Exercise blood metabolomics analyses help unveil the metabolic landscape of physical activity, with the caveat that interpretations must be considered in light of exertion level, type and length of exercise, and time frame of measurements (e.g., see Brugnara et al., 2012; Chorell, Moritz, Branth, Antti, & Svensson, 2009; Lewis et al., 2010; Davison et al., 2018; Hansen et al., 2015; Mueller‐Hennessen et al., 2017; Nieman et al., 2014, 2015; Nieman, Sha, & Pappan, 2017; Peake et al., 2014; Zafeiridis et al., 2016). Our recent papers (Grapov et al., 2019; Zhang et al., 2017) and the results herein represent some of the first data that relate the exercise metabolome with improved metabolic health and fitness in the same cohort, and provide first‐ever information related to exercise metabolomics in women. In these studies, within the same cohort of women metabolite patterns were compared between a less healthy state (obesity, insulin resistance, and sedentary lifestyle) and a more healthy state characterized by increased insulin sensitivity and cardiorespiratory fitness following a weight loss and training intervention (described in: Campbell et al., 2014).

Previously in this cohort, we demonstrated that an acute submaximal exercise bout leads to rapid and robust changes in blood concentrations of acylcarnitines and small molecules reflective of intermediary metabolism (Grapov et al., 2019; Zhang et al., 2017). The current experiment extends those results through to OxL and eCB lipid mediators. Very few studies have considered OxL patterns during exercise (e.g., Giordano et al., 2011; Gollasch, Dogan, Rothe, Gollasch, & Luft, 2019; Nieman et al., 2019; Stanford et al., 2018) and no studies, to our knowledge, have determined how improved metabolic health and fitness in previously sedentary overweight individuals influences OxL or eCB patterns. Our results point to several important take‐home messages.

First, consistent with our previous results for acylcarnitines and small molecule patterns (Grapov et al., 2019; Zhang et al., 2017), exercise led to a rapid and significant change in concentrations of many blood OxL and eCB metabolites over the first 10 min of exertion, regardless of pre‐ or post‐weight loss and fitness intervention status. This speaks to the quick, adaptive shifts in metabolism that must take place to meet the energy needs associated with initiation of aerobic muscle work (e.g., ramp‐up of fat oxidation blended with glucose oxidation). Based on plasma metabolomics patterns, we have previously proposed that acute aerobic exercise leads to reduction of nonoxidative metabolic pathways for glucose (Grapov et al., 2019) and possibly reduces branched chain amino acid mitochondrial oxidation, at least transiently (Zhang et al., 2017). Since the current study design did not allow for flux determinations, the main tissue drivers of OxL and eCB blood profiles during acute exercise (0–10 min) or with recovery remain to be established.

Second, regardless of the pre‐ or post‐intervention state, exercise generally led to reductions in OxL metabolites, most notably members of the HETE, DiHETrE, and PG families as well as 13‐HOTE, 17,18‐DiHETE, and 19,20‐EpDPE. A survey of all metabolite patterns individually (Materials S1) further highlights acute reductions in most OxLs, remaining low or rising with time to concentrations about the same as the pre‐exercise sample (even if not statistically significant due to high variability). This contrasts with at least one previous report using a 75 km time trial model in male and female cyclists, in which OxL concentrations were generally increased immediately following the exercise challenge (Nieman et al., 2019). Qualitatively similar to our results, following a 40 min workout test at 70% of heart rate reserve in healthy men and women, Stanford et al. found that 13 blood OxLs were significantly reduced: 9‐oxoODE, 13‐oxoODE, 9‐HOTrE, 13‐HOTrE/13‐HOTrE(r), 13‐HODE, 9(10)‐EpOME, 9‐HEPE (hydroxyeicosa(penta)enoic acid), 12‐oxoETE, 14(15)‐EET, 12‐oxo‐LTB4, 8‐HDHA, 19(20)‐EpDPE and LTC4 (Stanford et al., 2018). In a different study, following 40 min cycling at 60% of maximal exercise intensity in 14 moderately active healthy individuals, there were no clear changes in OxL concentrations except for modest increases in 14(15)‐EET and14(15)‐DHET (Giordano et al., 2011). Taken together, the results from the literature and the current results indicate that exercise‐associated OxL blood patterns are likely dependent on exercise intensity or other related factors (e.g., sympathetic activation, stress or cytokine release), yielding variable results depending on study cohort phenotypes and experimental design.

The initial reduction in OxL concentrations in response to exercise in our cohort does not reflect limited availability of substrate fatty acids from the blood pool. Indeed, during acute exercise the plasma glycerol concentration increased (Grapov et al., 2019) and fatty acid concentrations dropped during exercise and rebounded in the recovery phase in this cohort (Grapov et al., 2019); and see Materials S1 for OxL‐related PUFA precursor patterns. Such a pattern is consistent with exercise‐associated lipolysis driven by sympathetic nervous system activation, coupled to accelerated muscle (and heart) LCFA uptake and combustion (Brooks, 1997; Hansen et al., 2015; Romijn et al., 1993). Another possibility is that enzymatic synthesis of OxL metabolites is reduced with acute aerobic exercise. One hypothesis is that plasma OxL reductions in the face of accelerated fatty acid flux with exercise is due in part to greater flow of precursor fatty acids toward oxidation in lieu of nonoxidative fates. The DiHETrEs, DiHETEs and DiHOMEs are produced through soluble epoxide hydrolase (sEH)‐dependent metabolism of antiinflammatory factors, like the cytochrome p450 (CYP) product 19,20‐EpDPE synthesized in the vascular endothelium and multiple peripheral tissues (Newman, Morisseau, & Hammock, 2005; Spector, 2009). On the other hand, HETEs, HODEs, LTB, LXA and HpODEs are products of lipoxygenases, cyclooxygenases, and/or the actions of reactive oxygen species (Mallat et al., 1999; Powell and Rokach, 1851). Gollasch et al. (2019), analyzing the plasma of a healthy nonobese group of five men and one woman undergoing treadmill exercise (2.7 km/h, 5% grade to exhaustion), did not observe differences in product/precursor ratios for metabolites reflective of sEH or CYP pathways in samples collected coincident with a heart rate of 150 beats per minute. In a post hoc evaluation based on product:precursor ratios for several epoxide/diols pairs (Table S6), there was a trend toward exercise‐associated increases in apparent sEH activity over the t0–t20 min time frame, followed by reduction at the t50 min recovery time point. Thus, there is little evidence that down‐regulation of OxL enzyme systems explains reduced blood OxLs with acute submaximal aerobic exercise in humans. The recovery‐associated dip in sEH surrogate indices appeared to be greater post‐intervention (Table S6), suggestive of a regulation of sEH biochemistry by metabolic health status.

The third key observation from the current study relates to 12,13‐DiHOME, an OxL previously reported as increased following a submaximal exercise bout (40 min workout at 70% of heart rate reserve or a 75 km time trial model in adults) (Nieman et al., 2019; Stanford et al., 2018) and mice (Stanford et al., 2018). This metabolite has been implicated in promotion of muscle fatty acid uptake, C2C12 myotube and whole‐body fatty acid oxidation, and increased myotube O₂ consumption capacity in C2C12 myotubes (Stanford et al., 2018). Moreover, 12,13‐DiHOME has been shown to reduce myotube glucose uptake and inhibit insulin‐dependent Akt phosphorylation, and increase Ccl2 (i.e. MCP1) secretion (Huang et al., 2018). Brown adipose tissue is an important producer of this metabolite in mice, and administration of 12,13‐DiHOME promotes fatty acid uptake in brown fat (Lynes et al., 2017). Since in our exercise model there was no clear rise in plasma 12,13‐DiHOME, and there was no apparent change in the exercise study (Giordano et al., 2011) of R.M. Giordano et al. (JWN, personal communication), a systemic increase in this OxL is not a sine qua non for tissue fat uptake and combustion during aerobic exercise in humans. Based on results to date, the physiological role and regulation of this OxL remains equivocal. That said, weight loss and fitness intervention lowered fasting‐to‐exercise‐to‐recovery concentrations of 12,13‐DiHOME in the current study, which points to some type of connection between metabolic health status and 12,13‐DiHOME biochemistry.

Fourth, we previously described several OxL metabolite species to differ in overnight‐fasted T2DM women compared to nondiabetic women (Grapov et al., 2012); also see Section 1), so it is of interest to consider if those metabolites were also altered in the current cohort as their metabolic health improved with intervention. Preintervention, this study cohort was insulin resistant, sedentary and overweight; thus, metabolites increased in T2DM might be reduced by weight loss and fitness intervention. For instance, epoxides were higher by ~44% to 127% in T2DM (Grapov et al., 2012). However, no clear, significant differences were observed following intervention in the current cohort. T2DM was marked by ~30% to 40% increases in some diols, and herein we did observe lowered concentrations of several diols following intervention. Most convincingly, concentrations of 9,10‐DiHODE, 12,13‐DiHODE, and 9,10‐DiHOME were lower postintervention and displayed a statistically significant intervention effect. One hypothesis is that plasma diols reflect changes in sEH activities that associate with shifts in insulin sensitivity and metabolic health (see Xu et al., 2016). While a complete understanding of the role that sEH plays in glucose homeostasis has yet to emerge, sEH ablation increases pancreatic islet size and insulin signaling in high fat fed mice (Luria et al., 2011). Conversely, overexpression of this enzyme is associated with diabetes‐associated comorbidities including diabetic retinopathy and impaired wound healing (Hu et al., 2017; Sun et al., 2018). Despite the possibility that changes in sEH activity play a role in pre‐ versus post‐intervention OxL differences in our cohort, the sEH product/precursor ratios (diols:epoxides) for 9,10‐DiHODE, 12,13‐DiHODE, and 9,10‐DiHOME were not convincingly altered (~4%, 12%, 11% reductions, respectively). Another plasma metabolite, 13‐KODE, was higher by ~86% in T2DM women (Grapov et al., 2012); intervention in the current cohort led to a reduction in 13‐KODE at the overnight‐fasted and 10‐min exercise time points, but this was not statistically significant. Notably, the precursor metabolite 13‐HODE was also reduced at these time points postintervention. From these results, one can speculate that improved metabolic health associates with lower flux through (at least some) LOX pathways. We conclude that plasma concentrations of several OxL classes in the overnight‐fasted state appear to track metabolic health status in women, perhaps reflective of differences in substrates and sEH and LOX enzyme activities that are regulated by factors such as insulin action or lipid availability.

Finally, in addition to OxL metabolites, ω3 and ω6 fatty acids (as well as C16:0, C18:0, and C16/C18 MUFAs) can give rise to a variety of eCBs (Fezza et al., 2014; Tsuboi, Uyama, Okamoto Y, & Ueda, 2018). A number of studies have evaluated the impact of acute aerobic exercise on circulating eCBs in healthy adults, but no clear patterns have emerged. Further, analyses typically have not covered the entire spectrum of eCBs and derivatives. Sparling, Giuffrida, Piomelli, Rosskopf, and Dietrich (2003) reported ~50% increased AEA but no change in 2‐AG following 50 min of treadmill or cycling at 70%–80% of maximum heart rate in young men. In a study of male well‐trained cyclists, at 60 min of exercise at 55% of maximum power output on an ergometric bicycle, 2‐AG and AEA were unchanged, but there was a very modest increase in PEA and higher OEA (Heyman et al., 2012). In the same study, the moderate exercise was immediately followed by an intense time trial paradigm (work equal to 30 min at 75% of maximum power output), which did not further raise PEA but increased OEA and caused a small rise in AEA. Raichlen, Foster, Seillier, Giuffrida, and Gerdeman (2013) evaluated the impact of 30 min at varying work intensities (~45%–92% of maximum heart rate) in regular runners (men and women), and reported no change in 2‐AG but a significant increase in AEA at ~72% of maximum heart rate, other intensities displayed no change or equivocal results. In our study, there were reductions in ω6 fatty acid‐derived eCBs during the acute exercise phase, including LEA, 1‐AG, 2‐AG and AEA. Even when considering the 20 min and recovery patterns (Materials S1), there is little‐to‐no evidence for an exercise‐induced increase in plasma eCBs and derivatives. From this, and a review of the literature, we conclude that modest aerobic exercise in adults does not, by definition, trigger eCB increases in blood. It is possible that exercise intensity, training status, diet and energy balance, body composition/adiposity, time of collection, or exercise type impact select eCBs. More studies are needed that systemically address fitness level, exercise type, and temporal patterns during stepped workloads to evaluate fully the role of acute exercise on eCB biochemistry and physiology.

In conclusion, we observed that acute (0–10 min) submaximal exercise in women leads to generalized reductions in plasma concentrations of OxL and eCBs, which tend to recover to baseline fasting concentrations later in exercise and recovery but are not increased overall by exercise. There was no increase in 12,13‐DiHOME, a bioactive lipid previously hypothesized to be involved in exercise‐associated muscle fatty acid uptake and energetics. It is acknowledged that measurement of these lipid mediators in whole blood plasma may not report on localized or tissue‐specific regulation (i.e., in red blood cells, peripheral blood mononuclear cells, vasculature or microvasculature, adipose or liver), and the latter may be very important in OxL or eCB actions on inflammation responses, vascular tone, or other outcomes. Furthermore, the potential for lipoproteins to ferry and sequester complex lipids cannot be ignored, and is not addressed by measuring the total plasma pool. Future experiments can address these factors during exercise or with changes in metabolic health. Altogether, the results herein and those from the literature highlight that acute exercise drives significant changes in circulating pools of OxLs and eCBs, raising the possibility that that some of the acute and long‐term effects of exercise on whole‐body physiology involve local or systemic actions of these classes of complex lipids.