relationship between the reduction in ATP cost of
contraction during muscle exercise and the level of
fatigue. We mainly report three novel findings: (i) at the
onset of the stimulation period, no signs of fatigue were
observed in the six protocols and ATP cost of contraction
calculated from the total anaerobic ATP turnover (PCr
hydrolysis plus glycolysis) remained independent of the
stimulation frequency; (ii) at the end of the fatiguing
protocols, ATP cost of contraction was reduced to the
same extent (35 ± 3 % of initial value) whatever the level
of fatigue; (iii) at the end of the non-fatiguing protocol,
ATP cost of contraction was also reduced but to a lesser
extent (67 ± 9 % reduction) when compared with the
fatiguing protocols.
Effect of stimulation frequency on ATP cost of
contraction before fatigue development
At the onset of the stimulation period, no signs of fatigue
were observed in the six protocols and ATP cost of

Figure 5. Relationship between isometric force and ATP cost of contraction at the end of 6 min
repeated isometric contractions produced during fatiguing stimulation protocols at 2, 3.2, 4, 5.2
and 7.6 Hz
Regression analysis (continuous line) performed on pooled data from the fatiguing protocols (2, 3.2, 4, 5.2
and 7.6 Hz) did not show any significant correlation (P = 0.87, r 2 = 0.01).
912 B. Giannesini and others
contraction did not differ significantly among protocols
in spite of increasing stimulation frequency. These
findings are in accordance with several previous
experiments showing that the ratio of energetic demand
to force output remained independent of the stimulation
intensity during short duration exercises performed
without any fall in force output. For instance, a thermo-
dynamic study in isolated fibres from Xenopus laevis
reports that muscle efficiency (defined as the ratio of
mechanical power to total energy output) of 1.5 s
repeated isometric contractions was not altered when
stimulation frequency increased from 10 to 40 Hz
(Buschman et al. 1995). Furthermore, 31P-MRS studies
have demonstrated in vivo that ATP cost of contraction
remained unaltered at the onset of a series of repeated
isometric contractions whatever the stimulation frequency,
ranging from 0.5 to 2 Hz in flexor digitorum superficialis
of human forearm (Blei et al. 1993), from 1 to 8 Hz in rat
gastrocnemius muscle (Foley & Meyer, 1993) and over
60 Hz in rattlesnake tail muscle (Conley & Lindstedt,
1996).
It is of interest to stress that ATP cost of contraction was
calculated in these previous 31P-MRS studies at the very
start of muscle stimulation from the initial burst of PCr
breakdown, which was assumed to be the only pathway
contributing to the total ATP turnover at this stage of
stimulation; in the present experiments, our ability to
measure the glycolytic ATP turnover throughout the
stimulation period allowed us to demonstrate that the
ATP cost of contraction also remained independent of
the stimulation frequency in the latter stage of the
stimulation period, after 0.25 min of repeated isometric
contractions, when no signs of fatigue were observed.
The independence of ATP cost of contraction from
stimulation frequency at the onset of the stimulation
period may be explained by the combination of two
processes. Firstly, the isometric force per twitch did not
differ significantly between the six protocols. Indeed, for
each experiment, stimulation parameters and muscle
length were optimized at rest in order to produce maximal
isometric force during stimulation (see Methods).
Secondly, total ATP turnover per twitch calculated from
the sum of anaerobic contributions (PCr breakdown plus
glycolysis) also remains unaffected by the increase of
stimulation frequency, which roughly coincides with the
view that anaerobic metabolism is considered to provide
enough energy to cope with a high ATP demand at the
onset of exercise, including brief intense exercise (Foley &
Meyer, 1993; Rico-Sanz et al. 1998; Sahlin et al. 1998;
Walter et al. 1999). Our data showing that the decrease in
the relative contribution from net PCr breakdown was
compensated by the increase in the glycolytic contribution
illustrates that the high precision of ATP production is
regulated to meet ATP utilization throughout muscle
exercise (Hochachka & McClelland, 1997; Sahlin et al.
1998; Stary & Hogan, 2000).
J. Physiol. 536.3
An interesting observation from the present study is that
the total ATP turnover per twitch averaged 0.21 ±
0.01 mM twitch_1 over the six protocols at the onset of the
stimulation period, which is strongly similar to previous
data showing that the rate of ATP hydrolysed per twitch
measured at the onset of muscle stimulation was
0.25 ± 0.01 mM twitch_1 in rat gastrocnemius muscle
(Foley & Meyer, 1993) and 0.21 ± 0.01 mM twitch_1 in cat
biceps (predominantly composed of fast-twitch fibres)
(Harkema et al. 1997).
Reduction in ATP cost of contraction associated with
muscle stimulation
In the six protocols, ATP cost of contraction was reduced
between the onset and the end of 6 min repeated isometric
contraction. In other words, muscle paradoxically needed
less ATP to produce the same amount of isometric force at
the end of the stimulation period compared with the
onset.
Alteration of activation and relaxation processes have
been proposed to play a role in the reduction in ATP cost
of contraction throughout muscle exercise (Hogan et al.
1998; Smith et al. 1999). During repeated contractions,
ATP is used for both contractile and non-contractile
processes, the latter process being related to ion transport
associated with the activation–relaxation cycle of muscle
contraction. Therefore, any reduction of one of these
processes throughout muscle activity would explain the
reduction in ATP cost of contraction. This assumption is
supported by the fact that, for a given duration of
contraction, ATP cost of contraction is lower for a single
sustained isometric contraction when compared with
repeated isometric contractions (de Haan et al. 1986;
Chasiotis et al. 1987; Bergstrom & Hultman, 1988; Spriet,
1989; Newham et al. 1995; Hogan et al. 1998): during a
single sustained isometric contraction, ATP is mainly
hydrolysed to maintain tension, whereas additional ATP
consumption is required for ion transport between
contractions during repeated isometric contractions; it
has been reported previously that 20–50 % of the ATP
utilized during contraction would be used for Ca2+
pumping across the sarcoplasmic reticulum (SR)
membrane (Bergstrom & Hultman, 1988; Lou et al. 1997;
Hogan et al. 1998). In the present study, muscle
relaxation was not fully achieved between consecutive
twitches produced during the fatiguing protocols at 5.2
and 7.6 Hz, hence suggesting an alteration in Ca2+ fluxes
across the SR membrane. Consequently, at least for these
levels of fatigue, the reduction in ATP cost of contraction
throughout the stimulation period could be due to the
reduction in the relative contributions of non-contractile
processes.
Another attractive possibility to explain the reduction in
the ATP cost of contraction might lie in the shift of the
pattern of fibre type recruitment throughout the
fatiguing protocol. Rat gastrocnemius muscle is a mixed
muscle composed of 93 % of fast-twitch fibres and 7 % of
J. Physiol. 536.3 ATP cost of contraction and fatigue
slow-twitch fibres (Armstrong & Phelps, 1984). At the
onset of electrical stimulation, all the fast- and slow-
twitch fibres are recruited to produce muscle contraction;
with fatigue development, because fast-twitch fibres are
less fatigue resistant than slow-twitch fibres (Chasiotis et
al. 1987), the proportion of fast-twitch fibre recruitment
could, however, decrease compared with the proportion
of recruited slow-twitch fibres. Given that slow-twitch
fibres are known to contract more economically compared
with fast-twitch fibres (Sawka et al. 1981; Crow &
Kushmerick, 1982; Harkema et al. 1997), the recruitment
of a higher proportion of slow-twitch fibres could
therefore account for the reduction in ATP cost of
contraction during the fatiguing protocols.
Reduction in ATP cost of contraction during non-
fatiguing protocols
The reduction in ATP cost of contraction during the non-
fatiguing protocols indicates that this phenomenon is not
always associated with fatigue development, and further
suggests that at least one additional mechanism could
intervene. To our knowledge, potent alteration in ATP
cost of contraction in the later stages of the non-fatiguing
protocols has never been reported. The muscle–tendon
complex is composed of contractile and series elastic (non-
contractile) elements. During contraction, a muscle uses
energy to shorten against series elastic elements (SEE),
which produce a mechanical work external to contractile
elements (Newham et al. 1995). This external work has
been demonstrated to affect muscle energetics during
stretch–shorten cycles (Ettema, 1996). To a lesser extent,
it has been suggested that even under isometric conditions
some energy utilized for contraction was wasted to
overcome SEE (de Haan et al. 1986; Newham et al. 1995;
Newcomer et al. 1999). The observation that during a
single contraction, external work occurs only at the onset
of contraction but not during subsequent maintenance of
force, could explain why ATP cost of contraction has been
found to be greater at the onset of a single contraction
(Newham et al. 1995; Newcomer et al. 1999). Consequently,
it is possible that in the present study, although muscle
was stretched to its optimal length before stimulation,
external work could affect ATP cost of contraction at the
onset of stimulation. Therefore, external work could be
attenuated with successive contractions, thereby
accounting for the lower ATP cost of contraction
measured by the end of muscle stimulation whatever the
stimulation frequency. It is important to emphasize that
this mechanism could also participate in the reduction in
ATP cost of contraction during the fatiguing protocols.
Relationship between fatigue level and ATP cost of
contraction
We can now ask why the reduction in ATP cost of
contraction measured at the end of the five fatiguing
protocols is not related to the level of fatigue? According
to the view that limitation in energy availability is a
classical hypothesis to explain muscle fatigue (Fitts, 1994;
913
Sahlin et al. 1998; Ward et al. 1998), it is legitimate to
consider that muscle needs to use ATP sparingly to avoid
or at least attenuate fatigue development. Reduction in
ATP cost of contraction throughout muscle activity
might represent a way to optimize ATP utilization to
produce force. Our hypothesis is supported by the
observations that the reduction in ATP cost of
contraction (i) was larger during the fatiguing protocols
when compared with the non-fatiguing protocol and
(ii) reached the same extent whatever the level of fatigue.
These results might mean that ATP utilization is
systematically optimized throughout muscle activity up
to a threefold level, which could be associated with
fatigue development. Further investigation is still
required to understand the cause-to-effect relationship
between fatigue development and reduction in ATP cost
of contraction.
In conclusion, this study reports for the first time the
effects of increasing fatigue levels on ATP cost of
contraction. We have demonstrated that ATP cost of
contraction was reduced throughout non-fatiguing and
fatiguing protocols. During the non-fatiguing protocol,
the reduction in ATP cost of contraction may be
attributed to ATP consumption by non-contractile
elements. During the fatiguing protocols, the larger
reduction in ATP cost of contraction may be due to
additional mechanisms such as shift in fibre type
recruitment and alteration of activation–relaxation
processes. The fact that the reduction in ATP cost of
contraction during in vivo stimulation of rat gastrocnemius
muscle is not related to the fatigue level leads us to
propose that fatigue development could be associated
with an optimization in ATP utilization.
APPENDIX
The ATP turnover rates (in mM s_1) by PCr hydrolysis,
oxidative phosphorylation and glycolysis, were calculated
from the time-dependent changes in [PCr] and pHi (Kemp
& Radda, 1994; Kemp et al. 1996; Ratkevicius et al. 1998;
Walter et al. 1999).
The ATP turnover rate by net PCr hydrolysis (D) was
directly calculated from the time-dependent changes in
PCr during muscle stimulation:
D = _dPCr/dt. (A1)
The oxidative ATP turnover rate (Q) was calculated at
the end of muscle stimulation using the initial rate of PCr
recovery (Kemp et al. 1994):
Q = k PCrcons, (A2)
where PCrcons is the PCr concentration at the cessation
of muscle stimulation. The pseudo-first order rate
constant of PCr recovery (k) was estimated by fitting the
PCr recovery profile to a monoexponential function
(PCrt = PCrrest + PCrcons e_kt, where PCrrest is the resting
PCr concentration).
914 B. Giannesini and others
The glycolytic ATP turnover rate (L) was inferred from
the number of moles of protons (P) generated by lactate
formation, considering that glycolytic ATP production is
related to lactate synthesis with a stoichiometry of
1.5 moles of ATP per mole of lactate (Hochachka &
Mommsen, 1983) (L = 1.5P). P was calculated from the
observed changes in pHi taking into account the number
of moles of protons (i) associated with PCr hydrolysis (H+PCr),
(ii) passively buffered in the cytosol (H+b ), (iii) leaving the
cell (rate of net proton efflux, H+efflux), and (iv) produced
by oxidative phosphorylation (H+ox) (Kemp & Radda,
1994; Walter et al. 1999):
P = H+PCr + H+b + H+efflux _ H+ox.
+ _1
H (mM s ) was calculated from the time-dependent
PCr
changes in [PCr] and from the stoichiometric coefficient
q = 1/(1 + 10 (pH _6.75) ), which represents the number of
i BOSKA, M. (1994). ATP production rates as a function of force level
protons associated with PCr hydrolysis (Wolfe et al. 1988):
H+PCr = _qD.
H +b was calculated from the apparent buffering
capacity btotal (in slykes, millimoles acid added per unit
change in pH i) and from the variations in pH i
(∆pHi = pHobserved _ pHrest):
H+b = _btotal ∆pHi.
btotal takes into account the buffering capacity of Pi (bPi )
and the buffering capacity of tissue btissue
(btotal = bPi + btissue). bPi was calculated from the
concentration of Pi with a pKa of 6.75 (Wolfe et al. 1988):
2.303 [Pi]
bPi = —————————————.
(1 + 10 (pH _6.75))(1 + 10 (6.75_pH ) )
i i
btissue has been demonstrated to vary linearly between
pH 7 (16 slykes) and pH 6 (37 slykes) (Adams et al. 1990);
according to these data, btissue was calculated as follows:
btissue = _21pHi + 163.
+ _1
H (mM s ) was calculated during muscle stimulation
efflux
using the proportionality constant g relating proton
efflux rate to pHi (Kemp & Radda, 1994; Newcomer et al.
1999):
H+efflux = _g∆pHi.
g (in mM s_1 (pH unit)_1) was determined during the initial
recovery period using eqn (A6) and the rate of net proton
efflux calculated at this stage of stimulation:
H+efflux = qD + btotal dpHi/dt.
+ _1
H (mM s ) was calculated from Q (eqn (A2)) and from
ox
the stoichiometric coefficient m = 0.16/(1 + 10 (6.1_pH )),
which represents the number of moles of protons
produced in association with oxidative phosphorylation
(Kemp & Radda, 1994):
H+ox = m Q.
J. Physiol. 536.3
ADAMS, G. R., FOLEY, J. M. & MEYER, R. A. (1990). Muscle buffer
capacity estimated from pH changes during rest-to-work
transitions. Journal of Applied Physiology 69, 968–972.
ARMSTRONG, R. B. & PHELPS, R. O. (1984). Muscle fibre type
composition of the rat hindlimb. American Journal of Anatomy
171, 259–272.
ARNOLD, D. L., MATTHEWS, P. M. & RADDA, G. K. (1984). Metabolic
recovery after exercise and the assessment of mitochondrial
function in vivo in human skeletal muscle by means of 31P NMR.
Magnetic Resonance in Medicine 1, 307–315.
BERGSTROM, M. & HULTMAN, E. (1988). Energy cost and fatigue
during intermittent electrical stimulation of human skeletal
muscle. Journal of Applied Physiology 65, 1500–1505.
(A3) BLEI, M. L., CONLEY, K. E. & KUSHMERICK, M. J. (1993). Separate
measures of ATP utilization and recovery in human skeletal
muscle (erratum appears in Journal of Physiology 475, 548 (1994)).
Journal of Physiology 465, 203–222.
in the human gastrocnemius/soleus using 31P MRS. Magnetic
Resonance in Medicine 32, 1–10.
(A4) BUSCHMAN, H. P., ELZINGA, G. & WOLEDGE, R. C. (1995). Energetics
of shortening depend on stimulation frequency in single muscle
fibres from Xenopus laevis at 20 °C. Pflügers Archiv 430, 160–167.
CHASIOTIS, D., BERGSTROM, M. & HULTMAN, E. (1987). ATP
utilization and force during intermittent and continuous muscle
contractions. Journal of Applied Physiology 63, 167–174.
(A5) CONLEY, K. E., KUSHMERICK, M. J. & JUBRIAS, S. A. (1998).
Glycolysis is independent of oxygenation state in stimulated
human skeletal muscle in vivo. Journal of Physiology 511,
935–945.
CONLEY, K. E. & LINDSTEDT, S. L. (1996). Minimal cost per twitch in
rattlesnake tail muscle. Nature 383, 71–72.
CROW, M. T. & KUSHMERICK, M. J. (1982). Chemical energetics of
slow- and fast-twitch muscles of the mouse. Journal of General
(A6) Physiology 79, 147–166.
DE HAAN, A., DE JONG, J., VAN DOORN, J. E., HUIJING, P. A.,
WOITTIEZ, R. D. & WESTRA, H. G. (1986). Muscle economy of
isometric contractions as a function of stimulation time and
relative muscle length. Pflügers Archiv 407, 445–450.
ETTEMA, G. J. (1996). Mechanical efficiency and efficiency of
(A7) storage and release of series elastic energy in skeletal muscle
during stretch-shorten cycles. Journal of Experimental Biology
199, 1983–1997.
FITTS, R. H. (1994). Cellular mechanisms of muscle fatigue.
Physiological Reviews 74, 49–94.
FOLEY, J. M. & MEYER, R. A. (1993). Energy cost of twitch and
tetanic contractions of rat muscle estimated in situ by gated
(A8) 31
P NMR. NMR in Biomedicine 6, 32–38.
GIANNESINI, B., BENDAHAN, D., IZQUIERDO, M., LE FUR, Y. &
COZZONE, P. J. (2000). In vivo 31P-MRS evidences a paradoxical
reduction of ATP cost of contraction associated with muscular
fatigue in rat gastrocnemius muscle. 8th Meeting of the
(A9) International Society for Magnetic Resonance in Medicine,
in Proceedings of the ISMRM. Denver, USA.
HARKEMA, S. J., ADAMS, G. R. & MEYER, R. A. (1997). Acidosis has
i
no effect on the ATP cost of contraction in cat fast- and slow-
twitch skeletal muscles. American Journal of Physiology 272,
C485–490.
HOCHACHKA, P. W. & MCCLELLAND, G. B. (1997). Cellular metabolic
homeostasis during large-scale change in ATP turnover rates in
(A10) muscles. Journal of Experimental Biology 200, 381–386.
J. Physiol. 536.3 ATP cost of contraction and fatigue
HOCHACHKA, P. W. & MOMMSEN, T. P. (1983). Protons and
anaerobiosis. Science 219, 1391–1397.
HOGAN, M. C., INGHAM, E. & KURDAK, S. S. (1998). Contraction
duration affects metabolic energy cost and fatigue in skeletal
muscle. American Journal of Physiology 274, E397–402.
KEMP, G. J. & RADDA, G. K. (1994). Quantitative interpretation of
bioenergetic data from 31P and 1H magnetic resonance
spectroscopic studies of skeletal muscle: an analytical review.
Magnetic Resonance Quarterly 10, 43–63.
KEMP, G. J., SANDERSON, A. L., THOMPSON, C. H. & RADDA, G. K.
(1996). Regulation of oxidative and glycogenolytic ATP synthesis
in exercising rat skeletal muscle studied by 31P magnetic resonance
spectroscopy. NMR in Biomedicine 9, 261–270.
KEMP, G. J., THOMPSON, C. H., BARNES, P. R. & RADDA, G. K. (1994).
Comparisons of ATP turnover in human muscle during ischemic
and aerobic exercise using 31P magnetic resonance spectroscopy.
Magnetic Resonance in Medicine 31, 248–258.
LOU, F., CURTIN, N. & WOLEDGE, R. (1997). The energetic cost of
activation of white muscle fibres from the dogfish Scyliorhinus
canicula. Journal of Experimental Biology 200, 495–501.
MAZZEO, A. R. & LEVY, G. C. (1991). An evaluation of new
processing protocols for in vivo NMR spectroscopy. Magnetic
Resonance in Medicine 17, 483–495.
NEWCOMER, B. R. & BOSKA, M. D. (1997). Adenosine triphosphate
production rates, metabolic economy calculations, pH,
phosphomonoesters, phosphodiesters, and force output during
short-duration maximal isometric plantar flexion exercises and
repeated maximal isometric plantar flexion exercises. Muscle and
Nerve 20, 336–346.
NEWCOMER, B. R., BOSKA, M. D. & HETHERINGTON, H. P. (1999).
Non-P(i) buffer capacity and initial phosphocreatine breakdown
and resynthesis kinetics of human gastrocnemius/soleus muscle
groups using 0.5 s time-resolved (31)P MRS at 4.1 T. NMR in
Biomedicine 12, 545–551.
NEWHAM, D. J., JONES, D. A., TURNER, D. L. & MCINTYRE, D. (1995).
The metabolic costs of different types of contractile activity of the
human adductor pollicis muscle. Journal of Physiology 488,
815–819.
RATKEVICIUS, A., MIZUNO, M., POVILONIS, E. & QUISTORFF, B. (1998).
Energy metabolism of the gastrocnemius and soleus muscles
during isometric voluntary and electrically induced contractions
in man. Journal of Physiology 507, 593–602.
RICO-SANZ, J., HAJNAL, J. V., THOMAS, E. L., MIERISOVA, S.,
ALA-KORPELA, M. & BELL, J. D. (1998). Intracellular and
extracellular skeletal muscle triglyceride metabolism during
alternating intensity exercise in humans. Journal of Physiology
510, 615–622.
915
RYSCHON, T. W., FOWLER, M. D., WYSONG, R. E., ANTHONY, A. &
BALABAN, R. S. (1997). Efficiency of human skeletal muscle in
vivo: comparison of isometric, concentric, and eccentric muscle
action. Journal of Applied Physiology 83, 867–874.
SAHLIN, K., TONKONOGI, M. & SODERLUND, K. (1998). Energy supply
and muscle fatigue in humans. Acta Physiologica Scandinavica
162, 261–266.
SAWKA, M. N., PETROFSKY, J. S. & PHILLIPS, C. A. (1981). Energy
cost of submaximal isometric concentrations in cat fast and slow
twitch muscles. Pflügers Archiv 390, 164–168.
SMITH, S. A., MONTAIN, S. J., MATOTT, R. P., ZIENTARA, G. P.,
JOLESZ, F. A. & FIELDING, R. A. (1999). Effects of creatine
supplementation on the energy cost of muscle contraction: a
31
P-MRS study. Journal of Applied Physiology 87, 116–123.
SPRIET, L. L. (1989). ATP utilization and provision in fast-twitch
skeletal muscle during tetanic contractions. American Journal of
Physiology 257, E595–605.
STARY, C. M. & HOGAN, M. C. (2000). Phosphorylating pathways and
fatigue development in contracting Xenopus single skeletal
muscle fibers. American Journal of Physiology 278, R587–591.
WALTER, G., VANDENBORNE, K., ELLIOTT, M. & LEIGH, J. S. (1999).
In vivo ATP synthesis rates in single human muscles during high-
intensity exercise. Journal of Physiology 519, 901–910.
WARD, C. W., SPANGENBURG, E. E., DISS, L. M. & WILLIAMS, J. H.
(1998). Effects of varied fatigue protocols on sarcoplasmic
reticulum calcium uptake and release rates. American Journal of
Physiology R99–104.
WOLFE, C. L., GILBERT, H. F., BRINDLE, K. M. & RADDA, G. K.
(1988). Determination of buffering capacity of rat myocardium
during ischemia. Biochimica et Biophysica Acta 971, 9–20.
Acknowledgements
We would like to thank Ferdinand Tagliarini for his technical
assistance. This work was supported by grants from CNRS
(UMR 6612), ADEREM (Association pour le Développement des
Recherches Biologiques et Médicales au CHR de Marseille) and
Ministère de la Santé (PHRC 1997).
Corresponding author
D. Bendahan: Centre de Résonance Magnétique Biologique et
Médicale (CRMBM), UMR CNRS 6612, Faculté de Médecine de
Marseille, 27 Boulevard Jean Moulin, 13005 Marseille, France.
Email: david.bendahan@medecine.univ-mrs.fr