Fibrosis is a pathological condition characterized by excessive deposition of extracellular matrix (ECM) proteins, leading to the disruption of normal tissue architecture and function [1]. Among various factors contributing to fibrosis, lysophosphatidic acid (LPA) has emerged as a potent pro-fibrotic lipid mediator and has been extensively studied [2]. Research has demonstrated that LPA directly promotes fibrosis by stimulating the activation, proliferation, and differentiation of fibroblasts [3], the major cell type responsible for ECM production in fibrotic tissues [4,5]. Specifically, LPA activates fibroblasts through various signaling pathways, including RhoA/ROCK [6], ERK1/2 [7], and PI3K/Akt pathways [8], and induces the expression of pro-fibrotic genes such as collagen I and α-smooth muscle actin (α-SMA) [9].

Lysophosphatidylcholine (LPC), a phospholipid secreted by hepatocytes into the plasma [10] in complex with albumin, serves as a precursor for LPA biosynthesis [11]. Autotaxin (ATX), an extracellular zinc metalloenzyme, hydrolyzes the choline headgroup of LPC to generate LPA [12]. Interestingly, LPC levels are elevated in the serum and tissues of patients with fibrosis [13,14], and ATX expression is upregulated in some fibrotic tissues [15], contributing to the overproduction of extracellular LPA. Several studies have suggested that ATX plays a pivotal role in the pathogenesis of fibrosis [16], making the reduction of cellular LPA levels through ATX inhibition a promising therapeutic strategy. While numerous ATX inhibitors [17] have been developed by both academic [18,19] and industrial [[20], [21], [22]] researchers, none have received FDA approval to date. The primary barriers include challenges in demonstrating efficacy and addressing safety concerns. Achieving meaningful clinical outcomes, such as improved survival or organ function, remains difficult due to the slow progression of fibrosis and the need for long-term clinical data [23]. Furthermore, the multifactorial nature of fibrosis—encompassing overlapping pathways like inflammation, oxidative stress, and fibroblast activation—makes it unlikely that targeting only the ATX-LPA axis alone will fully address the disease [24]. Additionally, existing animal models, such as chemically induced models of fibrosis, have limitations in replicating human disease progression and therapeutic responses [25,26].

Recent clinical outcomes of ATX inhibitors illustrate the challenges of translating diverse inhibition strategies into safe and effective therapies. For example, the Phase 3 trial of the allosteric, non–zinc-binding inhibitor ziritaxestat (I) was discontinued due to safety concerns [27]. However, this outcome does not represent a class-wide limitation of ATX inhibition. The allosteric ATX inhibitor IOA-289 has received FDA Orphan Drug Designation (NCT05586516), and the active-site–directed ATX inhibitor BBT-877 advanced to clinical evaluation for idiopathic pulmonary fibrosis [28], although its Phase 2 study did not meet the primary efficacy endpoint. Together, these clinical experiences suggest that distinct ATX inhibition strategies entail context-dependent challenges in balancing potency, selectivity, and safety, rather than indicating inherent limitations of the target. Nevertheless, ATX remains a compelling therapeutic target, as the unmet medical need for effective fibrosis therapies persists and the clinical benefit of FDA-approved agents such as nintedanib [29] and pirfenidone [30] remains limited.

Moreover, the therapeutic relevance of the ATX–LPA axis has expanded beyond fibrotic disease, with emerging evidence indicating that ATX inhibition can remodel the tumor microenvironment to promote immune cell infiltration [[31], [32], [33]]. Accordingly, addressing these challenges will require the development of novel ATX inhibitors with low toxicity, robust efficacy, and favorable physicochemical properties to maximize their translational potential in both fibrotic disease and immuno-oncology applications.

ATX possesses a T-junction-shaped drug-binding site located in the catalytic phosphodiesterase (PDE) domain. This site comprises a hydrophobic channel, a hydrophobic pocket, and the metal binding catalytic site (Fig. 1A). The catalytic site contains two zinc ions: one fully coordinated by residues of two aspartates, a histidine, and a threonine, and the other coordinated by two histidines and an aspartate. The latter zinc ion plays a critical role as the essential catalytic metal for LPC hydrolysis [39]. While some strategies, such as those employed with Ziritaxestat (Fig. 1B–I), have targeted the hydrophobic channel and hydrophobic pocket [34,35], most efforts have focused on targeting the zinc catalytic site [20,[36], [37], [38]] (Fig. 1B, II - VIII). Zinc binding is frequently associated with strong catalytic inhibition of ATX activity [40], and many highly potent ATX inhibitors incorporate zinc-chelating pharmacophores such as imidazole, triazole, benzoxazolone, benzimidazole, hydroxyamide, or boronic acid. Consistent with this concept, benzoxazolone-based inhibitors II and III exhibit low-nanomolar potency in an in vitro ATX LPC inhibition assay, with IC50 values of 1.7 nM and 2.5 nM, respectively, while a benzotriazole analogue IV shows an IC50 of 2.5 nM. A triazole-based inhibitor V displays exceptional potency, with IC50 values ≤ 1.7 nM in the same in vitro LPC assay [41]. Notably, these inhibitors also maintained substantial inhibitory activity under more physiologically relevant conditions, as demonstrated in a human plasma ex vivo ATX inhibition assay, with plasma IC50 values of 42 nM for II and III, 29 nM for IV, and ≤2.2 nM for V [41].

Despite their high potency, a potential concern associated with many zinc-binding ATX inhibitors is the presence of free-rotating linkers connecting the zinc-chelating moiety to the lipophilic pocket-binding group. As shown in Supplementary Fig. S1, compounds II–VI contain two or more freely rotatable bonds between the zinc-binding motif and the hydrophobic pocket binder, which may confer substantial conformational flexibility and enable multiple binding conformations. Although an increased number of rotatable bonds has not yet been directly linked to adverse effects or off-target liabilities in ATX inhibitors, we reasoned that excessive conformational flexibility could potentially increase the likelihood of unintended interactions with other metalloproteins, particularly those sharing conserved metal-coordination environments [42]. Accordingly, the development of zinc-binding ATX inhibitors with increased conformational constraint at the substrate-binding site was hypothesized to represent a viable strategy to potentially improve ATX selectivity, while mitigating potential metalloprotein-related off-target interactions.

Previously, we synthesized a series of zinc-binding compounds and demonstrated that a triazole derivative effectively inhibits ATX activity in vitro [43] (Fig. 1F). In a mouse model of lung fibrosis, this compound significantly attenuated fibrotic progression, as evidenced by reduced collagen deposition, decreased expression of LPAR1 and p-ERK1/2 in lung tissue, and suppressed the expression of other key fibrotic markers.

In this study, we utilized ATX 3D structure–based computer-aided drug design (CADD) to identify novel inhibitors. For constructing the linker that connects the hydrophobic pocket–binding region to the zinc-binding group, we analyzed the binding modes of previously reported ATX inhibitors. Although earlier studies commonly utilized either a carbamate linker or a pyrimidin-2-amine scaffold to attach lipophilic tails to the zinc-binding motif, we selected the pyrimidin-2-amine core because it provides a more rigid and stronger linkage. Specifically, pyrimidin-2-amine engages ATX through two hydrogen bonds and a π–π stacking interaction (Fig. 1D), whereas carbamates form only a single hydrogen bond (Fig. 1C). Nevertheless, as illustrated in Fig. 1E, further optimization of the hydrophobic pocket–filling moiety was deemed necessary.

To link the triazole zinc-binding group to the pyrimidin-2-amine scaffold, we conducted docking studies to evaluate various linker architectures and functional group combinations. In our previous studies, compounds employing phenyl-based linkers with 1,4- or 1,3-substitution patterns (series VII), as well as linkers containing two rotatable bonds (series VIII), often resulted in flexible and less well-defined binding poses. In contrast, introduction of a cyclic linker at the same position improved the geometric alignment of the two pharmacophoric moieties by restricting conformational freedom and pre-organizing the ligand for productive binding. Notably, incorporation of an azabicyclo[3.1.0]hexane unit enabled the two connected ring systems to adopt a pseudo-linear yet vertically offset arrangement, which was consistently observed in docking analyses. On the basis of these observations, an oxadiazole unit combined with an azabicyclo[3.1.0]hexane ring was selected as a linker motif to provide a rigid connection between the triazole zinc-binding group and the pyrimidin-2-amine core (Supplementary Fig. S2). This design rationale guided the development of a series of N-arylalkyl–substituted 5-(5-(6-(1H-1,2,3-triazol-4-yl)-3-azabicyclo[3.1.0]hexan-3-yl)-1,3,4-oxadiazol-2-yl)pyrimidin-2-amine derivatives evaluated in this study.