The Complete Blue Horizon Sessions

>> i'm pleased to welcome you to the second day of our meeting. i have some announcements to make before we get started. so first, there were 200-300 people on line yesterday. and we expect a larger number today. i thank all the people on line

for attending. and please use our twitter feed and gmail account to send us questions and those links are on the main screen and throughout the meeting they will be on the side screens as well. also, the nih institutes who fund much of the work that you

do, have prepared fact sheets about their interest in microbiome research and those are on the table by the registration desk. so please stop by and pick up copies. we are very fortunate to have jeff gordon here to enlighten us

this morning. jeff is truly one of the founding figures of modern human my row biome research. those of us old enough have been reading his microbiome papers for 20 years and all the younger people are trying to catch-up. we have all been influenced by

jeff in how we design our studies and just very basically, how we think about the microbiome. and i think the most important route of jeff's influece on the field has been through the people whom he has trained. and i would just like everyone

here who spent time in jeff's lab to stand up. okay. keep standing for a minute. now you'll notice a number of the speakers among the people who are standing. everyone here who has or is currently spending time in the

labs of these individuals, would you please all stand up? you didn't bring your labs with you? you can sit down now, thank you. you won't have to remain standing through jeff's entire lecture. the people you see or who stood

up, are familiar to all of us and have transmitted the wisdom of the microbiome and are continuing to do so as the people who are there training so jeff, in many ways, has shaped the entire human microbiome research field to which we all owe him a debt and

our presence here. so, without further adieu, i'd like to welcome jeff to the podium to speak to us about the gut microbite ota and childhood under nutrition. [ applause ] >> dr. gordon: is steve banyan under there?

thank you, bob, for your kind words. and just in the spirit of what bob was saying, there is an african proverb ink captures the spirit of science. it says that if you want to go fast go alean. if you want to go far, go

together. and it's been my privilege to be inspired by a magnificent group of students, postdocs and staff and collaborators over the years. this is a snapshot of the lab that i'm at presently. this is their story and i'll

highlight a few individuals during the course of the next 2 1/2 hours that i will be speaking to you. is that right? we are funded in large part by the bill and melinda gates foundation, which i think has had enormous impact on science.

it's inspired students to look beyond self and beyond family, beyond community, beyond nations to look at the conditions that human beings live in this defining century for humanity where we are tested, our capacity to evolve psychologically and will

biologically to the point where we can care for one another and respect one another. we can promote this. so i thank the foundation for their funds and also for the hope they bring to all of us. this field brings together in different sets of lenses looking

at a set of problems that are very vexing and challenging. that coming together is a precious moment not just for participants but also for the universities that are fortunate enough to attract them and it creates an opportunity to evolve the way we do science.

i really believe that the field of microbiome research creates an opportunity for experimentation, not only educationally but also in the structure of science. and the bringing together of people into ecosystems where high-end technology is demock

tides, made available on the bench so that students can pull a hypothesis out of themselves, design well-controlled experiments g directly to the machines themselves, look you should the hood, gather data, massive amounts of data and mine the data and inform themselves

about what algorithms currently exists or what has to be developed to be able to go to a cluster nearby and to be able to do their own jobs, install theironal go rhythms, install the data and the next experiment. and scales in science, it's a

opportunity to bring together people who can co-evolve computational and experimental biology and one of the major focuses of this field as we speak with common and similar voices within our institutions is to encourage our institutions to make those investments in

these special ecosystems where direct access to technology, not only sequencing technology but mass spectrometry and others can be experienced. create laboratories where people can tinker and think beyond what is currently available and can know their own data.

i think that's the key to success in many respects for the scientific journey. so, as i said, we live in a very challenging century and i won't make any political statements about the challenge of the moment. but, one of the most vexing,

tragic challenges that are faced deals with the healthy growth of children and highland hood under nutrition, which is the leading cause of death worldwide children under 5. there have been a lot of epidemiological studies that have shown that the risk for and

manifestations of childhood under nutrition are the result of a number of interacting factors. not food insecurity alone. there have been remarkable improvements, therapeutically in terms of treatment of severe acute malnutrition where ready

to use therapeutic foods are administered for short perceived. children gain a certain amount of weight and they and their families are given nutritional counseling and if the infrastructure of the health care system is sufficient, they

can be followed up. typically that is not the case but nonetheless, these therapeutic foods reduced mortality but not repaired the long term sequel i, including stunting and immune dysfunctions. we are not treating the systems

in a way that restore them to full functioning. and what might those systems be? and the hypothesis we are pursuing is described in the current slide. so, it postulates that the gut microbiota has characteristic and shared features of the

developmental program; that this healthy development of the microbite ota is linked to healthy development of infants and children and there is say disruption in this developmental process that is causally related to the pathogenesis of malnutrition.

and then you can envision some of the consequences of disrupted maturation of the microbiota. insufficient execution or execution of this program could influence the ability of pathogens to find niches or jobs in the microbiota, could effect the codevelopment of the gut

mucosal immune system and impaired development could effect pathogen burden if the developing microbiota is able to produce a variety of signaling molecules and metabolites that are instrumental in the proper development of post-cell lineages and impairment in this

developmental program, could fact organs well beyond the gut. so, that's the hypothesis. the sub text is that if we knew how to be good microbial farmers, we could have a way of not only monitoring the trajectory of development of the gut microbiota but to install a

series of prescribed steps that may ensure the healthy development of this organ. and i think all of you who are fortunate enough to have children know some of the dilemmas that are faced, including the decision after a period of exclusive milk

feeding, what kind of complementary feeding practices you shall apply to your child. so, let's go through the journey that all of us go through, which is defining normal populations and determining whether that definition of normal generalizes the other population, developing

amendmentrix to quantify deviations, determining whether those deviations are associated with disease. if there is a significant association to at least preclinically begin a journey to determine those deviations are simple leave effects or causes

of pathology, and if causality is established, throne design interventions that could repair digs function, repair dysfunction in a safe way, both in a short-term and long term. and do so perhaps in ways that generalize across different populations.

if there are deviations from normal in terms of this developmental program that are causally related to childhood under nutrition, then the ability to repair these deviations from normal, this dysfunction, may have long-lived consequences to the metabolic,

face logic and immunologic and neurological phenotypes of a child. so we turn to bangladesh and our collaborators in bangladesh. this is a slide, to give you a few definition its, will define stunting as a height less than or more than two standard

deviations from a w.h.o. reference cohort. underweight, weight for score, wasting wac score and say that sam will be defined arbitrarily as a score of more than know -3 standard deviations and modern acute malnutrition, -2 to -3. so defining normal in bangladesh

we turn to an urban slum b55,000 people living in approximately 5 square miles. our wonderful collaborators at icbd, the international center for die real disease research, to trust the mothers for the health care providers is enormous and that is of course

absolutely key for one important facet of microbiome research, the ability to do time series studies and also long-term follow-up of interventions. so a birth cohort we studied those individuals who had consistently health anth pom tree were selected for an

analysis of the monthly fecal samples by 16s sequencing, although ongoing efforts are using a whole host of other tools but let me just focus on the question of who is there as a function of time in their relative abundance. applying machine learning

algorithms you can identify the most age discrimy tory strains and develop a sparse model of the most age discriminatory strains and plot the results in the term of a heat map shown at the lower right of the slide where each row represents one of these strains.

we are looking at the first two years of life. each column is a monthly age band an you can see the changing patterns of representation of these 24 different tax a if you take a vertical slice through there, you can look at the relative

abundance of these age discriminatory strains and assign an age or state of maturation to the mic row biota. this age of microbiota calculated in this fashion is correlated to chronological age and i'll give you data about that.

so, when we developed -- together who drove that analysis was a very talented m.d., ph.d. student whose picture is shown in the lower left, a number of questions immediately emerge. is this model generated from the children and generalizable to other populations?

so the children here have this shared feature defined in this fashion in community assembly or succession. they have other birth cohorts at other states in know one of the gates funded network of cohorts distributed around the globe representing individuals who

have different cultural traditions and different linear tradition. what is the relative maturity of those with sam compared to those who have healthy growth phenotypes? and finally eare these age discriminatory strains simply

biomarkers of this succession that kurds in this and other communities or some of them mediators of healthiness? so i will tell you that these are distributed across populations and we have created random forrest derived models of gut microbiota development from

a number of sites. they are shown in the upper left of this particular slide. and there is a plot of chronological age on the x axis in the upper left hand corner and microbiota in the y axis. and this is the bangladesh random forrest model applied to

members of a cohort with healthy growth phenotypes living in a community in india. and the correlations between microbiota age and chronological age are shown in that particular case, .75 but all these reciprocal comparisons -- so site-specific rf models applied

to the population living there but also to other populations and we see in this fashion, that there are shared features of this community development. so, we define normal one population and looked for its generalize built neother populations n this case,

children with healthy growth, and now we can develop a metric for quantifying deviations for normal shown in the lower left. so this the children had healthy growth. hitudes random forrest-derived model and computer their testimonied microbiota age for

all the children, develop a standard deviation and you can see the correlations and for those who are not red and green color-blind, there is it say red dot on the lower left hand graph. that is say child who has severe acute malnutrition who presented

to the nutritional rehabilitation unit at the hospital with a chronologic age of 18 months. but a microbiota age of 8 months. this child has significant microbiota immaturity. and this is not unique to this

you'll see in a moment children with sam have microbiota immaturity. what is quite remarkable is the failure of current therapeutic foods to repair this. those foods were never designed with knowledge of the normal program of maturation of a gut

microbiota. so, this is a clinical study that was done by our colleagues at icdrb. they were asked to provide samples of children of sam prior to any treatment. they were randomized to a peanut-based ready to use

therapeutic food, which is surprise lentil-based, which has powdered milk and micronutrients and they were treated for a couple of weeks and they gained a certain amount of weight and then they were followed following discharge. so this is the calculation of

the scores. this is a healthy cohort of children. the value is around zero. this is the results obtained from the plumpy nut, the name of that arm. they present with significant microbiota immaturity, they are

in treatment, no effect. maybe a slight improvement post-discharge, and in this case somewhat of a regression. he also failed to repair this so these children are walking around with a persistent developmental ab normality effecting an organ, it happens

to be the gut microbiota and persistent stunting and persistent abnormalities effecting bone, et cetera. so, i said something earlier on about the importance of remember being able to study a cohort of individuals at a site where there is enormous trust between

mothers and the health care providers and that certainly is the case at icddrb. so this is the result of a follow on study we found to be very informative, a study where children were randomized to three different types of ruts. they were followed prior to and

during treatment for 12 months after discharge. monthly for six months and then every other month for another six months. and you notice there was a blood sample obtained together with the fecal sample at the time of enrollment, there is also a

blood sample at the time of discharge and a third blood sample at 6 months post intervention. and these are waz scores and haz scores and you can see that there is no effect on stunting. these children were severely stunted.

there is some improvement but not the normal waz scores. these children, this is their maz scores. this is with sam. this is post-sam. these are these children during this period of time. and persistent microbiota

so you can learn a tremendous amount about the biological features of these children at these three points. it's a time series study of a child that represents their own control. one of the needs of the field is to be able to reasonably hike

through quantitative serum proteomics and we have been extremely captivated by soma scan technology, i don't know how many people are familiar with this. it's expensive but you can monitor the abundances and levels of 1305 serum proteins.

the technology is going to advance. how accessible will be a matter of conjecture at this pointed in time. i thought i'd spend of the 2 1/2 hours i have been allocated, the next hour-and-a-half going through that matrix of proteins

where we correlated the indexes serum metabolites and a variety of other features with the representation of proteins to get a whole host of biomarkers. those have been extremely informative. and i just give you a few examples.

carry is very interested, a poverty dock in the lab, in the effects of the gut microbiota and bam biology. these children come in with a biomarker increased osteoblastic activity, that is the ctx or c terminal single peptide or type i collagen shown in the lower

left here. after treatment, that value is reduced on average, sis months of follow-up. and osteoblastic biomarker pro collagen type i, pro peptide, doesn't change. igf1 improves. serum leptin improves at the

time of discharge but regresses. there are a lot of alterations but we can see that there are many facets of host biology that are not repaired by these rutfs. so, our invited coconspirator testified to the extraordinary nature of this individual.

and the importance of targeted mass spec. i know that the ability to do non-targeted mass spec is very alluring in this field and important but you need the chemistry backunto be bab up to be confident -- we together with chris looked at

the metabolic features of these children at enrollment. there is rapid fatty acid oxidation and at the end of the intervention, there is say sharp lowering of carnitine levels and ketones and also pronounced increases in amino acids and metabolites, kilo amino acids.

so they are switching their substrate utilization and then 6 months after discharge, they start losing ground, amino acid levels decline. there is sustained lowering of fatty acids. so they haven't gone back to the rampant lipolysis maybe because

their insulin levels and igf1 levels are maintained above enrollment levels. so, just a flavor to define normal. generalizability of normal. a metric maz score for deviations to normal and correlations between microbiota

immaturity and severity of acute malnutrition and phenotyping of the children prior to, during and after currents intervention. is that immaturity effect of or is it causally related to the pathogenesis of under nutrition and manifestations? microbiota transplants were

performed by some very talented individuals in the laboratory. we're going to move from bangladesh to malawi for this particular study also conducted as described in bangladesh. in this case, chronologically age matched children, 6 months, 18 months, in each age bin,

healthy growers are those stunted and underweight and have significant microbiota each microbiota into a group of young, just-we understand typical malawian diet and just we understand and the mice are followed over time -- weaned -- i'll show you the experimental

design in a second theimportance of freezing fecal matters is very important. we use charged chriogennic canisters like nitrogen temperatures that can maintain that temp for up to 3 weeks. the average time at the sites between the generation of a

fecal sample and its emersion into that canister is seven minutes, which for us, is very, very important if we want to do metabolomics, rna-seq analysis, et cetera. so not all sites around the world are capable of that. although i think they are

generalizable. that allows us to create culture collections, methods that were championed and developed by andy goodman where we were so fort not have him in the laboratory. so each individual, we can capture a significant amount of the bacterial diversity in

middley plates where each well has a clonal out growth from a single starting cell and sequence the genomes. and you can ask, does the untacked microbiota transmit if that is the case, you can go into a room and maintaining humility, still shout, eureka

because you're on your way to dissecting out which organisms might be critical for transmitting those phenotypes. so, we published this so i'll just go over this quickly. this is laura, a graduate student extraordinaire who is now at the white head in david

page's lab. this is the experimental design. again, the key elements just weaned mice prototypic diet of the population of interest administered, colonization, time series studies, and then at the time of sacrifice, multiple tissue harvest for a variety of

analysis, including mass spec, take the bone out for microcomputed demography, do serial quantitative magnetic resinence to look at body composition as a function of time and then we have ways of taking mice out of thizes later and do an mr and bring them back

without exposure to environmental microbes. bottom line, there are transmiscible growth phenotypes in groups of mice that are fed the same prototypic ma allowian diet but the image microbiota transmit are faltering defined by reduced weight gain, reduced

lean body mass gain. there are abnormalities in bone growth and also there is impaired metabolism evident in multiple tissue that is can be characterized in part by a loss of metabolic flexibility. laura took the fecal samples from the recipient animals and

applied for it and asked whether she could identify growth dissem tory taxa. the organisms whose representation correlates with lean body mass gain or features of bone growth. and that was a very exciting she cull tedder these organism

and for a number of them -- culture -- put them together with immature microbiota from a stunted weight individual and showed their representation meliorated some of these phenotypes that i just described. okay, so that is say tested of

causality. i won't have much time to talking about this but i think it's an important challenge for the field which is to develop metrics for assessing the codevelopment of the gut mucosal immune system and the so andy, who i was very

fortunate to know whether he was a postdoc in the lab and who is in the audience. andy? are you here? did i see you? there he is. very modest but very courageous. so, he was able to apply fact

sorting to segregate out iga targeted mike robes in the gut microbiota from those who didn't have bound iga. he was able to sort out these organisms into an iga plus and negative fraction in way that is preserve the viability of a number of anna robes.

so take the iga plus fraction and numerate who is there based on 16s. culture them as he did but you also can introduce them directly into recipient animals fed diets representative of the microbiota donor and that's what andy did. in the case of sames, severe

acute malnutrition. so he mined the microbiota through the lenses of the host iga response. found iga plus consortium which he cultured, could transmit interropathy to recipient animals shown in the upper left or right portion of this slide.

km stands for another name for ma-malawi and you can see that disruption in the small intestine and epithelial cell adhesion molecule there. that's sustaining. and we see this as well. this results in rapid weight loss and taxia phenotypes and

lethality and andy found that he could mine the microbiota of a healthy child using the host iga response and identify some organisms that could oppose the transmission of this phenotype. that leads me to this slide which i think is important for a number of reasons.

largely the spirit of a slide because it's densely factor slide, which is how do we explore the microbiota along the length of the gut? it's still inconceivable to me that we haven't been able to recruit bioengineers who have been able to create capsules

that function autonomous leor semi-autonomously to sample the length of the gut and to monitor a variety of biological features, talking about metabolic features along the length. i know there are attempts to do so but i think we should really

encouraged those. so in this particular case, i'm talking about biogeography along the length of the gut. i believe there have been some talks related to looking at biogeography across multiple scales. gary is here.

he spent a lot of his time in the last several years developing methods for looking at the degree of add mixture of microbes in the gut not only in the lumen, per se but especially with food particles and mucosa and i think the results are fascinating and should be linked

to a variety of hypothesis directed studies where we looked at the cooccurrence of organisms not only in the way that is commonly done but also the cooccurrence spatially. and new technology like rob's minibioreactors to follow-through with that and

those minibioreactors are terrific potential upstream tools to design and hypothesis the best way studies&in the systems as well on our way to human studies. i want to talk about the loss of height, epithelial surface area with associated infiltrative

immune cells in the lamina. ee does. postulated to be one of the conpowders that contibutes to the persistence of stunting and other manifestations of there are a series of biomarkers of eed but they have not been really validated by concurrent histopathologic analysis of the

upper gi tract. so, in an extraordinary study that has begun that we are part of, done in bangladesh, as well as at the university called bead and seem. children who were refractary to current interventions for under nutrition, are being end scoped.

there was a lot of time and attention paid to gettinged approval for this but this endoscopy allows us to determine whether they have eed and to look at the microbial consortia associated with the mucosa as well as in the loomen and determine whether they transmit

a pathology you see here shown in human biopsies. so, the correlations that are being tested are there eed discriminatory taxa in the proximal small intestine? and are those taxa seen in fecal microbiota? are there distinguishing of the

microbiota that are only evident in the proximal gut and are eed associated taxad shared across different sites and can we do a test of causality by introducing consortia associated with my row biota in animals. that is being done right now. and a graduate who is in a

program, physician scientist training program where you fasts track, and i know a number of institutions have this through the clinical training. she is a trained pathologist in the laboratory. so, the first food, we are very interested in the relationship

between breast milk,al go saccharides and the microbiota and how it might transmit signaling molecules that might derive from these hmos to the host. we published it last year and our wonderful colleagues at uc davis, mass spec was applied to

breast milk obtained from malawian mothers 6 months postpartum who had healthy infants or seriously stunted underweight. we found differences in the representation of number of hmos. we were interested in the

sileiated hmos. we will be talking about those during the course of this meeting. but the infant isn't able to make much of the enzymes and biosynthesis expressed in low levels of the liver. the mother provides salic acid

in the form of hmos and mark, a very talented student captured niece a stunted under relate malawian infant and took a prototypic diet and supplemented with eitheral go saccharides and captured from the waste streams of food manufacturing. the way waste stream contains

lots of bovine millal go saccharides that look like human milk ol go saccharides and another group received another basic diet. and voila, he found that isocaloric diets given to these three groups of mice consumed in similar amounts showed markedly

different effects on growth. the enhancing the lean body mass growth dramatically shown here and these are the control arms. this is microbiota dependent not seen in germ-free animals and the exon host biology with dramatic exon bone biology effects on metabolic flexibility

and promotion of metabolic flexibility. that ask ask a whole series of studies -- that is say whole series of studies -- effecting features of under nutrition. but i'm going to finish up my talk talking about another therapeutic approach that

illustrates a pipeline that may or may not be useful or informative to you, which is predicateod hypothesis that traditional complementary foods are culturally acceptable, affordable and available and that they represent the results of many generations of imperic

observations by morgues who try correlate foods they give their children to healthy growth. so why not try to mine the complementary foods that are commonly consumed and try to identify combinations of complementary foods that might promote representation of these

under represented age discriminatory strains. so could we develop microbiota-directed complementary foods that would repair microbiota immaturity? if we could, then those complementary foods may be part of a more generally-prescribed

way of sponsoring good farming of a developing microbiota. so, a very talented group of people in the laboratory said bangladesh see - [ inaudible ] overseeing many of our tomorrow studies, rolled up their sleeves so to speak and did the

following: this is thepipeline that represents a marriage we think is important. the marriage is between these pre-clinical models, food science, and human studies. you don't need a hubble telescope. i'll use in words what this

shows. that is you take your targeted microbes, age and growth discriminatory strains in this case. we are going to screen different combinations of complementary foods for their effects on the representation express functions

of these targeted organisms. we are also going to spike in in the same animal models, organisms bad actors, enriched in children with severe acute we are going to get leads and prioritize the leads, leads being complementary foods linked to particular targeted microbes;

and pause and ask are these complementary foods available? how do they link to agricultural initiatives? how are they affordable? and processing considerations? is the way to incorporate them into foods that have adequate organoleaptic properties going

to be possible? if those things are considered, then we can develop complementary food prototypes which we can test to see with after food processing they maintain their bioactivity and advanced the define community and then go into microbiota in

the target population and see if they work. and then make a decision. you go from mice to humans or get a second species? and test your inside a second but then you proceed directly after that decision is made to get to a clinical study using

knowledge gained from the pre-clinical model that incorporated the microbes from the first go around. i'm going quickly go through this because i don't know how good i am for time, but -- seven minutes? excellent news.

not. so, this is the approach we used for screening. the rules are on the right. each group is colonized with a combination of age and discriminated strains and each group of animals receives a different sequence of diets and

initially composed of six ingredients, talking about a complementary foods and they are shown in the bubble plot for each diet. one is dominant per diet and we do analysis to make sure that the levels are not correlated with one another because we are

trying to link these foods to the response of a taxa. take a breath it's different orders of presentation that are important because we want to avoid -- effect. so we do this and then a follow-up screen and then a follow-up screen.

so, 80 plus diets later, we have lead microbiota-directed complementary foods that we can test. we use monotonous feeding of animals that have these sam strains and and with the patrolo type versus existing therapy. -- and these germ-free animals

as a way of measuring the availability of nutrients from these diets in the absence of microbes in different regions of the gut. we use mass spec and amino acids and b vitamins and look at monosaccharides and disaccharides and compare and

contrast the levels of these nutrients in the absence of microbes and in these particular dietary context and in the presence of these microbes. you can take that data by the way, and correlate the results with ra seq analysis to look at which genes which metabolic

pathways for utilization are being used by these different actors on stage and look at transportedders t is really quite informative looking at niches and you can also correlate with the predictive prototrophies these organisms. bottom line right shows

superiority in terms of abundance of these targeted organisms in the presence 6 advance showing that there are diet and microbiota dependent effects on serum igf levels. this is the combination of the m dcf and the presence of all 14 actors here.

compared to germ-free: igf is elevated, akt in liver is elevated and effects on bone biology. this is tomography of femurs. we go on and see similar results with intact microbiota and then decide whether to invest in a second speeds ease model.

so 2 1/2 years later a development of a robust protocol in pig lets by c-section is germ free in colonization. why pig lets? because they are more similar to humans and rapid growth phenotype, 250 grams a day. and then the opportunity of

looking atrophy tusser that are more difficult to delineate in mouse models, including features of biogeography opportunities to develop new technologies including ways of sampling those things along the gut for monitoring metabolism. you can see a formulation in the

maker row microbiota. and only two of these ingredients. this is the day of the experiment and the animals are rederived as germ-free bottle-fed with a formulation that looks like sow's milk. it is called sowina.

and then they are weaned and then they are followed up to 30 days. i can go through the interesting challenges of anesthesia. the 350 kilo gram sow is brought in and she'll have up to 15-18 pig lets. you have to anesthetize the

sow's pig lets when they are removed from the uterus. don't take a single breath until they get into the isolater and put through a dip tank because if they take a sinkel breath they will be contaminated by microbes but they can't be so deeply anesthetized you can't

revitalize them. so the bottom sideline -- two minutes. is there any debate about that? [ laughs ] so i'm almost done, folks. so, you have improved growth ice caloric diets and you have increased representation of the

targeted microbes. you have marked effects on host metabolism, including increased hepatic and diminished beta oxidation. so now we are at the stage of doing a human study begun in lang la desh and that study is summarized here it's a bake off

between our lead and nrdcs it's following a population prior to the intervention of blood and fecal samples that they -- at the beginning of treatment and blood and fecal samples at the end of treatment and fecal samples during the endpoint biomarkers are maz scores, serum

proteomics, pathogen burdens, a varietiesy of other studies of immune function including iga targeting and targeted mass spectrometry of serum. we have a unique opportunity determining whether we can repair microbiota immaturity, whether we have control over the

rate of repair in the long and short-term impact of such repair. we values an opportunity of these sorts of early pre-poc clinical studies, that not only select mdcf but also determine is response of these parameters linked to certain configurations

of the microbiota? so are we going t be able to daughter-in-law whether specific clinical outcomes are correlated with the representation of specific bacterial strains? and what we'll learn from the non responders and what additional approaches might be

useful, including direct administrationful pro biotic consortium though the regulatory route is much more complicated which is why -- discern discern challenges, define underpinnings of microbial succession. the challenge is great but there are models for being able to do

this but i think the lessons learned could be great in terms of understanding the determinants of cooperation and competition between microbes and studies that are intended to develop next generation presymbiotics. understanding how to modulate

pathogen invasion and also potentially restoration microbiota including antibiotics. and the last thing is an issue i promise to take no more than 30 seconds but the words are at the threshold of a revolution in our understanding of the

molecular components of food with mass spectrometry demonstrated earlier on looking at the landscape of human ol go saccharides and that is a historic moment for us as consumers of the food and also as us as observers of microbes. so i think that this marriage

that i alluded to between science and understanding how these ingredients are transformed or not by food processing and how to link structural features of a food with the response of microbes and hosts and view the microbial community as transforming food

into a metabolic output that has profounds effects on many cell lineages and many features of our biology. i thank you for your patients and i appreciate being here. >> bob karp: we have a short time for questions. >> hi.

excellent talk, dr. gordon. i wonder when you watch primates wean their infants, typically there is say mast indication process in the foods. do you think that may be applicable here for recovery in some of these microbiomes in these children if the mothers

were to pre-chew the food and then offer to the infants? >> dr. gordon: interesting thought. i have no specific data one way or the other. i think any -- i think your point is an important and general one.

number 1, am partner with anthropologists to see how different societies apply child rearing practices. two, consider all forms of processing of foods in terms of how they effect the molecular composition how that might impact the microbiota.

i'll also say as a parenthetical comment, that the investment in science of the type i just described, in itself is incomplete if you don't invest the time to consider ethical, social and cultural and regulatory issues. and i think that if you don't

spend the time on that, then moving from the discovery and development to deployment will be severely impeded. so a lot of the work that we are doing is on the ese issues and regulatory issues and food based intervention and pro biotic interventions initially.

other questions? >> audience member: that was great. pouter. i have two questions, one technical and one broader. the first one is, whether do you think is the best time for neocolonnization in mice when

you want to study infants and the pediatric population? i notice a lot of people do neocolonnization at 4 weeks but there is already a lot of development. in my lab we are doing it more 3-5 days of life. so that's more the technical

question and the other broader question which is something i think you highlighted is, how can we be better at getting samples and understanding the microbiome internationally from all over the world? because i think we are missing an opportunity or many

opportunities to have good hypothesis-driven research because of differences in diet and environment and the way we wean our children. >> dr. gordon: as the latter point, i couldn't agree with you more. and i think that is one of the

philosophies of the gates foundation which is to take lenses of anthropology and apply to the world and see populations that are represented of different lifestyles and make investment where the trust that i described is great between the population and health care

providers and to be able to embark on a journey that i tried to illustrate. second of all, studies of biology involve admission to the types of experiments we also colonization of dams so that they can transmit their microbiota to their pups

beginning at -- and i think exemplary microbial communities, either intact or defined culture collections, which are more sustainable, need to be deployed to look at the effects of exposure on different types of host biology. so i applaud and try to

reinforce the statement you made about the importance of early exposures. >> bob karp: we have time for one last question. >> audience member: thank you. as someone who practicessed in india and seen these children right front and center, my

perception is that a lot of die precipitates all of these malnutritions in these children. i was wondering in light of the new study in nature, which said that the probiotics reduced the effect or the area, how did you adjust for the interaction from these multiple diets and whether

these inner ventions you had in mind would it also have outcome with reducing the diet in these children? >> dr. gordon: so you raise a very important set of points. and in our birth cohorts, we have a collaboration margaret at hopkins, a study of peruvian

birth cohort included in this network. so asking questions like die real days, the nature of the interpathogen burden and expose and you are complimentive feeding practices including issues of food diversity anded when feeding is initiated.

what are those effects on overall diversity? what are the effects on maz scores? again many of these parameters that have to be considered. one of the things about these transplant experiments you is can do those sorts of surveys

and then take well archived fecal samples representing different states of the child's health and be able to model the effects of those states on host biology and also interventions. you can enroll a child in all arms of a clinical trial stow speak boobie taking pretreatment

microbiota and transplanting different groups of mice and re-enacting the clinical study or trying new different types of therapy. it's a important part of our discovery pipeline. i'm finished with time and the program officer at nidds rising

and i learned to pay attention t him over the years. he's been very supportive but very informative. so thank you. >> dr. riscuta: good morning. thank you for coming to this important presentation. so we are now at session 5,

which is host-microbiota interactions. so we have agreed by now that we want to know what is there but we really want to know is how our body interacts with the microbiota and if there is a possibility for us to modulate it in a way that would be

beneficial for human body. we have four exceptionally good speakers today. dr. andrew gewirtz from georgia state university and he will talk about gut microbiota, diet, chronic inflammation, and metabolic syndrome. dr. dana philpott from

university of toronto. and she will present nlrp6: current challenges on studying the impact of the microbiome on host responses. dr. nita salzman from medical college of wisconsin. she will present antimicrobial peptides shape intestinal

bacterial niche creation and competition. and dr. ramnik xavier from harvard university ed will talk about microbes, molecules and mucosal immunity. so i invite all the speakers and give them applause. >> dr. gewirtz: thank you for

the introduction and opportunity to be here today. in the next 15 minutes, i'm going to try to convey four take home messages we learned in studying micobiota, diet and inflammation over the last few so this is an artist illustration of the gut

microbiota and beautiful as it may look, it will not police itself. rather it requires active insights and adaptive immune system to keep it in check. the first thing we learned in this area is that failure to keep a microbiota in check can

result in multiple flavors of gut inflammation, including colitis, perhaps under nutrition, and also metabolic syndrome. and i think this is a pretty generalizable message that failure to keep a microbiota in check can result in inflammation

but we learned it from study of mice that lacked the flagell angry ceptor. 5 allows direction of any bacteria that breaches mucosa, be it pathogens or non pathogenic, and in the absence of toll-like receptor 5, these bacteria are not cleared quickly

and this makes mice prone to developing spontaneous colitis. it is not all the mice, and this one of the phenotypes that is environmental dependent and we don't see it in every facility. and it can be removed by redashation of the mice into another microbiota.

we see it in the knockouts and not in the litter mates in the same microbiota. the more common phenotype of these mice is that they develop metabolic syndrome and this has been seen across multiple facilities but not all. in any case, these mice develop

a low grade inflammation that correlates with increased weight gain and adiposity, and other parameters ever metabolic the extent of gut inflammation in these mice is proposal to their phenotype. in other words, using fecal lypocall-in two which we have

shown is a reliable non-invasive dynamic marker of inflammation, we see a lesser degree of inflammation in those that have metabolic syndrome whereas on this log scale, the levels are much higher in the mice that have colitis. and this is proposal to the

extent of their microbiota alteration. in other words, the non clitic knockouts that have this syndrome exhibit moderate but nonetheless significant alteration in the microbiomes whereas the clitic tlr5s exhibit a more severe alteration

in the microbiota composition. in general-free conditions, these phenotypes go away entirely and will in other words, the inflammatory marker becomes unwell detectable in both wildtype and tlr5 knock-out mice under germ free conditions and any or all phenotype

disappears and any indicators of inflammation and metabolic phenotype as well. and the microbiota composition is sufficient to transfer the phenotype to germ-free mice. so wildtype germ free nice they got the knockout microbiota develop the metabolic syndrome.

so what are the features of this microbiota that makes it pro inflammatory and can transfer these phenotypes? we see a lot of differences in species that i won't go through but one general characteristic of this altered microbiota it seems to be more aggressive and

it can encroach upon the host o here using not dehydrating stains and measuring how close are the closest bacteria, we see in mice na lack toll-like receptor 5 or that receive the transplant the mice have bacterial encroachment into the mucus.

and general phenotype, this is true in humans. so this is a slide generated from patients of the local va in atlanta and has nothing to do with tlr5, per se but in humans with metabolic syndrome, they are featuring microbiota encroachment, in other words as

they have increasing levels of disglysemia and hemoglobin a1c, there is say closer distance between bacteria and the epithelium. but certainly for a disease like metabolic syndrome that had huge increase among constant host genetics, we started to think

more about what might have changed the microbiota and/or how it interacts with the host and inflammation. and we focused on dietary factors mainly. we thought about changes in macro nutrient content and micronutrient content.

so the 1st thing we did, like many -- and that brings me to lesson number 2. a healthy gut requires proper nourishment of the microbiota. so when we began exploring diet like many researchers, we compared mice given a quote/unquote high-fat diet to

the standard used in most animal facilities which is chow. so, as was discussed yesterday, the high-fat die set an open-source diet. it's assembled from very well defined partially purified ingredients such as shown here whereas chow is a conglomerate

of food scraps. you can see the list. ground corn, beat pulp, fish meal and this makes it not very amenable for well-controlled types of studies of how various food ingredients might affect the microbiota. and it does have one thing going

for it, this conglomerate of food scraps keeps the mouse intestine healthy, specifically these are pictures of showing what a mouse intestine looks like on a normal chow diet. many are familiar with this. on any of the open-source defined diets, intestine is

unhealthy. it's fragile, there is a major loss of mass and this is irrespective of fat content. this is a feature of all of the defined diets we looked at so far. and this correlates with the -- this is just showing the same

thing by histology that on any of the purified diets, you see a significant loss of not only colon weight but the villa shank is shorter. and working with research diets, we found we can recover if we supplement the purified diet with a fermental fiber but not

the insoluble fiber, cellose. this correlates with differences in proliferation, simply on the purified diets we would call gut atrophy, intestine doesn't proliferate very well. and this correlates with microbiota encroachment which we see on the high-fat diet and

again correctable by supplementation of that diet with imlin. all of this is microbiota dependent. in other words, whether one looks at colon weight orvilley length, the atrophy we see from the purified diets, we don't see

this in germ-free mice nor can we increase the villa in germ-free mice by supplementing diets with imlin. so what changes in the microbiota is the purified diet having and how can imlin correct this? one of the biggest differences

is one of the most basic, just measuring total bacterial loads in the feces by pcr. we see purified diet results in approximately one log drop in the total number of bacteria and we can recover this by supplementing the diet with imlin.

we don't recover the same microbiota composition, specifically high-fat diet or purified diet causes a dramatic change in composition you can see here. not much of an effect with cellose but you supplement the diet with imlin and not only do

you increase the total numbers of bacteria but you change the composition quited dramatically. so it's really not any closer to what we would see in a chow-fed diet. altered the composition further? we seem to recover the intestinal phenotype t looks

relatively normal. and supplementing the diet in this manner protects against high-fat diet induced metabolic here showing impressive effect of imlin and reducing adiposity and we see similar effects on other parameters of metabolic so maybe differences don't

really matter all that much. but, i remembered something that jeff had said at microbiome meeting last year, which was lesson number 3 and that is to be really careful about telling people what to eat. and these mice on these imlin supplemented diet that their

basal metabolism seems okay but if we administer the compound dss, we see an extreme colitis like we never seen before shown in these histological images. even at low amounts of dss is just a very severe colitis and the colon is affected in a couple of days and very severe

bleeding and death. and even if we don't challenge the mice with dss, if the mice are simply maintained on these imlin supplemented diets for a longer period of time, such as six months, they start to develop liver cancer. and this is a combination of

dysbiosisis and attempting to repair it with these imlinen riched diets. so if it's not safe to tell people what to eat, can we at least tell them what not to eat? or some things they should avoid? and that brings me to lesson 4,

which is that some common food additives can perturb the microbiota and promote inflammation. specifically the class that we have studied are emulsifiers, detergent-like stabilizing mixtures. they are very common food

additives incorporated into all sorts of processed foods. we focused on two synthetic emulsifiers, compounds that don't exist in nature, polysore bait 80 and carboxy methyl cellose that has been used since the mid 20 century and increasing incidents of some

chronic inflammatory diseases such as ib. and metabolic syndrome. -- ibd and we found that both of these emulsifiers have quite a strong impacts on the altering its comp and promoting microbiota and encroachment. so again, this is non

dehydrating affixative to look at where our microbiota are located in the colon and hopefully appreciate that in mice fed either one of these, bacteria are encroaching upon the host and getting closer to the epithelium. so what is the consequence of

this microbiota encroachment? if mice are prone to developing colitis such as il10 knock-out mice, they are more likely develop it if they have consumed emulsifiers. these are the biomarkers we used to assess colitis and this is the brief summary.

in our facility, il10 knock-out mice are prone to developing colitis so 40% will develop any way if they maintain an emulsifier that incidents of colitis doubled. in wildtype mice, encroachment seems to be promoting metabolic syndrome in mice fed either a

emulsifier increase body weight mostly in form of adipose tissue increased food consumption and disglysemia. if the emulsifiers are given in germ-free conditions, there is flow effect on phenotype, no effect on metabolic phenotype or any of the other parameters we

can measure. and well in subsequent experiments using ex-vivo systems, our results are suggesting the emulsifiers are acting at least in part on the microbiota directly rather than directly on the host. so summarizing the emulsifier

work which was published last year, is that this class of food additives seems to be promoting bacteria encroachment and misbehavior to a place they shouldn't be. they may like it there but i don't think it's very good for the host.

and we have a, in collaboration with gary, we just got a clinical trial funded so we'll see how applicable this is in humans hopefully in the not too distant future. so, to finish with the gap, i think the general gap i'd like to highlight is the need for

better understanding of how a range of major and minor food components impact microbiota composition and gene expression. and to me, one critical specific gap that is going to make this a lot hardner mice is that at present, the lack of a compositionally defined --

open-source diet that can maintain a healthy intestine reminiscent of a chow-fed mice. i think it's nice standardize with diet, with facilities and microbiota but i worry that in the risk of standardizing, we get much further from physiology.

so i think very clean mouse facility is moving away from relevant physiology and then open-source diet at least the ones currently available are very much moving away from relevant physiology as well. so lastly, i just want to thank the people who did the work.

everything i showed today was generated by these three very talented individuals: [ reading ] thank you. >> dr. riscuta: thank you very much. so we'll have panel discussion at the end of the session.

our next speaker is dana philpott from university of toronto. and she will talk about nlrp6, >> dr. philpott: thank you for the introduction and also for the invitation to be here i'm probably the only canadian here. which you'll notice when i talk

about certain things. i'm going to talk about the impact of the microbiome and intestinal inflammation. so in my lab, we study the receptors and these are a set of receptors that are present within the cytosolof the cell that detect bacteria and trigger

inflammatory responses. the character features are these c terminal series of repeats which have been shown to be involve in ligand sensing in some molecules especially for nod 1 and nod 2. and then on the end terminus of these proteins are different

protein interaction domains that allows them tow connect to different pathways to simulate these inflammatory signals. so, this also defines them into different sub families so the nod-sub family has this recruitment domain, which allows them to link up to the nf cap b

signaling pathways. these other molecules, which we study also in my lab, have a similar domain and we don't really know the binding partners of these molecules. the nlrp sub family have this purine domain and this allows them to link up and form

inflammasome complexes important for the production and secretion of inflammatory cytokines il1 beta and il18. so nlrp6 is a classic nlrp structure. it has again the purine domain on the end terminis and repeats on the c terminis.

but i think for now we really lack a lot of biochemical information about the nlrp6 potential involvement in the inflammasome. there has been one study to show that it interacts with asc but whether it really forms, it has not really been formerly

demonstrated. and also i guess importantly also in terms of ligand, we don't know what triggers the activation of nlrp6 is. we were interested in nlrp6 because my lab studies nod like receptors in the context of intestinal disease, inflammatory

bowel disease and nlrp6 is interesting to us because it's highly expressed in the gastrointestinal track in humans and mice especially in the small intestine and this shows an rna-seq we did in most -- these are epithelial cell populations, looking at different expressions

of nlrs and you can note that nlrp6 has one of the highest expressions compared to the other nod-like receptors including the ones we like to study nod 1 and nod 2. and interesting expression that is absent or beyond the level of detection, nlrc5 and nlrp3 and

these epithelial cells. so, there has been a lot of interest in nlrp6 over the last few years and this just summarizes its potential role in an inflammasome dependent way and its involvement in intestinal homeostasis as wel as colitis.

so, it's presumed that nlrp6 has this inflammasome dependent pathway that leads to the production of the cytokine il18. and il18 we know is important in intestinal homeostasis so it helps to maintan epithelial barrier integrity. but on the flip side if you have

low production of il18 then you have more susceptibility to colitis as well as colorectal cancer. so there seems to be this link between nlrp6, il18 and role in intestinal homeostasis. to summarize past studies looking at the regulation of

colitis and also of today gut microbiome, it's been shown that colitis and tumorigenesis susceptibility of nlrp6 mice as well as downstream potential effectors in this pathway in il18, it's a transmissible phenotype to wildtype mice so they show that if you take the

microbiome from these nice and put them into naive mice, then the wildtype mice show this increased susceptibility to cancer and in line with that, studies also have shown that there is an well altered fecal biome in these deficient mice characterized by the expansion

of these representative back torial fila as well as tm upon 7. but the potential problem that we have been going through, we have done this study with nod 2 deficient mice. we questioned whether this pro microby on the tick phenotype,

perhaps it was acquired stow castedically in that sort of facility and was potentially genotype independent. so we noted to revisit the role of nlrp6 in the regulation of so we took heterozygous nlrp6 mice, brid them to derive litter mice of wildtype pets and

kockout nlrp6 deficient boys and they were separated at weaning and i'll show you the data from mice that were put into single cages but we also saw the similar data if we put them in co-caged with the -- depending on sex, of course. and these mice were then again

separated and weaning and left for 5-6 weeks. and then we isolated different samples from the intestines and looked at distalil yum mucosal scrapings and colon scrapings as well as the fecal pellet. we extracted dna and then performed sequencing to analyze

the microbiome of these different animals. so i'm not an expert microbiome analysis so we have lots of help on this part but we wanted to do the most robust test to look at the different compositions of these animals. so this is just showing

principle components and analysis generated using this dissimilaritiy matrix and we also showed this with other distance measures including weighted and unweighted and i guess it's pretty clear from these studies if we look at the feces, colon scrapings andil yum

scrapings, we saw no difference. so this shows us that there is no significant effect of nlrp6 phenotype on the community, composition -- and we also showed using looking at different taxoid in these mice no difference in bacteria taxa when it is between these two

groups. and i should mention also that we were -- there is say paper that just came out in immunity, that showed a similar finding to us. they used germ-free mice and rederived the p6 mice into germ-free conditions and gave

these mice similar microbiota and showed there was no impact ever nlrp6 on community structure when they looked at x germ free mice in these conditions. so, what we did find perhaps not surprisingly, is there was a genotype independent but sex

dependent effect on the microbiota and again this is something that already has been shown but just to show you in our animals, so this is shown in the three gut compartments analyzed looking that the matrix. when we studied the male mice in

squares versus female nice circles, we saw a separation between the female animals versus male animals and this was completesly independent of genotype. using analysis, you can see the differential taxa associated with males versus females.

so males had a greater abundance of these organisms including -- and the females had abundance of a lot of organisms. so again as i said, sex also has a influence on bacterial community structure. so looking now at the role of

nlrp6 on colitis severity. so the similar set up, you get heterozygous breeding pairs and derived litter plates from these animals. i should mention we did a study that had just less mice to study these were signal housing and it didn't make a difference.

these mice were left 4-6 weeks and then challenged with dss. this is say typical model to study colitis in animals. so these mice were given dss for five days and then water for the remainder of the study and we monitored clinical signs daily, including weight loss for

example and physical sites, diarrhea. the colon is 7 days as well as collected fecal samples for a marker of inflammation and as well as did a number of measurements of cytokines from plants of these mice there was a phenotype and sex dependent

impact. so looking at the male mice, you can see wildtype versus nlrp6. we didn't see is any difference between genotypes although there was definitely an effected of dss. they were losing weight during the study.

what was interesting to susthat for the female mice, we saw a genotype dependent impact on colitis where the nlrp6 deficient animals had a greater severity of colitis in the model compared to the wildtype female it was also notable that the female mice have less severe

colitis than male mice and again this is something that has been previously shown. this shows the disease activity scores, including stool, rectal bleeding and weight loss. and we saw the nlrp6, female mice had increased severity compared to their wildtype

counterparts. so these studies show that female mice are partially protected against chemically-induced colitis and the role of nlrp6 in acute colitis winning this within this model and only reveals within female models.

i'm going to talk about a conclusion. so many of these statements were already brought up yesterday, especially, but just to go over our particular findings we showed it doesn't have significant impact on intestinal microbiota homeostasis.

the use of litter mate mice is the gold standard for normalizing genetics and also environmental variability in animal models. sex hormones are known to influence the microbiome. and finally nlrp6 deficient female highs have increased

severity to induced colitis. there seems to be an interplay in sex but independent of the microbiome that increases colitis severity. so getting to the critical challenges and understanding the role of the microbiome in most models, this study demonstrates

that the use of litter mate mice from heterozygous greetings is imperative for well-controlled studies. i think the things that we need to think about is that the microbiome influences most phenotypes not just restricted to the gut.

there is a lot of studies showing the gut microbiome can influence other disease models in animals independent of gut. so how can we better control these studies? there is a lot of studies out there that co-house mice. is this sufficient to normalize

the microbiome? our data says it is not. you really need to use litter mate mice from mothers or related mothers to have a well-controlled study. and the other challenge we have is how to control microbiome effects and complex backgrounds?

so many people like us for example, we have double knock-out mice. some people have triple how do you litter mate controls for this kind of complex genotypes in mice? lab to lab, it's a huge issue so multicenter validation should be

contact leagues and see if they can validate our findings and other facilities. will we need to move to a standardization of the microbiome using defined communities? and also as we have recently shown, there is other microbes

that impact phenotypes. we had a paper showing that parasite -- has humanige pact on gut at baseline. so parasites and other organisms should be also considered. improving this, this is taken from a recent review.

we need to report source and sex so genetic background, control for microbiome and discuss the breeding strategy we use to generate litter mate mice. animal numbers and we snowed our studies include sex within this and also the cooperation among stakeholders and funding

institutions and editors and reviews need to enforce these reporting rules. so i'll end there. so thank you very much. >> our next speaker is nita salzman from medical college of wisconsin and she will present anti-microbiota peptides shape

intestinal bacteria, creation computation. >> thing might be more from the side of the microbe than the the study in my lab has been antimicrobial peptides and factors that shape the intestinal microbial niche. we know that has been reported

in the last day and will continue to discuss there are a number of factors that shape the intestinal microbial niche and the one i'm going to speak about today is antimicrobial peptides which are immune effectors. there are a number of sources of intestinal microbials in the gi

tract. we know that panna cells produce a wide variety of antimicrobial peptides including defense ins. we know that intrasites also produce antimicrobials, the liver produces bioacids, anti-pry croakial and microbe -- microbes are a rich source of

antibiotics in the gut through the production of metabolic byproducts and other toxins. work from my lab a number of years ago looked at the role of defense ins in the regulation of the intestinal microbiota and this is looking at panna cells. this is an.

m from andy's lab of a panna cell found at the base of the small in testedinal crypt. paneth cells line mucosal surfaces and they produce and secrete a wide array of immune effectors and antimicrobials including defense ins and work from our group showed if you

change the composition of antimicrobial peptides within the paneth cell using litter mate controls with either reciprocal models of removing production of any paneth cell defense ins, we see shifts in the microbiome and then we also found that if you alter the

composition of paneth cell antimicrobial peptides, you alter composition of the microbiota and we and many other groups shown if you alter the microbiota you alter mewicose at immune responses and the animal then we turn to another question which was, what is the role of

antimicrobials produced by commensal bacteria in either forming or competing for an intestinal niche? and for there are we turned to enter cock us, it's a facultative anna robe and a commensal of animals and the human gi tract and it is quite

broadly found as a commensal. it's intrinsically resistance to foreign antibiotics and extremely good at picking up antibiotic resistance genes. this is resulted in the development of multi-drug resistant strains which are acquired infections and they

have been designatessed serious public health threats by the cdc. so our interest was to investigate how antimicrobial peptides contribute to this fitch formation and competition in the gi tract. to do this, we needed to

generate animal model that did not disrupt the gi tract with antibiotics because if you do that, you're no longer studying colonization, you're studying infection. we did this feeding mice with this in their drinking water for two weeks and after withdrawing

it from the drinking water we followed the colonization in the feces and then at harvest of gi tract portions. and we founds, this is showing over four weeks but we have done this as far as 11 or 12 weeks and we find continuous shedding of this from the gi tract of

these mice. and when we sacrifice the mice, we find that there is consistent stable colonization of the mouse gi tract without prior antibiotic disruption. so now we are able to get a lab strain of intercock us to colonize the mice and this

allows us to look at how it formings niche and how it competes for its niche. and it turns out that it can harbor plasmids fair moan responsive plasmids that produce bacterio sins. the plasmid of our interest is p pd-1 and it harbors bacterias in

21. and what we find is that bacterio sin,nents microbial peptides, can specifically target and kill bacteria of the same jean us and they act by damaging cell membranes. this is an example of as48, essentially the same as bacterio

sin 21. it distributes cell membranes and broadly anti-gram positive in-vitro even though they are considered to be specific, this a relatively pro muskuous bacterio sin. so a strain that harshos this encodes immunity peptide to

protect itself from being killed by its own bacterio sin. they produce this to give them a competitive advantage and alternatively under the correct circumstances instead of killing another organism with its bacterio sin, it encodes also what is called a conjugation

apparatus stow directly can contact another enter cock us and transmit that bacterio sin to an enter cock us that lacks it. and this requires. [ cell phone ringing ] contracts and we use that to our advantage to understand the

niche in the gi tract. what we found from 234 experiment, where we were looking at culture of this in the gi tract from mice that did not receive enter cock us, nice got our intercock us strain that lacked the bacterio sin go we colonize without bacterio sin,

it colonized at a low-level but persisted ently. if we colonized at a higher levels and essentially over takes the entire niche of enter cox us in the gi tract. if we knockout the bacterio sin, here we have it colonizing and here we have the knockout.

it no longer can colonize. it turns out if we restore the bacter sin on a separate plasmid it regains the ability to colonize but doesn't regain the ability to conjugate. we use that also later. so, the other thing we noted is we are looking at what happens

whether conjugation or competition is what is going on so we first colonize mice with intercoxide that lacks bacterio sin and then chase in what we would call our -- our regular strain that produces bacter sin and what we found -- this is looking at transconjugates.

at the first organisms and which ones picked up the bacterio sin plasmid. over a period of 5 weeks, 100% of strains remaining picked up this suggests that when delivered sequential, it can over take the entire niche in the gi tract.

and this also tells us something very important which is that enter cox eye local ice to a specific physical niche and directly physically interact with others in the gi tract. we thought we could employ this therapeutically and so we did. we looked at vre and up here we

colonized mice fine with multidrug resist ent. and we used our complimented strain because we didn't want to conjugate bacter sin activity to aability drug resistant strain to give it another mechanism to colonize. so we chased with a strain that

could produce bacter sin but couldn't conjugate it and we see over several weeks after treating mice, the mice we can eliminate almost colonization by vrw administration of bacter sin producing strain. we find that because we used the complimented strain, we could

not conjugate the plasmid to the vre strain. what was also interesting is we looked at the microbiome and we saw that when we compared the microbiota of intercock us colonized mice versus p pd-1 colonized mice, there were significant but subtle

differences in the microbiome. they were drink by changes in muse spur lumt turns out that it is not a grand positive. so we believe that these changes are indirect changes and not directly caused by direct killing from the bacter sin. and we just compared this to

conventional antibiotics. this is mice treated with streptomycin and we see much more significant differences in colonization between mice treated with september price in than compared to - audio september mice in streptomycin. -- bacter son 21 primarily

impacts the populations and this suggests that we can actually therapeutically target a specific niche in the gi tract by engineering organisms. this takes advantage of two aspects, one is it takes advantage of the native biology of the intercock us to

graphitate to the right place in the gi tract and then takes advantage of the specificity of the bacter sin is to specifically eliminate the multi-drug resistant strain d turns out this strain works on as well. but then we found a little hitch

in what we were doing and this leads to my discussion of the knowledge gaps that need to be addressed before we can advance these types of treatments. we looked at gene expression from these mice from the guts of these mice that were treated either with nothing or with

intercock us alone or intercaucus that harbor bacter son. we isolated rna from the gi tramp can tract and then looked at the microarray and did component analysis is to look at the microarray and i'm going to take you through the small

intestine. we see that the control mice are clustered very closely together. the ones that have just been colonized by our lab strain of intercaucus cluster quite separately from our controls. and those that received the enter caucus harboring p pd-1

cluster separately as well. so this is that the we have differential gene expression driven by the colonization and also by the p pd-1. and this is actually a major knowledge gap if we want to therapeutically intervene in changing the microbiome.

this is an important translational goal intervention but it may lead to unintended consequences. and i apologize for the stick drawing and all the arrows everywhere but essentially we don't know or have the slightest idea what have has effected what

in the gi tract? these microbes are producing this toxin. it's killing other microbes. are we signaling the host through the production of antimicrobial peptides? it's organism signaling the are we changing directly

changing other members of the microbiota or host response we know we have unintended consequences here and we need the knowledge for per cise so we are knowledge gaps. what i consider a major knowledge gab gap is the better understanding of interaction in

the gi tract which will allow more precise manipulation of this will entail the development of good systems, genetic systems for genetic manipulation of commensal species. a greater understanding of the bacterial physiology of and also the identification and

characterization of physical and metabolic sub-niches within the gut. many have been raised over time through many of the speakers over the last day. the collateral benefits other than precise intervention is just having a better

understanding. it will enhance our ability to make sense of our large datasets in our current studies and also it should allow more appropriate sequence annotation and improve our database accuracy which is clearly of concern for understanding our current

studies and finally, the people who did this work are here: >> dr. riscuta: 33 seconds left out of 15 minutes. so we have our last speaker for this session, ramnik xavier from harvard university. and he will present microbes, molecules and mucosal immunity.

>> dr. xavier: i want to thanks the organizing committee for this invitation what i'll try to do in the next 15 minutes is to give you some highlights of how we are approaching the gut microbiome in health and disease. so broadly the gut is primarily

involved to absorb nutrients and despite years of understanding of this process, i think there is still a deficiency in trying to understand what are the receptors that set these nutrients. at the same time the gut evolved a complex array of various

receptor systems that sends pathogens and today we still need to understand how these commence cells respond. the first we need to talk to you about is this i believe increasing recognition of how this contribute to the metabolic signaling pool that you have gut

immunity through a large panel of receptors that result in innate and adaptive immunity. so broadly one way to look at the gut, there are many firewalls that restrict access of these commensal microbes to encounter and signal through the ep legal yum.

a number of adaptive systems control these responses and we are interested in how do these microbes or commence cells interact with the layer and then contribute to signaling and homeostasis? so they took a very simple approach.

they asked the question, were there utilizing genes? so are there enzymes and transporters that we can manually curate and then try to identify using computational approaches which express these enzymes and transporters. so elustrate the principal pel

if you look at the pep 2 peptide is has a carbohydrate chain that undergoes cleavage of these sites and then recognized by these. so the initial nam sis that was done to identify these en giles and transporters computationally search the human microbiome one

database identify microbes that selectively express transporter genes whether enzymes or transporters. and what they observed was a clear enrichment in these class of mike robes and you can so here four of these species robustly utilize this energy

source. so with this background, what we asked is using a simple epithelial model, so the dss model that both an grew and dana just alluded to, to ask the question, if these individual bugs could we then alter the susceptibility to epithelium

injury with the idea that if you don't have a barrier then you're predisposed to additional injury so what we identified was expected was attenuating the epithelial injury by dss shown by many labs and also observed another member called vector straight caucus also increased

weight when in this setting of this model. examining sort of the strains that were identified in this setting, we identified five strains of vector streptococcus and by sequencing, there were 2 strains that expressed a gene cluster --

so using these three microbes we tried to identify metabolites that could be made by these bugs that could potentially signal in this new program. so what they identified were molecules that were differentially produced by the bacteria that contained the 15al

cluster. so, they produce prop antic acid and acrylic acid whereas there was a non differential production of another acid metabolite. when comparing sequences of ip too. ia was a presence of this double

bond which i'm going do get to in some of the signaling experiments. so with the availability of these two molecules, then we did two simple experiments, one tow ask the questioned how do organizers respond to these molecules and then how to

organoids respond to these micromolecules? both ip and ia used this production suggesting they signal to these cells and increase the production of il10, a cytokine that attenuates signaling. so using a system of both

organoids with and without macrophages they would show that they both have the properties of increasing the peptide and the increase in il10. the phenotype of ia as well as ipa was also seen in blood cells showing they both attenuated the production of il1 beta and il6.

if you then took these mol tools and exposed them to river blood cells, you can see that ia and ipa to a lesser amount increased the expression of these genes that i listed here. and motivation for this type of experiment was the presence of an ia compared to ipa suggesting

this might be a better electrophilic. so we asked what is the function or the human disease relevance for this observation? we went back to the metagenomic datasets that have been generated in the lab with both support and other datas, to ask

what happens to these two clusters in patients who have this boil disease. as you can see both patients with cones disease and ulcerative colitis have reduced incidents of this cluster compared to the controls. so, what i have tried to show is

we used a simple computational approach to identify transporters and enzymes that are important in utilizing metabolizing. we focused in on pep streptococcus and showed that this gene cluster here produces a tryptophan metabolite and gone

on to show that ia has an influence on the epithelial barrier as well as immune cells and then going on to show the function implications of this observation that this cluster and these microbes are depleted in patients with inflammatory bowel disease.

the second approach -- so what do we need going forward? i think eric will talk about this, is that we need a better collection of microbe that is come from humans with and without disease. we needed to try to identify what do these microbes make and

then sort of embark on some of the signaling experiments that will we have outlined. the second approach that is broadly taken is to work with well defined cohorts in most experiments that have been done using a cohort in about 1000 children who were born in

finland, estonia who basically carried these alleles. these children were followed for about three years and the reason for pressing on this cohort is children born in finland have a very high incidence of type i diabetes as well as various allergies to foods.

compared to children born in russia who have incidence of significantly lower similar to what is seen in the developing world and asked the question by sampling these children over three years with monthly microbiomes, can we try to begin to understand, are there

microbial factors that might tell us about this end phenotype? so tommy, a graduate student in the lab, set about studying this and what i'm showing you here are these plots that you can clearly see is that if you were born in finland, you have a

higher relative abundance of -- as kids born in russia had high incidents of ack tin bacteria. so what are the pathways that are abundant in these two cohorts of children show that one of the things that emerged from this analysis, a differential abundance of

biosynthesis and polysaccharide. so this could be one that could be differential because of the microbes compared to the kids born in russia. we then next asked with this hypothesis, what are the bacteria that contribute to the different types of lpsf you were

born in russia most lps came from e.coli whereas if you were born in finland, it came from ecola as well as a large group of mack roids. so with this observation we purified the lps that was made by each one of thieves microbes and found that we made lps that

came from e.coli, you activated pro-inflammatory signaling pathways whether you read out with nf cap b or various cytokine read outside whereas an abundant found in the finnish kids, clearly had a silent response to these pathways.

highlighting the value of cohort studies, focusing on humans with serial sample collection. so, we then asked about the tips of lps made by e.coli versus this as expected. the e.coli makes a 6 chain lps whereas the other suggesting there might be certain

structural features for this lack of signaling in the context of pro in pro-inflammatory cytokine production. another postdoc fellow went on to do a series of functional experiments and what i'm showing you here are cells that are stimulated with the e.coli lps

treated with increasing doses of the doreilps. and as you can see here, they counter act the immune stimulation mediated by the lps recognized by the tlr receptor in blood mononuclear cells and moan oocytes isolated from healthy individuals.

and this phenotype is seen both in children as well as in adults. to goat what might be sort the immunological consequences, i will show you two pieces of data shown here the experiment where they looked at the ability to induce endotoxin tolerance and

by both experimental systems the lps fails to induce end toxin tolerance as well as prevent the establishment of e.coli mediated tolerance and finally, the lab is primarily interested in understanding the human microbiome, the reviewers forced us to do one experiment and we

slowed that the e.coli lps was a important educator of the immune system whereas the dorei-lps fails to signal and increase the incidence of type i diabetes or the hypoglycemic state. i have told you by looking at well phenotype cohorts focusing on finland and russia, we have

incidents of type i diabetes very different. we have been able to go on to show that there is an important component of immune education that is required in the first year of life. immune development and maturity and if these children lack these

immune education signals, one is increased risk for immune disease in this context using type i diabetes as an example. the thirds type of experiment that we have been interested in the slab to start to do pertubation experiments again in humans to try to understand how

pertubations might tell us about disease onset and progression. and a gastroenterologist in the december broadly intereted in how can we think about the microbiome in response to the current approved drugs for inflammatory bowel disease. so despite huge progress, if you

take any of the anti-tnf or other agents about 40% of individuals respond to these drugs and what we were interested in was to take individuals who were exposed to a biological agent, so in this case the alph 4 beta 7 intergrain blocker and ask the

question, what are microbial features in individuals who respond to the drug and stay responsive to one year compared to those who response for a short period of time, 2-3 months and then lose response. and are there features clinical and microbial that might tell us

something about sustained response in these individuals. i'm going summarize all the data in three slides. so we collected multiple time points and what ashley was able to observe, there were clearly microbes in this case a short chain fatty acid.

the presence of this is associated with a better response and most sustained response to intergrain age and went on to identify a series of microbial factors when combined with clinical factors could give us a preliminary insight into those individuals who would

responsive compared to those who respond and then lose response. hopefully by the end of this year we will be able to report on something like 450 children with newly-diagnosed ulcerative colitis. these were kids who enrolled before they got treatment and

then were bending to the standard treatment regimen that is children with osteocolitis would be treated with and sort of begin to ask the question how many of those children had severe disease, mild disease and intermediate disease and how one might be able to correlate this

with host fact expose microbial -- host factors and -- i summarized the work of three talented -- two postdocs and a graduate student. most of the microbiome work is led by: those were great presentations. to remind you, this session was

to highlight the mechanistic and specific function of examples of host microbiome interaction and what we still don't understand about those interactions. and i just before the questions, i just want to give a few highlight that is i noticed. so andrew just mentioned that we

should not tell people what to eat but he says what we should imagine there are other examples of what not to eat. and we also understand from data that the importance of sex differences and probably we should look in menopause and this kind of sex hormones.

we understand a lot about the necessity of microbe-microbe interaction and microbe fungy and microbe interactions and it was explained to us the importance of commensal bacteria , the importance of host genetic variants as we are predisposed towards certain

microbiomes and also geographical differences into the microbiome which could also predispose towards certain so now, we are open for questions. >> audience member: hi, phil dashner, nci. my question is directed mainly

at andrew. so i'm interested in your observation of htc and the imlin supplemented knock-out mice. do you have any ideas of what is driving those hepatic tumors? interactions with a short-chain fatty acids or something else going on?

>> dr. gewirtz: i do want to say is that work is being done in vj kummar's lab at penn state. i'm a collaborator on it. essentially, it is mediated by short chain fatty acids and it can be prevented if the mice are administered a yeast product -- i'm sorry, a hops product, beta

acids. so indicated it is mediated by short-chain fatty acids and something about the context of the purified diet dysbiosis and then short-chain fatty acids driving it. beyond that we don't have a great sense.

>> audience member: matthew stole. uab. this is for dr. xavier. a very interesting talk. you showed the abundance was very predictive response and you also looked to see what bacterial predictor response to

inhibitors and what are the prospects we can use to guide which therapy to use. >> dr. xavier: so, to study that is published, we have done a small cohort of anti-tnf reeffects with the response the hope is that in this large pediatric cohort will tell us

how to solve for things effect the microbiome and what is the treatment response and steroids to step up therapy to try to understand what the effects might be. so i think a larger cohort will hopefully identify additional factors that have a drug

response. >> audience member: the question for dr. salzman. so, it appeared to me that spacial aggregation of species is playing a role in defining the niche here. so, do you think that formation -- and if the mucosa

layer place a role in that? >> dr. salzman: i'm sorry. can you repeat the last part of your question? >> audience member: so if the mucus layer helps in this particular species you looked at-- >> need neat it's an dr. salzman: it's an interesting

question. when we look at hybridizationing to try and identify a physical niche, we are not really very successful at funding specifically where the organisms are. and they don't seem to be associated with sort of large

groups of organisms in biofilms although they are associated with the mucus. and i think the problem actually to some extent, we don't have much architectural associations that we can make. it's not typical or basis or anything like that.

and the other problem is the numbers are relatively low. even though they seem like they are measurable and culturallable compared to the populations. they are actually so low we are having difficulty identifying a specific place. so it is hard to say at this

point, the role that biofilm would have. >> audience member: i have a question regarding the use of dss induced -- so two talks maybe related. it is actually a very tricky model and we found that the load-- the pathogen load in the mice and from bench to bench and

facility varies a lot and that effects the severity of the so, some of the results like donna, you see no contribution of genetic mutation -- so just gender and contribution not to the mic row biome. >> i agree with you. the problem i think we all have

with this model is the lab to lab variability depending on the source of dss. the concentration of dss and the length of time and all those things. i think again, need to be reported in papers so everyone knows, and everyone is on the

same page in terms of the type of dss they are using. i guess the other thing to consider too is that something i didn't bring up as well is in terms of animal -- like if they are on a solidified -- acidified water, we seen a huge impact on susceptibility to dss.

for me it's just making sure that we report all these things i our papers so that other labs can try to reproduce them. how long they are on dss, in terms of transplant microbiota transplant experiments, how long that microbiota has been in those animals.

i think we should have some standardized time where those microbes are in general-free mice for example, how long they should be there before you can say that they are stably colonized. i think all those things we still need to define in this

field. >> just to add to that. it would be nice have reproducibility across facilities but it would also be nice if we were working with mice that as much as possible were a good model for humans. and i think a paper last year

from david, and i think it was in nature showed, if you want mice within an immune system that resembles humans you should not acquire them from jackson or taconic or any other commercial vendor but from your local pet store. >> audience member: university

of michigan. i have a question. so very nice presentation. i was wondering as far as vre, you tried to use these tow kill vre but what about the opposite that the vre will spread to the gene -- and also could it be audioses a way of vre to

emerge -- how much diversity is within these bacteria? how many genes they have? how much diversity of these molecules within the genomes? >> dr. salzman: there are a number of different bacteriosomes harbored by enter

caulk us that produce a variety of bacter sons. the risk, if you don'ts use a crippled complimented strain to transmit the immunity gene, it is encoded on the plasmid. so if you can't conjugate the plaid plasmid, you shouldn't be able to conjugate the immunity

gene. so that would reduce the risk. clearly, there is always a risk a single therapeutic, that a certain population of vre will already harbor that plasmid and that immunity gene and therefore your therapeutic with bacter son 21 will not be effective.

so is there is probably a therapeutic to actually combine a variety of enter sons and bacter sons together to limit that risk. in the native population, obviously these strains were pulled from the native population.

in the mouse gut, which is where we have looked, we have never found the presence of plasmid and we haven't found it in human isolated strains of vre either. it doesn't mean it doesn't exist. >> you mean immunity gene? >> dr. salzman: or the bacter

son plasmid. usually they are encoded on the same plasmid. >> audience member: hi, alexandria shubert from the fda and my question is also for dr. sals man. i wanted to know or just get clarification really on your

you said you didn't want to use antibiotics because then you're looking not at colonization but and i was wondering if you could clarify because i thought that you could pre-treat with antibiotics and it wouldn't necessarily lead to an infection by vre.

it's not really, i thought, a gut pathogen, per se. it's more of a problem when it gets out of the gut. >> dr. salzman: what we find, and we have limited to sep la spore in antibiotics to mimic a human system, that you get a translocation.

you don't get infection, per se. in other words, the gut does not show inflammation when you get sort of expansion and translocation of intercox eye in response to antibiotic treatment. you get systemic spread. but if you use antibiotics to

establish vre in the gut, you have actually disrupted the standard interactions of the so you will have disrupted any aspects of colonization resistance, bacterial host that's what we wanted to avoid. we wanted to understand to the extent we could native, the

native interaction of enter cox nye its environment without having massively disrupted the other colonizers. when we colonize these lack we don't find any changes in the microbiome at all. so is that is probably a better system to understand true

mechanisms of colonization. >> audience member: peter, ucsf. i'm wondering whether you have thought about ways of dealing with these -- many examples now like the inflammasome studies that don't replicate and in one lab whereas there was originally a really interesting finding.

like how do we or how are you thinking about tackling that? is there a way to reconstitute the microbiome from the original studies and try to rebuild it in toronto or belgium or wherever else these studies are taking place? can you get the microbe system

from the original investigators? not only reconstitute the transgenic model but the microbiome it started with? what are you thinking about doing? >> a great point because i think-- there are studies out there with nlrb6 knockout mice that show that they do have an

increased severity in colitis. we never seen it in our most colony but people have seen it. and likely because within the facility there is say path that seems to trigger the inflammation in the particular genotype compared to the wildtype mouse.

so, one way to get around this, i mean it would be pretty difficult, would be to try to use defined communities of microbes that are healthy defined as a healthy microbe versus one that is pro clingenic and try to tease out the factors from those organisms that lead

to the triggering of disease in that particular genotype. that's one thing we have been thinking of but i think we are far from that at the moment. and then again, to try to see if other labs can verify your results. i mean that means more

collaboration and discussion with other groups but can other people replicate what you have seen in their particular facility with their microbiota? are bee going to have to report the whole microbiota of all our mice that might be part of what we have to do in the future?

again, i need to reiter that it you need to also been others microbes. we had a paracytic infection that flipped the intestinal phenotype in our animals. so it's not only the bacterial mycobiome but others we have to think about as well.

>> i thought a lot of the things were outlined that will be helpful but they won't happen unless incentives are changed. if someone is going to primarily publish papers that says i examined there in my facility and found the same thing, that person will quickly be unfunded

and probably unemployed. so what the current incentives in place where there is say focus on novelty over rigor, at both top journals and funding agencies that really rely on those journals, we can say it would be nice for these things to happen but they won't happen.

>> audience member: i wanted to direct the question to dr. salzman and dr. xavier. do you think it's possible when you talk about are we losing old friends, that in the virulence and antibiotic resistance, the virulent organisms in their genes that are producing

antimicrobials, the argues nel of aren't microbials is just a war going on between the virulent and the more commensal and those commensal especially in the bacterium lack of vasill i we use in probiotics are going into dormant states that are sort of endospore, going into a

sporealation state in these horizontal transfer of losing gene, adding gene, and losing gene, as this interaction goes on in evolutionary biology, and that when we really get to addressing the antibiotic resistance problem, it will give us clues as to how to restore

better the commensal that you think we are possibly losing. and it's not just an issue of diet but it may be an issue more of what we are calling pro biogenomic. >> dr. salzman: the first thing i wanted to say is i wouldn't describe antibiotic prescription

by bacteria to be equivalent to virulence because the production of these bacter sins by commensals or other antibiotics because bacter sins are kneely not the only antibiotics produced by bacteria, are for colonization more than virulence.

so it's actually more a benefit for colonization as opposed to invasion. in terms of restoration, i don't know if you want to speak to that a little bit more directly in terms of restoring commensals. >> xavier xavier the model

system. >> dr. xavier: tracking back is a hard experiment. some of the concepts you raised are very relevant. i think we need to find systems to look at bacterial states and host states and see how one might sort of push in both

directions. a round of applause for our speakers and we will be back at 10:55 for the next session. >> welcome to the second part of this session. my name is roberto floresc. i'm program director and the nutritional science research

group in the division of cancer prevention and the national cancer institute. please. so again my name is roberto, i'm the program director at the division of cancer prevention in the national cancer institute. so this is say continuation of a

continuum of this microbe-microbe interactions within the human microbiome and the session purpose to highlight concrete examples of how microbe-microbe and interkingdom interactions either effect or regulate human and health and then identify the gaps of

knowledge in understanding about the role of microbe-microbe interactions in these so we have an excellent panel of speakers. our first speaker is gautam dantas washington university in st. louis. and his talk title is

understanding and predicting the impact of antibiotic therapy on the developing pediatric microbiota and resis tome. >> dr. dantas: thank you for that introduction. thank you also to anita and the rest of the organizers for putting this on and for the

invitation and a personal thank you for funding all of us with this god work. i will also follow in jeff's lead. start out first by acknowledging that the incredible group of folks i have the privilege of working with i'll highlight

those who led the particular areas that are mentioned today but just to start by saying anything of importance i say today was done by these folks. i took the prompt from the end of the day yesterday to switch out the order of the challenge question to see the context of

my talk and i thought an interesting challenge to consider, which is a theme we heard before here ask, how, in a real world of a hypothesis of microbiome study, how do we accurately define the omics studies required to answer that specific biological question?

this might seem like a really simple thing but in this day and age where we really are on a systems biology field where we are tempted to collect just as much data as we can and then post-hoc analyze lies. so what standards we need? is it appropriate to think of

that in terms of what is the appropriate resolution to answer specific biological questions? we heard this at least in specifically with dna and rna omics, the type of method used. do you use shotgun sequencing and single cell of population, do you do longitudinal studies

and where do you call particular molecular entities and is it really important for the question? now, the flip side of this as i alluded to is, one aspect of this, you do have this opportunity for yourself as well as the rest of the field to do

post-hoc analysis and so, what might you want to consider in terms of maybe diving deeper than the question you want needs to enable those studies. and you know, how does that work in the way you write your grant application? and then finally eto go one step

further, not only do we want to understand the systems we are studying, we want to be able to predict. power of systems biology is this hope you can pattern recognize so the next study really builds off the study you have and then you can do things in higher

through put. so this comes down to a definition of resolution of what you're studying such that hopefully in the future you and others might be able to predict. that's my challenge slide. we and others, i think this is a team that goes almost all the

talks here, are very globally interested in being able to predict how microbe communities respond to pertubation. we heard about this in virtually every talk. we are very, very fortunate, important to note that many of the tunes we use come from

quantitative models that borrow from macro ecology. so you'll see through my talk this notion of being able to understand how either short pertubation or long pertubations result in changes in states and what is the acute nature of that or persistent nature of that?

that's the global framing. and the particular pertubation that my lab has focused on the for the last many years to different communities including the human microbiome, it should be obvious why because antibiotics are perhaps the greatest pertubation you can

think of in terms of microbial life because they are designed to kill microbes. what is important to note, that even though we think of these magic bullets as targeting specific parts of life in pathogens, all of those particular targets are shared

across the only tire domain of the kingdom. so there have been lots of studies that looked at over the large span of human life from early infancy after birth up to old age, how various types of antibiotic pertubations can lead to changes in host phenotypes

and we are interested in understanding what are the specific natures and specific antibiotics that can be used to understand the responses of microbial communities and again hopefully in the future understand what collateral damage that various antibiotics

might engender so you have a choice that can be made from a personalized medicine perspective. now, one of the major consequences obviously of antibiotic uses and resistance, these are stats that hopefully you seen before.

i really like to e size them though because antibiotic resistance in pathogens is our greatest infectious disease challenge right now. resistant infections claim about 700,000 lives around the planet. it's been estimated if we don't change the game in terms of

resistance understanding and new antibiotics, that number will bloom to 10 million deaths by 2050 around the planet. imagine this meeting being held in 2050. over the course of the meeting, about 86,000 people would die from this.

in addition to the huge human burden, there is also the cost, the estimate of this problem to the u.s. economy is on the route of 55 million dollars and we scale this 10 million deaths, the cost to the global economy couple hifly, about 100 trillion dollars.

100 trillion dollars of the last six years of the entire usgdp wiped out. and unfortunately, to make this worse, at exactly the time we have got these rising infections, we have fewer and fewer drugs coming to market. this is the motivation why we

want to study this problem, both from a pathogen perspective and all the communities of microbes, like the microbiome they interact with. now we all realize that any scientist worth salt associates themselves with their own olt. the we have our resistome.

i'm using that phrase to describe any collection of resistance in a particular community of sample. and we recognize we can turn to culture to look at the resistance genes in that population and we can compliment the culture work with shotgun

sequencing to look at all the microbes in the community and these methods can ask questions among antibiotic resistance. the problem, is that this is say self fulfilling professory. and so what we have done standing on the shoulders of giants like joel handelsman we

have turned to this method called functional -- so, we use these type of methods along with the other methods available to us to establish such things like the soil resis tome to the resistome of the soil is extensive but fortunately most of it appears

to be not as easily set up horizontal gene thansfers and pathogens. that's a good part of the equation. however, we found that there are particular microbes that defy those strains that appear to be facile at crossing habitat

barriers and also gene transfer. so this allows us to start zooming in and maybe the vehicles of these particular transmission devices. we have also then looked at colocalized systems of both environments as well as human mycobioma so this study led by

erika pearson in our lab, we looked at a village in el salvador and a slum outside of lima, peru, which in terms of lifestyles in people represented two-thirds of the human population, and asked questions across both their fecal samples as well as environmental samples

around the houses and as well as the waste water treatment plants. for instance, how these people got my crow bite ota and how are they structured? so we fine confirming hypothesis others put out before that human my crow bite ota seems to be

structured by lifestyle as opposed to simply geography. they seem to be structured by habitat. there is a signature of the type of bugs that are in the gut of the human that are different from the type of bugs that are in the gut of a chicken and the

resistance genes generally follow that trend too. so on inference from that, perhaps there is not a ton of horizontal gene transfer going across habitats or at least it's not overwhelming the system in terms of sequencing. so what we do then is go into

this and ask for which of the genes that break those trends because those are the ones that will be hotspots for transmission. this is two examples of things that erika and pablo found. chicken coupes inel valva dore and waste wafter treatment

system in peru seems to be enriched from these gene transfer that is make bugs compatible with the environment also compatible with the human this has obvious public health benefits in terms of knowing this exists because for instance in lima, lima is a desert so all

this clean water enriched with genes is thennitudes irrigate every park in lima. so that is the end of the sort of survey part of the resistome. what we try to do to compliment the basic science find suggests turn to various areas of the we try toet go the predictive

models of how the microbiota responds. we focused on the developing pediatric microbiota. you heard many stories as to why that is critical. we know from jeff and others who worked in the first three years of life it is critical time

window when the ecosystem is established so you can argue if you care about pertubations that is really where we want to look because then we have gone even one step further back and that is looking at when the system is even more vulnerable and that is in the state of prematurity.

arguing that your gut and your immune system is immature and premature and so is your moreover, the reason we care in this case about preterm infants is because of the high-risk of infections. they just going to assault of our cohort of kids who are born

10 weeks too early almost every single one of them receive antibiotics for the first couple of days of life. the top two drugs these kids get are antibiotics. the top 4 drugs they get are the top 8 also antibiotics. so if you think about it, shear

san ecosystem that is bombing with things that will disrupt him. sewed my lab published on the first phase of our investigation of what happens to these kids guts in terms of their composition, who is there, how many as well as specifically

their antibiotic resistance genes across a set of 84 preterm infants. we had samples in collaboration with, during roughly two and a half months in the neonatal icu. we stratified this score in terms of kids who only had antibodies in first few days of

life and others. your suite of omics methodding and analysis, here are some of the general trends we found. so she confirmed that the gut microbiota of these preterm infants is reduced in terms of diversity compared to age-matched term infants.

but more importantly, the type of bugs that you see in the guts dominating preterm infants are not what you might associate with health. associate with being on the cdc's list of stuff that should scare you. and obviously you would argue

it's on a good toss have in a preterm kid's gut. and then we dive into this possibility of finding antibiotic specific effects. so one of things that molly sees is for very specific antibiotics, those that are dominant go up and down.

we can also extend this into the collateral damage in terms of gene enrichment. again, in terms of specific antibiotics, you can see specific types of antibiotic resistance genes enriched in specific states and then the model.

statistical associations not just with the microbiome data but all the rich metadate you have with these kids applies generalize linear mix model to understand how species richness, a measure of microbiota and health changes and able to identify three antibiotics that

clearly depressed the diversity in these kids guts and then also identified other correlates which increased diversity. age as well as maternal -- consistent with things you heard about today. and then finally in the cases where we got this really nice

sample before and after antibiotic treatment, was able to apply a more than learning approach where she was able to identify the two major antibiotics nudes this population just four variables, two antibiotic resistance genes and two bacteria that allow you

with 85% accuracy to really know what direction this is going. so early days but it has again the hope of personalized medicine approach to measure the impact of antibiotics. one more slide and i'm going to steal a few seconds of time. what we are doing right now

thanks to nih and cdc support, we have a nice picture of the acute pertubations but what happens later in life with these antibiotic treatments? so is with that expanded area, drew finished analyzing samples from 60 individuals where we have samples on two years of

life and added on a healthy court, no antibiotics to match healthy control. we now screened everyone of these samples with improvements with our functional genomics genes and also added on a culture incompetent to see whether enriched drug resistant

organisms persist aftera couple of years of life and then finally using models you heard about before, transgenerational humanized system where we can take these acute pertubations and test our computational models under scrutiny using mice.

so i'll there. thank the folks who have funded this and leave you with this. >> dr. flores: thank you. our next speaker is mahmoud ghannoum. his title is bacteriome and microbiome, polymicrobial interactions and define health

and disease: a new paradigm. >> dr. ghannoum: first, thank you very much the organizer, we appreciate everyone who asked me to come here so it is really great opportunity. what is really important, i think, i'm going to convince you that it is not only bacteria,

it's not only fungus, they work together to cause problems and that is how we are going to start talking about. so to start with, how do i know that? i want to tell you a personal story. in 1974 my supervisor gave me a

paper where rabbits treated with antibiotics developed fungal it became clear to me that fungi have negatively impact us and also they live in our body and they are sometimes very inefficient. and also it became clear to me that if you get rid of one

community, what is going to happen? the other will take over. and guess what? it was time flies. 43 years ago. so, i tried to -- with all the hype and everything going on with the microbiome, in 2010, i

wrote an article saying we need to look at fungi as well. i need to start working on this. so what happens, nobody of course listened to me. so, in 2016, i went again and brought an article in science and there i called it, the microbiome, which is really the

ignored kingdom n that article, we showed that if you look at 2015, that there were merely 800 articles, 757 of them were epithelium and viruses and i'm not biased against it. just remember we need to think of virus as well. but only 12 articles on the

fungi. so to me, i owe pined that it is really very important we should look at both in the same sample because this is going to help us how to really treat with the good news, people are listening and we are starting to see a number of papers where

they characterize the microbiomes different body size and i clearly want to look at the possibility about the gastrointestinal tract. so, i'm going to tell you the story of the interaction between the bacteriome and microbiome as it is told through crohn's

now, as you know, crohn's disease say multifactorial. so you have genetic factor. you have intestinal microbiota as well as the dysregulation. so three of them work together and it is well established and we heard the role of bacteria in crohn's and ulcerative colitis.

the first step of that is to look at the fungi as a port player in crohn's disease came from a group that looked at the dss models for the colitis and they found in the gut, it is colonized by 10 fungal species and by far the most predominant one was c.

tropicalis. their conclusion and not bacteria, is responsible for aggravating the severity and inflammation in these patients at fleet mouse models i started with groups of friends in france and we looked at crohn's disease patients.

instead of having crohn's disease and unrelated controls, we thought better to look within the family. so we looked at people with cd, their relative who lived with them so that -- and to try to reduce at least some extent the genetic as well as the diet

factor but also we look at other families where they also are healthy but lived in the same region. we only looked at the fungi, even though i told you i like fungus. but we look at bacteriome and bacteriome as well.

there is no particular distribution here or clustering. as you look at the microbiome, red and blue, and the c. and their relatives, they are very close together. so, in a way, if you want to try to compare the microbiota and people who are not related to

them, you are not really getting a lot of the picture. it will be better to look at within the family. and here is the heat map and you see the bacteriome, microbiome, similar really. the usual suspect as you expect. and the fungi you see a lot of

the -- one point which is lacking a lot in the fungal we don't have enough databases. this is one thing that we need to work on. but you can see about four different ones. if you look more specifically at taxa with bacteriome, this is

cd. the relatives and the other control group and look at how close they are, the cd and their members who are non cd. again, across all different taxa elements, the same applies for the microbiome and again they are very close.

so it's good idea to look at within the family as i mentioned. we did this here and this is the bacteria. these are the taxa. you can see between the cd and relatives, it's more spiral in the again the healthy relatives

and this is no surprise to anybody, it is really has been shown by others. when we looked at the fungi, you can see species. there is a trend in favor of where you have higher levels of this in healthy, also when you take all this together, there is

not all species there. you don't see is any difference. however, we found that they showed in their animal study that the increase was particularly higher in these this is all the significant work tropicalis is higher in cd another important observation is

there is a test called sas ca. this is usually used to look at cd patients as a diagnostic markers and the elevation. and that's what we saw in our study. but really what we saw interestingly is that as cawas correlated with the tropicalis.

we then went into bacterial and bacterial fungal interaction interaction -- - look at the colors. blue and red are correlated. when you look at the fungi, you can see for example, this is in cd patients. when you look at the relatives

there is no association between them. so then we looked at the correlation between these and other bacteria and we found that there is three different oganisms. whereas particularly significantly associated in

positive way and also especially increased. because these organisms in the gut can form as we started looking at the three organisms, tropicanis, and two other and we put them together. so tropicalis alone usually is yeast form.

you start with equilly or put the three together and look what happened to it. you have a lot of hype. as you know, the ability to form is a factor for this organism. so it's there. they form this and guess what? they start to have high

finnation where it's allowed to invade the epithelial. so, so we measured the thickness of this and you can see here it is much easier to see at ct alone and with ec and sm and you put the three together it's very, very thick using confocal microscopy.

so it forms a robust area. and this is some picture from this is ct alone. really beautiful yeast. this is with e.coli and surprised it was fused with can deet tropicalis. whereas different interaction this is the three together and

just fat on top as you look under microscopy. we looked at transmission microscopy and look what happens. this is the candida and this is all different organisms. you find that with e.coli, it's very near diffused whereas, this

produces filaments which bind not only to e.coli but also to candida so it is forming a bridge to stabilize and make it stick better. this is all in vitro? people say you don't have the environment of the gut and the so develop a model and this is

just recent data where we had cd no organisms really. you see mucus and epithelium and then look at the large biofen forming in the colon of the mouse plus the three organisms you have. so we showed both in-vitro and in-vivo.

now other thing, when this organism comes together, do they do anything with respect to metabolites? so we -- the question was, does this interaction alter metabolites and secrete it? so we did all of these like alone, medium alone and then

this sort of thing and different combinations and two and 3 together and look at what we found. it's really very, very interesting so this is medium alone. this is -- e.coli and then when you put the three together you

can see different metabolites are produced. so not only they come together but also they alter the metabolites secreted. and when we look at that in total, we found 299 metabolite detected and 8 of them were significant.

one of them was 150-fold higher than when you have them alone like candida and others. so this is say way in which we know we are looking at only with respect to how they interact between microbe-microbe and also how this is effected by the host so, this is a summary of the

model which is enhancing the gut you have mice home stays sis. no cell damage. you make this and then they start to secrete muse ins and you break the muse in and then fungi form these and start to in strayed and break the tissue. so this is our working model.

we published this paper and we didn't publish the metabolites but this is with respect to the interaction between the bacteria and microbiome. what is nice about it, it was received very, very good attention and encouraged us a lot.

the last thing i want to tell but this corporation, i don't think corporation only occurs on crohn's disease patients. another paper from university of pittsburgh where they askedny write the bio and guess what? they saw exactly what we are seeing in crohn's disease where

the bacteria and fungi cooperate together so we need to start to think about how can we manage how can we take this whether we use fungal, probiotics, because there is no way you can get these organs. and now we just did a study which is have we just submited

in cancer patients with where we are seeing similar sort of so bacteria interactions and fungi really spans different diseases. so to me, this is a, what i believe is happening, is that the organisms have evolved to form a comparative strategy to

try to attack really, benefit themselves as well as harm the fungi will gain factors and i showed you the experimentation. bacteria living inside the biobecome antibacterial and also they start secreting enzymes which guess what? they dysfunction of gut barrier

and really cause damage. the host is impacted negatively where you have under the influence of these organisms, you have components of the fungal system increase in the pro-inflammatory cytokines and causing damage and cell death. so this is the gaps, i think we

need to start looking at the microbial organisms and also we need to -- you have so much information. how can we harness this information through the human microbiome project how can we look at it and start to focus? to me studying -- is not the

way. not investigating all microbiota at once. for example, i put a grant and they say it is nice to have three organisms but you should study all the microbiota. i don't know what they were smoking but this is not going to

happen because when you put three organisms they tell,is very complex. you say, okay. but now put all microbiota. well, let's not go there. and now the need to develop multiple resources, imaging and in-vivo models to address this

microbial organism, define mechanism underlying interaction between the organisms and also between the organisms and the and also define how metabolites interact with these and this is i'm going finish with this last slide. this is the way of the future.

we always look at the mother, which is very important and to me, the mother is the fungus. and of course, we need to look at the babies as well. thank you very much. >> dr. flores: our next speaker is michiko taga from the university of california in

berkeley. her talk title is the decoding nutritional interactions in microbial communalities. >> what i'm most interested in is the molecular interactions that occur between microbes within their communities. so we are doing studies mostly

of single bacteria and one of the challenges here is to make generalizations about what is happening in the community-based on what we are doing experimentally with single and related to that, probably within the next year, we are going to have 100,000 bacterial

genomes. and also many, many metagenomes. so with all of this abundant sequence data, we are still limited by our ability to decode these genome sequences to make the most of the sequence data that we have to make generalizations of our

experimental data, and understand the bigger picture of what is going on in microbial communities. and so, one of the keys that i think will aid in this transition is to incorporate bio informatics using the sequence data we have when we are doing

these experimental studies using my crow biology, genetics, biochemistry, et cetera. b12 has a cobalt metal at its center. we are most interested in the lower ligand which for reasons that will become obvious soon. so core noticed are the vitamin

b12 family of molecules. and these are critical metabolites in bacteria. so, genome analysis tells us that 86% of bacteria use coreinoids for some aspect of their metabolism. only 14% do not use coreinoids. so in for example, the human

gut, coreinoids are important cofastors for a variety of different metabolisms. two-thirds of bacteria in human gut have a corinoid dependent -- methyl transferase. what makes us think we can learn something microbial interactions based on studying corinoids, the

majority of corinoid dependent bacteria cannot make their own corinoids. most bacteria have at least one corinoid-dependent enzyme encoded in their genome but they are unable to make corinoids. instead, these bacteria rely on corinoids produced by the

bacteria that do produce so to make matters more complicated, corinoids are structurally variable. and specifically there is variation in the structure of the lower ligand of corinoids and here i'm showing you some examples of different lower

ligands found in corinoids& produced by different bacteria. and then the important thing about this is that these corinoids with different lower ligand structures are not functionally interchangeable. so this means that bacteria that are in the community with

corinoids have to somehow obtain the specific corinoids that function in the metabolism. so, we see that multiple corinoids are present in various microbial community samples so these are corinoid profiles measured by lcns. so in these plots, each color

represents a different cornode structure and the height of each bar represents the relative abundance of those corinoids. these samples were done by bob allen and these were done by us. you can see multiple corinoids are present in different amounts.

and so, we know that there are multiple corinoids in these microbial communities and what we are interested in is understanding corinoid sharing and corinoid specificity. so breaking it down, we are interested in specifically, we have done a lot of work on

specificity in corinoid biosynthesis. so whatta is it about the genome of these bacteria that make it causative to produce specific corinoid molecules? we have also looked at corinoid transport in collaboration with andy goodman.

there is also specificity in the modification enzyme for the upper ligand which i won't talk about. as well as modification of the lower ligand and cornode dependent regulation. so first the lower ligand modification.

this is one example of a strategy that a corinoid dependent bacterium uses to get the corinoid that is function in its metabolism from the environment where there are multiple corinoids. and so we found can only use 3 of the 9

corinoids we tested and these are just the lower ligands that we looked at. so it's very particular for the corinoids it needs. but it can take up any corinoid as long as it has also a lower ligand of one of these available and it is able to remove the

lower ligand it cannot use and attach the lower ligand it can use to produce a corinoid that functions in the metabolism. so this is one strategy of one bacterium showing how it can get the corinoids it needs. so, now i'll talk about some unpublished work on what we are

doing studying specificity in corinoid dependent enzymes. so we are using the enzyme co-enzyme a mutates as a model enzyme because it is very widespread in all three domains of life and one of the two corinoid dependent or b12 dependent enzymes that humans

use and this is the structure of the human mcm. so, we know that corinoid dependent enzymes from organism to organism have different specificities for different corinoids and we want to know the molecular basis of that specificity.

so we are interested in figuring out specifically in the lower ligand abiding pocket what is it about those proteins that make the corinoid dependent enzyme bind and use specific sets of corinoids? and so, olga who is say graduate student in my lab, is looking at

mcm orthologs in-vitro and one of the assays she is doing is binding assay that is dependent or based on the question of intrinsic fluorescence of the protein. so, she looked first at the mcm enzyme. and this organism normally

produces you can see from these fluorescence binding ass says, that the mcm has highest affinity for native corinoids we can use these curves to calculate the binding constants. a structurally very similar corinoid slightly lower affinity

and corinoids with lower ligands bind with lore affinity. it's one mcm. showing the binding affinity of the protein matches the organism producing. and also looked at two other orthologs and found the same pattern where the corinoid

produced by the organism is the same one that has highest binding affinity for the mcm. so using those kinds of examples, we are trying to do or use bio informatics to identify the specific sequence motifs in this binding pocket that confer and so in this alignment, we

separated the sequences from bacteria that produced or are thought to use corinoids with the ligand like b12 versus those with pewine lower ligands. you can see covariation in these residues highlighted in the structure here. so we are -- this is very new so

i don't have data yet but we are doing mutagenesis in looking at the affinities of mutant proteins for corinoids based on this kind of analysis. so, finally i'll talk about regulation by corinoids. so, many corinoid biosynthesis genes and transporter genes are

regulatessed by an rna element called a riboswitch. so these are -- so in general, riboswitches are regulatory elements that when the ligand is not present or the ligand like for example corinoid is in low concentration, the rna is in the is able to be expressed.

when the ligsand at higher concentration, it binds the rna and in this confirmation, terminative hair pin is allowed to form. or a hair pin that is the ribosome binding site. in this confirmation, the downstream gene is not expressed

and this is the way the bacteria regulate expression of genes for example, like a said, for biosynthesis or transport of ligand. so we think that riboswitches 14 have specificity for the corinoid that is function for the organisms so that they shut

off gene expression only when the functional corinois are present in the cell. so, another graduate student is at the beginning of this work where first he made riboswitch reporters with m cherry and expressed them in e.coli. and we wanted to know first of

all, is there specificity in riboswitches for a different corinoid? so i'll show you data from two riboswitches. this shows you that different riboswitches have different corinoid affinities. and so, as we increase the

concentration of corinoid you can see that this corinoid and these corinoids are able to robustly shut off gene expression of the reporter. and that these two riboswitches have different corinoid specificities. this is kind of curious to us

because these two riboswitches exist in the same cell and they actually coregulate the same genes. which i don't have time to talk about and we don't know the reason anyway. so, those are some of the examples that we are using

experimental microbiology and molecular biology and biochemistry, to study speckity in corinoid -- specificity in corinoid biology. getting back to these critical challenges, what we hope in the long term is that we will learn enough about corinoid

specificity from these experimental studies that we will be able to use bio informatics to -- in conjunction with these experimental studies, to be able to predict corinoid metabolism. and so, the sort of long term dream is to be able to look at

any bacterial genome and be able to know which corinoids are needed by that organism for which kind of metabolism. and if we are able to do that, we don't have to do experimental studies on every single organism. and instead, we can use bio

informatics and even long-term, we hope that we can use that knowledge to some day perhaps create mechanisms to modulate a microbiome based on some kind of corinoid base manipulation. so, with that, i'd like to thank the members of my lab who contributed to this work as well

as the collaborators that i mentioned here and i also want to thank the nih for supporting the work that i talked about and the work in my lab in general. >> dr. flores: we have one minuted now for q&a. so our last speaker in this session is forest rohwer from

san diego state university. his talk title is how phages create an immune system: bam immunity and transcytosis. >> dr. rohwer: sorry about that. i have a horrible voice. so what i'm going to do, this is my favorite system that i study, which are coral reefs and this

is ma lin yum island. and it's about 7000 feet of calcium carbonate sitting on the top of a mountain under the ocean calfs built by the chorals. and chorals are pretty amazing because they are actually not only do they build the biggest

structures ibut are also immortal and some of the first animals ever to evolve. how do that do i this? and how do they interact with their organisms? this is where i normally work but then i got drug into the world of feces by this fecal

matter fanatic and he can sell you anything. so somehow he convinced me to look at fecal matter and this was at an older study where basically we are building off the twin study, you know about. where jeff and that group show that people who are related, in

particular in families, have very similar bacteria in the and we just did our typical prep where we woo can get the viruses away from everything else and then we sequence them and just ask one, do you have the same viruses as you travel through time if you're a person?

and these are almost all phages. and two, how different are they per person? and what is interesting about it is, we know for the most part the viruses you have today, you'll have later. so you'll carry them through and the other thing that is

really cool is that every individual actually has very different viral communities and particular phages. that is particular whether you're a person or if you're a coral. all right. now the other part of this story

is the mucus. and what is interesting about mucus is that it creates these exchange surfaces that have to be renewed so you can get rid of nutrients or exchange nutrients or waste products, et cetera. and it is actually a glycoprotein that really is

different between everyone. and that is because there is kind of a combinatorial chemistry thing that goes on to add the sugar groups to the protein itself. there is all these -- think about mucus but remember that it is essentially individualized.

all right, and the starting point here for this study is that we have been looking at viruses or viral and microbial communities living on corals and we would go around and suck off mucus and count the number of bacteria and viruses and compare that number against what we

found in the water column. and then we started doing it on other surfaces because we got these wierd bacteria or phage to bacteria ratios. by wierd, i mean that they were elevated. so we have about 40 phage for every microbial cell when

looking at a mucosal surface. and that is very different than what we see in the water of the lumen where you will only have about 10 phages for every microbial cell. and you can imagine that since phage for the most part, they are not actively hunting, so

they are not able to use anything do move, their interactions with the bacterial host is driven by mass action. so if you push up this number, you're more likely to run into a so this is interesting. this was our sticking point. we started doing things like

this where you take tissue culture cells with and without we add t4 phage, this becomes important later -- and we would wash it across the surfaces with and without muc us and then we would take or put e.coli over the top of it and then count the number of plaques.

what you find is that wherever you have a mucus, you would get the phage to hold on to the mucus layer and it actually protects underlying epithelial cells against the bacteria. so this gives you an antibacterial response. and so is this is what we

started calling bacterial phage adherence to mucus or bam. and it's basically an acquired immune system because what is going on, any time you pick up or a microbial community comes in, it will bring the phage, and the phage gets stuck in the muc us and they start killing

bacteria because they are higher concentrations. the question is it specific or non specific? and this is when we got really lucky. we got a great paper by a lab that came out in 2012. and what rick's group did, they

took a bunch of fecal phage communities from humans and sequenced very deeply and they found that there were these reegence that were hypervariable in the verio. and these are things that we know a lot about because the way they are made is that they

actually use a reverse transcription mechanism to induce basically an alanin scan. and you guys have heard of different versios of them because they are used things like this are used in phage display systems and so forth. what is really cool about them

is that am they are one, hypervariable and maintain the original code, why they become hypotheticaller variable. they have all this massive diversity of about 10-13th possibilities which is close what antibodies and tcrs do and what was exciting about

rick's study is these guys were ig-like. so they looked like immunoglobulin to me, this hypervariable. which is of course if you are an immunologist, gets you all excited. so here is the capsid of the t4

phage and this is one of these things that is related to these, which is called a hock domain and they just stick out on the head. and they have high affinity because there is a lot of them. so our working hypothesis is phage will be adhering muse in

through these hypervariable things because the muc as is changing. so this is the same sort of you just have the phage plus or minus the ad-hoc domain and if you knockout the hock domain, it doesn't stick to the mucus and if you leave it there it does

stick. so, when we get left with through this and a bunch of other stuff, is that basically all these mucosal surfaces have these phage living in a particular place in them and they are sticking to them and you're acquiring them with your

and they are specific because they are adapting to your mucus and they are also specific to the bacteria they are killing. i just put in this in here to show you as you guys, we have so one of the things that is hard in the microbiome so far you is don't have very good

dynamical models of the system. this gives you a fill of one approach that we use where we make these microfluidic systems and we have epithelial cells producing muc us and then we'll run different concentrations of bacteria and phage across this and then we just measure how

many bacteria we get as well as cell survival. and then we put it into these models and these models become really important because if you want to starts controlling these systems you have to have a model like this to work against even if the model isn't right.

in this case, we are dragging everything by this diffusion and it turns out this is a very exciting parameter but i'm not going to go through it. but if you take this, you actually get why this system works. and that's because it's an

emergent property of just the dynamics itself. if you're a phage living in that muc us and you're holding on and your bacteria is coming in like this, you're much more likely to find a host. about 15 times more. so that is a very nice selection

pressure. and it is bad for the microbes. so if you're trying to get in and you're running into this reeve of phage, you get killed off at about 14 times more likely to die. so you can see is that these guys are sitting there in the

mucus and they are effectively serving as a barrier to protect the epithelial layer underneath. so, what about inside the body? so we have seen this stuff with the mucus but there is old observations that there are tonse of phage in your blood. so we find them in blood

whenever we do a blood -- so that sudden one of the newer questions we have been looking at and this is how we have been doing it. so this is a transcytosis assay and what you do is you put the cells -- these kidney cells that will polarize when you put them

in these chambers and so you get a basal side and you pult phage on one side and you see the phage come through. and as put them on the basal side, you see they go this direction. what we always find is that the phage move from to the basal

side. they don't go the other and .1% of the phage are functionally transsighitosed and i mean that so one out of 1000 make it all the way around there killing bacteria on the other and we know that 1% of them are endosighitosed but most die

within the cells. it's fast so we see close of saturation in about 30 minutes and we know that it's relatively concentration independent. so is it doesn't look like there is a specific receptor for it. we asked if they are getting through and they do and it's

always in this direction. so what does that mean? well, if you just make a rough calculation, there is going to be about 30 billion phage events per day in the average human. and that is roughly equivalent to how many white blood cells so you have got as many phage

running around there that can potentially kill bacteria. our working hypotheses is you're constantly sampling this and you're getting them into your lymphatic tissue initially, and then this may serve as an acquired systemic immune system remember something like a coral

that doesn't is have anything we would call an acquired system. this may be how they are doing the other thing that i think is important for as we move forward, is that actually this is probably a tokerrization thing so an unusual thing about phage.

so if i inject them into you, you'll develop a massive immune response to them. but if you are getting them through your gut, you usually don't get an immune response and it's probably because this transcytosis is taking them directly into the lymphatic

system without a danger signal so we are tolerized to our so here are the conclusions that we have these two different versions of this model, which would be innate and specific. they are -- we think it is important for individualizing the my crow by onlies.

we know that the tuning of these proteins that stick out can be used to put in particular places which should be important for drug delivery and phage therapy. we know that phage are also common throughout the body and may be serbing as a systemic immune system that nobody ever

seen before. there is tons of people i highlighted in yellow that people are in my lab that have been working on this project and of course we have the math group and all of these people, including gordon's lab and littler who have been great and

very supportive of us doing some human stuff. these are the people that paid for various parts of it and then, i'm just going to -- supposed to give you something. so you guys need to study phage. you guys won't listen to me. i give these talks and there is

also a token phage person invited to every microbiome meeting and everybody goes those are so cool and then you forget. you should really study them because especially if you're young people. they are really just -- there are more of them than anything

else out there. most of the genetic diversity is actually encoded in the phage. we know that most of the unknowns, like when you look at a vary ohm it -- like we get them out of lungs that is 99% unknown genes. so these are just gigantic

things that we know nothing what we do know the genes, it's usually something cool like antibiotic resistance or pathogenicity or something that is being moved around. in the human, as far as we can tell, every cell, your cells as well as your microbial cells are

lysojens. so they are all carrying temperate phages. and there is probably this new immune system. and finally, this is a field that is really suffering. so we have lots of phage biologists with lots of

knowledge and then they got pulled into other fields really under -- nixon was the war on brought these guys into cell biology and stuff and so we really are losing a large knowledge base at this moment. so it's a good place to be if you're a person and i should

mention roger died monday. >> dr. flores: a fascinating topic. i think we can start the q&a session. and. of pennsylvania. thank you to all the speakers. i thought it was a really

interesting session. i have a couple of technical so gautam dantas you're aware of the controversy about antibiotic resistance genes and their transportation on phage. traditional view is not very much and then a paper saying maybe antibiotic treatment

causes resistance genes to prophage and then they move around like crazy and then there was a much less prominent paper published pretty recently saying the annotations were probably much too permissive in those studies and probably lots of other studies and if you look

really carefully and do it well, then you see much -- way fewer so, what advice would you have for us in this area? what is best practices? your view? >> dr. dantas: a great question. i think it's not just antibiotic resistance t just happens to be

a selective feature. but in the microbiome field, annotation creep is one of the things that hampers the ability to make inference. so antibiotic resistance, we tried to advocate for a slightly better way of doing it in terms of taking protein evolution into

accountability, for instance models as opposed to your standard -- most people annotate by blasting database and the gene you hit is probably 60% sequence identical and then something else that someone did one biochemical assay in the 1970s.

right? so i think there is an opportunity to use more sophisticated annotation and you have to be lucky with resistance highly selectable. i think we just sneed to do more biochemistry. additional assays on genes to

move that they do what they do. so i agree. i think it's a very legit concern that many of these studies are fraught with over annotation of function. >> audience member: and then, in a different field, dr. mahmoud ghannoum at penn we have done

various profiles of microbiome from iv is d subjects and we seen fungi that were notable in a couple of studies but they were different from yours. yours seemed candida one of the candida species. we saw some examples of candida but different species and then

also some other species as well. and looking over the literature, there is starting to be a few of these papers now and your hard pressed to find any two that are calling the same fungi furthermore, there is the complication of the fungi neighbor much more commonly

transient in food than we usually see in bacteria. so masking effect or danger of getting fooled. so what is your take on all of as far as i can tell it's up in the air as to which are the most important fungi though a lot of people seem to be calling them.

>> began began there are a couple of groups working on this area. >> dr. ghannoum: and usually at least we agree all of us that sack row miceis is there in healthy people. they see candida and we saw tropicalis as i mentioned.

i think it does not mean that candida is not also present in because as you saw in one of the slides, you can -- it did not come out to the same level of significants as candida prop callis. the other thing, i tried to point out it is different

control group. when you have people who are not related and compared to those who are ran the same family having the same -- live together and the genetic animal mucosa, that is where we found the difference compared to the others.

so the differences could be different control groups that are being used. now, the other thing is with respect to the other fungi, i agree with you that a lot of -- we eat a lot of mushrooms and other -- cheese and these sort of things.

so some of it could be transient. usually we try to look at co-microbiome to try to eliminate these and i showed that this type of thing will be published where we looked at a fungi that should be present in 20% of the people, at least the

sample size, which controls it. now, the proof in the pudding is in the eating as you say. we took the next step and we are starting to see that this is what happens when thisanch mall so hopefully if you have enough -- animal model -- we can going back from like host to

theilable back to the host but i think exactly still a lot to be done. >> audience member: marty. i want to ask forrest your talk was fascinating. and as you know in the oceans, phage are involved in the kill the winner biology where there

is an advantage, it selects for a diverse population. that's an open ecosystem. how about in the coral or in the human in those what we think are more closed ecosystems? with your diffusion constant, is that suggesting that we have got kill the winner behavior there

too? >> dr. rohwer: yes. -- no. i think what is more important in these systems are actually wha we call piggyback the winner, which is the temperate basically forming a mutualistic symbiosis with that bacteria,

with microbiome bacteria on the corals and in the gut. and they are bringing in functions of course to enhance their host, right? because they are so crammed together under those conditions. when you get this behavior, they can sweep through the whole

so really the viruses and the phage are bringing it anti-other phage mechanisms as well as things protect them against things like the coral system, which is probably where we get virulence factors from. that's my guess. >> audience member: so thesis

are mostly temperate phage on the coral. >> dr. rohwer: so on the surface it is more temporal phage and in the gut it is essentially yes, more temperate phage and all of the cells are carrying pro it doesn't mean -- you definitely have lipid phage

running through the system but not as much as you get in the marine environment. >> audience member: brandon. this question is for dr. fowler again. i have a hygiene hypothesis question for you. so there is phage everywhere

obviously and a lot of these are in source water and things. we have a pretty interesting habit of sterilizing or at least treating much of the water we consume and most western cultures. do you think that there is some contribution to our microbes or

dysregulation between host and microbe due to the fact of consuming this purified water and not getting more interesting diversity of phage in our diet? >> dr. rohwer: no. i think we get quite -- i mean they are everywhere. i think we get a lot of them --

we know they are in the breast milk so we know we are getting them at the beginning. you get them of course off the skin of your mother and all of yes, we get a lot of phage. the interesting event is i think just driven by what you guys already know.

when the stomach actually really becomes acidified and closes off during the weaning time, those phage seem to get locked in to your gut at that point and that may be why we have such individualized verios because you're mostly stuck with those viruses.

>> audience member: so you believe there is no integration once the gut is closed off? my question is for forrest. so forrest, if i'm understanding correctly, i think you're implying a little bit that the host mucosa actually recruits phages in a way.

have you looked at muck tow mutant mouse or any condition where the mucus is degraded and seen whether this change in homeostasis of the phage? that's one question. the second is, is there an ig response friday host to the phage response from the host to

the phages. >> dr. rohwer: so no, we have not done those experiments. that's all jeff's fault. he is supposed to do everything with the mice. so we have actually done nothing along those lines. and i don'ts know of any

and what was -- >> aid aid uga. >> foretorso no, for the most part you don't get -- so you can get a strong improve response if you in just a second but when we are goating them from our blood, we did this study a long time ago because we have all these

phage we kept finding in blood. and we looked for antibodies to them in the blood and we never found any. even though we had tonse of phage in the blood samples. question for the panel. given the problem of antibiotic resistance what is the viability

of phage therapy in conjunction or instead of antibiotic therapy and where we are in the field? >> dr. rohwer: so phage therapy in theory, could work. in practice, it needs a lot of science. it's a very -- anybody that has ever tried to raise a phage and

bob knows they are really finnicky and you have to get the right conditions from the bacteria to grow. i mean, there are many examples of where we have phage that we cannot grow in the lab but we can grow in the mouse or vice versa.

we can grow them in the mouse and not in the -- yes. in the lab and not in the mouse. so you have lots of this going on. and that has been really to make phage a really viable option for antibiotic resistant treatment we will have to invest in the

same way we have in drugs, to tell you the truth. and nobody is willing to spend that money so far. >> there are a number ever other alternatives, biologics that people are considering outside of phage in terms of being adjuvants to antibiotics.

we heard of probiotics a little bit. that's a big deliverable where if we recognize that infectious agents want to occupy, you could potentially knock them out by probiotics and prebiotic combinations and then there is also just reversing the entire

paradigm of how we go after infectious agents. we used the warfare analogies. the thing we learned from the microbiome it's probably a little bit more cross talk and dialogue between microbes and if we adopt that then perhaps we can use things like symbiotics

where perhaps more targeted therapies for instance go after virulence factors which will have resistance but not the same burden as antibiotics. >> i must agree that in terms of for example, as you said, the probiotics, if we can think of ways to selects the strains that

are able to antagonize the bad bugs or resistant organs, that could be a good way to do it. and i think there is some progress in that direction. >> dr. flores: i think it's time to feed our gut microbiota. so thank you for your presentations.

>> my name is cindy davis, program director in the office of dietary supplements at nih and it's my pleasure to welcome you to this session focusd on the intimate relationship between diet and microbiome. the purpose of the session is to highlight current knowledge on

the directional influence of diet and dietary components on microbial and host metabolism and how the metabolites influence host physiology, what is still not understand about microbe interactions and human health disease and what is needed to make progress in this

we'll have five speakers. the first will be justin sonberg from stanford -- sonnenburg trillion stanford university, leaping back to build a strong foundation and dietary interventions in humans. andrew patterson, penn state university will talk about

intestinal epithelial cell receptors as modulators of host communication. gary wu, university of pennsylvania that will talk about diet, gut microbiome and metabalome as a therapeutic probe in ibd. nicole koropatkin will talk

about how gut bacteria eat your veggies. and johanna lampe from fred hum cancer certaintier testing diet gut microbiome interaction use of controlled feeding studies in humans so my pleasure to introduce justin sonnenburg. >> thanks.

i want to start by thanking people associated request nih and beyond that put together this great workshop, it's been spectacular. one thing that's perplexed me relates to the conversation yesterday about wherables. mike schneider my colleague

mentioned at the time of his talk he was actually wearing eight wearables. he went on to show all kinds of microbiome data and nobody asked him where he is wearables were. and we heard about the toilet of the future that monitor the microbiome, we can cobetter than

it may have to do with repurposing, many of you may remember google glass, it was sidelined buy bigoogle, i think there's still potential here. with a slight tweak to both design and name. this is a real time microbiota sampling device.

may not be the exact design but it's combing. so enjoy your freedom and court now. on -- comfort now. on to the real talk. so critical gaps in the field, title of my talk is leaving limbo, we understand the

potential of the microbiome and the field right now is grappling with how to translate this. and i worry that we have left the basic science a little too rapidly and a little too far behind. so i think one step of limbo needs to be back towards basic

mechanism, we need imaging more genetic tools, use of other model systems like with heard about in this meeting and funding specifically for non-disease related interactions in the gut. i think the step forward is to study humans more particularly

with dietary interventions. manipulations are safe by and large for humans so we can do experiments right away, we have seep great success in the field in publishe study and the preliminary work we saw published yesterday and so we need to think about embracing

this approach and going wholeheartedly developing humans as a model system for studying microbiome and its importance in human biology. my lab studies the dynamics of the gut microbiome in response to diet and we particularly are enamored with the microbiome

fuel that feed the microbiota. we use this term to distinguish carbohydrates they utilize versus the once ones they don't utilize as they pass through the we eat a high in sugar fat, leaving the microbes starving. we heard about microbes turning to eat us in as tense of

sustenance from the diet. insighting inflammation, doing bad things. if we eat complex car boy hydrates that feeds the microbiota because we can't degrade the carbohydrates in the in the microbiome it's clear one of the primary professions is

degrading complex carbohydrates. one of the gaps we have been trying to fill is one in imaging. i think it's important to recognize how important spatial organization is in anything. this is an image taken from an artist and comedian who likes to

organize things, in this case organized the components of fur branch you can see the little branches and needles here. this is similar to performing a 16s analysis and you can see how much more information is gained when you have the spatial we have been working very hard

to develop tools to enable the field to move rapidly towards studying spatial organization of we recently reported software and saw host microbe enable people to perform quantitative studies in imaging. we have also developed fluorescent backroidees species,

each glowing with a different fluorescent signature. in this case embedded in a piece of plant cell wall material within the gut. these tags allow us to differentiate the species and help -- hopefully help propel understanding spatial

organization. we are interested in incompatibility with the human genome and the microbiota. the human genome and microbiome with asymmetric plasticity over evolutionary time and it leads to the possibility that when pressed by selective forces,

things like extreme dietary changes, we have moved to a very low fiber diet in the industrialized world, introduction of antibiotics we may end one a microbiota than the human genome has come to expect and perhaps shaped the human genome during our

evolution. so it will be wonderful to have a time machine to sample the microbiome in different parts of the planet. we can't get the approximate i mr.cation funded. so what ewer doing instead is studying traditional populations

of humans. these are the hadza of tanzania, the last full time hunter gatherers in africa they spend all their time hunting and gathering food they eat each day. they're continuation of humanities ancient past, the modern people serve as modern

approximation of what our microbiome looked like during 190,000 years of our species and perhaps longer. this is them sitting around a fire, they have fibrous tumors, very different than the ones we eat. very gracious to participate in

our research study and we have been sampling them over time. we see patterns of seasonality coincident with seasonality of foods they eat. great diversity in their gut microbiome and new taxa we don't see in the industrialized world. so wonderful former student who

not only generated the data and analyses it, brought together as much data as he could from industrialized populations microbiomes and put all that on to one pico a plot. that's shown here. each dot is an individual microbiota and he compressed the

data from the first principle coordinate and provided a bar plot of different populations. what you can see immediately is a stark pattern all the traditional populations over here, the industrialized populations over here. so it does at no time matter

whether in new guinea, or africa, if you live a traditional lifestyle you youhave a microbiome certainty to their populations and very different from us. so is it's not just diversity thing. if we actually map abundance of

taxa on to pico 1 and look at the fractional representation we see groups like spiro coats, greatly ebb enriched with the traditional populations. alternatively in the industrialized populations what we see is a bloom of veruca microbeia, mucin loving bug.

we can get the same bacteria to bloom in mouse experiments if we deprive those mice of knacks or dietary fiber. so this is -- we're reconstituting a finding here on a global microbiome scale we can see in mice by depriving them of fiber.

what does this mean for us the fact we have lost species and gained others we have -- we likely shifted our microbiome recently in evolution. what does this mean for biology how did we lose microbe? data probably play ad role. we have undergone loss of fiber

in the western diet and we wanted to test this in mice. so we gave mice a human microbiota and gave them a high fiber diet and split into two the low mack group expected the diversity would crash over course of the two month period and ask the question how many

lost species could come back if we reintroduced dietary might be it is too late for us or if we started eating better could we recover our microbiota. we did see a lot of species come back actually unexpectedly, not all but a lot. we thought maybe this is a

multi-generational affect. and so we repeated this experiment over the course of four generations, this is the high mack diet control microbiome data from the 200 most abundant species over four generations so each row here is a different mouse at a different

time point. stable species more variable and some species here go he can tenth over the course of four generations but what happens on the low mac diet? huge decimation of microbiota. so each generation we see this loss of diversity and very

interesting pattern. so we see actually an early time point here, microbes that we can't detect we wean pups to the next generation and put the parents back on a high fiber diet and some continue back but those are not passed to the next generation.

that happens at each generation so you press microbes to level they have lore probability of being passed to the next so elimination of dietary fiber depletes the microbiome. i want to return to the data, one of the interesting things we saw was the seasonal variation.

we saw a cyclic pattern of microbiome configuration in the hadsa so as food disappear, microbes disappear and the following season when that they reappear the microbings disappear. this is the data from the eight individuals surveyd over three

seasons so this is dry wet dry season and 2013 and 14. and what we see is prevetella spirochetes species disappear and reappear the following season, that doesn't happen for the seasonally stable species. what is notable to how similar to the emperiment.

p what would happen if we lock the hadza in the wet generation diet for multiple generations would the species disappear? also notable is the seasonally roll until specieses are the ones that correspond to those that have been lost in all of so could this be a diet driven

effect. this is our hypothesis that diet depleted microbiota. is this meaning for biology? we don't know. should we recover the species are they help testing a, b, c, d, another test for the captioner.?

do we -- can we do it with diet, do we have to have deliberately reintroduce them? these things to study over the coming years. we know our we know western diseases are on the rise. could these be linked. microbes in our gut control the

immune system and most western diseases are driven by low levels of infa nation. could it be we're sitting in this room with similar information driving us towards one of these diseases because we have the wrong gut microbes. that's the hypothesis.

so we want to test this and we have started the center for human microbiome study at stanford to do dietary interventions and starting to look at these questions. idea is fairly simple enroll a cohort of people perform an intervention and look who

responds to whatever you're looking at. perform omics data, for the immune system, for the microbiome over the course of the experiment and then identify the predictive elements that tell you which individuals at baseline are likely to respond.

use machine learning to identify predictive elements. do this again and again. and you end up not only with precision interventions for these individuals to deal with whatever they're dealing with, chronic inflammation, obesity, autoimmune diseases.

but what you also get is a road map for what mechanisms we should be studying. so we have traditionally thought about mechanism to animal model to human translational path we have to think about humans as the first step just because of the sheer complexity of the

why not let studies in humans guide the priority list for what mechanisms out of this infinite list we should be studying. this is approach that definitely works, there are great publish studies we have seen more studies presented here where it's clear longitudinal studies

in humans if allowed to collect omics data are really effective way to hone in on important biology and the gut microbiome. so we're starting this. we ran a pilot intervention over the past year, this is a high fiber high fermented foods two arms.

we enrolled somewhere in the order between 40 and 80 people, high fiber, we got some between 60 and 80-grams of fiber from plant based foods over the course of the ten week people eating 6 to 8 helpings of fermented foods trying to load them with probiotic like

microbes panned over the course we're performing microbiota profiling at the time points, performing immune profiling using things like 370 immune parameters to see if we can move the inflammatory status of the and also performing clinical behavioral cognitive tests along

the way. this is a collaboration with christopher gardener whose group performs dietary interventions and we're collaborating with mike schneider mark davis and several others at stanford to look at different cohorts and you can different tech nols to

monitor these cohorts longitudinally. so i just wanted to finish with a slide of something that i think we should think about. i think many of us in the biomedical endeavor think about solving mechanisms to create drugs to treat diseases that

something to spend more time on is get people to eat differently to prevent diseases that are completely preventable. this is doable, it will be difficult. this is shown by hadsa food preferences. they eat a hugement of fiber not

because they want to but because plants are abundant in what they can get every day. if you ask them what they prefer eating, it's meat and honey. those are the two foods they would eat all the time if they could, put them in restaurant or convenience store they make the

same bad decisions that we make. so how do you get, this is a beautiful picture that shows a hasma man holding a piece of honey comb, he had to get stung to get the honey comb so there's a cost to getting the sweet substance. they can't get meat every day so

it's a different ecosystem and the question arises how to get people to eat healthy in afro society. policy education, food technology, addressing the farm bill, things like this. that data should feed into and be cross felterlyze by nih.

finally i want to acknowledge the people that did the work. sam smith and ericssonburg did the microbiome work. synthetic technology and glowing bacteria. kristin did the imaging and steven does a fantastic job of running the mouse facility.

we have collaborators that drove the hadsa project. thanks funding nih and stanford and just to mention my wife and i wrote a book about the microbiome, if friends or relatives are interested in learning more. thanks for your attention.

[applause] >> our next speaker is andrew patterson who will talk about intestinal epithey willthelial cell receptors as modulators of host microbiota communication. >> thank you. i would also like to thank cindy for organizing and also to the

organizers of the general, i really had a great time and got to meet interester interesting people and see interesting so today i want to talk allot will bit slightly different topic a little more mechanistic understanding how the microbes perhaps interact with the host.

and in this case by understanding the small molecule that exchange that occurs in the various systems. before we try to think future it's good to step back and think about the past. this is a quote taken from pharmacologist from 50 years

ago. where they began to appreciate the understanding or have a good understanding of microbes couldn't themselves contribute to potency or efficacy of drugs but didn't have a good understanding or way to measure quantify these microbes and what

they were doing to manipulate these effects. so going forward we have developed a lot of really great tools that have given us this opportunity to do this in a much more mechanistic scale. so seen lots of excellent examples of sequencing based

approaches, excellent model system, but i think one thing to emphasize, i think a place metabalomics have a strong foothold is here in the microbiota community, trying to understand the products or the metabolism of and put more broader view how this health and

disease and metabolic phenotype, this is not made possible without significant bioinformatics capabilities, that's impressive display i have seen over the last few days how important this is driving this peeled and it's more important to convicts people to move away

from tagc and focus on carbon hydrogen oxygen and phosphorus. so anybody out there willing to do that, i think it's a ripe opportunity. this is giving a an opportunity to understand the chemical dialogue between the host, this is signally mechanism signal to

microbes vice versa. a fewment examples so when i started thinking how to design this talk or what i would talk about is something that i think is really close to post-doctoral training, understanding the signaling pathway that occurs between the liver and the

so we can think about it in terms of exchange of nutrients, drugs, environmental chemicals, but importantly we also see a lot of examples bacterial metabolites signal back and forth between within portal circulation. within these organ systems we

have a network of sensors or receptors that are there to signal these changes to the liver. i broadly classified them as metabolic sensors, you see them enriched in the liver and gi and they are very important for mediating signals back and forth

between organs. over the past several years we appreciate a lot of the ligands generatedd of microbial originser generated specifically by microbes or metabolized by microbes can modify the activity of these receptors and it's becoming more a parent these are

clearly there to sense and signal these metabolic changes from changes in the microbes. so i want to focus on two, there's not a lot of time to but i will talk about the farnesyl x receptor and end with a story about the hydrocarbon receptor. so fxr it's a great example of

the impact of the microbes with the hepatic circulation, this is a process of exchange of material between the liver and the gut and back and forth. yes think about bioassay metabolism, produced in the liver from oxidation of cholesterol secreted after a

meal to solublize or mobilize dietary compounds but i think we're moving beyond this appreciation this that these are just detergents when bioassays are serving more as important signaling molecules that allow host or microbes interact or signal back and forth between

each other. a lot of data from several decades really understood that this process was dramatically influenced by gut microbial processes and it's becoming more apparent once we have had the development of mouse models that allowed us to investigate more

carefully. so with if,xr we have been able to appreciate this change that hock curs in bioassay metaboism using mouse models that w can manipulate fxr as well as manipulate the microbiome so one example how they're transformed by bacterial

processes, in this case i have shown an example of hydrolase activity of lactobacillus for clostridia, as well. this is a common process that occurs on the row dented model as glycene so more common to humans. bioassays and remarkably in

spite all we know about that we don't know necessarily what benefit it provides to the microbes itself, it could perhaps provide tolerance to bioacid pools secreted, could help microbe scavenge nutrients in doreen as a sulfur source. furthermore it probably more

than likely helps them maintain gi persistence, those that don't have significant biohydrolase activity don't have good gi persistence. so when we began to look at this carefully we did study with antibiotic. i have shown free radical

scavenger and we found that there was a dramatic change in bioacid pools with these compounds that we treated with. and we saw a major shift from unconjugated to those more con conjugated. this gave the idea that perhaps the drugs we were thinking

regulartively innocuous were manipulating the microbiome as we found with temple. so what we saw as dsh activity dropped we saw increase in t beta mca levels or doreen conjugated bioacids and this had a tremendous impact on fxr activity so this was

demonstrated by our lab along with frank gonzales as well, elegantly demonstrated by -- group as well. these compounds, bioacids typically can be can serve as agonists of receptor, shown on the left i show reporter icey increased fxr activity with the

bioacid tca. but you note when we treat with t bit the fxr activity goes down suggesting that indeed this is probably some way manipulating fxr activity and if we look at target gene activation in the intestin, we see fxr activity is reduced.

not dramatically knocked down to zero but we see 50% reconduction in activity. -- reduction in activity. so gone on to demonstrate t beta mca as well as variety of other bioassays themselves serve as potent fxr antagonists so this helped us appreciate this

dramatic affect occurring with massive pool mixing around in our guts and if there are ways manipulating this pool one way or another, perhaps we can use this as a tool for or a means to modulate fxr activity in vivo. this sum rideses a loot of work into one flow diagram but we

found with bioacid metabolites intensifies fxr signaling and it showed the affect was profoundly observed in the liver as well. so we found this was really beneficial for modulating fatty liver disease, as well as gnash, we found observations related to insulin resistance and obesity

and remarkably this appeared driven through ceramide signaling which i won't have time to go over today. so i want to switch in the last five minutes to a different receptor, not a typical nuclear receptor, this is a protein, hydrocarbon receptor typically

or long time identified as being the dioxin receptor to bind tcdd as well as variety of other environmental compounds like benzopyrene and others but other the last decade began to appreciate a bigger role for receptor that serves as dietary receptor so it binds the

flavinoids. it's been shown to be a gut microbial sensor so binds to products of bacteria themselves, in this case benzenes, it binds to endogenous metabolites manipulated by the host, metabolites as well as other cholesterol metabolites, we

started to think this balance of ah receptor activity is modulated by things that we consume in our diet as well as things we maybe exposed to in terms of environmental contaminants. so i want to run through very brief summary how trip toe fan

gets catabolized in the importance for ah receptor activity. starting with trip toe fan here. we see a variety of bacterial species are able to import through variety of trip toe fan gets metabolized to a variety of things including asee dick acid

and car box lace produces things like it's important to note this is not entirely mediated by in the case of endol. produced by bacterial property bus these can be further metabolized by the host. in this case by (inaudible) or sulfur transferase, you may know

as typical kidney toxins. the interesting thing ant this, if we go and investigate their ability to activate or bind to ah receptor we can see they have profound and important properties for doing so. i'm showing a simple reporter assay at the ability of

compounds to activate receptor, there by activate reporter so using the prototypical dioxin we see a rather dramatic increase in ah receptor activity shown in the green bar we can see as we try a variety of different trip toe fab catabolites they in some cases

activate receptor as well as tcdd thought to be one of the most potent for ah receptor. this goes to show or emphasize the important properties of microbiota in metabolizing these compounds to activate ah receptor. one thing that's interesting is

when we look at harksh receptor activity in terms of ability modulated not by dietary factor it is but also by environment a.m. compounds this is a really strong contribution by the gut microbiota to modulate basal i'm showing a read out of target gene ink presentation, ilium of

mice treated with environmental contaminant. so cn stands for conventional mice, germ free mice and ah receptor knockout mice we expect not to respond. we can see with treatment of tcdd in blue or red bar we see increase in, ah receptor

i want to point out or emphasize that the basal activity of ah receptor is dramatically lower in germ free mice supporting this notion that the microbes themselves are able to produce important ah receptor ligands. so we can see this as a read out in the metabolic response of

various organ systems so here i'm showing the change in the metabalome, this is a pseudonmr spectra where metabolites that are increased are pointing towards the treatment group or those decrease point away, so in the case of tcdf treated mice we see increase in lipid

accumulation in the liver. very typical of what you see with ah receptor or high ah receptor activity. a but what i want to point out is when we compare these changes between conventional and germ free mice we see a different supporting this idea that basal

activity is important or is importantly modulated by gut microbial processes. so in preparing this talk, i really tried to think hard about what were some of the challenges or gaps that are facing us and moving this type of work forward.

that is truly trying to understand what are these metabolic products or what are these metabolites of the microbiome itself that are serving as signaling intermediates between the host and the microbe. so there was a lot of excellent

talks yesterday metabalomics is helpful to capture a catalog, the true full or the full repertoire of metabolites produced or modified by the and i think to add one important addition to that is to try to understand these in a functional context.

can we show these indeed are signaling through or binding to these receptors and they're providing a means for signals to be transmitted to the host. in chosing if we look at what a typical or typical view of the gut microbiome, in this case a human stool sample we profiled

by mass spectrometry and using molecular networking through gmps library searching we saw great examples of that yesterday by peter. we see array of things to identify we can do pretty good trying to identify what they r. bioassays lipiddings aeromatic

compounds and drugses and environmental chemicals these individuals were perhaps on, these are t no fully annotated and there's compounds yet to identify in fact this is very great example of where the field of metabalomics is today and that's trying to further

development of identification of these metabolites such that we then can do functional studies on them. so the development a lot of these databases are helping to get tows this point. gmps for example, mass bank hmdd pub chem are all excellent

resources helping to get us towards this elucidation or illumination of chemical dark matter. so with that i would like to thank collaborators, at my university ads well as national so lehman, a talented post-doc as well as my long time

collaborator gary purdue and excellent collaboration with gonzales group here at nci as well. >> the next speaker is garry wu who will talk about diet, the gut microbiome and metabalome as a therapeutic probe in ibd. >> thank you so much.

first i want the thank cindy and rest of the organizing committee for inviting me to give a presentation here. this is really been a fabulous so i'm going to start off with challenges and gap slide here. so there are challenges and characterizing the effects of

diet on the human gut microbiome, humans are poorly adherent to dietary regimens. current tools to characterize dietary corp signatures and intake are imperfect. the reciprocal nature of dietary composition to maintain isocaloric consumption make it

difficult to determine factor responsible for about served outcome. diet can have a profound impact on a host biology independent of the gut microbiota, both sensesive controlled feeding experiments and large outpatient cohort studies are expensive and

challenging to complete. it's for that reason there's enormous utility to study animal models when thinking about the interaction between the diet and the gut microbiome. for the following reasons you can control defined diets for long periods of time, multiple

biological replicates. germ free animals can be used to examine effect of diet independent of the gut define microbial consortia or complete gut microbiota studies performed in mice and most importantly you can determine cause and effect relationships

in gut microbiota. how far there is a challenge because mice and other model organisms are not totally equivalent in humans so that study, i did many, many years ago, studying impact of diet and genotype, on the gut microbiome so, we had a wild type and what

you can see here is on a high fat diet there was massive consistent effect hove diet on composition of gut microbiota with a bloom of fermicutes and proteobacteria. so very, very significant we then published a paper a couple of years later in a

controlled feeding experiment healthy human volunteers where we randomize to high fat or low fat diet, this was a ten day inpatient stay, meals each day collected school samples each each ball is a different day of the study and for each color, there's always one ball that's

different from all the other balls, that ball different from other balls is couple of other points on the slide, first the colors are different from each other. this is intersubject variability, the largest source of variants in most human

studies, that's why humans are noisy when thinking about human gut microbiota. so the affect within 24 hours is highly statistically significant but very, very modest, relative to intersubject variability. the other point is though the individuals were on the same

diet, they didn't converge over period of ten days and a movement from day one to the other days is seemingly stochastic, it wasn't consistent in a way observed in mice. maybe in a question-and-answer period there is biological relevance and information what

seems to be noise. we can discuss later on in question-and-answer period. bottom line it appears the human gut microbiota in terms of composition is more resilient to environmental stress of change in diet than is a mouse. we have observed this in another

study published on looking at the composition of the gut microbiota metabalome in vegans and onmivore. you can imagine they both have different diets. the plasma metabalome is very, yet the composition of the gut microbiota by weighted analysis

or unweighted analysis is very, very similar between two. not necessarily replicating what we see in mice. we're in a stage, i personally believe where we have a lot of wonderful animal model data showing functional affects of we have human association

studies p. one goal is to get down here. thinking novel diagnostics, we are using stool for treatment of c diff infection so this is evolution of microbiome field here down to here and we feel as justin pointed out in his talk. studying humans is important and

trying to get down this access to understand the biology of humans and the impact of various types of interventions on gut microbiota in humans and also to determine if what we see and observe in animal models is relevant in human biology. so we began as a group at penn

in microbiome field studying inflammatory bowel disease and the current treatments for inflammatory bowel disease focus on deep immunosuppression of the so if you ask the question how much of inflammatory is due to the host and in other words what's the genetic distribution

to development of inflammatory bowel disease is actually relatively modest. so a lot of inflammatory bowel disease in terms of risk is inferred by the environment and just one exampl of this is the increasing incidence of inflam bowel disease associated with

industrialization. we see with many other types of inflammatory mediated disease processes, this began in north america, europe, israel, and about a decade after that we see a rise in japan and now beginning to see rapid rise associated again with

industrialization, incidence on inflammatory bowel disease in china and in india. so we think that that's enormous opportunity, think about how you might be able to engineer the environment of the gut, maybe through diet or change in the microbiota that might make these

types of interventions work better. or be stand alone modality for if we use diet antibiotics to engineer the environment to treat inflammatory bowel disease there's important questions to address. was is the optimal method of

administration, can modalities combine with immunosuppression to optimize therapy. are optimal clinical settings these intervention can be used. and if you do a human subjects study, you should be collecting a lot of biospecimens that see how the human host responds to

your microbial or dietary there is a dietary intervention documented to be effective in treatment of crohn's disease and this is a defined formula diet and most simple terms is elemental diet, amino acids, triglycerides and simple carbohydrates.

for that reason there are actually many of these are not palatable so they have to be delivered by nasal gastric tube. despite high level efficacy we don't ups how these diets, that particular diet works so the paradigm is here. we know diets are

epidemiologically associate with the inflammatory bowel disease and we know gut microbiota plays a fundamental role in pathogenesis of ibd. we also know or heard a lot about impact of diet on composition of microbiota as well as small molecules.

so our strategy was to use a diet we already know works and inflammatory bowel disease as therapeutic probe to understand the mechanisms behind this. so we can understand proof of concept mechanisms we can define better diets for patients with so i'm a physician scientist,

i'm a big police officer astute observations in the clinical domain, it's important to know define formula diets for treatment of crohn's disease are the most effective when consumed pleatly in place of normal diet that's why we call it een or exclusive internutrition.

let you know how little we understand of how these diets work, there's two fundamental questions implicit in this clinical observation. does een provide something good for patients? with ibd, that's in the abundant in regular diet or consumption

exclude something that is bad for patients with ibd and the regular diet. we're beginning to see a body of literature arise suggesting things added to food supply over past decades change the composition of the microbiome in mice and lead to inflammatory

artificial sweetners, dietary not saying sweetners and emulsifiers cause inflammatory bowel disease, i am saying our diets have changed over the past several decades and it maybe one of those many environmental factors that maybe playing a role in increase incidence of

various types of immune mediated disease like inflammatory bowel keep in mind however these are house studies. so we don't really know whether or not these observations are so we're working together, jim lewis and myself and penn working togethe with andrew and

he talked about this dietary emulsifier work. in a controlled feeding experiment in humans to really determine whether any of the phenotypes that andrew observed in mice and/or other types of biomarkers can be reproduced in human biology which again is

fundamentally important. thinking about the relevance of these different types of observations. thinking impact of elemental diets on composition of gut microbiota, our first step was this study, please pediatric longitudinal study of elemental

diet and microbiota composition. these were 90 pediatric crohn's disease patients with anti-tnf therapy, a longitudinal prospective cohort study, big fan of doing prospective longitudinal studies to try to reduce and identify signal among a very noisy background of

intersubject variability. this was previously published information so i won't show the data. i'm going to say we learned a lot, in doing this study, turns out this is not simply one thing, it is a compilation of many things, use of antibiotics

alterations by diet and whether or not you simply have diagnosis of ibd or not, in combination with this biosis, elemental diets had a particular microbiome signature and one thing we observed was that there was a decrease in fungus. i want to show you unpublished

data integrating taxonomy, the structure of microbiota determined by shall be and meta genomics sequencing and metabolite production. so we think the intersection looking not only taxonomy small molecules they make tell us more about function and diet is a

good example because diet as i mention al through composition of microbiota but also as a source of substrate for the microbiome to produce small molecules that we actually end up absorbing. the study is called food and resulting microbit metabolites

by farm, to determine relation between dietary composition, gut microbiome composition and metabolic products that are present in the gut my lumin and plasma of humans. this was a two week intensive in patient controlled feeding experiment in vegans, onmivores

randomized to consume exclusive nutrition, the product was mod lant. so these are healthy human subjects, impact of diet on composition of microbiome and did intensve intervention to try to map the metabolites that are in plasma and feces coming

through the microbiome. we have done something similar in crohn's disease and are please study and the intersection between these two large data sets will give us some indication by which diet alters the composition of the gut microbiome as metabalome and

host metabalome reducing disease this is part of the crohn's colitis microbiome initiative, that was designed to identify new therapeutic targets. our contribution was actually to use diet as that therapeutic probe to find new therapeutic targets.

let me show you about two slides about what this study was about. so there were three phases in this study, the first phase was looking at impact of diet. and half wei through we did a deep intervention to reduce microbial load using antibiotics and gut purge called

polyethylene glycol. then we followed, reconstitution of the microbiome at the end here. our rationale looking from this time point to this time point give us some idea of diet induced changes in metabolites and microbiome, if we compare

this time point to this time point, the difference is would identify metabolites produced or cob psalmed by microbes. as the microbiota reconstitutes we can correlate increase or decrease in metabolites with recurrence of microbes an diet to identify key drivers. just

to show you the intervention was robust, i show you here, the shotgun meta genomic reads of each day of all the individuals in this study. all the green are microbial reads and the blue are human reads you can see you start out and all -- almost all the reads

in stool are microbial. with the intervention we largely extinguish most of the microbial reads and largely human reads but then the human reads disappear and you begin to see microbial reads as the microbiota reconstitutes itself. we did cfu counting and 16s copy

number pcr so we estimate we reduce bacterial load in in humans by five logs, about 100,000 fold so this really was a very meaningful intervention. so what do we see in terms of shannon diversity? on the western diet you can see shannon diversity is disrupted

and decreases with the intervention and recovers over the next couple of days. on a vegan diet you see that decrease in diversity but it doesn't decrease to same degree and recovers to higher level which mean it is vegan microbiota is more resilient to

this intensive intervention and on a mod lap diet you can see significant decrease in diversity but it actually does not recover. so what happens composition of each diet is a different symbol. each day at the study is color coded here so walking through

this diagram here. so pre-intervention before we use the antibiotics in peg, everybody is over here. the vegans and onmivores are clustered here so substantiates what we previously seen by 16s sequencing the vegan and onmivore microbiota are similar

to each other. but the mode lynn microbiome switched and changed after five days from here to here. so modulin has an effect on composition of the human upon reconstitution, the vegans and omnivores clustered together and they're switching over here

but now modulin that microbiome is distinctive and not completely separate so we have engineered the microbiota into a different configuration. along with the shotgun meta genomics we have done metabalomics in plasma and feces and this was done by john helped

support this, i don't have time to really show you any of that data but let me show you this final slide about the concept of engineering the intestinal environment for treatment of just like any other chronic disease we believe that there probably will need to be an

inductions therapy and maintenance therapy phase so what we are doing is taking our most refractory patients with inflammatory bowel disease that do not respond to any types of pharmacologic immunosuppressive agents. and we have a study that we're

recruiting for right now called holiday we're using antimicrobials in a different fashion to try to get these individuals in to remission. and we do -- we're cautiously optimistic we will see interesting things. we see a subset of sings that

significantly respond to this intervention, two possible outcomes. one is that they're better for a listening period of time because we reset something. the other possibility is that the disease comes back when you stop the anti-buy microbials. that's a notion that you're

going to have to have something to maintain remission. so maybe you can engineer the microbiota in some different configuration to maintain remission. jocie in our laboratory has a paper thinking about bacterial nitrogen flux as a way doing

this, i showed you previous slie diet able to do. this other notion might be diet as maintenance therapy, maybe it's a combination of defined formula diet with safe whole foods to eat. but we don't know what the safe whole foods are.

we're taking a systematic i talked about the farm study. we have two other large intervention studies ongoing that define what those say for whole foods might be. so i want to acknowledge, i represent a very large team. we began with rick jim and

myself but with support from the hmp our group has gone through very large size and i want to acknowledge tremendous collaborators and many different types of projects. >> our next speaker is nicole car kin who will talk about how gut bacteria eat your veggies,

molecular details of glycan scavenging at the cell surface. nicole koropatkin. >> just as soon as i get this up we'll begin. first i want to say thank you so much for the invitation to speak here today. let me explain myself first.

i'm a structural boyologist and so i really think about the details of how proteins and carbohydrates interact and i'm interested in using that information to understand how the gut microbiota specifically responds to diet. so the way i have structured my

talk actually is to spend less time telling you about the specific details of my work but hopefully trying to impress upon you why we need to return a little bit more to understanding some of the basic biochemistry of these organisms that live within us and how this

information by taking a detailed structural biochemical approach and understanding how gut bacteria perform certain metabolic processes i will tell you about carbohydrate degradation but you can take some of the examples i will give you and extrapolate that to

almost my metabolic process, why's important to understand details and how they better inform various data omics sets and maybe allow us to predict the way gut microbiomes are going to respond to different types of environmental perturbations.

so of course as justin and others have mentioned, the gut microbiota responds to diet in very different ways. these organisms are extremely prolific in their capacity to degrade different types of carbohydrate nutrition. they have many different

categories of carbohydrates to choose from, some degrade different dietary plant cell wall polysaccharides, other types of carbohydrates that we find in our diet and then some can turn and degrade host mucosal glycan, some of the protective lining of the tract

that keeps us healthy. so a lot of carbohydrates to choose from. this means bacteria actually have quite a difficult task at hand when it comes to recognizing these discreet glycan structures and deciding which are going to eat.

so how are they able to select and scavenge and process these different carbohydrates and the reason they can is because they're genomes are full of carbohydrate active enzyme, cazymes. as humans we can't digest many we can't digest most, only three

really. sucrose, lactose and starch. but these guy consist really degrade just about everything else, a single gut bacterium can have hundreds of enzymes for the degradation of carbohydrates and its environment. so it's been used to look at the

assume contempt of bacterium or genomic data set to try to understand what kind of carbohydrates is this community capable of utilizing. but there are caveats relying on strictly these annotations even very well established databases like car boy hydrate active

enzymes database which is cell curated. for example, i will take you through a list of examples why we need to delve deeper and beyond annotations in order to understand function. in the glycoside hydrolase family enzymes this was thought

of a group of enzymes that only degrade cellulose. but from the list here enzymes in this family degrade a wide variety of different so protein sequences and structures that are all going to look alike, some degrade cellulose, some are

endoglocknases. and how do you know which can perform which activity. receive you say break into subfamilies but it requires detailed understanding of ha specific amow know acids and specific proteins lend this specific functionality.

another beautiful am exam is from matt's lab, they were studying a group of enzymes in the human gut, gut bacterial enzyme able to deglucagon date so it can get reactivated in the colon and cause diarrhea. so individuals taking this drug had to come off, if it was

reactivated in colon because of side effects. the thing is, is that almost all gut bacteria have these types of enzymes, they have glycoside hydrolase family enzymes and a lot look like this. what his group did is bend to distinct classes and they found

subtle changes in the active side site, it comes down to different loops surrounding the completely conserved active site or what dictated whether or not certain bacteria, certain bacterial enzymes were deglucagon dating these drugs. we can never get at this if we

just look at sequences and annotations alone, it takes drilling down to individual features of protein structures. another problem with annotation is some enisyls actually lack for example, within the backroides genome there's a locust toker break down of cell

wall polysaccharide zylain, there's a protein in the locust that looks just like a sigh lenase however missing the catalytic base, it's not an enzyme, one single amino acid change. there's another example from the locust, this is another plant

cell wall there is a surface within this locust that looks like it should be really a robust degrader of cellulose. and yet it has almost no enzyme attic activity. 's a glucan just a little bit. yet it seems to be important for glucan utilization, it acts like

a binding protein. the point is that there are these protein sequences that unless we drill down into the details we don't really know exactly what they're doing. of course the other problem, as many individuals have pointed out, katherine pollard mentioned

yesterday that something like 53% genetic information really can't be assigned functionality. and certainly when it comes to looking at whether or not a back fear yum or a community can degrade carbohydrates we have the same problem with an dation. nature always find a way to

surprise us. and so these were some surprises that came out of a study where this group in harry gilbert's lab looked how into rg 2 considered the most complex polysaccharide on earth can be broken down by single gut bacterium.

and as they study the sequential break down of this molecule, they uncovered six new enzyme families within this genome and discovered two existing enzyme families that have separate catalytic activity. again, we don't have this information, we don't know these

things until we take a step back and do hard biochemical studies. so my group studies the molecular details of glycan out youthlyization. we're interested how proteins recognize carbohydrates in the gut, how to process and import the reason we think it's

important to drill down to these biochemical details is so we can understand why diet changes community how it does. we see studies that some respond to diet and some way and some respond to diet in another way. we can say their microbiotas are different.

of course they are different. but if we drill down and look at the proteins that are present in the communities, can we by using biochemistry find specific molecular signatures within protein families that lets us know whether or not capacity is there to recognize or degrade a

particular carbohydrate. so moving beyond the annotation. another reason it's important to understand the biochemistry how carbohydrates are broken down to the intestinal tract is we can stop deleterious activities from occurring. we can stop from bing eating the

things we don't want them eating. bacteria has a choice different car behydrates that they eat. this was one study that came out of eric martin's lab they had mice on a fiber rich diet or fiber free diet and you can see when gut bacteria didn't have

any complex polysaccharides to eat from the diet, they degraded the mucosal lining there. which bacteria are performing the task which specific enzymes are they using. rather than use a antibiotic to get rid of bad bugs perhaps find targets to prevent this specific

ativity from happening in absence of polysaccharides or in certain disease backgrounds. so i think understanding which enzymes are participating in degradation activity can provide us a window into changing the metabolism of the gut so i want to address i hope that

i demonstrated a little bit through these examples with what critical knowledge gap, we need to identify individual species, regulatory pathways and protein to explain predict and develop personalized ways to shape the gut microbiome with diet. how do we fill this gap?

in the words of bernie (inaudible) the curator of the active enzymes database, we need in a lot of ways those who do this work to get back to the bench. and to do that hard biochemistry work in order to assign functionality to proteins and

enzymes we don't understand. define molecular signatures. we need to bridge collaboration between microbial ecologists who have large omics data sets who have the data that shows how microbiome respond to diet and pull more biochemistry and structural biologists into this

to bridge the gap and figure out why these changes occur. why some respond to diet and why some people don't. so like i said, we need to do the protein characterizations. particularly in field of study carbohydrates we need better tools for studying carbohydrates

for identifying them, quantifying them because this type of glycomics technology is severely lacking behind technologies that we have for assessing bacterial community structure and identifying other metabolites present in the gut microbial community.

so in the last little bit of time that i have, i'll tell you a little bit about the work that i'm doing in my lab, we are interested in the anity gritty details how bacteria recognize and process different carbohydrates in do gut. we look at a number of

the particular fiber we look at the most is starch. starch is a carb that humans can digest themselves but there's a portion termed resistant starch that we can't process and only a subset of gut bacteria can process resistant starch. the thing is, if you were to

look through the genomes of every bacterial species whether degrade in starch or not have the right enzymes for degrading starch so which can degrade resistance starch and which can't degrade resistant starch? that's one question we're trying to answer.

but the thing is, all these bacteria not only have different enzymes and different enzyme specificityies, they have organized their binding proteins for carbohydrates with enzymes in different ways surface of the cell that changes the way they see carbohydrates in their

environment so the bacteria dede iss break down in peptides on different other sides of the cell to work together and for other gram positive organisms there maybe a large catalytic protein with binding modules that hold on to a fiber breaks down into pieces and a separate

transporter that imports it and we're finding unique systems which is like a molecular leg go system where there's a scaffolding protein and the different enzymes can dock on to it so you have this massive hydrolytic activity for break down of difficult to degrade

fibers including resistant starch so i will tell you briefly about some of the stories in these different systems that we worked on. we have been particularly interested in how systems are organized so this is a system that's involved in soluble

starch degradation and bacteriadede omicron. we characterize the proteins on the surface of the bacterial cell, there's one protein that works with the transporter, seemed to be incredibly important for getting pieces of starch into the cell though it

turns out this is a starch binding protein, it doesn't bind star touch get things into the cell. there's additional starch binding proteins on the surface of the cell including amylase and somehow these proteins all work together.

we wanted to understand how these proteins work together to process import carbohydrates. so we decided to take a single molecule approach with julie and post doc chris and what we did is substituted this carbohydrate binding module we see in sus g this alpha amylase with a helo

tag protein we can give fluorescent ligand to. these are live an aerobic bacteria you see moving the suffer surface of the bacterial cell is an individual protein molecule of sus g. we have cells fixed, these proteins don't move at all.

we did this in an iterative fashion where we look at the movement of our label sus g and presence or absence of starch. so here you can see that when starch is present, you can see sus g is adhering to the starch particle and coming off so it's really the dynamic assembly and

movement of enzyme on the cell surface. so we were interested how mobility affected when other members of the system were present. so we can look at diffusion co-efficient of sus g in absence of starch and see how the

diffusion co-efficient change when we added starch and took other components of the system. we can see it slows down in presence of starch, it speeds up ad we move other players to the system so there are dynamic interactions, assembly going on while this system is degrading

and catabolizing starch. so not just the specific enzymes it has but how they're organized with binding proteins, how organized on the cell surface can make a difference in the way an organism is able to utilize a particular hydrate. we look at fermicutes.

i'll skip that. we looked at the fermicutes to see how some of their systems are different and again like i mentioned what they do instead is they have one or a few enzymes on cell surface that have not only a catalytic domain for breaking down a fiber but

also several carbohydrate binding modules along with it. and the interesting thing about this system is all in this cbms associated with this are specific for corn starch but don't recognize granular potato starch so starch is not starch, it comes down on molecular level

how a protein actually recognizes specific motifs on a we have been looking into the am zone type systems molecular leg go type system of proteins that assemble on scaffold. and we have been interested in how some of these enzymes can break down alpha 16 linkage,

branch points we find in starch. here we crystallize one of these proteins with two different oligosaccharides just as if this branch point and starch had been broken and we think that these types of enzymes represent a weak point where certain gut bacteria break down resistant

starch because this is a linkage that human enzymes simply can't break down. so with that, i hope i have in some way impressed upon you there is room for biochemistry in the microbiome and that maybe taking a step back and doing some biochemical experiments and

looking for molecular signatures within omics data set can improve our predictive capacity to understand how microbiomes work. so i have a small group of wonderful people, i'll pointed out that all the work i do is in collaboration with eric march

tens, i mention julie with the single molecule imaging. darryl co-burn who is now professor at penn state university, did a lot of work we recollecttile as well as pneumococcus romie, ryan and matthew are also members of my group that have contributed to

these projects. >> our next speaker will be johanna lampe who will talk targeting gut diet the use of controlled feeding studies in >> thank you very much to cindy and the organizing committee for the opportunity to participate what's really been exciting day

and a half and i expect this fun will continue as we move i will wrap up talking about some of the issues involved in using controlled feeding studies and testing diet microbiome we have heard through the course of the last day a number of elegant examples of these

approaches. so i'll fill in some of the gaps with regards to hows and whys, challenges and where we go from we use the word diet very loosely. i think this is one of the things sometimes catch us unaware or trip us up in that we

are talking about an extremely complex exposure of thousands of compounds, particularly human diet for those who work in drosophila or some of the more simple models you can simplify the diet extensively, we have yet to find study participant who will put up with long term

feeding ton same food every day so we're talking complex group of mixtures, which we can characterize on many levels we can talk about dietary patterns, talk about specific foods or classes of foods, there are certain sources of compounds or we can focus on individual

compounds whether they be nutrients we have identified as essential for life or we have identified as contributing to improvemented health or increase in disease risk. and particularly in western countries we also have a whole slew of components of our diets

that are supplements either as food or as specific compounds. so when we talk controlled feeding studies, we are speaking about situation which we provide food the study participants. these are diets designed prepared to specifications related to the hypothesis that

you're interested in testing. in the case of gary wu's example you may conduct studies in context of metabolic ward where people are sequestered in particular space for a certain tenth length of time. those are elegant. complex studies to conduct

because they're expensive. you have to find study participants who are willing to be locked up, often it's graduate students writeing a thesis in the case of healthy individuals which it does limit the generalizability of your activities.

one of the things to keep in mind, these studies are designed to measure biologic physiologic effects of dietary manipulation. they are not designed to look at application of a dietary intervention in real world so we're really as my colleagues would say you're doing rat

studies in humans. if that's how you want to think about it sure, but it does speak to the idea of trying to get at the biology, trying to address specific hypotheses as far as mechanisms to the extent we can in intact humans. as you might imagine useful

components of controlled feeding study but also limitations that come with both approaches and as well as working in humans. the nice thing about this is from standpoint of thinking what you can control. you are not only controlling the intervention ie what is a

specific food components of food, et cetera, to test but also in the context of a controlled background diet. putting everybody on high fat diet or in the context of particular designation relation combination carbohydrate protein fat.

you can set these up to test dose response where you're either providing specific doses across the group of individuals or you're actually running this on a continue one of dosing on the basis for example of body weight or other factors that allow you toe examine this.

as i said with emphasis on biologic physiologic properties you're focusing on biomarkers of we can use these studies to develop biomarkers of dietary exposure and in many cases also particularly in relation to this conference, really thinking about biomarkers of bacterial

of course there are always some limitations, these are typically short term interventions, nobody likes to be provided food for weeks to months on end. i will say there maybe a few who maybe willing to do this but most case we typically start to see mutant after six to eight

weeks so this is something that we rely on short term effects. we also have to rely on itermediate biomarkers. we can't carry tout the point of development of disease at the in that context we're slightly hand strung. became way of example i will

show you couple of studies, feeding studies to test dietary pattern microbiome interactions. these are probably most complex of the approaches, compared to something where you're looking at addition of a specific supplement or particular food. we really are looking at several

questions, what is impact of dietary pattern gut microbiome and how does diet pattern microbiome interaction affect disease risk biomarkers and the why both we are dietary patterns? at this point most humans eat foods in a variety of different

foods and we have come to recognize from the standpoint of making dietary recommendations and public health recommendations long term behavior change or long term viability of approaches but we need to focus on proposing food choices over the broad continuum

of types of foods that are available to us. from a research standpoint it helps to offset limitations of a single nutrient approach. where we know from the standpoint of epidemiologic studies or even context of foods in if general, nutrients are

often intercorrelated types of foods consumed together often correlated. this also helps preserve multi-dimensional aspects of foods as opposed to feeding specific nutrients. we can include all components of the diet both nutrients spanned

non-nutrients. and we can take different approaches to studying dietary patterns you you may run across sum studies, comparisons between traditional dietary patterns compared to western diets, mechanism based indices so we know diet patterns are

associated with increased innamation, less inflammatory diet versus more inflammatory diet and also comparing typical diets against a established recommended pattern if we will tell people we should make these dietary choices is what is the impact of that diet

pattern in contrast to what people are usually do. there's a number of elegant short term feeding studies looking at rather extreme differences between diets and this is work out of peter turnbau lab a number of years ago, across five days of either

a very heavy animal food based diet compared to a diet devoid of animal foods and predominantly plant food, compared to individuals baseline traditional habitual diet. you can see percentage of energy for fat varies across these animal diet being very high in

energy from fat. high in protein and in contrast less percent energy from fat. lower protein intake as well as substantial intake in fiber. compared to the anal as well as individuals habitual diet. as they show when you look at affect of diet on gut microbiome

particularly in context of looking at meta transcript omics you can see substantial differences when individuals are on the plant biased diets versus the animal diet. in contrast, we have seen this over course of several presentations, is that we often

don't see the same affect from the standpont of changes in the relative abundance of particular taxa. here the black dots indicate really you see no separation of the same individual independent or dependent whether or not they ae on the animal or plant diet.

as indicated we see differences in microbes involved in bile tolerance, changes in microbes that are metabolizing polysaccharides as you might expect given fiber content. but really the idea that in the context of humans and contrast to many rodent work is we often

don't see those same beautiful separation on two different diets. steven o'keeffe also looking again at substantially different diets feeding african americans a native south african diet ie going from a diet for the african americans at 14-grams of

fiber per day after 55-grams of fiber and substantial reduction in dietary fat. in contrast taking the native south africans reducing them from their usual 66-grams of dietary fiber down to around 12 and upping the fat. in the context of this you see substantial changes in

microbiome activity from the standpoint of the fiber fermentation story as well as conversion to secondary bile additional component of this is they took colon biopsies at end of intervention period, short term intervention and saw substantial changes in co-le

rectal cancer relevant components in the colollic epithelium, decrease in in proliferation, in the african americans put on the south african diet and in contrast africans on the african america diet substantial increase proliferation.

marked changes in inflammatory markers, in relation to changes in the diet. suggesting combination of effect on microbiome as well as on host, maybe contributing to some of the risks that we see in increase in colorectal cancer risk in u.s. african americans.

from wrap up here, couple of examples from our work using controlled feeding studies that was designed to look at low and high glycemic load diet patterns, in healthy individuals, these individuals were not insulin resistant, normal blood fast glucoses and

looking to choose food sources that had low glycemic indices or high glycemic indices fed at u caloric diet levels to maintain weight over span of four weeks and the low glycemic load diet ended up to be substantially high in fiber, about 55-grams per day, hyperglycemic lode

though about half of that, it's still a relatively high fiber intake for the typical u.s. as part of the initial study we were interested in a host of circulating metabolic inflammation biomarkers, this was part of the transdisciplinary research

energetics and cancer and this is also by way of encouraging those of you who don't necessarily do feeding studies and don't see yourselves doing feeding studies to engage colleague whose do or even if you read a study say i would like to ask this question in the

context of this intervention, hemoglobin to contact those individuals we squirrel away samples in order to be able to continue to work in these areas so provides us good opportunities. so by way of the affects of these diets as far as glucose

you see substantially lower area under the curve for glucose as well as insulin as you expect in this type of study. so we're now using this as -- in a multi-omic approach to look at effects of these two diets the impact of gut microbiome in relation to the plasma

metabalome and ultimately biologic standpoint risk. in feeding studies cross over design, we have stool samples interrogated for 16s meta genomics, meta transcript omics at the end of each intervention period, plasma metabalome couple of platforms and proteome using

antibody array approach for analysis. so just by way of example, when we look at 16s we don't see any difference between effects of the lower high gl diet or comparison to baseline, these black dots are all our qc control clustered together in

one spot. we do see differences in certain taxa with regards to low and high gl diet. changes in relation low gl diet with paraback roy disease. and high gl diet a couple of members in the clostridia group. as we move to meta transcript

omics and meta proteomic we'll have an opportunity to look at these additional questions. from the standpoint of metabalomics we separate slightly better low gl diet with triangles as opposed to high gl there's a number of targeted compounds that are different,

some higher in relation to the low gl diet and some are lower ie either higher on the high gl diet so integrating two pieces will give us more information in this area. who talked phenotypes with regards to microbial metabolism, particularly in the area of

metabolism in plant compounds there is a lot of heterogeneity how individuals microbial communities handle these, this is by way of example, the ligands which are compounds found in plant foods typically whole grains fruits and veggies and we actually which undergo

metabolism by a consortia of bacteria to produce the bioactive that's been shown to be associated with lower risk of several cancers. if you give everybody particular doses of ligands you don't necessarily see a nice dose response as far as lactone

excretion. this is not unusual but again speaks to the heterogeneity in we find in context of this feeding study, that actually there are several secondary bioacids that typically are positively associated with lactone excretion.

these are plasma bile acids. suggesting differential response based on capacity to produce lactone. some of you may have had a chance to see poster yesterday of rob considering issues of effects of gut microbiome on production of interlactone and

the variation in the capacity to produce this. this was a study looking at effects of ligand supplement on gene expression and colonic epithelium we took sigmoid colon biopsies at end of each intervention and we look at response to intervention.

we found that though there was a modest response to the intervention it really varied depending on whether or not the individuals were low or high producers. more to follow on that. knowledge gaps needs and challenges, recapitulates what's

said several times, strong need for well annotated metabalomics platforms that capture the range of endogenous and exogenous compounds and metabolites and also as we get to the point establishing all the genomes on the various types of bacteria, to be able to improve our

approaches to characterize microbial metabolism that involve bacteria. we focus on specific microorganisms but to find ways to integrate those in order to get through consor shall and i'll skip testimony this slide for reasons of time.

i want to thank meredith huller a microbial ecologist working with our group the last decade as well as collaborators at texas arc and m., and collaborators at the hutch. >> the session is now open for >> hello, brandon acre. excellent discussions.

i don't know if anybody know there's a raging debate of all things dietary fiber going on in the fda right now. and i think that there's a lot of challenge around trying to define what dietary fibers are and i would like to post to the panel and the group of you we

need to mind the fiber gap and the colleagues at fda need assistance from the scientific community to help understand the more about what that does structure function for the host. and set an example. so things like resistant starch are not currently considered a

dietary fiber on list of approved fibers. so there's probably several in this room that disagree with and i was wondering i want to take it to a further step and try to find way to define microbiotic carbohydrate ads an essential nutrient.

that's challenges around essential nutrient because it's something that we must have for life. and i think there's bit of argument against gains that in the fda currently so want to hear your opinions on defining dietary fiber pass essential

nutrient and how to do that future. >> great question. i'll be curious to see how debate plays out right now. there eat a lot of questions we saw with andrew's talk this morning. i think there's context,

determines a lot what microbes you have in your gut, what other diet you maybe eating, how rapidly things are fermented. where fermented in the colon so there's stale lot we have to learn. it really validates the approach that i and others pitched here

which is to study some of these in a controlled setting to assess what they're doing in i think we don't really have a good grasp of how complex this is and i don't think there's going to be a simple answer to that question. it's a bit of a bill out answer

but cure curious to hear what others have to say. >> that's a good point. one of the challenges relates to the fact in human studies we don't take into consideration the structure of the food from the standpoint of jus tip's approach looking what's

microbial accessible carbohydrate, it depends on whether or not it's covered in brand layer or chopped up into pieces so from the standpoint of thinking about fiber, you could argue anything from the standpoint of a carbohydrate that gets to lower gut is twoing

to act as fiber. and it really does muddy waters when we try to come up with a particular definition of specific compounds, et cetera. >> i would just add gut microbiota is deeply fermented and has enzyme attic activity as any coal pointed out so we focus

on fiber keeping in mind the human diet is to a certain degree reciprocal. in other words, high plant high fiber diets tend to be lower in fats and meat. and vice versa. and meat and fat we do know has affects direct on host maybe

independent of gut microbiota so i want to make the point that it's not just one thing that changes when you change a diet, it's probably other things that change reciprocal fashion that may affect on health. (indiscernible) ucsd. two questions.

one is four labs have shown the gut microbiome is cyclical and has circadian changes and in fact controls host circadian rhythms. bile acids are circadian in terms of which exist in lieu minute. the effects are downstream,

transcripts are circadian. how can we this is true for practically anything the gut absorbing. so how can we understand the system when not looking ate as a dynamic system. that's number one. number two, 95% of the became

acids are absorbed in the ilium so wouldn't intestinal microbiome play a more important role in digesting fibers unless it gets down to short chain fatty acids or digesting proteins or digesting fats, all these things, more so than anything we're measurening the

stool. >> i'll give it a shot at one or two of those things. i think what you are saying in terms of bile acids, i think bile acid is important emulsifycation and absorption of fat soluble vitamins lipids and things like that.

i don't know i don't have good knowledge how that affects fiber digestion by the microbiota. having said that, jeff mentioned this morning unknown domain for largely is the small intestinal microbiota and i have made the point that i think that when i think about nutrient intake and

response of the microbiota to diet, i think though not a lot of evidence much more dynamic and you think about receptors in the small intestine, the enteric or moans in the small intestine, fxr, bile acids that type of biology is largely unknown and another area of significant gap

that jeff pointed out that do agree with. the circadian rhythm thing, it's a fascinating area. some of that is dependent on when you eat and perhaps what you eat. it's a challenge. i have had people talk to me

about studying circadian rhythms in microbiome in humans. it is a challenge. mice you can take out of cage, put in a box and -- they'll poop all the time but humans don't do that in a way that you can stull dicircadian rhythm so there's inherent challenges, interested

knowing how others figured out way to do that, i'm in the smart enough to think about a resolution to that problem. >> one quick add on to that. the concept of sensors not just flowing through the die yes, sirtive track but perhaps tethered to sites you can

measure what's happening at a given site over period of time can be help. trillion dollars >> not here to talk about census. when i will listen to your talksit's impressive, but since it's a panel discussion we're seeing

one part of the elephant. may after this, may after that. we need a full on project no less powerful and impressive than the genome project, lay out map systematically go after the variables, i haven't thought about circadian though i do on breakfast whether the effect is

different at lunch. but point aside we need systematic method, if you will o approach, to go after this in a comprehensive fashion where we can tear this apart. what are your thoughts there in q. it sounds lovely, trying to get it funded would be extremely

think about trying to integrate possible combinations you think about this as totality of diet, to do each separately is going to give you one little snap shot but it won't cover the totality of diet which is what people are >> you probably have to do it in commontorrics in some fashion

and tear apart with multi-variant analysis. i recognize the issues with but nonetheless i would argue if we go the way we're going, it will take several decades to sort this out. and i think that's too long. if you look at the epidemic that

exists, we're just not approaching this right. again, given the magnitude of the problem. this problem is no less than terms of diabetes, obesity. we're going about it in a trivial fashion, one at a time, it's just not going to solve the

problem. i think. >> i think the reason i presented the work on dietary patterns is that really is a direction that the nutritional field is also taking from this standpoint of being able to make recommendations.

we know that looking at large scale epidemiologic studies with thousands of people that individuals who follow a diet pattern that as recommended higher in fruits and vegetables, low in red meat, et cetera, have lower risk of cardiovascular disease, lower risk of several

types of cancer in prospective studies so starting where you say okay, what does that diet look like, what are then some of the biologic physiologic as well as microbiome pieces would be one way. >> i agree with johanna. we ooher not starting from zero.

inflammatory bowel disease is an example. so we have diets that work inflammatory bowel disease, antibiotics much as cipro have been long team treatment for crohn's disease and we see pico transplantation with numerous innoculations can have a modest

so these tell us that these things are actually effective in humans treatment but don't work as robustly and significantly and as consistently as we want them to work. so that's adding the newest technologies to try to come up with better ways of actually

doing what we actually see to certain degree. so i don't think we're starting from zero, we're trying to improve something we have seen at least in some disease. i have to say in terms of nutritional biologist, as you know, there's a trans-nih

initiative now thinking about next decade of research on nutritional biology, i don't know if program offices want to mention this, the same issue applies there. nutritional biologist is exceedingly complex, how do you study, we have now wonderful

technologies maybe we can actually better define biomarkers what people eat because measuring using dietary questionnaires is the best we have right now but a poor surrogate in terms of reliably quantifying what people actually >> let me add a comment.

i don't think the problem is any less complex when people talk about the genome project back in early '# 0s, it sounded -- '80s. it sounded impossible. but because it was put out as a goal people came up with ways to go at it and new technologies

came along the way. to me that's how solve a complex >> i think that one issue is treatments for specific disease but the other way to think about the diet microbiome access as this incredible lever on human biology to toggle so many aspects of metabolism and immune

function, central nervous system, but we know very little about that wiring. you are getting at a systematic effort to figure out that wiring so if somebody needs immunotherapy we node what to feed them to get the microway chrome to move in the right way,

move the immune system, alternatively that same person fighting autoimmune disease it would be a different treatment in a different direction. thinking along those lines figure uing how to do it systematically would be a wonderful things to do.

i was at a diet microbiome workshop earlier this summer at the nih and there's really no consensus in many different methods how to measure what people are eating. just to be able to come together and find what best practices are for that, that's a major hurdle

in this field because that's what everything is based on downstream analysis is relying so i think there really could be huge improvements rapidly in a nice big effort would be great. >> we have to wrap up but just in response to what gary was talking about, one of the things

that nih is working on now is a ten year strategic plan in regards to nutrition. and if anyone has thoughts or suggestions please feel free to contact if anyone needs his contact information we'll get it to you but i really -- please join me in showing our

appreciation for all the speakers in this session. and in order to get back on schedule, the break will be 20 minutes and we'll be starting again at 3:50. >> okay in the interest o time, i'm going to start if everybody will sit pretty much there.

so welcome to session number eight, the role of the microbiome in the disease initiation and acceleration, moving beyond association. my name is lis caler, program lecturer at the heart lung and blood institute and area of lung the purpose of this session was

to highlight concrete cause and effect mechanistic example of the microbiome role in disease initiation or excessive. what is not understood about the microbiome, how it initiates or contributes to disease and hopefully we can get enlightening clues from our

wonderful speakers. we have four speakers. >> sorry about that. so i was saying we are in session eight, the role of the microbiome in disease initiation or exacerbation. moving beyond associations and misname is liz caler, program

director at nhlbi division of lung disease. the purpose of this session is effect mechanistic examples of initiation or exacerbation, what is not yet understood about how the microbiome initiates or accelerates disease and what is we have four wonderful speakers,

andrew goodman from yale university. wei jia from university of hawaii, cindy sears from johns hopkins university and gary huffnagle university of michigan. we'll start with the first speaker, andrew goodman, title

of the talk, efforts to understand causation in microbiota host interaction, two >> it's actually just going to be one example due to uncertainties of our budget climate or due to lack of time. so i will jury in to approaches to get at causation in host

microbiome interaction. maybe some of you have seen this before, so this is a correlation you can see two types of data here that are very well correlated with each other. 91% correlation according to data from the usda and the american bar association.

so you can see these two measurements are. are the amount of sour cream that's consumed in the u.s. per capita. and number of lawyers in new york. what you can see is though there's a nice correlation

between these two increase, i hope you can realize that if you find yourself eating sour cream it's not because there's more lawyers in new york. in context of microbiota field this can be a challenge because we can see very nice trends and i think part of our challenge is

to think of different ways to disentangle and differentiate this example from the types we're interested in. one way to do that is if we have that ability to manipulate variables related to each other. we can change one of the two variables and ask whether the

other one changes in response, then we can get a sense of whether there's a causative element one way other the other. that's what i will talk about today is our efforts to manipulate elements of the microbiome, in this case specific genes an enzymes within

gut bacteria in order to identify causation. the example i will talk about, i will give an example how to apply approach to understand this observation that there are many pathogens that seem to be associated with antibiotic depletion of the microbiota so

this is a pattern seen in many examples but the underlying mechanisms for how antibiotics change the microbiota to facilitate pathogen invasion are still being discovered. so before i get to this example i want to tell you about this challenge we have.

i will tell you approaches for looking at thebacteroid es, the prominent genus in the gut, not the only one and in certain individuals not the most dominant one so right now i'm going to say that, propose if i convince you of utility of approaches we're taking none of

them work in the fermicutes and i would say that one of our challenges is to spend the time toe develop these approaches for this important majority or major phyla in the gut. so this wish list that we had for working with member hearsay of the microbiome in general was

pretty simple. we had for many years wished that we had ability to turn genes off and on in the we wanted to do that using some kind of signal or inducer that wasn't already in a mouse or already in the diet of a mouse so we can really control what's

going on in a gut commensal while in the gut of a live wake animal. i also don't need to tell you, though these tools may seem modest especially if you work on e. coli or proteobacteria, e-coli is a rare member of the microbiome and phylogenetically

different so it's a general pattern that tools developed for e. coli don't work in for examplebacteroidedes which is a different phylum entirely. so the approach we set out to take was to use tools developed in e. coli and completely rebuild them from scratch in

order to use them inbacteroidedes so we used this teat r based gene regulation system that has three parts. the first is an debt operator sequence here, dna element that is you can put into the promoter of a gene of interest. this operator sequence is bound

by tet r represser. this protein here and when the t represser binds to teto it binds gene expression. the third part of the system is tetracycline or analog such ads an hydrotetracycline a synthetic molecule that binds to teat r and sequesters it away from the

debt o operator so this double negative means that atc is inducer of gene expression. so again systems had been developed in e. coli but none work in bacteroides. we started from scratch. we measure by putting a reporter gene downstream of engineered

promoter so in many cases i'll show you data using the nanoluke luciferase to measure luminescence produced in direct correlation to how much of this protein is made over many orders of magnitude. we started with a genome approach we took highly

expressed promoters, 180 highly expressed promoters from across the phylum bacteroides and made a consensus here, this showed us regions of high conservation but also quite variable regions that potentially we can insert these debto operator sequences in. so with trial and error we did

that, we found two locations of insertion or partial substitution in a specific orientation able to get these debt o operators into -- tet o operators into a representative promoter without killing the activity of the promoter entirely.

from and we did this quite a bit, some of this in vitro, i won't show you in vitro data just in vivo data so we took first mice colonized with 14 species of human gut microbes including engineered the of bacteroides nanopromoter. when we put atc in drinking

water of the mice on day four we get this increase in luminescence that increases by orders of magnitude when we add atc, when we remove atc from the water bacteria in the gut turn off the reporter and we can see luminescence goes down. there's nothing in the mouse or

mouse diet or other species that's interfering with this inducible system. that's because is this inducer is synthetic, not present in the gut unless we add it. we can do this in spf animal, when we do quantitative pcr analysis of the immunity,

community, there's no specific atc differences in microbiome suggesting this inducer is not impacting other microbes in this defined community. we can do the same thing in conventional mice with complete microbiota of their own. so in this case the engineered

theta strain makes 0.1 to 0.5% entire microbiota yet when we add atc to the drinking water we can turn on one gene inside the genome of this one bacteria that's .5% of the entire you can see that in the absence of inducer we don't get any expression.

say again telling us that there's nothing in the complete microbiota either turning on gene expression when we don't want it to or preventing ability of inducer to activate gene expression at a specific time we use a high throughout put capture to capture bacteroides

to healthy donors. wild isolates and found in every case the system works the same as it does with our type strains. now i will come back to them question about trying to understand the mechanisms how antibiotics and microbiome and

pathogen infection may interact using the system. and i'll start with an example that was nicely worked on by justin sonnenburg in 2030 looking at one of these mucous glyco profanes that have acid as terminal residue. what's clear is there are many

commensals and also pathogens that can con psalm acid once liberateed from the glyco proteins in the gut epithelium, there's commensals with enzymes that have this acid liberation capacity but even though these enzymes are ubiquitous in the microbiome the levels of acid

are fairly low in the lumen. that's because there's also other commensal microbes that consume the acid. however if this balance of things is disrupted then the acid serves as nutrient for pathogens including clostridium did i have sill.

so what his group showed in 2013, they can see c diff colonization levels dependent on ability in at least nodo biotic mice to encode the sialdase, corpmented by restoring the s ialidae or the i a sid directly. if they gave them antibiotics to disrupt the microbiota and the

lumen of the gut went up significantly but it wasn't clear how this was occurring. you can imagine different explanations, could be there's a certain microbe in the gut that responds to antibiotics could be specific antibiotic necessary to cause this effect or in general

we see expect to translate to a different mouse colony or maybe even to other organisms including humans. so i think that this ability to control gene expression can help us address these kind of questions experimentally. so what which did is we took

this engineered system and instead of couple to a reporter like luciferase reporter, we couple to sialidase responsible for releasing the acid to the lumen. we call this btrs for regulated sialida serks. then we colonize germ free mice

with this engineered strain. we gave them a drop of atc in their drinking water then followed measurements over time. we measured the inducer itself, atc and you can see that after we add to water you can see anytime fecal samples collected from the animals and disappears

within two days. the next thing we wanted to measure was how much sialidase protein and active protein produce as we turn on and off gene expression using engineered system so to do this twee developed e vivo assay we took samples from the mice and

inbaited with artificial substrate with sialic acid. this is the acid here, it cleaves here released in the nigh troll phenol group we can measure so we developed enzyme assay on fecal samples to convince ourselves provide substrate in excess, measuring

the initial velocity of this reaction. as we dilute the fecal samples we have different reaction rates ads expect for typical biochemical enzyme assay except it's done on feces. so then we can confirm take fecal samples colonized the

activity is consistent and high. and take mice colonized with mutant the x vivo activity is low as we expect. then we can do after we add inducer activity measured by x vivo assay is wild type organism and goes back down to zero. you can measure acid itself

released from the epitheme yum to the lumen and measure that in fecal samples shown here. these taught us two things. the first thing is that when we look at when the protein is present and when the product of this 'packs is present you can see not exactly concordant and

there's a window of time here where we have the product the acid is present in the gut but bacteria no longer has the activity because we turn it i will off and we measured that it was turned off. this is the first time we started thinking about dimension

of time when we look at microbiome phenotypes so when you see that there's a certain metabolite that's higher or lower in a certain sample, is it necessarily an at this particular time activity the microbiome is doing at that time or a legacy effect of what the

community was doing earlier. in this case this is several days where the product of the action stay around even when the activity isn't there. the second thing we learned is looking at dose response relationship. here is the same data plotted

against each other. each is a mouse activity and x axis and amount of product object y axis. what you can see is there's a part of the curve where the amount of activity you have tells you how much to produce. that makes sense.

there is another part of the curve where the release of the acid is not determined by how much enzyme activity you have. in other words,, if you have high level activity or low levels of salidase activity the amount of product is the same. in this part of if dose response

curve the release is not enzyme limited. this is again something that we at least in my lab haven't thought about in context of we see a gene upregulated does that mean the consequence will be more. and i think it may well not be

the ways. and in conventional mice we can see that's a activity is very high, well into this range where changes in the amount of salidase activity if we deplete how much producing microbes are there is going to have no affect how much acid is released.

so we can think about this, within context of antibiotics in terms of supply and demand. we know acid is not determined by salidae levels because we can tune and see the response but the demand is determined by how many microbes are there because conventional mice don't deplete

acid again to zero. in other words when we think what antibiotics do, decrease demand but won't affect the supply. that explains in a general way why the acid levels >> up with antibiotic. we don't need a specific microbe

or antibiotics to explain what happened. to summarize, we can model this using parameters from the experimental data showing how much antibiotic effect would be expected to increase sighialic acid. the first thing to summarize to

leave you with is new york lawyers do not cause more sour cream consumption, you should think about that in terms of microbiota studies will, it's becoming possible to tune microbiome gene expression to help connect genes to functions. for many of the things we're

thinking about with the microbiome whether or not we're considering dose response curves and times is something that should be considered. this was recently published and i will stop and thank the group especially the post-doc who did the project.

>> wonderful our second speaker is dr. wei jia from university of hawaii, bile acid move cross talk and its effects on liver good afternoon. very glad to have the opportunity to talk about our research in the area of bile acid and microbiome cross talk.

and in fact -- their effect on liver cancer. i'm wei jia professor university of hawaii cancer center. the uh cancer center is national cancer institute designated cancer center so the cancer programs are supported by several well established culting

edge core fa sits at the karen center which which is metabalomics core facility. i am faculty director of metabalomics core. one of the highlights of our lab is that recently we have developed a microbial metabalomics protocol that

quantitatively profiles somewhere between 200 and 250 small molecule metabolites that are derived from host microbe co-metabolism. we believe this is a very important tool that provides unprecedent opportunities for functional characterization of

gut microbiome in addition to the meta genome sequencing. in a review paper that we published along with dr. jeremy nicholson and other collaborators, we summarized a number of important metabolic pathways involved in the gut liver and gut liver brain

metabolic interactions one of which is bile acid nuclear receptor act seize. that we are going to focus on personally i like to study bile acid metabolism because i believe that is the best model, the best pathway to study the interactions between gut

microbiome and liver because the bile acids are produced in liver in hepatocytes to what we call primary bile acid. ca and cdca and will be stored in bile. after they are conjugated with amino acids glycene or torene. then they will be released to

small intestine where they're extensively metabolized by intestinal bacteria through the conjugation, through hydroxylation all the way to what we call secondary bile acid co-lick acid and lethal co-lick and these secondary bile acids are comparatively speaking much

more cytotoxic. than primary bile acids and then they're transported to the liver through a number of interestal transporters like ost alpha, beta mrp 3 through portal vein and then hepatic npcp transporter to complete internal hepatic bile acid circulation.

so we know that biosynthesis and metabolism has been considered a common etiological factor for almost all human liver diseases including alcoholic fatty liver disease, and hepatic cellular carcinoma. we also know that the intracellular accumulation of

bile acids especially microbial metabolized secondary bile acids they are more cytotoxic and it has been proposed as a mechanism of colo static liver injury. we suspect increased hepatic bile acids maybe a mechanism of liver carcinogensis. so we recently conducted a

targetedded bioacid profiling study of liver fibrosis patients where we recruited 1,006 participants including 504 patients with biopsy confirmed chronic hep b or associated with fibrosis of various stages as well as 502 age and gender matched healthy subjects.

so bile acid profiles tell us that they can clearly differentiate different liver disease phenotypes we have healthy control, black points here, we have chronic hep b+ fibrosis. we have cirrhosis. and this results were verified

in 339 patients with chronic heap b with liver fibrosis cirrhosis and hcc. one of the striking findings is they're substantially in patients conjugated with doreen or glycene are increased by a factor of 175 in this case, 156, compared to healthy controls

which suggests high levels of bile blood and hepatic bile asits may not only be result of gut microbiome and liver function they may also contribute to pathological development of liver disease given cytotoxicity of these bile acids in nature.

so we conducted animal study to find out if there's such a mechanism, the animal model we use we develop is mice model inject very low dose into the baby mice shortly after birth and then after four weeks we feed them with high fat diet and at five weeks, immediately a

week after the fatty liver phenotype shows up and at seven weeks the -- becomes evidence and nine weeks fibrosis phenotype becomes evidence and 12 weeks nodule evidence and 16 weeks hcc is developed. so by the time we sacrifice all animals, that's 20 weeks or the

all the male mice developed hcc. so this is a very nice animal model that it produces liver cancer phenotype and gives a very nice window of different stages of liver diseases to allow us to take samples of different time points to find out what changes are involved

including gut microbial alterations. so compared with the controls, the stz, high fat diet group exhibited significantly lower diversity based on number of observed species and we also observed extensive species interactions and strong

co-exclusion and co-occurrence relationship between the phylotypes. and the correlation study suggests that alters microbiota at different stage of liver disease is significantly correlated to hepatic and fecal levels of bile acids.

so we see significant increase in abundance of gram negative such as bacteroides and spp leading to increased lps in plasma, liver and stools so in liver of model mice we found that key bile acid transporter and number of enzymes were significantly inhibited one of

the major transporter called bsep which is responsible for transporting hepatic bile acids into the bile into the gallbladder is significantly inhibited. you can imagine that when this major bile acid pump is shut down we will see increase

intrahepatic bile acid concentrations which in turn promote progression of liver diseases. one critical step towards development of liver fibrosis is actually the activation of hepatic stem cells. we observed the bile acids

particularly glycene conjugated secondary bile acid, can readily activate hepatic cells. these microbial metabolized bile aids is when conjugatedded with glycene or doreen have the most significant effect of activating liver fibrosis and promoting hsc proliferation. why male mice

successfully develop cancer phenotype almost none of the female mice developed liver when i say almost, i think one out of ten female mice developed tiny tumor here. it is significantly decrease in terms of formation of liver cancer what is the reason behind

we found they were significant difference correlated to bile acid metabolism between normal male and female mice and such differences were amplified when mice of both sexes were exposed to stz in high fat diet. as a result hepatic bile acids are significantly increased in

male mice compared to the female mice upon stz high fat dyes so we concluded it was increased level of secondary conjugate bile acids accumulatedded in liver that cost the problem so we tried to remove these or reduce production of intestinal bile acid by using an exchange

and to our surprise the therapeutic effect of this treatment is very good. none of the treatment group, none of the treatment group has developed liver cancer. increased level of bioacid liver were attenuated after the treatment and relative balance

o significantly altered bacterial genome such as clostridium and et cetera. were normalized after treatment. so the clear we are getting a clear picture now, so high glucose high fat diet hits gut microbiome causing the microbial compositional change leading to

increase secondary bioacids, bc lc which in turn circulate back to liver. and conjugate with glycene or torene to induce hepatic cells and develop to promote development of liver fibrosis which leads to further progression of liver disease

such ads cirrhosis and liver cancer. we thought this was a novel strategy to study disease mechanism involving organ interactions but realize this is a challenge in such an omic driven mechanistic research that involves a large number of factor, we have microbes here,

large panel of metabolites to also we have multiple protein targets involved in mechanistic pathway that all colocated in multiple organs and tissues so that is technologically challenging for a research group of average size to completely figure out what the picture is.

so with that i will stop here by acknowledging my group members and collaborator and more importantly, the nih grant supports. next speaker is dr. cindy sears from johns hopkins university, she will talk about biofilms as a risk factor for human colon

>> so i too want to thank liz the chance to speak at this very thought provoking meeting. i think you're going to find what i say builds on concepts introduced by harris waning yesterday and then andrew gerwitz and justin sonnenburg i too want to thank the nih, i

have a career because of support from the nih and most of what i'm going to present has been supported by the nih. so what i'm going to do is tell you why study colon cancer in one slide. and then i will talk to you about biofilms as we observed in

sporadic colon cancer, tell you a little bit about biofilms in hereditary colon cancer sum up and tell you what gaps i would suggest be worked on. so hypothesis and goal of this work is thought by defining the microbial drivers of colon carcinogenesis we can design

knew approaches to the prevention of human colon cancer we would argue that's important and there's several domestic health reasons why that's so when diagnosed early colorectal cancer which is a leading cause of cancer death men and women is a completely

preventable disease. yet there will be 135,000 new cases in the u.s. this year in a rather alarming report published by the nci in february, it appears there's now rising rates in both men and women under the age of 50 years of age of colorectal cancer.

if we wish to work on healthcare disparities and indeed we do, the highest incidence in death rates for colon cancer and blacks in the united states. we also argue this is important from a global health perspective in colon cancer rates are rising globally with 80% increek by

2035 when anticipated we may see 2.4 million cases of colon cancer around the globe. if we think about global economies, not everybody on the globe will get a screening colonoscopy at age 50 or younger. so we need new approaches to try

to identify those at risk. so microbial colon cancer field toggles between two extremes one single species may play a role in the initiation progression of colon cancer, there are several organisms that targeted i show pictures from the mouse colon of mice colonized with an organism

we worked on as an example you can show single organism to complex microbiota and indeed susceptible mouse model you get a dramatic change in the but there are similar though less robust data for the pks producing e. coli or containing e. coli among other

streptococcus gataliticus. on the other end of the spectrum is the idea there's something about the complex microbiota is able to be pro carcinogenic in the colon. and it's this end of the spectrum i will talk about today focusing on the power of

community to be onco genic in so you understand the pictures that i'm going to show you, i want you to understand how we have gotten the samples that we have worked on. so it's taken a large group of people and coordination with large group but what we do is go

to pathology sweet with that the pathologists take surface of tumors then we take the normal colon mucosa as as far from the tumor as we can get in that particular resection. those are the cancer related specimens. and they're studied as a pair.

in the colonoscopy suite individuals undergoing colonoscopy, we have random biopsies, right colon, left colon of normal mucosa. this is the bottom line of what we saw and work initiated by kristin and carried on by julia drew, this is the sum of samples

from johns hopkins an second cohort that was in exactly the same way. in malaysia with (indiscernible) at the university of malawi, all dots are biofilm covered tumors, ends flank normals and the blue dots don't have a biofilm. what we initially learned at

hopkins to our surprise is that essentially every tumor proximal had these complex communities. what we learned in malaysia was that one difference between ma lay shan population and the u.s. population was there was increase in left sided tumors with biofilms in malaysia.

such we recognize sigmoid area this appears to o cur. when biofilms are present both cancer and normal tissues taken as pairs are nearly always concordant for biofilms and doing this requires in the pathology suite we're putting the tissues in to a fixative

which fixes the spatial organization of the mucous and microbiota relevant to the mucosa. what do i mean by biofilm? this is a cartoon from gunter hanson and here is the dense inner mucous layer that coats the entire length of the colon

and largely enpenetrable most part exclude bacteria and there's another mucous layer with lots of bacteria in it right adjacent to this layer and then of course there's the so when i say a biofilm what it means is we detected bacteria in mucous layer at high density and

going at least 200-microns along the length of the colog. and in addition, in that mucous layer they're swathed in polysaccharide capsules as well as glycans within the mucous. when fluorescent bacterial, bacterial probe is used, this is what a biofilm positive tumor

looks like. so the nucleus of the cells and there you can see dense layer of bacteria adjacent to the epithelial cells. if you go to high power you can see the individual bacteria. here is this tumor's flanking normal you see exactly the same

and you don't see that in these tumors from the left colon, where you can visualize empty mucous layer here. this is confirmed by scanning electron microscopy but not seen in sample that was thought to be biofilm negative by the microscopy approach.

we went to use a dozen bacterial probes different groups and even down to species level and determined in sporadic colon cancer these are polymicrobial and a feature we picked up in more detail was a portion in this community invade every tumor that is biofilm positive.

we turn to those individuals undergoing screening colonoscopy do same types of analyses. what we learned is that about 15% of individuals undergoing screening colonoscopy will have similar polymicrobial biofilms on top of their epithelial cells but they don't tend to be as

thick or as thick really as those in the cancer situation. other thing i wanted to say, in this case the biofilmings are random throughout the colon so this geography on the right side is seen in the cancer patient but in the healthy patient they're actually randomly

distributed and present in the right and left colon at that point in time. so as justin sonnenburg said we're looking at the spatial organization of the microbiota in this setting or the biogeography would be another word.

they come in three microbial flavors if you will. most common is those dominatedded by bacteroides ads shown here. there's about 30% that also have bacteroides and embedded in there had are balloons of fuso the third type which is rare,

are those that are dominated by rodeo bacteria, and i can't tell imlach more about that at this point but it's certainly a focus going forward trying to unravel what that means. i don't have time to show you primary data but this is a summary how to put the data

together as we pursue hypothesis, biofilms and healthy individuals would increase the risk of developing adenomas or colon cancer. change barrier function as we saw changes or loss in e cadherin. there was immune activation

within the mucosa as detected by the proonco genic cytokine il 6, a well known activator of transcriptional factor highlighted in cancer verge stat 3, in addition working with gary zusdac and carolyn johnson scripps institute we identified a rare polyamine metabolite

increased in setting of biofilms and n 1 n 12 diacetyl spermen molecule. the data suggested this question came up yesterday, the data suggested this was being made both by the bacterial biofilm and by epithelial cells. these factors were associated

with clonic epithelial cell proliferation, this is taken from a left colonoscopy control biopsy just to show you an example, this is ki 67 staining as measure of proliferation. and you can see that there was increase proliferation biofilm positive compared to biofilm

negative sample aggregate data and to remind to say any biology we have seen on right sample has been the same on the left sided sample. now to try to extend this work to understand epidemiology and significance of biofilms in healthy individuals with the

support of the nci, we're in the midst of a 2000 person longitudinal colonoscopy study, we enrolled 600 patients the first year and as consistent with some of the discussions about longitudinal studies yesterday, we estimate about 50,000 samples in the freezer to

do additional work to try to understand relationships between biofilms and human biology. a question is colon biofilms carcinogenic. to try to answer that question, we collaborated with christian yobin, university of florida in their germ free facility.

what we did is created an inoculum of biofilm positive tissues either tumor flanking normal or colonoscopy biopsies. as a negative control or negative sample, the biofilm negative colonoscopy biopsies. these were inoculated into mice. that have -- apc heterozygous

and therefore prone to get colon tumors, multiple intestinal neoplasia mice or il-10 knockout what we saw aggregate data here, the biofilm positive samples but not biofilm negative samples induced tumors in the coo lon of the type of mouse dinner to the tumor result.

and surprise was each of the biofilm positive samples were approximately as onco genic which was a surprise in these thinner biofilms in healthy host were carcinogenic as those tumors in the cancer host. so we view this as a great opportunity to be microbial

sleuths to try to understand what within those communities actually is contributing to oncogenesis in the colon. this is just one slide, a snap shot of our data from hereditary this is data from the syndrome known as familial polyposus. apc heterozygous individuals.

these individuals will get polyps in their colisten and will get colon cancer. so they eventually all lose their colons which are removed to prevent colon cancer. over six years we had the opportunity to collect samples from four as shown here.

the red dots biofilm positive, blue are not. throughout the access in the colon, each patient there were biofilms detected. in addition i'm just showing you species specific probes here because to our surprise the biofilms differed from what we

saw in sporadic colon cancer. in that you see there's a lot of red, there was e. coli and there was at leastbacteroid fragelis. similar to sporadic colon cancer we saw invasion of the bacteria into the tissue and when biofilms are cut out and pcr done, we were able to detect the

colypactin gene virulence factor of e. coli with biogenesis with the tox tic gene with the tumors with etbf. lastly, we were able to go to burt vogelstein's bank of samples, 25 years in the making and identify another 25 patients and controls and using now

microbiology we're able to recover pkse e. coli and etbf with half the samples having both organisms and overall more common in the tumor patients than in the controls. so how do we put this together? we know etbf and pks e. coli commonly colonize children

before age 5. we know much less what happens in the epidemiology of those organ subsequently though a small amount of data supports that at least some teenagers and young adults remain colonized. as discussed extentatively here, there's -- extensively here

there's many things that modify microbiome over time so we might suggest host bud collaborations occurring and we have mouse data that suggest these two organize. s are synergistic in capacity to form colon tumors. we would further suggest biofilms may develop over time.

late sporadic colon cancer we see biofilms in half tumors. so the gap to be addressed is not novel, it's been discussed here and that is defined causality or not. microbiome in human disease and specifically in the setting of colon tumors does biofilm

formation or presence of any specific microbe or community consistently perceived development of colon polyposis in humans. this is a challenge because we know the time from the clonal expansion of that cancer cell in the colon to the visualization of polyp is ten

years and maybe up to 40 years. so these chronic slow diseases which being discussed here pose a particular challenge to try to link microbiome to disease. terribly forgot, i show you pictured of who did the work over time but i really do want to thank the entire lab and all

of our collaborators as well as again nih which has provided most of the support for this. >> our last speaker is dr. gary huffnagle from the university of he will talk about the role of the microbiome in respiratory health and disease. >> okay.

cool. first i would like to thank lis and lida for the invitation to come speak today. and so also to represent the unrepresented mucosal site in the microbiome which is the lung. so this statement and this book

and many similar to this, has not only cut me up -- kept me up many nights, seeing my grants sometimes go down in flames, and driven my research the past seven years, which question is that a cause effect that's a different story. the other hat i wear at the

university of michigan besides being a researcher is i now teach introductory microbiology to the undergraduate campus. 200 students in roman coliseum style. if there's a message i want them to take away, from that course, it's this.

there's virtually no environmental niche on the earth that's so extreme that microbial communities cannot be found. so really beggings the question for that statement, why did we ever -- begs the question why did we ever think that a warm moist environment that is just a

few centimeters away over which air continuously flows would be completely free of microbes, would be sterile? so that would be a whole 'nother long discussion and i'll show you the data and things like but i then if you're going to believe me at that stage then

let me take it a step further. while the lung are not sterile, they are not a great environment when healthy for microbes to grow. especially colonization for an arobes gram positive back fearia and others. why?

first off if you're an an arobe in the mouth and you get whipped out and flown in the lungs it's an aerobic environment. so you're not a happy camper that way. there is also cilia constantly pushing the upper airway mucous back up so you have to fight

we think about the body temperature between 37-degrees turns out that the upper part of the lower airways is actually not 37-degrees, it's lower temperature so number of bifurcationses before the wall hits 37-degrees so you have then lower temperatures in your

trachea first few bronchi. we can't dismiss surfactin. so it's lipid, is a detergent so you have microbes from mucous rich environment being thrown into lipid rich environment and especially if you're a gram positive organism it will blow a hole in you.

finally, there's very little mucin, there's food, very little food in the alveoli and we don't think about the fact that the lungs are the only place in the body in which there is always a leukocyte that is outside the external surfaces, is a patroller, is alveolar

macrophage patrols that alveolar external surface things are coming down. in other words it's a predator. if we use ecological models it's like having a protozoa. so you have bacterial protozoa interactions essentially. not a great phrase when you're

healthy for organisms seeded from the upper part of the air wei to come down. so thus the lung is a low microbial biomass site when you're healthy. i want to point out, i think typically learn this in chemistry 101 between accuracy

and precision. the precision is you can make the same measurement again and again and again, accuracy is are you correct. so when we talk about things being below signal and assay for microbe that is not the same thing as being sterile.

just because you can measure many times i don't see it doesn't mean that's not necessarily there. we heard a number of presentations about technical limitations, of our assays. reagent issues again, i can go on for long period of time and

bring rick and ron up here to talk about reagent and kit noise especially when dealing with low biomass samples. are you using the correct assay, et cetera, et cetera. so i bring this up because there's a paper that came out in two months ago or month ago that

claim mice their lung moye, a portion have a sterile lung so i want to point out that it's not sterile, it's below limit of assays and by the way, we see the same thing in mice also. i will get to that at the end of the talk. mice have an even lower biomass

when healthy in lungs than humans do. major challenge for this field is study of lung microbiome has been and always will be the lower airway sampling is invasive and not amenable to serial sampling over time. this whole idea of the elephant

cartoon we have seen come up many times is really, really relevant to the study lung many weighings we can look at the lung microbiome, bronchoscopic brushes sputum in postmortem tissue all give information not a single one gives us everything.

so it's the collection of putting them together that then give you the big picture what's going on. let me give an example of one of those sampling techniques. this is lavage so this is a study we had that was supported by nhlbi.

it's 28 subjects. and we look -- we took samples from the nose, the mouth, gastric and bronchiolar alveolar lavage. and analyzed the microbiome. so this is an odder nation, if we plot samples on here and the first thing you can see which is

exactly described in hmp is the nose is distinct from the mouth. the gastric is very similar to the mouth. which makes perfect sense saliva flow from the way we were sampling this. if you impose on top of that, lavage samples what you will see

is they're indeed distinct statistically you can show these are samples are distinct from oral samples gastric samples and nasal samples and but you can see it's not separating until you get to this second component it fits the model of the lung microbiome being a target

community whose source is coming from the upper part of the airways. but dig deeper into the data, it will tell you more about human variability in the lung microbiome which is if we actually look at similarities and dissimilarities between oral

communities of an individual compared to what we get from ba, also each individual we can use them as their own control. what we see is some individuals in which the bal community is much more similar, look at this one down here, to what you get out of their mouth, and others

in which it's very distinct. go to the nose, as a source community as air flows through, most individuals there's no similarity but indeed out of 28 we have 8 there, that you can see some similarity to it. there is variability of between what we're looking at in the

lungs. so this is a type of chart we have seen in the microbiome reviews so i just throw this up here because now we can put a pie chart for the lung. so we know -- we have known for a long time that microbial communities -- let me put it

this way. when you have respiratory disease you quite often isolate bacteria from it. what we have learned over the past again five to ten years is the microbiome we can measure by culture independent analysis differs during disease states

compared to what we see in healthy. the note i put up here is the magnitude of these changes that we're seeing here is -- is not historically captured by standard culture techniques that's been kind of a shock, like wow, there is more

difference during disease than what we expect based upon what we can culture. i want to give you one example of this, this that we have been following which is if we look by culture independent analysis at the supermonas, these are taxonomic units, we have six

ones, so all these trait tax no, ma'amically identified as pseudomonas we get diversity of that in the human lung, no t a lot when healthy but definitely during disease states, so these are lung transplant recipients symptomatic and asymptomatic for subsequent disease, these are

bronchial lavages from fibrosis these are brushes from chronic obstructive pulmonary disease. then you see a number of pseudomonas. one is originosa. the other two are not. so throw out there, well, it might be contamination or some

environmental sort of thing, and that was definitely something we dealt with. i don't have time to go into it here but i can it will you these indeed are real. it's just that we're not finding the right way to culture them and we recently have identified

a way to culture these out. and it's actually again, telling us more about what that environment looks like these organisms are flying into. so what surprised me is unlike studies of the gi track microbiome culture dependent analyses of the airways have not

identified significant numbers of uncultured bacteria. i expect to find new bugs an things like that, it's still the same ones we have known for however we're not growing them out all the time. so you can go back to the literature, we have seen these

all through discussion of respiratory disease of chronic colonization like pseudomonas then it reappears so what's so so what we think it mean it is there's a change in vitro conditions required to culture these and reflects a change in metabolic activity of the

bacteria, a change in nutritional environment in the lungs and subversion of host defenses. so our context with which we're looking at all these is the following model. we believe the constitution of the human lung microbiome

determinedded by three factors. the first which is immigration. so organisms coming in and we think the main thing when you're healthy is microaspiration. you can have inhalation of bacteria, you can have direct mucosal spread and obviously during disease you can also have

blood and lymphatic immigration. that is balanced by microbial elimination so clearance in adaptive host responses, so i want to point out for that last innate host defense, i have just completed a study we look at 40 inbred mice different capables different inventorytors, a lot

of information about that. the one thing i want to point out to is the following graph; these are completely healthy mice if we look at diversity of microbiome and their lungs we compare to concentrations of il 1 alpha in the alley owe lar lavage, we get a wonderful

correlation. and actually really stands out if i put quadrants on here, look down here, it's basically saying that when the community becomes less diverse we no longer have low levels of il-1 alpha, in fact the only thing that ever find are higher levels of il-is

alpha, it's of interest because it's a pre-formed cytokine, it's a fast released cytokine. so our interpretation of this when something tries to colonize it is pushed out quickly by innate immunity. in fact, this diversity of what's going down from above is

actually the normal state. that was in mice. this fits perfectly with what the nyu group led by leo sagal found, human lung microbiome that when you start to get domination by organisms, you also then get low level inflammatory cytokines being

made. in healthy individuals. the other part of the model is all this once you have a balance between elimination and immigration and elimination, once a microbe is down there regional growth determinations determine what you're going to

see down there. so in relevance to this session, what are the mechanisms we need to be looking at, one thing that we have been looking at and others have been looking at but borrowing from research especially number of things pointed out here is inflammation

in and of itself, changes this environment. so you can see emergence of pathogenic strains from non-biofilms or potentially the outgrowth of transient microbes. you have host derived factors in metabolites that we think are being generated by inflammatory

response, catecholamines reactive oxygen species, triptofan metabolites so this results in shifts in microbial physiology and i will argue can result in changes in culturibility. why is that of interest? because that that's what we use

to detect organisms. so in the end when healthy it's immigration and elimination. once you get down to the lungs bacteria can grow. but once you start to have lung disease then the growth conditions in the lungs themselves change.

so my final point involves high-tech animation is to ask question when we talk about animal models and studying the human microbiome does this lung ecology model apply to other species? you think air flow and humans coming in and going down, what

happens if we take our subject creature here, turn airway sideways, change their breathing patterns. so mostly going through the mouth obligate nasal breeder the mast vast majority goes through the nose. which is an even lower microbial

biomass site than the mouth and then you add in extended complex nasal determinant. what are you going to end up with lower immigration rates in the lungs. in other words if you're going to use mice to study the lung microbiome don't be surprised if

you're not going to quote unquote find any bacteria in the lungs of healthy mice. because it's there. we have got that data. it's just really low because same thing, same when we talk about pigs because there's study about pigs in the microbiome.

so the model does fit. so in summary, the lungs during health are not free from bacteria, rather than harbor a low level of bacteria derived from the oral cavity and nasal cavity, so there's individual variation in degree similarity between lung microbiome, and in

the same individual. this lung microbiome that's dynamic, represents a balance between influx elimination and growth but we know changes in the lung microbiome associated with respiratory disease. so the final point is then what's the gap and is the gap

trying to solve here, it's really acute in study of lung microbiome which is are changes in the lung microbiome cause of airway inflammation? or the result of airway innamation or both? so our goal has -- as a research community move associative

studies to respiratory disease. with that i would like to thank my laboratory at the university of michigan plus a number of wonderful collaborators that made this work possible and then of course my funding source the nhlbi. >> we have about ten minutes for

>> i have a question for dr. jia. in your model you suggested hydrophobic bile acids responsible for notch and the development of hepatic carcinoma in this mouse model. the (inaudible) with this interpretation is that based on

(indiscernible) specific, binds to many hydrophobic molecules in the gut in addition to several bile acids. in addition clostridium mice change microbiota in major way so many molecules are expected to be changed in treated mice. so you heard any other data that

supports the model that similar bioacids are responsible for notch and hepatic cancer in this model? >> this is a great question. i don't believe the hydrophobic secondary are causing the fatty liver in nash. what i was saying is these

hydrophobic bioiceys circulated back to liver after they conjugated with lysine or torene they are more ohio they can activate hepatic cells which induces the liver fibrosis. that's a major mechanism for the bioacid induced liver disease. so the treatment removing or

reducing the level -- intestinal levels of secondary bioacids would -- bile acids would help elimination of the hepatic bile acids, especially those secondary acids that pretty much stop fibrosis process i believe. it's not stopping the nash and fatty liver because we feed the

mice with that high fat diet. so i wouldn't be surprising that the liver will become -- the enzymes are enzymes to remove through pxr and other receptors use enzymes in the liver in the hepatocytes to remove toxicity from bile aid sides. before or after they're

so i feel it's more complicated. >> that's another way to do it. to -- by changing the activity of some of the transporters you accelerate bile acid circulation you can also reduce intrahepatic levels of bile acids that way. >> my question is for dr. sears, (inaudible) columbia university.

nice talk, cindy. my question is in your germ free mice study, where you inoculate a human biopsy, have you looked what kind of bacteria do you recover for those mouse tumors? >> we're just starting. we don't have any comments. christian did sequence them and

found similarities across the samples that induced tumorigenesis but the correlation work and culture work hasn't been done as yet. >> do you see a difference in potency of the biopsy to generate tumor whether it's -- what kind of biofilm from the

>> the inoculum was actually a composite of five patients and we put some patients with each type of biofilm in the tumor inoculum. in the healthy patients, we only see one phenotype and that is the polymicrobial. we don't detect any fuso

bacterium in those and there's no proteobacteria document nans leucine to date. predominantly in -- but again we haven't -- we're just developing a strategy as part of our prospective colonoscopy study to try to get at this in a little more detail.

>> phil dasher, quick question for cindy. given what we heard this morning dr. dr. gonham on the fungi in biofilms, have you had any -- have you looked for any fungi in the biofilms that are tumor associated? >> no.

we have not specifically used probes would pick that up. it's a good suggestion. >> it's only we're trying to design a study to do in parallel 16s meta genomics rna seq to try to look across the spectrum and in that setting should pick it up, i believe, if it's there.

>> if no more questions, thank you very much. >> so i will kick off this round table discussion with an introduction to who i am. i'm a program officer in the division of microbiology and infectious diseases in the national institute of allergy

infectious diseases. my name is ryan ranallo and i will guide you through this next 30 minutes. hopefully we'll have both a series of conversations as well as discussions so one of the things that i definitely wanted to do is thank everybody,

certainly speakers, i see many faces and seen you here for numerous workshops that we have had over the years, and i feel like some of you need an office here you're ear here so often. but nonetheless, so a year ago when she started planning the meeting one thing i asked her is

what's the purpose. that is outline and speakerrings have done a great job identifying gaps, technical hurdles and new opportunities that potentially lead to future interventions and new approaches so those four things i'm focused i'm happy to have extended q&a

but i'm working for lita today and now. that is my goal to focus on those four things and the speaker versus done a great job. something i thought would be an interesting idea rather than focus perhaps speakers focus on their own challenges or things

they brought during presentation perhaps to think about what other speakers highlighted and maybe echo those comments that, might be effective way to get questions and to maybe help us certainly as program officers and administrators thinking towards the future.

as hmp winds down an institutes contemplate their activity this is this area. i tried toe generate a brief list, just to jog everybody's memory about maybe some gaps. so some of the gaps perhaps repeated is really we had a significant focus on mechanism

and determination of causality. gaps in terms of annotation and metabolites, and look at and takes. we had two wonderful talks on other so phage viruses an fungi as a gap and i challenge everybody at least to say why are you still ignoring them the

question is why is that. the other thing that i heard in terms of the gap is generalizability. of findings and model systems. so some technical hurdles i wrote down, i thought maybe that was interesting the look at validating results in different

lab, comment of novemberty over rigor, is is that a cost issue. this is a big technical hurdle, what's in food. i heard that a lot this afternoon. is that a key challenge. new approaches spatial geography, biofilms.

new model systems fecal bioreactors are engineered tissues such as organoids sampling the entire gut culturing the unculturable. so that's what i have listed i would like to hear speakers talk we'll go with that. otherwise we are happy to take

questions and answers. so i'll look at the speakers and see if they have any thoughts on what i just said and perhaps maybe echoing that. >> and dy goodman. -- andy goodman. i can start. to be controversial, i'm not

sure we could all look at everything. i think there's a theme of that viruses and fungi and eukaryots in the microbiome is certainly a fact. but i don't know necessarily most effective way to move forward is to have everybody

look at everything all the time and another model would be yes we agree these are there but that certainly who focus expertise in certain areas. i don't know if people have comments about that, whether we should look a these things or another acceptable way forward

is not do that. >> i have a comment based on studies we have done in the we had this crazy phenotype and in mice microbuyialsis and shown bacteria up or down in this model what we found is a parasite inducing and if we would have -- if you ignore that

and focus on the bacterial side of things, we would have -- we would have made different so i agree with you, you can't be the expert in everything but you also cannot ignore the contributions of other kingdoms looking at again most model specific.

>> i really like that question because it really speaks to this issue of whether there's group think or some prescribed orthodoxy of religion around the analysis should be informed by the boy logical question, the biological system. if's novelty in analysis it's

driven by questions. i home the next phase of microbiome research is directed at a thousand garages rather than a few tall towers. that was a challenge and problem with the first hpp project limited explore ration of diverse populations and systems.

it wasn't that it was an error in thinking it was just a way it began. but i do worry about prescription and the way people do science. but i also think subsumed under your question is what needs to be co-evolved in the field and

conditions for innovation. and conditions for innovation is the field microbiome, it brings together we heard nicole's talk, we heard your talk, you can find structural biology and genetics in elegant ways and other people's talks. what kind of learning

environments, we have to create for students in this area. where you can encourage people to come together within academic units, where there's enough support kindness and generosity so people around afraid to say i don't understand and you can teach one another but also where

you bring together the capacity to apply rigorous statistical and computational tools together with whatever other biochemical metabalomic, et cetera. so i worry about the education and how we're missing an opportunity to ensure our in that sense. the last part at

least my response and my gut response to your point is there was a lot of appropriate plaintiff and not so plaintiff pleas for better understanding of gene function. and that comes back to something that you complained about in the port now sense but rather we

don't have genetic tools to deal with a lot of these organisms. that's a shame. it's a shame because it's hard, innovation because of gene manipulation, with crisper, cas, if we reserve that investment now, for the future, we're really going to be handicapped.

know your targets, look at mechanism but be able to manipulate the system whatever ways you think are wise. >> i want to speak to what you said we do better, what you both said we do better to develop expertise in our area. and then bring that expertise to

collaboration. rather than us trying to do everything in our system. >> so i would add one other thing to that which is then also the need to continue to develop expertise or support individual whose are indeed the jack of all trades.

because if everybody has their areas of expertise which we absolutely need, especially interdisciplinary field, it's helpful to have people who research is constantly moving about back and forth to stimulate ideas and help sew all the parts together. thinking

ant a broad land scape versus being somewhat more focused, there is a divide between basic science world, we have wonderful technologies really understanding mechanisms and clinical relevance of disease pathogenesis. i think it actually justin and i

were at a meeting at nih and we had a conversation with one of the program directrs. the conversation oriented towards having a clinician that understands what's relevant in what are the biggest gaps and biggest struggles that physicians and surgeons have in

treatment of their patients, having a conversation with basic scientists and physiologic and mechanistic understandings bridging that gap. having the scientific field focus on things maybe more low hanging fruit or things much -- the most highest priorities that

the scientific field can use their best technologies to try to address. >> (inaudible) we heard a couple of people talk about reproducibility. and in the metabolism -- this is more for mouse research, the metabolism aging field there's

debate about housing and diet for example temperature, whether animals should be housed at thermal neutral versus the common room temperature which is cold for the mice. i was wondering if this debate has moved on to the microbiome field and whether temperature is

studied with in the context of microbial research. the other question or the other parameter diet. we heard about standard chow and we saw picture of the beautiful dissected gutted and what the healthy gut looks like. my colleagues though seem to

always complain about standard chow and mouse studies. so i'm just wondering how do we know what evidence do we have standard chow is optimal or great diet and the only specific question i have in that context, what does the gut look like in wild caught mouse and when you

take that wild caught mouse and standard chow for standard of time or multiple generations does the gut still look good, different, or not? >> i guess i'll make a comment because there's just silence i had a conversation with somebody in the audience during

one of the breaks. it gets to the question of whether or not you define diet that's control whether or not you get a study wild type mice like mice in the field, because they're going to be more physiologic than the inbred mice we use in the laboratory.

and my comment was i'm not sure that's the right way to go. because as we discussed, humans are exceptionally noisy. there's a lot of intersubject variability. and there's a reason why we study mice and we study nodobotic mice and i study

inbred mice and as we can lower the noise level so we can see a signal. sew if you wanted to do -- if you wanted to do nice mice in the field and undefined diet with you're defeating the you were p of using the model system because noise is so great it

will be very difficult or expensive or take a long time to see some type of signal. i think the reason we use these defined model systems is see that signal and then extrapolate and figure out is that large enough or is there some way we can actually study this in human

but if you don't -- if you can't see the signal, you're not -- you actually can't take the first step from a mechanistic standpoint. >> from to follow-up on that, finding what the standard is you're going to ailed something and look at the effect you have

to keep in mind if you start a high fat diet your chance of impacting that is preventive component, maybe higher if you start what might be a more normal diet. in that context unless you do additional studies you're under the impression that you're

dietary supplement or phytochemical of interest is going to be a great magic bullet. but in reality under normal circumstances may have no effect at you will. -- at all. trying to standardize but really

understand what is your standard diet doing. >> a quick comment standard this is a significant topic of conversation the earlier workshop this summer related to the microbiome and diet but it was it is a complex subject one thing we grapple with in the

field is trying to maintain a community of microbes that's diverse and representative of a community that we might see in and to do this with a defined diet that's very -- to do this in a -- with elemental diet or diet constructed from the ground up is very difficult.

eric tried this with diverse fibers he put together and he hasn't been able to reconstitute something that mimics the complexity found in that list of ingredients you saw today for chow diet. it's in the as reproducible, there's ill defined things in

there but it serves a purpose for particularly large number of >> components of mass chow diet are ill defined and from a vendor so lack of definition applies to vendor as well as recipient. i don't know if that we would be able to cob strain sourcing from

the vendor, that would be difficult, it's helpful to use germ free animals fed a diet that you are applying to other con recognized animals to see what the bioavailable contents are of that diet in the absence of microbes. so targeted mass spec for amino

acids certainly micronutrients, the thing that is a huge gap that people were alluding to was the ability to define molecular composition. and i think there should be a campaign and there's been individuals who raised their hands and starting to do this

not just in words but in deeds to look at macro molecular content of the major food staples of the world. a thousand staples, different cultivars of a young staple that start out primarily looking at glycans protein content and micronutrients.

what i was getting at is it's unbelievable to me that advances in nanolc mass spec applied to breast milk. haven't been advanced, largely because glyco biologists are endangered species to plant based glycans and host based glycans, it's not just

monosaccharide andy saccharide it's the linkage analysis. tools are available now the start doing that work. the nih with enthusiasm, and focus and with nutrients. ie cash. should i think sponsored that campaign to be able to quantify

the non-saccharide and linkage characteristics of major sources of glycans. if we restrict diet to one major standard chow, you're going the lose a lot of biology but you have to know what you put in. as well as what comes out. >> may the force be with us.

>> i want to speak to the issue of temperature. i don't know if snib studied this or published this but anecdotally temperature is a very good point temperature is a serious issue and temperature shifts cause profound shifts in the microbiome so that probably

should be addressed in a more systematic way. or note whack the temperature to maintain mice. >> so you are talking about mouse chow and how it's standard mouse chow, at least there is some i would argue uniformity then you talk about human

studies and you do interventions with high fat low fat whatever. i would argue that there has to be some standardization there too and what you consider, as jeff, dr. gordon just now, if the bioavailability of different greens is different, though you can say calorically it's high

fat it seems to me a bit of apples and oranges. the difficulty for me asen editor is if i get two papers, yes, mandating the reporting of everything these days but science is a process of building on existing science. you have to have a way to say

something, some comparison in human studies, this came up and i think michael was trying to make a comment about this. the diet session. so even i would argue even in human studies for diet intervention, some standardization it doesn't have

to be what's given to patients or ulcer raytive colitis that's not palatable but it's great if the community came together and said if we say high fat interventions for humans are high fiber interventions, this is what we mean and defin it more precisely.

-- and define it more reprecisely. >> the fact that the microbiota transforms ingredients in foods forced us to have a higher resolution view what the ingredients of foods are. so the point i was trying to make is it's important to

characterize the diets you put in. i only worry there's just one diet used by everybody for their experiments because though that's a form of standardization, i'm not sure that long term consequences of that may be deleterious.

but it is critical to define the autoclaving of chow for nodo biotic experiments and everybody knows how valuable it is but it varies over time. whatever settings there are. so how is food transformed by processing, either processing to create the raw chow and the

processing that occurs when we irradiated for autoclaving so i would go back again to the issue of at least in a nodo biotic setting of trying to characterize what nutrients are available in regions of the gut in absence of microbes as way of determining bioavailability to

that mouse genetic setting in your vivarium at whatever temperature whatever time of year and whoever is president. >> i was wondering along with diet discussion if there's an idea of the best methodologies or procedures we would use to determine functional change.

what i mean by that so standardization diet and where along the gi track we measure for changes as well as do we do fecal matter versus allele scrapings. are we at the point we can identify what the best things are to do to see whether or not

there will be functional change in the microbiome or in the >> i would argue no. so i can tell you as an immunologist, one of the areas the entire microbiome field largely ignored is biology of the small intestine in terms of microbiology.

even as you read reviews, it's like the gi tract, the intestine is one long tube but it's not, it's not only a major feature difference, there's small differences yet we can't -- we haven't got to be the point of actually placing microbial community structure, microbial

metabolites on top of these specialized cells and structures at different points intestine that inform the rest of the immune system how to act. so i think a lot of what's the way things are driven, is largely by cost effectiveness and reproducibility but i think

there's still a gap in terms of there's no one special site or special recommendation how to do it's back to what jeff your question your biology has to drive how you do your sampling. >> i would say you have to understand your particular

organism for -- because they will each vary how they affect various portions of the gi and i think you should presume that until you prove otherwise. i would also say, i don't think the small intestine of the mouse is a place to study biology accept perhaps totally model

system mice are copper foggic so if you're interested in the biology of the small intestine, you are witted to very difficult >> there is a lot of enthusiasm and potentially hype around impact of microbial communities on brain biology. i think this is a very profound

area of science that we should approach with sobriety. so the issue of regions of the brain that are analyzed in careful description of where you do the dissection is important. it's also important, i use this as an excuse to bring up that topic, to invest money to

determine how we can apply at least noto biotic systems standard procedures, we're looking at learning and other facets of behavior. how much can you import into a noto biotic setting? what are improvements that we can generate be able to test

physiological responses within a nodo boyiotic setting. one is learning. how we can monitor physiology over time with implanted sensors and transmitters. to be able to capture in state of ox yes nation activity of the animal, all things we can think

that's an investment we can bring together bioengineers and this field in ways that would be productive. a related -- final comment to make, had to do something brought up in the course of the day is what are people's definition of biogeography.

and at what scale is the relationship between microbes and members of microbial consortia being quantified. and what imaging procedures can we apply. it's daunting but important. i mention in my talk garry spent a lot of time in that justin

made investments, it's really time for imagination and creativity. if we don't determine beyond computational aappropriate of co-occurrence taking a fecal sample and asking which organisms travel together to understand the nature of the

journey at a micron scale and determine degree of mixture is going to be the foundational knowledge to determine how microbes can collude and collaborate with one another, whether those are bacteria, whether those microbes are across different domains of

so i would plea -- make a plea to the nih at this moment in time to think about how you sponsor initiatives sponsor in a way that would evoke creativity and again this idea of how much funding should be for sort of garage like behavior as op posed to towered behavior.

huge consortia, which sometimes has a very valuable outcomes but also you want to have degree of versatility and nimbleness and bring people together around these compelling problems. >> my name is ab broadbandy glasscock from virginia commonwealth university, i'm

representing the twitter verse specifically for dr. gordon and dr. wu. twitter person asks have you considered antimicrobial residues in food such ads meat and milk products and their impact on diversity of the in the u.s. levels of

antimicrobial residues are legal in food animal tissue and how this impacts studies that are performed here in the united states. >> there's an advantage to try to formulate human diets that are representative of those consumed by the target

population you're studying. and apply them to some of the pre-clinical models. i think your question is a good one and are different sources of food ingredients is food science apply to formulate those ingredients into diet. components in the microbiota,

not structurally but functionally so is it's possible to directly address that important question and there are means an path forward for doing so. gary, what do you think? >> we formally haven't studied it but my notion is the levels

of antibiotics used in food may have a long term effect short term effects maybe relatively modest to the degree that it maybe difficult to actually visualize in some way. though i do or i are about the antibiotic resistance because that is a listening term effect.

also really varies by type of antibiotic we find in both mouse studies but also in human studies certain antibiotic we use at pharmacologic levels with minimal effect on composition. i don't know anything about the function but the composition of the human microbiota.

so again, it's the dose, the duration and the type of antibiotic that can lead to specific effects, so i do worry more about antibiotic resistance and the long term effects than just short term effects. >> decommissioned gastroenterologist it meant risk

for logoria so i'll try to avoid it but i interpret the question of the twitter user, as more general. which is to apply analytic techniques to food ingredients or diets including mass spectrometry prior to its administration, and within

knowledge can you do correlation analysis to see whether responses of community are related to the ingredients in the foods and do so in unbiased wa as much ads possible to create testable hypotheses that could be carried forward to a noto biotic.

one of the bioactve ingredients in foods. the raws, the food formulations and certain processing and how that relates to the bioactivity of that food. >> i think you then have to layer the fact nanobiotics are one of the most commonly used

drugs in the community as well as proton pump inhibitors all which have been shown to affect the microbiome and depending on your antibiotic you need to keep in mind individuals interesting example is when very large studies we're done with veiningmycin in individuals with

deep seeded serious staph aureus the iv veiningmycin over time begins to leech into the gut enough to precipitate clostridium did i have. it becomes affect long term antibiotic administration so dose duration, frequency and these drugs are very common.

so it's very hard to i think to -- it is a great question. it becomes a very hard problem to study because of the complexity of care. just simple healthcare. in the united states. >> one thing especially about jeff said with bioactivity, that

is it's absolutely true we think of antibiotics as therapeutic concentrations causing enrichment of resistance but antimicrobials are carbon they're food source in their own, they have various biodegradation pathways and there's literature in model

systems to show break down products of antibiotics may not have antimicrobial activities but they have antimicrobial patient type activities that subpoena another concern the twitter user could have been alluding to, the way diagnostics currently exist, for instance,

antibiotic use in dairy farming, very, very specific anolytes that are analyzed based on what we think bioactivity is. so i think there is a huge amount of work to be done antibiotic being one of those chemicals to what happens to the -- along the pathways of

biodegradation, microbially mediated what are the breakdown products of the infrastructure. tetra cycline, degradation products not only perturb communities but in some cases increase frequency of horizontal gene transfer of tetracycline resistance genes.

much much more complicated. >> i'm (inaudible) from the cdc. i want to bring up a question maybe we'll think about it tomorrow because this is a big -- translation, translating to the clinic and public health. i heard one question today about something, i think it was one of

the food sessions if we keep going this way it will be 30 years before we translate. we have to keep in perspective here, i'm not talking about the ip issues around product development, drug development and regulatory hurdles either. it's not the focus here but just

around these questions, i'm a little worried, and philosophical, back in the sequencing human genome, after that, there was a long time of development of getting anything translated. part of that was technologies that we're catching up but now

we're on the the trajectory of technologies and just see all this enthusiasm we know we could answer and find functional answers to things. and couple more years, couple more years, and the question becomes from the research community, because you speak to

the industry in a way, checktively, that is to try to figure enough understanding is adequate to go ahead do something. we have seen associations we have seen outcomes, as long as something is repeatable over and over again and you have the

desired outcome, disco we have to understand. i just -- there's something to think about and how we start asking. i think the focus here was to define gaps. i agree with that but maybe a gap is to understand when

something could go to translation. and where is that break point. so maybe something to think >> this was (inaudible) from bcu. there's a gap that was called out through email and this is something i'm going to read that

came through from thomas schneider senior investigator at nih. i have been watching the human excellent biology emerging themes at the #st century talks today. i have been analyzing a lot of data but most case it is, otu

counts are unavailable hard to find or are only partially published. incomplete data for example top 20 species is useless for careful data availability of sequence data is nice but analyzing it adds another layer of computation so critical need is to have full

microbiome otu counts broad data processed by scientists, simple files should be provided so all dta are accessible and permanently available and there was a question that came in from anna edwin also related to data sets and she's a jcbi. here she said when working with

the assembly of sequencing libraries from hmp one realizes quickly there are shallow libraries sometimes only 3 million and she called out a need to sequence more in depth to understand the human moibiota and asks can we set a common standard for sequence step of

community shotgun libraries representing human microbiome. >> i'll answer the last i think a lot of us have been saying the the same thing. you need to define your resolution define your experiment based on biology you want to answer.

i would strongly resist any kind of community enforced standard or sequencing death. if i'm sequencing i have shouldn't be held to the same standard as remote hunt irgatherrer microdata with ten times accuracy. it's important to define up

front why we chose what we did. make the data available that completely agree, so others analyze as needed and do the raw data level rather than process data level people can reanalyze the data but predefining sequencing depth is dangerous. >> just the simple point of

making o thetus, -- otus there's different way it is do it that give different answers and you can pick different cut values for piling up your otus. so there would be a danger if you're pooling for much several different studies that they were

made in quite different ways so i'm afraid annoying business of going back to sequence data making otu de novo is probably the best solution. >> i would like to add would be good if journals reinforce typical policies of depositing the -- making sure the data does

get deposited with associated metadata required to do analyses, that is generally highly inconsistent in work we're trying to do at present and requires that months of effort to go to the authors and request data needed to analyze. brandon acre from mutual aids.

in my personal research i have been embracing the idea that there is variability in the microbiome and that's actually the cool thing about the microbiome and explain reasons why our experiments and clinicals don't work sometimes. in fact, i was in a conference

for the ag microbiome a few months ago, one thing that they were doing is to get to the field faster, the companies going to real life situations, situation will be humans in our case, we're actually getting to learn and understand and capture that varyn 't and make products

and get to market. and solve problems going down that path. i take it with excitement and when we start to do clinicals for nutritional studies it's right into humans and the cool thing here is it's food so we're not feeding things dangerous

things to people. and realize we are all different, we have different microbiomes. i can give everybody in this room a single polyphenol or phytonutrient. 30% will have a response and 60% won't.

that explanation comes from the i think the future of this space and the ability to personalize nutrition, is to grab that by the horns, and start to understand why it is that i'm different than someone else. and start to formulate things or drive towards understandings in

medicine to tease apart the differences so when we offer solutions to a population, it's not the fallacies we have with drugs. drugs have three cohorts, we have drugs that work for 30% of the pop language, drug that do nothing and drugs that hurt

people in a negative fashion. that was my general comment. i think that instead of trying to wipe out the variation and normalize everything down, i think the cool thing here is to try to understand it at great depth, get to humans, i like justin's commentary let's figure

how the foods work in populations and we can solve some of these nutritional issue us both from the western side of the thing where we overeat and consume lots of not food foods to nutritional gaps we see in other parts of the world. >> with that i think -- comment.

sorry. go i a head. >> sorry. i just wanted to agree with that comment. variability can be viewed as a negative thing as noise but it can be very informative so i showed you that principle

analysis how people's microbiota changes in response to diet within 24 hours, it seems to be stochastic but in fact that's actually quite informative. and an example of that would be the work from the wiseman institute making predictions about post friendial glycemic

response and the reason they pick it up is there's ooh a high level intersubject variability and glycemic responses. and i talked to them about we actually observed many years before about the intersubject variability in response to diet. that was one of the features

that was very meaningful and coming up with that postprandial glycemic response indicator was exactly as you say, the >> you might think nih has plenty of money we have a limited budget for this workshop and we become pumpkins at six o'clock.

otherwise -- we get charged more for this room. thank you, everybody.