Streptomyces bacteria produce tunicamycin, a compound that shows promise as an antibiotic because it can inhibit cell-wall synthesis in other bacteria. Unfortunately, tunicamycin isn’t suitable for clinical use because it inhibits the human enzyme DPAGT1, which plays an important role in protein production. To find safe alternatives, researchers synthesized tunicamycin derivatives and studied the structure and function of DPAGT1. A team led by Benjamin G. Davis at the University of Oxford added lipid chains of varying lengths to two positions on the core tunicamycin scaffold. To understand the functional significance of these changes, Elisabeth P. Carpenter’s team at Oxford solved structures of the human enzyme alone as well as bound to its natural substrate and tunicamycin (Cell 2018, DOI: 10.1016/j.cell.2018.10.037). The human protein has a narrow tunnel that can bind only one lipid chain on tunicamycin, while the bacterial target has a wide groove that can accommodate two. The most potent analogs, called TUN-8,8 (shown, lipid chains in red) and TUN-9,9, had lipid chains eight and nine carbons long, respectively. In mouse studies, the analogs safely and effectively treated tuberculosis.