The Human Microbiome Project at Baylor College of Medicine: How we did it
Carrying out a project as complex as the Human Microbiome Project took considerable efforts and coordination among the 80 institutions involved. Baylor College of Medicine and its collaborating institutions were integrally involved in much of the work including:
Defining who is healthy, normal
Dr. Wendy Keitel, professor of molecular virology and microbiology, and director of the Vaccine Research Center at BCM, said the upfront process defining who could take part in the study involved a tremendous amount of time and input in the form of conference calls, meetings and expert panel discussions.
"What is normal?" she said. "This has to be clearly defined and regarded as a reference set."
A panel of experts in microbiology, infectious diseases, and the microbiome were assembled several years in advance to strategize on the ideal number of body sites and subjects to sample, and the best way to do so.
"Jim Versalovic and Wendy Keitel and their group came up with the protocols needed to handle collection of samples from human subjects," said Dr. Sarah Highlander, associate professor of molecular virology and microbiology at BCM. She was chair of the working group that came up with a manual of procedure on how to process those samples in the laboratory. Four centers came together to determine the best way to extract DNA to ensure consistent results.
"Defining what was a "normal" subject for the Human Microbiome Project was a bit of a challenge," said Dr. Kjersti Aagaard, associate professor of obstetrics and gynecology at BCM. "What we were really seeking to find were 'reference' subjects, or folks that had likely minimally perturbed microbial communities. For these reasons we recruited men and women who were not obese, not on medications, and did not have chronic health problems. While this is perhaps not what is regarded as truly reflective of the current state of health or lifestyle for many, we felt that it would yield the most robust reference cohort available."
Keitel and her group had significant experience in recruiting subjects for studies, based on their vaccine work.
In the end, they recruited 150 of the 300 subjects needed for the study. Researchers at Washington University recruited the other 150. Results from 242 subjects were reported in the two Naturepapers.
As the researchers interviewed people who wanted to take part, they had to discuss the risks– which were minimal. One concern is privacy. Would their microbiomes be identifiable in the public databases where they would be placed? Dr. Amy McGuire, associate professor in the Center for Medical Ethics and Health Policy at BCM, was a leader in the Ethical, Legal and Social Implications (ELSI) of the Microbiome Project and helped define the risks in that effort. One mitigating factor was that while the sequences of the subjects' microbiomes with no personal identifiers would be available on the public database, their genomes or personal genetic code would not be.
Dr. James Versalovic, professor of pathology & immunology, pediatrics, molecular and human genetics and molecular virology and microbiology, helped design the methods of clinical sampling.
"I was just the facilitator," he said. Keitel, Dr. Kjersti Aagaard, Dr. Shital Patel, Dr. Jim Katancik, chair of periodontology at University of Texas Dental School at the time, and others were in the trenches, taking samples.
Subjects had to be instructed on the use of topical medicine, soaps, mouthwashes, hand sanitizers, bathing, showering and swimming before they came in for sampling. Researchers wanted them to have had as similar experiences as possible before the sampling, and they even gave them a bag of hygiene products to use in the days prior to the sampling.
At each visit, nine specimens were collected from the oral cavity and oropharynx: saliva, buccal mucosa (cheek), gums, palate, tonsils, throat and tongue soft tissues, and the tooth biofilmabove and below the gum. Skin specimens were collected from the creases behind each ear and the two inner elbows, and one specimen for the nostrils. Three vaginal specimens were collected from different areas of the vagina.
Subjects brought in a stool specimen in a special jar, which represented the microbiota of the lower gastrointestinal tract. Many people were asked to return for additional sampling to find out how stable the microbiome is over time.
At BCM, the work occurred in a single clinic area, making it easier for subjects to appear for their sampling procedures. Aagaard said the clinicians arrived at the clinic two afternoons a week for two years in order to carry out the procedures.
"It was a labor of scientific love", she said. "How many times in one's life do we have the opportunity to work on a 'big team science project' which will literally write scientific history? We had an incredible team of physician scientists, and an amazing group of subject volunteers."
Processing the samples
"We really had a superb atmosphere in which to do this," said Highlander. She credits Keitel, Versalovic and Dr. Janet Butel, chair of molecular virology and microbiology, for setting up the physical infrastructure that made that possible. The laboratory in which the samples were logged, processed and DNA extracted and catalogued before being sent for sequencing is just around the corner from the clinic. The close proximity of clinic and sample processing lab permitted rapid transfer of samples and easy communication between the two groups.
Graduate student Bonnie Youmans in Highlander's laboratory and Matthew Ross in the lab of Dr. Joseph Petrosino, assistant professor of molecular virology and microbiology, worked together to set, organize and code the thousands of samples that resulted. With the help of two technicians they extracted DNA from all the samples and packaged and sent them for sequencing at the four Genome Sequencing Centers that were part of the project.
A good deal of that sequencing took place at Baylor College of Medicine's own Human Genome Sequencing Center. Before that could begin in earnest, protocols for sequencing microbiome samples were needed.
Crucial to that was a mock community involving 20 bacteria developed by Highlander, said Dr. Kim Worley, associate professor in the Human Genome Sequencing Center. That community enabled experts in the Human Microbiome Project to develop analysis methods that enabled them to define the least and most common members of the microbial communities in samples where the DNA in the community varied ten-thousand fold between the most abundant and least abundant species. It also helped them fine tune methods for processing raw sequencing data.
"From some microbes, we did whole genome sequencing to get a better idea about the genes and therefore the functions provided by the different kinds of bugs," said Worley. "What's important? Is it the different species or is it that species in the mix have genes that carry out this or that function?"
In many instances, they used the conserved 16s ribosomal RNA gene to identify the different types of species of bacteria in a sample, which involved less sequencing.
"We participated in different phases of the sequencing", she said, "During the early Jumpstart phase, we finished a number of bacterial strains to use as references for this phase of the project." said Worley. "In total, we completed 191 bacterial genome sequences for the project that will be used as references for many ongoing and future human studies of the impact of the microbiome in health and disease. We also contributed to 16S and whole genome shotgun sequencing of the DNA from human samples."
Applications of the results to other analysis are ongoing. Versalovic is looking at how the microbiome is involved in chronic abdominal pain found in children and comparing the microbiomes in the intestines of children and adults as part of his work. He has published one study on the issue already and is hoping that the new microbiome information will further inform that work.
Aagaard has described the vaginal microbiome in pregnancy in a paper that appears today in the journal PLoS One.