The present 5-day study was part of a series of bed rest studies organized by the European Space Agency (ESA), starting with a short-term bed rest in preparation for more long-term studies. Details of the design have been presented elsewhere (Mulder et al. 2014). In short, a total of three bed rest campaigns were scheduled. Each campaign consisted of 5 days of baseline data collection (BDC-5 through BDC-1), 5 days of bed rest in 6° head-down tilt (HDT1 through HDT5), and 6 days of recovery (R + 0 through R + 5). The washout period between the end of campaign 1 and the start of campaign 2 was 50 days; the washout period between the end of campaign 2 and the start of campaign 3 was 94 days. Each subject randomly performed bed rest only (CON), bed rest with 25 min of daily upright standing (STA), or bed rest with 25 min of locomotion replacement training (LRT). In bed, the subjects maintained the 6° HDT for 24 h/day (except for 25 min in the LRT and STA interventions). The study design was approved by the Ethics Committee of the Northern Rhine medical association in Düsseldorf, Germany and was organized by the DLR Institute of Aerospace Medicine.
10 male subjects who had given their written consent completed the study. Baseline characteristics are provided in Table 1. One subject discontinued the study on BDC-3 of the first campaign and was instantly replaced by a backup volunteer. This subject performed nonetheless all experiments (including familiarization sessions) that were planned for BDC-5 and BDC-4.
During the entire study, the subjects received a strictly controlled and individualized diet, and all meals were completely consumed. The individual energy intake (total energy expenditure, TEE) was calculated by multiplying resting metabolic rate (RMR), measured by indirect calorimetry (Deltatrac II MBH 200 metabolic monitor, Datex-Ohmeda) with a physical activity level of 1.4 (during ambulatory phase) and 1.1 (during HDT) for physical activity plus 10 % for diet-induced thermogenesis (DIT). 29.7 ± 0.2 % of the daily energy intake was consumed as fat; 54.9 ± 2.2 % as carbohydrates, and protein was taken in, in the amount of 1.21 ± 0.01 g/kgBM per day. The daily diet was also constant for calcium (1,085 ± 62 mg), potassium (3.9 ± 0.3 g), sodium (2.3 ± 0.1 mmol/kg BW), and water (50 mL/kgBM) intake. Additional fluid and energy intake was administered in the form of water and diluted-apple juice following physically demanding experiments to compensate for sweat and energy loss. To assess sweat loss, subjects were weighed before and after the MVC and cycle ergometry tests. Any loss in mass was assumed to be due to loss of sweat. Only the energetic cost of the cycle ergometry test was incorporated, for which the following formula was used: gross mechanical efficacy (%) = mechanical power [(W) × 0.01443 (kcal/W) × 100]/Total metabolic power input (kcal). Gross mechanical efficacy was fixated for all subjects at 23 % and hence the total metabolic power input could be calculated from the wattage and the duration of the various stages of the cycle ergometer test. Due to the absence of sunlight exposure, the subjects were daily supplemented with 1,000 IU (international units) of vitamin D.
Interventions and control condition
Locomotion replacement training (LRT)
Subjects executed the upright 25-min LRT session daily during the HDT phase. This session consisted of a combination of heel raises, squats and hopping exercises in the upright position (see Mulder et al. (2014) for details). In brief, subjects performed three blocks: block one consisted of 20 bilateral heel raises, 20 squats (90°) and 4 sets of 6 reactive jumps; block two consisted of 2 × 12 unilateral heel raises, 12 deep squats (60°) and jumping as above. Block three consisted of 2 × 12 unilateral heel raises, shallow squatting (120°), and cross hopping and finished with a static squat (90°). One minute of upright pause was incorporated between blocks. A Smith Machine with fixed rails (PTS-1000 Dual Action Smith™ Cage, Hoist Fitness Systems, San Diego, USA) was used to guide the heel raise and squat exercises. Squats and heel raises were performed against body weight plus the additional weight of the barbell (15 kg). The heel raises were performed with straight knees and without ankle dorsiflexion. The shallow squats were performed continuously for 3 min. The reactive jumps and the cross hopping (left–right-left–right, etc.) exercises were performed without Smith Machine. The reactive jumps were performed with the ball of the foot (heels not touching the ground) at ~3 repetitions per second separated by 15-s rest every six jumps. Cross hopping was performed continuously for 3 min at a frequency of 1.3 repetitions per second. The duration of the exercises was unchanged during the study, except for the static squat, which increased from 45 s at HDT1 to 70 s at HDT5 for motivational purposes.
Though gravitational loading per se (i.e., standing) partially preserved orthostatic tolerance during bed rest (Vernikos et al. 1996), the general consensus is that ‘static loading’ is ineffective to maintain bone and muscle integrity [e.g., (Lanyon and Rubin 1984)]. The standing condition was implemented as an ‘active control condition’ to test whether the effects of LRT were related to the exercise per se, or related to the fact that the exercises were performed in the upright (i.e., gravitationally loaded) posture. For this purpose, each subject stood upright directly next to the bed for 25 min. Both feet were in contact with the floor during and any type of physical activity (e.g., heel raise, squatting or walking) was prohibited.
Control condition (CON)
Subjects remained in HDT 24 h/day for 5 days and refrained from any type of physical exercise and/or upright posture.
The maximum CSA of knee extensor and plantar flexor muscles from the right limbs were assessed once before (BDC-2), and once following BR (R + 0) using Magnetic Resonance Imaging (Siemens Sonata scanner) at 1.5 Tesla using a spin echo sequence (TR = 28.00 ms, TE = 4.78 ms). Axial images with 3 mm (thigh) or 2 mm (lower leg) slice thickness were acquired with a matrix of 256 × 224 pixels of 1.0 × 1.0 mm pixel size. Subjects were positioned with their thighs in the horizontal plane, and foot restraints were used for fixation. To prevent fluid shifts from influencing CSA secondary to a change in body position, subjects remained supine for 30 min before imaging started. The knee extensor and plantar flexor muscle were each manually encircled by one operator blinded to both the session and intervention, and CSA was calculated using semi-automated SliceOmatic 4.3 software (Tomovision, Magog, Canada). Sliding averages of CSA values were calculated for three successive slides (Mulder et al. 2006) and the highest mean value was used as maximum CSA for further evaluation.
The maximal voluntary contraction (MVC) of the knee extensor, and plantar flexor muscle groups was assessed as the highest attained torque value before (BDC-1) and following bed rest (R + 0). MVC measurements were obtained from the left leg using the Biodex-3 system (Biodex Medical Systems, Shirley, New York, USA). Care was taken to align the axis of rotation of the dynamometer with the respective joint axes. After individual adjustments, the fixed positions were maintained for seat, shin pad, and dynamometer axis positions during subsequent tests. Subjects were firmly strapped to the examining chair before the measurements started. Isometric knee extension and knee flexion MVCs were obtained at knee angles of 90, 80, and 70° from full knee extension (0°). Plantar flexion and dorsal flexion MVCs were obtained at angles of −10, 0, +10, +20, and +30° from ankle neutral position (0°). At each joint angle, which was assigned in random order, the subjects performed a 5–7 s maximum isometric extension followed by a maximum flexion after 30 s of rest, and 30 s thereafter by another pair of extension and flexion contractions until three complete sets of extension/flexion contractions were obtained. A 2-min rest was incorporated before proceeding to the next randomized joint angle. In this paper we present only data from the knee extensions and plantar flexions because electromyographic activity (see below) was recorded only from m. vastus lateralis and m. gastrognemius medialis. Subjects received loud verbal encouragement during the performance of the maximal contractions.
Knee extensor muscle fatigability was assessed before (BDC-1) and following BR (R+ 0) at a knee angle of 80°, using a 90-s sustained submaximal isometric contraction. The target torque was set at 50 % of the highest torque achieved in the knee extension isometric MVC test at 80° at the day of testing. Before each test, two to three practice contractions were performed until the subject managed to reach the visualized 50 % MVC target torque without difficulty. Following a 2-min rest, the subjects were instructed to quickly reach the target torque and maintain it for 90 s without interruptions. Verbal encouragement was provided to the subjects to reach the target torque until the finish time. As some subjects could not sustain the 90-s contraction without interruptions, the time to task failure was assessed as the time until the torque declined >5 % of the initial value for a period longer than 2 s.
Motor unit activity in the m. vastus laterals during knee extension and in the m. gastrocnemius medialis during plantar flexion was recorded with bipolar differential electromyography (EMG) using a Noraxon MyoSystem 1400A and Ag–AgCl surface electrodes directly connected with pre-amplifiers. Before applying the electrodes, the skin was shaved, cleaned and scrubbed with sandpaper. Skin–electrode resistance was checked for being lower than 10 kΩ and the skin was re-prepared if needed. The correct position of the electrodes was verified by M-wave assessments and this position was retained across the study using cutaneous ink marks. To also ensure consistency across campaigns, individual EMG templates (transparent plastic sheets with identified landmarks and electrode locations) were constructed during the initial pretest and subsequently used prior to all tests. Torque and EMG signals were digitized at 1,000 Hz using Noraxon software and processed by customized software. For each MVC trial, the peak torque value of a 0.5 s interval was assessed and EMG amplitude (RMS) was averaged for that time. The contraction that yielded the highest peak torque value at any of the angles was finally selected for comparisons. To discriminate between changes in RMS due to alterations in neural drive from changes due to alterations in peripheral factors, we normalized the RMS to the RMS of the M-wave (Arabadzhiev et al. 2010). RMS and median frequency (FM) during the knee extensor fatiguing contraction were assessed over the first 2 s and the 2 s immediately preceding task failure.
The countermovement jump test was assessed before (BDC-1) and following bed rest (R + 0). Subjects refrained from any type of exercise on the days of testing. Subjects stood on a ground reaction force plate (Leonardo, Novotec Medical GmbH, Pforzheim, Germany) with their hand on their hips. When stipulated by the Leonardo software, subjects flexed their knees and subsequently jumped as high as possible. During the jump, the hands remained on the hips. Trials were repeated when the subjects landed outside the platform, or had to be actively supported in maintaining balance following landing. The procedure was repeated until three valid trials had been acquired. The assessment of maximal force, maximal velocity and maximal jump height and countermovement depth (i.e., the lowering of the center of mass during the countermovement) was performed using the ground reaction forces, software provided by the manufacturer as well as customized software.
Before the actual functional testing procedures at BDC-1, each subject was familiarized with the equipment and the proper techniques during dedicated sessions scheduled at BDC-3. This familiarization session was identical in setup as the actual testing sessions, but the data are not included in the comparisons.
Biological sample collection
Fasting blood samples were taken on days BDC-3, BDC-1; HDT2, and HDT5; and R + 1 and R + 5 in the supine or HDT position under standardized conditions at ~7:00 a.m. shortly after subjects awakened. Whole blood was centrifuged after coagulation (3,000 rpm, 4 °C, 10 min), and serum was distributed in small aliquots and immediately frozen at −80 °C until analysis. Urine was collected as 24-h urine pools on all study days from ±7:00 a.m. to ±7:00 a.m. on the following day. Single voids were stored under darkened and cooled conditions until final pooling into 24-h volumes. The subsequently obtained aliquots were stored at −20 °C.
Serum concentrations of bone formation markers bAP and P1NP, as well as urinary bone resorption markers NTX and CTX, were determined with commercially available assays in the in-house laboratory of the Institute of Aerospace Medicine (bAP: Tandem R, Ostase, Hybritech, Liege, Belgium; PINP: Orion Diagnostica, Finland; NTX: Osteomark, Wampole Laboratories, Princeton, NJ; CTX: Crosslaps, Osteometer BioTech, Herlev, Denmark). Interassay and intraassay variations were as follows. Interassay were bAP, 8.8 %; PINP, 3.5 %; NTX, 4.0; CTX, 5.5 %; intraassay were bAP, 7.4 %; PINP, 3.5 %; NTX, 1.5 %; CTX, 2.5 %. Urinary calcium concentrations were analyzed in duplicate by flame photometry (EFOX 5053, Eppendorf, Germany).
Total urinary nitrogen was determined by highly sensitive chemiluminescence with a TNM-1 automated analyzer (Total Nitrogen Measuring Unit, Shimadzu, USA) and sample injector ASI-V (Shimadzu, USA). Within a series of control analyses performed each day using freshly prepared calibrators the coefficient of variation of this method was 1.00 % and the recovery was 103.68 %. Nitrogen balance was estimated as nitrogen intake (protein/6.25) minus urinary nitrogen excretion. Because nitrogen losses through skin and feces are very low and regarded as constant (Frings-Meuthen et al. 1985), these were not taken into account when calculating nitrogen balance.
In all parameters the effects of interventions (CON, STA, LRT) and sessions (BDC, HDT1-HDT5, R) were analyzed using Linear mixed effect (LME) models with sessions and intervention as fixed effects and subject ID as random effect. Variances were allowed to differ between participants, and LME models were optimized according to the Akaike information criterion [see p.353 and p.652 in (Crawley 2007)]. Data were box–cox transformed where indicated by test on normal distribution [Kolmogorov–Smirnov (SPSS) or Shapiro (R)], non-linear quantile–quantile plots or in case of heteroscedasticity. Initially, all HDT days, as well as the recovery days, were tested against the lumped BDC days. For bone resorption, day HDT1 was lumped with the BDC data. This was done because bone resorption is known to increase in the second day of bed rest. Hence, the first day of bed rest, from a bone physiology point of view, pertains to BDC rather than to HDT. The data of CSA were only obtained pre and post bed rest. Models were then further simplified in a step-wise manner when no significant intervention effects were found. First, data from all HDT days were lumped together, and so were data from all recovery days to yield the HDT phase and the REC phase. When there were still no significant effects, then STA and CON interventions were pooled to yield NO_LRT vs LRT comparisons. The last step of model simplification was the deletion intervention effects, so that pure phase effects were analyzed as the simplest model. Where ANOVA indicated significant effects, these were followed up with treatment contrasts with BDC as the reference. In addition GLMs were calculated to evaluate specific differences among certain sessions. Statistical analyses were carried out in SPSS v 20.0 and in the “R” statistical environment (version 2.9.2, www.r-project.org). Data are given as means and standard errors (SE) if not indicated otherwise. The level for statistical significance was set to α < 0.05.