Modular Synthesis of Pyritide‐Inspired Macrocycles Featuring Bipyridine Motifs

Abstract

Macrocycles represent a promising class of drug‐like scaffolds with unique structural features and the ability to engage challenging targets such as protein–protein interactions. Inspired by structural characteristics of pyritides, we constructed a library of 27 diverse macrocycles via a build/couple/pair approach, enabled by efficient synthesis of bipyridine‐based triaryl building blocks through azaindole cleavage. Kinetic and cheminformatic analyses confirmed both reactivity trends and structural diversity. From this library, we identified a potential ferroptosis inhibitor, 6paW, with clear structure–activity relationships, validating our diversity‐oriented synthesis platform. This strategy offers a robust approach to macrocycle library design, expanding opportunities for targeting previously inaccessible biological space.

A pyritide‐inspired build/couple/pair strategy yields a diverse macrocycle library. This approach expands chemical space and enables the discovery of 6paW, a potent anti‐ferroptotic agent.

Introduction

Macrocycles—ring structures containing 12 or more atoms—constitute a distinctive class of biologically active molecules with properties that extend beyond the traditional drug‐likeness described by Lipinski's Rule of Five.^{[ 1} , ² , ³ , ^{4 ]} Their unique physicochemical characteristics allow macrocycles to occupy otherwise inaccessible regions of chemical space. Owing to their conformationally pre‐organized structures, macrocycles can mitigate entropic penalties upon target engagement, while retaining sufficient conformational flexibility to confer high affinity and selectivity toward biological targets. Their large molecular size and extended surface area make them particularly well‐suited for modulating protein–protein interactions (PPIs), which are generally considered undruggable by traditional small molecules due to their comparably large and flat binding surfaces, as well as the absence of defined pockets. In peptide‐based therapeutics, macrocyclization further enhances proteolytic stability and cell permeability, thereby improving the drug‐like potential.^{[ 5 ]}

Macrocyclic scaffolds are abundant in natural products, exemplified by the immunosuppressant rapamycin^{[ 6 ]} and the antibiotic vancomycin^{[ 7 ]} (Figure 1a, above). These natural macrocycles exhibit ring sizes ranging from 12 to over 50 atoms, with inherent stereochemical complexity and structural diversity that have inspired the development of first‐in‐class drugs.^{[ 8} , ⁹ , ¹⁰ , ^{11 ]} Recently, a new subclass of ribosomally synthesized and post‐translationally modified peptides (RiPPs),^{[ 12 ]} termed pyritides (e.g., thiocillin, lactazole), has been identified.^{[ 13} , ¹⁴ , ¹⁵ , ^{16 ]} Their extensive post‐translational modifications (PTMs) confer high structural diversity and broad biological activity, making them attractive scaffolds for targeting PPIs. Indeed, more than 100 macrocycle‐based drugs are currently approved or in clinical development, with 80%–90% originating from natural products.^{[ 2} , ^{17 ]} However, the structural complexity of natural macrocycles often hampers synthetic modifications and pharmacokinetic optimizations, limiting their direct utility in drug discovery.

Figure 1.

Pyritide‐inspired macrocycles featuring bipyridine motifs. a) Representative natural macrocycles, including pyritides, alongside FDA‐approved synthetic macrocycles with diverse bioactivities. b) Design strategy for constructing a library of pyritide‐inspired macrocycles. c) Thiazole‐conjugated bipyridine scaffolds with ortho‐, meta‐, and para‐connectivity, synthesized via azaindole cleavage followed by Negishi coupling.

Synthetic macrocycles have emerged as a powerful alternative (Figure 1a, below). Compared to their natural counterparts, synthetic macrocycles typically have smaller ring sizes and are more amenable to structural modification, which can impact pharmacokinetic properties.^{[ 18} , ^{19 ]} These advantages have expanded accessible chemical space and enabled the exploration of new biological pathways. Clinically approved examples highlight their potential: lorlatinib, a macrocyclic analog of the anticancer drug crizotinib, achieved about an 80‐fold potency increase and enhanced blood–brain barrier (BBB) penetration through macrocyclization,^{[ 20 ]} while the HCV drug simeprevir gained improved biological activity, pharmacokinetics, and safety from cyclization of its initial lead structure.^{[ 21 ]} The therapeutic success of such synthetic macrocycles underscores the urgent need for efficient strategies to generate structurally diverse macrocycle libraries, thereby accelerating the discovery of drugs against previously intractable targets.

We set out to design a novel macrocycle library that combines the strengths of natural and synthetic macrocycles (Figure 1b). Four principles guided our design: 1) high biological relevance, 2) structural diversity, 3) modular synthesis, and 4) high expandability. To achieve these key features—particularly biological relevance—we drew inspiration from pyritides, a class of natural macrocyclic peptides. Thiopeptides, the best‐known pyritides, display antibacterial,^{[ 22 ]} anticancer,^{[ 23} , ^{24 ]} and antifungal activities,^{[ 25 ]} and their characteristic thiazole‐pyridine conjugates served as the foundation for our macrocycle design. We adopted this motif as a core structural element to enhance biological relevance and molecular diversity.

To systematically construct and diversify the library, we applied the build/couple/pair (B/C/P) strategy, which is one of the principal strategies in diversity‐oriented synthesis (DOS). Since its introduction by Schreiber,^{[ 26} , ^{27 ]} DOS has enabled the discovery of numerous bioactive small molecules, and our group has further advanced it through a privileged substructure‐based DOS (pDOS) approach that leverages structural motifs recurrent in bioactive natural products and therapeutic agents.^{[ 28} , ²⁹ , ^{30 ]} This strategy has proven effective in modulating biomacromolecular interactions and identifying molecules with broad biological activities. The modularity of the B/C/P strategy is particularly well‐suited for diverse macrocycle synthesis: building blocks are prepared in the build phase, combinatorially assembled in the couple phase, and cyclized intramolecularly in the pair phase to yield diverse scaffolds.^{[ 31} , ³² , ^{33 ]} Guided by retrosynthetic analysis, we designed three classes of building blocks: 1) thiazole‐conjugated bipyridine structures, 2) chiral cyclic triamines, and 3) alkyne‐functionalized amino acids.

In pyritide biosynthesis, pyridine formation via [4 + 2] cycloaddition between dehydroalanine (Dha)‐like residues restricts macrocycle connectivity to the ortho position.^{[ 13 ]} To overcome this constraint and prove how pyridine connectivity influences macrocycle conformation and consequent biological activity, we designed three distinct thiazole‐conjugated bipyridine structures with ortho‐, meta‐, and para‐connectivity (Figure 1c). These novel triaryl building blocks were expected to modulate backbone rigidity and structural diversity. We have developed an efficient synthetic route to these distinctive meta‐bipyridines: pyridine formation from 3‐formyl azaindole through aldol‐type addition, followed by intramolecular cyclization, C–N bond cleavage, and re‐aromatization.^{[ 34} , ³⁵ , ^{36 ]} This method allows broad substituent incorporation—including halo and boronic ester groups—often difficult to access via conventional cross‐couplings. The bipyridine frameworks derived from this method has already been successfully employed in the development of selective anti‐inflammatory agents, highlighting its bio‐relevance.^{[ 36 ]} Furthermore, the hydrophobic bipyridine unit is expected to play a pivotal role in enhancing membrane permeability, a critical factor for modulating intracellular targets.^{[ 37 ]} Utilizing this strategy, we introduced thiazole at various positions of the halo‐substituted bipyridine core through a Negishi coupling reaction, thereby completing the synthesis of the three key bipyridine‐based triaryl building blocks.

This strategy enabled the construction of a structurally diverse macrocycle library spanning ring sizes of 21–25 atoms, a relatively unexplored range in synthetic macrocycles. The resulting library demonstrated significant potential for biological applications, as evidenced by the identification of a ferroptosis inhibitor. Overall, the B/C/P‐based approach provided modularity and expandability, with the integration of a pyritide‐inspired framework enabling access to previously underexplored chemical space with promising therapeutic potential.^{[ 9 ]}

Results and Discussion

Library Design Strategy

Our initial synthetic approach sought to incorporate a thiazoline heterocycle, a structural motif found in many pyritides (Figure 2a). In natural biosynthesis, thiazoline is formed from cysteine by cyclodehydratase and can undergo subsequent oxidation to thiazole through dehydrogenase‐mediated PTM.^{[ 13 ]} To introduce this functionality into our bipyridine scaffold, we employed a nitrile‐aminothiol (NAT) click reaction between a nitrile‐substituted bipyridine and cysteine (S5 to S6).^{[ 38} , ³⁹ , ⁴⁰ , ^{41 ]} We anticipated this approach would generate two chiral thiazoline isomers, depending on cysteine stereochemistry, and potentially yield thiazoles upon further oxidation, thereby enhancing structural diversity. However, the poor reactivity of substrate S5 in the NAT click reaction necessitated harsh conditions, resulting in low functional group tolerance, inseparable byproducts, and frequent uncontrolled thiazoline oxidation during the amide coupling step (S6 to S7),^{[ 42 ]} ultimately leading to low yields (Figure S1). These limitations prompted a strategic redesign.

Figure 2.

Modular synthesis of pyritide‐inspired macrocycles via the B/C/P strategy. a) Initial synthetic route for incorporating thiazoline into the bipyridine scaffold. b) Optimized route for constructing the pyritide‐inspired macrocycle library. c) The build phase of the B/C/P strategy, illustrating the synthesis of three key building blocks: thiazole‐conjugated bipyridine structures, chiral cyclic triamines, and alkyne‐functionalized amino acids. Reagents and conditions: i) Ethyl propiolate, ammonium acetate, zinc triflate, EtOH, 120 °C; ii) (4‐(tert‐butoxycarbonyl)thiazol‐2‐yl)zinc(II) bromide solution (∼0.21 M in dry DMA), Pd₂(dba)₃, Xphos, dry DMA, 80 °C; iii) Propargyl amine, PyBOP, DIPEA, DMF, r.t. Abbreviations: zinc triflate = zinc trifluoromethanesulfonate; DMA = dimethylacetamide; Pd₂(Dba)₃ = tris(dibenzylideneacetone)dipalladium(0); XPhos = dicyclohexyl[2′,4′,6′‐tris(propan‐2‐yl)[1,1′‐biphenyl]‐2‐yl]phosphane; PyBOP = (benzotriazol‐1‐yloxy)tripyrrolidinophosphonium hexafluorophosphate; DIPEA = diisopropylethylamine; DMF = dimethylformamide.

To achieve a more modular and reliable approach, we incorporated a thiazole‐functionalized bipyridine as the core structural motif for subsequent diversification (Figure 2b). Thiazole offers greater chemical stability and rigidity compared to thiazoline, while maintaining biological relevance, making it an ideal scaffold for macrocycle construction.^{[ 43 ]}

Our revised synthetic strategy was built upon three classes of modular building blocks: 1) thiazole‐conjugated bipyridine structures with ortho‐, meta‐, and para‐connectivity to modulate backbone rigidity and molecular diversity; 2) chiral cyclic triamines to enhance structural complexity and interaction specificity; and 3) alkyne‐functionalized amino acids to enable further functional diversification (Figure 2c). Halo‐substituted bipyridine precursors (7o , 7m , and 7p ) were synthesized on gram scale via azaindole cleavage, then coupled with thiazoles using Negishi coupling to yield bipyridine‐based triaryl building blocks (1o , 1m , and 1p ), each bearing two orthogonally protected carboxylic acid groups for downstream functionalization.

To introduce skeletal diversity and improve drug‐likeness, we prepared chiral cyclic triamines (2a–2c), derived from a common intermediate (8). These six‐membered cyclic structures, reminiscent of pipecolic acid found in natural macrocycles such as rapamycin,^{[ 44 ]} were functionalized with amine and azide groups in varying orientations, enabling broad diversification during the couple and pair phases (Figures S2 and S3). Orthogonal protection of amines allowed selective transformation, and each chiral center was generated in an enantiomerically controlled manner, maximizing the scope of skeletal diversification.

To further enhance functional diversity, we introduced alkyne‐functionalized amino acids (3F, 3L, and 3W), prepared via amide coupling of propargyl amine with phenylalanine (F), leucine (L), and tryptophan (W)—amino acids known to play critical roles in biomolecular recognition and highly enriched at PPI hotspots, where they substantially contribute to binding energy.^{[ 45} , ^{46 ]}

Macrocycle Synthesis and Optimization

Final macrocycle synthesis proceeded in three stages (Figure 3). Bipyridine‐based triaryl building blocks (1o , 1m , and 1p ) were conjugated with chiral cyclic triamines (2a–2c) through tert‐butyl ester deprotection and sequential amide coupling, yielding nine key intermediates (4oa–4pc). These intermediates were then coupled with the alkyne‐functionalized amino acids (3F, 3L, and 3W) via saponification and amide coupling, affording 27 linear precursors. Final macrocyclization was achieved using copper‐catalyzed azide–alkyne cycloaddition (CuAAC), a robust click reaction widely regarded as a powerful method for macrocyclization. Beyond its efficiency, CuAAC incorporates a peptide‐bond‐mimicking 1,4‐disubstituted 1,2,3‐triazole, thereby reinforcing both structural rigidity and pharmacophoric relevance.^{[ 31} , ⁴⁷ , ^{48 ]} Owing to this dual advantage—efficient macrocyclization and amide‐analogous triazole formation—CuAAC was particularly well suited to our strategy, ultimately furnishing a 27‐member macrocycle library with substantial structural and functional diversity.

Figure 3.

The couple and pair phase of the B/C/P strategy. Reagents and conditions: i) 4 M HCl in dioxane, DCM, r.t.; ii) Chiral cyclic triamines, HATU, DIPEA, DMF, r.t.; iii) LiOH(aq), THF, MeOH, H₂O, r.t.; iv) Alkyne‐functionalized amino acids, HATU, DIPEA, DCM, DMF, r.t.; v) CuBr, sodium ascorbate, DIPEA, 2,6‐lutidine, DMF (1.0 mM), r.t. Abbreviations: DCM = dichloromethane; HATU = hexafluorophosphate azabenzotriazole tetramethyl uronium; DMF = dimethylformamide; DIPEA = diisopropylethylamine; CuAAC = copper‐catalyzed azide‐alkyne cycloaddition.

The overall synthesis proceeded in moderate to good yields (Figure 4). However, macrocycles incorporating a para‐oriented triaryl building block 1p (6paF–6pbW) displayed significant dimer formation, resulting in reduced yields. Lowering the reaction concentration by threefold effectively suppressed dimerization and improved product formation, thereby confirming the critical role of dilution in macrocyclization. A similar challenge was observed in macrocycles derived from the para‐positioned chiral triamine 2c and the ortho‐oriented triaryl building block 1o (6ocF–6ocW). These species, constrained to form a 22‐membered ring—the smallest in the series—likely suffered from ring strain, which promoted dimerization under standard conditions. LC‐MS analysis of 6mcF revealed incomplete macrocyclization, with a stalling point at 84% conversion, yielding only 47% of the product. In contrast, complete conversion was achieved for leucine (L) and tryptophan (W) analogs under the same conditions (6mcF–6mcW), suggesting that the amino acid component also influences macrocyclization efficiency in the meta‐oriented triaryl/chiral triamine (1m /2c) series.

Figure 4.

Library of 27 pyritide‐inspired macrocycles featuring bipyridne motifs. Reagents and conditions for final click reaction: CuBr (1.0 equiv.), sodium ascorbate (3.0 equiv.), DIPEA (10 equiv.), 2,6‐lutidine (10 equiv.), DMF (1.0 mM), r.t. 1–24 h. ^a)Starting material concentration: 0.33 mM; ^b)Isolated recovery yield.

Among all combinations, macrocycles incorporating both the para‐oriented triaryl building block 1p and the chiral cyclic triamine 2c (6pcF–6pcW) predominantly underwent dimerization, with the desired macrocycles detected only in trace amounts or not at all. This outcome is likely due to the combined effect of two para‐oriented building blocks, which induce substantial ring strain and strongly disfavor productive macrocyclization. Attempts to optimize reaction conditions, including changes in solvents, catalysts, and reaction temperatures, failed to improve yields (data not shown).

Kinetic Analysis of Structural Variations

To investigate how variations in bipyridine‐based triaryl (1o , 1m , and 1p ) and chiral cyclic triamine (2a–2c) building blocks influenced reactivity, we performed kinetic analyses of the CuAAC macrocyclization (Figure 5a). Using representative examples 6oaW, 6maW, and 6paW, we found that ortho‐ (1o , k = 0.0003 s⁻¹) and meta‐oriented (1m , k = 0.0002 s⁻¹) triaryl building blocks showed comparable reaction rates, with the ortho configuration slightly faster. In contrast, the para‐oriented triaryl building block (1p ) showed a much slower macrocyclization rate (k = 2.87 × 10⁻⁵ s⁻¹) and a pronounced tendency toward dimerization (k = 0.465 L·mol⁻¹·s⁻¹), consistent with the low yields and significant dimer formation observed in 6paF–6pbW. These findings align with the probability of forming a favorable head‐to‐tail transition state, which is highest for ortho‐, intermediate for meta‐, and lowest for para‐configuration.

Figure 5.

Reactivity differences arising from building block variations. a) Kinetic analysis of macrocyclization using different bipyridine building blocks in 6oaW, 6maW, and 6paW ( o : red, m : blue, p : green; starting material concentration: 0.33 mM). Reactivity trends for chiral cyclic triamines were also evaluated in 6maW, 6mbW, and 6mcW (a: blue, b: dark red, c: dark green, right; starting material concentration: 1 mM). Data points indicate starting material (SM, square), product (P, circle), and multimeric species (dimer & trimer, double circle). b) A one‐carbon elongation strategy enabling successful macrocyclization of previously inaccessbile combinations (6pcF, 6pcL, and 6pcW). Starting material concentration: 0.33 mM.

We next examined the effect of cyclic triamine connectivity (2a–2c) using a representative set of 6maW–6mcW. The reaction rate followed the order 2a (k = 0.0013 s⁻¹) > 2b (k = 0.0007 s⁻¹) > 2c (k = 0.0004 s⁻¹), indicating that macrocyclization slows as connectivity shifts from ortho to para around the piperidine ring. The slowest reacting 2c building block also exhibited minor dimer formation, further supporting the correlation between reduced macrocyclization kinetics and increased backbone rigidity imposed by the para‐connectivity.

Structural Insights and Optimization Strategy

Our results indicate that slow macrocyclization kinetics caused by specific building blocks promote dimer formation. The failed reaction of 6pcF, 6pcL, and 6pcW can be attributed to extremely slow intramolecular cyclization, arising from both the low head‐to‐tail probability of the substrates and ring strain in the designed macrocycles. To further support this, we performed energy‐minimized structural calculations for 6pcF, which failed to cyclize efficiently. The triaryl segment, in which the bipyridine and thiazole are para‐connected, adopted a bent conformation with a dihedral angle of 154° (Figure S4). This distortion was alleviated upon elongation of the alkyne carbon chain, which extended the dihedral angle to 180° and relieved ring strain. Based on this insight, we implemented a one‐carbon elongation strategy (Figure 5b), which enabled macrocyclization of the previously unsuccessful combinations (6pcF ′, 6pcL ′, and 6pcW ′).

Cheminformatic Analyses

To assess structural diversity, we analyzed the 27 synthesized macrocycles using Tanimoto similarity, 3D structural overlays, principal moment of inertia (PMI) analysis, and principal component analysis (PCA) (Figure 6). As the flexibility of each macrocycle varies intentionally according to its building block combination, 3D structural overlays and PMI analysis utilized MM2‐level energy‐minimized structures for all compounds to ensure consistency and a fair comparison.

Figure 6.

Cheminformatic analyses of the designed macrocycle library. a) Violin plots of Tanimoto similarity scores based on ECFP4 and ECFP6 fingerprints. Upper and lower lines represent the quartiles, while the center line indicates the median. b) Structural overlay of energy‐minimized macrocycles: 6ocL, 6mcL, and 6pcL (left), and 6maF, 6mbF, and 6mcF (right), aligned by their common 2‐aminopyridine substructure. Comparisons highlight differences arising from bipyridine connectivity ( o : red, m : blue, p : green, left) and chiral cyclic triamine scaffolds (a: yellow, b: light green, c: purple, right). c) Principle moments of inertia (PMI) analysis. Left: PMI distribution of the 27 synthesized macrocycles (red dots), 24 natural macrocycles (blue diamonds), and 20 pyritides (mint squares). Middle: Clustering by bipyridine connectivity ( o : red, m : blue, p : green). Right: Clustering by chiral cyclic triamines (a: pink, b: yellow, c: purple). d) Principle component analysis (PCA) of eight drug‐like properties condensed into three principal components (PC‐1, PC‐2, and PC‐3). The 27 synthesized macrocycles (red) are compared with 2124 FDA‐, EMA‐, and Health Canada‐approved drugs from DrugBank database (grey), shown in both 3D and 2D projections. Properties analyzed include AlogP, molecular weight, hydrogen bonding donors/acceptors, rotatable bonds, total rings, aromatic rings, and fractional polar surface area.

Tanimoto Similarity: Pairwise comparisons using ECFP4 and ECFP6 fingerprints^{[ 49} , ^{50 ]} revealed relatively low similarity scores despite the isomeric nature of the library (median ECFP4: 0.54; median ECFP6: 0.44), highlighting significant diversity even among closely related scaffolds (Figure 6a; Tables S1 and S2).

3D Structural Overlays: Energy‐minimized overlays aligned on a 2‐aminopyridine substructure showed substantial spatial variation depending on bipyridine connectivity ( o , m , p , left) and chiral cyclic triamine type (a, b, c, right), underscoring how building block combinations strongly influence the three‐dimensional shape of macrocycles (Figure 6b).

PMI Analysis: The PMI plot^{[ 51 ]} revealed that our synthetic macrocycles span a broad range of rod‐like, disc‐like, or sphere‐like geometries compared to 24 natural macrocycles and 20 pyritides (Figure 6c, left). Macrocycles clustered when grouped by bipyridine or triamine unit, but not by amino acid (Figure 6c, middle/right; Figure S5), suggesting the former drives most conformational variation.

PCA Drug‐likeness Evaluation: PCA of key physicochemical parameters—including molecular weight, AlogP, the number of hydrogen bonding donors and acceptors, and polar surface area (PSA)—placed our macrocycles primarily within the same chemical space as 2124 FDA‐approved drugs (DrugBank dataset) (Figure 6d).^{[ 52 ]} This confirms that, in addition to structural diversity, the library exhibits favorable drug‐like properties, supporting its potential for pharmaceutical applications.

Discovery of a Ferroptosis Inhibitor

To evaluate the biological potential of our macrocycle library, we performed phenotypic screening assays. In a cell viability assay using RAS‐selective lethal 3 (RSL3),^{[ 53 ]} a potent ferroptosis inducer, we identified 6paW as a potential ferroptosis inhibitor (Figure 7). Ferroptosis is a regulated form of necrotic cell death driven by iron‐dependent lipid peroxidation (Figure 7a).^{[ 54 ]} Although the concept was only defined in 2012,^{[ 55 ]} ferroptosis has since emerged as a critical process implicated in neurodegenerative diseases (NDDs),^{[ 56 ]} ischemia/reperfusion injury,^{[ 57 ]} and type 1 diabetes.^{[ 58 ]} While ferroptosis inducers are being mainly investigated as anti‐cancer agents, inhibitors are actively pursued as potential therapeutics for refractory diseases.^{[ 59 ]}

Figure 7.

Discovery of ferroptosis inhibitor 6paW. a) Schematic representation of ferroptosis and its associated diseases. b) Screening results from the EZ‐Cytox cell viability assay against the macrocycle library. RAW 264.7 murine macrophage‐like cells were co‐treated with library compounds (10 µM) and RSL3 (1 µM) for 24 h. Control (DMSO) treatment resulted in 18% viability. Data are mean ± SD (n = 3). c)–f) Dose dependent ferroptosis inhibition assays. RAW 264.7 cells were co‐treated with RSL3 (1 µM) and compounds (0.3–20 µM, log scale) for 24 h. Relative viabilities were normalized to DMSO. Data are mean ± SD (n = 3). c) 5paW(○) versus 6paW(●). d) 6oaW(□), 6maW(△), and 6paW(●). e) 6paW(●), 6pbW(▽), and 6pcW′(◇). f) 6paF(★), 6paL(✳), and 6paW(●). g) Representative fluorescence images of nuclei (blue, Hoechst 33342) and lipid peroxide (red, BODIPY 665/676). HeLa cells were treated with DMSO, 6paW (40 µM), RSL3 (1 µM), or both for 3 h, followed by staining with Hoechst 33342 (2 µg/ml) and BODIPY 665/676 (5 µM) for 0.5 h. Scale bar: 25 µm. h) Quantification of BODIPY 665/676 fluorescence intensity from (g), expressed relative to DMSO‐treated cells. Data are mean ± SD (n = 21). i) Mean fluorescence intensity (MFI) from flow cytometry analysis of RAW 264.7 cells treated with DMSO (red), RSL3 (1 µM, blue), or co‐treatment of 6paW and RSL3 (purple) for 4 h, followed by staining with BODIPY 665/676 for 0.5 h. Data are mean ± SD (n = 3). j) Representative flow cytometry histograms corresponding to (i). (*: P < 0.01, ****: P < 0.0001; one‐way ANOVA test).

In the screening assay for potential ferroptosis inhibitors, RAW 264.7 murine macrophage‐like cells treated with RSL3 (1 µM) showed a sharp decrease in viability (18%). Co‐treatment with library compound (10 µM) revealed that 6paW restored viability to 61% (Figure 7b). Importantly, none of the macrocycles exhibited significant cytotoxicity across multiple cell lines, including RAW 264.7 (Figure S6).

Structure‐Activity Relationships (SARs)

Dose‐response assays (0.3–20 µM with 1 µM RSL3) revealed several key SAR trends (Figure 7c–f).

Macrocyclization: Linear precursor 5paW showed limited inhibition of ferroptosis, while macrocyclic 6paW was active, demonstrating that macrocyclization introduces distinct biological properties (Figure 7c).

Triaryl Connectivity: Regioisomers 6oaW, 6maW, and 6paW exhibited markedly different anti‐ferropototic efficacies, despite having identical atom compositions. More extended (para‐oriented) connectivity correlated with more potent inhibition (Figure 7d), underscoring the importance of our triaryl design strategy.

Cyclic Triamine: Activity varied with triamine connectivity (Figure 7e). 6paW (the shortest linker, 23‐membered ring) showed the highest potency. Between 6pbW and 6pcW′ (both 25‐membered), 6pcW′ was slightly more effective, possibly due to differences in connectivity‐induced skeletal rigidity (transannular versus peripheral).

Amino Acid Side Chain: The indolyl side chain in 6paW conferred the most vigorous activity, whereas the benzyl (6paF) and the isobutyl (6paL) analogs showed progressively weaker effects (Figure 7f). This highlights amino acids as valuable diversification handles for further optimization.

To confirm that the protective effect of 6paW derived from ferroptosis inhibition, we monitored lipid peroxidation using BODIPY 665/676, a fluorescent probe sensitive to lipid peroxides.^{[ 60} , ⁶¹ , ^{62 ]} Imaging and flow cytometry showed that RSL3 markedly increased lipid peroxide levels, while co‐treatment with 6paW significantly reduced them (Figure 7g–j). 6paW alone did not affect basal lipid peroxides. Flow cytometry histograms further confirmed this effect, with the RSL3‐induced peak shift reversed upon 6paW co‐treatment (Figure 7i–j). Comparative analysis also confirmed differential efficacy among regioisomers (6oaW, 6maW, and 6paW; Figure S7). Together, these results demonstrate that 6paW rescues RSL3‐treated cells by suppressing lipid peroxidation and ferroptosis. Compared to ferrostatin‐1, a representative ferroptosis inhibitor (EC₅₀ = 40 ± 4 nM in RAW 264.7 cells^{[ 63 ]}), the efficacy of 6paW was only moderate, which limited the feasibility of conducting more detailed mode‐of‐action studies. But their clear SAR trends highlight the importance of macrocyclization, triaryl connectivity, triamine topology, and amino acid side chain composition in shaping biological activity. Leveraging the high expandability of our synthetic strategy, ongoing studies aim to discover compounds with enhanced activity and to extend its application to the detection of diverse new bioactivities.

Conclusion

In this study, we established a structurally diverse and biologically relevant macrocyclic library inspired by pyritides, incorporating a novel thiazole‐conjugated bipyridine scaffold. Using a modular B/C/P strategy, we achieved the streamlined synthesis of 27 distinct macrocycles. A central innovation was the azaindole cleavage reaction, which provided efficient access to diverse thiazole‐conjugated bipyridine building blocks, enabling structural diversification while maintaining synthetic feasibility.

Integration of chiral cyclic triamines and alkyne‐functionalized amino acids further expanded the library's chemical diversity. Notably, our azaindole cleavage methodology uniquely enables meta‐selective functionalization of the bipyridine core with biologically relevant linkages (amide, phosphonamide, sulfonamide). At the same time, the 2‐aminopyridine moiety provides a versatile handle for heterobifunctional designs, allowing various applications, including PROTAC development. The chiral triamine scaffold permits stereochemical inversion to access all four diastereomers, while selective amine deprotection supports late‐stage functionalization. In parallel, the alkyne‐functionalized amino acid module—compatible with both natural and unnatural residues, including fluorescent analogs—offers a broadly adaptable diversification point. Moreover, the CuAAC employed for macrocyclization can be readily extended to ruthenium‐catalyzed azide–alkyne cycloaddition (RuAAC) simply by changing the catalyst, thereby contributing to structural diversity. In addition, by modifying only the terminal functional groups of the building blocks, the strategy can be further expanded to various macrocyclizations, highlighting the high expandability of our approach.

Kinetic analyses revealed distinct reactivity profiles among the building blocks, which correlated with the late‐stage macrocyclization efficiency. Cheminformatic studies confirmed the structural diversity and favorable drug‐like properties of this macrocycle library. Despite similar compositions, the rigid thiazole‐conjugated bipyridine scaffolds enforced discrete three‐dimensional conformations, ranging from compactly folded to extended architectures.

Biological validation identified 6paW as a potential ferroptosis inhibitor. A detailed SAR analysis highlighted the critical importance of macrocyclization and demonstrated how variations in triaryl connectivity, cyclic triamine topology, and amino acid side chains strongly influence biological activity. Ongoing efforts are focused on further expanding the library to identify compounds with superior activity. Additionally, the permeability, a critical drug property, will be evaluated using methods such as the PAMPA (parallel artificial membrane permeability assay)^{[ 64 ]} or the Caco‐2 permeability assay,^{[ 65 ]} to further investigate the working mechanism of our macrocyclic ferroptosis modulators.

Altogether, this DOS‐based pyritide‐inspired macrocycle platform represents a powerful and expandable approach for generating structurally diverse and tunable macrocyclic libraries. The current library not only provides a valuable resource for screening against a wide range of challenging biomolecular targets, including protein–protein interactions, but also opens new opportunities for the development of next‐generation therapeutics for previously intractable diseases.

Supporting Information

The authors have cited additional references within the Supporting Information.^{[ 66} , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ^{72 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Lee Ji H., Yi S., Bang J., Choi H., Park S. B., Angew. Chem. Int. Ed. 2025, 64, e202518622. 10.1002/anie.202518622

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.