NMR Tool Fast-Tracks Antifungal Discovery in Fungal Co-Cultures

Deep inside a papaya or a pineapple, fungi wage a silent chemical war. Now researchers at the University of São Paulo have devised a way to eavesdrop on that war and identify the weapons, using nuclear magnetic resonance spectroscopy and a statistical model that narrows dozens of candidate molecules down to a testable few. The work, published in Frontiers in Natural Products, offers a faster, less biased route to discovering new antifungal compounds at a time when drug resistance is rising and new antimicrobials are badly needed.

The team, led by Mustapha N. Abubakar, Vitor S. Mazucato and Paulo C. Vieira, worked with three fruit-derived fungi: the endophyte Neofusicoccum ribis, isolated from papaya, and two pineapple-derived pathogens, Fusarium sacchari and Fusarium verticillioides. These species inhabit protease-rich niches shaped by enzymes like papain and bromelain, and both Fusarium species carry large repertoires of biosynthetic gene clusters predicted to encode an array of specialized metabolites, including the mycotoxins known as fumonisins. The researchers reasoned that fungi competing in such demanding microenvironments would be a rich, underexplored source of antifungal chemistry.

Co-cultivation has become a popular strategy in natural product research because forcing microbes into direct confrontation can switch on otherwise silent biosynthetic pathways. But translating the resulting chemical complexity into actual lead compounds remains a bottleneck. The traditional approach, bioassay-guided fractionation, is slow and iterative and tends to favor abundant constituents, often overlooking low-abundance molecules produced only during an interaction. Mass spectrometry-based metabolomics helps, but it can struggle when the goal is de novo identification of novel or poorly annotated scaffolds.

The Brazilian team’s solution centers on proton NMR, which produces quantitative, reproducible fingerprints of complex extracts without the ionization biases that plague mass spectrometry. They grew the three fungi alone, in all three pairwise combinations, and as a three-way consortium on rice substrate, extracting each culture with ethyl acetate. Crude extract yields ranged from 1.65 grams for axenic N. ribis to 8.70 grams for the trio culture, which produced the highest biomass of any condition. Each extract was then profiled by 1H NMR at 500 MHz and fed into an OPLS-DA model, a supervised statistical method that separates variation correlated with a defined response from orthogonal noise.

The biological results were striking. In axenic culture, F. verticillioides produced metabolites that inhibited N. ribis by roughly 70 percent, indicating strong baseline allelopathy. Yet when the two fungi were grown together, that inhibition collapsed dramatically, a statistically significant attenuation that suggests contact with its target suppresses the pathogen’s chemical arsenal. Meanwhile, a genuinely emergent phenomenon appeared only in the three-way culture: the trio extract inhibited F. sacchari by about 15 percent, even though this species had been resistant to every axenic and pairwise extract. The researchers call this operational synergy, meaning inhibition absent from all constituent single-species cultures, and they interpret it as the combined outcome of ecological competition and interaction-triggered secondary metabolism.

On the computational side, the OPLS-DA model compared the NMR spectra of the seven bioactive crude extracts with spectra of 31 purified metabolites previously isolated from the same cultures. The model, built with one predictive and eight orthogonal components, showed strong internal diagnostics, with a predictive ability statistic Q2 of 0.748 and a significant cross-validated ANOVA p-value of 8.2 times 10 to the minus 4. Variable importance analysis and S-plot decomposition converged on a spectral signature dominated by olefinic protons around 5.35 ppm and aliphatic signals near 1.31 ppm, features consistent with the unsaturated and terpenoid or polyketide-like frameworks typical of fungal natural products.

By ranking the 31 purified compounds according to their spectral similarity to the bioactive extract cluster, the model generated a shortlist of five candidates. The top-ranked compound, number 27, identified as 7-hydroxy-3-(hydroxymethyl)-5-methyl-1H-isochromen-1-one, had been isolated exclusively from the trio co-culture, the only condition that inhibited F. sacchari, giving it strong ecological plausibility. Compound 5, a constitutively produced isocoumarin from N. ribis, ranked close behind. The other prioritized compounds were numbers 16, 6 and 11.

Preliminary antifungal screening at a single concentration of 50 micrograms per milliliter gave initial support to the prioritization. All five compounds showed statistically significant activity against at least one target fungus. Compound 27 inhibited F. sacchari by 20.9 percent, aligning with its exclusive provenance from the trio culture that produced emergent inhibition of that same species. Compounds 11, 5 and 6 were actually more potent against F. verticillioides, with compound 11 achieving 41.3 percent inhibition. That pattern suggests the emergent trio phenotype likely arises from a combination of metabolites rather than a single molecule, and it underscores the workflow’s value in generating multiple testable hypotheses simultaneously.

The authors are careful to frame the model as a hypothesis-generating triage tool, not a validated bioactivity predictor. Because it compares complex mixtures with pure compounds, class separation may partly reflect differences in sample complexity rather than bioactivity itself. The team also acknowledges a list of candid limitations: the small number of bioactive extracts relative to the number of spectral variables raises overfitting risk, permutation testing was not performed, spectra of pure compounds were acquired in different solvents than the crude extracts, the bioassays lacked positive controls and blinding, and all activity testing was done at a single concentration. Dose-response assays will be needed before any functional relevance or potency claims can be made.

Even with those caveats, the study demonstrates a generalizable and rational path from ecological observation to targeted compound discovery. By combining ecologically faithful co-culture design with quantitative NMR fingerprinting and supervised prioritization, the workflow reduced a 31-compound library to two or three chemically explicit hypotheses, a task that conventional fractionation would have tackled far less efficiently. As antifungal resistance continues to climb and the World Health Organization highlights the urgent need for new agents, approaches that squeeze more discovery out of every microbial confrontation may prove increasingly valuable to chemists and microbiologists alike.

Subject of Research: NMR-based metabolomics for prioritizing antifungal metabolites from fungal co-cultures

Article Title: 1H NMR-based profiling and candidate prioritization of interaction-associated metabolites from fungal co-cultures

Article References: 1H NMR-based profiling and candidate prioritization of interaction-associated metabolites from fungal co-cultures. (n.d.). https://doi.org/10.3389/fntpr.2026.1879578

Image Credits: AI Generated

DOI: 10.3389/fntpr.2026.1879578

Keywords: fungal co-culture, 1H NMR metabolomics, OPLS-DA, antifungal compounds, natural product discovery, chemical ecology, Fusarium, Neofusicoccum ribis, secondary metabolites, biosynthetic gene clusters, endophytes, operational synergy