Amlexanox

Amlexanox is a small-molecule drug, 2-amino-7-isopropyl-5-oxo-5H-[1]benzopyrano[2,3-b]pyridine-3-carboxylic acid, first reported in 1985 as the antiallergic compound AA-673 (Nohara et al., 1985). A 2020 review reports that Japan has authorized it for asthma since 1987, and for allergic rhinitis and conjunctivitis (Dosanjh and Won, 2020).

Since 2013, labs have studied it as an inhibitor of the protein kinases TBK1 and IKK-epsilon (IKKε) (Reilly et al., 2013).

PubChem CID 2161 lists the same formula, weight and CAS number and supplies the aliases and structure name.

Published work on amlexanox falls into six areas, each summarized below with its model. Outside the human trials, the evidence comes from purified proteins, cells and rodents.

S100 protein binding

Used as affinity ligands on bovine lung extract, amlexanox, cromolyn and tranilast bound two calcium-binding proteins, S100A12 and S100A13, but scarcely calmodulin, and the recombinant proteins bound immobilized amlexanox; the authors speculated that this binding blocks mast cell degranulation (Shishibori et al., 1999). S100A13 also forms part of the complex that releases fibroblast growth factor 1 (FGF1) from stressed cells. NMR and calorimetry placed amlexanox on the FGF1-binding site of S100A13, where it prevents that complex from forming (Rani et al., 2010). Models: bovine lung extract and recombinant proteins.

TBK1 and IKK-epsilon inhibition

A screen of 150,000 compounds against purified IKKε picked out amlexanox, which inhibited IKKε with a half-maximal inhibitory concentration (IC50) of about 1 to 2 µM and TBK1 at similar concentrations. It competed with ATP and spared IKKα, IKKβ and a broad kinase panel (Reilly et al., 2013).

TBK1 crystal structures showed that its carboxylic acid and the nearby Thr156 residue are needed for full potency; a tetrazole analog reached 200 nM against IKKε and 400 nM against TBK1 but did no better than amlexanox in adipocytes (Beyett et al., 2018). Amlexanox also inhibits GRK5 at micromolar concentrations (Homan et al., 2014). Models: purified kinases, crystal structures and cultured cells.

Adipose tissue and liver signaling in obese mice

In C57BL/6 mice on a 45 percent fat diet, daily gavage with 25 or 100 mg/kg amlexanox held body weight at control-diet levels without changing food intake, and in obese mice it reversibly lowered body weight and adipose tissue mass. Oxygen consumption and brown-fat UCP1 rose, insulin sensitivity increased in insulin tolerance tests and clamps, and hepatic steatosis reversed; ob/ob mice responded similarly and normal-diet mice did not (Reilly et al., 2013).

In 3T3-L1 adipocytes, IKKε and TBK1 activated phosphodiesterase PDE3B, and amlexanox restored the cAMP signaling that TNF-α had blunted (Mowers et al., 2013). Acutely, amlexanox raised cAMP in subcutaneous fat, releasing IL-6 that activated hepatic STAT3 and suppressed gluconeogenic genes; an IL-6 neutralizing antibody blocked this (Reilly et al., 2015). In knockout mice, the rise in energy expenditure required UCP1 and adipocyte FGF21 (Reilly et al., 2021). Models: mice, including knockout strains, and 3T3-L1 adipocytes; animal and cell data only.

The type 2 diabetes trial

The same group then studied people with type 2 diabetes, obesity and nonalcoholic fatty liver disease. A 12-week open-label pilot in 6 people, using tablets made by Takeda, had tolerability, with attention to hypoglycemia, as its primary endpoint; HbA1c fell in 4. A randomized, double-blind, placebo-controlled trial then enrolled 42 people, and 38 completed 12 weeks. It met its primary endpoint: HbA1c fell close to 0.5 percentage points more with amlexanox than with placebo, and fructosamine also fell.

Responders, with an HbA1c fall of at least 0.5 percentage points, showed greater insulin sensitivity, a transient rise in serum IL-6 and a distinct inflammatory gene signature in fat biopsied at baseline; liver fat fell only in responders, in an MRI subset too small for a firm conclusion. Rash was the main adverse finding, in 2 of the 6 pilot participants and in 7 reports from the randomized trial. Fasting glucose rose again after the drug stopped (Oral et al., 2017). Model: human trials.

Other cell and animal work

A screen of 1,200 marketed drugs for inhibitors of nonsense-mediated mRNA decay picked out amlexanox. In Calu-6 lung cancer cells, Duchenne muscular dystrophy myoblasts and 6CFSMEo- cystic fibrosis cells, it raised nonsense-containing mRNAs and yielded full-length p53, dystrophin and CFTR, each with a functional readout (Gonzalez-Hilarion et al., 2012).

In mouse bone marrow macrophages it blocked RANKL-induced osteoclast formation, and at 20 mg/kg it prevented bone loss in ovariectomized mice (Zhang et al., 2015). In mice sensitized through the nose with house dust mite plus cGAMP, airway eosinophilia needed IL-33, STING and TBK1, and amlexanox reduced the allergic inflammation (Ozasa et al., 2019). These are cell and mouse data only. Models: human cell lines, mouse cells and mice.