Never say never. Screening natural product extracts for new antibiotics has been a diminishing-returns exercise for quite a while now, which is too bad, since basically every single important antibiotic class came via that route originally. Bacteria compete with each other (as do plenty of other organisms), and seeing what they’ve come up with after a billion years or so of infighting has been very productive.
But when you run such screens now, you tend to find things that we’ve already found – either the exact compounds, or as close as makes no difference. (That’s as opposed to random screening of compound collections, when you tend to find those and a bunch of things that are nonselectively toxic, which does you no good, either). Digging your way through the pile of unusable hits is a challenge, and there’s no guarantee that there will be anything left after you do.
So I’m particularly surprised and impressed by this new paper (press release here, C&E News article here), which reports the discovery of a new variety of antibiotic entirely, and by good old soil-sample screening, at that. (Update: here’s the full paper). It’s a collaboration between several research groups in Milan and a group at the Waksman Institute at Rutgers, whose namesake would be glad to hear the news. The compound is pseudouridimycin, a nucleoside analog that inhibits bacterial RNA polymerase. That’s the same target as rifampicin, but this molecule binds (according to an X-ray structure) to a slightly different region of the protein, and is in fact additive with it when co-administered. Finding a compound that is such a good inhibitor of bacterial RNAP without activity at the human enzyme is another welcome surprise.
It’s an active-site mimic of UTP, uridine triphosphate, which is not an easy feat. Phosphates are, of course, extremely important in biochemistry, but making small molecules that bind to their recognition sites is difficult. Look, for example, at the wide world of kinase inhibitors – out of all the compounds described in that area, only a tiny fraction seem to avail themselves of the phosphate-binding region of the proteins. The same goes for many other classes (phosphatases, naturally, as well as helicase inhibitors, etc.) It’s the concentration of negative charges, surely. The protein binding sites are (naturally enough) very polar in the opposite direction, and don’t seem to care much about the kinds of molecules that normally populate screening collections. But at the same time, they’re quite picky and directional, so just throwing a bunch of polar stuff at them indiscriminately doesn’t get you anywhere, either.
But pseudouridimycin manages it. Coming off a 6-amino pseudouridine core is a glutamine-glycine chain, with the glycine being N-hydroxylated at one end and functionalized with an N-formamidine at the other. Binding to the nucleoside triphosphate site makes it harder for the enzyme to mutate its way out of trouble, and indeed, development of resistance in culture seems about ten times slower for this compound than is does for rifampicin, which is good news. It’s active against drug-resistant stains of mycobacteria, which is something that we could very much use, considering the foul strains of tuberculosis that are at large.
This looks like a promising drug candidate all by itself, and I hope that discovery of this structural class will lead to interesting analogs as well. Past that, I wonder if that whole modified peptide side chain can be repurposed to hit triphosphate binding sites in general – I don’t think that will be easy, but considering how hard it is to make headway in that area, I think it’s probably worth a shot. . .