When a patient takes an oral dose of a drug, you can picture it as being a little like one of the those Japanese pachinko machines, where a ball drops into a maze of complicated bouncing paths on its way to the bottom. There are a lot of things that happen once a pill hits the gut, the broad features of which process are described by the acronym ADME (absorption, distribution, metabolism, and excretion). The metabolism part, for example: there are many chances for one enzyme or another to alter the structure of your active drug. There are digestive fluids in the stomach and intestine, enzymes in the gut wall, enzymes in the blood plasma once a compound is absorbed through it, and a ton of enzymes in the liver, which is of course the first stop for anything absorbed out of the intestine. A drug’s structure can get hydrolyzed, oxidized, dealkylated, hydroxylated, conjugated, reduced, and more, and most pharmaceuticals produce a spray of such metabolites that can be picked up in the blood, urine, and so on. Some of these will also be active against your drug’s original target, albeit with totally different half-lives and distribution. Some of them will totally lose that original activity, but will turn out to be active against other things entirely. Good luck tracking those down, and welcome to ADME!
It’s long been recognized that the gut bacteria can be another source of metabolic degradation, but this new paper from a team at Yale suggests that it’s even more wide-ranging than we’ve realized (commentary at Nature here). The authors did a pretty extensive survey: 271 known drugs against 76 different sorts of human microbiota. This is definitely the most systematic look at the issue yet, since many of the known examples are one-offs discovered in the clinic. It was done by a pooling strategy – the 271 drugs were spread out among 21 pools, such that each one was represented in four different groups, but shares any pool with another given drug at most twice. That gives you 3,840 interactions, measured in quadruplicate at 12-hour anaerobic incubation points.
For two-thirds of the drugs tested, the amount of the parent compound was reduced by at least 20% by at least one bacterial strain. So bacterial metabolism is not some odd exception – and given how variable the gut microbiome can be from person to person, you can see yet another source of the noise in human clinical data and medical practice. As was already well appreciated, nitro and azo functional groups were particularly labile (for many varieties of bacteria), but quite a few other things showed up as well. Bacteroidetes species, for example, were particularly good at amide and ester hydrolysis, but the whole range of chemical changes mentioned above were present in various cases. The paper goes on to a number of experiments with gnotobiotic (bacteria-free) mice, showing that colonization of their intestines with various bacterial strains can reproduce the in vitro effects on metabolite profiles, and the team also started to untangle some of the genetic signatures of the various metabolizing enzymes across the different species.
This paper last year showed that a number of (non-antibiotic) drugs can have effects on the gut microbiota in turn, and it’s also been shown that treatment with metformin in Type II diabetes patients alters their gut bacterial population, and in a way that actually potentiate’s the drug’s effects. So all this comes together into a coherent (if wildly complicated) story. The drugs are affecting the bacteria, the bacteria are affecting the drugs, and both of those are affecting the patients and their responses. We have quite a ways to go!