Vol. 172, No. 7JOURNAL OF BACTERIOLOGY, JUlY 1990, p. 3609-36190021-9193/90/073609-11$02.00/0Copyright © 1990, American Society for Microbiology

Diversity and Origin of Desulfovibrio Species: PhylogeneticDefinition of a Family

RICHARD DEVEREUX,1t SHAO-HUA HE,2 CAROLYN L. DOYLE,1t SILVIA ORKLAND,3 DAVID A. STAHL,'JEAN LEGALL,2 AND WILLIAM B. WHITMAN3*

Department of Veterinary Pathobiology and Microbiology, University of Illinois at Champaign-Urbana, Urbana, Illinois61801,' and Departments ofBiochemistry2 and Microbiology,3 University of Georgia, Athens, Georgia 30602

Received 18 January 1990/Accepted 9 April 1990

The different nutritional properties of several Desulfovibrio desulfuricans strains suggest that either thestrains are misclassified or there is a high degree of phenotypic diversity within the genus Desulfovibrio. Theresults of partial 16S rRNA and 23S rRNA sequence determinations demonstrated that Desulfovibriodesulfuricans ATCC 27774 and "Desulfovibrio multispirans" are closely related to the type strain (strain Essex6) and that strains ATCC 7757, Norway 4, and El Agheila Z are not. Therefore, these latter three strains ofDesulfovibrio desulfuricans are apparently misclassified. A comparative analysis of nearly complete 16S rRNAsequences in which we used a least-squares analysis method for evolutionary distances, an unweighted pairgroup method, a signature analysis method, and maximum parsimony was undertaken to further investigatethe phylogeny of Desulfovibrio species. The species analyzed were resolved into two branches with origins deepwithin the 8 subdivision of the purple photosynthetic bacteria. One branch contained five deep lineages, whichwere represented by (i) Desulfovibrio salexigens and Desulfovibrio desulfuricans El Agheila Z; (ii) Desulfovibrioafricanus; (iii) Desulfovibrio desulfuricans ATCC 27774, Desulfomonas pigra, and Desulfovibrio vulgaris; (iv)Desulfovibtio gigas; and (v) Desulfomicrobium baculatus (Desulfovibrio baculatus) and Desulfovibrio desulfuri-cans Norway 4. A correlation between 16S rRNA sequence similarity and percentage of DNA relatednessshowed that these five deep lineages are related at levels below the minimum genus level suggested by Johnson(in Bergey's Manual of Systematic Bacteriology, vol. 1, 1984). We propose that this branch should be groupedinto a single family, the Desulfovibrionaceae. The other branch includes other genera of sulfate-reducingbacteria (e.g., Desulfobacter and Desulfococcus) and contains Desulfovibrio sapovorans and Desulfovibriobaarsii as separate, distantly related lineages.

The obligately anaerobic, dissimilatory, sulfate-reducingbacteria utilize sulfate as their terminal electron acceptorand derive their energy for growth from the oxidation oforganic compounds and hydrogen gas (27). Among thegenera of sulfate-reducing bacteria, perhaps the most thor-oughly studied species are classified in the genus Desul-fovibrio. The Desulfovibrio species which have been char-acterized are nutritionally similar; they are readily enrichedwith lactate or hydrogen and utilize only a few other sub-strates (38). Diagnostically, Desulfovibrio species test posi-tive (with some exceptions) for the sulfite reductase desulfo-viridin and contain the unique tetraheme cytochrome C3 (27).These observations have contributed to the notion thatalthough the sulfate-reducing bacteria are ecologically verysignificant, they comprise a small and nutritionally limitedgroup. However, recent isolations and descriptions of newgenera have dispelled this view (38). At present, more than10 genera of sulfate-reducing eubacteria (one of which isaffiliated with gram-positive bacteria) and an extremelythermophilic sulfate-reducing archaebacterium have beendescribed (36, 39).While the nutritional versatility and phylogenetic diversity

of the sulfate-reducing bacteria have been established, ques-tions of relatedness among species of the genus Desul-fovibrio remain unanswered. Postgate has stated that the"taxonomic picture" of the genus is "imperfect" (27).Indeed, a number of observations suggest divergent evolu-

* Corresponding author.t Present address: Microbial Ecology and Biotechnology, U.S.

Environmental Protection Agency, Gulf Breeze, FL 32561.

tionary relationships among the species presently included inthe genus. An early rRNA-DNA hybridization study showedthat Desulfovibrio vulgaris rRNA is 97% related to Desul-fovibrio desulfuricans rRNA and only 57% related to De-sulfovibrio africanus rRNA (25). The guanine-plus-cytosinecontents for the genus range from 49 to 66 mol% (38).Cytochromes C3 from various Desulfovibrio desulfuricansstrains exhibit variability in their antigenic determinants(33); and a comparison of the six amino acid sequencesknown for this cytochrome has shown that only 20% of theresidues are conserved and that the conserved amino acidsare mainly involved in heme attachment (16). Furthermore,surface antigens of strains of the same species are oftenhighly specific (1, 34).Comparisons of 16S rRNA sequences have placed the

gram-negative sulfate-reducing eubacteria within the deltasubdivision of the purple photosynthetic bacteria along withsulfur-reducing bacteria, myxobacteria, and bdellovibrios(40). Ecological, comparative physiology, and taxonomicstudies of Desulfovibrio species would benefit from a moredetailed understanding of the phylogeny of these organisms.In this paper we describe phylogenetic relationships amongDesulfovibrio species as determined by comparisons ofnearly complete 16S rRNA and partial 23S rRNA sequences.

MATERIALS AND METHODSStrains. We used the following strains of bacteria in this

study: Desulfovibrio desulfuricans Essex 6 (= DSM 642),MB (= ATCC 27774), ATCC 7757, Norway 4 (= NCIB8310), and El Agheila Z, (= NCIB 8380); Desulfovibriovulgaris Hildenborough (= NCIB 8303); Desulfovibrio gigas

3609

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3610 DEVEREUX ET AL.

NCIB 8332; "Desulfovibrio multispirans" NCIB 12078 (cul-tures of this organism were obtained from the original strainisolated in the laboratory of J.L.); Desulfovibrio baculatus X(= DSM 1743); Desulfovibrio salexigens ATCC 14822; De-sulfovibrio baarsii 2st14 (= DSM 2075); Desulfovibriosapovorans Lindhorst lpa3 (= ATCC 33892); and De-sulfomonas pigra ATCC 29098.Growth responses. The ranges of substrates utilized by

strains of Desulfovibrio spp. were determined in a basalmedium supplemented with carbon sources (10 mM) andwith 10 mM sulfate or 10 mM nitrate as the terminal electronacceptor. The gas phase was argon except when a hydrogen(80%, vol/vol) or carbon dioxide (20%, vol/vol) atmospherewas used.The basal medium contained (per liter of distilled water)

2.0 g of NH4Cl, 1.65 g of MgCl2 6H20, 0.5 g of K2HPO4,0.2 g of CaCl2, 0.007 g of FeCl2 - 4H20, 10 ml of a tracemineral solution, and 1 ml of a 0.2% resazurin solution. Thetrace mineral solution was modified from that described byBadziong and Thauer (2) and contained (per liter) 12.8 g ofnitrilotriacetic acid, 0.21 g of FeCl2 - 4H20, 0.1 g ofMnCl2 * 4H20, 0.17 g of CoCl2 6H20, 0.10 g of ZnCl2, 0.02g of CuCl2, 0.01 g of H3BO3, 0.01 g of NaMoO4, 0.017 g ofNa2SeO3, and 0.028 g of NiCl2 .6H20. The nitrilotriaceticacid was dissolved in 200 ml of distilled water, and thesolution was adjusted to pH 6.5 with 30% KOH. Water andthe remaining salts were added to a final volume of 1 liter.Cysteine hydrochloride (0.2 g/liter) was added as the reduc-ing agent, and sodium bicarbonate was used to adjust the pHto 7.2 to 7.6.

After inoculation (1%, vol/vol), cultures were examineddaily for an increase in A6.. An increase in absorbance wasconfirmed by microscopic examination of the cultures. Cellgrowth was identified as positive if the absorbance of aculture incubated at 37°C exceeded 0.10 within 14 days.rRNA sequencing. 16S and 23S rRNA sequences were

determined by using the dideoxynucleotide chain-termi-nating procedure with reverse transcriptase and rRNA as thetemplate (14). Sequencing reactions were primed with syn-thetic oligonucleotides hybridized to conserved rRNAtracts. The sequences of the 16S rRNA primers were com-plementary to the following regions (Escherichia coli num-bering; 3' position of rRNA target; parentheses indicatemixed positions): regions 350 [CTGCTGC(GC)(CT)CCCGTAG], 536 [ACCGCGGC(GT)GCTGGC], 691 [GAT(AC)TCTACG(GA)ATTTCAC], 915 [GCCCCC(TC)CAATTCCT], 1110 [AGGGTTGCGCTCGTTG], and 1400 [ACGGGCGGTGTGT(GA)C]. Partial 23S rRNA sequences wereobtained by using a primer having the sequence CTTTCCCTCACGGTA, which was complementary to region 473 ofE. coli rRNA.

Phylogenetic analyses. The nucleotide sequences werealigned with sequences representative of eubacteria andmajor groups of sulfate-reducing bacteria. Percentages ofsimilarity and distance values were calculated for eachsequence pair (23). The positions at which any sequence hadan unknown or ambiguous nucleotide were omitted from thecalculations. Phylogenetic trees were constructed by usingthe least-squares method from distance data (9) and by usingunweighted pair group method-determined analysis(UPGMA) (22). In addition, a parsimony analysis was per-formed by using the DNAPARS program from the PHYLIPstatistical package of J. Felsenstein (University of Washing-ton).

Relationship rRNA sequence similarity to DNA relatedness.The equation used for comparing the level of similarity of the

16S rRNA sequences (S) with the DNA relatedness value orpercentage of hybridization (% DNA) was developed fromthe equations described by Kimura (13) for comparison ofamino acid substitutions in homologous proteins. Because ofthe generality of this treatment, it should apply equally wellto substitutions in homologous nucleic acids. Thus, if Kna isthe average number of base substitutions per site betweentwo nucleic acids, from the Poisson distribution the proba-bility that no substitutions occur at a site is eK_ . For closelyrelated sequences, the number of back mutations is small,and

e-Kna Pd

Kna = -2.3 log1o (1 - Pd)

kna = Kna/2Twhere Pd iS the fraction of differences in the rRNA or DNA,kna is the rate of base substitutions, and T is time. If S or tDNA is substituted for 1 - Pd and these equations arecombined, upon rearrangement the following equations areobtained:

T = 2.3 log1o (S)12k1T = 2.3 log1o (% DNA)/2k2

where k1 and k2 are the constants for the rates of substitutionfor S and % DNA, respectively. Because T is the same forcomparisons of S and % DNA for the same pair of organ-isms, these equations can be combined to obtain the follow-ing equation:

log1o (S) = (k1lk2) log1o (% DNA)To correct for back mutations in the rRNA sequences, theJukes-Cantor correction may be applied and 1 - d may besubstituted for S, where d is the evolutionary distance or-0.75 loge [1 - 1.333(1 - S)].

RESULTS

Polyphyletic origin of the species Desulfovibrio desulfuri-cans. An examination of the nutritional properties of strainsof Desulfovibrio desulfuricans revealed an unexpectedamount of heterogeneity (Table 1). The growth response ofthe type strain (strain Essex 6 [= ATCC 29577 = DSM 642])differed from the growth responses of some of the otherstrains previously assigned to the species. The type strainutilized hydrogen, formate, ethanol, and lactate as electrondonors, and it utilized sulfate and nitrate as electron accep-tors. Other strains utilized butyrate or were unable to utilizehydrogen, ethanol, or nitrate. However, it was not clearwhether the variable growth responses in the species weredue to misclassification of strains or due to a high degree ofvariability within the species for these characteristics.To resolve this question, portions of the 16S rRNA and

23S rRNA were sequenced. For the 16S rRNA, sequencingwas initially performed by using the reverse transcriptasemethod and the 1400 region primer (14). The sequences ofvarious strains were compared with the sequence of the typestrain, with the following results (number of substitutionsfound/number of sequence positions determined): strainATCC 27774, 0/181; "Desulfovibrio multispirans," 0/163;strain ATCC 7757, 5/127; strain Norway 4, 14/124; and strainEl Agheila Z, 18/151. For the 23S rRNA, sequencing wasinitiated at the 473 region primer. This primer was chosenbecause preliminary studies indicated that a high degree ofvariability was found in this region. The numbers of substi-

J. BACTERIOL.

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PHYLOGENY OF DESULFOVIBRIO SPP. 3611

TABLE 1. Growth responses of strains of Desulfovibrio spp.a

Desulfovibrio Desulfovibrio "De .br Desulfovibrio Desulfovibrio Desulfovibrio besiiMedium compositionb desulfuricans desulfuricans eultspiras desulfuricans desulfuricans baculatus X baarsi

Essex 6 ATCC 27774 multispirans El Agheila Z Norway 4 baculatus X 2st14

Lactate + S042- + + + + + +Lactate + N03- + + +Ethanol + S042- + + + +Formate + S042- + CO2 + + + + + + +Butyrate + So42- - - - (+) - - +Acetate + So42 - - - - - - (+)Acetate + S042 + H2 + + + + +Acetate + S042 + CO2

a Growth responses were scored as follows: +, increase in A6w was greater than 0.10 within 14 days; -, no growth was detected microscopically or bymeasuring absorbance within 14 days; (+), growth was detected microscopically within 14 days but the increase in absorbance was less than 0.10.

b Basal medium was supplemented with the potential electron donors or carbon sources and electron acceptors listed. The gas phase was argon unless indicatedotherwise.

tutions found compared with the type strain were as follows(number of substitutions found/number of sequence posi-tions determined): strain ATCC 27774, 5/85; "Desulfovibriomultispirans," 1/117; strain ATCC 7757, 19/116; and strainNorway 4, 31/115. These results suggested that strain ATCC27774 and "Desulfovibrio multispirans," two organismswhose nutritional properties were identical to those of thetype strain, are related at the species level. It is interestingthat the ability to use nitrate as a respiratory substrate waslimited to these strains and was not a widespread character-istic in the genus Desulfovibrio (18, 32). Moreover, some ofthe strains assigned to the species Desulfovibrio desulfuri-cans, especially strains ATCC 7757, Norway 4, and ElAgheila Z, diverged greatly in their 16S and 23S rRNAsequences.Sequence data for a larger portion of the 16S rRNA were

obtained by using a full complement of primers for some ofthe strains mentioned above, as well as other Desulfovibriospecies, and the results of these studies refined and con-firmed the conclusions described above. Evolutionary dis-tance and similarity values for pairs of 16S rRNA sequencesare shown in Tables 2 and 3, respectively; the valuesencompass about 950 positions, beginning a few nucleotidesfrom the 5' terminus and ending about 200 nucleotides fromthe 3' terminus, with a gap between nucleotide positions 890and 972 (E. coli numbering). Sequence information was notobtained for this gap for some species because of prematuretermination of the region 1110-primed reverse transcripts.From these studies we found that Desulfovibrio desulfu-

ricans Norway 4 and Desulfovibrio baculatus have virtuallyidentical 16S rRNA sequences (99.6% sequence similarity)(Table 3). This result was consistent with the identicalnutritional responses of these organisms (Table 1), the factthat they both contain desulforubidin instead of desulfoviri-din as their dissimilatory sulfite reductase (15), and thesimilarities of their cytochromes c3 (20). In addition, De-sulfovibrio desulfuricans ATCC 7757 was found to be a closerelative of Desulfovibrio vulgaris Hildenborough. No substi-tutions in 129 positions in the region 1400-primed 16S rRNAsequence were observed for these two strains. Likewise, inthe region 473-primed 23S rRNA sequence, we found only 3substitutions among 146 positions compared. Finally, strainEl Agheila Z was not closely related to the type strains ofother species of the genus.Phylogeny of the genus Desulfovibrio. Phylogenetic trees

based on the 16S rRNA sequences from the Desulfovibriospecies were constructed by using two methods, a least-squares analysis of the evolutionary distances (9) andUPGMA (22). In addition, these trees were compared with

the results of maximum parsimony and sequence signatureanalysis (40). Outgroups were chosen on the basis of previ-ously published sequence data for representatives of othergenera of gram-negative sulfate-reducing bacteria, the othermain subdivisions of the purple bacteria, and Myxococcusxanthus (7, 24, 40). Separate studies have shown that thethermophilic sulfate-reducing genus Thermodesulfobacte-rium is not related to the 8 subdivision of the purple bacteria(C. R. Woese, personal communication). Therefore, thistaxon was not included in this study. The phylogenetic treesconstructed by using the three methods were generallyconsistent and suggested that the mesophilic and gram-negative sulfate-reducing bacteria evolved from a commonorigin deep within the 8 subdivision of the purple bacteria.This result is consistent with the 8 subdivision signaturesequence present in this group (40).A comparison of the 16S rRNA catalogs suggested that the

mesophilic gram-negative sulfate-reducing bacteria are com-posed of two major branches (10). The sequencing dataconfirmed this conclusion (Fig. 1 and 2). The first branch,which includes many Desulfovibrio species, as well as De-sulfomonas pigra, is also well defined by signature analysis(Table 4). The second branch contains the remaining generaof previously characterized sulfate-reducing bacteria. Dis-tinctive features of the Desulfovibrio signature include G atposition 360, A at position 366, G at position 409, C atposition 433, G at position 689, C at position 698, G atposition 822, C at position 834, G at position 852, C atposition 1214, and A at position 1233.

Within the Desulfovibrio branch, Desulfovibrio desulfuri-cans ATCC 27774 and Desulfomonas pigra were closelyrelated (95% sequence similarity). Desulfovibrio vulgariswas also related to these two species (91% sequence simi-larity). Likewise, Desulfovibrio salexigens and Des-ulfovibrio desulfuricans El Agheila Z represented a deeplineage (c90% sequence similarity). The other species in thisbranch, Desulfovibrio baculatus plus Desulfovibrio desulfu-ricans Norway 4, Desulfovibrio gigas, and Desulfovibrioafricanus, all represented deep lineages whose branchingorder was not clear. The results of both the UPGMA and theleast-squares analysis suggested that Desulfovibrio africanusis related to Desulfovibrio salexigens. However, in neithercase was this relationship strong, and it must be consideredtentative at this time. Likewise, the results of the least-squares analysis and the UPGMA suggested that Des-ulfovibrio gigas is a deep branch of the lineage leading toDesulfovibrio desulfuricans. However, the results of theparsimony analysis suggested that this organism is related toDesulfovibrio salexigens. The results of the UPGMA of the

VOL. 172, 1990

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3612 DEVEREUX ET AL. J. BACTERIOL.

suazofawns Ntunu9aj3vqo&. o

snzluvx V o o

C)(N (N

?/00' O000

s-uv.ioAtl1nu4 C0N V

smn.ovo.olnsaa o4 o oSUVJOA.llIfWn 0ONN "t

SfdddOlfS2dq O '° XON (N 4 Nrwnxilfdoijon - 0 N 4it

wnudalvoqofjn.saa -' '

Q~~~~~~~~~~~..

o~q1(lofjnsvaa 666666!CZ

O2 SUIfUAdVS 'no (NO- 'r N̂

CZ ouoqn4o7nsa 66 66 66 6. . .

L1 oyq!AofjnSsaa . . . .

.4 zsue)osg ot oN _ N o00 ( Ng s-uvqymlnvap g -4 a-, 00 C( (oN (N (N (NoLuq;Aof,nsaa 66666666

0

'0 s S d?JlJvlsSap ,tt ,- o-(N (N (N (N (N (Nouqitoflnsaa 666666666

8) > sndvq _ ^ 0000ttsnvgN0 0 ()oA0

0. oul *izss W__ n_ n_ _ _ _ q en \CN

oL/q?iofjnsaa 66666666666r44 4 q

o SflUVdLfJV Cf) CD OC O-O' (NCOOO

00 NrN(Nr-rq n al4 -"r)ON-4It enOooCD

_ _ON ON C( N WICfooy/q?Aofunsaa 6 6 6 6 6o o

0

osvuoofunsaa 66666666-6ogooooo _ - ) riuVLL DV 4C r-d 000,T'Ct00~~n 1,(0r-

t'LLLZJJXLVW QC\-4 nN" 0In00N0000r4r)( tO-. C)£.Uvqyqilffsap 0 0 -4r- r-r- 4-- --4 ---4r- r- i" > (

OUq!Ao,fnsaa °666666 6666o CD° °

0

0N 0'

N U)~~~~~~~~~~~0

(N .

Q .0t i Q g

" o2L~i -., 00.::7S

- - - - -geeegat2- - - - - - - - j;, .iq Z:SZ ,; :t : : : U;S°a LX- co X) X) X0 X- X0 X) Li X.- XnC..

aRaRaRaRaRaR%)4 aW)4)j ;

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

VOL. 172, 1990 PHYLOGENY OF DESULFOVIBRIO SPP. 3613

m Z0 0a

CD. 1-

N P

,0

O O0 ~0O -

-- ON --000Q

p . p

00 00 0000 \0 £0

4- CD .4 WC 0 k) ) 0%

00000000

(-.o. . w

00 0 00 0 00<00

0000 0000 00

' a -4 -4 -4 00

o ~A W P.o rK-- -

--CO 0C (-0000 0000000

0%--A--~ ~- -tQ w0 .t -

-4ON O-(4 -A 0o (A

ONO0 .P. .P .- . .w

00 00 00 00 00 00 00 00Q 0 0 C-4 00a) -4 -4

-4 O 0(-A W(4. -- .

COCDC OC) C) C) C> C) C

000 Q 00 00 00 00 00 00-4 (- O 0000 ) 0% 0%k (

C)~~ 0 CD C) Q C)C ) O C> C) C

0000000 00 0000\00 00 00-4- 00 (-.w -4 O- ~

-. .~.~ (.4 (.

000 0000 0000 0000 0000 00000

00 .. .-I .

00 .- .~ . .4

0000000000000oo o o o o00000 000 0000000 ~ 00k)- (-40£ 000 -

o- (-.o (- - \0 -4 -4 (-4. -4 (-

00000000000000o o o o o osH>~HobFooo04oP00~~00~ 00e0 \0 0 ~0 0

0%)o0%04oOk) 00%(-k)

000000000000000< op< o00-4 -4 -4 -4 -4 -4- -4 -4 -4- -4--

.- .) . .)k) ..-k .~ .4 .4 .~ .% .I

Oa'

Desulfomonaspigra

Desulfovibriovulgaris

Desulfovibriogigas

Desulfovibrioafricanus

DesulfovibriodesulfuricansNorway 4

Desulfovibriobaculatus

DesulfovibriodesulfuricansEl Agheila Z

Desulfovibriosalexigens

Desulfovibriosapovorans

Desulfovibriobaarsii

Desulfobacteriumautotrophicum

Desulfococcusmultivorans

E. coli

M. xanthus

Argobacteriumtumefaciens

Pseudomonastestosteroni

9T10CDCDD

co0

C.

CD

CD

0

3

cn

10

0'CD

D.0

0

CDo0CD

cn

CD

CD

0.CD

r-CD

0

(A

H

mIJ4

I-.4zCD.0

0

CD0

0

C-CD

U,

0

0

u,

CD

0._

CDCD0

CD

I.

CD0T

a'

0

CD

a

(A0

r+

m0

CD

CDaQ

cn

:t

D-o.0

C-

co

0CDCD

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3614 DEVEREUX ET AL. J. BACTERIOL.

A.yxococcus xanthusAgrobacteriur tumfac/ens

Desuhfb'OME deSifr/cns Norway 4Desuh4obm n bacudatus

Desu/foWbrlo gas

Desuwfov o KAris

Desuffowbro dedfunas 27T4Desufononas pgr

-Defsufoo* slexionsDesuhfoptbrio desulfulkans El Aghelila Z

Desuffoibrio afrkanus

Desuh'ococcus mnxtlvrnsDesuffowibrlo sapovorans

Desu/fobacterbm autotropcum0.02 Knuc

Desuh'omWo bearsli

Esche,chla co/I

Pseudom.ons testostwroniFIG. 1. 16S rRNA phylogenetic tree of Desulfovibrio species, constructed by using the least-squares method. Distance measurements

were taken from Table 2. The scale bar is in units of fixed nucleotide substitutions per sequence position. Bacillus subtilis and Halococcusmorrhuae were used as outgroups.

sequences obtained from region 473 of the 23S rRNA wereconsistent with this latter interpretation (data not shown).Therefore, while Desulfovibrio gigas represented a deeplineage of this branch, its phylogenetic position was uncer-tain. Similarly, Desulfovibrio baculatus plus Desulfovibriodesulfuricans Norway 4 also represented a deep lineage ofthis branch whose phylogenetic position was not consis-tently determined after least-squares analysis and UPGMA.Thus, this branch of the sulfate-reducing bacteria containsfive lineages with an average level of similarity of 88%.These lineages are represented by the following organisms:(i) Desulfovibrio salexigens and Desulfovibrio desulfuricansEl Agheila Z; (ii) Desulfovibrio africanus; (iii) Desulfovibriodesulfuricans, Desulfomonas pigra, and Desulfovibrio vul-garis; (iv) Desulfovibrio gigas; and (v) Desulfovibrio bacu-latus and Desulfovibrio desulfuricans Norway 4. However,the branching order within this group is uncertain.The second major branch of the gram-negative sulfate-

reducing bscteria contained two Desulfovibrio species inaddition to a number of other genera (7). The results of boththe UPGMA and the least-squares analysis supported theinterpretation that Desulfovibrio sapovorans and Desulfo-coccus multivorans are related but not closely. In addition,Desulfovibrio baarsii also represented a deep lineage of thisbranch. This branch also contained species belonging to thegenera Desulfobacterium, Desulfobulbus, Desulfococcus,Desulfobacter, and Desulfosarcina. The signature of theseorganisms was distinctive and included A at positiop 360, Uat position 409, G at position 433, U of position 554, C atposition 1201, U at position 1214, and G at position 1233(Table 4).

Relationship of rRNA sequence similarity and DNA relat-edness data. Bacterial species are currently defined in part onthe basis of levels of DNA relatedness (37). Likewise,Johnson has proposed that the species in a genus shouldexhibit more than 20% DNA relatedness (12). The relation-ship between the level of similarity of 16S rRNA sequencesand the level of DNA relatedness has not been describedpreviously, but a description of this relationship would beuseful for interpreting the rRNA sequence data of thesulfate-reducing bacteria. In search of the literature, wefound both rRNA sequence and DNA relatedness data forstrains belonging to six eubacterial genera, including Acti-nobacillus (5, 8), Arthrobacter (35), Bacteroides (11, 26),Campylobacter (3, 28), Micrococcus (4), and Staphylococ-cus (17). Except for the genera Actinobacillus and Campylo-bacter, the rRNA data were determined by using oligonu-cleotide cataloging methods and were expressed as thebinary coefficient SAB. When the SAB was less than 0.30, thedata were not included because the DNA relatedness datawere not considered reliable for very distantly relatedstrains. For DNA relatedness values ranging from 1 to 40%,

log10 SAB = 0.209 (log1o % DNA) - 0.409The correlation coefficient was 0.815, which was significantat P = 0.01 for n = 29. The rRNA similarity values werecalculated fron the SAB values by using the followingformula: S = S/A (40). The data for the genera Actino-bacillus and Campylobacter were then combined with thelarger set of data from the SAB values (Fig. 3). The relation-ship between rRNA similarity values and levels of DNArelatedness was expressed as follows:

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PHYLOGENY OF DESULFOVIBRIO SPP. 3615

Myxococcus xanthus

Desulfovibrio desulfiricans Norway 4

Desulfovibrio baculatus

Desulfovibrio gigasDesulfovibrio vulgaris

Desulfovibrio desulfuricans 27774

Desulfomonaspigran_L.-.:- 1^"

Desulfovibrio desulfuricans El Agheila Z

Desulfovibrio africanusDesulfococcus multivoransDesulfovibrio sapovorans

Desulfobacterium autotrophicumDesulfovibrio baarsii

Agrobacterium tumefaciens

FIG. 2. Phylogenetic tree of Desulfovibrio species, constructed by using UPGMA. The dotted lines indicate the standard error for eachof the branch points. The scale (d) indicates evolutionary distance, which was calculated from the data in Table 2 after correction for deletionsand insertions.

loglo S = 0.0350 (log1o % DNA) - 0.0698

The correlation coefficient was 0.728, which was significantat P = 0.0001 for n = 34. From this formula, rRNA similarityvalues at levels of DNA relatedness of 70 and 20% werecalculated to be 0.988 (standard error, 0.990 to 0.986) and0.946 (standard error, 0.955 to 0.937), respectively. Therelationship between the level of DNA relatedness and theevolutionary distance (d), which corrected for back muta-tions, was also determined. The results were very similar(Fig. 3):

log1o (1 - d) = 0.0396 (log1o % DNA) - 0.0770

The correlation coefficient was 0.722, which was significantat P = 0.0001 for n = 34. At DNA relatedness levels of 70and 20%, the evolutionary distances were calculated to be0.0191 (S = 0.981) and 0.0666 (S = 0.936), respectively.On the basis of the relationship between rRNA similarity

values and levels of DNA relatedness, most Desulfovibriospecies are related more distantly than the genus level in thesense of Johnson (12). For instance, the evolutionary dis-tance between most of pairs of species was greater than 0.1,which corresponds to a level of DNA relatedness of about7%. Likewise, the measured value for the level of DNArelatedness between Desulfovibrio baculatus and Desul-fovibrio africanus was 0 to 2% (21). The calculated valuefrom the rRNA similarity value of 0.861 was 1.4% (range, 1to 2%). In contrast, the rRNA similarity value for Des-ulfovibrio desulfuricans and Desulfomonas pigra was 0.951(Table 3). This value corresponded to a level of DNArelatedness of 23%. Therefore, the rRNA sequence evidencesupported the assignment of these species to a single genus.The rRNA similarity values for Desulfovibrio vulgaris and

Desulfovibrio desulfuricans and for Desulfovibrio vulgarisand Desulfomonas pigra were 0.913 and 0.910, respectively.These values corresponded to a level of DNA relatedness of7% (range, 5 to 10%), which is close to the minimum genuslevel suggested by Johnson. Therefore, the rRNA sequenceevidence would support the assignment of Desulfovibriovulgaris to a new genus if additional phenotypic or geneticevidence suggested that it was desirable. Moreover, reas-signment of Desulfovibrio vulgaris to a new genus would beobligatory if Desulfomonas pigra is not reclassified as a

Desulfovibrio species.Phylogeny of the genus Myxococcus. On the basis of the

sequence of its 16S rRNA, M. xanthus has been placed in the8 subdivision of the purple bacteria along with the sulfate-reducing bacteria and the genus Bdellovibrio (40). With theavailability of many more 16S rRNA sequences from thesulfate-reducing bacteria, we performed analyses to deter-mine whether the myxococcal line of descent preceded thesulfate-reducing bacteria. The results of both UPGMA andleast-squares analysis supported the assignment of this or-

ganism to a very deep lineage of the 8 subdivision of thepurple bacteria. The results of UPGMA and parsimonyanalysis consistently grouped the genus Myxococcus withother members of the purple bacteria outside the sulfate-reducing bacteria (Fig. 2) (data not shown). In the least-squares analysis, the genus Myxococcus appeared to ariseclose to the root of the two major branches of gram-negativesulfate-reducing bacteria (Fig. 2). However, its exact root,whether inside or outside the sulfate-reducing bacteria,depended greatly on the choice of the outgroup sequencesused to root the tree. Therefore, while the results of ananalysis of the rRNA sequences supported the hypothesis

d 0.10.2

m tEscherichia coliPseudomonas testosteroni

VOL. 172, 1990

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

J. BACTERIOL.3616 DEVEREUX ET AL.

0uu~~~~u~~~~~~~

o.<Q<Q<QQ00 0

00~~~~~~~~~~~~~

gN<0U0000000Q 0

00~~~~~~~~~~~

UDDDDbUU

O <QrX.)QQ<Q<QQ r.

0000

00~~~~~~~~~~~~~

00~~~~~~~~~~~~

00 U U U .-

z~~~~~~~~~~~~~0~~~~~~~~~~~

0 C.)

Noox6 < <2 U< UU c_UE

0~~~~~~~~~~~~~~

oV«<¢<O;<O s0

0D O~ Q D D ;00 Q o

0 <0 t< D U < D < n *

0 00o

_O O 0

_n 1- D = Q Cn =

U~~~~~~~~~~~~~~U U~~~~~~.

0 0

zZ.0~~~~~~~~~~~~~~

i 0<fi -c E-a

0)Q80E. E E ' E o = , 5 <

fi ,D.eC) -O

-Uc0

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PHYLOGENY OF DESULFOVIBRIO SPP. 3617

cn

0.9

0.8

-0.02

U)0

"'- -0.04

0

-0.06

-0.02

-0.040

0

v-

0)

0 -0.06

-0.08

DNA Relatedness (%)

1 2

log10 (% DNA)

FIG. 3. Relationship of 16S rRNA similarity values and evolu-tionary distances to levels of DNA relatedness. Symbols: A, Acti-nobacillus; A, Arthrobacter;*, Bacteroides; Campylobacter;*,Micrococcus; 0, Staphylococcus. (A and B) The solid line repre-

sents the least-squares fit to the data: log1o S = 0.035 (log1o % DNA)0.0698. The dashed lines indicate the range of the standard error

of the slope. The F value for the analysis of variance was 36, whichwas significant at P = 0.0001. The standard error of the slope was

0.00582. The standard error of the y-axis intercept was 0.00687. (C)The solid line represents the least-squares fit to the data: log1o (1

d) = 0.0396 (log1o % DNA) 0.0770. The dashed lines indicate therange of the standard error of the slope. The F value for the analysis

of variance was 35, which was significant at P = 0.0001. Thestandard error of the slope was 0.00671. The standard error of they-axis intercept was 0.00792.

that M. xanthus represents a line of descent which precededthe radiation of the gram-negative sulfate-reducing bacteria,a comparison of more myxococcal sequences will be neces-sary to remove any ambiguities concerning the line ofdescent of this organism. Certainly, the great physiologicaldifferences between the myxococci and the sulfate-reducingbacteria are consistent with the hypothesis that the myxo-cocci preceded the radiation of sulfate-reducing bacteria.

DISCUSSION

The assignment of a number of unrelated strains to thespecies Desulfovibrio desulfuricans, as well as the assign-ment of a number of distantly related species to the genusDesulfovibrio, is strongly reminescent of the early classifi-cation of the methane-producing bacteria. In both cases,laborious anaerobic methodology hindered detailed pheno-typic characterization, and a great deal of diversity was notrecognized. Thus, the utility of molecular techniques isreadily apparent because a large amount of discriminatorydata can be collected very rapidly. Likewise, in the absenceof other information, the taxonomic significance of morphol-ogy tended to be overemphasized. Thus, the close phyloge-netic relationship between Desulfovibrio desulfuricans andDesulfomonas pigra was obscured by differences in mor-phology and motility (19). Similarly, classification accordingto morphology tended to group very dissimilar species (e.g.,Desulfovibrio desulfuricans and Desulfovibrio baarsii).

With the exception of Desulfovibrio baarsii and Desul-fovibrio sapovorons, the desulfovibrios represent a phyloge-netically coherent or monophylogenetic group. This conclu-sion is consistent with the distinctive rRNA signature andthe nutritional similarities of the group (27, 38). All of thesespecies incompletely oxidize lactate to acetate. All utilizeH2, formate, and ethanol. None uses fatty acids as growthsubstrates. However, the diversity within the group isgreater than the diversity normally encountered within eu-bacterial genera. Therefore, the results of the phylogeneticanalysis support the reassignment of Desulfovibrio bacula-tus to the new genus Desulfomicrobium (21, 29, 30). Thisconclusion is consistent with the morphology of this organ-ism (which is rod shaped), the absence of desulfoviridin, andthe lack of significant DNA hybridization with Desulfovibrioafricanus. Presumably, reclassification of the other fourdeep lineages of the desulfovibrios into separate generawould also be justified. However, it is not appropriate toreclassify these species solely on the basis of rRNA se-quence data and in the absence of distinguishing phenotypicfeatures.

If many of the desulfovibrios are reclassified into separategenera, it is important that the monophylogenetic origin ofthe group continues to be recognized at a higher taxonomiclevel. Thus, grouping of these organisms into a single family,the Desulfovibrionaceae, appears to be appropriate. On thebasis of rRNA similarities, such a family would be approx-imately equal in depth to other families currently recognizedin the purple bacteria. For instance, the families Rhizobi-aceae, Enterobacteriaceae, and Vibrionaceae form DNA-rRNA hybrids with thermal denaturation values of at least68°C (6). This value corresponds to an rRNA similarity valueof about 0.85 (31). The rRNA similarity values among thedesulfovibrios are greater than 0.87. Likewise, the remaininggram-negative sulfate-reducing bacteria, such as the generaDesulfobacterium, Desulfobulbus, Desulfococcus, Desulfo-bacter, and Desulfosarcina, may represent an additionalfamily (7).

C~~~~~~~~~~~~~~~~~~~

A *7 A7 /

/ / A

VOL. 172, 1990

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3618 DEVEREUX ET AL.

ACKNOWLEDGMENTSThis study was supported by grants PCM-8351355 and DMB-

8602789 from the National Science Foundation and by a grant fromthe Georgia Power Company (to W.B.W.) and by grant EPA-CR-812496 from the United States Environmental ProtectionAgency (to D.A.S.).We thank C. R. Woese for initially providing some of the primers

used and A. Popadic for performing the parsimony analysis. We alsothank J. Arnold for helpful discussions.

LITERATURE CITED1. Abdollahi, H., and D. B. Nedweli. 1980. Serological character-

istics within the genus Desulfovibrio. Antonie van Leeuwen-hoek J. Microbiol. Serol. 46:73-83.

2. Badziong, W., and R. K. Thauer. 1978. Growth yields andgrowth rates of Desulfovibrio vulgaris (Marburg) growing onhydrogen plus sulfate and hydrogen plus thiosulfate as the soleenergy sources. Arch. Microbiol. 117:209-214.

3. Belland, R. J., and T. J. Trust. 1982. Deoxyribonucleic acidsequence relatedness between thermophilic members of thegenus Campylobacter. J. Gen. Microbiol. 128:2515-2522.

4. Brooks, B. W., R. G. E. Murray, J. L. Johnson, E. Stacke-brandt, C. R. Woese, and G. E. Fox. 1980. Red-pigmentedmicrococci: a basis for taxonomy. Int. J. Syst. Bacteriol.30:627-646.

5. Chuba, P. J., R. Bock, G. Graf, T. Adam, and U. Goibel. 1988.Comparison of 16S rRNA sequences from the family Pasteurel-laceae: phylogenetic relatedness by cluster analysis. J. Gen.Microbiol. 134:1923-1930.

6. De Smedt, J., M. Bauwens, R. Tytgat, and J. De Ley. 1980. Intra-and intergenic similarities of ribosomal ribonucleic acid cistronsof free-living, nitrogen-fixing bacteria. Int. J. Syst. Bacteriol.30:106-122.

7. Devereux, R., M. Delaney, F. Widdel, and D. A. Stahl. 1989.Natural relationships among sulfate-reducing eubacteria. J.Bacteriol. 171:6689-6695.

8. Escande, F., F. Grimont, P. A. D. Grimont, and H. Bercovier.1984. Deoxyribonucleic acid relatedness among strains of Acti-nobacillus spp. and Pasteurella ureae. Int. J. Syst. Bacteriol.34:309-315.

9. Fitch, W. M., and E. Margoliash. 1967. Construction of phylo-genetic trees: a method based on mutational distances asestimated from cytochrome c sequences is of general applica-bility. Science 155:279-284.

10. Fowler, V. J., F. Widdel, N. Pfenmig, C. R. Woese, and E.Stackebrandt. 1986. Phylogenetic relationships of sulfate- andsulfur-reducing eubacteria. Syst. Appl. Microbiol. 8:32-41.

11. Johnson, J. L. 1973. Use of nucleic-acid homologies in thetaxonomy of anaerobic bacteria. Int. J. Syst. Bacteriol. 23:308-315.

12. Johnson, J. L. 1984. Nucleic acids in bacterial classification, p.8-11. In N. R. Krieg and J. G. Holt (ed.), Bergey's manual ofsystematic bacteriology, vol. 1. The Williams & Wilkins Co.,Baltimore.

13. Kimura, M. 1983. The neutral theory of molecular evolution, p.69-70. Cambridge University Press, New York.

14. Lane, D. J., B. Pace, G. J. Olsen, D. A. Stahl, M. L. Sogin, andN. R. Pace. 1985. Rapid determination of 16S ribosomal RNAsequences for phylogenetic analyses. Proc. Natl. Acad. Sci.USA 82:6955-6959.

15. Lee, J.-P., C.-S. Yi, J. LeGall, and H. D. Peck, Jr. 1973.Isolation of a new pigment, desulforubidin, from Desulfovibriodesulfuricans (Norway strain) and its role in sulfite reduction. J.Bacteriol. 115:453-455.

16. LeGall, J., and G. Fauque. 1988. Dissimilatory reduction ofsulfur compounds, p. 587-639. In A. J. B. Zehnder (ed.),Biology of anaerobic microorganisms. John Wiley & Sons, Inc.,New York.

17. Ludwig, W., K.-H. Schleifer, G. E. Fox, E. Seewaldt, and E.Stackebrandt. 1981. A phylogenetic analysis of staphylococci,Peptococcus saccharolyticus and Micrococcus mucilaginosus.J. Gen. Microbiol. 125:357-366.

18. Mitchell, G. J., J. G. Jones, and J. A. Cole. 1986. Distribu-tion and regulation of nitrate and nitrite reduction by Desulfo-vibrio and Desulfotomaculum species. Arch. Microbiol. 144:35-40.

19. Moore, W. E. C., J. L. Johnson, and L. V. Holdeman. 1976.Emendation of Bacteroidaceae and Butyrivibrio and descrip-tions of Desulfomonas gen. nov. and ten new species in thegenera Desulfomonas, Butryrivibrio, Eubacterium, Clostridium,and Ruminococcus. Int. J. Syst. Bacteriol. 26:238-252.

20. Moura, I., M. Teixeira, B. H. Huynh, J. LeGall, and J. J. G.Moura. 1988. Assignment of individual heme EPR signals ofDesulfovibrio baculatus (strain 9974) tetraheme cytochrome C3.Eur. J. Biochem. 176:365-369.

21. Nazina, T. N., A. B. Poltaraus, and E. P. Rozanova. 1988.Estimation of genetic relationship of rod-shaped asporogenicsulfate-reducing bacteria. Microbiology (Engl. Transl. Mikrobi-ologiya) 56:669-672.

22. Nei, M., J. C. Stephens, and N. Saitou. 1985. Methods forcomputing the standard errors of branching points in an evolu-tionary tree and their application to molecular data from humansand apes. Mol. Biol. Evol. 2:66-85.

23. Olsen, G. J., D. J. Lane, S. J. Giovannoni, N. R. Pace, and D. A.Stahl. 1986. Microbial ecology and evolution: a ribosomal RNAapproach. Annu. Rev. Microbiol. 40:337-365.

24. Oyaizu, H., and C. R. Woese. 1985. Phylogenetic relationshipsamong the sulfate-respiring bacteria, myxobacteria and purplebacteria. Syst. Appl. Microbiol. 6:257-263.

25. Pace, B., and L. L. Campbell. 1971. Homology of ribosomalribonucleic acid of Desulfovibrio species with Desulfovibriovulgaris. J. Bacteriol. 106:717-719.

26. Paster, B. J., W. Ludwig, W. G. Weisburg, E. Stackebrandt,R. B. Hespell, C. M. Hahn, H. Reichenbach, K. 0. Stetter, andC. R. Woese. 1985. A phylogenetic grouping of the bacteroides,cytophagas, and certain flavobacteria. Syst. AppI. Microbiol.6:34-42.

27. Postgate, J. R. 1984. The sulfate-reducing bacteria, 2nd ed.Cambridge University Press, Cambridge.

28. Romaniuk, P. J., B. Zoltowska, T. J. Trust, D. J. Lane, G. J.Olsen, N. R. Pace, and D. A. Stahl. 1987. Campylobacterpylori,the spiral bacterium associated with human gastritis, is not atrue Campylobacter sp. J. Bacteriol. 169:2137-2141.

29. Rozanova, E. P., and T. N. Nazina. 1976. A mesophilic, sulfate-reducing, rod-shaped, nonsporeforming bacterium. Mikrobi-ologiya 45:825-830.

30. Rozanova, E. P., T. N. Nazina, and A. S. Galushko. 1988. A newgenus of sulfate-reducing bacteria and the description of its newspecies, Desulfomicrobium apsheronum gen. nov., sp. nov.Mikrobiologiya 57:634 641.

31. Schleifer, K. H., and E. Stackebrandt. 1983. Molecular system-atics of prokaryotes. Annu. Rev. Microbiol. 37:143-187.

32. Seitz, H.-J., and H. Cypionka. 1986. Chemolithotrophic growthof Desulfovibrio desulfuricans with hydrogen coupled to am-monification of nitrate or nitrite. Arch. Microbiol. 146:63-67.

33. Singleton, R., Jr., J. Denis, and L. L. Campbell. 1984. Antigenicdiversity of cytochromes C3 from the anaerobic, sulfate-reducingbacteria, Desulfovibrio. Arch. Microbiol. 139:91-95.

34. Singleton, R., Jr., J. Denis, and L. L. Campbell. 1985. Whole-cell antigens of members of the sulfate-reducing genus De-sulfovibrio. Arch. Microbiol. 141:195-197.

35. Stackebrandt, E., V. J. Fowler, F. Fiedler, and H. Seiler. 1983.Taxonomic studies on Arthrobacter nicotianae and related taxa:description of Arthrobacter uratoxydans sp. nov. and Arthro-bacter sulfureus sp. nov. and reclassification of Brevibacteriumprotophormiae as Arthrobacter protophormiae comb. nov.Syst. Appl. Microbiol. 4:470-486.

36. Stetter, K. O., G. Lauerer, M. Thomm, and A. Neuner. 1987.Isolation of extremely thermophilic sulfate-reducers: evidencefor a novel branch of archaebacteria. Science 236:822-824.

37. Wayne, L. G., D. J. Brenner, R. R. Colweli, P. A. D. Grimont,O. Kandler, M. I. Krichevsky, L. H. Moore, W. E. C. Moore,R. G. E. Murray, E. Stackebrandt, M. P. Starr, and H. G.Truper. 1987. Report of the Ad Hoc Committee on Reconcilia-

J. BACTERIOL.

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

VOL. 172, 1990 PHYLOGENY OF DESULFOVIBRIO SPP. 3619

tion of Approaches to Bacterial Systematics. Int. J. Syst. 39. Widdel, F., and N. Pfennig. 1984. Dissimilatory sulfate- orBacteriol. 37:463-464. sulfur-reducing bacteria, p. 663-679. In N. R. Krieg and J. G.

38. Widdel, F. 1988. Microbiology and ecology of sulfate- and Holt (ed.), Bergey's manual of systematic bacteriology, vol. 1.sulfur-reducing bacteria, p. 469-585. In A. J. B. Zehnder (ed.), The Williams & Wilkins Co., Baltimore.Biology of anaerobic microorganisms. John Wiley & Sons, Inc., 40. Woese, C. R. 1987. Bacterial evolution. Microbiol. Rev. 51:New York. 221-271.

on July 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from