3.7 Bacteria in 10 Billion Survive This New Antibiotic. It Hits a Ribosome Target No Drug Has Ever Touched.

3.7 × 10⁻¹⁰.

That is the spontaneous resistance frequency of Escherichia coli exposed to manikomycin, the antibiotic published in Nature this month by teams at the University of Illinois Chicago, McMaster University, and the University of Hamburg. In plain language: for every 10 billion E. coli cells that encounter the drug, roughly 3.7 survive to pass on genetic resistance. For the multidrug-resistant pathogen Klebsiella pneumoniae, classified by the WHO as a critical-priority threat, the number is 1.1 × 10⁻⁸, roughly one in every 100 million.

To understand how remarkable those numbers are, compare them to the antibiotics clinicians actually use when nothing else works:

Antibiotic	Target	Typical Resistance Frequency (E. coli)
Colistin (last resort)	Outer membrane	~10−7 to 10−6
Ciprofloxacin	DNA gyrase	~10−8 to 10−9
Meropenem	Cell wall (PBPs)	~10−8
Manikomycin	Ribosome E-site (novel)	3.7 × 10−10

Manikomycin is 100 to 1,000 times harder for E. coli to resist than antibiotics doctors consider weapons of last resort. The reason, according to the paper, is structural novelty: the drug binds to a part of the ribosome that no antibiotic in history has occupied, so bacteria lack any preexisting defense.

A 75-Year-Old Bacterium’s Secret

Streptomyces rimosus has been studied since 1950, when it gave modern medicine oxytetracycline, one of the most widely prescribed antibiotics of the 20th century. Researchers at McMaster University’s Wright Actinomycetes Collection screened methanolic extracts from 255 bacterial strains using improved fractionation techniques designed to separate compounds whose activity is normally drowned out by the dominant antibiotic a strain produces. In S. rimosus WAC 7405, after removing the oxytetracycline fractions, they found antimicrobial activity in two previously ignored fractions. Neither contained any compound in GNPS, SciFinder, or PubChem databases. Something genuinely new.

The compound was a cyclic depsipeptide, nine amino acids linked through an ester bond, with five variants identified (MKM-A through MKM-E). The team named it manikomycin, from the Hindi and Punjabi word manik, meaning “precious gem.” They validated the responsible 67 kb biosynthetic gene cluster by heterologous expression in Streptomyces coelicolor, confirming the compound is a genuine natural product of the man gene cluster.

The E-Site: A Binding Pocket Nobody Knew Was Available

Approximately one-third of all prescribed antibiotics target the bacterial ribosome. They bind the A-site (aminoglycosides), the peptide exit tunnel (macrolides), or the peptidyl transferase center (chloramphenicol, linezolid). None bind the E-site of the 50S large subunit, the pocket where deacylated tRNA exits after delivering its amino acid. Until now, no one had a drug that could.

Cryo-EM structures at 2.4 Å resolution show manikomycin wedged into the E-site at the junction of three 23S rRNA helices (H13, H21, and H88). The drug physically blocks the CCA-end of deacylated tRNA from entering the pocket, preventing the ribosome from completing translocation, the step where the next amino acid is brought into position. Translation stalls, and the bacterium stops making the proteins it needs to survive.

The eukaryotic ribosome has a protein, eL42, that physically occupies the space where manikomycin binds in bacteria. This structural difference explains why the drug inhibits bacterial translation at 0.6 µM but requires 9.2 µM to affect mammalian translation (a 15-fold selectivity gap) and shows no toxicity to human cell lines at concentrations up to 256 µg/mL. The selectivity is not engineered. It is an accident of 3 billion years of ribosomal evolution, a structural divergence that no drug designer could have predicted or introduced on purpose.

Why Resistance Is So Rare

Three factors compound to make resistance difficult. First, the E-site has never been under antibiotic selection pressure, so no resistance gene for this target exists in the global resistome of clinical pathogens. Second, manikomycin enters cells through at least two independent transporter systems (SbmA and YejABEF); knocking out either one increases the minimum inhibitory concentration only 2- to 8-fold, insufficient for clinical resistance. Third, the self-resistance mechanism used by the producing bacterium (S. rimosus) involves a specific methyltransferase, ManE, that modifies a single nucleotide (C2395) in the 23S rRNA. This gene exists only in S. rimosus strains that carry the manikomycin gene cluster and is not found in any human pathogen.

“Bacteria need to jump through hoops to find resistance,” Alexander Mankin, distinguished professor at UIC’s Retzky College of Pharmacy and a co-author, said in a university statement.

The Efficacy Data (and the Problem)

In time-kill assays, manikomycin was bactericidal against both E. coli and K. pneumoniae. An ex vivo model using human blood inoculated with K. pneumoniae showed approximately 1,000-fold reduction in bacterial load after six hours of exposure at five times the minimum inhibitory concentration. In a Caenorhabditis elegans infection model, 55 to 60 percent of nematodes survived to day six with manikomycin treatment, compared to 10 to 30 percent in untreated controls (P < 0.0001).

Then comes the pharmacokinetics. Mice tolerated doses up to 220 mg/kg/day with no acute toxicity. But at 50 mg/kg subcutaneous, peak plasma concentration reached only 9.13 µg/mL and the terminal half-life was approximately 36 minutes. The drug cleared too fast to maintain therapeutic levels. Gone in half an hour. No efficacy was observed in mouse infection models. The paper is transparent about this: “Insufficient plasma exposure, rather than inherent pharmacological inactivity, accounts for the lack of efficacy observed in our experiments.”

How Many Hidden Antibiotics Are Still Buried?

This is where the implications extend beyond one compound: the Wright Actinomycetes Collection at McMaster screened 255 bacterial strains, and exactly one yielded a novel mechanism. If that hit rate (roughly 1 in 255, or 0.4%) holds across the estimated 10,000+ actinomycete strains maintained in major collections worldwide (the DSMZ German Collection alone catalogs over 4,000 Streptomyces species), the arithmetic suggests 40 or more entirely novel antibiotics could be hiding in existing banks, masked by better-known compounds in crude extracts.

That is a back-of-the-envelope number with enormous uncertainty. Not every hidden compound will be therapeutically relevant, and many will fail preclinical hurdles far more mundane than a novel ribosomal binding site. But the core insight is durable: 80 years of antibiotic discovery relied on crude extracts that amplify dominant compounds and drown minor ones. Improved fractionation changes the denominator of what is detectable. “There is likely so much still to be discovered through fractionation,” Manpreet Kaur, first author and postdoctoral fellow in Gerry Wright’s lab at McMaster, said. “Revisiting the extracts of even well-studied bacteria like Streptomyces may lead to similar discoveries in the future.”

Limitations

Several caveats belong on the table. Manikomycin’s spectrum is narrow: it kills E. coli, K. pneumoniae, and mycobacteria, but not Staphylococcus aureus, Pseudomonas, or most other Gram-positives. The binding site is conserved across bacterial species, so the gap is probably uptake, not target affinity, but that does not help the patient with a staph infection. The 1,000-fold kill in human blood is an ex vivo result; the C. elegans model, while statistically robust, is a nematode, not a mammal. The 36-minute half-life in mice is a hard pharmacokinetic reality that chemical modification may or may not overcome. The polycationic nature of the molecule raises nephrotoxicity concerns common to this class. And the 0.4% hit-rate estimate for hidden antibiotics in strain collections is extrapolation from a single screen. The actual yield could be higher or lower depending on fractionation quality, the diversity of the strain library being screened, the rigor of the dereplication workflow used to distinguish novel compounds from known ones, and plain luck in which extracts cooperate with a given separation technique.

Strongest Counterargument

Derek Lowe, a medicinal chemist who has spent decades in pharmaceutical R&D and writes the In the Pipeline column for Science, described manikomycin as “a really good example” of an overlooked compound, but added bluntly that “it is not going to be a wonder drug.”

He is probably right, at least about manikomycin-A itself. The history of antibiotic development is littered with compounds that killed bacteria brilliantly in test tubes and failed in animals, compounds whose discoverers believed they had found the next penicillin only to watch them dissolve in the bloodstream or poison the kidneys. Teixobactin, discovered in 2015 with a novel mechanism, generated enormous excitement and remains in preclinical development over a decade later. Platensimycin, clovibactin, and darobactin all showed novel mechanisms in Nature papers and have not yet reached patients. The pharmacokinetic valley between “kills bacteria” and “cures infection,” the stretch of development where a promising compound must demonstrate it can reach the site of infection at therapeutic concentrations for long enough to matter, has a failure rate exceeding 90% for antibiotic leads. Manikomycin’s 36-minute half-life puts it squarely in that valley.

Where this counterargument weakens is in what the compound teaches. Even if manikomycin-A never becomes a drug, it is the first chemical probe for the bacterial ribosomal E-site. The 2.4-Å cryo-EM structure provides a blueprint for designing derivatives and entirely new scaffolds optimized for that binding pocket. The paper explicitly notes that the scaffold “is amenable to chemical modification” and that analogue efforts are underway. The value may not be the molecule. It may be the map.

The Bottom Line

Antimicrobial resistance kills an estimated 1.27 million people per year directly, and 4.95 million in association with resistant infections. Carbapenem-resistant K. pneumoniae, one of manikomycin’s primary targets, is the bacterium driving some of the worst hospital outbreaks globally. The antibiotic pipeline has delivered only a handful of truly novel mechanisms in the past two decades: daptomycin (2003), fidaxomicin (2011), cefiderocol (2019), and lefamulin (2019). Each targets a site where resistance already exists, meaning bacteria had decades of evolutionary practice at evading them before patients ever swallowed the first dose.

Manikomycin opens a new front. Whether it becomes a drug depends on chemistry that has not been done yet. But its existence proves two things simultaneously: that the bacterial ribosome still has unoccupied binding pockets after 80 years of study, and that the most-studied organisms in the strain collections of the world are still hiding compounds we missed because our analytical tools were too crude to find them.

What you can do with this: If you work in antibiotic discovery, the fractionation strategy described here is directly reproducible. The WAC library at McMaster is accessible to academic collaborators. If you fund antimicrobial R&D, this paper makes the case for re-screening existing strain collections with modern fractionation and metabolomics, not just hunting for new organisms. If you are a clinician tracking the AMR pipeline, bookmark the E-site as a new pharmacological target class and watch for manikomycin analogues with improved pharmacokinetics. The mechanism works. The delivery does not, yet.