Muscle-liver substrate fluxes in exercising humans and potential effects on hepatic metabolism

Context The liver is crucial to maintain energy homeostasis during exercise. Skeletal muscle-derived metabolites can contribute to the regulation of hepatic metabolism. Objective We aim to elucidate which metabolites are released from the working muscles and taken up by the liver in exercising humans and their potential influence on hepatic function. Methods In two separate studies, young healthy males fasted overnight and performed an acute bout of exercise. Arterial-to-venous differences of metabolites over the hepatosplanchnic bed and over the exercising and resting leg were investigated by capillary electrophoresis- and liquid chromatography-mass spectrometry metabolomics platforms. Liver transcriptome data of exercising mice were analyzed by pathway analysis to find a potential overlap between exercise-regulated metabolites and activators of hepatic transcription. Results During exercise, hepatic O 2 uptake and CO 2 delivery were increased two-fold. In contrast to all other free fatty acids, FFA with 18 carbon atoms and more and a high degree of saturation showed a constant release in the liver vein and only minor changes by exercise. FFA 6:0 and 8:0 were released from the working leg and taken up by the hepato-splanchnic bed. Succinate and malate showed a pronounced hepatic uptake during exercise and were also released from the exercising leg. The transcriptional response in the liver of exercising mice indicates the activation of HIF-, NRF2-, and cAMP-dependent gene transcription. These pathways can also be activated by succinate. This work consists of two separate exercise studies to assess metabolite fluxes over the hepato-splanchnic bed and over the resting and exercising leg. The studies have been described in detail previously (12,13), were executed in accordance with the Helsinki Declaration and were approved by the Scientific Ethics Committee of the capital region of Denmark. Subjects participating in the two studies were not identical. All subjects provided written informed consent to participate. In both studies, the subjects were recreationally active and asked to refrain from strenuous exercise 24 h before the experimental day. Blood was collected in EDTA containing tubes, placed on ice and immediately spun. Plasma aliquots were stored at -80 o C until analysis. dissolved in plasma. Finally, rates of O 2 uptake and CO 2 release were calculated by multiplying the arterial-to-venous difference with the hepatic blood flow. The effect of exercise on metabolite fluxes over the leg was studied by analyzing plasma samples from a one-legged knee extensor study where the exercising leg was compared to the other, resting leg (13). Briefly, after an overnight fast nine healthy male subjects (age: 20.9 ± 0.5 years and BMI: 22.6 ± 0.8 kg/m 2 ) performed continuous one-legged knee-extensor exercise for 2 hours at 50% of maximum workload on a modified Krogh ergometer while the contra lateral leg was resting. Catheters were inserted retrogradely into both femoral veins as well as into one femoral artery. The femoral retrograde placement avoids a contribution from the veins draining the lower abdominal adipose tissue (v. epigastrica superficialis) and leg cutaneous and subcutaneous tissues (v. saphena magna), which is important for the study of lipid metabolism (15). The subjects were fasted until 3 hours after the exercise bout. Blood flow was determined for both the resting and the exercising leg using ultrasound Doppler in three subjects and the mean value was used to calculate net skeletal muscle release/uptake for all subjects. Three blood samples were drawn simultaneously at each time point. Differences in metabolite flux over the hepato-splanchnic bed were detected using one-way ANOVA (time as fixed and subject as random effect) followed by Bonferroni post-hoc test. Flux data were UV-scaled for the heatmap since mean-centering could lead to loss of directional information. The subset of metabolites exhibiting a significantly different hepatosplanchnic flux over time according to one-way ANOVA was further assessed in the one-leg exercise study, employing two-way ANOVA (time and resting/exercising leg flux as fixed, subject as random effect) followed by Bonferroni post-hoc test. Statistical analyses were performed using JMP 13.0 (SAS Institute Inc, Cary, North Carolina, USA). A p-value < 0.05 was considered statistically significant. Data are presented as mean ± standard error (SEM).


Introduction
Since several decades, the working skeletal muscle has been in the focus of research to understand how the muscle can achieve the pronounced increase in substrate oxidation that is necessary to provide ATP for muscle contraction and movement. However, physical exercise is not possible without an orchestrated cooperation of several tissues in order to support muscular work and to maintain metabolic homoeostasis. The central organ in this energydemanding condition is the liver. Hepatic metabolism is crucial for glucose and lipid homoeostasis, for the recycling of metabolites and for the detoxifying of metabolic waste (1).
Knowledge of the hepatic contribution to metabolism during acute exercise is mainly based on mouse studies and on only a few human exercise studies (2). In humans, collection of blood samples from the hepatic vein and peripheral arteries has been employed to analyze arterial-to-hepato-splanchnic differences of selected metabolites and thus assess their hepatosplanchnic fluxes during exercise. These studies verified the exercise-induced increase in hepatic glucose output, the increased uptake of the gluconeogenic precursors lactate, pyruvate and glycerol, of some amino acids, and of palmitate and oleate (3)(4)(5). The change in the hepato-splanchnic substrate fluxes is dependent on the duration and intensity of the exercise bout (6). In a recent study, we expanded the knowledge on exercise-regulated metabolite fluxes over the hepato-splanchnic bed by performing targeted liquid chromatography-mass spectrometry (LC-MS) analysis of acylcarnitines in the hepatic vein, the femoral vein of the exercising and the resting leg, and peripheral arterial samples (7). Exercise amplifies the release of medium chain acylcarnitines from the exercising leg and the uptake in the hepatosplanchnic bed, reflecting the huge increase in fatty acid oxidation in the working muscle and the function of the liver to buffer excess circulating metabolites by their uptake and conversion. In contrast, the liver constantly releases C2-and C3-carnitine during exercise, indicating an excess production of acetyl-CoA and propionyl-CoA (7).
A c c e p t e d M a n u s c r i p t 6 All these data suggest that exercise is an energy-demanding metabolic challenge for the liver that goes along with the need for increased ATP production to maintain biosynthesis of substrates and recycling of metabolites. Well in line, analyses of hepatic tissue of mice after one acute bout of exercise revealed a pronounced change in the hepatic energy state with a drop in ATP, a huge increase in AMP, and activation of the AMP-activated kinase (8,9).
Transcriptome analyses in mice revealed that this is accompanied by profound alterations in hepatic transcript levels related to adaptive responses in glucose and fatty acid metabolism (10). Notably, studies on the release of hepatokines from the liver during or immediately after exercise in humans mirror the exercise-induced increase of the respective transcripts in exercising mice (11).
In humans, collection of hepatic tissue biopsies for molecular investigations before and immediately after exercise is not feasible. But by applying two complementary metabolomics analyses, capillary electrophoresis time-of-flight mass spectrometry (CE-TOF/MS) and ultra high-performance liquid chromatography quadruple-TOF MS (UHPLC-Q-TOF/MS), we can obtain a global picture of exercise-induced changes in the hepatic release or uptake of metabolites and thereby deepen our understanding of the hepatic contribution to exercise metabolism. First, we analyzed the metabolome in arterial and hepatic vein samples obtained at 8 time points during and after a 120 min aerobic exercise trial at 60% VO2max to identify metabolites exhibiting exercise-induced changes. Next, we investigated which of these metabolites are released or taken up by the exercising skeletal muscle in order to obtain information on substrate fluxes between liver and muscle. When compared with the transcriptional response observed in livers of exercising mice, our data indicate that substrates delivered from the working muscle may not only support hepatic metabolism, but also contribute to the regulation of signaling pathways and gene transcription in the liver in response to exercise.

Experimental design
This work consists of two separate exercise studies to assess metabolite fluxes over the hepato-splanchnic bed and over the resting and exercising leg. The studies have been described in detail previously (12,13), were executed in accordance with the Helsinki Declaration and were approved by the Scientific Ethics Committee of the capital region of Denmark. Subjects participating in the two studies were not identical. All subjects provided written informed consent to participate. In both studies, the subjects were recreationally active and asked to refrain from strenuous exercise 24 h before the experimental day. Blood was collected in EDTA containing tubes, placed on ice and immediately spun. Plasma aliquots were stored at -80 o C until analysis.
In the hepato-splanchnic exercise, ten healthy male subjects (age: 22.9 ± 0.8 years and BMI: 22.6 ± 0.5 kg/m 2 ) reported to the laboratory after an overnight fast (12). Catheters were inserted into the brachial artery of the non-dominant arm and into a liver vein via the right femoral vein. The subjects performed a 2-h cycling exercise at 60% VO2 max in semi supine position and remained fasting throughout the whole experimental period. Blood samples were obtained in pairs at each time point. Hepatic blood flow was measured by indocyanine green clearance (12). Samples obtained at baseline (0), 60, 120, 150, 180, 240, 300 and 360 min were used for metabolomics analysis. Samples for blood gas analysis were taken every 15 min throughout exercise and every 15 or 30 min during the recovery period. The content of oxygen (O2) and carbon dioxide (CO2) were measured using an ABL (Radiometer, Copenhagen, Denmark). Whole blood content of CO2 was calculated as described in (14) and whole blood O2 content was calculated by addition of O2 bound to hemoglobin with O2 A c c e p t e d M a n u s c r i p t 8 dissolved in plasma. Finally, rates of O2 uptake and CO2 release were calculated by multiplying the arterial-to-venous difference with the hepatic blood flow.
The effect of exercise on metabolite fluxes over the leg was studied by analyzing plasma samples from a one-legged knee extensor study where the exercising leg was compared to the other, resting leg (13). Briefly, after an overnight fast nine healthy male subjects (age: 20.9 ± 0.5 years and BMI: 22.6 ± 0.8 kg/m 2 ) performed continuous one-legged knee-extensor exercise for 2 hours at 50% of maximum workload on a modified Krogh ergometer while the contra lateral leg was resting. Catheters were inserted retrogradely into both femoral veins as well as into one femoral artery. The femoral retrograde placement avoids a contribution from the veins draining the lower abdominal adipose tissue (v. epigastrica superficialis) and leg cutaneous and subcutaneous tissues (v. saphena magna), which is important for the study of lipid metabolism (15). The subjects were fasted until 3 hours after the exercise bout. Blood flow was determined for both the resting and the exercising leg using ultrasound Doppler in three subjects and the mean value was used to calculate net skeletal muscle release/uptake for all subjects. Three blood samples were drawn simultaneously at each time point.

Sample preparation
Metabolite extraction for CE-TOF/MS-based metabolomics analysis was performed as previously described (16). Briefly, 50 µL of plasma was mixed with 450 L MeOH solution containing ISS1, followed by the addition of 500 L of CHCl3. Subsequently, 200 μL of water was added, and samples were vortexed and centrifuged to form a two-phase system. phosphazene was delivered at 10 μL/min. CE-TOF/MS data acquisition of plasma samples was carried out in both cation and anion mode. The detailed parameters of these two scan modes were previously described (16). FFA containing less than 12 carbon atoms were analyzed by CE-TOF/MS.

Ultra high performance liquid chromatography quadruple-time-of-flight mass spectrometry
FFA containing 12 or more carbon atoms were analyzed using a Waters ACQUITY-UHPLC system (Waters Corp, Milford, USA) coupled to AB SCIEX Triple Q TOF 5600 plus System (AB SCIEX, Framingham, USA) operated in negative ion mode as previously described (18) with slight modifications. The separation was performed on a 2.1×100 mm ACQUITY TM 1.8 µm T3 column (Waters, Milford, MA, USA) and the mobile phase consisted of 6.5 mM ammonium bicarbonate in water (A) and 6.5 mM ammonium bicarbonate in 95% MeOH and water (B). The gradient elution started at 98% eluent A and was maintained for 1 min, then linearly changed to 100% eluent B within 18 min and maintained for 4 min, and finally reverted back to 98% B and equilibrated for 2 min. Flow rate was 0.35 mL/min, and the column temperature was kept at 55 °C. The ion spray voltage was set to 4500 V. Interface heater temperature was 500 °C. Curtain gas, ion source gas 1 and ion source gas 2 were set to

Calculation of metabolite fluxes
Hepato-splanchnic and leg fluxes (negative=release or positive=uptake) of a metabolite (a) were calculated by multiplying the arterial-to-venous difference by the plasma flow:

Pathway analysis of liver transcriptome data of mice
Upstream regulator analysis using Ingenuity Pathway Analysis (Qiagen, Redwood City, CA, USA) was performed with recently published whole genome array data of liver tissue of exercising male C57BL/6N mice (GEO database at NCBI; GSE110747) (19). Data of the control group of chow-fed mice were used for analysis. Livers were obtained immediately after 1 h of treadmill running (13 m/min and 14 uphill slope) (19). Transcripts with a limma ttest p-value < 0.05 and median fold change >|1.5| between exercised mice and sedentary mice were included in the analysis.

Statistics
To identify metabolites with plasma concentrations affected by exercise on a systemic level, a one-way ANOVA was performed on arterial samples from the liver study (with time as fixed and subject as random effect). A Benjamini-Hochberg correction was performed with a false discovery rate (FDR) of 1%. A heatmap from the resulting subset of data was generated by unsupervised hierarchical clustering, using the open-source MultiExperiment Viewer software (20) and employing unit variance (UV)-scaled, mean-centered data.
A c c e p t e d M a n u s c r i p t 12 Differences in metabolite flux over the hepato-splanchnic bed were detected using one-way ANOVA (time as fixed and subject as random effect) followed by Bonferroni post-hoc test.
Flux data were UV-scaled for the heatmap since mean-centering could lead to loss of directional information. The subset of metabolites exhibiting a significantly different hepatosplanchnic flux over time according to one-way ANOVA was further assessed in the one-leg exercise study, employing two-way ANOVA (time and resting/exercising leg flux as fixed, subject as random effect) followed by Bonferroni post-hoc test.
Statistical analyses were performed using JMP 13.0 (SAS Institute Inc, Cary, North Carolina, USA). A p-value < 0.05 was considered statistically significant. Data are presented as mean ± standard error (SEM).

Increase in hepato-splanchnic oxygen consumption during exercise
O2 uptake and CO2 release over the hepato-splanchnic bed were significantly increased immediately after the commencement of exercise and remained at this elevated level until the end of the exercise bout ( Figure 1A,B) with a similar hepato-splanchnic blood flow before (1.81 L/min) and during exercise (1.68 L/min). Before exercise, O2 uptake and CO2 release from the hepato-splanchnic bed were 3 mmol/min. The mean O2 uptake over the hepatosplanchnic bed during exercise was 6 mmol/min, paralleled by a 5 mmol/min CO2 release. In

Exercise increases the proportion of monounsaturated FFA in plasma
Exercise did not cause an equal increase in the arterial plasma concentration of all FFA species. Immediately at the end of the exercise bout, no increase of saturated FFA with a chain length of 18 and more carbon atoms was observed ( Figure 2B). This resulted in an increase in the proportion of several mono-and polyunsaturated FFA, which were elevated after exercise. This effect was particularly pronounced for the abundant monounsaturated FFA species (29±6% before vs. 44±4% after exercise; Figure 2C), while the proportion of the  Figure 3B). This opposite regulation of long and very long chain saturated FFA was also observed before exercise in the fasting state (Table 1).
An increased uptake during or after exercise was also observable for succinate, malate, lactate, and pyruvate, and for the amino acids arginine, glutamine and lysine ( Figure 3A).
Threonate and N-acetylasparagine only showed an uptake after 360 min. Of all these metabolites, only succinate and FFA 8:0 were significantly taken up in the fasting state ( A c c e p t e d M a n u s c r i p t

Increase in substrate fluxes between skeletal muscle and liver during exercise
Next, we studied which of the metabolites with exercise-induced changes in the hepatosplanchnic flux showed an oppositely changed flux over the exercising leg compared to the resting leg in the one-legged exercise study. This was, as expected, the case for lactate, which was released from the exercising leg after 60 min of exercise ( Figure 4A). The exercising leg also contributed to the systemic increase of malate, succinate, FFA 6:0 and 8:0, which may support the uptake into the hepato-splanchnic bed ( Figure 4B-E). Thus, these metabolites may circulate from the contracting muscle to the liver. In particular TCA cycle metabolites malate and succinate showed similar exercise-regulated kinetics of their arterial concentration and fluxes with a pronounced hepatic uptake after 60 and 120 min of exercise ( Figure 4B,C).

Fatty acids, cAMP, HIF1A, and NRF2 are identified as upstream regulators of the hepatic transcriptional response to exercise
Metabolites can be shuttled into metabolic pathways, but also act as signaling molecules: by binding to membrane or nuclear receptors and to intracellular proteins, or as substrate for protein modifications with impact on transcriptional activation and epigenetic regulation of gene expression (21)(22)(23). In order to investigate which of the metabolites with an increased hepato-splanchnic uptake during exercise are potentially implicated in the regulation of hepatic transcripts, we analyzed the acute transcriptional response of the liver to exercise.
Since the acute regulation of hepatic transcripts is hardly accessible in humans, we made use of data from exercising mice recently published by our group (19).  (Tables 3,4), which is in accordance with previous data (10).
Transcription data also indicated pronounced cAMP-dependent activation of transcription factors (CREB, CREM, and FOXO) and activation of the hypoxia-induced factor (HIF)1A, as well as activation of nuclear factor erythroid 2-related factor 2 (NRF2)-dependent transcription. Fatty acids and cAMP were among the identified endogenous substances implicated in the transcriptional activation of genes (Table 4).

Discussion
In this study we provide, for the first time to our knowledge, a comprehensive metabolomics analysis of human plasma samples collected at 8 time points from the artery and the hepatic vein during and after one acute bout of exercise. The participants performed aerobic endurance exercise at 60% VO2max for 120 min. At this intensity, hepatic blood flow was not reduced and an increase in hepatic oxygen uptake was observed. An approximately 2-fold increase of oxygen uptake has also been reported in other studies performed at a moderate exercise intensity, e.g. during exercise at 30% VO2max for 240 min (3), at 55% VO2max for 60 min (24), and at 60% VO2max for 120 min (25). Even with a 50 % reduction in hepatic blood flow during intense exercise a slightly increased oxygen uptake was reported (6), which underlines the capacity of the hepato-splanchnic region to increase oxygen extraction in order to maintain metabolic processes during states of blood flow redistribution towards the working muscle (26). Together with the increase in CO2 release this is a clear indication of higher metabolic activity in the liver of exercising humans and underlines an increased demand for ATP as reported in the liver of exercising mice (8).
An important ATP-consuming metabolic process in the liver during exercise is the production of glucose from glucogenic precursors. In the overnight fasted participants of our study, net hepato-splanchnic glucose production increased from 1 mmol/min to approximately 3 A c c e p t e d M a n u s c r i p t mmol/min (12). Gluconeogenesis accounts for 20-25% of total glucose production when exercise is performed after an overnight fast at 45 or 65 VO2peak, conditions similar to our study (27). Increased hepato-splanchnic uptake of the glucogenic substrates lactate, pyruvate, and lipolysis-derived glycerol during exercise has been reported previously (4,28). Covering also polar metabolites by our CE-TOF/MS we confirmed the exercise-induced increase in the hepato-splanchnic uptake of lactate and pyruvate. The exercise-dependent regulation of glycerol was not detectable with our approach since the filters used for sample preparation contain glycerol, which cannot be completely washed out and cannot be distinguished from plasma glycerol. Two of the amino acids that exhibited an increased hepato-splanchnic uptake during the trial, arginine and glutamine, are glucogenic. Other glucogenic amino acids such as alanine showed already at baseline, in the fasting state, a positive value for the hepato-splanchnic flux and thus an uptake ( Table 2). The findings demonstrate that our approach is suitable to analyze substrate fluxes over the hepato-splanchnic bed.
A constant finding of our study is that the chain length and degree of saturation has a strong influence on the hepato-splanchnic exchange of FFA during exercise, but also in the baseline  which is in accordance with previous reports (32). The data also indicate that the contribution of other non-hepatic intraabdominal tissues to the measured hepato-splanchnic fluxes must be considered.
A novel finding is the efflux of FFA 6:0 and FFA 8:0 from the exercising muscle and their increased uptake into the hepato-splanchnic bed. Since the respective C6:0 and C8:0 acylcarnitine esters are strongly increased in exercising muscle tissue (14-and 8-fold, respectively) (7), fatty acid chain shortening during β-oxidation followed by hydrolysis of acylcarnitine esters is a likely explanation for the increased release from the exercising leg.
Medium chain fatty acids are considered as easy accessible fuels since they can be metabolized independent of proteins for binding and transport (33). muscle, our data revealed the uptake of malate and succinate into the hepato-splanchnic bed during exercise and their release from the exercising leg. There was no or only a small detectable uptake before exercise (Table 1) suggesting that the pronounced increase in arterial plasma is the driver of the increased uptake. Arterial concentrations and fluxes of malate and succinate showed similar kinetics during the study. Both are intermediates of the citric acid cycle and succinate is converted to malate via succinate dehydrogenase (SDH), which catalyzes the reduction to fumarate and fumarase, which catalyzes the hydration to malate.
Since SDH couples the TCA cycle to the respiratory chain, succinate is an electron donor and driver of oxidative phosphorylation, but it is also substrate for protein modifications and has signaling and paracrine/endocrine effects (36). Succinate has structural similarity to 2ketoglutarate and can compete for the binding to 2-ketoglutarate-dependent dioxygenases (2-KGDD), causing inhibition of these enzymes (37). One prominent target of 2-KGDD is the transcription factor HIF, which was activated in the livers of exercising mice according to transcriptome analysis ( Table 4). The inhibition of these enzymes by succinate causes stabilization of HIF, increased activation of HIF-target genes (38), and links the accumulation of succinate to HIF-induced IL1beta expression (39). High fumarate concentrations cause S-(2-succino)-cysteine modifications of proteins, a process called succination (40). The succination of Kelch-like ECH-associated protein-1 (KEAP)1 releases the transcription factor NRF2 from its complex with KEAP1 and enables its nuclear translocation and activation of target genes (41). NRF2 was also identified as transcription factor activated in the livers of exercising mice (Table 3, 4). Another mechanism how elevated succinate concentrations could influence hepatic signaling pathways was described recently in mice, where the succinate-induced HIF activation leads to accumulation of cAMP by inhibition of its degradation (42). Notably, cAMP is a master regulator of the transcriptional response to exercise in the liver which is responsible for the activation of the CREB1, CREM and FOXO A c c e p t e d M a n u s c r i p t 20 transcription factors (Table 4). Thus, accumulation of succinate has been linked to the activation of certain transcription factors and to cAMP-dependent transcriptional activation, and these succinate-dependent pathways share a great overlap with the pathways which are found to be acutely activated in the liver of mice after exercise (43). This overlap remains speculative without experimental validation. Moreover, we performed this comparison of metabolomics with transcriptomics data across different species, which can be seen as another limitation. But human transcriptome data obtained from liver tissue under exercise conditions are hardly accessible. In conclusion, we suggest that succinate can be a candidate for the list of metabolic upstream regulators of the hepatic exercise response, such as fatty acids as PPAR activators and the glucagon/insulin ratio as cAMP inducer. Our data are a first hint that succinate accumulation in hepatic tissue can support the acute transcriptional response to exercise. Notably, the more than 6-fold increase in the hepato-splanchnic uptake of succinate may cause tissue concentrations sufficient to inhibit 2-KGDD (44,45).
It is intriguing to hypothesize that the working leg is a major source of the increased hepatosplanchnic uptake of the TCA metabolites, lactate, and FFA 6:0 and FFA 8:0. However, the metabolomics data were obtained from two different exercise studies, and a direct comparison is not possible since the exercise modalities are different. In particular the ongoing hepato-splanchnic uptake in the second exercise phase after the leg release peaked at 60 min raises the question of other exercise-dependent sources of these metabolites. To the best of our knowledge, no human data are available to support the contribution of other tissues. It can be speculated that tissues with increased metabolic activity during exercise such as the heart can contribute with an increased net release of TCA metabolites. Up to now, the release of succinate from the heart was mainly reported during ischemic conditions (46).
Another possible candidate is the adipose tissue, which is postulated to show increased net release of lactate during exercise (47). The results underline the essential function of the hepatic metabolism during exercise and support the relevance of the crosstalk of working muscles and the liver to exchange metabolite substrates, but also signaling molecules, that support and mediate compensatory and adaptive processes during exercise.     Negative values indicate a hepato-splanchnic release, positive values an uptake. The flux was assessed by a t-test against 0 (no uptake/release), a p<0.05 was considered significant A c c e p t e d M a n u s c r i p t Negative values indicate a hepato-splanchnic release, positive values an uptake. The flux was assessed by a t-test against 0 (no uptake/release), a p<0.05 was considered significant.  A c c e p t e d M a n u s c r i p t Table 4 Upstream regulators implicated in the activation of genes in the liver of mice after a 1htreadmill run Ingenuity upstream regulator analysis of transcripts that were significantly different (limma ttest p < 0.05 and median fold change >|1.5|) between exercised and sedentary mice immediately after a 1h treadmill run (GSE110747; (19)). Regulators with z-score >2.0 (i.e., predicted to be activated) and with a p-value of overlap <0.05 are shown (other regulator groups are not included). Transcription regulators discussed as succinate-regulated factors are formatted bold.