In the present work, we demonstrated that the systemic inflammatory response induced by LPS in healthy volunteers significantly alters the plasma metabolome, with the most profound changes taking place 4 h following LPS administration. Endotoxemia mainly led to an increase of several nucleosides, nucleotides, and derivatives, and a decrease in many glycerophospholipids. Before LPS administration, trained subjects practicing the learned breathing exercises exhibited higher levels of lactate and pyruvate compared with a control group who did not practice any exercise, and concentrations of these metabolites correlated with the profoundly enhanced levels of the anti-inflammatory cytokine IL-10 observed in trained individuals following LPS administration. We subsequently validated these findings in vitro, by showing that co-incubation with lactate and pyruvate enhances LPS-induced IL-10 release and attenuates pro-inflammatory cytokine production by primary human leukocytes.
Two of the most profoundly enhanced metabolites after the LPS challenge were deoxyuridine triphosphate and especially, deoxycytidine, both involved in the pyrimidine metabolism pathway, which was enriched following LPS administration. Increased plasma deoxycytidine levels result from DNA degradation [13
]. We have previously shown that plasma levels of both nuclear and mitochondrial DNA show a transient increase during endotoxemia, reaching their maximum levels 3 h after LPS administration, after which they gradually return to baseline in the following hours [14
]. Hence, it might be speculated that the breakdown of plasma DNA is one of the main drivers of the increased deoxycytidine concentrations observed 4 h after LPS administration. Many metabolites belonging to the class of glycerophospholipids were distinctly decreased following administration of LPS. Related to this, various pathways related to lipid metabolism were enriched, which is in accordance with previous findings obtained in healthy volunteers undergoing experimental endotoxemia [15
]. Glycerophospholipids are the main components of cell membranes and function as precursors to signaling molecules involved in many cellular and physiological processes [16
]. The majority of the decreased metabolites belonging to this class consisted of phosphatidylcholines. Previous work from our group has shown that plasma levels of secretory phospholipase A2 (sPLA2
), the principal catalysts of glycerophospholipid hydrolysis, greatly increase during endotoxemia [17
], which may be an explanation for the reduced plasma glycerophospholipid concentrations. The Warburg effect is another pathway of interest that was enriched during systemic inflammation, with four significantly increased (6-phosphonoglucono-D-lactone, glucose, lactate, pyruvate) and one significantly decreased (l
-glutamine) metabolite. This process entails the shift from oxidative phosphorylation (OXPHOS) as the primary energy source towards aerobic glycolysis and was recently shown to play a critical role in mounting (LPS-induced) immune responses, by, among other things, facilitating rapid production of inflammatory cytokines [18
Following the training program, trained subjects exhibited lower plasma levels of L-carnitine compared with untrained controls. This relative depletion of carnitine from plasma may indicate an overall increment in Acyl-shuttling mechanism in the mitochondria by CPT1/CPT2 and might be related with lipid beta-oxidation intensification [21
]. The most pronounced changes in the plasma metabolome of trained subjects compared with controls were, however, observed during the period in which the breathing exercises were carried out. This was not unexpected given the major changes in cardiorespiratory parameters, markers of autonomic nervous system activity, and inflammatory molecules and symptoms observed during this period, as reported elsewhere [1
]. At T = 0, when the trained individuals had been practicing the breathing exercises for 30 min but just prior to LPS administration, the Warburg effect was one of the top enriched pathways in trained individuals compared with controls. Plasma levels of four metabolites (lactate, 6-phosphonoglucono-D-lactone, pyruvate, and isocitric acid) of this pathway were increased in the trained group, whereas concentrations of l
-glutamine were lower. This finding may be counterintuitive, as the Warburg effect is mainly associated with pro-inflammatory effects, whereas trained subjects exhibited a distinct anti-inflammatory phenotype [1
]. The fact that LPS, a strong inducer of the Warburg effect [20
], had not yet been administered at this early timepoint, renders this finding biologically implausible as well. Finally, as alluded to before, the Warburg effect encompasses a shift from oxidative phosphorylation (and thus usage of the citric acid cycle) towards glycolysis, but the citric acid cycle pathway was also enhanced in the trained group. Therefore, we hypothesize that the observed effect is not due to a true Warburg effect (aerobic
glycolysis), but rather driven by the profoundly increased lactate levels in this group as a result of classical anaerobic
glycolysis in the muscles. The latter may have been caused by the combination of vigorous breathing (including repeated muscle tightening), intermittent hypoxia, and epinephrine-induced vasoconstriction in this group [1
]. The breathing exercises could certainly be regarded as exercise, given that, similar to exercise, they resulted in profuse sweating and exhaustion. The fact that lactate and pyruvate levels were highly intercorrelated provides a strong indication of increased gluconeogenesis (i.e., the generation of glucose from non-carbohydrate substrates such as lactate, with pyruvate as the intermediate metabolite, which mainly takes place in the liver). Gluconeogenesis was also one of the top enhanced pathways in the trained group during practicing of the breathing exercises. Because epinephrine is a well-known strong inducer of gluconeogenesis [22
], the profoundly increased plasma epinephrine levels observed shortly after initiation of the breathing exercises [1
] likely play a pivotal role in this effect. Combining the findings of enhanced anaerobic glycolysis with increased gluconeogenesis, our data provide evidence of increased activation of the Cori cycle in trained subjects during practicing of the breathing exercises, in which lactate is produced by muscle cells and subsequently converted back, via pyruvate, into glucose in the liver [23
]. However, it needs to be acknowledged that we can only infer activation of the Cori cycle, as we did not perform isotype labeling experiments.
Correlation analysis between differentially regulated metabolites and inflammatory cytokine levels in trained subjects revealed that both lactate and pyruvate were related to plasma concentrations of the key anti-inflammatory cytokine IL-10. Of note, production of this cytokine was profoundly accelerated and enhanced in trained subjects practicing the breathing exercises, and its plasma levels were strongly correlated to the subsequent attenuation of pro-inflammatory cytokine responses [1
]. Our PBMC stimulation experiments demonstrated that pyruvate alone, but especially the combination of both pyruvate and lactate, exert a robust enhancing effect on LPS-induced IL-10 production. These results may substantiate the correlations found during endotoxemia in vivo. Furthermore, both lactate and pyruvate individually, as well as the combination of both metabolites, attenuate IL-1β production, and lactate mitigated the release of IL-6, another pro-inflammatory cytokine. Both lactate and pyruvate have been demonstrated previously to exert anti-inflammatory effects. For instance, very recent work demonstrates that addition of lactate to LPS-stimulated primary human monocytes causes a distinct metabolic shift by decreasing aerobic glycolysis and increasing oxidative phosphorylation, a metabolic state characteristic for anti-inflammatory responses [24
]. In the same study, lactate was shown to decrease LPS-induced production of pro-inflammatory cytokines by human PBMCs [24
]. The authors proposed that immunomodulatory effects of lactate may serve as a feedback signal to limit excessive inflammatory responses of highly glycolytic pro-inflammatory immune cells [24
]. Pyruvate has been shown to decrease mRNA expression and protein levels of pro-inflammatory cytokines TNFα and IL-6 in LPS-stimulated canine PBMCs, whereas IL-10 expression and production was increased [25
]. Furthermore, administration of pyruvate significantly lowered IL-6 and enhanced IL-10 plasma concentrations in LPS-treated rats, leading to prolonged survival, and incubation with ethyl pyruvate blocked activation of nuclear factor (NF)-κB, a critical pro-inflammatory transcription factor, in LPS-stimulated murine macrophages [26
]. These data strengthen the notion that also in the human in vivo situation, the observed increases in lactate and pyruvate at least partly account for the immunomodulatory effects observed in trained subjects.
Several limitations and aspects of the present work deserve attention. First, only male volunteers were included. In earlier work, we observed that the endotoxin-induced pro-inflammatory immune response is more pronounced in females than in males [27
]. Furthermore, menstrual cycle-induced variation in hormone levels can also impact immune parameters [28
], thereby further increasing variation. Therefore, inclusion of both sexes yields more inter-individual variation and would necessitate larger group sizes. Because endotoxemia studies are very labor-intensive and costly, the choice was made to only include males, which nevertheless limits this study’s generalisability.
A second limitation, which applies to all plasma metabolomics studies, pertains to the uncertainty of the source of the measured metabolites. Virtually all identified metabolites can be produced by a wide variety of cells in many organs. Especially in immunological studies like this, combining plasma metabolomics with the determination of intracellular metabolites in immune cells (e.g., leukocytes) would represent a more powerful approach. Unfortunately, no samples were stored for this purpose.
Third, the current study does not allow us to deduce which (combination) of the three elements of the training program is responsible for the observed effects, and several of our ongoing studies are aimed at elucidating this question. Nevertheless, as alluded to in a previous section, the breathing exercises probably play a pivotal role.
Fourth, because metabolomic profiling was performed on samples stored for three years at −80 °C, sample degradation could be an issue. Nevertheless, the high correlation between lactate measured during the experiments using a point-of-care analyzer and by LC-MS indicates that, at least for this metabolite, no significant degradation occurred during storage. Furthermore, because storage time was virtually identical across samples (all were collected within one month), the extent of degradation of certain metabolites would have been similar across all samples. Provided that metabolites did not completely degrade, this would therefore have little impact on the results.
Fifth, as the PBMCs stimulation experiments with lactate and pyruvate contained a mixture of different cell types (predominantly monocytes and lymphocytes), we cannot be certain about the cellular origin of the effects observed. However, it is well established that monocytes are the main cytokine producers in short-term whole blood LPS stimulation assays (which next to PBMCs also contain granulocytes) [30
]. Furthermore, as discussed earlier on, anti-inflammatory effects of lactate and pyruvate have been reported in primary human monocytes [24
] and murine macrophages [26
], the latter of which show many similarities to monocytes. Therefore, it is likely that the effects of pyruvate and lactate we observed can predominantly be ascribed to monocytes.
Finally, and importantly, all training procedures described in this study were conducted in the presence of medical personnel. Because of profound physiological effects of the breathing exercises (e.g., acid-base shifts, intermittent hypoxia) and exposure to cold [1
], potential health risks while practicing these elements of the training program should be considered.
In conclusion, the present study extends our previous findings regarding the effect of a training intervention consisting of cold exposure, breathing exercises, and meditation on the LPS-induced immune response in healthy volunteers. Practicing the breathing exercises acquired during the training program results in enhanced activity of the Cori cycle, and next to the previously established relationship between epinephrine and IL-10 induction [1
], the current data indicate a role of lactate and pyruvate in the enhanced production of this key anti-inflammatory mediator and in the overall anti-inflammatory phenotype observed in trained subjects.