Ly et al.

Microbiome (2016) 4:64

DOI 10.1186/s40168-016-0212-z

RESEARCH Open Access

Transmission of viruses via our

microbiomes
Melissa Ly1, Marcus B. Jones2, Shira R. Abeles3, Tasha M. Santiago-Rodriguez1, Jonathan Gao1, Ivan C. Chan1,
Chandrabali Ghose4 and David T. Pride1,3*

Abstract
Background: Bacteria inhabiting the human body have important roles in a number of physiological processes and
are known to be shared amongst genetically-related individuals. Far less is known about viruses inhabiting the
human body, but their ecology suggests they may be shared between close contacts.
Results: Here, we report the ecology of viruses in the guts and mouths of a cohort and demonstrate that
substantial numbers of gut and oral viruses were shared amongst genetically unrelated, cohabitating individuals.
Most of these viruses were bacteriophages, and each individual had distinct oral and gut viral ecology from their
housemates despite the fact that some of their bacteriophages were shared. The distribution of bacteriophages
over time within households indicated that they were frequently transmitted between the microbiomes of
household contacts.
Conclusions: Because bacteriophages may shape human oral and gut bacterial ecology, their transmission to
household contacts suggests they could have substantial roles in shaping the microbiota within a household.
Keywords: Saliva, Gut, Microbiome, Microbiota, Antibiotic perturbations, Antibiotic courses, Antibiotics, Virus, Virus
transmission, Virome, Bacteriophage

Background inhabiting the human virome may have a significant

Emerging data suggest that viruses are integral members capacity to shape our cellular microbiomes.
of the human microbiome, inhabiting the surfaces of the How the microbiome responds to perturbations such
skin [1], mouth [2], gut [3], and urinary tract [4]. Many as antibiotics may determine our susceptibility to patho-
of these viruses are bacteriophages that infect the gen colonization and/or to the development of certain
numerous bacteria that also inhabit these surfaces, and diseases [6–8]. Viral communities also respond to
may have antagonistic relationships (where they perturbations, although in different ways than might be
frequently kill their hosts) or mutualistic relationships predicted based on the responses of their bacterial hosts.
(where they integrate into their host’s genome and The gut viromes of mice and humans have more homo-
potentially provide beneficial gene functions) with their logues to genes involved in antibiotic resistance after
host bacteria. Further study of these viromes has demon- antibiotic perturbations [9, 10], but a recent study
strated that these bacteriophages carry numerous gene indicates that most of these genes might not be involved
functions including complement and immunoglobulin in antibiotic resistance [11]. Viral communities do not
degradation [2], and platelet binding [5] among other appear to have altered diversity in the long-term in
gene functions that might provide significant benefits to response to antimicrobial therapy, potentially because
their bacterial hosts. Whether they kill their hosts viruses that exit the community after perturbations are
rapidly or provide selective advantages to their hosts replaced by other eukaryotic viruses [10].
through the process of transduction, bacteriophages The finding that bacteriophages are highly prevalent
members of the human microbiome led to the hypothesis
* Correspondence: dpride@ucsd.edu that viruses were transient in the human microbiome [2,
1
Department of Pathology, University of California, San Diego 92093, USA
3
Department of Medicine, University of California, San Diego 92093, USA
12], as they likely identified their hosts rapidly, killed
Full list of author information is available at the end of the article them, and spread their progeny to other susceptible hosts.
© The Author(s). 2016 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Ly et al. Microbiome (2016) 4:64 Page 2 of 10

Recent studies now have shown that bacteriophage mem- (vitamin C). The four separate controls did not receive
bers of the oral [12] and gut [3] microbiomes can be any therapy. Each subject was sampled on day 0 (day
highly persistent. That persistence can have consequences, prior to antibiotics), day 3 (on the third day of antibi-
as one study suggests that phages may be shared amongst otics), day 7, week 8, and at 6 months.
close contacts [13]. If these phages can be transmitted be-
tween close contacts, the persistence provides further op- Individual-specific patterns of viruses in the gut and
portunities for the dissemination of gene functions such mouth
as complement or immunoglobulin degradation carried in We isolated viruses from all 97 fecal and 95 saliva
human viral communities. samples according to our previously described protocols
In this study, we recruited and sampled the saliva and [2], which involved sequential filtering to remove cellular
feces of a cohort of genetically unrelated individuals debris, CsCl density gradient ultracentrifugation, and
from different households to discern whether viral mem- DNA extraction from intact virions. Resulting DNA was
bers of the human microbiome may be transmitted sequenced using Semiconductor Sequencing [14] for a
between subjects as a result of close contact. We gave total of 118,527,761 reads with a mean length of 215
an antibiotic to one household member and placebo to nucleotides. There were 63,092,083 fecal reads with a
the other household member to decipher whether the mean GC content of 39.9% and 55,435,678 salivary reads
use of antibiotic perturbations may affect the direction with a mean GC content of 43.0%. We sequenced an
of the sharing of bacteriophages in each household. average of 5,926,388 reads per subject and an average of
1,234,664 reads per each time point.
Results We characterized the ecology of the viruses in each of
Study cohort the eight households to decipher whether the individuals
We recruited a cohort of 20 subjects and sampled their in each household could be distinguished based on their
saliva and feces over a 6-month period. Of the 20 fecal or oral viruses and whether the use of antibiotics
subjects enrolled, there were eight separate households might disrupt individual-specific patterns of viruses
consisting of two individuals and four separate controls within each household. Using principal coordinates
not enrolled with a housemate (Additional file 1: Table analysis (PCOA) to visualize the beta diversity between
S1). In each household, one individual received treat- household members, each individual could be distin-
ment with an antibiotic (azithromycin or amoxicillin) for guished based on their gut (Fig. 1) and oral viromes
7 days and the other individual received a placebo (Additional file 2: Figure S1). These individual-specific

Fig. 1 Principal coordinates analysis of beta diversity present in the feces of all households. Panels a–h represent households 1–8, respectively.
Subjects who received placebo are represented by squares, amoxicillin are represented by circles, and azithromycin are represented by triangles.
Specimens collected from day 0 are represented in red, day 3 in blue, day 7 in green, week 8 in yellow, and month 6 in orange
Ly et al. Microbiome (2016) 4:64 Page 3 of 10

patterns were not disrupted by the use of either amoxi- Household-specific patterns of viruses in the gut and mouth
cillin (Panels A–D) or azithromycin (Panels E–H). The We also tested whether household-specific patterns
individual-specific patterns were significant in the feces of viral ecology may exist in the cohort in addition
of 13/16 subjects and in the saliva of 7/16 subjects when to individual-specific patterns of viruses we observed
using a permutation test to compare the viruses within (Fig. 1). We utilized PCOA to compare all subjects
an individual over time with that of their housemates in households that received amoxicillin or azithro-
(Table 1). mycin and observed patterns that suggested shared

Table 1 Viral homologues within and between individuals

House Subject Percent homologous within subjectsa Percent homologous between subjectsa p valueb
Feces
House 1 CA05 73.55 ± 6.33 14.27 ± 5.13 <0.0001
CA06 51.91 ± 11.30 12.48 ± 7.15 0.0058
House 2 CA13 61.49 ± 14.60 8.27 ± 6.01 <0.0001
CA14 44.27 ± 20.86 2.96 ± 2.68 <0.0001
House 3 CA23 45.23 ± 19.70 3.23 ± 4.12 <0.0001
CA24 43.87 ± 26.21 3.76 ± 4.56 0.0009
House 4 CA27 35.38 ± 29.00 3.61 ± 6.56 0.0550
CA28 36.64 ± 23.35 1.47 ± 3.11 0.0007
House 5 CA11 20.35 ± 21.07 1.00 ± 2.07 0.0646
CA12 27.99 ± 28.24 1.61 ± 2.94 0.0893
House 6 CA15 8.06 ± 11.04 3.87 ± 3.81 0.2800
CA16 51.27 ± 6.92 3.30 ± 3.71 <0.0001
House 7 CA39 47.19 ± 14.19 10.20 ± 5.12 0.0100
CA40 68.41 ± 3.30 11.35 ± 11.18 <0.0001
House 8 CA47 57.75 ± 5.80 14.74 ± 6.56 <0.0001
CA48 60.67 ± 8.20 16.84 ± 7.64 <0.0001
Saliva
House 1 CA05 14.11 ± 8.17 4.86 ± 3.71 0.1100
CA06 28.25 ± 17.26 5.79 ± 5.90 0.1000
House 2 CA13 39.51 ± 14.08 3.57 ± 3.89 0.0001
CA14 19.19 ± 17.37 6.24 ± 6.27 0.1635
House 3 CA23 26.85 ± 13.38 4.86 ± 6.57 0.0459
CA24 29.54 ± 9.62 4.29 ± 5.86 0.0099
House 4 CA27 24.03 ± 9.35 11.35 ± 9.14 0.1798
CA28 38.60 ± 26.63 16.75 ± 16.43 0.1275
House 5 CA11 29.57 ± 24.08 7.50 ± 5.85 0.1552
CA12 31.71 ± 26.37 8.75 ± 7.28 0.1579
House 6 CA15 24.33 ± 22.43 8.59 ± 9.49 0.1438
CA16 14.56 ± 18.61 7.09 ± 7.79 0.2335
House 7 CA39 9.51 ± 13.11 0.34 ± 1.19 0.0505
CA40 47.95 ± 5.12 0.24 ± 0.45 <0.0001
House 8 CA47 35.80 ± 15.61 9.77 ± 8.34 0.0529
CA48 39.48 ± 9.52 14.17 ± 8.09 0.0211
a
Based on the mean of 10,000 iterations. Per iteration, 10,000 random contigs were sampled
b
Empirical p value based on the fraction of times the estimated percent homologous reads within each subject exceeds that for different subjects. p values ≤0.05
are represented in italics
Ly et al. Microbiome (2016) 4:64 Page 4 of 10

fecal viral ecology within the households for each house- proportions of shared viruses between the members
hold evaluated (Fig. 2). These patterns observed were of each household. We found similar patterns of
statistically significant in the feces for five out of eight shared viruses in the gut (Additional file 2: Figure
households as was observed through permutation analysis S5, Panel A), and in the mouth (Panel B). There
comparing the shared viral ecology within a household were generally more shared viruses within each
with the shared viral ecology between different households household in the gut than were found in saliva, but
(Additional file 1: Table S2). Similar patterns were observed the difference did not meet statistical significance
in the mouths of each of the household members, but were (Additional file 2: Figure S6). The majority of the
only significant in three out of eight households (Additional viruses that were not shared in the guts of house-
file 1: Table S2 and Additional file 2: Figure S2). mates were found in those individuals who took a
placebo (Panel A), while roughly equal numbers of
unique oral viruses in each household were found in
Household viruses are persistent and shared subjects who received either an antibiotic or placebo
A prior study indicated that oral bacteriophages are (Panel B).
highly persistent and that the gut viruses of a single We also measured Sorensen distances between the
individual were persistent over time [13]. We evaluated individuals in each household and compared them
the persistence of viruses in our cohort over time to de- with individuals from different households to decipher
termine whether their persistence may be altered by the whether there were significant similarities between the
use of antibiotics. We found that fecal viruses were individuals in each household that indicated that
highly persistent with greater than 70% found at least viruses were shared within a household. We measured
4 days apart, and 30% identifiable at 5 to 6 months Sorensen rather than Bray Curtis distances because
(Additional file 2: Figure S3). There was no identifiable Sorensen distances are useful for measuring similarity
effect the antibiotics had on the persistence of phages, as between viromes, while Bray Curtis distances measure
similar patterns were observed in amoxicillin and dissimilarity. In the gut, we found significantly higher
azithromycin households that were observed in the levels of similarity amongst household members than
controls who took no therapy. The levels of persistence were found between individuals who resided in
in the gut far exceeded those observed in the mouth different households (Fig. 3, Panel A). We also
(Additional file 2: Figure S4). identified higher similarity in the saliva for household
Because we found patterns that suggested shared members, but these results were not statistically
household viral ecology (Fig. 2), we quantified the significant (Panel B).

Fig. 2 Principal coordinates analysis of beta diversity present in the feces of all subjects. Panel a represents subjects who received amoxicillin,
their housemates who received a placebo and control subjects who received no therapy. Panel b represents subjects who received azithromycin,
their housemates who received a placebo and control subjects who received no therapy. Controls who received placebo or no therapy are
demonstrated by black outer circles and subjects who received antibiotics are represented by gray outer circles. Outlines are drawn around all time
points representing individual households
Ly et al. Microbiome (2016) 4:64 Page 5 of 10

Fig. 3 Bar graphs comparing Sorensen distances (± standard error) between subjects within a household and between subjects from different
households for feces (panel a) and saliva (panel b). p values are represented above the bars

Putative transmissions and directionality of transmissions amplified and sequenced from the 6-month time point.
in each household While there were a few polymorphisms in the phage
We found bacteriophages that were shared in the mouth over the course of time, the phage only had one poly-
and the gut of housemates from all households (Fig. 3, morphism between subjects CA39 and CA40 at the 6-
Additional file 2: Figure S5, Figure S6). We examined month time point. These data suggest that between week
the structure of some of these phages and found that 8 and month 6, the phage was transmitted from subject
some were present in both housemates on most days CA40 to subject CA39. We characterized the presence/
examined (Additional file 2: Figures S7–S10). We identi- absence of viruses in all households over the course of
fied virulence factors within the genome structures of the study and defined any virus that was present in one
many of these phages, including an enterotoxin, toxin- subject, absent in their housemate, and later appeared in
antitoxin system, and platelet binding protein (Add- at least two time points in the housemate as a putative
itional file 2: Figure S7), mucin and m23 peptidase (Add- transmission. Using these criteria, we found that of the
itional file 2: Figure S8), and a beta-lactamase that could 23.7 ± 0.05% of fecal viruses that were shared, 9.8 ±
be involved in the degradation of beta-lactam antibiotics 0.03% had a pattern consistent with a putative transmis-
such as amoxicillin (Additional file 2: Figure S9). We sion between housemates (Fig. 5). When examining the
also identified a phage with significant similarity to directionality of the putative transmissions, the majority
crAssphage [15] shared in the members of household 6 (7.4 ± 0.04%) were consistent with transmission from the
(Additional file 2: Figure S10). Finding genes poten- subject taking the placebo to the subject taking antibi-
tially involved in antibiotic resistance in the structures of otics, while the minority (2.4 ± 0.01%) went in the other
phage genomes shared between household contacts sug- direction. Similar results were identified in terms of the
gests that the sharing of these phages could be a means number of putative viral transmissions in the mouth, but
by which antibiotic resistance spreads through close there was little difference in the directionality of these
contact with others. transmissions between housemates (Additional file 2:
We identified numerous fecal phages whose presence/ Figure S12).
absences on certain days between the housemates
suggested that they had been transmitted from one Discussion
household member to the other. For example, we could It has been well established that genetically related indi-
identify portions of a phage present in all days in the viduals and some spouses who share a household also
fecal viromes of one subject in household 7 (includes share some portion of their microbiota [13, 16, 17]. Most
subjects CA39 and CA40), but only after 6 months in of these studies have focused on the bacteria that inhabit
their housemate (Additional file 2: Figure S11). We the human microbiome, while ignoring the estimated
verified the structure of this virus by PCR amplification 1015 viruses that inhabit the human body. We utilized
and Sanger sequencing of its genome from all days in CsCl density gradients to purify viruses which often
subject CA40 (Fig. 4). We attempted the same in the results in highly purified virus fractions, but introduces
housemate (subject CA39), but the phage could only be biases when examining the relative abundances of some
Ly et al. Microbiome (2016) 4:64 Page 6 of 10

Fig. 4 Diagram of contig71 assembled from Sanger sequences from all time points in subject CA40 and from the month 6 time point in subject
CA39. The contig was not identified on days 0, 3, 7, or week 8 in subject CA39. Putative ORFs and their directions are indicated by the arrows at
the top of the diagram. ORFs that had significant homologues (BLASTP E-score <10−5) are indicated by the text above each arrow. The location of
polymorphisms (when compared to the day 0 time point in subject CA40) are indicated by orange vertical lines

viruses [18]. Our data indicate that neither a genetic for how long the household members had lived together
relationship nor a spousal relationship is required to prior to the initiation of this study, but had households in
share viruses, and that unrelated individuals who share a this study who had lived together for as little as 1 month
household also will share some proportion of their prior to this study, so it may be a matter of days or weeks
viruses. This sharing is not based on intimate contact, as before viruses are shared between household members. In
both romantic couples and roommates were enrolled in general, there were greater levels of shared viruses found in
this study and all of them shared viruses. Because six of the gut than in the mouths of our subjects, which we
the eight households enrolled were romantic couples, we believe is related to the higher persistence of viruses in the
could not discern statistically whether there were more gut and suggests less turnover and greater opportunities to
shared viruses amongst the romantic couples; however, share fecal viruses within a household. A prior study
there was no observed trend of greater sharing of viruses reported that individuals on similar diets converge upon
amongst the households studied (Additional file 2: similar gut virome phenotypes [3]. None of the subjects re-
Figure S5). We did identify a non-significant trend of ported any changes to their diets during this study, but we
greater shared bacterial biota amongst romantic couples did not monitor whether housemates shared meals, which
in a study using a larger cohort [19]. We did not control could have resulted in greater shared virome contents.
Our finding of both virulence factors (Additional file 2:
Figures S7 and S8) and an antibiotic resistance homologue
(Additional file 2: Figure S9) within the structures of bacte-
riophages suggests that their sharing could have conse-
quences for the microbiota of housemates to which these
genes are shared. The sharing of antibiotic resistance genes
such as a beta-lactamase could result in resilience to amoxi-
cillin perturbations in the recipient; however, we have not
demonstrated that the homologue is involved in antibiotic
resistance. A recent study suggests that most of the homo-
logues identified to antibiotic resistance genes in viromes
may not be involved in antibiotic resistance; thus, func-
tional assays need to be performed to demonstrate
antibiotic resistance in viromes. While only about 10% of
Fig. 5 Bar graph representing the mean proportions of fecal viromes the shared viruses were identified as putative transmissions,
(± standard error) shared between housemates, putative transmissions the fact that we did not control for the length of time
between housemates, putative transmissions from subjects taking individuals had cohabited prior to the beginning of the
antibiotics to subjects taking placebo, and putative transmissions from
study suggests a larger portion (about 15%) of the viruses
subjects taking placebo to subjects taking antibiotics
may have been transmitted between subjects prior to the
Ly et al. Microbiome (2016) 4:64 Page 7 of 10

start of the study. A more controlled study where subjects received 7 days of an antibiotic and the other subject
are sampled both prior to and after they cohabit would be received 7 days of the placebo (vitamin C). The dose of
necessary to gain a more detailed understanding of the amoxicillin was 500 mg twice daily, and the dose of vita-
dynamics of viral transmissions between individuals. min C was 500 mg twice daily. The dose of azithromycin
Despite the use of antibiotics, we could still identify was 500 mg on the first day, and 250 mg daily thereafter
each subject based on their virome contents regardless (this dosing was used to be consistent with the
of the time point studied. These results are similar to a commonly prescribed Z-Pak). In the azithromycin arm,
prior study in which we noted that human oral viruses the placebo was given at 500 mg once daily. Each subject
were highly persistent over time [12]. The use of rela- enrolled donated saliva and feces on day 0 (day prior to
tively narrow spectrum oral antibiotics such as amoxicil- antibiotics), day 3 (3 days after initiation of antibiotics),
lin and azithromycin may have contributed to the day 7, week 8, and month 6. Of the eight households en-
stability observed in virome contents, as we observed rolled, one of those households was lost to follow-up and
much greater turnover in virome contents when broad did not provide specimens at the month 6 time point.
spectrum intravenous antibiotics were used in a separate Each subject provided at least 3 ml of unstimulated saliva.
cohort [10]. We believe that there is a relatively stable All subjects were encouraged to provide specimens in the
core of viruses that inhabit an individual’s microbiome, AM prior to breakfast and freeze them at −20 °C prior to
and that there are other viruses more susceptible to transporting on ice to the study site, where they were fro-
being replaced as individuals are exposed to perturba- zen at −80 °C until use in this study. Exclusion criteria in-
tions such as viruses transmitted from their close con- cluded prior antibiotic use for 1 year prior to the initiation
tacts. Further studies are necessary to demonstrate such of the study, and preexisting medical conditions such as
a phenomenon. diabetes, inflammatory bowel disease, and organ trans-
plantation that might result in significant immunosup-
Conclusions pression. All subjects self-reported their health status and
There now are quite a few studies characterizing the were genetically unrelated.
viruses that inhabit human body surfaces. Those studies
have demonstrated that viral communities are highly Analysis of viromes
diverse, carry substantial pathogenic gene functions, are Fecal viromes were prepared by diluting 0.4 g of feces in
populated by viruses that persist over time, and have 4 mL of SM buffer, vortexing for 40 min to separate viral
individual- and body surface-specific ecology [1, 3–6, 12, particles, spun at ×4000g for 10 min to pellet the
20]. Our data indicate that viral communities in the remaining solid material, and the supernatant was
mouth and gut are readily shared within a household, treated in an identical manner to saliva specimens. Saliva
likely through transmissions from one subject to an- samples and fecal supernatants were filtered sequentially
other. While it is not clear whether most of the sharing using 0.45 and 0.2 μm cellulose acetate filters (GE
is from direct viral transmissions rather than lysogen Healthcare Life Sciences) to remove cellular and other
(bacteria with un-induced prophage) transmissions, the debris, and then purified on a cesium chloride gradient
sharing results in viromes formed through interactions according to previously described protocols [2]. Only the
with individuals and their environments. We do not yet fraction with a density corresponding to most known
fully understand the extent to which bacteriophages in bacteriophages [21] was retained, further purified on
humans may help to shape the ecology of the bacteria in Amicon YM-100 protein purification columns (Milli-
our microbiomes, but the data presented here suggest pore, Inc.), treated with two units of DNase I, and sub-
that our viruses not only have the capacity to shape the jected to lysis and DNA purification using the Qiagen
natural history of our microbiomes on multiple body UltraSens Virus kit (Qiagen). Recovered DNA was
surfaces, but also have the potential to shape the natural screened for the presence of contaminating bacterial nu-
history of the microbiomes of our close contacts. cleic acids by quantitative 16S rRNA gene PCR using
primers 8 F (AGAGTTTGATCCTGGCTCAG) and 357R
Methods (CTGCTGCCTYCCGTA) in Power SYBR Green PCR
Cohort design Mastermix (Thermo Fisher Scientific) [10]. No products
Sixteen subjects were enrolled in the study in pairs, with were detected in any of the viromes after 35 cycles, which
two individuals living in each household. An additional does not exclude the presence of contaminating bacterial
four individuals were enrolled without a housemate and nucleic acids, but indicates that they were not present at
received no therapy over the course of the 6-month dominant levels. Viral DNA then was amplified using
study. Of the household pairs, four pairs were placed GenomiPhi Hy MDA amplification (GE Healthcare Life
into the amoxicillin arm and four pairs were placed into Sciences), fragmented to roughly 200 to 400 bp using a
the azithromycin arm. In each household, one subject Bioruptor (Diagenode), and utilized as input to create
Ly et al. Microbiome (2016) 4:64 Page 8 of 10

libraries using the Ion Plus Fragment Library Kit (Thermo Analysis of shared sequence similarities present in
Fisher Scientific) according to manufacturer’s instructions. each virome was performed by creating custom BLAST
Libraries then were sequenced using 316 chips on an Ion databases for each virome, comparing each database
Torrent Personal Genome Machine (Thermo Fisher with all other viromes using BLASTN analysis (E-value
Scientific). We trimmed sequence reads according to <10−10), and these compiled data used to calculate
modified Phred scores of 0.5 using CLC Genomics Work- Sorensen distances using Ion Assist (http://www.thepri-
bench 8.5 (Qiagen), removed any low complexity reads delaboratory.org/). Sorensen distances are measured on
with ≥8 consecutive homopolymers, and removed any a scale of 0 to 1, where 0 represents no sharing and 1
reads with substantial length variation (<50 nucleotides or would represent identical viromes. These distances were
>300 nucleotides) or ambiguous characters prior to fur- determined for all pairs of housemates and compared
ther analysis. Each virome was screened for contaminating with distances between subjects from different house-
human nucleic acids using BLASTN analysis (E-value <10 holds. Statistical significance was determined by the
−5
) against the human reference database available at ftp:// Mann–Whitney U test using MaxStat Pro (http://
ftp.ncbi.nlm.nih.gov/genomes/H_sapiens/. Any reads with www.maxstat.de/). Bray Curtis distances (equivalent to 1
significant sequence similarities to human sequences were minus Sorensen values) also were determined and used
removed prior to further analysis using Ion Assist (http:// as input for principal coordinates analysis using QIIME
www.thepridelaboratory.org/). All reads were assembled [23]. We determined persistence of viruses by construct-
using CLC Genomics Workbench 8.5 (Qiagen) based on ing global assemblies from all contigs within a subject
98% identity with a minimum of 50% read overlap, which over time, and using the contribution of each time point
were more stringent than criteria developed to discrimin- to the assemblies to decipher the time points that con-
ate between highly related viruses [22]. Because the short- tributed to the construction of each virus, as has previ-
est reads were 50 nucleotides, the minimum tolerable ously been described [12]. We utilized this technique to
overlap was 25 nucleotides, and the average overlap was decipher those contigs that were unique to a subject
no less than 100 nucleotides depending on the character- within a household and those shared between house-
istics of each virome. The consensus sequence for each mates. We also created global assemblies from both sub-
contig was constructed according to majority rule, and jects within a household to determine the presence/
any contigs <200 nucleotides or with ambiguous charac- absence of viruses at each subject and time point and
ters were removed prior to further analysis. Contigs were identify viruses that may have been transmitted between
annotated using BLASTX against the NCBI NR database housemates. We defined any virus that was present in one
with an E-value cutoff value of 10−5. Specific viral subject, absent in their housemate, and later appeared in
sequences were identified using Ion Assist (http:// at least two time points in the housemate as a putative
www.thepridelaboratory.org/) by parsing BLASTX results transmission. Statistical significance was determined by
for known viral genes including replication, structural, comparisons between groups by the Mann–Whitney U
transposition, restriction/modification, hypothetical, and test using MaxStat Pro (http://www.maxstat.de/).
other genes previously found in viruses for which the E- To assess whether viromes had significant overlap
value was at least 10−5. Each individual virome contig was within or between subjects, we performed a permutation
annotated using this technique; however, if the best hit for test using Ion Assist (http://www.thepridelaboratory.org/
any portion of the contig was to a gene with no known ) based on resampling (10,000 iterations) [13]. We simu-
function, lower level hits were used as long as they had lated the distribution of the fraction of shared virome
known function and still met the E-value cutoff. The se- homologues from two different time points within indi-
quences of contig71 were generated by PCR amplification vidual subjects that were randomly chosen across all
of overlapping fragments using Platinum PCR SuperMix time points. For each set, we computed the summed
(Thermo Fisher Scientific) with specific primers (Add- fraction of shared homologues using 1000 random con-
itional file 1: Table S3). Each resulting amplicon was in tigs between randomly chosen individual time points
both directions using Sanger sequencing. Contigs were as- within different subjects, and from these computed an
sembled interactively using Sequencher (Gene Codes empirical null distribution of our statistic of interest (the
Corp), ORFs predicted using FGenesV (Softberry Inc, fraction of shared homologues). The simulated statistics
Mount Kisco, NY), and ORF putative functions assigned within each subject across all time points were referred
by BLASTP homology (Escore <10−5). If the best hit was to the null distribution of inter-subject comparisons, and
to a gene with no known function, lower level hits were the p value was computed as the fraction of times the
used for the annotation as long as they had known puta- simulated statistic for the each exceeded the observed
tive function and still met the E-score cutoff (10−5). Poly- statistic. We utilized a similar technique to determine
morphisms were identified using day 0 as the index whether there was significant overlap between the
sequence. viromes of the different households.
Ly et al. Microbiome (2016) 4:64 Page 9 of 10

Additional files University of California, San Diego 92093, USA. 4Bioharmony Therapeutics Inc,
New York, NY 10065, USA.
Additional file 1: Table S1. Study Subjects. Table S2. Viral homologues
Received: 9 July 2016 Accepted: 24 November 2016
within and between households. Table S3. Primer sequences used in
this study. (PDF 795 kb)
Additional file 2: Figure S1. Principal coordinates analysis of beta
diversity present in the saliva of all households. Figure S2. Principal References
coordinates analysis of beta diversity present in the saliva of all subjects. 1. Hannigan GD, Meisel JS, Tyldsley AS, Zheng Q, Hodkinson BP, SanMiguel AJ,
Figure S3. Bar graph (± standard error) demonstrating the relative time Minot S, Bushman FD, Grice EA. The human skin double-stranded dna
intervals between the earliest and latest time point that formed each virome: topographical and temporal diversity, genetic enrichment, and
fecal viral contig. Figure S4. Bar graph (± standard error) demonstrating dynamic associations with the host microbiome. MBio. 2015;6:e01578-15.
the relative time intervals between the earliest and latest time point that 2. Pride DT, Salzman J, Haynes M, Rohwer F, Davis-Long C, White 3rd RA,
formed each salivary viral contig. Figure S5. Bar graphs representing the Loomer P, Armitage GC, Relman DA. Evidence of a robust resident
relative proportions of viromes that are unique or shared within each bacteriophage population revealed through analysis of the human salivary
household. Figure S6. Bar graphs representing the mean proportions of virome. ISME J. 2012;6:915–26.
viromes (± standard error) that are unique or shared within a household. 3. Minot S, Sinha R, Chen J, Li H, Keilbaugh SA, Wu GD, Lewis JD, Bushman FD.
Figure S7. Assembly of contig 39 from the feces in all time points from The human gut virome: inter-individual variation and dynamic response to
the subjects in household 1. Figure S8. Assembly of contig 142 from the diet. Genome Res. 2011;21:1616–25.
feces in all time points from the subjects in household 2. Figure S9. 4. Santiago-Rodriguez TM, Ly M, Bonilla N, Pride DT. The human urine
Assembly of contig 41 from the feces in all time points from the subjects virome in association with urinary tract infections. Front Microbiol.
in household 3. Figure S10. Assembly of contig 44 from the feces in all 2015;6:14.
time points from the subjects in household 6. Figure S11. Assembly of 5. Willner D, Furlan M, Schmieder R, Grasis JA, Pride DT, Relman DA, Angly FE,
contig 71 from the feces in all time points from the subjects in McDole T, Mariella Jr RP, Rohwer F, Haynes M. Metagenomic detection of
household 7. Figure S12. Bar graph representing the mean proportions phage-encoded platelet-binding factors in the human oral cavity. Proc Natl
of salivary viromes (± standard error) shared between housemates and Acad Sci U S A. 2011;108 Suppl 1:4547–53.
putative transmissions. 6. Ly M, Abeles SR, Boehm TK, Robles-Sikisaka R, Naidu M, Santiago-Rodriguez
T, Pride DT. Altered oral viral ecology in association with periodontal
disease. MBio. 2014;5:e01133–01114.
Abbreviations 7. Norman JM, Handley SA, Baldridge MT, Droit L, Liu CY, Keller BC, Kambal A,
PCOA: Principal coordinates analysis Monaco CL, Zhao G, Fleshner P, et al. Disease-specific alterations in the
enteric virome in inflammatory bowel disease. Cell. 2015;160:447–60.
Acknowledgements 8. Lewis JD, Chen EZ, Baldassano RN, Otley AR, Griffiths AM, Lee D, Bittinger K,
Not applicable. Bailey A, Friedman ES, Hoffmann C, et al. Inflammation, antibiotics, and diet
as environmental stressors of the gut microbiome in pediatric Crohn’s
Funding disease. Cell Host Microbe. 2015;18:489–500.
This study was supported by the Burroughs Wellcome Fund and the Doris 9. Modi SR, Lee HH, Spina CS, Collins JJ. Antibiotic treatment expands the
Duke Charitable Foundation to DTP. resistance reservoir and ecological network of the phage metagenome.
Nature. 2013;499:219–22.
Availability of data and materials 10. Abeles SR, Ly M, Santiago-Rodriguez TM, Pride DT. Effects of long term
All sequences are available for download in the Sequence Read Archive antibiotic therapy on human oral and fecal viromes. PLoS One. 2015;10:
under accession number SRP077685 (https://www.ncbi.nlm.nih.gov/sra/ e0134941.
?term=SRP077685). Contig71 sequences are deposited in GenBank (https:// 11. Enault F, Briet A, Bouteille L, Roux S, Sullivan MB, Petit MA. Phages rarely
www.ncbi.nlm.nih.gov/genbank/) under accession numbers KX501134- encode antibiotic resistance genes: a cautionary tale for virome analyses.
KX501139. Ion Assist can be found at http://www.thepridelaboratory.org/ ISME J. 2016;10. doi:10.1038/ismej.2016.90. [Epub ahead of print]
and runs on Windows XP or higher. 12. Abeles SR, Robles-Sikisaka R, Ly M, Lum AG, Salzman J, Boehm TK, Pride DT.
Human oral viruses are personal, persistent and gender-consistent. ISME J.
Authors’ contributions 2014;8:1753–67.
SRA, MBJ, CG, and DTP conceived and designed the experiments. ML, TSR, 13. Robles-Sikisaka R, Ly M, Boehm T, Naidu M, Salzman J, Pride DT. Association
JG, and ICC performed the experiments. DTP analyzed the data. SRA and between living environment and human oral viral ecology. ISME J. 2013;
DTP contributed the reagents, performed the examinations, and recruited 7(9): 1710-24.
the subjects. MBJ, CG, and DTP wrote the manuscript. All authors read and 14. Rothberg JM, Hinz W, Rearick TM, Schultz J, Mileski W, Davey M, Leamon JH,
approved the final manuscript. Johnson K, Milgrew MJ, Edwards M, et al. An integrated semiconductor
device enabling non-optical genome sequencing. Nature. 2011;475:348–52.
Competing interests 15. Dutilh BE, Cassman N, McNair K, Sanchez SE, Silva GG, Boling L, Barr JJ,
The authors declare that they have no competing interests. Speth DR, Seguritan V, Aziz RK, et al. A highly abundant bacteriophage
discovered in the unknown sequences of human faecal metagenomes. Nat
Consent for publication Commun. 2014;5:4498.
The consent forms signed by each participant included their consent to 16. Song SJ, Lauber C, Costello EK, Lozupone CA, Humphrey G, Berg-Lyons D,
allow us to publish our findings. Caporaso JG, Knights D, Clemente JC, Nakielny S, et al. Cohabiting family
members share microbiota with one another and with their dogs. Elife.
Ethics approval and consent to participate 2013;2:e00458.
Subject recruitment and enrollment were approved by the University of 17. Lax S, Smith DP, Hampton-Marcell J, Owens SM, Handley KM, Scott NM,
California, San Diego Human Research Protection Program under project Gibbons SM, Larsen P, Shogan BD, Weiss S, et al. Longitudinal analysis of
#130807. All subjects signed written informed consent indicating their microbial interaction between humans and the indoor environment.
willingness to participate in this study and willingness to allow us to publish Science. 2014;345:1048–52.
our findings. Each subject was compensated for their continued participation 18. Kleiner M, Hooper LV, Duerkop BA. Evaluation of methods to purify virus-like
in this longitudinal study. particles for metagenomic sequencing of intestinal viromes. BMC Genomics.
2015;16:7.
Author details 19. Abeles SR, Jones MB, Santiago-Rodriguez TM, Ly M, Klitgord N, Yooseph S,
1
Department of Pathology, University of California, San Diego 92093, USA. Nelson KE, Pride DT. Microbial diversity in individuals and their household
2
Human Longevity, Inc, San Diego, CA 92121, USA. 3Department of Medicine, contacts following typical antibiotic courses. Microbiome. 2016;4:39.
Ly et al. Microbiome (2016) 4:64 Page 10 of 10

20. Willner D, Furlan M, Haynes M, Schmieder R, Angly FE, Silva J, Tammadoni S,

Nosrat B, Conrad D, Rohwer F. Metagenomic analysis of respiratory tract
DNA viral communities in cystic fibrosis and non-cystic fibrosis individuals.
PLoS One. 2009;4:e7370.
21. Murphy FA, Fauquet CM, Bishop DHL, Ghabrial SA, Jarvis AW, Martelli GP,
Mayo MA, Summers MD. Virus taxonomy: sixth report of the International
Committee on Taxonomy of Viruses, vol. Supplement 10. New York:
Springer-Verlag; 1995.
22. Breitbart M, Salamon P, Andresen B, Mahaffy JM, Segall AM, Mead D, Azam
F, Rohwer F. Genomic analysis of uncultured marine viral communities. Proc
Natl Acad Sci U S A. 2002;99:14250–5.
23. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello
EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, et al. QIIME allows analysis of
high-throughput community sequencing data. Nat Meth. 2010;7:335–6.

Submit your next manuscript to BioMed Central

and we will help you at every step:
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit