Abstract

Background & Aims

   The constitutive androstane receptor (CAR) is a nuclear receptor that
   binds diverse xenobiotics and whose activation leads to the modulation
   of the expression of target genes involved in xenobiotic detoxification
   and energy metabolism. Although CAR hepatic activity is considered to
   be higher in women than in men, its sex-dependent response to an acute
   pharmacological activation has seldom been investigated.

Methods

   The hepatic transcriptome, plasma markers, and hepatic metabolome, were
   analysed in Car^+/+ and Car^-/- male and female mice treated either
   with the CAR-specific agonist
   1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP) or with vehicle.

Results

   Although 90% of TCPOBOP-sensitive genes were modulated in a
   sex-independent manner, the remaining 10% showed almost exclusive
   female liver specificity. These female-specific CAR-sensitive genes
   were mainly involved in xenobiotic metabolism, inflammation, and
   extracellular matrix organisation. CAR activation also induced higher
   hepatic oxidative stress and hepatocyte cytolysis in females than in
   males. Hepatic expression of flavin monooxygenase 3 (Fmo3) was almost
   abolished and was associated with a decrease in hepatic
   trimethylamine-N-oxide (TMAO) concentration in TCPOBOP-treated females.
   In line with a potential role in the control of TMAO homeostasis, CAR
   activation decreased platelet hyper-responsiveness in female mice
   supplemented with dietary choline.

Conclusions

   More than 10% of CAR-sensitive genes are sex-specific and influence
   hepatic and systemic responses such as platelet aggregation. CAR
   activation may be an important mechanism of sexually-dimorphic
   drug-induced liver injury.

Impact and implications

   CAR is activated by many drugs and pollutants. Its pharmacological
   activation had a stronger impact on hepatic gene expression and
   metabolism in females than in males, and had a specific impact on liver
   toxicity and trimethylamine metabolism. Sexual dimorphism should be
   considered when testing and/or prescribing xenobiotics known to
   activate CAR.

   Keywords: Sexual dimorphism, Hepatic xenobiotic metabolism, Lipoprotein
   metabolism, Platelet aggregation, Trimethylamine-N-oxide

Graphical abstract

   [69]graphic file with name ga1.jpg
   [70]Open in a new tab

Highlights

     * •
       Sex-independent vs. sex-biased CAR-dependent hepatic pathways were
       defined.
     * •
       CAR activation influenced circulating lipoproteins in a
       sex-independent manner.
     * •
       CAR activation influenced liver metabolism in female mice more than
       in males.
     * •
       CAR activation inhibited TMAO synthesis, which could influence
       platelet aggregation.

Introduction

   The liver is a highly sexually dimorphic organ. There is increasing
   evidence for sexually-dimorphic regulation of xenobiotic clearance,
   responses to drugs and drug-induced liver injury. We postulated that
   the mechanisms underlying such dimorphism may involve the constitutive
   androstane receptor (CAR, NR1I3). CAR is a liver-enriched member of the
   nuclear receptor superfamily that controls ligand-dependent regulation
   of gene expression. Upon ligand-binding, CAR translocates to the
   nucleus, heterodimerises with retinoid X receptor α, and binds the
   xenobiotic response element located on DNA, upstream of the promoter
   sequences of its target genes. CAR was first described as a xenobiotic
   receptor that recognises a wide variety of drugs, foods, and
   environmental pollutants.[71]^1^,[72]^2 Later studies then unveiled
   that CAR can also be activated by endobiotics such as bilirubin, bile
   acids, and steroid hormones.[73]^3^,[74]^4 Upon activation, CAR
   regulates the expression of critical enzymes involved in phase I, II,
   and III xenobiotic metabolism pathways,[75]^5^,[76]^6 thereby playing a
   central role in xenobiotic detoxification and clearance. Moreover, CAR
   is involved in glucose and lipid homeostasis, although its exact role
   in hepatic metabolism remains controversial.[77]^7

   Hepatic expression of Car and its main target genes Cyp2b9 and Cyp2b10
   is higher in female mice than in males.[78]^8^,[79]^9 Moreover,
   treatment with nonylphenol, a moderate CAR activator, induced
   expression of cytochromes P450 (CYPs) more strongly in the female mouse
   liver than in males.[80]^10 Similarly, treatment with
   1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), the prototypical
   pharmacological agonist of murine CAR (mCar),[81]^11 increased liver
   proliferation in female more than in male mice.[82]^12 Our previous
   study showed that deletion of CAR expression had a stronger impact on
   female hepatic gene expression than on males; however, CAR-deleted
   males developed spontaneous steatosis during ageing while females did
   not.[83]^13 Therefore, CAR is thought to impact rodent liver gene
   expression in a sex-dependent manner, with higher CAR activity and
   higher sensitivity to CAR activation in females. Interestingly, in
   humans, the expression and activity of CYP2B6, the prototypical target
   gene for human CAR (hCAR), were higher in the liver of women compared
   with men, indicating that sexual dimorphism of CAR activity was also
   transposable to humans.[84]^8 Despite the recognition of its sexually
   dimorphic activity, in vivo studies conducted so far on both male and
   female mice have focused on the impact of CAR activation on the
   regulation of CYPs,[85]^14 on genes involved in cell cycle and
   hepatocarcinogenesis,[86]^15^,[87]^16 or on long non-coding RNA.[88]^17
   A genome-wide comparison of the effects of CAR activation in male and
   female mice is still lacking.

   In this study, we used hepatic microarray and metabolomics analysis of
   wild-type (Car^+/+) and whole-body knockout littermate (Car^-/-) male
   and female mice treated with either Corn oil (CO, vehicle) or TCPOBOP
   to elucidate the potential sex-dependent impact of CAR activation.

Materials and methods

Animal models

   In vivo studies were performed in a conventional laboratory animal room
   following the European Union Guidelines for Use and Care of Laboratory
   Animals. This project was approved by an independent ethics committee
   (CEEA-86 Toxcométhique, authorisation number 2019123014045837). The
   animals were treated humanely with due consideration to the alleviation
   of distress and discomfort. All mice were housed at 22 °C ± 2 °C on a
   12-h light (ZT0–ZT12) 12-h dark (ZT12–ZT24) cycle, where ZT indicates
   Zeitgeber time; ZT0 is defined as the time when the lights are turned
   on. Animals were allowed free access to food (Teklad Global 18% Protein
   Rodent Diet) and tap water. Car^-/- mice (backcrossed on the C57BL/6J)
   were engineered by Wei et al.[89]^1 and were bred for 15 years in our
   animal facility. Car^+/- mice were co-bred and gave birth to true
   Car^+/+ and Car^-/- littermate mice, which were then separated by sex
   and genotype at 4 weeks of age and were randomly allocated to the
   different experimental groups. Nine-week-old male and female mice
   included in TCPOBOP groups received a daily intraperitoneal injection
   of 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP, Sigma Aldrich)
   diluted in CO used as vehicle at 3 mg/kg for 4 days, whereas CO mice
   received CO only (Sigma Aldrich). One cage housing n = 6 mice per group
   was used. At ZT16 (6 h after the last TCPOBOP injection), mice were
   anaesthetised with isoflurane and xylazine (2%, 2 mg/kg) then blood
   from the vena cava was collected into lithium heparin-coated tubes (BD
   Microtainer, Franklin Lake, NJ, USA). Plasma was prepared by
   centrifugation (1,500×g, 15 min, 4 °C) and stored at −80 °C. Following
   euthanasia by cervical dislocation, the liver and perigonadal white
   adipose tissue were removed, weighted and snap-frozen in liquid
   nitrogen, and then stored at −80 °C until use.

   To confirm Fmo3 regulation upon TCPOBOP treatment an independent
   experiment was conducted using the same experimental groups but with
   different timing of TCPOBOP treatment leading to the same total dose of
   TCPOBOP: intraperitoneal injection either with TCPOBOP diluted in CO at
   3 mg/kg every 2 days for 10 days or with CO, at ZT10. Mice were
   euthanised by cervical dislocation at ZT8 and liver was removed,
   weighed and snap-frozen in liquid nitrogen and stored at −80 °C.

   For thrombus formation analysis, another set of 7-week-old C57BL/6J
   mice were purchased from Charles River laboratories, acclimatised for 2
   weeks, then randomly allocated to the different experimental groups:
   Female Corn Oil (F CO, n = 10), female TCPOBOP (F TCPOBOP, n = 10),
   male Corn Oil (M CO, n = 10), male TCPOBOP (M TCPOBOP, n = 10) (two
   cages per group). Then, mice were fed with 1% choline-enriched diet
   (D13090101, Research Diets) for 10 days. Mice included in the TCPOBOP
   groups received an intraperitoneal injection of TCPOBOP diluted in CO
   at 3 mg/kg for the last 4 days of diet, whereas CO mice received CO
   only, at ZT0. Between ZT3 and ZT8 whole blood was drawn from the
   inferior vena cava of anaesthetised mice (100 mg/kg ketamine, 10 mg/kg
   xylazine) into heparin sodium (10 IU/ml) and mice were euthanised by
   cervical dislocation and liver was removed, weighed and snap-frozen in
   liquid nitrogen and stored at −80 °C until use.

Gene expression

   Gene expression profiles were performed at the GeT-TRiX facility
   (GénoToul, Génopole Toulouse Midi-Pyrénées, Toulouse, France) using
   Agilent Sureprint G3 Mouse GE v2 microarrays (8 × 60K, design 074809)
   following the manufacturer’s instructions. Data acquisition and
   statistical analyses were performed as previously described.[90]^18 A
   correction for multiple testing was applied using the
   Benjamini-Hochberg procedure to control the false discovery rate (FDR).
   Probes with fold change (FC) ≥1.5 and FDR ≤0.05 were considered to be
   differentially expressed between conditions. The enrichment of Gene
   Ontology (GO) biological processes was evaluated using
   Metascape.[91]^19 Data are available in NCBI’s Gene Expression Omnibus
   and are accessible through GEO Series accession number [92]GSE228554.

   For real-time quantitative polymerase chain reaction (RT-qPCR), 2 μg
   RNA samples were reverse-transcribed using the High-Capacity cDNA
   Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA).
   [93]Table S1 presents the SYBR Green assay primers. Amplifications were
   performed using an ABI Prism 7300 Real-Time PCR System (Applied
   Biosystems). RT-qPCR data were normalised to TATA-box-binding protein
   (Tbp) mRNA levels.

Proton nuclear magnetic resonance (^1H-NMR)-based metabolomics

   Plasma samples and liver polar extracts were prepared and analysed
   using ^1H-NMR-based metabolomics. All spectra were obtained on a Bruker
   DRX-600-Avance NMR spectrometer (Bruker) on the AXIOM metabolomics
   platform (MetaToul). Details on experimental procedures, data
   pre-treatment and statistical analysis were described
   previously.[94]^18 Parameters of the final discriminating orthogonal
   projection on latent structure-discriminant analysis (O-PLS-DA) are
   indicated in the figure legends. To identify metabolites responsible
   for discrimination between the groups, the O-PLS-DA correlation
   coefficients (r^2) were calculated for each variable. Correlation
   coefficients above the threshold defined by Pearson’s critical
   correlation coefficient (p <0.05; |r| >0.7; for n = 6 per group) were
   considered significant. For illustration purposes, the area under the
   curve of several signals of interest was integrated and significance
   tested with two-way ANOVA as described below. For metabolite
   identification ^1H–^13C heteronuclear single quantum coherence (HSQC)
   spectra were obtained on one representative sample for each biological
   matrix. Lists of metabolites measured are presented in [95]Table S2 and
   S3.

Multi-omics analyses

   Bidirectional correlations between plasma metabolites and hepatic
   transcripts were investigated using N-integration discriminant analysis
   with DIABLO, an algorithm that aims to identify a highly correlated
   multi-omics signature discriminating several experimental groups using
   the R package Mixomics v6.10.9.[96]^20 We used two components in the
   models, and for the estimation of model parameters, the
   cross-validation procedure method was used. For the correlation
   networks, only correlations with a Spearman’s rank correlation
   coefficient >0.96 were plotted.

Analyses of plasma markers

   Alanine aminotransferase (ALT), phosphatase alkaline (ALP), total
   cholesterol, high-density lipoprotein (HDL-cholesterol), triglycerides,
   and free fatty acids were determined using an ABX Pentra 400
   biochemical analyser (Anexplo facility, Toulouse, France). Blood
   glucose levels were measured from the vena cava with an AccuCheck
   Performa glucometer (Roche Diagnostics).

Trimethylamine-N-oxide targeted LC-MS/MS measurement

   For trimethylamine-N-oxide (TMAO) extraction and measurement, see
   details in the supplementary methods.

Publicly available datasets and databases

   Four independent gene expression datasets were found on the Gene
   Expression Omnibus data repository accessed in September 2019.
   [97]GSE149229 compared hepatic transcriptome of humanised CAR mice
   (hCAR) fed a control diet or a phenobarbital (PB)-enriched diet.
   [98]GSE98666 compared hepatic transcriptome of hCAR mice treated with
   CO, TCPOBOP or
   6-(4-chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde-O-(3,4-dic
   hlorobenzyl)oxime (CITCO). [99]GSE149228 and [100]GSE57056 compared the
   hepatic transcriptome of chimeric mice with most human hepatocytes fed
   a control diet or a PB-enriched diet. Values for Fmo3 gene expression
   were calculated using the GEO2R tool for microarray data and using
   GREIN[101]^21 for RNA sequencing data.

Thrombi formation under flow

   Biochips microcapillaries (Vena8Fluoro+, Cellix) were coated with a
   collagen fibril suspension (50 μg/ml) and saturated with a solution of
   0.5% bovine serum albumin in phosphate-buffered saline without
   Ca^2+/Mg^2+. Mouse blood was transferred into heparin (10 IU/ml), and
   DIOC6 (2 μM) was used to label platelets in whole blood. Using a
   syringe pump (PHD-2000; Harvard Apparatus) to apply a negative
   pressure, labelled blood was then perfused through a microcapillary for
   indicated time at a wall shear rate of 1,500 s^-1 (67.5 dynes/cm^2).
   Thrombus formation was visualised with an ×40 oil immersion objective
   for both fluorescent and transmitted light microscopy; the light source
   was provided by Colibri (Zeiss) and was recorded in real-time (one
   frame every 20 s). Thrombi volumes were calculated by thresholding of
   the surface covered by thrombi on a slice of Z-stack images using
   IMARIS software.[102]^22

Other statistical analyses

   All univariate statistical analyses were performed using GraphPad Prism
   v.9 (GraphPad Software, San Diego, CA, USA). Outliers were identified
   using the ROUT method. The Kolmogorov–Smirnov test of normality was
   applied to all data. Two-way ANOVA was performed within each genotype
   using sex (male or female) and treatment (CO or TCPOBOP) as fitting
   factors for the models and a p value representing interactions were
   reported. If p[sex∗treatment] was significant, Sidak’s multiple
   comparisons test was used as a post-hoc test to determine which group
   differed from its appropriate control, otherwise p values representing
   the main effects from the ANOVA model (namely sex or treatment) were
   reported. For platelet aggregation measures, a mixed-effect model was
   fitted using time and treatment as fixed effects. A p <0.05 was
   considered significant. Results are given as the mean ± SEM.

Results

Analysis of hepatic transcripts revealed a majority of sex-independent
CAR-target genes upon TCPOBOP treatment

   To investigate the potential sex-dependent consequences of CAR
   activation, we treated Car^+/+ and Car^-/- male and female mice with
   TCPOBOP ([103]Fig. 1A). TCPOBOP did not affect the total body,
   decreased perigonadal white adipose tissue and increased liver weights
   in Car^+/+ male and female mice ([104]Fig. 1B). We confirmed Car
   deletion and observed increased expression of Cyp2b10 and Cyp2c55, two
   prototypical CAR target genes, in TCPOBOP-treated Car^+/+ males and
   females. Cyp2c55 induction by TCPOBOP was significantly higher in
   females ([105]Fig. 1C). We characterised the impact of CAR activation
   by TCPOBOP on hepatic gene expression using microarrays. TCPOBOP was a
   very specific CAR agonist since there was no significantly regulated
   gene in Car^-/- male mice, and only two in Car^-/- female mice
   ([106]Fig. S1), we thus continued our analysis using Car^+/+ mice only.
   Principal component analysis (PCA) of the entire expression data set
   from Car^+/+ mice revealed that individuals clustered separately
   according to treatment on the first axis and to sex on the second axis
   ([107]Fig. 1D), illustrating a major effect of CAR activation on the
   liver transcripts. Comparison of the lists of differentially expressed
   genes (DEGs) upon TCPOBOP treatment in males vs. females Car^+/+
   demonstrated that more than half of TCPOBOP-modulated genes were common
   to males and females ([108]Fig. 1E and [109]Fig. S2). Using all 4663
   TCPOBOP-sensitive genes, we highlighted two gene clusters that
   exhibited sex-independent responses ([110]Fig. 1F). Genes upregulated
   upon TCPOBOP (cluster 1, 2,366 genes) were mainly involved in the ‘cell
   cycle’ (p = 10^-46) and ‘cellular response to xenobiotic stimulus’ (p =
   10^-25, [111]Fig. 1F and [112]Table S4B), whereas downregulated genes
   (cluster 2, 2297 genes) were enriched for ‘carboxylic acid catabolic
   process’ (p = 10^-17) and ‘regulation of lipid metabolic process’ (p =
   10^-14, [113]Fig. 1F and [114]Table S4D). We focused on well-described
   CAR target genes involved in the cell cycle ([115]Fig. 1G) and
   carbohydrate and lipid metabolism ([116]Fig. 1H) and confirmed that
   regulation upon TCPOBOP treatment was similar in both males and females
   ([117]Table S5).

Fig. 1.

   [118]Fig. 1
   [119]Open in a new tab

   Modulation of classical CAR-controlled pathways is sex-independent.

   (A) Experimental design. (B) Body weight, perigonadal WAT, and liver
   weights. (C) Hepatic gene expression. Results are given as the mean ±
   SEM. ∗Treatment effect, ^#sex effect. ∗ or ^#p <0.05, ∗∗ or ^##p <0.01,
   ∗∗∗ or ^###p <0.001 (two-way ANOVA). (D) PCA of the whole liver
   transcriptomic dataset in Car^+/+ mice. (E) Venn diagram representing
   the number of genes significantly modulated by TCPOBOP in the liver of
   Car^+/+ male and female mice (p[adj]<0.05 and fold-change >1.5). (F)
   Hierarchical clustering and pathway enrichment analysis using the 4,663
   genes significantly regulated upon TCPOBOP in males or females (p[adj]
   <0.05 and fold-change >1.5). (G) Heatmaps representing the
   log(fold-change) of gene expression between TCPOBOP- and CO-treated
   Car^+/+ males and females and (H) for genes involved in cell cycle. and
   carbohydrate and lipid metabolism. CAR, constitutive androstane
   receptor; CO, corn oil; PCA, principal component analysis; TCPOBOP,
   1,4-bis[2-(3,5-dichloropyridyloxy)] benzene; WAT, white adipose tissue.

Role of CAR in sexually-dimorphic regulation of hepatic gene expression in
response to TCPOBOP

   TCPOBOP impacted a much higher number of genes in the liver of females
   vs. males (∼40% more DEGs in females vs. males, [120]Fig. 1E).
   Accordingly, PCA of the microarray data projected on the second and
   third principal components showed a distinct clustering of TCPOBOP- and
   CO-treated females, whereas males from both groups were merged
   ([121]Fig. 2A). To identify genes with sex-dependent regulation upon
   TCPOBOP, we next focused on the DEGs with a significant interaction
   between sex and treatment ([122]Table S6). These 486 sex-specific DEGs
   clustered within four distinct expression profiles ([123]Fig. 2B and
   C). Female-specific upregulated genes (cluster 1, 215 genes) were
   involved in xenobiotic metabolism (p = 10^-12) and extracellular matrix
   remodelling (p = 10^-6) and contained genes encoding for collagens
   (Col4a1), extracellular matrix degrading metalloproteinases (Mmp12,
   Mmp13) and proteinases involved in the processing of procollagens
   (Adamst2, Adamst4, Adamst14, and Adamst15), whereas female-specific
   downregulated genes (cluster 3, 170 genes) were involved in phase I
   xenobiotic metabolism (p = 10^-8) and flavin monooxygenase
   (FMO)-dependent oxidations (p = 10^-6) ([124]Fig. 2D–F). As indicated
   above, male-specific CAR target genes were fewer. The male-specific
   upregulated genes (cluster 4, 37 genes) were involved in ‘steroid
   metabolism’ (p = 10^-5.5), whereas no significantly enriched metabolic
   pathway was found using the male-specific downregulated genes (cluster
   2, 64 genes) ([125]Fig. 2D and E and [126]Table S6). Overall, this
   analysis provided evidence that 10% of the TCPOBOP-sensitive genes were
   regulated in a sex-dependent manner ([127]Fig. 2G).

Fig. 2.

   [128]Fig. 2
   [129]Open in a new tab

   Identification of sex-dependent CAR-sensitive genes.

   (A) PCA of the liver transcriptomic dataset in Car^+/+mice. (B)
   Hierarchical clustering using the 486 genes with significant
   p[sex∗treatment](p[adj] <0.05 and fold-change>1.5). (C) Gene expression
   profiles in each cluster. (D) Pathway enrichment analysis. (E) Heatmaps
   representing the log(fold-change) of gene expression between TCPOBOP-
   and CO-treated Car^+/+ males and females for the top 10 genes in each
   cluster. (F) Hepatic gene expression of genes involved in inflammation
   and fibrosis derived from microarray data. ∗Treatment effect, ^#sex
   effect. ∗ or^#p[adj] <0.05, ∗∗ or^##p[adj] <0.01, ∗∗∗ or^###p[adj]
   <0.001 (linear models). (G) Genes modulated by TCPOBOP treatment in a
   sex-dependent and -independent way. CAR, constitutive androstane
   receptor; CO, corn oil; PCA, principal component analysis; TCPOBOP,
   1,4-bis[2-(3,5-dichloropyridyloxy)] benzene.

CAR regulated plasma lipoprotein metabolism in a sex-independent manner

   We next explored the systemic consequences of CAR activation using
   plasma metabolomics. PCA analysis of the whole plasma metabolic
   profiles showed a separation of male vs. female mice on the first
   principal component, illustrating a constitutive sexual dimorphism in
   plasma metabolite levels, while TCPOBOP-treated Car^+/+ were
   discriminated from Car^-/- mice and from Car^+/+ CO-treated mice on the
   second principal component, illustrating a significant effect of
   TCPOBOP on plasma metabolites ([130]Fig. 3A). As seen for hepatic
   transcripts, there was no significant differences between metabolic
   profiles from TCPOBOP vs. CO-treated Car^-/- mice ([131]Fig. S3). In
   Car^+/+ males, TCPOBOP treatment decreased glucose, increased lactate
   levels, and had a major impact on circulating lipoproteins: cholesterol
   and several broad lipid peaks were strongly decreased upon TCPOBOP
   treatment, while other lipid peaks were increased ([132]Fig. 3B). In
   Car^+/+ females, cholesterol and lipid signals were changed in a
   similar manner than in males ([133]Fig. 3C). Putative assignment of
   these differential peaks revealed that the decreased cholesterol peak
   reflected HDL-cholesterol, decreased lipid peaks belonged to LDLs and
   increased lipids belonged to VLDLs. The area under the curve for
   selected HDL-cholesterol, lipid-LDL and lipid-VLDL signals further
   illustrated this CAR-dependent impact of TCPOBOP treatment on
   circulating lipoproteins in both sexes ([134]Fig. 3D). Total plasma-,
   HDL-cholesterol, plasma triglycerides, free fatty acids, and glucose
   were quantified through additional classical biochemical assays and
   confirmed the strong CAR-dependent impact of TCPOBOP on plasma
   metabolites observed by metabolomics ([135]Fig. S4).

Fig. 3.

   [136]Fig. 3
   [137]Open in a new tab

   Sex-independent impact of CAR activation on lipoprotein metabolism.

   (A) PCA of plasma metabolomic dataset. (B) Coefficient plots related to
   the PLS-DA models discriminating between plasma spectra from TCPOBOP
   vs. CO Car^+/+ males. Parameters of the PLS-DA model: Q^2Y = 0.84, p =
   0.001. (C) Coefficient plots related to the PLS-DA models
   discriminating between plasma spectra from TCPOBOP vs. CO Car^+/+
   females. Parameters of the PLS-DA model: Q^2Y = 0.89, p = 0.001. (D)
   AUC of the ^1H-NMR spectra was integrated for glucose, lactate,
   cholesterol, lipid CH[3](CH[2])[n] (mainly in Ldl), lipid
   CH[3]CH[2]CH[2]C = (mainly in Vldl) signals. Results are given as the
   mean ± SEM. ∗Treatment effect, ^#sex effect. ∗ or ^#p <0.05, ∗∗ or ^##p
   <0.01, ∗∗∗ or ^###p <0.001 (two-way ANOVA). (E) Multi-omic integrative
   analysis performed on plasma metabolomic and hepatic transcriptomic
   datasets (DIABLO model). (F) Correlation network between hepatic
   transcripts and plasma metabolites (R^2 > 0.97, DIABLO model). (G)
   Fold-change (TCPOBOP- vs. CO-treated Car^+/+ mice) of hepatic
   expression for genes involved in lipoprotein metabolism. ∗Treatment
   effect, ^#sex effect. ∗p[adj] <0.05, ∗∗p[adj] <0.01, ∗∗∗p[adj] <0.001
   (linear model). CAR, constitutive androstane receptor; CO, corn oil;
   PB, phenobarbital; PCA, principal component analysis; PLS-DA,
   orthogonal projection on latent structure-discriminant analysis;
   TCPOBOP, 1,4-bis[2-(3,5-dichloropyridyloxy)] benzene; Vldr, very
   low-density protein receptor.

   Plasma metabolomics and hepatic transcriptomic data statistical
   integration revealed a strong correlation between plasma metabolites
   and liver transcripts (R^2 = 0.94) regardless of sex ([138]Fig. 3E).
   The correlation network highlighted a strong positive correlation
   between Apoc1 mRNA and plasma LDL-cholesterol and HDL-cholesterol
   ([139]Fig. 3F). This led us to investigate more closely the hepatic
   expression of genes involved in hepatic cholesterol metabolism
   ([140]Fig. 3G). CAR activation significantly decreased the expression
   of most apolipoproteins. Expression of the Ldl receptor (Ldlr), which
   is responsible for LDL clearance was unchanged; however, the expression
   of Vld receptor (Vldlr) was increased by a factor of four upon TCPOBOP
   treatment. Mttp, the protein that transports triglycerides and
   cholesterol esters in the endoplasmic reticulum for VLDL synthesis was
   also significantly increased. Moreover, TCPOBOP treatment impacted
   cholesterol and bile acid metabolism with decreased expression of genes
   involved in cholesterol transport, decreased expression of genes
   involved in bile acid synthesis and increased expression of genes
   involved in bile acid detoxification and transport. All significant
   changes in hepatic mRNA and plasma metabolites related to cholesterol
   metabolism were CAR-dependent and were similar in both sexes
   ([141]Fig. S5 and S6). Altogether, these results illustrate that CAR
   activation deeply modulates hepatic and systemic cholesterol metabolism
   in a sex-independent way.

TCPOBOP induced liver oxidative stress and toxicity in a sex-biased way

   We next performed metabolic profiling of hydrophilic metabolites in
   liver tissue. PCA of the entire metabolic profile revealed a distinct
   clustering of Car^+/+ mice treated with TCPOBOP vs. all other mouse
   groups on the first principal component, whereas male and female mice
   were separated on the second component, revealing once again a major
   impact of CAR activation on liver metabolites ([142]Fig. 4A).
   TCPOBOP-treated males displayed significant changes in hepatic levels
   of many amino-acids (increased glutamine and glutamate and decreased
   valine, leucine, and isoleucine), energy-related metabolites (increased
   lactate and 3-hydroxybutyrate, and decreased succinate), cell membrane
   constituents (decreased choline and glycerophosphocholine and increased
   phosphocholine) and metabolites involved in oxidative stress (decreased
   hypotaurine) ([143]Fig. 4B). In females, most of these metabolites
   followed the same pattern, with perturbations of metabolites involved
   in oxidative stress being more pronounced than in males (significant
   increased levels of reduced [GSH], oxidised [GSSG], and total [Gsx]
   glutathione) ([144]Fig. 4C). AUC for glutathione signals illustrated
   that TCPOBOP induced a more pronounced hepatic oxidative stress in
   females than in males ([145]Fig. 4D). All significant changes in
   hepatic metabolites related to oxidative stress upon TCPOBOP were
   CAR-dependent ([146]Fig. S7). Finally, sex-dependent TCPOBOP hepatic
   toxicity was confirmed through biochemical quantification of plasmatic
   markers ([147]Fig. 4E–F). Circulating levels of ALT were significantly
   higher in TCPOBOP-treated Car^+/+ females vs. males, whereas plasma ALP
   was significantly increased in both sexes upon TCPOBOP with a tendency
   to higher levels in males (P[sex∗treatment] = 0.12). Overall, our
   results demonstrate a sex-dependent impact of CAR activation on liver
   metabolism and on the toxicity profile.

Fig. 4.

   [148]Fig. 4
   [149]Open in a new tab

   TCPOBOP treatment increases liver oxidative stress and toxicity in a
   sex-biased manner.

   (A) PCA of the liver metabolomic dataset. (B) Coefficient plots related
   to the PLS-DA models discriminating liver extract spectra from TCPOBOP-
   and CO-treated Car^+/+ males. Parameters of the PLS-DA model: Q^2Y =
   0.89, p = 0.001. (C) Coefficient plots related to the O-PLS-DA models
   discriminating between liver extract spectra from TCPOBOP- and
   CO-treated Car^+/+ females. Parameters of the PLS-DA model: Q^2Y =
   0.91, p = 0.001. (D) AUC of the ^1H-NMR spectra was integrated for the
   glutathione signals (GSH, reduced form; GSSG, oxidised form; GSx, total
   glutathione) and for hypotaurine. (E) Plasma alanine aminotransferase.
   (F) Plasma alkaline phosphatase. Results are given as the mean ± SEM.
   ∗Treatment effect, ^#sex effect. ∗ or^#p <0.05, ∗∗ or ^##p <0.01, ∗∗∗
   or^###p <0.001 (two-way ANOVA). CAR, constitutive androstane receptor;
   CO, corn oil; GSH, reduced glutathione; GSSG, oxidized glutathione;
   GSx; PLS-DA, projection on latent structure-discriminant analysis; PCA,
   principal component analysis; TCPOBOP,
   1,4-bis[2-(3,5-dichloropyridyloxy)] benzene.

CAR modulated liver trimethylamine metabolism through regulation of Fmo3 gene
expression mostly in females

   Another intriguing sex-dependent hepatic impact of TCPOBOP was the
   female-specific, CAR-dependent increased level of trimethylamine (TMA)
   ([150]Fig. 4, [151]Fig. 5A). TMA is a gut microbiota-dependent
   metabolite that is metabolised to TMAO by the liver-specific flavin
   monooxygenase 3 (FMO3).[152]^23 This result was in accordance with the
   female-specific decrease of Fmo3 mRNA expression observed previously in
   the microarray data ([153]Fig. 2E) and was confirmed here with RT-qPCR
   ([154]Fig. 5B). Finally, we quantified hepatic TMAO and confirmed a
   twofold decrease of this metabolite in TCPOBOP-treated Car^+/+ females
   compared with vehicle-treated females ([155]Fig. 5C). Female-biased
   inhibition of Fmo3 mRNA by CAR activation was reproducible in an
   independent study in which mice were treated with TCPOBOP or CO every 2
   days for 10 days ([156]Fig. 5D). Next, we analysed Fmo3 hepatic
   expression in several publicly available gene expression datasets. In
   the first experiment, Car^-/- mice were knocked-in with human CAR
   coding sequence[157]^24 and were fed diets containing 0 (CTRL) or
   1,000 ppm PB, an indirect activator of both mCAR and hCAR.[158]^25 We
   found that PB-fed hCAR mice had significantly lower expression of
   hepatic Fmo3 mRNA compared with control mice ([159]Fig. 5E). The second
   study compared wild-type (WT) mice treated with TCPOBOP vs. CO-treated
   mice and hCAR mice treated with CITCO (a specific agonist of hCAR) vs.
   CO-treated hCAR mice.[160]^26 Both TCPOBOP and CITCO-treated mice had
   significantly lower Fmo3 hepatic mRNA compared with their relative
   controls ([161]Fig. 5F). Finally, the last two studies were conducted
   in chimeric mice with human hepatocytes treated with
   PB.[162]^25^,[163]^27 Again, we found a significant decrease in Fmo3
   hepatic gene expression in response to PB in both datasets
   ([164]Fig. 5G and 5H). It is worth noting that all publicly available
   studies were conducted in male mice only, which could explain why the
   decrease in Fmo3 expression seen upon CAR activation by PB or CITCO was
   of lower magnitude than that seen in our own in vivo experiments in
   females. Finally, we took advantage of the only available ChIP-seq
   analysis of hCAR binding in vivo to date[165]^28 and observed that,
   among the 6,364 unique genes associated with high-confidence
   hCAR-binding genes, Fmo3 was found as a hCAR-direct binding gene with a
   p value = 1.26 × 10^-29 ([166]Fig. 5I). Thus, in females, CAR
   activation by TCPOBOP perturbated the metabolism of TMAO from TMA by
   inhibiting the expression of Fmo3 ([167]Fig. 5J) and this result might
   be relevant to humans.

Fig. 5.

   [168]Fig. 5
   [169]Open in a new tab

   CAR modulates liver TMA metabolism by regulating Fmo3 gene expression,
   mostly in females.

   (A) AUC of the ^1H-NMR spectra for TMA. (B) Hepatic gene expression.
   (C) Hepatic content of TMAO. (D) Hepatic gene expression in independent
   confirmatory experiment. Results are mean ± SEM. ∗Treatment effect,
   ^#sex effect. ∗ or ^#p <0.05, ∗∗ or ^##p <0.01, ∗∗∗ or ^###p <0.001
   (two-way ANOVA). (E–H) Hepatic mRNA expression derived from publicly
   available datasets. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001 (one-way ANOVA or
   Student t test). (I) Proposed model for direct binding of hCAR to Fmo3
   regulatory DNA sequences based on Niu et al.[170]^28 (created with
   [171]BioRender.com). (J) Proposed impact of CAR activation on TMA
   metabolism (created with [172]BioRender.com). (K) Experimental design.
   (L) Hepatic gene expression. Results are mean ± SEM. ∗Treatment effect,
   ^#sex effect. ∗ or ^#p <0.05, ∗∗ or ^##p <0.01, ∗∗∗ or ^###p <0.001
   (two-way ANOVA). (M) Representative images of platelet adhesion. (N)
   Quantification of platelet adhesion to a microfluidic chip surface ∗p
   <0.05, ∗∗p <0.01 (one-way ANOVA). CAR, constitutive androstane
   receptor; GSH, reduced glutathione; GSSG, oxidised glutathione; GSx,
   total glutathione; PCA, principal component analysis; PLS-DA,
   orthogonal projection on latent structure-discriminant analysis; TMA,
   trimethylamine; TMAO, trimethylamine N-oxide.

CAR activation decreased platelet hyperactivity induced by dietary choline
supplementation

   FMO3 activity and TMAO levels have been shown to modulate platelet
   hyper-responsiveness and thrombosis potential.[173]^29^,[174]^30 We
   thus wondered whether CAR activation could also influence platelet
   function. To enhance platelet responsiveness, female and male mice were
   fed a choline-enriched diet before treatment with TCPOBOP
   ([175]Fig. 5K). We confirmed significant hepatic CAR activation in both
   sexes ([176]Fig. 5L) and examined thrombi formation. Whole blood from
   female mice treated with TCPOBOP formed smaller thrombi over time
   compared with blood from control mice. In males, TCPOBOP treatment did
   not affect thrombi formation. Thus, in vivo TCPOBOP treatment modulates
   platelet activation and reduces the thrombotic risk of females
   specifically ([177]Fig. 5M and N).

Discussion

   The liver appears to be one of the most sexually dimorphic organs and
   expression of genes involved in drug metabolism is sexually dimorphic
   in rodents and humans.[178]^8^,[179]^31 CAR is the target of many drugs
   and is widely involved in the control of expression of xenobiotic
   metabolising genes. However, a genome-wide description of sex-specific
   CAR-dependent sensitive genes was lacking. Here, we provide an
   exhaustive study of the transcriptomic impact of acute pharmacological
   CAR activation in male vs. female mice and novel insights into the
   metabolic impact of this activation.

   First, most TCPOBOP-modulated genes were regulated in a similar manner
   in male and female livers, especially those involved in the cell cycle.
   Our results are consistent with other studies revealing that chronic
   activation of CAR using TCPOBOP promotes tumour formation in rodents in
   a CAR-dependent manner.[180][32], [181][33], [182][34] The underlying
   mechanisms depend on direct CAR-dependent induction of Mdm2, a primary
   inhibitor of P53 and induction of the transcription factor
   FoxM1,[183]^35 which is essential for the initiation of
   carcinogen-induced liver tumours, thus resulting in modulation of many
   genes implicated in cell proliferation, cellular growth, apoptosis, and
   cell differentiation, such as Gadd45b, Ccnd1, Ccnb1, C-myc, and
   Yap.[184]^35^,[185]^36 Few studies have highlighted that sex could
   influence TCPOBOP-induced liver proliferation but showed inconsistent
   results. Some studies described that female mice were more sensitive to
   TCPOBOP-induced liver proliferation,[186]^12^,[187]^37 whereas others
   showed no tumour development in female mice treated with the genotoxin
   diethylnitrosamine followed by TCPOBOP, compared with males.[188]^38
   Similarly, after a single injection of TCPOBOP, male mice displayed a
   deeper disturbance of key cell cycle genes.[189]^16 Unlike these
   studies, we did not reveal any sexual dimorphism on hepatomegaly and
   cell cycle gene modulation upon CAR activation. However, a long-term
   analysis of TCPOBOP-induced liver tumours conducted in male and female
   mice in parallel would be required to further investigate this
   discrepancy.

   We next observed a strong impact of TCPOBOP treatment on plasma
   lipoproteins, with decreased total-, HDL- and LDL-cholesterol measured
   in both sexes. This result is consistent with previous findings whereby
   TCPOBOP decreased circulating levels of plasma HDL in both WT and
   transgenic mice expressing human apolipoprotein A-1, at least in part
   through downregulation of ApoA-1 hepatic gene expression.[190]^39
   Similarly, in Ldlr^-/- mice fed a Western-diet, TCPOBOP decreased
   circulating levels of ApoB-containing lipoproteins (mainly VLDL and
   LDL) and reduced the development of atherosclerotic lesions,[191]^40
   presumably through CAR-mediated induction of the VLDL receptor, a
   receptor involved in the clearance of VLDL and LDL as a backup for the
   LDL receptor.[192]^41 Here, we confirm that CAR activation results in
   decreased ApoA-1 and increased Vldlr hepatic mRNA levels, which could
   both participate to the observed decrease of plasmatic HDL and LDL
   levels. Moreover, we also observed a strong decrease in hepatic
   expression of other major lipoprotein-coding genes (namely ApoB, Apoc1,
   and ApoE), and of Lecithin cholesterol acyl transferase (Lcat, another
   constitutive component of HDL) which could also play a role.

   Another well-known function of CAR is its ability to promote bile acid
   detoxification during cholestasis.[193]^42^,[194]^43 As previously
   described, we found that the expression of genes involved in
   hydroxylation, sulfation, and excretion of bile acids was significantly
   enhanced upon CAR activation, whereas expression of genes involved in
   bile acid synthesis and cholesterol transport was decreased in both
   sexes. The emerging role of CAR in cholesterol homeostasis represents
   new perspectives in the treatment of hypercholesterolaemia and
   atherosclerosis.[195]^39^,[196]^44 The current study confirms and
   extends these previous studies reporting the effects of TCPOBOP on
   hepatic expression of genes involved in bile acid, cholesterol and
   lipoprotein metabolism, as well as those on lipoprotein concentrations,
   are sex-independent, at least in mice. Our findings may have clinical
   relevance. Indeed, a recent study combining genome-wide analysis of
   cholestatic mice genetic models and data-mining of human patient
   cohorts with various liver diseases unravelled a significant enrichment
   of CAR-sensitive genes in cholestatic livers specifically.[197]^45
   Moreover, CAR activation in cholestasis leads to alterations of drug
   metabolism with significant effects on drug-induced hepatotoxicity.
   Drug-induced liver injury (DILI) is still a serious clinical concern
   and one of the most common drug adverse reactions. DILI clinical
   phenotype is influenced by age and sex.[198]^46^,[199]^47 Here, we
   found that, upon TCPOBOP, 385 genes displayed a female-specific vs. 101
   genes with a male-specific response. Many female-specific genes were
   involved in extracellular matrix organisation. We also observed
   stronger perturbations of hepatic metabolites involved in glutathione
   metabolism in livers of females compared with males, reflecting higher
   hepatic oxidative stress. Finally, we highlighted a sexually dimorphic
   impact of CAR activation on clinical markers of liver toxicity, namely
   significantly higher levels of ALT in females compared with males, and
   a trend toward higher ALP levels in males. This result is consistent
   with the sex-influence on DILI clinical phenotype with cytolytic damage
   being more frequently observed in women, whereas cholestatic damage
   presented a male predominance.[200]^46^,[201]^47 It is well known that
   women experience higher rates,[202]^48 and more severe[203]^49 adverse
   drug reactions than men. However, mechanistic explanations for these
   observations are often lacking. Our present results suggest that drugs
   interacting with CAR may be considered with particular attention before
   their use in women.

   Limitations of our study include the use of only one drug (TCPOBOP),
   whereas DILI has been shown to depend both on patient characteristics
   and on drug properties.[204]^50 Our results therefore need to be
   confirmed with other drugs that act as CAR agonists.

   Another novel finding from our study was the strong increase in hepatic
   TMA upon TCPOBOP administration observed in female mice specifically.
   TMA is a product of the gut microbial metabolism of
   phosphatidylcholine, choline, and L-carnitine. It is transported from
   the gut to the liver via the portal vein and N-oxidised into TMAO by
   host FMO3.[205]^51 Analysis of natural genetic variation in inbred
   strains of mice indicate that FMO3 and TMAO are significantly
   correlated and explain more than 10% of the variation in
   atherosclerosis.[206]^51 Since then, it has been confirmed that high
   circulating levels of TMAO are linked to increased thrombotic and
   cardiovascular risks in animal and human studies, even after adjustment
   for known cardiovascular risk factors.[207]^52^,[208]^53 Consistent
   with increased TMA, hepatic Fmo3 mRNA expression and TMAO concentration
   were both strongly decreased in Car^+/+ females treated with TCPOBOP.
   We confirmed the CAR-dependent regulation of Fmo3 mRNA in publicly
   available cohorts that used different hCAR models and different mCAR
   and hCAR agonists, therefore suggesting that the regulation of Fmo3
   expression is not dependent on the CAR agonist used and might be
   relevant in humans. In rodents, hepatic Fmo3 knockdown was sufficient
   to decrease diet-dependent platelet responsiveness and thrombotic
   potential.[209]^29^,[210]^30 Here, we observed that, in conditions of
   diet-induced platelet hyper-responsiveness, CAR activation was indeed
   sufficient to significantly modulate thrombus growth in female mice
   specifically. We postulate that this effect is, at least in part,
   attributable to the CAR-mediated downregulation of Fmo3 expression and
   activity in females. Nowadays, drugs represent the main cause of
   platelet dysfunction.[211]^54^,[212]^55 Our results suggest that drugs
   or other xenobiotics (such as pollutants, foods) that interact with CAR
   could decrease thrombus formation in a pro-thrombotic context. These
   compounds may provide a beneficial effect by modulating platelet
   activation and thrombosis. We, therefore, highlight a new axis between
   hepatic xenobiotic metabolism and blood haemostasis. We suggest that
   this axis may be especially relevant in women. However, there are
   important species-specific sex-based differences in FMO3 expression:
   its expression is female-specific in mice as a result of modulation by
   sex steroids,[213]^56 whereas its abundance was significantly
   associated with females in humans but to a much lower extent than in
   rodents.[214]^57 Thus, the gender-specificity and clinical relevance of
   this CAR–FMO3–TMAO–platelet axis deserves further investigation. In
   addition to platelet function and thrombotic risk, an increase in the
   TMAO plasma concentration has also been shown to increase the risk of
   impaired glucose tolerance,[215]^58 colorectal cancer,[216]^59 chronic
   kidney disease,[217]^60 and overall mortality.[218]^61 Whether drugs
   interacting with CAR could also influence these TMAO-dependent
   endpoints deserves further investigations.

   In summary, the present study provides an exhaustive description of the
   sex-independent and sex-dependent CAR-sensitive genes and demonstrates
   a stronger impact of CAR pharmacological activation on hepatic
   transcriptome and metabolism of the female. Additionally, CAR
   activation impacted the TMA–FMO3–TMAO pathway in females, which might
   link drugs and environmental xenobiotic exposure with platelet
   aggregation and other TMAO-sensitive physiological responses.

Financial support

   MH is the recipient of a PhD grant from the French Ministry of
   Research. This work was supported by Région Occitanie (Hepatomics to
   NL) the Fondation pour la Recherche Médicale (grant number
   ENV202109013962 to HG) and grants from the French National Research
   Agency (Hepatomorphic ANR-20-CE14-0035 to HG; Gadget
   ANR-22-CE34-0005-01 to SE-S).

Authors’ contributions

   Conceptualisation: MH, FL, NL, SE-S.Methodology: MH, FL, MG, BE, AP,
   CC. Software and statistical analysis: MH, AP, YL, SE-S. Investigation:
   MH, FL, MG, BE, JB, TF, CM, CR, AF, CN, GG, ER-B, SE-S.
   Writing—original draft preparation: MH, SE-S. Writing—review and
   editing: MH, BE, AF, ER-B, UR-K, BP, EB-R, LG-P, HG, NL, SE-S.
   Supervision: YL, ER-B, LD, UR-K, MB, BP, NL, HG, SE-S. Project
   administration, NL, HG, SE-S. Funding acquisition: MB, BP, NL, LG-P,
   HG, SE-S. Read and approved the final manuscript: all authors.

Data availability statement

   Gene expression data generated for this study with microarray are
   available in NCBI's Gene Expression Omnibus and are accessible through
   GEO Series accession number [219]GSE228554. Other data are available
   from the corresponding author upon reasonable request.

Conflicts of interest

   The authors declare no conflicts of interest.

   Please refer to the accompanying ICMJE disclosure forms for further
   details.

Acknowledgements