Edit: deleted incorrect information about b-lactamase.
Nature asked me for $$$ to read the article, so ... from your description, that's leaving out the hard, fast test of just culturing several billion of those, adding a lethal concentration and seeing if there are any survivors. Which is how I gather one found spontaneous transport mutations, at least circa 1977. The state of the art has likely improved, and these guys used novel microbiological methods to find the bacteria that produce teixobactin in the first place.
ADDED: thanks to betatim's link to the full text, I've skimmed it and read the discussion, and teixobactin sound quite promising. They haven't found any protein to which it binds, and they think it binds to an "Achilles's heel" in the outer cell wall. The method described to generate resistance was their most extreme attempt, so I assume they tried the fast way, and I can see why it didn't work.
Ecologically, they believe there's little gene (e.g. plasmid) transfer between these soil bacteria and human pathogens, and the "30 year" bit came from experience with vancomycin, to which it has a lot of similarity. And they've done lots of tests for human toxicity and effectiveness in mammals. It's still early in the process, but a degree of fuss is warranted, especially due to their discovery method.
>We were unable to obtain mutants of S. aureus or M. tuberculosis resistant to teixobactin even when plating on media with a low dose (4 X MIC [minimal inhibitory concentration]) of the compound. Serial passage of S.aureus in the presence of sub-MIC levels of teixobactin over a period of 27 days failed to produce resistant mutants as well (Fig. 2d, Supplementary Discussion)
From the Supplementary information:
>Cells were added to teixobactin present at 0.25xMIC, 0.5xMIC,1xMIC,2xMIC and 224xMIC. At 24 hour intervals, the cultures were checked for growth. Cultures from the second highest concentrations that allowed growth (OD600≥2) were diluted 1:100 into fresh media containing 0.25xMIC, 0.5xMIC,1xMIC,2xMIC and 4xMIC of teixobactin. This serial passaging was repeated daily for 30 days. Any cultures that grew at higher than the MIC levels were passaged on drug free MHA plates and the MIC was then determined by broth microdilution. No resistant mutants were obtained. This experiment was repeated, and produced the same negative result. In order to maximize the chance of obtaining a resistant mutant, we performed an additional experiment with very small incremental increases in the drug concentration. Cells were added to a series of tubes with small differences in the concentration of teixobactin (0.06xMIC, 0.25xMIC, 0.5xMIC, 0.75xMIC,1xMIC, 1.25xMIC, 1.5xMIC, and 2xMIC). At 24 hour intervals, cultures from the highest concentration that allowed growth to a minimum OD600 of 0.2 were diluted 1:100 into fresh medium containing 0.06xMIC, 0.25xMIC, 0.5xMIC, 0.75xMIC,351xMIC, 1.25xMIC, 1.5xMIC, and 2xMIC. This passaging was repeated for 27 days. Cultures that grew at levels higher than the MIC were passaged on drug free MHA plates, and the MIC was determined. For teixobactin, there were no mutants with an MIC greater than the parent S. aureus ATCC 29213.
Moreover, there's a strong biochemical argument against resistance: the region that the antibiotic targets is highly conserved, which means that it's probably functionally necessary. The bug is therefore unlikely to evolve away from the threat. This is important, because it's that most likely avenue for antibiotic resistance, energetically speaking: it doesn't cost anything for an organism to mutate a weakly conserved gene, so they do it once, and pass it on to every subsequent generation with no penalty.
The antibiotic is just a peptide, so in theory you could see some sort of specialized peptidase evolve from an existing gene...but the problem is that the bug would have to then carry around that gene and express it constantly (or even less likely: evolve a sensing system that allows for selective expression). It's a highly unlikely thing, and virtually impossible to maintain over multiple generations without constant selective pressure. Bacteria do not like to hold on to genes that they don't need.
Nothing is impossible over evolutionary time, of course, but the researchers don't seem to be overstating their case here.
It's a highly unlikely thing, and virtually impossible to maintain over multiple generations without constant selective pressure. Bacteria are quickly out-competed by genetic variants with fewer expressing genes that they don't need.
BTW: do you have any links/data re the relative 'cost' of expressing unneeded genes?
Empirically, everyone who has done lab work with bacteria knows that they'll quickly (i.e. within a few generations) kick out plasmids unless there's selective pressure to keep them. Genes are more likely to stick around than plasmids (since they're directly integrated in the organism's genome, and there's less of an energetic cost to accumulating DNA sequence), but stable genome transformation is slow/unlikely, and bacteria will still find a way to disable un-needed genes in a relatively short period of evolutionary time.
Beyond that, even if they are out competed, that's not sufficient. They need to be out competed to extinction.
There's also regulatory genes that may be present, there are 6 known varieties of the whole gene cassette, with 5 further subtypes based on these regulators, so in the absence of β-Lactam antibiotics it sounds like a lot of them will be fit.
Nasty.
However, that does not exclude the possibility of the bacterium acquiring a plasmid or phage carrying an enzyme that inactives teixobactin. This particular mode of acquiring antibiotic resistance is quite common.
Nevertheless, this finding does seem to be quite a big deal. A novel broad-spectum antibiotic where spontaneous resistance is unlikely is a pretty powerful addition to the pharmacopeia.
Right, which is why I said the bit about bacteria kicking out genes that they don't need. Bacteria will "eject" a plasmid within a generation or two if they no longer need the gene(s) in question. Replicating a big gob of easily-ejectable DNA is not something bacteria do unless there's a good reason to do it.
Basically, for plasmid resistance to propagate, you need to have constant exposure to the antibiotic in question, or the resistance gene needs to be stably integrated into the bacterial genome. The former doesn't happen until an antibiotic is in extremely wide use, and the latter is one more rare step on top of an already unlikely chain of events.
It's not impossible, it's just unlikely.
Plasmids are indeed a source of resistance, but plasmids just contain genetic code. They are considered mobile since they are easily transferable between bacteria (including different species). So for a plasmid to encode resistance to this new antibiotic, it would need to contain the code for a protein which disrupts the action for the antibiotic. For example, the plasmid would encode for an enzyme that digests the antibiotic at a faster rate than it can work, or bind to the target of the antibiotic with a higher affinity.
So maybe what you were getting at is that a plasmid exists out in the world that encodes the protein for resistance, it just didn't happen to exist in the researchers niche world, or doesn't rely on a random mutation to become a resistance product.
Per cowsandmilk the senior author of the publication "is one of the world's foremost experts on antibiotic resistance". The paper discusses how there seems to be little gene transfer between the "biome" of soil bacteria from which this and vancomycin come and the relevant pathogens. They specifically cite that it took 30 years for any resistance to develop to vancomycin.
Whereas I'd add β-Lactam antibiotics seem to be pretty common (they are derived from at least 5 different organisms that I just counted in Wikipedia), and β-Lactamases are all too common; per Wikipedia https://en.wikipedia.org/wiki/Beta-lactamase the first was discovered in 1940 before penicillin was in clinical use.