 Good. Thank you very much, Carl. And I'd like to thank Lita and the organizers of this meeting for the opportunity to present. It's just terrific to see the progress in this field. Going on day three of the conference, many very important topics have been covered. And so my overview is going to be very brief. We can think about a series of biological questions about the human microbiome. What is the identity of the microbes that populate their host? What are they doing? How is the host responding to them? What are the forces that maintain equilibrium among the populations? And what are the unique characteristics of each individual? And we can add a sixth question. How can we manipulate it? Or in other words, another way to ask it is how is the microbiome being manipulated every day? Now our views of the microbiome are based on this schematic of the interaction between a co-evolved colonizing microbe in the host. And the basic idea is that co-evolution has selected for organisms that know how to signal the host and receive host signals back. And the concordance of the phylogenies between animals and their microbiota indicate a very high degree of co-evolution. And mathematical modeling has shown this kind of equilibrium to be robust and resilient. And I want to talk about what happens when it's perturbed, especially by antibiotics. And the focus of my talk is going to be about early life. And I'm going to show this slide again as many people have shown from the great studies from Jeff Gordon's lab with Rob Knight and Maria Gloria Dominguez, where they asked when does the adult gut microbiome become established? This axis is age. Here's unifract distance from adult. Babies are born with a very different microbiota from adult. And then gradually they become more and more adult-like. It's the same in all three populations. And the main point is that it becomes adult-like at the age of three. So this is the period in which the microbiome is the most dynamic. And it's also the period in which babies are developing. And this, I think, is a very important window for us to consider. Now I'd like to focus on obesity, as was my charge, and remind you of obesity trends in U.S. adults from the CDC, a marker of changing physiology. In 1989, no state had more than 14 percent of adults obese. By 2010, no state had less than 20 percent obesity. The main point is that it's happening everywhere. And this is only 21 years. It's happened very, very rapidly. Now I want to focus on antibiotics and remind you that when we look at the top eight prescriptions in U.S. children in 2010, five of them are antibiotics, accounting for more than 40 million courses. Recently, the CDC published data looking at outpatient antibiotic use in 2010. And the bottom line is that they found that there were 258 million courses of antibiotics prescribed in the United States, or 833 per thousand population. The highest rates were in young children. The rates went down through middle life and then went up again. And considering that antibiotic use has been relatively stable or possibly declining, we can do a back of the envelope calculation and estimate that the average child in the United States has received nearly three courses of antibiotics in the first two years of life, more than 10 courses by age 10 and more than 17 courses by the age of 20. This seems a lot, but in fact it's consistent with studies in other developed countries. Now the CDC data also included this map of the geography of antibiotic usage. And the national average was 833. Northeast and Midwest are not much different. West is a lot lower and South is about 50 percent higher than the West. And we don't think that bacterial diseases are so much different in the South and the West. This must represent practice or culture. Now here we compare the two geographic maps, obesity and antibiotic use. And it looks like there are a lot of parallels. The observation is striking. But remembering what Jesse Goodman said, remember this is observational data. And there are many things that could affect these observations. So let's see if we can go further with experiments. And so we've taken advantage of the fact that farmers have found more than 60 years ago that they can feed low doses of antibiotics to livestock, what are called subtherapeutic antibiotic treatments. And this will promote livestock growth. They found that many different agents do it regardless of class, target or spectrum, antibacterials but not antifungals or antivirals. And importantly they found the earlier they start the effect on growth rate and feed efficiency. And that's the conversion of food calories into body mass. So we decided to try to model this experimentally in mice in which we would feed mice subtherapeutic antibiotics in their drinking water or not, look at phenotypes and look at the microbiome. As was discussed this morning by Jonathan, our first work in this field was published last year. Il Sung Cho led the effort. In his initial studies we didn't find any difference in growth rate but we found that the mice put on more fat. And we could find this consistently across the antibiotics. We found that there were big changes in intermediate metabolism as shown by these microarrays of the liver, looking at differentially regulated lipogenesis genes, up regulation in stat of genes and central carbohydrate metabolism, lipid production and lipid transport out to the periphery. This slide summarizes some of the major findings in that study. We've hypothesized that the antibiotics mediate differential selection of colonic microbiota. We have found changed composition of the microbiota, altered representation of short chain fatty acid synthesis genes, increased short chain fatty acid synthesis, induction of lipogenic genes in the liver, incorporation of lipids into adipose tissue and increased adiposity. Now we've progressed since there and I want to highlight the work of Laurie Cox who's here at the meeting. She's presenting poster number six. Laurie's student to graduate with her PhD and has done really a remarkable series of studies. First I'll point to the Dynastat study where we've looked at the dynamics of stat phenotype development in which mice received either 30 weeks of penicillin or no antibiotics. We've looked at male and female mice. Here we're looking at body fat percent. The black line is the control. We see a significant increase in body fat. In males it starts at 16 weeks and it continues for life. There's the expected decrease in PYY and the expected increase in leptin. In females we see a similar phenotype. It starts a little later. We don't see a significant PYY or leptin effect. We studied the microbiome in the male mice. This slide shows a left-see analysis of four week fecal pellets from the males. We looked at four weeks because it is before the development of the phenotype. Left-see showed that there were significant associations of firmacutes and TM7 with control and bactiroidetes, proteobacteria and tenorocutes with stat. It also identified particular taxa that were associated with control such as SFB and alobaculum, a member of the erycephalotrichiae, and also taxa associated with stat such as ocelospira and ruminococcus. We also looked at the data using Spearman correlations. In the Spearman correlation, negative associations are shown in red or orange. Positive associations are blue. The closer you get to align, the stronger the association. In this slide we're looking at those taxa that were associated with body fat percentage at the end of the study looking at fecal samples at four weeks, 16 weeks, 26 weeks, cecum and ilium. These are the taxa. The asterisks indicate those taxa that are significantly different. We're only including those that show significance at any of these time points. We find, again, certain taxa that are associated inversely with body fat such as SFB, alobaculum, lactobacillus, and organisms positively associated at this four-week time with body fat such as ocelospira and ruminococcus. When we looked at leptin, we have a smaller number of taxa, but the same organisms and the same relationships are shown as for body fat. When we looked at PYY, there were more taxa, and the associations are in the opposite direction just as we would expect. This gives us a group of candidate organisms that are associated with the later development of phenotype from an earlier time period. Now, Lori went on to do an experiment that we call fat stat, where she looked at the relationship between antibiotics and putting mice on a high fat diet. So at 17 weeks, she put half the mice on a high fat diet. We looked at males and females, total mass, lean mass, and fat mass. Here's the control group, normal child, no antibiotics, putting them on antibiotics, increased mass, putting them on high fat, increased mass more, high fat plus antibiotics, even more. Antibiotics increased lean mass. If we look at fat mass, we see the high fat increases mass dramatically, antibiotics, even more. Female mice, similar findings, a little different. If we just look at the fat mass on high fat diet, these mice have five grams of body fat, high fat, plus stat antibiotics, more than 10 grams. They've more than doubled their fat. We see an additive or synergistic effect of high fat diet and antibiotics. So next we asked, those studies involved lifelong antibiotics. Next we asked, is a shorter course of antibiotics significant for the growth promotion effect? So in addition to lifelong antibiotics, Lori studied eight weeks or four weeks of antibiotics. Here are the phenotypes in terms of total mass, lean mass, and fat mass. Black lines are the control. All the stat groups showed increase in all three of these markers. And there's no difference between four weeks and 28 weeks. So four weeks was sufficient for the phenotype. Shingo Yamanishi, a visiting postdoc from Japan, did microarrays of the liver of the terminal ilium. Here are keg pathways involved in immunity. T cell activation, immune system process, toloic receptor signaling pathways, down regulation and staff. Many important genes in innate immunity were particularly interested in SAA from other studies, but a lot of decreases in toll receptors. Here are QPCRs that Shingo did focusing on T helper cells, looking at the transcription factors for Th1, Th2, Treg, and Th17. Just to summarize a lot of data, we've been most impressed by the consistent decreases in Th17 here looking at the transcription factor in male and females. We've also seen in cytokines that are Th17, the same phenomenon. Jackie Leung and Ping Lokes lab has looked at Th17 by focytometry in the small intestine and the large intestine. Again, looking at cells marked by IL17 and IL22, decrease in STAD in both small and large intestine. A consistent decrease in IL17. So what about the microbiota? This is the first of several slides looking at the fecal community structure from MySeq analysis, more than 2 million sequences in total in the experiment. We're looking at three weeks. And at three weeks, there are only two groups, control and antibiotics. All of these mice are receiving antibiotics. Here's control and black, STAD in orange, they're beginning to split, they're significantly different. Now at eight weeks, there are three groups, no antibiotics, continued antibiotics, and four weeks of antibiotics and then stopped. Black control and orange STAD are separated more, further antibiotic effect, but the blue group has now reverted, the community structure has reverted to normal. So the effect on community structure is transient, yet the phenotype is permanent. And by 28 weeks, all the antibiotic groups have returned to normal. So Lori did the experiment where she asked whether the growth phenotype can be transferred by the microbiota. She harvested sequel contents from control females and STAD females at 18 weeks. Gavaz germ-free mice that we've received for teconic and now conventionalize these mice. The experimental design is again shown here. The control microbiota or the microbiota from antibiotic receiving mice was given to germ-free recipients. They received no further antibiotics and they were followed for the next five weeks. Here are the metabolic phenotypes. Black line is control, body mass increased in STAD, no change in lean, increase in fat. So the microbiota is sufficient to transfer the phenotype. When we look at IL-17 gene transcription, looking at ROR gamma and two IL-17 cytokines, the donors, even three mice in each group shows a significant effect. We see that we've transferred the phenotype with ROR gamma. We see a trend here, but it's not significant. Now when we look at the alpha diversity of the transferred communities, they're similar at the time of transfer. Then there's a big bottleneck for both of them. The control group is gradually coming back, but the STAT transfers never recover their full alpha diversity. So there's a decrease in biodiversity. What about the beta diversity? Laurie looked at the inocula of the control and the STAT. These are the inocula. Now one day later, they've both moved out across the PC1 really in parallel directions. Here's nine days after transfer, the controls are coming back toward where the inoculum was. The STAT are coming back more slowly. It seems two groups at 34 days near the end of the experiment. These guys are almost back. These guys are lagging and they're more separated. So what about the fecal microbiota after transfer? Each of these boxes looks at the taxa that are present in the control recipients and the STAT recipients over time at five different taxonomic levels. We can see lots of differences. We are currently analyzing this, but we're finding some of the same organisms that were present in some of the earlier experiments. Up to this point, it's been the farm model of low dose antibiotics, but human children get high doses of antibiotics for their ear infections and their throat infections. So we developed a second model called PAT. This was worked by Yale Noble, a student. Yale's hypothesis was that a series of short therapeutic dose pulses of antibiotics administered early in life will sufficiently change the gut microbiome to alter body composition. The doses were devised so that they would mimic the pharmaconetics of antibiotics in children. The experiments involved female pups who did not receive antibiotics or three pulses of amoxicillin, the most widely used antibiotic in childhood, or thylacin, the macrolide or a mixture of the two. The first pulse was at day ten of life. So they got their antibiotic through their mother's milk. They were weaned, two more pulses, and then everyone was started on a high fat diet. Here we're looking at the phenotypes. Here are the three pulses. When we look at scale weight, the black line is the control. There's an early increase in scale weight in the PAT mice. When we look at growth rates, we see a significant increase in the growth rate of the thylacin mice. In amoxicillin, it's not clear cut. When we use DEXA, we see this increase in total mass, lean mass, mimicking the farm, fat, intermediate, especially in the thylacin group. We also see increases in bone density, bone mineral content, bone area, lifelong effects in amoxicillin. We've seen this in other studies. TH17, again, decreases control, amoxicillin, thylacin, same for all these three markers of TH17. Now we work with George Weinstock and Erica Sodigren to look at the taxa that are present. Here we're looking at alpha diversity across the experiment. Here are the species richness in the mothers, about 800 OTUs. The mothers never received any antibiotics. Now after one antibiotic dose, here are the control mice. They have the same richness as the moms. They didn't receive antibiotics. But the mice that got amoxicillin or thylacin have reduced diversity. And this continues through the end of the doses into the high-fat diet to the end of the experiment. They never recover their diversity. So this is a lifelong effect from these three pulses of antibiotics. Alex Alexenko and Informatocyst in the group did a Kalinsky analysis, found that four clusters best describe the data. Here's a PCOA showing the four clusters. This is the mom cluster. This is the high-fat cluster. These are the antibiotic cluster. And here we can look at a timeline of the clustering. Here's the time axis from the first dose to the end of the experiment. Each line is one mouse. In the control mice, they all begin life with the mom cluster type. And then with the high-fat diet, there's selection for the high-fat OTUs. Amoxicillin looks similar to control. It is perturbed, but not terribly so. And there's even a late perturbation. In contrast, thylacin is perturbed from the very beginning. It's a massive perturbation. And finally, the strong selective effects of high-fat can overcome this effect that ended here. So there's a very long-term effect by the thylacin. So we can think of a series of models of early life development. There are stem cell populations of interest to us or mesenchymal-like stem cells that have to decide how many times to go around and should they become fat, bone, or muscle. There's also immunological development, lots of focus on diet and calories. We think that the microbiome and the cells that they're signaling through are important in this regard to provide the context for these developmental steps. This is the ancient development. Here's the modern development with antibiotics and other interventions changing the composition of the microbiota, changing the composition of the responding cells, and thus creating a new context of signals for these important developmental processes. As Jonathan Braun pointed out, we consider that there are different kinds of obesity. There's physiologically induced obesity, gene-induced obesity such as elliptin knockouts, diet-induced obesity. They have both microbe independent and microbe dependent effects. But we're interested in a primary microbe-induced obesity where the disruption of the microbial community is primary, such as through antibiotics or C-section, or both combined. So I want to recognize many of the people who've done this work. I've mentioned them as we've gone along, funding from a number of sources, especially from the NIH. And I want to finish just with some comments about the overall context. If we apply economic theory to infectious disease, we remember that nothing is free, sometimes costs are low. Low cost does not mean free and low costs may be cumulative. And so I'd like to raise the question whether there's collateral damage to children from the antibacterial activities that we're doing. And using the mouse models affects maybe on metabolic and immunologic processes as well as others. In the old days, moms, kids got their trans-, their microbes from their moms starting at birth, vaginal, cutaneous, mammary, and oral. That's how it was. But things are changing. Moms aren't the way they used to be. They're subject to antiseptics and antibiotics and diets and other factors. Babies aren't the same. 33% are born by C-section today in the United States. Bottle feeding, early life antibiotics. This is consistent with an idea that we've had for many years of a disappearing microbiota due to changes in human ecology, changing the composition of the microbiota, changing human physiology. And what we believe is most important is the loss of ancestral bacteria that usually are acquired early in life. Because it affects a developmentally critical stage. A few years ago, Stan Falco and I postulated that there's an effect of maternal status on the resident microbiota of the next generation in which it is stepping down at each generation because of the bottleneck at birth. And is that, in fact, is that true if we go back to the Yatsunenko paper and look at the total diversity in the adult fecal microbiota, comparing people in the US with Amerindians and Malawians, people in the US have less diversity, 15 to 25% less diversity today. And Maria Gloria's more recent studies are consistent with these observations. What do we need to do? This is a two-year to 30-year plan. We need to do research about antibiotic consequences in both humans and in model systems. If these observations are correct, we need to educate the public in the profession about risks of early life antibiotics. We need, as a research community, to develop narrow spectrum treatments, which mean better diagnostics and specific agents. Develop plans for remediation, identifying and replacing acutely lost actors. The new probiotics, the real ones, enhance depleted actors with prebiotics. Try to reverse by archiving vanishing organisms and replacing them and monitor to see if it's working. Thank you for your attention. Thank you, Marty. I think we have time for one or two questions here in the middle. I guess the use of low-dose antibiotics in the in the agriculture industry must have been studied. And so I don't suppose they have been giving antibiotics to get high fat levels. They do it presumably to get lots of lean mass. Is that correct? They're, in general, it's done to increase body mass, lean most of body mass is lean, but there are effects on fat mass as well, and there are effects on bone. The animals are bigger. But it is, I mean, is a predominant increase in the fat or in the lean, and is it also associated with a different diet? Because in your experiments you're using high fat diets in combination with antibiotics. Whereas what's happening when you're trying, when they're using in agriculture? So in these experiments we use both normal chow and high fat, and we, in the one experiment I showed you, we compared the two. And again, these are model systems that we have done to test the hypothesis that antibiotics will change a number of steps in development. They don't have to be exact analogies, and in fact, none of these models are exactly like humans, but we are building a body of data that are indicating that early life antibiotics are changing developmental phenotypes in terms of muscle, and we're studying muscle. Actually, I have a graduate student, Chase Demeris, is working on that. We're studying fat, we're studying bone. I was curious if you could comment on the slide, you showed this intriguing slide, the prevalence of obesity and antibiotic use in the US. Were the temporal profiles of these prevalence changes similar? So I mean, the answer is another way, was the rate of increase of antibiotic use paralleled by the rate of increase in obesity, which would obviously strengthen potential causative relationship. Yeah, so we don't know the answer to that question. This was a new study, I've been thinking about this for many years, and then this paper from the CDC just appeared two months ago in the New England Journal, and when I saw the map, I thought it really is quite striking. I'm not sure what to do with it. It could be reverse causation. We're not sure, but it certainly is a striking association. And one final question in the back. Thank you very much. It was just an excellent talk. In addition to the antimicrobial effects of antibiotics, antibiotics are also known to have direct effects on mitochondrial function. Some are mitochondrial toxins, and some affect carnitine absorption. Again, you've elegantly showed transfer of some of the phenotypes, but those pharmacologic or toxological effects of these antibiotics could cause abnormalities and fat metabolism that also could contribute to obesity. Are there any germ-free studies with antibiotics looking at any effects on mitochondrial function or fat metabolism? I'm not aware of such studies, but the agricultural scientists have done studies in germ-free chickens, and they have shown that antibiotics in germ-free chickens do not have a weight phenotype. So you need the microbiota to get the effect that this was studies done decades ago, and of course, as you point out, the transfer shows that we can get the effects irrespective of the presence of antibiotics. Could there be independent effects from antibiotics? Very possibly. Thank you. Thank you, Marty.