The batch used in this study was chemically characterized using the Folin–Ciocalteu method (total polyphenol estimation), the Dubois method (total sugar estimation), the Phospho-Vanillin method (total lipid estimation) (31, 32), the O-phthalaldehyde method (total protein estimation) (33), and HPLC-UV/visible/MS (Agilent Technologies Infinity series 1,200 and 1,260, Santa Clara, CA, USA) with a C18 column (250 × 4.6 mm, 5 μm, Phenomenex, USA) and a HILIC Silica column (150 × 4.6 mm, 5 μm, Waters, The Netherlands) (quantification of potential compounds of interest). The results of this analysis are presented in Table 1.

All animal procedures were approved by the local ethics committee (C2E2A, Auvergne, France) and comply with the ARRIVE guidelines. Male C57BL/6JOlaHsd mice were purchased from Envigo (Gannat, France). All mice were housed at Valbiotis animal facility (Valbiotis R&D Center, Riom, France) at 22°C under standard 12 h light-12 h dark cycle. Upon arrival, 56 mice were fed a normal diet (ND) (D14042701, Research Diets, New Brunswick, NJ, USA) for 2 weeks of acclimatization until the age of 8 weeks. Then, animals were randomly assigned to four groups (n = 14 mice per group) with similar average body weight and fat mass: (i) ND, (ii) WD (41 Kcal % fat, 30 Kcal % sucrose and 0.21% w/w cholesterol; D12079B, Research Diets, New Brunswick, NJ, USA), (iii) WD supplemented with 1 g/kg of Totum-070 (WD-T070 1 g/kg), and (iv) WD supplemented with 3 g/kg of Totum-070 (WD-T070 3 g/kg). Totum-070 solution was prepared in vehicle (1% Tween-20 in tap water) at 100 g/L or 300 g/L. Totum-070 was administered to WD-fed mice by gavage daily. Control groups (WD and ND) received only vehicle by gavage. Food and water were supplied ad libitum. Supplementation was for 6 weeks. Body weight was recorded weekly, and food intake was measured 3 times per week. The ND and WD energy densities were 3.9 kcal·g⁻¹ and 4.7 kcal·g⁻¹, respectively. Plasma lipid parameters were monitored before the study start (week 0), after 3 weeks of supplementation (week 3), and at the study end (week 6). Mice were fasted for 6 h and blood was collected from the tail vein into EDTA-treated capillaries. Blood was centrifuged at 2,000 g at 4°C for 10 min. Plasma was harvested and stored at −80°C until analysis. Body composition was assessed with an echo MRI system (Zynsser Analytic, Frankfurt, Germany). At the end of the 6 weeks of supplementation, animals were fasted for 6 h, anaesthetized with isoflurane, and euthanized by cervical dislocation.

Cholesterol levels in the plasma were quantified enzymatically using the CHOD-PAP colorimetric assay kit (Biolabo SAS, Maizy, France). Plasma triglyceride concentration was measured using a colorimetric assay kit (Cayman chemical, Ann Arbor, MI, USA).

Liver triglyceride level was determined with the triglyceride assay kit by Cayman Chemical. Mouse livers (100 mg) were homogenized in 1 ml of a solution containing 5% NP40 in water followed by centrifugation (10,000 g at 4°C for 10 min) to recover the supernatant. Triglyceride concentration was normalized to the liver sample weight and expressed as mg/g liver weight.

At the study end, liver tissue was fixed in paraformaldehyde at 4°C for 48 h and kept in 70% ethanol for 2 to 3 h before automated tissue processing. Samples were embedded in paraffin and 4μm sections were cut with a microtome. Sections were dewaxed and rehydrated in successive xylene and ethanol baths. Then, they were stained in Gill II hematoxylin for 4 min, and after removal of excess stain in an alcohol acid bath, they were transferred to a 0.25% eosin Y bath for 45 s. Finally, sections were dehydrated with ethanol and xylene, and slides were mounted with the Eukitt medium. Other liver tissue samples were frozen in OCT medium, and 10 μm tissue sections were cut with a cryostat and fixed in cold 10% formalin for 5 min. Slides were stained with Oil Red O at 60°C in an oven for 8 min. Oil Red O areas were quantified using the ImageJ software. Stained areas of at least 10 μm² and with a circularity between 0.1 and 1 were selected for quantifying the Oil Red O surface as percentage of the total surface. At least three images per liver were analyzed.

Total mRNA was extracted from snap-frozen tissues using TRIzol® (Invitrogen, Life Technologies). cDNA was synthesized from 2 μg RNA with the High-Capacity cDNA transcription kit (Applied Biosystems, Life Technologies). PCR amplification was carried out using the CFX system (Bio-Rad, Marnes-la-Coquette, France) with Taqman probes (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) and the ΔΔCt method was used to quantify mRNA levels. Gene expression was normalized to the housekeeping genes Gapdh (liver samples) and Polr2a (intestine samples).

Liver samples were homogenized in 20 µl/mg RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS) on ice using a glass potter. Just before use, protease (P8340, Sigma, Saint-Louis, MO, USA) and phosphatase (88,667, Thermo Fisher Scientific, Whaltham, MA, USA) inhibitors were added to the buffer. Homogenized samples were centrifuged at 14,000 g at 4°C for 10 min, and supernatant was collected. Protein content was determined using a commercial Lowry-based DC protein assay (Bio-Rad, Marnes-la-Coquette, France). All samples were adjusted to a standard concentration (15 µg) and then diluted with 2× Laemmli buffer. Western blotting was performed as described by Chavanelle et al. (34) with the following specificities: membranes were incubated overnight at 4°C with primary antibodies against LDL Receptor (Biovision, Milpitas, CA, USA) diluted to 1:1,000. After overnight incubation, membranes were washed with Tris-buffered saline/0.5% Tween-20 and incubated with anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1:2,000) at room temperature for 1 h. Then, membranes were washed three times in Tris-buffered saline/0.5%Tween-20 before addition of an enhanced chemiluminescent solution (Clarity Western ECL; Bio-Rad) for 1 min. Images were acquired with the Bio-Rad ChemiDoc system, and band density was determined using Image Lab V6.0 (Bio-Rad). Stain-free blot images were used as total protein loading control for data normalization (35).

For each cage (1 animal per cage), feces were collected over a week at the study end. Feces were dried at 60°C for 72 h. Fecal lipid quantification was performed using 0.5 g of dried sample per cage by chloroform/methanol lipid extraction as described previously (36). The lipid phase was placed in a glass recipient and solvent was evaporated. The total lipid mass per g of feces was calculated by subtracting the empty glass recipient weight.

Lyophilized feces (± 50 mg) were mixed in 750 µl of 0.04 M sulfuric acid using an Ultra-Turrax® (Ika, Staufen, Germany). Samples were centrifuged (16,000 g, 4°C, 15 min) and the supernatants were filtered (0.22 μm). The three major SCFAs (acetate, propionate and butyrate) were quantified by high-performance liquid chromatography (HPLC, Elite LaChrom, Merck HITACHI, USA) coupled to a UV detector (λ = 205 nm). Elution was done using an ICE-COREGEL 87H3 9 µm 150 × 7.8 mm column and its pre-column (Concise Separations, San Jose, CA, USA) at 50°C with 0.04 M sulfuric acid as mobile phase (0.6 ml/min). Data were analyzed with the EZChrom Elite software. SCFA concentrations were calculated from standard curves established with a known equimolar concentration solution of acetate, propionate, and butyrate (0, 10, 25, and 40 mM).

Lipase (type II, from porcine pancreas, Sigma, Saint-Louis, MO, USA) hydrolytic activity was assessed by monitoring the conversion of 4-methylumbelliferyl oleate (4-MUO, Sigma, Saint-Louis, MO, USA) into 4-methylumbelliferone (4-MU). The enzymatic activity was determined by measuring 4-MU (Sigma, Saint-Louis, MO, USA) fluorescence signal at 460 nm (after excitation at 355 nm) over time. A commercial reference inhibitor, orlistat (Sigma, Saint-Louis, MO, USA), was used as positive control, and citrate-phosphate buffer (0.1 M, pH 7.4) as reaction buffer. A series of 4-MU dilutions in buffer was used to generate the standard curve to convert the absorbance values into 4-MU concentrations. A 4 g/L solution of lipase in buffer was centrifuged at 5 000 g, 10°C for 10 min, then the supernatant was aliquoted. The final in-well concentration used in the assay was 250 mg/L. In each microplate well, 20 µl of test products diluted to 10:90 (v:v) in a DMSO:water mixture (or 20 µl of this solvent mixture for controls), 20 µl of lipase in buffer (or 20 µl of buffer for the blanks), and 110 µl of buffer were incubated at 37°C in a microplate reader for 10 min. Then, 10 µl of 4-MUO substrate (48 µM in DMSO) was added to each well to initiate the reaction. Fluorescence was measured at 355 nm/460 nm (excitation/emission) using a FLUO Star Omega (BMG LabTech) 96-well microplate reader, maintained at 37°C, every 30 s for 15 min. The relative lipase activity was calculated using Equation 1. The IC₅₀ of each compound corresponded to the lowest concentration at which the lipase activity was halved.(1)%lipaseactivity=ResidualactivityNegativecontrolactivity×100

Lipidomic analysis was performed at the Center for Proteomics and Metabolomics, Leiden University, with feces collected as described above. Lipids were extracted from 50 mg of dry feces using the methyl-tert-butylether method, and analyzed using Lipidyzer™, a direct infusion-tandem mass spectrometry (DI-MS/MS)-based platform (Sciex, Redwood City, USA), as described previously (37, 38). Lipid concentrations were expressed as pmol/mg of dry feces.

Fresh feces were collected during week 6 of the study and immediately frozen in liquid nitrogen. DNA was extracted from 50 to 100 mg of sample using the FastDNATM Spin Kit for feces (MP Biomedicals, Illkirch, France) and FastPrep-24 5G (MP Biomedicals, Illkirch, France) following the manufacturer's instructions. DNA was eluted in 50 µl of elution buffer, pre-heated at 55°C, and normalized to 20 ng/µl. Microbial 16S amplicon libraries were prepared at the Plateforme Génome Transcriptome de Bordeaux, Bordeaux, France, by amplification and sequencing of the V3–V4 region of the 16S rRNA amplicon on an Illumina MiSeq machine, using the 2 × 250 base pair Illumina v2 kit. Quality filtering and processing were done as described previously (39) with cutadapt v4.0 (40) and Python v3.9.9, usearch v11.0.667 (41, 42) and mothur v1.48.0 (43).Classification of quality filtered reads was performed by comparison with the ribosomal database project (RDP trainset 14) (44, 45).

Prism V.7.0 and 8.0 (GraphPad Software) were used for statistical tests and figure drawing. The Shapiro–Wilk normality test was used to determine whether data followed a Gaussian distribution. If data were not normally distributed, the Kruskal–Wallis nonparametric test was used followed by the Dunn test for post hoc comparison. When the normal distribution was confirmed, data were compared with the one-way or two-way ANOVA and Tukey's test for multiple comparisons. For measurements repeated over time, differences between groups and time points were evaluated using a repeated-measure two-way ANOVA followed by the Tukey's post hoc test for multiple comparisons. If missing data did not allow using the repeated-measures two-way ANOVA, a mixed-effects analysis was performed. Values are presented as the mean ± SEM, unless specified otherwise. Differences were considered significant at p < 0.05.

For microbiota analyses, relative abundance tables were imported and analyzed in R v4.1. Alpha (Chao and Shannon indices) and beta diversity (Bray-Curtis and Jaccard distances) were estimated using vegan 2.6–2 (46) and compared using the Kruskal-Wallis and MANOVA tests, respectively. Indicator taxa were selected using the KW test and multiple testing correction with False Discovery Rate (FDR) correction. For indicator taxa, an additional linear model was tested to determine the type of observed effect. Correlations between the relative abundances of taxa and arbitrary values were tested using the Pearson correlation coefficient. Arbitrary units were allocated to the experimental groups and six models were tested: (1) Totum-070 induces a return to normal that goes beyond the ND value (ND = 25, WD = 75; WD-T070 3 g/kg = 10), (2) Totum-070 induces a return to normal (ND = 25, WD = 75; WD-T070 3 g/kg = 25), (3) Totum-070 induces a partial return to normal (ND = 25, WD = 75; WD-T070 3 g/kg = 50), (4) Totum-070 does not rescue the changes induced by WD (ND = 25, WD = 75; WD-T070 3 g/kg = 75), (5) Totum-070 accentuates the WD effect (ND = 25, WD = 75; WD-T070 3 g/kg = 100), (6) Totum-070, but not WD, induces a change (ND = 100, WD = 100; WD-T070 3 g/kg = 10). Indicator taxa were identified by selecting taxa significantly associated with the supplementation group in the Kruskal-Wallis test, and for which at least one of the linear models was significant. If multiple linear models were significant, the model with the highest effect size was reported in Table 2. Correlations between indicator taxa and metabolic parameters were assessed using the Pearson correlation with FDR correction for multiple testing.

To establish a model of lipid metabolic disorder, mice were fed a WD (n = 42) or a normal diet (ND; n = 14) for 6 weeks. Animals fed a WD were subdivided in three groups: WD alone (WD; n = 14), WD supplemented with 1 g/kg of Totum-070 (WD-T070 1 g/kg; n = 14), and WD supplemented with 3 g/kg of Totum-070 (WD-T070 3 g/kg; n = 14). Totum-070 (or vehicle alone in the WD and ND groups) was administered by gavage daily. The final body weight was significantly higher in the WD groups than ND group (Figure 1A). Body weight was similar between the WD and WD-T070 1 g/kg groups throughout the study (Figures 1A,B). Conversely, after 4 weeks of supplementation, body weight increase was reduced in the WD-T070 3 g/kg group compared with WD mice (Figure 1A). This resulted in a 30% lower (p < 0.01) body weight gain (Figure 1B) and fat mass gain (Figure 1C) compared with the WD group at the supplementation end. No change was observed in lean mass between the WD group and mice supplemented with Totum-070 (both concentrations) (Figure 1D). During the study, the daily caloric intake was higher in the WD groups compared with the ND group (Figure 1E), and was similar among the WD groups (WD and WD-T070) (Figures 1E,F). Therefore, Totum-070 gavage did not affect energy intake in WD mice.

At the study start (week 0), plasma cholesterol concentrations were identical in all groups (Figure 2A). After 3 weeks, plasma cholesterol was significantly increased (+38%) in the WD group, and this level was maintained at the study end (+36%) (Figure 2A). Totum-070 (1 g/kg and 3 g/kg) dose-dependently reduced plasma cholesterol concentration after 3 weeks (−8.7%, p < 0.05 and −12.7%, p < 0.01 vs. WD group, respectively), and after 6 weeks of supplementation (−7.3%, p = 0.08 and −11%, p < 0.01 vs. WD group, respectively) (Figure 2A). Circulating triglyceride concentration was similar in all groups throughout the study (Figure 2B).

Histological analysis of liver samples revealed abnormal accumulation of lipid droplets induced by WD that was reduced by Totum-070 supplementation (Figure 3A). In agreement, quantification of Oil Red O staining in liver sections confirmed in the WD-T070 3 g/kg group, the reduction of neutral lipid accumulation to the level observed in the ND group (Figure 3B). Moreover, hepatic triglyceride content was reduced by 17% (p < 0.05) in the WD-T070 1 g/kg group and by 40% (p < 0.0001) in the WD-T070 3 g/kg group (Figure 3C) compared with the WD group. Analysis of the expression of genes implicated in liver lipid and lipoprotein metabolism showed that Totum-070 supplementation influenced the expression of some genes involved in cholesterol metabolism (Figure 3D). Specifically, Totum-070 supplementation limited the upregulation of the cholesterol transporters Abcg5 and Abcg8 induced by the WD. Conversely, it did not have any effect on the expression levels of Srebf2 and the downstream genes Ldlr and Hmgcr that were decreased in all WD groups. WD and Totum-070 supplementation did not have any effect on LDL receptor (LDL-R) protein expression in liver (Figure 3E). Totum-070 supplementation did not have any effect on the expression of genes implicated in de novo lipogenesis, whereas it decreased (vs. WD) the expression of genes involved in fatty acid oxidation (Ppara, Cpt1a, Acox1), but only when used at the concentration of 3 g/kg (Figure 3F). Totum-070 further decreased the expression of genes involved in lipoprotein metabolism, such as Mttp (both doses compared with the WD and ND groups) and also Apoc3 and Angptl3, but only at the higher dose (Figure 3G).

To better understand the regulation of lipid homeostasis by Totum-070, total lipid content was quantified in feces from mice in the WD and WD-T070 3 g/kg groups (Figure 4A). This showed that fecal lipid content was increased by 31% in the supplemented group (p < 0.05). Lipidomic analyses in stool samples highlighted the increase of specific lipid classes in the WD-T070 3 g/kg group compared with the WD group: cholesterol ester (CE) (+437%, p < 0.05), lysophosphatidylethanolamine (LPE) (+98%, p < 0.001) and phosphatidylethanolamine (PE) (+197%, p < 0.05). Conversely, triglycerides (TG) were decreased by 49% (p < 0.05) in feces from the WD-T070 3 g/kg group compared with the WD group (Figure 4B). Genes implicated in intestinal cholesterol absorption were assessed for differential expression between groups in the duodenum (Supplementary Figure S1A), jejunum (Figure 4C) and ileum (Figure 4D). Most investigated genes were not influenced by Totum-070 in the three intestinal segments. Apob mRNA level was repressed specifically in the jejunum (Figure 4C). The transcript for Abcg5 was downregulated by Totum-070 in the ileum only (Figure 4D).

To complete the characterization of pathways linked to cholesterol metabolism, the concentration of the major bile acid species was measured in feces. With the exception of cholic acid (CA), all bile acid classes were significantly reduced in feces from mice in the WD-T070 3 g/kg group compared with the WD group: α-muricholic acid (αMCA) −41%; β-muricholic acid (βMCA) −60%; ursodeoxycholic acid (UDCA) −92%; hyodeoxycholic acid (HDCA) −54.5%; and deoxycholic acid (DCA) −63% (Figure 4E). Based on these data, the expression of genes associated with bile acid regulation was assessed in liver and ileum (Supplementary Figure S1B,C). In liver, Cyp8b1 and Cyp27a1 were significantly downregulated in the WD-T070 3 g/kg group compared with the WD group as well as Slc10a1 and Slco1a5, which encode transporters implicated in the regulation of bile acid uptake by hepatocytes. In ileum, none of the genes investigated was affected by Totum-070 administration.

Lastly, an in vitro pancreatic lipase activity inhibition test was used to directly assess Totum-070 effect on intestinal lipid digestion. Comparison of Totum-070 and orlistat (the reference lipase inhibitor drug) effects on lipase activity (Figure 4F and Supplementary Figure S1D) showed that Totum-070 inhibited lipase activity (IC₅₀ = 305.6 µg/ml), but not as strongly as orlistat (IC₅₀ = 0.0084 µg/ml.

Next, 16S rRNA amplicon sequencing was used to identify the microbiota structure in feces collected from mice in the ND, WD and WD-T070 3 g/kg groups at the study end. At the phylum level, evenness (Shannon index) was significantly decreased in the WD group (p < 0.05), compared with the ND group, but not in the WD-T070 3 g/kg group (Figure 5A). At the genus level, species richness (Chao index) was comparable among groups, whereas evenness (Shannon index) was increased in both WD and WD-T070 3 g/kg groups compared with the ND group (Figure 5B,C). Principal coordinate analysis of the relative abundance of intestinal microbial communities at the phylum level using the Bray-Curtis dissimilarity index followed by PERMANOVA showed that microbial composition differed significantly in the three groups (Bray-Curtis, adonis: p < 0.05 R_NDvsWD = 016, p < 0.01 R_{NDvsWD−T070 3 g/kg} = 0.24, p < 0.1 R_{WDvsWD−T070 3 g/kg} = 0.12) (Figure 5D). The same analysis at the genus level confirmed that the WD strongly affected microbial composition compared with the ND (Bray-Curtis, adonis: p < 0.0001, R_NDvsWD = 0.32; Figure 5E and Jaccard, adonis: p < 0.0001, R_NDvsWD = 0.20; Figure 5F). The microbial composition also differed between the WD-070 group and the ND group (Bray-Curtis, adonis: p < 0.0001 R_{WDvsWD−T070 3 g/kg} = 0.39; Figure 5E and Jaccard, adonis: p < 0.0001, R_{WDvsWD−T070 3 g/kg} = 0.21; Figure 5F) and also the WD group (Bray-Curtis, adonis: p < 0.001, R_{WDvsWD−T070 3 g/kg} = 0.26; Figure 5E and Jaccard, adonis: p < 0.0001, R_{WDvsWD−T070 3 g/kg}= 0.20; Figure 5F). Given the supplementation effect on the alpha and beta diversity measures and the visible differences in the average composition at the phylum and genus levels, an indicator species analysis was performed to identify the taxa that were specifically affected in the different conditions. Data are shown at the phylum level (Table 2) and genus level (Figure 6 and Table 3) as results of the two extremes of the taxonomic levels.

At the phylum level, supplementation with Totum-070 (3 g/kg) led to an increase in the relative abundance of Proteobacteria compared with the WD group (+89%, p < 0.05) and the ND group (+195%, p < 0.01). Moreover, the Firmicutes to Bacteroidetes ratio remained identical between WD and ND (Table 2) but was 20.5% lower in the WD-T070 3 g/kg group compared with the ND group (p < 0.05).

At the genus level (Figure 6 and Table 3), the proportions of uncl. Coriobacteriia (p = 0.036), Phocaeicola (p = 0.0111), uncl. Muribaculaceae (p = 0.0162), Muribaculum (p = 0.0096), Parabacteroides (p = 0.0162), Ruthenibacterium (p = 0.0096), Oscillibacter (p = 0.0181), uncl. Alphaproteobacteria (p = 0.0199) and Desulfovibrio (p = 0.0116) were significantly increased in the WD-T070 3 g/kg group compared with the WD group. Conversely, Paramuribaculum (p = 0.0096), Odoribacter (p = 0.0181), uncl. Bacteroidales (p = 0.0096), uncl. Bacteroidetes (p = 0.0181), Lactococcus (p = 0.0116) and uncl. Lactobacillales (p = 0.0193) were significantly decreased in the WDT070 3 g/kg group compared with the WD group. Noteworthy, some WD-linked alterations were reversed by Totum-070 supplementation. Specifically, Muribaculum richness was significantly reduced in WD-fed mice (−79%, p < 0.01, WD vs. ND), but was increased by Totum-070 (+207%, p < 0.01, WD-T070 3 g/kg vs. WD). Moreover, the Pearson's correlation analysis showed a significant negative association between Muribaculum richness and plasma cholesterol (p = 0.0026). Conversely, Lactococcus richness was significantly increased by WD (+375%, p < 0.01, WD vs. ND, respectively) and reduced by Totum-070 supplementation (−84%, p < 0.05, WD-T070 3 g/kg vs. WD). The relative abundance of Lactococcus was positively correlated with plasma cholesterol (p = 0.0322) (Table 3).

Our analysis showed that the cholesterol-lowering effect of Totum-070 was related to gut microbiota modifications. As fibers and some herb-derived polysaccharides are fermented into SCFAs (mainly acetate, propionate, and butyrate) by intestinal bacteria (47) with potential beneficial effects on lipid disorders, Totum-070 effect on SCFA production was investigated at the study end. Quantification of SCFA species in the feces revealed a significant global increase of the total fecal SCFA amount in the WD-T070 3 g/kg group compared with the WD group (+102.8%, p < 0.05). This effect was mainly driven by fecal acetate production (Figure 7A). Post-hoc analysis indicated a significant increase of acetate in feces from WD-T070 3 g/kg mice compared with WD mice (+86%, p < 0.05). Propionate was not detected in any sample and butyrate only in samples from WD-T070 3 g/kg mice. The total SCFA amount in feces was negatively correlated with plasma cholesterol levels in the WD and WD-T070 3 g/kg groups (Figure 7B, p < 0.05).