To corroborate the presence of metabolic alterations in TNF-induced SIRS, functional enrichment analyses were performed on a liver bulk RNA sequencing (RNASeq) dataset retrieved 18h after injection of a lethal TNF dose in mice (viz. 25 µg), i.e. during the acute phase of TNF-induced SIRS (46). TNF challenge resulted in 1502 upregulated genes (log fold change (LFC) > 1 and p < 0.05) and 1310 downregulated genes (LFC < -1 and p < 0.05) ( Figure 1A ). As expected, Enrichr and Metascape analyses of the upregulated genes demonstrated a clear enrichment related to pro-inflammatory transcription factors and immune responses ( Supplementary Figures S1A–D ). Notably, Enrichr analysis of the downregulated genes identified metabolism as the top hit, clearly suggesting the presence of impaired metabolic processes in TNF-induced SIRS ( Figure 1B ).

Further analysis specifically exhibited a significant downregulation of lipid metabolism, as indicated by the association of downregulated genes with retinoid X receptor (RXR), peroxisome proliferator-activated receptor alpha (PPARα) and liver X receptor (LXR), three transcription factors involved in regulating lipid metabolism ( Figure 1C ), and their enrichment for pathways like metabolism of lipids and fatty acid metabolism ( Figure 1B ). To validate these transcriptome findings, we measured several lipid metabolic parameters 18h after TNF injection ( Figure 1A ). Both plasma FFA and glycerol levels were significantly elevated in TNF-treated mice compared to the PBS controls, suggesting the presence of increased white adipose tissue (WAT) lipolysis and/or reduced FFA oxidation ( Figures 1D, E ). However, no significant decrease in body weight nor inguinal WAT (iWAT) weight of TNF-treated mice could be observed ( Supplementary Figures S1E, F ), implying that the normally observed TNF-induced lipolysis is not markedly activated at that time point or is too limited to be detected by weight of the WAT (23–25). Examination of the relative expression levels of genes involved in FFA oxidation revealed a surprising upregulation of genes specifically associated with fatty acid uptake and mitochondrial import (e.g. Cd36 and Slc25a20), whereas mitochondrial FFA β-oxidation genes (e.g. Acsl1) were significantly downregulated upon TNF challenge ( Figure 1A ). This suggests a potential impairment in FFA oxidation, particularly at the mitochondrial β-oxidation level. We next performed Seahorse analysis to assess the FFA β-oxidation capacity of isolated liver mitochondria from PBS- and TNF-treated mice ( Figure 1G ). Palmitoylcarnitine-driven oxygen consumption rate (OCR) in TNF mitochondria was severely impaired compared to PBS mitochondria ( Figure 1H , Supplementary Figure S1G ). These findings support that TNF-induced SIRS is characterized by compromised fatty acid metabolism, reflected by diminished FFA β-oxidation leading to elevated FFA levels.

To affirm the Enrichr results, we next performed Metascape analysis on the downregulated genes, which interestingly identified monocarboxylic acid metabolic process as the top enriched term ( Figure 1I ). This encompasses all chemical reactions and pathways involving monocarboxylic acids, including subprocesses such as fatty acid metabolism, but also pyruvate and lactate metabolism. This potentially suggests a disruption in carbohydrate metabolism, consistent with previous studies showing the presence of Warburg-like metabolism upon TNF treatment (14, 18, 19). We substantiate this finding, noting significantly diminished relative expression levels of genes involved in pyruvate-driven TCA cycle capacity and demonstrating reduced pyruvate-driven OCR in mitochondria after a lethal TNF dose ( Figures 1A, J, K , Supplementary Figure S1H ). This goes along with the presence of an increased glycolytic flux, suggested by enhanced relative expression levels of glycolytic genes upon TNF challenge ( Figures 1A, L ). This was further supported by the presence of profound hyperlactatemia, hypoglycemia and depleted liver glycogen stores, proving enhanced glucose utilization through glycolysis after a lethal TNF dose ( Figures 1A, M–O ). Additionally, the relative expression levels of genes encoding enzymes in the lactate-mediated GNEO pathway (Ldhb, Pck1 and G6Pc) were significantly diminished upon TNF injection ( Figures 1A, L ). This suggests that the observed severe hypoglycemia and hyperlactatemia are not only driven by increased glucose consumption and lactate production via glycolysis, but also by impaired glucose production and lactate usage through the GNEO pathway.

Overall, TNF-induced SIRS is marked by profound metabolic reprogramming characterized by diminished mitochondrial FFA β-oxidation, increased glycolysis, impaired GNEO and reduced mitochondrial respiration.

To identify a potential, coherent mechanism underlying all metabolic disturbances in TNF-induced SIRS, we focused on the ‘Elsevier Pathway Collection’ in Enrichr for TNF-downregulated genes (1310, LFC < -1 and p < 0.05). Remarkably, this revealed a clear hit for citrullinemia ( Figures 2A, B ). Citrullinemia is a rare autosomal recessive urea cycle disorder caused by mutations in either ASS1 (type 1), encoding argininosuccinate synthetase which is an urea cycle enzyme, or SLC25A13 (type 2, also called citrin deficiency), encoding the aspartate-glutamate carrier (citrin) which results in insufficient aspartate supply to the urea cycle (47). Since citrin is a crucial component of the MAS, citrin deficiency is also characterized by impaired MAS activity, disrupting the cytosolic and mitochondrial NAD⁺ and NADH levels and affecting multiple NAD⁺/NADH dependent metabolic pathways, including glycolysis, GNEO, TCA cycle and FFA β-oxidation ( Figure 2C ), all pathways that we found to be significantly affected in TNF-induced SIRS ( Figure 1 ) (43).

With this, further Enrichr analyses displayed that the TNF-downregulated genes (1310, LFC < -1 and p < 0.05) were also related to key metabolites, including NADP⁺, NAD⁺, ATP and ADP, coenzyme A, and to the mitochondrial matrix as the most affected cellular component ( Figures 2D, E ). This implies diminished NADP⁺ and NAD⁺ levels and impaired mitochondrial energy homeostasis. Combined with the association of the downregulated genes with citrullinemia, this proposes the presence of citrin deficiency leading to MAS inactivity and disrupted NAD⁺ and NADH levels. Indeed, the relative expression levels of the MAS genes were markedly diminished, with Slc25a13 showing the most significant downregulation upon TNF challenge ( Figures 2F, G ). This was associated with significantly diminished liver SLC25A13 and SLC25A11 protein levels ( Figures 2H, I , Supplementary Figures S2A–C ). Moreover, NAD⁺ and NADH levels in total liver lysates were significantly diminished after a lethal TNF dose ( Figures 2J, K ). Since the cytosolic fraction contains significantly more NAD⁺ of the total NAD⁺ pool and the mitochondrial fraction contains significantly more NADH of the total NADH pool, these reductions likely reflect diminished cytosolic NAD⁺ and mitochondrial NADH levels. Hence, these findings demonstrate that TNF-induced SIRS is characterized by impaired MAS activity resulting in a deficit of NAD⁺ and NADH, which eventually could lead to metabolic disturbances.

On the other hand, to identify the mechanism underlying Slc25a13 downregulation in TNF-induced SIRS, we focused on the ‘transcription factor – loss of function (LOF)’ dataset in Enrichr for TNF-downregulated genes (1310, LFC < -1 and p < 0.05). This identified HNF4α, a key nuclear transcription factor, as the top hit ( Figure 2L ), very far beyond other factors. Moreover, the promoter regions of these downregulated genes were significantly more enriched for the HNF4α motif ( Figure 2M ). This clearly shows the presence of HNF4α LOF in TNF-induced SIRS. Moreover, hepatocyte-specific HNF4α knockout (Hnf4α^Liver-i-KO) mice were significantly more sensitive to a TNF dose that was sublethal in Hnf4α^fl/fl (control) mice ( Figure 2N ), demonstrating the crucial role of hepatic HNF4α in surviving TNF-induced SIRS. Interestingly, Slc25a13 is among the genes associated with HNF4α LOF, and GeneCards indicates that its promoter region contains a HNF4α binding site. This was confirmed by HNF4α CHIP-Seq identifying a distinct HNF4α peak in the Slc25a13 promoter region ( Supplementary Figure S2D ). Additionally, exploring the liver mRNA counts of Slc25a13 in a previously published liver Hnf4α^Liver-i-KO dataset (GSE260635; Van Dender et al. (44)) showed significantly reduced liver Slc25a13 mRNA counts in Hnf4α^Liver-i-KO livers compared to Hnf4α^fl/fl mouse livers proving that Slc25a13 expression is HNF4α dependent ( Supplementary Figure S2E ). Hence, this strongly suggests that Slc25a13 downregulation is mediated by HNF4α LOF in TNF-induced SIRS.

To investigate the potential role of MAS dysfunction in metabolic reprogramming and lethality in TNF-induced SIRS, we generated Slc25a13^-/- mice. Full knockout validation was performed by measuring liver SLC25A13 protein levels via western blotting, confirming the total absence of SLC25A13 protein in Slc25a13^-/- mice ( Supplementary Figures S3A, B ). Strikingly, Slc25a13^-/- mice showed acute lethality in response to a TNF dose that was sublethal in Slc25a13^+/+ mice ( Figures 3A, B ). This was accompanied by more pronounced hypothermia and hyperlactatemia over time in Slc25a13^-/- mice after TNF injection ( Supplementary Figures S3C, D ). Moreover, both NAD⁺ and NADH levels in total liver lysates were already significantly reduced in Slc25a13^-/- mice 8h post-TNF injection, a difference that was not yet detected in Slc25a13^+/+ mice ( Figures 3C, D ). Interestingly, no basal differences in liver NAD⁺ and NADH levels are notable between Slc25a13^+/+ and Slc25a13^-/- mice as during homeostasis this is compensated by glycerol-3-phosphate shuttle activity, as evidenced by other studies (48, 49). Overall, this shows that citrin deficiency impairs MAS activity leading to a more rapid NAD⁺ and NADH depletion in TNF-induced SIRS, and more importantly, this emphasizes a crucial role of MAS in resisting TNF lethality.

To examine the mechanisms underlying the increased TNF sensitivity of Slc25a13^-/- mice, RNASeq was performed on livers of Slc25a13^+/+ and Slc25a13^-/- mice 8h after PBS or TNF injection, a timepoint chosen based on the NAD⁺ and NADH effects ( Figure 3A ). Citrin deficiency alone resulted in only four significantly differentially expressed genes (Slc25a13, Map1a, Asns and Zbed), indicating no notable baseline differences between Slc25a13^+/+ and Slc25a13^-/- mice. When plotting the log fold changes (LFCs) of all significantly upregulated or downregulated genes (p < 0.05) in both Slc25a13^+/+ and Slc25a13^-/- mice upon TNF challenge, a clearly more pronounced transcriptional response to TNF was observed in Slc25a13^-/- mice compared to Slc25a13^+/+ mice, especially for the downregulated genes (Upregulated genes: Figure 3E , red line slope = 1.246; downregulated genes: Figure 3F , red line slope = 2.210). We detected 752 and 1942 significantly upregulated genes and 69 and 2227 significantly downregulated genes (LFC > 1 or < -1 and p < 0.05) by TNF in Slc25a13^+/+ and Slc25a13^-/- mice, respectively ( Figure 3G ). Hence, TNF caused a substantially stronger transcriptional response in Slc25a13^-/- mice, with a 2.6- and 33-fold increase in the number of up- and downregulated genes, respectively, compared to Slc25a13^+/+ mice.

Enrichr analysis of the 1227 uniquely upregulated genes by TNF in Slc25a13^-/- mice demonstrated a clear enrichment in pro-inflammatory responses ( Supplementary Figure S3E ). More specifically, heatmap depiction of the LFCs of genes involved in inflammation and endothelial activation demonstrated a stronger upregulation in Slc25a13^-/- mice compared to Slc25a13^+/+ mice 8h post-TNF injection ( Supplementary Figure S3F ). These results imply that citrin deficiency enhances the transcriptional inflammatory response upon TNF stimulation. However, the difference in differential gene expression upon TNF injection was much more pronounced for the downregulated genes compared to the upregulated genes between Slc25a13^+/+ and Slc25a13^-/- mice ( Figures 3E, F ). Therefore, we focused on the uniquely 2164 significantly downregulated genes by TNF in Slc25a13^-/- mice to explore potential mechanisms contributing to their increased TNF sensitivity. Interestingly, Enrichr analysis of these downregulated genes revealed metabolism as the top hit, followed by ‘metabolism of lipids’ and ‘fatty acid metabolism’ ( Figure 3H ). Transcription factor analysis (ChEA 2022) also revealed a clear enrichment for RXR, PPARα and LXR, clearly suggesting a significant downregulation of lipid metabolism ( Supplementary Figure S3G ). We next performed further functional analyses on the 327 downregulated genes specifically associated with metabolism. Both Enrichr and Metascape analyses revealed a clear hit for lipid, fatty acid and monocarboxylic acid metabolic processes, and the mitochondrial membrane as the most affected cellular component ( Figures 3I–K ). Clearly, these functional enrichment analyses 8h after TNF injection in Slc25a13^-/- mice closely resembled the functional enrichment analyses observed 18h after lethal TNF injection in Slc25a13^+/+ mice ( Figure 1 ). When plotting the LFCs of all differentially expressed genes (p < 0.05) in Slc25a13^+/+ upon 18h TNF treatment versus their LFC in Slc25a13^-/- mice upon 8h TNF treatment, a similar transcriptional signature response to TNF was observed ( Supplementary Figure S3H , red line slope = 0.7814 versus black diagonal). Hence, these data strongly suggest that citrin LOF and subsequent MAS inactivity play a crucial role in the development of metabolic dysregulations, i.e. fatty acid and carbohydrate metabolism, and that this contributes to lethality in TNF-induced SIRS.

To substantiate the importance of citrin LOF in the development of a disrupted lipid metabolism in TNF-induced SIRS, we measured several lipid metabolic parameters 8h after TNF challenge in Slc25a13^+/+ and Slc25a13^-/- mice ( Figure 4A ). As expected, TNF injection led to significantly more severe hypothermia in Slc25a13^-/- mice ( Figure 4B ). Both plasma FFA and glycerol levels were significantly increased in TNF-treated Slc25a13^-/- mice compared to PBS controls, a difference that was not (yet) observed in Slc25a13^+/+ mice ( Figures 4C, D ). This indicates the earlier presence of a dysregulated lipid metabolism in TNF-treated Slc25a13^-/- mice. We next determined the potential presence of enhanced lipolysis responsible for the more severe increase in plasma FFA and glycerol levels in TNF-treated Slc25a13^-/- mice by determining loss of fat tissue. The total body weight and iWAT weight of both Slc25a13^+/+ and Slc25a13^-/- mice were not altered upon TNF injection ( Figures 4E, F ). Additionally, gene expression of Lipe (encodes hormone sensitive lipase enzyme involved in the hydrolysis of triglycerides into FFA and glycerol) in iWAT was even significantly reduced in Slc25a13^+/+ upon TNF injection compared to PBS controls. This decrease was also observed in both PBS-treated and TNF-treated Slc25a13^-/- mice ( Figure 4G ). This suggests no clear detection of enhanced lipolysis in TNF-treated Slc25a13^-/- mice, despite other studies showing elevated lipolysis upon TNF treatment (23–25).

Given the more pronounced reduced liver NAD⁺ and NADH levels in TNF-treated Slc25a13^-/- mice, and the dependence of mitochondrial FFA β-oxidation on a proper NAD⁺/NADH ratio, we next explored whether citrin LOF contributes to impaired FFA β-oxidation in TNF-induced SIRS. Heatmap depiction of the LFCs of genes involved in FFA β-oxidation showed a more severe downregulation in Slc25a13^-/- mice compared to Slc25a13^+/+ mice 8h post-TNF injection ( Figure 4H ). We assessed mitochondrial FFA β-oxidation capacity via Seahorse analysis on isolated liver mitochondria of PBS- and TNF-treated Slc25a13^+/+ and Slc25a13^-/- mice ( Figure 4I ). Palmitoylcarnitine-driven OCR was already markedly diminished in Slc25a13^-/- mitochondria compared to Slc25a13^+/+ controls under basal conditions ( Figure 4J , Supplementary Figure S4A ). Following TNF injection, this reduction became significantly more pronounced in Slc25a13^-/- mitochondria, whereas no such decline was observed in Slc25a13^+/+ mitochondria ( Figure 4J , Supplementary Figure S4A ). Hence, citrin deficiency aggravates mitochondrial FFA β-oxidation supposedly leading to the more rapid FFA accumulation in the blood in TNF-induced SIRS. When fatty acid levels are elevated in circulation, the liver stores them in lipid droplets (50). We performed LipidTOX staining to examine the occurrence of ectopic lipid accumulation in livers of PBS-treated and TNF-treated Slc25a13^+/+ and Slc25a13^-/- mice. An increased number of lipid droplets were detected in livers of Slc25a13^-/- mice under basal conditions, likely reflecting their diminished mitochondrial FFA β-oxidation capacity ( Figures 4K, L ). Interestingly, liver lipid droplet count was increased in Slc25a13^+/+ mice upon TNF challenge which was even more augmented in Slc25a13^-/- mice ( Figures 4K, L ). Overall, these data illustrate that citrin LOF-mediated MAS dysfunction impairs mitochondrial β-oxidation, ultimately resulting in FFA accumulation and lipid droplet formation in TNF-induced SIRS.

Since TNF-treated Slc25a13^-/- mice were characterized by more prominent hyperlactatemia ( Supplementary Figure S3D ) and adequate NAD⁺ and NADH levels are crucial for sustaining glycolysis and oxidative phosphorylation, we next investigated the impact of citrin LOF on carbohydrate metabolism in TNF-induced SIRS. Heatmap visualization of LFCs of genes involved in glycolysis demonstrated a stronger upregulation in Slc25a13^-/- mice compared to Slc25a13^+/+ mice 8h post-TNF injection ( Figures 5A, B ). This was accompanied by more severe hypoglycemia and hyperlactatemia in TNF-treated Slc25a13^-/- mice, which was not yet evident in TNF-treated wildtype controls ( Figures 5C, D ) at this 8h post TNF timepoint. Liver glycogen depletion, already observed in TNF-treated Slc25a13^+/+ mice, was significantly more pronounced in TNF-treated Slc25a13^-/- ( Figure 5E ), suggesting accelerated glucose mobilization and glycolytic flux. To further assess glycolytic capacity, a glucose tolerance test was performed upon TNF challenge ( Figure 3F ). Glucose injection resulted in a significant blood glucose increase in TNF-treated Slc25a13^+/+ which was remarkably less pronounced in Slc25a13^-/- mice, indicating quick enhanced glucose utilization ( Figure 5G ). This was accompanied by more severe lactate production in TNF-treated Slc25a13^-/- compared to Slc25a13^+/+ controls ( Figure 5H ), demonstrating the presence of enhanced aerobic glycolysis. Pyruvate-driven OCR was also significantly more diminished in isolated liver mitochondria from Slc25a13^-/- mice compared to Slc25a13^+/+ controls following TNF injection ( Figure 5I , Supplementary Figure S4B ). Together, these findings indicate that citrin absence – and the resulting MAS inactivity - contributes to the metabolic switch toward a Warburg-like phenotype in TNF-induced SIRS. Specifically, the accelerated decline in NAD⁺ levels in Slc25a13^-/- mice likely drives lactate dehydrogenase A (LDHA)-mediated lactate production to sustain NAD⁺ regeneration for glycolytic flux. Simultaneously, diminished mitochondrial NADH availability may impair pyruvate-driven respiration, further enhancing aerobic glycolysis and excessive lactate production.

In addition to exacerbating lactate production, we propose that citrin LOF also impairs lactate-mediated GNEO in TNF-induced SIRS. Heatmap depiction of genes involved in lactate-mediated GNEO (Ldhb, Pck1 and G6Pc) were notably more downregulated in Slc25a13^-/- mice compared to Slc25a13^+/+ mice 8h post-TNF injection ( Figure 5B ). To functionally evaluate lactate clearance, lactate tolerance was assessed in TNF-treated Slc25a13^+/+ and Slc25a13^-/- mice. Lactate administration caused a significant increase in blood lactate in the TNF-treated Slc25a13^+/+ mice. This increase was remarkably more apparent in TNF-treated Slc25a13^-/- mice ( Figure 5J ). Hence, it is suggested that MAS dysfunction due to citrin absence leads to diminished lactate clearance in TNF-induced SIRS by a cytosolic NAD⁺ shortage as the LDHB-catalyzed conversion of pyruvate to lactate is NAD⁺ dependent.

Interestingly, a previous study showed that elevated blood lactate levels, in the absence of efficient clearance, can induce toxicity through lactate-mediated VEGF production, promoting vascular permeability and ultimately resulting in lethal vascular collapse (51). Based on these findings, we hypothesized that the heightened sensitivity to TNF-induced SIRS observed in Slc25a13^-/- mice is driven by lactate-mediated toxicity. Indeed, TNF treatment resulted in significantly higher VEGF levels ( Figure 5K ) and a stronger upregulation of genes related to endothelial activation and damage in Slc25a13^-/- mice compared to wildtype controls ( Supplementary Figure S4C ). This was associated with more vascular permeability for most organs ( Figure 5L , Supplementary Figures S4D–G ). To corroborate these findings, lactate-induced toxicity was assessed in Slc25a13^+/+ and Slc25a13^-/- mice under basal conditions ( Figure 5M ). Slc25a13^-/- mice displayed significantly higher mortality for a sublethal lactate dose compared to wildtype controls ( Figure 5N ). This was correlated with diminished lactate clearance, reflected by significantly elevated blood lactate levels over time in Slc25a13^-/- mice compared to wildtype controls ( Figure 5O ). Overall, this shows that citrin deficiency-induced hyperlactatemia is a key driver for lactate-mediated VEGF production and vascular permeability, thereby contributing to lethality in TNF-induced SIRS.

C57BL6/J (wildtype) mice were purchased from Janvier (Le Genest-St. Isle, France). Mutant Hnf4a^fl/fl mice were generated by Dr. Frank Gonzalez (NIH, Bethesda, USA) and formally called B6.129×1(FVB)-Hnf4atm1.1Gonz/J (93). The mice had been backcrossed into C57BL/6J background and were provided by Dr. Iannis Talianidis (University of Crete, Heraklion (Greece), by courtesy of Dr. Frank Gonzalez, and were under protection of an MTA. AlbCreERT2^Tg/+ mice were kindly provided by Dr. D. Metzger & Dr. P. Chambon (Igbmc, France) (94). Hnf4α^fl/fl;AlbCreERT2^+/+ (Hnf4α^fl/fl) and Hnf4α^fl/fl;AlbCreERT2^Tg/+ (Hnf4α^Liver-i-KO) were generated by crossing as described by Van Dender et al. (44). Hnf4α^fl/fl and Hnf4α^Liver-i-KO were intraperitoneally (IP) injected with 1 mg tamoxifen in a 1:8 ethanol:oil solution for 5 consecutive days to induce hepatic HNF4α depletion, which was observed 3 days after the final tamoxifen injection. Slc25a13^-/- mice, in a C57BL6/J background, were generated in house by the transgenic core facility (IRC-VIB, Ghent University, Belgium) and formally called Slc25a13^em1Clib. A deletion of 83 base pairs was induced in exon 6 via Crispr/Cas. Slc25a13^-/- mice were backcrossed to C57BL/6J (wildtype) mice, purchased from Janvier (Le Genest-St. Isle, France), to establish Slc25a13^-/- and Slc25a13^+/+ mice by Slc25a13^+/- intercrosses. All mice were housed in individually ventilated cages at conventional housing conditions (22°C, 14/10h light/dark cycle, dark phase starting at 9 p.m.) in a specific pathogen-free facility. Food (chow diet consisting of 18% proteins, 4.5% fibers, 4.5% fat and 6.3% ashes, Provimi Kliba SA) and water were given at libitum, unless if otherwise stated. Male and female mice were used at the age of 8–20 weeks. Approval for all experiments was granted by the Institutional Ethics Committee for animal welfare of the Faculty of Sciences, Ghent University, Belgium. The methods were in conformity with the relevant guidelines and regulations.

All injections were administered IP and injection volumes were adapted based on the bodyweight of the mice. Mice were injected with recombinant mouse TNF in a volume of 200 µL/20g. A dose of LD₅₀ or LD₁₀₀ of TNF was determined in advance and varies depending on the mouse strain, animal house, type of experiment and the TNF batch. Doses are depicted in the figure legends. Recombinant mouse TNF was generated in Escherichia coli and purified at our facility in the absence of detectable endotoxin contamination.

During lethality experiments, the rectal body temperature of the mice was frequently monitored, and when it dropped below 28°C, the mice were euthanized via cervical dislocation (humane endpoint). For sampling experiments, blood (for plasma isolation) was collected via cardiac puncture after anesthetizing the mice with ketamine (100 mg/kg) and xylazine (10 mg/kg) mixture. Mice were euthanized via cervical dislocation at the indicated timepoints and organs were isolated. Organs were preserved in RNA later (Life Technologies Europe) or snap-frozen in liquid nitrogen for further analysis.

To assess glucose and lactate tolerance, glucose monohydrate (2 g/kg; Sigma 49159) or sodium-L-lactate (2 g/kg; Sigma 71718) were both dissolved in PBS and administered 8h post TNF injection. To determine lactate toxicity, mice were injected with sodium-L-Lactate (≥ 3 g/kg; dissolved in PBS; Sigma 49159) and lethality was monitored.

Blood lactate and glucose levels were monitored in tail blood using the Lactate Plus meter (NOVA Biomedical) and the OneTouch Verio glucose meter, respectively. Plasma FFA (Abnova KA1667), plasma glycerol (Cayman Chemical 10010755-96), liver glycogen (Abcam ab65620) and liver NAD⁺ and NADH (Abcam ab65348) levels were measured via a colorimetric assay kit. Plasma VEGF levels (Bio-Techne MMV00) were quantified via ELISA. All assays were executed according to the manufacturer’s protocol.

For the detection of SLC25A13 and SLC25A11, total protein was isolated from snap-frozen liver tissue with RIPA lysis buffer containing a protease inhibitor cocktail (Roche). Protein concentration was determined by Bradford assay (Bio-Rad). Protein samples (50 µg protein) mixed with loading dye were separated by electrophoresis on a 8% gradient SDS-polyacrylamide gel, followed by transfer onto a nitrocellulose membrane (pore size 0.45 µm). The membranes were blocked with a ½ dilution of Starting Block/PBST0.1% (Thermo Fisher Scientific) followed by an overnight incubation at 4 °C with primary antibodies against SLC25A13 (1:1000; NBP1-33380, Novus Biologicals) or SLC25A11 (1:1000; PA5-27510; Thermofisher Scientific), and β-ACTIN (1:5000; MA5-15739, Thermo Fisher Scientific) or β-TUBULIN (1:1000; 2146S, Cell Signaling Technology) as loading control. After washing with PBST0.1%, the blots were incubated with Amersham ECL anti-mouse antibody (1:2000, GENA931, GE Healthcare Life Sciences) or Amersham ECL anti-rabbit antibody (1:2000; GEN934, GE Healthcare Life Sciences) for 1h at room temperature. The blots were washed with PBST0.1% and immunoreactive bands were detected and quantified using the WesternBright ECL kit (Advansta Inc.) and an Amersham Imager 600 (GE Healthcare Life Sciences).

White adipose tissue was isolated, snap-frozen and stored at -20°C. The Aurum Total RNA mini Kit (Bio-Rad) was used for total RNA isolation following the manufacturer’s instructions. RNA concentration and quality were measured using the Nanodrop 1000 (Thermo Scientific) and 1000 ng RNA was used for cDNA synthesis with the Sensifast cDNA Synthesis Kit (Bioline). cDNA was 10 time diluted in nuclease-free water. RT-qPCR was executed using the Bioline SensiFAST SYBR No-ROX mix (Bioline) and the Roche LightCycler 480 system (Applied Biosystems). Genorm was used to determine the stability of the housekeeping genes and qPCR data were analyzed with qBase+ software (Biogazelle). Results are depicted as relative expression values normalized to the geometric mean of the housekeeping genes. Used qPCR primers are depicted in Table 1 .

Liver mitochondria were isolated following the protocol in Frezza et al. (101). In short, the whole liver was isolated, washed and minced in ice-cold isolation buffer (1 M sucrose, 0.1 M Tris/MOPS and 0.1 M EGTA/Tris). A Dounce tissue grinder set (Sigma D9063) was used to homogenize the minced liver. After a centrifugation step (600 g, 10min, 4°C), the supernatans was collected and recentrifuged at 7000 g, 10 min, 4°C. The resuspended pellet in ice-cold isolation buffer was recentrifuged at 7000 g, 10 min, 4 °C to collect purified mitochondria. The total mitochondrial protein content was measured via a Bradford protein assay (Bio-Rad).

The OCR of isolated liver mitochondria was assessed using a Seahorse Bioscience XF96 Analyzer (Agilent), following the protocol by Luso et al. (102). Mitochondria (10 µg) in mitochondrial assay solution (MAS, 70 mM sucrose, 220 mM mannitol, 10 mM KH₂PO₄, 5mM MgCl₂, 2 mM HEPES, 1 mM EGTA and 0.2% fatty acid-free BSA) were delivered/well in a 96-well Seahorse microplate (Agilent). After centrifugation (20 min, 2000 g, 4°C), prewarmed MAS with specific respiratory substrates (40 µM palmitoylcarnitine and 0.5 mM malate or 10 mM pyruvate) were administered to the mitochondria, followed by a 10 min incubation at 37°C. The OCR flux was determined over time after the sequential injection of ADP (40 mM), oligomycin (25 µg/ml), FCCP (40 µM) and rotenone (20 µM) with antimycin A (AA; 40 µM).

Isolated liver tissue was fixed in antigenfix (DiaPath) at 4 °C for 1-2h. Liver tissues were washed with PBS, followed by an overnight incubation in 34% sucrose at 4 °C, and were mounted with O.C.T. compound (Tissue-Tek). Cryostat sections of 20 µm thickness were rehydrated in PBS and were blocked for 30 min at room temperature (RT) with blocking buffer (2% BSA, 1% fetal calf serum, 1% goat serum, in 0.5% saponin). The liver sections were incubated with a primary antibody mix (LipidTOX Deep Red (1:400; Life Technologies Europe B.V.); Acti-stain 488 Phalloidin (1:150; Cytoskeleton Inc.)) for 2h at RT. The slides were washed with PBS and the nuclear staining (DAPI (1:1000, Sigma-Aldrich N.V.)) was added for 15 min at RT. After washing with PBS and bidi to remove residual salts, the slides were mounted with PVA including DABCO. For each liver section, Z-stacks of 8 areas were imaged using a spinning disk confocal microscope (Zeiss), with a Plan-Apochromat 40x/1.4 oil DIC (UV) VIS-IR M27 objective lens at a pixel size of 0.275 µm and at optimal Z-resolution (240 mm). Z-stacks were analyzed in Arivis and the amount of lipid droplets relative to the tissue volume was quantified.

8h after TNF injection, mice were intravenously injected with FITC-dextran 4 kDa (25 mg/kg, TDB labs 60842-46-8). One hour post-injection, mice were transcardially perfused with 0.2% EDTA in PBS after anesthetizing the mice with ketamine (100 mg/kg) and xylazine (10 mg/kg) mixture. Organs were isolated, cut in small pieces and overnight incubated (37°C; shaking) in 100% formamide (Sigma) for FITC-labeled dextran extraction. After centrifugation (15 min, 14–000 rpm), the supernatans was 1/20 diluted in PBS and fluorescence was measured with a FLUOstar OMEGA plate reader (BMG Labtech, Germany). Fluorescence was normalized to the respective Slc25a13^+/+ - PBS control per tissue.

Statistical and graphical data analysis were performed using GraphPad Prism software (version 10.3.1). qPCR data was log-transformed to obtain normal distribution. An unpaired Student’s t-test was performed to compare two group means. A two-way ANOVA with a Tukey’s multiple comparisons test was used for experimental setups with a second variable. Kaplan Meier survival curves were compared using a Log-Rank (Mantel-Cox) test or one-side chi-squared test.

Statistical significance was defined by a P-value of < 0.05. **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05, ns: not significant. All data are expressed as means ± standard error of the means (SEM). N represents the number of biological replicates used in the experiments. Group sizes were chosen based on prior experience. Statistical details can be found in the figure legends.