Mitochondrial bioenergetic pathways in blood leukocyte transcriptome decrease after intensive weight loss but are rescued following weight regain in female physique athletes

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC 4.0 https://creativecommons.org/licenses/by-nc/4.0/ Mitochondrial bioenergetic pathways in blood leukocyte transcriptome decrease after intensive weight loss but are rescued following weight regain in female physique athletes © 2021 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology Published version Sarin, Heikki V.; Pirinen, Eija; Pietiläinen, Kirsi H.; Isola, Ville; Häkkinen, Keijo; Perola, Markus; Hulmi, Juha J. Sarin, H. V., Pirinen, E., Pietiläinen, K. H., Isola, V., Häkkinen, K., Perola, M., & Hulmi, J. J. (2021). Mitochondrial bioenergetic pathways in blood leukocyte transcriptome decrease after intensive weight loss but are rescued following weight regain in female physique athletes. FASEB Journal, 35(4), Article e21484. https://doi.org/10.1096/fj.202002029r 2021 The FASEB Journal. 2021;35:e21484. | 1 of 15 https://doi.org/10.1096/fj.202002029R wileyonlinelibrary.com/journal/fsb2 Received: 31 August 2020 | Revised: 18 January 2021 | Accepted: 15 February 2021 DOI: 10.1096/fj.202002029R RESEARCH ARTICLE Mitochondrial bioenergetic pathways in blood leukocyte transcriptome decrease after intensive weight loss but are rescued following weight regain in female physique athletes Heikki V.Sarin1,2 | EijaPirinen2 | Kirsi H.Pietiläinen3,4 | VilleIsola5 | KeijoHäkkinen5 | MarkusPerola1,2 | Juha J.Hulmi5 1Genomics and Biobank Unit, The Department of Public Health Solutions, National Institute for Health and Welfare, Helsinki, Finland 2Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, Finland 3Obesity Research Unit, Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, Finland 4Obesity Center, Abdominal Center, Endocrinology, Helsinki University Hospital, University of Helsinki, Helsinki, Finland 5Faculty of Sport and Health Sciences, Neuromuscular Research Center, University of Jyväskylä, Jyväskylä, Finland This is an open access article under the terms of the Creative Commons AttributionNonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2021 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology. Eija Pirinen and Kirsi H. Pietiläinen contributed equally to this work. Markus Perola and Juha J. Hulmi contributed equally to this work. Abbreviations: CR, caloric restriction; DEG, differentially expressed genes; ETC, electronic transport chain; FDR, false discovery rate; GEE, generalized estimating equations; IPA, ingenuity pathway analysis; METh, metabolic equivalent hours; NMR, nuclear magnetic resonance; OXPHOS, oxidative phosphorylation; ROS, reactive oxygen species; TCA, tricarboxylic acid. Correspondence Juha Hulmi, Faculty of Sport and Health Sciences, Neuromuscular Research Center, University of Jyväskylä, Jyväskylä, PO Box 35, Jyväskylä FI40014, Finland. Email: [email protected] Heikki Sarin, Genomics and Biomarkers Unit, Department of Public Health Solutions, National Institute for Health and Welfare (THL), P.O. Box 30 (Mannerheimintie 166), Helsinki FIN00271, Finland. Email: [email protected] Funding information Academy of Finland (Suomen Akatemia), Grant/Award Number: 269517, 286359, 275922, 314383 and 266286; Academy of Finland, Centre of Excellence in Research on Mitochondria, Metabolism and Disease, Grant/Award Number: 272376; Novo Nordisk Foundation Center for Basic Metabolic Research, Grant/Award Number: NNF16OC0020866, NNF17OC0027232 and NNF10OC1013354; Juho Vainion Abstract Prolonged periods of energy deficit leading to weight loss induce metabolic adaptations resulting in reduced energy expenditure, but the mechanisms for energy conservation are incompletely understood. We examined 42 healthy athletic females (age 27.5±4.0years, body mass index 23.4±1.7kg/m2) who volunteered into either a group dieting for physique competition (n=25) or a control group (n=17). The diet group substantially reduced their energy intake and moderately increased exercise levels to induce loss of fat mass that was regained during a voluntary weight regain period. The control group maintained their typical lifestyle habits and body mass as instructed. From the diet group, fasting blood samples were drawn at baseline (PRE), after 4to 5month weight loss (PREMID), and after 4to 5month weight regain (MIDPOST) as well as from the control group at similar intervals. Blood was analyzed to determine leukocyte transcriptome by RNASequencing and serum metabolome by nuclear magnetic resonance (NMR) platform. The intensive weight loss period induced several metabolic adaptations, including a prominent suppression of transcriptomic signature for mitochondrial OXPHOS and ribosome biogenesis. The upstream regulator analysis suggested that this reprogramming of cellular energy metabolism may be mediated via AMPK/PGC1α signaling and mTOR/eIF2 2 of 15 | SARIN et Al. 1 | INTRODUCTION Every year, many individuals engage in intense diet and exercise training programs in an attempt to pursue fat mass loss and improved aesthetic appearance. Results of these diets are often temporary, thus leading to a cycle of weight loss and subsequent regain. This so called yoyo dieting/ weight cycling has been regarded as a risk for sustained reduced energy expenditure at least in overweight individuals.14 This decline in energy expenditure (ie, adaptive thermogenesis) is caused by a number of coordinated actions of metabolic, behavioral, neuroendocrine, and autonomic responses designed to maintain body energy stores at optimal levels.1 Adaptive thermogenesis is considered a biologically meaningful survival mechanism that conserves energy in the face of starvation and dangerously low energy supplies.5 Dieting and weight cycling studies have thus far focused on overweight individuals. Weight cycling in normal weight individuals has been ignored perhaps due to the questionable ethics of subjecting normal weight individuals to a combination of very high dose of exercise with lowcalorie diet, that is, very low energy availability.6,7 However, even previously normalweight athletes pursue lower levels of adiposity and a leaner phenotype as it has been considered beneficial in terms of superior performance and aesthetic appearance.811 A period of very highvolume exercise and low energy availability leading to weight loss followed by a voluntary weight regain period is a very common practice in many sports.6 In popular aesthetic sports such as fitness and physique sports, athletes after years of training, go through intensive dieting periods (>10weeks) with reduced energy intake (<30kcal/ kg)10 preceding competitions. Through intensive precompetition dieting, athletes aim to improve their muscular definition and aesthetic appearance by reducing body fat mass accomplished by high volume of both resistance and endurance training and a lowenergy intake.10,12,13 The weight loss period is usually followed by a voluntary weight regain period during which energy intake, exercise, and body fat mass levels are restored back to baseline levels aiming to recover from the intensive competition preparation diet.10 To date, only a few studies on physique athlete weight regain period has been conducted, but it has been recommended that energy intake is increased up to 4248kcal/kg.14 Thus, these individuals are ideal candidates for investigating the effects of metabolic adaptations during periods of voluntary weight loss and weight regain in previously normal weighed physically active individuals. Although metabolic adaptation following caloric restriction (CR) is a wellestablished phenomenon, its molecular mechanisms are not completely understood. Given that mitochondria act as central bioenergetic organelles in the cells, they are thought to play an important role in CRinduced metabolic adaptations. Previous data from animal models indicate that the decreased mitochondrial respiration and oxygen consumption due to a decline in uncoupled respiration, that is, improved bioenergetic efficiency of mitochondria, might contribute to decreased energy expenditure during CRinduced weight loss in muscle, thus making weight maintenance and further weight reduction more challenging.9,15,16 This metabolic adaptation may also make individuals more prone to regain the already lost fat mass after otherwise successful weight loss. We have previously shown that intense fat mass loss and regain have a distinct effect on hormonal system regulating energy metabolism.10 In the present study,10 we aimed to investigate the magnitude of mitochondrial and energy metabolism adaptations during a period of intense weight loss and voluntary weight regain at a systemic level in a population of normalweight, healthy physique athletes.10 More specifically, pathways of energy metabolism were examined by utilizing an integrative omics approach consisting of leukocytederived transcriptomics and serum nuclear magnetic resonance (NMR) metabolomics as minimally invasive markers of mitochondrial and metabolical adaptation. Säätiö; Orionin Tutkimussäätiö (Orion Research Foundation); The European Union's FP7 programme, Grant/Award Number: HZ2020 633589 (Ageing with Elegans); Yrjö Jahnsson Foundation; Suomen Lääketieteen Säätiö (Finnish Medical Foundation); Finnish Diabetes Research Foundation; Signe ja Ane Gyllenbergin Säätiö (Signe and Ane Gyllenberg Foundation); Sigrid Juselius Foundation; Helsinki University Hospital Research Funds; Government Research Funds; Helsingin Yliopisto ja Jyväskylän yliopisto (University of Helsinki and Jyväskylä) signalingdependent pathways. Our findings show for the first time that prolonged energy deprivation induced modulation of mitochondrial metabolism can be observed through minimally invasive measures of leukocyte transcriptome and serum metabolome at systemic level, suggesting that adaptation to energy deficit is broader in humans than previously thought. KEYWORDS diet, exercise, leukocytes, oxidative phosphorylation, ribosomes | 3 of 15 SARIN et Al. 2 | MATERIALS AND METHODS 2.1 | Study participants and design Young normalweight female physique athletes (n = 60) were recruited10 to participate and volunteer in a selfadministered weight loss and weight regain regimen (Figure1). By the end of the study period (PREPOST), a total of 10 athletes failed to complete the study regimen in a required manner. Specifically, the study dropouts consisted of (a) one control not arriving to baseline testing, (b) three diet group participants failing to follow the diet program and thus did not compete, and (c) six controls were not able follow the control period or for other unknown reason to finish the study. In addition, participants with missing dietary data from at least one point of the study (n=8) were excluded from the current omics study. After these exclusions, based on completion of the diet/control period and availability of data from all study points, 42 healthy athletic females (age 27.5±4.0years and body mass index [BMI] 23.4±1.7kg/m2) were included in the current followup omics study. These individuals had volunteered into either a group dieting for physique competition (n=25) or a control group (n=17). Due to the intense nature of the study regimen, and ethical restrictions for randomization, the participants were allocated to diet and control groups based on their own choice. The effects of intense fat mass loss and regain were examined from the diet group (n=25, age 27.2±4.2years, BMI 23.5±1.8kg/m2) at three test time points: (a) baseline tests were conducted before the weight loss regimen began (PRE), (b) immediately after the weight loss period (PREMID, 20.2 ± 3.3 weeks), and (c) after the weight regain period (MIDPOST, 17.8±2.4weeks) (Figure1). The weight loss period (PREMID) is usually characterized with low energy availability together with increasing the amount of aerobic exercise training while frequency and intensity of resistance training is kept at a relatively constant level to minimize the amount of muscle mass lost. The vigorous progressive weight loss routine is usually followed by a voluntary weightregain period (MIDPOST), during which exercise training and energy intake levels are reverted back to normal levels. This subsequent weight regain has been considered mandatory and beneficial to restore potentially disrupted metabolic homeostasis caused by prolonged low energy availability, intense exercise training, and extremely low levels of fat mass. The control group of physique athletes (n = 17, age 27.8±3.8years, BMI 23.1±1.4kg/m2) was instructed to maintain their typical weight and fitness lifestyle throughout the whole study period (PREMID, 22.4± 2.5weeks; MIDPOST, 19.2±3.5weeks). At each experimental time point (PRE, MID, and POST), blood samples were obtained, and all the participants went through a series of the same anthropometric and clinical tests that were described FIGURE 1 Study design and workflow. The study design and workflow are illustrated in a flowchart to demonstrate the utilized study protocol. *Overall, 10 participants failed to complete the study regimen (PREPOST) in a required manner of the 60 individuals who volunteered for the study. A number of participants lacking complete dietary records (n=8) were also excluded from the current omics study. Only individuals with minimal missing information were included in the final study sample of 42 participants (diet group n=25, control group n=17), for which bioinformatic analyses were conducted. Study period lengths (PREMID, MIDPOST) are reported as mean (standard deviation) of both the diet and control group individuals. GEE, generalized estimating equations; IPA, ingenuity pathway analysis; NMR, nuclear magnetic resonance 4 of 15 | SARIN et Al. earlier (Figure1).10 The following methods on reported nutrient intakes, exercise levels, and muscular fitness have been described in full detail earlier by Hulmi et al.10 The Ethical Committee at the University of Jyväskylä approved the study protocol, and all participants gave written informed consent in accordance with the Declaration of Helsinki. 2.2 | Anthropometric measurements In the current study, body composition and anthropometrics (ie, fat mass, visceral fat mass, and lean mass) were assessed with dualenergy Xray absorptiometry (DEXA, Lunar Prodigy Advance, GE Medical Systems, Lunar, Madison, WI, USA).10 2.3 | Dietary intake The physique study participants recorded nutrient intake repeatedly using selfreported food diaries from representative days throughout the study (a) at baseline (PRE), (b) after the weight loss period (MID), and (c) after the weight regain period (POST).10 The diet group participants followed strict dietary routines during the weight loss period (PREMID), where majority of the food ingested was weighed and reported. The control group reported food diaries over three weekdays and one weekend day (a) at the baseline (PRE), (b) in the middle of the study (MID), and (c) during the last part of the study (POST). Nutritional supplements (ie, sport supplements) were included in the dietary analysis. The food diaries provided by the study participants were analyzed by dietary analysis software (Aivodiet, Flowteam Oy, Oulu, Finland). 2.4 | Physical activity and muscular fitness The duration and intensity of daily physical activity and exercise training were also reported by the physique study participants throughout the study period (PRE, MID, and POST) from which overall physical activity (METh/week) level was calculated.10 As reported earlier, the training regimen of the study participants (n=42) included (a) resistance training 4.8±0.7 (diet group) and 3.9±1.9 (controls) times per week at the baseline (PRE), after which the resistance training frequency was maintained throughout the study period (MID and POST) in both groups. In contrast, the aerobic training frequency was increased from 3.8±2.8 to 4.8±3.0 times per week for the duration of the weight loss period (PREMID), which was mostly due to the increased amount of steady state aerobic training, and partly due to increased amount of HIT (high intensity training)- exercise.10 The level of aerobic training decreased during the recovery period (MIDPOST) in these participants down to 2.3±2.2 times per week, whereas the controls maintained their levels of aerobic training relatively constant.10 Changes in lower and upper body strength (ie, maximal voluntary contraction) was determined using isometric tests consisting of (a) a horizontal leg press extension dynamometer and (b) maximal isometric bilateral smith bench press together with (c) maximal explosive strength using a vertical counter movement jump. See further information on the training regimens and measures of muscular fitness from Hulmi et al.10 2.5 | Venous blood sampling and analysis Fasting blood samples were taken from both the diet and control group after at least 8hours of fasting at each time point. All the participants were asked to sleep for at least 8hours during the preceding night and were required to refrain from strenuous physical activity for at least 24hours before venous blood sampling at baseline (PRE), after the weight loss period (MID), and after the weight regain period (POST). Venous blood samples were taken from the antecubital vein into serum tubes (Venosafe; Terumo Medical Co., Leuven, Hanau, Belgium) using standardized laboratory procedures.10 2.6 | Transcriptome 2.6.1 | Library preparation, sequencing, read alignment, and batch effect Transcriptome was quantified in peripheral leukocytes extracted from the samples collected in PAXgene blood RNA tube. The sequencing RNA library of each sample was processed using Illumina TruSeq according to the protocol provided by the manufacturer (https://www.illum ina.com). The utilized Illumina protocol was paired end and strand specific, and the applied read depth for library preparation was set to 2× 100bp. Sequencing of the RNA libraries was carried out with the Illumina HiSeq2000 sequencing platform. 2.6.2 | Differential expression analysis We further processed sequence alignments with the DESeq2 software to assemble transcripts, quantify the expression levels, and analyze differentially expressed genes (DEGs). DESeq2 applies its own normalization methods and independent filtering to raw read RNASequencing data. Before statistical analysis, some of the low expressed genes were excluded from the analysis if they matched the criteria: (a) | 5 of 15 SARIN et Al. gene had zero read counts across samples, and (b) gene had lower than five read counts in at least five samples. We further excluded additional samples from the analysis according to three criteria: (a) prior information on the incompetence to follow the study protocol; (b) sample outliers based on Cook's distance, pairwise MA plots, and sample distance heat maps; and (c) subjects without all three timepoint measurements. After applying these exclusions, the final differential expression analysis set of samples included 111 samples from 37 participants (diet group n=24, control group n=13) in the differential expression analysis. 2.6.3 | Statistical analysis of transcriptome Primary genomewide expression analysis of the leukocyte transcriptome suggested suppressed mitochondrial bioenergetic profile following the intense weight loss period.13 Thus, for the purposes of our study, we aimed to investigate mitochondrial bioenergetic pathways further and performed a targeted differential expression analysis for gene sets associated with mitochondrial function (eg, electronic transport chain [ETC], tricarboxylic acid [TCA] cycle, and βoxidation,), mitochondrial biogenesis, and associated energy metabolism pathways and ribosomal metabolism. Human MitoCarta 2.0, Reactome, KEGG, and WikiPathway databases were used as the main gene lists for targeting genes of interest in the aforementioned pathways. In total, 1,700 genes were selected for the differential expression analysis (TableS1). To identify DEGs among these gene sets, we used the Likelihood ratio test to conduct a nested timecourse study with DESeq2 (H0=Group+Time+Group*Subject, H1 =Group+Time+Group*Subject+Group*Time). We investigated whether genes were differentially expressed between the diet and control group across any of the time points when accounting for the betweensubject variability. Posthoc analysis of the Likelihood ratio test was also conducted for the diet and control group only (H0=Subject, H1 = Subject + Time) to further explore withingroup changes. In addition, Wald tests were applied within the DESeq2 interface for testing contrasts for the between/ within group comparison across any two individual time points. A Q value of .05 for false discovery rate (FDR) was used to adjust for multiple testing. 2.6.4 | Pathway analysis of gene level data Downstream pathway analysis of the DEGs was conducted to identify the affected biological pathways. The Ingenuity Pathway Analysis (IPA, QIAGEN Bioinformatics, Aarhus, Denmark) tool was used for analysis as it combines a wide set of integrated databases. 2.6.5 | Analysis of transcriptome centroids and their association with phenotype We also determined how phenotype characteristics and their changes (eg, anthropometric and exercise level) associated with the observed changes in energy metabolism gene pathway centroids using correlation analysis. For each mitochondrial and energy metabolism gene pathway, centroids were calculated and normalized as standard deviation (SD) change from the reference Zscore (PRE) within each sample. In a similar manner, phenotype variables were calculated and normalized as SD change from the reference Zscore (PRE) within each sample. For gene pathway and phenotype variable centroids, statistically significance was calculated using Pearson correlation, where significance threshold was set as P value<.05. 2.7 | Metabolomics: NMR data preparation, quality control, and management A highthroughput serum NMR metabolomics platform was used for the absolute quantification of serum metabolites. The full process and methods of sample preparation and quantification have been described previously.17 The NMR metabolome assay yields a total of 228 different metabolites, including an array of lipoprotein subclasses, apolipoproteins, serumfree fatty acids, and a wide variety of small molecules such as glycolysis precursors, amino acids, and inflammation biomarkers. Of these metabolites, we selected for the present article a total of 22 metabolites known to be associated with mitochondrial function (eg, oxidative phosphorylation, and βoxidation) and energy metabolism (TableS2). For this aim, we analyzed the metabolome of 42 physique athletes (diet group n=25, control group n=17) measured at three time points (PRE, MID, and POST). Prior to the analysis, we assessed data skewness, normality and outliers with dot plots and histograms. To dispose of excess variance caused by outliers, metabolite values were excluded from the analysis if SD was greater or less than four (±4) from the mean. Statistical analysis of the metabolomic data was performed using generalized estimating equations (GEE). 3 | RESULTS 3.1 | Phenotype characteristics of study participants throughout the study As reported previously,10,12,13 after the 20week weight loss period (PREMID), the diet group achieved a 52% decrease in total body fat mass with only minor changes in lean mass, muscle size and maximal and explosive 6 of 15 | SARIN et Al. strength.10 Aforementioned decrease in fat mass measures was accomplished by a ~19% decrease in energy intake (mainly from carbohydrates) and a ~15% increase in total volume of exercise, as measured by metabolic equivalent hours per week (METh/week).10 Following the weight regain period (MIDPOST), all measures on fat mass were reverted back to baseline levels as reported previously.10 In the controls, no significant changes were observed in the anthropometric traits measured throughout the study (PREMIDPOST).10 3.2 | Overview of the targeted transcriptome analysis of mitochondrial and energy substrate metabolism First, we examined the transcriptomic profile of leukocytes in the diet group using a targeted approach with ~1,700 preselected genes with importance in mitochondrial function and energy metabolism. Of those, a total of 303 genes were differentially expressed (161 downregulated and 142 upregulated) (FDR<0.05) following the weight loss period (TableS3). All of the intense weight loss induced changes in gene expression levels were reverted back to baseline levels during the subsequent weight regain period in the diet group (TableS4). Following the entire weight cycling period, the expression levels of four genes were differentially expressed (FDR < 0.05) as (a) ODC1 and ATP5F1E were reverted above baseline levels after the weight loss period, whereas (b) the expression levels of MCUR1 and ELOVL7 were increased only by the weight regain period when compared with baseline levels (Figures2E and S1). Thus, this shows that none of the downregulated genes after the diet routine remained downregulated following the weight regain and the same was true for the upregulated genes. In the controls, no significant (FDR>0.05) changes were observed in the investigated gene expression profiles measured throughout the study (TablesS5 and S6). Pathway analysis of the aforementioned 303 DEGs in the diet group following the intense weight loss period (PREMID) revealed significant (Q value < .05) changes in a number of canonical pathways connected to the regulation of mitochondrial oxidative function and energy metabolism (TableS7). More specifically of the DEGs enriched in canonical pathways, IPA predicted (a) eIF2 signaling (Zscore = −5.20, Q value = 1.00 × 10−38), (b) oxidative phosphorylation (OXPHOS) (Zscore = −2.84, Q value = 2.69× 10−9) pathways to be most strongly (Zscore>|2|) suppressed, whereas NRF2mediated oxidative stress response (Zscore=2.24, Q value=5.25×10−2) pathway was suggested to be activated after the weight loss period (TableS7). In addition, other DEG enriched pathways with tendency (Zscore<|2|) to be suppressed included, (c) mTOR signaling (Zscore=−1.34, Q value=7.94×10−15) pathway whereas activation was suggested for (a) Sirtuin signaling (Zscore=0.83, Q value=5.89×10−10) and (b) AMPK Signaling (Zscore=1.39, Q value=5.62×10−6) (TableS7). 3.3 | Intensive weight loss is characterized by transcriptomic and metabolomic signatures of decreased mitochondrial bioenergetics Given the high enrichment of DEGs in pathways associated with mitochondrial oxidative function and energy metabolism upon weight loss, we aimed to characterize changes taking place in the major cellular energyyielding pathways, glycolytic metabolism and mitochondrial OXPHOS, in more detail. First, enhanced glycolytic metabolism was indicated through moderate upregulation of genes participating in reactions and an activation of glycolysis (FDR<0.05: PGAM2, LDHAL6A, P<.05: HK3, PFKM, PGK1, PKLR, and LDHD). In addition, diminished utilization of glycolysis end product, pyruvate, within mitochondria was suggested through increased expression (FDR<0.05) levels of HIF1A13 that is a central metabolic switch driving glycolytic metabolism and mitochondrial oxidative quiescence (Figure S2B). This observation was corroborated by altered expression profile of several genes (FDR<0.05: PDK4, P<.05: MCP2, PDK1) responsible for (a) pyruvate transport to mitochondria and (b) pyruvate conversion to acetylCoA that fuels the TCA cycle (FigureS2B; TableS3). However, only nominal downregulation was observed in two rate limiting key enzymes of TCA cycle (eg, IDH2 and SDHA) (P<.05) without further changes in other enzymes (TableS3). Second, we aimed to understand how the predicted suppression of OXPHOS pathway affected mitochondrial bioenergetics by mapping expression of all genes encoding the five mitochondrial respiratory complex subunits in the ETC pathway (Figure 2AE; Table S3). Downregulatory (FDR < 0.05) effect was observed for several nuclear genes coding the mitochondrial respiratory chain subunits, where the most distinct suppression of transcriptomic profile was detected in the Complex V (Figure2E; Table S3). Suppression of mitochondrial oxidative metabolism was further indicated through downregulation of genes responsible for regulating mitochondrial fatty acid βoxidation (FDR<0.05: CPT1A, SLC25A20, ETFA, and ACAA2; P < .05: ACADVL and ACADS) (Figure S3B; TableS3). Consequently, together with the leukocyte transcriptomic signatures for suppressed OXPHOS and fatty acid oxidation— also systemic level mitochondrial bioenergetic quiescence was suggested through elevated levels | 7 of 15 SARIN et Al. FIGURE 2 Transcriptomic activity of oxidative phosphorylation (OXPHOS) gene pathway following the intense weight loss period. Here, we demonstrate how the prolonged period of lowenergy availability and intense exercise leading to substantial weight loss modulated gene expression profile of genes associated with OXPHOS pathways. Specifically, transcriptomic changes of genes in Complex I (panel A), Complex II and TCA cycle (panels B), Complex III (panel C), Complex IV (panel D), and Complex V (panel E) have been demonstrated with heatmaps. The intense weight loss period (PREMID) had the most distinct downregulatory effect on the expression levels of Complex V (panel E) genes. Gene expression levels are represented as standard deviation (SD) change from the reference Zscore. Baseline calculated Zscore values (PRE) from both the diet and control groups were separately set as the reference level to which each individual group/timepoint level was compared. On the heat map, blue colors indicate decrease and red colors increase in gene expression level compared with the calculated reference value as depicted with the color key. Statistically significant change is annotated with asterisks (*) within heatmap cells using FDR adjusted Pvalues (false discovery rate [FDR]<0.05) for withingroup timepoint comparisons. Levels of substrates (serum alanine, pyruvate, lactate, and glucose) reflecting the balance between oxidative and glycolytic metabolism during the study period are depicted in panel F (A) (B) (C) (E) (D) (F) 8 of 15 | SARIN et Al. of serum alanine (β=0.143±0.011, FDR=1.0×10−16), pyruvate (β = 0.015 ± 0.006, FDR = 0.02), and lactate (β=0.149±0.062, FDR=0.04) (Figure2F; TableS8). These clinical markers for mitochondrial dysfunction are subsequently increased in circulation when glycolysis end product, pyruvate, is not utilized in sufficient manner for energy production in mitochondria. Third, it was also investigated in more depth how changes in the clinical phenotype10 correlated with the aforementioned changes in the leukocyte transcriptomic bioenergetic profile following the intense weight loss and weight regain period (Table1). Subsequently, significant correlations (r=0.30.5, P<.05) between changes in fat mass parameters (eg, weight, fat mass, android fat mass, and BMI) and mitochondrial oxidative metabolism signature were observed when data from all time points were analyzed (Table 1). Physique athletes with the most drastic drop in fat mass experienced also the most distinct suppression of mitochondrial oxidative metabolism signature, together with a similar positive correlation between weight regain and increase in mitochondrial oxidative bioenergetic signature (Table1). Thus, collectively, our results suggest that a major systemic hallmark of intensive weight loss is the decline in mitochondrial bioenergetics, whereas the weight regain has the potential to rescue these changes. 3.4 | Intensive weight loss is characterized with increased regulatory signaling for mitochondrial biogenesis Consistent with previous CR studies,18 intensive weight loss lead to the aforementioned upregulation of pathways (Q value<.05: Sirtuin signaling and AMPK signaling) regulating mitochondrial biogenesis (TableS7). Closer examination of individual genes involved in “Sirtuin and AMPK signaling” pathways revealed a significant upregulation (FDR<0.05) of three AMPK subunits, whereas no significant changes in the expression of genes coding Sirtuins 17 themselves were detected (Figure3A,B, TableS3). The key downstream mediator of positive effects of AMPK on mitochondrial biogenesis and cellular bioenergetic metabolism— PGC1α (PPARCG1A) was observed to be upregulated (FDR<0.05) upon weight loss. Similarly, nominal changes were observed in its transcriptional targets (a) NRF1 (P < .05), the transcription factor involved in the activation of mitochondrial transcription factor A (TFAM) and transcription of mtDNA encoded genes, and (b) in NFE2L2 (aka. NRF2) (P<.05), the key driver of transcriptional activity of oxidative stress response genes (Figure3A,B; TableS3). These results suggest that intensive weight loss triggers systemic AMPK/PGCαcoordinated induction in mitochondrial biogenesis and oxidative stress defence signaling. 3.5 | Intensive weight loss is characterized with increased signature for cellular exposure to ROS As we observed positive upstream signals by AMPK/PGC1α/ NRF2 and predicted the activation of NRF2mediated oxidative stress response pathway (TableS3), we further analyzed transcription signatures of reactive oxygen species (ROS) and antioxidant defence system. We observed suppressed transcriptomic profile for genes coding major hydrogen peroxide (ROS)- scavenging enzymes (FDR<0.05: GPX1, TXN2, PRDX2, PRDX5, and PRDX6) (FigureS4; TableS3). In addition, enhanced formation of hydrogen peroxide (ROS) was suggested by upregulation (FDR<0.05) of (a) gene coding mitochondrial superoxide dismutase 2 (SOD2) (FigureS4C) and (b) genes regulating initial stages of peroxisomal βoxidation (ACOX1 and ACOXL) (FigureS4B). Taken together, these results imply that longterm energy deficit combined with highvolume exercise training leading to weight loss results in increased cellular transcriptomic signature potentially suggesting increased exposure to ROS and exhaustion of ROS defense system. 3.6 | Intensive weight loss is characterized with transcriptomic profiles of suppressed capacity for cytosolic protein translation Lastly, given that CR has been previously shown to regulate protein synthesis pathways, we assessed more in detail pathways involved in protein metabolism. Suppressed cytosolic and mitochondrial ribosome biogenesis and translation capacity was suggested through (a) the predicted inactivation of the major nutrient sensing pathways regulating protein translation and ribosome biogenesis (Q value<.05: eIF2 signaling and mTOR signaling) that were specifically highly enriched with genes encoding cytosolic ribosomal proteins, (b) repression of genes coding key regulatory factors of protein translation (FDR<0.05: EIF2AK1, EIF2A, EIF3, eEF2),13 (c) decreased gene expression of cytosolic and mitochondrial ribosomal proteins (Figure4A,B; TableS7). Overall, our results depicted that intensive weight loss triggers reduction in the gene expression signatures important for global protein synthesis and translation, which is known to be an energetically expensive process.19 4 | DISCUSSION The current integrated highthroughput omics study, utilizing a unique group of female physique athletes who voluntarily performed high volumes of exercise together with a prolonged period of low energy availability,10,12,13 represents an | 15 of 15 SARIN et Al. 46. Kramer PA, Ravi S, Chacko B, Johnson MS, DarleyUsmar VM. A review of the mitochondrial and glycolytic metabolism in human platelets and leukocytes: Implications for their use as bioenergetic biomarkers. Redox Biol. 2014;2:206210. SUPPORTING INFORMATION Additional Supporting Information may be found online in the Supporting Information section. How to cite this article: Sarin HV, Pirinen E, Pietiläinen KH, et al. Mitochondrial bioenergetic pathways in blood leukocyte transcriptome decrease after intensive weight loss but are rescued following weight regain in female physique athletes. The FASEB Journal. 2021;35:e21484. https://doi.org/10.1096/ fj.20200 2029R