The TBK1/IKKε inhibitor amlexanox improves dyslipidemia and prevents atherosclerosis

Cardiovascular diseases, especially atherosclerosis and its complications, are a leading cause of death. Inhibition of the noncanonical IκB kinases TANK-binding kinase 1 and IKKε with amlexanox restores insulin sensitivity and glucose homeostasis in diabetic mice and human patients. Here we report that amlexanox improves diet-induced hypertriglyceridemia and hypercholesterolemia in Western diet–fed (WD-fed) Ldlr–/– mice and protects against atherogenesis. Amlexanox ameliorated dyslipidemia, inflammation, and vascular dysfunction through synergistic actions that involve upregulation of bile acid synthesis to increase cholesterol excretion. Transcriptomic profiling demonstrated an elevated expression of key bile acid synthesis genes. Furthermore, we found that amlexanox attenuated monocytosis, eosinophilia, and vascular dysfunction during WD-induced atherosclerosis. These findings demonstrate the potential of amlexanox as a therapy for hypercholesterolemia and atherosclerosis.


Introduction
Metabolic diseases have become a worldwide epidemic (1). Atherosclerosis and its complications, including heart attack and stroke, are the leading causes of death (2,3). The origins of atherosclerosis are complex and multifactorial, and often linked via common underlying mechanisms. For example, hypercholesterolemia and hypertriglyceridemia are frequently associated with chronic inflammation, leading to excessive accumulation of monocyte-derived macrophages in the arterial wall that contributes to the development of atherosclerotic plaques (2,(4)(5)(6). Atherosclerosis is currently treated primarily with statins, ezetimibe and PCSK9 inhibitors to decrease plasma cholesterol (7)(8)(9)(10). Niacin was also shown to decrease LDL-cholesterol and increase HDL-cholesterol (11)(12)(13). However, the use of these agents is not always optimally efficacious and at times associated with problems. Some individuals cannot tolerate statins, the most widely used agents, due to myopathies and occasionally increased blood glucose and 3 insulin resistance (14)(15)(16). Other drugs that reduce triglycerides (Fibrates) or decrease bile acid reabsorption (bile acids sequestrants) are not as effective as statins, and carry other liabilities (17)(18)(19).
While novel convertase subtilisin kexin type 9 (PCSK9) inhibitors, alirocumab and evolocumab, have recently been introduced to control cholesterol in patients who do not respond to statins, these drugs are expensive (20)(21)(22). Thus, there is a need for new safe and effective drugs to combat this devastating disease.
Obesity is characterized by low grade, persistent inflammation in adipose tissue and liver, involving the recruitment and activation of pro-inflammatory immune cells (23)(24)(25). These inflammatory events are characterized by activation of the transcriptional factor NFκB in both immune cells and metabolically active hepatocytes and adipocytes, linking obesity to both cardiovascular and metabolic disease (26)(27)(28)(29).
Studies from our laboratory on the NFkB pathway in adipose tissue and liver from obese mice revealed that both the noncanonical IkB kinases (IKKs), IKKe and TANK-binding kinase 1 (TBK1) are elevated in obesity due to NFkB activation, and further that both proteins play a role in suppressing energy expenditure in the obese state (28,30). These findings led us to discover amlexanox as a specific inhibitor of both kinases (30). This drug was developed in the mid-1980's to treat asthma and allergic rhinitis (31,32), and has an excellent record of safety. We demonstrated that amlexanox substantially improved glucose tolerance, fatty liver and insulin sensitivity, and reduced hepatic steatosis in genetically obese and diet-induced obese (DIO) mice (30,33,34), and significantly reduced HbA1c levels in a subset of diabetic patients with high basal levels of systemic inflammation (35). Mechanistic studies revealed that amlexanox reduced expression of pro-inflammatory cytokines genes Ccl2, Ccl3, and attenuated inflammation (30). Moreover, amlexanox inhibits IKKe-induced activation of phosphodiesterase 3B (PDE3B) to elevate cAMP levels and p38 phosphorylation in adipocytes, and thus increases catecholamine sensitivity and energy expenditure via increased adipose tissue browning and thermogenesis (36,37). However, it is unknown whether amlexanox could affect other diet-induced metabolic diseases, especially atherosclerosis. In this study, we assessed the effects of amlexanox on 4 Western diet (WD)-induced atherosclerosis in Ldlr -/mice. We examined its effects on lipid metabolism, inflammation and vascular dysfunction, and demonstrate that amlexanox systemically ameliorates three major pathogenic mechanisms that promote atherogenesis. Given its beneficial effects in obesity, diabetes and fatty liver diseases, here we demonstrate the potential of amlexanox as a simultaneous treatment for atherosclerosis and other metabolic diseases, including diabetes and fatty liver disease.

Amlexanox improves dyslipidemia and protects against atherosclerosis.
Amlexanox is a selective inhibitor of the protein kinases TBK1 and IKKe (30), and its administration to obese rodents or humans improved energy and glucose metabolism (30,35). To examine whether amlexanox exerts a beneficial effect on Western diet (WD)-induced atherosclerosis, we fed Ldlr -/mice with WD for 3 weeks, and then orally gavaged the mice with vehicle or amlexanox for 8 weeks with the continuation of WD feeding ( Figure 1A). Consistent with our previous findings, amlexanox improved diet-induced obesity, indicated by significantly reduced body weight and adipose tissue weight in WDfed mice (Supplementary Figure 1B-1D). After 11 weeks of WD feeding, aortas were collected to evaluate lesion development. En face staining demonstrated that amlexanox substantially reduced the area of aortic lesions ( Figure 1B-1C). Staining of aortic roots also showed that amlexanox significantly reduced the size of lesions ( Figure 1D-1E). Together, these data demonstrated that amlexanox reduced atherogenesis in Our previous studies demonstrated that amlexanox reduced blood glucose and improved insulin sensitivity in ob/ob and high fat diet (HFD)-fed mice (30). We thus examined the impact of amlexanox on dietinduced dyslipidemia, including hypertriglyceridemia and hypercholesterolemia. We found that mice gavaged with amlexanox had clear serum, while serum from mice in the vehicle group was milky ( Figure   1F Figure 2E). Together, these data demonstrated that amlexanox markedly improved dyslipidemia in WD-fed Ldlr -/mice.

Amlexanox increases bile acid synthesis and cholesterol excretion.
Our previous study demonstrated that amlexanox significantly increases energy expenditure to improve obesity and hypertriglyceridemia (30). To elucidate the mechanism by which amlexanox ameliorates hypercholesterolemia in WD-fed Ldlr -/mice, we systematically examined cholesterol metabolism. Mice 6 were fed WD for 8 weeks with 4 additional weeks of feeding along with vehicle or amlexanox administration. 4 weeks of amlexanox treatment significantly reduced circulating levels of cholesterol and triglyceride in WD-fed Ldlr -/mice (Figure 2A-2B). We assessed the rate of cholesterol absorption by assaying serum 14 C radioactivity 2 hrs after gavage of cold cholesterol mixed with 14 C-cholesterol. These data demonstrated that amlexanox did not affect cholesterol absorption ( Figure 2C). Because liver is the major site of cholesterol synthesis, we also examined hepatic cholesterol synthesis rate in response to amlexanox administration. Mice gavaged with vehicle or amlexanox were injected with 3 H-acetate, and 3 H radioactivity in the liver sterol fraction was determined after 2hrs. Our results demonstrated that amlexanox did not affect radioactivity in the liver sterol fraction ( Figure 2D), indicating no effect on cholesterol synthesis.
Given that amlexanox does not affect cholesterol absorption or synthesis, we examined whether the drug might control cholesterol excretion to attenuate hypercholesterolemia. WD-fed Ldlr -/mice received vehicle or amlexanox, followed by oral gavage with cold cholesterol mixed with 14 C-cholesterol. After 21hrs, radioactivity was measured in serum, feces and liver lysates. Interestingly, radioactivity in serum was significantly reduced in mice gavaged with amlexanox, but significantly increased in the liver lysates, bile, and feces, indicating a dramatic increase in cholesterol clearance ( Figure 2E-2H).
To understand the mechanism by which amlexanox increases cholesterol excretion, we performed RNAseq on liver tissue of vehicle or amlexanox-treated mice. Our data revealed that amlexanox induced profound transcriptional changes in the livers of WD-fed Ldlr -/mice ( Figure  Cholesterol is mainly excreted in the form of bile acids. To ascertain whether amlexanox affects bile acid synthesis and excretion, we measured bile acid levels in the feces of mice gavaged with vehicle or amlexanox. Amlexanox significantly increased the amount of fecal bile acids ( Figure 2L). Together, these data suggest that inhibition of the protein kinases TBK1 and IKKe by amlexanox upregulates bile acid production to increase cholesterol excretion, and thus ameliorates hypercholesterolemia.

Amlexanox reduces circulating monocytes and lesion macrophage.
Inflammation is a major pathogenic factor for atherosclerosis. Under atherosclerotic conditions, macrophages infiltrate into the blood vessel wall and promote the formation of the necrotic core, which is a defining feature of unstable plaques (38). Staining of plaques with the macrophage marker macrophage antigen-3 (Mac-3) showed a significant attenuation of macrophage in aortic lesions ( Figure 3A). To understand the underlying mechanism, we did a complete blood count to evaluate circulating immune cells. Interestingly, we found that amlexanox significantly reduced the number of monocytes and eosinophils, while neutrophil, lymphocyte, and basophil numbers were unaffected ( Figure 3B-3F).
Circulating monocytes are the major source of infiltrated macrophages, and monocytosis has been linked to atherosclerosis (38). These data indicate that amlexanox attenuates monocytosis, which in turn would lead to decreased numbers of monocytes recruited to the artery wall, contributing to the decrease of atherosclerosis. In contrast, amlexanox had no effect on the number of blood monocytes in mice fed normal chow diet (Supplementary Figure 4). We note that amlexanox was originally developed as an asthma treatment (39). Since eosinophilia is a major pathogenic factor in the development of asthma, the anti-eosinophilic effects of the drug might provide insight into the mechanism of amlexanox's anti-asthma function.

Amlexanox protects against vascular dysfunction in vitro and in vivo.
Vascular dysfunction is a major contributor to the pathogenesis of atherosclerosis (40)(41)(42)(43). Endothelial cells that are dysregulated in atherosclerosis trigger monocyte adhesion, and thus promote macrophage 8 infiltration into the aortic vessel (3,4,(41)(42)(43). Increased proliferation and migration of smooth muscle cells (SMCs) is also an indispensable component of plaque formation (44,45). To understand whether amlexanox affects monocyte-endothelial cell adhesion, we labeled THP-1 monocytes with the radiometric pH indicator BCECF, and treated human aortic endothelial cells (HAECs) with vehicle or TNFa. After removal of BCECF and TNFa, monocytes were incubated with HAECs. Non-adherent monocytes were washed away, and monocytes adhering to endothelial cells were visualized by fluorescence microscopy.  Figure 4E). We also utilized a Transwell cell migration assay to study amlexanox's effect on SMC migration. Amlexanox significantly reduced the migration of SMCs induced by the addition of serum from WD-fed Ldlr -/mice ( Figure 4F). Together, these findings demonstrate that amlexanox attenuates the proliferation and migration of smooth muscle cells, which could contribute to the amelioration of atherogenesis.
To elucidate the underlying mechanism by which amlexanox prevents cardiovascular dysfunction to protect against atherosclerosis, we dissected the aorta from WD-fed Ldlr -/mice gavaged with vehicle or amlexanox and performed RNA-seq analysis on whole aorta. We observed 157 genes with levels of expression in aorta of amlexanox-treated mice that are at least 1.5-folder less than the vehicle-treated aorta at an FDR of 0.05 and TMP greater than 16 ( Figure 5A, Supplementary Figure 5A). Ontological analysis of genes downregulated by amlexanox demonstrated significant functional enrichment of atherogenic related categories, such as inflammatory response, cell chemotaxis, smooth muscle cell proliferation, and cell migration ( Figure 5B). The expression of genes involved in inflammation and smooth muscle cell proliferation and migration are shown in Figure 5C-5D. Gene set enrichment analysis for differentially expressed transcripts in aorta revealed a significant downregulation of the inflammatory response, TNFa signaling, TGF signaling, and apoptotic pathways by amlexanox ( Figure 5E, Supplementary Figure 5B).
a-SMA (alpha-smooth muscle actin) staining showed a significant reduction of smooth muscle cells within the aortic lesions ( Figure 5F), confirming that amlexanox attenuated SMC migration into plaques.

Discussion
Atherosclerosis and its complications, like heart attack and stroke, are the leading causes of death in modern society (2,3). Atherogenesis results from hypercholesterolemia (46,47), systemic chronic inflammation (3,48,49), and aortic cell dysfunctions (40)(41)(42)(43)50), conditions that often appear together in cardiovascular disease patients. In this study, we found that amlexanox improves all three aspects of this syndrome to prevent atherosclerosis. Hypercholesterolemia, especially high blood concentrations of VLDL and LDL cholesterol, is essential for the development of aortic plaques (46). We showed that amlexanox significantly improved dyslipidemia and reduced both VLDL-cholesterol and LDL-cholesterol in WD-fed Ldlr -/mice. Chronic inflammation is characterized by penetration of monocytes into aortic vessels, and monocyte-derived macrophages form the necrotic core of atherosclerotic lesions (3,51).
Infiltration of monocytes/macrophages modulates the development of lesions, and affects their stability through cytokine secretion and cross-talk with artery wall cells (5). Oral gavage of amlexanox significantly reduced circulating monocytes as well as macrophage in aortic lesions. Additionally, endothelial cell dysfunction results in an up-regulation of pro-inflammatory cytokines and adhesion molecules, which further increase monocyte adhesion (4). The phenotypic switch of smooth muscle cells from "contractile" to "synthetic" increases the proliferation of these cells (44). Increased expression of matrix metalloproteinases (MMPs) in aortic vessels cells results in extracellular matrix (ECM) remodeling, which promotes SMC migration into lesions (45). Our study demonstrated that amlexanox not only attenuates monocyte-endothelial adhesion, but also reduces SMC proliferation and migration to ameliorate the dysfunction of aortic vessel cells. Taken together, amlexanox improves all three aspects of this syndrome to protect against atherogenesis.
To understand the mechanism by which amlexanox improves these pathogenic aspects during atherogenesis, we profiled the transcriptome in liver and aorta. Bioinformatic analyses indicated that amlexanox significantly increased the expression of genes mediating bile acid synthesis in the liver, while attenuating the expression of inflammatory genes and the genes involving smooth muscle cell proliferation and migration. Our previous study demonstrated that amlexanox inhibits IKKe/TBK1 to attenuate inflammation (30,35). Here, we showed that amlexanox significantly reduced the expression of inflammation genes in the liver. Given that inflammation, especially via the production of TNFa, could repress the expression of Cyp7a1 (52,53), we speculate that the improvement of systemic inflammation in WD-fed Ldlr -/mice is responsible for the increased expression of bile acid synthesis genes. A recent study showed that amlexanox inhibits IKKe/TBK1 to repress TGFb production (54). Consistent with this work, we found that amlexanox downregulated TGFb signaling in aorta, which could result in the attenuation of smooth muscle cell proliferation and migration.
While current anti-atherosclerosis drugs are effective, their use is limited in certain populations. Most patients with metabolic syndrome present with hypercholesterolemia, hyperglycemia and insulin resistance. Although the use of statins in these population reduces the risk of CVD, their use maybe associated with the risk of a drug-induced increase in blood glucose, potentially exaggerating diabetes in certain patients (14)(15)(16). Therefore, there is a need for new anti-atherosclerotic medications for a subgroup of patients with metabolic syndrome. Our previous studies showed that amlexanox significantly improves insulin sensitivity and glucose metabolism, while reducing hepatic steatosis in obese mice (30,33,34). A proof-of-concept clinical study also demonstrated that 12 weeks of amlexanox treatment was safe, and significantly decreased HbA1c levels, an effect that was more pronounced in a subset of patients with high systemic inflammation at baseline (35). The profound effects of the drug prompted us to evaluate amlexanox in a mouse model of atherosclerosis. Our findings demonstrate that amlexanox substantially attenuates diet-induced hypercholesterolemia and reduced aortic lesion area. Mechanistic studies suggest that amlexanox increases cholesterol excretion, prevents monocytosis, and attenuates aortic vessel cell dysfunction, indicating that amlexanox may represent a novel approach to the treatment of hypercholesterolemia and atherosclerosis in statin-resistant or metabolic syndrome patients with hyperglycemia, dyslipidemia and fatty liver diseases. More importantly, these data illustrate a potentially important role for the amlexanox targets, IKKe and TBK1, as a driver of cardiovascular disease.

Mice
Mice were used in accordance with the Guide for Care and Use of Laboratory Animals of the National Institute of Health. Mice were housed in a specific pathogen-free facility with a 12-h light, 12-h dark cycle, and given free access to food and water, except for fasting period. At 8 weeks of age, 18-20g male Ldlr −/− mice (The Jackson Laboratory, #002207) were matched for age, body weight and total cholesterol and placed on a Western diet (WD) (Research Diets, D12079Bi). Mouse were orally gavaged 25mg/kg amlexanox (Abcam, Cat. ab142825) or vehicle every other day with the continuation of WD feeding.
During the feeding, mice were weighed monthly. At sacrifice, tissues were weighed. Blood samples were collected. Complete blood count was done by hematology core at UCSD. Liver H&E staining was carried 12 out by UCSD histology core. Stained tissue was visualized with NanoZoomer Slide Scanner at UCSD School of Medicine microscopy core. Lipidomic study was performed at UCSD lipidomics core.

Atherosclerotic plaque analysis
The aortas were dissected under a microscope and fixed in 4% formalin-sucrose, opened, flattened pinned and stained with Sudan IV, and images of the aortas were captured and quantified by analysis of the entire en face aorta as previously described (55). Aortic root cross-sectional lesion areas were quantified using Lipoprotein profiling was performed on terminal blood samples from a pool of 5 mice using fast performance liquid chromatography equipped with a Superose 6 column, and total cholesterol and triglycerides levels in each fraction were determined as described above.

Fecal bile acid measurement
Feces were weighed out. Bile acids were extract using 90% ethanol with 0.1N NaOH. Bile acids were determined using Mouse Total Bile Acids Assay Kit (Crystalchem, Cat. 80471) according to the manufacturer's instruction.

Gene expression analysis
Analysis of gene expression was performed as previously described (56,57 AGCGTGGAATTGGTCCCCTCA.

Cell migration assay
Mouse vascular smooth muscle cells (MOVAS) (ATCC, Cat. CRL-2797) were cultured within the Transwells. Cells were treated with PDGF-BB or serum in the presence of vehicle or amlexanox for 24hrs.
MOVAS that migrate to the lower side of Transwells were stained with 0.1% Crystal Violet and visualized by brightfield microscropy. Crystal Violet was eluted by 10% acetic acid. Absorbance was measured at 590nm for quantification.

RNA-Seq Analysis
RNA-seq analysis was conducted as previously described (59,60 Metascape was used (64). The accession number for the transcriptomic data reported in this paper is GEO: GSE209621.

Statistics
All data in animal studies are shown as mean ± SEM., while data from in vitro studies are shown as mean ± SD. Replicates are indicated in figure legends. N represents the number of experimental replicates. Ftest was performed to determine the equality of variance. When comparing two groups, statistical analysis was performed using a two-tailed Student's t-test, except when the f-test suggested that variances are statistically different. For analysis of more than two groups, we used analysis of variance (ANOVA) to determine equality of variance. Comparisons between groups were performed with Tukey-Krammer posthoc analysis. For all tests, P<0.05 was considered statistically significant (56).

Study approval
The animal studies were approved by the Institutional Animal Care and Use Committee of (IACUC) of UCSD and UTHSCSA.