doi:10.1111/j.1365-2052.2011.02238.x PGC-1a encoded by the PPARGC1A regulates oxidative energy in equine during

S. S. Eivers*, B. A. McGivney*, J. Gu*, D. E. MacHugh*,†, L. M. Katz‡ and E. W. Hill* *Animal Genomics Laboratory, UCD School of Agriculture, Food Science and Veterinary Medicine, UCD College of Life Sciences, University College Dublin, Belfield, Dublin 4, Ireland. †UCD Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Belfield, Dublin 4, Ireland. ‡University Veterinary Hospital, UCD School of Agriculture, Food Science and Veterinary Medicine, UCD College of Life Sciences, University College Dublin, Belfield, Dublin 4, Ireland

Summary Peroxisome proliferator-activated receptor-c 1a (PGC-1a) has emerged as a critical control factor in skeletal muscle adaptation to exercise, acting via transcriptional control of responsible for angiogenesis, fatty acid oxidation, oxidative phosphoryla- tion, and muscle fibre type composition. In a previous study, we demonstrated a significant increase in mRNA expression for the gene encoding PGC-1a (PPARGC1A) in Thoroughbred horse skeletal muscle following a single bout of endurance exercise. In this study, we investigated mRNA expression changes in genes encoding transcriptional coactivators of PGC-1a and genes that function upstream and downstream of PGC-1a in known canonical pathways. We used linear regression to determine the associations between PPARGC1A mRNA expression and expression of the selected panel of

genes. Biopsy samples were obtained from the gluteus medius pre-exercise (T0), immediately

post-exercise (T1) and 4 h post-exercise (T2). Significant (P < 0.05) expression fold change

differences relative to T0 were detected for genes functioning in angiogenesis (ANGP2 and VEGFA); Ca2+-dependent signalling pathway (PPP3CA); carbohydrate/glucose metabolism (PDK4); fatty acid metabolism/mitochondrial biogenesis (PPPARGC1B); haem biosynthetic process (ALAS1); insulin signalling (FOXO1, PPPARGC1A and SLC2A4); mitogen-activated kinase signalling (MAPK14 and MEF2A); and myogenesis (HDAC9). Gene expres- sion associations were identified between PPARGC1A and genes involved in angiogenesis, mitochondrial respiration, glucose transport, insulin signalling and transcriptional regu- lation. These results suggest that PGC-1a and genes regulated by PGC-1a play significant roles in the skeletal muscle response to exercise and therefore may contribute to perfor- mance potential in Thoroughbred horses.

Keywords athletic performance, , PGC-1a, thoroughbred.

receptor-c coactivator-1a (PGC-1a) has emerged as a fun- Introduction damental turning point in understanding the molecular Skeletal muscle has a remarkable ability to respond to the contributions leading to exercise-induced phenotypic adap- metabolic stresses imposed by physical activity. The meta- tations in mammalian skeletal muscle, including oxidative bolic demand for fuel in response to exercise in skeletal phosphorylation, mitochondrial biogenesis, muscle fibre- muscle is limited by ATP availability, activity of oxidative type switching and angiogenesis (Handschin et al. 2003; enzymes, the availability of oxygen, and mitochondrial Arany 2008). Exercise is a powerful inducer of PPARGC1A content. The discovery of peroxisome proliferator-activated gene and PGC-1a protein expression in human and mouse skeletal muscle (Pilegaard et al. 2003; Russell et al. 2005; Address for correspondence Wende et al. 2005). During exercise, a number of signalling pathways are Emmeline W. Hill, Animal Genomics Laboratory, UCD School of Agriculture, Food Science and Veterinary Medicine, UCD College of activated that have been identified as key regulators of PGC- Life Sciences, University College Dublin, Belfield, Dublin 4, Ireland. 1a activity and function. The (Handschin et al. E-mail: [email protected] 2003), p38 mitogen-activated protein kinase (MAPK) Accepted for publication 28 March 2011 (Akimoto et al. 2005) and adenosine monophosphate

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 153 154 Eivers et al.

(AMP)-activated protein kinase (AMPK) (Jager et al. 2007) Children (Ireland), and ownersÕ consent was obtained for all signalling pathways influence PGC-1a activity directly or horses. via PGC-1a coactivation. PGC-1a is a powerful activator of a number of different transcriptional coactivators, such as Subjects nuclear respiratory factor 1 and 2 (NRF-1 and NRF2). These bind to and interact with a number of mitochon- The study cohort comprised eight 4-year-old untrained drial genes in the cell nucleus, resulting in increased mito- Thoroughbred horses (castrated males) raised at the same chondrial biogenesis (Lin et al. 2005). Regulation of farm for the previous 12 months and destined for National mitochondrial fatty acid oxidation occurs via PGC-1a Hunt racing with the same trainer. The horses had a mean interaction with peroxisome proliferator-activated receptor (±SD) height of 165.25 ± 1.44 cm and a mean pre-exercise alpha. In addition, PGC-1a has been shown to influence weight of 565.75 ± 13.71 kg. control of fatty acid oxidation via PDK4 by its interaction Details on subject physiological and biochemical charac- with oestrogen-related receptor alpha (ERRa) (Zhang et al. teristics are presented in Table 1. 2006; Wende et al. 2005). The forkhead transcription fac- tor FOXO1 also directly regulates PDK4 through binding Exercise protocol sites in the PDK4 gene promoter region (Kwon et al. 2004; Lin et al. 2005). The eight untrained Thoroughbred horses participated in a Aside from its regulatory role in fatty acid metabolism standardized incremental-step exercise test (Rose et al. and mitochondrial function, Arany et al. (2008) have 1990, 1990; Woodie et al. 2005) on a high-speed equine identified PGC-1a as a major regulator of angiogenesis treadmill (Sato; Sato AB). The treadmill was set to a 6° through coactivation of ERRa under pathologic ischaemia incline. The warm-up consisted of 2 min at 2 m/s, followed in addition to physiological conditions such as exercise by 2 min at 4 m/s and 2 min at 6 m/s. Warm-up was fol- (Hudlicka et al. 1992). In addition to the role of PGC-1a in lowed by an increase in treadmill velocity to 9 m/s for 60 s mitochondrial biogenesis and fatty acid oxidation, it also and then a 1 m/s increase in treadmill velocity every 60 s functions to regulate muscle fibre-type switching via tran- until the animal was no longer able to maintain its position scriptional coactivation of the myocyte enhancer factor 2 on the treadmill at that speed or until the heart rate reached

gene family (MEF2A, MEF2B, MEF2C and MEF2D), a plateau (HRmax). Heart rate was measured continuously inducing a fast-to-slow fibre-type switch (Wu et al. 2001). before, during and after exercise by telemetry (Polar Equine In a previous study, we observed the activation of the S810i heart rate monitor system; Polar Electro Ltd). Fol- pAMPKa and PGC-1a proteins in Thoroughbred horse lowing warm-up, the exercise test comprised an average of skeletal muscle following a single bout of exercise (Eivers six (range 5–7) incremental steps achieving a mean maxi- et al. 2009). Furthermore, we reported a significant increase mum velocity of 12.4 ± 0.2 m/s and a mean distance of in expression of the gene encoding PGC-1a (PPARGC1A). 4362.9 ± 102.7 m for an average duration of 8.77 ± The PGC-1a response and interaction with molecular sig- 0.5 min. Details on exercise parameters are presented in nalling pathways in the phenotypic adaptation of equine Table 1. skeletal muscle to exercise has not previously been reported. However, it is well established that PGC-1a plays a pivotal role in energy metabolism and resistance to fatigue in hu- Table 1 Exercise test details. mans and rodents. There have been numerous reports in Mean SD (±) the literature describing the effects of exercise on PPARG- C1A gene expression in these species (Pilegaard et al. 2003; Resting HR (bpm) 32.63 1.42 Maximum HR (bpm) 217.50 3.32 McGee & Hargreaves 2004; Akimoto et al. 2005; Russell Velocity at maximum HR (m/s) 12.43 0.24 et al. 2005; Vissing et al. 2005; Wright et al. 2007). Distance of treadmill exercise (m) 4362.87 102.71 Therefore, the purpose of the present study was to evaluate Pre-exercise muscle biopsy T0 1:19 0.03 the hypothesis that a single bout of exercise in Thorough- (hours–minutes pre-exercise)

bred horses leads to changes in mRNA expression for genes Post-exercise muscle biopsy T1 6:46 0.07 encoding transcriptional coactivators of PGC-1a and genes (minutes–seconds post-exercise)

that function upstream and downstream of PGC-1a in Post-exercise muscle biopsy T2 4:14 0.02 known canonical pathways. (hours–minutes post-exercise)

An incremental-step exercise test to maximum HR was performed for Materials and methods n = 8 subjects. The mean distance of the exercise tests was 4.4 km. Skeletal muscle biopsy samples were collected before exercise (T0), All animal procedures were approved by the University immediately post-exercise (T1) and 4 h post-exercise (T2), for gene College Dublin, Animal Research Ethics Committee; a expression analyses. licence was granted from the Department of Health and HR, heart rate.

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 PGC-1a a key regulator of oxidative energy metabolism 155

Muscle biopsy sampling MassARRAYÒ Quantitative Gene Expression (QGE) analysis Percutaneous needle muscle biopsies (Lindholm & Piehl 1974) were obtained from the dorsal compartment of the Real-time competitive PCR coupled with product resolution gluteus medius muscle according to the methods of via matrix-assisted laser desorption/ionization time-of-flight Dingboom et al. (1999) using a 6-mm-diameter, modified (MALDI-TOF) mass spectrometry (MassARRAYÒ QGE; Bergstrom biopsy needle (Jørgen KRUUSE, Veterinary Sequenom) was performed as previously described (Ding & Supplies). Biopsy samples were preserved in RNAlaterÒ Cantor 2003) in order to estimate copy number values for (Ambion/Applied Biosystems). each of the 24 gene transcripts. The copy number estimates Muscle biopsy samples were collected at rest pre-exercise for each gene transcript/sample combination were deter- Ò (T0), immediately post-exercise (T1) and 4 h post-exercise mined using MassARRAY QGE Analyser software v3.4 (T2). The time-points were chosen to examine both the (Sequenom). Data were plotted and analysed based on the immediate and the delayed responses to endurance exercise EC50 of standard curve titrations with known competitor (Mahoney et al. 2005). The time (mins) at which the sam- concentrations per assay vs. a fixed amount of cDNA tem- ples were collected are given in Table 1. plate. Normalization factors per sample were calculated using the geometric mean of the most stable combination of reference RNA isolation, purification and cDNA synthesis genes, determined by the measure of their pairwise variation Total RNA was extracted from approximately 100 mg of as calculated using the geNorm software package (Vande- tissue, using a protocol combining TRIzolÒ reagent (Invi- sompele et al. 2002). Normalization of copy numbers be- trogen), DNase treatment (RNase-free DNase; Qiagen Ltd) tween samples for each assay was performed using two and RNeasyÒ Mini-kit (Qiagen Ltd). RNA was quantified reference genes (GAPDH and TTN) and the geNorm software. using a NanoDropÒ ND1000 spectrophotometer V 3.5.2 (NanoDrop Technologies), and RNA quality was assessed Data analysis using the 18S/28S ratio and RNA integrity number (RIN) on an Agilent Bioanalyzer with the RNA 6000 Nano Lab- The StudentÕs t-test (parametric, paired, two-tailed test) was Chip kit (Agilent Technologies Ireland Ltd). One microgram used to identify significant differences in gene expression of total RNA from each sample was reverse transcribed to between samples at the various time-points. Statistical sig- cDNA using random hexamer (50 ng/ll) primers using a nificance was determined at a < 0.05. Normal distribution SuperScriptÒ III first-strand synthesis SuperMix kit of the gene expression data was examined using the one- according to the manufacturerÕs instructions (Invitrogen sample Kolmogorov–Smirnov test (K-S test), and all but Ltd). three of the genes (ANGPT2, PPARGC1A and UCP3) had normal distributions (P > 0.05). For ANGPT2 and UCP3, the log transformation was successfully used [log Assay design 10 10 (mRNA copy number)] to normalize the data. For PPARG- Where possible, equine transcript sequences were extracted C1A, the reciprocal was successfully used (X)1) to normalize from the Ensembl genome browser (Spudich et al. 2007) for the data. The K-S test was performed using SPSS version 14.0 each study gene. However, when equine gene transcript (SPSS Inc). Tests of significance were confirmed by one-way sequences were unavailable (SLC2A4 and TTN), Homo ANOVA and post-hoc tests (Table S2). sapiens transcript sequences were extracted from Ensembl. Linear regression and correlation analyses were used to The equine nucleotide sequence with greatest similarity to determine associations between mRNA gene expression of the human transcript was identified along with its chro- PPARGC1A and all other genes using GRAPHPAD INSTAT version mosomal location by BLAST searching horse genome 3.00 (GraphPad Software Inc). The correlation coefficient (r), sequence version 2 [EquCab2.0] (Altschul et al. 1997). 95% confidence interval and significance value (P) for each Multiplexed primer and competitive template designs were association were calculated, as well as the best-fit linear constructed using the MassARRAYÒ QGE Assay Design regression line [including 95% confidence interval bands]. software v1.0 (Sequenom). Forward and reverse primers were tagged with a 10-base oligo tag sequence (ACGTTGG Results ATG) to assist the amplification process and ensure that the PCR primer pairs produced mass signals. To preclude Skeletal muscle mRNA expression following a single amplification of genomic DNA, PCR primers were chosen for bout of exercise each amplicon so that at least one of the pair spanned an exonic boundary. The 24-gene panel used for this study was The RNA isolated from skeletal muscle tissue had an aver- designed across two multiplex reaction plates, and details on age RIN of 8.43 ± 0.08 (range 8.0–9.0). Mean fold differ- the primer sets used are presented in Table S1. ences between pre- and post-exercise time-points (T0 vs. T1;

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 156 Eivers et al.

T0 vs. T2) were calculated for the 24 genes (Table 2). NRF1, PRKAA2, SLC2A4, UCP3 and VEGFA. Both PRKAA2 Analyses of mRNA copy numbers identified significant and SLC2A4 displayed a significant correlation with 2 (P < 0.05) fold change differences immediately post-exer- PPARGC1A at rest (T0)(PRKAA2 (R = 0.840, P = 0.004); 2 cise (T0 vs. T1) for six of the 24 selected genes. At 4 h post- SLC2A4 (R = 0.662, P = 0.026)) [Fig. 2]. A relationship

exercise (T0 vs. T2), 10 of the 24 selected genes had signif- was observed between PPARGC1A and HDAC9 immediately 2 icant (P < 0.05) differences relative to pre-exercise levels. post-exercise (T1)(R = 0.523, P = 0.043). At 4 h post- Graphical representations of copy number per lg cDNA in exercise, significant associations between HDAC4, NRF1,

response to exercise are shown normalized to the reference VEGFA and PPARGC1A mRNA levels were detected (T2) genes (TTN and GAPDH) in Fig. 1. (HDAC4, R2 = 0.748, P = 0.003; NRF1, R2 = 0.484, P = 0.037; VEGFA, R2 = 0.647, P = 0.009) [Fig. 2]. The strongest correlation was observed between PPARGC1A and PPARGC1A mRNA gene–gene interactions 2 UCP3 mRNA abundance 4 h post-exercise (T2)(R = 0.878, We investigated the relationships between PPARGC1A P < 0.001). These data have not been corrected for multiple mRNA abundance and mRNA expression for genes encoding testing. Therefore, 50% of the associations (i.e. 3 of 69 transcriptional coactivators of PGC-1a and genes that func- tests) may represent false positives. However, correction for tion upstream and downstream of PGC-1a in known multiple testing may not be such an issue for studies in canonical pathways. Linear regression and correlation which a priori hypotheses exist (Perneger 1998). analyses were performed between mRNA copy numbers for PPARGC1A and the entire panel of 24 selected genes for each Discussion time-point (T0, T1, T2). Details on gene–gene interaction regressions and correlation analyses are detailed in Table S3. This study has demonstrated a delayed exercise-induced Significant correlations with PPARGC1A mRNA copy regulation of the gene encoding PGC-1a (PPARGC1A)4h number were detected for seven genes: HDAC4, HDAC9, post-exercise in skeletal muscle in the horse. Upstream

Table 2 Relative gene expression fold changes (FC) in equine skeletal muscle between sampling time-points (T0 vs. T1; T0 vs. T2) following treadmill exercise.

T0 vs. T1 T0 vs. T2

Gene symbol FC P-value FC P-value

Angiogenesis ANGPT2 1.29 0.046 4.35 0.018 VEGFA 1.13 0.323 2.08 0.012 Ca2+ signalling pathway CREB1 1.31 0.112 )1.24 0.255 PPP3CA 1.43 0.079 1.57 0.039 Carbohydrate/glucose metabolism PDK4 1.80 0.009 2.49 0.001 Fatty acid metabolism/mitochondrial biogenesis PPARGC1B )2.29 0.003 )2.76 0.009 Haem biosynthetic process ALAS1 1.42 0.037 1.24 0.136 Insulin signalling pathway FOXO1 2.17 0.008 3.69 0.002 PPARGC1A 1.01 0.921 4.46 <0.001 PRKAA1 1.18 0.206 )1.18 0.464 PRKAA2 )1.18 0.296 )1.11 0.572 SLC2A4 1.16 0.255 1.42 0.017 MAPK signalling pathway MAPK14 1.19 0.219 1.31 0.032 MEF2A 1.20 0.043 1.27 0.057 MEF2B 1.07 0.638 )1.20 0.322 MEF2C 1.01 0.829 1.09 0.191 NFATC2 1.36 0.129 1.24 0.131 Mitochondrial biogenesis NRF1 1.24 0.21 )1.15 0.202 Myogenesis HDAC4 )1.17 0.392 1.73 0.189 HDAC9 )1.39 0.204 )6.20 0.006 Myogenesis/PGC-1a regulation HDAC5 1.19 0.237 )1.41 0.067 Oxidative phosphorylation uncoupler activity UCP3 )1.24 0.544 3.79 0.076 PGC-1a regulation ESRRA )1.18 0.62 )1.74 0.221 PPARA signalling pathway PPARA 1.17 0.326 1.02 0.914

Genes were selected for gene expression evaluation on the basis of a priori relationship with PGC-1a signalling in exercise in other species. Significance values (P-value) were calculated by performing two sets of paired StudentÕs t-tests and are indicated in bold. MAPK, mitogen-activated protein kinase.

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 PGC-1a a key regulator of oxidative energy metabolism 157

3 000 000 (a) ALAS1 400 000 (b) ANGPT2 * *** 300 000 2 000 000

200 000 1 000 000 100 000 Copy number per µg cDNA

0 Copy number per µg cDNA 0 T0 T1 T2 T0 T1 T2

4 000 000 (c) FOXO1A 600 000 (d) HDAC9 ** 3 000 000 400 000 ** 2 000 000

200 000 1 000 000 ** Copy number per µg cDNA Copy number per µg cDNA 0 0 T0 T1 T2 T0 T1 T2

8 000 000 (e) MAPK14 3 000 000 (f) MEF2A * 6 000 000 * 2 000 000 4 000 000

1 000 000 2 000 000 Copy number per µg cDNA

Copy number per µg cDNA 0 0 T0 T1 T2 T0 T1 T2

10 000 000 (g) PDK4 *** 30 000 000 (h) PPARGC1A *** 7 500 000 ** 20 000 000 5 000 000

10 000 000 2 500 000 Copy number per µg cDNA

Copy number per µg cDNA 0 0 T0 T1 T2 T0 T1 T2

4 000 000 (i) PPARGC1B 8 000 000 (j) PPP3CA * 3 000 000 6 000 000

** 4 000 000 2 000 000 **

1 000 000 2 000 000 Copy number per µg cDNA Copy number per µg cDNA 0 0 T0 T1 T2 T0 T1 T2

9 000 000 SLC2A4 * (k) 20 000 000 (l) VEGFA * 16 000 000 6 000 000 12 000 000

8 000 000 3 000 000 4 000 000 Copy number per µg cDNA 0 Copy number per µg cDNA 0 T0 T1 T2 T0 T1 T2

Figure 1 Significant changes in mRNA expression immediately (T1) and 4 h (T2) post-exercise compared to resting levels (T0). (a) ALAS1 (b) ANGPT2 (c) FOXO1 (d) HDAC9 (e) MAPK14 (f) MEF2A (g) PDK4 (h) PPARGC1A (i) PPARGC1B (j) PPP3CA (k) SLC2A4 (l) VEGFA. Asterisks denote P-values: *P < 0.05; **P < 0.01; ***P < 0.001; determined by paired StudentÕs t-tests.

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 158 Eivers et al.

Figure 2 PPARGC1A mRNA (x-axis) expression relationships with mRNA abundance for genes (y-axis) with significantly altered gene expression. Correlation coefficients (R2) between PPARGC1A and mRNA gene expression are given with significance of the association (P-value) for relationships

between (a) PPARGC1A T2 vs. HDAC4 T2 (b) PPARGC1A T1 vs. HDAC9 T1 (c) PPARGC1A T2 vs. NRF1 T2 (d) PPARGC1A T0 vs. PRKAA2 T0 (e) PPARGC1A T0 vs. SLC2A4 T0 (f) PPARGC1A T2 vs. UCP3 T2 (g) PPARGC1A T2 vs. VEFGA T2.

activators of PGC-1a and transcriptional coactivators of expression increased (+4.5-fold, P < 0.001) 4 h following PGC-1a involved in the regulation of downstream genes exercise, we observed a significant decrease in transcript were also activated. We have observed correlations between abundance for the PGC-1 family member PPARGC1B (per- PPARGC1A mRNA abundance and expression of genes in- oxisome proliferator-activated receptor gamma, coactivator 1 volved in angiogenesis, mitochondrial respiration, glucose beta) immediately ()2.3-fold, P = 0.003) and 4 h ()2.8- transport, insulin signalling and transcriptional regulation. fold, P = 0.009) post-exercise. This contrasts with previous studies that have reported PPARGC1B expression in hu- mans (Mortensen et al. 2007) and rats (Meirhaeghe et al. PGC-1 family members 2003) to be unaffected by acute exercise following a period We have previously reported PGC-1a protein activity post- of training. Mathai et al. (2008) found a significant decrease exercise in this cohort of horses (Eivers et al. 2009). The in PGC-1b protein content 24 and 52 h after exhaustive observed increase in PPARGC1A mRNA in this study is in exercise in human skeletal muscle. In addition, this decrease accordance with previous studies in humans (Pilegaard was greater in a restricted low carbohydrate diet cohort et al. 2003; Norrbom et al. 2004), mice (Akimoto et al. because glycogen was not resynthesized to resting levels. 2004) and horses (Eivers et al. 2009). However, it is Conversely, Mortensen et al. (2007) observed that uncertain whether the mRNA abundance is a direct PPARGC1B mRNA content was reduced at rest following reflection of PGC-1a protein content. While PPARGC1A 10 weeks of knee extensor endurance training in humans;

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 PGC-1a a key regulator of oxidative energy metabolism 159 they concluded that PPARGC1B was a long-term training way in horses requires further investigation; however, the adaptive response gene. In vitro studies performed by Meir- increase in PPP3CA mRNA in Thoroughbred horse skeletal haeghe et al. (2003) found that the PGC-1b protein func- muscle during recovery from exercise implies a role in the tions to regulate the mitochondrial biogenesis required to adaptive response. maintain basal energy levels in tissue. Therefore, the Calcineurin is a known activator of the PGC-1a tran- increase in PPARGC1A and down-regulation of PPARGC1B scriptional coactivator MEF2 via phosphorylation of the in Thoroughbred horse skeletal muscle following exercise class II HDACs (Wu et al. 2001). We observed a large may suggest a complementary switching mechanism gener- decrease in HDAC9 (histone deacetylase 9) mRNA ()6.2-fold, ated by the increased energy demands as a result of exercise. P = 0.006), with HDAC5 (histone deacetylase 5) showing a similar trend ()1.4-fold, P = 0.067) 4 h post-exercise. A previous study has focused on HDAC5 repression in PPARGC1A upstream signalling pathways and response to exercise (McGee & Hargreaves 2004); however, coactivators the significant decrease in HDAC9 mRNA copy number A number of upstream signalling pathways including the indicates a role for this gene in the response to exercise. It Ca2+-dependent calcineurin signalling pathway and the may be that this observation is related to the control of kinase signalling pathways – p38 MAPK and 5-AMPK – muscle differentiation; HDAC9 has previously been reported have been identified as exercise-induced activators of PGC- to be a transcriptional target of MEF2, and together they 1a expression. During exercise, AMPK partly influences control skeletal muscle development in a negative-feedback PGC-1a transcriptional activity, and during chronic energy loop (Haberland et al. 2007). deprivation, AMPK is required for increased mitochondrial The MEF2 isoforms MEF2C and MEF2D have been iden- biogenesis (Zong et al. 2002). We have previously reported tified as important targets of PGC-1a in coactivation of type I pAMPKa activity immediately post-exercise in this study myofibrillar proteins in cultured cells (Handschin et al. cohort (Eivers et al. 2009). While we did not observe sig- 2003). We observed no change in MEF2B (myocyte nificant post-exercise alternations in gene expression levels enhancer factor 2B)orMEF2C (myocyte enhancer factor 2C) for the AMPK isoforms PRKAA1 (protein kinase, AMP-acti- mRNA following exercise. However, we did observe a sig- vated, alpha 1 catalytic subunit gene) and PRKAA2 (protein nificant increase in MEF2A mRNA abundance immediately kinase, AMP-activated, alpha 2 catalytic subunit gene), we did post-exercise (+1.2-fold, P = 0.043). MEF2A has been pre- identify a significant linear relationship between PPARG- viously reported to be an important of GLUT4 C1A and PRKAA2 at rest. The significance of this relation- expression (Mora & Pessin 2000); however, in this study, ship remains unclear. Exercise is a powerful activator of there was no evidence for a relationship between MEF2A MAPK pathways, and we found a significant increase in the and SLC2A4 [solute carrier family 2 (facilitated glucose gene encoding p38 MAPK (MAPK14, MAPK 14 gene) 4 h transporter), member 4] mRNA expression (data not shown). post-exercise (+1.31-fold, P = 0.032). The p38 MAPK pathway gene (MAPK14) regulates PGC-1a activity via PPARGC1A and glucose metabolism upstream activation of transcription factors such as MEF2 (McGee & Hargreaves 2004), and a direct link has been PGC-1a is a key regulator of insulin sensitivity, controlling established between PPARGC1A mRNA expression and p38 glucose transport through coactivation of PDK4, a mediator MAPK pathway activation in cultured cells (Akimoto et al. of glucose metabolism, by regulation of glucose transport 2005). Although further investigation is required to estab- (SLC2A4) and by phosphorylation of the pyruvate dehy- lish evidence of p38 MAPK activation of PGC-1a, the in- drogenase complex to reduce glucose oxidation and increase crease in expression of MAPK14 suggests the p38 MAPK mitochondrial fatty acid oxidation (Wende et al. 2005). A pathway to be an important activator of PGC-1a in response significant increase in SLC2A4 expression was detected 4 h to exercise. post-exercise (+1.42-fold, P = 0.017), which was correlated Calcineurin is activated in response to endurance-type with a concomitant increase in PPARGC1A mRNA abun- exercise and has been found to enhance binding of the dance. PGC-1a regulation of PDK4 is mediated by coacti- myocyte enhancer family (MEF) to the promoter region of vation of ERRa in skeletal muscle (Wende et al. 2005). We PPARGC1A. Handschin et al. (2003) have suggested that observed an increase in mRNA abundance for PDK4 the stimulated increase in Ca2+-dependent calcineurin (pyruvate dehydrogenase kinase, isozyme 4) immediately post- results in an autoregulatory loop that potentially stabilizes exercise (+1.80, P = 0.009) and a further significant in- the induction of PGC-1a, leading to a stable increase in crease 4 h post-exercise (+2.5-fold, P = 0.001). oxidative muscle fibre-type adaptations in transgenic mice. There was no change in mRNA expression for ESRRA We observed a significant increase in the gene encoding (oestrogen-related receptor alpha), which encodes ERRa; calcineurin (PPP3CA, protein phosphatase 3, catalytic subunit however, there was a pronounced increase in FOXO1 and alpha isozyme) 4 h (+1.6-fold, P = 0.039) post-exercise. (forkhead box O1) expression immediately post-exercise The link between PGC-1a and the calcium signalling path- (+2.17, P = 0.008), with a further increase in mRNA

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 160 Eivers et al.

abundance 4 h post-exercise (+3.7-fold, P = 0.002). The by an increase in NRF1 transcription, which is consistent FOXO1 regulates glucose metabolism by with findings from previous studies in humans (Pilegaard increasing fatty acid oxidation through activation of PDK4 et al. 2003; Norrbom et al. 2004). We found a significant and lipoprotein lipase in skeletal muscle of transgenic mice negative relationship between PPARGC1A and NRF1 and in vitro (Kamei et al. 2003; Kwon et al. 2004). The mRNA abundance 4 h post-exercise. Previous work has PDK4 gene promoter contains a binding site for the FOXO1 demonstrated that the NRF-1 binding protein increases transcription factor (Araki & Motojima 2006), and the activity following a single bout of exercise in rats (Baar et al. PDK4 promoter binding site sequence is conserved in the 2002), and therefore it has been suggested that NRF-1 horse. The increase in PDK4 and FOXO1 mRNA abundance regulation is post-transcriptional (Norrbom et al. 2004). following exercise and the linear relationship observed be- ALAS1 (encoded by ALAS1 – aminolevulinate delta syn- tween PPARGC1A and SLC2A4 gene expression suggests a thase 1) is a mitochondrial protein that provides haem for mechanism by which PGC-1a regulates glucose oxidation cytochromes in the respiratory chain (Ajioka et al. 2006). while increasing mitochondrial fatty acid oxidation in PGC-1a is an important regulator of mitochondrial respi- Thoroughbred horse skeletal muscle, particularly during ration. ALAS1 gene expression has previously been shown recovery. to precede the increase in PGC-1a protein expression in rats (Wright et al. 2007). We observed a significant increase in ALAS1 gene expression immediately post-exercise (+1.4- PPARGC1A and angiogenesis fold, P = 0.037), which may reflect an increase in oxidative PGC-1a has emerged as a powerful hypoxia-inducible factor respiration following exercise. (HIF)-independent regulator of the angiogenic pathway via It has been proposed that PGC-1a may be an important coactivation of the orphan ERRa on both mediator in the activation of the skeletal muscle-specific the promoter region and the enhancer in the first intron on uncoupling protein UCP3 (Jiang et al. 2009). The strong the VEGFA gene (Arany et al. 2008). Arany and colleagues relationship observed between PPARGC1A and UCP3 have demonstrated previously that PGC-1a upregulates mRNA expression supports a key role for PGC-1a regula- transcription of VEGFA mRNA (Arany et al. 2008). Con- tion in mitochondrial respiration through interaction with sequently, coordinated expression of angiopoietin 2 and UCP3 in Thoroughbred horses in response to exercise. The VEGFA induces angiogenesis in cultured cells and skeletal exact role of UCP3 in muscle remains unclear, but in muscle in vivo in mice (Arany et al. 2008). Although there addition to a role in uncoupling of respiration, it has was no change in mRNA expression for ESRRA, we did been implicated in the regulation of observe a modest increase in ANGPT2 (angiopoietin 2) and in mitochondrial fatty aid transport (Schrauwen & expression (+1.29-fold, P = 0.046) immediately post-exer- Hesselink 2002). cise and a further increase in mRNA abundance for both ANGPT2 (+4.4-fold, P = 0.018) and VEGFA (vascular Conclusion endothelial growth factor A gene) [+2.1-fold, P = 0.012] 4 h post-exercise. Interestingly, there was a strong positive The exercise-induced changes in gene expression observed correlation between PPARGC1A and VEGFA mRNA abun- in this study highlight the role of PGC-1a in the regulation dance 4 h post-exercise. The prolonged increase in ANGPT2 of mitochondrial biogenesis and oxidative metabolism in and VEGFA expression indicates that angiogenesis is an skeletal muscle in the horse. PPARGC1A gene expression important adaptive exercise response in Thoroughbred was found to be significantly correlated with genes involved horse skeletal muscle. In the absence of a detectable ERRa in angiogenesis, mitochondrial respiration, glucose trans- gene response, it may be that VEGFA and ANGPT2 are in- port, insulin signalling and transcriptional regulation. duced by a PGC-1a-independent mechanism. However, the Together, these molecular activities may lead to increased positive correlation observed between PPARGC1A and oxidative capacity and a greater resistance to fatigue. This VEGFA mRNA supports the hypothesis that PGC-1a may study indicates that PGC-1a and genes regulated by PGC-1a have a role in the regulation of angiogenesis in Thorough- contribute in part to the exercise response in horse skeletal bred horses. muscle and may therefore constitute regulators of perfor- mance potential in Thoroughbred horses. PPARGC1A and oxidative metabolism Acknowledgements Mitochondrial biogenesis is induced by PGC-1a transcrip- tional activation of the nuclear-encoded nuclear respiratory The authors thank Mr P.J. Rothwell and Mr G. Burke for factor 1 (NRF1) and nuclear respiratory factor 2 (GABPA) access to horses. We thank the University Veterinary Hos- genes, resulting in increased oxidative phosphorylation and pital yard staff for assistance with exercise experiments. This enhanced oxidative respiration (Baar et al. 2002). The project was funded by a grant to EH: Science Foundation induction of PPARGC1A mRNA expression was not followed Ireland PIYRA 04/YI1/B539. The funding agency had no

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 PGC-1a a key regulator of oxidative energy metabolism 161 part in the study design; in the collection, analysis and Hudlicka O., Brown M. & Egginton S. (1992) Angiogenesis in interpretation of data; in the writing of the report; or in the skeletal and cardiac muscle. Physiological Reviews 72, 369– decision to submit the article for publication. 417. Jager S., Handschin C., St-Pierre J. & Spiegelman B.M. (2007) AMP- activated protein kinase (AMPK) action in skeletal muscle via References direct phosphorylation of PGC-1alpha. Proceedings of the National Academy of Sciences of the United States of America 104, 12017– Ajioka R.S., Phillips J.D. & Kushner J.P. (2006) Biosynthesis of 22. heme in mammals. Biochimica et Biophysica Acta 1763, 723–36. Jiang N., Zhang G., Bo H., Qu J., Ma G., Cao D., Wen L., Liu S., Ji L.L. Akimoto T., Sorg B.S. & Yan Z. (2004) Real-time imaging of per- & Zhang Y. (2009) Upregulation of uncoupling protein-3 in oxisome proliferator-activated receptor-gamma coactivator-1alpha skeletal muscle during exercise: a potential antioxidant function. promoter activity in skeletal muscles of living mice. American Free Radical Biology & Medicine 46, 138–45. Journal of Physiology. Cell Physiology 287, C790–6. Kamei Y., Mizukami J., Miura S., Suzuki M., Takahashi N., Kawada Akimoto T., Pohnert S.C., Li P., Zhang M., Gumbs C., Rosenberg T., Taniguchi T. & Ezaki O. (2003) A forkhead transcription P.B., Williams R.S. & Yan Z. (2005) Exercise stimulates Pgc-1al- factor FKHR up-regulates lipoprotein lipase expression in skeletal pha transcription in skeletal muscle through activation of the p38 muscle. FEBS Letters 536, 232–6. MAPK pathway. The Journal of Biological Chemistry 280, 19587– Kwon H.S., Huang B., Unterman T.G. & Harris R.A. (2004) Protein 93. kinase B-alpha inhibits human pyruvate dehydrogenase kinase-4 Altschul S.F., Madden T.L., Schaffer A.A., Zhang J., Zhang Z., Miller gene induction by dexamethasone through inactivation of FOXO W. & Lipman D.J. (1997) Gapped BLAST and PSI-BLAST: a new transcription factors. Diabetes 53, 899–910. generation of protein database search programs. Nucleic Acids Lin J., Handschin C. & Spiegelman B.M. (2005) Metabolic control Research 25, 3389–402. through the PGC-1 family of transcription coactivators. Cell Araki M. & Motojima K. (2006) Identification of ERRalpha as a Metabolism 1, 361–70. specific partner of PGC-1alpha for the activation of PDK4 gene Lindholm A. & Piehl K. (1974) Fibre composition, enzyme activity expression in muscle. The FEBS Journal 273, 1669–80. and concentrations of metabolites and electrolytes in muscles of Arany Z. (2008) PGC-1 coactivators and skeletal muscle adapta- standardbred horses. Acta Veterinaria Scandinavica 15, 287–309. tions in health and disease. Current Opinion in Genetics & Develop- Mahoney D.J., Parise G., Melov S., Safdar A. & Tarnopolsky M.A. ment 18, 426–34. (2005) Analysis of global mRNA expression in human skeletal Arany Z., Foo S.Y., Ma Y. et al. (2008) HIF-independent regulation muscle during recovery from endurance exercise. FASEB Journal of VEGF and angiogenesis by the transcriptional coactivator PGC- 19, 1498–500. 1alpha. 451, 1008–12. Mathai A.S., Bonen A., Benton C.R., Robinson D.L. & Graham T.E. Baar K., Wende A.R., Jones T.E., Marison M., Nolte L.A., Chen M., (2008) Rapid exercise-induced changes in PGC-1alpha mRNA Kelly D.P. & Holloszy J.O. (2002) Adaptations of skeletal muscle and protein in human skeletal muscle. Journal of Applied Physi- to exercise: rapid increase in the transcriptional coactivator PGC- ology 105, 1098–105. 1. FASEB Journal 16, 1879–86. McGee S.L. & Hargreaves M. (2004) Exercise and myocyte enhancer Ding C. & Cantor C.R. (2003) A high-throughput gene expression factor 2 regulation in human skeletal muscle. Diabetes 53, 1208– analysis technique using competitive PCR and matrix-assisted 14. laser desorption ionization time-of-flight MS. Proceedings of the Meirhaeghe A., Crowley V., Lenaghan C. et al. (2003) Character- National Academy of Sciences of the United States of America 100, ization of the human, mouse and rat PGC1 beta (peroxisome- 3059–64. proliferator-activated receptor-gamma co-activator 1 beta) gene in Dingboom E.G., Dijkstra G., Enzerink E., van Oudheusden H.C. & vitro and in vivo. The Biochemical Journal 373, 155–65. Weijs W.A. (1999) Postnatal muscle fibre composition of the Mora S. & Pessin J.E. (2000) The MEF2A isoform is required for gluteus medius muscle of Dutch Warmblood foals; maturation striated muscle-specific expression of the insulin-responsive and the influence of exercise. Equine Veterinary Journal 31, 95– GLUT4 glucose transporter. The Journal of Biological Chemistry 100. 275, 16323–8. Eivers S.S., McGivney B.A., Fonseca R.G., MacHugh D.E., Menson Mortensen O.H., Plomgaard P., Fischer C.P., Hansen A.K., Pilegaard K., Park S.D., Rivero J.L., Taylor C.T., Katz L.M. & Hill E.W. H. & Pedersen B.K. (2007) PGC-1beta is downregulated by (2009) Alterations in oxidative gene expression in equine skeletal training in human skeletal muscle: no effect of training twice muscle following exercise and training. Physiological Genomics 40, every second day vs. once daily on expression of the PGC-1 83–93. family. Journal of Applied Physiology 103, 1536–42. Haberland M., Arnold M.A., McAnally J., Phan D., Kim Y. & Olson Norrbom J., Sundberg C.J., Ameln H., Kraus W.E., Jansson E. & E.N. (2007) Regulation of HDAC9 gene expression by MEF2 Gustafsson T. (2004) PGC-1alpha mRNA expression is influenced establishes a negative-feedback loop in the transcriptional cir- by metabolic perturbation in exercising human skeletal muscle. cuitry of muscle differentiation. Molecular and Cellular Biology 27, Journal of Applied Physiology 96, 189–94. 518–25. Perneger T.V. (1998) WhatÕs wrong with Bonferroni adjustments. Handschin C., Rhee J., Lin J., Tarr P.T. & Spiegelman B.M. (2003) BMJ 316, 1236–12380. An autoregulatory loop controls peroxisome proliferator-acti- Pilegaard H., Saltin B. & Neufer P.D. (2003) Exercise induces vated receptor gamma coactivator 1alpha expression in muscle. transient transcriptional activation of the PGC-1alpha gene in Proceedings of the National Academy of Sciences of the United States human skeletal muscle. The Journal of Physiology 546, 851–8. of America 100, 7111–6.

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162 162 Eivers et al.

Rose R.J., Hendrickson D.K. & Knight P.K. (1990a) Clinical exercise begins before the increase in muscle PGC-1alpha expression. The testing in the normal thoroughbred racehorse. Australian Veteri- Journal of Biological Chemistry 282, 194–9. nary Journal 67, 345–8. Wu H., Rothermel B., Kanatous S. et al. (2001) Activation of MEF2 Rose R.J., Hodgson D.R., Bayly W.M. & Gollnick P.D. (1990b) by muscle activity is mediated through a calcineurin-dependent Kinetics of VO2 and VCO2 in the horse and comparison of five pathway. The EMBO Journal 20, 6414–23. methods for determination of maximum oxygen uptake. Equine Zhang Y., Ma K., Sadana P., Chowdhury F., Gaillard S., Wang F., Veterinary Journal suppl. 9, 39–42. McDonnell D.P., Unterman T.G., Elam M.B. & Park E.A. (2006) Russell A.P., Hesselink M.K., Lo S.K. & Schrauwen P. (2005) Reg- Estrogen-related receptors stimulate pyruvate dehydrogenase ki- ulation of metabolic transcriptional co-activators and transcrip- nase isoform 4 gene expression. The Journal of Biological Chemistry tion factors with acute exercise. FASEB Journal 19, 986–8. 281, 39897–906. Schrauwen P. & Hesselink M. (2002) UCP2 and UCP3 in muscle Zong H., Ren J.M., Young L.H., Pypaert M., Mu J., Birnbaum M.J. & controlling body metabolism. Journal of Experimental Biology 205, Shulman G.I. (2002) AMP kinase is required for mitochondrial 2275–85. biogenesis in skeletal muscle in response to chronic energy Spudich G., Fernandez-Suarez X.M. & Birney E. (2007) Genome deprivation. Proceedings of the National Academy of Sciences of the browsing with Ensembl: a practical overview. Briefings in Func- United States of America 99, 15983–7. tional Genomics & Proteomics 6, 202–19. Vandesompele J., De Preter K., Pattyn F., Poppe B., Van Roy N., De Paepe A. & Speleman F. (2002) Accurate normalization of Supporting information real-time quantitative RT-PCR data by geometric averaging of Additional supporting information may be found in the multiple internal control genes. Genome Biology 3, RE- online version of this article. SEARCH0034. Table S1 Equine oligonucleotide primers for MassARRAY Vissing K., Andersen J.L. & Schjerling P. (2005) Are exercise-in- Quantitative Gene Expression analysis. duced genes induced by exercise? FASEB Journal 19, 94–6. Wende A.R., Huss J.M., Schaeffer P.J., Giguere V. & Kelly D.P. Table S2 One-way ANOVA and post-hoc test results. 2 (2005) PGC-1alpha coactivates PDK4 gene expression via the Table S3 Correlation coefficients (R ) and significance of orphan nuclear receptor ERRalpha: a mechanism for transcrip- associations (P-value) for relationships between PPARGC1A tional control of muscle glucose metabolism. Molecular and Cel- mRNA expression and candidate genes. lular Biology 25, 10684–94. As a service to our authors and readers, this journal pro- Woodie J.B., Ducharme N.G., Hackett R.P., Erb H.N., Mitchell L.M. vides supporting information supplied by the authors. Such & Soderholm L.V. (2005) Can an external device prevent dorsal materials are peer-reviewed and may be re-organized for displacement of the soft palate during strenuous exercise? Equine online delivery, but are not copy-edited or typeset. Technical Veterinary Journal 37, 425–9. support issues arising from supporting information (other Wright D.C., Han D.H., Garcia-Roves P.M., Geiger P.C., Jones T.E. & than missing files) should be addressed to the authors. Holloszy J.O. (2007) Exercise-induced mitochondrial biogenesis

Ó 2011 The Authors, Animal Genetics Ó 2011 Stichting International Foundation for Animal Genetics, 43, 153–162