Bile acids regulate lipid metabolism through selective actions on fatty acid absorption

Summary

Intestinal lipid absorption, the entry-point for fats into the body, requires the coordinated actions of bile acids and lipases. Here, we uncover distinct yet cooperative roles of bile acids in driving the differential uptake of dietary fatty acids. We first decreased bile acid pool size by disrupting the rate-limiting enzyme in bile acid synthesis, Cyp7a1, using liver-directed gene editing in mice. Compared to lipase inhibition, reduced bile acids prevented diet-induced obesity, increased anorectic hormones, suppressed excessive eating, and improved systemic lipid metabolism. Remarkably, decreasing bile acids selectively decreased absorption of saturated fatty acids, but preserved polyunsaturated fatty acids. By targeting additional bile acid enzymes, we identified specific functions of individual bile acid species. Mechanistically, we show that cholic acid preferentially solubilizes polyunsaturated fatty acids into mixed micelles for intestinal uptake. Our studies demonstrate that bile acids can selectively control fatty acid uptake, revealing insights for future interventions in metabolic disease.

Graphical Abstract

eTOC blurb

As detergents, bile acids govern the intestinal uptake of dietary fats. By manipulating the bile acid pool size and composition, Chan et al. uncover how bile acids selectively regulate fatty acid absorption and gut hormone secretion to prevent obesity and metabolic disease.

Introduction

Dietary fat is a fundamental, energy-dense macronutrient necessary for survival. The absorption of dietary fat has evolved to become a highly efficient process, with a 95% absorption rate reported in humans.^1,2 However, in the Western world, where high-fat foods are readily available, this conserved efficiency for absorption can lead to excessive lipid accumulation, contributing to the development of obesity and associated metabolic disorders. A typical Western diet is high in fat with a predominance of long-chain saturated fatty acids over unsaturated fatty acids.³ Saturated fatty acids, which lack double bonds, contribute to inflammation and are frequently implicated in metabolic disease.⁴ In contrast, monounsaturated and polyunsaturated fatty acids, which contain one or more double bonds, are considered cardio- and hepatoprotective.^4,5 Fatty acids can also vary by acyl chain length, which confers specific physiological functions.⁶ Despite differences in structure and function, it remains uncertain whether—and to what extent—the absorption of these fatty acids differ.

The intestinal tract is the frontier for dietary fat absorption. As food enters the intestinal lumen, bile acids emulsify triglycerides, which represent the majority of dietary fats, into smaller lipid droplets. Gastrointestinal lipases hydrolyze triglycerides to liberate monoglycerides and free fatty acids that are then solubilized by bile acids into mixed micelles for uptake across the intestinal epithelium (Figure 1A). In the enterocyte, lipids are re-esterified and packaged into chylomicrons for delivery to peripheral tissues.⁷ Although absorption of dietary lipids is highly efficient, approximately 10% of food energy is lost by fecal excretion in healthy humans.⁸

Figure 1. Manipulating absorption by distinct mechanisms leads to differences in energy balance.

(A) Overview of dietary fat digestion and absorption.

(B) Diagram (left) of AAV-CRISPR tool. U6, U6 promoter; gRNA, guide RNA; HLP, hybrid liver promoter; SaCas9, Staphylococcus aureus Cas9. Western blot analysis (right) of hepatic CYP7A1 expression in representative mice.

(C) Total bile acid pool from gallbladder, liver, and small intestine in Cyp7a1 CRISPR mice. n = 9–10 mice per group. Lagodeoxycholic acid and 7-ketolithocholic acid not shown in the legend but included in the total bile acid pool analysis.

(D) Thin layer chromatography of fecal lipids from representative orlistat-treated mice. STD, standard; TG, triglyceride; FFA, free fatty acid; DG, diacylglyceride; MG, monoglyceride.

(E) Experimental schematics to assess energy balance in Cyp7a1 CRISPR (top) and orlistat-treated mice (bottom).

(F) Body mass for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 9–10 mice per group.

(G) Fecal energy loss for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 7–10 per group.

(H) Continuous (left) and average (right) energy expenditure during dark (grayed background) and light (white background) cycles over final one week of the experiment for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 7–8 mice per group. For the average energy expenditure graph, each data point represents one half-day timepoint (seven dark cycles, six light cycles).

(I) Food intake for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 7–8 mice per group.

MCA, muricholic acid; LCA, lithocholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; CA, cholic acid; UDCA, ursodeoxycholic acid; HDCA, hyodeoxycholic acid; T, taurine-conjugated.

Data are represented as mean ± standard error of mean (SEM). *p < 0.05, **p < 0.01, ***p < 0.001, by two-sided t-tests (C, G), multiple t-tests (C), two-way ANOVA with Šídák’s multiple comparisons test (F), ANCOVA using body mass as a covariate (H, I).

See also Figure S1.

As detergents, bile acids play a crucial role in the efficient uptake of dietary lipids.⁹ Primary bile acids are synthesized from cholesterol in the liver through a series of enzymatic steps that culminate with conjugation to an amino acid, like taurine or glycine, increasing their solubility.¹⁰ The bile acids can then be further transformed by the intestinal microbiome into secondary bile acids (Figure 1A).¹¹ The resulting group of bile acids are structurally diverse¹², but the precise function of individual bile acids species as detergent molecules is not well understood. In this study, we manipulated the size and composition of the bile acid pool to identify the specific roles of distinct bile acids in controlling dietary fat absorption. We show that modulations to the bile acid pool selectively alter fatty acid absorption and induce physiological protection against obesity and steatotic liver disease that are distinct from those achieved through lipase inhibition. These findings reveal a previously underrecognized mechanism by which bile acids regulate dietary fat uptake, leading to systemic shifts in lipid metabolism, and highlight the therapeutic potential of bile acid modulation for metabolic health and disease.

Results

Manipulating absorption by distinct mechanisms leads to differences in energy balance.

To study the role of bile acids in dietary fat absorption, we delivered Clustered Regularly Interspaced Short Palindromic Repeats Cas9 using adeno-associated virus (AAV-CRISPR) to disrupt cholesterol 7 alpha-hydroxylase (Cyp7a1), the first and rate-limiting enzyme in bile acid synthesis, in adult mice (Figure 1B).¹³ This approach enables targeted gene disruption in adult wild-type mouse livers, while also avoiding developmental abnormalities that can be associated with lifelong deletions in whole-body knockout mice.¹⁴ Cyp7a1 disruption was robust on a low-fat diet (Figure S1A). To assess how loss of this enzyme altered the bile acid pool, we measured bile acids from the liver, gallbladder, and small intestine, which together contribute the majority of bile acids to the total bile acid pool.¹² Cyp7a1 disruption reduced all bile acid species, including their taurine-conjugated derivatives, leading to an overall decrease in the total bile acid pool size (Figure S1B). Despite efficient loss of CYP7A1, a residual amount of bile acids remained likely due to induction of the alternative bile acid synthetic pathway (Figure 1A). To contrast the effects of bile acid pool modulation, we used the lipase inhibitor tetrahydrolipstatin (orlistat) as an alternative strategy to modify dietary fat absorption. Because orlistat blocks the hydrolysis of fatty acids from triglycerides, fatty acids cannot be absorbed.¹⁵

To explore how these manipulations alter energy homeostasis, we performed parallel experiments with Cyp7a1 CRISPR mice and orlistat-treated mice fed a Western diet for eight weeks. Loss of CYP7A1 was confirmed by western blot (Figure 1B), and bile acid pool size was reduced by 50% (Figures 1C and S1C). Consistent with its mechanism, orlistat-mediated lipase inhibition led to increased fecal diacylglycerides and triglycerides, which were nearly undetectable in controls (Figures 1D and S1D). Orlistat treatment increased the total bile acid pool size (Figures S1E and S1F) yet did not change bile acid signaling (Figure S1G), likely reflecting intestinal sequestration of bile acids due to lipase inhibition.¹⁶ After eight weeks of Western diet, Cyp7a1 CRISPR mice were protected from diet-induced body mass and fat mass gain, while maintaining lean mass. In contrast, orlistat-treated mice exhibited body composition comparable to controls (Figures 1E, 1F, S1H, and S1I). To assess absorption, total fecal energy output and density were quantified using oxygen bomb calorimetry (Figures S1J and S1K). Loss of CYP7A1 and orlistat treatment both increased fecal energy excretion (Figure 1G), suggesting reduced absorption. Despite these changes, energy expenditure was unchanged across groups (Figures 1H and S1L). The respiratory exchange ratio was higher in orlistat mice during the dark active cycle (Figure S1M), indicating decreased lipid utilization.¹⁷ Locomotion and water intake were similar across all groups (Figures S1N and S1O). Food intake was comparable between Cyp7a1 and control CRISPR mice, but higher in the orlistat mice compared to their controls (Figure 1I). Since reducing bile acids decreased energy absorption without affecting intake, we concluded that loss of CYP7A1 prevents weight gain primarily due to changes in absorption. In contrast, orlistat did not protect against obesity, likely because the reduced energy absorption was offset by the increased food intake. Reduced absorption should normally trigger a compensatory increase in food intake¹⁸, thus it was surprising that the Cyp7a1 CRISPR mice did not eat more, unlike orlistat-treated mice.¹⁶

Given the weight protection conferred by loss of CYP7A1, we next tested whether it could also treat obesity. After eight weeks of Western diet, obese mice were administered either control or Cyp7a1 CRISPR (Figure S1P). Cyp7a1 CRISPR treatment blunted further increases in body and fat mass, which persisted for eight weeks after AAV-CRISPR injection (Figure S1Q–S). Fecal energy loss was greater in Cyp7a1 CRISPR mice (Figure S1T), suggesting that decreasing bile acids with loss of CYP7A1 not only prevents, but also effectively treats diet-induced obesity through increased energy excretion. These findings, together with our metabolic cage studies showing that Cyp7a1 CRISPR mice did not exhibit a compensatory increase in food intake, prompted us to investigate how alterations in intestinal absorption might modulate appetite.

Unabsorbed lipids promote enteroendocrine incretin hormone secretion.

To determine what factors might be controlling the differences in food intake between Cyp7a1 CRISPR and orlistat-treated mice, we measured circulating levels of hormones known to regulate hunger and satiety (Figure 2A). The orexigenic hormone, ghrelin, increased during fasting and dropped after ad libitum refeeding in all treatment groups (Figure 2B). In contrast, the anorectic hormone, leptin, was suppressed during fasting but amplified after refeeding in all groups (Figure 2C). Since leptin is produced by adipose tissue in proportion to body fat¹⁹, the lower leptin levels in the Cyp7a1 CRISPR mice during both fasting and refeeding likely reflect their reduced body fat mass (Figure S1H). The postprandial secretion of incretin peptides, glucagon-like peptide-1 (GLP-1) and peptide YY (PYY), was significantly augmented with loss of CYP7A1, while GLP-1 was completely unaffected and PYY only modestly increased with orlistat (Figures 2D and 2E). Given the known satiety effects of these hormones²⁰, this may explain why the Cyp7a1 CRISPR mice did not increase food consumption, while the orlistat-treated mice did.

Figure 2. Unabsorbed lipids promote enteroendocrine incretin hormone secretion.

(A) Experimental schematics to measure plasma hormone levels in Cyp7a1 CRISPR (top) and orlistat-treated mice (bottom) following fast and refeed.

(B-E) Plasma concentrations of (B) ghrelin, (C) leptin, (D) GLP-1, and (E) PYY during fasted and refed states for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 6–8 mice per group.

(F) Visualization of neutral lipid accumulation in cross sections of small intestinal Swiss rolls from representative mice. Schematic (left) and confocal immunofluorescence imaging of intestinal Swiss roll (right). LipidTOX (red); DAPI (blue). Scale bar is 1000 μm.

(G) Experimental schematic to measure plasma GLP-1 in Gpr120^−/− mice treated with Cyp7a1 CRISPR.

(H) Western blot analysis of hepatic CYP7A1 expression in representative Gpr120^−/− mice treated with Cyp7a1 CRISPR.

(I) Plasma concentration of GLP-1 in Gpr120^−/− mice treated with Cyp7a1 CRISPR following fast and refeed. n = 10 mice per group.

(J) Experimental schematic to profile the microbiome in Cyp7a1 CRISPR mice by metagenomic shotgun sequencing.

(K) Relative abundance of taxa at the phylum level in microbiomes of Cyp7a1 CRISPR mice. n = 9–10 mice per group.

(L) Alpha diversity at the species level was assessed by Shannon index (left) and number of observed species (right) in microbiomes of Cyp7a1 CRISPR mice. n = 9–10 mice per group.

H, hour; Sac, sacrifice; F, fasted; R, refed; GLP-1, glucagon-like peptide-1; PYY, peptide YY.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, two-sided paired t-tests between fasted and refed states within groups (B-E, I), multiple t-tests (K), two-sided unpaired t-tests (L). ^#p < 0.05, ^##p < 0.01, ^###p < 0.001, two-sided unpaired t-tests between control and treatment groups for fasted and refed states (B-E, I). Color of the significance annotation indicates the group with the greater mean value.

See also Figure S2.

To investigate how the release of these hormones may be regulated, we considered specific dietary nutrients that are known to influence the secretion of gut hormones. Secretion of GLP-1 and PYY is stimulated by carbohydrates²¹, therefore we tested whether the glucose-dependent release of these hormones was different between the Cyp7a1 CRISPR and orlistat-treated mice (Figure S2A). Our results showed increased GLP-1 levels following the glucose challenge in all groups, but no differences between the Cyp7a1 CRISPR and orlistat groups (Figure S2B). Interestingly, despite similar GLP-1 responses to glucose, Cyp7a1 CRISPR mice displayed improved glucose tolerance (Figure S2C), demonstrating that a reduced bile acid pool has metabolic benefits beyond incretin release. Finally, the glucose challenge had minimal effects on ghrelin or PYY secretion (Figures S2D and S2E). Taken together, these findings suggest that other dietary components, likely lipids, may be driving the enhanced gut hormone secretion in Cyp7a1 CRISPR mice.

Since lipid-sensing receptors regulating incretin secretion are regionally distributed along the gastrointestinal tract²², we next investigated whether site-specific differences in lipid uptake could explain the differences in GLP-1 and PYY secretion. To test this, we performed lipid fluorescent staining (LipidTOX) on small intestinal sections to visualize postprandial neutral lipid uptake in mice with reduced bile acids or lipase inhibition. In the control groups, lipid uptake was primarily observed in the proximal and mid segments of the small intestine. However, loss of CYP7A1 led to lipid uptake in the more distal regions, indicating delayed uptake. In contrast, orlistat treatment showed minimal, scattered lipid accumulation throughout the intestine, consistent with a broad decrease in lipid uptake (Figures 2F, S2F, and S2G). Given increased amounts of free fatty acids observed in the feces of Cyp7a1 CRISPR mice (Figure S2H), we hypothesized that the smaller bile acid pool would promote the delivery of free fatty acids to the distal intestine, where GLP-1 and PYY-secreting L-cells are most concentrated.²³

Free fatty acids are known to trigger gut hormone secretion by activating G protein-coupled receptors (GPCRs) on enteroendocrine cells.⁶ Although GPCRs respond to different types of fatty acids, many share ligands, suggesting a degree of functional redundancy that might depend on the type and intestinal location of unabsorbed lipids. Among these, GPR120 is a key receptor for long-chain fatty acids and is highly expressed in distal intestinal L-cells.²⁴ To test whether the enhanced GLP-1 response observed in Cyp7a1 CRISPR mice were mediated by GPR120, we evaluated hormone secretion in Gpr120^−/− mice with Cyp7a1 CRISPR (Figure 2G and 2H). In the absence of GPR120, the increase in GLP-1 seen in Cyp7a1 disruption (Figure 2D) was significantly blunted (Figure 2I), suggesting that GPR120 is at least partially responsible for this response.

In addition to enteroendocrine cells, the intestinal microbiome is a key responder to luminal nutrients.²⁵ Moreover, the microbiota are responsible for chemically modifying host bile acids into secondary bile acids.¹² To assess whether the influx of luminal lipids and the decrease in bile acids by loss of CYP7A1 would influence microbial composition, we performed metagenomic shotgun sequencing in cecal contents from control and Cyp7a1 CRISPR mice (Figure 2J). At the phylum level, we found that loss of CYP7A1 modestly increased Firmicutes and significantly decreased Verrucomicrobia (Figure 2K and S2I). These phylum level shifts resemble the microbial signatures observed in high-fat diet feeding^26,27, suggesting that the microbiota of mice with loss of CYP7A1 are likely a reflection of the increased luminal lipid exposure. At the species level, we observed no significant changes in alpha diversity and observed species (Figures 2L and S2J), and no clear clustering between individual microbiomes when measuring beta diversity by Bray-Curtis dissimilarity (Figure S2K), indicating that the overall microbiota community remained largely unchanged. Despite this, differential abundance analysis using MaAsLin2²⁸ identified Adlercreutzia equolifaciens as the most significantly changed taxa that has been fully characterized in this dataset. Adlercreutzia equolifaciens is an anti-inflammatory commensal bacterium that is depleted in patients with metabolic dysfunction-associated steatotic liver disease (MASLD).²⁹ Interestingly, its abundance was increased with loss of CYP7A1 (Figure S2L), suggesting a potentially favorable microbial shift. However, since both alpha and beta diversity analyses do not highlight clear differences, we conclude that the microbiome likely responds, but does not contribute, to the host metabolic phenotype seen with loss of CYP7A1.

Taken together, these data highlight distinct mechanisms of intestinal lipid absorption that differentially stimulate endogenous gut hormone secretion. By reducing bile acids, we can enhance the influx of free fatty acids to the distal intestine, increasing the strength of the fed response and inducing GLP-1 and PYY to promote satiety. The GLP-1 response is driven, at least in part, by free fatty acid activation of GPR120. While orlistat treatment also leads to high amounts of unabsorbed lipids (Figure S2H), these lipids are likely sequestered within the undigested lipid mass, preventing interaction with enteroendocrine cells and thus limiting hormone release as they pass through the intestine, an observation reported previously in healthy humans after oral administration of orlistat.³⁰

Although bile acids, such as lithocholic acid (LCA), can also activate the G-protein-coupled bile acid receptor 1 (GPBAR1, also known as TGR5) on enteroendocrine cells to secrete incretin hormones³¹, this is unlikely due to the decreased bile acids caused by CYP7A1 loss. While the gut microbiota has also been shown to modulate GLP-1 regulation through microbial metabolites³², no landmark changes in the microbiota composition were observed, making a microbiota-driven mechanism less likely. Collectively, our studies reveal that bile acid reduction via CYP7A1 loss and lipase inhibition exert distinct effects on body weight, food intake, and gut hormone secretion through unique mechanisms of action on fat absorption.

Targeting dietary fat absorption modulates systemic lipid metabolism.

After establishing the effects of fat absorption on energy balance and gut hormones, we investigated its impact on liver lipid metabolism, the central hub for lipid uptake and redistribution.⁷ Histological analysis of livers with loss of CYP7A1 showed decreased hepatic fat deposition with no signs of hepatocyte injury (Figures 3A and S3A). Liver lipidomic profiling also showed reduced total liver lipids, particularly triglycerides (Figure 3B). Further analysis of the individual triglyceride species revealed that loss of CYP7A1 decreased 311 species, but to our surprise, increased 22 triglyceride species. Of the increased triglyceride species, nearly all contained polyunsaturated fatty acids, specifically arachidonic acid (C20:4) (Figures 3C and S3B), which is an important lipid that is largely synthesized from the essential fatty acid, linoleic acid (C18:2).³³ Analysis of total liver fatty acid composition further corroborated this increase in total C20:4 (Figures 3D and S3C). Similar to loss of CYP7A1, orlistat also reduced liver triglyceride accumulation on both histological and lipidomic analyses (Figures 3E and 3F) without hepatocyte injury (Figure S3D). Detailed analysis of triglyceride species demonstrated a decrease of 398 triglyceride species by orlistat compared to control (Figure 3G), but unlike loss of CYP7A1, no triglyceride species were increased. Further quantification of fatty acid composition revealed that all fatty acids, including the polyunsaturated fatty acids, were either unchanged or decreased by orlistat (Figures 3H and S3E).

Figure 3. Targeting dietary fat absorption modulates systemic lipid metabolism.

(A and E) Histological images of representative liver sections stained with hematoxylin and eosin from (A) Cyp7a1 CRISPR and (E) orlistat-treated mice. Scale bar is 100 μm.

(B and F) Lipidomic analyses of hepatic lipid classes from (B) Cyp7a1 CRISPR and (F) orlistat-treated mice. n = 8–10 mice per group. Directionality of triangle indicates significantly increased or decreased compared to respective controls.

(C and G) Lipidomic analyses of hepatic TG species from (C) Cyp7a1 CRISPR and (G) orlistat-treated mice. Dashed line on volcano plot is the threshold at which differences reach statistical significance. TG species that are significantly different are colored yellow, TG species not significant are colored grey. PUFA-containing TG species that are significantly increased in Cyp7a1 CRISPR mice are colored blue. n = 8–10 mice per group.

(D and H) Bubble plot of fold change in hepatic fatty acid abundance from (D) Cyp7a1 CRISPR or (H) orlistat-treated mice. n = 8–10 mice per group. Bubble color and size indicate significance.

(I and L) Hepatic mRNA expression of Acaca, Fasn, and Scd1 in (I) Cyp7a1 CRISPR and (L) orlistat-treated mice. n = 7–10 mice per group.

(J and M) Western blot analyses of hepatic ACC, FASN, and SCD1 expression in representative (J) Cyp7a1 CRISPR and (M) orlistat-treated mice.

(K and N) Percentage of total newly synthesized deuterium-labeled fatty acids in livers of (K) Cyp7a1 CRISPR and (N) orlistat-treated mice. n = 10 mice per group.

(O-Q, S) Lipidomic analyses of plasma and perigonadal white adipose tissue for (O, Q) total TG concentration and (P, S) TG species abundance by fold change in Cyp7a1 CRISPR mice. n = 7–9 mice per group.

(R) Histological images of representative perigonadal white adipose tissue sections stained with hematoxylin and eosin from Cyp7a1 CRISPR mice. Scale bar is 100 μm.

TG, triglyceride; SM, sphingomyelin; PS, phosphatidylserine; PI, phosphatidylinositol; PG, phosphatidylglycerol; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PA, phosphatidic acid; LacCER, lactosylceramide; LPE, lysophosphatidylethanolamine; LPC, lysophosphatidylcholine; HexCER, hexosylceramide; FFA, free fatty acid; DG, diacylglyceride; Cer d18:1, ceramide 18:1; Cer d18:0, ceramide 18:0; CE, cholesterol ester; PUFA, polyunsaturated fatty acid; MUFA, monounsaturated fatty acid; SFA, saturated fatty acid.

Data are represented as mean ± SEM (B, F, I, K, L, N, O, Q), fold change (C, D, G, H, P, S). *p < 0.05, **p < 0.01, ***p < 0.001, by two-sided t-tests (B, F, I, K, L, N, O, Q), multiple t-tests (B, D, F, H).

See also Figure S3, Table S1, Table S2.

Polyunsaturated fatty acids possess integral structural and signaling properties for a wide array of physiological processes.³³ In the liver, polyunsaturated fatty acids can inhibit lipogenesis by blocking the proteolytic processing of sterol regulatory element-binding protein-1c (SREBP-1c), a master regulator of lipogenesis.^34,35 We therefore hypothesized that the relative deficiency of hepatic polyunsaturated fatty acids by orlistat may promote lipogenesis, while the increase in hepatic polyunsaturated fatty acids, specifically C20:4, from loss of CYP7A1 may not. To evaluate lipogenesis in the liver, we first measured the mRNA and protein expression of key lipogenic enzymes acetyl-CoA carboxylase (ACC), fatty acid synthase (FASN), and stearoyl-CoA 9-desaturase (SCD1).³⁴ In line with our hypothesis, loss of CYP7A1 did not significantly increase lipogenic mRNA or protein expression (Figures 3I and 3J, S3F). To directly measure de novo lipogenesis, newly synthesized fatty acids were quantified using deuterium labeling (Figure S3G).

Cyp7a1 disruption resulted in only a slight increase in deuterated fatty acids (Figures 3K and S3H). In contrast, orlistat treatment markedly upregulated lipogenic mRNA and protein levels (Figure 3L and 3M, S3F), accompanied by a substantially greater increase in newly synthesized fatty acids (Figures 3N and S3I). To determine whether the increase in de novo lipogenesis by orlistat was mediated by SREBP-1c, we treated wild-type (Srebp-1c^+/+) and Srebp-1c^−/− mice with orlistat and measured lipogenesis after Western diet feeding (Figure S3J). The increases in mRNA and protein levels and newly synthesized fatty acids observed with orlistat were abolished with loss of SREBP-1c (Figures S3K–O). Since insulin resistance is also known to activate SREBP-1c–mediated lipogenesis³⁴, we measured plasma insulin levels and found them to be comparable between groups (Figure S3P). This suggests that the increase in lipogenesis observed in orlistat-treated mice was independent of insulin signaling and likely driven by the depletion of polyunsaturated fatty acids. This contrasts with Cyp7a1 disruption, which enriches the liver with polyunsaturated fatty acids and suppresses lipogenesis, underscoring their opposing effects on hepatic lipid composition and synthesis.

The shifts in the hepatic lipidome prompted us to next ask whether these metabolic changes resulted from changes in the hepatic transcriptome, specifically with loss of CYP7A1, since bile acids are known to influence gene expression by binding to nuclear receptors.¹² To assess this, we performed bulk liver RNA sequencing. Differential gene expression between Cyp7a1 and control CRISPR mouse livers was identified using DESeq2³⁶ (12,031 genes), revealing subtle changes of only 11 downregulated and 18 upregulated genes that were significant (Figure S3Q). Additionally, gene set enrichment analysis (GSEA)^37,38 revealed minimal enrichment in hallmark gene sets³⁹, with only four pathways exceeding our normalized enrichment score (NES) threshold of 1.5. (Figure S3R). Since cholesterol homeostasis ranked highest in our analysis and relates to lipid absorption, we examined which individual genes contributed to this pathway’s enrichment. We found that nearly all cholesterol biosynthesis genes were upregulated (Table S1). Using the same deuterium-based assay described earlier, we confirmed that hepatic cholesterol biosynthesis was significantly increased (Figure S3S). Given the role of bile acids in promoting cholesterol absorption⁴⁰, we hypothesized that the increase in synthesis was a compensatory response to impaired absorption from decreased bile acids. Supporting this, fecal cholesterol was elevated in Cyp7a1 CRISPR mice (Figure S3T). Plasma cholesterol was also elevated, likely reflecting increased hepatic cholesterol output from enhanced synthesis (Figure S3U). This increase in export may also account for the reduced hepatic cholesterol content (Figure 3B). These findings align with the report of hypercholesterolemia in human patients with homozygous Cyp7a1 mutations⁴¹, though the cholesterol changes that we observed were only marginal.

To determine whether the metabolic changes in the liver were also reflected in peripheral tissues, we examined the lipid composition in plasma and adipose tissue of Cyp7a1 CRISPR mice. Plasma lipidomic analysis revealed overall decreased triglyceride levels (Figure 3O), but saturated fatty acid-containing triglycerides were reduced much more than polyunsaturated fatty acid-containing triglycerides, some of which were also increased in Cyp7a1 CRISPR mice (Figure 3P). In perigonadal white adipose tissue, total triglyceride levels were also reduced (Figure 3Q), consistent with histological analyses (Figure 3R). Although nearly all triglyceride species decreased, those containing saturated fatty acids were disproportionately reduced, while polyunsaturated species were relatively spared (Figure 3S). These patterns in plasma and adipose tissue mirrored the liver, indicating a systemic enrichment of polyunsaturated over saturated fatty acids in the setting of reduced bile acid levels.

Selective fatty acid absorption following the reduction of bile acids.

Given the coordinated changes in fatty acid composition in the liver, plasma, and adipose tissue, we next investigated whether these shifts were due to specific changes in fatty acid absorption. To interrogate this, we incorporated a nonabsorbable lipid, sucrose polybehenate (SPB), into the Western diet to measure the absorption of individual fatty acid species over time (Figure 4A). SPB resembles a triglyceride but contains predominantly behenic acids (C22:0) esterified to sucrose instead of glycerol.⁴² Because its fatty acids cannot be hydrolyzed by lipases or absorbed, SPB serves as a marker to normalize fecal fatty acid measurements and food intake. We first quantified the cumulative total fatty acid content of the feces over the course of the experiment. Mice with loss of CYP7A1 caused a striking increase in cumulative fecal fatty acid content, and orlistat treatment displayed an even greater cumulative fecal fatty acid content (Figure 4B). These findings were consistent with previous fecal thin layer chromatography, calorimetry, and fecal lipidomics data (Figures 1D, 1G, S2H), establishing that the observed energy loss likely stemmed from decreased dietary fat uptake.

Figure 4. Selective fatty acid absorption following the reduction of bile acids.

(A) Experimental schematics to measure fatty acid absorption in Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. Sucrose polybehenate was added to the diet at timepoints indicated by chevrons.

(B) Cumulative fatty acid excretion for Cyp7a1 CRISPR (top) and orlistat-treated (bottom) mice. n = 6–10 mice per group.

(C and D) Absorption of (C) total fatty acids (FA) and (D) saturated, monounsaturated, and polyunsaturated fatty acids in Cyp7a1 CRISPR (top) and orlistat-treated mice (bottom). n = 7–10 mice per group.

Data are represented as mean ± SEM. **p < 0.01, ***p < 0.001, by two-way ANOVA with Šídák’s multiple comparisons test (B), two-sided t-tests (C, D). Color of significance annotation indicates the group with the greater mean value.

See also Figure S4.

We then determined the percent absorption of individual fatty acids. The absorption of total fatty acids was significantly decreased in both Cyp7a1 CRISPR and orlistat-treated mice at all time points (Figures 4C and S4A). However, analysis of individual fatty acid absorption exposed key differences between bile acid- versus lipase-mediated absorption. Cyp7a1 CRISPR led to large reductions in the absorption for saturated fatty acids with differences magnifying as carbon chain length increases (C14:0, C16:0, C18:0). In contrast, there were smaller reductions in absorption for monounsaturated fatty acids (C16:1, C18:1) and almost no apparent changes for polyunsaturated fatty acids (C18:2, C18:3) (Figure 4D). These data, which were observed for every time point (Figure S4B), indicate a selectivity in fatty acid absorption, with increased absorption efficiency for fatty acids with more double bonds and shorter acyl chain length. The fecal fatty acid composition did not correlate with their abundance in the diet (Figure S4C), suggesting that the specific changes in fatty acid absorption were not simply the result of dietary lipids overloading the intestinal absorptive capacity. In contrast to the bile acid manipulation, orlistat treatment led to a more pronounced reduction of all fatty acids, especially the polyunsaturated fatty acids (Figure 4D). These changes in fatty acid absorption were consistent across all time points (Figure S4D). Thus, although both orlistat and loss of CYP7A1 reduced the absorption of lipids, there were clear differences in the type of fatty acid being absorbed. Orlistat treatment decreased the absorption of all fatty acids, whereas loss of CYP7A1 resulted in almost normal absorption of polyunsaturated fatty acids but disproportionately decreased absorption of saturated fatty acids, underscoring a selectivity in fatty acid absorption. These findings were corroborated by further fecal lipidomic analysis, which revealed significantly greater fecal losses of unsaturated fatty acids in the orlistat-treated mice compared to the mice lacking CYP7A1 (Figure S4E). The differential absorption of fatty acids by bile acid modulation mirrors the changes in liver fatty acids (Figure 3D), particularly for C20:4-containing lipids in the liver, which likely accumulated due to preserved uptake of polyunsaturated fatty acids in the intestine. However, since C20:4 is derived from C18:2⁴ and present in only trace amounts in the diet (Figure S4C), we reasoned that the increase in hepatic C20:4-containing triglyceride species was due to the retained intestinal uptake of C18:2 and subsequent conversion of C18:2 into C20:4 in the liver. Consistent with this, our bulk liver RNA sequencing analysis identified upregulation of specific genes (Fads1, Fads2, Elovl2, Elovl5) involved in the elongation and desaturation steps leading to C20:4 synthesis (Table S2).

Fatty acids are differentially solubilized in bile.

We next sought to determine the molecular mechanism by which fatty acids are differentially absorbed. In the small intestine, micellar solubilization increases the aqueous concentration of fatty acids by 1,000-fold and accelerate their diffusion by 100-fold.⁴³ This led us to hypothesize that the differential absorption of fatty acids is mediated by physiologic differences in their incorporation into mixed micelles that also contain bile acids. To test whether bile itself is required for selective fatty acid absorption, we developed an in vitro fluorescent-based micelle formation assay to test how much physiologic bile is required to solubilize different fatty acids (Figure 5A). We pooled bile from gallbladders of wild-type mice (Figures 5B and S5A) to mimic the bile that is secreted into the proximal gut for lipid absorption.⁴⁴ We then incubated different amounts of gallbladder bile with a set amount of individual fatty acids (C14:0, C16:0, C18:0, C18:1, C18:2, C18:3) to assess the bile’s capacity to solubilize the distinct lipids. In line with prior results, the solubility of fatty acids varied with acyl chain length and degree of unsaturation. Specifically, saturated fatty acids required a greater volume of physiologic bile to micellize than mono- and polyunsaturated fatty acids, confirming that this natural selectivity in fatty acid solubilization is conferred by mouse bile (Figures 5C, 5D, S5B).

Figure 5. Fatty acids are differentially solubilized in bile.

(A) Experimental schematic to determine the volume of mouse or human gallbladder bile required to solubilize equal amounts of fatty acids varying in acyl chain length and degree of unsaturation. Above a critical volume of bile, the bile, fatty acid, and fluorophore aggregate and increase fluorescence emission upon micelle formation.

(B and E) Pie charts of bile acid composition from (B) gallbladder bile pooled from wild-type mice (n = 37) and (E) human gallbladder bile (n = 1). Lagodeoxycholic acid and 7-ketolithocholic acid are not shown in the legend but included in the mouse bile acid pool analysis.

(C and F) Scatter plots of (C) mouse and (F) human bile volume against fluorescence intensity. Inflection point (red) represents the minimum volume of bile required to micellize in an aqueous solution.

(D and G) Table (top) of critical volumes of (D) mouse and (G) human bile mixed with free fatty acids C18:0, C18:1, C18:2, and resulting scatter plots (bottom).

MCA, muricholic acid; CA, cholic acid; CDCA, chenodeoxycholic acid; LCA, lithocholic acid; DCA, deoxycholic acid; HDCA, hyodeoxycholic acid; UDCA, ursodeoxycholic acid; T, taurine-conjugated; G, glycine-conjugated.

See also Figure S5.

Given known differences in bile composition between mice and humans¹², we subsequently explored whether similar selectivity exists in human bile. We obtained gallbladder bile from an otherwise healthy human patient following cholecystectomy and quantified the bile acid composition. Unlike mice, humans lack muricholic acids (MCAs) and contain more glycine-conjugated bile acids and secondary bile acids (Figure 5E). The human sample required less bile than mice to self-aggregate into micelles (Figure 5F). We then incubated the bile with different fatty acids and again found that saturated fatty acids needed more bile to solubilize than unsaturated fatty acids for the same acyl chain length (Figures 5G and S5C). Together, these findings show that the differential solubilization of fatty acids in bile is conserved in humans. However, it is noteworthy that significantly less human bile was required to achieve comparable solubilization of fatty acids, suggesting that the components in human bile may possess superior solubilizing properties relative to mouse bile. This distinction underscores potential interspecies differences in bile composition that may influence lipid absorption efficiency.

Fatty acid absorption changes with alterations in bile acid pool composition.

As bile acids are the major constituents of bile⁴⁵, we hypothesized that the selectivity in fatty acid absorption observed in vivo is driven primarily by bile acids. However, bile acids are a heterogeneous group of molecules, each with distinct molecular structures and functions, which may differentially influence fatty acid uptake. To determine whether a specific bile acid species is responsible for this selectivity, we modulated the composition of the bile acid pool in vivo to assess the impact of distinct bile acids on absorption. To achieve this, we targeted specific bile acid synthetic enzymes using AAV-CRISPR to modulate the bile acid pool composition. We first targeted cytochrome P450, family 8, subfamily B, polypeptide 1 (CYP8B1), a 12α-hydroxylase that is required for the synthesis of cholic acid (CA) and dictates the ratio of 12α-hydroxylated to non-12α-hydroxylated bile acids.⁴⁶ We also targeted cytochrome P450 family 2 subfamily c polypeptide 70 (CYP2C70), a rodent-specific 6β-hydroxylase that is responsible for synthesizing MCAs from chenodeoxycholic acid (CDCA).⁴⁷ Lastly, we targeted cytochrome P450, family 2, subfamily A, polypeptide 12 (CYP2A12), a 7α-hydroxylase that converts the microbially-derived secondary bile acids, deoxycholic acid (DCA) and LCA, from enterohepatic circulation back into the primary bile acids, CA and CDCA, respectively (Figure 6A).⁴⁸

Figure 6. Fatty acid absorption changes with alterations in bile acid pool composition.

(A) Overview of hepatic bile acid synthesis pathway.

(B) Experimental schematic (top) to measure fatty acid absorption in Cyp8b1 CRISPR, Cyp2c70 CRISPR, and Cyp2a12 CRISPR mice. Donut chart of total bile acid pool from the gallbladder, liver, and small intestine for each respective group (bottom). n = 9–10 mice per group.

(C) Absorption of total fatty acids in Cyp8b1 CRISPR, Cyp2c70 CRISPR, and Cyp2a12 CRISPR mice. n = 7–10 mice per group.

(D) Experimental schematic (top) to measure fatty acid absorption in Cyp7a1 CRISPR and Cyp7a1 + Cyp2c70 CRISPR mice supplemented with CA or CDCA. Donut chart of total bile acid pool from the gallbladder, liver, and small intestine for each respective group (bottom). Size of donut charts are proportional to total bile acid pool size. n = 4 mice per group.

(E) Absorption of total fatty acids in Cyp7a1 CRISPR and Cyp7a1 + Cyp2c70 CRISPR mice supplemented with CA or CDCA. Text inside each bar denotes what dietary bile acid was supplemented. For Cyp7a1 CRISPR mice supplemented with CDCA, the CDCA is converted to MCA by endogenous CYP2C70, which is reflected in the total bile acid pool (see Figure 6D). n = 3–4 mice per group.

MCA, muricholic acid; CA, cholic acid; CDCA, chenodeoxycholic acid; LCA, lithocholic acid; DCA, deoxycholic acid; HDCA, hyodeoxycholic acid; UDCA, ursodeoxycholic acid; T, taurine-conjugated. Lagodeoxycholic acid and 7-ketolithocholic acid are not shown in the legends but included in the total bile acid pool analyses (B, D).

Data are represented as a percentage (B, D), mean ± SEM (C, E). ***p < 0.001, by one-way ANOVA. ^#p < 0.05, ^###p < 0.001, by one-way ANOVA between control CRISPR without dietary bile acids and all treatment conditions. Color of the significance annotation indicates the group with the greater mean value.

See also Figure S6.

To confirm efficient knockout, we measured hepatic protein levels of CYP8B1 and CYP2C70 and found undetectable levels in their respective knockout groups (Figure S6A). Because an antibody specific to CYP2A12 does not exist, we used gene expression analysis to confirm efficient loss of Cyp2a12 (Figure S6B). All three AAV-CRISPR treatments resulted in the expected changes in bile acid pool composition. Loss of CYP8B1 caused large reductions in CA and shifted the bile acid composition toward more α/β-MCAs and CDCA without altering overall pool size (Figures 6B and S6C). Loss of CYP2C70 decreased MCAs and increased CDCA, suggestive of a more humanized bile acid pool (Figures 6B and S6D). Loss of CYP2A12 resulted in a bile acid pool with a predominance of DCA (Figures 6B and S6E).

To determine how these different bile acid compositions altered lipid absorption, we measured fatty acid absorption in these mice at endpoint using the SPB assay described earlier.⁴² Loss of CYP8B1 moderately reduced total fatty acid absorption (Figure 6C), although the magnitude was smaller than the reduction observed in loss of CYP7A1 (Figure 4C). It also mirrored the pattern seen with CYP7A1 loss (Figure 4D), showing preserved mono- and polyunsaturated fatty acid absorption with a disproportionate reduction in saturated fatty acids (Figure S6F). Despite lower conjugated CA and MCA, loss of CYP2C70 maintained normal total absorption (Figure 6C) with only slight decreases in mono- and polyunsaturated fatty acid absorption (Figure S6G), suggesting that the increase in CDCA may compensate for the bile acids that decrease. Loss of CYP2A12 similarly had no effect on overall fatty acid absorption (Figure 6C) and only subtle decreases in mono- and polyunsaturated fatty acid absorption (Figure S6H) despite the increase in DCA at the expense of CA. These findings reveal that individual bile acid species vary in their capacity to promote fatty acid absorption, underscoring the importance of both bile acid pool size and composition in this process.

To further dissect the contribution of individual bile acid species to fatty acid absorption, we systematically decreased the bile acid pool in vivo using AAV CRISPR targeting Cyp7a1 and then administered back bile acids that were deficient in the pool in the diet (Figure 6D). To isolate the effect of CA, we supplemented it into the Western diet of Cyp7a1 CRISPR mice. Loss of CYP7A1 reduces the overall bile acid pool, particularly CA, while preserving MCA. Repleting CA restored the bile acid pool and fatty acid absorption to profiles comparable to control mice (Figures 6D, 6E, S6I–K), confirming its strong detergent capacity in promoting dietary fat uptake. In contrast, supplementing CDCA, which is rapidly converted to MCA in the bile acid pool, had minimal impact on absorption (Figures 6D, 6E, S6I–K), suggesting that MCA is a weaker detergent and insufficient to compensate for CA deficiency in mice with loss of CYP7A1. To directly test the role of CDCA, we supplemented it into the Western diet of mice with co-disruption of Cyp7a1 and Cyp2c70. In this model, CA is depleted and CDCA accumulates due to impaired conversion to MCA (Figures 6D, S6I, S6J). CDCA rescued total fatty acid absorption (Figure 6E, S6K), indicating that CDCA, like CA, enhances fatty acid uptake. Together, these findings suggest that CA and CDCA are effective detergents for dietary fat absorption, whereas MCA is unable to mediate selective fatty acid absorption.

Fatty acids are differentially solubilized by bile acids into mixed micelles.

To determine which specific bile acids were driving the selective uptake of fatty acids, we modified the previous micelle formation assay (Figure 5A). Here, we interrogated the effects of each fatty acid on bile acid micelle formation by calculating a critical micelle concentration (CMC) for each bile acid (Figure 7A).⁴⁹ The CMC is the minimum concentration of detergent needed to form micelles in aqueous solution.⁹ Below the CMC, detergent molecules disperse; above the CMC, they aggregate into micelles. A lower CMC indicates more efficient solubilization and thus a more potent detergent. First, we measured CMCs of individual bile acid species, without a fatty acid present. The hydroxyl groups on a bile acid are thought to significant influence the CMC because they disturb the hydrophobic interactions between individual bile acid molecules.⁵⁰ The range in CMC values for the different bile acid species indicate that the number, position, and orientation of the polar hydroxyl groups in bile acids are indeed important for micellization (Figure S7A).^50,51,52,53 Next, we mixed individual bile acids with equal amounts of fatty acids varying in double bond number and acyl chain length. We focused on taurine-conjugated CA, β-MCA, CDCA, and DCA, the major constituents of the bile acid pool in mice (Figure 5B).¹² For conjugated CA, the CMCs incrementally increased when the bile acids were mixed with longer chain fatty acids (C14:0 < C16:0 < C18:0), suggesting that conjugated CA preferentially micellizes shorter-chain fatty acids. Contrarily, the CMCs of conjugated CA decreased when fatty acids of the same chain length but increasing double bonds were added (C18:0 > C18:1 > C18:2 or C18:3) (Figures 7B and 7C). This suggests a selectivity towards unsaturated fatty acids, indicating that conjugated CA is a more effective solubilizer of unsaturated fatty acids. Conjugated DCA and CDCA showed no specificity and micellized all fatty acids with equivalent efficiency. Meanwhile, conjugated β-MCA was a poor solubilizer of all fatty acids, compared to other bile acids (Figures 7C and S7B). Because humans have more glycine-conjugated bile acids than mice (Figure 5E)¹², we also determined the CMCs for these bile acids. The CMCs of glycine-conjugated CA, DCA, and CDCA were similar to their taurine-conjugated counterparts, with lower CMCs for DCA and CDCA compared to that of CA (Figure S7A), consistent with prior literature.^51,52 Like taurine-conjugated CA, glycine-conjugated CA also demonstrated a preference towards solubilizing unsaturated over saturated fatty acids, suggesting that the selectivity is preserved across the different conjugated CA species (Figures 7C and S7B).

Figure 7. Fatty acids are differentially solubilized by bile acids into mixed micelles.

(A) Experimental schematic to determine the influence of fatty acids on the CMC of specific bile acids. Above a CMC, the bile acid, fatty acid, and fluorophore aggregate and increase fluorescence emission upon micelle formation.

(B) CMC plots for T-CA mixed with the fatty acids C14:0, C16:0, C18:0, C18:1, C18:2, and C18:3. CMC (red) was determined by the intersection of the two tangents created in the plot.

(C) Table of CMC values of various bile acids mixed with specific fatty acids. Bile acids highlighted in blue differentially solubilize fatty acids.

(D) Experimental schematic to determine the effect of T-CA on the uptake efficiency of saturated, monounsaturated, and polyunsaturated fatty acids in human enteroids.

(E) Rate of deuterium-labeled fatty acids C18:0, C18:1, and C18:2 uptake in enteroids over 15 minutes. n = 3 technical replicates.

CMC, critical micelle concentration; CA, cholic acid; MCA, muricholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; T, taurine-conjugated; G, glycine-conjugated.

Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. Color of the significance annotation indicates the group with the greater mean value.

See also Figure S7.

Finally, to determine whether this differential micellar solubility translates into selective uptake of fatty acids in the intestine, we loaded mixed micelles containing taurine-conjugated CA and equimolar amounts of deuterium-labeled (-d) C18:0, C18:1, and C18:2 fatty acid on the apical surface of human enteroids (Figure 7D). Because the exogenous fatty acids were labeled with deuterium, they could be distinguished from endogenous lipids within the enteroids. Analysis of intracellular deuterated fatty acids over a 15-minute time course revealed a clear hierarchy in enterocyte uptake, with C18:2-d taken up most efficiently, followed by C18:1-d, and lastly C18:0-d (Figure 7E, S7C). This pattern mirrors our micellar solubility data (Figure 7C), in which unsaturated fatty acids are preferentially incorporated into mixed micelles. Together, these data demonstrate that the selective incorporation of fatty acids into micelles translates directly into selective uptake by enterocytes. Given the high abundance of CA and findings from these in vitro studies, we posit that conjugated CA is likely the main detergent bile acid and primary driver of the selective fatty acid absorption in both mice and humans.

Discussion

Dietary fat absorption is an indispensable physiological process that can contribute to metabolic health and disease. Although intestinal absorption is historically viewed as a passive process driven by concentration gradients⁷, we show that bile acids, as crucial gatekeepers of dietary fats, can be modulated to control intestinal lipid absorption and subsequent downstream metabolism. We demonstrate that the bile acid pool preferentially solubilizes polyunsaturated fatty acids into mixed micelles for intestinal uptake. Since certain polyunsaturated fatty acids, like C18:2, are essential and must be obtained from diet⁴, it is conceivable that this selectivity in polyunsaturated fatty acid absorption evolved to ensure efficient uptake of these vital nutrients. By reducing the bile acid pool size, we found that saturated fatty acids, particularly longer-chain ones, are inefficiently solubilized and disproportionately malabsorbed. This also results in the passage of unabsorbed fatty acids into the distal gut, triggering incretin release that can regulate appetite and energy balance. This combination of reducing absorption, while also enhancing gut hormone secretion, may prove to be a powerful strategy to tackle metabolic disease, particularly in a pharmacologic era that demands multiple therapeutic targets.

This differential uptake of fatty acids is a natural phenomenon that occurs in both mice and humans. Early animal studies using radioactive tracers showed that saturated long-chain fatty acid species required a greater length of intestine for absorption compared to unsaturated fatty acids.⁵⁴ Further research in chronic bile diversion models confirmed that unsaturated fatty acids are less dependent on bile for uptake than the more hydrophobic saturated fatty acids, despite longer chain length.⁵⁵ In humans, clinical studies similarly show that fatty acid absorption depends on acyl chain length and saturation, with polyunsaturated fats absorbed more efficiently than saturated ones.⁵⁶ Our studies reveal that both the size and composition of the bile acid pool play essential roles in regulating the differential uptake of fatty acids.

Using AAV-CRISPR tools, solubility micelle assays, and intestinal enteroids, we further demonstrate that individual bile acid species differ in their detergent capacities and distinctly influence fatty acid absorption. Among these, conjugated CA acts as the key regulator of fatty acid absorption. Prior reports have also demonstrated the central role of CA in mediating intestinal lipid absorption^57,58,59, but our study extends these findings by showing that CA is selective and preferentially solubilizes unsaturated fatty acids for uptake across the intestinal epithelium. Since CA is a major bile acid species in both mice and humans¹², we propose it is likely the main physiological driver of differential fatty acid absorption. Since humans do not produce MCAs, the predominance of these bile acids in mice may explain the heightened protection of mice from metabolic diseases.⁶⁰ This may also suggest that the reduced absorption in mice with loss of CYP8B1^57,58,59 may not be directly mirrored in humans with CYP8B1 variants.⁶¹ However, human patients with homozygous mutations in CYP7A1 do demonstrate increased fecal fat excretion⁴¹, suggesting that absorption may at least partly be regulated by fluctuations or genetic variants in CYP7A1 that alter the whole bile acid pool. As we show here, loss of CYP7A1 in mice has the most pronounced effects on selective absorption.

In contrast to the selective fatty acid absorption achieved through bile acid manipulation, lipase inhibition results in equivalent but indiscriminate reductions in absorption of all fatty acids. The consequences of this broad decrease in absorption by orlistat were evident in the liver, where polyunsaturated fatty acids were markedly decreased—a lipidomic pattern also observed in patients with MASLD.⁶² Given the reports of severe liver injury of unclear origin associated with orlistat use⁶³, our findings suggest that hepatic polyunsaturated fatty acid depletion, together with increased lipogenesis, may contribute to this pathology. Collectively, these results highlight distinct mechanisms of fatty acid absorption and underscore the need for careful consideration of therapeutic strategies targeting dietary fat uptake.

Emerging evidence supports the replacement of saturated fats with unsaturated fats to reduce cardiometabolic risk.⁶⁴ Such dietary shifts have been shown to induce changes in the plasma lipidome that are strongly linked with lower rates of cardiometabolic disease.⁶⁵ Our findings extend this concept by showing that bile acid modulation can also enrich liver and adipose tissues with polyunsaturated fatty acids by limiting absorption of saturated fatty acids. Furthermore, reducing bile acids stimulates incretin hormone secretion and promotes satiety, a response currently utilized in obesity treatments with medications such as GLP-1 and glucose-dependent insulinotropic polypeptide receptor agonists.⁶⁶

Beyond metabolic diseases, our study opens opportunities to leverage the solubilizing capacity of different bile acids in other disease settings. In malabsorptive disorders, such as short bowel syndrome and pancreatic insufficiency, stronger detergent bile acids could be used to enhance dietary fat absorption and prevent essential fatty acid deficiencies—an especially important consideration in the more vulnerable neonatal and pediatric populations who require key polyunsaturated fatty acids, like docosahexaenoic acid (C20:5) and C20:4, for growth and neurodevelopment.⁶⁷ In cholestatic liver diseases, where impaired bile flow causes fat malabsorption, different bile acids can also be considered. Ursodeoxycholic acid, a bile acid commonly used to improve bile flow, has limited impact on lipid solubility and absorption.⁶⁸

A growing number of medications under investigation or approved for other therapeutic purposes have been demonstrated to alter the bile acid pool.⁶⁹ For example, ileal bile acid transporter inhibitors, approved for use in cholestatic disorders such as Alagille syndrome and progressive familial intrahepatic cholestasis^70,71, prevent bile acid reuptake in the ileum and thereby decrease the overall pool size. Thyroid hormone receptor β agonists for MASLD⁷² have been shown to remodel the bile acid pool by decreasing 12α-hydroxylated bile acids, impairing intestinal lipid absorption.⁷³ Other classes of drugs, including bile acid analogs, like norucholic acid for primary sclerosing cholangitis⁷⁴; nuclear receptor agonists, such as obeticholic acid for primary biliary cholangitis⁷⁵; fibroblast growth factor 19 analogs for MASLD⁷⁶; and sterol-like molecules, such as 7-ketocholesterol⁷⁷, can also modulate bile acid metabolism. These pharmacologic agents may directly impact bile acid-mediated dietary fat absorption, warranting further investigation.

In this study, we employ AAV-CRISPR as a proof-of-concept strategy to mimic pharmacologic inhibition of bile acid synthetic enzymes to modulate fatty acid absorption. Our approach will inform the development of small-molecule inhibitors that fine-tune the bile acid pool in a selective, dose-dependent manner without triggering compensatory responses. This contrasts with bile acid sequestrants, which broadly bind intestinal bile acids. Bile acid sequestration triggers compensatory increases in bile acid synthesis and very low-density lipoprotein secretion that can lead to hypertriglyceridemia.⁷⁸ Together, these pharmacologic interventions underscore the translational relevance of our work and highlight the therapeutic potential of existing and emerging strategies to manipulate the bile acid pool and to regulate dietary fat absorption across a spectrum of conditions—from malabsorptive disorders to cardiometabolic diseases.

Limitations of Study

A key limitation to any bile acid research is the difference in bile acid metabolism between mice and humans. Mice synthesize bile acids that are not present in humans, complicating direct translational relevance. Despite this limitation, our findings reveal that the selective solubility of fatty acids in bile acid micelles is a conserved phenomenon across species. This effect appears to be primarily mediated by conjugated CA, a major bile acid that is shared by both mice and humans.

Resource Availability

Lead Contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Thomas Q. de Aguiar Vallim (tvallim@mednet.ucla.edu).

Materials Availability

All unique/stable reagents generated in this study are available from the lead contact without restriction.

Data and Code Availability

All data and materials reported in this paper will be shared by the lead contact upon request. All original code is available here: https://github.com/tarlingvallimlab/Absorption-CYP7A1. Raw sequencing data is available on NCBI BioProject via the accession code PRJNA1302405 for microbiome metagenomic sequencing and on NCBI GEO via the accession code GSE304800 for bulk liver RNA-sequencing. Source data is available in Data S1.

STAR Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice.

Experimental protocols were approved by the UCLA Animal Research Committee. C57BL/6J male mice (000664) were obtained from The Jackson Laboratory at six weeks of age and acclimated at UCLA until experimental start at eight weeks of age. Srebp1c^−/− mice (004365) were previously purchased from The Jackson Laboratory and backcrossed onto a C57BL/6J background for over 15 generations. Gpr120^−/− mice were a gift from Dr. Daisy Sahoo at the Medical College of Wisconsin, under material transfer agreement (MTA2023–1916) from Dr. Vincent Poitout.⁸³ All mice were maintained in temperature-controlled conditions (20–24°C) with a 12 hour:12 hour light:dark cycle (dark, 18:00–06:00) in the vivarium until eight weeks of age, which marked experimental start point for all experiments. They were fed ad libitum on a standard, low-fat, diet (LabDiet, PicoLab^® Mouse Diet 20, 5053) for one week following experiment onset and then transitioned to a typical Western diet (Research Diets, D12079B; 43% kcal from carbohydrate, 40% kcal from fat, 17% kcal from protein, 0.21% cholesterol). For initial AAV-CRISPR validation experiments while on standard diet, mice were group housed by genotype. For metabolic and absorption studies, mice were singly housed to avoid potential cross-contamination between animals through coprophagy. AAV-CRISPR were administered to eight-week-old mice by intraperitoneal injection with 5 × 10¹¹ genome copies per mouse. Validation experiments were performed (data not shown) to ascertain efficient editing by western blot and DNA analysis, which was used for all experiments in this paper. For Western diet experiments, mice in the AAV-CRISPR groups were also dosed with 5 × 10¹¹ genome copies per mouse and fed standard diet for one additional week after AAV-CRISPR injection to accelerate turnover of the bile acid pool and then switched to Western diet for the remainder of the study. For the obesity treatment experiment, eight-week-old mice were group housed and fed Western diet for eight weeks to induce obesity. After this period, mice were administered either control or Cyp7a1 AAV-CRISPR, individually housed, and maintained on Western diet for an additional eight weeks. Loss of CYP7A1 was confirmed by western blot and bile acid profiling at the end of experiments. In the case of inefficient loss of protein expression, mice were excluded from analysis. Mice in the orlistat group were treated with orlistat at 10 mg/kg/day⁸⁴ via a modified Western diet containing orlistat (TCI America, O0381) at 0.01% weight (Research Diets, D22050401). Body weight and composition (whole body fat and lean mass) were measured using the EchoMRI^™ (Houston, Texas, USA).

METHOD DETAILSAAV-CRISPR Design, Cloning, and Production.

For the targeted disruption of Cyp7a1, Cyp8b1, Cyp2c70, and Cyp2a12, we utilized AAV to deliver a complete CRISPR system based on Staphylococcus aureus CRISPR/Cas9 as previously reported.^79,80 To design 21-mer guide RNAs (gRNAs) targeting these genes, the coding sequence was downloaded from NCBI and regions most likely to be essential to gene function with limited homology to related proteins were identified. Up to 10 gRNAs were hand annotated, and off targets were queried using CRISPR Off-target Sites with Mismatches, Insertions and/or Deletions (COSMID).⁸¹ In order to produce the highest number of possible off targets, a NNGRR protospacer adjacent motif (PAM) was queried instead of NNGRRT, and relaxed search terms allowing three mismatches or two insertions/deletions were used. Of the available gRNAs, only those with the fewest off targets and no predicted off targets in exons were selected for cloning and in vivo testing. The gRNAs were cloned into 1313 pAAV-U6-BbsI-MluI-gRNA-SA-HLP-SACas9-HA-OLLAS-spA (Addgene, 109304), which was a gift from Dr. William Lagor at Baylor College of Medicine using the BbsI cloning site.⁸⁰ This backbone contains AAV2 inverted terminal repeats surrounding a U6 promoter driving gRNA expression, and a hybrid liver-specific promoter (HLP) driving expression of Staphylococcus aureus Cas9 (SaCas9). The 1313 parent backbone encodes a gRNA with two BbsI cut sites, which expresses, but does not return off targets for SaCas9 with COSMID. For the production of AAV2/8 for the in vivo testing of the AAV-CRISPR constructs, the triple transfection method followed by cesium chloride gradient was used as previously described⁸², with modifications. Plasmids for AAV2/8 production, the pAdDeltaF6 helper plasmid (PL-F-PVADF6 p0047) and pAAV2/8 (PL-T-PV0007 p2123) were obtained from the University of Pennsylvania Vector Core and produced by Puresyn, Inc (Devault, PA, USA). These plasmids were co-transfected into HEK293T (ATCC, CRL-3216) with parent plasmids targeting Cyp7a1, Cyp8b1, Cyp2a12, Cyp2c70 or 1313 pAAV-U6-BbsI-MluI-gRNA-SA-HLP-SACas9-HA-OLLAS-spA for Control to package AAV-CRISPR. A single cesium chloride gradient was sufficient to separate the AAV. Highest titer fractions were determined by quantitative PCR and then desalted by dialysis in phosphate-buffered saline (PBS) and concentrated using four milliliter centrifuge concentrators with a 100,000 kDA cutoff (Sartorius, VS04T22). A final titer was completed by qPCR following DNase treatment to account for unencapsidated viral genomes, and the AAV was aliquoted and stored at −80°C until injection.

Metabolic profiling studies.

Metabolic caging was conducted using the Promethion Core^® system (Sable Systems International, Las Vegas, NV, USA). All mice were individually housed for these studies. A one-week acclimation period in the metabolic cage system was completed before the start of the experiments. Food intake, physical activity (beam breaks), oxygen consumption (VO₂), carbon dioxide production (VCO₂), respiratory energy ratio (RER), and energy expenditure data were collected every 5 minutes. Energy expenditure was calculated using the Weir equation: kcal/day = 1440 × (3.9 VO2 + 1.1 VCO2).^17,85 The metabolic studies spanned a total duration of seven or eight weeks.

Tissue collection.

Tissues were collected after eight weeks of Western diet for most experiments. However, tissues were collected after four weeks in Cyp7a1 CRISPR mice on low-fat diet (Figure S1), after four weeks in Cyp8b1, Cyp2c70, and Cyp2a12 CRISPR mice (Figure 6), and after one week of bile acid supplementation in Cyp7a1 and Cyp7a1+Cyp2c70 CRISPR mice (Figure 6). Mice were fasted for four hours prior to sacrifice, unless otherwise stated. Plasma was collected under anesthesia by retroorbital collection. Following sacrifice, bile, liver, cecal contents, and small intestine were collected. For bile acid pool analysis, bile volume was measured by micropipette, intestine (including its contents) was weighed by scale, and liver was weighed by scale, allowing for calculation of total bile acid content per tissue (μmol). All samples were snap frozen with liquid nitrogen.

Human bile collection.

The human bile sample was obtained from an otherwise healthy adult patient, who underwent partial hepatectomy and cholecystectomy for a benign hepatic mass. Bile was aspirated directly from the excised gallbladder, placed in a sterile container, de-identified, and stored at −80°C until analysis. The study procedures were exempt from UCLA Institutional Review Board review as the sample was collected during standard clinical care.

Western blot.

Liver tissues (50–100 mg) were homogenized using radioimmunoprecipitation buffer supplemented with protease inhibitor cocktail tablets (MedChem Express HY-K0011), phenylmethanesulfonyl fluoride (Sigma Aldrich, 93482–50mL-F), and calpain inhibitor (Sigma Aldrich, A6185–25 mg). Debris was pelleted by centrifugation at 5000 rpm at 4°C for 15 minutes. The supernatant was collected, and protein concentration was determined using a standard bicinchoninic acid protein assay (Fisher, PI23227). Samples were prepared for western blotting by dilution with 4x sample buffer (Thermo Fisher, NP0007) and warming at 37°C. Equal concentrations of protein were loaded for each sample and separated by electrophoresis using Biorad Any kD^™ gels (4569033) in Tris Glycine buffer. Proteins were transferred to a polyvinylidene fluoride membrane and blocked in 5% non-fat milk in tris buffered saline with 0.01% Tween-20 (TBST). Following blocking, antibodies to target proteins in Table S3 were incubated with the membrane overnight at 4°C. Following incubation, membranes were washed with TBST and incubated with donkey anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody at room temperature for one hour. Following washing in TBST, blots were developed with HRP substrate (Fisher, WBLUF0500) on Amersham Imager 600 (GE Healthcare, Chicago, IL, USA).

Bile acid extraction and quantification.

Bile acids were extracted from mouse gallbladder, liver and small intestine as described previously⁸⁶, with some adjustments described as follows. Glyco-chenodeoxycholic acid (Sigma-Aldrich, G0759) was spiked into Optima methanol (Fisher Scientific, A456–1) as a recovery standard for bile, liver, and intestine at 1, 4 and 90 nmol respectively. For bile from human gallbladder, tauro-β-MCA (Cayman Chemical, 20289) was spiked in at 1 nmol as the recovery standard. For gallbladder, bile was diluted 1:1000 (mouse) or 1:2000 (human) in Optima Methanol, centrifuged at 15,000 rpm for 5 minutes then diluted 1:2 into an autosampler vial, prior to analysis. For liver and intestine, samples were homogenized in Optima methanol, vortexed and incubated at room temperature overnight. Samples were centrifuged at 4000 rpm for 10 minutes the next day, and supernatant used for bile acid analysis. Prior to analysis, an aliquot of supernatant was centrifuged at 15,000 rpm for 5 minutes. Intestinal samples were further diluted 1:20, whereas liver supernatant was filtered through a 0.45 μm syringe driven filter unit (Millipore, SLLHR04NL) directly into autosampler vials. Bile acid analysis was performed as described⁸⁶, with the following modifications: the Ultimate3000 system was equipped with either an ACE Excel C18-PFP 1.7 μm (100 mm × 2.1 mm, MAC-MOD Analytical, EXL-1710–1002U) or C18-PFP 2 μm (150 mm × 2.1 mm, MAC-MOD Analytical, EXL10101502U) UHPLC column, connected to a TSQ Quantiva Mass Spectrometer (Thermo Scientific); bile and intestine samples were injected at 2 μL per sample, and liver samples were injected at 5 μL via the WPS-3000 TSL autosampler; the following additional purified standards were detected in the bile acid run: (A) ω-muricholic acid (ω-MCA; Isosciences, 18015; Precursor m/z = 407.2802), (B) Ursodeoxycholic acid (UDCA; Cayman, 15121; Precursor m/z = 391.2853), and (C) T-UDCA (Millipore-Sigma, 580549; Precursor m/z = 498.2894, Product 1 m/z = 124.0067, Product 2 m/z = 79.9578). The following lists all the bile acid species analyzed: α-muricholic acid (MCA), β-MCA, ω-MCA, γ-MCA, cholic acid (CA), chenodeoxycholic acid (CDCA), lithocholic acid (LCA), deoxycholic acid (DCA), hyodeoxycholic acid (HDCA), ursodeoxycholic acid (UDCA), lagodeoxycholic acid (3a12b), 7-ketolithocholic acid (3a7k), and their glycine and taurine-conjugated derivatives.

Thin layer chromatography.

To determine the type of lipid excreted in the feces, lipids from 25 mg of feces, which were collected from mice after two weeks of Western diet, were extracted using the Folch method⁸⁷ for thin layer chromatography. Lipid extracts were dried under nitrogen gas, resuspended in 1 mL of chloroform, spotted on silica plates, and separated using the solvent systems chloroform:methanol (2:1) and heptane:isopropyl ether:acetic acid (15:10:1). The plates were sprayed with 8-anilino-1-naphthalenesulfonic acid and bands were visualized with a Transilluminator FBTI-88 (Fisher Scientific, Hampton, NH, USA).

Oxygen bomb calorimetry.

To determine fecal energy loss, feces over the final seven-day period of experiments were weighed and measured for energy content (heat of combustion) using a 6400 isoperibol calorimeter with an 1138 oxygen bomb (Parr Instruments Co., Moline, IL, USA). Total fecal energy loss was calculated by multiplying cumulative fecal mass (gram) with energy density (kilocalorie/gram) to calculate total fecal energy loss (kilocalories).

Glucose tolerance tests.

Glucose tolerance tests were performed in control and Cyp7a1 CRISPR mice after three and seven weeks of Western diet. Mice were fasted for four hours and administered D-glucose (Sigma, G8769) by oral gavage dosed at 1.5 g/kg body weight. Blood glucose was measured from tail vein blood at 0, 15, 30, 45, 60, 90, and 120 minutes after gavage using a OneTouch Ultra Mini glucometer (LifeScan, Milpitas, CA).

Hormone assays.

For the hormone studies after refeeding, mice were fed Western diet for eight weeks and fasted overnight for 16 hours, then refed ad libitum for 2 hours until blood collection to minimize feeding variability and maximize hormone responses under physiologic conditions. The 2-hour refeeding duration was chosen based on the expected gastrointestinal transit time through the small intestine during normal feeding.⁸⁸ Plasma was collected in ethylenediaminetetraacetic acid (EDTA) tubes in both the fasted and refed states. For the hormone studies after a glucose challenge, mice were fed Western diet for four weeks and fasted overnight, then administered an oral gavage of d-glucose in water dosed at 2 g/kg (Sigma-Aldrich, G8270).⁸⁹ Blood was collected immediately before and 10 minutes after glucose gavage. The blood samples were immediately mixed with aprotinin 1,000 KIU/mL (Sigma-Aldrich, A6279) and dipeptidyl peptidase 4 (DPP-4) 0.1 mM (Sigma-Aldrich, D38225MG, final concentration) to prevent degradation of certain hormones. Plasma was collected after centrifugation for 10 minutes at 1,000 × g at 4°C. Plasma samples were frozen at −80°C and thawed on ice on the day of hormone analyses. Insulin, total GLP-1, PYY, ghrelin, and leptin were measured by sandwich enzyme immunoassay using the U-PLEX Custom Metabolic Group 1 (Meso Scale Discovery, K152ACM-1) with samples diluted 1:4 in the Metabolic Assay Working Solution, according to the manufacturer’s protocol.

Plasma Alanine Aminotransferase (ALT) assay.

Plasma ALT was analyzed using Teco Diagnostics ALT Liquid Reagent Kinetic Method Kit (A524–150) according to the manufacturer’s instructions, with modifications for microplate. Ten microliters of 1:5 diluted plasma were pipetted in triplicate in a 96-well clear bottom plate. Five volumes of prewarmed reagent 1 and one volume of reagent 2 were combined, and 100 μL of working solution was rapidly pipetted on top of the diluted plasma using a multichannel repeater pipette. The plate was read in the spectrophotometer at 340 nm every minute for 10 reads. To determine units per liter of activity, the average change in absorbance was calculated from the read range where change in absorption was linear. This value was used to calculate ALT using the manufacturer-provided equation: (Change in absorbance)/0.00622 × (Total Volume)/(Sample Volume).

Histological analysis.

To determine specific regions along the gastrointestinal tract where lipids are taken up, small intestines were collected from mice following a 3-hour refeed of a Western diet (Research Diets, D12079B) or a Western diet containing 0.01% orlistat (Research Diets, D22050401) after a 16-hour overnight fast. The 3-hour refeed duration was selected to allow food to reach the end of the small intestine.⁸⁸ Small intestines were dissected between the end of stomach to cecum and flushed with cold PBS until luminal content was cleared, then followed by flushing 10 mL 4% paraformaldehyde to flash-fix the luminal epithelium. The intestinal tissue was then longitudinally opened and rolled into a Swiss roll with luminal facing out and embedded in tissue cassette. Tissue fixation was performed with 4% PFA for 24 hours, then dehydrated in 20% sucrose in PBS for 48 hours in a 4°C rocker before flash freeze in optimal cutting temperature (OCT) compound on dry ice. Embedded tissues were sectioned at 18 μm thickness and attached to Superfrost Plus slides. For tissue staining, slides were further fixed with pre-chilled methanol in −20°C for 20 minutes, air-dried and rehydrated with PBS containing 0.05% Tween-20 (PBST).

Blocking was performed in PBS + 0.3% Triton X-100 + 10% normal goat serum block (BioLegend) for one hour at room temperature and stained overnight at 4°C in blocking/permeabilization buffer containing anti-EPCAM-AlexaFluor 488 (G8.8; 2.5μg/ml). Slides were washed 3 to 5 times with PBST, stained with DAPI (ThermoFisher; 1/1000) and LipidTOX Red (ThermoFisher; 1/1000) for one hour at room temperature. LipidTOX stains the neutral lipids, like triglycerides, that were resynthesized from fatty acids taken up by the intestine. Slides were further washed 3 to 5 times in PBST and mounted with Prolonged Diamond anti-fade mounting reagent and cured overnight before imaging. To assess histological lipid accumulation, liver and perigonadal white adipose tissues were collected from mice after eight weeks of Western diet following a 4-hour fast and fixed in 4% paraformaldehyde for 24 hours. After fixation, the samples were immersed in 70% ethanol solution until the liver tissues were processed, paraffin-embedded, serially sectioned and stained with H&E by the UCLA Translational Pathology Core Laboratory. Brightfield and fluorescence microscopic images were taken using a KEYENCE BZ-X810 Fluorescence Microscope (El Segundo, CA, USA) at total magnification of 40x and 200x.

Microbiome analysis.

Cecal contents of singly housed mice after eight weeks of Western diet were processed for metagenomic shotgun sequencing. Genomic DNA was extracted from each sample using the DNeasy PowerSoil Pro Kit (Qiagen, 47126) per manufacturer’s instructions, except with two adaptations. First, the cecal contents were suspended in 1 mL of Solution CD1 from the kit, then 800 μL of the suspension was transferred to the PowerBead Pro Tube for sample input. Second, the homogenization step was carried out using an OMNI Bead Ruptor Elite (OMNI International, Kennesaw, GA, USA) at the speed of 5 meters/second for 1 minute, for 10 cycles, with samples resting on ice for 5 minutes in between each cycle. Extracted genomic DNA was then quantified using Qubit^™ 1X dsDNA HS Assay Kits (Invitrogen, Q33231). Sequencing libraries were generated using a custom, low-volume protocol based on the Illumina DNA Prep workflow (Illumina, 20060059), which included tagmentation, PCR amplification, and library cleanup steps. The genomic DNA concentration from each sample was normalized to 15 ng/μL prior to the start of the workflow. Unique index adapters were used for each sample (Illumina, 20091654). Libraries were pooled to yield 1000 ng of final pooled product, then cleaned up using KAPA Pure Beads (Roche, 07983271001) for library size selection, which removed fragments shorter than 250 bp and longer than 1,500 bp. The final library pool was checked for quality, size distribution, and concentration using the 4200 TapeStation System (Agilent Technologies). The pooled libraries were then submitted to the UCLA Technology Center for Genomics & Bioinformatics for sequencing. Sequencing reads were generated using the Novaseq X Plus 25B Flow Cell (Illumina, San Diego, CA, USA) in the 2×150 bp configuration. Twenty million paired end reads were targeted for each sample.

To prevent host contamination, reads that aligned to the mouse genome (GRCm39) were removed from each sample using Bowtie2.^90,91 Then, index adapters and PhiX sequences (a common Illumina spike-in) were removed using BBDuk.⁹² With the filtered reads, metagenomic taxonomic profiling was performed using MetaPhlAn (v4.1.0) and the mpa_vOct22_CHOCOPhlAnSGB_20221 database as reference.⁹³ The resulting absolute abundances output was used for downstream analyses. Alpha diversity at the species level was determined by Shannon’s index, Simpson’s index, inverted Simpson’s index, and observed number using the vegan (v2.6.4) package in R (v4.2.1). Beta-diversity at the species level was determined by performing principal coordinate analysis (PCoA) with Bray-Curtis dissimilarity using the tidyverse (v2.0.0) and vegan (v2.6.4) packages in R. The PCoA plot was generated using the ggplot2 (v3.5.0), cowplot (v1.1.3), and ggpubr (v0.6.0) packages in R. Beta diversity was also assessed using the PERMANOVA test with adonis2 and 1000 permutations from the vegan package in R. Lastly, significant differential taxa at the species level were determined using the MaAsLin2 (v1.12.0) package in R.²⁸

Lipidomic analysis.

Lipidomic analyses were performed on plasma, liver, and perigonadal white adipose tissue from mice after eight weeks of Western diet and on feces after two weeks of Western diet. Plasma was collected following a 16-hour fast and 2-hour ad libitum refeed. For plasma samples, 5 μL was used for extraction. For homogenized samples, 50–100 mg of liver and perigonadal white adipose tissues or feces were collected in a 2 mL homogenizer tube pre-loaded with 2.8 mm ceramic beads (OMNI #19–628). 0.75 mL PBS was added to the tube and homogenized with 3 cycles of 10 seconds at 5 m/s with a 10 second dwell time in the OMNI Bead Ruptor Elite (OMNI International). Homogenate containing 2–6 mg of original tissue was transferred to a glass tube for extraction. A modified Bligh and Dyer extraction⁹⁴ was carried out on all samples. Prior to biphasic extraction, an internal standard mixture consisting of 70 lipid standards across 17 subclasses was added to each sample (AB Sciex, 5040156; Avanti, 330827; Avanti, 330830; Avanti, 330828; Avanti, 791642). Following two successive extractions, pooled organic layers were dried down in a Thermo SpeedVac SPD300DDA using ramp setting 4 at 35°C for 45 minutes with a total run time of 90 minutes. Lipid samples were resuspended in 1:1 methanol/dichloromethane with 10 mM Ammonium Acetate and transferred to robovials (Thermo 10800107) for analysis.

Samples were analyzed on the Sciex 5500 with DMS device (Lipidyzer Platform) with an expanded targeted acquisition list consisting of 1450 lipid species across 17 subclasses. Differential Mobility Device on Lipidyzer was tuned with EquiSPLASH LIPIDOMIX (Avanti 330731). Data analysis performed on an in-house data analysis platform comparable to the Lipidyzer Workflow Manager. Instrument method including settings, tuning protocol, and MRM list were reported.⁹⁵ Quantitative values were normalized to mg of tissue.

Quantitative PCR analysis.

RNA was extracted from snap frozen liver pieces (approximately 50 mg) from mice after 8 weeks of Western diet. Tissue was homogenized in QIAZOL (Qiagen) with the OMNI Bead Ruptor (OMNI International). RNA was extracted using chloroform (Thermo Scientific) and purified using RNeasy mini kit per manufacturer’s instructions (Qiagen, 74104). Purified RNA (500 ng) was then utilized for cDNA synthesis, which was subsequently diluted at a 1:5 ratio for gene expression analysis. Gene expression levels were assessed using specific primers for each gene (primer sequences available in Table S4) with a qPCR Blue Mix LR (Azura Genomics). Relative quantification was determined using an efficiency-corrected method (Roche Diagnostics) on a Lightcycler 480 II qPCR machine (Roche Diagnostics). All genes were normalized to housekeeping gene 36b4.

Cholesterol and fatty acid analysis.

Cholesterol and fatty acid methyl esters (FAMEs) were extracted from feces and liver and quantified using gas chromatography-mass spectrometry (GC-MS), as previously reported.⁸⁶ For feces, 5 mg of samples were prepared from fecal pellets powdered in liquid nitrogen. For liver, 0.3 mg of samples were prepared from frozen liver homogenized in ice-cold PBS using the OMNI Bead Ruptor (OMNI International). Cholesterol and FAMEs were extracted using mild acid methanolysis with a mixture of hydrochloric acid/methanol/toluene supplemented with 5α cholestane (Sigma, C8003) as an internal standard for sterols and trinonadecanoin (C17:1) (Nu-chek Prep, T-165) for FAMES, and then incubated overnight at 45°C. Resultant cholesterol and FAMEs were extracted in 1 mL of hexane and 10 μL from liver and 10 μL of a 1:10 dilution of fecal extracts were analyzed by GC-MS using an Agilent Technologies (Santa Clara, CA, USA) 7890B/5977A with DB-WAX UI column (Agilent Technologies, 122–7032 UI). Complete GC-MS configurations and method programs are available upon request. Quantification of all ions was conducted using custom Python Tkinter software as previously described⁸⁶ (https://github.com/gcalmettes/labUtils?tab=readme-ov-file#msanalyzer) by comparison of sample data with serial dilutions of cholesterol (Sigma, C3045) and fatty acid standard mixes (Nu-Check-Prep, GLC96 for liver; Nu-Check-Prep, GLC96C for feces) containing various long-chain fatty acid species (methyl myristate, palmitate, palmitoleate, stearate, oleate, linoleate, linolenate, arachidonate, and behenate). Integration of all ions (samples and standards) was performed on MassHunter Quantitative Analysis Program (Agilent Technologies). Fatty acids were calculated as either amount (nmol) or concentration (nmol/mg of sample).

In vivo de novo lipogenesis and cholesterol biosynthesis.

To determine de novo lipogenesis and hepatic cholesterol biosynthesis, deuterium was added to the drinking water of wildtype and Srebp1c^−/− mice at a concentration of 5% for the final three days of the lipid synthesis experiments (Sigma-Aldrich, 151882). For experiments with the Srebp1c^−/− mice, newly synthesized deuterium-labeled fatty acids were measured in livers of mice after a 16-hour fast and 2-hour refeed. Lipid synthesis was measured by GC-MS of deuterium-labelled isotopic enrichment of fatty acid and cholesterol methyl esters in the liver. The hydrogens are heavier by a mass of +1 and are incorporated into newly synthesized lipids. These minor changes in charge do not impact the retention time of the fatty acids and cholesterol, but do change the mass-to-charge ratio, m/z, which can be detected by GC-MS. The recorded values were corrected for the natural abundance of stable isotopes using the modern least-squares implementation of the skewed matrix correction method as published previously using the same custom Tkinter software described above.^86,96

Bulk RNA-sequencing analysis.

Total RNA was extracted from flash-frozen tissue samples from mice after eight weeks of Western diet using the RNeasy Mini Kit (Qiagen, 74104). For quality control of RNA integrity and library synthesis products, the RNA ScreenTape and D1000 ScreenTape kits were used with the 4200 TapeStation system (Agilent Technologies). mRNA libraries were prepared from 800 ng RNA using the KAPA RNA Hyperprep Kit (KAPA Biosystems, KK8541), with KAPA Unique Duel-Indexed Adaptor Kit (KAPA Biosystems, KK8727). The pooled libraries were then submitted to the UCLA Technology Center for Genomics & Bioinformatics for sequencing. Paired-end sequencing was performed on an NovaSeq X Plus platform (Illumina) in the 2×50 bp configuration.

Raw FASTQ files were trimmed for adapter sequences and low-quality bases using Trim Galore (https://github.com/FelixKrueger/TrimGalore). Trimmed FASTQ files were mapped to the mm10 mouse assembly using STAR.⁹⁷ Gene counts were quantified using featureCounts (Subread v2.0.6).⁹⁸ Genes with fewer than 500 total counts across all samples were excluded from downstream analyses. Differential expression analysis was conducted using DESeq2⁹⁹ (v1.48.1) in R (v3.6.3). Genes were determined as differentially expressed between two conditions if they passed these thresholds: −1.5 ≤ fold change ≥ 1.5 and adjusted p-value ≤ 0.05. Differentially expressed genes were plotted using EnhancedVolcano (v1.16.0), RColorBrewer (v1.1.3), and ggplot2 (v3.5.0) packages in R (v4.2.1). Gene set enrichment analysis (GSEA) was performed using the clusterProfiler^100,101 package (v4.16.0) with MSigDB gene sets in R (v3.6.3).¹⁰²

Fat absorption studies.

Fecal fatty acid excretion was measured by GC-MS, as above. Cumulative fecal fatty acid excretion was measured over seven weeks on Western diet and calculated by multiplying the estimated total fecal mass output over the week with the fecal fatty acid concentration for the respective week. The percentage of fatty acid absorption was measured by GC-MS of feces samples. A non-absorbable lipid, sucrose polybehenate (SPB; mainly C22:0/behenic acid conjugated to a sucrose backbone) (CarboMer, Inc., 56449-50-4) was incorporated into both Western diet and Western diet containing Orlistat at 1.5% weight, along with a non-absorbable blue dye (Research Diets, D22050608 and D22072704). The fatty acid composition of SPB was determined by GC-MS quantification to be myristic acid (C14:0), 0.05%; palmitic (C16:0), 0.7%; palmitoleic acid (C16:1), 0.005%, stearic (C18:0), 2.3%; oleic (C18:1), 0.05%; linoleic (C18:2), 0.01%; arachidic acid (C20:0), 7.7%; eicosenoic acid (C20:1), 0.02%; and behenic (C22:0), 87.1% (not shown). Since lipases cannot hydrolyze behenic acid and other fatty acids from the sucrose backbone, SPB remains unabsorbed and serves as an internal control for measuring other fatty acids.⁴² To measure absorption, the SPB diet was given to mice for 24 hours and feces were collected over the next 24 to 36 hours. Only blue-colored feces were collected, as the blue dye indicates the presence of SPB. The feces were powdered with a mortar and pestle in liquid nitrogen. FAMEs were then extracted from the diet and feces using mild methanolysis, as described above. The concentration of behenic acid and the major fatty acids were quantified by GC-MS. Fatty acid absorption was characterized as either amount of fecal fatty acids normalized to mg of feces (nmol/mg) or fractional absorption (%). The fractional absorption of each fatty acid was corrected for the non-behenic fatty acids in SPB and calculated using the following equation⁴²:

$$\frac{\frac{F_d}{B_d}-\left[\frac{F_f-\left(B_f\times F_{\mathit{\text{SPB}}}\right)}{B_f}\right]}{\frac{F_d}{B_d}}$$

F_d = sum of the masses of all dietary fatty acids (or individual fatty acid) excluding behenic acid. B_d = mass of dietary behenic acid. F_f = sum of the masses of all fecal fatty acids (or individual fatty acid) excluding behenic acid. B_f = mass of fecal behenic acid. F_SPB = mass of individual fatty acid in SPB.

Fat absorption following bile acid supplementation.

The bile acid diets were prepared by supplementing Westen diet (Research Diets, D12079B) with either cholic acid (Sigma-Aldrich, C9282) or chenodeoxycholic acid (Sigma-Aldrich, C9377) at 0.05% and 0.1% of the total diet by weight. Bile acids were dissolved in 70% ethanol, mixed into the powdered Western diet, dried for 24 hours under a negative pressure hood, and pressed into food pellets using a pellet press (Parr Instrument Co., 2811). Mice were fed the bile acid diet for one week to allow sufficient turnover of the endogenous bile acid pool. Following this, the diet was switched to the Western diet containing SPB at 1.5% weight (Research Diets, D22050608) and feces were collected after 14 hours to determine how specific bile acids affected fatty acid absorption. Because the higher 0.1% dosage markedly increased the total bile acid pool size, only experiments using the 0.05% bile acid dosage were presented in the study.

Bile acid micelle formation assay.

The micellar assay was adapted and modified to determine the volume of bile or concentration of bile acids required to solubilize fatty acids.^49,103 Six uM coumarin 6 (Sigma-Aldrich, 442631) in dichloromethane (Fisher Scientific, AC610050040) was added to 0.2 mL tubes and evaporated in the chemical hood for 30 minutes. Coumarin 6 is a hydrophobic fluorescent dye that self-aggregates and increases its fluorescence emission upon micelle formation. For the physiologic bile assay, bile was pooled from wild-type C57Bl/6 mice after eight weeks of Western diet (n = 37 mice; mean gallbladder bile volume, 10.7 μL per mouse) or collected from the gallbladder of an otherwise healthy human patient post-cholecystectomy. Bile samples were serially diluted over a range of concentrations. Mouse bile was diluted from 1:16 (highest concentration) to 1:7,800 (lowest concentration), and human bile was diluted from 1:5 (highest concentration) to 1:2×10⁶ (lowest concentration). For the in vitro bile acid assay, a series of concentrations of unconjugated, taurine-conjugated, and glycine-conjugated bile acids (CA, CDCA, DCA, UDCA, T-CA, T-β-MCA, T-CDCA, T-DCA, T-UDCA, G-CA, G-CDCA, G-DCA; Table S5), ranging from 0.0001 mM to 40 mM were prepared in PBS at 37°C. For all assays, 50 μL of the bile acid solution was transferred to the mini tubes with the fluorescent dye. Next, 2 mM of each fatty acid (C14:0, C16:0, C18:0, C18:1, C18:2, C18:3) in 100% ethanol was added to each bile acid solution, vortexed for 30 minutes in the dark, and sonicated for 15 minutes at 37°C. Afterwards, 15 μL of the solution was transferred to a black 96-well plate (Greiner Bio-One, 784076) in duplicates, and fluorescence intensity at 488±14/535±30 was measured using a CLARIOstar Plus plate reader (BMG LABTECH, Offenburg, Germany). The resultant fluorescence intensity was plotted against the logarithm of the corresponding bile volume or bile acid concentration, and the volume or CMC was determined by the intersection of the two tangents using segmental linear regression.

Culture and expansion of human enteroids.

De-identified human jejunal enteroids were collected in accordance to UCLA Institutional Review Board, IRB #10–001653, and provided as a gift from the laboratory of Dr. Martin Martin at UCLA and maintained in human enteroid culture medium containing 50% Advanced DMEM/F12 (Gibco 12634–010), 50% L-WRN conditioned medium¹⁰⁴, 2 mM GlutaMAX, 10 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin, 1X N2 supplement, 1X B27 supplement, 50 ng/ml EGF (Peprotech 315–09), 500 nM A83–01 (Med Chem Express HY-10432), 10 μM SB202190 (Med Chem Express HY10295), 2.5 μg/ml fungizone, and 200 μg/ml normocin. Enteroids were suspended in Geltrex or Matrigel and expanded/passaged every 7–10 days with the addition of 10 μM Y-27632 (Med Chem Express HY-10071) for the first few days after passage.

Culture of human intestinal epithelial cells on transwell membranes.

Human intestinal epithelial cells (IECs) were derived from basolateral-out human jejunal enteroids grown in human enteroid culture medium with the addition of 2.5 μM CHIR99021 and 10 mM nicotinamide. To culture IECs as a monolayer on transwell membranes, basolateral-out enteroids were extracted from Matrigel or Geltrex with cold PBS and incubated in cell recovery solution for 20 minutes at 4°C. The enteroids were pelleted and the supernatant was removed. The pellet was resuspended in TrypLE for 10 min at 37°C to obtain single cells. After breaking the enteroids by pipetting vigorously, medium was added to stop the reaction and the mixture was centrifuged to pellet the cells. The cells were resuspended in human enteroid culture medium with additional supplementation: 100 ng/ml IGF-1 (Med Chem Express HY-P70788), 50 ng/ml FGF2 (Med Chem Express HY-P7004), 10 nM Gastrin (HY-P2671), 10 μM Y-27632 and seeded onto transwells at 10⁵ cells per well. The same medium was added to the basolateral compartment and exchanged every other day. After ~5 days, when cells were confluent, the IECs were differentiated for 3 days by withdrawing L-WRN media and EGF, while adding BMP-2 and BMP-4. Human differentiation medium contained Advanced DMEM/F12, 2 mM GlutaMAX, 10 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin, 1X N2 supplement, 1X B27 supplement, 250 ng/ml r-spondin1, 500 nM A83–01, 100 ng/ml IGF-1, 50 ng/ml FGF2, 10 nM gastrin, 50 ng/ml BMP-2 (Med Chem Express HY-P7006), 50 ng/ml BMP-4 (Med Chem Express HY-P7007), 2.5 μg/ml fungizone, and 200 μg/ml normocin.

Fatty acid uptake in human intestinal epithelial cells on transwell membranes.

To prepare mixed micelles containing the deuterium-labeled free fatty acids, cholesterol (Sigma, C3045), 1,2-dioleoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, 850375), C18:0-d2 (Cayman, 28150), C18:1-d17 (Cayman, 9000432), and C18:2-d11 (Cayman, 9002193) were combined into a glass vial and dried under nitrogen gas. After the lipids were completely dried, sodium taurocholate (Sigma, T4009) in PBS was added, vortexed, and sonicated at 37°C for 10 minutes. The mixture was brought to a final concentration of 0.05 mM cholesterol, 0.2 mM 1,2-dioleoyl-sn-glycero-3-phosphocholine, 2 mM sodium taurocholate, 0.5 mM C18:0-d2, 0.5 mM C18:1-d17, and 0.5 mM C18:2-d11 in the treatment media. As a negative control, mixed micelles were also prepared without deuterium-labeled fatty acids. The apical compartment of the IEC monolayer was incubated with the mixed micelles in triplicates and harvested after 5, 10, and 15 minutes to assess the rate of fatty acid uptake. At each timepoint, the media was aspirated, and cells were washed twice with PBS. The cells were detached using trypsin and centrifuged at 1,200 rpm for 7 minutes. Cell pellets were resuspended in 60 μL PBS. A 40 μL aliquot was used for fatty acid analysis by GC-MS, as previously described in the methods for fatty acid analysis. The remaining 20 μL was used for DNA extraction using a Qiagen DNeasy Blood and Tissue Kit and the concentration was quantified. Following GC-MS analysis, the intracellular fatty acids were normalized to DNA content and reported as nanomoles of fatty acid per microgram of DNA.

Statistical analyses.

Statistical analyses were performed using Prism (GraphPad v10), unless otherwise noted. All data are represented as mean ± standard error of the mean (SEM), unless otherwise noted. Physiological measurements were taken from individual biological replicates. In vitro measurements were performed in duplicate or triplicate. A two-sided Student’s t-test was used for single variable comparison between two groups. An ordinary ANOVA or two-way ANOVA followed by Šídák’s or Tukey’s multiple comparisons test was used to examine interactions between multiple variables, as appropriate. ANCOVA was used to correct for the influence of total body mass as a covariate on energy expenditure, food intake and water intake; analyses were performed in CalR.⁸⁵ A p-value < 0.05 was considered statistically significant.

Supplementary Material

Supplemental Data

Document S1. Figures S1–S7. Table S1–S5.

Data S1. Unprocessed data underlying the display items in the manuscript, related to Figures 1–7, S1–S7.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-CYP7A1	Novus	Cat#NBP3–04836 lot 5500012145; RRID: AB_2255011
Rabbit polyclonal anti-CYP8B1	Abcam	Cat#ab191910; RRID: AB_2828000
Rabbit polyclonal anti- rat CYP2C22 (mouse CYP2C70)	Gift from Edward T. Morgan (Emory University)	N/A
Rabbit monoclonal anti-PDI	Cell Signaling Technology	Cat#3501S; RRID: AB_2156433
Rabbit monoclonal anti-ACC	Cell Signaling Technology	Cat#3676S; RRID: AV_2219397
Rabbit monoclonal anti-FASN	Cell Signaling Technology	Cat#3180S; RRID: AB_2100796
Rabbit monoclonal anti-SCD1	Cell Signaling Technology	Cat#2794S; RRID: AB_2183099
Donkey polyclonal anti-Rabbit IgG	Cytiva	Cat# NA934–1ML; RRID: AB_772206
Mouse monoclonal anti-EpCAM (G8.8)	DSHB	Cat#G8.8; RRID: AB_2098655
Bacterial and virus strains
pAAV-U6-SA-BbsI-MluI-gRNA-HLP-SACas9-HA-OLLAS-spA	This Paper	vETV_0015
pAAV-U6-SA-mCyp7a1-gRNA6-HLP-SACas9-HA-OLLAS-spA	This Paper	vETV_0028
pAAV-U6-SA-mCyp8b1-gRNA1-HLP-SACas9-HA-OLLAS-spA	This Paper	vETV_0032
pAAV-U6-SA-mCyp2c70-gRNA4-HLP-SACas9-HA-OLLAS-spA	This Paper	vETV_0033
pAAV-U6-SA-mCyp2a12-gRNA6-HLP-SACas9-HA-OLLAS-spA	This Paper	vETV_0057
Biological samples
Human Bile	Human	De-identified
Chemicals, peptides, and recombinant proteins
Sodium cholate	Sigma-Aldrich	Cat#C9282; Cas: 206986-87-0
Sodium taurocholate	Sigma-Aldrich	Cat#T4009; Cas: 345909-26-4
Chenodeoxycholic acid	Sigma-Aldrich	Cat#C9377; Cas: 474-25-9
Taurochenodeoxycholic acid (sodium salt)	Cayman Chemical	Cat#20275; Cas: 6009-98-9
Tauro-β-muricholic acid (sodium salt)	Cayman Chemical	Cat#20289; Cas: 145022-92-0
Sodium deoxycholate	Sigma-Aldrich	Cat#D6750; Cas: 302-95-4
Taurodeoxycholic acid, sodium salt	Millipore	Cat#580221; Cas: 1180-95-6
Ursodeoxycholic acid	Sigma-Aldrich	Cat#U5127; Cas: 128-13-2
Tauroursodeoxycholic acid, sodium salt	Millipore	Cat#580549; Cas: 35807-85-3
Sodium taurolithocholate	Sigma-Aldrich	Cat#T7515; Cas: 6042-32-6
Sodium glycocholate	Chem-Impex	Cat#27530; Cas: 863-57-0
Sodium glycochenodeoxycholate	Chem-Impex	Cat#21506; Cas: 16564-43-5
Glycodeoxycholic acid, sodium salt	Chem-Impex	Cat#21779; Cas: 16409-34-0
Glyco-chenodeoxycholic acid	Sigma-Aldrich	Cat#G0759; Cas: 16564-43-5
Aprotinin	Sigma-Aldrich	Cat# A6279; Cas: 9087-70-1
Dipeptidyl peptidase 4 (DPP-4)	Sigma-Aldrich	Cat# D38225MG
LipidTOX Red	ThermoFisher	Cat#H34476
5α cholestane	Sigma Aldrich	Cat# C8003; Cas: 481-21-0
trinonadecanoin (C17:1)	Nu-chek Prep	Cat# T-165; Cas: 26536-13-0
cholesterol	Sigma Aldrich	Cat# C3045; Cas: 57-88-5
GLC96 Lipid Standards (mixed lipid standards)	Nu-chek Prep	Cat# GLC96
GLC96C Lipid Standards (mixed lipid standards)	Nu-chek Prep	Cat# GLC96C
Deuterium oxide	Sigma Aldrich	Cat# 151882; Cas: 7789-20-0
Sucrose Polybehenate	CarboMer	Cas: 56449-50-4
coumarin 6	Sigma Aldrich	Cat#442631; Cas: 38215-36-0
1,2-dioleoyl-sn-glycero-3-phosphocholine	Avanti Polar Lipids	Cat#850375; Cas: 4235-95-4
C18:0-d2	Cayman	Cat#28150; Cas: 19905-58-9
C18:1-d17	Cayman	Cat#9000432; Cas: 223487-44-3
C18:2-d11	Cayman	Cat#9002193
GlutaMAX	ThermoFisher Scientific	Cat#35050061
N2 supplement	Fisher Scientific	Cat#17502–048
B27 supplement	Fisher Scientific	Cat#17504404
EGF	Peprotech	Cat#315–09; Cas: 62253-63-8
A83–01	Med Chem Express	Cat#HY-10432; Cas: 909910-43-6
SB202190	Med Chem Express	Cat#HY10295; Cas: 152121-03-7
Y-27632	Med Chem Express	Cat#HY-10071; Cas: 146986-50-7
CHIR99021	Stem Cell Technologies	Cat#72054; Cat: 252917-06-9
IGF-1	Med Chem Express	Cat#HY-P70788
FGF2	Med Chem Express	Cat#HY-P7004
Gastrin	Med Chem Express	Cat#HY-P2671
BMP-2	Med Chem Express	Cat#HY-P7006
BMP-4	Med Chem Express	Cat#HY-P7007
Orlistat	TCI America	Cat# O0381; Cas: 96829-58-2
Protease inhibitor cocktail tablets	Med Chem Express	Cat# HY-K0011
Phenylmethanesulfonyl fluoride	Sigma Aldrich	Cat#93482–50mL-F; Cas: 329-98-6
Calpain Inhibitor	Sigma Aldrich	Cat# A6185–25 mg; Cas: 110044-82-1
Critical commercial assays
U-PLEX Custom Metabolic Group 1	Meso Scale Discovery	Cat#K152ACM-1
Teco Diagnostics ALT Liquid Reagent Kinetic Method Kit	Teco Diagnostics	Cat# A524–150
DNeasy PowerSoil Pro Kit	Qiagen	Cat#47126
Qubit™ 1X dsDNA HS Assay Kit	Invitrogen	Cat#Q33231
Illumina DNA Prep workflow	Invitrogen	Cat#20060059
KAPA Pure Beads	Roche	Cat#07983271001
RNeasy mini kit	Qiagen	Cat#74104
KAPA RNA Hyperprep Kit	KAPA Biosystems	Cat# KK8541
KAPA Unique Duel-Indexed Adaptor Kit	KAPA Biosystems	Cat# KK8727
Bicinchoninic acid protein assay	Fisher	Cat#PI23227
Deposited data
Microbiome Sequencing Data	NCBI BioProject	PRJNA1302405
Bulk Liver RNA Sequencing Data	NCBI GEO	GSE304800
Source Data	This Paper	Data S1
Experimental models: Cell lines
HEK293T	ATCC	Cat#CRL-3216
De-identified human jejunal enteroids	Dr. Martin Martin, UCLA	N/A
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	Jax: 000664
Mouse: Srebp1c−/− (B6;129S6-Srebf1tm1Mbr/J	The Jackson Laboratory	Jax: 004365
Mouse: Gpr120−/−	Dr. Daisy Sahoo via Dr. Vincent Poitout83	N/A
Oligonucleotides
Primers: See Table S4
Recombinant DNA
1313_pAAV-U6-BbsI-MluI-gRNA-SA-HLP-SACas9-HA-OLLAS-spA	Li et al80	Addgene Plasmid #109314
pETV_0065-U6-SA-mCyp7a1-gRNA6-HLP-SACas9-HA-OLLAS-spA	This Paper	N/A
pETV_0067-U6-SA-mCyp8b1-gRNA1-HLP-SACas9-HA-OLLAS-spA	This Paper	N/A
pETV_0073-U6-SA-mCyp2c70-gRNA4-HLP-SACas9-HA-OLLAS-spA	This Paper	N/A
pETV_0149-U6-SA-mCyp2a12-gRNA6-HLP-SACas9-HA-OLLAS-spA	This Paper	N/A
pAdDeltaF6	Puresyn Inc.	Cat#PL-F-PVADF6 p0047
pAAV2/8	Puresyn Inc.	Cat#PL-T-PV0007 p2123
Software and algorithms
Graphpad Prism V10	Graphpad	N/A
Excel	Microsoft	N/A
The R Project for Statistical Computing (v3.6.3 and v4.2.1)	www.r-project.org	N/A
Bowtie2	https://bowtie-bio.sourceforge.net/bowtie2/index.shtml 90,91	N/A
BBDuk	https://docs.seqera.io/multiqc/modules/bbduk 92	N/A
MetaPhlAn (v4.1.0)	https://huttenhower.sph.harvard.edu/metaphlan/ 93	N/A
Vegan (v2.6.4) package	https://cran.r-project.org/web/packages/vegan/index.html	N/A
tidyverse (v2.0.0)	https://tidyverse.org	N/A
ggplot2 (v3.5.0)	https://ggplot2.tidyverse.org	N/A
cowplot (v1.1.3)	https://cran.r-project.org/web/packages/cowplot/vignettes/introduction.html	N/A
ggpubr (v0.6.0)	https://rpkgs.datanovia.com/ggpubr/	N/A
MaAsLin2 (v1.12.0)	https://huttenhower.sph.harvard.edu/maaslin/ 28
Python (v3.6)	Python	N/A
Custon Python TKinter for GC-MS analysis	G. Calmettes: https://github.com/gcalmettes/labUtils?tab=readme-ov-file#msanalyzer	N/A
Trim Galore	https://github.com/FelixKrueger/TrimGalore	N/A
STAR	https://github.com/alexdobin/STAR 97	N/A
Featurecounts Subread v2.0.6	https://subread.sourceforge.net 98	N/A
DESeq2 (v1.48.1)	https://bioconductor.org/packages/release/bioc/html/DESeq2.html 99	N/A
EnhancedVolcano (v1.16.0)	https://bioconductor.org/packages/release/bioc/html/EnhancedVolcano.html	N/A
RColorBrewer (v1.1.3)	https://cran.r-project.org/web/packages/RColorBrewer/index.html	N/A
clusterProfiler package (v4.16.0)	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html 100,101	N/A
Other
Western Diet	Research Diets	Cat#D12079B
Custom Western Diet with 0.01% orlistat	Research Diets	Cat#D22050401
Custom Western Diet with 1.5% sucrose polybehenate	Research Diets	Cat#D22050608
Custom Western Diet with 1.5% sucrose polybehenate and 0.01% orlistat	Research Diets	Cat#D22072704
Commented code, packages used in the methods, some data, and other information from the studies in this paper	This Paper	https://github.com/tarlingvallimlab/Absorption-CYP7A1

Highlights.

• Bile acids selectively regulate the differential uptake of dietary fatty acids

• Reducing bile acids enhances gut hormone secretion

• Polyunsaturated fatty acids are preferentially micellized for intestinal uptake

• Bile acids can be leveraged to target lipid absorption in metabolic disease