Unlocking therapeutic potential of amlexanox in MASH with insights into bile acid metabolism and microbiome

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD), characterized by excessive lipid accumulation in the liver, is the hepatic manifestation of metabolic disorders associated with obesity and type 2 diabetes^1,2. This condition affects approximately 25% of the global adult population and a striking 40% of adults in the U.S^2,3. Metabolic dysfunction-associated steatohepatitis (MASH) is an advanced form of MASLD, marked by chronic inflammation, liver injury, and frequently, fibrosis. Left untreated, MASH can progress to cirrhosis, liver failure, and hepatocellular carcinoma (HCC), necessitating liver transplantation¹. Despite the growing burden of MASH, the development of MASH drugs has proven challenging, in part due to the complexity of the disease^4,5,6,7. Hence, there is an urgent need to gain a comprehensive understanding of the underlying pathomechanisms of MASH and to uncover new therapeutic strategies capable of addressing its multifaceted pathologies.

Obesity induces a persistent state of low-grade inflammation in the liver and adipose tissue, accompanied by increased circulating pro-inflammatory cytokines^8,9. These inflammatory events are characterized by the activation of the transcriptional factor NF-κB in both immune cells, as well as metabolically active hepatocytes and adipocytes^10,11. It has been previously reported that NF-κB sensitive genes, including noncanonical IκB kinases (IKKs), IKKε and TANK-binding kinase 1 (TBK1), are elevated at both the mRNA and protein levels in adipose tissue and livers of obese mice^11,12. Mice lacking IKKε show resistance to diet-induced obesity and insulin resistance while exhibiting increased energy expenditure¹¹. These findings led to the discovery of amlexanox as a specific inhibitor of both kinases¹². Originally developed in the mid-1980s to treat asthma and allergic rhinitis, amlexanox has an excellent record of safety¹³. We demonstrated that amlexanox substantially improved glucose intolerance, insulin resistance, and hepatic steatosis in genetically obese and diet-induced obese mice^12,14,15,16. Additionally, it significantly reduced hemoglobin A1c (HbA1c) levels in diabetic patients with high systemic inflammation¹⁷. Most recently, we found that amlexanox improves diet-induced dyslipidemia and attenuates atherogenesis in Western diet-fed Ldlr^−/− mice¹⁶. However, further investigation is needed to understand its potential impact on advanced liver diseases such as MASH and associated HCC, especially in its therapeutic potential to reverse established diseases.

Bile acids (BAs) play a crucial role in regulating lipid metabolism and maintaining metabolic homeostasis. BA synthesis and secretion is the main pathway for cholesterol excretion. Proper lipid metabolism relies on tightly controlled BA metabolism, including synthesis, secretion, and reabsorption. BA sequestrants have been used as medications to attenuate hypercholesterolemia. Dysregulation in BA metabolism has emerged as a significant factor in the pathogenesis of both acute and chronic liver diseases¹⁸. Studies have shown that MASH patients experience changes in BA composition and related signaling pathways, suggesting a potential connection between BA metabolism and liver disease¹⁸. Moreover, studies investigating the gut microbiota in MASH have revealed significant alterations, which in turn impact BA metabolism and homeostasis^19,20,21,22. The interaction between BAs and the gut microbiota, governed by the gut-hepatic axis, plays a crucial role in metabolic health and disease. Disrupted BA metabolism, caused by gut microbiome dysbiosis, exacerbates hepatic inflammation and steatosis, leading to the pathogenesis of MASH^19,20,21,22. Therefore, understanding the dynamic interrelationships between BA metabolism and gut microbiome in MASH is critical for developing effective treatments.

Here, we explore the therapeutic potential of amlexanox in treating MASH and associated HCC. Using a robust preclinical model of GAN diet-fed Ldlr^−/− mice, we evaluated the effects of amlexanox on dyslipidemia, BA metabolism, gut microbiota, and MASH-associated manifestations, such as steatosis, inflammation, and fibrosis. Our findings reveal that amlexanox shows promise as a therapeutic intervention for MASH and associated HCC, while significantly ameliorating atherosclerosis at the same time. Transcriptomic profiling demonstrated that amlexanox downregulates the expression of inflammatory, fibrogenic, and tumorigenic genes, while upregulating genes involved in BA metabolism. Mass spectrometry revealed that amlexanox increased BA synthesis and excretion while reducing BA reabsorption by increasing sulfated BAs. 16S rRNA sequencing showed that amlexanox significantly reprogrammed gut microbiota and enriched Akkermansia muciniphila, a probiotic bacterium exhibiting major metabolic beneficial function. Correlation analysis revealed a strong positive association between Akkermansia muciniphila and bile sulfate salts. Altogether, our findings shed light on the impact of amlexanox on the interplay between bile acid and microbiome, offering new insights into potential therapeutic strategies for MASH and associated HCC.

Results

Amlexanox improves lipid metabolism to attenuate hepatic steatosis

To assess the efficacy of amlexanox in MASH therapy, we fed Ldlr^−/− mice with a GAN (Gubra-Amylin NASH) diet for 20 weeks to establish MASH. Following this, the mice received oral gavage with either a vehicle or amlexanox every other day for an additional 12 weeks while continuing on the GAN diet (Fig. 1a). After a total of 32 weeks on the GAN diet, mice treated with amlexanox exhibited a significant reduction in body weight, liver weight, and adipose tissue weight compared to control mice (Fig. 1b–e), consistent with our previous observations in mice on a high-fat diet (HFD). Hematoxylin and eosin (H&E) staining showed improved hepatic steatosis (Fig. 1f). Quantification of hepatic lipid content confirmed a substantial reduction in triglyceride and cholesterol levels in GAN-fed Ldlr^−/− mice treated with amlexanox (Fig. 1g and h). Additionally, serum triglyceride and cholesterol levels were significantly reduced, indicating a marked improvement in dyslipidemia (hypertriglyceridemia and hypercholesterolemia) compared to control mice (Fig. 1i and j). We also evaluated atherosclerotic plaque development in the aortas. Remarkably, amlexanox not only exhibited therapeutic effects on MASH, but also attenuated atherosclerosis. En face staining showed a significant reduction in aortic lesion area following amlexanox treatment (Supplementary Fig. 1a and b), and staining of aortic roots revealed a substantial decrease in lesion size (Supplementary Fig. 1c). The attenuation of atherosclerotic lesion development suggests that amlexanox has the potential to alleviate systemic consequences of dyslipidemia and metabolic dysfunction, including MASH and cardiovascular complications. Next, we examined the expression of key genes involved in lipid metabolism in the livers of vehicle- and amlexanox-treated mice. Notably, amlexanox treatment downregulated genes associated with fatty acid synthesis (Elovl7, Elovl6, Hacd4, Acsl4, Pla2g4a, and Scd2) and triglyceride synthesis (Mogat1) (Fig. 1k). Conversely, genes involved in fatty acid catabolism (Acox2, Acsl1, Cyp4a10, Cyp4a14, Cyp4a32, Ehhadh, and Slc27a2) and cholesterol catabolism (Cyp7a1 and Lcat) were largely upregulated (Fig. 1l). These data collectively indicate that amlexanox improves lipid metabolism and attenuates hepatic steatosis in GAN diet-fed Ldlr^−/− mice.

Unlocking therapeutic potential of amlexanox in MASH with insights into bile acid metabolism and microbiome — **Fig. 1: Amlexanox improves lipid metabolism to attenuate hepatic steatosis.**

Amlexanox modulates bile acid metabolism

To better understand how amlexanox promotes lipid metabolism in MASH, we performed a transcriptomic analysis on the livers of vehicle- or amlexanox-treated mice. Our data demonstrated that amlexanox treatment induced profound transcriptional changes in the livers of GAN-fed Ldlr^−/− mice. Specifically, we identified 786 genes were upregulated, and 1718 genes that were downregulated by amlexanox treatment (> 1.5-fold change, P-adj < 0.05, Fig. 2a). Gene Ontology analysis (GO) revealed that the upregulated genes in amlexanox-treated livers were predominantly involved in lipid metabolism, including fatty acid, triglyceride, cholesterol, and BA metabolism pathways (Fig. 2b). Further functional analysis using the KEGG pathway highlighted significant enrichment in primary BA biosynthesis and secretion (Fig. 2c). Among the upregulated genes were Cyp7a1 (rate-limiting enzyme), Cyp7b1, Cyp27a1, Cyp8b1, Hsd3b7, and others, which encode enzymes crucial for BA synthesis and secretion (Fig. 2d and e). BA synthesis serves as a major pathway for hepatic cholesterol catabolism²³. To validate amlexanox’s effects on BA synthesis and excretion, we measured the weight of bile collected from the gallbladders of mice treated with vehicle or amlexanox. The results showed that amlexanox significantly increased bile weight, confirming the induction of BA synthesis and secretion by amlexanox (Fig. 2f).

To comprehensively assess the effect of amlexanox on BA homeostasis, we examined BA profiles in serum, bile, and feces using ultra performance liquid chromatography mass spectrometry (UPLC-MS) (Supplementary Fig. 2). Notably, we found elevated total bile and fecal BA levels but reduced serum BA concentrations following amlexanox treatment. This indicates that amlexanox upregulates BA synthesis and excretion, while reducing hepatic BA content and BA reabsorption (Fig. 2g–j). Specifically, amlexanox decreased serum levels of multiple BA species (Supplementary Fig. 3, Dataset 1), with tauroursodeoxycholic acid (TUDCA) being the most significantly altered species (Supplementary Fig. 3d). Bile BA profiles demonstrated a predominance of taurine-conjugated BAs, with notable increases in taurocholic acid (TCA), tauro-α-muricholic acid (TαMCA), tauro-β-muricholic acid (TβMCA), among others (Supplementary Fig. 3e–h, Dataset 1). Amlexanox significantly increased the amount of sulfated BAs (DCA-3S, CA-3S, AlloCA-3S, and TDCA-3S) as well as taurine and glycine-conjugated BAs (Fig. 2l, Supplementary Fig. 3g, h). Concurrently, we observed an upregulation of genes responsible for hepatocyte BA and bile salt transport/secretion (Fig. 2k). In feces, levels of bile acid sulfates, particularly cholate-3-sulfate (CA-3S) and allocholate-3-sulfate (AlloCA-3S)-two predominant BAs in feces, were significantly elevated by amlexanox (Fig. 2m, Supplementary Fig. 3i–l). Consistent with these findings, liver transcriptome data showed an upregulation of BA sulfotransferase genes (Sult2a7 and Sult2a8) (Fig. 2n). Sulfation is a process that detoxifies and eliminates BA by increasing their solubility, reducing toxicity and intestinal absorption, and promoting their excretion²⁴. Under normal condition, 95% of BAs undergo intestinal reabsorption via both active transport and passive absorption. Sulfation increases the hydrophilicity and water solubility of BAs, especially hydrophobic BAs, which leads to their excretion rather than reabsorption within the intestine. Our observations indicate that, besides upregulating BA synthesis and secretion, amlexanox also induces BA sulfation to increase their excretion while reducing intestinal reabsorption.

To further clarify whether the reduction in bile acid reabsorption also resulted from altered expression of BA transporter genes in the intestine, we performed RNA-seq analysis on the ileum of both vehicle- and amlexanox-treated mouse groups (Supplementary Fig. 4). Our data revealed significant differences in gene expression associated with BA and bile salt transport, BA metabolism, and fatty acid metabolism (Fig. 2n). Surprisingly, key transporters such as apical sodium-dependent bile salt transporter (Asbt) and ileal bile acid binding protein (Ibabp), which are crucial for active BA reabsorption and transport in enterocytes^24,25, showed significant upregulation in the ileum following amlexanox treatment (Figs. 2o and 2p). This upregulation might be compensatory, induced by the increased BA synthesis and secretion due to with amlexanox treatment. These results indicated that amlexanox upregulates transporter genes involved in BA reabsorption via active transport. Taken together, the increased BA excretion induced by amlexanox is likely mediated by a reduction in passive absorption due to the enhanced BA sulfation.

Amlexanox changes microbiota and increases Akkermansia muciniphila

BAs are critical mediators of interactions between the host and microbiota. Regulation of the intestinal microbiota holds promise as a potential strategy for treating various diseases. To investigate if amlexanox impacts the microbiota, we performed 16S rRNA sequence analysis on fecal samples collected from vehicle- and amlexanox-treated mice. Principal coordinate analysis (PCoA) showed a clear separation between the amlexanox-treated and vehicle groups (Fig. 3a). We then analyzed alpha diversity indices and observed reduced microbial diversity in the amlexanox-treated group, as evidenced by decreased Shannon and Simpson indices (Fig. 3b, c). Hierarchical clustering at the genus level showed distinct patterns between the two groups, suggesting that amlexanox alters the microbial community structure (Supplementary Fig. 5a). At the phylum level, we observed a significant increase in the relative abundance of Verrucomicrobia and a decrease in Firmicutes following amlexanox treatment (Figs. 3d and e; Supplementary Fig. 5b, 6a). Particularly noteworthy was the significant elevation in the relative abundance of Akkermansia muciniphila (A. muciniphila) in the amlexanox-treated group (Figs. 3f and g; Supplementary Fig. 5c, 6b). A. muciniphila is known to play a crucial role in regulating metabolic homeostasis and maturation and function of the immune system^26,27. This suggests that A. muciniphila could be a key player in the beneficial effects of amlexanox on lipid metabolism. Additionally, we observed a reduction in the relative abundance of genera such as Bacteroides, Lactobacillus, and Parabacteroides following amlexanox treatment (Fig. 3f, Supplementary Fig. 6b). These genera produce bile salt hydrolases (BSH), enzymes involved in BA metabolism by deconjugating glycine- or taurine-conjugated BAs into unconjugated BAs²⁸. This finding suggests that amlexanox might modulate BA metabolism by influencing these microbial communities. Through LDA Effect Size (LEfSe) analysis, we identified several potential microbial biomarkers. In the amlexanox-treated group, Akkermansia, Blautia, and Bacteroidetes were notably enriched, whereas Firmicutes were more prevalent in the control group (Fig. 3h, Supplementary Fig. 6c). These findings underscore the profound impact of amlexanox on microbiota composition and suggests potential implications for host health and therapeutic efficacy.

Correlation analysis between microbiota and bile acid species

BAs exert a profound influence on the structure and function of the gut microbiome and impact the abundance, diversity, and metabolic activity of microbial communities in the gut^24,29,30. To investigate if there is an association between microbiota composition and BA metabolism, we performed a correlation analysis between 16S metagenomic data and BA profiles. Significant associations were found between several bacterial genera and BAs. Specifically, certain bacterial taxa, including Akkermansia (otu3), Eubacterium (otu8), Alistipes (otu13), Desulfovibrio (otu15), Lachnoclostridium (otu22), Enterococcus (otu23), and Escherichia-Shigella (otu26), showed positive correlations with the levels of BA sulfates (AlloCA-3S and CA-3S) in feces (Fig. 4a–c, Supplementary Table 1). The fecal concentrations of AlloCA-3S and CA-3S were markedly increased in amlexanox-treated mice (Fig. 2m). These data suggest that these bacteria might affect BA metabolism and excretion processes by regulating sulfation pathways. Moreover, we observed a significant positive correlation between the abundance of Faecalibaculum bacteria (out28) and conjugated BAs, including TαMCA, TβMCA, TCDCA, and TCA (Fig. 4a). This suggests that Faecalibaculum bacteria might be involved in the metabolism and conversion of conjugated BAs, thereby influencing BA circulation and metabolism. Our data indicate that amlexanox affects the interactions between microbiota composition and host BA metabolism in MASH.

**Fig. 4: Correlation analysis of fecal microbiota and bile acid species.**

Amlexanox attenuates MASH-associated inflammation, liver injury, and fibrosis

To explore whether amlexanox alleviates the manifestations of GAN diet-induced MASH after disease onset, we examined liver sections from vehicle- and amlexanox-treated groups. Amlexanox-treated mice showed reduced liver macrophage infiltration, as evidenced by macrophage marker F4/80 staining (Fig. 5a and b). Liver transcriptome analysis revealed significant downregulation of genes involved in inflammatory responses, immune system processes, and interleukin-1 beta production regulation in amlexanox-treated mice (Fig. 5c, Supplementary Fig. 7a). Gene set enrichment analysis (GSEA) of differentially expressed genes (DEGs) identified “chemokine signaling pathway” and “cytokine-cytokine receptor interaction” as the most highly enriched terms (Fig. 5d). KEGG functional enrichment analysis of downregulated genes also highlighted pathways including cell adhesion molecules and focal adhesion (Fig. 5e, Supplementary Fig. 7b–e). Amlexanox significantly reduced the expression of key inflammatory genes, such as macrophage markers Adgre1, Itgax, and chemokines and receptors Ccl2, Ccr2, Il10, Il10ra, Il10rb, Itgax, Tnf (Fig. 5f). Furthermore, serum activities of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) were significantly decreased in amlexanox-treated mice, suggesting improved liver damage (Fig. 5g–i). As shown by sirius red/fast green staining, hepatic fibrosis was substantially decreased in livers of amlexanox-treated mice (Fig. 5j). Consistent with reduced fibrosis, TGF-β signaling and ECM-receptor interaction pathways were significantly enriched among downregulated genes (Fig. 5k). Amlexanox significantly downregulated the expression of fibrotic genes, including Acta2, Col1a1, Col3a1, Itgav, Pdgfra, Pdgfrb, Tgfb1, Tgfbi, Tgfbr1, Timp1, and Timp2 (Fig. 5l). Collectively, these findings indicate that amlexanox therapeutically improves liver inflammation, fibrosis, and injury in MASH, even after disease onset.

Amlexanox prevents the progression of MASH to HCC

To investigate whether amlexanox also prevents the progression from MASH to HCC, Ldlr^−/− mice were fed a GAN diet for 20 weeks, followed by oral gavage with either vehicle or amlexanox for an additional 32 weeks, while continuing GAN diet feeding (Fig. 6a). After 52 weeks of GAN feeding, liver tissues were collected for histological analysis. Mice treated with vehicle developed a significant number of tumors within the liver, whereas the majority of mice treated with amlexanox showed no tumor (Figs. 6b and c). This robust protective effect of amlexanox was further supported by a notable reduction in tumor volume in the liver (Fig. 6d), suggesting that amlexanox effectively prevents the occurrence of MASH-associated HCC.

Our molecular analysis revealed significant downregulation of signaling pathways associated with transcriptional misregulation in cancer and cell cycle in the livers of amlexanox-treated mice (Fig. 6e). This indicates that amlexanox exerts its anti-tumorigenic effects by modulating these key pathways, in addition to its established effects on attenuating MASH-associated hepatic steatosis, inflammation, liver injury, and fibrosis (Figs. 1 and 5). In exploring potential biomarkers and genetic indicators for HCC, epithelial cell adhesion molecule (Epcam), a well-established cell marker for the tumorigenic hepatic stem and progenitor cells (HSPC)^31,32, showed a significant positive correlation with both tumor number (R = 0.9, P = 0.0022) and volume (R = 0.87, P = 0.0051) (Figs. 6f and g). Notably, Epcam gene expression was markedly decreased in the livers of amlexanox-treated mice (Fig. 6h). Consistently, the expression of alpha-fetoprotein (Afp), Survivin (Birc5), Ctnnb1, Mmp7, Hspa14, and Golm1, known biomarkers for HCC, were markedly reduced following amlexanox treatment (Fig. 6i). Furthermore, we observed a significant downregulation of genes related to cell cycle regulation and proliferation (e.g., Ccna2, Ccnd1, Ccnd2, Ccnb1, Mki67) in amlexanox-treated MASH mice (Fig. 6j). Genes associated with the tumor microenvironment, including Cd68, Mrc1, Msr1, and Cd14, also showed significant downregulation in the livers of mice treated with amlexanox (Fig. 6k). Taken together, these findings underscore the potential of amlexanox as a promising agent in preventing MASH to HCC progression through targeting key signaling pathways and molecular mechanisms implicated in hepatocarcinogenesis. Additionally, its multifaceted effects on MASH-associated dyslipidemia, inflammation, liver injury, and fibrosis highlight its therapeutic versatility across disease stages.

Discussion

As a central organ regulating lipid metabolism, the liver oversees circulating triglycerides and cholesterol by managing the secretion and uptake of lipoproteins. Given the profound impact of circulating lipids, especially cholesterol, on the development of atherosclerotic cardiovascular diseases, it is imperative to consider the intricate crosstalk between lipid metabolism in the liver and the cardiovascular system when addressing MASH. Previous trials, such as those involving obeticholic acid, an FXR (farnesoid X receptor) agonist aimed at inhibiting cholesterol-derived bile acid synthesis for MASH therapy, have revealed potential risks, including elevated circulating cholesterol and LDL-cholesterol levels³³, heightening the risk of atherosclerosis. Conversely, Lomitapide, an inhibitor of MTTP (microsomal triglyceride transfer protein), effectively lower circulating cholesterol levels by reducing the production of Apo-B and thereby diminishing the secretion of chylomicrons, VLDL, and LDL. However, Lomitapide also comes with drawbacks, such as hepatic lipid accumulation and fluctuating serum aminotransferase levels³⁴. In light of these complexities, developing a novel therapy that simultaneously addresses fatty liver diseases and atherosclerosis poses a significant challenge yet remains crucial. Our study presents compelling evidence that amlexanox effectively reduces both circulating and hepatic lipid levels, proving its potential as an effective therapy for managing both MASH and atherosclerotic cardiovascular diseases.

To gain deeper insights into how amlexanox affects MASH, we performed transcriptomic analysis of both the liver and intestine. Our results revealed a significant upregulation of key genes involved in BA synthesis and transport, notably Cyp7a1 and Abcb11, upon amlexanox treatment. Further analysis suggested that amlexanox enhanced BA synthesis in the liver and promoted BA excretion in feces. This dual mechanism leads to improved lipid metabolism and diminished hepatic cholesterol accumulation. BA serves as a primary route for cholesterol excretion. However, approximately 95% of BAs undergo reabsorption in the intestine. BA sequestrants that enhance BA excretion through reducing reabsorption have been used as cholesterol-lowering medications. Our study found that amlexanox not only boosts BA synthesis but also augments BA excretion, thus yielding more potent cholesterol-lowering effects.

Moreover, we observed that amlexanox exerts a beneficial effect on microbiota composition, particularly by increasing the abundance of A. muciniphila. Gut microbiota plays an important role in regulating various metabolic functions, including glucose homeostasis, inflammation, and nutrient absorption³⁵. Numerous studies have indicated that a lower abundance of A. muciniphila is associated with various diseases in both mouse models and humans²⁷. A. muciniphila has been proven to improve obesity, diabetes mellitus, hepatic steatosis, and intestinal inflammation²⁷. Supplementation with live A. muciniphila reversed high-fat diet-induced metabolic disorders, including fat-mass gain, metabolic endotoxaemia, adipose tissue inflammation, and insulin resistance²⁷. Additionally, A. muciniphila positively impacts metabolic health through diverse mechanisms. It ameliorates inflammation and improves glucose tolerance in obese and diabetic patients^26,27,35. However, there remains a scarcity of pharmaceutical interventions aimed at increasing A. muciniphila, with options limited to dietary polyphenol-rich extracts and direct supplementation of viable A. muciniphila³⁶. Although metformin showed moderate effects on restoring A. muciniphila in diet-induced obese mice^36,37,38, the quest for potent interventions persists. In this study, we observed that amlexanox markedly increased A. muciniphila abundance, while concurrently improving pathological hallmarks of MASH. This discovery provides a novel therapeutic strategy that boosts A. muciniphila in metabolic disorders.

Furthermore, we observed a strong positive correlation between microbiota and specific BA salts, indicating a complex and intimate interplay between microbiota and BA metabolism. Studies have shown that microbial BA metabolism pathways can diversify the BA pool and regulate its excretion, thus influencing BA homeostasis in the human body²⁴. Among these metabolic pathways, sulfation orchestrated by intestinal bacteria such as Clostridium, Enterococcus, Clostridioides, and Pseudomonas²⁴, plays a crucial role. This sulfation process serves to detoxify and clear BAs by enhancing their solubility, reducing toxicity and intestinal absorption, and facilitating their excretion²⁴. Specifically, we found that specific bacterial taxa, including Akkermansia and Desulfovibrio, positively correlate with the levels of bile sulfate salts (AlloCA-3S and CA-3S) in feces. Notably, the fecal concentration of AlloCA-3S and CA-3S was substantially increased in amlexanox-treated mice. These findings strongly suggest that these bacteria might intricately modulate BA metabolism and excretion processes through the regulation of of BA sulfation pathways.

BAs serve as critical regulators of lipid metabolism and overall metabolic homeostasis. Mounting evidence suggests that disruption in BA composition and signaling pathways are implicated in MASH patients, suggesting a potential association between BA metabolism and liver disease¹⁸. The dynamic interplay of BAs in the gut-hepatic axis emphasizes their crucial role as signaling molecules, underscoring the complex crosstalk between BAs and the gut microbiome in shaping metabolic health and disease. Growing evidence indicates that disruption of the gut-liver axis is a key driver in the progression of numerous chronic liver diseases²¹. This disruption manifests in various ways, including alterations in gut microbiota composition and damage to the intestinal barrier, resulting in shifts in serum BA levels²¹. Elevated serum BAs caused by dysbiosis, exacerbate hepatic inflammation and steatosis, thereby contributing to the pathogenesis of MASH^19,20,21,22. Furthermore, the microbiota can produce large amounts of secondary BAs, such as deoxycholic acid (DCA), which antagonize FXR and lead to enhanced BA synthesis and reabsorption. Consequently, this contributes to increased serum concentrations of both primary and secondary BAs^39,40. Remarkably, our study reveals that that amlexanox differentially regulated BA metabolism by increasing total BA levels in bile and feces while decreasing serum BA levels concurrently. The reduction of both primary and secondary BAs is achieved through enhancing BA sulfation and facilitating their excretion, despite increasing BA synthesis. By doing so, amlexanox not only enhances cholesterol-derived BA synthesis but also mitigates potential adverse effects associated with increased BA synthesis. Thus, our study provides new insights into potential therapeutic strategies targeting BA metabolism in the gut-liver axis for MASH treatment.

Previous studies have consistently shown that amlexanox has several beneficial effects, including increased energy expenditure, reduced inflammation, improved dyslipidemia, and attenuation of obesity and diabetes^12,16,41,42. Notably, amlexanox has been found to reduce food intake transiently within the first three days of treatment but does not significantly affect long-term food intake⁴¹. Our current study uncovered a novel aspect of amlexanox’s impact on bile acid metabolism and the microbiome in the context of MASH and associated HCC. Our findings also reveal that amlexanox reverses hepatic steatosis and inflammation while significantly ameliorating fibrosis. These beneficial effects on MASH are likely mediated by the influences on different cell types, including hepatocytes, immune cells (particularly macrophages), and hepatic stellate cells. Moreover, our preliminary studies observed that a single treatment of amlexanox does not induce bile acid synthesis gene expression in primary hepatocytes or cell lines like AML12 or HepG2. Notably, there was a mild upregulation of Cyp7a1 expression one day after starting amlexanox treatment. A significant upregulation of Cyp7a1 expression and moderate changes in the expression of other bile acid synthesis genes after three days (daily gavage of amlexanox). These observations suggest that a complex systemic or multi-organ interaction may be crucial in mediating the effects of amlexanox. Therefore, further mechanistic investigation focusing on the multi-organ crosstalk is still needed to elucidate the comprehensive effects of amlexanox.

Methods

Animals

All procedures involving mice were conducted in accordance with the Guide for Care and Use of Laboratory Animals of the NIH (National Academies Press, 2011). All animal use was approved by the Institutional Animal Care and Use Committee (IACUC) of University of Texas Health Science Center San Antonio. Ldlr^−/− mice were purchased from Jackson Laboratory (stock #002207) and maintained on a C57BL/6 J background. All mice were housed in the UTHSCSA pathogen-free animal facility with a 12-hour dark/12-hour light cycle and given free access to water and food, except for the fasting period. Male Ldlr−/− mice were fed a GAN diet consisting of 40 kcal% Fat, 20 kcal% Fructose, and 2% Cholesterol (Research Diets, D09100310) starting at 8 weeks of age for 32 weeks to establish the MASH model and for 52 weeks for the HCC model. Amlexanox treatment commenced after the development of hepatic inflammation and liver damage after 20 weeks of GAN diet feeding. Mice were orally gavaged 25 mg/kg amlexanox (Abcam, Cat. ab142825) or vehicle every other day. Feces were collected 2 days before the sacrifice. Ad libitum-fed mice were temporally single-housed and monitored for feces collection. Fresh feces were immediately collected after being produced. At sacrifice, mice were fasted overnight and then euthanized to collect blood and tissues. All samples were stored at −80 °C until analysis.

For the evaluation of experimental neoplasia, including constraints on tumor size (with individual tumors not exceeding 2 cm) and monitoring parameters, tumor volumes were computed using the formula (width² × length)/2. In cases of multiple liver tumors, the volumes of each tumor were summed to determine the total tumor volume^43,44.

Triglyceride and cholesterol measurement

Serum and tissue triglyceride levels were measured using the Triglyceride Quantification Colorimetric/Fluorometric Kit (Abcam, Cat. ab65336). Total cholesterol and cholesterol ester levels were determined using the Total Cholesterol and Cholesterol Ester Colorimetric/Fluorometric Kit (Abcam, Cat. ab65359), following the manufacturer’s instructions. Tissue triglyceride/cholesterol levels were normalized to tissue weight.

Atherosclerosis analysis

The aortas were dissected under a microscope, fixed in 4% formalin-sucrose, opened, flattened, pinned, and stained with Sudan IV. Images of the aortas were captured and quantified by analyzing the entire en-face aorta as previously described^16,45. Aortic root cross-sectional lesion areas were quantified using serial cross-sections taken at 100 μm intervals between 100 μm and 900 μm, from the first appearance of the first leaflet of the aortic valve to the last leaflet. The mean lesion size at each 100 μm section in each animal was determined by computer-assisted morphometry on serial 10 μm paraffin sections. Modified van Gieson elastic stain was employed to enhance the contrast between the intima and surrounding tissue. Cross-sectional plaque area and plaque morphology were evaluated blindly. The results are presented as the mean of all values for each interval plotted versus the distance from the first leaflet, and the overall extent of aortic root lesions was determined by AUC analysis of all serial sections in each group. Mac-3 staining on aortic roots was performed using anti-Mac-3 antibody (Santa Cruz Biotechnology, catalog sc-20004).

Histology

Liver tissues from Ldlr^−/− mice were fixed in 10% neutral buffered formalin, embedded in paraffin, and sectioned and processed by hematoxylin and eosin (H&E) or immunohistochemistry (IHC) staining. For H&E staining, tissue sections were deparaffinized and rehydrated, the nuclei were stained with hematoxylin for 5 min. Sections were then rinsed in running tap water and stained with eosin for 1 min, dehydrated, and mounted.

For IHC staining, paraffin-embedded tissue sections were deparaffinized and rehydrated and then immersed in 95 °C antigen retrieval buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) for 30 min. Tissue sections were treated with 3% H₂O₂ for 30 min to block endogenous peroxidase activity and then incubated with 5% bovine serum albumin (BSA, Sigma-Aldrich) for 30 min to block non-specific antibody binding. The samples were incubated with primary antibody at 4 °C overnight. Anti-F4/80 (Bio-Rad, MCA497) was used to detect macrophage content. After washing, sections were incubated with HRP-conjugated secondary antibody for 1 h. Color detection was carried out using 3,3’-diaminobenzidine (DAB), and nuclei were counterstained with hematoxylin solution. Quantitative image analysis was performed using Image J software.

For fibrosis staining, paraffin-embedded liver sections were subject to de-paraffinization and then stained with Sirius Red (Sigma Aldrich)/Fast Green (Fisher Scientific). Images were taken with the NanoZoomer Slide Scanner.

ALT, AST, ALP activity measurement

Serum ALT activity was measured with Alanine Aminotransferase Activity Colorimetric/Fluorometric Assay Kit (Abcam, Cat. ab105134). AST activity was measured with Aspartate Aminotransferase Activity Colorimetric Assay Kit (Abcam, Cat. 105135). ALP was measured with Alkaline Phosphatase Activity Colorimetric Assay Kit (Abcam, Cat. ab83369), according to the manufacturer’s instructions.

QPCR gene expression analysis

Tissues were homogenized in TRIzol reagent (Life Technologies). RNA was isolated with PureLink RNA Mini Kit (Life Technologies). A total of 500 ng of purified RNA was used for reverse transcription PCR to generate cDNA using PrimeScript RT Kit (Takara). ΔΔCt real-time PCR with Power SYBR Green PCR Master Mix (Life Technologies) and BioRad CFX Opus Real-Time PCR System were used to analyze cDNA. Primers are listed in Supplementary Table 2.

RNA-seq library preparation

Total RNA was isolated from mice livers and ileums homogenized with TRIzol reagent and purified using Quick RNA mini prep columns and RNase-free DNase digestion according to the manufacturer’s instructions (Life Technologies). RNA quality was measured by a Tapestation system. Sequencing libraries were prepared in biological replicates from polyA-enriched mRNA. RNA-seq libraries were prepared from poly(A)-enriched mRNA as previously described⁴⁶. Libraries were size selected and purified by Speedbeads, quantified using a Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), and sequenced on a Hiseq 4000 (Illumina) according to the manufacturer’s instructions.

RNA-seq analysis

RNA-Seq analysis was performed as previously described^44,47. FASTQ files from sequencing experiments were mapped to the mouse mm10 genome. STAR with default parameters was used to map RNA-Seq experiments⁴⁸. To compare differential gene expression between indicated groups, HOMER’s analyzeRepeats with the option rna and the parameters -condenseGenes, -noadj, and -count exons was used on at least three replicates per condition⁴⁹. Each sequencing experiment was normalized to a total of 10⁷ uniquely mapped tags by adjusting the number of tags at each position in the genome to the correct fractional amount given the total tags mapped. Sequence experiments were visualized by preparing custom tracks for the UCSC Genome Browser. Differential gene expression was assessed with DESeq2 using HOMER’s getDiffExpression.pl with the parameters –P-adj 0.05 and -log₂ fold 0.585 (for 1.5-fold differently expressed genes)⁵⁰. For all genes, the TPM values were plotted and colored according to fold change. For various ontology analyses, HOMER, Metascape and KEGG were used⁵¹.

BA mass spectrometry

LC-MS system: An Agilent 1290 UHPLC coupled to a Sciex 4000 QTRAP mass spectrometer was used. The MS instrument was operated in the multiple-reaction monitoring (MRM) mode with negative-ion (-) detection. A Waters BEH C18 column (2.1 mm I.D. x 15 cm, 1.7 μm) was used for LC separation with a mobile phase composed of (A) 0.01% formic acid in water and (B) 0.01% formic acid in acetonitrile for binary-solvent gradient elution. Detailed LC and MS operation parameters and analytical sensitivities were described in Anal Chem⁵².

Standard solutions and calibration curves: A mix containing standard substances of all the measured bile acids was dissolved in 50% methanol at 10 nmol/mL for each. This solution was further diluted step by step at a same dilution ratio of 1 to 4 (v/v) with the same solvent to have standard solutions of each sample. 50 μL of each sample was mixed with 50 μL of an internal standard (IS) solution containing 14 D-labeled bile acids. 15 μL of each solution was injected to run UPLC-(-) ESI-MRM/MS. Linear-regression calibration curves were constructed using analyte-to-internal standard peak area ratios (As/Ai) versus molar concentrations (nmol/mL) of the standard solutions for each bile acid. The results are provided as the spreadsheet in supplementary materials.

Mouse bile: 20 μL of bile was mixed with 980 μL of methanol. The mixture was vortex and sonicated for 3 min before clarification by centrifugation at 15,000 rpm in an Eppendorf 5420 R centrifuge for 15 min. 200 μL of the supernatant was mixed with 200 μL of the IS solution. 15 μL was injected for quantitation of bile acids by UPLC-MRM/MS. For quantitation of high-concentration of bile acids, the solution was diluted 10-fold for a second round of injection.

Mouse serum: 50 μL of mouse serum was mixed with 50 μL of the IS solution and 100 μL of acetonitrile in an Eppendorf tube. After vortex mixing for 15 s and sonication for 2 min in an ice-water bath, the tube was centrifuged at 15,000 rpm and 10 ÅãC for 15 min in an Eppendorf 5420 R centrifuge. The supernatant was taken out and mixed with 830 μL of water. The mixture was loaded onto a reversed-phase Strata-X SPE cartridge (60 mg/L mL). After sample loading, the cartridge was washed with 2 Å~ 1 mL of water, and bile acids were eluted with 3 Å~ 1 mL acetonitrile. The collected fraction was dried in a nitrogen evaporator under a gentle nitrogen gas flow. The residues were dissolved in 100 μL of 50% methanol. 15 μL was injected for quantitation of bile acids by UPLC-MRM/MS.

Mouse feces: The fecal sample was weighed out and 70% acetonitrile was added to make the concentration as 100 μL/mg raw material. After homogenization, sonication, and centrifugation, 20 μL of the supernatant was mixed with 100 μL of the IS solution and 880 μL of water. The mixture was loaded onto a reversed-phase Strata-X SPE cartridge (60 mg/L mL). After sample loading, the cartridge was washed with 2 Å~ 1 mL of water and bile acids were eluted with 3 Å~ 1 mL acetonitrile. The collected fraction was dried in a nitrogen evaporator under a gentle nitrogen gas flow. The residues were dissolved in 200 μL of 50% methanol. 15 μL was injected for quantitation of bile acids by UPLC-MRM/MS.

16S rRNA sequencing

Fecal samples from vehicle and amlexanox groups were randomly selected for microbiota 16S rRNA analysis. Microbial genomic DNA was extracted and purified using a QIAamp DNA Mini Kit (Qiagen, Germany). Agarose gel electrophoresis was used to determine the DNA quality. The V4 region was selected for 16S rRNA amplification, and the universal primers used were 515 F (5’-GTGCCAGCMGCCGCGGTAA-3′) and 806 R (5’-GGACTACHVGGGTWTCTAAT-3′). PCR products were purified by AMPure XP beads (Beckman Coulter Genomics, Danvers, MA, USA) and quantified by Qubit (Invitrogen, USA). After evaluation using an Agilent 2100 Bioanalyzer (Agilent, USA) and Illumina (Kapa Biosciences, Woburn, MA, USA) library quantification kits, the purified PCR products were sequenced on the Illumina platform according to standard protocols. >100,000 reads were collected for each sample using illumina Miseq (https://earthmicrobiome.org/protocols-and-standards/16s/). For analysis, DATA2 v1.18 (https://github.com/benjjneb/dada2)⁵³ was used.

Following sequencing, bacterial Operational Taxonomic Units (OTUs) were counted for each sample to assess the richness of bacterial species. Bacterial OTUs were generated using QIIME analysis (http://qiime.org/scripts/pick_otus.html). Taxon-dependent analysis was conducted using the Greengene database. Alpha diversity was analyzed using Observed species, Shannon index, and Simpson index and calculated with QIIME2. Beta diversity was determined by QIIME2.

Statistical analysis

All data are shown as mean ± SEM. Replicates are indicated in figure legends. N represents the number of experimental replicates. An f test was performed to determine the equality of variance. When comparing 2 groups, statistical analysis was performed using a two-tailed Student’s t test, except when the f test suggested that variances were statistically different. For analysis of more than 2 groups, we used ANOVA to determine equality of variance. Comparisons between groups were performed with Tukey-Kramer post hoc analysis. For all tests, P < 0.05 was considered statistically significant.