Proc. Nati. Acad. Sci. USAVol. 91, pp. 1609-1613, March 1994Microbiology

Remarkable archaeal diversity detected in a Yellowstone NationalPark hot spring environment

(archaebacteria/phylogeny/thermophfly/molecular ecology)

SUSAN M. BARNS, RUTH E. FUNDYGA, MATTHEW W. JEFFRIES, AND NORMAN R. PACE*Department of Biology and Institute for Molecular and Cellular Biology, Indiana University, Bloomington, IN 47405

Contributed by Norman R. Pace, November 17, 1993

ABSTRACT Ofthe three primary phylogenetic domainsArchaea (archaebacteria), Bacteria (eubacteria), and Eucarya(eukaryotes) -Archaea is the least understood in terms of itsdiversity, physiologies, and ecological panorama. Althoughmany species of Crenarchaeota {one of the two recognizedarchaeal kingdoms sensu Woese [Woese, C. R., Kandler, 0. &Wheelis, M. L. (1990) Proc. Nadl. Acad. Sci. USA 87, 4576-4579J} have been isolated, they constitute a relatively tight-knitcluster of lineages in phylogenetic analyses ofrRNA sequences.It seemed possible that this limited diversity is merely apparentand reflects only a failure to culture organisms, not theirabsence. We report here phylogenetic characterization ofmanyarchaeal small subunit rRNA gene sequences obtained bypolymerase chain reaction amplification of mixed populationDNA extracted directly from sediment of a hot spring inYellowstone National Park. This approach obviates the needfor cultivation to identify organisms. The analyses documentthe existence not only of species belonging to well-characterizedcrenarchaeal genera or families but also ofcrenarchaeal speciesfor which no close relatives have so far been found. The largenumber of distinct archaeal sequence types retrieved from thissingle hot spring was unexpected and demonstrates that Cre-narchaeota is a much more diverse group than was previouslysuspected. The results have impact on our concepts of thephylogenetic organization of Archaea.

Microbiologists have long understood the limitations of cul-tivation techniques in assessing the diversity of naturallyoccurring microbial communities. Commonly, only a smallfraction of organisms observed microscopically can be cul-tivated using standard methods. Recently, sequence-basedphylogenetic techniques have been used to alleviate therequirement for cultivation to identify microorganisms. Suchstudies have detected the presence of previously unknownorganisms in each instance of their use (for review, see ref.1). These techniques sample microbial populations directlythrough isolating and sequencing specific genes from theenvironment. Phylogenetic comparative analysis of thesesequences is then used to determine evolutionary relation-ships between members of the community and cultivatedspecies. The results allow inference of some properties ofotherwise unknown organisms in the environment, based onthe properties of their studied relatives. In addition, thesequences can be used to design oligonucleotide probes fordetermination of morphotype and abundance of particularorganisms and for assistance in cultivation efforts.We have employed molecular phylogenetic techniques to

investigate the diversity of Archaea in a hot spring in Yel-lowstone National Park. Small-subunit rRNA genes wereamplified by polymerase chain reaction (PCR) from DNAextracted directly from sediment, by using primers designed

to amplify archaeal and eucaryal genes selectively. Amplifi-cation products were then cloned and the nucleotide se-quences of the inserts were determined.t rDNA sequencesobtained were aligned with and compared to an extensivedata base of rRNA sequences from cultivated species. Thisreport describes the recovery, from this single hot spring, ofrDNA clones from a remarkable variety of archaeal types,many of them crenarchaeal species with no known closerelatives.

MATERIALS AND METHODSBiomass Collection and DNA Extraction. Samples of the

upper 1-10mm of sediment were collected, frozen on dry ice,and stored at -70'C until processed. Nucleic acids wereextracted from sediment samples by a direct lysis procedureadapted from several methods (2-6) and designed to obtainDNA from a broad range of cell types. Approximately 5 mlof sediment was resuspended in buffer A (500 mM Tris-HCl,pH 8.0/100 mM NaCl/1 mM sodium citrate) in the presenceof polyadenosine (100 pg/ml) and lysozyme (5 mg/ml) andincubated for 1 hr at 370C with occasional agitation. Protein-ase K was then added to 2 mg/ml, and the mixture wasincubated for a further 30 min. At the end of incubation, 8 mlof lysis buffer [200 mM Tris HCl, pH 8.0/100 mM NaCl/4%(wt/vol) SDS/10% (wt/vol) 4-aminosalicylate] was added,and the solution was mixed gently by inversion. Three cyclesof freezing in a dry ice-ethanol bath and thawing in a 650Cwater bath were conducted to release nucleic acids. Themixture was then extracted with an equal volume of phenol[saturated with 100 mM Tris-HCl (pH 8.0)], followed byextraction with phenol/chloroform/isoamyl alcohol, 24:24:1(vol/vol). Four grams of acid-washed polyvinylpolypyrroli-done (PVPP) (Sigma) (6) was added to the aqueous phase, andthe mixture was incubated 30 min at 37TC. The PVPP waspelleted from the mixture by centrifugation, and the resultantsupernatant was filtered through a 0.45-gm (pore size) filterto remove residual PVPP. Bulk nucleic acids were precipi-tated from solution with isopropyl alcohol and centrifugation.The resulting pellet was resuspended in 500 jLd ofTE (10 mMTris HCl, pH 8.0/1 mM EDTA), 0.1 g of anhydrous ammo-nium acetate was added, and the solution was mixed quicklyand then centrifuged immediately for 30 min at 40C in amicrocentrifuge. Nucleic acids were precipitated from thesupernatant by addition of 1 vol of isopropyl alcohol, incu-bated on ice for 10 min, and centrifuged for 30 min. Afterresuspension in TE, high molecular weightDNA was isolatedfrom the extract by purification on Sephadex G-200 (Phar-

Abbreviation: RDP, Ribosomal Database Project.*To whom reprint requests should be addressed.tThe sequences reported in this paper have been deposited inGenBank data base (accession nos. pJP 6, L25306; pJP 7, L25307;pJP 8, L25309; pJP 9, L25308; pJP 27, L25852; pJP 33, L25300; pJP41, L25301; pJP 74, L25302; pJP 78, L25303; pJP 81, L25304; pJP89, L25305).

1609

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

Proc. Natd. Acad. Sci. USA 91 (1994)

macia) columns as described (2) or by size selection on 1%low-melting-point agarose gels (SeaPlaque GTG, FMC), ac-cording to the manufacturer's protocols.PCR Amplition of rDNA. Bulk DNA from Sephadex

fractions or in agarose gel slices was titrated in amplificationreaction mixtures to empirically determine the optimal DNAconcentration for maximum synthesis of 1- to 1.5-kb prod-ucts. rRNA genes were amplified by PCR under conditions asdescribed (7), with inclusion of acetamide to 5% (wt/vol) andthe substitution of a Tricine-containing buffer (300 mMTricine, pH 8.4/500 mM KCI/15 mM MgCl2) (8) for the lOxreaction buffer. After an initial 5-min denaturation at 94TC,during which the DNA polymerase was added, thermalcycling conditions were as follows: denaturation at 940C for1.5 min, annealing at 55TC for 1.5 min, and extension at 720Cfor 2 min, repeated for a total of 40 cycles. The oligonucle-otide primer sequences used were 1391R (9) (5'-GACG-GGCGGTGTGTRCA-3') and 23FPL (5'-GCGGATCCGCG-GCCGCTGCAGAYCTGGTYGATYCTGCC-3'), where R isa purine and Y is a pyrimidine.

Purifiction and Clning of PCR Products. Amplified DNAfrom 5 to 10 reaction mixtures was pooled, heated to 650C tomelt agarose if necessary, then extracted sequentially withphenol/chloroform/isoamyl alcohol and chloroform/isoamylalcohol, and precipitated with ethanol. After centrifugation,DNA pellets were resuspended in TE and products of theexpected size (1.4 kb) were purified on 4% polyacrylamidegels, eluted into 300 mM sodium acetate/0.1% SDS, andprecipitated with ethanol. Products were then digested withrestriction endonuclease Not I (New England Biolabs) (10),purified on low-melting-point agarose gels (1% SeaPlaque,FMC), and cloned into pBluescript KS+ (Stratagene), allaccording to manufacturers' directions. Clones containing1.4-kb inserts were identified by SDS/agarose gel electro-phoresis of overnight cultures (11).

Sequencing and Phylogenetic Analysis. DNA from plasmidpreparations of clones (10) was denatured and sequenced bythe dideoxynucleotide chain-termination method using Se-quenase 2.0 (United States Biochemical) by the manufactur-er's recommendations. Universal rRNA-specific (9) and M13forward and reverse primers (10) were used in sequencing theclones. The following four archaeal-biased sequencing prim-ers were also designed and used for sequence analysis:340RA, 5'-CCCCGTAGGGCCYGG-3'; 744RA, 5'-CCS-GGGTATCTAATCC-3'; 765FA, 5'-TAGATACCCSSG-TAGTCC-3'; 1017FA, 5'-GAGAGGWGGTGCATGGCC-3'(primer numbering corresponds to the Sulfolobus acidocal-darius nucleotide to which primer 3' nucleotide is comple-mentary).Sequences were manually aligned with rRNA sequence

data from the Ribosomal Database Project (RDP) (12) basedon primary and secondary structural considerations, by usingthe GDE multiple sequence editor distributed by the RDP (12).Sequences were submitted to the CHECK-CHIMERA programof RDP to detect the presence of possible chimeric artifacts(12, 13). Phylogenetic analyses were restricted to nucleotidepositions that were unambiguously alignable in all sequences.Least-squares distance matrix analyses were performed us-ing the algorithm ofDeSoete (14), with correction for multipleundetected mutations (15). Neighbor joining analysis wasaccomplished using the PHYLIP package (version 3.5; ob-tained from J. Felsenstein, University of Washington, Seat-tle), while parsimony trees were constructed using PAUP(version 3.1.1; obtained from D. L. Swofford, SmithsonianInstitution). Maximum likelihood analyses were performedusing FASTDNAML (version 1.0; distributed by RDP; ref. 12).Bootstrap methods (16) were used to provide confidenceestimates for tree topologies in neighbor joining, parsimony,and maximum likelihood methods. To avoid potential biasintroduced by order of sequence addition, taxon addition

order was randomized in neighbor joining, parsimony, andmaximum likelihood analyses.

Location and Chemical Analysis ofHot Spring. The locationof the hot spring analyzed in this study was determined usinga PYXIS IPS-360 global positioning system receiver (Sony).Ten readings at 30-s intervals were taken on 2 days andaveraged to give position of site. Water and sediment sampleswere analyzed for chemical composition and pH by WWAnalytical Sciences (Cleveland, TN) and by H. Huber (Uni-versity of Regensberg, Germany).

RESULTSThe hot spring analyzed in this study, "Jim's Black Pool," islocated in the Mud Volcano area of Yellowstone NationalPark, Wyoming, "-0.75 miles (1 mile = 1584 m) southsouth-west of Black Dragon's Cauldron, at 440 36' 35.4" ± 1.3" Nand 110° 26' 20.6" ± 1.3" W. The pool of the spring isapproximately 3 x 9 m in size, with several boiling sourceareas (930C). The water and sediment of the spring are deepblack in color due to a fine black particulate material,obsidian sand, and possibly iron sulfide, which also accumu-lates on the periphery of the pool. Chemical analysis ofwaterand sediment samples taken from this pool indicates that it issimilar in chemistry to other Yellowstone hot springs (see ref.17), although the sediment contains an unusually high ironcontent (415,600 mg/kg). Sulfide is present. Temperaturevaries across the pool and increases rapidly with increasingdepth through the sediment. The sediment was =740C at thesite of sampling, while pH was approximately neutral (waterpH = 6.7 at 18TC and sediment pH = 7.6 at 20TC).DNA was isolated directly from sediment samples, and

rRNA genes were amplified by PCR and cloned. One of theprimers for amplification, 23FPL, was chosen to amplifyarchaeal and eucaryal small subunit rRNA genes selectively,to the exclusion of bacterial genes. Use of this primer, inconjunction with the "universal" 1391R primer (9), resulted inthe recovery of archaeal rDNA clones almost exclusively. Asingle bacterial-type clone was retrieved (out of98 screened),in addition to 12 clones containing a common sequence thatdoes not appear to be ribosomal (data not shown). No eucaryalrDNAs were detected. The high G+C content ofthe bacterialand nonribosomal sequences may have allowed their amplifi-cation by the 23FPL primer, itself w68 mol% G+C. Second-ary structure models ofthe cloned rDNAs used in this analysisindicate that they derive from functional rRNA genes.

Initially, m250 nt of sequence was obtained from each ofthe 98 insert-containing clones. This information was used toidentify unique sequence types. Clone types pJP6, pJP27, andpJP8 were the most common (23, 8, and 7 occurrences,respectively), while the remaining types were represented1-4 times in the collection. This distribution may reflect thepopulation present in the original sample (other investigatorshave suggested correlation between prevalence of rDNAclone types and population structure; refs. 1 and 18); how-ever, the potential selectivity of DNA recovery, amplifica-tion, and cloning prohibits confidence in such quantitation.An additional six clones contained inserts highly similar insequence to that of pJP27, but differing in 1-4 positions (outof =250 nt sequenced) from that clone. Similarly, a sequence-98% similar (over =250 nt of sequence) to that ofclone pJP9was found in 4 clones of the library (data not shown). Suchmicroheterogeneity in rRNA gene clones obtained frompure-cultured organisms as well as from mixed natural pop-ulations has been observed previously (for review, see ref. 1)and may be attributable to allelic variation within or betweenmembers of the population, incorporation errors by polymer-ases, and/or cloning and sequencing errors. That such het-erogeneity was not observed in the duplicates of the otherclone types suggests that these pJP27- and pJP9-related

1610 Microbiology: Barns et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

Proc. Natl. Acad. Sci. USA 91 (1994) 1611

sequences were not generated by PCR, cloning, or sequenc-ing errors and probably arose from distinct but closely relatedtemplate types in the original population.The 5'-terminal 450 nt of sequence from each unique clone

type was then determined and used for approximate phylo-genetic analyses. Sequences of clones pJP6, pJP7, pJP8,pJP9, pJP74, and pJP81 showed high similarities with 16SrRNA sequences of cultivated Archaea (similiarity values of0.89-0.98 to closest cultured relatives) but were not identicalto any. Phylogenetic analysis of these sequences by maxi-mum likelihood, neighbor joining, and maximum parsimonyall resulted in trees with similar topologies (Fig. 1A). Thespecific relationships of the sequences of pJP7 and pJP74with those of Desulfurococcus mobilis, Pyrodictium occul-tum, and Sulfolobus acidocaldarius were not resolved in thisanalysis, as indicated by low bootstrap values for thesebranches. The sequences of clones pJP27, pJP41, pJP78, andpJP89 bore no close similarity to those of any Archaea yet

A

cultured (similarity values less than 0.87). To infer moreaccurately the phylogenetic affiliation of the organisms rep-resented by these rDNAs, nearly full-length sequences(.1330 nt) were determined. Analyses of these data by allthree phylogenetic methods showed that the sequences clus-ter specifically with known Crenarchaeota (Fig. 1B).The phylogenetic placement of the sequences represented

in Fig. 1B was further assessed by nucleotide signatureanalysis and determination of the sensitivity of tree topologyto sequence selection and base composition. An intradomainnucleotide signature analysis (19) of these sequences is givenin Table 1. In agreement with the inferred phylogenetic trees,pJP33, pJP41, and pJP89 all share a majority of sequencesignature features with Crenarchaeota. The rDNA sequencesof clones pJP27 and pJP78, however, have about as manysignature features in common with Euryarchaeota as withCrenarchaeota (8 vs. 6 features). This suggests that theselineages branch deeply in the archaeal tree, perhaps suffi-

Desulfurococcus mobilispJP74

Sulfolobus acidocaldariuspJP7

occultum pNp 889, Pyrobaculum islandicumP 7Pyrobaculum aerophilumThermoproteus tenax

99-pJP6NP1 6Thermofilum pendens

-pJP 81

dleri,occus celer__L10~0~ Archaeoglobus fulgidus

Npp 9.10

B - Pyrodictium occultum55 671 Sulfolobus acidocaldarius

88 Thermoproteus tenaxThermofilum pendens

pJP 33pJP 41

pJP89pJP 78pJP27

Methanopyrus kandleri69 | Thermococcus celer66 Methanococcus vannielii

80 Methanobacterium thermoautotrophicum82 901 Methanosarcina barkeri

Haloferax volcanli'°°I Aquifex pyrophilus100 Thermotoga maritima

.10

FIG. 1. (A) Phylogenetic tree of archaeal rDNA gene clones obtained from Jim's Black Pool, illustrating close affiliations of these sequences(designated pJP) with those of cultivated Archaea. Tree was inferred by maximum likelihood analysis of 397 homologous positions of sequencefrom each organism or clone. Scale bar represents 10 mutations per 100-nt sequence positions. The percentage of 100 bootstrap resamplingsthat support each topological element in maximum likelihood (above line) and maximum parsimony (below line) analyses is indicated. No valuesare given for groups with bootstrap values less than 50%. [Note: The sequence of Sulfolobus acidocaldarius rRNA used in this analysis wasoriginally reported as having derived from Sulfolobus solfataricus. Recent DNA hybridization and sequence data suggest that the sequence iscorrectly attributed to Sulfolobus acidocaldarius (P. Dennis, personal communication).] (B) Phylogenetic tree of archaeal rDNA gene clonesinferred from maximum likelihood analysis of 1170 homologous positions of sequence from each organism or clone. Scale bar represents 10mutations per 100-nt sequence positions. The percentage of 100 bootstrap resamplings that support each topological element in maximumlikelihood (above line) and maximum parsimony (below line) analyses is indicated.

Nficrobiology: Bams et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

Proc. Natl. Acad. Sci. USA 91 (1994)

Table 1. Intradomain nucleotide signature analysis forpJP sequences

Position(s)27 55628 5553055334 550289-311501-544503 542504-541513-538518658-7476929651074*10831244-12931252

CrenC.GC-GGCCOGG-CCGGCG-yU-AU

GCC

GGNUR-YC

Eury

GCG-yY-RU.GC.GR-YC-GY-RC.GC

Y-RU

yACY-RU

Gp. I

CrenCrenCrenEuryEuryEuryCrenEuryCrenCrenCrenEuryCrenCrenCrenEury

pJP33CrenCrenCrenCrenEuryCrenCrenCrenCrenCrenCrenCrenCrenCrenCrenND

pJP41CrenCrenCrenCrenEuryCrenCrenCrenCrenCrenCrenCrenCrenCrenCrenCren

pJP89

CrenCrenEuryEuryEuryEuryCrenEuryGCCrenCrenEuryCrenCrenCrenEury

pJP27EuryEuryA.U*CrenEuryEuryCrenEuryCrenEuryCrenEuryEuryCrenEuryCren

pJP78EuryEuryA.U*CrenEuryEuryCrenEuryCrenGCrenEuryEuryCrenEuryCren

Intradomain nucleotide signature analysis of the rDNA sequencesof clones pJP33, pJP41, pJP89, pJP27, and pJP78. Sequence signa-tures for the two archaeal kingdoms, Crenarchaeota (Cren) andEuryarchaeota (Eury), are taken from Winker and Woese (19). Gp.I, marine Crenarchaeota signature features, is as described byDeLong (20) and Fuhrman et al. (18). Eucaryal signatures aredenoted by an asterisk, and undetermined nucleotides are denoted asND. Numbering (nucleotide position) is based on the Escherichia coli16S rRNA sequence.

ciently deeply that they should not be defined as Crenar-chaeota or Euryarchaeota.The correct topology of a phylogenetic tree should be

insensitive to the sequences used and the method of analysis(15). Phylogenetic trees differing in composition of archaealtaxa and rooted with various sets of bacterial or eucaryalsequences, or unrooted, were constructed by maximumlikelihood, distance matrix, and maximum parsimony meth-ods. With a single exception (sequences from marine Ar-chaea, see below), no perturbation of the topology shown inFig. 1B was observed for the sequences of clones pJP33,pJP41, and pJP89 (analyses not shown). Parsimony analysisplaced the divergence of the pJP27/78 lineage prior to thebranching of the Crenarchaeota from the Euryarchaeota withsome selections of taxa (data not shown). However, boot-strap support for this topology was low (<50%) and may bea result of systematic error due to a relatively high rate ofnucleotide substitution in the pJP27/78 lineage (21).Base composition disparities between sequences have

been shown to promote artifacts in phylogenetic analyses(22). The high G+C content of the sequences of these clones(0.60-0.70) is comparable to that of the rRNAs of cultivatedCrenarchaeota and extremely thermophilic Euryarchaeota(0.62-0.67; ref. 22). Transversion distance-matrix and trans-version parsimony analyses (22) (data not shown), however,support the branching order of the clone sequences given inthe trees of Fig. 1, suggesting that base compositional biashas little, if any, influence on this topology.Hybrid rDNA clones, composed of rDNA from different

organisms, can arise during PCR amplification of mixed-population DNAs (13). One possible source of such chimericrDNAs is that partially elongated DNA products formedduring one round of amplification may serve as primers inanother round of amplification with a template rDNA from adifferent organism. Inspection of predicted secondary struc-tures, phylogenetic analyses of different portions of thesequences, and evaluation by the CHECK-CHIMERA programof the RDP indicated that four of the original 98 clonescontained such chimeric sequences. These tests all indicate

that the sequences analyzed in Fig. 1 are free of chimericartifacts (data not shown).

DISCUSSIONThe great phylogenetic diversity of archaeal rDNA clonesrecovered from this single hot spring was unexpected and iswithout precedent in previous studies (1). Moreover, it isunlikely that the 98 clones inspected exhaust the diversity ofthis archaeal community, since many of the sequence typeswere recovered only once. Gene sequences were obtainedhere that indicate the presence of both close evolutionaryrelatives of cultivated species and several organisms withoutknown close relatives. Although phylogenetic placement ofsome of the sequences in Fig. 1A is inexact in detail (asindicated by low bootstrap values), representatives ofmost ofthe major groups of cultivated Crenarchaeota are evidentlypresent in this environment. Sequences affiliated with theDesulfurococcus/Pyrodictium clade (pJP7 and pJP74), Py-robaculum sp. (pJP8), and Thermofilum pendens (pJP6 andpJP81) were recovered. In addition, an rDNA sequence(pJP9) highly similar to that ofArchaeoglobusfulgidus (97%similarity) was obtained. The first four ofthese named generahave been isolated from environments having temperatureand pH ranges that overlap those of this hot spring, whereasall prior isolates of Archaeoglobus sp. have been obtainedfrom marine environments (23).

Several of the cloned rDNAs show no close phylogeneticaffinity to cultivated species (pJP33, pJP41, pJP89, pJP27,and pJP78). Each of these lineages diverges from the cren-archaeal stem closer to its root than do any previouslycharacterized crenarchaeal rDNA sequences. Because ofthegreat evolutionary distance separating these sequences fromthose of cultivated Crenarchaeota, these lineages could rep-resent organisms having fundamentally distinct physiologies.The rDNA sequences of one group in particular, the pJP27/pJP78 clade, suggest both early divergence from other ar-chaeal groups and relatively rapid evolution after that diver-gence.

Recently, two other archaeal lineages have been detectedby phylogenetic analyses of rDNA clones obtained frommarine bacterioplankton populations (18, 20, 24). The se-quences of one of these groups are crenarchaeal in affiliation,although their rapid evolution and relatively low G+C con-tent make their phylogenetic placement problematic. Whenincluded in the present analysis, these sequences most oftengrouped with those of clones pJP27 and pJP78 in distance-matrix, maximum parsimony, maximum likelihood, andtransversion analyses. However, bootstrap values for thisgrouping were '55% in all cases, and the marine sequencesfrequently affiliated with the pJP41 and pJP89 sequences aswell, depending on which sequences were used in the anal-ysis (data not shown). The apparent rapid evolution of thesemarine sequences may account for their erratic behavior inphylogenetic analysis. Table 1 confirms that the marineorganisms are Crenarchaeota in that they have 10 of 16crenarchaeal signature features (20). However, the pJP27 andpJP78 sequences have only 6 of 16 crenarchaeal signatures,consistent with the hypothesis that they constitute a lineagedistinct from that of the marine species and diverge closer tothe root of the crenarchaeal branch than do the marinesequences. Further phylogenetic and phenotypic analyseswill be necessary to determine the precise evolutionaryrelationships among these organisms.The ubiquity of thermophilic lineages and their predomi-

nence as the deepest and most slowly evolving branches ofthe archaeal tree led to the proposal that thermophily may bean ancestral character of Archaea (25). The G+C contents ofthe sequences analyzed in this study are largely within therange of those previously reported for the rRNAs of culti-

1612 Microbiology: Barns et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020

Proc. Natl. Acad. Sci. USA 91 (1994) 1613

vated thermophilic Archaea, supporting this hypothesis (22).In addition, studies of bacterial phylogeny suggest an inversecorrelation between rate of rRNA sequence evolution andretention of ancestral phenotypic characters (25, 26). Severalof the organisms identified in this hot spring are among thedeepest and most slowly evolving of known lineages andmay, therefore, provide additional perspective on the ances-tral phenotype of Archaea.rRNA gene sequences recovered in this study document

the occurrence ofmany more crenarchaeal lineages than havebeen recognized previously through cultivation. The resultsalter our understanding of the phylogenetic organization ofArchaea in a fundamental way. Previous analyses of rRNAsequences led to the conception of the archaeal domain astwo distinct evolutionary lineages, Crenarchaeota and Eur-yarchaeota (27). A relatively large evolutionary distance wasseen to separate known Euryarchaeota from the phylogeneticcluster ofcultivated Crenarchaeota, enhancing their apparentdistinction. Such discrete separation of the two groups hascontributed to speculation that Crenarchaeota might be spe-cifically related to Eucarya rather than to Euryarchaeota,while Euryarchaeota are specifically related to Bacteria (28).Although all but one of the archaeal lineages discovered inthis study appear to be Crenarchaeota, some of them branchfrom near the root of the crenarchaeal stem. This, togetherwith the recognition of similarly deeply diverging Euryar-chaeota (e.g., Methanopyrus; ref. 29), blurs the prior sharpdistinction between the two archaeal groups. The expandedarchaeal tree lends further support to the phylogenetic co-herence ofArchaea as a whole. If the wealth of diversity seenin the present study extends to other environments, it seemslikely that additional major lineages of Archaea will bediscovered. What has been taken to be two distinct archaeallineages may even become a bush of lineages arising from theroot of the archaeal tree.Molecular methods such as described here allow study of

microbial ecosystems without the requirement for cultivationand description of specific organisms. There is a genuineneed for sequence-based surveys of microbial communities:our knowledge of the types and distributions of microorga-nisms in the environment is rudimentary. Studies of naturalmicrobial populations undoubtedly will continue to revealorganisms and fields for further investigation. Environmentalsurveys of microorganisms also are likely to expand furtherour understanding of biological diversity and the evolution-ary processes that have led to it.

We gratefully acknowledge Steve Koch, Anna-Louise Reysen-bach, and Claire Woodman for their excellent field assistance; CarlWoese, Ed DeLong, and Mitchell Sogin for unpublished sequencedata; Robert Barns for generously lending the global positioningsystem receiver; Chuck Delwiche, Niels Larsen, and Gene Wickhamfor computer assistance; and Jim Brown, Gene Wickham, andparticularly Carl Woese for extensive helpful comments on themanuscript. We also thank Dr. Bernadette Pace for supplyingThermus aquaticus DNA polymerase, and the staff of Yellowstone

National Park for their enthusiastic cooperation. This material isbased upon work supported by a National Science FoundationGraduate Fellowship to S.M.B. and a Department of Energy grant(DE-FG02-92ER-20088) to N.R.P.

1. Ward, D. M., Bateson, M. M., Weller, R. & Ruff-Roberts,A. L. (1992) Adv. Microbial Ecol. 12, 219-286.

2. Tsai, Y.-L. & Olson, B. H. (1992) Appl. Environ. Microbiol. 58,2292-2295.

3. Steffan, R. J., Goksoyr, J., Bej, A. K. & Atlas, R. M. (1988)Appl. Environ. Microbiol. 54, 2908-2915.

4. Tsai, Y.-L. & Olson, B. H. (1991) Appl. Environ. Microbiol. 57,1070-1074.

5. Torsvik, V., Goksoyr, J. & Daae, F. L. (1990) Appl. Environ.Microbiol. 56, 782-787.

6. Holben, W. E., Jansson, J. K., Chelm, B. K. & Tiedje, J. M.(1988) Appl. Environ. Microbiol. 54, 703-711.

7. Reysenbach, A.-L., Giver, L. J., Wickham, G. S. & Pace,N. R. (1992) Appl. Environ. Microbiol. 58, 3417-3418.

8. Ponce, M. R. & Micol, J. L. (1992) Nucleic Acids Res. 20, 623.9. Lane, D. J. (1991) in Nucleic Acid Techniques in Bacterial

Systematics, eds. Stackebrandt, E. & Goodfellow, M. (Wiley,New York), pp. 115-175.

10. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab.Press, Plainview, NY).

11. Sekar, V. (1987) BioTechniques 5, 11-13.12. Larsen, N., Olsen, G. J., Maidak, B. L., McCaughey, M. J.,

Overbeek, R., Macke, T. J., Marsh, T. L. & Woese, C. R.(1993) Nucleic Acids Res. 21, 3021-3023.

13. Liesack, W., Weyland, H. & Stackebrandt, E. (1991) Micro-bial. Ecol. 21, 191-198.

14. DeSoete, G. (1983) Psychometrika 48, 621-626.15. Olsen, G. J. (1987) Cold Spring Harbor Symp. Quant. Biol. 52,

825-833.16. Felsenstein, H. (1985) Evolution 39, 783-791.17. Brock, T. D. (1978) Thermophilic Microorganisms and Life at

High Temperatures (Springer, New York).18. Fuhrman, J. A., McCallum, K. & Davis, A. A. (1993) Appl.

Environ. Microbiol. 59, 1294-1302.19. Winker, S. & Woese, C. R. (1991) Syst. Appl. Microbiol. 14,

305-310.20. DeLong, E. F. (1992) Proc. Natl. Acad. Sci. USA 89, 5685-

5689.21. Felsenstein, J. (1978) Syst. Zool. 20, 401-410.22. Woese, C. R., Achenbach, L., Rouviere, P. & Mandelco, L.

(1991) System. Appl. Microbiol. 14, 364-371.23. Burggraf, S., Jannasch, H. W., Nicolaus, B. & Stetter, K. 0.

(1990) System. Appl. Microbiol. 13, 24-28.24. Fuhrman, J. A., McCallum, K. & Davis, A. A. (1992) Nature

(London) 356, 148-149.25. Woese, C. R. (1987) Microbiol. Rev. 51, 221-271.26. Achenbach-Richter, L., Stetter, K. 0. & Woese, C. R. (1987)

Nature (London) 327, 348-349.27. Woese, C. R., Kandler, 0. & Wheelis, M. L. (1990) Proc. Natl.

Acad. Sci. USA 87, 4576-4579.28. Lake, J. A. (1987) Cold Spring Harbor Symp. Quant. Biol. 52,

839-846.29. Burggraf, S., Stetter, K. O., Rouviere, P. & Woese, C. R.

(1991) System. Appl. Microbiol. 14, 346-351.

Microbiology: Barns et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

4, 2

020