The sweet invasion: sugar molecule disrupts bacterial cell wall

With antibiotic resistance in mind, the search for new ways to neutralise bacteria continues. An interesting target in Gram-negative bacteria is the cell wall, which consists of two membranes with a peptidoglycan layer in between. On the outside of the outer membrane sits a further lipopolysaccharide (LPS) layer (the red filaments in the photograph).

Disrupt the cell wall from within, and the bacterium becomes more susceptible to antibiotics. Zeynep Su Ziylan, Marthe Walvoort and colleagues at the University of Groningen have achieved this using a derivative of a remarkable sugar molecule, 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo). Well — almost achieved it.

A beast

‘The first paper on the use of the Kdo derivative 8-azido-Kdo appeared ten years ago already,’ says Marthe Walvoort, associate professor of chemical glycobiology. Bring that derivative into contact with a bacterium and the bacterium metabolically incorporates it into the LPS. ‘It’s a very peculiar sugar with eight carbon atoms. In Zeynep’s PhD project, she created variations on Kdo with the aim of disrupting the assembly of LPS in this way.’ The hypothesis was that it would yield truncated LPS, undermining the bacterium’s structural integrity, and hopefully leading to a permeable membrane.

But the synthesis ‘is an absolute beast, just dreadful,’ laughs Walvoort. ‘It took much longer because all sorts of isomers were forming and purification was difficult.’ Nevertheless, they succeeded, and from the small synthesised library Ziylan had assembled, 5-epi-Kdo-8-N₃ proved the most successful.

Text continues below the image

This molecule was indeed incorporated by the model bacterium (E. coli) into the lipopolysaccharide structure. ‘Just not at the crucial position! It turns out that E. coli contains two Kdo units — one at the critical site where the LPS is further elaborated, and one adjacent to it,’ explains Walvoort. ‘Our molecule was most likely incorporated at position two, meaning it had no impact on the sugars that still need to be added further along the LPS.’ The initial goal was therefore not achieved, but there is now a greater understanding of the labelling process and why the bacterium does incorporate this derivative whilst rejecting the others (those lacking the hydroxy group at position 5).

Proving the point

Once the manuscript had been submitted and was going through the review process, something in an NMR spectrum turned out not to add up. Walvoort happened to be in Vienna at a conference, where she bumped into Paul Kosma — ‘one of the founding fathers of sugar chemistry’. ‘He was a tremendous help; he discovered that in one of the NMR spectra we had a completely different molecule from what we thought.’

One of the labelled Kdo variants had undergone recombination. ‘The reviewer had picked up that the assignment was incorrect, and Paul Kosma had noticed the same thing by coincidence,’ recounts Walvoort. ‘He then helped to assign the correct structure. During the review phase I genuinely felt supported by the community — it’s gratifying when you sense that people want to help improve the paper rather than feeling as though you have to prove yourself.’

Walvoort also finds it ‘brilliant that you can still collaborate at that stage of publication’. ‘It taught me that you can simply approach people who have long since retired and ask for their help. I should have reached out much earlier, but better a goal in injury time than no score at all, as I see it.’

Not such a good model

The interdisciplinary nature of this work is important to Walvoort. ‘We received faculty funding specifically for people from two different institutes — that’s what makes this kind of blue-sky research possible.’ And for the future too. ‘You create fertile ground for future collaborations that way. There are many things you cannot measure but which are enormously valuable, and this is a wonderful example of that.’

This paper also marks the end of Ziylan’s PhD, and no follow-on funding is immediately available. ‘I’d very much like to continue,’ says Walvoort. ‘But there’s no one to carry the work forward.’ But if she could continue? ‘Then I’d label Haemophilus influenzae bacteria with our molecule. That bacterium has only one site where Kdo is present in the LPS, so it simply must produce truncated LPS.’ With a laugh: ‘In hindsight, E. coli wasn’t really such a good model bacterium after all.’

‘That’s what I find so fascinating about this field,’ concludes Walvoort. ‘The tension between wanting your sugar molecules to resemble the original sugars as closely as possible so that they are taken up and incorporated, whilst simultaneously wanting your molecule to exert a disruptive effect.’

Ziylan, Z. et al. (2025) JACS Au, DOI: 10.1021/jacsau.5c00338