 I'd like to thank Leader Proctor and Owen for inviting me here and I'm very glad to be in the basic biology session because I'm going to be telling you some basic biology. I was actually hoping that David Relman would talk a bit more about babies. We should have maybe coordinated on that because I sort of expected him to. So I'll touch on microbiome, colonization and assembly in babies but then move on to a few other aspects of basic biology that we've been working on that I think you might find interesting. So there's rapidly developing data coming out about what sort of environment we actually develop in when we're in the womb. I think we've assumed for a long time that it's been sterile but studies that are in progress and coming out are indicating that it might be slightly more microbially sort of influenced that we might previously have thought but it's still a relatively protected space where we develop until we come to the outside where we then rapidly colonized by the microbiota of the biosphere which of course are very diverse and vast and so one of the first things that happens even before you take your first breath I think is that you start to be colonized by microbes and we need microbes for proper developments and this has come from years of research that goes back more than 100 years with not abiotic animals. Animals that have been raised under germ-free conditions differ quite substantially from animals that are raised normally in the normal microbial world in that they have a different kind of immune system that's not fully developed if you will because that requires microbial influence they have different physiologies they have different behaviors there's lists and lists and lists of ways in which animals that are raised under extremely artificial conditions without microbes can differ from animals that are raised under with the microbial influence. So we've done some work looking at how this colonization happens this is a case study of a single individual over two and a half years which is quite a long time for current studies although more of these are coming now so that so that we can sort of compare the patterns that we saw in this particular individual and what you're looking at is bar graphs of taxa this is very simple way of looking at the data each one of these bars represents a day over two and a half years and there's certain events that associate with changes in the distribution of these taxes such as a fever here the introduction of table foods here which brought about a bloom in Bacteriades and all the while there was a very slow increase over this time period in in the diversity that was there so the richness showed a very slow incline over time increase over time and then various events bring about these big changes in the relative abundances of different taxa and the way this particular individual responded to these events is something and here's an antibiotic event for example is something that we can characterize but whether whether this these responses are typical of all individuals and whether these we can start to to go from these associations that we see in a case study to general principles is something that's going to require following many many babies and this is something for instance that David relevance group has been doing and so we can we can make some logical sense as to why things like Bacteriades increase when those table foods introduced based on what we know about their genomes and how they have the genes necessary to break down these foods and so changing the the milieu here in the gut is definitely going to bring about a bloom in the kinds of organisms that can that that use those substrates but but the kinds of microbes that are present in one individual to another is something that we're still trying to understand where they come from how much regional differences might affect the kinds of microbes are colonized with food yes we have some idea of how how food influences these things but there's still a lot of basic biology though that still needs to to be understood about this in this particular individual breaking it down and looking at the kinds of sort of consortia if you will this is again the same bar graph showing you how the phyla change but within those phyla we see discrete consortia that are members of those phyla and we see them change very gradually over time so these are the early days we see this sort of group and then later on this group and later on this group these early taxa came back after antibiotics we can see that here and then a whole new consortium replaced them afterwards and these these we know that these associations that we see like this these groups and non random but again it's going to take profiling many different types of babies over time for instance to to understand how if there's some general principles we can we can derive from these kinds of data or if or if different perturbances are acted differently in different individuals when we think about how the microbiota develop over an entire lifetime this is something that we can only piece together from looking at current studies of of individuals that have these different ages because at this time there hasn't been a single individual it's been followed over the course of a lifetime like this and so we have data on on very young people and we have data on adults most of the data is on adults and then we have some data on elderly people but but again no one person has been studied for this long and so although we we can see some sort of general themes like for instance it looks like there is differences in very early colonization based on how babies are born and it looks like stability sort of comes in maybe an early childhood and adult microbiota may be even more stable than that these are all these principles are basically derived from stitching together very different studies on very different cohorts and so going forward I think it's going to be very important to have large long-term cohorts so that we can really test whether what we think is happening based on looking at current studies is actually holds up when you when you look at people over the long term and one of the questions that the people and sort of the probiotics industry and so on are particularly interested in is whether or not what happens early on affects microbiome later in life for instance so if you have certain certain microbiota early here are they going to affect what comes later do we have distinct successional stages and do the do the what you have in these later stages depend on what comes before and that's not something I think we have the answer to at this point so what does impact the microbiome later in life or during adulthood in general this this is a very fuzzy diagram it was deliberately fuzzy that Pete Turnbaugh and colleagues wrote back in 2007 asking what kinds of influences are there on the microbiome both what's shared and what's not shared between people and you can think of this as lineages presence or the functional genes that are present and you can come up with things that you you know or think might influence this like the lifestyle and the immune system and the environment but the reason all these arrows are sort of the same size and don't intersect is because we don't have a good idea of which are the strongest how they might interact and so on and one of the areas that we have the least understanding of at this time is host genotype and I'm going to tell you now about a study that we've been doing to address how host genotype might influence the composition of the gut microbiota so there's a need for genetic studies in humans uh there've been some studies in mice using qtl mapping that reveal associations between certain taxa abundances for instance and certain loci in the genome these loci can be quite big and contain sometimes hundreds of genes and so actually getting in finer details to which genes are actually driving the associations um can is this sort of the next step for for animal studies as well um there's been studies with candidate genes in humans so for instance if you have a particular allele that's associated with a particular disease like nod 2 and crones you can select for people with different versions of that and look at microbiota and that's one example but you have to know what genes you're looking for to do this uh but there haven't been any published studies yet on genome-wide associations in humans there has been a work going back more than a decade now looking at twins and and twins are particularly sort of intriguing uh this is not actually a mirror this is a human mirror this is an improv group that got monozygotic twins and got them to line up down subway cars just to see people's reactions and so this really drives home the point that monozygotic twins are really remarkably similar phenotypically and the question is are they got microbiota is also more similar than than you might expect for say fraternal twins which might be these two although they're not and so again this goes back uh people have started to use twins uh this is an example of a paper from 2001 using DGGE fingerprinting and based on the relatedness of the people where one was the monozygotic twins they saw a trend towards more similar microbiotas in the monozygotic twins and then Pete again when he was working in uh Jeff Gordon's lab published this study so these are unifract distances this is based on 16s RNA pyro sequencing uh when the bar is smaller it means the microbiota is more similar and so now you're looking at twin pairs mz versus dz and there was a hint of greater similarities for mz's although probably because of the relatively low number of twins used here it didn't actually come out as significant so we've we've decided to look at twins and we're looking at twins that are genotypes so we can we can actually do a GWAS study as well as uh as use some of the twin based techniques to look at this and we're working with Tim Spector who runs the King's College of London uh twin registry there and it has about 15 000 twins in it that they've been working with for many years and they're part of many genome uh space studies so in this particular study we have access to about 6000 genotype twins and we're collecting stool from as many as we can so basically we're asking them to send it to us and uh as you know that's not always something that people want to do but so far we've had about a thousand send-in samples we have 250 dz twin pairs about 160 mz pairs unrelated individuals and these people are mostly middle age they're average age of 64 and they're mostly women and that's not because the twins in the cohort are mainly women it's well they are actually it's just that women seem to be more willing to be involved in this kind of study for whatever reason um there's several ways that we can we can analyze data like this so we've started with a 16s survey for and we have data for a thousand samples now and one thing that we can do is when we have abundances of different taxa we can look within twin pairs how well the abundances correlate and if you you might expect a tighter correlation for twins that are mz uh for for the abundances of taxa compared to dz twins if there's a genotype effect and we can calculate these correlation coefficients and we can do that at every single node in the phylogenetic tree and then take a look at the distribution of the coefficients the correlation coefficients and what we see when we do this is that um this is a histogram is that on average the mz twins tend to have stronger correlation coefficients than the dz twins which is shown here in the the dark compared to the light so that's telling us that there is a signal um that the mz that there is a genotype effect we can do the same kind of analysis that i showed you from pete turnbow's paper and we get the same result so if we look generally across microbiota's using the unifrack metric we don't see a difference between mz and dz's but if we look within specific families and these are the two dominant families of the fermicutes which is the dominant phylum in this in this study we do see that the mz's have a significantly more similar microbiota's compared to their dz's compared to dz's within uh within these particular families um we can also use twin based models to look at these kinds of data so this is using the ace model which is a which is a twin based sort of statistical model that partitions the variation in the data to genotype effects and environmental effects and what we're showing here is the as the effect of genotype with a heritability and we're showing it as a as a color so the strength of the heritability is on this scale with warmer colors meaning stronger and we've painted it on two parts of the phylogeny so that you can see that there's specific branches within the phylogeny that have particularly high heritability values and again they're within these two families the rumenicocaceae and the lachnospheraceae such as this part in here and there's other parts of the tree that have particularly cool colors like the bacturidides and this is telling us that bacturidides are particularly not heritable so this is allowing us to zoom in on parts of the the bacterial phylogeny that is most likely to be under host um genetic control and i just point out down here that we have methanogens and they're also coming out as um with some heritability which is kind of interesting and that corroborates what's been shown before with methanogens. We can use these techniques and look across other previously published studies so we pulled data from the Turnbaugh 2009 paper and a more recent paper from the Gordon lab that also had some twins and we do actually find some of the same patterns with these big families in the firmacutes showing us some heritability and much less heritability in the bacturidides shown there and then and i don't want you to write this down because this would probably change because as you know you know a thousand samples in a GWAS study is still a very small GWAS study and and we're in the middle of this it'll it'll expand over the next year or two but using the data that we have using the SNP data from the twins we're able to run genome-wide association tests and we we're finding things like this where for example u-bacterium shown here is has a correlation with the genotype of a particular gene so this particular SNP is coming up in association with you with a particular species of u-bacterium here and it there is actually a risk allele for crones in in this gene but but again it's early days it's particularly underpowered at this point and so it will update we'll be able to update this going forward but in terms of gaps we still don't have an understanding of how the host genotype determines the microbiome and we don't know at this point how the microbiome interacts with host genotype to determine risk susceptibilities but i think incorporating microbiome into GWAS studies for particular diseases is going to be very fruitful we already know for instance that different kinds of helicobacter pylori pose particular risks to particular kinds of to people with particular genotypes and so it it would just makes intuitive sense that this is also going to be the case with microbiome and at this point it's it's going to be interesting to find out how much variation in any kind of host trait could be explained by microbiome component alone or in combination with the genotype and and i i see this as as i really look forward to seeing microbiome incorporated into all kinds of GWAS studies in the future especially those that have some kind of inflammation basis that the microbiome might be driving so now this kid wants his own personalized microbes or at least he he wants to be able to determine you know the the abundance of certain kinds of things that you might get colonized by and what does he get well there's some thought that he might be getting his mother's microbes at the very beginning and and these are data that i think Maria Dominguez-Bello might talk about later on that very early on we know that that it looks like microbiota come from the mother and then other sources come in later but i'm going to use this to segue into pregnancy because and this this nicely shows it a pregnant woman who's just given birth is still not exactly like a pregnant woman a non-pregnant woman and you can see this like Kate for example would not have been looking like this a year ago so pregnant there's profound changes that happen during pregnancy and and persists to a certain degree after pregnancy and we were interested in those and i'm going to talk about some work we've done on pregnancy now and the microbiota that are associated with pregnancy we know that when women go from the first to the third trimester of pregnancy several things happen they gain adiposity their blood glucose levels are higher and their insulin sensitivity is decreased and these are thought to help grow a baby by keeping the blood glucose levels higher and then the increased adiposity is is thought to help prepare for breastfeeding later but these these these actually these things fat mass blood glucose insulin sensitivity have also been shown to be regulated by microbes in non-pregnant animal models so we were curious if microbiota had any role in this so we worked with Erika Ausolari and Seppo Salmonen from the University of Turku who had ongoing studies with pregnant women and they provided us with stool samples and diet data and clinical data and and baby stool samples from 91 women in Finland and first and third trimester and what we saw was the first view of the data they got was this so here are the same women sampled first and third trimester and the first trimester samples are very similar to one another and this again this is based on 16th but the diversity increases expands quite dramatically in the third trimester and the purple dots are postpartum one month so this point we didn't know what was normal the tight clustering or the not tight clustering so we put it in the context of the human microbiome male and female obviously non-pregnant people here and although there's a shift along this axis that might have to do with geography because these are Finnish people what what we see is that this big expansion of beta diversity is really not really not normal so we tried to associate the pattern with different things we had obesity and overweight data for prior to pregnancy but that didn't explain the patterns and neither did gestational diabetes status because some of these women develop gestational diabetes but but this didn't seem to map on to the expansion at all what did however seem to be driving these patterns were the the gradients of how much of the bacteriides and proteobacteria were in these samples so cool colors mean less and warm colors mean more so on this axis we have samples separating according to the bacteriides abundances and proteobacteria here indicating that the third trimester had many of the samples had more proteobacteria present and that was true for about 60 percent of the samples so if we look a little deeper and in use machine learning techniques to find the ot use that can discriminate the two trimesters we find that there's more short chain fatty acid producers in the first trimester which is kind of the normal looking state and the third trimester we saw things that reminded us of opportunistic pathogens things like proteobacteria which seem kind of odd and they're the kinds of types of bacteria that are typically associated with inflammation we also noticed a drop-off in the alpha diversity so the richness is reduced in the third trimester so in other words everybody has a reduction in the diversity but all in their own way because they become quite different from one another and this persists one month postpartum which is shown here so this is another way of showing the clustering tight clustering is a shorter bar and then you see the expansion of between subject diversity it stays one month postpartum and then when we looked in the kids you see that higher beta diversity in the children and then by four years of age they look like first trimester or normal people and I just note on the side here that we couldn't actually match children to their mothers at any time points even though the greater similarities between kids and their own mothers was for the four-year-olds in the first trimester but we couldn't actually match the one month or the six months old to their own mothers they weren't more similar to their own mothers than to anybody else's mother so we see this increase in proteobacteria for instance and other kinds of bacteria that don't look like they should be sort of that dominant in a normal setting and so we thought is there inflammation we'll look for it and we measured greater levels of inflammatory cytokines in the third trimester stool shown here some of these like IL-6 TNF alpha and IFN gamma so then we thought well what happens are the microbiota driving these these phenotypes at all so we took microbiota from the first and third trimester and and transplanted them to germ-free animals and what we found and now these are data for the recipient mice is that we after two weeks in the mice you could see a difference between first and third trimester microbiota and now looking at mouse cytokines we saw that the third trimester microbiota induced a higher cytokine load in the so so there was a sort of sign of inflammation in the third trimester recipients we saw higher blood glucose in the third trimester recipient mice and finally we got fatter mice um those that received a third trimester microbiota had greater adiposity gains after two weeks and these were swiss websters germ-free sort of same genotype that we used for both obviously for both inocular here so what we what we think is happening from this study is that the first trimester has a relatively normal microbiota and then as you go to third trimester there's an altered microbiota and what might be driving this transition is something that we still have to determine it might have to do with changes in immunity uh that that change mucosal immunity and when we transplant these microbiotas into germ-free animals it's enough to see some of these phenotypic differences recapitulated in in the uh in the recipients here so that we have a fatter mice with insulin desensitization which is actually what we see in the third trimester host when she's pregnant so these metabolic changes that are induced by by microbiota it's something that we've seen that's been talked about in the context of a metabolic syndrome in human beings and it's possible that they have that this kind of host microbiota interaction has maybe evolved in the context of pregnancy where it's really much more adaptive and maybe microbiota kind of used as a link in the chain here so that when there's a change in hormones or immune state that could alter the microbiota and then they altered microbiota then has an effect on the host but in this context it's actually beneficial as opposed to this other context where maybe it's less beneficial and so one of the things we're doing now is trying to compare you know do we see the same kind of host microbial interactions underlying these um these metabolisms in the metabolic inflammation that might be happening here and it's intriguing to think of whether these kinds of host microbial interactions might have actually come out of the context of reproduction from an evolutionary standpoint so I'm gonna this is my last slide I'm gonna finish with uh what I think some of the big knowns and unknowns are and um where I think there's there's some areas that that might be be very interesting to go into and that's thinking about what the extent to what extent do microbiota affect host phenotype so we know from these germ free studies for instance with transplantation that have been done in various labs and pioneered by the gordon lab that microbiota can affect metabolism immunity there's there's work coming out showing that microbiota can affect behavior but there's many many other things that we haven't actually looked at yet such as aspects of fertility uh aspects of longevity how much activity animals or people engage in um basic physiology uh there's a lot more to look at the extent to which microbes can do this how they do this on what kinds of microbes do this and and and how much can you push it so and this is a picture that's stolen from garland's lab at UCR where they've been selecting for my say that enjoy running a lot but I think that we could do these kinds of studies where we select for microbes that confer these kinds of behaviors and it's going to be very interesting to see going forward just how much microbes influence phenotype and then I think we can use that as a basis for therapeutics later so I'd like to finish up by thanking the people in my group specifically uh Armie Coran who worked on the pregnancy study and Julia Goodrich who's been working on the twins UK study um I have a number of collaborators just as just some of them um including Tim Spector and Andy Clark um who work with me on the uh on the GWAS um on the GWAS study and Erica and Seppo who worked with us we're very generous with the uh pregnancy study that I mentioned and we have some other ongoing things but finally I'd like to thank uh from MIH the new innovator award which has really allowed me to do creative things and not worry about it um this is a fantastic program and then Bob Carpenter IDDK for funding our twins UK study and thank you okay if somebody we have time for one quick question especially if you're standing near a microphone and also as uh the questionnaire is going there remember especially people online that the email address is hnvisionatmail.nih.gov so with respect to one of your earlier slides saying that nobody's been followed from birth to old age what am I wrong no no no okay no but I mean uh cesarean versus natural birth is something that everybody remembers and it's a relatively easy thing you could correlate at later age has anybody done those studies well that's interesting with our twins we've we've been uh asking them were you born by c-section and were you born uh vaginally and funnily enough not everybody remembers or or knows but we're trying to use use this cohort because because we can't ask them all sorts of things that's something that we're we're trying to do okay let's thank Ruth again let's thank Ruth again thank you okay our next talk is uh Jacques Revelle um microbiome dynamics in adults and here's Jacques