Penicillin kills bacteria through a four-membered ring under enormous strain — the shared weapon of all beta-lactam antibiotics. How this ring is assembled has puzzled scientists for over forty years. The answer, published in *Nature Catalysis*, uses an old technique in a new way: filming a chemical reaction like a movie.
The Problem: Intermediates Are Too Short-Lived
The beta-lactam ring formation is catalyzed in a single step by isopenicillin N synthase (IPNS), an iron-dependent enzyme. The overall reaction has been known since the 1940s — but not the intermediate steps. The reaction involves extremely unstable intermediates that exist for only fleeting moments, while traditional structural biology relies on static snapshots of stable states.
An important clarification: the X-ray pulses are femtosecond-scale, but the reaction timescale is seconds. The femtosecond pulse exists to record a diffraction image before radiation damage destroys the sample — it does not mean the reaction was filmed at femtosecond intervals. The accurate description: researchers used second-scale controlled oxygen exposure combined with femtosecond-pulse measurements to assemble the structure frame by frame, with snapshots taken at 1, 2, 3, 4, 6, 8, and 10 seconds after oxygen exposure.
The setup is elegantly simple. Over a thousand droplets containing anaerobic enzyme microcrystals were deposited on a 2-millimeter-wide moving tape. As the tape passed through an oxygen chamber, oxygen rapidly diffused into the crystals, synchronously starting the reaction across the whole sample. By precisely controlling tape speed, researchers set how long each batch of crystals had been reacting when it reached the X-ray interaction point — all at physiological temperature and pressure, with no freezing and no chemical trapping.
What Was Captured
The paper's title is direct: identification of unexpected intermediates. Two previously never directly observed intermediates were captured:
- A thioaldehyde intermediate, appearing before beta-lactam ring formation. At 3 to 6 seconds after oxygen exposure, the team observed evidence of this intermediate, with its sulfur no longer coordinating to the iron atom at the enzyme's center.
- A monocyclic beta-lactam intermediate, the first cyclic structure on the path to the complete penicillin scaffold. Follow-up snapshots using deuterium-labeled substrate revealed this intermediate, along with restoration of the iron-sulfur bond.
- No new antibiotic was produced
- No testing against drug-resistant bacteria
- No animal experiments
- No clinical trials
- No product timeline
This is the watershed of the whole mechanism: sulfur is first oxidized, then detaches from iron, then re-bonds. Iron plays an active role in this pathway, not a passive one.
Another finding: water molecules inside the enzyme deliver protons at critical moments, while subtle movements of the protein backbone help position these waters relative to the substrate. The enzyme's shape and its chemistry work together.
Why This Paper Is More Credible
Structural biology has a long-standing problem: electron density maps are ambiguous — a density peak does not equal a uniquely determined atomic configuration.
This paper collected two data types simultaneously: time-resolved serial femtosecond crystallography (tr-SFX) gives ordered atomic positions, while X-ray emission spectroscopy (tr-XES) probes the electronic environment of all iron atoms in the beam. Both datasets come from the same sample and the same X-ray pulse.
The crystallographic structures must remain consistent with the iron oxidation states. When multiple molecular states coexist at the same time point, the spectroscopic data constrains the crystallographic conclusions. This is methodologically key: crystallography does not speak alone; the two probes limit each other's degrees of freedom.
The paper explicitly notes that coexisting molecular states exist at some time points. So this remains a mechanistic model pieced together from multiple lines of evidence — not a continuous movie of every atomic motion.
The Weight of Forty Years
The team composition itself footnotes this history. The Ineos Oxford Institute for Antimicrobial Research at Oxford University led the work. More than eighty years ago, Florey and Chain turned penicillin into a usable antibiotic at Oxford's Sir William Dunn School of Pathology; earlier still, in 1945, Nobel laureate Dorothy Hodgkin solved penicillin's molecular structure by X-ray crystallography at Oxford.
This paper follows the structure Hodgkin solved eighty years ago, filling in the intermediates of the synthesis pathway. Collaborators include the XFEL Hub at Diamond Light Source, Lawrence Berkeley National Laboratory, and the PAL-XFEL and SLAC accelerator facilities — the product of more than a decade of international collaboration. First author Patrick Rabe is a Wellcome Career Development Award researcher; the senior author is Oxford chemistry professor Christopher Schofield.
The paper was published in *Nature Catalysis*, DOI: 10.1038/s41929-026-01618-4. Related work goes back further: a 1999 *Nature* paper studied the IPNS reaction cycle, and in 2021 a team used X-ray free-electron lasers to study protein motions during catalysis.
What This Paper Is Not
This must be stated clearly, because penicillin as a topic is easily read as more optimistic than reality:
WHO data shows why this matters: globally, one in six laboratory-confirmed bacterial infections is resistant to antibiotic treatment. The World Health Organization's 2025 evaluation of the antibacterial pipeline also found it remains insufficient relative to the burden of resistant infections.
These two numbers explain why the research matters — but they do not turn it into a drug development result. The enzyme family involved also extends beyond antibiotics: IPNS belongs to the large family of iron-dependent oxygenases, which includes members involved in human oxygen sensing. Mechanistic insights may extrapolate elsewhere, provided those enzymes share similar reaction mechanisms.
The Remaining Step
Eighty-plus years after Hodgkin solved the structure, the bridge to capturing intermediates was built by technology rather than ideas. X-ray free-electron lasers began operating in the late 1990s; in 2021 the same collaborators were already using one to study this enzyme. The final step was completed in 2026: making the reaction proceed controllably to a specified moment before the femtosecond pulse arrives.
The method's value extends beyond this one enzyme. It offers a path for any catalytic reaction whose intermediates exist for far shorter than the capture window of traditional methods.