The Human Microbiome Has Parallels With Forest Ecology

By Deborah Borfitz

July 15, 2019 | Using multi-omic analysis of different measurement types over time, the recently concluded Human Microbiome Project (HMP) of the National Institutes of Health (NIH) successfully captured host and microbial properties linked to disease in a trio of longitudinal studies specific to pregnant women and preterm birth, inflammatory bowel diseases (IBD) and type 2 diabetes. Across studies, health-associated microbiome interactions manifested in highly individualized ways with effects at both the local and systemic level.

To microbiome researchers, comparisons with forest ecology can be irresistible since it's a lot easier to see a tree than a bacterium. Most microbial communities in the body are inhabited by hundreds of different types of microbes, like a forest with many species of birds but just a handful of dominant ones who might have arrived first, says George Weinstock, PhD, professor at The Jackson Laboratory in Farmington, Connecticut, and one of the diabetes project leads. The nasal microbiome might similarly have early colonizers that dominate until overthrown by other organisms during disruptive events such as the common cold, he muses.

"I often think a microbial ecosystem is easier to think about in terms of a forest ecosystem," says Curtis Huttenhower, professor of computational biology and bioinformatics at the T.H. Harvard Chan School of Public Health, associate member at the Broad Institute and senior author of the IBD study. If 100 different forests burned down, "they'd probably recover over time but there'd be a lot of variation depending on what the weather pattern happened to be at that location, what other environments were nearby, and what seeds or animals arrived to recolonize the area." As with people after a big shock of inflammation, "usually everything recovers but sometimes things come back differently, or they don't come back right at all."

The Backstory

The first phase of the program (HMP1) launched in 2007 when the human microbiome was just emerging as a scientific field of study with relevance across all institutes within the NIH. The NIH Common Fund therefore made more than $150 million available to "benchmark" the microbiomes found in different parts of the body, explains Weinstock.

A major portion of that study did microbial profiling on 300 healthy adults at 18 specific body sites (15 for men). Over 11,000 samples were collected in total. Demonstration projects also looked at 15 diseases that in each case determined whether microbiomes looked different in individuals who had a condition of interest compared to healthy subjects, says Weinstock. More than half of the NIH's 27 institutes have since funded other microbiome research to further study specific diseases or organs. To date, nearly a billion dollars have been invested in these projects.

Connecting the Dots

Microbial activity doesn't happen in isolation but in parallel with various host immune and inflammatory responses, says Weinstock. The second phase of the HMP (HMP2) therefore endeavored to uncover those inter-relationships over time. The unlikely pairing of type 2 diabetes with IBD and pre-term birth was to quickly get at answers inside of a brief three-year window, allowing NIH funds to rapidly share new resources with the community, he explains. Conditions were favorable in these therapeutic areas for recruiting patients and collecting massive amounts of data.

"The zoo of microorganisms in the body each have their own unique biochemistry, some of which are highly inflammatory and others that do nothing antagonistic, and the goal of HMP2 was to understand as much as possible about their role in the time allotted," Weinstock says. Researchers took a first pass at identifying associations between host and microbiome and the impact of routine events such as gaining or losing weight and catching a cold.

Many different measurements were taken of microbes, human cellular activity and chemicals to understand host/microbiome interactions "with as many different lenses and views as possible… to learn how things are going wrong in the crosstalk between ecosystems," says Huttenhower. "HMP1 was like building a parts list, cataloguing what one typically finds in the microbiome, and with HMP2 we assembled some of those parts into a machine and identified the important interfaces between bugs, immune molecules and biochemistry that seem to matter during disease."

The decade-long undertaking generated a lot of big data the computational world now must figure out how to analyze, notes Weinstock. Rather than taking two or three samples per subject, as in HMP1, participants in the HMP2 longitudinal time series studies had samples taken "many times over many years to gauge the stability of the microbiome."

All the raw data and several processed data products are now publicly available, along with the metadata—information about the samples and participants, including their clinical activity, medications, diet and demographics, says Huttenhower. There should be no shortage of interest in the output, now the microbiome is widely appreciated as a tool for therapeutic development.

The testing of therapies based on microbiome science has mushroomed over last two years, Huttenhower notes. A recent query of indicates 765 such studies are underway and more than 400 have been completed. This includes a flurry of early-stage biotech activity around at least three different kinds of microbiome therapeutics—human-derived and synthetic fecal transplantation (aka "really smart probiotics"), microbiome-derived therapies and, most speculatively, synthetically engineered microbes or microbial communities. "Larger pharma companies are also interested in microbiome-related therapies, which I think is fantastic because it's difficult for academic researchers to bridge the last gap between potential therapies and the market."

Disruptions in the Gut

Like many other chronic inflammatory diseases, the prevalence of IBD is highest in developed Westernized countries and the incidence rate is rising, says Huttenhower. It has been known for more than 50 years that IBD has a complex association with the microbiome. The problem is that many of the plausible modifiers of the immune microbial balance—including diet, early-life exposures, working environment, stress level and aging management—tend to change simultaneously at the population scale.

One interesting finding to emerge from the recently completed IBD study is that flares of disease activity are associated with a previously uncharacterized group of microbes, suggesting a means to better predict and perhaps interfere with disease activity, says Huttenhower. A group of approximately a half dozen bugs related to bacteria in the genus Subdoligranulum are carried by almost everyone, and they are among the organisms depleted during inflammation.

Specifically, as discovered during HMP2, they're also the ones more often lost during periods of IBD-related inflammation, he continues. One possibility is that these organisms are involved in maintaining a "healthy" gut.

The IBD study also found an association between circulating antibody levels and disruption in the gut microbiome, Huttenhower says. Known clinical biomarkers of disease activity that help clinicians discriminate between Crohn’s disease and ulcerative colitis were elevated specifically during periods when the microbial ecosystem was also out of whack. "This type of immune system activation is thus one way in which information about problems in the gut is transmitted out to the rest of the body."

Different kinds of instability could also be seen in the microbiome of IBD patients compared to healthier control subjects. The gut microbiome of people with IBD is predictably fragile, while the healthy microbiome is "personalized and relatively stable," says Huttenhower. "The structure of your gut will tend to look a lot like itself a couple months from now, and even several years from now. But if you look at the trend for individual IBD patients, they will often go off into these excursions during disease activity… something is clearly not normal about the structure of their gut microbiome."

As with trees in a forest ecosystem disrupted by fire, he says, some individuals will return to their microbial status before the flareup while others will find themselves with an entirely new community of microorganisms in their gut.

More closely related to previous work was the catalog of chemical compounds found to become more abundant in the gut, or less so, during inflammation, Huttenhower says. This lays the groundwork for researchers to follow up on molecules that might work as therapies to either prevent flareups or block inflammation.

Case for Precision Medicine

Two striking findings of HMP1 made it apparent that a study of the vaginal microbiome in pregnancy was worth doing, says Jennifer Fettweis, PhD, director of the Research Alliance for Microbiome Science (RAMS) Registry of Virginia Commonwealth University (VCU) and lead on the pregnant women and preterm birth HMP2 study. The first was that differences in the human microbiome were seen according to race and ethnicity. The second was variances existed between women who were and were not pregnant.


Using new molecular approaches, VCU researchers are hopeful they can better understand bacterial communities associated with different subtypes of preterm birth, says Fettweis. Perhaps then antibiotics could be used in a more targeted fashion to improve treatment outcomes for women with bacterial vaginosis, which has long been known to increase the chances of preterm delivery.

"Preterm birth has been a stubborn problem and rates have not budged despite much effort," says Fettweis. About one in 10 women in the US delivers preterm in the US, but among African American women the odds are closer to one in six. Differences in who bears the burden can be seen globally, she adds, where about 15 million babies arrive preterm. "The numbers are rising and, from a global health perspective, it's a leading cause of death for children under 5."

For the HMP2 study, more than 1,500 pregnant women were followed longitudinally, she says. The primary trial sites were associated with VCU and the RAMS Registry, plus another three via collaboration with the Washington State Global Alliance to Prevent Prematurity and Stillbirth. VCU also teamed up with the Medical College of South Carolina, which had a vitamin D trial underway with an African American cohort.

Improved biomarker predictive tests are needed, especially for asymptomatic women, and findings of the HMP2 study on pregnant women included microbial signatures early in pregnancy that could potentially be developed into a diagnostic tool, Fettweis says. "It has been known for some time both that organisms in the vagina can affect the upper reproductive tract, and that many other factors are at play [e.g., fetal and maternal genetics, post-immune responses and preeclampsia] …. The question is which bacterial species might be causative. We're not there yet, but in the future, we might be able to get answers from a simple vaginal swab sample."

Women of African ancestry who represented 80% of those who delivered spontaneously preterm are in dire need of a reliable screen, says Fettweis. Other study findings similarly highlighted the need for a precision medicine approach. For example, the presence of beneficial bacteria offering protection from preterm birth varied across groups.

Black women with lower levels of vitamin D during pregnancy had more Megasphaera, a type of bacteria linked to bacterial vaginosis and preterm birth, while white women with higher levels of vitamin D had more of the protective Lactobacillus bacteria, says Fettweis. The microbial ecosystem of many women, particularly if they had African ancestry, also had different trajectories between the start and end of pregnancy.

The challenge for researchers is to better risk-stratify pregnant women and get each person to their healthiest possible state, she continues. Still unknown are all the relevant factors, and which are modifiable. "We found annual household income correlated with race and ethnicity, but we have yet to untangle host and genetic factors in terms of what can be done."

Among less surprising discoveries of the HMP2 study was that the microbial profile associated with preterm birth correlated to a greater presence of proinflammatory proteins called cytokines. The next steps there include figuring out how to make correlations with molecular measurements that use different scales and how cytokines potentially affect preterm birth.

An important observation during attempts to do meta-analysis was the need for harmonization across studies, says Fettweis, including the definition of preterm birth. Her group included only spontaneous preterm births that were not medically indicated and occurred before 37 weeks of gestational age. In other studies, a significant proportion of the preterm births were medically indicated. Additionally, one of the studies involving women predominantly of African American ancestry included only people who had been treated with progesterone while another excluded them completely.

Communication Mystery

Like IBD, type 2 diabetes is a multi-factorial disease associated with the Western lifestyle and its prevalence parallels rising rates of obesity, cancer and cardiovascular disease, says Weinstock. The American Diabetes Association reports that it afflicts over 29 million Americans. Diabetes being a metabolic disease, its potential relationship to the microbiome in the gut is more obvious. But chemical reactions occurring elsewhere in the body during metabolism, in concert with other physiological functions, may prove just as important, he says.

The study, which The Jackson Laboratory conducted with Stanford University, gathered multi-faceted clinical data from more than 100 people—many of them at risk for diabetes—over the course of four years. "It was a very limited study," notes Weinstock. "I don't think we understand the cure to diabetes, or what causes it or how to predict it, but we do now have datasets and made some connections between various manifestations of diabetes in the body and what happens to the microbiome."

Of interest to Weinstock is what happens in the nasal microbiome when people catch a cold. He pictures a settlement of bacteria, some more numerically dominant than others, being thrown into chaos by the virus and all the subsequent mucus secretions, nose blowing and Afrin sniffing. The bacterial community eventually reforms, but its structure changes significantly during the disruptive event—as was consistently demonstrated in a few dozen people who got colds during the HMP2 study.

This may provide hints about how the microbiome gets built, says Weinstock. The big mystery is how viruses in the upper respiratory tract communicate with bacteria in the gut—a parallel in ecology being how a lake and a forest could interact and impact one another. Understanding that might explain why people experiencing a cold sometimes complain of a stomachache. "That human virus, even if it could get to your gut, doesn't necessarily infect cells there or affect anything [directly]," he says. Rather, different parts of the body "seem to be always talking to each other" as part of a complex network.


The research also highlighted differences between insulin-sensitive and insulin-resistant groups both at baseline and during infection, says Weinstock. The resistant group had an impaired and delayed immune response to viral infection, as well as a milder response in their gut microbiomes, he reports. Changes in immune signaling molecules and gut microbial diversity, and weight gain, were noted in one individual newly diagnosed with type 2 diabetes during the study.

Next Steps

For IBD researchers, the big task ahead is to learn which of the many potential microbiome-related therapeutic routes are going to work out for the broadest set of patients struggling on current medicines, says Huttenhower. In addition to microbial live cell therapies in clinical development, microbial gene products or microbially-derived small molecules may also prove to be anti-inflammatory. He predicts that in another few years many different options will be added to the mix for patients suffering from this "heterogeneous, highly personalized disease," including more targeted use of currently marketed therapies based on their gut microbiome.

Huttenhower says he is working on NIH-funded projects looking at individual bioactive products in the gut microbiome "in IBD and otherwise." Several foundations are also sponsoring studies zooming in on bioactive compounds specific to IBD.

Fettweis says she is "very optimistic" that biomarkers will be developed for preterm birth, although her study of the vaginal microbiome is now expanding to a wider range of pregnancy outcomes. She recently teamed up with a pediatric endocrinologist to look at how a mother’s vaginal and gut microbiome, and obesity risk, may get transmitted to her offspring.

Other areas of interest are harmonization of microbiome research from a women's health perspective, and "broadly applicable" method development to pull together study findings over time. "Our study alone generated data on more than 12,000 samples that are linked to rich data and now accessible to other researchers," Fettweis says.

The goal of many follow-up diabetes studies is to link the depletion or overgrowth of organisms that correlate with disease to come up with new diagnostic tools and treatment therapies, says Weinstock. The skin microbiome in the extremities of people with diabetes is different than in healthy individuals, for example, which may open discovery opportunities for topical agents to prevent amputations. The gut microbiome of people with type 2 diabetes have recently been found to measurably benefit from Metformin, although the search continues for the mechanism of action.

"We have another wave of analysis we want to do with [the HMP2 study] data to, for example, determine the stability of microbiomes under ordinary conditions," says Weinstock. "Does the microbial community in the gut pretty much stay the same or does it flip from time to time, like when we feel less energetic or out of sorts but not really sick? We'd like to borrow some of the tools of ecologists who have been looking at these kinds of questions in their various macrohabitats for 50 to 60 years and are doing a lot of really cool analysis."