Like Scotch Eggs

In the 1880s, the Danish bacteriologist Hans Gram was working in the morgue of the Berlin city hospital, trying to identify bacteria in sections of lung tissue under the microscope. But there was so much blood that the bacteria were ‘impossible to see’. He used a dye – gentian violet – to stain the whole sample, then rinsed it with alcohol to wash out the purple colour. The bacteria appeared ‘an intense blue (often almost black)’ while the human cells were unstained.

The dye molecules had flooded through the system but stuck only to the bacteria. (This sort of selective binding prompted Paul Ehrlich to conceive of a ‘magic bullet’ against syphilis.) Gram noticed, however, that some bacteria didn’t take up the violet colour. The stain offered a way of classifying bacteria into two broad types: those that took up the dye (Gram-positive) and those that didn’t (Gram-negative).

A full explanation for the difference was published in the 1960s. Bacteria are built like scotch eggs. They all have an inner membrane made of fatty acids, surrounded by a firmer wall of a substance known as peptidoglycan. The peptidoglycan is what keeps the cell’s shape; it’s also what binds to gentian violet. In Gram-positive bacteria, the peptidoglycan wall is thick and exposed, easily soaking up the dye. In Gram-negative bacteria, the peptidoglycan wall is much thinner and shielded from the outside world by another membrane, which the stain doesn’t penetrate.

Getting something into a Gram-negative bacterium isn’t easy. Though the sum total of its boundary layers is thinner than for a Gram-positive, the outer and inner membranes are separated by a no-man’s land known as the periplasmic space (which contains the thinner peptidoglycan wall). The arrangement makes the cell much less permeable, apart from small water channels. Antibiotics that can get into and kill Gram-negative bacteria are therefore often small molecules, around a hundredth the size of a typical protein.

One example is meropenem, which has been used against Gram-negative bacteria for nearly thirty years. Unfortunately many hospital strains of A. baumannii are now resistant to it, making the species a popular target for recent antibiotic development. Historically, a lot of this development has involved taking small molecules as a starting point. But there are alternatives. Antibiotics can also target the ways that Gram-negative bacteria move molecules between the inner and outer membrane. You then don’t have to get your molecule all the way inside the cell – just far enough into no-man’s land to disrupt the transport mechanisms that shuttle things back and forth.

Transport mechanisms don’t only bring molecules in – they take them out, too. The fatty molecule that the outer membrane is made of, lipid polysaccharide (LPS), has to be made inside the cell. To maintain the bacterium’s outer membrane, molecules of LPS have to be transported across no-man’s land. To do this, A. baumannii constructs a protein bridge: a ‘transporter’ made of seven proteins bound together that spans from the inner to the outer membrane.

Last week, researchers from Roche reported a new antibiotic that targets this transporter. They’ve called it zosurabalpin (I don’t know why). Their collaborators also demonstrated how zosurabalpin works: it binds and traps LPS in place in the transporter, ‘stalling’ the machine. A. baumannii can live without its outer membrane, but it can’t survive the build-up of all the excess LPS inside the cell. Its own cellular processes destroy it in a glut of fat and sugar.

It's unlikely Roche knew much about this mechanism beforehand. As the commentary in Nature explains, the research started from a class of molecules known as macrocyclic peptides. These are chunkier than most antibiotics (though one of the few antibiotics left for multidrug-resistant A. baumannii is colistin, another cyclic peptide, discovered over seventy years ago). After screening 45,000 molecules, Roche found a new one that protected mice from infection with A. baumannii. But when injected into rats – physiologically closer to humans – the promising molecule led to ‘a rapid decrease … in lipid parameters’. Bubbles made of lipids and proteins were precipitating out of the plasma, thickening the blood and killing the rats: injecting it into humans was also likely to end badly. Tweaking the molecular structure to reduce the amount of plasma precipitation led to zosurabalpin.

The story shows why it isn’t enough simply to find a molecule that kills a bacterium and run with it. Drug development typically involves a lot of tweaking, to balance a molecule’s positive and negative properties. And it isn’t quick. It’s been known for about a year that Roche had a new antibiotic in development, known as RG6006 and in a phase-I clinical trial since December 2022. Roche first filed a US patent for macrocyclic peptides against A. baumannii in September 2016, more than seven years ago – and the drug is probably still years away from being available for doctors, if it ever makes it.

The research is a convincing demonstration of a new mechanism against Gram-negative bacteria, and zosurabalpin seems highly specific to A. baumannii, which is particularly welcome. It’s also a reminder that 150 years after Gram learned how to distinguish two broad classes of bacteria, there’s still a lot left to discover about the ways their cellular architecture operates. With luck, the design of future antibiotics might make more of that knowledge.