Mycobacteria, I think I can say without fear of contradiction, are a real pain in the behind, scientifically speaking. Don’t get me wrong: bacteria in general are no fun to develop drugs against. Whatever fun was available in that area drained away by the early 1970s at the latest as the major classes of antibiotics were discovered and as the bacteria themselves set about busily developing resistance to them. The rate of drug discovery in the area has famously slowed down, with most of the advances being improvements on existing scaffolds. Meanwhile, the bacterial resistance problem has not slowed down appreciably at all, and the intersection of these two trends has been the subject of a lot of worry and a lot of warnings over the years.
But mycobacteria in particular are a tough problem to crack. Gram-positive bacteria are generally the easiest to kill with drug therapies because of their single-membrane structures, although please note that this “ease” is on a relative scale. Most readers will have heard of MRSA (“mer-sa” in the lingo), which is the source of some extremely unwelcome infections that are very hard to treat and in which the underlying Staphylococcus aureus organism is Gram-positive. These strains are able to resist a broad spectrum of beta-lactam-based antibiotics, although there are (for now) some other types that are still useful for treatment (linezolid, clindamycin, vancomycin and others).
Gram-negative bacteria, though, have a double-membrane structure with a thin peptidoglycan cell wall in between, and that’s a more formidable barrier to getting antibiotics inside them at all. These membranes are well stocked with efflux-pump proteins, and those are a big part of the problem. Many are the compounds that can kill off efflux-pump-crippled engineered bacteria in the lab, but unfortunately none of us are going to be infected with any of those. And the great majority of such compounds, when exposed to real Gram-negative pathogens, barely even ruffle their bacteria hair. Finding a really active compound against these is a real accomplishment.
And so is finding one against the Mycobacteria. Those guys have an arrangement all their own: a cell membrane, on top of which is a periplasmic space capped by a layer of peptidoglycan gunk, and on top of that is a layer of arabinoglycan (a unique feature). On top of that is yet another unique feature, though, a double layer of mycolic-acid-based gorp with various surface lipids and proteins imbedded in it. This is a really tough gauntlet to run for a small-molecule antibiotic, and it’s fortunately that most Mycobacteria are not pathogenic. The bad part is that the ones that are cause tuberculosis and leprosy, with the former being present in maybe a third of the entire world population (!) In many of these people the M. tuberculosis infection is just sitting around latent in the lung tissue, growing very slowly. This growth rate is seen in culture, too - even if you have the right medium for them (and many of the common ones don’t work), it can take weeks to grow visible waxy colonies of the things. As a human pathogen (and we’re the only animal that’s a reservoir for them), the bacteria are extremely resistant to being killed by macophages because of that coating.
There are antibiotics that work, although of course there are now plenty of resistant strains out there, particularly in garden spots like the Russian prison system. Resistance is showing up and increasing in countries around the world, though, and finding new antibiotics is a real world health priority. I thought this paper made an interesting contribution to that. The authors are doing wide-ranging structural studies on model peptides to see what factors are more likely to get these compounds past those thick multi-level defenses.
A first takeaway is that peptides themselves can actually get through at all - the prevailing idea has been that you need smaller and more hydrophobic molecules to have a real chance. A second lesson is that the best modification to make is cyclization, although you do need to pay particular attention to the overall ring size and the structures that you’re using to close the rings. But this seems to be the category that showed the most notable success, and the differences between the cyclized compounds and their linear counterparts is often impressive. The second-best strategy is N-methylation of the peptide, but that has a lot of variability in it, for reasons that are not really clear (or at least aren’t to me). The paper demonstrates improvement on an antibiotic candidate by adopting these features, and also shows that removing them from an existing compound (griselimycin) significantly weakens its activity.
We need plenty of these sorts of insights to deal with drug-resistant tuberculosis, because the only reason that it’s not an even bigger problem is that slow growth rate mentioned above and thus its relatively slow spread through the human population. But that tends to bring on complacency, because it’s not just ripping through the population in real time like a new respiratory virus (you remember those, right?) The last thing we need is another plague, even a slow one.