Ecology, Evolution, And Rrna

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

Ann. Rev. Microbiol. 1986. 40:337~5 Copyright© 1986by AnnualReviewsInc. All rights reserved

MICROBIAL ECOLOGY AND EVOLUTION: A RIBOSOMALRNA APPROACH Gary J. Olsen, David J. Lane, Norman R. Pace

Stephen

J. Giovannoni,

and

Department of BiologyandInstitute for Molecular andCellularBiology,Universityof Indiana,Bh~,omington, Indiana47405 David A. Stahl Department of VeterinaryPathobiology,Universityof Illinois, Urbana, Illinois 61801

CONTENTS INTRODUC’I’ION ..................................................................................... MOLECULAR PHYLOGENY ANDMICROBES ............................................. RibosomalRNAsas Indicatorsof Phylogeny................................................ Analysis of Population Contents by RibosomalRNASequences......................... INFERRINGRELATIONSHIPSFROMMOLECULAR SEQUENCES ................... Methodsqf PhylogeneticTreeInference..................................................... Distance~4atrix Methodof PhylogeneticTree Inference ................................. Alternativesto PhylogeneticTrees............................................................ RIBOSOMAl_, RNASEQUENCE DATA BASE............................................... CurrentlyAvailableDataCollections......................................................... RapidDeterminationof Additional 16SrRNASequences................................. NATURAL POPULATION ANALYSIS ......................................................... Populations Inspected............................................................................ 5SrRNA Analysis................................................................................. PopulationAnalysis by 16S RibosomalRNAGenes....................................... IN SITU HYBRIDIZATION FOR COUNTINGAND IDENTIFYING ORGANISMS ............................................................................ SUMMARY AND FUTURE PROSPECTS ......................................................

338 338 338 341 342 342 344 348 348 348 349 351 351 353 357 359 361

337 0066-4227/86/1001-0337502.00

Annual Reviews www.annualreviews.org/aronline 338

OLSEN, LANE, GIOVANNONI,PACE & STAHL

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

INTRODUCTION Nucleic acid sequencing technology is bringing a much-neededphylogenetic perspective into microbiology. There is no morefundamentaland straightforward way to classify and relate organisms than by appropriate nucleic acid sequence comparisons. The simple morphologyof most microbes provides few clues for their identification; physiological traits are often ambiguous. The microbial ecologist is particularly impededby these constraints, since so many organisms resist cultivation, which is an essential prelude to characterization in the laboratory. In this article we describe the application of rapid nucleic acid sequencing and recombinant DNAmethodologies to the analysis of phylogenetic and quantitative aspects of mixedmicrobial populations. The analysis is based on nucleotide sequence comparison of ribosomal RNAs(rRNAs)or their genes, extracted from naturally occurring biomass. Because the analysis is phylogenetic, population membersare related to knownorganisms in terms of their fundamentalbiochemicalproperties and potentials. Also, because molecules rather than organismsare isolated, the methodis not limited to species that are amenableto laboratory cultivation. Characterization of unknownorganisms by rRNAsequences requires a reference collection of sequences from knownorganisms. Although substantial, the available reference collection is far from comprehensive,largely because nucleotide sequence determinations have been the province of specialized laboratories. However,recently developed sequencing methodsare sufficiently simple that the routine incorporation of rRNAsequence information into the systematic description of microorganismsis nowfeasible. A rapid expansionof the reference sequence collection can therefore be anticipated. Wediscuss these developments and review approaches for relating organisms by macromolecular sequences.

MOLECULAR PHYLOGENY AND MICROBES Ribosomal RNAs as Indicators of Phylogeny The use of macromolecularcomparisons to infer phylogenetic relationships (101) is now well established. Comparisons may be based either on experimental measurementsof "molecular similarity" (e.g. antibody crossreactivity, DNA-DNA hybridization, and ribosomal RNA-DNA hybridization) or on mathematical analyses of molecular sequence data. The former methodsrequire the pairwise experimental comparisonof most, or preferably all, organismsconsidered. In contrast, sequencedata are readily accumulated, creating a "data base" that can be referred to for phylogeneticanalysis of new sequence data as they becomeavailable.

Annual Reviews www.annualreviews.org/aronline ECOLOGY,EVOLUTION,ANDrRNA

339

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

What molecule is to be used for mapping phylogenetic relationships? Studies colnparing c-type cytochromes, globins, and other commonproteins have been :rewarding, especially amongthe "higher" eukaryotes (35). On the other hand,, both prokaryotic and eukaryotic microbes are so phylogenetically and biochemicallydiverse that even the identification of homologous proteins is not straightforward (20). For the analysis of natural microbial populations, in whichunknowndiversity must be anticipated, there are several reasons to focus on the rRNAs(88). 1. The rRNAs,as key elements of the protein-synthesizing machinery, are functionally and evolutionarily homologousin all organisms. 2. The rRNAsare ancient molecules and are extremely conserved in overall structure. Thus, the homologousrRNAsare readily identifiable, by their sizes. 3. Nucleotide sequences are also conserved. Somesequence stretches are invariant across the primary kingdoms,while others vary. The conserved sequences and secondary structure elements allow the alignment of variable sequences so that only homologousnucleotides are employedin any ¯ phyloge, netic analysis. The highly conserved regions also provide convenient hybridization targets for cloning the rRNAgenes and for primerdirected sequencing techniques (see below). 4. The rRNAsconstitute a significant componentof the cellular mass, and they are’, readily recoveredfrom all types of organismsfor accumulationof a data base of reference sequences (see below). 5. The rRNAsprovide sufficient sequenceinformation to permit statistically significant comparisons. 6. The rRNAgenes seem to lack artifacts of lateral transfer betweencontemporaneousorganisms. Thus, relationships between rRNAsreflect evolutionary relationships of the organisms. There are three rRNAsin bacteria, 5S (-120 nucleotides), 16S (-1600 nucleotides), and 23S (~3000 nucleotides). Eukaryotes commonlycontain fourth rRNA,5.8S (-160 nucleotides), which is homologousto the 5’ end the bacterial 23S. The size of rRNAsvaries somewhatamongthe organisms inspected, but we use the above values as generic designations. The 5S and 16S rRNAshave been used most for rRNA-basedphylogenetic characterizations, largely for historical and technical reasons. The 5S rRNA, because it is relatively small, was amenableto sequenceanalysis by the late 1960s. However,its paucity of independently varying nucleotide positions limits its ultimate phylogenetic usefulness (see below). The 16S rRNAis appropriate size for broad phylogenetic analyses, but it was too large for complete sequence determinations until the developmentof DNAcloning and sequencing: protocols. Instead, the 16S rRNAwas subjected to partial se-

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

340

OLSEN, LANE, GIOVANNONI,PACE & STAHL

quence analysis, so-called "oligonucleotide cataloging," which characterizes ~25%of the sequence as unique RNase T1 oligonucleotides. The first comprehensivephylogenyof prokaryotes (indeed of life on the earth) emergedfrom studies of 16S rRNApartial sequences (oligonucleotide catalogs) by Carl Woeseand his colleagues (33). Earlier, the eukaryotes were thought to be relative latecomers generated by the fusion of bacterial cosymbionts. Although the rRNA-basedstudies confirmed that the mitochondria and the chloroplasts are related to contemporaryprokaryotes, the eukaryotic line of descent, i.e. the nuclear genotype, was found to be as ancient as the bacterial genotype(see Figure 1). Morestrikingly, the Woesestudies (33, revealed that extant life on earth consists of not two, but three primarylines of evolutionary descent; there are two phylogenetically distinct groups of prokaryotes, termed by Woese"eubacteria" and "archaebacteria." It was established, initially from rRNAsequencesand later from manyadditional lines of evidence, that the archaebacteria are as evolutionarily distinct from the eubacteria as either is from the eukaryotes (reviewed in 98). Figure 1 shows a quantitative phylogenetic "network" (an unrooted phylogenetic tree) of somefamiliar organismsin the three primary kingdoms, based on 16S rRNAsequences. The lengths of the line segments are pro-

ARCHAEBACTERIA i Halococcu~ morrhua S~lfolob....

~" Ifataricus~/Me~l°atbo~ot~filmvilfil°li]

0.1

FUB~CTFRI,~

FUKN~YOTE$ Figure 1 An unrooted phylogenetic

tree,

based on 16S rRNAsequences (10, 15, 23, 36, 37, 47,

60, 67,69, 71,76, 78,82, 92,93, 99),illustratingthe threeprimary linesof descentandsome of their majorsubgroups. Thetree wasinferredbya distancematrixmethod (23,31,67);the lengths of the line segments reflect the estimatednumber of fixed mutationalevents.Thescalebar corresponds to an evolutionary separation of 0.1 accepted pointmutations persequence position.

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION, AND rRNA

341

portional to evolutionary distances (evolutionary distance increases with time, but not necessarily linearly; see below). All of the --400 organisms inspected (mostly bacteria) are clearly affiliated with one of the three primary groups. Some of the major subgroups in each of the primary kingdoms have also been delineated (71, 97). About ten-subgroups in the eubacteria and two in the archaebacte, ria have been defined. So few eukaryotic rRNAs have been sequenced that major phylogenetic units remain to be defined; there is, in particular, tremendous uncharacterized diversity among the eukaryotic microbes (85). More detailed evolutionary relationships have been established within some of the familiar subgroups of each primary kingdom, but the task of inferring the phylogeny of life on earth has hardly begun.

Analysis of Population Contents by Ribosomal RNA Sequences Two general methods have been used for exploring natural microbial populations (outlined in Figure 2). In one approach (Figure 2a), suitable populations of limited complexity, 5S rRNAis isolated from naturally occurring biomass and the various species-specific molecules are sorted by highresolution ~gel electrophoresis. Unique 5S rRNAtypes are then sequenced, and with reference to other 5S rRNAsequences, the phylogenetic affinities of the contributing organisms are defined.

POPULATION. ANALYSIS USING 5SrRNAs

POPULATION ANALYSIS USING 16SrRNAGENES BIOMASS (mixed population)

BIOMASS (mixed population)

~

break cells; phenol extract

flULKRNA I-low-resolution polyaerylamide ~gelelectrophoresis MIXI--D 55rRNAs

broak ceils; phenol extract

1

PURIFIED 5SrRNAs

for analyses

"shotgun clone"into bacteriophage lambda

screen byhybridization with

~"mxedkingdom"16SrRNAprobe 16 rRNAGENE CLONES

determine nucleotide sequence; I.compare with other5SrRNA" ,~, sequences PHYLOGENETIC CHARACTERIZATION OFPOPU.ATION MEMBERS

l

RECOMBINANT DNAUBRARY

32pend-label RNAs; high-resolution polyacrylamide gelelectrophoresis

Figure 2 Flow charts populations.

~

TOTAL DNA

sequence with 165rRNA-specific primers;compare with other 6S rRNAsequences

~

PHYLOGENETIC CHARACTERIZATION OFPOPULATION MEMBERS

of (a) 5S rRNAs and (b) 16S rRNA genes from natural

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

342

OLSEN, LANE, GIOVANNONI,PACE & STAHL

In the second approach, which seems not to be limited by population complexity, 16S rRNAgenes are "shotgun cloned" using DNApurified from collected biomass (Figure 2b). It does not matter if the original DNAwas from a mixedpopulation of organisms; the individual rRNAgenes are clonally isolated as recombinantbacteriophage. The different types of cloned rRNA genes are then sorted in the laboratory and are subjected to limited sequence analysis by a technique that affords direct access to regions of the 16S rRNA gene that are particularly useful for phylogenetic evaluations (see below). Again, by referring to existing collections of completeand partial sequences, it is possible to infer the phylogenetic affinities of the organismsin the original population. The cloned rRNAgenes can also be used as hybridization probes to quantitate the corresponding organisms in the population or to identify similar organismsin other environments. Inference of the phylogenetic relationships of the organismsin a microbial population from those in a data base is not just an exercise in genealogy. In particular, closely related organismsmust have similar fundamentalbiochemical properties. However,relatively few "fundamentalproperties" have been identified. Certainly the translation apparatus, componentsof the DNA and RNAsynthetic machineries, ion pumps, and other central functions are fundamental, i.e. conserved. On the other hand, manyovert physiological traits, such as heterotrophy and autotrophy, are very scattered phylogenetically (see e.g. 33). As the details of microbial phylogenyare elucidated, it will becomeincreasingly clear which biochemical properties may be accurately predicted about organismsthat are well-defined phylogenetically but have not been cultivated or otherwise characterized in the laboratory.

INFERRING RELATIONSHIPS FROM MOLECULAR SEQUENCES Phylogenetic trees provide the most incisive summaryof phylogenetic relationships inferred from molecular sequencedata. Weoutline several different tree inference methodsand describe one of the methodsin detail. This is followed by descriptions of two alternatives to tree analysis.

Methodsof Phylogenetic Tree Inference Several quite different tree inference methodsare in use (reviewed in 29); unfortunately, they do not necessarily arrive at the sameconclusions. Thus, it is important to distinguish the conceptual bases of three commonlyused inference methodsand a fourth which is not in general use. CLUSTER ANALYSIS Cluster analysis (86) is a rapid method of assigning taxa to groupson the basis of similarities. It is applicable to any formof data

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION,ANDrRNA

343

for which a meaningful measure of similarity can be defined, and it is therefore the methodof choice for analyzing rRNAoligonucleotide data (33). Cluster anailysis has also been applied to 5S rRNAand 16S rRNAsequence data (e.g. 55, 63). Phylogeniesinferred from cluster analysis are expected be valid whtenthe rates of sequencedivergenceare sufficiently similar in all lineages (16). Whenthe rates of sequence divergence are unequal (see below), the most rapidly changing sequences tend to branch too deeply (too early) in the inferred phylogeny.Of the commonly used tree-inference techniques, cluster analysis is the mostproneto this inaccuracy. However, at least one variatie,n of cluster analysis, the "present-dayancestor" method(53, 61), partially avoidsit. MAXIMUM PARSIMONY Tree inference methods based on the principle of parsimonyare frequently used for analyses of sequencedata (e.g. 14, 30, 54). These methodsseek the tree branching order (topology) that requires the fewest mutational events. Proponents emphasize that the method independently examinesthe evolution of every sequence position, rather than reducingthe data to averagesimilarities (as in cluster analysis) or distances (see below). Felsenstein (27) has demonstrated that the methodaccurately reconstructs,; phylogenieseither if the rates of evolutionin the various lineages are similar or if the sequence changes in the most rapidly evolving lineages are not too numerous.However,manyinteresting groups, such as the mycoplasmas (96) and mitochondria (e.g. 11, 55, 67, 99), exhibit unusually evolutionaEy rates and deep divergences. In such cases, the most rapidly evolving lineages tend to branchtoo deeply in the inferred phylogenetictree. DISTANCE MATRIX METHODS Distance matrix methods (e.g. 31, 81) use the differences between pairs of sequences to estimate the "evolutionary distance" (usually expressed as the average numberof accepted point mutations per sequence position) separating the sequence pairs. This step can include a correction for the superposition of multiple mutations at a single sequence p,asition. After the evolutionary distances separating every pair of sequences ihave been estimated, the phylogenetic tree is found that most faithfully represents the pairwise distance estimates (according to any one of several mathematicaldefinitions of "faithfully") (29). To the extent that methodcornpensates for multiple changes at sequence positions, it is more reliable than parsimonyin situations that involve large amountsof sequence divergence and lineage-to-lineage variations in evolution rate. Schwartz & Dayhoff (81) have presented evidence that distance methodsmight also less susceptible to statistical errors (see below)resulting fromthe finite length of the sequences being compared.

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

344

OLSEN, LANE, GIOVANNONI,PACE & STAHL

MAXIMUM LIKELIHOOD Felsenstein (28) has pointed out that the current methodsof tree inference are not directly derived froma specific modelof the evolutionary process. Heproposes that tree inference should find the tree with the maximum likelihood of explaining the data in the context of a concrete evolutionary model. He illustrates his proposal with a simple model that assumes that all sites are equally and independently mutable and have the same average base usage. Although maximum likelihood has been applied to problems with few organisms (e.g. 28, 42), the methodis computationally demanding and hence has not been frequently used. A GENERAL CAVEAT Site-to-site differences in mutability present a largely unsolved problem in tree inference. In combinationwith lineage-to-lineage variations in mutation rate, the site-specific variations introduce systematic errors into all of the tree inference techniques. There is abundant evidence (14, 18, 30, 34, 38, 64, 67) for sufficiently great position-to-position variation in mutability to give dramatic errors in evolutionary distance estimates (34). This error places the branch points of rapidly evolvinglineages too early in the inferred trees. The effect can be partially alleviated if there are more organisms represented in the tree, which increases the chance of having a moderatelyevolvingsequencethat is specifically related to the rapidly evolving sequence(99). It is also possible to estimate the actual distribution of rates over the length of the sequences and to appropriately compensate for the variations (34, 43; G. J. Olsen, unpublished), though muchwork remains be done. Randomerrors arise because comparisons of finite length sequences provide a limited sampling of evolutionary history. That is, the number of sequence differences observed from the finite number of nucleotides compared is subject to a binomial-typecounting error. The magnitudeof this error can be minimized by examining a larger number of independently evolving sequencepositions. This is illustrated in Figure 3. Distance

Matrix Method of Phylogenetic

Tree Inference

Althoughit is beyondthe scopeof this reviewto detail all of the tree inference methods mentioned above, we discuss one for additional perspective. The majority of the discussion is applicable to all methods. The process of phylogenetictree inference can be divided into three distinct components. First, the various sequences must be "aligned" so that the evolutionarily homologousnucleotides of each sequence are in registry with one another. Second, the data are fit into a given tree branching order, and criteria are defined for evaluating its faithfulness in representing the data. Third, alternative branching orders are tested to seek the tree that most faithfully reflects the sequencedata by the criteria definedin step two. Cluster

Annual Reviews www.annualreviews.org/aronline ECOLOGY,EVOLUTION,ANDrRNA

345

¯ Nominal values 2.0

95~Confidence Umits:

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

I I6SrRNA ISSrFINA

0 ILl 0.5

1,0

0,2

Sequence Homology Figure 3 The relative statistical uncertainties (52) of evolutionary distance estimates (50) based on 5S and 165. rRNAs. Because not all sequence positions are independently variable, the 5S and 16S rRNAs have been treated as 80 and 1000 sequence positions, respectively. Adapted from (72).

analysis does not require this search of alternative branchingorders, so it is the most rapid of the various methods. SEQUENCE ALIGNMENT The drawing of evolutionary conclusions from nucleotide sequence data requires the assumption that each set of compared nucleotides (one from each sequence) is derived from a single ancestral nucleotide by a concrete (thoughgenerally unknown) series of mutationalevents. Conversely,.if nucleotides do not share a common ancestry, then their comparison provides no valid evolutionary information. The process of sequencealignmentdefines this one-to-one correspondenceof the residues in each sequence with their evolutionary homologsin each of the other sequences. The basis of an alignment lies in the recognition of regions that are more similar amongthe various sequences than could be expected at random. Initially, the most clearly homologoussequence regions are aligned. Later, regions of less substantial homologyare aligned. In addition to primary structure (the linear sequence), conservedsecondarystructures (base pairings) provide additional guidance for the alignment of transfer RNAand rRNA sequences. The process would be simple if there were no insertions or deletions in the mutational history of the molecules. However,a significant fraction of the mutationsin genes of interest lead to changesin the length of

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

346

OLSEN, LANE, GIOVANNONI,PACE & STAHL

the molecule. Becauseof these internal insertions and deletions, the alignment of the sequences is only regional and must be constantly reconsidered along the length of the molecules. It is usually necessary to introduce some numberof "alignment gaps" in order to align all of the homologoussequence regions in a collection of molecules. The greater the numberof homologous features used to define a sequencealignment, the better-defined the locations of the alignment gaps become. After a set of sequences has been aligned there are generally regions in whichthe alignmentis ambiguous,particularly if the sequencesare diverse. Because of the importance of comparingonly homologousnucleotides, it is necessary to exclude these regions of ambiguousalignment from evolutionary analyses. FITTING ALIGNEDSEQUENCEDATATO A TREE BRANCHING ORDERBecause this step differs with each major tree inference method, the following discussion will mostly be limited to the distance matrix methodused in our laboratory (23, 70; G. J. Olsen, unpublished). In outline, the number nucleotide differences is used to estimate the numberof mutationalevents that separate each pair of aligned sequences. The estimates of evolutionary separation are fit to an assumedphylogenetic branching order; we find the optimal tree branch lengths and then calculate how well this phylogenetic tree represents the evolutionary distance estimates (see below). Whenquantitating the differences between aligned sequences, the treatment of alignment gaps must also be defined. In sequence regions for which we are sufficiently certain of the alignment to be confident that only homologousnucleotides are compared(see above), there are generally so few alignment gaps that the details of the treatment have little influence on the phylogenetic conclusions. Juxtaposition of an alignment gap in one sequence with a nucleotide in a second sequence has been treated as 0--2 sequence differences in estimating evolutionary divergence. Because multiple mutations occur at single sequence positions, the observed numberof nucleotide differences generally underestimates the number of mutational events that have occurred since the separation of the corresponding genes. A variety of formulas (e.g. 34, 43, 50, 51) have been proposed for estimating the average numberof fixed mutational events per sequence position (i.e. the "evolutionary distance") separating the two sequences. The resulting values are properly referred to as estimates because they are limited by both randomand systematic errors (see above). Havingestimated the evolutionary separations of all pairs of sequences, one must then answer the question, howaccurately can a given phylogenetic tree branchingorder represent the pairwise distance data? Wedefine the error of the representation as the sumof the squares of the differences betweenthe

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION,ANDrRNA

347

pairwise evolutionary distance estimates and the corresponding tree reconstructions of those distances, with each difference weighted by the corresponding statistical uncertainty (70; for other weighting factors see 29). Kimura& Ohta (52) have calculated the variance (the square of the expected counting en’or due to finite sequence length) of the Jukes &Cantor (50) evolutionary .distance estimates. Witha given tree branching order, it is necessary to compute the branch lengths that minimize tree error. These branch lengths define the minimum-errortree consistent with the branching order. Whenbranch lengths are assigned by such a least-squares approach, it is possible for lengths to assumenegative values (this problemdoes not arise in the other methodslisted above). Thesenegative values do not correspond to meaningful evolutionary phenomena,but rather are the mathematical responseto c,~rtain suboptimalbranchingorders. In trees that contain negativelength segments we currently change the negative values to zero (without makingan3, compensatoryadjustments in the lengths of the other branches) and then reevaluate the overall tree error (G. J. Olsen, unpublished). This simple trea~Imentprovides meaningfulvalues for relative tree errors, even in the presence of one or more short negative-length segments. Usually, some minorrearrangementof the tree branching order yields a tree that does not contain an3, negative-length segmentsand that also has a lower tree error. FINDINGTHEBESTTREEBRANCHING ORDER Even with a quantitative definition o,f howwell the sequence data are accomodatedby a given branching order (see above), as yet no process is guaranteed to find the "best" branching order of a large phylogenetic tree in a practical amountof computational time. In essence, methodslook instead for "better" trees by systematically testing a tractable numberof alternatives to the current "best known branchingorder." Whenthe process stops finding better trees, the current tree is assumedto be the "best." To find better tree branching orders we use a computerprogram(70) that examinesaill rearrangementsof the tree that can be achievedeither by moving any single group of organisms(subtree) to every alternative location in the tree or by interchanging any pair of subtrees. For trees of 10, 20, or 30 organismsthis requires testing about 250, 1600, or 4200 alternatives to the current tree. The programexaminesall of these tree rearrangements before taking the most improved tree as the starting point for another cycle of searching. Twocriteria suggest that this methodconsistently finds the best tree. First, the methoddoes not dependon the starting point of the search for better trees; randominitial tree configurations always converge on the same solution. Second,extending the vicinity of the search for improvedtrees by examining

Annual Reviews www.annualreviews.org/aronline 348

OLSEN, LANE, GIOVANNONI,PACE & STAHL

rearrangementsin the vicinities of near-optimal branchingorders (as opposed to the vicinity of the presumptivebest tree) does not yield a better solution.

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

Alternatives

to Phylogenetic

Trees

As can be seen, the casting of phylogenetic trees from sequence data is not a trivial undertaking. What are the alternatives to phylogenetic trees? Two simpler, particularly useful alternatives are described here. Thepossibility that a novel sequenceis very closely related to a previously knownsequenceis easily investigated. Thesimilarity of the novel sequenceto each of the previously knownsequences is determined; if the novel sequence is muchmore similar to one of the knownsequences than either are to the remainder of the sequences, then it mayimmediately be concluded that a specific relative of the novel sequence has been identified. If no individual sequenceis particularly similar to the novel sequence,it is natural to ask whether the sequence fits into any previously defined major phylogenetic division. Frequently, the novel sequence can be examinedfor "signature" sequences (or signature nucleotides) that have been compiled the basis of their ability to distinguish rRNAsfromeach of the majorbacterial groups (19, 97). By definition, signature nucleotides change sufficiently rarely that they are generally constant within each major group, but fortuitously vary between one or more groups. These nucleotides can be considered a classification key to the phylogenetic divisions of the corresponding rRNA sequences. Thus, by determining the nucleotide identities at a limited number of specific sequencelocations, we can fit novel sequencesinto the predefined phylogenetic groups. As the frameworkof reference sequences increases, it will be possible to increase the resolution with which placements can be made. RIBOSOMAL Currently

RNA SEQUENCE

Available

DATA

BASE

Data Collections

Animportant aspect of the analyses described here is the correlation of rRNA sequences derived from natural populations with those in existing reference collections. Three rRNAsequence collections are of particular value: 5S rRNAcomplete sequences, 16S rRNAoligonucleotide catalogs, and 16S rRNAcomplete and partial sequences. The following discussion outlines the various methodsof generating these rRNAsequences, the current status of the correspondingdata collections, and the prospects for significant expansionof these collections using new sequencingstrategies. 5S rRNAsare -120 nucleotides in length. Complete 5S rRNAsequences are determined by enzymatic(21) and chemical (74) sequencingprotocols. brief, gel-purified, 32p-end-labeled,5S rRNAis cleaved in separate reactions

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY, EVOLUTION, AND rRNA 349 with base-~,Ipecific enzymesor reagents such that, on average, one break per RNAmolecule is introduced. The partial digestibn products are then separated by size on high-resolution polyacrylamide gels. Whenadenosine-, cytidine-, guanosine~, and uddine-specific reactions are electrophoresed in adjacent ge.l tracks, a "ladder" is producedon an autoradiograph of the gel from wliich.the nucleotide sequence can be read. Weare aware of nearly 400 complete5S rRNAsequences; most or. all of these will eventually worktheir wayinto published compilations (25) and will be availabl~ for phylogenetic comparisons. The 16S rRNAs,which are -13 times the size of 5S rRNAs,contain more phylogenetic information, but determination of their sequenceswas a greater technical claallenge. Three methodswereused’in obtaining the majority of the available lt6S rRNAdata: oligonucleotide cataloging, gene cloning and sequencing:, and reverse transcription from 16S rRNA. Woeseand his colleagues have used oligonucleotide catalogs to determine evolutiona~3~relationships amongorganisms(32). Uniformly32P-labeled 16S rRNAisolated from cells grown in the presence of [3ZP]brthophosphate is ~ digested e~:haustively with ribonuclease T~I. RibonucleaseT1 cuts at the 3 side of all guanylic acid residues, and so generates a set of oligonucleotides, each with a single guanylic acid residue at its 3’ end. Theseare separated by two-dimen:sional electrophoresis according to size, base composition, and sequence(’79, 95). Individual oligonucleotides are recovered and. then sequenced: using a variety of enzymatic and chemical techniques (87). A 16S rRNAcatalog, i.e. the sequences of the ribonuclease Tl-generated 16S rRNA oligonucleotides, is characteristic of the source organismand maybe compared to l~e catalogs derived from other organisms. Approximately 400 prokaryotic 16S rRNAshave been cataloged (97). Each catalog contains about 400 nucleotides (25% of the 16S rRNA)in pentanucleotide or larger fragments. Becauseof its broad samplingof organisms, the catalog collection is particul~xly useful as a source of "signature" sequences (see above). Rapid Determination

of Additional

16S rRNA Sequences

The 16S rPJ~IAcatalog collection will continue to be a valuable resource for phylogenetic information; however, more expedient techniques~for 16S rRNA sequence determination are becomingavailable. One is aimed at determining complete16S .rRNAgene sequences, and another is designed to 0btain~sizable blocks of phylogenetically useful sequence data from the RNAitself. Both approaches makeuse of dideoxynucleotide sequencing protocols (80) and 16S rRNA-specific oligodeoxynucleotide primers (56). In the dideoxynucleotide chain termination sequencing protocol a DNA strand cortkplementary to the template nucleic acid is synthesized from a "priming oligonucleotide" (primer) which has been annealed to its specific

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

350

OLSEN, LANE, GIOVANNONI,PACE & STAHL

"priming site" on the template (80). The use of DNApolymerase or reverse transcriptase permits the sequencing of DNAor RNAtemplates, respectively. Chainelongation from the primer is randomlyterminated in separate reactions at either adenosine, cytidine, guanosine,or thymidineresidues by inclusion of low levels of the respective 2~,3~-dideoxynucleoside triphosphate in the reaction mixtures. Reaction products are detected by inclusion of a radioactively labeled nucleoside triphosphate (e.g. [a-35S]dATP)in the reaction (5) or by initiating synthesis with a radioactively labeled primer. Theproducts are then resolved on a polyacrylamide gel, and the nucleotide sequence is read from an autoradiograph of the gel. For the determination of complete 16S rRNAsequences, DNArestriction fragments containing all or part of the 16S rRNAgene are typically cloned into one of the single-stranded bacteriophage M13vectors. In the standard M13sequencing system, the priming site is a unique stretch of DNA adjacent to the M13cloning site (68). Chain elongation thus proceeds from this vector-specific site into the cloned segment of DNA.Since only about 300-500nucleotides of sequence can be read from a given primer, a variety of subcloning and "trimming" strategies have been devised to bring different parts of larger cloned DNAsinto juxtaposition with the priming site (68). Considerablemanipulationof the original clone is required for a gene the size of the 16S rRNAgene. Analternative strategy, which is specific for the determination of 16S rRNAgene sequences, takes advantage of the fact that 16S rRNAsare not uniformly variable in sequence (38). Several 15-20-nucleotide regions sequence with little or no variation are found within every 16S rRNAexamined. Because of their length and their universality, a synthetic DNA oligonucleotide that is complementary to one of these sequenceswill specifically anneal to the corresponding site in the 16S rRNAgene from any source organism, thereby providing a specific priming site for dideoxynucleotideterminated sequencing (56). Because these priming sites are in the gene sequenceitsel,f,, there is no needto manipulatethe clonedgeneso that it abuts on the vector-specific priming site mentionedabove. Additional oligonucleotides that are homologouswith (rather than complementaryto) the universal sequences anneal to the opposite DNAstrand of a cloned gene and permit sequence determination in the opposite direction. Thus, cloned 16S rRNA genes maybe rapidly sequenced by working outward in both directions from the universal 16S rRNAsequences (23, 99). About 80 complete 16S rRNAgene sequences representing the three primary kingdomshave been determined; a steady expansion of this collection is anticipated. Phylogenetic characterization of most organisms will not require complete sequences. A protocol for rapidly generating large blocks of 16S rRNA

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION,ANDrRNA

351

sequence data, which requires neither isolation of purified 16S rRNAnor the cloning of its gene, has been described recently (56). The 16S rRNAin bulk, cellular RNApreparations is selectively targeted for dideoxynucleotideterminated sequencing with reverse transcriptase and the 16S rRNA-specific primers mentionedabove. Three primers, complementaryto three strategically positioned regions of conservation in 16S rRNA,have been found particularly useful[ for phylogeneticevaluations becauseof their general applicabifity (56). Figure 4 showsthe location of these three primingsites in representative eubacterial, archaebacterial, and eukaryotic small-subunit rRNAs,and indicates the extent of nucleotide sequence routinely accessed from each. These primers routinely yield 800-1000 nucleotides of sequence from each 16S rRNAmolecule, i.e. three 250-350-nucleotide "blocks" (Figure 4). Phylogenel:ictrees inferred fromthese limited regions of the 16S rRNA (either singly or together) have topologies identical to those obtained from complete sequences(56, 67, 70). At present our collection contains ~75 such "partial" 16S rRNAsequences, --50 from eubacteria and --25 from eukaryotes. The phylogeneticusefulness of the data obtained and the relative simplicity of the approachnaakethis technique attractive for classifying microbesof uncertain affiliation. It requires only about four days of effort to evaluate the phylogenetic statu:~ of a cultured organism.Weanticipate, therefore, that the 16S rRNApartial sequence data base will expandrapidly. NATURAL

POPULATION

Populations

Inspected

ANALYSIS

So far, three natural populations have been characterized by 5S rRNAanalysis: (a) the bacteria inhabiting OctopusSpring in YellowstoneNational Park; (b) the eukaryotic and bacterial componentsof the unusual symbiosesinvolving the "gutless" marine invertebrates Riftia pachyptila Jones, Calyptogena magnifica Boss & Turner, and Solemya velum Say; and (c) the bacteria inhabiting a leaching pondatop a copper recovery dumpat the Chinomine in ~ ~

A ~ A~

B B

C ~

C

E.COil

H.VO/CatE/ C O. discoideum Figure4 Hybridization sites (A, B, andC)andapproximate amounts of sequenceaccessible (arrows)fromthree small-subunit rRNA primers,shown on linear representations of the 16S rRNAs fromEscherichia coli (a eubacterium), Halobacterium volcanii(anarchaebacterium), and Dictyosteliurn discoideum (a eukaryote). Solidboxesalongthe sequence lines markregionswith sufficientintrakingdom homology to be generally usefulin the inferenceof phylogenies. Reproducedfrom(56). A

B

Annual Reviews www.annualreviews.org/aronline 352

OLSEN, LANE, GIOVANNONI, ,PACE & STAHL

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

southern NewMexico (57, 58, 89, 90). Octopus Spring, in addition, has served as a test environment for the 16S rRNAgene-cloning approach :described below. THE OCTOPUS SPRING COMMUNITY One of the better studied thermal springs in YellowstoneNational Park is OctopusSpring (9). Abundant,.growth of microorganisms is evidenced by fibrous pink tufts on the wails of the slightly alkaline, 9 I°C source pool and the immediaterun-off channel, and by microbial mats in the cooler, peripheral waters. There ’has been no reported cultivation of microorganismsresembling those observed in the source pool (9). Appreciablenucleic acid could not be extracted from pink tuft material, probably because most of it is dead and leached by the ~hot water. How.ever, since contact (microscope) slides immersed in the 91°C source pool--are rapidly colonized (9), an alternative strategy was to-introduce growthsurfares for collection. Cotton and fiberglass battings sewn into nylon screens were visibly colonized after 7-10-day immersion, and provided ample biomass for analysis. THE SULFUR-OXIDIZING SYMBIONTS The submarine hydrothermal vent systems:are associated with crustal spreading centers at the Mid-OceanRidge, which extends over about 70,000 km of the earth’s surface (22). It estimated that a volumeof water equivalent to the world’s oceans percolates through these systems about once every eight million years. The chemistry of the ocean water is markedlyaltered during convective passage; it is depleted of somecompounds(e.g. Ca2÷, Mg2÷, SO42-) and charged with others (e.g. metallic sulfides, H2S, CH4,and CO). Nourished by the reduced compounds in the exudate, dense populations of chemolithotrophic bacteria develop in and around manyof 1the vents (45). These, in turn, support rich animal communities that graze and filter-feed upon them. In manycases, sulfuroxidizing bacteria have formed symbiotic associations with macrobiota to produce, in effect, chemoautotrophicanimals. The giant vestimentiferan tube worm,.Riftia pachyptila Jones, and the giant clam, Calyptogenamagnifica Boss & Turner, are two such hydrothermalvent-associated invertebrates (13, 26). Solemyavelum Say is a musselthat inhabits sulfide-laden mudsof tidal marshes(12). All three share the anatomical peculiarity of partially or completely lacking mouths and digestive systems. They are nourished by their chemoautotrophic endosymbionts, which colonize a specialized organ in Riftia (the trophosome) and the gill tissues of the bivalves. Attempts cultivate these bacterial symbionts have so far been unsuccessful or have yielded equivocal results (H. Jannasch & C. Cavanaugh, personal communication). Small amountsof homogenizedtissue (< 1 g from each organism) served as starting material for the rRNAsequence analyses of these two-componentpopulations.

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION,AND rRNA

353

THE CHINOCOPPER-LEACHING PONDAbout 10% of total commercial copper recovery in the United States relies, uponmicrobial leaching of lowgrade ore (7). At the Chino.mine,a microbially conditioned, low-pHleaching fluid is pumpedinto pondsatop large dumpsof coarsely crushed’; sulfide-rich copper ore. Sulfuric acid produced by microbial’ chemolithotrophic metabolism in the pondspercolates.throughthe ore stack, solubilizing the copperas a sulfate sail.. Thiobacillusspp. are the most abundantisolates from these sites and; so generally are considered to be the primary effectors of leaching. However,given the traditional difficulty in cultivating chemoautotrophic bacteria and the knownpresence of physiologically distinct bacteria in these environments (7), the ecology of such leaching communities remains incompletely described. Samplesof pond water and surface mudunderlying one pH2.5, iron-rich leaching pond at the Chino mine were anal~czed. Between 109 and 1010 cells, as estimated, by microscopi~inspection, were recoveredby centrifugally washing --1 kg of mud. 5S rRNA Analysis .ACID EXTRACTION AND SEQUENCING The 5S rRNA analysis requires relatively few cells. From.108-109ceils of each species present will usually suffice for the complete5S analysis outlined in Figure 2a. Collection and extraction are complicated by other factors, including the presence of materials, that interfere with. nucleic acid recovery’ or, manipulationand the possibility of bias introducedat either the collection or extraction steps. In most cases, standard methodsfor the isolation of nucleic acids (73) will suffice. Cells maybe. lysed, by any of a variety of enzymaticand/or mechani~ cal techniques, including lysozyme and/or protease treatment; passage through a French pressure cell; or, most simply, direct extraction with detergent and phenol. In our experience, a good yield of low-molecularweight RNAscan be recovered from most bacteria (both gram-positive and gramnegative) without cell breakageby subjecting the bacteria to several freezethaw cycles with dry ice, followed by extraction at 60°C against buffersaturated phenol (90). The’5S :rRNAsare i~olated’ from total nucl~i~acid~ by polyacrylamidegel. electropho:resis. Theyare then radioactively labeled at their 3’ termini using [5’-32p]cyfidine bisphosphate and RNA ligase, or at their 5’ termini (following removal’ of preexisting-terminal phosphates with alkaline phosphatase) using [T-3~:P]ATPand polynucleotide kinase (89). Species-specific 5S rRNAs are separated on high-resolution (sequencing-type) polyacrylamide gels. only very small amountsof RNAare available, the total RNApopulation can be 3’-end-labeled’ before~ gel electrophoresis, and the preliminary fractionation of unlabeled rRNAscan be omitted. Followingtheir recovery from gels, each of the purified 5S rRNAsare sequenced as described above. NUCLEIC

Annual Reviews www.annualreviews.org/aronline 354

OLSEN, LANE, GIOVANNONI,PACE & STAHL

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

RESULTS OF 5S rRNA ANALYSES The

organisms comprising the natural populations discussed here span the three phylogenetic kingdoms(Figure 1). The Octopus Spring source-pool community contains three dominant members,two eubacteria (comprising -50%of the total 5S rRNArecovered) and one archaebacterium (90). The two eubacterial 5S rRNAsmost closely resemble those of the two Thermusspecies (T. aquaticus and T. thermophilus) in our reference collection. The archaebacterial 5S rRNAis distantly related to those from the "sulfur-metabolizing"branch of the archaebacteria, a physiologically diverse group which includes Sulfolobus and Pyrodyctiutn (91). Somespecies grow heterotrophically or as sulfur-oxidizing autotrophs, while other membersof this branch grow by sulfur-dependent respiration of hydrogen or organic compounds.It has recently been demonstrated that certain Sulfolobus spp. grow anaerobically by the sulfur-dependent respiration of hydrogen(83, 100). Since OctopusSpring is a low-sulfide environment, this archaebacteriumand the Thermus-like cohabitants are likely to grow heterotrophically (90). The Chino mine leaching pond yielded two distinct 5S rRNAs(57). The more abundant species is identical in sequence to that of the Thiobacillus ferrooxidans strain in our reference collection (ATCC 19859). This result consistent with previous observations that T. ferrooxidans is the predominant bacterial species cultivated from this environment,followed by Thiobacillus thiooxidans and the moderatelythermophilic, iron-oxidizing THstrains (6, 8: J. A. Brierley, personal communication). The second 5S rRNA(Chino 2) not specifically related to any organismin our reference collection, but like T. ferrooxidansand T. thiooxidans, it is affiliated with the fl subdivision of the purple bacteria (see below). Lackingreference sequences from THstrains, cannot assess its relationship to these bacteria. The invertebrate symbionts each contained two distinct 5S rRNAs:one eukaryotic, i.e. the host, and one eubacterial (58, 89). In each case, the eukaryotic 5S rRNAis very similar to those of mollusks. Althoughthis was expected of the bivalve sequences, the tube wormRiftia had been classified as a memberof a different phylum(48). Becauseof the generally small amount of sequence variation observed amongthe invertebrate 5S rRNAs, it is difficult to assess the significance of this result. Indeed, previous 5S rRNAbased analyses (54) had grouped members of three additional phyla (Brachiopoda, Coelenterata, and Porifera) amongthe mollusks. The bacterial endosymbiontsare all affiliated with the "purple bacteria" group (phylum) of the eubacteria. Oligonucleotide catalog comparisonshave delineated three main subdivisions of the purple bacteria group: 0¢, represented in Figure 1 by Agrobacteriumtumefaciens; ~, represented by Pseudomonastestosteroni; and y, represented by Escherichiacoli (97). The bacterial

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY, EVOLUTION, AND rRNA

355

symbionts all fall into the 7 subdivision, as do several strains (A7, G3, HSP, and E7) obtained as end-point dilution isolates from Riftia trophosome (46, 58). Figure 5 presents a 5S rRNA-based phylogeny of the 7 subdivision of the purple bacteria, illustrating the inferred affiliations of all these organisms. The Riftia symbiont is placed, by 5S rRNAsequence comparisons, very near the root of the tree shownin Figure 5. Its closest knownrelative (89%5S rRNAsequence similarity) is "Thiobacillus ferrooxidans" strain ml, an iron-oxidizing, Thiobacillus-like bacterium that does not appear to oxidize reduced sulfur compounds (41). None of the bacteria isolated by end-point dilution of trophosome tissue has a 5S rRNAsequence identical to that extracted from the sample ofRiftia trophosome material (58). Isolates G3 and HSP have nearly identical 5S rRNAsequences and are affiliated with the well-defined Vibrio cluster, represented in Figure 5 by Photobacterium phosphoreum and Vibrio harveyi. Isolate E7 and Pseudomonas fluorescens have

Figure5 A phylogenyof the gamma subdivisionof the purple bacteria (see text) basedon rRNAsequences.Thetree wasinferred as in Figure 1. Theanalysis also included several organisms oullsideof the group(not shown),whichcontributedto the placement of the root of this tree in the vicinity of the opencircle. Reproduced from(58).

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

356

OLSEN, LANE, GIOVANNONI,PACE & STAHL

identical 5S rRNAsequences. The 5S rRNAsmost Closely related to that Of isolate A7 are from other ’hydrothermal vent-associated prokaryotes, the -Calyptogenasymbiont(88%similarity)and Thiomicrospira strain ~L-12(see below). Recentlyan evencloser relative of isolate,A7 (at least 97%similarity) has been identified, Acinetobacter calcoaceticus (58). The Solemyagill ~prokaryote is affiliated, as a deep branching, with the fluorescent pseudomonadcluster. Its closest knownrelative (87%5S rRNA sequencesimilarity) is Thiothrix nivea, ,a sheathed, filamentousorganismthat accumulates sulfur whengrownin the presence of sulfide or thiosulfate. The relative abundance of each unique 5S rRNAextracted from an environment offers someindication of the relative amountof the corresponding microorganismin the population (89, 90). Because 5S rRNAsaccept an end label (see above) with varying efficiency (primarily owingto secondarystructure differences), the amount radioisotope incorporated into the various intact 5S rRNAsis an unreliable measure Of their relative abundance. However,this bias maybe avoided by first digesting the mixture of 5S RNAsto completion with RNaseT1, and then labeling the resultant oligonucleotides at their 5’ termini with [3/32p]ATPand polynucleotide kinase. The oligonucleotides are fractionated by two-dimensionalhigh-voltage paper electrophoresis (79) and the radioactive content of each is determined. Since the oligonucleotides-are efficiently and uniformly labeled, the relative amounts of label incorporated into those oligonucleotides that are unique to a specific 5S rRNA.providean estimate of its relative abundancein the original population. RELATIVE ABUNDANCEOF UNIQUE 5S rRNAs

LIMITATIONS OF THE 5S rRNA-BASED ANALYSES .PQpulation

analysis by

5S rRNAsequences has two major limitations: (a) Complex mixtures similarly sized molecules must be separated, and (b) relatively little phylogenetic information is available in the molecule (--120 nucleotides). Current fractionation techniques rely upon high-resolution gel electrophoresis. Althoughthe fractionation of --10 5S rRNAsis feasible, more complexmixtures are problematic. The contribution by individual organisms of 5S rRNAswith terminal ’or internal length heterogenei.ty may,also confound fractionation and analysis of somepopulations. Althoughimprovedtechniques of nucleic acid fractionation can be anticipated, the precision Of 5S rRNA-basedphylogenetic analysis will remain limited by the small size of the molecule. One exampleof the problems that can arise from.this is the placementof Thiomicrospirastrain E-12, an isolate from a sulfide diffusion-gradient enrichment inoculated with water and par,ticulate,matter’.from,the GalhpagosRift (77). In previous 5S rRNAcharacterizations (without the Riftia trophosomeend-p0int dilution.isolates), .T.hiomicrospira strain L-12 had been placed, along with Thiomicrospira pelophila,

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY,EVOLUTION,AND rRNA

357

as a peripheral affiliate of the fluorescent pseudomonad cluster (57, 89); this placementwas consistent with their striking morphologicaland physiological similarities (77). However, in the absence of the A7 rRNAsequence, Thioraicro.’~pirastrain L-12 could be movedfrom the vicinity of the fluorescent pseudomonadcluster to the Calyptogenasymbiont branch (as shownin Figure 5) with a minimal(<0.1%)increase in the mean-squaretree error. .addition of the A7sequenceto the tree appears to cementthe latter placement, owingto its similarity to both’the Calyptogenasymbiontand Thiomicrospira strain L-12, sequences(58). The lesson here is that the accuracy of branching order inference is limited by statistical’fluctuations in such a small molecule’s reflection of evolution. Rearrangementsin the vicinity of short tree segments can somet!imes be effected by single nucleotide changes in the sequences, especially whentree segments are inferred from small differences between large evolutionary separations. Phylogenetic branching orders based on 5S rRNAare.also of limited accuracy if the comparedsequences are too similar (i.e. vary at too few positions). However,5S rRNAanalysis should remain useful adjunct to 16S rRNA-basedanalysis. It can also provide an independent assessment of population complexity, composition, and change through tiine. Population

Analysis

by 16S Ribosomal

RNA Genes

The Octopus Spring microbial communityand the invertebrate symbionts have relatively few major constituents, and thus their 5S rRNApopulations could be fully resolved in vitro..However, the 5S rRNAmethod is not sufficiently sensitiveto detect the minorcomponentsof mixedpopulations or to resolve complexpopulations. Analternative approach involves cloning the ’16S rRNAgenes from DNAisolated from environmental samples. Although this approach is still under development(using DNAfrom OctopusSpring), requires fewer cells than 5S rRNAanalysis and offers greater potential for detecting r~ainor:population constituents. As discussed above, the 16S rRNA also permits moreprecise characterization of evolutionary affiliations. Population analysis using 16S rRNAgenes (rDNA) involves technical considerations somewhatdifferent from those involved in 5S rRNAanalysis (compare ]Figures 2a and 2b). Because 5S rRNAis small and resistant mechanicalshear, a goodyield of intact moleculesis fairly easy to isolate. ,DNA,in contrast, is highly susceptible to mechanicalshear and, to be suitable for cloning, should be recovered in relatively high-molecular weight form. Mostof the manyDNA extraction protocols in the literature (73) are applicable .to the microbial material obtained from environmental samples; cell breakage slaould be monitored by microscopy. DNAof 25-50 kb (determined by agarose gel electrophoresis) was.preparedfrom the OctopusSpring source pool biomassby treatment with lysozyme,digestion with proteinase K in the presence of sodium dodecyl sulfate, and extraction with phenol at pH 9.

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

358

OLSEN, LANE, GIOVANNONI,PACE & STAHL

The sample DNAis "shotgun cloned" into a phage lambda cloning vector, producing a recombinant"library" whichcan be amplified and probed for 16S rDNA--containingclones (Figure 2b; for a general discussion of gene cloning see Reference 66). The Octopus Spring DNAwas partially digested with the restriction endonuclease Sau3A, and 10-15-kb fragments were selected by agarose gel electrophoresis. Since Sau3Arecognition sites are located, on average, at 256-bp intervals, the products of the partial digestion were a nearly random collection of overlapping DNAfragments. These fragments were ligated into the BamHIcloning site of the phage lambdavector L47.1 (62), and the resulting recombinant phage DNAwas "packaged" into infectious lambdaphageparticles in vitro (24, 75). A numberof hybridization probes have been tested for detection of rDNA clones on membrane filter replicas of phageplaques (4). In current use (72) a "mixedkingdom"probe, i.e. a mixture of 16S rRNAs,one from each of the primary kingdoms, which have been alkali-fragmented and radioactively labeled (65). Becausesubstantial portions of 16S rRNAsequences are similar among all known organisms, low-stringency hybridization permits the identification of rRNAgenes from unknownorganisms. Initial characterizations have used a mixture of E. coli (eubacterial), Sulfolobus solfataricus (archaebacterial), and Dictyostelium discoideum (eukaryotic) 16S rRNAs. Hybridization conditions that minimizebackgroundhybridization to nonribosomal DNA,yet that are of sufficiently low stringency that even crosskingdom 16S rRNA/rDNA hybridization takes place, have been empirically determined (72). Of the recombinant phage derived from the Octopus Spring source-pool DNA,0.2-0.3% contained 16S rDNAinserts. This is about the percentage of rDNAin the typical bacterium, which suggests that rDNAwas uniformly recovered from the mixed population. Subsequent to the identification of the 16S rDNA-containingclones, redundant genes are identified, and the sequences are determined for each unique gene (presumablyone per source organism). Several approaches to the sorting and sequencing steps are being explored. The methodsin common use (sorting by a combinationof restriction enzymeanalysis and hybridization, followed by subcloning into single-stranded bacteriophage M13for sequencing) are cumbersomefor screening large numbers of clones (66). Shotgun cloning directly into M13 wouldexpedite these steps of the analysis, since the 16S rDNAsequences could immediately be determined using the universal rRNAprimers discussed above. However,large DNAinserts are unstable in this vector (68; D. J. Lane, unpublished). Hong (44) has reported a procedure for direct dideoxynucleotideterminated sequencing using double-stranded phage lambda DNAtemplates, DNApolymerase (Klenow fragment), and 20-nucleotide synthetic oligodeoxynucleotide primers. Wehave so far been unable to reproduce this result with rDNA-containing lambda DNAtemplates and the 16S rRNA-specific

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

ECOLOGY, EVOLUTION, AND rRNA 359 primers. However,adaptation of the protocol for use with reverse transcriptase did permit 100-150 nucleotides of the sequence to be read from many rDNA-containing lambda clones, although the legibility of the sequence information varied amongclones (D. J. Lane, unpublished). While this amountof the sequence is suboptimal for phylogenetic purposes, it does provide a relatively rapid meansof sorting the lambdaclones into uniquesets representing each of the different membersof the sampledpopulation. These are then subcloned and sequenced in phage M13for more detailed analysis, again using the 16S rRNA-specific primers, As with the 5S rRNAanalysis, the derive,d 16S rRNAsequences are comparedto others in the reference collection of complete and partial 16S rRNAsequences (see above), and the organismsthat contributed these sequences are phylogenetically classified. Data so f~r available from the Octopus Spring rDNAclones are consistent with the results of the 5S rRNAanalysis, but promise to yield moredetail. Manyaspects of the cloning approachto natural populationanalysis are still being dew:loped and evaluated. Assumptionsabout the uniform "clonability" of heterogeneousrRNAcistrons are still largely untested, and certain of the required technical manipulations remain cumbersome for routine use. Historically, rec.overy of recombinantrRNAcistrons has not proven particularly difficult, although cloning of exceptionally active promoters or highly modified DNAsequences can be troublesome. The restriction enzymeSau3A,’ which is iinsensitive to DNAmethylation and which generates relatively random fragments of the sample DNA,is used to avoid these potential pitfalls. Tlae cumbersomeaspects of the sorting and sequencing steps are technical and there seemslittle doubt that future improvements in bifunctional, cloning;/sequencingvectors will further streamline these operations.

IN SITU HYBRIDIZATION FOR COUNTING AND IDENTIFYING ORGANISMS The analysis of cloned rRNAgenes from a mixed population of microorganisms offer:s phylogenetic characterization of the resident organisms, but it does not provide a goodestimate of their abundance.This information can be derived by quantitating the hybridization between each of the rDNAclones and the DNAextracted from the population of organisms. A novel alternative approachntowbeing developeduses in situ hybridization techniques to detect organismospecific or group-specific sequences in the rRNA.This should permit the direct visual identification and counting of the corresponding organisms. The methodwouldbe somewhatanalogous to the use of fluorescently labeled antibodies against cell surface antigens, but without the prerequisite for organismcultivation prior to raising antibodies. In addition, the phylogenetic breadth of the probe can be chosen (see below). In situ hybridization (reviewed in 1) uses nucleic acid probes to detect

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

360

OLSEN, LANE; GIOVANNONI,PACE & STAHL

complementarysequences inside fixed cells or in fixed, thin sections of tissues. It has been widely used for detecting mRNAs (1) and viral nucleic acids (39) in cells, anff for localizing genes in chromosomes(40). We apply the methodsto natural microbial populations using contact slides .or environmental samples. Microscopic autoradiography has frequently been used to detect the binding of radioactively labeled (3H or 35S) nucleic acid probes. This methodis sensitive, and bound probes can be quantitated , by countingsilver grains in a photographicemulsionoverlayingtl~e fixed cells or tissue. Nucleic acid probes can also be labeled with fluorescent compounds and subsequently detected by fluorescence microscopy(2, 3, 59, 84).. Some methodsfor detecting nucleic acid probes by fluorescence, have relied upon the binding of fluor-coupled avidin to biotinated nucleotides (e.g. 59). Since fixed cells maynot be freely permeable to large molecules, methodsusing avidin (MW= --68,000) may not be generally useful for the analysis natural microbial populations, in whichmuchvariability in cell wall structure must be expected. However,fluorescent compoundscoupled directly to the nucleic acid probes should be usable. The rRNAsare uniquely attractive targets for in situ hybridization probes because of their abundance; each cell contains --10,000 ribosomes. While it is possible to detect --50 copies of a nucleotide sequencein a cell (17), the presence of manycopies increases the signal obtained, reducing problemsdue to nonspecific, backgroundbinding. Abundanttarget sequences also allow the use of fluorescently labeled probes, which are not as efficiently detected as radioactive probes. Twogeneral approaches toward the identification of microorganismsby in situ hybridization are being developed. Oneuses organism-specific hybridization probes, whichcan be produced,in several, ways. Restriction. fragments. containing all or part of the rRNAgenes can be cloned and labeled by nick translation, generating a duplex, or "symmetric,"probe (1). Alternatively, the restriction fragments can be cloned adjacent to the bacteriophage promoterin a "transcription vector" and used to producetranscripts complementary to the rRNA(an "asymmetric" probe) (17). An asymmetric probe specific for rRNAof an organism in culture can be efficiently produced by reversetranscribing cellular RNAfrom one of the "universal" 16S rRNA-specific primers (see above). With any of the above probes, high-stringency hybridization is used to identify organismsthat specifically bind them. The secondapproach uses synthetic oligodeoxynucleotidesthat are specific for restricted phylogenetic groups of organisms. As discussed above in the context of "signature" sequences, someshort sequences are diagnostic of 16S rRNAsfrom particular phylogenetic groups, so oligonucleotides complementary to those sequences should help identify organisms from those groups in the context of mixedpopulations. Current experience is limited to oligonucleotide probes that are specifiC at. the. kingdom,level. Based on

Annual Reviews www.annualreviews.org/aronline ECOLOGY, EVOLUTION, AND rRNA

361

complete 16S rRNAsequences, we have synthesized probes complementary to sequences diagnostic of the eukaryotic, eubacterial, or archaebacterial rRNAs. These indeed allow the determination of the kingdom affiliation of unknown cells from natural populations fixed to microscope slides. Future directions include the development of more selective probes for more limited phylogenefic groups.

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

SUMMARYAND FUTURE PROSPECTS The techniques of molecular biology are constantly opening new domains to microbiology. Molecular shylogeny and macromolecular sequencing have made possible the natural classification of microorganisms, allowing us to substitute a molecular record for the fossil record used to characterize metazoan evolution. The ribosomal RNAs provide a measurable connection among all organisms, permitting the development of a universal phylogeny. By analyziing mixtures of rRNAs (or rRNAgenes), it is now possible phylogenetically identify the component organisms of mixed populations. The phylogenetic characterization of microorganisms that have not been grown in the laboratory provides an opportunity to explore beyond the culture collections in assessing the diversity of terrestrial life. The techniques discussed above will facilitate other studies as well. For example, tlae 5S rRNAoligonucleotide analysis of population composition is well-suited to study of the stability of populations through time, or of the changes following perturbation. Also, the rapid sequencing techniques discussed in n~lation to 16S rRNAare equally applicable to the 23S rRNA(74a). Finally, the data gathered for phylogenetic studies is fodder for comparative analyses ol." rRNAsecondary and tertiary structure (38, 70). ACKNOWLEDGMENTS This work was supported by National Institutes NRP. Literature

of Health Grant GM34527to

Cited

1. Angerer,R. C., Cox, K. H., Angerer, L. M. 1985. In situ hybridization to cellular RNAs. In GeneticEngineering, ed. J. K. Setlow, R. Hollaender,7:4365. NewYork/London:Plenum 2. Banman,J. G. J., Weigant, J., van Duijn, P. 1981. Cytochemical hybridization with fluorochrome-labeledRNA. I. Development of a methodusing nucleic acids boundto agarose beads as a model. J. Histochem. Cytochem. 29:227-37 3. Bauman,J. G. J., Weigant, J., van Duijn, P. 1983.Thedevelopment, using.

poly(Hg-U)in a modelsystem,of a new method to visualize cytochemical hybridization in fluorescencemicroscopy. J. Histochetn.Cytochem.34:57178 Benton, W. D., Davis, R. W. 1977. Screeninglambda-gtrecombinantclones by hybridization to single plaques in situ. Science 196;180-82 Biggin,M.D., Gibson,T. J., Hong,G. F. 1983. Buffer gradient gels and 3sS label as an aid to rapid DNA sequence determination. Proc. Natl. Acad.Sci. USA80:3963-65

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

362

OLSEN,

LANE,

GIOVANNONI,

6. Brierley, J. A. 1978. Thermophilicironoxidizing bacteria found in copperleaching dumps. Appl. Environ. Microbiol. 36:523-25 7. Brierley, J. A. 1985. Use of microorganisms for mining metals. In Engineered Organismsin the Environment: Scienti~c Issues, ed. H. O. Halvorson, pp. 141-46. Washington DC: Am. Soc. Microbiol. 239 pp. 8. Briedey, J. A., Norris, P. R., Kelly, D. P., Le Roux, N. W. 1978. Characteristics of a moderately thermophilic and acidophilic iron-oxidizing Thiobacillus. Eur. J. Appl. Microbiol. Biotechnol. 5:291-99 9. Brock, T. D. 1978. Thermophilic Microorganisms and Life at High Temperatures. NewYork: Springer-Vedag. 465 PP. 10. Brosius, J., Dull, T. J., Sleeter, 13. D., Noller, H. F. 1981. Gene organization and primary structure of a ribosomal RNAoperon from Escherichia coli. J. Mol. Biol. 148:107-27 11. Brown, W. M., George, M. Jr., Wilson, A. C. 1979. Rapid evolution of animal mitochondrial DNA.Proc. Natl. Acad. Sci. USA 76:1967-71 12. Cavanaugh, C. M. 1983. Symbiotic chemoautotrophicbacteria in marine invertebrates from sulfide-rich habitats. Nature 302:58-61 13. Cavanaugh,C. M., Gardiner, S., Jones, M. L., Jannasch, H. W., Waterbury, J. B. 1981. Prokaryotic cells in the hydrothermal vent tubewormRiftia pachyptila Jories: Possible chemoautotrophicsymbionts. Science 213:340-42 14. Cedergren, R. J., Sankoff, D., LaRue, B., Grosjean, H. 1981. The evolving tRNAmolecule. CRCCrit. Rev. Biochem. 11:35-104 15. Chao, S., Sederoff, R., Levings, C. S. III. 1984. Nucleotide sequence and evolution of the 18S ribosomal RNA gene in maize mitochondria. Nucleic Acids Res. 12:6629-44 16. Colless, D. H. 1970. The phenogramas an estimate of phylogeny. Syst. Zool. 19:352-62 17. Cox. K. H., DeLeon, D. V., Angerer, L. M., Angerer, R. C. 1984. Detection of mRNAs in sea urchin embryos by in situ hybridization using asymmetric RNAprobes. Dev. Biol. 101:485-502 18. Curtiss, W. C., Vournakis, J. N. 1984. Quantitation of base substitutions in eukaryotic 5S rRNA:Selection for the maintainence of RNAsecondary structure. J. Mol. Evol. 20:351-61 19. Delihas, N., Anderson, J. 1982. Generalized sa-uctures of the 5S ribosomal

PACE & STAHL RNAs. Nucleic Acids Res. 10:732320. Demoulin,V. 1979. Protein and nucleic acid sequence data and phylogeny. Science 205:1036-38 21. Donis-Keller, H., Maxam,A., Gilbert, W. 1977. Mappingadenines, guanines, and pyrimidines in RNA.Nucleic Acids Res. 4:2527-38 22. Edmond, J. M., Von Damm,K. 1983. Hot springs on the ocean floor. Sci. Am. 248:78-93 23. Elwood,H. J., Olsen, G. J., Sogin, M. L. 1985. The small-subunit ribosomal RNAgene sequences from the hypotrichous eiliates Oxytrichanovaand Stylonychia pustulata. Mol. Biol. Evol. 2:399--410 24. Enquist, L., Sternberg, N. 1979. In vitro packaging of lambda Damvectors and their use in cloning DNAfragments. Methods Enzymol. 68:281-98 25. Erdmann, V. A., Wolters, J., Huysmans, E., De Wachter, R. 1985. Collection of published 5S, 5.8S and 4.5S ribosomal RNAsequences. Nucleic Acids Res. 13:r105-r153 (Suppl.) 26. Felbeck, H., Childress, J. J., Somero, G. N. 1981. Calvin-Benson cycle and sulfide oxidation enzymes in animals from sulfide-rich habitats. Nature 293: 291-93 27. Felsenstein, J. 1978. Cases in which parsimonyor compatibility methodswill be positively misleading. Syst. Zool. 27:401-10 28. Felsenstein, J. 1981. Evolutionarytrees from DNA sequences: A maximum likelihood approach. J. Mol. Evol. 17: 368-76 29. Felsenstein, J. 1982. Numerical methods for inferring evolutionary trees. Q. Rev. Biol. 57:379~404 30. Fitch, W. M. 1976. The molecular evolution of cytochromec in eukaryotes. J. Mol. Evol. 8:13-40 31. Fitch, W. M., Margoliash, E. 1967. Construction of phylogenetic trees: A methodbased on mutational distances as estimated from cytochrome c sequences is of general applicability. Science 155:279-84 32. Fox, G. E., Pechman, K. R., Woese, C. R. 1977. Comparativecataloging of 16S ribosomal ribonucleic acid: Molecular approachto procaryotic systematics. Int. J. Syst. Bacteriol. 27:44-57 33. Fox, G. E., Stackebrandt, E., Hespell, R. B., Gibson, J., Maniloff, J., et al. 1980. The phylogeny of prokaryotes. Science 209:457-63 34. Golding, G. B. 1983. Estimation of DNAand protein sequence divergence:

Annual Reviews www.annualreviews.org/aronline ECOLOGY,

35.

36.

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

37.

38.

39.

40.

41.

42.

43.

44.

45. 46.

47.

An examination of some assumptions. Mol. Biol. Evol. 1:125-42 Goodman,M., ed. 1982. Macromolecular Sequences in Systematic and Evolutionary Biology. NewYork/London: Plenum. 418 pp. Green, C. J., Stewart, G. C., Hollis, M. A., Void, B. S., Bott, K. F. 1985. Nucleotide sequenceof the Bacillus subtilis ribosomal RNAoperon, rrnB, Gene 37:261--66 Gupta, R., Lanter, J. M., Woese,C. R. 1983. :~equence of the 16S ribosomal RNAfirom Halobacterium volcanii, an archaebacterium. Science 221:656-59 Gutell, R. R., Weiser, B., Woese, C. R., Noller, H. F. 1985. Comparative anatomy of 16S-like ribosomal RNA. Prog. Nucleic Acid Res. Mol. Biol. 32:155--216 Haase, A. T., Walter, D., Stowring, L., Ventura, R., Eerballe, A., et al. 1985. Detection of viral genomes in single cells by double-label hybridization in situ and color microradioautography. Science 227:189-91 Harper, M. E., Saunders, G. F. 1981. Localization of single copy DNAsequence.,; of G-banded humanchromosomesby in situ hybridization. Chromosoma 83:431-39 Harrison, A. P. Jr. 1984. The acidophilic thiobacilli and other acidophilic bacteria that share their habitat. Ann. Rev. Microb~!ol. 38:265-92 Hasegawa, M., Iida, Y., Yano, T., Takaiwa, F., Iwabuchi, M. 1985. Phylogenetic relationships amongeukaryotic kingdoms inferred from ribosomal RNAsequences. J. Mol. Evol. 22:32-218 Holmqttist, R., Pearl, D., Jukes, T. H. 1982. Nonuniform molecular divergence: The quantitative evolutionary analysis of genes and messengerRNAsunder selective constraints. See Ref. 35, pp. 281-315 Hong, G. F. 1982. Sequencing of large double-stranded DNAusing the dideoxy sequencing technique. Biosci. Rep. 2: 907-12 Jannasch, H. W., Mottl, M. J. 1985. Geomicrobiologyof deep-sea hydrothermal veuts. Science 229:717-25 Jannasch, H. W., Nelson, D. C. 1984. Recent progress in the microbiology of hydrothermal vents. In Current Perspectives in Microbial Ecology, Proc. 3rd Int. Symp. Microb. Ecol., East Lansing, Mich., 1983, ed. M. J. Klug, C. A. Reddy, pp. 170-76. Washington DC: Ara. Soc. Microbiol. Jarsch, J., Brck, A. 1985. Sequence of

EVOLUTION,

AND rRNA

363

the 16S ribosomal RNAgene from Methanococcusvannielii: Evolutionary implications. Syst. Appl. Microbiol. 6: 54~59 48. Jones, M. L. 1985. On the Vestimentifera, new phylum: Six new species, and other taxa, from hydrothermal vents and elsewhere. In The Hydrotherreal Vents of the East Pacific Rise: An Overview, Bull. Biol. Soc. Wash., No. 6, ed. M. L. Jones, pp. 117-58. Vienna, Va: Infax. 547 pp. 49. Deleted in proof 50. Jukes, T. H., Cantor, C. R. 1969. Evolution of protein molecules. In MammalianProtein Metabolism, ed. H. N. Munro, pp. 21-132. NewYork: Academic 51. Kimura, M. 1981. Estimation of evolutionary distances between homologous nucleotide sequences. Proc. Natl. Acad. Sci. USA78:45zl.-58 52. Kimura, M., Ohta, T. 1972. On the stochastic modelfor estimation of mutational distance between homologous proteins. J. Mol. Evol. 2:87-90 53. Klotz, L. C., Blanken, R. L. 1981. A practical methodfor calculating evolutionary trees from sequence data. J. Theor. Biol. 91:261-72 54. Komiya, H., Hasegawa, M., Takemura, S. 1983. Nucleotide sequence of 5S rRNAsfrom sponge Halichondria japonica and tunicate Halocynthia roretzi and their phylogeneticpositions. Nucleic Acids Res. 11:1969-74 55. Kiintzel, H., KGchel, H. G. 1981. Evolution of rRNA and origin of mitochondria. Nature 293:751-55 56. Lane, D. J., Pace, B., Olsen, G. J., Stahl, D. A., Sogin, M. L., Pace, N. R. 1985. Rapid determination of 16S ribosomal RNAsequences for phylogenetic analyses. Proc. Natl. Acad. Sci. USA 82:6955-59 57. Lane, D. J., Stahl, D. A., Olsen, G. J., Heller, D. J., Pace, N. R. 1985. A phylogenetic analysis of the genera Thiobacillus and Thiomicrospira by 5S ribosomal RNAsequences. J. Bacteriol. 163:75-81 58. Lane, D. J., Stahl, D. A., Olsen, G. J., Pace, N. R. 1985. Analysis of hydrothermal vent-associated symbionts by ribosomal RNAsequences. See Ref. 48, pp. 389-400 59. Langer, P. R., Waldrop, A. A., Ward, D. C. 1981. Enzymatic synthesis of biotin-labeled polynucleotides: Novel nucleic acid affinity probes. Proc. Natl. Acad. Sci. USA 78:6633-37 60. Leffers, H., Garrett, R. A. 1984. The nucleotide sequence of the 16S ribosom-

Annual Reviews www.annualreviews.org/aronline

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

364

OLSEN,

LANE,

GIOVANNONI,

al RNAgene of the archaebacterium Halococcus morrhua. EMBOJ. 3:161319 61. Li, W. H. 1981. A simple method for constructing phylogenetic trees from distance matrices. Proc. Natl. Acad. Sci. USA 78:1085-89 62. Loenen, W. A., Brammer, W. J. 1980. A bacteriophage lambda vector for cloning large DNAfragments made with several restriction enzymes. Gene 10: 249-59 63. MacDonell, M. T., Colwell, R.. R. 1985. Phylogeny of the Vibrionaceae, and recommendation for two new genera, Listonella and Shewanella.Syst. Appl. Microbiol. 6:171-82 64. MacKay, R. M., Spencer, D. F., Schnare, M. N., Doolittle, W. F., Gray, M. W. 1982. Comparative sequence analysis as an approach to evaluating structure, function, and evolution of 5S and 5.8S ribosomal RNAs.Can. J. Biochem. 60:480-89 65. Maizels, N. 1976. Dictyostelium 17S, 25S, and 5S rDNAslie within a’38,000 base pair repeated unit. Cell 9:431-38 66. Maniatis, T., Fritsch, E. F., Sambrook, J. 1982. Molecular Cloning: A Laboratory Manual. New York: Cold Spring Harbor Lab. 545 pp. 67. McCarroll, R., Olsen, G. J., Stahl, Y. D., Woese, C. R., Sogin, M. L. 1983. The nucleotide sequence of the Dictyostelium discoideum small subunit ribosomal RNAinferred from the gene sequence: Evolutionary implications. Biochemistry 22:5858-68 68. Messing, J. 1983. New MI3 vectors. for cloning. Methods Enzymol. 101:2079 69. Messing, J., Carlson,~J., Hagen, G:, Rubenstein, I., Oleson!: A. 1984. Cloning and sequencing 6f the ribosomal RNAgenes in maize: The 17S region. DNA3:31--40 70. Olsen, G. J. 1983. C~arative analysis of nucleotide sequence dam. PhD thesis. Univ. Colo. H6alth Sci. Center, Denver. 163 pp. 71. Olsen, G. J., Pace, N. R., Nuell, M., Kaine, B. P., Gupta, R., Woese, C. R. 1985. Sequence of the 16S rRNAgene from the thermoacidophilic archaebacterium Sulfolobus solfataricus and its evolutionary implications. J. Mol. Evol. 22:301-7 72. Pace, N. R., Stahl, D. A., Lane, D. J., Olsen, G. J. 1986. The analysis of natural microbial populations by ribosomal RNAsequences. Adv. Microb. Ecol. In press 73. Parish, J. H. 1972. Principles andPrac-

PACE & STAHL tice of Experimentswith Nucleic Acids, pp. 104-11. London: Longman 74. Peattie, D. A. 1979. Direct chemical method for sequencing RNA. Proc. Natl. Acad. Sci. USA 76:1760-64 74a. Qu, L.. H., Michot, B., Bachellerie,. J.-P. 1983: Improvedmethodsfor structure probing in large RNAs:a rapid ’heterologous’ sequencing approach is coupled to the direct mappingof nuclease accessible sites. Application to the 5’ terminal domain of eukaryotic 28S rRNA. Nucleic Acids Res. 11:5903-20 75. Rosenberg, S~ M., Stahl, M. M., Kobayashi, I., Stahl, F. W. 1985. Clean and simple one-strain in vitro packaging of bacteriophage lambda DNA. ASM News 51:386-91 76. Rubtsov, P. M., Musakhanov, M. M., Zakharyev, V. M., Krayev, A. S., Skryabln, K. G., Bayev, A. A. 1980. The structure of the yeast ribosomal RNAgenes. I. The complete nucleotide sequence of the 18S ribosomal RNA gene from Saccharomyces cerevisiae. Nucleic Acids Res. 8:5779-94 77. Ruby, E. G., Jannasch, I4. W. 1982. Physiological characteristics of Thiomicrospira sp. strain L-12 isolated from deep-sea hydrothermalvents. J. Bacteriol. 149:161-65 78. Salim, M., Maden, B. E. H. 1981. Nucleotide sequence of Xenopuslaevis 18S ribosomal RNAinferred from gene sequence. Nature 291:205-8 79. Sanger, F., Brownlee, G. G., Barrell, B. G. 1965. A two-dimensional. fractionation procedure for radioactive nucleotides. J. Mol. Biol. 13:373-98 80. Sanger, F., Nicklen,~S., Coulson, A. R. 1977. DNAsequencing with chainterminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463-67 81. Schwartz, R. M., Dayhoff, M. O. 1978. Origins of prokaryotes, eukaryotes, mitochondria, and chloroplasts: Aperspective is derived from protein and nucleic acid sequence data. Science 199:395-403 82. Schwarz, Zs., K6ssel, H. 1980. The primary structure of 16S rRNAfrom Zea mays chloroplast is homologous to E. coli 16S rRNA. Nature 283:739-42 83. Segerer, A., Stetter, K. O., Klink, F. 1985. Two contrary modes of chemolithotrophy in the same archaebacterium. Nature 313:787-89 84. Smith, L. M., Fung, S., Hunkapiller, M. W., Hunkapiller, T. J., Hood,L. E. 1985. The synthesis of oligonucleotides containing an aliphatic amino group at the 5’ terminus: Synthesis of fluorescent DNApalmers for use in DNAsequence

Annual Reviews www.annualreviews.org/aronline ECOLOGY,

85.

86.

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

87.

88.

89.

90.

91.

92.

analysis. Nucleic Acids Res. 13:2399413 Sogin, M. L., Elwood, H. J., Gunderson, J. H. 1986. Evolutionary diversity of eukaryotic small subunit rRNAgenes. Proc. Natl..Acad. Sci. USA 83:1383-87 Sokal, R. R., Sheath, P. H. A. 1973. Numerical Taxonomy. San Francisco: Freem~aa.573 pp. Stackebrandt, E,, Ludwig, W., Fox, G. E. 1985. 16S ribosomal RNAoligonucleotide cataloging. In Methods in Microbiology, ed. G. Gottschalk, 18:75-107. NewYork: Academic Stackebrandt, E., Woese, C. R. 1981. The evolution of prokaryotes. In MolecMar and Cellular Aspects of Microbial Evolution, ed. M.J. Carlisle, J. R. Collins, B. E. B. Moseley, pp. 1-31. Cambridge: Cambridge Univ. Press Stahl, 13. A., Lane, D. L, Olsen, G. L, Pace, N. R. 1984. Analysis of hydrotherma]t vent-associated syrnbionts by ribosomal RNAsequences. Science 224:40!)-11 Stahl, D. A., Lane, D. J., Olsen, G. J., Pace, N. R. 1985. Characterization of a Yellowstone hot spring microbial community by 5S ribosomal RNA sequence:~. Appl. Environ. Microbiol. 49: ~1379-84 Stetter, K. O., Zillig, W. 1985. Thermoplasma and the thermophilic sulfurdependent archaebacteria. See Ref. 98, pp. 85--170 Tomioka, N., Sugiura, M. 1983. The complete nucleotide sequence of a 16S ribosomal RNAgene from a blue-green alga, Anacystis nidulans. Mol. Gen. Genet. 191:46-50

EVOEUTION,

AND rRNA

365

93. Weisburg, W. G., Oyaizu, Y., Oyaizu, H., Woese, C. R. 1985. Natural relationship between baeteriodes and flavobacteria. J. Bacteriol. 164:23036 94. Woese, C. R., Fox, G. E. 1977. .Phylogeneticstructure of the prokaryotie domain: The primary kingdoms. Proc. Natl. Acad. Sci. USA 74:5088-90 95. Woese, C., Sogin, M., Stahl, D., Lewis, B. J., Bonen, L. 1976. A comparison of the 16S ribosomal RNAsfrom mesophilic and thermophilic bacilli: Some modifications in the Sanger method for RNAsequencing. J. Mol. Evol. 7:197-213 96. Woese,C. R., Stackebrandt, E., ’Ludwig, W. 1985. What are mycoplasmas: The relationship of tempo and modein bacterial evolution. J. Mol. Evol. 21: 305-16 97. Woese, C. R., Stackebrandt, E., Macke, T. J., Fox, G. E. 1985. A phylogenetic definition of the major eubacterial taxa. Syst. Appl. Microbiol. 6:143-51 98. Woese,C. R., Wolfe, R. S., eds. 1985. The Bacteria, Vol. 8, Archaebacteria. Orlando: Academic. 582 pp. 99. Yang, D., Oyaizu, Y., Oyaizu, H., Olsen, G. J., Woese, .C.R. 1985. Mitochondrial origins. Proc. Natl. Acad. Sci. USA82:4443-47 100. Zillig, W., Yeats, S., Holz, I., B~ck, A., Gropp, F., et al. 1985. Plasmidrelated anaerobic autotrophy of the novel archaebactedumSulfolobus ambivalens. Nature 313:789-91 101. Zuckerkandl, E., Pauling, L. 1965. Molecules as documentsof evolutionary history. J. Theor. Biol. 8:357-66

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.

Annu. Rev. Microbiol. 1986.40:337-365. Downloaded from arjournals.annualreviews.org by 192.244.210.205 on 10/25/05. For personal use only.