Graded reductions in pre‐exercise glycogen concentration do not augment exercise‐induced nuclear AMPK and PGC‐1α protein content in human muscle

What is the central question of this study? What is the absolute level of pre‐exercise glycogen concentration required to augment the exercise‐induced signalling response regulating mitochondrial biogenesis? What is the main finding and its importance? Commencing high‐intensity endurance exercise with reduced pre‐exercise muscle glycogen concentrations confers no additional benefit to the early signalling responses that regulate mitochondrial biogenesis.


INTRODUCTION
The concept of deliberately commencing endurance exercise with reduced muscle glycogen (i.e. the train-low paradigm, Burke et al., 2018) is now recognised as a potent nutritional strategy that is able to modulate acute skeletal muscle cell signalling Wojtaszewski et al., 2003;Yeo et al., 2010) and transcriptional responses Pilegaard et al., 2002;Psilander, Frank, Flockhart, & Sahlin, 2013). Furthermore, repeated bouts of train-low exercise can subsequently augment many hallmark muscle adaptations inherent to the endurance phenotype. Indeed, the strategic periodisation of dietary carbohydrate (CHO) in order to commence exercise with low muscle glycogen (during 3-10 weeks of training) enhances mitochondrial enzyme activity and protein content (Hansen et al., 2005;Morton et al., 2009;Yeo et al., 2008) and whole body and intra-muscular lipid metabolism (Hulston et al., 2010), and, in some instances, improves exercise capacity (Hansen et al., 2005) and performance (Marquet et al., 2016a(Marquet et al., , 2016b. As such, the train-low paradigm and wider CHO periodisation strategies have subsequently gained increased recognition amongst athletic populations (Burke et al., 2018;Impey et al., 2018;Stellingwerff, 2012).
Skeletal muscle glycogen appears to exert its regulatory effects primarily through the AMP-activated protein kinase (AMPK)peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) signalling axis, whereby exercise-induced AMPKα2 activity (Wojtaszewski et al., 2003), phosphorylation (Yeo et al., 2010) and nuclear abundance (Steinberg et al., 2006) are all augmented under conditions of reduced pre-exercise muscle glycogen. These effects may be partly mediated through the glycogen binding domain present on the β subunit of AMPK McBride, Ghilagaber, Nikolaev, & Hardie, 2009). This allows for the physical tethering of AMPK to the glycogen granule and may subsequently inhibit its translocation to the nucleus and its activation of transcriptional regulatory proteins. In this way, the subcellular localisation of AMPK may play an important role in the signal transduction pathway regulating train-low responses, whereby its translocation to the nucleus could allow for interaction with transcriptional regulatory proteins, such as PGC-1α, to control gene expression (McGee et al., 2003). Indeed, endurance exercise also appears to increase the nuclear abundance of PGC-1α (Little, Safdar, Cermak, Tarnopolsky, & Gibala, 2010;Little, Safdar, Bishop, Tarnopolsky, & Gibala, 2011), and may constitute the initial phase of exercise-induced adaptive responses.
Despite the potential regulatory role of muscle glycogen availability in exercise-induced cell signalling, the absolute concentrations of muscle glycogen required to facilitate such responses are currently unknown (Impey et al., 2018). Recent suggestions propose the potential existence of a muscle glycogen threshold, a metabolic window of absolute muscle glycogen concentration whereby the augmented signalling and transcriptional responses associated with train-low models are particularly evident (Impey et al., 2018). However, using a sleep-low, train-low model, we recently demonstrated that commencing exercise with stepwise reductions in pre-exercise muscle glycogen concentrations below 300 mmol (kg dw) −1 (within the range

New Findings
• What is the central question of this study?
What is the absolute level of pre-exercise glycogen concentration required to augment the exercise-induced signalling response regulating mitochondrial biogenesis?
• What is the main finding and its importance?
Commencing high-intensity endurance exercise with reduced pre-exercise muscle glycogen concentrations confers no additional benefit to the early signalling responses that regulate mitochondrial biogenesis.
Such findings may be related to the fact that commencing exercise with <300 mmol (kg dw) −1 is already a critical level of absolute glycogen (as suggested by Impey et al., 2018) that is required to induce a metabolic milieu conducive to cell signalling. We therefore suggested that future research should investigate stepwise reductions in preexercise muscle glycogen within a wider range of pre-exercise muscle glycogen concentration (i.e. 600-200 mmol (kg dw) −1 ) in order to investigate the existence of a potential glycogen threshold and allow for a better definition of its potential upper and lower limits.
With this in mind, the aim of the present study was to test the hypothesis that graded pre-exercise muscle glycogen concentrations modulate the exercise-induced nuclear abundance of AMPK and PGC-1α protein content as well as the transcription of genes with putative roles in the regulation of mitochondrial biogenesis. To achieve our model of graded pre-exercise muscle glycogen concentrations, we utilised an experimental protocol previously studied in our laboratory (Impey et al., 2016) consisting of an amalgamation of trainlow protocols whereby participants perform a glycogen depletion protocol on the evening of Day 1, consume a modified CHO intake throughout Day 2 and then perform fasted exercise on the morning of Day 3. In this way, we studied trained male cyclists who commenced an acute bout of work-matched non-exhaustive highintensity interval cycling (HIT) with low (∼200 mmol (kg dw) −1 ), moderate (∼350 mmol (kg dw) −1 ) or high (∼550 mmol (kg dw) −1 ) preexercise muscle glycogen concentrations.

Ethical approval
All subjects provided written informed consent and all procedures conformed to the standards set by the Declaration of Helsinki (2008).

F I G U R E 1
Schematic overview of the experimental protocol. Following the completion of an evening bout of glycogen-depleting cycling exercise subjects received three graded levels of CHO in order to manipulate pre-exercise muscle glycogen on the day of the main experimental trial. Following an overnight fast, subjects completed a high-intensity intermittent cycling exercise (8 × 5 min at 80% peak power output). Muscle biopsies were obtained pre-, post-and 3 h post-exercise

Participants
Eight endurance-trained amateur male cyclists (mean ± SD: age, 30 ± 10 years; body mass 72.6 ± 9.4 kg; height, 177.0 ± 8.9 cm) took part in this study. Mean peak oxygen consumption (V O 2 peak ) and peak power output (PPO) for the cohort were 60.4 ± 7.7 ml kg −1 min −1 and 338 ± 45 W, respectively. Cyclists were classed as trained according to guidelines for subject classification in research (De Pauw et al., 2013;Jeukendrup, Craig, & Hawley, 2000). A priori sample size determination was performed, assuming an effect of muscle glycogen on PGC-1α mRNA of 1.34 , an α-value of 0.05 and a power of 0.80. None of the subjects had any history of musculoskeletal or neurological disease, nor were they under any pharmacological treatment during the course of the testing period.

Experimental design
In a repeated measures design, with each experimental trial separated by a minimum of 7 days, subjects undertook an evening bout of glycogen depletion exercise followed by the consumption of graded quantities of CHO (L-CHO: 2 g kg −1 , M-CHO: 6 g kg −1 , H-CHO: 14 g kg −1 for low, medium and high groups, respectively) across a ∼36 h period so as to manipulate pre-exercise muscle glycogen prior to a bout of high-intensity interval exercise (8 × 5 min at 80% PPO). All trials were performed in a randomised and counterbalanced order. Skeletal muscle biopsies were obtained from the vastus lateralis immediately before, post-and 3 h post-exercise. An overview of the experimental protocol is shown in Figure 1. as previously completed in our laboratory (Hearris et al., 2019;Impey et al., 2016;Taylor et al., 2013). The pattern of exercise and total time to exhaustion in the subject's initial trial was recorded and replicated in all subsequent trials. Subjects were permitted to consume water ad libitum during exercise, with the pattern of ingestion replicated during subsequent trials.

Phase 2: carbohydrate re-feeding strategy
To facilitate the goal of achieving graded differences in muscle glycogen concentrations between trials, subjects were provided with varying amounts of CHO during the ∼36 h recovery period prior to the subsequent bout of high-intensity interval exercise. Within the immediate ∼4 h recovery period following glycogen-depleting exercise, subjects in the H-CHO trial were provided with CHO at a rate of 1 g kg −1 h −1 for 3 h from a mixture of CHO drinks and gels (Science in Sport, Nelson, UK) followed by a high carbohydrate meal providing a further 1 g kg −1 CHO. Subjects in the M-CHO trial consumed 1 g kg −1 CHO immediately following exercise whilst subjects in the L-CHO trial refrained from CHO intake throughout the remainder of the evening. Across all trials, subjects also consumed 30 g of whey protein isolate (Science in Sport) mixed with 500 ml of water immediately following the cessation of glycogen-depleting exercise to reflect realworld practice as per current nutritional guidelines (Thomas et al., 2016). Over the course of the following day (Day 2), subjects consumed either 10 g kg −1 (H-CHO), or 5 g kg −1 (M-CHO) or 2 g kg −1 (L-CHO) carbohydrate (i.e. between 09.00 and 21.00 h). In this way, total CHO intakes in the H-CHO, M-CHO and L-CHO trials equated to 14, 6 and 2 g kg −1 CHO, respectively, over the course of the ∼36 h following glycogen depletion. In all trials, subjects also consumed 2 g kg −1 protein and 1 g kg −1 fat with fluid intake allowed ad libitum. Total energy intake over the course of the ∼36 h equated to 5306 ± 687, 2978 ± 386 and 1816 ± 235 kcal in the H-CHO, M-CHO and L-CHO trials, respectively.

Phase 3: high intensity interval cycling
Subjects arrived at the laboratory between 08.00 and 09.00 h on the morning of Day 3 (in a fasted state), where a venous blood sample was collected from the antecubital vein and a muscle biopsy taken from the vastus lateralis. Subjects then completed the high-intensity interval (HIT) cycling protocol, consisting of 8 × 5 min intervals at 80% PPO, interspersed with 1 min rest. During exercise, heart rate (HR) was continuously measured and the final HR for each 5 min interval was recorded whilst ratings of perceived exertion (RPE) were recorded upon completion of each interval (Borg, 1982). Expired gas was collected via a mouthpiece connected to an online

Blood analysis
Venous blood samples were collected in vacutainers containing K 2 EDTA, lithium heparin or serum separation tubes and stored on ice or at room temperature until centrifugation at 1500 g for 15 min at 4 • C. Samples were collected immediately prior to and after exercise. Plasma was aliquoted and stored at −80 • C until analysis. Samples were later analysed for plasma glucose, lactate, non-esterified fatty acids (NEFA) and glycerol using commercially available enzymatic spectrophotometric assays (RX Daytona Analyser, Randox Laboratories, Crumlin, UK) as per manufacturer instructions.

Muscle biopsies
Skeletal muscle biopsies (∼60 mg) were obtained from the vastus lateralis immediately prior to exercise, immediately upon completion of the exercise bout and at 3 h post-exercise. Muscle biopsies were obtained from separate incision sites 2-3 cm apart using a Bard Monopty Disposable Core Biopsy Instrument (12-gauge × 10 cm length, Bard Biopsy Systems, Tempe, AZ, USA) under local anaesthesia (0.5% Marcaine) and immediately frozen in liquid nitrogen and stored at −80 • C for later analysis.

RNA isolation and analysis
Muscle samples (∼20 mg) were homogenised in 1 ml TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA) and total RNA isolated by phenol-chloroform extraction. Concentrations and purity of RNA were assessed by UV spectroscopy at optical densities of 260 and 280 nm, using a NanoDrop 3000 (Thermo Fisher Scientific, Rosklide, Denmark) with an average 260/280 ratio of 1.93 ± 0.08. A quantity of 50 ng total RNA was used for each 20 μl PCR reaction. conditions. As such, the relative gene expression levels were calculated using the comparative C t (2 −ΔΔC t ) equation (Schmittgen & Livak, 2008) where the relative expression was calculated as 2 −ΔΔC t , where C t represents the threshold cycle. mRNA expression for all target genes was calculated relative to the reference gene (B2M) within the same subject and condition and relative to the pre-exercise value in the H-CHO condition. The primers used are shown in Table 1.

Subcellular fractionation
Isolation of nuclear and cytosolic fractions was performed using

Statistical analysis
All statistical analyses were performed using SPSS Statistics Version

Skeletal muscle glycogen
Muscle glycogen displayed main effects for condition (P < 0.001), time (P < 0.001) and a treatment × time interaction effect (P = 0.001).
As such, the exercise and nutritional strategy employed was successful in achieving graded levels of pre-exercise muscle glycogen

Physiological and metabolic responses to exercise
Average exercise intensity across the HIT session equated to 85 ± 8%, 85 ± 8% and 89 ± 9%V O 2 peak for H-CHO, M-CHO and L-CHO trials, respectively. Subjects' average heart rate ( Figure 3a

Correlation between muscle glycogen and cell signalling responses
The absolute concentration of muscle glycogen immediately postexercise demonstrated a significant correlation with PGC-1α mRNA expression 3 h post-exercise (P = 0.048, r = −0.416; Figure 6). In contrast, no correlation was observed between post-exercise muscle

DISCUSSION
The aim of the present study was to test the hypothesis that graded pre-exercise muscle glycogen concentrations modulate the exerciseinduced nuclear abundance of AMPK and PGC-1α protein content as well as the transcription of genes with putative roles in the regulation of mitochondrial biogenesis. Using trained male cyclists, we demonstrate that commencing an acute bout of work-matched and non-exhaustive HIT cycling with graded pre-exercise muscle glycogen (within a range of 600-200 mmol (kg dw) −1 ) does not modulate such early signalling responses. In the context of manipulating CHO availability around training, our data suggest that the metabolic stress of HIT exercise may override any potential effect of pre-exercise muscle glycogen and induce negligible modulatory effects on skeletal muscle that is already subjected to the local metabolic challenge of high-intensity exercise.
To address our aim, we utilised an experimental protocol consisting of a purposeful amalgamation of previous train-low protocols whereby participants perform a glycogen depletion protocol on the evening of Day 1, consume a modified CHO intake throughout Day 2 and then perform fasted exercise on the morning of Day 3. This approach has been studied previously in our laboratory (Impey et al., 2016) and provides an extended recovery period following glycogen-depleting exercise (∼36 h) during which larger amounts of CHO can be consumed when compared to the 12 h period associated with the traditional sleep-low, train-low protocol (Hearris et al., 2019;Lane et al., 2015).
Indeed, this extended feeding period circumvents the inability to fully restore muscle glycogen to normative resting concentrations after  (Areta & Hopkins, 2018). Whilst we acknowledge that the manipulation of CHO intake also results in considerable differences in total energy intake, such short-term differences in energy intake do not appear to influence the molecular response to exercise (Hammond et al., 2019). In contrast to our previous work (Hearris et al., 2019), we also employed a workmatched, non-exhaustive HIT exercise protocol in an attempt to prevent the depletion of muscle glycogen to the absolute concentrations normally associated with exhaustion (i.e. <100 mmol (kg dw) −1 ; Hearris et al., 2019;Impey et al., 2016;Taylor et al., 2013). Despite differences in absolute glycogen utilisation between trials, we also observed a comparable relative utilisation (∼50%) between trials, a magnitude of use that is consistent with that previously observed using this exercise protocol (Stepto, Martin, Fallon, & Hawley, 2001). As such, participants in all trials finished the HIT protocol with absolute glycogen concentrations <300 mmol (kg dw) −1 .
In relation to post-exercise signalling, the exercise-induced increase in nuclear AMPK protein content was not augmented in response to stepwise reductions in pre-exercise muscle glycogen concentrations, which is in contrast to previous reports of enhanced nuclear AMPK content when exercise is commenced with low pre-exercise muscle glycogen (Steinberg et al., 2006). Given that nuclear AMPK translocation in response to exercise may be partly regulated by absolute glycogen concentrations (due to the physical tethering of AMPK to the glycogen granule; Steinberg et al., 2006)  CHO, 100 ± 42) when compared with that of the low glycogen trial in previous work (17 ± 6 mmol (kg dw) −1 ) (Steinberg et al., 2006). In fact, post-exercise muscle glycogen concentrations in the present L-CHO trial are comparable to the control condition (111 ± 35 mmol (kg dw) −1 ) of previous work (Steinberg et al., 2006) and may suggest that the augmentation of nuclear AMPK may only occur at extremely low absolute glycogen concentrations normally associated with exhaustion.
Irrespective of muscle glycogen, the use of high-intensity endurance exercise within the present study is known to induce significantly greater metabolic stress and augments the phosphorylation and activation of AMPK when compared with low-intensity exercise models (Combes et al., 2015;Egan et al., 2010;Fiorenza et al., 2018;Wojtaszewski, Nielsen, Hansen, Richter, & Kiens, 2000). Given that such metabolic fluctuations ultimately regulate signalling kinase activity, the completion of high-intensity endurance exercise may provide a sufficient local metabolic challenge to skeletal muscle whereby reducing pre-exercise muscle glycogen induces negligible further modulatory effects on skeletal muscle. Although augmented signalling in response to high-intensity exercise commenced with low muscle glycogen has been previously observed (Yeo et al., 2010), However, as PGC-1α is regulated by numerous post-translational modifications, it is unclear whether an increase in its nuclear protein content would have been observed later in recovery within the present study.
In relation to exercise-induced gene expression, we demonstrate that the expression of genes with putative roles in mitochondrial biogenesis and substrate utilisation is not augmented under conditions of reduced muscle glycogen concentrations. These candidate genes were chosen upon the basis of their reported time course of expression and to allow comparison with previously studied trainlow investigations (Hammond et al., 2019;Impey et al., 2016;Lane et al., 2015;Psilander et al., 2013). The exercise-induced increase in PGC-1α, p53 and CPT-1 are consistent with previous reports following a similar bout of high-intensity running (Hammond et al., 2019). However, the lack of change in other mitochondria-related genes is unclear given they display increased expression within the chosen time course across a range of exercise modalities (Fiorenza et al., 2018;Impey et al., 2016). In contrast to the previously reported effects of muscle glycogen concentrations on the regulation of gene expression (Pilegaard et al., 2002(Pilegaard et al., , 2005Bartlett et al., 2013;Psilander et al., 2013), these data support our previous findings ( concentrations (∼100 mmol (kg dw) −1 ), despite marked differences in pre-exercise concentrations (600 vs. 300 mmol (kg dw) −1 ) and exercise duration (Impey et al., 2016). Furthermore, these findings are supported by data that demonstrate post-exercise muscle glycogen concentrations regulate exercise-induced PGC-1α mRNA expression (Fiorenza et al., 2018;Pilegaard et al., 2005;Psilander, 2014) and total protein abundance (Mathai, Bonen, Benton, Robinson, & Graham, 2008).
In summary, we provide novel data by demonstrating that graded pre-exercise muscle glycogen (within a range of 600-200 mmol (kg dw) −1 ) does not modulate the exercise-induced nuclear abundance of AMPK or PGC-1α nor does it affect the expression of genes with regulatory roles in mitochondrial biogenesis and substrate utilization. Practically, these data suggest that the additional stress of low pre-exercise muscle glycogen may not be required when performing high-intensity exercise that already subjects skeletal muscle to a sufficient metabolic challenge and may be better suited during conditions that do not elicit such cellular perturbations (e.g. prolonged low-intensity exercise completed below lactate threshold). Restriction of CHO availability for the latter training sessions would also circumvent the impairment in selfselected training intensity observed when high-intensity exercise is performed with reduced muscle glycogen (Lane et al., 2013;Yeo et al., 2008). Given that post-exercise muscle glycogen concentrations were reduced to low levels across all trials (i.e. 100-250 mmol (kg dw) −1 ), our data raise the possibility that the absolute post-exercise muscle glycogen concentration may also be an important factor in regulating exercise-induced skeletal muscle signalling responses associated with mitochondrial biogenesis, and further work is now required.

COMPETING INTERESTS
None declared.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.