Over the past decade, more than US$1.7 billion has been spent on human microbiome research. Major projects are under way in the United States, the European Union, China, Canada, Ireland, South Korea and Japan.

This investment has confirmed the importance of the microbiome to human health and development. It is now known, for instance, that newborns receive essential microorganisms from their mothers¹. Moreover, the sugars in breast milk that infants cannot digest nourish babies’ developing microbiomes², which in turn shape their immune systems³.

Now is a good moment for reflection. The biggest investment made (around $1 billion) comes from the United States. Some 20% of this has gone to two phases of the Human Microbiome Project (HMP), which is creating the research resources needed for studying the human microbiome (see ‘Big spend’). A review⁴ of what that decade of investment in human microbiome research has achieved was published in February (see ‘Big wins’). And findings from the second phase of the HMP are published in this week’s Nature^5–8.

In my view, most of the research so far has placed too much emphasis on cataloguing species names. We’ve been characterizing the human microbiome as if it were a relatively fixed property to be mapped and manipulated — one that is separate from the rest of the body. In fact, I think that interventions that could help to treat conditions such as diabetes, cancer and autoimmune diseases will be discovered only if we move beyond species catalogues and begin to understand the complex and mutable ecological and evolutionary relationships that microbes have with each other and with their hosts.

Beyond inventories

The HMP, funded by the US National Institutes of Health (NIH), has clearly catalysed human microbiome research in the United States and globally. The same is true of other projects with similar aims, such as the European Union’s Metagenomics of the Human Intestinal Tract (MetaHIT) consortium (in partnership with China) and other European projects⁹; the Irish Metagenomics of the Elderly programme (ElderMet); the Canadian Microbiome Initiative; and the Japanese Human Metagenome consortium¹⁰, to name a few.

One of the main goals of the HMP, launched in 2007, was to create a toolbox of reference data sets, computational techniques, analytical methods and clinical protocols. This seems to have been a success: around 75% of the 2012–16 NIH grant recipients for microbiome research outside the HMP — working on more than 100 diseases — cited reliance on HMP data and tools in their funding applications⁴.

This progress in microbiome research has excited industry. The current value of human-microbiome-based products and interventions for diagnostic and therapeutic use is estimated to be between $275 million and $400 million worldwide. This is expected to increase to between $750 million and $1.9 billion by 2024.

Yet even with this considerable public and private investment, many fundamental questions about the human microbiome remain.

Researchers don’t yet agree on what constitutes a healthy microbiome or how to define an impaired one. There is still uncertainty about which microbiome properties will be the most informative biomarkers in clinical and epidemiological studies. And little is known about how the microbiomes of different body regions, such as the mouth, gut or skin, interact.

Biogeochemical habitats that are unique to each body region have been discovered through the analysis of metagenomic data. These DNA sequences, derived directly from environmental samples, can be used to characterize the microbial communities present and their metabolic capabilities¹¹. For example, the main metabolic process used by microbes in the mouth is anaerobic respiration, because oxygen is limited. By contrast, in the oxygen-free gut, the dominant process is microbial fermentation (the extraction of energy from carbohydrates in the absence of oxygen). Yet researchers have not investigated the factors that drive these shifts in microbial processes — such as oxygen concentrations, pH levels and nutrient sources.

Furthermore, it is becoming clear that microbes are needed to support human development and maturation, and to activate and maintain stability in the immune system and metabolism. But we do not understand how these fundamental biological phenomena involving human cells and microbes co-evolved. What’s more, some ecological concepts are not yet commonly considered in studies of the human microbiome. These include how communities of microbes operate as a whole; how ‘keystone species’ can pave the way for others by altering local conditions; and the predator–prey relations between different microbes.

Holistic view

Some researchers have studied the microbiome as if it were an organ¹². But even this approach is not entirely satisfactory, because the cardinal property of the microbiome is its mutability — during development, over a lifetime and in response to stressors or disease. This means that it does not demonstrate typical organ-system biology.

For all these reasons, I think that the most effective route to discovering microbiome-based remedies will be to establish which microorganisms — and which assemblages of them — play a major part in dictating local conditions, or in affecting important cellular processes.

Much could be learnt about human–microbe associations if researchers investigated the mechanisms underlying the development of these associations in the well-characterized animal models commonly used in biomedical research, such as mice and rats. Indeed, factoring the microbiome into animal preclinical studies might drastically alter the conclusions¹³.

Evolutionary biologists have argued for decades that human microbiome research would benefit from the evolutionary understanding provided by symbiosis research¹⁴. Certainly, human–microbiome systems share some of the features of highly regulated symbiotic associations. As just one example, a class of molecules produced only by bacteria (short-chain fatty acids) has a central role in host–microbe interactions. These molecules provide an energy source for the cells lining the human gut (most other cells depend on glucose). And they mediate interactions between different gut microbes, and between microbes and human cells.

Two fronts

Developing a new conceptual framework and applying it to the human microbiome will require much more collaboration between investigators working across disparate fields, including evolution, ecology, microbiology, biomedicine and computational biology. It will also demand significant changes in how data and other resources are distributed between scientists, and in how currently disparate areas of microbiome research inter-relate.

Here I address what’s needed in the United States. These changes must happen elsewhere as well.

Data standards. Microbiome researchers have not yet broadly embraced quality-control practices for their data in a way that would make results more reproducible, and that would facilitate the analysis and interpretation of data across multiple studies.

Studies based on characterizing genetic material, proteins or metabolites using high-throughput analyses will remain the norm for the foreseeable future. To produce useful results, however, researchers must adopt better data-sharing practices.

The Genome Standards Consortium, established in 2005, has developed standards and templates for reporting metagenomics data, as well as for environmental measurements and various clinical metadata. These have been adopted by the Data Coordination Center of the HMP, the public repository for everything produced by the project. But this is insufficient on its own. Funding agencies and journals must also promote the use of these standards in reporting microbiome data in databases and publications — much as was done for RNA microarray studies in the early 2000s¹⁵.

Coordination and collaboration. Currently, 21 of 27 NIH institutes provide extramural funds for human microbiome research. Any coordination that does occur is mediated by the trans-NIH Microbiome Working Group — a committee of programme directors established in 2012. More than 40 staff members gather each month to discuss key developments in the field. However, the committee has no budget and no authority to make funding decisions.

In my view, the big investment in human microbiome research should be formally managed. The research community has pushed for this kind of formalized coordination before¹⁶. Indeed, the EU, Canada, Ireland and Japan have arguably done better than the United States when it comes to coordinating human microbiome research; for instance, by mandating partnerships between researchers in academia and government agencies or industry.

Recognizing that many disciplines are needed to study the microbiome, 33 universities, research institutions and medical schools in the United States have now formed microbiome centres. In principle, these could champion data-sharing practices. Researchers at the centres could agree to adopt such practices and advocate for them at meetings. In partnership with journals and funding agencies, this network of centres could identify and promote shared resources, such as biobanks, analytical and computational standards, protocols and public databases.

Encouraging signs

Another government agency, the US National Institute of Standards and Technology, is leading efforts to develop analytical standards for microbiome sequencing. In the next few months, discussions will take place on how to build on the lessons learnt by the US microbiome centres. A research coordination network could emerge.

Outside the United States, the Canadian Microbiome Initiative is developing national core resources for microbiome research, such as public data repositories and analytical centres. The International Human Microbiome Consortium (IHMC) has been raising awareness of the importance of data sharing and standards internationally by holding conferences around the world since 2008. But the IHMC, a 13-country organization formed to coordinate microbiome research, has never had a budget and relies on volunteers, so its powers are limited.

Microbiome researchers should take inspiration from the many examples of other disciplines that are advancing thanks to collaboration. Take my former field of oceanography. Studying an ecosystem that spans 70% of the planet’s surface requires expensive research vessels, satellite data and high-speed computing. Oceanographers have had to share ships, instruments, hardware and other resources to further their own lines of enquiry. They have also had to collaborate across physical, chemical, biological, geological and meteorological approaches to assess what drives oceanic physical, biogeochemical and marine food-web dynamics. These oceanographic studies now form a foundation for global climate science.

The fruits of a coordinated effort in microbiome research that is grounded in ecological and evolutionary principles could be similarly significant.