All blood samples used to isolate PBMCs were collected from purebred German Landrace pigs raised at the experimental pig farm (EAS) of the Research Institute for Farm Animal Biology (FBN). The blood samples used to make PBMC pools were obtained from conventionally-reared slaughter pigs (~160 days old), sampled as part of a previous study (17). A total of four pools representing in total 34 pigs of both sexes were used. In addition, individual PBMC samples were obtained from 160 days old pigs reared in conventional conditions (designated TW1 according to German animal husbandry welfare label; n=14 balanced for sex) and from 160 days old pigs reared in enriched conditions (designated TW3 according to German animal husbandry welfare label; n=10 balanced for sex) (for more information on the husbandry conditions please see Supplementary Material ). The Animal Care Committee of the FBN and the State Mecklenburg-Western Pomerania (Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei; LALLF) approved the animal experiment (Approval number 7221.3-1-029/21).

Approximately 50 mL of trunk blood was collected from each pig during exsanguination using 50 mL conical tubes containing 1 mL of 0.5 M EDTA to prevent blood coagulation, and put on wet ice. Porcine PBMCs were isolated through a density gradient centrifugation method using Histopaque-1077 (Sigma-Aldrich, Taufkirchen, Germany) and cryopreserved as previously described (22).

To explore the influence of PBMC metabolism on response to stimulation, the pooled PBMC samples (n=4, set-up as mentioned above, for each combination of pretreatment and treatment) were seeded in 96-well cell culture plates (Faust Lab Science, Klettgau, Germany) at a density of 2.5 × 10⁵ cells per well in standard complete medium consisting of RPMI1640 medium (PAN Biotech, Aidenbach, Germany) supplemented with 10% fetal calf serum (PAN Biotech), 2 mM L-glutamine (PAN Biotech), 100 U/mL penicillin/100 µg/mL streptomycin (PAN Biotech), and cultured in a 37°C, 5% CO₂ environment overnight. The next day, the cells were pretreated for 30 minutes with increasing concentrations of either oligomycin (ThermoFisher, Darmstadt, Germany) - a mitochondrial ATP synthesis inhibitor (0, 1.25, 2.5, 5, and 10 µM), 2-deoxy-D-glucose (2-DG) (ThermoFisher) - a glycolysis inhibitor (0, 0.625, 1.25, 2.5, and 5 mM), or KC7F2 (Bio-Techne, Wiesbaden-Nordenstadt, Germany) - a HIF-1α inhibitor (0, 6.25, 12.5, 25, and 50 µM). Following this, the cells were stimulated for 24 hours with either vehicle, 100 ng/mL LPS (Sigma, Escherichia coli O111: B4 serotype), 100 nM DEX (Sigma), or a combination of both. Finally, after the stimulation, the supernatant was carefully collected by centrifugation at 515 g for 7 minutes and used for bioassays as described below.

For metabolic flux analyses, individual PBMC samples were resuscitated in standard complete medium at 37°C, 5% CO₂ environment overnight. Subsequently, the cells were harvested and reseeded onto a Seahorse XFe96 PDL cell culture microplate (Agilent, Waldbronn, Germany) at 2 × 10⁵ cells/well (the cell number was determined in preliminary experiments as recommended by the manufacturer using either 0.5 × 10⁵, 1 × 10⁵, 1.5 × 10⁵, or 2 × 10⁵ PBMCs/well) in standard complete medium, and cultured in a 37°C, 5% CO₂ environment. Before bioenergetics analysis, the cells were challenged using the same stimulants and concentrations as described above. Details of the metabolic flux analyses are described in section 2.5.

The in vitro experiment for gene expression analysis was performed using 3 × 10⁶ cells/well from each of the pooled PBMC samples in 6-well cell culture plates (Faust Lab Science), essentially as described above. After overnight incubation, the cells received no pretreatment or were pretreated with 12.5 µM of KC7F2 for 30 minutes. Subsequently, the non-pretreated cells were stimulated for 24 hours with either vehicle, 100 ng/mL LPS, 100 nM DEX, or a combination of 100 ng/mL LPS and 100 nM DEX. The pretreated cells were stimulated using either 100 ng/mL LPS alone or in combination with 100 nM DEX. After stimulation, the cells were harvested, lysed in TRI-Reagent (Sigma), and stored at -80°C until further processing for RNA extraction. Transcriptional response of selected genes to treatment was measured as described in section 2.6.

Since the modulators of cell metabolism can influence cell vitality, cytotoxicity assays were performed by measuring extracellular lactate dehydrogenase (LDH) concentration in the collected supernatants using the colorimetric CyQUANT LDH Cytotoxicity Assay (ThermoFisher). The concentration of TNF-α in the supernatant was assessed using a widely used (23) porcine ELISA (Bio-Techne). Both assays were performed as per manufacturer protocol, and the resulting absorbances were measured using a FLUOstar Omega microplate reader (BMG LABTECH, Ortenberg, Germany).

After 24 hours of PBMC stimulation, as described in section 2.3, the metabolic flux was measured on a Seahorse XFe96 instrument (Agilent). The cells were washed twice with either Mito Stress Test Assay Medium (XF RPMI medium, pH 7.4, supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose) or Glycolysis Stress Test Assay Medium (XF RPMI medium, pH 7.4, supplemented with 2 mM glutamine), the final volume (180 µL) of corresponding assay medium was added, and cells were incubated at 37°C without CO₂ for 45 minutes to allow equilibration. Mitochondrial and glycolytic activity were assayed using the Mito Stress Test kit and Glycolysis Stress Test kit, respectively, according to manufacturer instructions (Agilent). Briefly, mitochondrial respiration function was assessed by measuring the oxygen consumption rate (OCR) after consecutively injecting oligomycin (1 µM), FCCP (an electron transport chain accelerator, 2 µM), and antimycin A (a complex III inhibitor, 0.5 µM) along with rotenone (a complex I inhibitor, 0.5 µM). The glycolytic rate was quantified by measuring the extracellular acidification rate (ECAR) after the injections of glucose (10 mM), followed by oligomycin (2 µM), and finally, 2-DG (50 mM). The concentration of the injected compounds was optimized along with cell density in preliminary experiments. Data were normalized to protein content per well obtained using Pierce BCA Protein Assay Kit (ThermoFisher) (13).

Expression of selected genes was measured using quantitative real-time PCR (qPCR) as described previously (24). Briefly, aqueous phase was obtained from cells lysed in TRI-Reagent, and total RNA was purified (including on-column DNase digestion) with the help of the RNA Clean&Concentrator-5 Kit (Zymo Research, Freiburg, Germany). Reverse transcription was performed in a reaction containing 500 ng total RNA, 500 ng random hexamers (Promega, Mannheim, Germany), 500 ng of oligo d(T)13 VN, 40 units of RNasin Plus (Promega), and 200 units of SuperScript III reverse transcriptase (ThermoFisher). The qPCR was carried out using the LightCycler 480 SYBRplus Green I Master kit on a LightCycler 480 System (Roche, Mannheim, Germany). For absolute quantification, a standard curve based on a serial dilution of a gene-specific PCR fragment was generated for each target. The results were normalized using EXOC1 as internal reference gene. Information on primers used for qPCR is summarized in Supplementary Table S1 .

Statistical analysis was performed using GraphPad Prism 9.5.1 (GraphPad Software, San Diego, CA, USA). Statistical tests with a p < 0.05 are reported as significant. The influence of different pretreatments on the response to stimulation (TNF-α concentration) or on metabolic flux parameters was analyzed using two-way ANOVA. The differences between means were tested using the Tukey´s multiple comparison test or the Dunnett´s multiple comparison test (for metabolic flux), respectively. The effects of treatment, sex, and husbandry conditions on metabolic flux in individual samples were tested using a mixed-effects model and the Tukey´s test. Gene expression data obtained by qPCR were log2 transformed before analysis. The transcriptional response to different treatments was analyzed by one-way ANOVA. Preplanned comparisons were performed using the Šidák´s test. Principal component and correlation analysis was performed on the log2 transformed gene expression matrix. Whole transcriptome data were obtained from our previous study (accession number E-MTAB-9808) and re-analyzed based on the Ensemble Gene version 111 as described previously (25).

To examine whether there is an interaction between energy metabolism and immune phenotype of porcine PBMCs, we pretreated the cells in vitro with increasing concentrations of 2-DG, a glycolysis inhibitor, or oligomycin, a mitochondrial ATP synthesis inhibitor, before stimulating the pro- and anti-inflammatory responses by LPS and/or DEX. To reduce the typically considerable individual variation in responses to immune challenges, and to have sufficient material for follow-up experiments, we used pooled PBMC samples. The immune response was monitored based on the concentration of TNF-α, a master regulator of pro-inflammatory cytokines whose secretion is influenced by the glycolytic pathway (26). As expected, there was a significant effect of treatment on TNF-α concentration (p < 0.0001), which increased above the low level of untreated cells (C, Figures 1A, B ) only in the presence of LPS (LPS and LPS+DEX, Figures 1A, B ) (for an overview of the statistical analysis results, see Supplementary Table S2 ). The stimulatory effect of LPS was significantly reduced by co-treatment with DEX (LPS+DEX, Figures 1A, B ). Pretreatment with 2-DG dose-dependently reduced the LPS-induced TNF-α production, but had no significant influence on the anti-inflammatory effect of DEX in the LPS+DEX treatment ( Figure 1A ; Supplementary Table S2A ). In contrast, pretreatment with oligomycin showed no statistically significant influence on the effect of either LPS or DEX on TNF-α production ( Figure 1B ; Supplementary Table S2B ). These results confirmed that the pro-inflammatory response is connected to the glycolytic pathway in porcine PBMCs, similar to humans and rodents. On the other hand, inhibition of mitochondrial ATP production does not appear to have a noticeable influence on the inflammatory state of the porcine PBMCs.

To further characterize the interaction between immune and metabolic activity in PBMCs, we examined their bioenergetics following LPS and/or DEX treatment in vitro using the Seahorse extracellular flux analyzer. For these experiments, we utilized individual samples, which also allowed us to study the impact of sex and husbandry conditions on the immunometabolic phenotype of porcine PBMCs.

To investigate the influence of LPS and DEX on glycolytic metabolism, we conducted the Glycolysis Stress Test. Figure 2 displays real-time curves of extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) (inset) from the test ( Figure 2A ), as well as the derived glycolytic parameters ( Figures 2B–D ), depending on the treatment. There was a significant treatment effect (p < 0.0001), whereby stimulation with LPS alone significantly enhanced glycolysis, glycolytic capacity as well as glycolytic reserve compared to control ( Figures 2B–D ). Co-treatment with DEX significantly mitigated the LPS-induced increase in all glycolytic parameters and reduced them to the level of untreated cells ( Figures 2B, C ) or even below ( Figure 2D ). Moreover, cells treated with DEX alone tended to show the lowest glycolytic parameters ( Figures 2B–D ). We found no statistically significant effect of sex or husbandry conditions on the outcome of the Glycolysis Stress Test ( Supplementary Table S2C ).

Next, we examined how mitochondrial respiration is altered by LPS and DEX using the Mito Stress Test. The results are summarized in a similar way to the Glycolysis Stress Test in Figure 3 . We found a statistically significant effect of treatment (p < 0.0001) on basal and maximal respiration, ATP production, and spare respiratory capacity ( Figures 3B–E ). Sex or husbandry conditions showed no statistically significant effects ( Supplementary Table S2D ). LPS stimulation caused a 1.4-fold increase in spare respiratory capacity compared to the control condition ( Figure 3E ), but neither this nor any other parameter showed a significant difference after LPS treatment compared to control ( Figures 3B–E ). In contrast, treatment with DEX alone significantly decreased all parameters of mitochondrial respiration compared to control, and also significantly reduced maximal respiration and spare respiratory capacity when co-applied with LPS compared to LPS applied alone ( Figures 3C, E ).

Taking the results of both tests together, LPS and DEX influence the metabolism of porcine PBMCs in opposite directions – LPS shifts towards enhanced glycolysis, and DEX induces an overall low energetic state ( Supplementary Material , Supplementary Figure 1 ).

As a next step, we investigated whether there is a connection between the function of HIF-1α and immune and metabolic actions of LPS and DEX in porcine PBMCs. HIF-1α is a transcriptional factor that plays a central role in immunometabolism, primarily by promoting glycolysis and enhancing interleukin 1β production (21). Furthermore, it has recently been suggested that the anti-inflammatory effect of glucocorticoids (DEX) in murine and human macrophages involves downregulation of HIF-1α mRNA and/or protein abundance, and thus counteracting its metabolic actions (19, 20).

To examine the role of HIF-1α in the regulation of the inflammatory state of porcine PBMCs by LPS and DEX, we performed a similar experiment as described above for 2-DG and oligomycin, but this time the cells were pretreated with increasing doses of KC7F2, a HIF-1α inhibitor that downregulates HIF-1α protein synthesis (20). The application of KC7F2 showed a similar influence on TNF-α concentration as 2-DG: the LPS-induced TNF-α production was dose-dependently lowered by KC7F2, but the inhibitory effect of DEX was not significantly affected ( Figure 4 ; Supplementary Table S2E ).

Having established the involvement of HIF-1α in the LPS response, we set out to investigate its role in the metabolic rewiring of porcine PBMCs by LPS and DEX. To this end, we examined the impact of KC7F2 on the bioenergetic effects of LPS and/or DEX treatment in the pooled PBMC samples. We applied the two lower KC7F2 doses (6.25 and 12.5 µM) because the higher doses showed a negative impact on cell vitality as indicated by the LDH cytotoxicity assay ( Supplementary Material , Supplementary Figure 2 ), and as also reported by Clayton et al. (20). Without KC7F2, both the Glycolysis and Mito Stress Tests ( Figures 5 , 6 ; Supplementary Tables S2F, G ) showed essentially the same effects of LPS and DEX as before in individual samples ( Figures 2 , 3 ), demonstrating the robustness of our findings. In the Glycolysis Stress Test, KC7F2 exhibited a treatment-dependent effect. The LPS-induced glycolytic shift was dose-dependently attenuated by KC7F2 addition, with a statistically significant reduction in glycolytic capacity and reserve ( Figures 5A–C ). Surprisingly, in the presence of DEX, KC7F2 tended to increase the glycolytic parameters, particularly in the combined LPS+DEX treatment, with statistically significant increases in glycolysis and glycolytic capacity ( Figures 5A, B ). Mitochondrial respiration, in turn, was consistently reduced by KC7F2 in a dose-dependent manner across treatments, with the 12.5 µM dose significantly reducing essentially all parameters regardless of treatment ( Figures 6A–C ). Our results show that in the LPS-induced pro-inflammatory state, KC7F2 has similar effects on the metabolism of PBMCs as DEX. In contrast, KC7F2 appears to counteract the suppressing effect of DEX on glycolysis, but its suppressing effect on mitochondrial respiration is independent of the treatment, i.e., the immune state. These observations thus support the hypothesis that the metabolic action of DEX in porcine PBMCs may be determined by an interaction between GR and HIF-1α signaling.

In order to gain insight into the mechanisms underlying the antagonistic action of LPS and DEX on the immunometabolic phenotype of porcine PBMCs, we re-analyzed our previously published whole transcriptome data (25) with particular focus on metabolic genes. Specifically, we extracted genes belonging to the glycolysis, TCA, and OXPHOS KEGG pathways to infer the direction of their regulation by LPS and DEX using Ingenuity Pathway Analysis (IPA). The analysis revealed changes in specific pivotal genes (e.g. PFKFB3, Supplementary Table S3 ), but the z-scores were rather low and lacked a clear trend ( Supplementary Table S4 ). It is important to note that in the previous in vitro experiment we used different conditions, particularly an early time point (2 hours), as well as different LPS and DEX doses, which may have contributed to this.

Therefore, and to investigate the function of HIF-1α in more detail, we challenged the four PBMC pools for 24 hours with LPS, DEX, LPS+DEX, LPS+KC7F2, and a combination of all three agents LPS+DEX+KC7F2, and analyzed the mRNA expression of selected candidates from relevant immune, metabolic, and signaling pathways using qPCR. The expression profiles of individual genes and the results of the statistical analysis are presented in Figure 7 and Supplementary Table S5 , respectively. With the exception of LDHA, EGLN1, and PDHA1, the overall treatment effect was significant for all other genes. Pro-inflammatory cytokines (IL1B, TNFA) and canonical LPS-response genes (SOD2, ACOD1) showed the expected significant upregulation by LPS, which was for most of them (except SOD2) effectively suppressed by co-treatment with DEX ( Figure 7A ). Surprisingly, considering the reduced production of TNF-α by KC7F2 pretreatment in the previous experiment (paragraph 3.3), KC7F2 significantly enhanced the LPS-induced expression of TNFA, and non-significantly increased IL1B ( Figure 7A ). The induction of ACOD1 by LPS, a potential target of HIF-1α (27), was decreased by KC7F2 approximately twofold ( Figure 7A ), but this effect did not reach statistical significance. On the other hand, KC7F2 effectively counteracted the suppressive effect of DEX in this group of genes (comparison between LPS+DEX vs. LPS+DEX+KC7F2; Supplementary Table S5 ). We measured the expression of RELB (encoding the RelB subunit of NF-κB) to examine if NF-κB, a central transcription factor in the inflammatory response, may be responsible for the observed expression changes. The profile of RELB was indeed similar and highly correlated to that of the pro-inflammatory cytokines ( Figure 7A ), particularly TNFA ( Supplementary Table S6 ). In addition, we also measured the expression of NFE2L2 encoding NRF2 - a transcription factor activated by itaconate. Itaconate is a metabolite produced by aconitate decarboxylase 1 (encoded by ACOD1) with anti-inflammatory properties mediated partly by NRF2 (28). However, the profile of NFE2L2 differs from ACOD1, and appears to be dependent on HIF-1α, as it was increased by KC7F2 ( Figure 7A ).

The analyzed glycolysis-related genes each showed a distinct response pattern ( Figure 7B ). Whereas the profile of the glucose transporter SLC2A1 (encoding GLUT1) was similar to TNFA, the profile of SLC2A3 (encoding GLUT3) was more similar to ACOD1, which is also reflected by the correlation analysis ( Supplementary Table S6 ). Overall, SLC2A3 featured higher abundance and larger fold changes between treatments than SLC2A1, but only the LPS-induced upregulation reached statistical significance. PFKFB3, encoding an enzyme (6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3) promoting glycolytic flux and pro-inflammatory switch (8), was significantly upregulated by the LPS treatment, but the effect of LPS does not appear to be influenced by KC7F2. Unlike LPS, the effect of DEX was context-dependent; DEX alone tended to upregulate the expression of PFKFB3, but blunted its induction by LPS, which in turn was abolished by KC7F2 addition.

The expression profile of the two TCA cycle-related genes IDH1 and SDHB ( Figure 7C ) indicates the widely described break in the TCA cycle caused by LPS (26). DEX blocked the LPS-induced downregulation of IDH1 as well as SDHB, but this effect was offset by KC7F2. However, the differences between the treatments only reached statistical significance for SDHB ( Supplementary Table S5 ). Interestingly, the transcriptional profiles of IDH1 and SDHB appear to be inversely related to that of TNFA and RELB, as evidenced also by their strong negative correlation ( Supplementary Table S6 ). The accumulation of citrate, caused by the break of the TCA cycle at isocitrate dehydrogenase, is coupled with itaconate production. Itaconate inhibits succinate dehydrogenase and thus contributes to succinate accumulation, which in turn promotes HIF-1α production (28). Despite the changes in expression of metabolic genes indicating altered activity of HIF-1α, none of the treatments significantly influenced mRNA expression of HIF1A and of EGLN1 encoding the prolyl hydroxylase PHD2, which is regulated, i.a., by citrate and succinate accumulation, and destabilizes HIF-1α. However, their profiles indicate a potential feedback regulation of HIF-1α expression at the mRNA level.

There is a bidirectional crosstalk between HIF-1α and GR signaling (29). Indeed, Stifel et al. (19) showed stronger induction of two anti-inflammatory targets of GR - GILZ (official designation is TSC22D3, but GILZ is commonly used) and DUSP1 by DEX in LPS-stimulated macrophages lacking HIF-1α compared to wild-type cells. In line with this finding, we observed enhanced upregulation of GILZ and DUSP1 by KC7F2 when the PBMCs were treated with DEX in the presence of LPS compared to DEX alone ( Figure 7D ). However, KC7F2 actually increased GILZ and DUSP1 expression not only in the presence of DEX (and LPS) but also when stimulated by LPS without DEX. The expression of DUSP1 was also induced by LPS treatment alone. In contrast, induction of the canonical GR target FKBP5 by DEX tended to be reduced by KC7F2. Likewise, the mRNA expression of the glucocorticoid receptor (referred to as GR, official designation is NR3C1), which was downregulated by DEX, tended to be increased by KC7F2. These results suggest, on the one hand, that KC7F2 in fact dampened GR signaling, which would explain the consistent loss of efficacy of DEX in this treatment. On the other hand, this also suggests that the upregulation of GILZ and DUSP1 by KC7F2 occurs by a GR-independent mechanism. In an attempt to uncover this mechanism, we measured the mRNA expression of two potential DUSP1 and GILZ regulators, IL10 and IRAK3, which may be influenced by HIF-1α modulation (11, 30, 31). Whereas the expression profile of IRAK3 was distinct from that of GILZ and DUSP1, the profile of IL10 showed some similarity ( Figure 7A ), particularly with DUSP1, as reflected also by the correlation analysis ( Supplementary Table S6 ). Thus, IL10 could be involved, at least partly, in the GR-independent upregulation of GILZ and DUSP1.

In order to get an overall picture of the relationships between the different functional groups of genes and the metabolic flux, we performed a principal component analysis in addition to the correlation analysis. As shown in Figure 8 , the first two principal components separate the treatment groups quite well ( Figure 8A ). The first principal component (PC1), characterized by high positive loading of features associated with inflammation and glycolysis on the one side, and high negative loading of those representing TCA cycle and respiration on the opposite side, clearly demonstrates the connection between the immune and metabolic state of the cells ( Figure 8B ). The treatment groups containing KC7F2 show the strongest rightward shift ( Figure 8A ), emphasizing enhanced LPS-response by KC7F2. The second principal component (PC2) is defined by the response to DEX, and is outlined mainly by the canonical GR targets. The PCA plot ( Figure 8B ) further highlights the close connection between the glucose transporter SLC2A3 with glycolysis and inflammation on the one hand, and of the TCA cycle genes SDHB and IDH1 with basal respiration and resting state on the other.