METAGENÓMICA VIRAL Delwart E.L. 2007

Rev. Med. Virol. 2007; 17: 115–131.
Published online 12 February 2007 in Wiley InterScience
(www.interscience.wiley.com)
Reviews in Medical Virology DOI: 10.1002/rmv.532
Viral metagenomics
Eric L. Delwart*
Blood Systems Research Institute, University of California, San Francisco, CA, USA
SUMMARY
Characterisation of new viruses is often hindered by difficulties in amplifying them in cell culture, limited antigenic/
serological cross-reactivity or the lack of nucleic acid hybridisation to known viral sequences. Numerous molecular
methods have been used to genetically characterise new viruses without prior in vitro replication or the use of virusspecific
reagents. In the recent metagenomic studies viral particles from uncultured environmental and clinical
samples have been purified and their nucleic acids randomly amplified prior to subcloning and sequencing. Already
known and novel viruses were then identified by comparing their translated sequence to those of viral proteins in
public sequence databases. Metagenomic approaches to viral characterisation have been applied to seawater, near
shore sediments, faeces, serum, plasma and respiratory secretions and have broadened the range of known viral
diversity. Selection of samples with high viral loads, purification of viral particles, removal of cellular nucleic acids,
efficient sequence-independent amplification of viral RNA and DNA, recognisable sequence similarities to known
viral sequences and deep sampling of the nucleic acid populations through large scale sequencing can all improve the
yield of new viruses. This review lists some of the animal viruses recently identified using sequence-independent
methods, current laboratory and bioinformatics methods, together with their limitations and potential improvements.
Viral metagenomic approaches provide novel opportunities to generate an unbiased characterisation of the viral
populations in various organisms and environments. Copyright # 2007 John Wiley & Sons, Ltd.
Received: 13 November 2006; Revised: 21 December 2006; Accepted: 22 December 2006
INTRODUCTION
Classical methods used for the identification of
viruses include in vitro viral amplification followed
by electron microscopy and the use of reference
serum from previously infected or vaccinated
hosts. Cell culture combined with visual observation
for cytopathic effects, followed by testing for
immunological cross-reactivity using large panels
of sera, is a powerful and relatively rapid method
when the unknown viral agents replicate in the
particular cell lines used and cross-reactive
reagents are available. Tentative identification of
the virus then allows the use of more specific
reagents, particularly degenerate PCR primers,
targeting the likely viral group for definitive genetic
characterisation [1–11]. Strictly molecular methods
that do not require in vitro replication and
scarce serological or antigenic reagents have also
been developed and successfully used to allow
the characterisation of numerous novel animal
viruses. Some of these methods are described
below with special emphasis on sequence-independent
amplification and plasmid library sequencing
of nucleic acids in biological fluids or
environmental samples collectively referred to as
viral metagenomics. The diversity of bacteriophage
obtained from seawater, in which viral particles
can be found at levels as high as 1010 per
litre of surface seawater [12], in near-shore sediments
and in human faeces has been recently
reviewed by the pioneering group in the rapidly
growing field of viral metagenomics [13,14].
THE NEED FOR BETTER VIRAL
DISCOVERY TOOLS
The human virome
The emergence of previously unrecognised viruses
as a result of improved transmission opportunities
R E V I E W
Copyright # 2007 John Wiley & Sons, Ltd.
*Corresponding author: E. L. Delwart, Blood Systems Research
Institute, University of California, 270 Masonic Ave., San Francisco,
CA 94118, USA. E-mail: delwarte@medicine.uscf.edu
Abbreviations used
AP-PCR, Arbitrarily Primed PCR; EST, Expressed-Sequence Tag;
HEV, Hepatitis E Virus; LASL, Linker Amplified Shotgun Library;
NHP, Non-Human Primate; RACE, Rapid Amplification Of cDNA
Ends; RCA, Rolling Circle Amplification; RDA, Representational
Difference Analysis; SISPA, Sequence-Independent Single Primer
Amplification.
and/or adaptive mutations underlines the need
for a better characterisation of the full range of
viruses replicating in humans (i.e. the human virome)
[15,16]. Highly prevalent infections with
anelloviruses [17–21] and GBV-C [10] have been
shown in humans. The only recent identification
of these two highly common chronic human infections
using molecular methods hints at the possibility
of a wider human viral flora. While initially
thought to induce hepatitis these two diverse viral
groups are now thought to be largely commensal
[22–27].
Recently identified pathogenic viruses of apparently
strictly human origin include the Norwalk
norovirus from a gastroenteritis outbreak in
1991 [28]; the human herpesvirus 8 from cases of
Kaposi’s sarcoma in AIDS patients in 1994 [29];
the metapneumovirus from children with a wide
range of respiratory symptoms in 2001 [30]; a new
human coronavirus in 2004 [8,31–33] and a new
human parvovirus in 2005 [15]. Improving ecological
opportunities, seen in the large numbers of
immunocompromised AIDS patients, has increased
the incidence of pathological infections with viruses
such as HHV8 and JCV (human polyomavirus)
[34,35] and provided fertile grounds for virus
spread and evolution. Newly characterised human
viruses of unknown pathogenicity include another
recently identified parvovirus and new anellovirus
variants in the blood of febrile patients [36].
The animal viromes
Frequent viral epidemics in crowded domesticated
animal populations as well as in wild animals
have potential spill-over effects into human population
and emphasise the need for epidemic surveillance
in animals. HIV1 and HIV2 originate
from Chimpanzees and Sooty mangabey, respectively,
and are thought to have entered the human
population during the last century through hunting
and consumption of non-human primate
(NHP) [37–39]. Central African bush-hunters
have been shown to be infected with simian foamy
virus [40] as well as the STLV3 related retrovirus
HTLV3 and the newly characterised HTLV4 [41].
Some NHP workers in the U.S. show signs of
infection with simian foamy viruses [42,43] and
SV40 [44] and have been deferred as blood donors
in Canada [45]. Other viruses recently transmitted
from animals to humans include the SARS
coronavirus from civet cats whose infection may
originate from bats [46]; the West Nile virus in
North America from mosquitoes feeding on
infected birds [47]; H5 influenza throughout Asia
and H7 influenza in Holland from poultry handling
[48]; hantaviruses from rodents’ urine in the
S.W. region of the U.S. starting in 1993 [49]; the
Ebola virus from NHPs possibly infected from
bats in the Democratic Republic of Congo and
Sudan in the late 1990s [50,51]; monkeypox viruses
from African rodent pets in the Midwestern U.S. in
2003 [52], and the Nipah virus in humans and pigs
in Malaysia also possibly transmitted from
infected bats [53,54].
Emerging virus surveillance
The characterisation of viruses in highly exposed
populations such as injection drug users, Central
African bush-hunters, zoo and NHP facility workers
or highly susceptible populations such as AIDS
and immunocompromised transplant patients
using non-biased methods may be used as a surveillance
tool for the early detection of emerging
viral infections. A large scale viral metagenomic
analysis of the less exposed but more easily obtainable
voluntary blood donors as a means of virus
surveillance was proposed [16]. Molecular analyses
of samples from patients suffering from
symptoms of unknown aetiology with a possible
infectious origin have recently yielded new human
viruses and a large fraction of encephalitis [55],
hepatitis [56–58], gastrointestinal diseases, myocarditis
as well as some auto-immune diseases
may be associated with infections by yet unknown
viruses [59]. As multiple cancers are now recognised
to result from viral infections (i.e. HBV,
HCV, HPV, HIV, HTLV-1, EBV and HHV8) it is
also conceivable that yet uncharacterised viruses
are involved in other forms of tumorgenesis [60].
Federal and state programmes in the U.S., in the
European community as well as numerous other
countries, are active in collecting specimens from
both symptomatic as well as highly exposed
human populations and testing them for known
infectious agents [61]. Similar studies of domesticated
and wild animal populations may also identify
newly emerging and established viruses
with the potential to cause economic or environmental
problems and cross-over to exposed
humans [62].
116 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
MOLECULAR METHODS OF
VIRAL DISCOVERY
Methods based on specific nucleic acid
hybridisation and antigenic cross-reactivity
Microarrays spotted with viral sequence oligonucleotides
have been used to genetically charaterise
the SARS-CoV from cell culture supernatant
[63,64], and a novel retrovirus from frozen human
prostate tissue [60]. Microarrays provide a powerful
tool for viral discovery provided the new
viruses are sufficiently related to those already
known to permit specific hybridisation. Degenerate
PCR primers are based on conserved
sequences within viral groups and have an
impressive track record having identified numerous
macaque herpesviruses [1–5], GBV-C [10,11],
HCoV-NH [18], animal retroviruses [6,7] and
picornaviruses from seawater [9]. This approach
is limited by each viral group requiring the use
of different degenerate primer sets and the use of
highly degenerate primers for highly variable viral
groups. Expression libraries generated using
sequence-independent amplification methods can
also be generated and screened using seropositive
plasma (which resulted in the identification of
HCV [65] and GBV-C [66]) or screened using virus
enriched labelled nucleic acid probes (which
yielded the Borna Disease virus [67]). These
powerful library screening methods therefore
require the use of specific reagents in which the
antibody or the unknown virus is already known
to be present.
Subtractive hybridisation
A number of related methods termed subtractive
hybridisation or representational difference analysis
(RDA) involve DNA hybridisation between tester
nucleic acids from infected tissue and
uninfected driver nucleic acids. Viral nucleic acids
are selectively amplified through several rounds of
hybridisation and purification of un-hybridised
single stranded tester nucleic acid using hydroxapatite
chromatography [67,68] followed by subcloning.
Plasmids expressing the antigen of interest
may be identified by immunoreactivity with appropriate
serum antibodies or by sequence homology
to known viruses. RDA therefore requires infected
and non-infected tissues from the same individual
and following multiple cycles of hybridisation and
PCR reamplification can result in the preferential
amplification of sequences unique to the tester
samples [69]. RDA has led to the discovery of
HHV8 [29], TTV [17], and GBV-A and B [70]. Subtractive
hybridisation-based techniques require
large amounts of infected and uninfected materials
from the same person and are technically demanding
therefore limiting the number of samples that
can be analysed.
SEQUENCE-INDEPENDENT NUCLEIC
ACID AMPLIFICATION
The biological samples available and the suspected
nature of the virus, dictate the most appropriate
method for virus discovery. A conceptually related
set of methods relies on sequence-independent
amplification, subcloning and sequencing of purified
viral nucleic acids followed by in silico
searches for sequence similarities to known
viruses. When applied to environmentally collected
samples or unmanipulated biological samples
this approach has been labelled viral
metagenomics [13,14,71]. The term viral metagenomics
may also be loosely used to describe the
general approach of non-specific amplification
and sequencing of viral nucleic acids from cell culture
supernatants where the presence of an
unknown virus is suspected based on the appearance
of cellular cytopathic effects. The primary
advantages of sequence-independent amplification
and sequencing methods for characterising
novel viruses are simplicity and relative speed,
the lack of bias towards any particular viral group
or requirement for specific reagents and the ability
to detect new viruses that are highly divergent
from those already known through conserved protein
motifs. A limitation of sequence-independent
nucleic acid amplification methods for viral discovery
is their general unsuitability with samples
in which host nucleic acids and viral nucleic
acids cannot be easily separated, such as tissue
biopsies and PBMC, since the resulting fraction
of viral sequences relative to host nucleic acids
would be extremely low. For studies of cellular
samples the use of microarrays, degenerate primer
PCR or subtractive hybridisation may be more
appropriate. Sequence-independent nucleic acid
amplification methods are particularly useful for
the study of samples from which cells can be
easily filtered and residual host nucleic acids
removed by enzymatic digestion while viral genetic
material remains protected within viral capsids.
Viral metagenomics 117
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
Therefore, plasma, serum, respiratory secretions,
cerebrospinal fluid, urine, faeces or filterable
environmental samples are most appropriate for
viral metagenomic studies.
Sequence-independent single primer
amplification (SISPA)
SISPA was originally developed to amplify low
copy number nucleic acids for human genomics
applications [72]. SISPA is based on endonuclease
restriction of target DNA followed by ligation of
adaptor linker complementary to the overhanging
bases on the target DNA. RNA viruses can also be
amplified by SISPA following random primed
cDNA synthesis followed by dsDNA synthesis
(using the DNA polymerase activity of reverse
transcriptase following RNAse H digestion of the
RNA component of the RNA/DNA hybrid). A
PCR primer complementary to the ligated linker
is then used to amplify the sequences located
between pairs of restriction sites. SISPA, combined
with immunoscreening of expression clones, was
used to genetically characterise Norwalk virus
from faeces [28] and a human astrovirus from culture
supernatants [73]. Hepatitis E Virus (HEV)
was also cloned using this method, combined
with differential hybridisation with labelled
nucleic acids from infected and non-infected tissues
to identify viral subclones [74,75]. The genetic
characterisation of a new human coronavirus from
a culture supernatant was performed using two
different SISPA primers annealing to different
restriction sites followed by a second round of
PCR using the same primers with an extra 30 base
to limit amplification to a subset of the original
amplification products [32]. This and other modifications
of SISPA were recently reviewed [76,77].
DNAse-SISPA is a modification of SISPA where
plasma samples are first filtered to remove bacteria
and eukaryotic cell sized particles, and then treated
with DNAse 1 to remove contaminating
human and other naked DNA [78]. Remaining viral
nucleic acids protected within their capsids are
then extracted and DNA or RNA (following conversion
to dsDNA using random primers) amplified
by SISPA and the amplification products
subcloned [78]. Plasmid inserts are sequenced
and analysed for sequence similarities to known
viruses. This method was successful in identifying
new parvoviruses in bovine sera [78] and human
plasma [36].
Linker amplified shotgun library (LASL)
A related method has been to physically shear
dsDNA from purified viruses at random sites
(using HydroShear from Genomic Solutions,
Inc.), repair the ragged ends with T4 DNA polymerase
and T4 polynucleotide kinase, ligate a
defined sequence linker to the extremities and
use a primer complementary to the linker to PCR
amplify the DNA fragments prior to plasmid subcloning
[79,80]. By first purifying the viral RNA
and performing dsDNA synthesis, this method
was also recently applied to RNA viruses in seawater
[81]. LASL and SISPA are related methods
with nucleic acids non-specifically PCR amplified
from attached linkers ligated at random sites versus
endonuclease restriction sites.
Arbitrarily primed PCR (AP-PCR)
This simple method is typically used to analyse the
differences between complex genomes (such as
strains of Staphylococcus) or to detect differences
in mRNA expression profiles. It takes advantage
of the ability of arbitrarily designed PCR primers
to initiate PCR at many different sites in a complex
mixture of nucleic acids when annealed at very
low temperatures [82–85]. Using a single PCR primer
pair (whose exact sequence is arbitrary), the
first round of PCR is performed at 40 C, allowing
the primers to initiate PCR at many partially complementary
sites, followed by 40 more PCR cycles
performed at a more stringent annealing temperature
( 60 C). PCR products are then typically
analysed by denaturing gel electrophoresis to
detect differences in the band patterns, but can
also be directly subcloned for sequencing. This
method was used to clone a new human pneumovirus
from cell culture supernatant [30].
Random PCR amplification
This method is based on the theoretical amplification
of all nucleic acids present using PCR primers
with a random nucleotide sequence at their 30 end
(size 4–8 N) and a defined sequence at their 50 end.
For RNA viruses reverse transcription is first performed
with such a primer at a low annealing temperature
of 37 C to allow randomly primed cDNA
extensions. Another single round of extension
with the same primer is then performed following
denaturation of the cDNA/RNA hybrid, and primer
annealing at low temperature followed by
118 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
Klenow DNA polymerase extension. Then, using a
PCR primer complementary to the defined 50 of
the initial primer, 30–40 cycles of PCR are performed
at high annealing temperature. For DNA
target amplification two rounds of low temperature
annealed primer extension are performed
before random PCR using the defined sequence
primer. Random PCR is the method of choice to
amplify and label probes with fluorescent dyes
for microarray analysis [60,64,86]. Random PCR
has been used to characterise a new human parvovirus
and identified numerous known RNA and
DNA viruses from respiratory secretions [15]
(Table 1), to amplify both RNA and DNA viruses
from cell culture [87] and to identify a short DNA
sequence with no sequence similarity in Genbank
whose prevalence in plasma is higher in non-A-E
hepatitis than control patients [88].
PhiX29 DNA polymerase based
amplification
The genomics field is often limited by the amount
of starting DNA available. The properties of
PhiX29 polymerase make it possible to amplify
the entire human genome, starting from as little
as 10 cells, until 20–30 mg of DNA are isothermically
produced [89,90]. This method is based on
the ability of bacteriophage PhiX29 DNA polymerase
to efficiently displace an annealed DNA strand
in front of its advancing 30 end coupled with its
very long processivity (>70 000 bases) resulting
in multiple displacement amplification reactions
[89,90] (Figure 1). The DNA polymerase is primed
with modified random hexamer oliogonucleotides
(resistant to the 30–50 exonuclease activity of
PhiX29). When single DNA strands are generated
by PhiX29, they can themselves be used as templates
(Figure 1). This method achieves unbiased
amplification at every human locus analysed
[90,91] (Figure 1). The high proof reading ability
of PhiX29 DNA polymerase also reduces artifactual
mutations [92]. Use of PhiX29 DNA polymerase
based amplification for viral discovery has
been recently reported, successfully amplifying
circular DNA anellovirus [19,93].
Rolling circle amplification (RCA)
This method has successfully amplified numerous
circular DNA viral genomes. When the PhiX29
polymerase comes full circle on a circular viral
genome it displaces its 50 end and continues to
extend the new strand multiple times around the
DNA circle. Random primers can then anneal to
the displaced strand itself and convert it to double
stranded DNA [94]. The long dsDNA products can
then be cut with a restriction enzyme, expected to
cut once within the circle sequence, to release linear
fragments the length of the circle (Figure 2).
This technique, generally used to amplify plasmid
libraries [94,95], was recently used to amplify the
circular genomes of human papillomaviruses in a
cervical keratinocyte cell line, a fibropapillomatous
wart [96] and in a Florida manatee [97]. RCA
initiated with random primers and the DNA of
various organs has also yielded full genome
sequences of polyomaviruses [98,99], anellovirus
[100], circoviruses [101] geminiviruses [102], plant
begomovirus [103] and wasp polydnavirus [104].
Figure 1. Principle of multiple displacement amplification using
PhiX29 DNA polymerase [164]
Viral metagenomics 119
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
Computational subtraction
A strictly informatics-based approach involves
computational removal of known human
sequences from expressed-sequence tag (EST)
sequence data. Foreign sequences can then be analysed
for sequence similarities against known
viruses. Various viral sequences were identified
in EST libraries derived from normal and cancerous
tissues (HBV, HCMV, Human papillomaviruses
18þ16, HHV8, HCV, EBV and human
spumavirus) [105]. When a cDNA library from a
post-transplant lymphoproliferative disorder tissue
was similarly analysed, 10 EBV sequences
were identified among >27 000 human cDNA
sequences [106]. This approach relies on sequence
data gathered for other purposes as the yield of
viral sequences is very low due to the predominance
of human sequences.
PURIFICATION OF VIRAL NUCLEIC ACIDS
The key to metagenomic based viral discovery is
increasing the levels of viral nucleic acids while
reducing background prokaryotic and eukaryotic
nucleic acids. For environmental samples available
in large volume such as seawater, virus concentration
and reduction of background nucleic acids
from prokaryotic and eukaryotic cells can be performed.
Three complementary approaches for
viral nucleic acid purification prior to their amplification
have been used: filtration, density gradient
centrifugation and enzymatic removal of non-particle
protected, nucleic acids.
Filtration and density gradient ultracentrifugation
based virus purification
A large body of literature has reported on the
removal of viral particles during the manufacture
of blood derived biologicals using filter pore size
as small as 35nm [107,108]. When the purpose is
concentration of viral nucleic acids a filtration
pore size of 160–450nm was initially used to
remove bacteria, eukaryotes and large aggregates
[15,79]. Since the largest known virus is the mimivirus
at 400nm in diameter, a size comparable to
that of a small bacterium [109], such a filter size
represents a compromise expected to allow the
flow through of all but the largest viruses. Further
filtration to concentrate viruses was used when the
starting sample volume was large as with faeces
[110,111] or seawater [79,9,81]. Studies analysing
seawater DNA bacteriophages used tangential
flow filtration with a cutoff of Mr 100 000 to concentrate
viral particles prior to ultra-centrifugation
in a cesium chloride step gradient to collect the
1.35–1.5 g/ml DNA phage density fraction [79].
CsCl density gradient ultra-centrifugation was
also used directly on plasma pools for the purification
of human DNA viruses [93]. Ultrafiltration
using a tangential flow filter with a cutoff of
Mr 30 000 was also used to purify viruses from filtered
seawater followed by ultracentrifugation to
pellet viruses [9,81,112]. Following the initial large
particle removal by filtration, viral particles were
further purified from lower density material by
centrifugation through a 30% sucrose solution
and resuspension of the viral pellet [87]. The overall
strategy for large volume samples is therefore
an initial filtration to remove bacteria size particles,
followed by concentration on a small pore
size filter (Mr 30 000–100 000) to reduce volume,
followed by high speed centrifugation through
sucrose or cesium chloride gradients.
Enzymatic digestion of non-capsid
protected nucleic acids
Host DNA is readily detectable in plasma and
even more so in serum [113] where it might be
Figure 2. Principle of rolling circle amplification using PhiX29 DNA polymerase [96]
120 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
non-covalently bound to histones [114] and can
therefore be a major source of background DNA
when using a total nucleic acid amplification
approach. DNAse I treatment was reported by
Allander et al. [78] as a key factor for the removal
of host DNA in serum prior to SISPA amplification
of viral nucleic acids. DNAse I treatment is
thought to remove naked DNA through exonuclease
digestion while DNA within viral capsids
(and within the lipid bilayer in the case of enveloped
viruses) are shielded from enzymatic activity.
Similar treatment using RNAse A has been
used to remove accessible RNA in viral concentrates
from faeces [80,110] and seawater [81]. Subsequent
to DNAse and RNAse treatments the
capsid protected viral and other nucleic acids are
then purified, using guanidinium isocyanate protein
denaturation followed by nucleic acid binding
to silica (e.g. Qiagen viral RNA purification), or by
phenol/chloroform extraction followed by ethanol
or isopropanol precipitation and CTAB cationic
detergent extraction [115]. If the focus is restricted
to DNA or RNA viruses, the extracted nucleic
acids can itself be further digested with the appropriate
nuclease.
An additional step to reduce non-viral background
nucleic acids has been to treat CsCl
banded viruses with diluted chloroform to disrupt
mitochondrial membranes and expose their DNA
to enzymatic degradation, however, this may disrupt
the stability of some lipid enveloped viruses
[93].
BIOINFORMATICS
Software or web sites to generate contigs of overlapping
sequences with variable number of mismatches
(due to variants of the same viral
species), starting with hundreds or hundreds of
thousands of input sequences (if using pyrosequencing),
exist but require a high level of user
expertise [116–118]. Using computationally
demanding search algorithms such as tBLASTx
to detect low-level translated protein similarities
to known viral sequences is also time consuming
[119,120]. The criterion for classifying sequences
into virus-like sequences is also arbitrary. A
tBLASTx E score of <0.001 to a known viral
sequence has frequently been used to define a
sequence as being of viral origin [79–81, 110,
111], although others have used a more stringent
cutoff of E<10
5 [15].
A fundamental problem is how to detect the presence
of novel viral sequences when they are so
highly divergent from those currently in the databases
that sequence similarities are not readily
detected using tBLASTx. A large fraction of
sequences (5%–30%) derived from animal samples
by sequence-independent amplification methods
and an even greater fraction of sequences derived
from environmental samples show no significant
nucleotide and amino acid sequence similarities to
any sequence, including viruses, currently in
Genbank. The origin of these nucleic acids is therefore
of great interest as they potentially represent
novel and highly distinct viruses. Several
approaches may improve the identification of
highly divergent viral sequences. The search for
conserved proteinmotifs is expected to help identify
distantly related viral protein sequences since some
viral groupings such as positive strand RNA viruses
encode a number of recognisable core protein functions
[121]. Viral hallmark genes encode viral functions
that are found in widely diverse virus groups,
have only distant homologues in cells, and whose
origins may predate cellular life [122]. For example,
genes encoding jelly-roll capsid protein structures or
superfamily 3 helicase functions are found in both
large and small DNA and RNA viruses [122].
Searching among the annotation of weak similarity
matches for viral hallmark gene keywords could
also focus further amplification and sequencing
efforts to potential viral sequences. The use of substitution
matrices, used to quantify protein sequence
similarities, generated from viral rather than eukaryotic
and prokaryotic protein alignments, may also
improve the detection of very low-level similarity to
current viral sequences. The computational generation
of theoretical ancestral sequences to the
numerous extant viral groupings and their subsequent
use in sequence similarity searches may
also improve the identification of highly divergent
viral sequences since the genetic distances of new
viruses to their common ancestor with extant species
will be reduced. Further bioinformatics
improvement could also be based on searching for
particular RNA folds related to those frequently
found in some RNA viruses [123] and viroid
RNA[124]. In silico protein structure predictions
using linear nucleic acid sequence, although not
yet commonly feasible, may also improve the detection
of divergent viruses encoding conserved viral
protein structures.
Viral metagenomics 121
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
Methods that are independent of BLAST based
alignments or predicted RNA and protein structure
based similarities would also assist in discriminating
among unclassifiable sequences those
likely of viral versus bacterial, archaeal or eukaryotic
origins. Dinucleotide sequence analysis takes
advantage of compositional biases in a sequenceindependent
fashion to establish genomic signatures.
The biological pressures that influence genomic
composition have been shown to be predictive
of evolutionary divergence [125]. For example, CG
dinucleotide under-representation in vertebrates
relative to bacteria has been attributed to high
levels of CpG methylation [126,127]. Differences
in dinucleotide frequencies have been used to successfully
identify regions of horizontal gene transfer
among bacteria [128–130], discriminate exonic
and intronic regions of human sequences [131]
and differentiate bacterial, plasmid and phage
sequences [132]. Virus genomes persist under
unique pressures that affect their nucleotide composition.
Rapid rates of replication have been proposed
to decrease CG dinucleotides due to
unfavourable high thermodynamic stacking energies
[126] and antiviral host factors, such as APOBEC3G
cytosine deamination, cause G to A
mutation in HIV [133] and can mold the genomic
composition of viruses. Dinucleotide composition
is therefore another potential tool to help discriminate
viral from other sequences and select from
among unclassifiable sequences those with the
most viral-like dinucleotide composition for
further studies. Analysis of tetranucleotide frequencies
has also been used to discriminate among
sequences from different bacterial species,
although its use for the short contigs typically generated
from viral metagenomics may be problematic
[134].
A more laboratory-based approach to identify
highly divergent viral sequences will be to search
for closely related nucleic acids repeatedly found
in different animal or human samples. The detection
of closely related yet unclassifiable nucleic
acids in different individuals, particularly those
containing long open reading frames, may reflect
the presence of highly prevalent viruses. Identification
of such prevalent sequences may target
further shotgun sequencing or specific chromosome
walking to particular samples or sequences
in order to generate larger contiguous sequences
allowing even weaker sequence or structure similarities
to be detected. Transmissibility and
ongoing replication of unclassifiable nucleic acids
may also be determined by analysing blood transfusion
recipients for the appearance and maintenance
of unclassifiable nucleic acids found in
transfused blood.
FLANKING SEQUENCE WALKING FOR
ACQUISITION OF FULL VIRAL GENOME
When the purpose of a metagenomic analysis is
the description of complete or nearly complete
new viral genomes, a high frequency of viral
nucleic acids relative to other amplifiable nucleic
acids is required to generate large contigs of overlapping
sequences. Samples containing low viral
concentration yielding only a single viral-like
sequence may be further analysed by simply
increasing the sampling of the nucleic acid mixture
(i.e. sequencing more library subclones or using
novel technologies such as pyrosequencing) or by
improving viral particle purification. Sequencing
costs and limited sample availability (particularly
of rare clinical samples) may preclude such
approaches and require that initial sequence
matches to known viruses be extended into larger
sequences using sequence-specific extension methods.
If two or more subclones show significant
sequence similarity matches to different regions
of the same virus, the regions between them may
be linked simply using long distance specific PCR
[15,36,78]. When only a single subclone shows significant
similarity to a known virus, this genetic
‘foothold’ may serve as a region to hybridise a specific
primer and acquire further flanking sequence
data using 50 or 30 Rapid amplification of cDNA
ends (RACE). Other chromosome walking methods
rely on linear PCR amplifications using a single
specific primer bound to the initial foothold
region followed by low temperature annealing
with a single randomly chosen primer and PCR
[135] or first non-specifically binding primers and
extending them followed by the use of a specific
primer [136]. Ligation of an adaptor linker to the
ends of dsDNA followed by PCR using linker
and foothold specific primers [137] or based on
the formation of DNA circles from which inverse
PCR using specific primers [138,139] or RCA can
take place [140] may also be used to acquire flanking
viral sequences. Replica plating of randomly or
specifically generated plasmid Escherichia coli
libraries may also be probed by colony Southern
122 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
hybridisation using the initial viral match
sequence as labelled probe. This traditional colony
lift method or PCR screening will identify the E.
coli subclones that contain the initial virus matching
sequence and flanking regions thereby fraction
of the viral genomes. Plasmid libraries may also be
screened for specific inserts by self-ligation of
inverse PCR products derived from plasmid
libraries [141,142].
NOVEL VIRUSES IDENTIFIED IN
METAGENOMIC STUDIES OF
UN-MANIPULATED BIOLOGICAL SAMPLES
Human and animal samples
Based strictly on sequence similarities to known
viruses, a number of animal viruses have recently
been identified. Allander et al. initially characterised
two novel parvoviruses in bovine sera
using DNAse-SISPA [78]. Studies of human plasma
from febrile patients using DNAse-SISPA [36]
and of nasopharyngeal secretions from patients
with respiratory symptoms using random PCR
[15] identified already known DNA and RNA
viruses as well as two new human parvoviruses,
PARV4 and human Bocavirus (Table 1). Both
new parvoviruses have since been repeatedly
detected and in the case of HBoV shown to be a
common infant respiratory pathogen [143–151].
A study of viruses in the plasma of healthy adults
using PhiX29 DNA polymerase amplification and
LASL identified numerous diverse anelloviruses
as well as significant matches (tBLASTx <0.001)
to other potentially new viral sequences [93].
A metagenomic study of human faeces using LASL,
focusing on DNA viruses, identified numerous
dsDNA Siphophage [111] whose gram positive
bacterial hosts make up the majority of bacterial
cells in human feces [152]. A later study of human
feces using SISPA and targeting RNA viruses
identified a large number of plant viral pathogens,
the large majority of sequences belonging to pepper
mild mottle virus [110]. Analysis of equine
feces DNA virus using LASL indicated that over
60% of subclones showed no similarity to any Genbank
sequence and greater than half of the remaining
sequences were also related to Siphophages
[80]. Using RCA, numerous circular viral DNA
genomes have been characterised from the blood,
tissue and feces of mammals, birds, insects and
plants [96,97–104].
Environmental samples
The landmark paper by Breitbart et al. analysing
viral communities present in seawater started
with 200 litres of seawater [79,111]. Viruses were
first purified by differential filtration and step gradient
CsCl density ultracentrifugation. Plasmid
libraries were constructed using LASL. DNA
shearing and PCR amplification, rather than direct
subcloning, were used to disrupt the potentially
toxic viral genes and to remove modified bases
often present in bacteriophage DNA which cannot
be cloned directly into E. coli. Sixty five per cent of
sequences derived from these linker-amplified
shotgun libraries were not related to any sequence
in the database (tBLASTx E scores >0.001). Extrapolating
from <1000 sequenced subclones and
the number of sequences with overlap, the
number of identifiable viral sequences in seawater
was estimated to number between 300–7000 new
viral types depending on contig assembly stringency.
Thirty–forty per cent of the significant
tBLASTx hits (E score <0.001) were for phage
sequences, followed by repeat and mobile elements
and bacteria, archaea and eukarya
sequences based on Genbank annotations [79]
(Figure 3). A similar study of dsDNA viruses in
near shore sediments indicated much phylogenetic
Table 1. Results of sequence similarity
searches of plasmid libraries derived from
respiratory secretions
Category Library 1 (%) Library 2 (%)
Human 84 (24) 110 (36)
Bacterial 202 (59) 65 (21)
Phage 6 (2) 2 (1)
Unknown 22 (6) 33 (11)
Virus 29 (8) 99 (32)
Influenza A virus 18 0
Adenovirus 6 0
Respiratory 0 10
syncytial virus
Metapneumovirus 0 1
TT virus 2 0
Coronavirus 1 26
Parvovirus 2 62
Total 343 309
Categorisations of DNA and RNA viruses were based
on tBLASTx E score <10
5 [15].
Viral metagenomics 123
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
Figure 3. Composition of seawater LASL libraries from two locations based on sequence similarities. (A) Sequences with any hits to
Genbank (E score <10
3). (B) Distribution among biological entities. (C) Families of phages. (D) Type of mobile elements [79]
124 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
overlap with seawater bacteriophages and the presence
of at least 104 distinct genotypes per kilogram
of sediments [153]. Unlike the findings
from the same group using human faecal matter
[111], bacteriophages known to infect gram positive
bacteria were almost completely absent in
marine samples.
A recent study of seawater purified and concentrated
viral particles through a succession of filtering
steps to remove cellular organisms prior to
pelleting viruses by ultracentrifugation [81]. The
resuspended pellet was then treated with RNAse
prior to extraction of nucleic acids protected within
viral particles. Extracted nucleic acids were then
treated with DNAse to remove viral DNA. Viral
RNA was then reverse transcribed using random
hexamers as primers and dsDNA generated using
RNAse H and E. coli DNA polymerase. Double
stranded DNA was then treated in a manner similar
to LASL. This approach yielded 60%–80% of
sequences without sequence similarities in Genbank
and among those with tBLASTx <0.001,
98% were related to positive strand ssRNA
viruses. No RNA bacteriophages were detected
indicating that most marine bacteriophages have
DNA genomes and that most hosts of marine
RNA viruses may be eukaryotes. Sequences
resembling viruses known to infect higher plants
and insects were detected alongside new picorna-
like viruses whose dominance in the sea-water
viral population allowed their genomes to be completely
assembled [81].
FUTURE DIRECTIONS
Viral metagenomic studies of environmental and
animal samples appear poised for rapid growth
driven largely by improved viral particle purification
methods, the reduced cost of DNA
sequencing, rapidly growing viral sequence databases
and improved bioinformatics tools. The
small genome size of most viruses allows new
genomes to be assembled from limited shotgun
sequencing data [81] aided in some cases by specific
PCR amplification based on initial partial
viral genome data [15,36,78]. It is likely that
new technologies will rapidly impact the field
particularly multiplex sequencing methods such
as pyrosequencing [154] and polonies sequencing
[155] which, until now were largely restricted to
bacterial metagenomics [156] and the analysis of
samples with very low levels of highly degraded
DNA such as frozen mammoth and Neanderthal
bone [157,158].
The further development of software to detect
low-level sequence similarities will greatly aid
data analysis [119,120]. The use of multiplex
sequencing tools generating up to 300 000 short
sequence reads (100–200 bp) per experiment will
also necessitate improvement in the methodologies
used for searching for low-level protein
sequence or RNA structure similarities. The development
of virus specific BLOSUM matrices used
to measure similarities between distantly related
proteins and the use of predicted ancestral
sequences may also improve the detection of
highly divergent viruses as will the development
of methods analysing metagenomic data using
di- or tri-nucleotide sequence composition.
The development of molecular biology reagents
certifiably free of amplifiable nucleic acids, particularly
from bacteria, will also reduce the
background noise of sequence-independent amplification
methods [159-163]. For example, we have
detected murine leukemia virus reverse transcriptase
nucleic acids in commercial reverse transcriptase
enzyme preparations as well as sequences
belonging to the widely used agricultural used
viral insect pest control agent Autographa
californica nucleopolyhedrovirus in other protein
reagents likely reflecting commercial reagent contaminations.
Collection and analysis of appropriate samples
from clinical cases of diseases with possible unidentified
viral aetiology will be key to the rapid
identification of new viral pathogens using metagenomics.
Samples collected early relative to onset
of disease or during a febrile episode prior to specific
symptom onset might contain the highest viral
loads to facilitate virus identification. Samples
from highly exposed populations such as injection
drug users and infection susceptible groups such
as AIDS patients and immunosupressed transplant
recipients will also help define the human
virome. Surveillance for animal virus transmission
into human populations will be helped by the
study of exposed African bush-hunters and workers
exposed to NHPs and other animals [40–44].
The study of viruses in animals, both wild and
domesticated, where environmental changes or
crowded conditions may accelerate virus transmission
and evolution is also ripe for application of
viral metagenomics techniques.
Viral metagenomics 125
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
With increasing use of sequence-independent
amplification and efficient sequencing methodologies
it seems likely that new viral species will be
identified at a rate considerably greater than the
knowledge of their biology. Determining whether
newly identified viruses are pathogens, even in a
subset of infections, together with their mode of
transmission and replication strategies in host cells
may require large-scale epidemiological, as well as
animal and detailed virological studies. A significant
fraction of the ever-evolving cast of viruses
infecting humans and animals, both pathogenic
and commensal, may still remain uncharacterised.
Viral metagenomic analyses of appropriate human
and animal samples will assist in the genetic characterisation
of these viruses facilitating subsequent
studies of their pathogenicity and possible means
of control.
REFERENCES
1. Rose TM, et al. Identification of two homologs of the
Kaposi’s sarcoma-associated herpesvirus (human
herpesvirus 8) in retroperitoneal fibromatosis of
different macaque species. J Virol 1997; 71: 4138–
4144.
2. Strand K, et al. Two distinct lineages of macaque
gamma herpesviruses related to the Kaposi’s sarcoma
associated herpesvirus. J Clin Virol 2000; 16:
253–269.
3. van Devanter DR, et al. Detection and analysis of
diverse herpesviral species by consensus primer
PCR. J Clin Microbiol 1996; 34: 1666–1671.
4. Rose TM, Henikoff JG, Henikoff S. CODEHOP
(COnsensus-DEgenerate Hybrid Oligonucleotide
Primer) PCR primer design. Nucleic Acids Res
2003; 31: 3763–3766.
5. Rose TM, et al. Consensus-degenerate hybrid oligonucleotide
primers for amplification of distantly
related sequences. Nucleic Acids Res 1998; 26:
1628–1635.
6. Wilson CA, et al. Type C retrovirus released from
porcine primary peripheral blood mononuclear cells
infects human cells. J Virol 1998; 72: 3082– 3087.
7. Osterhaus AD, et al. Isolation and partial characterization
of a lentivirus from talapoin monkeys (Myopithecus
talapoin). Virology 1999; 260: 116–124.
8. Esper F, et al. Evidence of a novel human coronavirus
that is associated with respiratory tract disease
in infants and young children. J Infect Dis
2005; 191: 492–498.
9. Culley AI, Lang AS, Suttle CA. High diversity of
unknown picorna-like viruses in the sea. Nature
2003; 424: 1054–1057.
10. Simons JN, et al. Isolation of novel virus-like
sequences associated with human hepatitis. Nat
Med 1995; 1: 564–569.
11. Leary TP, et al. Consensus oligonucleotide primers
for the detection of GB virus C in human cryptogenic
hepatitis. J Virol Methods 1996; 56: 119–121.
12. Fuhrman JA. Marine viruses and their biogeochemical
and ecological effects. Nature 1999; 399: 541–
548.
13. Breitbart M, Rohwer F. Here a virus, there a virus,
everywhere the same virus? Trends Microbiol 2005;
13: 278–284.
14. Edwards RA, Rohwer F. Viral metagenomics. Nat
Rev Microbiol 2005; 3: 504–510.
15. Allander T, et al. Cloning of a human parvovirus by
molecular screening of respiratory tract samples.
Proc Natl Acad Sci USA 2005; 102: 12891–12896.
16. Anderson NG, Gerin JL, Anderson NL. Global
screening for human viral pathogens. Emerg Infect
Dis 2003; 9: 768–774.
17. Nishizawa T, et al. A novel DNA virus (TTV) associated
with elevated transaminase levels in posttransfusion
hepatitis of unknown etiology. Biochem
Biophys Res Commun 1997; 241: 92–97.
18. Leary TP, et al. Improved detection systems for TT
virus reveal high prevalence in humans, nonhuman
primates and farm animals. J Gen Virol
1999; 80: 2115–2120.
19. Tanaka Y, et al. Genomic and molecular evolutionary
analysis of a newly identified infectious agent
(SEN virus) and its relationship to the TT virus
family. J Infect Dis 2001; 183: 359–367.
20. Umemura T, et al. SEN virus infection and its relationship
to transfusion-associated hepatitis. Hepatology
2001; 33: 1303–1311.
21. Hino S, Miyata H. Torque teno virus (TTV): current
status. Rev Med Virol 2007; 17: 45–57.
22. Tangkijvanich P, et al. SEN virus infection and the
risk of hepatocellular carcinoma: a case-control
study. Am J Gastroenterol 2003; 98: 2500–2504.
23. Barin F. The virus isolated from patient TT (TTV):
still an orphan 2 years after its discovery. Transfus
Clin Biol 2000; 7: 79–83.
24. Thomas DL, et al. Persistence and clinical significance
of hepatitis G virus infections in injecting
drug users. J Infect Dis 1997; 176: 586–592.
25. Kanda T, et al. GB virus-C RNA in Japanese patients
with hepatocellular carcinoma and cirrhosis. J
Hepatol 1997; 27: 464–469.
26. Chams V, et al. Is GB virus C alias ‘hepatitis’ G virus
involved in human pathology? Transfus Clin Biol
2003; 10: 292–306.
27. Kao JH, et al. GB Virus C Infection in hemodialysis
patients: molecular evidence for nosocomial transmission.
J Infect Dis 1999; 180: 191–194.
126 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
28. Matsui SM, et al. The isolation and characterization
of a Norwalk virus-specific cDNA. J Clin Invest
1991; 87: 1456–1461.
29. Chang Y, et al. Identification of herpesvirus-like
DNA sequences in AIDS-associated Kaposi’s sarcoma.
Science 1994; 266: 1865–1869.
30. van den Hoogen BG, et al. A newly discovered
human pneumovirus isolated from young children
with respiratory tract disease. Nat Med 2001; 7: 719–
724.
31. Esper F, et al. Association between a novel human
coronavirus and kawasaki disease. J Infect Dis 2005;
191: 499–502.
32. van der Hoek L, et al. Identification of a new human
coronavirus. Nat Med 2004; 10: 368–373.
33. Fouchier RA, et al. A previously undescribed coronavirus
associated with respiratory disease in
humans. Proc Natl Acad Sci USA 2004; 101: 6212–
6216.
34. Elphick GF, et al. The human polyomavirus, JCV,
uses serotonin receptors to infect cells. Science
2004; 306: 1380–1383.
35. Berger JR, Major EO. Progressive multifocal
leukoencephalopathy. Semin Neurol 1999; 19: 193–
200.
36. Jones MS, et al. New DNA viruses identified in
patients with acute viral infection syndrome. J Virol
2005; 79: 8230–8236.
37. Keele BF, et al. Chimpanzee reservoirs of pandemic
and nonpandemic HIV-1. Science 2006; 313: 523–
526.
38. Chen Z, et al. Genetic characterization of new West
African simian immunodeficiency virus SIVsm:
geographic clustering of household-derived SIV
strains with human immunodeficiency virus type 2
subtypes and genetically diverse viruses from a single
feral sooty mangabey troop. J Virol 1996; 70:
3617–3627.
39. Marx PA, et al. Isolation of a simian immunodeficiency
virus related to human immunodeficiency
virus type 2 from a west African pet sooty mangabey.
J Virol 1991; 65: 4480–4485.
40. Wolfe ND, et al. Naturally acquired simian retrovirus
infections in central African hunters. Lancet
2004; 363: 932–937.
41. Wolfe ND, et al. Emergence of unique primate
T-lymphotropic viruses among central African
bushmeat hunters. Proc Natl Acad Sci USA 2005;
102: 7994–7999.
42. Heneine W, et al. Human infection with foamy
viruses. Curr Top Microbiol Immunol 2003; 277:
181–196.
43. Switzer WM, et al. Frequent simian foamy virus
infection in persons occupationally exposed to nonhuman
primates. J Virol 2004; 78: 2780–2789.
44. Engels EA, et al. Serologic evidence for exposure to
simian virus 40 in North American zoo workers. J
Infect Dis 2004; 190: 2065–2069.
45. Heneine W, Kuehnert MJ. Preserving blood safety
against emerging retroviruses. Transfusion 2006;
46: 1276–1278.
46. Lau SK, et al. Severe acute respiratory syndrome
coronavirus-like virus in Chinese horseshoe bats.
Proc Natl Acad Sci USA 2005; 102: 14040–14045.
47. Lanciotti RS, et al. Origin of the West Nile virus
responsible for an outbreak of encephalitis in the
northeastern United States. Science 1999; 286:
2333–2337.
48. Koopmans M, et al. Transmission of H7N7 avian
influenza A virus to human beings during a large
outbreak in commercial poultry farms in the
Netherlands. Lancet 2004; 363: 587–593.
49. Nichol ST, et al. Genetic identification of a hantavirus
associated with an outbreak of acute respiratory
illness. Science 1993; 262: 914–917.
50. Leroy EM, et al. Fruit bats as reservoirs of Ebola
virus. Nature 2005; 438: 575–576.
51. Leroy EM, et al. Multiple ebola virus transmission
events and rapid decline of central African wildlife.
Science 2004; 303: 387–390.
52. Anderson MG, et al. A case of severe monkeypox
virus disease in an American child: emerging infections
and changing professional values. Pediatr Infect Dis
J 2003; 22: 1093–1096 (discussion 1096–1098).
53. Yob JM, et al. Nipah virus infection in bats (order
Chiroptera) in peninsular Malaysia. Emerg Infect
Dis 2001; 7: 439–441.
54. Wong S, et al. Bats as a continuing source of emerging
infections in humans. Rev Med Virol 2006; 17:
31–55.
55. Glaser CA, et al. In search of encephalitis etiologies:
diagnostic challenges in the California Encephalitis
Project, 1998–2000. Clin Infect Dis 2003; 36: 731–742.
56. El Gaafary MM, et al. Surveillance of acute hepatitis
C in Cairo, Egypt. J Med Virol 2005; 76: 520–525.
57. Desai SM, et al. Prevalence of TT virus infection in
US blood donors and populations at risk for acquiring
parenterally transmitted viruses. J Infect Dis
1999; 179: 1242–1244.
58. He Z, et al. Retrospective analysis of non-A-E hepatitis:
possible role of hepatitis B and C virus infection.
J Med Virol 2003; 69: 59–65.
59. Hajjeh RA, et al. Surveillance for unexplained
deaths and critical illnesses due to possibly infectious
causes, United States, 1995–1998. Emerg Infect
Dis 2002; 8: 145–153.
60. Urisman A, et al. Identification of a novel gammaretrovirus
in prostate tumors of patients homozygous
for R462Q RNASEL variant. PLoS Pathog
2006; 2: p.e25.
Viral metagenomics 127
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
61. King DA, et al. Epidemiology. Infectious diseases:
preparing for the future. Science 2006; 313: 1392–
1393.
62. Daszak P, Cunningham AA, Hyatt AD. Emerging
infectious diseases of wildlife-threats to biodiversity
and human health. Science 2000; 287: 443–449.
63. Wang D, et al. Microarray-based detection and genotyping
of viral pathogens. Proc Natl Acad Sci USA
2002; 99: 15687–15692.
64. Wang D, et al. Viral discovery and sequence recovery
using DNA microarrays. PLoS Biol 2003; 1: p.E2.
65. Choo QL, et al. Isolation of a cDNA clone derived
from a blood borne non-A, non-B viral hepatitis
genome. Science 1989; 244: 359–361.
66. Linnen J, et al. Molecular cloning and disease
association of hepatitis G virus: a transfusiontransmissible
agent. Science 1996; 271: 505–508.
67. VandeWoude S, et al. A borna virus cDNA encoding
a protein recognized by antibodies in humans
with behavioral diseases. Science 1990; 250: 1278–
1281.
68. Lipkin WI, et al. Isolation and characterization of
Borna disease agent cDNA clones. Proc Natl Acad
Sci USA 1990; 87: 4184–4188.
69. Lisitsyn, N, Wigler M. Cloning the differences
between two complex genomes. Science 1993; 259:
946–951.
70. Simons JN, et al. Identification of two flavivirus-like
genomes in the GB hepatitis agent. Proc Natl Acad
Sci USA 1995; 92: 3401–3405.
71. Riesenfeld CS, Schloss PD, Handelsman J. Metagenomics:
genomic analysis of microbial communities.
Annu Rev Genet 2004; 38: 525–552.
72. Reyes GR, Kim JP. Sequence-independent, singleprimer
amplification (SISPA) of complex DNA
populations. Mol Cell Probes 1991; 5: 473–481.
73. Matsui SM, et al. Cloning and characterization of
human astrovirus immunoreactive epitopes. J Virol
1993; 67: 1712–1715.
74. Reyes GR, et al. Isolation of a cDNA from the virus
responsible for enterically transmitted non-A, non-
B hepatitis. Science 1990; 247: 1335–1339.
75. Reyes GR, et al. Hepatitis E virus (HEV): the novel
agent responsible for enterically transmitted non-
A, non-B hepatitis. Gastroenterol Jpn 1991; 26: 142–
147.
76. Ambrose HE, Clewley JP. Virus discovery by
sequence-independent genome amplification. Rev
Med Virol 2006; 16: 365–383.
77. Jarrett RF, et al. Molecular methods for virus discovery.
Dev Biol (Basel) 2006; 123: 77–88 (discussion
119–132.)
78. Allander T, et al. A virus discovery method incorporating
DNase treatment and its application to
the identification of two bovine parvovirus species.
Proc Natl Acad Sci USA 2001; 98: 11609–11614.
79. Breitbart M, et al. Genomic analysis of uncultured
marine viral communities. Proc Natl Acad Sci USA
2002; 99: 14250–14255.
80. Cann AJ, Fandrich SE, Heaphy S. Analysis of the
virus population present in equine faeces indicates
the presence of hundreds of uncharacterized virus
genomes. Virus Genes 2005; 30: 151–156.
81. Culley AI, Lang AS, Suttle CA. Metagenomic analyis
of coastal RNA virus communities. Science
2006; 312: 1795–1798.
82. Welsh J, McClelland M. Fingerprinting genomes
using PCR with arbitrary primers. Nucleic Acids
Res 1990; 18: 7213–7218.
83. Welsh J, McClelland M. Genomic fingerprinting
using arbitrarily primed PCR and a matrix of pairwise
combinations of primers. Nucleic Acids Res
1991; 19: 5275–5279.
84. McClelland M, et al. Arbitrary primed PCR fingerprinting
of RNA applied to mapping differentially
expressed genes. Exs 1993; 67: 103–115.
85. McClelland M, et al. Arbitrarily primed PCR fingerprints
resolved on SSCP gels. Nucleic Acids Res 1994;
22: 1770–1771.
86. Bohlander SK, et al. A method for the rapid
sequence-independent amplification of microdissected
chromosomal material. Genomics 1992; 13:
1322–1324.
87. Stang A, et al. Characterization of virus isolates by
particle-associated nucleic acid PCR. J Clin Microbiol
2005; 43: 716–720.
88. Yeh CT, et al. Identification of a novel singlestranded
DNA fragment associated with human
hepatitis. J Infect Dis 2006; 193: 1089–1097.
89. Dean FB, et al. Comprehensive human genome
amplification using multiple displacement amplification.
Proc Natl Acad Sci USA 2002; 99: 5261–5266.
90. Hosono S, et al. Unbiased whole-genome amplification
directly from clinical samples. Genome Res
2003; 13: 954–964.
91. Luthra R, Medeiros LJ. Isothermal multiple displacement
amplification: a highly reliable approach
for generating unlimited high molecular weight
genomic DNA from clinical specimens. J Mol Diagn
2004; 6: 236–242.
92. Esteban JA, Salas M, Blanco L. Fidelity of phi 29
DNA polymerase. Comparison between proteinprimed
initiation and DNA polymerization. J Biol
Chem 1993; 268: 2719–2726.
93. Breitbart M, Rohwer F. Method for discovering
novel DNA viruses in blood using viral particle
selection and shotgun sequencing. Biotechniques
2005; 39: 729–736.
128 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
94. Dean FB, et al. Rapid amplification of plasmid and
phage DNA using Phi 29 DNA polymerase and
multiply-primed rolling circle amplification. Genome
Res 2001; 11: 1095–1099.
95. Detter JC, et al. Isothermal strand-displacement
amplification applications for high-throughput
genomics. Genomics 2002; 80: 691–698.
96. Rector A, Tachezy R, van Ranst M. A sequenceindependent
strategy for detection and cloning of
circular DNA virus genomes by using multiply
primed rolling-circle amplification. J Virol 2004;
78: 4993–4998.
97. Rector A, et al. Characterization of a novel close-toroot
papillomavirus from a florida manatee by
using multiply primed rolling-circle amplification:
trichechus manatus latirostris papillomavirus
type 1. J Virol 2004; 78: 12698–12702.
98. Johne R, et al. Characterization of two novel polyomaviruses
of birds by using multiply primed rolling-
circle amplification of their genomes. J Virol
2006; 80: 3523–3531.
99. Johne R, et al. Novel polyomavirus detected in the
feces of a chimpanzee by nested broad-spectrum
PCR. J Virol 2005; 79: 3883–3887.
100. Niel C, Diniz-Mendes L, Devalle S. Rolling-circle
amplification of Torque teno virus (TTV) complete
genomes from human and swine sera and identification
of a novel swine TTV genogroup. J Gen Virol
2005; 86: 1343–1347.
101. Johne R, et al. Genome of a novel circovirus of starlings,
amplified by multiply primed rolling-circle
amplification. J Gen Virol 2006; 87: 1189–1195.
102. Haible D, Kober S, Jeske H. Rolling circle amplification
revolutionizes diagnosis and genomics of
geminiviruses. J Virol Methods 2006; 135: 9–16.
103. Inoue-Nagata AK, et al. A simple method for cloning
the complete begomovirus genome using the
bacteriophage phi29 DNA polymerase. J Virol
Methods 2004; 116: 209–211.
104. Espagne E, et al. Genome sequence of a polydnavirus:
insights into symbiotic virus evolution.
Science 2004; 306: 286–289.
105. Weber G, et al. Identification of foreign gene
sequences by transcript filtering against the human
genome. Nat Genet 2002; 30: 141–142.
106. Xu Y, et al. Pathogen discovery from human tissue
by sequence-based computational subtraction.
Genomics 2003; 81: 329–335.
107. Burnouf T, Radosevich M. Nanofiltration of
plasma-derived biopharmaceutical products. Haemophilia
2003; 9: 24–37.
108. Burnouf T, et al. Nanofiltration of single plasma
donations: feasibility study. Vox Sang 2003; 84:
111–119.
109. La Scola B, et al. A giant virus in amoebae. Science
2003; 299: p. 2033.
110. Zhang T, et al. RNA viral community in human
feces: prevalence of plant pathogenic viruses. PLoS
Biol 2005; 4: p. e3.
111. Breitbart M, et al. Metagenomic analyses of an
uncultured viral community from human feces. J
Bacteriol 2003; 185: 6220–6223.
112. Suttle CA, Chan AM, Cottrell MT. Use of ultrafiltration
to isolate viruses from seawater which are
pathogens of marine phytoplankton. Appl Environ
Microbiol 1991; 57: 721–726.
113. Lee TH, et al. Quantitation of genomic DNA in
plasma and serum samples: higher concentrations
of genomic DNA found in serum than in plasma.
Transfusion 2001; 41: 276–282.
114. Rumore PM, Steinman CR. Endogenous circulating
DNA in systemic lupus erythematosus. Occurrence
as multimeric complexes bound to histone. J Clin
Invest 1990; 86: 69–74.
115. Sambrook J, Fritsch EF, Maniatis T. Molecular
Cloning: A Laboratory Manual. Cold Spring
Harbor Laboratory Press; Cold Spring Harbor:
1989.
116. Pop M, Kosack D. Using the TIGR assembler in
shotgun sequencing projects. Methods Mol Biol
2004; 255: 279–294.
117. Huang X, Madan A. CAP3: a DNA sequence
assembly program. Genome Res 1999; 9: 868–877.
118. Green P. PHRAP in http://bozeman.mbt.washington.
edu/phrap.docs/phrap.html. 1996.
119. Zhang Z, et al. Protein sequence similarity searches
using patterns as seeds. Nucleic Acids Res 1998; 26:
3986–3990.
120. Altschul SF, et al. Gapped BLAST and PSI-BLAST: a
new generation of protein database search programs.
Nucleic Acids Res 1997; 25: 3389–3402.
121. Koonin EV, Dolja VV. Evolution and taxonomy of
positive-strand RNA viruses: implications of comparative
analysis of amino acid sequences. Crit
Rev Biochem Mol Biol 1993; 28: 375–430.
122. Koonin EV, Senkevich TG, Dolja VV. The ancient
virus world and evolution of cells. Biol Direct
2006; 1: p.29.
123. Simmonds P, Tuplin A, Evans DJ. Detection of genome-
scale ordered RNA structure (GORS) in genomes
of positive-stranded RNA viruses: Implications
for virus evolution and host persistence. Rna 2004;
10: 1337–1351.
124. Diener TO. Discovering viroids—a personal perspective.
Nat Rev Microbiol 2003; 1: 75–80.
125. Karlin S, Ladunga I. Comparisons of eukaryotic
genomic sequences. Proc Natl Acad Sci USA 1994;
91: 12832–12836.
Viral metagenomics 129
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
126. Karlin S, Burge C. Dinucleotide relative abundance
extremes: a genomic signature. Trends Genet 1995;
11: 283–290.
127. Gentles AJ, Karlin S, Genome-scale compositional
comparisons in eukaryotes. Genome Res 2001; 11:
540–546.
128. van Passel MW, et al. An acquisition account of
genomic islands based on genome signature comparisons.
BMC Genomics 2005; 6: p.163.
129. van Passel MW, et al. An in vitro strategy for
the selective isolation of anomalous DNA from prokaryotic
genomes. Nucleic Acids Res 2004; 32:
p.e114.
130. Nakamura Y, et al. Biased biological functions of
horizontally transferred genes in prokaryotic genomes.
Nat Genet 2004; 36: 760–766.
131. Liu ZH, Jiao D, Sun X. Classifying genomic
sequences by sequence feature analysis. Genomics
Proteomics Bioinformatics 2005; 3: 201–205.
132. Reva ON, Tummler B. Global features of sequences
of bacterial chromosomes, plasmids and phages
revealed by analysis of oligonucleotide usage patterns.
BMC Bioinformatics 2004; 5: p.90.
133. Huthoff H, Malim MH. Cytidine deamination and
resistance to retroviral infection: towards a structural
understanding of the APOBEC proteins. Virology
2005; 334: 147–153.
134. Teeling H, et al. TETRA: a web-service and a standalone
program for the analysis and comparison of
tetranucleotide usage patterns in DNA sequences.
BMC Bioinformatics 2004; 5: p.163.
135. Aquino VH, Figueiredo LT. Linear amplification
followed by single primer polymerase chain
reaction to amplify unknown DNA fragments:
complete nucleotide sequence of Oropouche
virus M RNA segment. J Virol Methods 2004; 115:
51–57.
136. Tan G, et al. SiteFinding-PCR: a simple and efficient
PCR method for chromosome walking. Nucleic
Acids Res 2005; 33: p. e122.
137. Chenchik A, et al. Full-length cDNA cloning and
determination of mRNA 50 and 30 ends by amplification
of adaptor-ligated cDNA. Biotechniques 1996;
21: 526–534.
138. Yu Q, et al. Rapid acquisition of entire DNA polymerase
gene of a novel herpesvirus from green turtle
fibropapilloma by a genomic walking technique.
J Virol Methods 2001; 91: 183–195.
139. Huang JC, Chen F. Simultaneous amplification of 50
and 30 cDNA ends based on template-switching
effect and inverse PCR. Biotechniques 2006; 40:
187–189.
140. Polidoros AN, Pasentsis K, Tsaftaris AS. Rolling
circle amplification-RACE: a method for simultaneous
isolation of 50 and 30 cDNA ends from amplified
cDNA templates. Biotechniques 2006; 41: 35–36,
38, 40 passim.
141. Wan K, et al. High-throughput plasmid cDNA
library screening. Nature Protocols 2006; 1: 624–
632.
142. Hoskins RA, et al. Rapid and efficient cDNA library
screening by self-ligation of inverse PCR products
(SLIP). Nucleic Acids Res 2005; 33: p. e185.
143. Fryer JF, et al. Novel parvovirus and related variant
in human plasma. Emerg Infect Dis 2006; 12: 151–
154.
144. Choi EH, et al. The association of newly identified
respiratory viruses with lower respiratory tract
infections in Korean children, 2000–2005. Clin Infect
Dis 2006; 43: 585–592.
145. Arden KE, et al. Frequent detection of human rhinoviruses,
paramyxoviruses, coronaviruses, and
bocavirus during acute respiratory tract infections.
J Med Virol 2006; 78: 1232–1240.
146. Weissbrich BB, et al. Frequent detection of bocavirus
DNA in german children with respiratory
tract infections. BMC Infect Dis 2006; 6: p. 109.
147. Arnold JC, et al. Human bocavirus: prevalence and
clinical spectrum at a children’s hospital. Clin Infect
Dis 2006; 43: 283–288.
148. Kleines M, Scheitchauer S, Rackowitz A, et al. High
prevalence of human bocavirus detected in young
children with severe acute lower respiratory tract
disease using a standard PCR protocol and a novel
real time PCR protocol. J Clin Microbiol 2007 [Epub
ahead of print].
149. Bastien N, et al. Human bocavirus infection,
Canada. Emerg Infect Dis 2006; 12: 848–850.
150. Ma X, et al. Detection of human bocavirus in Japanese
children with lower respiratory tract infections.
J Clin Microbiol 2006; 44: 1132–1134.
151. Sloots TP, et al. Evidence of human coronavirus
HKU1 and human bocavirus in Australian children.
J Clin Virol 2006; 35: 99–102.
152. Harmsen HJ, et al. Extensive set of 16S rRNA-based
probes for detection of bacteria in human feces.
Appl Environ Microbiol 2002; 68: 2982–2990.
153. Breitbart M, et al. Diversity and population structure
of a near-shore marine-sediment viral community.
Proc Biol Sci 2004; 271: 565–574.
154. Margulies M, et al. Genome sequencing in microfabricated
high-density picolitre reactors. Nature.
2005; 437: 376–380.
155. Shendure J, et al. Accurate multiplex polony
sequencing of an evolved bacterial genome. Science
2005; 309: 1728–1732.
156. Edwards RA, et al. Using pyrosequencing to shed
light on deep mine microbial ecology under
extreme hydrogeologic conditions. BMC Genomics
2006; 7: p. 57.
130 E. L. Delwart
Copyright # 2007 John Wiley & Sons, Ltd. Rev. Med. Virol. 2007; 17: 115–131.
DOI: 10.1002/rmv
157. Poinar HN, et al. Metagenomics to paleogenomics:
large-scale sequencing of mammoth DNA. Science
2006; 311: 392–394.
158. Dalton R. Neanderthal DNA yields to genome
foray. Nature 2006; 441: 260–261.
159. Mohammadi T, et al. Removal of contaminating
DNA from commercial nucleic acid extraction
kit reagents. J Microbiol Methods 2005; 61: 285–
288.
160. Evans GE, et al. Contamination of Qiagen DNA
extraction kits with legionella DNA. J Clin Microbiol
2003; 41: 3452–3453.
161. Peters RP, et al. Detection of bacterial DNA in blood
samples from febrile patients: underestimated
infection or emerging contamination? FEMS Immunol
Med Microbiol 2004; 42: 249–253.
162. Meier A, et al. Elimination of contaminating DNA
within polymerase chain reaction reagents: implications
for a general approach to detection of
uncultured pathogens. J Clin Microbiol 1993; 31:
646–652.
163. Nikkari S, et al. Does blood of healthy subjects contain
bacterial ribosomal DNA? J Clin Microbiol 2001;
39: 1956–1959.
164. Lage JM, et al. Whole genome analysis of genetic
alterations in small DNA samples using hyperbranched
strand displacement amplification and
array-CGH. Genome Res 2003; 13: 294–307.
Viral metagenomics 131
Copyright