Hypoxia-inducible factors (HIFs) are members of the basic helix loop-helix PER-ARNT-SIM (bHLH-PAS) family of transcription factors, which require heterodimerization of an oxygen-dependent HIF-α subunit (HIF-1α, HIF-2α, or HIF-3α) and an aryl hydrocarbon receptor nuclear translocator (ARNT) subunit. Both HIF-α and ARNT subunits contain an N-terminal binding domain (bHLH) and two tandem PAS domains (PAS-A and PAS-B). The transcriptional activities of HIFs are affected by the allosteric transmission between their bHLH and PAS domains [1,2]. Small-molecule ligands have been found to regulate the transcriptional activities or protein fold stability of nuclear receptors (NRs) [3]. For example, analysis of co-crystal structures on nuclear receptor HNF4α homodimer showed that myristic acid stabilize the folded conformation of HNF4α through binding its ligand-binding domain (LBD), and physically interconnect with distal domains [4]. These results indicate that understanding the allosteric mechanism involved in HIFs transcriptional functions has significant importance for the discovery of potential drug candidates.

Like other members of the bHLH-PAS family, the PAS domains of HIFs are not only involved in heterodimer formation, but also harbor hydrophobic pockets [5,6]. Crystal structures of the HIF-α-ARNT heterodimers have revealed that each HIF-2α and HIF-3α PAS-B domain contains a pocket of approximately 300 Å3 in size, with the potential to accommodate small-molecule compounds [[7], [8], [9]]. While the relatively small PAS-B pocket (<200 Å3) in HIF-1α makes it difficult to identify HIF-1α binding ligands [10]. The endogenous oleoylethanolamide (OEA) was identified as a highly potent HIF-3α allosteric agonist through binding the PAS-B domain of HIF-3α, and it enlarge the pocket about 150 % [11,12]. HIF-2α PAS-B domain was widely studied in the past few years, and a variety of compounds, including THS-044, OX3 and PT-2385 were revealed as HIF-2α inhibitors [8,13,14]. In 2019, two benzisothiazole derivatives M1001 and M1002 were identified as HIF-2α agonists by Rastinejad group via affinity selective mass spectrometry screening [15]. Further interaction mechanism study has revealed these HIF-2α modulators bind to the same HIF-2α PAS-B pocket and bidirectionally regulate the stability of HIF-2α-ARNT heterodimerization in an allosteric manner [15].

HIF-2α has been considered as an attractive target to treat von Hippel-Lindau (VHL)-deficiency tumor and renal anemia. It acts as a key oncogenic driver in VHL-deficiency tumors, especially clear cell renal cell carcinoma (ccRCC). In 2021, the HIF-2α inhibitor Belzutifan (Fig. 1, 1) was approved to treat renal cell carcinoma, neuroblastoma, and pancreatic neuroendocrine tumors associated with VHL syndrome [16]. However, the clinical efficacy revealed such HIF-2α inhibitors have the potential risk in altering ventilation through direct actions on HIF-2α within the carotid body [17].

Additionally, HIF-2α is indispensable for the renal and hepatic expression of erythropoietin (EPO) [18], which is closely related to renal anemia caused by chronic kidney disease. Current treatment for renal anemia predominantly relies on EPO replacement [19] and prolyl hydroxylase domain (PHD) inhibitors such as Roxadustat (Fig. 1, 2) and Vadadustat (Fig. 1, 3) [20,21]. However, the former may be associated with an increased risk of cardiovascular events, and the latter may interfere with various physiological processes, including angiogenesis and iron metabolism, due to their lack of selectivity for the HIF-2 isoform [22,23]. Therefore, the development of selective HIF-2α agonists could have potential advantages over PHD inhibitors. Since the pioneering HIF-2α agonists M1001 (Fig. 1, 4) and M1002 (Fig. 1, 5) were disclosed, a series of benzisothiazole derivatives or analogs were developed with more potent HIF-2α agonistic activity and optimized drug-likeness [[24], [25], [26]]. Especially, compound ZG-2033 (Fig. 1, 6) demonstrated oral HIF-2α agonistic activity [25]. While the scaffold of HIF-2α agonists was limited, therefore, the identification of novel skeleton HIF-2α agonists would be a suitable approach to the treatment of renal anemia.

In our previous study, several 3,5-diaryl-1,2,4-oxadiazole derivatives, including 3-aryl-5-indazol-1,2,4-oxadiazole derivative 7, 3-aryl-5-pyrazol-oxadiazole derivative 8, and 3-aryl-5-triazolyl-oxadiazole 9, were designed as HIF-1α inhibitors (Fig. 2) [[27], [28], [29]]. More recently, we virtually screened a library of 1.5 million molecules and identified novel scaffold HIF-1α inhibitors based on ComABAN and molecular docking [30]. Compound 10 exhibited moderate HIF-1 inhibitory activity, while compound 11 exhibited HIF-1 agonism activity (Fig. 2).

Inspired by the successful discovery of HIF-1α modulators, a docking-based virtual screening was also conducted using crystal structures of HIF-2α inhibitors to approach novel scaffold HIF-2α modulators. Interestingly, a 3-phenyl-5-methyl-isoxazole-4-carboxamide derivative v19 was unexpectedly identified as a hit compound with HIF-2α agonistic effect, followed by further structure-based modifications on v19 to achieve a series of 3-aryl-5-methyl-isoxazole-4-carboxamide derivatives as novel scaffold HIF-2α agonists.