Metagenomic analysis of double-stranded DNA viruses in healthy adults

Abstract

Background: The Human Microbiome Project (HMP) was undertaken with the goal of defining microbial communities in and on the bodies of healthy individuals using high-throughput, metagenomic sequencing analysis. The viruses present in these microbial communities, the 'human virome', are an important aspect of the human microbiome that is particularly understudied in the absence of overt disease. We analyzed eukaryotic double-stranded DNA (dsDNA) viruses, together with dsDNA replicative intermediates of single-stranded DNA viruses, in metagenomic sequence data generated by the HMP. 706 samples from 102 subjects were studied, with each subject sampled at up to five major body habitats: nose, skin, mouth, vagina, and stool. Fifty-one individuals had samples taken at two or three time points 30 to 359 days apart from at least one of the body habitats.

Results: We detected an average of 5.5 viral genera in each individual. At least 1 virus was detected in 92% of the individuals sampled. These viruses included herpesviruses, papillomaviruses, polyomaviruses, adenoviruses, anelloviruses, parvoviruses, and circoviruses. Each individual had a distinct viral profile, demonstrating the high interpersonal diversity of the virome. Some components of the virome were stable over time.

Conclusions: This study is the first to use high-throughput DNA sequencing to describe the diversity of eukaryotic dsDNA viruses in a large cohort of normal individuals who were sampled at multiple body sites. Our results show that the human virome is a complex component of the microbial flora. Some viruses establish long-term infections that may be associated with increased risk or possibly with protection from disease. A better understanding of the composition and dynamics of the virome may hold important keys to human health.

Figure 1

The human virome in healthy, asymptomatic adults. (A) The histogram shows the number of individuals (x-axis) who were positive for a given number of different viral genera (y-axis). (B) The viral genera (x-axis) detected in each subject (y-axis) are represented by black bars. The virome of each individual is viewed by looking at the black bars in a given row.

Figure 2

The human virome in five body habitats. (A) All of the viruses detected in the five body habitats are shown. Each virus is represented by a colored bar and labeled on the y-axis on the right side. The relative height of the bar reflects the percentage of subjects sampled at each body site in whom the virus was detected. In this panel, the bar representing roseoloviruses in the oral samples reflects the maximum bar height, because 98% of the individuals who were sampled in the mouth harbored roseoloviruses. (B) This panel shows papillomaviruses included in the category ‘Other papillomaviruses’. The largest bar height shown represents the unclassified papillomaviruses found in skin samples from 65% of subjects. The percentages represented in this figure are shown in Additional file 1: Table S3.

Figure 3

Individuals have unique viral fingerprints. (A-D) Each panel represents the virome of a specific subject. The dark rectangles represent the presence of a particular virus (x-axis) in a body habitat (y-axis).

Figure 4

Stability of the human virome across two evaluations. The whole genome shotgun sequence data included a subset of samples from 51 individuals, which were collected at two time points. For viruses detected in five or more individuals, Panel A shows the percentage of subjects in whom the virus was detected at only one (gray bar) or two consecutive (black bar) visits. The complete census of viruses that were detected in the longitudinal data set is presented in Additional file 1: Table S4. Panels B-G display the viruses detected in samples from a subject at one or both evaluations. Gray rectangles signify that the indicated virus was detected in a sample collected in one of two evaluations; black rectangles signify that the indicated virus was detected in the samples collected at both evaluations. Data from six individuals are shown in this figure for the purpose of illustration. Additional file 1: Table S5 shows the data for all subjects with multiple visits.

Figure 5

Variation in the vaginal microbiome. All posterior fornix samples were clustered based on bacterial community structure. The 20 most abundant components of the bacterial communities are shown in the stacked bar charts. The presence of alpha-papillomaviruses is indicated with a red circle at the base of each stacked bar.