13C-Metabolic flux analysis detected a hyperoxemia-induced reduction of tricarboxylic acid cycle metabolism in granulocytes during two models of porcine acute subdural hematoma and hemorrhagic shock

Introduction Supplementation with increased inspired oxygen fractions has been suggested to alleviate the harmful effects of tissue hypoxia during hemorrhagic shock (HS) and traumatic brain injury. However, the utility of therapeutic hyperoxia in critical care is disputed to this day as controversial evidence is available regarding its efficacy. Furthermore, in contrast to its hypoxic counterpart, the effect of hyperoxia on the metabolism of circulating immune cells remains ambiguous. Both stimulating and detrimental effects are possible; the former by providing necessary oxygen supply, the latter by generation of excessive amounts of reactive oxygen species (ROS). To uncover the potential impact of increased oxygen fractions on circulating immune cells during intensive care, we have performed a 13C-metabolic flux analysis (MFA) on PBMCs and granulocytes isolated from two long-term, resuscitated models of combined acute subdural hematoma (ASDH) and HS in pigs with and without cardiovascular comorbidity. Methods Swine underwent resuscitation after 2 h of ASDH and HS up to a maximum of 48 h after HS. Animals received normoxemia (PaO2 = 80 – 120 mmHg) or targeted hyperoxemia (PaO2 = 200 – 250 mmHg for 24 h after treatment initiation, thereafter PaO2 as in the control group). Blood was drawn at time points T1 = after instrumentation, T2 = 24 h post ASDH and HS, and T3 = 48 h post ASDH and HS. PBMCs and granulocytes were isolated from whole blood to perform electron spin resonance spectroscopy, high resolution respirometry and 13C-MFA. For the latter, we utilized a parallel tracer approach with 1,2-13C2 glucose, U-13C glucose, and U-13C glutamine, which covered essential pathways of glucose and glutamine metabolism and supplied redundant data for robust Bayesian estimation. Gas chromatography-mass spectrometry further provided multiple fragments of metabolites which yielded additional labeling information. We obtained precise estimations of the fluxes, their joint credibility intervals, and their relations, and characterized common metabolic patterns with principal component analysis (PCA). Results 13C-MFA indicated a hyperoxia-mediated reduction in tricarboxylic acid (TCA) cycle activity in circulating granulocytes which encompassed fluxes of glutamine uptake, TCA cycle, and oxaloacetate/aspartate supply for biosynthetic processes. We further detected elevated superoxide levels in the swine strain characterized by a hypercholesterolemic phenotype. PCA revealed cell type-specific behavioral patterns of metabolic adaptation in response to ASDH and HS that acted irrespective of swine strains or treatment group. Conclusion In a model of resuscitated porcine ASDH and HS, we saw that ventilation with increased inspiratory O2 concentrations (PaO2 = 200 – 250 mmHg for 24 h after treatment initiation) did not impact mitochondrial respiration of PBMCs or granulocytes. However, Bayesian 13C-MFA results indicated a reduction in TCA cycle activity in granulocytes compared to cells exposed to normoxemia in the same time period. This change in metabolism did not seem to affect granulocytes’ ability to perform phagocytosis or produce superoxide radicals.

The theoretical model used for calculations is visualized in Figure S1, with abbreviations being assigned in Table S3 and the list below.Each node represents a metabolite with its isotope labeling and each arrow a flux contributing to the metabolite mass distribution of the respective node.Their inputoutput flux balances are presented in Table S3.The calculation of theoretical mass distribution, e.g. through condensation of two metabolites, was based on the elementary metabolite unit (EMU) approach (2)(3)(4)(5).To optimize for speed and to ensure efficient calculation, we reduced the model to a set of essential labeling patterns (oxaloacetate, acetyl-CoA), while the labeling on other metabolites were derived from the latter (6).We utilized multiple models: The first calculated ratios from CMDs, the second lactate production rates from measurements of the supernatant, and a third transformed ratios and production rates into absolute flux values.Parameter bounds and priors are listed in Table S5 (first model) and S6 (third model).For more detail about equations for EMU levels and modeling strategies, we refer to the already published supplement of Wolfschmitt et al. (7).

Supplementary Figures
Important tertiary model parameters and their priors.
Figure S1.Visualization of the model used for Bayesian sampling.Glyc: glycolytic flux.Q: flux within the PPP.F: flux within the TCA cycle.L: flux leaving the network.Abbreviations are explained in TableS3.

Figure S2 .
Figure S2.Effects of targeted hyperoxia on circulating granulocytes.Filled symbols indicate data from BMW animals, empty symbols FBM data.NormOx data are shown as blue bars, HyperOx as red bars.Absolute flux values of TCA cycle fluxes in isolated granulocytes including all measurement time points.F3: flux from αketoglutarate to oxaloacetate, F4: oxaloacetate to α-ketoglutarate.F8: glutamate to α-ketoglutarate, LOAA: oxaloacetate/aspartate leaving the modeled network.P-values are indicated in the graphs; Mann-Whitney U test for intergroup differences, Kruskal-Wallis rank sum test for time-related effects.Data are presented as median with IQR.

Figure S4 .
Figure S4.Effects of targeted hyperoxia on E.coli bioparticle-stimulated granulocytes.Filled symbols indicate data from BMW animals, empty symbols FBM data.NormOx data are shown as blue bars, HyperOx as red bars.Data are presented as median with IQR.A) Time-related behaviour of fluxes in HyperOx and NormOx groups in isolated granulocytes after stimulation.Data are either presented as absolute flux values or as the difference between measurement time points.F3: flux from α-ketoglutarate to oxaloacetate, F4: oxaloacetate to αketoglutarate.F8: glutamate to α-ketoglutarate.B) Phagocytic activity of E.coli bioparticle-stimulated granulocytes.Data is either presented as mean fluorescence intensity (nMFI) or fraction of phagocytic granulocytes within the granulocyte subset (6D10 + Phago + ).P-values are indicated in the graphs; Mann-Whitney U test for intergroup differences, Kruskal-Wallis rank sum test for time-related effects.

Figure S4 :
Figure S4: Effect of targeted hyperoxia on O2 •− production.Filled symbols indicate data from BMW animals, empty symbols FBM data.NormOx data are shown as blue bars, HyperOx as red bars.O2 •− production by PBMCs (circles) and granulocytes (triangles) and O2 •− concentration in whole blood (diamonds) as determined by ESR at the indicated measurement time points.Data are either presented as absolute flux values or as the difference between measurement time points.P-values are indicated in the graphs; Mann-Whitney U test for intergroup differences, Kruskal-Wallis rank sum test for time-related effects.Data are presented as median with IQR.

Figure S5 .
Figure S5.Effects of targeted hyperoxia on the mitochondrial respiration of circulating immune cells.Respiration was measured as Routine and ETS in PBMCs (circles) and granulocytes (triangles).Filled symbols indicate data originating from BMW animals, empty symbols FBM data.NormOx data are shown as blue bars, HyperOx as red bars.P-values are indicated in the graphs; Mann-Whitney U test for intergroup differences, Kruskal-Wallis rank sum test for time-related effects.Data are presented as median with IQR.

Supplementary Figures and Tables 2.1 Supplementary Tables Table S1:
Components of mitochondrial respiration medium MiR05.

Table S2 :
End concentrations of important components in the three different RPMI media.

Table S3 :
Input-Output relations at different metabolite pools.

Table S5 :
Important primary model parameters and their priors.Lowercase letters indicate transformed parameters.stdPrior = 0.3