Many of the most potent antibiotics originate from bacteria. A large proportion belong to a class of compounds known as nonribosomal peptides (NRP). They are produced by large bacterial enzyme machineries – the nonribosomal peptide synthetases, or NRPS for short. These function like molecular assembly lines: similar to the stations in a car factory, where new parts are added to the car at each step, several specialized modules string together amino acid by amino acid to build a peptide.
The finished product is used in nature as a weapon in microbial competition – and by us humans, chemically optimized, as an active ingredient in medicines. The research group of Helge Bode at the Max Planck Institute for Terrestrial Microbiology aims to engineer the NRPS machinery at the genetic level to create entirely new drug candidates.
Like ordinary proteins, NRP are subject to evolution – but the way they change is unusual. While ordinary proteins evolve mainly through random mutations of individual DNA building blocks, NRPS exchange entire gene segments with one another. This process is called recombination. The result is new combinations of modules – and thus new peptides with altered chemical properties.
How exactly this diversity arises and which forces drive it has been poorly understood, not least because frequent recombination pushes classical methods of evolutionary research to their limits. Standard tests that normally distinguish selection from random evolution do not work here, because they are designed for changes at individual DNA positions rather than the exchange of entire DNA segments.
A team led by Helge Bode and Georg Hochberg from the University of Marburg has now solved this problem through an elegant approach. They focused on the GameXPeptide (GXP) – an insecticide produced by the bacteria Xenorhabdus and Photorhabdus. These bacteria live in a complex three-way partnership: They live in the gut of nematodes, which in turn infect insect larvae. Once the host is infected, the bacteria are released by the nematode and kill the insect – partly through GameXPeptides. This system of predator, host, and parasite generates evolutionary pressure: insects develop resistance, and bacteria must adapt their toxins accordingly.
The responsible enzyme machinery, the GxpS synthetase, consists of five modules – making it manageable enough to reconstruct its full evolutionary history. Through phylogenetic reconstructions, the researchers discovered a surprisingly large diversity of previously unknown GXP variants. In total, they identified 49 distinct compounds, including shortened cyclic tetrapeptides and an extended hexapeptide.
Particularly striking were patterns of convergent evolution: several bacterial lineages that had evolved independently produced identical or very similar peptide structures by separate routes. To understand the order in which the different GameXPeptide variants arose, the research group reconstructed the last common ancestor of GxpS synthetase. To do so, they had to restore around 15,000 base pairs of genetic material, assemble them into a functional gene, and finally bring it to life in Escherichia coli. This allowed them to determine what the ancestral GxpS looked like – and in what order today's variants evolved from it.
"It felt like playing five-dimensional chess," says Georg Hochberg, describing the methodological challenge. "Frequent recombination obscures the evolutionary history of these enzymes – each recombination event overwrites a piece of their past. But it is precisely this complexity that makes the system so fascinating."
The detailed teamwork paid off: the analysis revealed that natural selection alone does not explain the diversity of GameXPeptides. At least three forces act together: natural selection, genetic drift, and a "recombination bias."
To distinguish natural selection from random evolution, the researchers used a statistical comparison. They identified recombinations that do not structurally alter the peptide – for example, when two modules with the same specificity are swapped for one another. These so-called homotypic recombinations served as a benchmark for what would be expected under neutral evolution. Heterotypic recombinations – those that actually change the amino acid sequence – hit significantly more often exactly those regions of the enzyme that are critical for substrate specificity. This was a clear sign of positive selection. Experiments with four different insect species confirmed it: different GXP variants kill different insects with different efficiency.
At the same time, the researchers found evidence for genetic drift – changes without any discernible selective advantage. This suggests that a substantial portion of GXP diversity arose through the random fixation of neutral variants. This effect is favored by the bacteria's life cycle: each nematode is initially colonized by only a small number of bacterial cells. These bottlenecks limit population size and amplify the role of chance in evolution.
The researchers also discovered that not every recombination is equally likely. In fact, recombinations occur preferentially between modules that are similar at the sequence level. Since similar modules are found mainly within the same gene cluster, NRPS in practice incorporate primarily those amino acids already represented in the cluster – even though they could theoretically draw on a repertoire of up to 300 different amino acid variants.
This bias is self-reinforcing and leads to an evolutionary tunnel vision: each recombination within the cluster makes the participating modules even more similar to one another, making future recombinations between the same partners increasingly likely. This mechanism may explain why the same variants so often evolve convergently – not because they are particularly effective, but because they are particularly easy to produce. "I was completely surprised by how many variants of GameXPeptides even exist," says first author Sebastian Kaiser. "And I think we are just at the beginning of understanding the evolution of natural products. There is so much more to discover."
Indeed, the implications of this work reach beyond basic research. Many well-known antibiotics are also nonribosomal peptides. If their evolutionary histories likewise conceal numerous, as-yet undiscovered recombination events, new bioactive variants may be hidden there – ones that overcome existing resistances or possess novel mechanisms of action. The methods developed in this study offer for the first time a systematic way to uncover this hidden diversity.
"Our results show that the conceptual framework we developed for the GameXPeptides can serve as a guide to decipher the evolutionary forces acting on other natural product families, such as the polyketide synthases, which like NRPS produce many clinically relevant natural products," says Helge Bode. "If we better understand the mechanisms underlying bacterial natural product diversity, it may become possible to discover new antibiotics that overcome existing resistances."
Nature Ecology & Evolution
Experimental study
Cells
Drift, selection and recombination bias drive the evolution of a non-ribosomal peptide.
21-Oct-2026