Non-Antibiotic Drugs Affect our Gut Bacteria


The microbiome is a totality of microbes, so bacteria, viruses, fungi, etc that live on us or in us. And it’s a huge
amount. There are more bacterial cells than human cells for example. And particularly
the gut microbiome it’s maybe 1 1/2 kilograms – more than our brain weighs. So it’s a lot of importance in terms of numbers here. But in terms of functionality, it’s very important because it can help producing vitamins that we cannot produce; it can help in degrading plants that we cannot degrade on our own to get actually nutrition out of them; and it protects also against infections. Many studies report these associations. That means if you have a disease as a person you have changed or an altered microbiome. So what we found accidentally in the past for type 2 diabetes where such an
association has been described, that the first line drug for type 2 diabetes, which is metformin, basically caused the whole effect. So the consumption of the drug changed the microbiome and not the disease itself, or the disease only to a certain degree or a smaller degree. So that of course triggers a question if it’s one drug can it not be me more drugs? And because if it’s the drugs that change it, there’s a question are there side effects associated with drug intake, caused by the microbiome, because the drugs modify the microbiome? Or is the microbiome even involved in a mode of action, so meaning how the drugs work on us. So that was one of the starting points where we met with colleagues here at EMBL, Kiran Patil, Nassos Typas, both had an interest in the microbiome from different perspectives, and whether we can team up and develop some in vitro platform – so platforms in the laboratory that mimic some of the things we see in these big association studies. The main aim of our study was to map the effects of 1,200 drugs on 38 human gut commensals, that are representative for the human microbiome. And this comes with two major challenges. The first one was that we had to handle a lot of different drugs and a lot of different bacteria and this is just possible with high throughput approaches. And the second is that most of these bacteria just grow under anaerobic conditions, and this required us to work in an
anaerobic chamber. Working under anaerobic conditions actually means that you work in the absence of oxygen, so all the buffers and all the media and all the plasticware that you are using have to sit in an anaerobic chamber for several days to be sure that they’re free of oxygen Up until now, previous studies have only
tested the effect of single drugs against gut bacteria, or tested a large number of drugs against pathogens. In this study we wanted to test a large
number of drugs against many bacteria in our gut to see how drug treatment effects
the bacteria. A typical experiment consists of three steps: The first step is to dilute the chemical compound library, in our case that’s the Prestwick library from the stock concentration to the screening concentration. We then rearrange the library to ensure we have sufficient control wells on each plate, containing the drug solvent DMSO. The process is automated and performed by liquid-handling robots. After rearrangement, one library aliquot
comprises 14 96-well microtiter plates All drug plates have to be pre-reduced overnight in the anaerobic chamber before we can start our experiment. For each run, which takes a whole day, we screen the effects of all 1,200 drugs on one gut commensal. Inside the anaerobic chamber, we grow the commensal strains starting from glycerol stocks. Depending on the strain, it might take up
to two days to see bacterial growth, i.e. when the medium turns turbid. From the overnight culture we dilute the strain in fresh medium so that the starting optical density is 0.01. We inoculate all 14 drug plates with
the diluted strain by using a smaller liquid-handling robot that fits inside
the chamber. Next we seal the plate with a breathable membrane. Performing these steps can be tricky as we need to wear the thick chamber gloves. This requires a bit of practice and patience at the beginning. Now we can start reading optical densities using a plate reader which we also do inside the chamber. The plate reader can only fit one plate at a time, but luckily we have a microplate
stacker to exchange the plates for us. The plate stacker is surrounded by an incubator, custom made by the EMBL workshop, so that all plates are incubated at 37 degrees. With the automation of the process it is possible to record growth curves up to 24 hours. After the experiment, I export the growth
curves and send them to Michael. The results of the experiment produce around 150,000 growth curves, which is of course far too many to analyse them individually. So my role was to analyse them on a computer. As all the bacteria behave differently we first have to define what normal growth is. And to do is we essentially take the integral under the growth curve which gives us a convenient single number that we can use to quantify growth for each bacteria. In most cases, the compounds do not affect the bacteria and therefore I can identify
reference compounds which have unperturbed growth to see how variable
the growth is for each replicate and for each strain. Once I have this, I basically get a distribution of growth and then I can check for any individual drug treatment, how different is it from this distribution of normal growth. So in some cases nothing grows and then it’s of course easy to say that this is a hit, but there’s also a grey area in between where we need to apply statistical methods to then really define if this is a hit or not. As expected we find that a lot of antibiotics in our screen had an effect on human gut commensals. But what was more surprising is that a quarter of the human targeted drugs, so with no direct target in bacteria, actually had a big effect on human gut commensals. We actually see that the drugs that have an effect in vitro, produce more antibiotic-like side effects in patients. We also compared our in vitro results
against results from patients which were monitored when they were taking certain
drugs. And again what we see is a concordance, so that if you see growth perturbations in vitro they also match growth perturbations in vivo. We find a general trend that if a strain is more resistant to antibiotics, it’s also more resistant against human targeted drugs. And that indirectly implies the risk of
acquiring an antibiotic resistance by being exposed to a human targeted drug. I think there are three major implications: If we’re in an era that we don’t have
antibiotics anymore and we don’t have antimicrobials that would act in clinics,
maybe some of those drugs that have unexpected activities could be at some
point be repurposed to do exactly that. And since those are approved drugs this can move faster in terms of their drug development. The second part is that we want
to understand if the effect they have on humans is part of the side effects or
part of the mode of action of the drug on humans, because this creates new strategies on how to go further. And on the third one, we want to see how
much they pose a risk for antibiotic resistance. I think they’re both risks and opportunities within this phenomenon. The risks are obvious – so if you get
resistance to antibiotics the same way you get resistance to human target drugs,
then by developing resistance to one you become automatically resistant to
the other one also. But there is also the scenario that some of the resistant mechanisms for antibiotics might get you more sensitive to human targeted drugs. And in that scenario you now have a tool which you can use now to drive off antibiotic resistance. So we’d like to study in the future whether those results are applicable broader to everybody or they’re individual specific. Our microbiomes are quite specific in terms of composition. So we do carry largely the the same number and amount of species. But the individuals of the species, that are called bacterial strains vary a lot between individuals. So the question is how those bacterial strains that are different between individuals will dictate whether the results of the drugs are the same across individuals. If they’re not that means are there opportunities for individual therapies. If they are it means that there’s opportunities for more broad therapies.

Leave a Reply

Your email address will not be published. Required fields are marked *