Bacteroides uniformis combined with fiber amplifies metabolic and immune benefits in obese mice

ABSTRACT Gut microbiota represents a therapeutic target for obesity. We hypothesize that B. uniformis CECT 7771 combined with wheat bran extract (WBE), its preferred carbon source, may exert superior anti-obesity effects. We performed a 17-week intervention in diet-induced obese mice receiving either B. uniformis, WBE, or their combination to identify interactions and independent actions on metabolism and immunity. B. uniformis combined with WBE was the most effective intervention, curbing weight gain and adiposity, while exerting more modest effects separately. The combination restored insulin-dependent metabolic routes in fat and liver, although the bacterium was the primary driver for improving whole-body glucose disposal. Moreover, B. uniformis-combined with WBE caused the highest increases in butyrate and restored the proportion of induced intraepithelial lymphocytes and type-3 innate lymphoid cells in the intestinal epithelium. Thus, strengthening the first line of immune defense against unhealthy diets and associated dysbiosis in the intestine. This intervention also attenuated the altered IL22 signaling and liver inflammation. Our study shows opportunities for employing B. uniformis, combined with WBE, to aid in the treatment of obesity.


Introduction
Obesity is a leading public health problem given its high prevalence and associated complications [e.g., type 2 diabetes (T2D) and cardiovascular diseases]. A deeper understanding of the multiple factors controlling body weight is urgently needed to tackle obesity. Gut microbiota has been identified as a key factor that, through interactions with the diet and intestinal immunity, influence metabolic health. [1][2][3] Western diets (low-fiber, high-fat and simple sugars) induce gut-microbiota imbalances (dysbiosis) 4,5 and damage intestinal barrier integrity and immune homeostasis. 6,7 Alterations to intestinal microbiota and leakage of their components (cells, proteins, and metabolites) across the intestinal barrier, permeabilized by unhealthy diets, contributes to obesityassociated low-grade inflammation and insulin resistance in metabolic tissues. 7,8 Moreover, investigations revealing interactions between diet, gut microbiota, and host metabolism 9,10 have allowed the identification of key intestinal bacteria that represent new targets to improve metabolic health. 11 For instance, Bacteroides uniformis CECT 7771 reduced the high-fat diet-induced metabolic alterations while improving antigen-presentation by dendritic cells, impacting on CD4 + T-cell proliferation in an obesity model. 12 On the other hand, fiber serves as fuel for commensal intestinal bacteria, promoting their growth and metabolic activity, which could also mediate some of the fiber-induced benefits on human metabolism. 13,14 Indeed, high intake of dietary fiber is known to help in weight maintenance and reduce the risk of coronary heart disease and T2D. 15 Furthermore, Bacteroides spp. thrive in fiber-enriched environments as they have complex enzymatic machinery to utilize oligo-and polysaccharides as nutrients. [16][17] B. uniformis CECT 7771 preferred carbon source was evaluated in vitro, thus maximizing potential benefits of their combined administration. This could also help lower the effective bacterial dose required, overcoming the challenge of producing intestinal anaerobes at high cell densities. Compared to other carbon sources, B. uniformis CECT 7771 thrived on wheatbran extract (WBE), enriched in arabinoxylan oligosaccharides (AXOS). 16 Intervention studies in humans reveal that wheat-derived AXOS have multiple metabolic benefits, 13 especially on the maintenance of glucose homeostasis 17 and also exert a bifidogenic effect and promote the abundance of butyrate-producing bacteria. 18 Here, we hypothesized that B. uniformis CECT 7771 combined with WBE could have an additive impact on restoring the metabolic and immune alterations of diet-induced obesity and exert greater effects than either WBE or the bacterium alone in mice. Specifically, we assessed whether combined intervention protects against the Western dietinduced intestinal immune impairment, which is one of the causative mechanisms underlying metabolic dysfunction in obesity.

B. uniformis and WBE amplified the reduction of weight-gain in mice fed HFHSD
At week 8 body-weight gain induced by high-fat highsugar diet (HFHSD) was lower when B. uniformis and/ or WBE were administered ( Figure 1a). Nonetheless, these effects diminished over time, especially in mice fed WBE alone, while the downward trend continued in those administered B. uniformis alone or combined with WBE until week 12, with B. uniformis and WBE inducing a greater impact on the prevention of weight gain comparatively until week 17 (Figure 1a and Figure S2A). Compared to controls, HFHSD-fed mice showed a slight increase in fasting triglycerides in plasma, cholesterol and glucose ( Figure 1b) and impaired OGTT (Figure 1c). WBE reduced cholesterol and normalized glycemia but did not affect triglycerides ( Figure 1b). By contrast, B. uniformis alone significantly reduced triglycerides in HFHSD-fed mice. Alone or combined with WBE, this bacterium also diminished plasma cholesterol and glucose ( Figure 1b) and ameliorated oral glucose intolerance of HFHSD-fed mice and the area under the curve (AUC) (Figure 1c). By contrast, compared to controls, AUC remained elevated in HFHSD-fed mice receiving WBE alone. This suggests that B. uniformis rather than WBE improved whole glucose clearance in the OGTT. Insulin plasma levels remained unaffected whatever the treatment ( Figure S2B). Neither glucagon-like peptide-1 (GLP-1) nor peptide YY (PYY) in plasma was significantly affected by HFHSD; however, the bacterium-WBE combination normalized GLP-1 levels compared to untreated HFHSD-fed mice and those receiving B. uniformis or WBE separately, while B. uniformis increased PYY in mice compared to controls ( Figure S2B).

Visceral fat thermogenesis normalized by bacteria-WBE combination in mice fed HFHSD
Thermogenesis in the brown adipose tissue (BAT) was enhanced in HFHSD-fed mice as suggested by the increased expression of the uncoupling protein-1 (Ucp-1) after HFHSD-feeding ( Figure 3a). WBE, but not B.uniformis, normalized the expression of Ucp-1 in HFHSD-fed mice. Food efficiency, an indirect measure of energy expenditure, was reduced by week-6 of HFHSD-feeding ( Figure 3b) which was normalized by WBE, combined or not with B. uniformis. Although Ucp-1 abundance was much lower in epididymal fat compared to BAT, its expression also showed a 3-fold increase in untreated mice fed HFHSD (Figure 3c). While Ucp-1 remained high in mice fed HFHSD receiving either B. uniformis or WBE, their combination Bravais-Pearson´s correlation between Ucp-1 or Cpt1-a expression in epididymal WAT to epididymal WAT weight or to Bw gain, N = 38 Data were represented as the mean ± SEM. Statistical differences between HFHSD-fed groups (receiving or not the bacterium and/or WBE) and controls (CD-veh) were analyzed using oneway ANOVA followed by post-hoc Tuckey # P < .05, ## P < .01, ### P < .001 vs control group. *P < .05 and **P < .01 indicate differences within HFHSD-fed groups. B. unif, Bacteroides uniformis CECT 7771; BAT, brown adipose tissue; Bw, body weight; CD, control diet; Cpt1a, carnitine palmitoyltransferase 1a; eWAT, epididymal white adipose tissue; HFHSD, high-fat high-sugar diet; Ucp-1, uncoupled protein-1; veh, vehicle; WBE, wheat bran extract.
interacts to normalize Ucp-1 (Figure 3c). The mRNA levels of carnitine palmitoyltransferase 1a (Cpt1-a), the rate-limiting enzyme of fatty acid oxidation and normally coupled with thermogenesis in the adipose tissue, was also increased in untreated mice fed HFHSD and remained high in either B. uniformis or WBE-treated mice and, to a lesser extent, in those receiving the combination of both (Figure 3c). Positive correlations between Ucp-1 or Cpt1-a to epididymal WAT or weight gain were identified (Figure 3d).

B. uniformis CECT 7771 and WBE administration modifies gut microbiota, while cecal abundance of fatty acids is mainly affected by B. uniformis CECT 7771 in mice fed HFHSD
Alpha diversity descriptor analysis did not show a major effect of HFHSD on the cecal microbiota structure compared to controls. However, the WBE reduced diversity while B. uniformis increased diversity, and their combination restored diversity metrics to control levels ( Figure S4A). The multivariate dbRDA analysis showed how treatments affected the microbial community structure at OTUs level. The dbRDA1 dimension better discriminated WBEtreated groups, whereas dbRDA2 showed the HFHSD shifted microbiota structure and this change was partially restored by B. uniformis administration ( Figure S4B). The interventions also influenced OTUs abundance (potential bacterial species) identified among groups (p < .001). Specifically, the HFHSD caused a depletion of Alistipes spp. (OTU117), Odoribacter spp. (OTU124), Roseburia spp. (OTU87) and Christensenella spp. (OTU114), which was restored efficiently only by B. uniformis ( Figure S4C). WBE increased Bifidobacterium species abundance (OTU102) and, particularly, Lachnospiraceae species including those belonging to Lachnospiraceae NK4A136 and the genus Blautia (OTU273). Concomitantly, WBE reduced the proportion of certain Ruminococcaceae and Oscillospiraceae species together with Bilophila spp (OTU22) ( Figure S4C). WBE and B. uniformis combined administration increased several Lachnospiraceae (OTU30, OTU180, OTU415, OTU76, OTU59, OTU155) and Muribaculaceae (OTU214) species, compared to the other groups (corrected p < .05).
In accordance with WBE increasing SCFAsproducing species (Lachnospiraceae NK4A136 group and Blautia genus), the actual concentration of acetate, propionate, and butyrate were increased while that of iso-butyrate was reduced in the cecal content of mice fed HFHSD receiving the WBE ( Figure 4a). Nevertheless, the butyrate increase was higher in HFHSD-fed mice receiving the combination of WBE and B. unformis, suggesting an additive effect in the production of this key metabolite ( Figure 4a).

B. uniformis CECT 7771 combined with WBE prevents the HFHSD-associated pro-inflammatory shift in the gut
To unravel mechanisms through which B. uniformis combined with WBE improved the metabolic phenotype of HFHSD-fed mice, we examined intestinal barrier integrity-related mediators. In colon, the expression of antimicrobial peptides, i.e. lysozyme 1 (Lyz1), regenerating islet-derived protein 3 gamma (Reg3ɤ), and phospholipase A2 group IIA (Pla2g2a) or the proliferator marker Ki67 remained unchanged regardless of treatment ( Figure S6A). In ileum, the analysis of tight junction protein expression showed reduced occludin in HFHSD-fed mice ( Figure S6B). Separately, neither B. uniformis nor WBE reversed this reduction but their combination increased occludin expression in HFHSD-fed mice. Occludin in ileum negatively correlated to weight gain and adiposity and positively to hepatic glycogen ( Figure S6C). As we demonstrated that the combination of B. uniformis and WBE has an additive effect preventing body weight gain and visceral adiposity, we further analyzed key intestinal immune cell populations in this intervention group and compared it to controls and untreated mice fed HFHSD.
Untreated obese mice showed increased intracellular levels of IFNɤ in the epithelium of small intestine (Figure 5a) indicating that HFHSD-feeding induced a pro-inflammatory shift in the gut. The analysis of INFɤproducing cells revealed that, compared to controls, HFHSD also led to a higher abundance of macrophages in lamina propria, with similar levels of type 1 and type 2 macrophages ( Figure S6D). In the intestinal epithelium, HFHSD-feeding did not impact on innate lymphoid cells (ILC) type 1 (Figure 5c). Nevertheless, intraepithelial lymphocytes (IEL) Statistical differences between HFHSD-fed groups (receiving or not the bacterium and/or WBE) and controls (CD-veh) were analyzed using one-way ANOVA followed by post-hoc Tuckey # P < .05, ## P < .01, ### P < .001 vs control group. *P < .05, **P < .01 and ***P < .001 indicate differences within HFHSD-fed groups. B. unif, Bacteroides uniformis CECT 7771; CD, control diet; Diunsat FAs, diunsaturated fatty acids; HFHSD, high-fat high-sugar diet; MUFAs, monounsaturated fatty acids; PUFAs, polyunsaturated fatty acids; veh, vehicle and WBE, wheat bran extract.
were altered, showing reduced levels of natural IEL but higher induced IEL (Figure 5d) which highlights a primary role of induced IEL on IFNɤ production as a consequence of HFHSDfeeding. B.uniformis combined with WBE seemed to alleviate the intestinal proinflammatory state since this intervention restored the levels of IFNɤ in the epithelium (Figure 5a) and normalized macrophages in the lamina propria ( Figure 5b) and the levels of induced IEL in the epithelium of the small intestine (Figure 5d)

B. uniformis CECT 7771 combined with WBE ameliorates HFHSD-associated systemic inflammation through IL22 signaling
The expression of il22 in the ileum was increased by the combination of B.uniformis and WBE compared to untreated mice fed CD or HFHSD (Figure 6a). In the small intestine, the abundance of ILC3, a major source of IL22 once activated, was lower in untreated mice fed HFHSD compared to controls and augmented in HFHSD-fed mice receiving B. uniformis combined with WBE represented as the mean ± SEM. one-way ANOVA followed by post hoc Tuckey was conducted for panels C-E. # P < .05, ## P < .01, ### P < .001 vs control group. *P < .05, **P < .01 and ***P < .001 indicate differences within HFHSD-fed groups (fed or not WBE). B. unif, Bacteroides uniformis; Bw, body weight; CD, control diet; epididymal WAT, epididymal white adipose tissue; HFHSD, high-fat high-sugar diet; IEL, intraepithelial lymphocytes; ILC, innate lymphoid cells; WBE, wheat bran extract.
( Figure 6b). Compared to controls, the detrimental effects of HFHSD on intestinal immunity were related to plasma IL22 reductions (Figure 6c), but not coupled to other changes in plasma proinflammatory (IFNɤ and IL12) and antiinflammatory cytokines (IL10 and IL4) (Figure 6c and Figure S7A). The serum concentration of IFNɤ was lower in HFHSD-fed mice receiving B. uniformis combined with WBE and IL22 reduction was attenuated (Figure 6c) compared to untreated mice fed HFHSD, while IL12, IL10, and IL4 remained unchanged ( Figure S7A). Expression of Il22r1 and Il10r2, the membrane receptor complex triggering IL22-mediated intracellular cascade, was unaffected in epididymal fat ( Figure S7B) while, in liver, Il10r2 was slightly higher in HFHSD-fed mice receiving B. uniformis combined with WBE compared to untreated mice fed HFHSD (Figure 6d), while no changes were observed in Il22r1 expression. Hepatic IL22 signal transduction tended to be attenuated in untreated mice fed HFHSD as shown by reduced pSTAT3/STAT3 ratio compared to controls (Figure 6e). This ratio was increased by the administration of B. uniformis combined with WBE to mice fed HFHSD. HFHSDfeeding altered hepatic inflammation, showing higher levels of IFNɤ and IL10 compared to controls ( Figure 6f). Again, B. uniformis combined with WBE normalized the increase in both cytokines in liver (Figure 6f), helping to maintain hepatic immune homeostasis in diet-induced obesity. Hepatic levels of IFNɤ or IL10 levels negatively correlated to glycogen levels (Figure 6g).

Discussion
This study provides a rationale to select dietary fibers that amplifies the benefits of specific intestinal bacteria to tackle obesity, enhancing our knowledge on the underlying mechanisms of action of both the bacterium and the fiber over the immune-metabolic axis. In particular, we demonstrate that body weight and adiposity reduction in diet-induced obesity is greater when B. uniformis CECT 7771 and WBE are administered in combination than separately. Furthermore, our study disentangles the respective contribution of B. uniformis and WBE to the benefits reported, indicating a key role of the bacterium in regulating oral glucose tolerance, promoting hepatic glucose storage, and influencing selective intestinal lipid absorption and serum triglyceride levels. The study also shows how B. uniformis combined with WBE contributes to maintain the intestinal immune homeostasis in obesity, partly by increasing butyrate production, which translates to reduced markers of systemic inflammation and may account for restoration of the metabolic phenotype in diet-induced obesity.
Herein, B. uniformis was administered at a lower dose than in previous pre-clinical trials 12 since we hypothesized its effects would be boosted in vivo by its combination with a fiber previously selected as its best carbon source in vitro. 16 Our results confirmed that while this bacterial dose or WBE administered separately failed to induce anti-obesity effects, their combination reduced body weight gain and adiposity. This was accompanied by improvements in energy metabolic routes and restoration of intestinal immune homeostasis. Nevertheless, each intervention separately ameliorated some metabolic alterations in obesity, independently of weight loss; i.e. reduced cholesterolemia and glycemia and improved glucose intolerance, although B. uniformis plays a greater role than WBE in restoring metabolic alterations.
The effects of the B. uniformis-WBE combination on adiposity were parallel to improvements of some insulin-regulated metabolic routes in adipose tissue and liver. Notably, HFHSD reduced lipogenesis in epididymal fat and hepatic glycogenesis, while both routes were normalized by the B. uniformis-WBE combination. In visceral fat, this intervention restored lipogenic markers (Acc and Fas) and the transcriptional factor regulating their expression Chrebpα. 19 The expression of lipogenic enzymes and Chrebp in WAT positively correlates to insulin sensitivity in obesity. In fact, the reactivation of de novo lipogenesis in the adipose tissue, which mainly uses glycolysis-derived metabolites as precursors of fatty acids, acts as an insulin sensitizer mechanism improving whole-body metabolism in obese subjects. [20][21][22] We also showed hepatic glycogen depots were restored by the B. uniformis-WBE combination, which were reduced by HFHSD, supporting a cross-talk between the visceral fat and liver. Furthermore, the combined intervention enhanced both, Gck expression, which is insulin-dependent 23 and its enzymatic product G6P, diminished in mice fed HFHSD. However, this effect was primarily driven by B. uniformis as it was also identified in group receiving bacteria alone supporting a major contribution of the bacterium restoring hepatic glycogenesis.
Overall, we identified specific metabolic routes through which the combination of B. unformis and WBE favors glucose disposal and improves wholebody insulin sensitivity; i.e. enhancing glucose utilization in the adipocyte of visceral fat to synthesize fatty acids without increasing adiposity and facilitating hepatic glucose storage via stimulation of diverse routes like glucose phosphorylation, which limits glucose output, and glycogenesis. 24 Targeting these metabolic routes, the B. uniformis-WBE combination would maintain physiological lipid storage in adipose tissue to limit ectopic fat accumulation in other organs under a Western diet. Thermogenesis, mainly fueled by fatty acids, 25 was increased in HFHSD-fed mice, as indicated by overexpression of Ucp-1 in BAT and in epididymal fat, whose thermogenic potential may increase under β 3 adrenergic stimulation. 26 Although increased adrenergic-induced thermogenesis associated to diet-induced obesity remains controversial, 27 it may be a mechanism to increase metabolic rate under a Western diet in order to maintain energy balance. 28 The overexpression of Ucp-1 in epididymal fat, coupled with increased Cpt1-a expression of mice fed HFHSD and their positive correlation to fat mass and weight gain, suggests excess energy from the obesogenic diet was partially counteracted by triggering fatty acid degradation by thermogenesis and fatty acid oxidation, albeit insufficient to prevent obesity. Also, the activation of this mechanism was unnecessary in mice receiving B. uniformis and WBE. Therefore, the beneficial effect of B. uniformis and WBE on adiposity was not mediated by the activation of energy dissipating metabolic routes using lipids as fuel, which suggests that the improved wholeenergy disposal was through a mechanism independent of adipose tissue.
Alternatively, we hypothesized that the B. uniformis-WBE combination modifies the metabolites resulting for diet-microbiota interactions, leading to the production of those beneficial for the host metabolism. Notably, WBE increased the concentration of acetate, propionate, and butyrate and the abundance of SCFAs-producing species (i.e. Lachnospiraceae species), which are intestinal bacteria associated with anti-inflammatory properties. 29 Moreover, the intervention with B. uniformis and WBE also led to a gain in several species belonging to the Lachnospiraceae family, which are enriched in oligo-and poly-saccharide degrading genes and SCFA metabolic pathways. 30 In fact, the B. uniformis-WBE combination caused the highest increase in butyrate, one of the key metabolites for improving both metabolism and immunity that could account for the amplified benefits in obesity.
Furthermore, the combined treatment may modify microbial metabolic pathways that, in turn, would modify the nutrients available in the intestinal lumen and thus their accessibility to the host.
Accordingly, and based on the hypothesis that B. uniformis combined with WBE would limit the access to HFHSD-associated excess energy by reducing intestinal lipid absorption, we analyzed the impact of each intervention on long-chain FAs, the most abundant dietary component of HFHSD in the cecum, which may indirectly reflect their intestinal uptake. B. uniformis combined with WBE increased saturated FAs abundance (stearic and arachidic acids) in the cecum, reflecting their reduced absorption. By contrast, B. uniformis, with or without WBE, reduced cecal unsaturated fatty acids (MUFAs, diunsaturated FAs and PUFAs) concentration, which may indicate increased absorption in HFHSD-fed mice. This suggests that B. uniformis combined with WBE would attenuate the damaging effects of saturated lipids on the host (i.e. impairment of intestinal integrity, insulin resistance and systemic inflammation). Previous studies support this, demonstrating the metabolic benefits of B. uniformis CECT 7771 were associated with reduced fat depots in the enterocytes in dietinduced obese mice. 12 Moreover, the bacterium with or without WBE would strengthen the metabolic and anti-inflammatory benefits associated with unsaturated FAs. In particular, increased intestinal uptake of oleic acid may improve glucose homeostasis by preventing the impaired β cell function and the inflammatory associated insulin resistance in liver and adipose tissue induced by palmitic acid. 31,32 In addition, higher absorption of ω3 fatty acids, such as α-linolenic acid, may prevent the disruption of the inflammatory cascade and reverse insulin resistance of diet-induced obese mice. 33 As the maintenance of the immune homeostasis in the gut protects against systemic inflammation and contributes to preventing onset of obesity and T2D, 34 we investigated whether this could be a mechanism whereby interventions improved the metabolic phenotype in HFHSD-fed mice. B. uniformis and WBE together restored occludin expression, a marker of gut integrity, strongly diminished by HFHSD-feeding. Occludin expression inversely correlated to weight gain, adiposity and liver glycogen, supporting a relationship between gut integrity and a healthier metabolic state. Furthermore, HFHSD-feeding induced a proinflammatory state in the gut showing higher levels of IFNɤ and increased macrophages infiltration into the small intestine. Additionally, we have described for the first time that HFHSD-feeding increased induced IEL and its effector cytokine IFNɤ, while decreasing natural IEL. Each IEL subset mainly diverges in its differentiation process; while naïve T cells differentiate into natural IEL in thymus to directly seed the intestine early in life, induced IEL differentiation occurs in lymphoid intestinal organs in response to local tissue damage. 35,36 The increased abundance of induced IEL of mice fed HFHSD could result from a response to enhanced bacterial antigen exposure and/or signals of epithelial cell damage under HFHS-feeding, potentially associated with HFHSD-induced deficits in specific microbial species (Alistipes spp., Odoribacter spp., Roseburia spp. and Christensenella spp). This, in turn, lowers natural IEL, mainly associated with surveillance and maintenance of intestinal homeostasis. 37,38 HFHSD also reduced ILC3, which play a critical role in containing commensal bacteria in the luminal side and replenishing epithelial cells in response to a damage [39][40] . Diminished Th17 cells in the lamina propria, the adaptive counterpart of ILC3, accounts for obesity onset by impairing intestinal surveillance which aggravates dysbiosis-mediated inflammation. 41 Similarly, we also found reduced ILC3 in the intraepithelial layer, the main cells producing IL22 in the intestine. 42 This suggests that the Western diet diminished the IL22mediated defense and mucosal reparation mechanisms accounting for the inflammatory state in obesity. 43 In parallel with the metabolic benefits, the B. uniformis-WBE combination completely normalized the abundance of induced IEL and ILC3, both immune cell populations are regulated by the intestinal microbiota and their dietary metabolic products and provide the first line of immune defense against microbes in the gut and their dysregulation causes loss of gut barrier integrity and inflammation. 42,44,45 Considering gut microbiota analysis, we could postulate that the interventionmediated restoration of IEL and ILC3 was partly due to the associated changes in gut microbiota composition and derived metabolites like SCFAs which are also present in the small intestine. 46 SCFAs enhance ILC proliferation 47 and, in particular, butyrate increases the production of IL22 by RORγt+ ILC3 and CD4 + T cells. 48 Here we showed that B.uniformis combined with WBE restored diversity and enrichment of Lachnospiraceae species that are butyrate producers and induced the higher increase of butyrate in the cecum, mechanisms through which this intervention may promote ILC3 proliferation.
Plasma IL22 was reduced similarly to the intestinal ILC3 in HFHSD-fed mice, suggesting a link between intestinal and systemic immune alterations. We also showed that hepatic IL22 signal transduction cascade was reduced in mice fed HFHSD along with increased IFNɤ and IL10 concentrations, while B. uniformis and WBE partially restored IL22 in plasma and IL22 signaling in liver. These findings suggest the combined intervention may attenuate the HFHSD-associated increase in pro-inflammatory IFNɤ-producing cells in liver, causing insulin resistance, through improved IL22 signaling. Supporting our findings, IL22 displays protective functions in response to hepatic tissue damage 49,50 reducing inflammation in an animal model of alcoholic liver 51 and improving hepatic insulin sensitivity in obesity. 43,50 In conclusion, we reveal the mode of action of B. uniformis CECT 7771 combined with WBE on the metabolic-immune axis in obesity. Limiting intestinal lipid absorption, lipogenesis-related glucose consumption in the adipose tissue, and glycogenesis in liver seemed to be the metabolic routes through which B. uniformis combined with WBE attenuated obesity progression in mice. Furthermore, ILC3 and IEL appeared to be critical intestinal immune mediators of the metabolic benefits of the combined intervention, whose regulation may provide protection against Western dietinduced intestinal immune dysregulation and systemic inflammation.

Bacterial growth conditions
B. uniformis CECT 7777, a bacterial strain originally isolated from healthy breast-fed infants and deposited in the Spanish Culture Collection (CECT), 53 was grown in Schaeldler broth at 37°C under anaerobiosis (AnaeroGen, ThermoFisher UK) as previously described. 12,52 The bacteria were harvested by centrifugation (6000g, 10 min), washed twice with phosphate buffer saline (PBS) solution, and finally resuspended in PBS containing cysteine (0.05%) and glycerol (10%). Aliquots were immediately frozen in liquid nitrogen and stored at −80°C until use. After freezing and thawing, the number of viable bacteria per milliliter was calculated by counting the colonyforming units (CFU) in Schaeldler agar medium after incubation for 48 h, at 37°C under anaerobic conditions. The bacterial identity was routinely confirmed during production and storage by partial 16S rRNA gene sequencing as described previously. 12

Mice and diets
The murine model comprised 6-8 week-old C57BL/ 6 J male mice (Charles Rivers, France), with adlibitum access to water and food except for oral glucose tolerance test (OGTT), housed in collective cages (5 mice per cage) under 12 h of light/dark cycle (lights on at 8:00) in a temperature-controlled room (23 ± 2° C). Mice were randomly allocated into 5 experimental groups by the animal healthcare technicians. For 17 weeks, mice were fed either control diet (CD; D12450K Research Diet NJ, USA; 10% of energy from fat and without sucrose; N = 10), high-fat highsugar diet (HFHSD, D12451 Research Diet NJ, USA; 45% of energy from fat and 35% from sucrose; N = 20) or HFHSD supplemented with 5% of wheat-bran extract (WBE, provided by Cargill, Belgium) (HFHSD-WBE; N = 20) (dietary FAs composition is detailed in Table S1). Both mice fed HFHSD alone or HFHSD+WBE were subdivided into two experimental groups. One of these groups received an oral dose of vehicle (PBS with 0.05% cysteine and 10% glycerol, N = 10) and the other an oral dose B. uniformis CECT 7771 (5x10 7 CFU per mouse, N = 10) daily. CD-fed mice received only vehicle (N = 10).
Body weight was determined weekly. Glucose, triglycerides, and cholesterol in plasma were measured at week 12 and an oral glucose tolerance test (OGTT) was conducted at week 14. At week 17, cardiac puncture was performed in isoflurane anesthetized mice for blood collection. Animals were then sacrificed by cervical dislocation and liver, epididymal white adipose tissue (WAT), ileum and colon (2 cm pieces both) tissues and cecal content were isolated, snap-frozen in liquid nitrogen and stored at −80°C until use. Whole small intestine (except the 2 cm-long distal part) was embedded in cold PBS and immediately processed for the analysis of intraepithelial mononuclear cells by flow cytometry. Another batch of animals were used to estimate food efficiency at week 6 of HFHSD intervention by calculating the ratio between the weight gain and the amount of energy consumed ad libitum during 24 h after 18 h of fasting (refeeding period) (experimental procedure detailed in Figure S1). All experimental procedures were performed in accordance with European Union 2010/63/UE and Spanish RD53/2013 guidelines and approved by the ethics committee of the University of Valencia (Animal Production Section, Central Service of Support to Research [SCSIE], University of Valencia, Spain) and authorized by Dirección General de Agricultura, Ganadería y Pesca (Generalidad Valenciana) (approval ID 2018/VSC/PEA/0090).

Oral glucose tolerance test (OGTT)
At week 14, an OGTT was conducted in 4-hour fasted mice. Blood from the saphenous vein was collected at 0, 15, 30, 60, and 120 minutes of being administered an oral glucose load (2 g/Kg). Glucose was measured through glucose test strips using a glucometer (CONTOUR®-NEXT meter, Bayer, Leverkusen, Germany).

Blood metabolic parameters
Blood from the mandibular vein of 4-hour fasted mice was collected at week 12 in microtubes with K3 EDTA (Sarstedt, Nümbrecht Germany) to determine glycemia, measured as conducted in the OGTT, and triglycerides and cholesterol in plasma (blood centrifugation at 400 g for 15 minutes at 4° C) using a colorimetric kit assay according to man-ufacturer´s instructions (Química Clínica Aplicada, Tarragona, Spain). Plasma from blood collected by cardiac puncture in 4-hour fasted mice at week 17 was used to measure insulin, glucagon-like peptide 1 (GLP-1), peptide YY (PYY) and the cytokine tumor necrosis factor α (TNFα), interleukins IL12, IL10, IL4, and IL22 and interferon ɤ (IFNɤ) using Multiplex Assays Using Luminex® (Milliplex, Merck group, Darmstadt, Germany) according to manufacturer´s instructions.

Isolation of intraepithelial mononuclear cells and flow cytometry analysis
Isolation of intraepithelial cells was conducted as previously described. 54,55 In brief, small intestine was washed with cold PBS and longitudinally opened and cut into small pieces. For the isolation of the epithelium, tissue was incubated twice in Hansk´s balanced salt solution with calcium and magnesium (HBSS; ThermoFisher Scientific, Massachusetts, USA) containing 5 mM EDTA (Scharlab), 1 mM DTT, 100 µg/ml streptomycin and 100 U/ml penicillin (Merck, Darmstadt, Germany) for 30 minutes at 37° C with orbital shaking. After each incubation, supernatant fractions were filtrated using 100 µm nylon cell strainers (Biologix, Shandong, China) and centrifuged to harvest cell suspensions. Remaining tissue was washed with PBS for being incubated twice with HBSS containing 0.5 mg/mL collagenase D (Roche Diagnostics GmbH, Mannheim, Germany), 3 mg/ mL dispase II (Sigma-Aldrich), 1 mg/mL DNase I (Roche Diagnostics GmbH), and streptomycin 100 µg/ml and penicillin 100 U/ml with orbital agitation for 30 minutes at 37°C. Cells from the lamina propria were collected by filtrating supernatant fractions with 70 µm nylon cell strainers that were then centrifuged to harvest cell suspensions.

Gene expression by RT-qPCR
Total RNA was isolated with TRIsure TM reagent (Bioline, London, UK) according to manufacturer´s instructions. For reverse transcription to cDNA, 2-1.5 μg of RNA were incubated with reverse transcription buffer, dNTPs (4 mM), random primers, and 50 units of MultiScribe TM Reverse Transcriptase at 25° C for 10 minutes followed by 120 minutes at 37°C and 5 minutes at 85°C. The qPCR was performed on a LightCycler® 480 Instrument (Roche, Boulogne-Billancourt, France). The reaction consisted of LightCycler 480 SYBR Green I Master mix (Roche, Branchburg, USA) and 300 nM of gene-specific primer pairs. The primer sequences of each gene are detailed in Appendix Table S2. Ribosomal protein L19 (Rpl19) was used as housekeeping gene. The qPCR program included the following steps: a denaturation step at 95°C for 10 minutes; followed by 45 amplification cycles (95°C for 10 seconds, 60°C for 30 seconds, and 72°C for 5 seconds) and a final melting stage at 95°C for 5 seconds and at 65°C for 1 minute. Variation of cDNA abundance was calculated according to the 2 −(ΔΔCt) method and represented as fold change expression relative to the control group.

Metabolites and cytokines in liver
Lipids from liver were extracted as previously described. 56,57 In brief, a solution of chloroform/ methanol (2:1) was used for tissue homogenization. After 3 h of shaking, milliQ water was added and the organic layer was separated by centrifugation (16000 g, 20 minutes), collected, and dried overnight. Then, triglycerides and free fatty acids were measured in the isolated organic layer using the Triglyceride colorimetric Assay Kit (Elabscience, USA) and the Free Fatty Acid Quantitation Kit (Sigma, Missouri, USA), respectively. Hepatic glycogen and glucose-6-phosphate (G6P) were quantified in the total homogenized tissue through the Glycogen Assay Kit (Sigma-Aldrich, Missouri, USA) and the Glucose-6-Phosphate Colorimetric Assay Kit (BioVision, California, USA), respectively. Cytokines (TNFα, IL10, IL6, and IFNɤ) were measured using ProcartaPlex multiplex immunoassays according to manufacturer´s instructions (Thermo Fisher Scientific, MA, USA)

Fatty acid analysis
Cecal content was homogenized by bead beating in 70%-isopropanol. 58 SCFA were quantified by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) upon derivatization to 3nitrophenylhydrazones. 58 Concentrations of total fatty acids were determined by gas chromatography coupled to mass spectrometry (GC-MS) after derivatization to fatty acid methyl esters (FAMEs). 59

Microbiota analysis
DNA of the cecal content was isolated using the QIAamp PowerFecal DNA kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. The DNA concentration was measured by UV methods (Nanodrop, Thermo Scientific, Wilmington, USA) and an aliquot of every sample was prepared at ~20 ng/µL with nuclease-free water for polymerase chain reaction (PCR). The V3-V4 hypervariable regions of the 16S ribosomal ribonucleic acid (rRNA) gene were amplified using 1 μL aliquot DNA and 25 PCR cycles consisting of the following steps: 95°C for 20 sec., 55°C for 20 sec., and 72°C for 20 sec. Phusion High-Fidelity Taq Polymerase (Thermo Scientific, Wilmington, USA) and the 6-mer barcoded primers, S-D-Bact-0341-b-S-17 (TAGCCTACGGGNGGCWGCAG) and S-D-Bact-0785-a-A-21 (ACTGACTACHVGG GTATCTAATCC), which target a wide repertoire of bacterial 16S rRNA genes, 60 were used for PCR. Dual-barcoded PCR products, consisting of ~500 bp, were purified from triplicate reactions with the Illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare, UK) and quantified through Qubit 3.0 and the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). The samples were multiplexed in one sequencing run by combining equimolar quantities of amplicon DNA (~50 ng per sample) and sequenced in one lane of the Illumina MiSeq platform with 2 × 300 PE configuration (Eurofins Genomics GmbH, Germany). Raw data were delivered in fastq files and pair ends with quality filtering were assembled using Flash software. 61 Sample de-multiplexing was carried out using sequence information from forward/reverse primers, the respective DNA barcodes, and the SeqKit suite of analysis. 62 After demultiplexing, the barcodes/primers were removed and sequences were processed for chimera removal using UCHIME algorithm 63 and the SILVA reference set of 16S sequences (Release 128). 64 Operational Taxonomic Unit (OTU) information was retrieved by using a rarefied subset of 17,000 sequences per sample, randomly selected after multiple shuffling (10,000X) from the original dataset, and the UCLUST algorithm implemented in USEARCH v8.0.1623. 65 Common alpha diversity descriptors including the Observed OTUs, Reciprocal Simpson's index, dominance, and phylogenetic distance (PD) were computed using QIIME v1.9.1. 66 For phylogenetic-based metrics, the OTUs sequences were aligned using the PyNAST algorithm implemented in QIIME and the SILVA aligned reference database, whereas the tree topology reconstruction was completed with the FastTree algorithm 67 using the generalized time-reversible (GTR) model and gamma-based likelihood. The evaluation of the community structure across the sample groups was performed with the Vegan R package through interpretative multivariate and constrained appraisal based on the distance-based redundancy analysis (dbRDA) (vegan:: dbrda function and "bray" method). Taxonomic identification of differentially abundant OTUs was achieved by submitting respective sequences to the SINA aligner web server (https://www.arb-silva.de/ aligner/) and using the SILVA database for classification.

Statistical analysis
G*Power 3.1.9.2 was used to calculate sample size and SPSS (IBM SPSS Statistics 24) to conduct statistics. Shapiro-Wilk test was employed to assess normality. Statistics of normal distributed data were performed using different parametric tests, including repeated measures two-way ANOVA for the analysis of multiple measures of the same variable taken on the same subjects over the time (body weight gain and OGTT); one-way ANOVA followed by post-hoc Tuckey test to compare controls (CD-fed mice orally administered the vehicle) vs HFHSD-fed mice, receiving or not the bacterium and/or WBE and two-way ANOVA restricted to HFHSD-fed mice to examine both, the main effect of each independent variable (B. uniformis or the WBE) and the interactions between them. For the analysis showing interactions between variables a post-hoc Tukey test was conducted. Kruskal-Wallis test followed by pairwise multiple comparisons was used to analyze non-normally distributed data. Bravais-Pearson´s correlation coefficient was used to test correlations between two variables. Only set of data with not significantly different slopes and intercepts, analyzed using linear regression, were combined to calculate a global correlation coefficient between variables.
Statistical analyses on microbiota data were done applying non-parametric methods such as Kruskal-Wallis and pairwise Wilcoxon Rank Sum tests (for unpaired samples) with Benjamini-Hochberg post hoc correction for multiple group comparison across the alpha diversity descriptors. Statistical differences in the microbial community structure were assessed by the permutation-based vegan::adonis function. Differential abundance of OTUs across groups was assisted by applying the Kruskal-Wallis test with Benjamini-Hochberg post hoc correction. Highly divergent OTUs in terms of abundance were selected when Kruskal-Wallis chi-squared test was ≥30 and corrected P ≤ 0.001. Additionally, pairwise Wilcoxon Rank Sum test with Benjamini-Hochberg correction was applied to identify precise sample groups showing up-or down-representation of different OTUs. Graphs and plots were drawn using ggplot2 and grid R v3.6 packages, Heatmap hierarchical clustering of OTUs (scaled read counts) was obtained using "correlation" as a distance and "complete" as clustering method. Results were not assessed by blinded investigators.