Male C57BL/6J mice were ordered from Janvier (Le Genest-St. Isle, France). Mutant Hnf4a^fl/fl mice were generated by Dr. Frank Gonzalez (NIH, Bethesda, USA) and formally called B6.129X1(FVB)-Hnf4a^tm1.1Gonz /J (41). Exons 4 and 5 were flanked by loxP sites using ES cell technology. 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. Subsequently, we crossed these mice with AlbCreERT2^Tg/+ mice, which were kindly provided by Dr. D. Metzger & Dr. P. Chambon (Igbmc, France) (42). The humanized liver mice were generated by Prof. Philip Meuleman (UGhent, Belgium) as described in (43). Briefly, the mice were generated by transplanting homozygous Alb-uPA^+/+-SCID mice (which suffer from spontaneous and chronic death of hepatocytes) with 0.7 x 10⁶ cryopreserved primary human hepatocytes (donor C342, Lonza), via intrasplenic injection. The human albumin concentration in plasma was determined 6 weeks after transplantation by ELISA (Bethyl Laboratories, Montgomery, Texas, United States) and was used as a marker of liver chimerism.

All mice were housed in individually ventilated cages (IVC) at standard housing conditions (22°C, 14/10h light/dark cycle) with food (chow diet consisting of 18% proteins, 4.5% fibers, 4.5% fat, 6.3% ashes, Provimi Kliba SA) and water ad libitum. The IVC were kept in a specific pathogen free facility. Both male and female offspring was used between 8 and 20 weeks. All experiments were approved by the institutional ethics committee for animal welfare of the Faculty of Sciences, Ghent University, Belgium (EC2024-059). The methods were carried out in accordance with the relevant guidelines and regulations.

Mice were subjected to CLP operation (44). Briefly, mice were anesthetized with isoflurane, followed by a 1 cm incision in the abdominal skin. The caecum was ligated for 75% and punctured twice with a 21-gauche needle, which allows the cecal content to leak into the abdomen and cause a systemic infection. The abdominal musculature and skin were closed with simple running sutures and metallic clips, respectively. Sham mice underwent the same procedure without ligating and puncturing the caecum. During lethality studies, mice were injected intraperitoneally (IP) with broad-spectrum antibiotics (25 mg/kg ceftriaxone and 12.5 mg/kg metronidazole) in phosphate buffered saline (PBS) 8h and 24h after CLP. Rectal BT was measured 3 times a day and mice were euthanized when their BT dropped below 28°C (= human endpoint).

To induce hepatic HNF4α depletion, Hnf4a^fl/fl;AlbCreERT2^Tg/+ and Hnf4a^fl/fl;AlbCreERT2^+/+ mice were IP injected with 1 mg tamoxifen in a 1:8 ethanol:oil solution for five consecutive days. HNF4α protein was completely depleted three days after the final injection (9). To investigate CAR activity, the CAR agonist TCPOBOP (12.5 mg/kg, dissolved in 50% corn oil and 50% DMSO) was administered via IP injection in non-operated wild-type mice, in Hnf4a^Liver-i-KO mice three days after the last tamoxifen injection, and in septic mice 6h or 24h after sham or CLP surgery. For survival studies, TCPOBOP (3 mg/kg, dissolved in 90% corn oil and 10% DMSO) was administered via IP injection or oral gavage for four consecutive days. To assess the functional consequences of CAR loss-of-function, the CAR inhibitor CINPA1 (12.5 mg/kg, dissolved in 90% corn oil and 10% DMSO) was administered via IP injection for four consecutive days. Additionally, the HNF4α agonist NCT was administered via IP injection for 7 days, with 1 injection during the first 3 three days, 2 injections during the following three days, and 1 injection on the day of CLP, at a dose of 200 mg/kg, dissolved in DMSO. Mice co-treated with NCT and CINPA1 also received 10 doses of NCT (200 mg/kg, dissolved in 2-hydroxypropyl-β-cyclodextrin (25 mg/mL in PBS)), including 2 additional injections during the first 2 days after CLP. They also received 4 doses of CINPA1 (12.5 mg/kg, dissolved in 90% corn oil and 10% DMSO), with 1 additional injection given 1 day post-CLP. NCT was synthetized by SynBioC (Dr. Bart Roman).

The liver was isolated and stored in RNAlater (Life Technologies Europe), after which RNA was extracted using the Aurum Total RNA Mini kit (Bio-Rad) according to the manufacturer’s-protocol. RNA concentration was measured, and quality was assessed using the Nanodrop 1000 (Thermo Scientific).

Total protein was isolated from snap-frozen livers using RIPA lysis buffer, supplemented with a protease inhibitor cocktail (Roche). Protein concentration was determined using the Bradford assay. Protein samples containing 30 µg of protein, along with loading dye, were separated by electrophoresis on an 8% sodium dodecyl sulfate (SDS)-polyacrylamide gel and transferred to a nitrocellulose membrane (0.45 µm pore size) by blotting. The membrane was blocked with a ½ dilution of Starting Block/PBST 0,1% (Thermo Fisher Scientific) and incubated overnight at 4°C with primary antibodies against CAR (1:667; ab186869, Abcam), (Ser/Thr/Tyr)-phospho (1:1000; ab15556, Abcam), CYP2B10 (1:1000; AB9916, Sigma-Aldrich), HNF4α (1:1000; PP-H1415-0C, R&D systems), and β-actin (1:5000; BA3R, Thermo Fisher Scientific) as an internal control. After washing with PBST 0,1%, the membrane was incubated for 1h at room temperature (RT) with either Amersham ECL anti-mouse antibody (1:2000; GENA931, GE Healthcare Life Sciences), Amersham ECL anti-rabbit antibody (1:2000; GEN934, GE Healthcare Life Sciences), or VeriBlot (1:2000; ab131366, Abcam) for the immunoprecipitation (IP) samples. The membrane was then washed again with PBST 0,1%, and immunoreactive bands were visualized and quantified using an Amersham Imager 600 (GE Healthcare Life Sciences) and the WesternBright Quantum kit (Advansta).

Since no antibody for phospho-CAR is available, we immunoprecipitated CAR from total liver lysates and used a general anti-phospo antibody for Western blot analysis. Total protein was isolated from 50 mg of snap-frozen liver tissue using RIPA lysis buffer, supplemented with a protease inhibitor cocktail (Roche) and phosphatase inhibitors (20 mM glycerophosphate, 1 mM NaF, 1 mM Na₃VaO₄). Meanwhile, Dynabeads™ protein A (Thermo Fisher Scientific) were washed three times with RIPA, then blocked with RIPA containing BSA for 10 min at RT. To reduce nonspecific binding to the beads, protein lysates were precleared by incubation with the blocked beads for 1h at 4°C. Protein concentration was then measured using the Bradford assay, and input samples were collected. The precleared lysates (containing 300 µg of protein) were incubated with primary antibodies against CAR (4.8 µg/IP sample; ab186869, Abcam) or rabbit GFP-IgG (4.8 µg/IP sample; AB3080, Sigma-Aldrich) as an internal control for 1h at 4°C. We worked with pooled IgG samples, meaning that all replicates for each condition were combined for incubation with the IgG antibody. The IP samples were then incubated with the BSA-blocked beads for 2h at 4°C. Afterward, bead-antibody-protein complexes were washed 5 times with RIPA, then lysed in RIPA with loading dye for 5 min at 95°C.

ChIP was performed on 50 mg of snapfrozen liver tissue, homogenized in PBS, and crosslinked with PBS + 2% formaldehyde for 20 min at RT. The crosslinking was quenched with 1M glycine (final concentration: 0.125M) for 10 min at RT. Liver homogenates were washed twice with PBS and then lysed with a buffer containing 0.1% SDS and 1% Triton-X-100. The liver lysates were sonicated (Bioruptor, Diagenode) at 4°C for 30 cycles (30s ON, 30s OFF) at high settings to yield 200–800 bp DNA fragments. Cell debris was removed by centrifugation at 10,000g for 5 min at 4°C, and input samples were collected. During sonication, nProtein G Sepharose™ 4 Fast Flow beads (Cytiva) were washed three times with incubation buffer containing 0.15% SDS and 1% Triton-X-100, then blocked with incubation buffer containing BSA for at least 2h at 4°C. Afterwards, sonicated samples were incubated with primary antibodies against HNF4α (5 µg/IP sample; PP-H1415-0C, Bio-Techne) or mouse IgG (5 µg/IP sample; sc-2025, SantaCruz) as an internal control for 2h at 4°C. These samples were then incubated with the BSA-blocked beads overnight at 4°C. The day after, bead-antibody-DNA complexes were washed twice with wash buffer containing 0.1% SDS, 0.1% sodium-deoxycholate, 1% Triton-X-100, 0.15M NaCl, 1 mM EDTA, and 20 mM HEPES; once with wash buffer containing 0.1% SDS, 0.1% sodium-deoxycholate, 1% Triton-X-100, 0.5M NaCl, 1 mM EDTA, and 20 mM HEPES; once with wash buffer containing 0.25M LiCl, 0.5% sodium-deoxycholate, 0.5% NP-40, 0.15M NaCl, 1 mM EDTA, and 20 mM HEPES; and twice with wash buffer containing 1 mM EDTA and 20 mM HEPES at 4°C for 5 min. Chromatin was eluted by incubating the IP samples with elution buffer (1% SDS, 0.1M NaHCO₃) 20 min at RT. Both IP and input samples were first incubated with proteinase K and RNase A for 2 hours at 56°C, then overnight at 65°C for reverse crosslinking. The DNA was purified in 50 µL elution buffer according to the manufacturer’s protocol by using the QIAquick PCR purification kit (Qiagen).

qPCR was performed on 2 µL of purified DNA from IP and input samples using the Light Cycler 480 system (Roche) and Sensifast Bioline Mix (Bio-Line) to assess HNF4α binding at the Nr1i3 promoter. qPCRs were carried out in duplicate. Sequences of the qPCR primers used: Forward primer (5’-3’) AACCAACACTTCTCGGGCA and Reverse primer (5’-3’) GCCTCTAGGTATCCTCGGTG. Results after immunoprecipitation were subtracted from the input and expressed as relative enrichment to the negative IgG control.

Nuclear protein lysates were prepared from 40 mg of snap-frozen liver tissue. First, the liver was homogenized in PBS supplemented with a protease inhibitor cocktail (Roche). Next, nuclei were isolated using a lysis buffer containing 0.5% NP-40 and 0.25% Triton-X100, with a protease inhibitor cocktail (Roche). The isolated nuclei were then washed and lysed in a buffer containing 0.5% N-lauroylsarcosine and 0.1% Na-Deoxycholate, again supplemented with a protease inhibitor cocktail (Roche). Finally, the samples were sonicated using a PIXUL Multisample sonicator (Active Motif) for 15 minutes with default settings (Pulse 50 cycles, PRF 1 kHz, Burst Rate 20 Hz). Protein concentration was determined using the Bradford assay.

Analysis of organ damage markers, ALT, AST and LDH, in plasma was performed at the University Hospital of Ghent. Plasma SAA protein levels (ICL lab; E-90SAA), plasma IL6 levels (Thermofisher; 88-7064-88), hepatic CAR protein levels (Antibodies-online; ABIN6969015), and the DNA binding activity of CAR (Abbexa; abx596559) in liver nuclear protein lysates (containing 50 µg of protein) were measured by ELISA according to the manufacturer’s instructions.

Plasma IL6 levels and qPCR data were log-transformed to obtain normal distribution. Normality was assessed using a normality test in Prism 9.0 (GraphPad Software, Inc), which was also used primarily for plotting. When comparing two group means, an unpaired student’s t-test was used. If the standard deviation (SD) between groups differed significantly, Welch’s correction was applied. For comparisons involving more than two groups, P-values were analyzed using a one-way ANOVA with the uncorrected Fisher’s LSD multiple comparisons test. Two-way ANOVA with either the uncorrected Fisher’s LSD or Tukey’s multiple comparisons test was used for analyses involving a second variable. Kaplan-Meier survival curves were compared using the log-rank test. XY-plots were analyzed using simple linear regression. P-values of < 0.05 were considered statistically significant. ∗∗∗∗P ≤ 0.0001, ∗∗∗P ≤ 0.001, ∗∗P ≤ 0.01, ∗P ≤ 0.05, ns: not significant. n represents the number of biological replicates used to generate the data, determined based on previous experiments. Each experiment shown in the main figures was repeated at least twice. Data are represented as dot plots with bars indicating the mean ± SD or SEM. Additional statistical details can be found in the figures and/or figure legends. To mitigate the influence of sex- or age-related factors, an equal number of animals of each sex and similar age were assigned to each treatment group. Treatments were also distributed across different cages. Data analysis was performed blindly to avoid subjective bias.

To investigate CAR activity in sepsis on a genome-wide level, sham and CLP-treated mice were injected with CAR agonist TCPOBOP or vehicle (corn oil + DMSO), 6h post-surgery ( Figure 1A ), when CLP mice display hypothermia and significant HNF4α, PPARα and GR activity problems (7–9). According to public RNA sequencing (RNA-Seq) data (GSE40120), injection of TCPOBOP in CAR KO mice showed no significant changes in gene expression, indicating its specificity for CAR. Given that CAR is mainly expressed in the liver, we isolated the livers of these mice 3h post-injection and performed bulk RNA-Seq. Only 36 genes were responsive to TCPOBOP in sham (padj < 0.05) ( Figure 1B ). When plotting the log2 fold change (LFC) of these 36 genes in sham mice upon TCPOBOP injection versus their LFC in CLP mice with TCPOBOP, a clear reduction in response to TCPOBOP was observed in the CLP group ( Figure 1C ). On average, only 20.7% of the gene expression changes observed in sham mice upon TCPOBOP remained in CLP mice upon TCPOBOP. For the 29 genes that were upregulated by TCPOBOP in sham, their LFC was significantly less high in CLP relative to sham ( Figure 1D ). Of these 29 genes, only 4 remained significantly upregulated by TCPOBOP in CLP (highlighted in red). Additionally, 2 genes were specifically upregulated and 1 was downregulated (padj < 0.05) by TCPOBOP in CLP and not in sham ( Figure 1B ). All 39 TCPOBOP responsive genes are shown in the heatmap ( Figure 1E ): some of the genes upregulated by TCPOBOP in sham showed a trend towards downregulation in vehicle CLP, except for Cyp2c29, which remained unchanged, and were either not induced or less induced by TCPOBOP in CLP (group 1). Others, which were upregulated by TCPOBOP in sham, showed a trend towards upregulation in vehicle CLP but exhibited no further increase in response to TCPOBOP in CLP, except for Gdf15 (group 2). Many of the genes in the first group are involved in hepatic metabolism. For example, Cyp2b10, Cyp2c29, and Cyp2c53 play a role in drug metabolism; Pdp2 mediates the dephosphorylation of pyruvate dehydrogenase, which is required for its activation and the subsequent conversion of pyruvate into acetyl-CoA; and Rarb regulates the transcription of genes involved in lipid metabolism. On the other hand, genes in the second group are involved in cell cycle regulation (Fam83d, Tubb6, Uck2), transport of ascorbic acid (Slc23a2), inorganic phosphate (Slc34a2), amino acids (Slc7a6), and cell communication and/or migration (Arhgef7, Sh3bp2, Synj, Tns1). Examples of genes that illustrate TCPOBOP resistance in CLP; we validated 3 genes by RT-qPCR ( Figure 1F ).

To investigate the mechanism responsible for the reduced CAR transcriptional activity during sepsis, we first examined whether Nr1i3 mRNA expression was affected. At 9h after CLP, hepatic Nr1i3 mRNA levels were indeed significantly decreased ( Figure 2A ). This decrease was consistent at multiple timepoints post-CLP, including 6h, 8h, 10h and 24h ( Supplementary Figure S1 ). Nr1i3 transcription is primarily regulated by HNF4α, PPARα and PGC1α (14, 55, 56). We previously reported that HNF4α progressively loses its function in the liver during sepsis, driven by reduced chromatin binding, leading to decreased H3K27 acetylation (H3K27ac) and, to a lesser extent, altered chromatin accessibility. The observed decreases in acetylation and chromatin accessibility may consequently contribute to the downregulation of various nuclear receptors, as shown for PPARα (9). We found reduced binding of HNF4α at the Nr1i3 promoter 8h after CLP, which was associated with a decrease in H3K27ac and H3K4 methylation (H3K4me3), but not with changes in chromatin accessibility ( Figures 2B, C ). Since H3K27ac marks active enhancers, while H3K4me3 marks active promoters, the losses of these marks suggest reduced enhancer and promoter activity, which is consistent with the decreased transcription of Nr1i3 observed in both CLP mice and mice with hepatic HNF4α depletion, three days after tamoxifen administration (Hnf4a^Liver-i-KO) ( Figure 2D ). The PPARα agonist GW7647 upregulates Nr1i3 mRNA expression in sham mice but not in CLP mice, indicating that both HNF4α and PPARα may contribute to the downregulation of Nr1i3 in CLP ( Figure 2E ). This lack of response to GW7647 has been previously attributed to the downregulation of Ppara in the liver 6h after CLP (8). Nevertheless, CAR protein was significantly downregulated in Hnf4a^Liver-i-KO mice, both in sham and CLP, compared to Hnf4a^flfl mice, but not in Hnf4a^flfl mice 8h after CLP ( Figure 2F ), but was decreased at later stages of sepsis, between 24h and 96h after CLP ( Figure 2G ). Since the decline in CAR transcriptional activity in response to TCPOBOP precedes it quantitative decline at the protein level, several mechanisms impacting its activity during sepsis might be at play.

To confirm the dependency of CAR activity on HNF4α, Hnf4a^flfl and Hnf4a^Liver-i-KO mice were injected with TCPOBOP or vehicle three days after tamoxifen administration, when livers were depleted of HNF4α, and livers were harvested 3h post-injection for RT-qPCR analysis ( Figures 2H, I ). None of the CAR target genes tested showed any response to TCPOBOP in the KO, indicating fully impaired CAR activity ( Figure 2J ). However, similar to CLP, Akr1b7 was basally upregulated in Hnf4a^Liver-i-KO mice. Given the loss of CAR protein in these mice, these findings suggest that transcription factors other than CAR may be responsible for Akr1b7 upregulation in CLP as a compensatory mechanism. PXR is a strong candidate, as it shares many target genes and binding partners with CAR and also upregulates Akr1b7 upon administration of its agonist PCN (57, 58). In conclusion, these findings suggest that the loss of HNF4α and PPARα function during sepsis disrupts Nr1i3 transcription, thereby impairing CAR activity. However, this reduction only partially accounts for the diminished transcriptional activity of CAR during sepsis, indicating that additional mechanisms are also involved.

Because total CAR protein levels remained unchanged 8h after CLP, we further investigated alternative mechanisms for the clearly impaired CAR functional decline, including its nuclear localization, phosphorylation status and chromatin binding. Consistent with previous reports, CAR could already be detected in the nucleus without ligand stimulation by MS analysis ( Figure 3A ). However, no differences were observed in the liver between sham and CLP at 8h. Despite its nuclear presence, phenobarbital-mediated activation of CAR is regulated by protein phosphorylation, as the PP2A-specific phosphatase inhibitor okadaic acid reduces CAR activity following phenobarbital treatment (36). To investigate CAR phosphorylation in CLP upon TCPOBOP, CAR was immunoprecipitated from livers of sham or CLP mice injected with TCPOBOP or vehicle 6h post-surgery, and isolated 3h post-injection, followed by Western blot using a general anti-phospho antibody ( Supplementary Figure S2A ). However, CAR phosphorylation was even reduced by TCPOBOP in CLP ( Supplementary Figures S2B–D ).

We then assessed the DNA binding activity of CAR. Due to the lack of suitable CAR antibodies for ChIP-Seq, we used an in vitro CAR DNA binding assay. In this assay, plates containing DNA with the consensus binding site of CAR were incubated with nuclear lysates prepared from liver samples collected 3h after TCPOBOP/vehicle injection (which was administered 6h after sham or CLP). CAR DNA binding was significantly decreased in CLP conditions, both basally and following TCPOBOP treatment ( Figure 3B ). Since the DNA probe remained unchanged in these studies, these data suggest that a posttranslational modification of CAR or an altered interaction may be responsible for the reduced DNA binding in sepsis. However, the chromatin region 100 bp upstream of the promoters of CAR target genes upregulated by TCPOBOP in sham (29 genes), as identified in our RNA-Seq analysis from Figure 1 , was less accessible in CLP, based on our ATAC-Seq data from livers 8h after CLP ( Figure 3C ) (9). Moreover, we overlapped this ATAC-Seq data with public CAR ChIP-Seq data from livers of YFP-mCAR transgenic mice collected 2h after the last TCPOBOP injection (59). To validate the ChIP-Seq data, 19 of the 29 genes responsive to TCPOBOP in sham were found near a peak. From this overlap, we found that chromatin sites to which CAR can bind were less accessible in CLP ( Figure 3D ). This suggests that a more closed chromatin signature in CLP might also contribute to the decreased CAR DNA binding and, consequently, its reduced transcriptional activity. CAR DNA binding sites were enriched for the Hnf4a motif, indicating a potential interaction between CAR and HNF4α ( Figure 3E ). Therefore, we also overlapped the CAR ChIP-Seq data with public ATAC-Seq data from livers of Hnf4a^Liver-i-KO mice (13) and found that CAR DNA binding sites were also less accessible in the KO, suggesting that the presence of HNF4α is crucial for maintaining an accessible chromatin state at CAR enhancers ( Figure 3F ). Consequently, these findings indicate that the loss of HNF4α function during sepsis not only decreases Nr1i3 mRNA expression, but also reduces chromatin accessibility at CAR binding sites, thereby contributing to its impaired DNA binding in CLP.

The CAR target genes identified in our RNA-Seq analysis provide initial insights into the biological processes affected by CAR during sepsis. However, due to the limited transcriptional response to a single TCPOBOP injection in our experiments, we expanded our analysis to other CAR dependent genes that have been identified by other researchers and are available as public RNA-Seq data (58). These authors identified 2100 CAR dependent genes (993 induced and 1107 reduced) based on repetitive TCPOBOP injections. Their approach, involving naive mice and multiple injections, yielded a stronger transcriptional response in the liver. We focused on their 993 TCPOBOP-induced (CAR dependent) genes and overlapped these with genes downregulated in the liver 24h after CLP at both the RNA and protein level ( Figure 4A ). The genes were identified using public RNA-Seq data 24h after sham or CLP (7), and the proteins were identified through MS proteomics on livers collected at the same timepoint. This timepoint was chosen because metabolic changes during sepsis are most pronounced at this stage, characterized by hypothermia, hypoglycemia, hyperlactatemia, and hepatic steatosis (7, 8). The 25 genes/proteins within the overlap ( Figure 4A ) are primarily involved in the metabolism of monocarboxylic acids ( Figure 4B ). These 25 are listed as mRNAs and proteins in Figures 4C, D , respectively, with their levels normalized to 100% in the sham mice. For example, Cyp2c29 and Cyp2c50 catalyze the epoxidation of arachidonic acid, Ephx1 facilitates the hydrolysis of epoxyeicosanoids, Aldh1a1 is involved in the metabolism of retinoic acid, and Lipa catalyzes the hydrolysis of triglycerides. Furthermore, Cyp4f15 participates in the catabolism of leukotriene B4, Thnsl2 contributes to the biosynthesis of 2-oxobutyrate, Akr1c19 plays a role in the metabolism of jasmonic acid, and Fmo5 contributes to the oxidation of short-chain fatty acids. Among the 25 genes, Slco1a4, Akr1c19, and Cyp4f15 are the most downregulated at the RNA level, with Slco1a4 being involved in the transport of bile acids and other monocarboxylic acids ( Figure 4C ). On the other hand, MBL2, CYP4F15, and SLCO1A4 are most strongly downregulated at the protein level, with MBL2 mediating complement activation ( Figure 4D ). We also identified TCPOBOP-reduced genes -genes that are significantly downregulated by TCPOBOP in the liver (1107 in total)- and overlapped them with genes upregulated in the liver 24h after CLP at both the RNA and protein level ( Figure 5A ). These 50 genes are primarily involved in the APR, including Stat3, Fn1, Itih4, Ilr1n, Cd14, Agt, Myd88, S100a9, Egfr, Dpp9, Saa1, Saa2, Saa3, Il1r1, Serpina3m, Serpina3n ( Figure 5B ). Among the 50 genes within the overlap, Cd14, Tifa, and Saa3 are most strongly upregulated at the RNA level, with Cd14 being mainly expressed by macrophages and interacting with LPS, while Tifa contributes to the activation of NFκB and MAPK signaling during inflammation ( Figure 5C ). On the other hand, CD14, TIFA, and SERPINA3M are most strongly upregulated at the protein level ( Figure 5D ). To conclude, CAR loss-of-function in sepsis disrupts hepatic metabolism, particularly monocarboxylic acid metabolism, while inducing the hepatic APR. Acute phase proteins (APPs) have anti-inflammatory and tissue repairing activities and are therefore essential in hepatic regeneration. In that regard, CAR loss-of-function in sepsis can be considered a way to promote the APR at the expense of certain liver metabolic functions.

Cyp2b10 is a key biomarker of CAR signaling due to the presence of a phenobarbital-responsive enhancer module within its promoter, which makes it highly responsive to TCPOBOP induction (60). It plays a role in drug and lipid metabolism in the liver and is downregulated 24h after CLP at the RNA level ( Figure 4E ). However, since CYP2B10 could not be detected by MS, we verified its protein levels 24h after CLP using Western blot analysis, which confirmed its downregulation ( Figures 4F, G ). Cyp2b10 mRNA expression also correlated with body temperature in mice subjected to CLP, suggesting that it could be a potential biomarker in sepsis ( Figure 4H ).

To further investigate the functional consequences of hepatic CAR loss-of-function in sepsis, mice were treated for 4 days with the CAR inhibitor CINPA1 or with vehicle ( Figure 6A ). Importantly, qPCR analysis of several CAR target genes, selected from the 993 TCPOBOP-induced genes from Figure 4 , showed that CINPA1 functions as only a partial inhibitor ( Supplementary Figure S3 ). As mentioned earlier, this might be explained by the fact that CAR and PXR share many binding partners and target genes, suggesting that they could compensate for each other’s activities upon inhibition, as reflected by the upregulation of several CAR target genes upon CLP or HNF4α depletion in the liver (57, 58). However, such compensation has not been described in the literature yet. Nevertheless, CINPA1 treatment significantly increased mortality in CLP compared to vehicle treatment, indicating that CAR plays a crucial role in the response to sepsis ( Figure 6B ). To further assess the reduced survival associated with CINPA1, liver and blood samples were collected 24h after sham or CLP surgery ( Figure 6C ). This timepoint effectively reflects the increased lethality of CINPA1 in CLP, as the body temperature was already further decreased ( Figure 6D ). Of the 25 TCPOBOP-induced genes previously shown to be downregulated 24h after CLP compared to sham, at both the RNA and protein level ( Figure 4A ), 6 representative genes were selected and showed further suppression by CINPA1 in CLP ( Figure 6E ). Using samples from untreated sham and CLP mice, we found that the expression of these genes correlated with body temperature at 24h after CLP ( Figure 6F ). As mentioned earlier, many of these genes are involved in monocarboxylic acid metabolism, except for Aco1 and Idh2, which play roles in the TCA cycle by catalyzing the isomerization of citrate to isocitrate and the oxidative decarboxylation of isocitrate to α-ketoglutarate, respectively.

IL6 is a proinflammatory cytokine and a marker of systemic inflammation, with an additional role in the priming phase of liver regeneration. Following partial hepatectomy, IL6 is rapidly produced by Kupffer cells, endothelial cells and hepatocytes, inducing APP production and stimulating hepatocyte proliferation (61, 62). IL6-deficient mice show impaired liver regeneration, characterized by liver necrosis and failure (63). Plasma IL6 levels were further increased with CINPA1 in CLP, suggesting enhanced systemic inflammation upon CAR inhibition and supporting a role for CAR in suppressing immune activation ( Figure 7A ). Consistent with this increase, plasma SAA protein levels were further elevated by CINPA1 in CLP ( Figure 7B ). SAA is one of the best validated APPs in mice. This result further confirms the involvement of CAR in suppressing the hepatic APR and suggests that its loss-of-function during sepsis contributes to APR induction, possibly indirectly by enhancing plasma IL6 levels. Although APPs promote liver regeneration by supporting tissue repair and limiting infection, CAR also regulates hepatocyte proliferation, with several proliferation markers upregulated following its activation by TCPOBOP. Consequently, CAR KO mice show increased mortality after 2/3^th partial hepatectomy due to impaired hepatocyte proliferation (20). The proliferation markers Mki67 and Pcna remained downregulated in CLP mice treated with CINPA1, with Mki67 already suppressed in sham mice upon CAR inhibition ( Figure 7C ). Furthermore, plasma alanine aminotransferase (ALT) levels, a biomarker of impaired liver function and hepatocellular injury (64), remained upregulated in CLP mice following CINPA1 treatment ( Figure 7D ). Both features suggest that, although APPs are upregulated following CAR inhibition in CLP, liver regeneration may not be improved. CINPA1 also led to increased organ damage in CLP, as indicated by elevated lactate dehydrogenase (LDH) levels in the blood ( Figure 7E ). However, ALT and aspartate aminotransferase (AST) levels were not further increased by CINPA1, suggesting that the increased damage is not specifically related to the liver, heart, or muscle, but is more generalized. To summarize, CINPA1 worsens sepsis survival, associated with the downregulation of genes involved in monocarboxylic acid metabolism, lipid metabolism, and the TCA cycle; increased systemic inflammation; and greater organ damage. Although the hepatic APR was further upregulated, no additional signs of improved liver regeneration were observed.

Since inhibiting endogenous CAR activity with CINPA1 sensitized mice to sepsis, we aimed to determine whether CAR activation by TCPOBOP could provide protection. Because CAR is unresponsive to TCPOBOP post-CLP ( Figure 1 ), mice were administered the CAR agonist TCPOBOP or vehicle for 4 days prior to CLP onset, with the final dose given on the day of CLP. Mortality was monitored for 10 days thereafter ( Figure 8A ). However, TCPOBOP did not improve sepsis survival, either when administered intraperitoneally ( Figure 8B ) or via oral gavage ( Figure 8D ), despite a clear increase in CAR target genes with both methods ( Figures 8C , E ). Taken together, our data suggest that CAR contributes to the host response during sepsis. Although ligand-mediated activation does not result in protection, the increased sensitivity observed upon CAR inhibition underscores its physiological significance.

NCT has previously been shown to specifically target HNF4α and provide protection in models of NAFLD and peritoneal sepsis (9, 65, 66). When NCT was administered for 7 days prior to CLP, including a final injection at the day of CLP, HNF4α chromatin binding was enhanced at target genes Hes6, Apoa2, Ugt2b1 and Ppara. NCT further increased hepatic Ppara expression, reduced plasma FFA levels and hepatic lipid accumulation, decreased systemic inflammation and improved the hepatic APR in CLP (9). To investigate whether NCT also improves Nr1i3 transcription in sepsis, mice were administered NCT or DMSO for 7 days, subjected to sham or CLP surgery, and the liver was isolated 24h thereafter ( Figure 9A ). NCT prevented the reduction of HNF4α binding to the Nr1i3 promoter following sepsis, as assessed by ChIP-qPCR, thereby enhancing its transcription in the liver during CLP ( Figures 9B, C ). We also evaluated CAR activity by injecting TCPOBOP or vehicle 24h post-surgery into mice that had received NCT or DMSO for 7 days and isolated the liver 3h thereafter ( Figure 9D ). CAR activation by TCPOBOP was clearly impaired 24h after CLP, as observed earlier at the 6h timepoint, and was improved by NCT ( Figure 9E ). From this, we conclude that NCT increases both Nr1i3 mRNA expression and CAR transcriptional activity in the liver during sepsis. To investigate whether NCT-mediated protection in CLP is dependent on CAR, mice were co-treated with NCT and CINPA1. NCT was administered daily for 7 days prior to CLP, including a final injection on the day of surgery and again 2 days post-CLP ( Figure 9F ). CINPA1 was given for 4 days prior to CLP, with the fourth injection on the day of surgery and an additional injection 1 day post-CLP. The protective effect of NCT was partially lost by CAR inhibition, indicating that CAR contributes to its protective role in CLP ( Figure 9G ).

In this study, both Nr1i3 mRNA expression and CAR transcriptional activity were found to be reduced in the liver 6h and 24h after CLP in mice. To translate these findings, we analyzed the expression of Nr1i3 and several CAR target genes, identified either in this study or in the literature, in RNA-Seq data from septic pigs and by qPCR in livers of humanized-liver mice 24h after sham or CLP. Septic shock in pigs was induced by intraperitoneal administration of autologous feces, resulting in fecal peritonitis. Liver samples were collected 18h after sepsis onset for bulk RNA-Seq (67). Like in mouse sepsis, Nr1i3 and all CAR target genes investigated were downregulated in pig sepsis, although the downregulation was generally less pronounced in pigs than in mice, except for NR1I3 and the CYP genes ( Figures 10A, B ). To generate the humanized-liver mice, severe combined immunodeficiency mice carrying the Alb-uPA transgene (uPA^+/+) were engrafted with human hepatocytes to replace the killed mouse hepatocytes ( Figure 10C ) (43). The repopulation of human hepatocytes in the liver was determined by the ratio of human to mouse albumin in the blood of the mice, which was 30% in our case. The 70% mouse hepatocytes express the uPA transgene but derive from constant proliferation and progenitor cell differentiation. Also in these mice, NR1I3 and the majority of CAR target genes were downregulated in CLP, in both mouse and human hepatocytes ( Figure 10D ).