REVIEW ARTICLESThe human gut microbiome:current knowledge, challenges,and future directionsMANEESH DAVE,PETER D.HIGGINS,SUMIT MIDDHA,and KEVIN P.RIOUXROCHESTER,MINN;ANN ARBOR,MICH;AND CALGARY,ALBERTA,CANADAThe Human Genome Project was completed a decade ago,leaving a legacy of pro-cess,tools,and infrastructure now being turned to the study of the microbes that re-side in and on the human body as determinants of health and disease,and has beenbranded‘‘The Human Microbiome Project.’’Of the various niches under investiga-tion,the human gut houses the most complex and abundant microbial communityand is an arena for important host–microbial interactions that have both local andsystemic impact.Initial studies of the human microbiome have been largely descrip-tive,a testing ground for innovative molecular techniques and new hypotheses.Methods for studying the microbiome have quickly evolved from low-resolution sur-veys of microbial community structure to high-definition description of composition,function,and ecology.Next-generation sequencing technologies combined withadvanced bioinformatics place us at the doorstep of revolutionary insight into thecomposition,capability,and activity of the human intestinal microbiome.Renewedefforts to cultivate previously‘‘uncultivable’’microbes will be important to the overallunderstanding of gut ecology.There remain numerous methodological challengesto the effective study and understanding of the gut microbiome,largely relating tostudy design,sample collection,and the number of predictor variables.Strategiccollaboration of clinicians,microbiologists,molecular biologists,computational sci-entists,and bioinformaticians is the ideal paradigm for success in thisfield.Meaning-ful interpretation of the gut microbiome requires that host genetic and environmentalinfluences be controlled or accounted for.Understanding the gut microbiome inhealthy humans is a foundation for discovering its influence in various important gas-trointestinal and nutritional diseases(eg,inflammatory bowel disease,diabetes,andobesity),and for rational translation to human health gains.(Translational Research2012;160:246–257)Abbreviations:GI¼gastrointestinal;IBD¼inflammatory bowel disease;IBS¼irritable bowelsyndrome;TRFLP¼terminal restriction fragment length polymorphism;UC¼ulcerative colitisFrom the Division of Gastroenterology and Hepatology,Mayo Clinic, Rochester,Minn;Division of Gastroenterology and Hepatology, Department of Internal Medicine,University of Michigan Medical Center,Ann Arbor,Mich;Division of Biomedical Statistics and Informatics,Department of Health Sciences Research,Mayo Clinic, Rochester,Minn;Department of Medicine,Division of Gastroenterology and Hepatology,Department of Microbiology and Infectious Diseases,Faculty of Medicine,University of Calgary, Calgary,Alberta,Canada.The authors have nofinancial disclosures relevant to this article.Submitted for publication December6,2011;revision submitted May 8,2012;accepted for publication May8,2012.Reprint requests:Kevin P.Rioux,University of Calgary,1863Health Sciences Centre,3330Hospital Drive NW,Calgary,Alberta,Canada T2N4N1;e-mail:kprioux@ucalgary.ca.1931-5244/$-see front matterÓ2012Mosby,Inc.All rights reserved.doi:10.1016/j.trsl.2012.05.003246Humans live in a biosphere where microbes are ubiqui-tous and have existed and evolved over3.8billion years.1 The intestinal tract of humans harbors a complex micro-bial community estimated to contain approximately100 trillion cells,exceeding the number of human cells by a factor of10.2-4This complex community of microbes in the alimentary tract(bacteria,archaea,eukarya,and viruses)is also known as the‘‘gastrointestinal(GI) microbiota.’’5,6Akin to the human genome,which is the sum of all human genes,the term‘‘microbiome’’refersto all of the microbiota in a defined microbial community,which are usually differentiated by their genetic elements.7Metagenomics refers to the functional and compositional analysis of an assemblage of microbes based on molecular study of their collective genomes.8,9 The gut microbiota performs a variety of beneficial functions(Table I),and given their importance in human health and disease,the Human Microbiome Project was launched by the National Institutes of Health to(1) characterize the microbial communities of various niches of the human body(eg,the nasal passages,oral cavity,skin,urogenital system,and GI tract),(2)to de-termine whether individuals share a core human micro-biome,and(3)to explore whether changes in the human microbiome cause or correlate with human disease.10,11 The distal GI tract contains the most abundant and diverse communities of microbes,in continuous interplay with the human host resulting in both local (mucosal and luminal)and systemic(metabolic and nutritional)effects.Therefore,of all microniches being explored by human microbiome researchers,the GI tract holds the most promise for discovery of important new concepts and understanding of the human‘‘superorganism’’and translation to clinical biomarkers and therapies.12There are excellent reviews that have focused on the temporal and spatial development of the human gut mi-crobiota13and their role in health and disease.14The goal of this article is to highlight the current state of knowledge of the human gut microbiome,discuss tech-nical and practical limitations of scientific devices and tools,and map out some of the knowledge gaps in this challengingfield that will be solved by thoughtful scien-tific approaches,advanced sequencing and computa-tional technology,and persistence.TECHNIQUES FOR ANALYZING MICROBIOTA Studies in the1970s using anaerobic culture–based techniques identified more than400to500distinct bac-terial species in the human gut.15Studying the human GI microbiota by cultivation methods has many draw-backs:It produces selective growth of some organisms and thus distorts composition of the natural community.Moreover,approximately60%to80%of gut microbes simply cannot be grown by conventional in vitro tech-niques.16,17In1977,Woese and Fox18described a tech-nique for molecular characterization of bacterial phylogeny based on ribosomal RNA sequence analysis. In particular,the16S rRNA is a molecule that is univer-sally present in bacteria and has highly conserved do-mainsflanking hypervariable sequences that can be used to distinguish bacterial groups.In the early 1980s,various culture-independent molecular tech-niques based on the16S rRNA became available,and during the last3decades they have been used extensively to assess the structure of complex or fastid-ious prokaryotic communities.1Examples of these techniques are terminal restriction fragment length polymorphism(TRFLP),denaturing gradient gel elec-trophoresis,andfluorescent in situ hybridization.19 These bacterial community profiling or‘‘fingerprint-ing’’techniquesfirst involve isolating bacterial DNA from environmental or biological samples.The DNA is amplified by polymerase chain reaction using univer-sal primers that target conserved regions of the16S rRNA gene,and the resulting amplicons contain vari-able regions that discern the constituent members of bacterial communities by electrophoretic or hybridiza-tion techniques.Cloning and then sequencing of the16S rRNA gene in an automated capillary sequencer is a higher-resolution method of studying bacterial phylogeny.20This tech-nique uses Sanger sequencing to produce a long read ( 800base pairs)of the16S rRNA gene,which enables identification of bacteria at a higher-level phylogenetic resolution(ie,genera and species).This relies on robust bioinformatics tools,such as the Ribosomal Database Project.21In the wake of the Human Genome Project,next-generation sequencing technologies have emerged that have increased the depth and speed of phylogenetic cov-erage and decreased the cost through massively parallel sequencing methods.There are various commercial platforms for next-generation sequencing.One of these newer techniques is pyrosequencing that identifies nu-cleotides by the amplitude of light signals generated when luciferin is converted to oxyluciferin during Table I.Biological effects of the gut microbiota on human hostDevelopment of innate and adaptive immunityIntestinal epithelial integrityEnergy sourceVitamin biosynthesis,bile salt transformation,catabolism of dietary glycans(eg,cellulose and pectins)Barrier to colonization by microbial pathogensXenobiotic metabolismTranslational ResearchVolume160,Number4Dave et al247DNA synthesis.22A limitation of this technique in its first iteration was short sequence reads of100to250 base pairs,which have been quickly surpassed by newer methods.There are several companies that manufacture these machines(eg,Roche454[Roche Diagnostics Corp,Indianapolis,Ind],Illumina/Solexa[Illumina Inc,San Diego,Calif],Applied Biosystems,Foster City,Calif).Although the16S rRNA gene sequences obtained by Sanger or next-generation sequencing en-able greater depth and resolution to identify and discern taxa,the identity and thus functional importance of community members may be missed because of a sub-stantial proportion gut microbes match ribosomal se-quences from uncultured bacteria.Also,there is currently no clear consensus on what constitutes a bacte-rial species at a molecular level,23so alternative methods of describing‘‘phylotypes’’are based on oper-ational taxonomical units,groups of sequences that are highly similar to a single reference sequence.24 Despite the enormous progress made in determining microbial community structure using the16S rRNA gene,the basic technique provides no information about bacterial physiology and ecological significance.1An-other novel approach to assess the diversity of microbes is the whole genome shotgun sequencing of community DNA.25This differs from16S RNA sequencing tech-niques in that entire genomes of microbes are sequenced and compared with previously characterized genes to build a picture of functional capability of the micro-biota.This approach helps not only with identification of rare species in a complex community but also with identification of microbial genes that code for metabolic or biologic functions.12The drawback with this tech-nique is that a large amount of DNA is needed(although it can be overcome by whole genome amplification), contamination with host DNA is problematic,and many genes are identified that do not yet have a known function.Hamady and Knight26have published a com-prehensive review of various sequencing techniques used in microbiome investigation.SEQUENCE DATA ANALYSIS AND CHALLENGESAs mentioned earlier,there are2main approaches to surveying the microbiome,16S rRNA sequencing and whole genome sequencing(Fig1).The former uses a small representative region as a marker or proxy and involves evaluating the informative sites of any of the 16S rRNA variable regions.27Similar organisms are represented by clustered reads and used to infer the phy-logenetic and taxonomic identity.The proportion of var-ious types of organisms is used to infer the structure of microbial communities using statistical analysis.The latter approach of assembly involves splitting the DNA into smaller pieces and sequencing them to later assemble into longer contigs.Those are then used to in-fer functional and metabolic pathway information.28 There are huge challenges in analyses of these data. All the next-generation sequencing technologies have improved vastly but still have higher error rates than tra-ditional Sanger sequencing.There are also systematic sources of error depending on the sequencing instru-ment used that can yield noisy data.29,30Qiime31and Mothur32are popular analysis tools for microbial se-quencing data,but there are many pre-and post analysis steps that have a bearing on the eventual results.It is thus vital to have quality checks and data-cleaning steps along the analysisflow to ensure the validity of the end results.The alignment of these short reads,especially to similar sequences,is error prone.A one-off read sup-porting evidence for a particular rare organism or group should thus be treated with caution.Some studies have come up with a leave-one-out analysis to recommend a minimum number of sequence-reads mapping to a par-ticular feature to be confident of the result.33These anal-yses are continually improving as the availability of complete microbial genomes increases.This would mean a larger number of reference genomes that can be used for accurate and efficient bioinformatics analy-sis and better functional interpretation of data.The current tools use traditional statistical techniques with some refinements that do not take into account the inherent complexity in biological systems and are lim-ited by the assumption that predictor variables(ie,gut microbes)are independent of each other.This is clearly not the case,because the human gut microbiota is a com-plex community whose constituents are linked together through complex homeostatic interactions.The solution to deciphering this complexity could be through the use of machine-learning algorithms that include neural net-works,support vector machines,and decision trees.Ma-chine learning is inherently more suited to the study of complex microbial communities because it uses pattern recognition,does not need identification of predictor variables in advance,and becomes increasingly better at prediction with larger volumes of data.34Another problem facing investigators is the huge volume of data that sequencing generates,which can overwhelm available computing resources of an institution.Cloud computing uses the storage and processing capabilities of a network of remote servers hosted on the internet, providing individual investigators paid access to power-ful computing resources suited to the massive data gen-erated by current sequencing methods.The Cloud BioLinux and Amazon Elastic Compute Cloud(http:// /ec2/)are examples of2currently available commercial resources for metagenomic anal-yses.35In addition,the Data Intensive Academic GridTranslational Research248Dave et al October2012is a free computational cloud sponsored by the National Science Foundation and available to researchers in aca-demic and nonprofit institutions.SAMPLING THE GUT MICROBIOTAAlthough the same bacterial phyla predominate in the stomach,small intestine,and colon,their relative com-position and abundance varies considerably.36Meth-odologic factors likely account for some of this variation,including patient selection,sample handling,and choice of molecular and bioinformatics techniques.Next,we provide data on some of the issues that may be confounding human gut microbiota studies.These con-cepts are summarized in Table II .1.What constitutes a normal,healthy human micro-biota?It has long been apparent that the intestinal micro-biota varies significantly from one individual to another,making it impossible to define the ‘‘normal’’or ‘‘healthy’’human gut microbiota.Indeed,it is difficult to define what constitutes a healthy human,and as-sumed absence of overt disease may not be the most re-liable indicator of a healthy microbiota for study purposes.Although it will be a difficult task to accom-plish,future studies could use a set of phenotypic char-acteristics,for example,the Adult Fitness Test launched as a part of President’s Challenge (http://www.adultfi/).This test incorporates aerobic fit-ness,muscular strength,endurance,flexibility,and body composition to determine overall health-related fitness.A ‘‘core microbiota’’has been proposed,37but it re-mains unclear what the essential constituents are,and,instead,a set of core functions of the microbiota may well emerge as the correct unifying concept.38Age has well-described influences on the fecal microbiota,for example,with substantial differences being de-scribed among healthy infants,adults,and the el-derly.39-42Gender differences have also been described.40On a collective scale,it is also apparent that the human gut microbiota differs between various ethnic groups,which likely reflects both genetic and en-vironmental influences.40Environmental influences that likely affect the GI microbiota are wide-reaching,dy-namic,and difficult to control for in the context of human studies.These include diet,medications (espe-cially antibiotic exposure,even if remote),stress,smok-ing,and GI infections.Early studies suggested that the intestinal microbiomes of healthy adults were stable over time,43,44but such studies generally used fingerprinting techniques with low sensitivity and relatively few points of observation over a short period of time.More recent studies using deep sequencing techniques have shown a surprising degree of species-level variability within an individual’s fecalFig rmatics analysis.Translational Research Volume 160,Number 4Dave et al 249microbiota over a short time span of days to weeks,with an individual core microbiota of perhaps only 10%re-maining unchanged in the long term.45This has obvious implications for basic and clinical studies in humans,because our new understanding of the human gut micro-biome as a moving target makes it difficult to ascribe significance to shifts in the microbiome in the presence of significant background temporal variation.2.Is fecal microbiota representative of the resident microbiota of the distal GI tract?A majority of the studies of the human gut micro-biome have analyzed the microbiota of human fecal specimens,given the ease of acquisition,minimal cost,and avoidance of invasive procedures such as colo-noscopy with requirement for fasting and laxative prep-aration.It is widely believed,however,that the highly evolved biofilm communities that are closely associated with the intestinal epithelium are biologically more rel-evant than planktonic microbes that exist in the lumen of the gut or associated with food residue,and,moreover,fecal specimens may not accurately portray the mucosal microbiota.Even within stool,there appears to be sig-nificant differences between microbiota of the liquid fraction compared with that associated with the solid phase (insoluble plant residue and mucin).46Lepage et al 47used a bacterial small subunit RNA fingerprinting technique to demonstrate that the dominant bacterial groups differ between fecal and mucosa-associated mi-crobiota in patients with inflammatory bowel disease (IBD)and healthy controls.In a deep sequencing study of 3healthy individuals,mucosal tissue samples from the cecum,ascending colon,transverse colon,descend-ing colon,sigmoid colon,and rectum were compared with fecal samples taken 1month after colonoscopy.48This study showed that the mucosa-associated micro-biota was different from stool microbiota within an in-dividual,and the authors hypothesized that the fecal microbiota represents a combination of shed mucosal bacteria and the nonadherent luminal population.In ad-dition,the investigators showed that mucosa-associated microbiota had a patchy and heterogeneous distribution within a subject.Rather than a subset of the fecal micro-biota,the mucosa-associated microbiota is composi-tionally distinct from that of stool.49Marteau et al 50compared cecal luminal contents (obtained by fluoro-scopic placement of an orocecal tube)with fecal sam-ples using both culture-based and culture-independent techniques,showing that the fecal flora was different from the cecal flora both qualitatively and quantita-tively.3.What is the optimal method of sampling mucosa–associated microbiota?In view of the limitations of studying stool specimens,investigators have turned to mucosal biopsies of the GI tract collected by endoscopic means as the source spec-imens for microbiome studies.However,endoscopic bi-opsies also have important limitations because fasting and colon cleansing are common prerequisites of endo-scopic biopsy collection,and tissue specimen them-selves are contaminated by luminal microbes in the process of endoscopic collection.During colonoscopy,for example,biopsy forceps are advanced and then with-drawn via the same channel used to suction stool resi-due.To determine the degree of contamination ofTable II.Technical and methodological challenges to studying the human gut microbiotaTechnicalContamination during sample collectionBiases due to varying efficiency of bacterial nucleic acid extractionInherent polymerase chain reaction biases leading to under-representation or failed detection of minor phylotypes 16S rRNA gene does not accurately reflect bacterial abundance (varying copy numbers)Sequencing errorsMolecular techniques cannot discern microbes as dead or alive,autochthonous or allochthonousMethodologicalBroad definitions of healthy human subject Lack of correlative host genotype dataConfounding influences of diet and drugs largely unaccounted forEffect of fasting and pre-colonoscopy bowel cleansing on relevant microbiotaValidity of mucosal biopsies as representation of ‘‘mucosa-associated’’microbiota Deep sequencing vs fingerprinting techniques How is operational taxonomic unit defined Moving target:temporal instabilityMajority of metagenomic studies to date have not yet assessed functional aspects of microbiome Statistical tools for devising adequately powered clinical studies are limitedTranslational Research250Dave et alOctober 2012biopsies during endoscopic collection,we deviseda study comparing standard and sterile sheathed biopsyforceps.51Sheathed forceps were constructed by cover-ing standard endoscopic biopsy forceps with polyethyl-ene tubing.Paired biopsies using standard and sheathedforceps were obtained from the terminal ileum of sub-jects,and mucosa-associated bacteria were character-ized by16S-rDNA polymerase chain reaction andTRFLP.Our study showed no difference in the micro-biota between specimens collected with standard or ster-ile sheathed biopsy forceps within individual subjects.In a unique study by Mai et al,52the fecal microbiotawas studied before and up to8weeks after colonoscopyin a small series of healthy individuals undergoing coloncancer screening.In some patients,there were persistentalterations in the fecal microbiota after colonoscopy,presumably due to the effects of fasting and the unspec-ified colon-cleansing agent.No similar studies havebeen published to replicate these preliminary data orto extend thefindings to a variety of common-use bowelpreparations or to the mucosa-associated microbiota.4.Do sample storage conditions and method of DNAextraction matter?As part of ongoing microbiome studies,samples arebeing collected from subjects and archived in biobanks.A relevant concern is whether storage conditions affectthe diversity of gut microbiota.A16S rRNA pyrose-quencing study showed that short-term storage of upto14days at20 C,4 C,220 C,or280 C did not sig-nificantly affect the microbial composition.53Likewise,Wu et al54showed that there was no significant differ-ence in bacterial composition determined by pyrose-quencing in fecal samples immediately frozen at 280 C or those stored on ice for24or48hours.54Other investigators have revealed concern about the stabilityof stool specimens stored at room temperature,particu-larly in excess of12hours.55The exact method of DNAextraction also seems to be important,with an initialbead-beating step improving the performance of bothphenol-based protocols54and proprietary DNA extrac-tion kits.56Bead beating refers to mechanical disruptionof bacterial cell walls by vigorous mixing with tinyglass beads and improves extraction of cytosolic com-ponents from difficult to lyse bacteria including Firmi-cutes,which are a major component of the human gutmicrobiota.COMPOSITION OF THE GUT MICROBIOTAIt is estimated that up to80%of the bacteria in the hu-man gut cannot be cultivated by conventional tech-niques,largely because of their fastidious nutrient andanaerobic requirements and their complex dependence on one another.17Molecular techniques have signifi-cantly advanced our understanding of the constituentmembers of the human microbiome and have largelysupplanted cultivation-based techniques for study ofthe gut microbiome.The major members of bacterialcommunities in various segments of the human GI tractare illustrated in Figure2.The most diverse and abun-dant microbial communities generally occur in theoral cavity and distal GI tract and have been the focusof most studies,because they are most amenable to sam-pling.However,the relatively simple indigenous micro-biota of the human esophagus,stomach,and smallbowel are relatively unstudied,and these niches presentthe added challenge of trying to discern allochthonousbacteria(‘‘passersby’’from upstream niches that arelikely irrelevant)from autochthonous microbes(repre-senting stable,functionally relevant community‘‘resi-dents’’).The oral cavity contains members of thephyla Firmicutes,Proteobacteria,Bacteroidetes,Acti-nobacteria,and Fusobacteria,which account for99%of all phyla present.The rare phyla belong to SR1,TM7,Cyanobacteria,Spirochaetes,Tenericutes,andSynergistetes.57A pyrosequencing study of samplesfrom distal esophagus showed the presence of membersof6phyla,Firmicutes,Bacteroides,Actinobacteria,Proteobacteria,Fusobacteria,and TM7.The bacteriaStreptococcus,Prevotella,and Veillonella were foundto be the most prevalent,and the community of the distalesophagus was similar to that of the oral cavity58;how-ever,the majority of esophageal organisms could be cul-tivated,unlike those of the oral microbiome.Given theextremely low pH in the stomach,it was originally be-lieved that,with few exceptions(eg,Helicobacter py-lori),the stomach harbored a few transient microbial species and did not have a complex community likethe distal GI tract.3However,in a study of gastric biopsyspecimens from23human subjects using the16S rRNAgene clone library construction,Bik et al12showed thatthere is a diverse community of gastric microbes domi-nated by members of phyla Proteobacteria,Firmicutes,Actinobacteria,Bacteroidetes,and Fusobacteria.Thismicrobial community was significantly different thanthe oral and esophageal community.Another interestingfeature of this study was that the composition of the bac-terial community was not apparently altered by the pres-ence of H.pylori.The microbiota of the human small bowel is relativelyunstudied.By using TRFLP to study samples obtainedfrom3individuals during autopsy,one group showedthat the microbiota of the jejunum and ileum was lesscomplex than that of the colon.The jejunal and ilealsamples contained more facultative anaerobes and noClostridium coccoides and leptum subgroups,in con-trast to cecum and rectosigmoid samples.59Translational ResearchVolume160,Number4Dave et al251Microbial abundance is greatest in the colon with ap-proximately 1011and 1012microbial cells per gram of stool,and the number of microbial genes is 2orders of magnitude greater than that contained by the human host genome itself.2Recent studies of human fecal and colonic biopsy specimens have revealed members of 9distinct phyla (Firmicutes,Bacteroidetes,Actinobacte-ria,Fusobacteria,Proteobacteria,Verrucomicrobia,Cy-anobacteria,Spirochaetes,and VadinBE97).5,48The phyla Firmicutes and Bacteroides are the dominant phyla in the colon.10,38,48In contrast,in the total biosphere there are up to 70bacterial phyla,which highlights the fact that the human GI microbiota is restricted to a small subset of phyla,but with a high degree of species richness and abundance within these core constituent phyla.By using the Illumina Genome analyzer platform,the Metagenome of Human Intestinal Tract Consortium as-sessed the metagenome of fecal specimens obtained from 124healthy patients,overweight patients,obese patients,and patients with IBD from Denmark and Spain.38This is the most comprehensive study of human microbiome from a large group of subjects to date.The investigators were able to estimate that human gut mi-crobiota harbors approximately 1150bacterial phylo-types or species.In addition,they noted 536,112unique genes in each sample,of which 99%were bacte-rial.Approximately 40%of the bacterial genes from each individual were shared with at least half of the in-dividuals of the cohort.These highly conserved genes are involved in degradation and digestion of complex sugars,production of short chain fatty acids,and bio-synthesis of vitamins.Although significant interindivid-ual differences exist in the composition of the gut microbiome of humans,there is an essential functional-ity of gut microbes that support human health and nutri-tion.Thus,the concept of the ‘‘human superorganism’’has emerged,a physiology and homeostasis that are the complex sum of human and microbial gene expres-sion.10One of the goals of the human microbiome project is to ascertain whether we have a core group of microor-ganisms that are shared between individuals.Evidence from recent studies suggests that,rather than acoreFig 2.The dominant phyla in various locations in the human GI tract.By permission of Mayo Foundation for Medical Education and Research.All rights reserved.(Color version of figure is available online.)Translational Research252Dave et alOctober 2012。
