Cefiderocol occupies a unique position in modern infectious disease medicine. Approved as a treatment for multidrug-resistant Gram-negative infections, it is a so-called siderophore-conjugated cephalosporin, a molecule engineered to exploit the bacterium’s own starvation reflex. Under iron-poor conditions—which prevail inside the human body—bacteria desperate for the metal import iron-chelating compounds called siderophores. Cefiderocol masquerades as one of these scavengers, binding iron on its catechol moiety and hitchhiking through the bacterium’s dedicated iron transport channels, the Cir, Fiu and Piu transporters, past the outer membrane barrier that defeats ordinary antibiotics. Once inside, the drug behaves as a classical beta-lactam: it binds and inactivates penicillin-binding proteins, the enzymes that assemble and cross-link the bacterial cell wall, halting construction of the organism’s protective exoskeleton. This Trojan-horse strategy gave clinicians hope that cefiderocol could succeed where carbapenems and cephalosporins had failed, particularly against carbapenem-resistant Enterobacteriaceae, which the World Health Organization ranks among the most critical pathogens for new antibiotic development.

Yet the drug’s Achilles heel was apparent from early clinical and surveillance studies. Because cefiderocol depends on iron transport systems to reach its target, bacteria can defend themselves simply by shutting down or mutating those entry doors. Many documented resistance cases involve mutations in the TonB-dependent transport machinery, loss of the PiuD channel, alterations in iron regulation through the Fur protein and the small RNAs RyhB and PrrF, or the amplification of beta-lactamase enzymes, including structural variants of enzymes such as CMY that hydrolyze the drug more efficiently. Plasmid-borne determinants can spread these defenses horizontally between strains, amplifying the threat. The Spanish team, however, found something strikingly different when they sequenced Klebsiella pneumoniae isolates collected from their patient before and after cefiderocol therapy.

Using whole-genome sequencing and comparative genomics, the researchers compared the pair of isolates at single-nucleotide resolution. The plasmid architecture of the bacterium—the constellation of extrachromosomal DNA elements that often carry resistance genes—remained conserved between the susceptible and resistant versions. There was no acquisition of a new beta-lactamase, no introduction of a foreign iron-uptake determinant, no dramatic genetic overhaul. Instead, the analysis revealed eighty-two chromosomal variants that had accumulated or been selected during therapy, and within this set two mutations emerged as the leading candidates for the resistance phenotype. The first, and most consequential, was a single missense substitution in the ftsI gene: a change from glycine to valine at position 306 of the FtsI protein, designated G306V. The second was E91K in AcrR, a regulator of the AcrAB-TolC multidrug efflux pump, hinting at a possible secondary contribution through enhanced drug export.

The significance of the FtsI mutation lies in the biology of the protein itself. FtsI, better known in the field as penicillin-binding protein 3, or PBP3, is an essential transpeptidase that catalyzes the cross-linking of peptidoglycan during cell division, building the septum that separates one daughter cell from another. Like all beta-lactam targets, it carries a catalytic serine residue—in this case S307—around which the antibiotic must nestle to form the covalent bond that permanently disables the enzyme. The glycine at position 306 sits immediately adjacent to this catalytic linchpin. Using structural modeling, the researchers showed that swapping the smallest amino acid, glycine, for the bulkier branched-chain valine produces a dramatic volumetric increase in the local side-chain architecture. The new valine juts into the active-site cavity, creating steric hindrance that could physically impede cefiderocol from approaching and properly positioning itself relative to the catalytic serine. In essence, the bacterium did not destroy the drug, pump it out, or block its entry through iron transporters—it simply altered the lock so that the key no longer fits.

This mechanism distinguishes the Spanish case from the predominant narratives in the global molecular epidemiology of cefiderocol resistance, and that distinction carries both scientific and clinical weight. Target-site alteration of PBP3 has long been recognized as a resistance route in Pseudomonas aeruginosa, where penicillin-binding protein mutations are a leading cause of cefiderocol failure, but it has remained insufficiently defined in Klebsiella pneumoniae. By demonstrating that a single amino-acid substitution adjacent to the catalytic serine can plausibly confer resistance in this organism, the study fills an important gap and suggests that surveillance programs may be looking in the wrong place—or at least not in enough places—when they screen for the emergence of treatment failure.

The team did not stop at a single case. To assess how common the G306V substitution is across the global bacterial population, they performed large-scale genomic screening of publicly available Klebsiella pneumoniae genomes, including isolates known to be cefiderocol-resistant. The result was reassuring but nuanced: substitutions affecting ftsI, including at this specific position, are extremely rare, even among resistant isolates. The researchers caution, however, that clonal expansion of a successful resistant lineage cannot be ruled out, meaning a rare mutation today could become a clinical problem tomorrow if the right selective pressure persists. The rarity also strengthens the inference that the mutation arose under therapy in this individual patient rather than being imported from the community or hospital environment.

The companion mutation in AcrR adds an additional layer of mechanistic intrigue. AcrR is the transcriptional repressor of the acrAB operon, which encodes one of the principal multidrug efflux pumps of Enterobacteriaceae. Loss-of-function or altered-function mutations in AcrR can de-repress the pump, flooding the periplasm with export machinery capable of lowering intracellular concentrations of a wide range of compounds, including beta-lactams under some conditions. An E91K substitution could plausibly perturb the repressor’s function, and the authors position it as a key candidate contributor that may act in concert with the PBP3 alteration. Such combined mechanisms—reduced drug access at the target and enhanced drug removal from the cell—are a recurring theme in the evolution of high-level resistance, and their convergence in a single treatment episode illustrates the flexible, heterogeneous nature of how bacteria respond to this last-line agent.

For clinicians and public health officials, the implications extend beyond basic biology. Cefiderocol is frequently deployed against infections caused by carbapenemase-producing organisms when no other option remains, and therapeutic drug failure in such settings can be fatal. The study’s central recommendation is that ftsI—specifically the region within or proximal to the active site—be incorporated into genomic surveillance frameworks for cefiderocol resistance. As whole-genome sequencing becomes faster and cheaper, real-time detection of target-site mutations could inform antibiotic stewardship decisions, alert infection control teams to the emergence of resistant subpopulations during therapy, and guide the development of diagnostic assays that flag resistance before clinicians lose precious time. The findings also resonate with the broader epidemiological picture in which cefiderocol resistance is not the product of a single dominant mechanism but of many parallel evolutionary solutions, each requiring its own watchfulness.

The research emerged from a close collaboration between the Clinical Microbiology Laboratory and the Infectious Diseases Department at Miguel Servet University Hospital, together with the University of La Rioja and international partners including the University of Debrecen in Hungary. The study was approved by the ethics committee of Aragón and conducted in accordance with the Declaration of Helsinki, with open-access funding provided through the CRUE-CSIC agreement with Springer Nature. The corresponding author, Alexander Tristancho-Baró, and co-authors including Carmen Torres and Antonio Rezusta acknowledge the laboratory technicians and clinical teams whose sample processing made the before-and-after genomic comparison possible.

As the arms race between antibiotics and bacteria enters its second century, the Spanish study is a reminder that evolution rarely follows a script. Cefiderocol was designed to outsmart one of the most formidable defense architectures in biology, and in many patients it does exactly that. But in at least one infection, a quiet change of a single molecular letter—glycine to valine, at the doorstep of the enzyme’s catalytic heart—was enough to turn the Trojan horse away from the gate. Whether such target-site mutations remain rare curiosities or become the next chapter in the Klebsiella resistance story will depend on how quickly surveillance science adapts to watch for them, and how judiciously the world’s remaining last-resort antibiotics are deployed in the years ahead.