October 15, 2015 (Vol. 35, No. 18)
Vicki Glaser Writer GEN
Researchers Are Consulting Metagenomic Maps and Equipping Themselves with Computational Models and Analytical Tools
The human microbiome comprises microbial communities that occupy habitats such as the gut, skin, mouth, airway, and vagina. The healthy human microbiome can vary substantially between individuals, with multiple potential host-related factors (genetics, diet, early microbial exposure) and environmental variables contributing to this diversity.
The link between changes in the microbiome and disease is an emerging area of research, and the possibility of targeting the microbiome to prevent or treat a range of disorders holds promise for the drug discovery field.
The development of therapeutics will require abundant data and deeper knowledge of microbiome diversity, normal versus abnormal variation, and host–microbiome interaction. Fortunately, data in abundance and knowledge in depth are becoming available through metagenomics, the study and profiling of genomes from microbial communities that constitute the microbiome. Metagenomics, then, promises to advance drug discovery.
To obtain a glimpse of the inner workings of the microbiome, researchers at the University of Washington decided to peruse the microbiome’s metabolic machinery. These researchers, led by Elhanan Borenstein, Ph.D., used genomic and metagenomic data to illuminate the interactions that microbiome members had with each other as well as with the human host.
Dr. Borenstein and colleagues published their results in an article published last year in Cell Metabolism. The article highlighted some of the main challenges and strategies in “untangling the microbiome,” in inferring the metabolic capacity of potentially “unculturable” members of complex microbial communities, and in studying the interactions between various community members.
The researchers focused on developing computational methods to analyze and model the microbiome, integrating data from different types of assays—from the gene to the cell, network, species, and community levels. They described, for example, a culture-independent metagenomic pipeline that utilizes variation in species and gene profiles across metagenomic samples to infer computationally the genomic content of community members.
In previous work, Dr. Borenstein and colleagues used a metagenomic systems biology approach to study the human gut microbiome. They demonstrated topological differences at both the gene level and at the level of microbiome-wide metabolic networks in individuals with obesity and inflammatory bowel disease.
As methods for studying the microbiome advance, researchers are moving closer to being able to develop accurate and predictive models of microbiome composition and function. “Ultimately, this will allow us to design microbiomes, to tailor specific microbiome interventions, and to test interventions intended to change the microbiome,” says Dr. Borenstein. “Such models should, for example, integrate the metagenomic and the metametabolomic, not only statistically, but also mechanistically.”
Recently, Dr. Borenstein and colleagues introduced Inter-MUSiCC (Inter-sample Metagenomic Universal Single-Copy Correction), an alternative normalization scheme that corrects the biases that stem from the traditional normalization method that converts measured levels of genes or pathways into relative abundances. The conventional approach, Dr. Borenstein advises, “introduces a lot of spurious variation” that can greatly bias comparative analyses across samples.
MUSiCC normalizes the abundance of all genes in a sample to a set of universal, single-copy genes—such as ribosomal genes—providing “a more accurate picture of what’s going on in the metagenome,” according to Dr. Borenstein. In an article that appeared in Genome Biology earlier this year, Dr. Borenstein and colleagues presented data demonstrating that “MUSiCC significantly improves downstream discovery of functional shifts in the microbiome.”
Oral Microbiome Profiling
Oral diseases including tooth decay and gum disease are among the most prevalent infectious diseases in humans. Dental plaque, a biofilm present in the oral cavity, contains a mixture of microbial species, and ongoing profiling of the microbial diversity and genomic variation in the mouth is aimed at identifying associations between specific pathogens, genes, and metabolic pathways with oral health and disease.
These associations are being studied by Ping Xu, Ph.D., and John Gunsolley, D.D.S., a pair of researchers at Virginia Commonwealth University. They propose a model that incorporates three levels of interactions in the oral microbiome that determine oral health or disease. This model, presented last year in Virulence, emphasizes that the roles of pathogenic species and functions of specific genes in oral disease development have been recognized by metagenomic analysis.
Research on the oral microbiome is advancing rapidly, Ping Xu tells GEN, but developing an understanding of how the oral microbiome is associated with disease is a complex and far from complete task. He attributes the acceleration in metagenomics research to quickly advancing sequencing technology: “We are able to do sequencing faster, with longer and deeper reads allowing for much higher coverage, and at lower cost.”
Most current oral microbiome profiling studies have relied on 16S RNA sequence analysis because of its low cost and less intensive computational requirements compared to metagenomics. But metagenomics approaches to analyze the DNA of oral microbial communities can provide more comprehensive data and will replace 16S RNA sequencing as the technology becomes faster and cheaper.
We are just beginning to be able to analyze how the different microbial populations in humans—the respiratory, oral, intestinal, skin, and vaginal microbiomes—are related and might affect each other. “We know there are some links,” notes Dr. Xu, and we are studying “how pathogens affect different locations and how microbiome populations interact with the host,” as well as how the host responds to shifts in microbial communities.
Another important implication for the oral microbiome is the ability for oral pathogens to get into the blood and colonize organs such as the heart. Oral pathogens are recognized as a risk factor for systemic diseases, such as infective endocarditis. “The human immune system can often clean up these pathogens,” remarks Dr. Xu, but the risk may be greater in immunodeficient or immunosuppressed individuals.
At present, Dr. Xu is investigating the causes of oral infections and developing an understanding of the relationship between the oral microbiome and the host. In the latter effort, Dr. Xu is taking a systems biology approach. Dr. Xu is also working to identify a core set of microbes in an oral sample that could be used as a diagnostic tool to detect disease.
A key challenge in the development of a diagnostic is that the microbiome subpopulation must be quite stable and the test highly accurate. It is particularly difficult to classify a reliable set of microorganisms for diagnosis at this time due to the dynamic nature of the oral microbiome, which changes with daily oral activities, before and after brushing teeth, for example.
Surveying Diverse Skin Habitats
Characterizing the human skin metagenome can be especially challenging due to the diverse microenvironments across the skin surface and its exposure to the external environment. Researchers from the National Human Genome Research Institute and the National Cancer Institute published a multisite metagenomic study of healthy human skin, identifying bacterial, fungal, and viral communities in the skin microbiome. This study, which appeared last year in Nature, emphasized how microbes demonstrate site specificity. It also noted that biogeography and individuality help shape the functional diversity and potential of microbial communities.
Investigating Mental Illness at the Gut Level
The literature supports an association between human obesity and increased prevalence of neuropsychiatric disorders, particular depression, dementia, and anxiety, as well as deficits in learning and memory. One or more factors could contribute to this link, including diet.
Diets high in saturated fat promote obesity and induce an inflammatory response in the intestines. The effects of a diet high in saturated fats on the core composition of the gut microbiome—and the neuropsychiatric implications of an altered microbiome—were studied by Annadora J. Bruce-Keller, Ph.D., J. Michael Salbaum, Ph.D., and colleagues at Louisiana State University (LSU) and LSU Health Sciences Center. These researchers considered whether dietary or pharmacologic manipulation of gut microbiota could attenuate the neurologic complications of obesity.
“There is significant descriptive data linking obesity and/or metabolic dysfunction to the prevalence and severity of a variety of psychiatric and neurologic disorders,” says Dr. Bruce-Keller. “The same is true for alternations in the gut microbiome.” For example, children with autism and individuals with major depression typically have dysbiosis, or an altered pattern of bacterial populations in the intestines.
The LSU researchers detailed their work in a paper that appeared earlier this year in Biological Psychiatry. The paper compared the behavior of two groups of normal weight mice that were both treated with microbiome depletion/transplantation. One group received a microbiome transplant from donor mice fed a high-fat diet (HFD); the other group received a microbiome transplant from donors fed a control diet.
The HFD donor transplant was intended to mimic the gut microbiome in obese mice in the absence of other metabolic and physiologic changes associated with obesity. “These mice were not obese, but they were not healthy,” clarifies Dr. Bruce-Keller. He adds that they had significant increases in markers of systemic and brain inflammation.
The mice that received HFD microbiome transplants demonstrated an overall pattern of neurobehavioral impairment, characterized by an increase in agitation, fear, and anxiety-like behaviors; decreased exploratory behavior; a tendency to be more cautious; and some evidence of memory impairment. The results of this study led the researchers to conclude that “therapeutic manipulation of the microbiome, which should be highly responsive compared with existing clinical targets, could dramatically mitigate the prevalence and/or severity of neuropsychiatric disorders.”
While the human microbiome varies across individuals, is that variation sufficient to allow for the development of unique microbiome “fingerprints” that persist over time and could it be used to distinguish individuals in a population? This question prompted an investigation by Eric Franzosa, Ph.D., a researcher affiliated with the Harvard T.H. Chan School of Public Health and the Broad Institute. Dr. Franzosa, a bioinformatics specialist, decided to develop an algorithm based on body site–specific metagenomic codes that can identify individuals from among populations numbering in the hundreds based on their microbiomes alone.
Along with colleagues from the University of Oregon, the Forsyth Institute, and Harvard Medical School, Dr. Franzosa described a “personal microbiome” methodology in an article that appeared earlier this year in the Proceedings of the National Academy of Sciences. The authors indicated that microbiome-based identification is feasible while highlighting “the important ethical implications for microbiome study design” and the “potential privacy concerns for subjects enrolled in human microbiome research projects.”
The researchers considered microbial DNA sequences from individuals sampled during the Human Microbiome Project and trained a computer algorithm to construct metagenomic codes based on multiple body sites and microbial measurements. Comparisons of samples from individuals taken at two time points (30 to 300 days apart) enabled quantification of metagenomic code stability and specificity.
Overall, based on comparisons with individuals’ follow-up samples, metagenomic codes could uniquely pinpoint about 30% of the individuals in the study population. The codes based on the gut microbiome exhibited the greatest stability and could pinpoint greater than 80% of individuals nearly a year later.
Early on, as they were applying the algorithm to construct unique metagenomic codes for individuals, Dr. Franzosa and colleagues found that there were not enough unique combinations at the species level; training the algorithm on strain-level variation, which includes the gain or loss of genes across members of a species, proved more successful. Whereas the human genome has been studied extensively and is relatively well characterized, the human microbiome, which consists of complex microbial communities, has only recently been the subject of profiling efforts.
“The method we developed for constructing these microbiome fingerprints is very general, and can work with a variety of different microbial measurements,” asserts Dr. Franzosa.
“With the measurements we have today, we can pinpoint an individual within a population of hundreds of people. As our ability to measure and describe the microbiome improves, we expect to be able to pinpoint individuals within progressively larger populations, such that the microbiome’s identifying power could one day rival that of our genomes.”