Microbiome-host communication via vaccenic acid modulates LRH-1/NR5A2 activity and ameliorates metabolic liver disease

Bifidobacterium animalis subsp. lactis B420 lipid extracts stimulate LRH-1 transcriptional activity

To explore the hypothesis that metabolites derived from intestinal bacteria in general, and probiotic in particular, may act as LRH-1 ligands and thereby modulate LRH-1-mediated transcriptional regulation of these processes, we first investigated whether lipid extracts from Bifidobacterium animalis subsp. lactis B420^TM (hereafter B420) could modulate LRH-1 activity using a reporter assay. We generated LS174T cells that overexpress human Flag-tagged LRH-1 upon induction with 4-hydroxytamoxifen (4-HT) (Fig. EV1A,B), and transiently transfected them with a luciferase reporter plasmid containing 5 repeats of the LRH-1 response element (5xRE). This colorectal epithelial cell line, unlike hepatic cell lines, has the advantage of low endogenous LRH-1 expression, making it well suited for investigating the effects of induced LRH-1 overexpression.

Induction of LRH-1 expression by 4-HT led to a dose-dependent activation of the 5xRE reporter (Fig. EV1C), whereas no significant effects were observed in a control cell line with 4-HT-induced expression of EGFP, or in cells transfected with the empty luciferase reporter control plasmid pGL3 (Fig. EV1D).

Lipid extracts isolated from B420 at a concentration of 50 μg/mL strongly enhanced LRH-1 reporter activity in LRH-1-overexpressing cells (Fig. 1A). This effect was minimal in EGFP-overexpressing cells and can be attributed to the low levels of endogenous LRH-1 in LS174T cells that can be detected by RT-PCR (Fig. EV1A). LRH-1 transactivation by B420 extracts was dose-dependent and more pronounced at higher levels of 4-HT-induced LRH-1 expression (Fig. EV1E).

To confirm the specificity of LRH-1 in the B420-induced transcriptional activation, we used two well-characterized LRH-1 pharmacological inhibitors, 3d2 and SR1848. 3d2 binds to the LRH-1 LBD, induces a conformational change that impairs its interaction with co-regulatory proteins (Benod et al, 2013) and reduces LRH-1 basal transcriptional activity in cell lines (Michalek et al, 2022). SR1848 promotes rapid translocation of LRH-1 from the nucleus to the cytoplasm (Corzo et al, 2015). Both molecules effectively reduced the B420-stimulated LRH-1 activity (Fig. 1B), suggesting that the stimulatory effect of B420 is dependent on LRH-1. Neither B420 extracts nor the LRH-1 inhibitors affected cell viability, as monitored by measuring cellular respiration by MTT assay (Fig. EV1F,G). To evaluate whether endogenous LRH-1 could also be activated by the bacterial extracts, we employed human Caco-2 cells, which express higher levels of endogenous LRH-1 (Sidler et al, 2011). Again, we could confirm that B420 extracts dose-dependently stimulated LRH-1 transcriptional activity, which was inhibited by 3d2 (Fig. EV1H).

In order to investigate whether the effect of B420 extracts on LRH-1 transcriptional activity was mediated via binding of potential ligands to the LBP, we made use of a mutant LRH-1, in which the mutations F342W and I416W prevent ligand interaction with the LBP (Ortlund et al, 2005; De Jesus Cortez et al, 2016; Lee et al, 2011). The Flag-tagged mutant LRH-1 F342W/I416W protein was induced in LS174T cells, resulting in comparable expression levels to those of wild-type LRH-1 after 4 h and overnight induction (Fig. 1C). Expression of mutant LRH-1 reduced basal activity and markedly impaired responsiveness to B420 extracts (Figs. 1D and  EV1I), indicating that B420 extracts contain ligands that directly activate LRH-1.

Interestingly, the LRH-1-activating capacity was not restricted to B420 nor limited to probiotic strains, as extracts from various Gram-negative and Gram-positive commensal and pathogenic bacteria also promoted LRH-1 transactivation. Among the strains tested (Lactobacillus rhamnosus, Proteus mirabilis, Enterobacter cloacae, Vibrio cholerae, Klebsiella pneumoniae, Enterococcus faecalis, Pseudomonas aeruginosa), all extracts modulated LRH-1 activity (Fig. 1E). LRH-1 transcriptional activity was stimulated by all bacterial extracts, with the exception of extracts from P. aeruginosa. Pre-treatment with the LRH-1 inhibitor 3d2 reduced this response to varying degrees. Additionally, lipid extracts from different Escherichia coli strains, including commonly used laboratory strains, also transactivated LRH-1 (Fig. EV1J). Interestingly, extracts from P. aeruginosa inhibited LRH-1 transactivation, which may be partly attributable to a decrease in reporter cell viability (Fig. EV1K). No major effects on cell viability were observed for the other bacterial extracts.

In summary, our findings demonstrate that lipid ligands capable of activating LRH-1 transcriptional activity are present in different human gut bacteria, including the probiotic strain B420.

Identification of vaccenic acid as a novel LRH-1-activating ligand

While an increasing number of compounds have been described to bind to and/or activate LRH-1, at present a definite physiological ligand remains elusive. Notably, the synthetic lipid DLPC was shown to bind to the LRH-1 LBD and to modulate its activity both in cell lines and mice (Lee et al, 2011; Musille et al, 2012). To explore whether DLPC might be responsible for the observed effects, mass spectrometry of B420 extracts was performed. The analysis of the chromatogram indicated the absence of DLPC, since the corresponding mass (622.44 m/z) could not be identified (Fig. EV2A).

In the attempt to isolate and identify the bioactive molecules present in the B420 extracts, we combined chromatographic separation of lipid classes with screening for activity on LRH-1 transactivation using the luciferase reporter assay (Fig. EV2B–E). The B420 extract was separated by silica flash column chromatography over 50 min, resulting in 120 vials. A small sample from every second vial was analyzed by thin-layer chromatography (TLC) and vials with similar retention factor (Rf) were combined resulting in ten pools. These were evaporated, resuspended in ethanol and used to screen for LRH-1 transactivation in Caco-2 cells transfected with the 5xRE LRH-1 luciferase reporter, which express high levels of endogenous LRH-1 compared to LS174T cells (Fig. EV2B). Pool 1, which induced the highest LRH-1 transactivation, was further fractionated with a second silica flash column chromatography. This active pool did not contain phospholipids (PLs), which were collected in pool 5 to 10 (Fig. EV2B). As in the previous step, individual vials were collected, combined into eight pools based on TLC analysis, and pooled fractions were tested for LRH-1 activity in Caco-2 cells (Fig. EV2C). The most active pool 1.5 was further purified by HPLC (Fig. EV2D). Samples were combined into six pools based on TLC analysis and tested for LRH-1 transactivation. The active pool 1.5.3 was further purified by a second HPLC, and fraction 1.5.3.47, eluting with a clear peak at 32.1 min (Fig. 2A) and showing a strong TLC spot (Fig. 2B), was analyzed by 1D NMR (¹H, ¹³C), 2D NMR (COSY, HMBC, HSQC) and GC-MS (Appendix Figs S1–S7). These identified the bioactive compound as vaccenic acid (VA, Δ¹¹-octadecanoic acid), a C18:1 fatty acid, present in both cis and trans isomeric forms (Fig. 2C).

To confirm the presence of cis- and trans-VA in active B420 extracts, we analyzed these fatty acids as FAMEs (fatty acid methyl esters) by GC-MS/MS. Comparison of retention times and fragmentation patterns with a reference mixture of cis- and trans-VA methyl esters confirmed the presence of both isomers in B420 extracts (Fig. 2D). We further investigated the presence of cis- and trans-VA in the other bacterial extracts that showed LRH-1 transactivation, namely P. mirabilis, E. cloacae, V. cholerae, E. faecalis, and P. aeruginosa. Cis-VA was found in all bacteria extracts tested, whereas the trans isomer was only detected in B420 (Fig. 2E).

As the active fraction 1.5.3.47 contained both cis- and trans-VA, the individual isomers were purchased and tested separately for LRH-1 transactivation and potential cytotoxicity (Fig. EV3A). Both VA isomers stimulated LRH-1 transcriptional activity in LRH-1-overexpressing LS174T cells, whereas no activation was observed in EGFP-expressing control cells (Fig. 2F). The response to VA was dose-dependent with lowest activities already detected at 0.1–1 μM and maximal effect at a concentration of 100 μM (Fig. EV3B,C). Basal activity and responsiveness to both cis- and trans-VA was reduced in mutant LRH-1 F342W/I416W (Fig. EV3D,E) compared to wild-type LRH-1, validating the hypothesis that VA directly engages the LBP of LRH-1.

The murine LRH-1 protein is considered constitutively active and less dependent on agonist binding compared to human LRH-1 (Sablin et al, 2003; Krylova et al, 2005). Nevertheless, many studies have demonstrated that LRH-1 modulation by agonistic molecules is also effective in mice (Cobo-Vuilleumier et al, 2018b; Mays et al, 2022; Lee et al, 2011). For this reason, we were interested in understanding whether murine LRH-1 activity could also be modulated by bacterial extracts and VA. In LS174T cells with inducible mLRH-1 overexpression (Fig. EV3F), transactivation of murine LRH-1 was induced by B420 extracts, and inhibited by 3d2 and SR1848 (Fig. EV3G). Both cis- and trans-VA stimulated murine LRH-1 activity, although to a lesser extent than human LRH-1 (Fig. EV3H).

In conclusion, our results identify the fatty acids cis- and trans-VA present in bacterial extracts as novel ligands for LRH-1, and demonstrate their ability to modulate LRH-1 activity in both human and murine models.

Binding to and modulation of LRH-1 activity by VA in comparison to other long-chain fatty acids

As bacteria can produce different fatty acids, we were interested in which other long-chain fatty acid species could be detected in B420 and whether they could also induce LRH-1 activity. Thus, fatty acids were extracted from B420 and analysed with GC-MS/MS after derivatization to their methyl-ester forms. Besides confirming the presence of cis- and trans-VA (C18:1), several other long-chain fatty acids, including palmitic acid (PA, C16:0), stearic acid (SA, C18:0), and oleic acid (OA, C18:1), were identified as the main lipid species (Fig. 3A). This profile is in accordance with that reported for other B. lactis strains (Oberg et al, 2013).

We then sought to determine whether unsaturated C18 fatty acids might exhibit similar LRH-1-stimulating activity to VA. OA contains one double bond between C9 and C10 in cis configuration, while elaidic acid (EA) is the trans-isomer of OA at the same position. Linoleic acid (LA) is a polyunsaturated fatty acid with two double bonds at position 9 and 12, both in cis configuration, whereas conjugated linoleic acid (CLA, cis-9, trans-11) is derived from trans-VA by the action of stearoyl-CoA desaturase in tissue (Reynolds et al, 2008) (Fig. 3B). Interestingly, none of these fatty acids stimulated LRH-1 activity in the LRH-1 reporter assay in LS174T cells (Fig. 3C), nor did they affect cell viability (Fig. EV4A). One possible explanation of the dissimilarity with VA could be that the position or configuration of the double bond, or the presence of an additional double bond, may hinder binding of these fatty acids to the LRH-1 LBP.

To further investigate this hypothesis, we expressed and purified the LBD of human LRH-1 as a His₆-tagged protein (Fig. EV4B,C) and quantified the binding affinity of the different long-chain fatty acids with microscale thermophoresis (MST). The analysis confirmed binding of cis- and trans-VA to LRH-1 LBD at dissociation constants (K_d) of 14.2 μM and 144.4 μM, respectively, whereas the other fatty acids tested did not show binding at relevant K_d (Figs. 3D–H and  EV4D).

Saturated and unsaturated fatty acids are well-known regulators/activators of the peroxisome proliferator-activated receptor (PPAR) transcription factors of the NR superfamily, namely PPARα (NR1C1), PPARβ/δ (NR1C2), and PPARγ (NR1C3). These NRs form heterodimers with their obligate partner RXR (retinoid X receptor), and bind to specific DNA sequences, known as PPAR response element (PPRE), thereby regulating lipid metabolism and inflammation (reviewed in (Varga et al, 2011)). In our experiments, VA failed to transactivate PPARγ/RXR heterodimers, but was able to stimulate transcriptional activity of PPARα, which was notably enhanced by cotransfection with RXR. No effect was observed on the transcriptional activity of the glucocorticoid receptor (GR, NR3C1) (Fig. EV4E–G). These results highlight the specificity of VA in modulating LRH-1 activity.

In vivo activation of LRH-1 by VA modulates intestinal and hepatic gene expression

LRH-1 is an established regulator of local intestinal GC synthesis via the transcriptional regulation of Cyp11a1 and Cyp11b1 (Coste et al, 2007; Mueller et al, 2006). To evaluate whether VA could function as an LRH-1 agonist in vivo, we administered a mixture of cis- and trans-VA (50 mg/kg) or its vehicle via intraperitoneal (i.p.) injection into male C57Bl/6 mice (Fig. 4A). Mice had ad libitum access to food and water. Tissue was collected either 8 h or 18 h post-injection, as indicated in the figure legends. In the small intestine, VA induced expression of both Cyp11a1 and Cyp11b1 (Fig. 4B), consistent with previous studies employing LRH-1 agonists (Mays et al, 2019, 2022) and describing Cyp11a1 and Cyp11b1 as LRH-1 targets in the intestine (Mueller et al, 2006, 2007). To assess whether these changes require the presence of LRH-1, we employed mice with LRH-1 deletion in intestinal epithelial cells (LRH-1 VilCre) and their littermates with floxed LRH-1 gene, but lacking Cre expression, as controls (LRH-1 L2L2) (Coste et al, 2007; Ahmed et al, 2023). Primary intestinal crypt cells isolated from the small intestine of these mice confirmed reduced Cyp11a1 expression in LRH-1 KO cells, and showed that VA incubation induced expression of Cyp11a1 in control LRH-1 L2L2, but not LRH-1 KO cells, although this difference did not reach significance at the condition used (Fig. EV5A).

Activation of LRH-1 in the liver has been associated with improvement of glucose and insulin homeostasis, likely through transcriptional regulation of de novo lipogenesis (Lee et al, 2011; Mays et al, 2022). Specifically, several key lipogenic genes, including sterol regulatory element binding transcription factor 1 (Srebf1, also called Srebp1-c), acetyl-CoA carboxylase 2 (Acacb), fatty acid synthase (Fasn), and stearoyl-CoA desaturase-1 (Scd1), are downregulated following treatment with the LRH-1 agonists DLPC and 10CA, an optimized derivative of the synthetic LRH-1 agonist RJW100 (Flynn et al, 2018). Similarly, treatment of HepG2 cells with the LRH-1 agonist 6N reduced the expression of SCD1, the rate-limiting enzyme in de novo lipogenesis (Mays et al, 2019; Cato et al, 2022).

To assess endogenous LRH-1 transcriptional activity in vivo in response to VA, we measured the hepatic expression of Scd1, as well as of the established hepatic LRH-1 targets Cyp7a1 and Cyp8b1 in C57Bl/6 mice following a single i.p. injection of VA. Consistent with prior studies, VA administration decreased expression of Scd1, as well as Cyp7a1 and Cyp8b1 (Fig. 4C). Notably, a single VA injection also significantly lowered serum glucose, although serum insulin levels remained unchanged under these conditions (Fig. 4D). These results indicate that VA can activate endogenous LRH-1 and regulate metabolic pathways in the liver. Interestingly, hepatic Shp and intestinal Fgf15 were upregulated after VA treatment (Fig. EV5B,C), suggesting involvement of FXR signaling and the existence of a complex cross-talk between LRH-1 and the FXR-SHP-FGF15 axis.

VA administration improves metabolic liver disease

To evaluate the potential benefits of VA-induced LRH-1 activation in metabolic disorders, we employed a mouse model of diet-induced obesity and hyperglycemia (diet-induced obese (DIO) mice), as LRH-1 has been previously shown to regulate glucose and lipid metabolism (Lee et al, 2011). Mice were randomly assigned to receive either a high-fat diet (HFD) or a standard diet (SD) for 15 weeks. By week 12, the HFD group exhibited significantly higher body weight (Fig. 5A), along with elevated blood glucose (Fig. 5B), serum cholesterol, triglycerides (TG), and the hepatic transaminase GPT/ALT (Fig. EV6A). Additionally, normalized liver weight was altered (Fig. EV6B), hepatic TG content was increased (Fig. EV6C), and histological analysis revealed a fatty liver with enhanced lipid deposition (Fig. EV6D) at the experimental endpoint.

After 15 weeks, mice were randomly assigned to receive daily i.p. injections of VA (50 mg/kg) or vehicle for 2 weeks, while maintaining their respective diets. VA treatment had neither a significant effect on body weight or weight gain during treatment (Fig. EV7A,B), nor did it induce hepatic toxic effects, as both serum GPT/ALT levels and normalized liver weights remained comparable to controls (Fig. EV7C,D). Blood glucose was measured via tail vein puncture right before the first i.p. injection (day 1) and after 7 and 13 days of treatment (Fig. 5C). VA administration reduced blood glucose in the HFD group at both day 7 and day 13 of treatment (Fig. 5D), with the latter approaching statistical significance (P = 0.058). A similar trend was observed in the SD group at day 13, whereas no change in glucose levels was observed in either group after vehicle administration. Serum glucose levels assessed at the end of the experiment were also lower in VA-treated HFD mice (Fig. EV7E), consistent with the tail vein blood measurements. VA also decreased serum insulin levels in the HFD mice, but not in the SD group (Fig. 5E).

LRH-1 plays a central role in liver lipid homeostasis by coordinating several metabolic pathways. Accordingly, disturbances in these LRH-1-regulated pathways result in the development of metabolic disorders, such as NAFLD and type 2 diabetes (Sun et al, 2021). Given that activation of LRH-1 by DLPC protected mice from hepatic steatosis (Lee et al, 2011), we examined the impact of VA administration on liver lipid accumulation in HFD-fed mice. Histological analysis of H&E- and Oil Red O (ORO)-stained liver sections revealed a mild reduction in lipid deposition following VA treatment (Fig. 5F), although no significant change in hepatic TG levels was detected (Fig. EV7F). We performed a more detailed analysis by quantifying macro- and microvesicular steatosis and hepatocyte hypertrophy separately (Fig. EV7G) and calculated the NAFLD activity score (NAS) (Liang et al, 2014). VA administration significantly improved the NAS value (Fig. 5G), predominantly by reducing the contribution of microvesicular steatosis and hepatocyte hypertrophy in comparison to vehicle-treated mice (Fig. EV7H). No effects were observed on collagen deposition and fibrosis, assessed by Sirius Red staining (Fig. 5H), or on serum cholesterol and TG levels (Fig. EV7I,J).

VA decreases the expression of lipogenic genes in the liver

To explore the molecular mechanisms underlying the improvements in glucose and lipid homeostasis induced by VA, we performed RNA-sequencing analyses on the liver samples from VA- or vehicle-treated HFD mice. Gene ontology analysis of significantly differentially expressed genes revealed that VA treatment modulates pathways related to lipid biosynthesis and fatty acid metabolism (Fig. 6A), alongside altered expression of LRH-1 target genes (Fig EV8 and Table EV1). Notably, VA treatment suppressed expression of key de novo lipogenesis genes, including Acacb, Fasn, and Scd1 (Fig. 6B,C). These findings align with previous studies showing downregulation of lipogenic genes following LRH-1 activation with DLPC or with the synthetic agonists 10CA and 6N (Lee et al, 2011; Mays et al, 2022; Cato et al, 2022). Additionally, the expression of Mid1ip1, a gene involved in the positive regulation of fatty acid biosynthesis via the nuclear receptor LXR (Izumi et al, 2021; Inoue et al, 2011), and Cd36, which encodes a fatty acid transporter, was also downregulated in VA-treated mice. Interestingly, VA treatment also decreased expression of the LRH-1 targets Cyp7a1 and Cyp8b1. RT-qPCR analysis of Cyp8b1 and Scd1 confirmed their downregulation upon VA treatment (Fig. 6D). Notably, while the downregulation of Scd1 was restricted to the context of the HFD, a reduction in Cyp8b1 expression was also detected in SD-fed mice. Overall, these results suggest that a reduction in lipogenic flux underlies the beneficial effects of VA on LRH-1-regulated glucose and lipid homeostasis.

Animal studies

All animal experiments were conducted according to the animal experimentation regulations of Germany and were approved by the Ethics Review Committee of the Regional Council Freiburg (reference number: 35-9185.81/G-17/91). Mice were bred and housed under specific pathogen-free conditions in individually ventilated cages at the animal facility of the University of Konstanz and were acclimated to a 12 h light/dark cycle with free access to food and water. Male C57BL/6 mice (originally obtained from Jackson Laboratory) aged 8–12 weeks were randomly assigned to groups fed either a standard diet (SD, SAFE A03, standard diet) or a high-fat diet with 45% calories from fat (HFD, SAFE 245HF, D12451) for 15 weeks. A mixture of cis- and trans-VA (1:1) dissolved in 40% PEG-400, 2% Tween-80 in PBS or the vehicle was administered daily to mice at 50 mg/kg by intraperitoneal (i.p.) injection (volume = 150 μl/mouse) for 2 weeks, while continuing the previously assigned diet. Blood glucose values were assessed right before the first i.p. injection (day 1) and after 7 and 13 days of treatment with a glucometer (GlucoCheck) by puncturing the tail vein. Mice were sacrificed 4 h after the last treatment (day 14) by CO₂ inhalation. Blood and tissues (liver, small intestine) were immediately harvested. Part of the tissues were snap-frozen in liquid nitrogen for RNA isolation, cholesterol and triglyceride (TG) analysis or formalin-fixed for paraffin sections. In parallel, tissue was embedded in Tissue-Tek O.C.T. Compound (Sakura) to prepare cryosections. Mice exhibiting a body weight gain of less than 45%, or greater than 75%, relative to their initial value were excluded from subsequent analyses.

Intestinal epithelial cell-specific LRH-1 knockout mice (LRH-1 VilCre) and the isolation of intestinal crypts have been described previously (Noti et al, 2010; Coste et al, 2007; Ahmed et al, 2023).

Serum analysis

After collection, blood was allowed to clot at RT for 30 min and was then centrifuged for 15 min at 13,000×g. Serum was collected and glucose, cholesterol, TG and ALT/GTP values were immediately analyzed with a Reflotron instrument (Roche). An aliquot was stored at −20 °C for determination of insulin content with the Ultra Sensitive Mouse Insulin ELISA Kit (CrystalChem) following the manufacturer´s instructions.

Histological analysis of liver sections

Tissue biopsies were fixed in formalin and embedded in paraffin. Paraffin sections of 4 µm were stained with hematoxylin and eosin (H&E) and evaluated blindly by light microscopy. Macrovesicular and microvesicular steatosis as well as hepatocellular hypertrophy were scored separately based on the percentage of the total area affected (0: area affected <5%, 1: 5-33%, 2: 34-66%, 3: >66%). Three images/mouse were inspected and the NAFLD Activity Score (NAS) was determined as the sum of the averaged scores, ranging from 0 to 9 (Liang et al, 2014).

Analysis of liver fibrosis

Connective tissue and collagen deposition in the hepatic tissue was visualized by Picro Sirius Red staining. De-waxed and rehydrated paraffin liver tissue sections were stained using 0.1% (w/v) Direct Red 80 (Sigma-Aldrich) dissolved in an aqueous solution of saturated picric acid for 1 h and then washed with 0.5% acetic acid and water. Stained slides were dehydrated in three changes of 100% ethanol (3 min for each step), mounted and examined by light microscopy with a Zeiss Observer Z.1 microscope. Quantification of fibrosis was determined by measuring the percentage of red-stained area in three non-overlapping images per mouse based on thresholds using ImageJ software (Schindelin et al, 2012).

Oil Red O staining

Lipid deposition in hepatic tissue was determined by Oil Red O staining (Mehlem et al, 2013). Frozen liver samples were cut with a Hydrax C500 cryostat (Zeiss) at −14 °C with a thickness of 10 μm, fixed in 10% formalin for 15 min and washed twice with water and once with 60% isopropanol for 5 min. Sections were stained with filtered Oil Red O solution (15 mL of 0.3% Oil Red O (Sigma) in pure isopropanol mixed with 10 mL H₂O) for 5 min, and washed with water until clear. Hematoxylin counterstaining was performed for 1 min, then slides were washed, mounted and sealed. Images were taken with a Zeiss Observer Z.1 microscope.

Liver triglyceride measurement

Lipid extraction from snap-frozen livers was adapted from Storlien et al and Frayn and Maycock (Frayn and Maycock, 1980; Storlien et al, 1991). Samples (50 mg) were mixed with 1 mL of chloroform/methanol (2:1, vol/vol) and homogenized using a TissueLyser (Qiagen) for 1 min. The mixture was then incubated on a rotator for 6–8 h. After incubation, 0.3 mL of double-distilled water was added, followed by vortexing and centrifugation at RT for 20 min at 2000 rpm. The lower organic phase was transferred to a new 2 mL tube and chloroform was evaporated overnight. A second extraction was performed by adding 0.5 mL of chloroform/methanol (2:1, vol/vol) to the remaining phases, followed by rotation and centrifugation. The samples from both the first and second extractions were combined and left overnight for complete evaporation. Samples were then vortexed in 200 μL of chloroform. Three microliters of the sample were pipetted in duplicate into a 96-well plate. Triglycerides (TG) were quantified using the Triglyceride Quantification Kit (MAK266, Sigma-Aldrich), following the manufacturer’s protocols.

RNA isolation and quantitative real-time PCR analysis

Tissues (50–100 mg) were harvested and snap-frozen in liquid nitrogen. RNA isolation from adherent cells was performed 24 h after seeding 0.6–1.2 × 10⁶ cells on a six-well plate. Total RNA was isolated with peqGOLD TriFastTM (VWR) or TRIzolTM Reagent (Invitrogen) and subjected to DNase I treatment (New England Biolabs) (PeqLab), or isolated with the SV Total RNA Isolation System (Promega) according to the manufacturer’s instructions. The cDNA was generated from 1 to 2 µg RNA using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Real-time quantitative PCR was performed using SYBR® Green (Applied Biosystems) with the StepOnePlusTM Real-time PCR System (Applied Biosystems). Gene expression was normalized using murine or human βActin or B2M reference genes, and represented as 2^−∆Ct (∆Ct = Ct target gene − Ct reference gene) or as fold change to untreated/mock-treated controls. Nucleotide sequences of primer sets used are indicated in Table 1. When possible, primers were designed to span an exon–exon junction.

RNA sequencing

RNA-Seq experiments were conducted at QIAGEN Genomic Services. The library preparation was done using TruSeq® Stranded mRNA Sample preparation kit (Illumina Inc). The starting material (500 ng) of total RNA was mRNA enriched using the oligodT bead system. The isolated mRNA was subsequently fragmented using enzymatic fragmentation. Then first strand synthesis and second strand synthesis were performed, and the double stranded cDNA was purified (AMPure XP, Beckman Coulter). The cDNA was end-repaired, 3’-adenylated, Illumina sequencing adaptors were ligated to the fragments ends and the library was purified (AMPure XP). The mRNA stranded libraries were pre-amplified by PCR and purified (AMPure XP). The libraries size distribution was validated and quality inspected on a Bioanalyzer 2100 or BioAnalyzer 4200 tapeStation (Agilent Technologies). High quality libraries were pooled in equimolar concentrations based on the Bioanalyzer Smear Analysis tool (Agilent Technologies). The library pool(s) were quantified using qPCR and the optimal concentration of the library pool were used to generate the clusters on the surface of a flowcell before sequencing on a NextSeq500 instrument (75 cycles) according to the manufacturer's instructions (Illumina Inc.). Differentially expressed genes (DEGs) were calculated with an adjusted P value < 0.05 using the R package DESeq2 (version 1.40.2), and GO analysis was performed with clusterProfiler (version 4.8.3). Visualization was performed using the R package “EnhancedVolcano” (version 1.18.0) and pheatmap (version 1.0.12).

Lipid extraction from bacteria, chromatographic separation and identification of lipids

Lyophilized Bifidobacterium animalis subsp. lactis 420 (B420^TM) was provided by Danisco Deutschland GmbH (Niebull, Germany). The other bacterial strains were grown overnight in 30 mL Luria-Bertani (LB) medium, except for L. rhamnosus, which was grown in De Man, Rogosa and Sharpe (MRS) medium and incubated for 24 h at 37 °C and 180 rpm. Bacteria were then pelleted for 5 min at 3000 rpm and the pellet was washed twice with PBS before proceeding to lipid extraction.

For lipid extractions, 10–100 g of fresh or lyophilized bacteria were rehydrated with 3 volumes of H₂O. After adding 3.75 volumes of chloroform/methanol (1:2) the mixture was stirred and incubated overnight. First 3 volumes of chloroform and then 3 volumes of 40 mM acetic acid were added, each step followed by 1 min vigorous shaking. The solution was filtered and transferred to a separatory funnel to allow separation of the organic and inorganic phases. The organic phase was collected and evaporated and lipid extracts were stored in glass flasks at −20 °C.

Phospholipid determination

Phospholipid (PL) estimation was performed with a colorimetric method based on formation of complexes between PLs and ammonium ferrothiocyanate (Charles et al, 1980). Bacterial extracts and PL standards were dissolved in 2 mL chloroform; 1 mL ammonium ferrothiocyanate (prepared by dissolving 27 g of ferric chloride (FeCl₃ · 6 H₂O) and 30 g of ammonium thiocyanate (NH₄SCN) in 1000 mL H₂O) was added and the mixture vortexed for 1 min. To separate the two phases, samples were centrifuged at 700×g for 10 min at RT. The lower phase was transferred to a quartz cuvette and absorbance at 488 nm was measured.

Lipid extraction from bacteria, chromatographic separation and identification of lipids

Assay-guided chromatographic separation of crude lipid extracts from B. lactis B420^TM was achieved in four steps. First, the crude extract was purified by silica flash column chromatography (Teledyne Rf combiflash, 40 g silica column) in chloroform by using a 0– 100% gradient of chloroform/methanol (7:3) in 50 min and a flow rate of 40 mL/min. Individual vials of 20 mL were collected and screened by thin layer chromatography (TLC) using phosphomolybdic acid stain for detection. Vials with similar content (determined by Rf) were pooled together, evaporated and the residue was solubilized in ethanol and tested for LRH-1 activity by luciferase-based reporter assays. Pool 1 was further fractionated by silica flash column chromatography (Teledyne Rf combiflash, 12 g silica column) using a 0– 100% hexane and ethyl acetate gradient in 50 min and a flow rate of 30 mL/min. Individual vials of 20 mL were collected and analyzed by TLC. Vials with similar content were pooled, evaporated, and tested again for LRH-1 activity. Pool 1.5, was then further purified by HPLC (Shimadzu LC20-AT) with a preparative Reprosil 100 silica column (5 μM, 250 × 10 mm) using 0–100% ethyl acetate in hexane over 90 min and a flow rate of 4 mL/min. Individual vials were collected and analyzed by TLC. Again, vials with similar content (determined by Rf) were combined, evaporated and tested for LRH-1 activity. Pool 1.5.3 was further purified by HPLC (Shimadzu LC20-AT) with a preparative Reprosil 100 silica column (5 μM, 250 × 10 mm) and dichloromethane (A)/methanol (B) as mobile phase (0–2.5% B in 45 min) in a flow rate of 4 mL/min. Fraction 1.5.3.47, eluting at 1% B after 32.1 min, was analyzed by 1D NMR (1H, 13 C), 2D NMR (COSY, HMBC, HSQC) and GC-MS (Appendix Fig S1–S7).

To detect DLPC in bacteria extracts, a MicrOTOF II ESI mass spectrometer was used with an electrospray ionization (ESI) source operating in positive ion mode. Parameters were set as follows: capillary voltage: 4500 V; nebulizer pressure: 12 bar; drying temperature: 180 °C; drying gas flow: 10 L/min; end plate offset: 80 V.

Lipid analysis by GC-MS/MS

Gas chromatography-mass spectrometry (GC-MS) was performed to assess the lipid profile of B. lactis B420. For this aim, lipids were extracted according to a modified Bligh and Dyer procedure (Jensen, 2008). Six hundred mg of lyophilized or fresh bacteria were dissolved in 1 mL 1 M HCl and heated to 80 °C for 1 h. Then, 1.5 mL methanol and 1 mL chloroform were added and the mixture was vortexed for 1 min. Thereafter, 1 mL water and 2 mL chloroform were added, and the mixture was vortexed again for 1 min. The mixture was centrifuged for 10 min at 4500 rpm and the chloroform phase was transferred into a glass vial and evaporated. For fatty acid methylation, 1 mg of extract was dissolved in 50 μL of hexane and 25 μL of trimethylsulfonium hydroxide (TMSH) solution (0.25 M in methanol) were added, and the solution was heated to 100 °C for 10 min. The resulting residue was dissolved in 100 μL hexane and subjected to GC-MS/MS measurement. Gas chromatography was performed on a Trace GC ultra (Thermo Fisher Scientific, Waltham, MA) using an OPTIMA 5 MS, 0.25 μm, 30 m, 25 mm ID column (Macherey-Nagel). Nitrogen was used as carrier gas. Throughout the measurements a constant gas linear velocity of 20 mL/min was maintained. The injection volume was 0.1 μL with a split ratio of 12:1. The initial oven temperature was set to 50 °C, immediately heated up to 80 °C at 20 °C/min, then ramped up to 280 °C with 6 °C/min and held for 2 min. MS/MS analysis was performed with a ISQ Single Quadruple GC-MS System (Thermo Fisher Scientific) mass spectrometer. The emission current was set to 50 μA and an ionization energy of 70 eV was applied. The ion source temperature was set to 200 °C. For ion fragmentation, methane was used as the collision-induced (CI) gas. Data was acquired in a mass range of m/z 41–600, and MS/MS acquisition of all compounds was performed in positive mode, fixed collision energy of 30 eV and dwell time of 0.2 s. Qualitative analysis was performed using Thermo Xcalibur 3.1.66.10 software and the NIST Database.

Protein purification

The ligand binding domain (LBD, residues 299-541) of human LRH-1 (Uniprot O00482) was cloned into the IPTG-inducible pETM-11 vector (EMBL collection) that incorporates a His₆-tag followed by a tobacco etch virus (TEV) protease cleavage site N-terminally to the inserted gene. This expression plasmid was transformed into E. coli Rosetta^TM 2 (DE3) (Novagen) and bacterial cultures were grown at 37 °C in LB medium to an OD₆₀₀ of 0.6. Protein expression was induced using 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and cultures were grown further at 18 °C for 16 h. Cells were harvested by centrifugation at 5000×g at 4 °C for 20 min and the pellets were resuspended in 30 mL of lysis buffer (300 mM NaCl, 20 mM Tris-HCl pH 7.4, 1 mM CHAPS, 10% Glycerol, 5 mM DTT) freshly supplemented with EDTA-free complete protease inhibitor (Roche). Bacteria were lysed via sonication on ice. Cell lysates were centrifuged (40,000×g for 90 min at 4 °C) and the supernatant was applied to a HisTrap HP column (Cytiva) pre-equilibrated in buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl, 1 mM CHAPS, 10% Glycerol, 5 mM DTT). The protein was eluted with an imidazole gradient.

Fractions containing LRH-1 LBD were pooled, treated with TEV-protease to cleave off the His₆-tag and dialyzed overnight at 4 °C against 500 mL of lysis buffer. The sample was then subjected to subtractive HisTrap chromatography. Fractions containing LRH-1 LBD were concentrated to 1.5 mL by centrifugal ultrafiltration using a 10 kDA Amicon Ultra tube (Millipore). The protein was further purified using a HiLoad 16/60 Superdex^® 200 PG column (GE Healthcare) pre-equilibrated in 20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM CHAPS, 10% Glycerol, 10 mM DTT. The protein was flash frozen and stored at −80 °C until use.

Microscale thermophoresis

To determine binding affinities, microscale thermophoresis (MST) was performed using the Monolith NT.115 instrument (NanoTemper Technologies). The purified LRH-1 LBD was labeled using the Monolith NT Protein Labeling Kit GREEN-NHS following the manufacturer´s procedure. Before labeling, protein buffer was exchanged to 20 mM HEPES, pH 7.4 by gel filtration chromatography with a PD SpinTrap G-25 column (Cytiva). Unbound dye was removed by overnight dialysis with a 3.5 KDa Slide-A-Lyzer MINI Dialysis Device (ThermoFisher Scientific). The labeled LRH-1 LBD concentration was adjusted to 60 nM in assay buffer (150 mM NaCl, 10 mM Tris-HCl pH 7.4, 5% glycerol, 0.1% N-octyl β-D-glucopyranoside, 0.05% Tween-20). Lipid stocks in ethanol were dissolved in assay buffer, vortexed and sonicated. A series of sixteen 1:1 dilutions of 20 μL was prepared for each lipid ligand. Each dilution was mixed with 20 μL of labeled LRH-1 LBD to a final concentration of protein of 30 nM and final lipid concentration ranging from 1 mM to 30.5 nM. Samples were incubated for 30 min at RT in the dark and loaded into Monolith NT.115 capillaries (NanoTemper Technologies). MST was measured using a Monolith NT.115 instrument at 22 °C. Instrument parameters were set to 60% LED power and 40% MST power. MST traces were analyzed with the MO. Affinity Analysis software version 2.2.4 (NanoTemper Technologies) selecting the K_d binding model. The cold region was defined as the 1 s region before the T-Jump while the hot region was defined as 0.5–1.5 s after the T-Jump to minimize possible temperature-dependent artifacts (López-Méndez et al, 2021).

Western blotting

For Western Blot analysis, cells seeded in 6-well plates were lysed in 100 μL RIPA Buffer (50 mM Tris pH 7.4, 150 mM NaCl, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate, freshly added Complete™ Protease Inhibitor Cocktail (Roche)) on ice for 15 min. Lysates were centrifuged for 20 min at 14,000×g and 4 °C, and total protein concentration of supernatants was determined with the Pierce™ BCA Assay Kit (Thermo Fisher). Samples were boiled in Laemmli buffer for 5 min at 95 °C and equal protein amounts (30–50 µg/sample) were separated on a 12% SDS-polyacrylamide gel. Then, proteins were blotted onto a PVDF membrane (Roche) in Blotting Buffer (25 mM Tris, 192 mM glycine, 0.1% SDS, 21.5% methanol). Non-specific binding was blocked in 5% non-fat dry milk/TBS-T (137 mM NaCl, 2.7 mM KCl, 15 mM Tris, 0.1% Tween-20) for 1 h and afterwards primary antibodies diluted in 5% BSA/TBS-T were incubated overnight at 4 °C. After washing three times for 5 min with TBS-T, incubation with peroxidase-conjugated secondary antibodies in 5% BSA/TBS-T was performed for 2 h at RT followed by three additional washing steps. An ECL solution (250 mM Luminol, 90 mM p-coumaric acid, 10 mM Tris) with 1% hydrogen peroxide was applied on the membranes and proteins were detected with an Image Quant LAS4000 (GE Healthcare).

Cell culture and treatments

The human cell lines Caco-2 (ATCC HTB-37) and LS174T (ATCC CL-188) were obtained from American Type Culture Collection (ATCC) and cultured in Iscove’s Modified Dulbecco’s Medium supplemented with 10% FCS, 4 mM L-glutamine and 10 μg/mL gentamicin (all from Sigma-Aldrich, Steinheim, Germany). HEK293T were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 5% FBS, 4 mM L-glutamine and 10 μg/mL gentamicin. Cells were grown at 37 °C and 5% CO₂ in a humidified atmosphere and passaged at 70-80% confluency every 2–3 days. Cell lines were regularly tested for mycoplasma contamination using the Venor^®GeM Classic kit according to the manufacturer’s protocol (Minerva Biolabs, Berlin, Germany).

The LRH-1 inhibitor 3d2 (ID: 16728690) was synthesized according to the initial publication from Benod et al (Benod et al, 2013) by ChemBridge Corp (San Diego, USA). SR1848 (SML 1513) was purchased from Sigma-Aldrich. Stock solutions of 3d2 and SR1848 were kept as 20 mM stocks in DMSO and stored at −20 °C or −80 °C. Lipids and bacteria extracts were dissolved in ethanol and stored at −20 °C.

Luciferase reporter assays

For luciferase-based reporter assays, cells were transiently transfected with expression and luciferase reporter plasmids using JetPei^TM (Polyplus Transfection). Plasmids used in this study are listed in Table 2. The LRH-1 reporter containing five copies of the LRH-1 response element in the pGL3 plasmid (LRH-1 5xRE) has been described previously (Schoonjans et al, 2002). The empty vector pGL3 basic was purchased from Promega. Co-transfection of the β-galactosidase expression plasmid (Invitrogen) served for normalization for transfection efficiency. Expression plasmids for nuclear receptors and their cognate luciferase reporters were: mouse PPARα, human FXRα, TK-PPREx3-Luc (kindly provided by David Moore), human PPARγ1, human RXR and human PPARγ1 luciferase reporter (kindly provided by Kristina Schoonjans). One day after transfection, cells were reseeded into 96-well plates, stimulated with medium control or 200 nM 4HT for 4–6 h to induce LRH-1 expression, and then treated for 16–18 h with the indicated substances in cell culture medium containing 10% charcoal-treated FCS, 4 mM L-glutamine and 10 μg/mL gentamicin. Working solutions of bacteria lipid extracts and lipids were prepared by diluting stock solutions with medium containing 10% charcoal-treated FCS, 4 mM L-glutamine and 10 μg/mL gentamicin, vortexed for 1 min and incubated in a bath sonicator for 10 min at 37 °C. For pharmacological inhibition of LRH-1, cells were preincubated with 3d2 (10 μM) for 24 h and the inhibitor was maintained in the medium during treatment with the tested substances. SR1848 (1 μM) was added to cells together with the tested substances 16–18 h before the assay. To assess luciferase and β-galactosidase activity, cells were lysed in lysis buffer (100 mM K₂HPO₄, 0.2% Triton X-100, 1 µM DTT, pH 7.8) for 15 min on ice, collected and centrifuged at 500×g for 10 min. Supernatants were transferred into white 96-well plates and luminescence was measured with an Infinite® 200 PRO plate reader (TECAN) by automatically injecting an ATP solution (10 mM ATP, 20 mM MgCl₂, 35 mM glycyl-glycine) and the luciferin substrate (470 µM firefly luciferin, 270 µM coenzyme-A, 20 mM MgCl₂, 35 mM glycyl-glycine). β-Galactosidase activity was assessed as described previously (Michalek et al, 2022). Luminescence was calculated as relative light units (RLU) by normalizing light units assessed with the luciferase assays to the values obtained with the β-galactosidase assay. Unless otherwise indicated, graphs show luciferase activity normalized to the vehicle-treated condition without LRH-1 induction.

Site-directed mutagenesis

The LRH-1 double mutant F342W/I416W, which displays reduced ligand binding activity, was generated with the QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies) in the pF-5xUAS-MCS-SV40 expression plasmid containing the human Flag-tagged LRH-1 in two steps, following the manufacturer´s instructions. Primers used are indicated in Table 1. Introduced mutations were confirmed by sequencing.

Generation of 4-hydroxytamoxifen-inducible cell lines

GEV16-mediated inducible overexpression of LRH-1 was achieved as described previously (Vince et al, 2007; Lambrecht et al, 2023). LS174T cells were transduced with HEK293T-derived lentiviruses containing the GEV16 plasmid (pHCA/GAL4(1-93).ERα.VP16) and a stable cell line was generated by selection with 1 mg/mL hygromycin B. The second step included transduction of GEV16-expressing LS174T cells with lentiviruses containing the pF-5xUAS-MCS-SV40-Puro-based expression plasmid with either human Flag-tagged LRH-1, the ligand binding mutant human Flag-tagged LRH-1 F342W/I416W or murine LRH-1. GEV16-LS174T cells transduced with EGFP-expressing lentiviruses were used as control. Stable cell lines were generated by selection with 8 µg/mL puromycin. Overexpression of LRH-1 or EGFP was achieved by treatment of cells with 200 nM 4-hydroxytamoxifen (4-HT) for 4 h, unless stated differently.

MTT cell viability assay

Cell viability after indicated treatments was estimated with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT, Sigma) viability assay. After removing the cell culture medium, cells in 96-well plates were incubated in 10% MTT/medium solution for 1 h. Generated violet formazan crystals in living cells were dissolved in DMSO and absorbance was measured at 562 nm with the Infinite® 200 PRO plate reader (TECAN). Values were normalized to those of vehicle-treated cells.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 7.0. If not stated differently, data are represented as mean ± standard deviation (SD), with the indicated number of independent experiments with 3 or 4 technical replicates. Statistical significance was calculated by unpaired two-tailed Student’s t test when comparing two groups, one-way or two-way ANOVA when comparing three or more dependent groups, respectively. Suitable post-hoc tests (Tukey’s or Sidak’s) were applied and are stated in respective figure legends. P values are indicated in the figures or figure legends.

Graphics

The visual abstract was created with BioRender.com.

Lotta K Stenman is an employee of DuPont Nutrition & Biosciences, Kantvik, Finland. The authors declare no competing interests.