Identification of a bacteria-produced benzisoxazole with antibiotic activity against multi-drug resistant *Acinetobacter baumannii*

The study addresses the urgent need for new antibacterial agents against multi-drug resistant Gram-negative pathogens, particularly Acinetobacter baumannii, an ESKAPE organism with high mortality in ICU infections and increasing resistance to carbapenems and colistin. Given the scarcity of new antibiotics, the researchers explored natural products and identified a benzisoxazole scaffold, a class known in pharmacology and with several FDA-approved derivatives. The research aims to discover and characterize a bacteria-produced 1,2-benzisoxazole active against A. baumannii, define its structure–activity relationships, and investigate its mechanism of action, hypothesizing interference with 4-hydroxybenzoate-related metabolic pathways critical for ubiquinone biosynthesis.

Benzisoxazole scaffolds, though infrequently reported from natural sources, are prominent in pharmacology with activities spanning anti-HIV, antimicrobial, antipsychotic, anti-inflammatory, antioxidant, anticancer, anticonvulsant, and antidiabetic applications. Several 1,2-benzisoxazole derivatives (zonisamide, risperidone, paliperidone, iloperidone) are FDA-approved. Earlier work (over 30 years ago) reported Gram-negative selectivity of 3,6-dihydroxy-1,2-benzisoxazole with low mammalian toxicity, but A. baumannii had not been tested. The background also reviews ubiquinone biosynthesis: in bacteria, 4-hydroxybenzoate (4-HB) is formed from chorismate by chorismate pyruvate-lyase (CPL) and is subsequently prenylated by 4-HB octaprenyltransferase (UbiA), pathways potentially targetable by small molecules resembling 4-HB.

- Source and initial screening: Crude organic extract from the exudate of a marine Bradyrhizobium denitrificans isolate (B158; 16S rRNA GenBank MF113387.1) was screened using the p-iodonitrotetrazolium chloride (INT) assay for MDR reversal activity against E. coli MDR strains MG1655 ABC/PABM and MG1655 ABC/pXYM, showing erythromycin potentiation. - Isolation and identification: Bioassay-guided fractionation of extract from 72 l culture yielded compound 1 as an amorphous white powder. HRESIMS ([M+H]+ 150.0204) indicated C7H5NO3. 1H NMR signals and formula supported identification as 3,6-dihydroxy-1,2-benzisoxazole. Only 1 mg was isolated from 14.1 g crude, so the compound was synthesized de novo; synthetic material matched the natural product by NMR. - Antibacterial testing: MICs were determined against Gram-negative panels including clinical A. baumannii strains (L1051, NR-13382, NR-17786, ATCC 19606) using Mueller-Hinton broth (MHB), and additional tests in minimal medium DM01 (defined, sodium pyruvate carbon source) to compare potency across media. MICs were also determined for E. coli, Pseudomonas aeruginosa, and Klebsiella pneumoniae in DM01 vs MHB. - Structure–activity relationships: A panel of analogs (2–10) was synthesized/purchased to probe substituent effects on the benzene ring and modifications to the isoxazole ring. Variants included replacement or modification of the C6 hydroxyl (e.g., amino, methoxy), additional hydroxyl at C4, N-methylation of the isoxazole nitrogen, and an oxazolone analog. - Phenotypic antagonism assays: Growth of A. baumannii Ab197 and P. aeruginosa PA14 in DM01 with or without compound 1 was assessed using Biolog phenotype microarray plates (PM1–PM5), with measurements at 17 and 23 h, to identify metabolic supplements antagonizing activity. Follow-up assays in PA14 tested 2-fold serial dilutions of 4-hydroxybenzoic acid and 4-hydroxybenzaldehyde (62.5–250 µg ml⁻¹) in the presence of 1. - Molecular modeling and docking: An A. baumannii CPL homology model was built using E. coli CPL (UniProt P26602; PDB template) due to lack of A. baumannii crystal structure. Superposition yielded RMSD 0.780 Å. Docking of compound 1 into the solvated model assessed binding affinity and interactions, comparing with 4-HB binding in E. coli CPL. Predicted binding energies and hydrogen-bond networks with conserved residues (Arg76/Arg78, Glu155/Glu157, peptide amide of Met34/Met35, Leu114/Leu116) were analyzed.

- Potent anti-A. baumannii activity: 3,6-dihydroxy-1,2-benzisoxazole (1) inhibited clinical MDR A. baumannii with MICs 6.25–50 µg ml⁻¹ in MHB; strongest against NR-13382 (6.25 µg ml⁻¹) and L1051 (12.5 µg ml⁻¹). - Media-dependent potency: Across Gram-negative species, activity was markedly greater in minimal medium DM01 than in MHB. Examples: E. coli ATCC 25922 MIC 0.25–0.5 µg ml⁻¹ in DM01 vs >500 µg ml⁻¹ in MHB; E. coli UNT156 0.25 vs 64 µg ml⁻¹; P. aeruginosa PA14 8–16 vs >500 µg ml⁻¹; K. pneumoniae BAA-2146 1–2 vs >500 µg ml⁻¹; A. baumannii UNT190 2 vs 128 µg ml⁻¹; A. baumannii UNT197 2 vs 16 µg ml⁻¹. - SAR insights: No analogs (2–10) surpassed 1 in potency. A hydrogen-bond donor at C6 is required; C6 hydroxyl is optimal. C6 methoxy or removal of H-bond donor caused large potency losses; C6 amino retained only modest activity. Adding a C4 hydroxyl abolished activity. N-methylation of the isoxazole (8, 9) and an oxazolone analog (10) eliminated activity. - Antagonism by 4-HB: 4-hydroxybenzoic acid and 4-hydroxybenzaldehyde antagonized the antibacterial effect of 1 in DM01, rescuing growth at 62.5–250 µg ml⁻¹. - Docking to CPL: Homology modeling and docking indicated favorable binding of 1 to A. baumannii CPL primary site with predicted affinity −5.8 kcal mol⁻¹, forming five hydrogen bonds with conserved residues analogous to 4-HB binding, suggesting potential competitive product inhibition of CPL. - Mechanistic leads: Data implicate two enzymes in the 4-HB/ubiquinone pathway—CPL (chorismate pyruvate-lyase) and 4-HB octaprenyltransferase (UbiA)—as potential targets for 1.

The findings identify a naturally derived 1,2-benzisoxazole with selective Gram-negative antibacterial activity, notably against MDR A. baumannii, and reveal strong media dependence that hints at metabolic pathway engagement. The reversal of activity by 4-HB and 4-hydroxybenzaldehyde points to interference with 4-HB-dependent ubiquinone biosynthesis, essential for aerobic respiration. Two mechanistic hypotheses are advanced: (1) product-competitive inhibition of bacterial chorismate pyruvate-lyase (CPL), supported by docking that reproduces the conserved hydrogen-bond network seen in 4-HB binding and a favorable predicted affinity; and (2) competitive inhibition at 4-HB octaprenyltransferase (UbiA), consistent with enzyme substrate promiscuity and SAR requirements (C6 H-bond donor and para-position H-bonding capability) aligning with known active-site interactions (Asp191 and Arg72). The SAR results align with these mechanisms: loss of the C6 hydroxyl or isoxazole integrity disrupts putative binding, while addition of a C4 hydroxyl mirrors known intolerance of ortho-disubstituted 4-HB analogs by UbiA. The Gram-negative specificity and strengthened activity under nutrient limitation are congruent with targeting respiratory cofactor biosynthesis. Given the increasing value of narrow-spectrum agents and diagnostic advances, a benzisoxazole targeting ubiquinone biosynthesis could complement current treatments for MDR A. baumannii while potentially limiting microbiome disruption.

This work reports 3,6-dihydroxy-1,2-benzisoxazole, produced by Bradyrhizobium denitrificans and accessible via concise synthesis, as a new chemotype with potent, media-enhanced activity against MDR Acinetobacter baumannii and other Gram-negative bacteria. Structure–activity studies indicate the natural scaffold is close to optimal, requiring a C6 hydrogen-bond donor and an intact 1,2-benzisoxazole ring. Antagonism by 4-HB and molecular docking implicate ubiquinone biosynthesis—via CPL and/or UbiA—as plausible targets. Future work should (i) biochemically validate CPL and UbiA inhibition by 1 and key analogs; (ii) elucidate resistance mechanisms and target engagement in cells; (iii) optimize substituents (e.g., explore 3-position and other ring positions) for potency and pharmacokinetics; (iv) assess in vivo efficacy and safety; and (v) investigate ecological roles and expression conditions in the producing bacterium.

Mechanistic conclusions rely on phenotypic antagonism and in silico docking without direct biochemical inhibition assays for CPL or UbiA. MIC improvements are media-dependent, and standard rich media show reduced activity, which may limit translational predictability. The SAR set is limited and did not improve potency. In vivo efficacy and contemporary toxicity assessments were not reported. The natural product yield was low, and activity was not evaluated against a broad Gram-positive panel in this study.