Scientist Stories: Lita Proctor, The Human Microbiome Project

Next, we have a report on the Human Microbiome Project, and that will be led by Lita Proctor. The NIH Common Fund Human Microbiome Project was established with the mission of generating research resources enabling comprehensive characterization of the human microbiota and analysis of their role in human health and disease. The information generated by HMP is made available worldwide for use by investigators and others in efforts to understand and improve human health. This is a 10-year project that is now completed. It's been managed by NHGRI, and Lita has been basically the program person from almost the beginning. So, she's here today to tell us about the accomplishments of the program. Leah?

[Speaker:LEAH_LIVESAY] Thank you, Betty. Good afternoon. As Betty mentioned, this is a 10-year program. It's sunsetting or has sunsetted. And many in council may not realize this, but the genesis of the Human Microbiome Project was really born out of this institute back when Dr. Collins was our director. So, when I had my first opportunity available at a council meeting, I asked to give this overview presentation to this council. I thought it was quite quite appropriate. I do want to cover sort of three areas. I'm sure many of you know what the human microbiome is, but for those of you who don't, I want to give just a brief overview so we can be talking about the same thing. I want to cover the Human Microbiome Project, its goals, its outcomes, and what was accomplished over these 10 years, and then close with talking about recent advances in human microbiome research. Even in this very young field. So, the human microbiome is the thousands of microbial species that possess millions of genes that live with us. They live in every epithelial surface in and on our— in and outside of our bodies. When you think of the gene complement of these microbes, that's— we generally use the word microbiome to include the gene complement of these microbes, but most of them are not culturable. So, in fact, for many, many years, we didn't really know much about the diversity, richness, or the vast array of these microbes. The numbers there next to the little cartoon is to give you a sense of— these are just bacterial numbers— get you a sense of the magnitude, the numbers of microbes that live on different body regions. Now, we acquire our initial inoculum of microbes at birth. This is a cluster analysis of microbiome makeup. Let's see, where's the pointer? Is that the pointer? Where's the pointer on this thing? I can't figure out where the pointer is on this thing. This is a cluster analysis of microbiome makeup in mother and baby born either by cesarean, which is in baby blue, or vaginally, which is in pink. And the bright red is the makeup of the mother's vaginal microbiome, and the bright blue is the makeup of the mother's skin microbiome. And then just for a data point, the oral microbiome of the mother. And the point I want to show by showing this figure is that babies are microbe magnets. So, when they're born vaginally, they acquire the mother's microbiome as an initial inoculum. However, if they're born cesarean, which is a very common birth mode these days, they still pick up microbes, but they tend to pick up microbes from skin or other sources in the environment. Then what happens is if a child is breastfed, breast milk turns out to be microbe food, not baby food. And so, the oligosaccharides in breast milk help the microbiome— is that the pointer? There it is. The microbiome mature over about 2 to 3 years of life. Of course, this is going on along the same time that the immune system is developing. These microbes encode a vast array of metabolic pathways, and it's thought that the microbiome either augments or extends the capabilities that are encoded in the human genome. So, I decided to use this graphic from a Scientific American article in which they very clearly show that here's the total number of human— genes in the human genome, and here are all the genes, unique genes, not duplicative genes, unique genes in just the gut microbiome, further driving home the point that the microbiome probably augments or extends capabilities in the human genome. So, there were a lot of different impetus for the Human Microbiome Project, but I'll just call out one, and this comes from the epidemiological literature in that epidemiologists noticed that that through the use of vaccines and antibiotics and so on, there was quite a successful drop in many kinds of infectious disease. But at the same time, they started to see this emergence of many different kinds of allergic or autoimmune diseases, almost like a mirror image of what was happening with the infectious disease. And so, the question to the community was, well, what's happening? I mean, is this the human genome changing at this kind of rate, or is there some other factor? And so, what was proposed is that there were changes in the microbiome associated with modern medical practices or just modern practices in society that might be somehow stimulating the emergence of these autoimmune diseases. The other really important impetus for the Human Microbiome Project, of course, was this program emerged out of the Human Genome Project. There are many aspects, many features of the Human Microbiome Project that really you can point back to the Human Genome Project as as its basis, but maybe most importantly is the fact that all these sequencing technologies were developed that now allowed us to study the human microbiome. Remember I mentioned that many microbes weren't culturable, and even if you cultured them, you've got them separated, not working in a community. If you can do sequence analysis, you can look at them at the community level. So, the Common Fund, which is an office under the Office of the Director, kickstarted this 10-year Human Microbiome Project at a level of $215 million solely as a community resource program, meaning that everything that was developed within this 10-year program was meant to be a resource, a toolkit for the larger research community to undertake microbiome research. So, the program goals were to develop these research resources, many, many kinds., and most importantly, to rapidly release these resources in every way possible. The other thing that's not often called out, but I think it's important to point out, is that the Human Microbiome Project also really birthed a research consortium across these 35 institutions around the country. And so, that really formed the foundation or the seed community, if you will, for this emerging research area. And then fourthly, I think a thing that hasn't also been called out very often is just the approach to this kind of research that the HMP and the resources that it developed really helped form a new paradigm for this research area and how to undertake this research. So, as is often the case with many Common Fund activities, they build these out as phases. So, there are two phases to the HMP. The first phase was a 6-year program, and the second phase was 4 The first phase was also a more richly funded component of this program. It was very basic. The question was, who's there? We want to do the survey of these microbes living in our bodies. And that survey work was done in a couple of different ways, only through human cohort studies. No animal models were done at all in any— oh, no animal model work was done at all within the HMP. There's a so-called famous healthy cohort study. We want to have a reference database in healthy people of what's a typical or characteristic microbiome in healthy people. So there was a very exhaustive healthy cohort study mounted in which 300 adult men and women were clinically examined across each of these body regions and verified to be free of disease before they were sampled. So clearly far more than you get in your annual exam. And these 5 major body regions were sampled simultaneously, and many of these cohort subjects were visited for up to 3 times. They also had other criteria. They couldn't take antibiotics. They couldn't take probiotics. They couldn't take immunomodulators. Really want— we really wanted an undisturbed microbiome from healthy people. On the other hand, we were quite interested in what was the role of the microbiome in disease. So, there was a suite of demonstration projects in order to demonstrate whether there was a characteristic microbiome associated with disease. There were about a dozen, a little over a dozen of these demonstration projects funded across basically three parts of the body: skin, GI, oral, and urogenital. All these projects utilized sequence-based analysis to characterize the microbiome in these cohorts. So, here's a kind of a classic graphic that you see whenever you hear about the microbiome. This is metagenomic analysis of microbiome community composition, again, another cluster analysis. And here are those 5 major body regions that I referred to. The main take-home point here on this figure is that microbial community composition in each body region is distinct, okay, GI different from oral and so on. But I want to zero in on the GI tract because there's another point that kind of caught us by surprise and is really forcing us to rethink what the microbiome is. And that is, if you look at the GI, This is a figure from another paper, but I used this because this particular investigator utilized the same color scheme. So, the dots in blue on this cluster analysis are healthy subjects, and then the dots in red and orange are overlaid gut microbiome composition of two forms of IBD patients, ulcerative colitis and Crohn's disease. And unfortunately, because we set out with the assumption that we could actually use community composition as a biomarker for phenotype. Our conclusion, though, unfortunately, was that large-scale community composition alone couldn't really be used to differentiate host phenotypes. But that really formed the basis of Phase 2 of this program, in that it's a smaller activity, $35 million rather than $180 million, but the goal here was to ask, "What are they doing?" We can't just say who's present; we have to ask the question, "What are they doing?" They undertook to fund a series of cohort studies in which these projects analyzed the multi-omic functional properties, not only of the microbiome, but of the host, and tracked them over time. The prior project were case control studies, and these are longitudinal studies. And then the quest was to interrogate these integrated datasets, hence the name of this phase, the Integrated HMP. Because it's quite pricey to undertake the analysis of all these multi-omic functional properties, we funded 3 projects. All 3 serve as models, so the data coming out of these 3 clinical studies would be of interest to much more than the communities that study these areas. So one of them was on the dynamics of pregnancy and preterm birth, so included both the vaginal and gut microbiome as the microbiome microbiome component of the project. The dynamics of IBD primarily focused on the gut microbiome, and the dynamics of type 2 diabetes, which included the dynamics of the microbiome in the nares, in the nasal area, as well as in the gut. And they collected 7 or 8 different data types, deposited all their data in these 7 or 8 different repositories. Let me just give you a picture of how much more complicated these 3 clinical studies were. This is just a snapshot from the IBD study. You can see these are all the different data types they collected from the host, all the different data types they collected from the microbiome, and I want to point out— I think I have it here, yeah— in Phase 1, the cohort studies only collected metagenome data and 16S data. The data types across the Phase 2 projects were far more complex. And remember what I told you, that we really could not use large-scale community composition to differentiate these host phenotypes. So, the analysis was far more complex. It's trying to take all these different omic data types and conduct a network analysis. Now, I don't expect you to make any sense of this network analysis web, but you have to trust me when the investigators have been able to show that certain properties or certain microbes are gained or lost in the disease cohorts compared to healthy controls. And so, the large take-home from Phase 2— and they're still sort of in the middle of the analysis— but the large take-home from Phase 2 is they found the loss or gain of specific microbes or specific metabolic pathways was characteristic of disease patients. So, far more kind of complex analysis would be needed in order to come up with a post-phenotype here. Now, I mentioned there were 7 or 8 different data types, 7 or 8 different public repositories. You know, since they're not federated or linked in any kind of way, if you as an investigator were interested in these studies, you would not be able to reassemble the study design of these projects. So, a really important component of this program was our HMP Data Analysis and Coordination Center, which houses all of the primary and derived datasets, the tools, the pipelines, everything that was produced as a resource in both phases of the program. They hold it— sorry about that— they hold it for Phase 1 as well as Phase 2, and just this year there are 4 major papers under review at Nature from Phase 2, as well as a series of 35 companion manuscripts, and the HMP DAC now houses all the multi-omic datasets, associated tools and pipelines, and all the walkthroughs, basically hand-holding your way through analysis of this kind of complex data type. So, to review then, here are all the resources that were developed in both phases of the program. Okay? Everything from sequence and other omic reference datasets from microbiome and host, computational and statistical tools and pipelines, for microbiome multi-omic data analyses, analytical protocols for doing microbiome sample analysis, all the critical clinical protocols for collection and storage of samples from these 5 major body regions, IRB protocols for actually conducting clinical studies of the microbiome, and a whole suite of ELSI issues that we evaluated that were related to the microbiome. And a couple of ways that I can point out that these have actually been a real resource to the larger community is So far, in HMP, there have been about 650 publications that have come out of this program. In addition, every year I've gone and tried to do a head count of how many other investigators are not funded by the HMP have actually used HMP data. And we've estimated between 25 and 50 papers per year come out— are published that utilize HMP reference data or tools, but are not funded by the HMP. So that's one example of how The HMP has acted as a catalyst, but I want to show you something else. I think this figure speaks volumes. What I want to show you here is the expansion of human microbiome research at this one institute over the— over 10 years. The x-axis is fiscal year, the y-axis total investment in millions of dollars, and I'm only talking about extramural. I'm an extramural program director, so the only data I have right now that I'm presenting is in the extramural arm of the NIH. The values in red are the investment per year in the HMP. The values in blue are the investment in every other institute's microbiome research activity. So, you can see right away that after about halfway through Phase 1 or so, you start to see this explosion of investment in human microbiome research across the NIH. Number 2, I didn't plot the data here, but this occurred during a time when there was either flat or even, in some cases, decreased funding in the NIH appropriations. And we have data to demonstrate that in the blue area, 89% of these grants were from investigator-initiated activities, not at all in response to any RFAs, any targeted RFAs. So, to give you another kind of perspective on this, at the beginning of HMP, there were about 5 or 6 institutes funding this area. Many of them were involved in the HMP., and maybe 50 investigators in this very, very small community. At the transition point between Phase I and Phase II, that total increased to about 15 institutes funding in this area— oops, what have I just done? There we go. And almost 300 investigators in this growing community. And towards the end of HMP, now we have 20-plus institutes institutes funding in this area with over 700 investigators. I head up a committee called the Trans-NIH Microbiome Working Group, and we've done sort of a deep dive into some of the trends that are happening in this field. So, I'm going to show you another data point for 5 years, and that is the relationship between studies of the microbiome and disease. We estimated over 5 years, fiscal years '12 through that the NIH has invested a total of $791 million in this area. About a third of that is on non-disease questions, host-microbiome interactions, and so on. But two-thirds is on disease, about $500 million. And we used the ICD, the International Classification of Diseases catalog, as our way of categorizing diseases, and found that about 100-plus classes of disease were understudied with relation to the microbiome over this 5-year period. And just to give you a sense, this is just a little kind of a sampling of different disease types. There are many, many areas of study, of course, in the GI tract, many diseases in the GI tract, but also the microbiome and cardiovascular disease, many kinds of cancers, and the microbiome in many kinds of cancers, many liver diseases, many vaginal conditions. Many skin diseases, many lung disorders or diseases, and neurological and mental health issues and the microbiome. So, there's a vast diversity of areas being studied with regard to the microbiome. So, let me quickly go over some recent advances that have happened in this field, in this young field. First of all, the first thing that's kind of emerged is apparent evidence of microbiome-based markers related to disease. One you've already heard about, I know, but it was talked about recently some more, and these are data coming from a 3-day trans-NIH microbiome workshop that we held last summer, so I'm pulling some highlights from that workshop. This notion that there is some type of transferable phenotype from the microbiome to host, and that if you take the feces from an obese mouse and you and, and gavage it into a lean mouse, it'll become obese. So there's some type of microbiome phenotype that's transferable. We don't know if it's a microbe, metabolic property, whatever it is, there's some type of a microbiome phenotype. Apparently, of the many metabolic byproducts that microbes make, particularly in the gut, many of them have epigenetic effects on the host. And so there's some very intriguing results showing that there may be a relationship between the epigenetic effects of the microbiome and the emergence of colorectal cancer. And a third example is this relationship apparently between gut microbiome bacterial metabolism of different types of diet, in this case meat, and the byproducts from that digestion and the close association with cardiovascular disease. So, different types of either bacterial or microbial metabolism biomarkers.. In addition, there have been a number, even in this young field, of recent microbiome-based therapeutic interventions that have been developed. Everybody's heard of, I'm sure, fecal microbiota transplantation, or FMT, but there are a number of FDA— there are a number of clinical trials underway right now for microbiome-based interventions that are being investigated, that are being monitored by FDA. They include more sophisticated consortia of microbes, or in this case, maybe some microbial metabolic byproducts that could act as an intervention to disease. In addition, there's been this whole new growth area of utilizing the microbiome itself as a source of new pharmaceuticals. So, instead of going to rainforests or going to the coral reef, you just go to a gut sample and try to look for new pharmaceuticals, and we have this great example from last year where a new innovator awardee named Michael Fischbach from UC San Francisco gave a wonderful NIH Council of Councils talk on how he mined the HMP metagenomic data to discover and develop new antimicrobials. But even though there are a lot of advances in this area, it's a young field, so there's certainly a number of gaps and challenges. And I'm also calling these out from this NIH workshop we held last year. First of all, model systems. Although a lot of NIH research is done using mice as animal models, they turn out to be in many cases, not all cases, in many cases not the best animal system because mice are coprophagic and we're not. And their biology is built around being coprophagic. So there is this need to explore new and different animal models. There's still this open question about cause and effect. We don't really know with the biomarker work or anything whether, in fact, are the microbes eliciting disease in the host or are they exacerbating an already developing condition? And that's still a quite open question. I didn't develop it very much in my talk here, but it's tied to this notion that microbiome community composition alone will not get us to host phenotype. And that's because there's a lot of work that's starting to show that maybe we need to think of the microbiome as like an organ system where all the members interact and they produce an emergent product, an emergent property, like an organ system. So, there's a lot of work trying to explore how to study the microbiome as an emergent property and not as just membership. Much of the early microbiome interventions has been around trying to treat for disease, but because the microbiome is a mutable, changeable property, a lot of it is driven by diet. There is some work now starting to come out around the— how do we develop microbiome-based interventions for supporting sustaining health? But it's still not a very well-developed area yet. And finally, this is the HMP logo. It's slightly ironic because it encompasses a microbial genome, but in fact, In this field, there's very, very little known with regard to the role of host genetics on the microbiome. So, to conclude, the human microbiome is made up of thousands of microbial species and millions— that contribute millions of microbial genes. And the microbiome is a metabolically diverse, active, and mutable, possibly microbial organ. The NIH Human Microbiome 10-Year Human Microbiome Project invested $215 million into rapidly deployed research resources, and also supported this 35-institute, 50-investigator research consortium, which formed the seed of this research community. And recent advances in this area include this really explosion of research support across the NIH, with now over $1 billion over 10 years, including the HMP funds. In this area, extensive research on host microbiome biology, and extensive study on the role of microbiome in 100-plus disease classes. And microbiome-based intervention and drug development is in its infancy but really exploding in growth. But I would conclude by saying that we still have a main challenge, which is really trying to understand the microbiome is far more than the sum of its microbial And my last slide is an acknowledgment slide. This has been a huge program involving many institutes, many investigators, so I really want to acknowledge all of my colleagues at NHGRI, the Common Fund, the HMP science co-chairs, which include Eric Green, Tony Fauci, Griff Rodgers, and Martha Sommermann. The 80-plus-member, 21-institute trans-NIH microbiome working group has been great as a sounding board for many things we're trying to do. Trying to develop in the program, as well as being program— some of them are program directors of some of the awards in the HMP. We had an external HMP advisory committee. And of course, I really have to acknowledge the creativity and enthusiasm of the HMP research consortium. Thank you.

Are there any questions?

Are there any questions, Felita? Thanks for the presentation. I noted that other institutes picked up 4 times the funding. That's right. Of the HMP. Who were the big players in that, and is that dynamic, or are some of those increasing while others are falling off?

Actually, what's happening is that sort of the 6 largest institutes are funding the most in this area. It's sort of proportional to their total budget size. But what's happening over time instead is there are many, many more institutes getting involved. So, when I first formed the TMWG, there were, I think it was 12 institutes. That was in 2012. Now we're at 21. So, there's other institutes and centers and offices joining in. So, I see that expanding across NIH. The TMWG was our effort to try to coordinate that activity, but that turns out to be, you know, easier said than done in some cases. I think we're seeing more institutes getting involved in this area, and we're seeing more disease classes being studied.

Anybody else? I would make the comment that there's often— Frequently there's discussion about the Common Fund in general at NI around the Institute Director's table because, you know, it's still, you know, evolving, and the Common Fund is, because of its forced— the restrictions that, you know, it turns over, it also gives an opportunity to try out different things, and should we have even simple things like larger numbers of smaller projects or smaller number of larger projects. These things constantly get described. So, this is always, I think, and it's always going to be sort of a playground of sorts. And while the NIH director makes the final decisions, the institute directors collectively are quite involved in vetting proposals, vetting ideas, you know, making decisions about starting new programs, renewing programs, and so forth. I would comment that the Human Microbiome Project is largely regarded as sort of a prototype. And maybe it's part of it, and there's a variety of reasons for that. First of all, it probably was never going to happen unless Common Fund jumped in, because it was not clear that any one institute would jump in at a very early stage, develop the resources, and technology, and approaches, and analytics that Lita just reviewed. But what it also did was catalyze investments, and that's what we were just talking about. That's not true of all Common Fund projects, but the fact that you see this huge— I think the graph, both in terms of total dollars, total numbers of institutes, jumping in and doing it was— is really a credit to the Microbiome Project. So, and again, you know, there's several others that are sort of put out there as these are prototypic, but I can guarantee you the Microbiome Project gets mentioned frequently as sort of a prototypic. And the other thing about it was that it had a logical first phase, it had a logical second phase, and while nobody likes to see programs end, it does make a lot of sense to end it as a Common Fund project because you see this uptake in both large number of institutes and amounts of dollars across NIH, oftentimes in a very disease-specific way.

So. Thank you. Thank you, Lita. Next, we will have a report from Sharon. Oh, okay.

Yeah, we can. I think now would be a good time for a short break. We've sort of been going a couple hours since lunch, and then we have a Sharon presentation, and then Jim Ostell. I saw he just got here, so then Jim Ostell, and then we just have a few little business items. So why don't we take a break and try to reconvene right at 3 o'clock? A 15-minute break, reconvene at 3 o'clock.