 Okay, well, thanks for bearing with me for a minute. Like I used a couple, hopefully not too much of my time. Okay, so it's a real pleasure to be here today, and I'm looking forward to sharing with you something we've been thinking about for the past few years in the role that microbes in the gut play in drug metabolism. And so, if you go to your favorite pharmacology textbook, you can usually find a brief, although interesting mention of microbial drug metabolism. In, for example, one that's widely used by pharmacologists in the industry, there's a short paragraph that says that the microflora, or microbiota, in the gastrointestinal tract can metabolize a wide range of drugs, which can reduce the amount available for absorption. And then they go through and mention a lot of the known reactions that can occur. In Goodman and Gilman's, which is the pharmacologic basics of therapeutics, sort of like the Bible for pharmacologists, there's actually just one sentence, but it says drugs in the gastrointestinal tract can be metabolized by enzymes of the intestinal flora before they gain access to general circulation. So the point being that we've known about these interactions between microbes and drugs for a long time. It's known to affect a variety of different compounds over 40 known drugs. But we know very little today about exactly how these reactions are taking place. And so at the suggestion of Lita, I'm gonna spend the beginning of the talk just giving you a few examples of what work has been done by other members of the field and then I'll end with a more in-depth story which we've recently published. And so you can sort of think about these interactions between gut microbes and drugs as breaking down into sort of two different types. So you think about drugs in the simplest case an orally administered compound going through the gastrointestinal tract. There's a variety of ways in which this drug prior to absorption can be directly modified by gut microbes and the case of prodrugs that are activated, compounds that are inactivated, converted to toxic byproducts and of course antibiotics that are intended to have an effect on pathogens but can of course have many unintended consequences for the standard residents of the gut microbiota. In addition to that, there's some very interesting indirect interactions where microbial colonization can actually result in certain microbial metabolites which go into circulation and interfere with drug metabolism in many sites throughout the body as well as effects of colonization on the expression of genes by the host. And so I'll just give you a few examples of some of the characterized cases where this has been shown to be the case. For example, in drug activation there's a very sort of famous example which is sulfa salazine. This is a drug that's used for inflammatory bowel disease and it's very conveniently activated by members of the gut microbiota and this actually proves to be very useful for the drug because you want it to locally reduce inflammation at the site of the problem. And so you administer this prodrug which has an azide bond shown by the arrow. Azo reductase is expressed in the microbiome cleave this drug in half and you get the active metabolite 5-ASA and not so conveniently sulfapuridine which is thought to contribute to the side effect of the drug. This is just a phylogenetic tree that we put together recently which sort of makes the point that this is an interesting enzyme and the fact that it's not just restricted to one particular type of bacteria. You can find enzymes that have actually been experimentally validated in a wide range of different bacterial phylin including the permacutes actinobacteria bacteroidides and proteobacteria, so it's one fuzobacterium. And this is sort of a neat assay because these same bonds are found in tattoo ink and so if you have undergrads or somebody that you want to search for azo reductases, all you have to do is put tattoo ink in your agar and you can find these enzymes. But what's kind of, so these are widely distributed and there have been papers where people actually exploit these enzymes to locally deliver other drugs in addition to sulfacylzene. And yet even though we know this enzyme and this has been known for a while, we still know very little about whether or not the inter-individual variations in the efficacy or toxicity of these compounds is mediated by variations in the gut microbiome. That remains to be explored. There's another really famous example which I think came up a little bit yesterday and this is work done by Matthew Redinbo at UNC and they were interested in the drug arenotican or CPT-11. And this drug has a ridiculously complicated route through the body. So what happens, this is used for cancer. It's administered by IV, makes it into the blood in an inactive form. It's then activated by carboxyl esterases to SM38. It then gets inactivated in the liver by being glucuronidated. And then this inactive form is released back into the intestine. And that's where it gets sort of interesting for the microbiologists where these enzymes, bacterial beta glucuronidases can actually reactivate the drug in a site that you don't want it. And that contributes to the dose limiting side effect of the drug or diarrhea. And so Matt decided to use what's known about the coli version of beta glucuronidase to screen a library of different small molecules and they identified four different inhibitors and this is just data from their paper in 2010 where they looked at beta glucuronidase activity on the left in E. coli. And you can see without the inhibitor, the enzyme is fine under anaerobic and aerobic conditions. And then when they add each of the inhibitors, it blocks the activity. And this is also the case for two other distantly related isolates, back to Rodie's vulgaris and clostridium. They then went on to show that this actually helps in mice, which is very exciting. So what they did was they took animals, fed them or gave them the vehicle, the inhibitor alone, CPT-11 or CPT-11 plus the inhibitor. And as you can see, the percent of mice with diarrhea was very high when given arena t-can or CPT-11 but the inhibitor successfully blocked that. And that was also the case when they just looked at histology and you can see this is the quantification histological score in this down here is just example pictures from the paper where you can see CPT-11 causes a major disruption to the inhibitor restores that. Okay, so the next type of interaction is the sort of more nebulous one where certain microbial metabolites can make it into circulation and black host drug metabolism. And there's sort of an interesting example of this which came from Jeremy Nicholson's group. And they were looking at the drug acetaminophen. So this is one of the most widely used drugs, probably many of you in the audience are taking it right now. And it's interesting because it gets detoxified by the body in two different ways. It can be sulfated or it can be glucoronidated. And they noticed that if you do metabolic profiling of individuals after being administered acetaminophen, they vary in the relative proportion of these two different forms of acetaminophen. And that was associated with the abundance of a microbial metabolite called P-cresol. So this is just the negative association where if you have a high level of P-cresol, you have a low level of the sulfated form of acetaminophen. And so this is potentially suggesting that what happens is that P-cresol which also gets sulfated may be titrating away the enzymes in the host that are responsible for this major pathway of drug detoxification in the host. Okay. We still don't really know the consequences of this, but there's a potential that the toxicity of acetaminophen could be linked to this particular metabolite. Okay. The other sort of interesting example is that it's been known for some time that if you compare germ-free animals to colonized animals, there are differences in gene expression that contribute to altered host drug metabolism. So this is a study from Sven Peterson's group and what they did was just looked at gene expression again and standard SPF mice compared to germ-free animals. And they interestingly found the down regulations or decreased levels of expression of multiple cytochromes in the liver. Then they went on to show this actually has a functional consequence. So if you dose these mice with barbiturates, the germ-free animals recover much faster from the anesthesia. All right. So the final example I'll share with you is one that we've been very interested in for the past couple of years. And this is another example that's been known for quite some time. So actually 30 years ago, a lot of the seminal work linking they got microbiota to de Jackson metabolism was done. But we still, at the time of when we started the study knew very little about how exactly mechanistically this was happening. So de Jackson is what's called the cardiac glycoside. It has these three sugar rings here, this large set of rings here. And you can see a single double bond in the lactone ring at the top right. And I should mention that all of this is work being done by Henry Hazer in the lab. He's a really talented postdoc and has a poster here. So de Jackson is a natural isolate from Fox Club plants. It's widely used still today for heart failure as well as a regular heartbeat. And the mechanism is that it inhibits a sodium potassium ATPase in the heart. It's been of interest and really widely studied for a long time due to the fact that it has a very narrow therapeutic range. And so you have to very carefully dose the drug to avoid a lot of the very toxic side effects while still providing enough to have the intended efficacy. And there's a range of side effects. I think just as a side note, it's sort of interesting to think about one of the common side effects which is blurry yellow vision. And this is inconvenient for most people but in one sort of famous example, which is Vincent Van Gogh, this is thought to be responsible for his art style. And you may wonder, how do we know that? And the reason is that Van Gogh painted a lot of self portraits. And so you can see him just sitting here staring at his Fox Club plant. All right, so moving on. So Egerthalolenta is the single gut actinobacterium that's commonly found in a lot of different people. And it was shown in 1983 to be the only organism that was culturable that was capable of inactivating the drug deoxan. And this is really based on some early work that had shown that there is a downstream metabolite of deoxan called dihydrodeoxan, which is not capable of binding the target sodium potassium ATPase. So it's an inactive form of the drug. It's found in a subset of patients. And if you administer antibiotics at the same time as deoxan, you elevate the amount of intact drug that makes it into serum. And so there was an argument to be made that gut microbes were involved. And Lindenbaum then went and proved that elenta was capable of doing this in the lab. And one of the interesting things about elenta is that unlike many of the organisms in the gut that are looking for carbohydrates, elenta really thrives on arginine. So it grows primarily using arginine as a source of carbon, nitrogen, and energy. And the interesting thing is that as you would expect as you add more and more arginine to the media, you can increase the overall growth of elenta. But at the same time, high levels of arginine inhibit the reduction of deoxan or inactivation. All right, so that was sort of the initial framework where we started. And we wanted to see whether or not understanding something about the mechanism of deoxan and inactivation would actually help us predict what's going on in the more complex situation in a complex human gut microbiota. And potentially design a way to test in animals whether or not we could actually prevent this from happening. And so we started looking at elenta in isolation, just grown in the lab in low and high arginine in the presence or absence of deoxan. And what I'm showing you are sort of RNA-seq results where just the height of the bars just represents the number of reeds that are aligned to a given place in the elenta genome. And as we look at, so in the absence of the drug, there's essentially no expression of these two genes. And when we add deoxan, they come on incredibly high. And interestingly, the two genes that are found in this operon are predicted bacterial cytochromes. And so they're potentially capable of accepting deoxan as an alternative electron acceptor. So this we got very excited about and we sort of tentatively called them the cardiac glycoside reductase operon. But we wanted to sort of look at this in a little bit more detail. And so the one thing that we noticed based on the RNA-seq and confirmed by quantitative PCR was that at high levels of arginine, the operon is actually suppressed. And so we see the highest level of expression when you add deoxan in low arginine and the levels are significantly lower in high arginine. So that fits with the idea that potentially high arginine is down-regulating this operon, reducing the amount of drug metabolism. The next thing Henry did was to just do an initial sort of structure function analysis. We wanted to know whether or not this was the only cardiac glycoside that could induce the operon or if this was a conserved feature of many related drugs. And so we took a small panel of deoxan and three other cardiac glycosides that vary in relatively subtle ways. You can see digitoxin here is just missing one of the oxygens. Ditoxigenin has the three sugars cleaved off and then wavane is actually somewhat distinct. And when you quantify the amount of induction of this operon, you find that all four of them are capable of inducing expression. Whereas if you chemically synthesize the reduced form of each compound, it's not able to trigger the expression of the operon. Okay. Unfortunately in Elinta, we don't have any bacterial genetics yet. So we didn't have a way to actually knock out the operon and prove to ourselves that this was actually responsible for digitoxin reduction. And so we took sort of a comparative genomic approach. And thanks to the HMP and other sort of large-scale genome sequencing projects, we actually had three closely related strains of Elinta that had all been fully sequenced. So we started by just comparing them in the lab to see whether or not they could all reduce the drug. And we found that only the type strain could inactivate didoxin. And these other two FAA strains were not capable of doing this. And when we compared the whole genomes of these three strains, we found that the non-reducing strains did not have the CGR operon. And we're also missing other interesting genes in particularly these two operons for potentially for the import of sugar in small metabolites. All right. So now that we were somewhat convinced that this operon was responsible for drug metabolism, we wanted to say, to ask the question of whether or not that would allow us to predict the efficiency of drug metabolism in a complex human microbial community. And so at this point, we're not looking at cardiac patients. Instead, we designed an assay where we collected fecal samples from 20 unrelated individuals. We took these samples and incubated them with the drug in the lab. And then we measured the amount of didoxin that was converted to dihydrodoxin. And then we compared that to the relative abundance of the CGR operon. And so what we found is that you could group people into sort of two main categories. There's individuals that are relatively poor at reducing the drug. And then there are microbial communities that are relatively efficient at this. Most of them actually reducing all of the drug that we exposed them to. And they were, as we expected, the level of the CGR operon was significantly higher in the high-reducing group of people. Okay, so next we turn to these neurobiotic animal models that have been mentioned in many of the talks today. To try to see whether or not, based on our knowledge of eulenta metabolism, we could actually prevent this from happening in vivo. And so the rationale was relatively simple. So we know that if you add arginine to the media in the lab, it blocks didoxin reduction. So the sort of obvious question is to ask, if you add more arginine in a mouse, do you also stop reduction? And so we did an initial experiment where we have germ-free animals that are all colonized with the type strain of eulenta. And then we transferred them from a standard chow into two different diets that were relatively extreme. So we have a 0% protein and 20% protein. And the prediction, again, is that we should see higher levels of didoxin on the 20% protein diet. And indeed, that's what we saw. So when we quantify didoxin by ELISA and the serum as well as the urine, we see dramatically higher levels on the high protein diet. We wanted to just make sure that this wasn't a non-specific effect of colonization or the two different diets. And so we designed a second experiment where germ-free animals are either colonized with the type strain or with one of these non-reducing FAA strains and placed on the same two diets. And the results made sense. So we saw, again, the higher levels of serum and urine didoxin and mice colonized with the type strain, but not in mice colonized with the non-reducing strain. And in the case of serum didoxin, we saw significantly lower levels comparing the two groups on the 0% protein diet. All right, so this is all sort of, we hope, building towards a model of didoxin metabolism that takes into account differences in colonization between individuals in the sense where some people, we think, may be colonized predominantly with strains that lack the CGR operon and are incapable of metabolizing the drug. And so the level of didoxin that's administered remains intact. Whereas other individuals that carry strains like the elenta type strain that have this operon are capable of converting the drug to dehydrogenoxin, yet at the same time, based on the amount of perhaps dietary protein and arginine that can inhibit this reaction. And so if you want more information, I would encourage you to go to poster P14 and also the paper which came out last week. Okay, so with that, I'll turn to some needs, gaps and challenges. I think my need is relatively simple and I think this has been covered by most of the talks today. So I don't have to go into more detail, but we've over the past, I guess nine years or so, been interested in combining these metagenomic studies of humans and animals with detailed and controlled studies of notabotic animals. And I think that's a useful sort of paradigm to go back and forth to develop useful hypotheses that are grounded based on human data that have the ability to test them in animals. And building on what Sarkis said, I think one of the real challenges for us is that in addition to genetic drift, we deal with the fact that there's no repository where you can obtain germ-free animals. And so we're really dependent on really one company, Tecanic, which provides wild type germ-free animals to researchers for a pretty hefty cost. And a few labs that sort of generously distribute these animals around the world. So there's unlike cultured bacterial isolates or normal mice, if one researcher re-derives a mouse of a certain genotype germ-free, there's no sort of central repository where you can put that mouse where other people will be able to share that after you publish your data. And that would be incredibly useful. All right, so these are really scientific gaps. I think the questions that we're very interested in are identifying the key bacterial or microbial Paxa and metabolic pathways that are involved in zeta-biotic metabolism. We still know very little about how microbial communities adapt to these very toxic compounds. This is obviously fairly well understood for antibiotics, but we know incredibly little about how other drugs that we take are influencing microbial community structure and function. And we also would like to get at estimating the relative contributions of host, microbial and environmental factors to pharmacokinetics and dynamics. And so if I can end with just sort of a grand challenge for pharmacology is that, you know, we obviously know a tremendous amount of pharmacology. Anyone that's ever looked at this book will be instantly odd by the amount that has been learned over the past hundred to so years about individual variation and how variations in the human genome influence metabolism, how exactly drugs traffic throughout the body and how they affect disease. But hopefully in the future, we may be able to move towards a metagenomic basis of therapeutic where we, you know, not only account for variations in the human genome, but we also have a deep understanding of how microbial colonization influences this process. All right, so with that, I just want to thank the Center for Systems Biology. It's been a wonderful place to be for the last three and a half years. And of course, the great group in the lab, Henry, who is here and presenting his poster, you know, really was responsible for all of the work into Jackson. I told you about our funding sources, including the NIH and just a brief plug for a Keystone Symposium, which is coming up in April of next year, which will go into drug metabolism, antibiotic resistance, but also, I think just represent a broader array of speakers that will address the importance of microbial metabolism for a lot of aspects of health and disease. And so hopefully I'll see a lot of you there. Thanks so much. So the doc is open for questions. Peter, not so much a question, but just addressing your need. There is an Office of Director, used to be NCRR comparative medicine funding to our Notabotic Facility at University of North Carolina that is designed as a relatively small source for people to use Notabotic mice. Now the problem I agree is central repository were limited by space, but certainly are trying to do frozen embryos. And that would be one cost saving way of having a central repository that could be brought out on demand. Not immediately like a bacteria, but still could be a, but I would encourage people that need Notabotic resources to contact me. Yeah, yeah, and I would agree. And thanks for all the help with the experiments I showed today. So we have no more questions. I wonder if you could speculate a little bit on what are the natural substrates for some of these microbial enzymes and how can you fit that into thinking about how these activities are being selected for in particular microbiomes? Yeah, that's a great question. And that was, I had sort of a laundry list of, obviously there's many more questions on the three I told you about, but that's something we're very curious about. And in the case of de Jackson, based on sort of predictions of protein structure for these two members of the operon and what's known based on sequence semology, I guess our working hypothesis is that one of these two proteins is potentially capable of accepting fumarate as its intended electron acceptor. And that based on sort of structural similarities in the unsaturated ring of de Jackson that may be able to sort of sneak in as an alternative compound. But we have not, it's all speculation at this point I think, but it's sort of tempting to, at least in this case, imagine that perhaps these enzymes are intending to target a more standard substrate that's found in the body that are sort of cross reacting with these drugs. So I'm just curious where de Jackson absorbs into the body and where he lentus colonizes. So those two weren't really put together well. No, that's a great question and something that we are planning to look at in more detail. We've looked at in mono associated mice, you can obviously detect a lentus throughout the length of the gut. But we still, based on sort of the classical view of the pharmacokinetics of de Jackson, the idea is that it's primarily absorbed in the small intestine, but we just haven't looked at that in enough detail yet in mice. Okay, Peter, thank you very much. And our final talk for this session is from Wendy Garrett, also from Harvard University, who's going to talk about the regulation of colonic T-cell homeostasis by microbial metabolites.