Arthritis Drug Turned Molecular Weapon Strikes a Rare Lung Pathogen

In a striking example of scientific serendipity guided by rational design, chemists have transformed diflunisal, a long-established anti-inflammatory painkiller, into a family of molecules that display astonishingly potent and highly selective activity against Mycobacterium kansasii, a lesser-known but increasingly troublesome relative of the tuberculosis bacterium. The study, published in Results in Chemistry by a team led by Václav Pflégr and Martin Krátký of Charles University, reports that two of the new diflunisal-derived ureas inhibit M. kansasii at concentrations as low as 0.03 micrograms per millilitre, a level of potency that rivals or exceeds first-line antimycobacterial drugs such as rifampicin, isoniazid, and ciprofloxacin in the same assays. Even more remarkable, the compounds are essentially inert against every other bacterium and fungus tested, suggesting they engage a molecular target unique to this particular pathogen.

The significance of this work lies in the growing crisis surrounding non-tuberculous mycobacteria, or NTM. Unlike Mycobacterium tuberculosis, NTM are ubiquitous environmental organisms that can colonise the lungs of patients with pre-existing structural damage, such as bronchiectasis or chronic obstructive pulmonary disease, and of those who are immunosuppressed. Their incidence is rising worldwide, and they are notoriously difficult to treat because of intrinsic resistance to broad classes of antibiotics. M. kansasii, in particular, is considered a relatively virulent pulmonary pathogen whose management typically requires prolonged multidrug regimens anchored by rifampicin. Diagnosis is complicated by nonspecific clinical presentations and by the slow, sometimes unreliable nature of culture-based detection. For many patients, the therapeutic window is narrow, tolerability is poor, and outcomes are variable, which makes the discovery of a new, selectively active chemotype both timely and clinically meaningful.

The researchers built on a strategy that has become one of the most productive in modern infectious-disease pharmacology: drug repurposing. Rather than searching for new molecules from scratch, repurposing examines approved drugs and established scaffolds for previously unrecognised antimicrobial properties. Because such molecules already possess well-characterised safety, pharmacokinetic, and toxicological profiles, early-stage development can proceed more rapidly than with entirely novel chemical entities. Non-steroidal anti-inflammatory drugs have proven particularly fertile ground. Several members of this class exhibit direct antibacterial, anti-biofilm, or antibiotic-potentiating effects, and linezolid, originally developed as an antibacterial agent, has in fact found its way into the treatment of NTM infections. Diflunisal, a difluorophenyl derivative of salicylic acid, had previously been shown to modulate virulence in Staphylococcus aureus, reduce biofilm formation, and inhibit tuberculosis bacilli in vitro with minimum inhibitory concentrations in the range of 25 to 50 micromolar, hinting that its biphenyl skeleton could be a versatile starting point.

To unlock that potential, the team systematically redesigned the diflunisal scaffold. The carboxylic acid group was esterified to boost lipophilicity, a physicochemical property frequently associated with improved antimycobacterial activity, while the phenolic hydroxyl was replaced with an amino group to enable the installation of urea moieties, a pharmacophore with a strong track record in antimycobacterial drug discovery. The urea N-substituents were chosen to probe chain length, lipophilicity, and topology, spanning short ethyl chains, longer hexyl and dodecyl chains, a cyclic cyclohexyl group, and aromatic phenyl, benzyl, and phenethyl variants. The synthetic route began with a palladium-catalysed Suzuki–Miyaura cross-coupling between commercially available (2,4-difluorophenyl)boronic acid and methyl 5-bromoanthranilate, executed at 90 degrees Celsius with tetrakis(triphenylphosphine)palladium(0) and potassium carbonate, which delivered the key aminobiphenyl intermediate in an excellent 89 percent yield. Reaction of this intermediate with various isocyanates furnished the asymmetric ureas in yields ranging from 28 to 97 percent, with the ethyl derivative proving especially difficult because of the volatility of ethyl isocyanate and the formation of a diethylurea side product.

The biological screening revealed a selectivity profile that surprised even the researchers. Against Mycobacterium tuberculosis H37Ra, the rapidly growing M. smegmatis and M. aurum, and even M. avium, the new derivatives showed little or no relevant inhibition. But against M. kansasii the results were dramatic. The ethyl-substituted urea inhibited the reference strain at 0.06 micrograms per millilitre and the hexyl analogue at 0.03 micrograms per millilitre, corresponding to micromolar potencies of just 0.18 and 0.077 respectively. The dodecyl and phenyl derivatives retained strong activity at 0.20 and 0.78 micrograms per millilitre, while diflunisal itself required 125 micrograms per millilitre to achieve the same effect against this organism. In matched comparisons, the most active analogues were therefore more than 4000-fold more potent than the parent drug against M. kansasii, and outperformed isoniazid and ciprofloxacin while at least matching rifampicin in the same assay.

Perhaps the most instructive structure–activity lesson came from a near-identical pair of compounds. The linear n-hexyl urea and its cyclohexyl counterpart differ only modestly in calculated lipophilicity, with clogP values of 6.47 and 5.85 respectively, yet constraining the flexible alkyl chain into a rigid ring abolished activity by more than 4000-fold. Similarly, inserting a methylene or ethylene spacer between the urea nitrogen and a phenyl ring, as in the benzyl and phenethyl analogues, was markedly detrimental. The authors concluded that substituent topology and conformational properties, rather than lipophilicity alone, are the critical determinants of activity, a finding that strongly implies a specific molecular interaction rather than a nonspecific membrane-disruptive effect. Supporting this interpretation, an isosteric butyramide analogue was completely inactive, and carbamate isosteres generally underperformed their urea counterparts, underscoring the importance of the two hydrogen-bond-donating NH groups of the urea linkage, although the cyclohexyl carbamate did retain moderate dual activity against both M. tuberculosis and M. kansasii.

Importantly, the two most active ureas did not lose their edge when tested against M. kansasii 6509/96, a clinical isolate resistant to isoniazid. Both compounds maintained a minimum inhibitory concentration of 0.125 micromolar at readout points after 7, 14, and 21 days of continuous exposure, indicating sustained growth inhibition without delayed regrowth. The potency also compares favourably with previously reported antimycobacterial ureas: sulfamethoxazole-based ureas active against M. kansasii typically reached low micromolar values of at least 4 micromolar, and recent thioureas achieved a best median inhibitory concentration of 1.5 micromolar against the same reference strain. Against the panel of Gram-positive and Gram-negative bacteria, including methicillin-resistant Staphylococcus aureus, Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa, the soluble derivatives were entirely inactive, as they were against eight fungal pathogens spanning Candida, Aspergillus, and dermatophyte species, confirming an exquisitely narrow species selectivity.

Safety signals in the preliminary screens were equally encouraging. When evaluated against MonoMac-6 human monocytic cells, a model relevant to the intracellular niche that pathogenic mycobacteria occupy, and against HepG2 hepatic cells, a metabolically active surrogate for systemic drug exposure, none of the urea, amide, or carbamate derivatives showed any cytostatic effect at concentrations up to 250 micromolar. Given that the hexyl urea kills M. kansasii at approximately 0.077 micromolar, this translates into a selectivity index exceeding 3250, meaning the antimicrobial effect is over three thousand times more potent than any measurable toxicity toward mammalian cells. The researchers are careful to note that the cytostatic assay used uninfected cells and therefore cannot confirm intracellular antimycobacterial efficacy, and that mechanism-of-action work remains outstanding, but the preliminary safety profile is nonetheless exceptional for an early-stage hit series.

What could explain such dramatic specificity for a single mycobacterial species? The authors propose that M. kansasii differs from M. tuberculosis and from rapidly growing mycobacteria in the composition and organisation of its cell-envelope lipids, including its distinctive lipooligosaccharides, which may alter the permeability, partitioning, or intracellular retention of amphiphilic fluorinated biphenyl derivatives. Interspecies variation in efflux pump capacity, drug-inactivating enzymes, and target susceptibility could further shape intracellular exposure. Whether the compounds act through preferential accumulation, through interaction with a species-dependent molecular target, or through interference with a resistance determinant cannot yet be distinguished with the available data. The team suggests that whole-genome sequencing of resistant mutants, combined with target-identification and uptake studies, could resolve the question in future work.

Fluorinated biphenyl ureas thus emerge from this study as a previously unrecognised chemotype for attacking M. kansasii, one that warrants systematic expansion. The path forward, as the researchers outline, involves broadening the structure–activity analysis, testing a larger panel of clinical isolates, establishing efficacy in intracellular infection models, and identifying the molecular basis of the species selectivity that makes these molecules so unusual. An unexpected synthetic bonus also came out of the work: during carbamate preparation, a side reaction yielded cyclic benzoxazine-dione scaffolds in just two steps, a shortcut to structures that normally require three-step routes. It is a fitting coda for a study built on repurposing, demonstrating that even the chemistry of an old painkiller can still yield surprises with the potential to reshape how we fight stubborn lung infections.

Subject of Research: Design of diflunisal-derived fluorinated biphenyl urea and carbamate derivatives with selective antimycobacterial activity against Mycobacterium kansasii

Article Title: Diflunisal-derived fluorinated biphenyl urea and carbamate derivatives with selective activity against Mycobacterium kansasii

Article References: Pflégr, V., Janďourek, O., Šikorová, E., Konečná, K., Bősze, S., & Krátký, M. (2026). Diflunisal-derived fluorinated biphenyl urea and carbamate derivatives with selective activity against Mycobacterium kansasii. Results in Chemistry, 31, Article 103965. https://doi.org/10.1016/j.rechem.2026.103965

Image Credits: AI Generated

DOI: 10.1016/j.rechem.2026.103965

Keywords: diflunisal, Mycobacterium kansasii, non-tuberculous mycobacteria, drug repurposing, urea derivatives, antimycobacterial agents, structure-activity relationships, NSAIDs, fluorinated biphenyls, antimicrobial resistance, medicinal chemistry, selectivity index