Advanceshttp://adr.sagepub.

com/
in Dental Research

Next-generation Sequencing Approaches to Understanding the Oral Microbiome

E. Zaura
ADR 2012 24: 81
DOI: 10.1177/0022034512449466

The online version of this article can be found at:

http://adr.sagepub.com/content/24/2/81

Published by:

http://www.sagepublications.com

On behalf of:
International and American Associations for Dental Research

Additional services and information for Advances in Dental Research can be found at:

Email Alerts: http://adr.sagepub.com/cgi/alerts

Subscriptions: http://adr.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Aug 16, 2012

What is This?

Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.

2012 International & American Associations for Dental Research

 Next-generation Sequencing Approaches to
Understanding the Oral Microbiome
E. Zaura
Department of Preventive Dentistry, Academic Centre for Dentistry
Amsterdam, University of Amsterdam and VU University Amsterdam,
Netherlands; e.zaura@acta.nl
Adv Dent Res 22(4):81-85, 2012
Abstract
Until recently, the focus in dental research has been on studying
a small fraction of the oral microbiomeso-called opportunistic
pathogens. With the advent of next-generation sequencing
(NGS) technologies, researchers now have the tools that allow
for profiling of the microbiomes and metagenomes at unprece-
dented depths. The major advantages of NGS are the high
throughput and the fact that specific taxa do not need to be tar-
geted. The relatively low cost and the availability of sequencing
facilities have contributed to nearly exponential growth of NGS
datasets. The quality and interpretation of the NGS data could
be undermined at numerous stepsfrom sample collection,
storage, and DNA extraction to PCR bias, sequencing errors,
choice of algorithms for data processing, and statistical analy-
ses. Making sense out of this data deluge is and will be the major
challenge. The community analyses based on systems ecology
principles will bring us closer to an understanding of the under-
lying forces that facilitate the stability (or imbalance) of the
microbiome. The next logical step will take us beyond the
microbiome. The integration of bacterial, viral, fungal meta-
omes such as the meta-transcriptome, meta-proteome, and
meta-metabolome, together with the host as a major co-factor,
should be the ultimate goal in unraveling the complexity of the
oral interactome.
S ince the introduction of Kochs postulates to the microbial
etiology of infectious diseases, the focus in medical micro-
biology has been on identifying pathogens and deciphering their
virulence. Oral microbiology followed in the footsteps of medi-
cal microbiology, and research efforts have resulted in a wealth
of knowledge on opportunistic pathogens within the commensal
oral microbiota. A milestone in microbiology was the under-
standing that bacterial cells are essential to the well-being of
their host. Since these microscopic inhabitants of our body out-
number our own cells, we are, in fact, walking microbes. For
decades, however, assessment of microbiota of the human body
DOI: 10.1177/0022034512449466
International & American Associations for Dental Research
Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.
2012 International & American Associations for Dental Research
was limited to a fraction of species and strains that could be iso-
lated from laboratory cultures. Since the 1980s, when it became
possible to determine the nucleotide order by sequencing, com-
plete microbial genomes and fragments (small subunit ribosomal
genes) have been sequenced. This has provided information on
the broad diversity of microbial taxa regardless of their culti-
vability. There are now twice as many oral microbial species
that have been identified only by cloning and sequencing than
by culturing efforts (Dewhirst et al., 2010). New possibilities
in microbiome studies were opened during the last decade with
the advent of high-throughput genome sequencing, also known
as next-generation sequencing (NGS) (for an extensive review
on the methodology and recent examples on applicability in the
oral field, see Siqueira et al., 2012). In this overview, the basic
principles of the technology and data analyses will be described,
followed by the most common pros and cons, and the potential
improvements and solutions to the encountered shortcomings.
Finally, the future research perspective beyond the microbiome
will be addressed.
Amplicon Sequencing for Taxonomic
Profiling of the Microbiome
Fragments of genomic DNA are sequenced in a massively paral-
lel way, resulting in gigabases of sequence data. The obtained
results are at an unprecedented sequencing depth if compared
with the conventional cloning and sequencing approach. For
bacterial community profiling (Fig., A), hypervariable regions
of a small subunit of the ribosomal gene, the 16S rRNA, are
used. These regions are flanked with conserved parts of the 16S
rRNA gene, which are used in primer designs to target as diverse
a bacterial population as possible (so-called universal bacterial
primers). The sequences of the hypervariable regions them-
selves are used to discriminate among different bacterial taxa.
The resulting sequence data require processing through a bioin-
formatics pipeline. This pipeline should ensure that low-quality
sequences are discarded and meaningful groups or clusters of
sequencesoperational taxonomic units (OTUs)are created.
The representative sequence of each OTU is then compared with
sequences found in publicly available databases, e.g., Ribosomal
Database Project (RDP) (Cole et al. 2009), and, when possible,
a consensus taxonomic lineage (genus, family, or higher taxon)
is given to the OTU.
Key Words
microbiome, metagenome, high-throughput nucleotide sequencing,
bioinformatics, data analysis, data quality.
81
82 Zaura Adv Dent Res 24(2) 2012
Currently, there are two approaches that are used in microbial
community sequence data analysis. In the analysis based on
OTUs, each OTU is treated equally. This way, poorly defined
microbial ecosystems can be compared without the need to
exclude the undefined, potentially novel phylotypes that cannot
be assigned to a consensus taxonomy lineage. Another approach
is based on phylogeny. Here, the phylogenetic information of
each OTU is used to account for the degree of divergence
between sequences. It has been shown that variances in 16S
rRNA sequences correlate positively with phenotypic variances
of microbiota (Nubel et al., 1999). By the phylogenetic approach,
different communities or sample types are compared based on
their evolutionary distances.
Regardless of which of the two approaches is used, there is a
broad range of diversity measures that can be applied to the
obtained sequence dataset. Alpha diversity regards diversity
within each sample, e.g., species richness, species evenness, and
diversity. The oral microbiome, for instance, represents a micro-
bial ecosystem with low evennessfew taxa such as strepto-
cocci, veillonellae, and prevotellae dominate the samples
obtained from dental plaque or saliva (Keijser et al., 2008; Cri-
elaard et al., 2011). Alpha diversity can be used to describe the
stability of the particular ecosystem. Highly diverse ecosystems
are considered more stable or healthier than communities that
are dominated by few taxa.
Beta diversity considers the differences (both qualitative and
quantitative) between different environments. Beta diversity tests
include statistical comparisons that allow for the assessment of
association of a certain microbial profile with a certain clinical
status. One of the widely accepted methods, UniFrac (Lozupone
and Knight, 2005), is a phylogeny-based method that can be used
to detect qualitative (Unweighted UniFrac) and quantitative dif-
ferences (Weighted UniFrac) among the different sample groups
(Lozupone et al., 2007). Making sense out of this data deluge is
and will be the major challenge. Next to microbiology, training in
bioinformatics and molecular microbial ecology will become
mandatory for the researchers of today and tomorrow.
PROS of Amplicon Sequencing by the Ngs
Approach
Amplicon sequencing by the NGS approach provides high-
throughput sequence information at an exceptional depth.
Hundreds of thousands of sequences can be obtained from a
single sample (Keijser et al., 2008), or, if required, specific
nucleotide barcodes can be added to mark each sample. This
will reduce not only the sequencing depth but also, simultane-
ously, the costs incurred per individual sample. There is a trade-
off, however. By reducing the sequencing depth, one might lack
the discriminatory power for effects of subtle interventions,
such as supplementation with pre- and probiotics. Major eco-
logical shifts, such as effects of potent antimicrobials, would
still be discernible at a relatively low sequencing depth
(Kuczynski et al., 2010). The actual number of reads per sam-
ple, however, needs to be validated per intervention.
The results of NGS are not biased by fast-growing, easily
cultivable taxa, and allow for hypothesis-driven research on
previously unknown and unclassified micro-organisms.
Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.
2012 International & American Associations for Dental Research
The information obtained by NGS is not targeted to a priori-
selected taxa as, e.g., in the microarray approach. This allows
for open-ended view on a whole breadth of the microbiome and
provides a new opportunity for oral-health-related studies.
Research questions on microbial stability and ecological shifts
due to environmental and host factors can now be addressed on
a full scale of community complexity.
CONs of Amplicon Sequencing in
Microbiome Studies and Ways to Increase
the Success Rate
The sequencing throughput of NGS provides exciting opportu-
nities for research. The down side of this enthusiasm is that the
quality of data generated is often undermined by failure to
address fundamental aspects of experimental design (Rogers
and Bruce, 2010). Only 18% of sequencing studies published in
2009 in major ecological journals analyzed replicate samples
(Prosser, 2010). This may be due to a belief that these cutting-
edge technologies are exempt from normal standards, and that
the costs involved justify the lack of proper design (Prosser,
2010). With the decreasing costs of sequencing and the avail-
ability of sample identification tags, lack of sample replicates
should not be accepted. As with any conventional studies, statis-
tics should be planned in advance, before the start of the study.
The major disadvantage of current NGS used for amplicon
sequencing is the short read length. This precludes accurate taxo-
nomic identification and results in low taxonomic resolution.
Most reads can be identified to genus level, but only a fraction to
species level. This depends on 16S rRNA gene sequence homol-
ogy among the members of the same genus. Depending on a
hypervariable region of the 16S rRNA gene that is targeted, dif-
ferent taxa are either missed or can be classified only at a higher
taxonomic level (family, class, or even phylum). The problem can
be diminished by in silico analyses of 16S rRNA sequences prior
to selection of the target region (Brandt et al., 2012). However,
some taxa, like several closely related streptococci, will be hard
or even impossible to distinguish by 16S gene sequence. In that
case, either a more specific gene instead of the small subunit
rRNA gene should be targeted, or a full metagenomic sequencing
(described below) should be applied.
As any other molecular-biology-based method, NGS suffers
from typical bias of DNA extraction and amplification (Hong
et al., 2009; Nadkarni et al., 2009), such as selectivity of prim-
ers and intrinsic differences in the amplification efficiency of
templates. Unfortunately, bias by DNA amplification by poly-
merase chain-reaction (PCR) can be avoided only by avoiding
the amplification step itself. This would mean choosing the full
metagenomic sequencing approach instead of amplicon sequenc-
ing of hypervariable regions of the 16S rRNA gene.
The quality of the DNA extraction protocol should be evalu-
ated in advance. In standard protocols, all DNA, including DNA
from dead or damaged cells and extracellular matrix, is used in
the downstream analysis. In intervention studies, where treat-
ment has an antimicrobial potential, it would lead to underesti-
mation of the effects of the intervention. The solution lies in the
removal of all DNA that does not originate from intact cells
before the DNA extraction process. One such approach is to
Adv Dent Res 24(2) 2012 Next-generation Sequencing and Microbiome 83

Figure. Flow diagram of two next-generation sequencing approaches for microbiome studies. (A) Sequencing of DNA amplicons targeting spe-
cific 16S rRNA gene fragments (hypervariable regions). The obtained sequence data are compared with sequences in small subunit ribosomal
RNA gene databases and are used for taxonomic profiling and diversity analyses. (B) Direct sequencing of random DNA fragments, also called
metagenomic shotgun sequencing. A sequence contig is a contiguous, overlapping sequence resulting from the re-assembly of the small DNA frag-
ments. The obtained sequence data are compared with full-genome reference databases and are used to describe the predominant functions of the
microbial communities, as well as to identify the microbial taxa. Steps indicated in gray differ between the two methods.

incubate samples with propidium monoazide (PMA) (Loozen
et al., 2011), where PMA will bind to extracellular DNA and to
DNA of cells that have lost their structural integrity. By expo-
sure to visible light, this reaction becomes irreversible and ren-
ders the PMA-bound DNA unable to act as a template for PCR
amplification (Rogers and Bruce, 2010). Membrane integrity,
however, is considered a conservative criterion for microbial
viability, and other approaches related to cell activity have been
proposed (Nocker and Camper, 2009).
Any sequencing method, but especially high-throughput
NGS technologies, suffers from sequencing errors (Kunin et al.,
2010; Balzer et al., 2011). Although newest-generation tech-
nologies are able to generate up to 1,000-bp-long reads, there is
a trade-off in positive correlation between the read length and
the error rate. Data pre-processing after the sequencing is used
to minimize the inaccuracy of the output. Low-quality reads,
reads below or above the certain length cutoff, and reads with
ambiguous base call and homopolymers are usually filtered out
from the dataset. After these pre-processing steps, the dataset is
frequently used for community analyses, while more recent
reports indicate the need for additional cleaning steps (Kunin
et al., 2010). PyroNoise (Quince et al., 2009) and Denoiser
(Reeder and Knight, 2010) are two examples of the tools that are
applied to remove the sequencing noise. In this process, how-
ever, valid, low-abundance sequences are also de-noised to
appear more similar to reads that are found at a higher abun-
dance and are assumed to be without errors. This, in turn, may
lead to clustering of the sequences into lower numbers of OTUs
and to underestimation of sample diversity.
Another cleaning step besides removing the noise is iden-
tification and removal of chimeric sequences (Edgar et al.,
2011; Haas et al., 2011). Chimeras are sequences that are cre-
ated in the PCR amplification process from two or more tem-
plates instead of a single parent template. Exact reasons for
chimera formation are not well-understood, but it has been
associated with PCR conditions, fragment length, and, possibly,
sample composition.
After the cleaning steps, the sequences are clustered into
OTUs at a predetermined similarity level, usually 97%. It has
been demonstrated that the specific choice of the clustering
algorithm affects the output by either under- or overestimating
the diversity of the sample (Sun et al., 2011). New, improved
algorithms appear as we speak, and preclude direct comparison
with earlier results. For that, all the steps of the pipelinefrom
sequence preprocessing until cleaning and clusteringshould
be performed at once.
Thus far, statistical data analyses have overlooked the fact
that microbiome profiles are not representations of absolute
measurements (e.g., microbial counts), but are typical examples
of compositional data, which are in the form of relative propor-
tions of taxa. These taxa can be either observed or missed,
depending on the sampling effort, amplification bias, and
sequencing depth. An increase in the relative abundance of some
taxa will be accompanied by a compositional decrease of other
taxa. This may lead to statistically significant, though spurious,
correlations without any biological dependence among the taxa
involved. Computational tools suitable for compositional data
analyses should be developed and implemented.

Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.

2012 International & American Associations for Dental Research

84 Zaura Adv Dent Res 24(2) 2012
Direct Shotgun Sequencing of Collective
Genome of the Microbiome (Metagenome)
Amplicon sequencing (sequencing of a targeted fragment of 16S
rRNA gene) provides information on only the microbial taxa
(the players) in the community. With NGS technologies, full
genomic DNA can be sequenced without the targeting step and
without PCR bias of the amplicon sequencing (Fig., B). Instead
of sequencing a single isolated clone (as with the traditional
cloning and sequencing approach), the genome of the entire
community (metagenome) is sequenced. For this, extracted
DNA is sheared into random fragments and sequenced directly.
Contaminations, e.g., human DNA, should be removed either
prior to (Hunter et al., 2011) or after the sequencing step by
filtering the data. Then, exhaustive bioinformatics steps follow,
where DNA fragments need to be assembled into genomes
against a reference genome database, followed by functional
gene annotation (Mitra et al., 2011; Simon and Daniel, 2011).
The major difficulty in the approach is genome assembly while
information is incomplete. First, the sampling is incomplete, and
most genomes are partially sequenced, if at all. Sequences
originating from predominant micro-organisms will dominate
the data, while less abundant species might be missed if
sequencing depth is not sufficient. Second, the information on
individual genomes (reference genomes) is often incomplete,
lacks accuracy, and has inconsistent annotations, and may pre-
clude the mapping of individual reads to the species of origin.
Nevertheless, impressive insights have already been obtained in
research on the gastrointestinal tract (Arumugam et al., 2011).
The stage in the oral field has just been set (Xie et al., 2010;
Belda-Ferre et al., 2011).
Toward the Interactome
Future developments should bring us toward translating the
genotype into phenotype (Henry et al., 2011). The question that
we should answer is What are they doing? instead of just
wondering Who is there? The expertise of researchers will
need to go beyond the field of microbiology and cariology, and
will have to apply systems ecology principles. A complex and
integrated systems biology and ecology approach should bring
us closer to understanding the underlying forces that facilitate
the stability (or imbalance) of the microbiome. The integration
of bacterial, viral, and fungal meta-omes such as the meta-
transcriptome, meta-proteome, and meta-metabolome, together
with the host as a major co-factor, should be the ultimate goal in
unraveling the complexity of the oral interactome.
Acknowledgments
The author received no financial support and declares no poten-
tial conflicts of interest with respect to the authorship and/or
publication of this article.
References
Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR,
et al. (2011). Enterotypes of the human gut microbiome. Nature
473:174-180; erratum in Nature 474:666, 2011).
Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.
2012 International & American Associations for Dental Research
Balzer S, Malde K, Jonassen I (2011). Systematic exploration of error
sources in pyrosequencing flowgram data. Bioinformatics 27:i304-i309.
Belda-Ferre P, Alcaraz LD, Cabrera-Rubio R, Romero H, Simon-Soro A,
Mira A, et al. (2011). The oral metagenome in health and disease. ISME
J 6:46-56.
Brandt BW, Bonder MJ, Huse SM, Zaura E (2012).TaxMan: a server to trim
rRNA reference databases. Nucl Acids Res [Epub ahead of print
5/22/2012] (in press).
Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, et al. (2009). The
Ribosomal Database Project: improved alignments and new tools for
rRNA analysis. Nucleic Acids Res 37:D141-D145.
Crielaard W, Zaura E, Schuller A, Huse S, Montijn R, Keijser BJ, et al.
(2011). Exploring the oral microbiota of children at various develop-
mental stages of their dentition in the relation to their oral health. BMC
Med Genomics 4:22.
Dewhirst FE, Chen T, Izard J, Paster BJ, Tanner AC, Yu WH, et al. (2010).
The human oral microbiome. J Bacteriol 192:5002-5017.
Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R (2011). UCHIME
improves sensitivity and speed of chimera detection. Bioinformatics
27:2194-2200.
Haas BJ, Gevers D, Earl AM, Feldgarden M, Ward DV, Knight R, et al.
(2011). Chimeric 16S rRNA sequence formation and detection in
Sanger and 454-pyrosequenced PCR amplicons. Genome Res 21:494-
504.
Henry CS, Overbeek R, Xia F, Best A, Glass E, Gilbert J, et al. (2011).
Connecting genotype to phenotype in the era of high-throughput
sequencing. Biochim Biophys Acta 1810:967-977.
Hong S, Bunge J, Leslin C, Jeon S, Epstein SS (2009). Polymerase chain
reaction primers miss half of rRNA microbial diversity. ISME J 3:1365-
1373.
Hunter SJ, Easton S, Booth V, Henderson B, Wade WG, Ward JM, et al.
(2011). Selective removal of human DNA from metagenomic DNA
samples extracted from dental plaque. J Basic Microbiol 51:442-446.
Keijser BJ, Zaura E, Huse SM, van der Vossen JM, Schuren FH, ten Cate
JM, et al. (2008). Pyrosequencing analysis of the oral microflora of
healthy adults. J Dent Res 87:1016-1020.
Kuczynski J, Costello EK, Nemergut DR, Zaneveld J, Lauber CL, Knights
D, et al. (2010). Direct sequencing of the human microbiome readily
reveals community differences. Genome Biol 11:210.
Kunin V, Engelbrektson A, Ochman H, Hugenholtz P (2010). Wrinkles in
the rare biosphere: pyrosequencing errors can lead to artificial inflation
of diversity estimates. Environ Microbiol 12:118-123.
Loozen G, Boon N, Pauwels M, Quirynen M, Teughels W (2011). Live/dead
real-time polymerase chain reaction to assess new therapies against
dental plaque-related pathologies. Mol Oral Microbiol 26:253-261.
Lozupone C, Knight R (2005). UniFrac: a new phylogenetic method for
comparing microbial communities. Appl Environ Microbiol 71:8228-
8235.
Lozupone CA, Hamady M, Kelley ST, Knight R (2007). Quantitative and
qualitative diversity measures lead to different insights into factors
that structure microbial communities. Appl Environ Microbiol 73:1576-
1585.
Mitra S, Rupek P, Richter D, Urich T, Gilbert J, Meyer F, et al. (2011).
Functional analysis of metagenomes and metatranscriptomes using
SEED and KEGG. BMC Bioinformatics 12(Suppl 1):S21.
Nadkarni MA, Martin FE, Hunter N, Jacques NA (2009). Methods for opti-
mizing DNA extraction before quantifying oral bacterial numbers by
real-time PCR. FEMS Microbiol Lett 296:45-51.
Nocker A, Camper AK (2009). Novel approaches toward preferential detec-
tion of viable cells using nucleic acid amplification techniques. FEMS
Microbiol Lett 291:137-142.
Nubel U, Garcia-Pichel F, Kuhl M, Muyzer G (1999). Quantifying microbial
diversity: morphotypes, 16S rRNA genes, and carotenoids of oxygenic
phototrophs in microbial mats. Appl Environ Microbiol 65:422-430.
Prosser JI (2010). Replicate or lie. Environ Microbiol 12:1806-1810.
Quince C, Lanzen A, Curtis TP, Davenport RJ, Hall N, head IM, et al.
(2009). Accurate determination of microbial diversity from 454 pyrose-
quencing data. Nat Methods 6:639-641.
Reeder J, Knight R (2010). Rapidly denoising pyrosequencing amplicon
reads by exploiting rank-abundance distributions. Nat Methods 7:668-
669.
Adv Dent Res 24(2) 2012 Next-generation Sequencing and Microbiome 85

Rogers GB, Bruce KD (2010). Next-generation sequencing in the analysis
of human microbiota: essential considerations for clinical application.
Mol Diagn Ther 14:343-350.
Simon C, Daniel R (2011). Metagenomic analyses: past and future trends.
Appl Environ Microbiol 77:1153-1161.
Siqueira JF Jr, Fouad AF, Rocas IN (2012). Pyrosequencing as a tool for
better understanding of human microbiomes. J Oral Microbiol [Epub
ahead of print 1/23/2012] (in press).
Sun Y, Cai Y, Huse SM, Knight R, Farmerie WG, Mai V, et al. (2011). A
large-scale benchmark study of existing algorithms for taxonomy-
independent microbial community analysis. Brief Bioinform 13:107-
121.
Xie G, Chain PS, Lo CC, Liu KL, Gans J, Qui F, et al. (2010). Community
and gene composition of a human dental plaque microbiota obtained by
metagenomic sequencing. Mol Oral Microbiol 25:391-405.

Downloaded from adr.sagepub.com at Sichuan University on September 17, 2012 For personal use only. No other uses without permission.

2012 International & American Associations for Dental Research