N u c l e i c A c i d - b a s e d M e t h o d s t o
`A s s e s s t h e C o m p o s i t i o n a n d
`F u n c t i o n o f t h e B o w e l M i c ro b i o t a
`
`Blair Lawley, PhDa, Gerald W. Tannock, PhDa,b,*
`
`KEYWORDS
` Bowel microbiota  Probiotic  Fecal microbiota  Microbiota compositional analysis
` Microbiota functional analysis  Nucleic acid-based analysis of microbiota
`
`KEY POINTS
` Over evolutionary time, people have developed an equilibrium with the microbial world,
`which consists of cloaking the body inside and out with microorganisms that are likelier
`to be friends than enemies.
` Nucleic acid-based methods of analysis are widely used to determine and monitor the
`composition of this metaphorical cloak of microbes (microbiota). This was originally
`because many of the members of microbiota had not yet been cultivated in the laboratory.
` Nucleic acid-based methods also facilitate logistical planning and execution of microbiota
`analysis for probiotic, clinical, and nutritional trials using human subjects.
`
`IS THERE A NEED TO KNOW ABOUT MICROBIOTA COMPOSITION IN PROBIOTIC
`STUDIES?
`
`Roy Fuller, a pioneer in the probiotic field, defined a probiotic as a “live microbial feed
`supplement which beneficially affects the host animal by improving its intestinal micro-
`bial balance.”1 This definition infers that consumption of the probiotic preparation will
`alter the proportions of the various populations that comprise the microbiota. There is
`little evidence that this happens, apart from small increases in the abundance of the
`taxonomic group to which the probiotic belongs. Moreover, the effect is transient,
`because the probiotic bacteria are only detected in feces as long as the probiotic is
`consumed.2 In other words, it is a temporary addition to the microbiota of the large
`bowel without displacement of resident populations. There is recent evidence that
`the biochemistry of the microbiota may change during probiotic consumption, but
`
`a Department of Microbiology and Immunology, University of Otago, Post Office Box 56, 720
`Cumberland Street, Dunedin, New Zealand; b Riddet Institute Centre of Research Excellence,
`Cnr University Avenue and Orchard Road, Massey University, PO Box 11 222, Palmerston
`North 4442, New Zealand
`* Corresponding author. Department of Microbiology and Immunology, University of Otago,
`Post Office Box 56, 720 Cumberland Street, Dunedin, New Zealand.
`E-mail address: gerald.tannock@otago.ac.nz
`
`Gastroenterol Clin N Am 41 (2012) 855–868
`http://dx.doi.org/10.1016/j.gtc.2012.08.010
`0889-8553/12/$ – see front matter Ó 2012 Elsevier Inc. All rights reserved.
`
`gastro.theclinics.com
`
`

`

`856
`
`Lawley & Tannock
`
`an impact on the immune system by consumption of probiotic bacteria seems to be
`the main outcome predicted.3,4 It can be argued, therefore, that probiotic stimuli are
`directed at the mucosal immune system of the small bowel as the bolus of bacteria
`progresses from stomach to colon after consumption.
`Until the situation is clarified, and the mechanisms that mediate probiotic activity are
`discovered, it seems wise to include analysis of the fecal or other microbiota as part of
`human probiotic trials. The more complete the picture that is generated, the more
`likely that mechanistic details will be revealed.
`While probiotic bacteria are allochthonous to the bowel biome, they should have the
`capacity to at least survive transit through the human gut after consumption to have
`some efficacy. Thus detection and quantification of the probiotic strain among the
`myriad of autochthonous commensals of the large bowel should be accomplished
`(see section on Why Detect Probiotic Bacteria Using Bulk DNA?).
`Human probiotic trials of appropriate statistical power are difficult and expensive to
`set up and run. The main outcome is usually a clinical read-out such as prevalence of
`eczema in a test group relative to a placebo control group. Recruits to clinical trials are
`not usually reluctant to provide fecal samples, and at the least, these can be archived
`in small aliquots for later microbiota analysis when initial results make this desirable, or
`additional funding becomes available.
`It is recommended that temporal studies of the microbiota be performed, necessi-
`tating the collection of specimens for microbiota analysis at several time points during
`the study. This will provide much needed information about the stability/variability of
`microbiota compositions over time. A major criticism of gut microbiota research is
`that human studies, and most experimental animal studies, are one-off; they are never
`repeated with another group of people or animals. Therefore the consistency of pro-
`biotic effects and microbiota composition are not known. This is in contrast to human
`drug trials, in which repetition of trials in more than one country is required before the
`drug is approved for general use.
`In summary, analysis of the microbiota should be performed because
` Probiotics may yet be shown to alter the microbial component of the bowel or
`other biome in a subtle yet significant way.
` Efficacy of probiotic consumption is presumably related to its presence in the
`bowel or other body site, and this should be demonstrated.
` More complete pictures of bowel ecology are required; human trials with probi-
`otics provide possibilities to acquire this knowledge.
`
`WHY USE NUCLEIC ACID-BASED TOOLS IN MICROBIOTA STUDIES?
`
`Much of the bacteriologic information about the bowel community has been generated
`by the use of nucleic acid-based methodologies.5 The phylogenetic analysis (deter-
`mining the phyla, families, genera, and species through molecular sequencing data
`and construction of data matrices) has mostly relied on comparisons of sequences
`of the small ribosomal subunit RNA genes (16S rRNA gene in the case of bacteria)
`present in DNA extracted from feces or other samples. A large database of about 2
`million 16S rRNA gene sequences is currently available and provides a cornerstone
`of bacterial detection and identification by molecular methods.6
`Nucleic acid-based methods of detection have indicated that most (about 90%)
`bacterial cells seen microscopically in terrestrial and aquatic samples, even
`accounting for the possibility that some are dead, have not yet been cultured in the
`laboratory.7 This observation was totally unexpected in relation to traditional experi-
`ences in medical bacteriology and has been called the great plate count anomaly.
`
`

`

`Analysis of the Bowel Microbiota
`
`857
`
`Operational taxonomic units (OTU, molecular species) never encountered in culture-
`based bacteriology but detectable by nucleic acid-based methods have revealed
`a vast, new bacterial world for investigation.
`The situation with human feces is not dire, since at least 50% of the cells seen in
`samples can be cultivated. Additionally, representatives of most of the metabolic types
`of bacteria in the human colon have been cultured.8 It is still convenient, however, to use
`nucleic acid-based methods to determine the composition and potential or real-time
`functioning of bowel communities. This is because of the diversity of bacterial types
`present in the bowel of people and the somewhat idiosyncratic nature of individual
`microbiota. Extraction of nucleic acids from feces, bowel digesta, or mucosal biopsies
`is now a standard procedure and is logistically much simpler than the preparation of
`a spectrum of selective media and anaerobic protocols that are required for culture-
`based investigations. Moreover, culture-based analysis requires the processing of
`the samples soon after collection, whereas materials for nucleic acid-based analysis
`can be frozen after collection and stored at low temperature (dry ice) during transport
`to the laboratory and while awaiting analysis. DNA, when stored at very low temperature
`
`(-80
`C), provides a useful archive of microbial genomes that might be used in the future
`when advanced methodologies may increase the amount of information that can be
`gleaned from bulk DNA.
`The starting point for nucleic acid-based analysis is the extraction of DNA or RNA
`directly from the fecal or other sample of interest, avoiding the need to cultivate any
`members of the community. A critical step in the procedure is the lysis of microbial
`cells to release the nucleic acids, because accurate representation of all of the micro-
`bial types in the sample is required. Although chemical methods have been used,
`mechanical disruption of the cells (bead-beating) is considered preferential, because
`even gram-positive bacterial cells, often resilient to other methods, will be lysed.
`The 16S rRNA gene sequence of bacteria has become the basis of bacterial
`phylogeny, because it contains regions of nucleotide base sequence that are highly
`conserved across the bacterial world.9 These conserved regions are interspersed
`with variable regions (V regions) that contain the signatures of phylogenetic groups
`even to species level. Therefore sequences of V regions of 16S rRNA genes are the basis
`of phylogenetic methods, as will be reinforced in subsequent sections of this article.
`It is important to remember, however, that bulk DNA has come from any cells that
`were present in the sample at the time of collection. The nucleic acids may have come
`from resident, living, metabolizing cells, or resident, living but relatively quiescent cells
`or spores, from dead bacteria, or living bacteria adventitiously present at the time of
`collection. Detection of a particular DNA sequence in a single sample does not equate
`to evidence of residence of the organism in the ecosystem. Collection and analysis of
`samples over time are necessary to provide this evidence.
`The capacity to sequence and assemble genomic information from fragments of
`DNA (metagenomics) or mRNA (transcriptomics) has extended the use of nucleic
`acids in researching the potential and actual metabolic capacity of the microbiota.
`These exciting approaches will be outlined in other sections of this article.
`In summary, nucleic acid-based analysis of fecal or other samples from people is
`useful because
` It enables descriptions to be made of the phylogenetic composition of the
`microbiota including its yet to be cultivated members.
` It has logistical advantages with respect to transport and storage of samples.
` It enables the genetic potential of the microbiota to be determined (metagenomics).
` It enables the expression of genes at a point in time to be revealed (transcriptomics).
`
`

`

`858
`
`Lawley & Tannock
`
`WHY DETECT PROBIOTIC BACTERIA USING BULK DNA?
`
`By current definition, probiotics are live microorganisms, which when administered
`in adequate amounts confer a health benefit on the host.10 Most probiotics are admin-
`istered as dietary supplements either in milk-based foods or as tablets or capsules.
`The detection of the probiotic bacteria in fecal or other samples is not routine in effi-
`cacy trials with people. However, inclusion of an assay to detect the presence or
`absence of the probiotic during the progress of the trial seems a wise step, since
`any absence of efficacy could be due to the inability of the probiotic to survive transit
`through the gastrointestinal tract. Even when probiotic efficacy has been established,
`a mechanistic explanation is often lacking. Hence the more detailed microbiological
`data available that can be interrogated statistically, the better.
`A study demonstrating the differential effect of 2 probiotics in the prevention of
`eczema and atopy provides a case in point.11 The aim of the study was to determine
`whether probiotic administration in early life could prevent the development of eczema
`and atopy at 2 years of age. A double-blind, randomized placebo-controlled trial of
`infants at risk of allergic disease was performed. Pregnant women were randomized
`to take Lactobacillus rhamnosus HN001, Bifidobacterium animalis subspecies lactis
`HN019, or placebo, daily from 35 weeks gestation until 6 months if breast-feeding,
`and their infants were randomized to receive the same treatment from birth to 2 years.
`Four hundred and seventy-four infants were included in the study. The infants’ cumu-
`lative prevalence of eczema and point prevalence of atopy (skin prick tests to common
`allergens) were assessed at 2 years. Infants receiving L. rhamnosus probiotic had
`a significantly reduced risk of eczema (hazard ratio [HR] 0.51, 95% confidence interval
`[CI] 0.30–0.85, P 5 .01) compared with placebo, but this was not the case for the
`B. animalis subspecies lactis group (HR 0.90, CI 0.58–1.41).
`Why was one probiotic effective but the other not? One possible explanation con-
`cerned the markedly different rates of detection of the probiotics in the stools of the
`infants during the trial. A DNA-based technique (polymerase chain reaction [PCR]
`coupled with an electrophoretic detection method; see section on Electrophoretic
`Methods to Screen Microbiota Compositions) showed that B. animalis subspecies lac-
`tis prevalence in fecal samples increased progressively over the course of the study,
`from 22.6% of infants at 3 months to 53.1% of infants at 24 months. In contrast, detec-
`tion of L. rhamnosus was greatest at 3 months, at 71.5% of babies, and slightly lower
`at 24 months, at 62.3% among infants administered this probiotic. Therefore, the effi-
`cacious probiotic (L. rhamnosus) was commonly present in the bowel of the infants
`from an early age, whereas the ineffective probiotic did not seem to transit the gut
`in about half of the infants.
`An indication of the numbers of probiotic bacteria in fecal samples could give addi-
`tional interpretative data in trials with people. However, it must be kept in mind that
`absolute guarantees of specificity (quantifying only a specific strain—the probiotic—
`in samples collected from nature) require a cautionary interpretation. Limited numbers
`of strains of the probiotic species are usually available, either in laboratory collections
`or in the form of genome sequences. Therefore, validation of specificity of the nucleic
`acid-based method is subject to relatively limited knowledge when the study was
`designed and performed.
`In summary, detection of probiotic strains in feces by DNA-based assays
` Demonstrates that the probiotic bacteria have at least survived transit through
`the gut (although viability cannot be demonstrated)
` May aid in the interpretation of efficacy outcomes in trials of probiotics with
`people
`
`

`

`Analysis of the Bowel Microbiota
`
`859
`
`ELECTROPHORETIC METHODS TO SCREEN MICROBIOTA COMPOSITIONS
`
`A relatively simple, semiquantitative screening method to compare the bacterial
`compositions of multiple samples is provided by PCR coupled with gradient gel elec-
`trophoresis.12 Either a chemical gradient (denaturing gradient gel electrophoresis,
`DGGE) or temperature gradient (temporal temperature gel electrophoresis, TTGE) is
`used. A V region of bacterial 16S rRNA genes is amplified from bulk DNA from samples
`using PCR. The PCR primers anneal with conserved sequences that span the selected
`0
`V region. One of the primers has a GC-rich 5
`end (GC clamp) to prevent complete dena-
`turation of the amplified DNA fragments during gel electrophoresis. The amplified DNA
`theoretically contains 16S rRNA gene sequences from all of the members of the micro-
`biota in the sample. All of the amplified fragments are of the same size. A polyacryl-
`amide gel
`is used to separate the fragments of 16S rRNA gene sequences that
`originated in different kinds of bacteria. The double-stranded fragments of DNA migrate
`in an electrical current through the polyacrylamide gel until they reach chemical or
`thermal conditions that represent their melting point (separation of polynucleotide
`strands). The migration rate of the partially denatured fragments (remember the GC
`clamp) is slowed. Because of the variation in the 16S rRNA gene sequences between
`phylogenetic groups, the DNA fragments from different kinds of bacteria have different
`melting points and hence show different migrations in the gel. Profiles of microbiota
`composition can be generated once the electrophoretic gel has been stained to reveal
`bands of DNA. Images of these profiles can be compared using computer software.
`For example, differential clustering of bowel biopsy-associated bacteria of specimens
`collected in Mexico and Canada was demonstrated using PCR-TTGE.13 Bacterial collec-
`tions associated with bowel biopsies, aspirates of residual fluid after pre-endoscopy
`cleansing, and feces from inflammatory bowel diseases (IBD) patients and healthy
`subjects in Edmonton, Canada, and Mexico City, Mexico, were investigated. PCR-
`TTGE produced profiles of the bacterial collections whose similarities were compared.
`Similarity analysis showed that the profiles did not cluster according to disease status,
`but the Canadian and Mexican profiles could be differentiated by this method. Compar-
`ison of biopsy, aspirate, and fecal samples obtained from the same subject showed that,
`on average, the profiles were highly similar. Therefore, biopsy-associated bacteria are
`likely to represent, at least in part, contaminants from the fluid, which resembles a fecal
`solution, that pools in the bowel after cleansing before endoscopy.
`Individual fragments of DNA can be cut from electrophoretic gels, further amplified
`and cloned in a surrogate host (Escherichia coli), then sequenced.14 The sequence can
`be compared with those in 16S rRNA gene databanks to obtain identification of the
`bacterium from which the sequence originated. PCR primers specific for particular
`groups of bacteria (for example lactic acid bacteria), rather than universal primers
`used to detect all bacteria, can be designed.15 These primers produce simpler profiles
`that can be useful in the detection of bacterial species commonly used as probiotics.
`In summary, electrophoretic methods such as DGGE and TTGE
` Provide comparative snapshots of microbiota
` Can detect, using appropriate PCR primer design, specific bacterial groups
`within the microbiota
`
`FLUORESCENT PROBES
`
`DNA probes (oligonucleotides) can be designed using information in DNA databases.
`They specifically target (anneal to) V region sequences in bacterial 16S rRNA or 23S
`0
`label consisting
`rRNA (large ribosomal subunit). The probes are synthesized with a 5
`
`

`

`860
`
`Lawley & Tannock
`
`of a fluorescent dye. The probes bind to target sequences within bacterial cells,
`enabling the bacteria to be detected and quantified. This method is referred to as ‘fluo-
`rescence in situ hybridization’ (FISH). Bacterial cells within which hybridization with
`a probe has occurred fluoresce when stimulated with appropriate wavelengths of light
`and hence can be detected and counted by epifluorescence microscopy or
`fluorescence-activated flow cytometry (FC).16 Permeabilization of the bacterial cells
`in samples is required to standardize intracellular access of DNA probes to their
`targets. Other technical considerations include
` The physiologic state of the bacterial cells, because the number of ribosomes per
`bacterial cell is greater the higher the metabolic activity; bacterial cells in a quies-
`cent state will have weak fluorescence and may not be detected
` Hybridization stringency determined by temperature, salt and formamide
`concentrations
` The need to check the specificity of probes by reference to databanks; new
`sequences are constantly added.
`
`Clostridium difficile infection (CDI) is the most common identifiable cause of diarrhea
`in hospitalized patients. Current therapies rely on the administration of metronidazole
`or vancomycin, which reduce vegetative populations of C. difficile in the bowel. Recur-
`rence of the disease when treatment with these antibiotics ceases indicates that metro-
`nidazole and vancomycin affect not only C. difficile but also commensal populations
`that normally mediate competitive exclusion. Fidaxomicin is a new antibiotic that
`inhibits C. difficile. PCR-TTGE and FISH/FC were used in a study of 23 patients with
`mild-to-moderate CDI to determine the different impacts of vancomycin and fidaxomi-
`cin on the composition of the fecal microbiota.17 Multidimensional scaling (production
`of a similarity matrix for pair-wise comparisons of TTGE profiles) indicated that the
`microbiota of fidaxomicin-treated patients was different from that of vancomycin-
`treated subjects. FISH/FC analysis showed that clostridial cluster XIVa and clostridial
`cluster IV populations increased during and after the fidaxomicin treatment period.
`Clostridial cluster XIVa populations were similar to those in the feces of healthy control
`subjects by day 10. In contrast, vancomycin treatment greatly reduced the proportions
`of the clostridial clusters, and also of bifidobacteria, by day 10 of treatment. Outgrowth
`of enterobacteria coincided with the decrease in other phylogenetic groups. Overall,
`vancomycin treatment was characterized by a decrease in the proportions of obligately
`anaerobic bacteria that normally populate the human colon, and an outgrowth of facul-
`tatively anaerobic and microaerobic bacterial groups during the treatment period.
`These findings help to explain the substantially reduced rates of relapse following treat-
`ment of CDI with fidaxomicin in recent clinical trials; fidaxomicin is sparing of the bowel
`commensals that normally regulate C. difficile population sizes in the gut.
`In summary, FISH/FC
` Detects and quantifies specific groups of bacteria (even to species level) in fecal
`and other samples
` Can be used to screen the microbiota for major alterations in composition
` Quantifies bacterial cells rather than DNA sequences and so is more easily
`related to traditional bacteriologic methods
`
`MEASURING ABUNDANCE OF BACTERIAL GROUPS BY QUANTITATIVE PCR
`
`PCR is a powerful tool for the detection of nucleic acid targets (both phylogenetic and
`functional). The utility of PCR can be further enhanced when it is used not only to
`detect a target but also to measure abundance of that target.
`
`

`

`Analysis of the Bowel Microbiota
`
`861
`
`PCR requires the following components: a target DNA molecule (template), primers
`(short single-stranded oligonucleotides), a thermostable DNA polymerase, and free
`nucleotides (dNTPs). The reaction begins with a denaturation phase in which the
`template DNA strands are separated. The primers then bind, 1 to each strand, brack-
`eting a region of DNA that may range from tens to thousands of nucleotides (the
`annealing phase). Using the primers as starting points, the DNA polymerase extends
`0
`0
`complementary DNA strands by incorporating dNTPs in a 5
`to 3
`direction (the exten-
`sion phase). Thus the region of interest is effectively duplicated. The whole process is
`repeated multiple times (usually 30–40 cycles), and the quantity of target DNA doubles
`with each cycle.
`PCR becomes quantitative when the amplified fragments are detected in real time
`after each cycle. To do this, most methods rely on either incorporation of a fluorescent
`dye or release of a fluorescent signal from a DNA-binding probe. The signal is recorded
`by a sensitive camera or spectrophotometer.18 When the starting quantity of a template
`is high, a signal will be detected at an early cycle, while if the initial quantity of a template
`is low, many cycles of PCR will be required before a signal can be observed. Absolute
`quantitation of a target sequence is achieved when a standard curve, prepared using
`known quantities of template, is generated and subsequently used to quantify unknown
`samples.
`The specificity of the reaction can be modified through careful design of the primer
`sequences. Primers can be designed to target individual species, genera, families,
`phyla, or kingdoms. Thus primer combinations may be used to quantify as broad or
`narrow a microbial group as is required.19
`Quantitative PCR (qPCR) can be time consuming and technically demanding (espe-
`cially primer design) and is a relatively low-throughput technique. As such, it is often
`used in conjunction with other methods as a confirmatory tool or a means of quanti-
`fying specific targets or groups within a microbiota.
`Hartman and colleagues20 assayed bacterial populations in the small bowel of patients
`who had undergone small bowel transplant. qPCR was used to measure changes in the
`relative proportions of 4 bacterial groups (Enterobacteriales, Lactobacillales, Clostri-
`diales, and Bacteroidales) in addition to total bacterial load. Post-transplant samples,
`before ileostomy closure, were considered to be abnormal, because they were domi-
`nated by facultative anaerobes or microaerobic bacteria (Enterobacteriales and Lactoba-
`cillales). After closure of the ileostomy, the relative proportion of strict anaerobes
`increased, thus returning the community to what was considered a more normal state.
`The authors concluded that the population shift was due to oxygen entering the small
`bowel through the ileostomy. This study implies flexibility within the ileal microbiota,
`whereby ecological factors can drive population change over relatively short time frames.
`In summary, qPCR
` Can be adapted to quantify narrow (species) to broad range (phyla) targets within
`a microbiota
` Is often used as an adjunct, confirmatory tool to support other quantitative or
`semiquantitative methodologies
`
`DNA CHIPS TO SCREEN MICROBIOTA COMPOSITIONS
`
`Microarrays have been a mainstay of gene expression studies for many years; however,
`the technology is also amenable to phylogenetic analyses of microbial communities.
`The first phylogenetic analysis arrays were used to identify nitrifying bacteria from envi-
`ronmental samples,21 but the approach has subsequently been adapted and used,
`especially in Europe, to study the human microbiota.
`
`

`

`862
`
`Lawley & Tannock
`
`Arrays are created by attaching specific nucleotide probes to a solid substrate,
`often a glass slide. Many thousands of probes are bound to the surface in a grid
`pattern. The design of the probes is critical to the specificity and quantitative nature
`of the assay, as cross-hybridization may lead to overestimation of species abundance.
`DNA is extracted from the sample to be tested, and the 16S rRNA gene is amplified
`using universal primers. The forward primer is modified to incorporate a T7 promoter
`sequence upstream of the 16S rRNA-specific region. Following the amplification step,
`RNA copies of the amplicons are generated via the T7 promoter sequence. Modified
`0
`uridine-5
`-triphosphate (UTP) molecules are included for the in vitro transcription step
`and post-transcription labeling performed with a fluorescent molecule. The labeled
`RNA molecules are then hybridized to the array. Hybridization of fluorescent target
`molecules to the array probes is detected with a highly sensitive scanner, and
`platform-specific software is used to normalize signals, generate quantitative data,
`and identify which phylogenetic targets are present in the sample.
`The Human Intestinal Tract Chip (HITChip) was developed using a curated database
`of 16S rRNA gene sequences retrieved from the human bowel and generated probes
`to the V1 and V6 regions.22 Probes (4809 in total) were designed to target 1140
`different human microbiota phylotypes at various phylogenetic levels ranging from
`order (10% of probes) to genus (20% of probes) to species (60% of probes). The assay
`was validated by identifying 40 clones from human microbiota clone libraries and was
`shown to be reproducible across samples. The array was also shown to be capable of
`relative quantification by testing artificial mixtures of phylotypes, with an estimated
`detection limit of 0.1% abundance. The authors subsequently demonstrated the
`quantitative potential of the array by reporting good correlation of data derived from
`HITChip experiments with data from FISH and qPCR studies. Finally the HITChip plat-
`form was used to analyze the temporal dynamics of the microbiota of young and
`elderly adults over a 2-month period. The results showed that the microbiota of an
`individual was relatively stable over a short period of time, while subjects clustered
`according to age, suggesting a definable change in microbiota with age.
`In summary, phylogenetic microarrays, although of relatively limited usage so far:
` Provide rapid, relatively low-cost phylogenetic analysis of defined microbial
`communities
` Can target down to species level and be modified to incorporate new species as
`they are discovered
` Provide quantitative estimates of microbial groups
`
`PHYLOGENETIC ANALYSIS—DETAILED MICROBIOTA COMPOSITION
`
`Improved DNA sequencing technologies allow phylogenetic analysis of microbiota
`using numbers of samples and depth of coverage not previously considered feasible.
`Next-generation or high-throughput sequencing (HTS) platforms generate thousands
`of sequences across hundreds of samples in a single analytical run.23 To date, most
`microbiota studies have utlilized 1 of 2 platforms (Roche/454 [454 Life Sciences, Bran-
`ford, CT, USA] pyrosequencing or Illumina bridge amplification [Illumina Inc, San Diego,
`CA, USA] sequencing). Early studies focused on sequencing PCR amplicons from short
`regions of the 16S rRNA gene.24 Genome sequencing of hundreds of bacterial strains
`isolated from human feces has led to shotgun sequencing of microbiota DNA and
`phylogenetic inference from databases containing these whole genome sequences.25
`Phylogenetic analysis of community composition frequently requires amplification of
`all or part of the 16S rRNA gene. An informed choice of primers is a critical step. Poorly
`designed primers can bias amplification reactions in favor of, or against, entire bacterial
`
`

`

`Analysis of the Bowel Microbiota
`
`863
`
`groups. An example is the study by Palmer and colleagues,26 in which universal bacte-
`rial primers were used to analyze human infant microbiota from birth through the first
`year of life. This study concluded that bifidobacteria were a minor and likely unimpor-
`tant phylogenetic group (contrary to most other studies). In retrospect, it is likely that
`the forward amplification primer, which contained several mismatches to bifidobacte-
`rial 16S rRNA gene sequences, led to an under-representation of this genus.
`HTS chemistry is capable of only short sequence reads (100–500 bases); thus anal-
`ysis of a short highly informative hypervariable region is the best choice (the V3 and V4
`regions of 16S rRNA genes are commonly used).27 16S rRNA gene primers are
`0
`end, with adapter sequences (specific for each platform) and
`appended, at the 5
`a short unique barcode sequence (usually 8–12 bases). Individual barcoded primer
`sets are then used to amplify DNA extracted from each sample. Amplicons from
`multiple samples are then pooled in equimolar amounts, and the entire primer pool
`is sequenced on the platform of choice. Several detailed bioinformatics pipelines
`are available for down-stream data analysis, with most following a similar format:
`sequence quality control, splitting of multiplexed (barcoded) sequences into individual
`sample pools, alignment of sequences, choice of operational taxonomic units (OTUs,
`phylotypes, molecular species) at a chosen level of specificity, selection of represen-
`tative sequences from each OTU, discernment of taxonomic identity of each OTU, use
`of OTU tables to generate diversity (alfa, within sample, and beta, between sample),
`and other metadata-driven analyses.6,28,29
`The value of amplicon sequencing can be seen in a study aimed at describing
`a putative core fecal microbiota in lean and obese twins.30 The authors characterized
`the fecal microbiota of adult female monozygotic (n 5 31) and dizygotic (n 5 23) twin
`pairs, concordant for leanness or obesity, and their mothers (n 5 46), and looked for
`effects of genotype, environmental exposure, and host adiposity on the composition
`of the fecal microbiota. Analysis of nearly 2 million partial 16S rRNA gene sequences
`revealed a distinct decrease in diversity of microbiota from obese subjects along with
`a shift at the phylum level toward a microbiota with lower Bacteroidetes and higher
`Actinobacteria proportions. Two samples (separated by approximately 57 days)
`were collected from each participant, and sequence analysis showed that microbiota
`was more similar within an individual over time than between individuals. Also, families
`had more similar microbiota than unrelated individuals; however, there was no differ-
`ence in the degree of similarity between monozygotic and dizygotic twins. Evidence of
`a phylogenetic core within the fecal microbiota was not obtained since no OTU was
`present at greater than 0.5% abundance within any sample, and no OTU was present
`in all samples. In addition to the phylogenetic analysis, the authors performed a meta-
`genomic screen (see section on Metagenomics