The exact number of bacteria living in or on humans isn't known, though it is estimated to be around a trillion. In any case, the number of microbial cells in the human body outnumbers the human ones by a factor of 10. Those symbiotic microbes endow humans with otherwise unattainable metabolic capabilities, but also may contribute to human disease. What is certain, however, is that humans may really be classified as super-organisms — and ones that are mainly bacterial at that.
"There's a huge amount of gene activity [in the human body] that's being carried out by microbes," says George Weinstock, associate director of the Genome Center at Washington University in St. Louis. "There just has to be a very large impact of all of that microbial genetic activity on both healthy body as well as when pathological conditions set in different parts of the body. I think we're right at the very, very beginning of an extremely exciting era where we're going to make connections between things that we waved our hands about in the past or didn't understand or may not have even focused on."
Researchers have been studying the flora living in and on the human body for decades. The problem, however, is that most of the bacteria that can be seen peering through a microscope stubbornly refuse to grow on Petri dishes in microbiology labs. Still, scientists slogged through and managed to link these micro-organisms to everything from cirrhosis to irritable bowel syndrome to autism. Recently, though, the human microbiome has been lavished with attention and funds. Now, genomics approaches are being applied to identify the bacteria lurking in and on the human body.
Knowledge from what is being called the "second Human Genome Project" will be able to inform clinicians previously stymied by the cause of a disease or a patient's response to a drug treatment. Organizations around the world have poured money into this work — an international consortium was formed to foster collaboration and data-sharing among member countries. In the United States, the National Institutes of Health recently launched a five-year, $140 million Human Microbiome Project, modeled after the Human Genome Project, to better understand how the human microbiome affects human health and disease.
This project had been in the wings for a while, Weinstock says, but technology and tools hadn't caught up to the idea. "Then, four years ago or so when next-generation sequencing methods started coming and they also became highly adaptable for microbial genomics, you had the development of the perfect storm where there was a large microbial project just waiting there to be done," he says.
The known biome
Despite the lack of funding focus on the human microbiome until recently, researchers have been whittling away at the microbiome's unknowns: How is it established? How similar is it between individuals? How diverse is it? How does it affect human health? "This was of great interest at least to me and my colleagues elsewhere prior to the NIH Roadmap Program," says Stanford University's David Relman.
In 2007, Relman and his colleagues followed 14 babies — including a set of fraternal twins — as well as their mothers and most of the fathers during the babies' first year. The researchers profiled 16S rRNA sequences from the participants' stool samples using a specialized micro-array and sequencing in an effort to determine how the microbiome is established. And it arises quickly. By the end of a week, Relman says, there are a surprising number of taxa present, and a prominent source of microbiota for each baby was its mother. At the end of a year, the babies' microbiome is adult-like and is, for the most part, distinct from each other and the adults. "The one pair of twins that we had in our group, they were shockingly similar at any given time during that first year, one to the other," Relman says.
Relman and his colleagues, including the J. Craig Venter Institute's Karen Nelson, also characterized the diversity of the adult microbiome. In 2005, they analyzed the 16S rDNA of mucosal and fecal samples from three healthy human adults, finding 244 novel bacterial phylotypes. Most of the organisms they found belonged to one of two phyla, the Firmicutes or the Bacteroides. "That work also pointed out that there was a lot of this micro-diversity within populations," Nelson says. It's been described as a tree with thick limbs that branch into tiny twigs at the ends.
Then in 2006, they kicked off the metagenomics age of microbial studies by shotgun-sequencing the distal gut microbiome of two healthy adults — a study affectionately known as the HuPoop Project. "Using Sanger-based approaches, [we] could see a significant amount of microbial diversity in those two samples," Nelson says. "I think that everybody realized from that study that there's a tremendous amount of diversity in human that has yet to be tapped into."
Naturally, studies haven't focused solely on the gut. The National Human Genome Research Institute's Julia Segre characterized the skin microbiome by taking samples from 20 different skin sites on 10 healthy people. Also from 16S rRNA, she and her colleagues found 19 bacterial phyla living on the skin, though most belonged to Actinobacteria, Firmicutes, Proteobacteria, or Bacteroidetes. Like Relman and Nelson, Segre saw greater diversity at the species level.
She also noted that some bacteria prefer certain body sites, such as sebaceous sites. "They are very different in terms of properties, whether they have hair, whether they have sweat glands, and it turns out that those factors end up really influencing what [is there]," Segre says.
By following up on five of the volunteers, Segre also found that the skin microbiome is fairly stable. The volunteers were significantly more similar to themselves than to the other participants.
In other research, Imperial College London's Jeremy Nicholson has found that the microbiome affects how people respond to drug treatment. "The bugs make compounds that are actually competitive with drugs for metabolism," he says.
Recently, in an August issue of PNAS, he and his colleagues found that the chemical p-cresol, produced by gut microflora, affects how people metabolize Tylenol as both p-cresol and acetaminophen undergo sulfonation. Some people have microflora that are more active in cresol production than others, and when they take acetaminophen, there isn't enough sulfur for acetaminophen to go down that metabolic pathway. "Even Tylenol is affected by your gut microbe activity," Nicholson says.
Of course, some drugs are meant to affect the microbiome. Antibiotics act broadly on bacteria and disturb the microbiome. Last year, Relman and his colleagues studied the effect of a five-day course of Ciprofloxacin on healthy human gut communities. "What we saw was a return to essentially the same community after about four weeks of time, but there are some exceptions," Relman says. "The exceptions might be important." There were microorganisms that hadn't returned by the end of four months and others that were more abundant than they had been. "It's hard to say if that's important or what the long-term consequence might be, but we're now looking at longer time periods of follow-up," Relman says.
As in-depth as the studies have been, they have merely skimmed the top layer of what's to be found in the microbiome. A deeper look is still ahead.
Sequencing the microbiome
One of the main tools being used in the Human Microbiome Project is the sequencer. The first phase of that project is to develop tools to support further research into the microbiome. For that phase, four sequencing centers — Baylor College of Medicine, the Broad Institute, JCVI, and Washington University — have been working on sequencing bacteria from healthy individuals and 200 bacterial reference genomes.
That first prong was originally to take samples from 250 healthy volunteers between the ages of 18 and 40 at various body sites — 15 sites for men and 18 for women — to try to get a description of the human microbiome. Now, that number of volunteers has been upped to 750.
"The first goal ought to be to have some description of human microbiome that could be comparable to the working draft of the human reference sequence," Weinstock says. "That description of the human microbiome should go far beyond anything that has been done so far and to really help to answer some of the very basic questions about our different microbiome communities in different parts of the body and how they vary between individuals." According to Baylor College of Medicine's James Versalovic who is coordinating the clinical efforts for this phase of the project, more than 150 people have been sampled.
The 12,000 samples from the participants are a mixture of bacteria, prokaryotic microorganisms, and viruses; the sequencing centers want to do a sort of census to determine what microbes are there and in what quantities. One way to do a headcount of the bacteria, Weinstock says, is to sequence all the 16S rRNA in the mixture. "[16S] is the functional equivalent of a barcode for an organism," says Doyle Ward at the Broad Institute. That barcode can then be analyzed and compared against reference genomes to determine what bacteria are present.
The centers have been working together to establish rigorous methods for deeply surveying community composition by sequencing and analyzing 16S sequence. In particular, Ward is developing methods to do this work on the 454 Titanium platform, as well as new analytical tools. One example is ChimeraSlayer, which detects chimeras in 16S data. He is currently working on adapting it for 454 data.
However, looking at 16S rRNA doesn't answer all the questions. Weinstock says that while 16S rRNA gives a good idea of a bacterium's relative abundance, it doesn't take vastly different strains of the same bacteria into account. For example, he says that through 16S sequencing, E. coli all look the same. Some E. coli strains, such as the K12, are benign but others are not — in 1993, the O157:H7 strain killed four people who ate at Jack in the Box restaurants in the western US. In their gene content, those two strains differ by about 1,000 genes.
"Work that we've done looking at other environments has shown that organisms can have identical 16S sequences, but have wide variation in their genome content," Nelson adds.
Shotgun analysis of the DNA samples, on the other hand, will catch those virulence factors and other genes of interest. Since assembling genomes from that mix is very hard to do, Weinstock says that each of those sequences is analyzed to determine what genes are present in the community. "That's what allows you to detect toxins and things like that," he says.
This approach also allows researchers to determine whether any eukaryotes or viruses are also present. "The metagenomic approach allows us to really delve into the question of how much genetic diversity there is in these populations," Nelson says, adding that it also avoids introducing biases inherent to other approaches.
Data derived from those healthy volunteers then needs to be compared to something to tell just what their microbiomes contain. In addition to sequencing people's microbiome, the four sequencing centers are also tasked with creating reference bacterial genomes. "We are going to sequence, overall, about 1,000 microbial references genomes," Nelson says. Through the Human Microbiome Project, JCVI has received $8.8 million in American Recovery and Reinvestment Act funding to do just that. Baylor received $3.7 million and WashU was awarded $16.1 million.
But which lucky microorganisms will be sequenced? "The centers in the consortium arrived criteria that we tried to apply to our selections," the Broad's Ward says. Those criteria emphasize phylogenetic uniqueness and whether or not related organisms have been sequenced, as well as its dominance or abundance within its niche and whether it is associated with health or disease. As the Broad has no microbiology lab, Ward says they also rely on the clinical expertise of their collaborators to determine what would be interesting to sequence. Through the initial phase and pilot programs, the Broad has sequenced 150 microbial genomes and recently received an award to sequence at least 200 more.
Data from these large-scale projects are continually being released and updated. "We actively try and release data as quickly as possible," Nelson says.
Funding the smaller scale
With the first large-scale phase of the Human Microbiome Project under way, the second phase — smaller-scale projects — is now beginning. NIH has awarded $21.2 million for these demonstration projects, and program officially kicked off in June of this year. Many of these new projects focus on health and disease, but some will also look into technology development. "Fifteen new principal investigators that are from different institutions in the United States are now funded by NIH to pursue specific clinical projects," says Baylor's Versalovic, who is the clinical coordinator of these projects.
For his own work, Versalovic is studying irritable bowel syndrome in children between the ages of seven and 12 with his $750,000 grant. "We think that bacteria have a role to play — not only in the immune response, immunity inflammation, but also in terms of pain perception," he says. "There are some data that some bacteria that normally live within us actually affect the nervous system and the ways that, for example, transmission of pain and reception of pain and pain signals may be processed."
The University of Michigan's Vincent Young and his colleague, microbial ecologist Thomas Schmidt at Michigan State, have both been studying the gut microbiome for a decade; they received a $1 million grant to study ulcerative colitis, a bowel disease marked by abdominal pain, gurgling, and diarrhea.
Nelson is working with New York University's Zhiheng Pei on esophageal cancer. "Our preliminary work has suggested that you can see a correlation with various microbial population and the onset and development of esophageal cancer," Nelson says. She is also collaborating with NYU's Martin Blaser to study psoriasis, a condition in which skin becomes salmon-colored, scaly, and raised.
Another project comes out of Segre's lab at NHGRI and will be studying the relationship between the human immune system and the microbiome. "The NIH Clinical Center has a lot of patients who are immunocompromised. So we're looking at patients who are immuno-deficient and seeing how specific immunodeficiencies affect their microbiome," Segre says. "We just don't know that much about our relationship between our human cells and our microbial cells."
On the tool development end of the spectrum, there is the University of Colorado, Boulder's Rob Knight and the University of Maryland's Mihai Pop. Knight has been given $1 million to develop computational tools. "What we have been doing is developing ways to compare communities and individuals with [a] large number of sequences from each individual," Knight says, as in the past few years the scale of the data has increased rapidly. He and his lab are developing tools to make a phylogenetic tree from 16S bacterial sequences and measure the distances between them.
Pop is working with Steven Salzberg on a $780,000 grant to develop assembly and analysis tools for all the new data. "Essentially there is very little bioinformatics infrastructure for metagenomics data," Pop says. "In particular, there is I would say zero software available for assembly. There are a lot of issues in gene finding and binning."
Also, Young and Schmidt will be coming full circle and developing ways to "culture the unculturable," particularly of organisms near the mucosal lining of the intestines that need low levels of oxygen. So far, they have been able to grow organisms that had only been identified through molecular surveys. "One of the reasons for doing that is that part of the Human Microbiome Project, they have a list of 600 organisms that they want to get full genome sequences, [but] they want more things to sequence," Young says. "We thought that by doing this, we could generate more cultivars so that we can do full-genome sequencing to understand how different members of a microbiota make their living."
The practice of medicine
The ultimate goal of diving into the depths of the human microbiome is to make people healthier. If there is a core microbiome that most people have that keeps them healthy (see sidebar, p. 41), then, as the story goes, perturbations to that community could be tracked and correlated to disease. Then the microorganisms behind the problem could be targeted by therapies, or prebiotics or probiotics could preventively encourage the establishment of a healthy microbiome.
"In my opinion, the ultimate goal of this research is for us to understand how the meta-system — host, microbes, viruses, other organisms, metabolites, environment, et cetera — works as a whole and to be able to predict how perturbations in the system impact the various parts," RIKEN's Todd Taylor says. "This research is still in its infancy. There is a long, unpaved road ahead, but with perseverance, hard work, and innovation on the part of the researchers involved, we will discover many fascinating things along the way."
With these new tools, Segre says that the practice of medicine and diagnosing and treating diseases will change. "Now that we have the tools to do it, we can start to look for disorders we previously hadn't expected to have microbial infections," Segre says. Baylor's Versalovic adds that they may be able to develop new diagnostic test.
However, Relman says it will be difficult to prove Koch's postulates for a microbial community. Those postulates set the criteria that need to be fulfilled for an infectious agent to be the cause of a disease. Namely, the agent or microorganism has to be present in all cases of the disease, isolated from someone with the disease, cultured, and then cause disease in a new host from which the agent is then recovered. "Causation is a really tough nut to crack," Relman says. "What has emerged so far is a bunch of associations that no one can really be sure that the different-looking community is responsible for, or whether it simply is a result of, disease."
Instead, he says, longitudinal studies will be needed in which people at risk for certain diseases, and their microbiomes, are followed over time. "You're looking for an early change that predates the pathology," Relman. "That's going to be tough. It's doable, but it's going to take some really smart clinicians getting together with some willing subjects and good scientists and then being patient."
In the long run, though, a swarm of prebiotics and probiotics could be developed to encourage a healthy microbiome or to have an effect on a dysfunctional microbiome and bring the community into a better balance. "You might lack this important microbe — and there's some evidence now already that suggests there are some key elements missing in patients with certain diseases — so we may have diagnostic tests that detect certain bacteria for human health and, when absent, indicate susceptibility to disease," Versalovic says.
"I really think that this is a big deal and I think it is going to have a very big impact," Weinstock says.
Sidebar: Is There a Core Human Microbiome?
Part of the Human Microbiome Project aims to determine whether there is a core human microbiome that all people share. Right now, whether there is one or not is a matter of some controversy. And then, if there is, there's the matter of resolution — how far down the taxonomy ladder do you have to go? What percentage of people must have the group for it to be "common"? A few preliminary studies have suggested there is a core microbiome, but they are far from the full picture.
JCVI's Karen Nelson isn't convinced there is a microbiome common to all people. Early studies, she points out, were limited to North Americans. "I think that once we start to get out and do broader sampling, we will have a better feel," she says.
"To some extent, the question is: how big is the core group of organisms?" says George Weinstock at Washington University in St. Louis. From years of medical microbiology, he says that "we know that for certain parts of your body, there are certain organisms that we're pretty sure are going to be there."
However, Baylor's James Versalovic and Stanford's David Relman add that there is not even a definition of "core" just yet.
RIKEN's Todd Taylor says the issue depends largely on resolution. "If you are talking globally, at the strain level, I'd guess no. If you are talking more locally, say within Japan, at the strain level, maybe, at the genus level, almost certainly," he says. "There are many factors that need to be considered in the core microbiome question, such as, one, are we talking about genes or species; two, if species, at what level: genus, species, or strain; or, three, if geographic, in what range: global, continental, country, state, local, or some other region?"
Versalovic says he thinks there is a core microbiome but agrees it is a matter of resolution. "When you say there is a core, does that mean that E. coli, for example, is present in more than 50 percent of people or does that mean that E. coli is present in 80 percent of people?" he says. "I take a somewhat liberal perspective on this. More than 50 percent is probably enough."
Now with sequencing, more species can be found than before using culture plates. "The question then becomes: is the core just a handful of organisms, and then there's a lot of variation after that, or are there really quite a few additional organisms that are non-culturalable that are also frequently found in body sites?" Weinstock says. Furthermore, he adds that there could be more than one core group for a particular body site, based on environmental factors.
Then for a particular body site, one set of people might have one group of organisms and another have a different set of organisms. "There might be more than one type of core that people can have," Weinstock says.
"In some way, shape, or form, it would be nice to know what, if anything, we share that determines the ability of our communities to promote health or be associated with health," Relman says.