Sprint-interval but not continuous exercise increases PGC-1α protein content and p53 phosphorylation in nuclear fractions of human skeletal muscle

Sprint interval training has been reported to induce similar or greater mitochondrial adaptations to continuous training. However, there is limited knowledge about the effects of different exercise types on the early molecular events regulating mitochondrial biogenesis. Therefore, we compared the effects of continuous and sprint interval exercise on key regulatory proteins linked to mitochondrial biogenesis in subcellular fractions of human skeletal muscle. Nineteen men, performed either 24 min of moderate-intensity continuous cycling at 63% of WPeak (CE), or 4 × 30-s “all-out” cycling sprints (SIE). Muscle samples (vastus lateralis) were collected pre-, immediately (+0 h) and 3 (+3 h) hours post-exercise. Nuclear p53 and PHF20 protein content increased at +0 h, with no difference between groups. Nuclear p53 phosphorylation and PGC-1α protein content increased at +0 h after SIE, but not CE. We demonstrate an exercise-induced increase in nuclear p53 protein content, an event that may relate to greater p53 stability - as also suggested by increased PHF20 protein content. Increased nuclear p53 phosphorylation and PGC-1α protein content immediately following SIE but not CE suggests these may represent important early molecular events in the exercise-induced response to exercise, and that SIE is a time-efficient and possibly superior option than CE to promote these adaptations.


Total work, performance parameters and blood lactate concentration ([La − ]) during the biopsy trial.
Total work during the biopsy trial was higher for CE compared with SIE (3.6-fold, P < 0.001; Table 1). In contrast, 1-s maximum and mean exercise intensity (expressed in Watts) were higher for SIE compared with CE (4.7-and 3.3-fold, respectively, all P < 0.001; Table 1). The exercise-induced increase in blood [La − ] for the two types of exercise (both P < 0.001) was also greater in SIE compared with CE (2.9-fold, P < 0.001; Table 1).

Muscle analyses. Representative immunoblots, cellular fractions enrichment, and antibody specificity.
Representative blots for the study are presented in Fig. 1a. The specificity of the p53 antibody was assessed by blotting samples beside a commercially-available, untagged, full-length p53 recombinant human protein (Aviva Systems Biology, AEP00002) (Fig. 1b). The same full-length p53 recombinant human protein, which was expressed in E. coli and was not phosphorylated, was also used as a negative control against the chosen p-p53 Ser15 antibody. Results from Fig. 1b show that the p-p53 Ser15 antibody did not recognise this protein, suggesting it is phospho-specific. The enrichment and purity of both nuclear and cytosolic fractions was confirmed by blotting the separated fractions against the nuclear protein histone H3, and the cytosolic protein lactate dehydrogenase A (LDHA). Histone H3 was only detected in nuclear fractions, whereas LDHA was only detected in cytosolic fraction (Fig. 1c), confirming the cellular fractionation protocol was successful. Images showing whole-lane Coomassie staining for both nuclear and cytosolic fractions, and histone H3 (nuclear) and GAPDH (cytosolic) immunoblots, were used to verify equal loading between lanes, and representative images are presented in Fig. 1d and e respectively. p53 protein content. There were main effects of time (nuclear: P = 0.002; cytosolic: P = 0.012), with the content of p53 protein significantly increased in both the nuclear (2.0-fold; P = 0.001) and cytosolic (1.8-fold; P = 0.014) fraction at + 0 h only. There was no interaction effect (nuclear: P = 0.283; cytosolic: P = 0.393), indicating no significant difference in p53 protein content between the two types of exercise in either the nuclear (1.5-fold; ES: 1.61; − 0.76, 3.97) or cytosolic (1.4-fold; ES: 0.57; − 0. 35, 1.48) fraction at + 0 h. There were no significant differences versus baseline or between the two types of exercise at + 3 h in either fraction. (Fig. 2a and b). PHF20 protein content. There were main effects of time (nuclear: P = 0.006; cytosolic: P < 0.001), with the content of PHF20 protein significantly increased in both the nuclear (1.9-fold; P = 0.004) and cytosolic (1.7-fold; P < 0.001) fraction at + 0 h only. There was no interaction effect (nuclear: P = 0.864; cytosolic: P = 0.399), Scientific RepoRts | 7:44227 | DOI: 10.1038/srep44227 indicating no significant difference in PHF20 protein content between the two types of exercise in either the nuclear (1.3-fold; ES: 0.22; − 0.83, 1.26) or cytosolic (1.1-fold; ES: 0.49; − 0.37, 1.34) fraction at + 0 h. There were no significant differences versus baseline or between the two types of exercise at + 3 h in either fraction. (Fig. 2c and d).
Phosphorylation of Acetyl-CoA Carboxylase (ACC) at serine 79 (p-ACC Ser79 ) protein content. p-ACC Ser79 was not detected in nuclear fractions (Fig. 1a). There was a main effect of time in the cytosolic fraction (P < 0.001), with p-ACC Ser79 content significantly increased (1.7-fold; P = 0.012) at + 0 h only. There was no interaction effect (P = 0.092), indicating no significant difference in p-ACC Ser79 content between the two types of exercise at + 0 h (1.2-fold; ES: 0.92; − 0.21, 2.05). There were no significant differences versus baseline or between the two types of exercise at + 3 h (Fig. 3a).
Gene expression. There was no change in p53 mRNA content throughout (main effect of time: P = 0.883; Fig. 4a). There was a main effect of time for PGC-1α mRNA content (P < 0.001; Fig. 4b), with a significant increase (5.6-fold, P < 0.001) at + 3 h only. There was no interaction effect (P = 0.910), indicating no significant difference in PGC-1α mRNA content between the two types of exercise at + 3 h (1.1-fold; ES: 0.147; − 0.85, 1.13); there were no significant differences versus baseline or between the two types of exercise at + 0 h (all P > 0.05). Finally, results relative to the mRNA content of apoptosis-inducing factor (AIF), dynamin-related protein 1 (DRP1), mitofusin 2 (MFN2), p21, superoxide dismutase 2 (SOD2), and cytochrome c are presented in Table 2.

Discussion
This study reports, for the first time, that only SIE, and not CE, was associated with a post-exercise increase in p-p53 Ser15 in the nuclear fraction of human skeletal muscle. Similarly, only SIE, and not CE, increased nuclear PGC-1α protein content post-exercise. Although the differential response to SIE and CE may be due to differences in the degree and time course of nuclear p53 and PGC-1α protein accumulation, our findings suggest that  Protein content of nuclear (a) and cytosolic (b) p53, of nuclear (c) and cytosolic (d) PHF20, and of nuclear (e) and cytosolic (f) p-p53 Ser15 before (Pre), immediately (+ 0 h) and 3 h (+ 3 h) after the CE and SIE trials. Open circles (CE) and closed triangles (SIE) represent individual values. Primary effects were analysed using two-way ANOVA with repeated measures for time followed by Tukey's honestly significant difference post-hoc test for pairwise comparisons. *P < 0.05 vs. Pre; ‡ P < 0.05 vs. CE at the same time point. Individual data points and mean bars are plotted. n = 9 for each type of exercise. SIE may represent a more potent stimulus than CE to induce some of the early molecular events associated with mitochondrial biogenesis, as previously suggested 31 . An additional novel finding was that the exercise-induced increase in the nuclear content of p53 protein in human skeletal muscle was also accompanied by an increase in the nuclear content of PHF20, a protein that stabilises and activates p53 29 , but was not differentially regulated by the two types of exercise investigated.
Nuclear (and cytosolic) p53 protein content was increased immediately post-exercise, with no difference between the two types of exercise. Previous research in human skeletal muscle also reported an increase in nuclear p53 protein content after a 60-min bout of CE 18 ; however, this was only reported after 3 hours of recovery (the only time point investigated in this previous study). Our results also agree with reports of increased nuclear p53 protein content immediately after a 60-min bout of continuous running in mouse skeletal muscle 25 , and 1 hour after a 20-min bout of intermittent eccentric skeletal muscle contractions in rats 32 . A decrease in nuclear p53 protein content has also been observed in one study after a 90-min bout of exhaustive exercise in mouse skeletal muscle 26 . This contrasting result is difficult to explain, but may relate to the type, age, and sex of the species investigated, and the different types of muscle analysed. We add the new finding that SIE also induced nuclear p53 protein accumulation, and that this increase was similar to that observed for CE. Our results are consistent with the well-accepted notion that cellular stress is associated with accumulation of p53 protein in the nucleus, a process that is mainly mediated by post-translational events such as nuclear/cytosolic shuttling, and/or increased p53 protein stability 24,33 .
p53 nuclear/cytosolic shuttling, a process requiring a tightly-regulated series of events 33 , cannot simply be assessed by subcellular fractionation and immunoblotting. We were therefore unable to determine the contribution of subcellular shuttling to the increase in nuclear p53 protein content observed with both types of exercise. However, some indirect measurements in the present study lend support to an exercise-induced increase in p53 stability. For example, and for the first time, we report a post-exercise increase in the nuclear content of PHF20, a multidomain protein that can bind to p53 and increase its stability 29 . In unstressed cells, nuclear p53 is bound to murine double minute-2 (MDM2), a p53 negative regulator inducing p53 ubiquitination and nuclear export, and subsequent cytosolic degradation by the proteasome 24,34 . Upon cellular stress PHF20 binds to p53, enhancing its stability by diminishing MDM2-mediated p53 degradation 29 . Although we were unable to directly measure the PHF20-p53 interaction, the concomitant increase in these two proteins may indicate an increased PHF20-p53 interaction induced by the cellular stress of exercise, and a subsequent increase in p53 stability 29 . Future research is required to confirm this hypothesis, and to better understand the molecular events regulating p53 stability, degradation, and nuclear/cytosolic shuttling following exercise.
An additional factor affecting p53 stability is phosphorylation at serine 15, which reduces the p53 interaction with its negative regulator MDM2 27 . In the present study, nuclear p-p53 Ser15 was increased immediately post-exercise only following SIE, an increase that persisted until 3 h of recovery. Cytosolic p-p53 Ser15 was also increased at both time points; however, the difference between the two types of exercise did not reach significance. While exercise duration and total work were greater during CE (1.7-and 3.6-fold, respectively), maximal and mean exercise intensity were 4.7-and 3.3-fold greater during SIE (Table 1). This suggests that exercise intensity, rather than exercise duration or total work, may be an important factor affecting exercise-induced changes in nuclear p-p53 Ser15 . The only two previous studies investigating p-p53 Ser15 following endurance exercise with humans reported whole-muscle p-p53 Ser15 to increase only after 3 h of recovery, regardless of exercise type  (continuous or intermittent exercise at the same average intensity) 28 or pre-exercise carbohydrate availability 35 . The earlier increase in nuclear p-p53 Ser15 in our study (+ 0 h), compared with that in whole-muscle p-p53 Ser15 in the two studies above (+ 3 h), raises the possibility that, similar to PGC-1α 14 , an increase in nuclear p-p53 Ser15 may contribute to the initial phase of the exercise-induced adaptive response. While multiple mechanisms probably contribute, the signalling kinases AMP-activated protein kinase (AMPK) 36 and p38 MAPK 37 have been reported to phosphorylate p53 at serine 15. In the present study, cytosolic phosphorylation of ACC, a downstream target and commonly used marker of AMPK activation 38,39 , was increased similarly between the two types of exercise immediately post-exercise. No ACC was detected in nuclear fractions, consistent with previous research 16 . Conversely, only SIE induced a significant increase in the phosphorylation of p38 MAPK in the nucleus, immediately post-exercise, consistent with the increase in nuclear p-p53 Ser15 only following SIE. This may suggest that signalling through p38 MAPK may be more closely associated with the exercise-induced modulation of p-p53 Ser15 . PHF20 can also act as a transcription factor, and has been shown to activate p53 gene expression 30 . Nonetheless, despite an increase in nuclear PHF20 protein content following both types of exercise, there was no significant change in p53 mRNA content within 3 h from the end of the bout. Studies in cells indicate that upregulation of p53 mRNA by PHF20 takes place only after 6 or 12 h 30 , suggesting that the 3 h post-exercise time point chosen in the present study might have been too early to detect a significant increase in p53 mRNA. Consistent with this, a small increase (~1.3-fold) has been reported following 3 h of recovery in human skeletal muscle with no difference between the three exercise intensities investigated 40 . Conversely, greater increases (2-to 2.5-fold) have been observed 4.5 and 7.5 h after the termination of the first running session of a twice a day exercise model 41 . The reasons for these divergent results may relate to the fact that participants engaged in eccentric exercise and that were fed before the post-exercise biopsies. Future research is required to clarify the controversial literature on this topic. p53 is an inducible transcription factor that directly regulates the transcription of, amongst others, PGC-1α , DRP1, MFN2, AIF, p21 and SOD2 -a series of genes implicated in mitochondrial biogenesis 19 , mitochondrial remodelling 21,22 , cell survival 42,43 and oxidative stress 44 . We report an exercise-induced increase in the mRNA content of these genes, although no significant differences between the two types of exercise were found. This is somewhat surprising considering the larger increase in nuclear p-p53 Ser15 (and PGC-1α ) after SIE compared with CE. Of note however, despite not reaching significance due to the large individual variability, the fold-change in p21 mRNA was almost twice as high following SIE compared to CE.
We observed an increase in nuclear PGC-1α protein content immediately after the termination of SIE (2.3-fold) and after 3 hours of recovery (1.7-fold). The increase we report after 3 h of recovery is similar to that observed in previous research in humans following an identical bout of SIE (1.7-fold) 15 ; however, in the present study this increase was also observed immediately post-exercise. In contrast to SIE, CE (24 min at 63% of W Peak ) did not provide a sufficient stimulus to induce a significant increase in nuclear PGC-1α protein content immediately post-exercise (1.4-fold), or after 3 hours of recovery (1.0-fold). Although these results may partially depend on the relatively short duration of the CE trial, they are consistent with the non-significant change (1.1-fold) in nuclear PGC-1α protein content reported immediately after 60 min of CE at 74% of  VO 2Peak 17 , the 1.5-fold increase observed immediately after a 90-min bout of CE at 65% of  VO 2Peak 16 , and the non-significant change (0.70-fold) recorded 3 hours after a 60-min bout of CE at 70% of  VO 2Peak 18 . Taken collectively, these results seem to indicate that CE at 60-75% of  VO 2Peak does not induce a large increase in nuclear PGC-1α protein content. However, by directly comparing participants of similar fitness levels within the same study, we report the novel finding that SIE is associated with larger increases in nuclear PGC-1α protein content than CE.
Similar to the exercise-induced increases in nuclear p-p53 Ser15 , and for the same reasons, the present results suggest that exercise intensity may also be an important factor affecting exercise-induced changes in nuclear PGC-1α protein content. Results from the cytosolic pool are also consistent with an exercise-intensity effect on the regulation of PGC-1α protein. Similar to the nuclear fraction, only SIE induced an increase in cytosolic PGC-1α protein content; however, this increase took place only after 3 h of recovery. The delayed increase in cytosolic compared with nuclear PGC-1α protein content provides further evidence that the initial phase of the exercise-induced adaptive response may indeed take place in the nucleus 14 .
The increase in the nuclear content of PGC-1α protein has previously been attributed to the exercise-induced translocation of PGC-1α from the cytosol to the nucleus [14][15][16]45 . While translocation remains possible, an alternative or additional explanation for the concomitant increase in nuclear and cytosolic PGC-1α protein content in the present study may relate to an increase in PGC-1α stability. An increase in p38 MAPK activity has previously been reported immediately post-exercise in humans [15][16][17]28 , and has been linked with greater PGC-1α stability 46 . The greater increase in the phosphorylation of p38 MAPK following SIE compared with CE in both subcellular fractions (both by 2.0-fold) occurred concomitantly with a greater increase in both nuclear and cytosolic PGC-1α protein content (both by 1.7-fold). It is therefore possible that exercise intensity may modulate PGC-1α protein content via an increase in PGC-1α stability, which may be mediated, at least in part, by greater phosphorylation of p38 MAPK 46 .
The PGC-1α protein has been shown to activate its own promoter through a feed-forward loop 47 ; as a result, increased content of nuclear PGC-1α protein should further enhance PGC-1α transcriptional activity before degradation 45 . However, despite a greater increase in nuclear PGC-1α protein content following SIE, when compared to CE, there was no difference for the increase in PGC-1α mRNA content following the two exercise types. This highlights a potential dissociation between exercise-induced increases in PGC-1α nuclear protein content and PGC-1α mRNA (and the mRNA of cytochrome c, a downstream target of PGC-1α 48 ). This may be related to other factors that influence PGC-1α transcriptional activity such as AMPK, a signalling protein inducing PGC-1α activation 39,49 , and its downstream target and commonly used biomarker ACC 38,39 , the phosphorylation of which was similarly increased following both exercise types. The similar increase in PGC-1α mRNA between the two types of exercise is also consistent with previous research comparing exercise at intensities below and above W Peak 40,50 , and is in agreement with the notion that the reported exercise intensity-dependent regulation of PGC-1α mRNA 51 is limited to submaximal (i.e., < W Peak ) exercise intensities 40 .
Although the between-subject design represents a potential limitation of this study, our findings add new insight into the early molecular events that regulate skeletal muscle remodelling in response to a single bout of exercise, and the role of exercise intensity in mediating these events. We report that a single bout of exercise induces nuclear accumulation of p53 protein, an increase that may relate to greater p53 stability, as suggested by the concomitant increase in PHF20 protein content. In addition, nuclear p-p53 Ser15 , a post-translational event also associated with enhanced p53 stability, and nuclear PGC-1α protein content, increased only following SIE, suggesting that exercise intensity may play an important role in the exercise-induced adaptations mediated by both p53 and PGC-1α . Results from the present study also indicate that increases in nuclear p-p53 Ser15 , as well as nuclear PGC-1α protein content, may represent important early events in the adaptive response to exercise. Our findings indicate that "all-out" SIE represents a valuable and possibly superior option to moderate-intensity CE for promoting the early molecular events leading to exercise-induced mitochondrial biogenesis, in a time-efficient manner.

Methods
Participants. Twenty healthy men aged 18-35 years, who were non-smokers, free of medications, moderately-trained (i.e., engaging in less than 3-4 hours per week of moderate, unstructured aerobic activity for 6 months prior to the study), and not regularly engaged in cycling-based sports, volunteered to participate in this research. Following medical screening participants were informed of the study requirements, benefits, and risks, before giving written informed consent. Approval for all the experimental protocols and the study's procedures, which conformed to the standards set by the latest revision of the Declaration of Helsinki, was granted by the Victoria University Human Research Ethics Committee. All experiments and procedures were performed in accordance with the relevant guidelines and regulations set by the above Human Research Ethics Committee.
Study design and testing. The experimental protocol consisted of two tests -a graded exercise test (GXT), and an exercise/biopsy trial. Participants were familiarised with both tests (with the exclusion of muscle biopsies) and were required to refrain from any strenuous exercise for the 72 h preceding each test, from alcohol and any exercise for 24 h before testing, and from food and caffeine consumption for 3 h before each test. After baseline testing, participants were ranked based on their W LT and assigned in reversed counterbalanced order (ABBA) to the CE or SIE group (both, n = 10), in a between-subjects study design. The between-subject design was necessary as this study was part of a longer training study in which participants then repeated their assigned exercise trial for four weeks, as previously described 12 . Nineteen participants completed the study, with one participant (SIE group) withdrawing due to time constraints. Participants' baseline physiological parameters are described in Table 1.

GXT.
A discontinuous graded exercise test was performed on an electronically-braked cycle ergometer (Lode Excalibur, v2.0, The Netherlands) to determine the HR Peak ,  VO 2Peak , W Peak , W LT (using the modified D Max method 52 ), and the exercise intensity for the biopsy trial, as previously described 12  and after each 4-min stage. Capillary blood [La − ] was determined using a pre-calibrated blood-lactate analyser (YSI 2300 STAT Plus, YSI, USA).
Biopsy trial. All trials were performed in the morning to avoid variations caused by circadian rhythms. To minimise variability in muscle gene and protein expression attributable to diet, participants were provided with a standardised dinner (55 kJ·kg −1 body mass (BM), providing 2.1 g carbohydrate·kg −1 BM, 0.3 g fat·kg −1 BM, and 0.6 g protein·kg −1 BM) and breakfast (41 kJ·kg −1 BM, providing 1.8 g carbohydrate·kg −1 BM, 0.2 g fat·kg −1 BM, and 0.3 g protein·kg −1 BM), to be consumed 15 and 3 h prior to the biopsy trials, respectively. While resting in the supine position, and after injection of a local anaesthetic (1% xylocaine) into the skin and fascia of the vastus lateralis muscle, three small incisions were made about 2-3 cm apart. A resting muscle biopsy (Pre) was obtained using a biopsy needle with suction before participants began either the CE or SIE protocol an electronically-braked cycle ergometer (Velotron, RacerMate, USA). A second muscle biopsy (+ 0 h) was obtained immediately upon completion of the exercise bout (< 5 s). A third muscle biopsy (+ 3 h) was obtained after 3 hours of recovery in the supine position (with no access to food and access to water ab libitum). Once obtained, muscle samples were immediately cleaned of excess blood, fat, and connective tissue, were rapidly frozen in liquid nitrogen, and stored at − 80 °C for subsequent analyses. Capillary blood samples to measure [La − ] were collected at rest, and immediately after the completion of exercise.
CE trial. Exercise consisted of 24 min of continuous cycling at a fixed power equivalent to 90% of W LT . This was preceded by a warm-up involving cycling for 6 min at 66% of W LT followed by 2 min of rest. Exercise intensity was set relative to W LT rather than W Peak as metabolic and cardiac stresses are similar when individuals of differing fitness levels exercise at a percent of the W LT , but can vary significantly when exercising at a percent of W Peak 53 . Participants received consistent verbal encouragement for the duration of the exercise bout. The overall duration of the CE exercise protocol (32 min of exercise inclusive of warm-up) was chosen based on the physical activity guidelines set by the ACSM position stand 8 .
SIE trial. Following the same warm-up procedure for CE, SIE consisted of 4 × 30-s "all-out" cycling bouts against a resistance set at 0.075 kg·kg −1 BM, interspersed with 4 min of rest (2 min of total exercise; 22 min of total session duration inclusive of warm-up). Participants received consistent verbal encouragement to keep the cadence as high as possible during the entire duration of the bout.
Skeletal muscle analyses. Subcellular fractionation. Nuclear and cytosolic fractions were prepared from 40-60 mg of wet muscle using a commercially-available nuclear extraction kit (NE-PER, Pierce, USA). Muscle samples were homogenised in CER-I buffer containing a protease/phosphatase inhibitor cocktail (Cell Signaling Technology [CST], 5872). Following centrifugation the supernatant was taken as the crude cytosolic fraction. Pellets containing nuclei were washed five times in PBS to remove cytosolic contamination, before nuclear proteins were extracted by centrifugation in high-salt NER buffer supplemented with the same inhibitors cocktail following manufacturers' instruction. Sufficient muscle was available to prepare subcellular fractions from nine participants in each group. Verification of subcellular enrichment is presented in the Results section.
Total RNA isolation. Total RNA was isolated from approximately 15-25 mg of muscle tissue using the RNeasy ® Mini kit (Qiagen, Canada) according to the manufacturer's instructions. Muscle samples were homogenised using the TissueLyser II (Qiagen, Canada), and total RNA was isolated from the aqueous phase following precipitation with 600 μ L of 70% ethanol using RNeasy ® Mini kit. On-column DNA digestion was performed. RNA concentration was determined by spectrophotometry (Nanodrop ND1000, Thermo Fisher Scientific, USA) by measuring the absorbance at 260 nm (A260) and 280 nm (A280), with A260/A280 ratios above 1.8 indicating high-quality RNA. Sufficient muscle was available to isolate and analyse total RNA from nine participants in each group.
Real-time RT-PCR. First-strand cDNA synthesis from 500 ng of total RNA was performed with random hexamer primers using a high-capacity cDNA reverse transcription kit (Applied Biosystems, USA), according to manufacturer's directions. All samples and reverse transcriptase (RT) negative controls were run together to prevent technical variation. Forward and reverse primers for the target and housekeeping genes (Table 3) were designed based on NCBI RefSeq using NCBI Primer-BLAST (www.ncbi.nlm.nih.gov/BLAST/). Specificity of the amplified product was confirmed by melting point dissociation curves. The mRNA expression of AIF, cyt c, DRP1, MFN2, p21, p53, PGC-1α and SOD2 were quantified by quantitative real-time RT-PCR (Mastercycler ® RealPlex2, Eppendorf, Germany), using a 10 μ L PCR reaction volume and SYBR Green chemistry (iTaqTM Universal SYBR ® Green Supermix, Bio-Rad, USA). All samples were run in duplicate simultaneously with Scientific RepoRts | 7:44227 | DOI: 10.1038/srep44227 template free controls, using an automated pipetting system (epMotion 5070, Eppendorf, Germany). The following PCR cycling patterns were used: initial denaturation at 95 °C (3 min), 40 cycles of 95 °C (15 s) and 60 °C (60 s). Relative changes in mRNA content were calculated using the normalised relative quantities (NRQs) method 55 . To account for the efficiency of RT and initial RNA concentration, the mRNA expression of four housekeeping genes was quantified, and their stability was determined using the BestKeeper software 56 . Cyclophilin, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and beta-2-microglobulin (B2M) were classified as stable, whereas TATA-binding protein (TBP) was reported as unstable and was therefore excluded. These results were confirmed by the Normfinder algorithm 57 .

Statistical analysis.
All values are reported as mean ± SD unless otherwise specified. Unpaired t-tests were used to assess differences between SIE and CE for Pre values in immunoblot analyses, and for age, height, body mass, HR Peak , W LT , W Peak ,  VO 2Peak , as well as 1-s max and mean exercise intensity, and total work during the biopsy trial. To investigate the influence of exercise type and time, and the interaction between both of these variables, two-way ANOVA with repeated measures for time were used. Where no interaction effects were observed, pooled values for time are reported. Significant interactions and main effects were further analysed using Tukey's honestly significant difference post-hoc test. Sigma Stat software (Jandel Scientific, USA) was used for all statistical analyses. The level of statistical significance was set a priori at P < 0.05. To assess the magnitude of effects, effect sizes (ES), assessed using Cohen's d, and 95% confidence intervals (95% CI), were also calculated and are reported as (ES; 95% CI) of the between-group difference (CE vs. SIE) scores.  Table 3. Primers used for real-time RT-PCR analyses of mRNA expression.