The Integrative Human Microbiome Project

Abstract

The NIH Human Microbiome Project (HMP) has been carried out over ten years and two phases to provide resources, methods, and discoveries that link interactions between humans and their microbiomes to health-related outcomes. The recently completed second phase, the Integrative Human Microbiome Project, comprised studies of dynamic changes in the microbiome and host under three conditions: pregnancy and preterm birth; inflammatory bowel diseases; and stressors that affect individuals with prediabetes. The associated research begins to elucidate mechanisms of host-microbiome interactions under these conditions, provides unique data resources (at the HMP Data Coordination Center), and represents a paradigm for future multi-omic studies of the human microbiome.

Fig. 1. The first and second phases of the NIH Human Microbiome Project.

The ten-year NIH Human Microbiome Project (HMP) program, organized into two phases (HMP1 and HMP2), developed reference sequences, multi-omic data sets, computational and statistical tools, and analytical and clinical protocols as resources for the broader research community. The HMP1 focused on the characterization of microbial communities from numerous body sites (oral, nasal, vaginal, gut, and skin) in a baseline study of healthy adult subjects, and included a set of demonstration projects that focused on specific diseases or disorders. The HMP2 expanded the repertoire of biological properties analysed for both host and microbiome in three longitudinal cohort studies of representative microbiome-associated conditions: pregnancy and preterm birth (vaginal microbiomes of pregnant women), inflammatory bowel diseases (gut microbiome) and prediabetes (gut and nasal microbiomes). These studies followed the dynamics of these conditions through multi-omic analyses of multiple measurement types over time, including changes in microbial community composition, viromics, metabolomic profiles, gene expression and protein profiles from both host and microbiome, and host-specific properties such as genetic, epigenomic, antibody, and cytokine profiles, along with other study-specific features. All sequences and multi-omic data, clinical information, and tools from both HMP1 and HMP2 are housed in the HMP Data Coordination Center (DCC) or referenced public or controlled-access repositories to serve as a central resource for the research community.

Fig. 2. The vaginal microbiome and its relationships with host factors in pregnancy and preterm birth.

The MOMS-PI project followed 1,527 pregnancies longitudinally and involved the collection of 206,437 biospecimens for analysis of host and microbial factors (16S amplicon, metagenomic, and metatranscriptomic sequencing; cytokine profiling; metabolomics; proteomics; genomics; and microbial isolate culture). Around 600 pregnancies were analysed in depth to assess features that lead to preterm birth; this analysis identified both host (for example, cytokine) and microbial (for example, ecological and specific strain) factors. As pregnancy progresses, with predictable changes in systemic oestradiol levels, the uterine and vaginal environments undergo various changes. The uterus switches from an early pro-inflammatory condition to an anti-inflammatory condition in the second trimester, and then back to a pro-inflammatory condition before the onset of labour. Meanwhile, specific changes in the microbiome of the vaginal lumen can be associated with preterm birth, possibly through mechanisms involving microorganisms travelling from the vagina to the uterus. The figure depicts an overview of longitudinal changes in the vaginal mucosal ecosystem and uterus during pregnancy.

Fig. 3. Host–microbiome dynamics in IBD.

The IBDMDB followed more than 100 participants with IBD (Crohn’s disease or ulcerative colitis), as well control individuals without IBD, for one year each to assess host and microbial molecular activity during changes in disease activity and gastrointestinal inflammation. Nearly all measured host–microbiome properties showed changes in either activity or stability during disease, including those shown here—not only microbial taxa and microbial transcription, but also host- and microbiota-derived small molecules in the gut, epithelial transcriptional responses at multiple points along the colon, and circulating antibody levels in peripheral blood serology.

Fig. 4. Differential host and microbial responses to dietary perturbations and infectious disease in individuals with prediabetes.

For integrated personal ’omics profiling of the microbiome during prediabetes, 106 participants were followed for up to four years, with samples (primarily blood and stool) collected quarterly and additional samples collected during periods of RVI and other stresses. The genomes of the participants were sequenced and measurements of transcriptomes, proteomes, metabolomes, and microbiomes taken at each visit, in addition to clinical details. Insulin-resistant individuals showed differences from insulin-sensitive participants in various measurements, both at baseline and in response to the stresses such as weight loss and RVI.

Fig. 5. Resources from HMP1 and HMP2 available at the DCC.

The HMP DCC (http://ihmpdcc.org) hosts raw and processed data from both phases of the Human Microbiome Project, comprising in total more than 42 Tb of multi-omic data. From the HMP1, these include 16S rRNA gene amplicons and metagenomes from the healthy human subjects (HHS) baseline cohort, as well as resources from demonstration projects (https://www.hmpdacc.org/health/projectdemos.php) and genomes of associated microbial isolates. From the HMP2, these include data that IRBs gave permission to be made publicly available from the pregnancy and preterm birth (MOMS-PI), inflammatory bowel disease (IBDMDB), and prediabetes (IPOP) projects, with links to raw data in other repositories. Additional data deposited elsewhere including microbial reference genomes, HMP1 human genomes, and controlled access data for all HMP2 projects is also linked from the DCC. Categories of data are colour coded, and the number of items in each dataset is indicated by the size of the circles.