- Email: [email protected]
The Importance of Using Outgroup Reference Organisms in Phylogenetic Studies: the Atopobium Case
System. App!. Microbiol. 17,39-43 (1994) © Gustav Fischer Verlag, Stuttgart· Jena . New York
The Importance of Using Outgroup Reference Organisms in Phylogenetic Studies: the Atopobium Case E. STACKEBRANDr'; and W. LUDWIG b aDSM-Deutsche Sammlung von Mikroorganismen, Mascheroder Weg 1B, 38124 Braunschweig, Germany, and fiir Mikrobiologie, Technische Universitat Miinchen, 80290 Miinchen, Germany
b Lehrstuhl
Received November 3, 1993
Summary Comparative 16S rDNA studies of lactic acid bacteria and some of their phylogenetic relatives of the Clostridium/Bacillus subphylum of Gram-positive bacteria indicated that three misclassified Lactobacillus and Streptococcus species, described as members of the novel genus Atopobium, formed separate sub line of descent within this subphylum. Inclusion of outside reference organisms such as members of the order Actinomycetales and a cyanobacterial species, serving as the root of Gram-positive bacteria, clearly showed Atopobium species to branch off within the actinomycetes line of descent. The importance of the use of an appropriate selection of reference organisms in phylogenetic studies is discussed.
Key words: Atopobium, Clostridium/Bacillus subphylum - Actinomycetales - 16S rRNA - phylogenetic analysis
Taxonomists that are not familiar with the interpretation of sequence data and the reconstruction of phylogenetic trees, assume not only that a phylogenetic tree of the 16S rRNA is based on optimal sequence alignment and a widely accepted treeing algorithm but also that the topology of a tree reflects by and large the correct position of the organisms analyzed. Although 16S rRNA sequences are known for only about 25% of described prokaryotic species, the approximate phylogenetic position of a taxon relative to its neighbor is believed to be rather stable and not to change when more species are included. This has been shown to be reasonable accurate when the topologies of dendrograms derived from 16S rRNA cataloging (Fox et al., 1980) were compared to the topologies of phylogenetic trees derived from complete 16S rRNA sequences (Woese, 1987). Certain changes were observed in the branching patterns especially in the lengths and the order of deep rooting lines of descent and certain corrections were made for rapidly evolving organisms. Differences in tree topologies have been shown to be caused not only by differences in the tempo of evolution (Weisburg et al., 1989; Woese et al., 1980), differences in the * Corresponding author
base composltlon of rRNA genes and unequal rates of evolution within different regions of the rRNA genes (Woese et al., 1991), but also by the selection of sequences stretches analyzed (Stackebrandt et al., 1991). Another important factor, well known to those directly involved in the generation of trees, is the influence of the number of organisms and the selection of reference organisms included in the analysis. The latter case has recently been demonstrated to result in misleading phylogenetic conclusions (Collins and Wallbanks, 1992; Lawson et al., 1993). In the .course of a broad survey on lactic acid bacteria (a term that reflects the production of lactic acid by Gram-positive bacteria and is by no means phylogenetically sound), the 16S rRNAs of Lactobacillus minutus, Lactobacillus rimae and Streptococcus parvulus were compared to those of members of various groups defined by lactic acid bacteria, such as members of Leuconostoc, Lactobacillus, Pediococcus, Streptococcus, Enterococcus and a few relatives. These three lactobacilli grouped neither with the core of Lactobacillus nor with any other lactic acid producing bacteria but formed "a hitherto unknown line of descent within the lactic acid group of bacteria" for which a new genus, Atopobium, was proposed (Collins and Wallbanks,
40
E. Stackebrandt and W. Ludwig
1992). More recently (Lawson et aI., 1993) these organisms were included as " ... representative low G+C Grampositive taxa known or considered to be related to clostridia ... " in a phylogenetic survey of members of Clostridium. In order to reevaluate the phylogenetic position of Atopobium the complete database of 16S rRNAs and 16S rDNAs (De Rijk et. aI., 1992; Olsen et aI., 192) of about 1500 sequences was analysed by a "rated parsimony" method (Ludwig, unpublished) (data not shown). A smaller selection of 13 species, seven from the Clostridium/ Bacillus subphylum and six from the actinomycetes subphylum, was then selected to cover a broad range of genetic diversity within the two subphyla. The following regions of alignment uncertainty and missing sequence information were omitted prior to the analysis: position 1 to 102, 179 to 224, 458 to 487, 930 to 968, 1028 to 1052, 1136 to 1154 and 1387 to 3' end (positions refer to the Escherichia coli numbering system [Brosius et aI., 1978]). The phylogenetic analysis was performed as described (Rainey and Stackebrandt, 1993 ). The phylogenetic tree is depicted in Fig. 1. The similarity values calculated between the Atopobium species and the reference organisms, that have been included in the previous study (Collins and Wallbanks, 1992) and this study, were very similar and differences can be explained by different stretches compared. However, with respect to the phylogenetic position of Atopobium species the comparison of the tree in Fig. 1 with that shown by Collins and Wallbanks (1992) and Lawson et al. (1993) demonstrated significant differences that are worth discussing:
,-----
Firstly, a proper selection of reference organisms would have avoided the misleading phylogenetic conclusion that Atopobium is either a relative of Erysipelothrix rhusiopathiae (Collins and Wallbanks, 1992) or of Clostridium leptum and C. sporosphaeroides (Lawson et aI., 1993). While E. rhusiopathiae is known to be a relative of the mycoplasmas and certain cell wall containing Gram-positive bacteria (including the misclassified Lactobacillus vitulinus and L. catenaforme) that were previously shown to be derived from Bacillus and relatives (Weisburg et aI., 1989; Woese et aI., 1980), the two clostridial species constitute an individual branch within the radiation of low G+C Gram-positives (Lawson et aI., 1993; Olsen et aI., 1992). Since the treeing algorithm does not determine whether or not the phylogenetic position of a deep branching organisms makes sense from a biological point of view, it is left to the microbiologist to do so. Bootstrap values, which were not included, would have been so low anyway that they would not have given a hint to whether the relationship between Erysiphelotrix or species of Clostridium on the one hand and Atopobium species on the other hand had any statistical significance. , The inclusion of a single sequence of the actinomycetes "proper" and a sequence of an outside reference organisms to define the root of the Gram-positive line of descent would have helped to solve the problem immediately. The root, here defined by a cyanobacterium, clearly separates the two main clusters of Gram-positive bacteria and the separate status is supported by high bootstrap values of 98%. The use of members from other bacterial phyla as outgroups do not change the topology of the bifurcation
- Arthrobacter g/obiformis
'---------
-
Actinomyces pyogenes
' - - - - - - Nocardioides simp/ex 51
'---------- Streptomyces coe/ic%r ' - - - - - - --
98
- Bifidobacterium bifidum
'-------------slrain TH3
Atopobium parvulum Atopobium rimae
100
' - - - - Atopobium minutum , -- - - - - -- Heliobacterium chlorum ' - - -- -, - - - --
- - - Acidaminococcus jermenlans -
-
-
Clostridium butyricum
, - - - - - - Erysipelothrix rhusiopathiae '------- --
-
-
Lactobacillus vilulmus
93 , - - - - - - Lactobacillus acidophilus ' - - - - - Bacillus subtilis 10%
Fig. 1. Dendrogram of 16S rONA relationships for the three Atopobium species and other members of the phylum of Gram-positive bacteria, generated by using the algorithm of DeSoete (1989) on dissimilarity values as calculated by Jukes and Cantor (1969). The sequence of the cyanobacterium Anacystis nidulans was used as the root for the two subphyla. Organisms in bold were shown as phylogenetic neighbors in reference (Collins and Wallbanks, 1992). Bootstrap values, based on the analysis of 1000 trees from 809 polymorphic sites were obtained using the programs NJFIND and NJBOOT (kindly provided by T.S. Whittam, Institut of Molecular Evolution and Genetics, Pennsylvania State University). Only values above 50% significance are indicated. Bar = 10% sequence divergence.
6.1 8.0 17.3 20.1 18.6 17.9 16.0 16.4 24.7 22.0 24.9 26.0 22.7 23.0 24.4 24.3
2.9 17.0 20.3 17.6 16.2 15.0 16.0 24.4 21.6 24.3 26.4 21.1 22.7 22.4 23.8
94.1 17.5 21.4 18.8 17.5 15.6 17.3 25.6 22.9 25.6 27.4 22.2 23.0 24.4 23.8
92.4 97.1
A.mi- A.par- A. nutum vulum rimae
18.8 16.4 16.4 14.8 14.7 22.5 20.2 24.2 25.3 23.0 22.0 19.4 25 .1
84.2 84.5 84.2
TH3
15.6 16.5 15.4 13.7 24.9 24.7 26.1 28.1 23.7 24.8 22.4 27.9
82.2 81.0 81.2 83.1
Bf. bi(idum
12.5 11.3 8.2 22.9 23.4 24.4 27.1 23.5 23.4 21.4 25.1
83.3 84.1 81.1 85.2 85.6
Ac. pyognes
10.1 9.0 21.8 23.3 22.9 27.7 23.0 23.9 21.1 26.6
83.9 85.2 84.2 85.2 84.9 88.4
N. simplex
8.0 19.7 20.2 21.3 24.3 21.3 22.5 19.5 23 .9
85.3 86.2 85.7 86.3 85.8 89.4 90.4
coelicolor
S.
20.1 21.1 21.3 25.2 21.8 23.0 18.8 23.5
85.1 85.4 84.3 86.6 87.2 92.1 91.4 92.3
Ab. globi(ormis
17.6 21.2 21.4 19.3 18.9 18.2 21.9
78.6 78.7 77.8 80.1 78.4 79.9 80.7 82.4 82.0
H. chlorum
21.2 22.3 19.4 21.4 17.7 21.3
80.5 80.8 79.8 81.9 78.6 79.5 79.6 82.0 81.3 84.1
Aa. (er(entans
15.9 21.9 18.3 16.9 23.2
78.5 78.8 77.9 78.9 77.6 78.8 79.9 81.1 81.1 81.2 81.2
23.7 20.8 19.4 25.6
77.7 77.4 76.6 78.1 76.2 76.8 76.4 78.8 78 .2 81.0 80.3 85.5
E. L. rhusio- vitupathiae tinum
18.6 17.7 24.2
78.7 79.8 78.9 78.5 77.8 78.1 78.4 79.7 79.4 81.1 81.0 79.0 77.7
C. butyricum
11.4 23.7
79.9 80.1 79.8 80.5 78.6 79.5 79.1 80.1 79.8 82.9 81.0 83.5 81.5 81.6
L.
acidophilus
21.2
78.8 80.2 78.8 82.6 80.3 81.0 81.3 82.5 83.0 83.5 83.9 84.6 82.7 82.3 89.3
B. subtitis
78.8 79.1 79.1 78.4 76.2 77.3 79.1 79.5 80.6 81.1 74.0 77.9 77.4 79.3 78.4 81.2
Synecchococcus
text for the regions compared. Abbreviations: A, Atopobium; Aa, Acidaminococcus; Ab, Arthrobacter; Ac, Actinomyces; B, Bacillus; Bf, Bi(idobacterium; C, Clostridium; E, Erysipelothrix; H, Heliobacterium; L, Lactobacillus; N, Nocardioides; S, Streptomyces
a The values on the lower left are evolutionary distances, which were calculated as described by Jukes and Cantor (1969). The values on the upper right are similarity values. See
A.minutum A.parvulum A.rimae Strain TH3 Bf. bi(idum Ac. pyogenes N.simplex S. coelicolor Ab. globi(ormis H.chlorum Aa. (ermentans E. rhusiopathiae L. vitulinus C. butyricum L. acidophilus B. subtilis Synecchococcus strain
Species/strain
Evolutionary distance or similarity value (%)a
Table 1. Evolutionary distances and similarity values in the regions compared for the 16S rDNAs of Atopobium species and various representatives of Gram-positive bacteria
-l:>
.....
'"
<1>
~
2
(Jl
::l "n·
r:tQ
0
::r '$.
'"
8. en 3 en S·
r:tQ
.,
..,0
<1>
()
::l
<1>
<1> <1>
..,.....
:;0
42
E. Stackebrandt and W. Ludwig
of the actinomycetes-and the Clostridium/Bacillus clusters. It is not important under this aspect to decide whether the two subphyla of Gram-positive bacteria are members of the same phylum (Woese, 1987) or whether they represent two individual phyla (Both et aI., 1992). Fig. 1. demonstrates that A. minutum, A. rimae and A. parvulum form one main subline of descent, equivalent in phylogenetic depth to those lines defined by the iron-oxidizing strain TH3 (Lane et aI., 1992) and the broad cluster of bifibobacteria (another group of "lactic acid bacteria") and the spore-and non-sporeforming actinomyctes. The branching point of A. minutum (in the past investigated as Lactobacillus minutus) is consistent to that published previously (Olsen and Woese, 1993; Rainey and Stackebrandt, 1993; Rainey et aI., 1993; Stackebrandt et aI., 1993). Evidence that members of Atopobium do not belong into the traditionally defined order Actinomycetales but constitute a deep branching line of descent within the actinomycetes subphylum is the low DNA G+C content of 35-46 mol% (Collins and Wallbanks, 1992), the presence of three signature nucleotides as indicated by Woese (1987) as being specific for bacilli and relatives (i. e., pos. 955 an U-residue in A rimae and A. parvulum, and an A-residues in position 1410 in all three Atopobium species) and the lack of a nucleotide in loop 1357-1365 which is present in all actinomycetes but which is also absent in strain TH3 and members of the low DNA G+C Gram-positive subphylum. Data that support the relationship of Atopobium species to the subphylum of actinomycetes are the significantly higher binary similarity values between Atopobium species and the actinomycetes shown in Fig.1 (82.283.3%) than between Atopobium species and members of the Clostridium/Bacillus line of descent shown in Fig. 1 (80.5-74.6). In addition, the 16S rRNA of the three Atopobium species have an A-residue in position 906, present in all actinomycetes but rarely found in members of the second subphylum of Gram-positive bacteria. The phylogenetic definition of the order Actinomycetales (Stackebrandt, 1982) to include Gram-positive organisms with a DNA base composition of more than 55 mol% G+C was described before the membership of Atopobium and strain TH3 to this branch was known. A revision of the description of Actinomycetales might be necessary if one wants to continue to embrace all members of this major subline in one order. Alternatively the two branches that lead to the Atopobium species and to strain TH3 could be considered to constitute two novel orders but at present the organismic sparesy of these lines prevents the neccessity to describe them. It should be noted that the branching pattern of reference organisms among each other may change slightly when sequences from a larger number of species are used to define more accurately the phylogenetic position of one of those genera from which a single species only was used in this study. Likewise, the relationship between reference genera, such as between Actinomyces and Arthrobacter or between Clostridium and Acidaminococcus, may change as well. The past has shown that for the investigation of the phylogenetic position of a novel taxon it is wiser to search for neighbors from a wide range of apparently re-
lated and apparently unrelated reference organisms then to restrict the selection to a large number of apparently related taxa. Secondly, the restriction of phylogenetic analyses to organisms with are defined by a common phenotypic property which itself has no phylogenetic relevance (i. e. lactic acid production) is a relapse into the "pre-phylogeny era". If the reference group would only consist of lactic acid bacteria any organism that produces lactic acid but is unrelated to Lactobacillus and relatives, e. g. a member of the actinomycete genus Bifidobacterium would neccessarily fall into the radiation of lactic acid bacteria and thus imply specific relationship. In order to avoid this dramatic bias of phylogenetic conclusion one should take advantage of the potential to measure phylogenetic relationships objectively by including a broader range of physiologically diverse types. Thirdly, despite the misleading indication of relationships, the conclusion of Collins and Wallbanks (1992) to describe Atopobium for the three misclassified "lactic acid bacteria", is justified. The phylogenetic distance to the nearest known neighbor is so large that there is no doubt that Atopobium represents a novel genus. The phenotypic properties determined as yet do not reflect this novelty and the unique phylogenetic position within the tree but additional epigenetic and genetic characters may be discovered.
References Both, B., Buckel, W., Kroppenstedt, R., Stackebrandt, E.: Phylogenetic and chemotaxonomic characterization of Acidaminococcus fermentans. FEMS Microbiol. Letters 97, 7-12 (1992) Brosius, j., Palmer, j. j., Kennedy, j. P., Noller, H. F.: Complete nucleotide sequence of a 16S ribosomal RNA gene from Escherichia coli. Proc. Nat!' Acad. Sci. USA 75, 4801-4805 (1978) Collins, M. D., Wallbanks, S.: Comparative sequence analysis of the 16S rRNA genes of Lactobacillus minutus, Lactobacillus rimae and Streptococcus parvulus: Proposal for the creation of a new genus Atopobium. FEMS Microbiol. Letters 95, 235-240 (1992) De Rijk, P., Neefs, j.-M., Van de Peer, Y., De Wachter, R.: Compilation of small ribosomal subunit RNA sequences. Nucl. Acids Res. 20, 2075-2089 (1992) De Soete, G.: A least squares algorithm for fitting additive trees to proximity data. Psychometrika 48, 621-626 (1989) Fox, C. R., Stackebrandt, E., Hespell, R. B., Gibson, j., Maniloff, j., Dyer, T. A., Wolfe, R. S., Balch, W. E., Tanner, R., Magrum, L., Hablen, L. B., Blakemore, R., Gupta, R., Bonen, L., Lewis, B. j., Stahl, D. A., Luehrsen, K. R., Chen, K. N., Woese, C. R.: The phylogeny of prokaryotes. Science 209, 457-463 (1980) Jukes, T. H., Cantor, C. R.: Evolution of protein molecules, pp.21-132. In: Mammalian Protein Metabolism, (Munro, H.N., ed.). Academic Press, New York 1969 Lane, D. J., Harrison, A. P., Stahl, D. A., Pace, B., Giovannoni, S. j., Olsen, G. j., Pace, N. R.: Evolutionary relationships among sulfur- and iron-oxidizing bacteria. J. Bacteriol. 174, 269-278 (1992)
Reference Organisms in Phylogenie Studies Lawson, P. A., P. A. Liop-Perez, Hutson, R. A., Hippe, H., Collins, M. D.: Towards a phylogeny of the clostridia based on 165 rRNA sequences. FEMS Microbiol. Lett. 113, 87-92 (1993) Olsen, G. j., Overbeek, R., Larsen, N., Marsh, T. 1., Xing, Y. Q., Woese, C. R.: The ribosomal RNA database project. Nucl. Acids Res. 20, 2199-2200 (1992) Olsen, G. j., Woese, C. R.: Ribosomal RNA: a key to phylogeny. FASEB 7, 113-123 (1993) Rainey, F. A., Stackebrandt, E.: Phylogenetic evidence for the classification of Acidothermus cellulolyticus in the subphylum of actinomycetes. FEMS Microbiol. Letters 108, 27-30 (1993) Rainey, F. A., Schumann, P., Prauser, H. Toalster, R., Stackebrandt, E.: Sporichtya polymorpha represents a novel line of descent within the order Actinomycetales. FEMS Microbiol. Letters 109, 263-268 (1993) Stackebrandt, E., Liesack, W., Goebel, B. M.: Bacterial diversity in a soil sample from a subtropical Australian environment as determined by 16S rDNA analysis. FASEB 7, 232-236 (1993)
43
Stackebrandt, E., Liesack, W., Witt, D.: Ribosomal RNA and ribosimal DNA sequence analysis. Gene 115,255-260 (1991) Stackebrandt, E.: What is an actinomycete? The Actinomycetes 16: 132-138 (1982) Weisburg, W. G., Tully, j. G., Rose, D. 1., Petzel, j. P., Oyaizu, H., Yang, D., Mandelco, 1., Sechrest, j., Lawrence, T. G., Van Etten, j., Maniloff, J., Woese, C. R.: A phylogenetic analysis of the mycoplasmas: basis for their classification. J. Bacteriol. 171,6455-6467 (1989) Woese, C. R., Achenbach, 1., Rouviere, P., 1. Mandelco, 1.: Archaeal phylogeny: reexamination of the phylogenetic position of Archaeoglobus fulgidus in the light of certain composition-induced artifacts. Syst. Appl. Microbiol. 14, 364-37l. (1991) Woese, C. R., Maniloff, j., Zablen, 1. B.: Phylogenetic analysis of the mycoplasmas. Proc. Nat!. Acad. Sci. USA 77, 494-498 (1980) Woese, C. R.: Bacterial evolution. Microbiol. Rev. 51, 221-271 (1987)
Professor Dr. E. Stackebrandt, DSM-Deutsche Sammlung von Mikroorganismen, Mascheroder Weg lB, 38124 Braunschweig, Germany. Tel. 0495312616352, FAX 0595312616418
The Importance of Using Outgroup Reference Organisms in Phylogenetic Studies: the Atopobium Case E. STACKEBRANDr'; and W. LUDWIG b aDSM-Deutsche Sammlung von Mikroorganismen, Mascheroder Weg 1B, 38124 Braunschweig, Germany, and fiir Mikrobiologie, Technische Universitat Miinchen, 80290 Miinchen, Germany
b Lehrstuhl
Received November 3, 1993
Summary Comparative 16S rDNA studies of lactic acid bacteria and some of their phylogenetic relatives of the Clostridium/Bacillus subphylum of Gram-positive bacteria indicated that three misclassified Lactobacillus and Streptococcus species, described as members of the novel genus Atopobium, formed separate sub line of descent within this subphylum. Inclusion of outside reference organisms such as members of the order Actinomycetales and a cyanobacterial species, serving as the root of Gram-positive bacteria, clearly showed Atopobium species to branch off within the actinomycetes line of descent. The importance of the use of an appropriate selection of reference organisms in phylogenetic studies is discussed.
Key words: Atopobium, Clostridium/Bacillus subphylum - Actinomycetales - 16S rRNA - phylogenetic analysis
Taxonomists that are not familiar with the interpretation of sequence data and the reconstruction of phylogenetic trees, assume not only that a phylogenetic tree of the 16S rRNA is based on optimal sequence alignment and a widely accepted treeing algorithm but also that the topology of a tree reflects by and large the correct position of the organisms analyzed. Although 16S rRNA sequences are known for only about 25% of described prokaryotic species, the approximate phylogenetic position of a taxon relative to its neighbor is believed to be rather stable and not to change when more species are included. This has been shown to be reasonable accurate when the topologies of dendrograms derived from 16S rRNA cataloging (Fox et al., 1980) were compared to the topologies of phylogenetic trees derived from complete 16S rRNA sequences (Woese, 1987). Certain changes were observed in the branching patterns especially in the lengths and the order of deep rooting lines of descent and certain corrections were made for rapidly evolving organisms. Differences in tree topologies have been shown to be caused not only by differences in the tempo of evolution (Weisburg et al., 1989; Woese et al., 1980), differences in the * Corresponding author
base composltlon of rRNA genes and unequal rates of evolution within different regions of the rRNA genes (Woese et al., 1991), but also by the selection of sequences stretches analyzed (Stackebrandt et al., 1991). Another important factor, well known to those directly involved in the generation of trees, is the influence of the number of organisms and the selection of reference organisms included in the analysis. The latter case has recently been demonstrated to result in misleading phylogenetic conclusions (Collins and Wallbanks, 1992; Lawson et al., 1993). In the .course of a broad survey on lactic acid bacteria (a term that reflects the production of lactic acid by Gram-positive bacteria and is by no means phylogenetically sound), the 16S rRNAs of Lactobacillus minutus, Lactobacillus rimae and Streptococcus parvulus were compared to those of members of various groups defined by lactic acid bacteria, such as members of Leuconostoc, Lactobacillus, Pediococcus, Streptococcus, Enterococcus and a few relatives. These three lactobacilli grouped neither with the core of Lactobacillus nor with any other lactic acid producing bacteria but formed "a hitherto unknown line of descent within the lactic acid group of bacteria" for which a new genus, Atopobium, was proposed (Collins and Wallbanks,
40
E. Stackebrandt and W. Ludwig
1992). More recently (Lawson et aI., 1993) these organisms were included as " ... representative low G+C Grampositive taxa known or considered to be related to clostridia ... " in a phylogenetic survey of members of Clostridium. In order to reevaluate the phylogenetic position of Atopobium the complete database of 16S rRNAs and 16S rDNAs (De Rijk et. aI., 1992; Olsen et aI., 192) of about 1500 sequences was analysed by a "rated parsimony" method (Ludwig, unpublished) (data not shown). A smaller selection of 13 species, seven from the Clostridium/ Bacillus subphylum and six from the actinomycetes subphylum, was then selected to cover a broad range of genetic diversity within the two subphyla. The following regions of alignment uncertainty and missing sequence information were omitted prior to the analysis: position 1 to 102, 179 to 224, 458 to 487, 930 to 968, 1028 to 1052, 1136 to 1154 and 1387 to 3' end (positions refer to the Escherichia coli numbering system [Brosius et aI., 1978]). The phylogenetic analysis was performed as described (Rainey and Stackebrandt, 1993 ). The phylogenetic tree is depicted in Fig. 1. The similarity values calculated between the Atopobium species and the reference organisms, that have been included in the previous study (Collins and Wallbanks, 1992) and this study, were very similar and differences can be explained by different stretches compared. However, with respect to the phylogenetic position of Atopobium species the comparison of the tree in Fig. 1 with that shown by Collins and Wallbanks (1992) and Lawson et al. (1993) demonstrated significant differences that are worth discussing:
,-----
Firstly, a proper selection of reference organisms would have avoided the misleading phylogenetic conclusion that Atopobium is either a relative of Erysipelothrix rhusiopathiae (Collins and Wallbanks, 1992) or of Clostridium leptum and C. sporosphaeroides (Lawson et aI., 1993). While E. rhusiopathiae is known to be a relative of the mycoplasmas and certain cell wall containing Gram-positive bacteria (including the misclassified Lactobacillus vitulinus and L. catenaforme) that were previously shown to be derived from Bacillus and relatives (Weisburg et aI., 1989; Woese et aI., 1980), the two clostridial species constitute an individual branch within the radiation of low G+C Gram-positives (Lawson et aI., 1993; Olsen et aI., 1992). Since the treeing algorithm does not determine whether or not the phylogenetic position of a deep branching organisms makes sense from a biological point of view, it is left to the microbiologist to do so. Bootstrap values, which were not included, would have been so low anyway that they would not have given a hint to whether the relationship between Erysiphelotrix or species of Clostridium on the one hand and Atopobium species on the other hand had any statistical significance. , The inclusion of a single sequence of the actinomycetes "proper" and a sequence of an outside reference organisms to define the root of the Gram-positive line of descent would have helped to solve the problem immediately. The root, here defined by a cyanobacterium, clearly separates the two main clusters of Gram-positive bacteria and the separate status is supported by high bootstrap values of 98%. The use of members from other bacterial phyla as outgroups do not change the topology of the bifurcation
- Arthrobacter g/obiformis
'---------
-
Actinomyces pyogenes
' - - - - - - Nocardioides simp/ex 51
'---------- Streptomyces coe/ic%r ' - - - - - - --
98
- Bifidobacterium bifidum
'-------------slrain TH3
Atopobium parvulum Atopobium rimae
100
' - - - - Atopobium minutum , -- - - - - -- Heliobacterium chlorum ' - - -- -, - - - --
- - - Acidaminococcus jermenlans -
-
-
Clostridium butyricum
, - - - - - - Erysipelothrix rhusiopathiae '------- --
-
-
Lactobacillus vilulmus
93 , - - - - - - Lactobacillus acidophilus ' - - - - - Bacillus subtilis 10%
Fig. 1. Dendrogram of 16S rONA relationships for the three Atopobium species and other members of the phylum of Gram-positive bacteria, generated by using the algorithm of DeSoete (1989) on dissimilarity values as calculated by Jukes and Cantor (1969). The sequence of the cyanobacterium Anacystis nidulans was used as the root for the two subphyla. Organisms in bold were shown as phylogenetic neighbors in reference (Collins and Wallbanks, 1992). Bootstrap values, based on the analysis of 1000 trees from 809 polymorphic sites were obtained using the programs NJFIND and NJBOOT (kindly provided by T.S. Whittam, Institut of Molecular Evolution and Genetics, Pennsylvania State University). Only values above 50% significance are indicated. Bar = 10% sequence divergence.
6.1 8.0 17.3 20.1 18.6 17.9 16.0 16.4 24.7 22.0 24.9 26.0 22.7 23.0 24.4 24.3
2.9 17.0 20.3 17.6 16.2 15.0 16.0 24.4 21.6 24.3 26.4 21.1 22.7 22.4 23.8
94.1 17.5 21.4 18.8 17.5 15.6 17.3 25.6 22.9 25.6 27.4 22.2 23.0 24.4 23.8
92.4 97.1
A.mi- A.par- A. nutum vulum rimae
18.8 16.4 16.4 14.8 14.7 22.5 20.2 24.2 25.3 23.0 22.0 19.4 25 .1
84.2 84.5 84.2
TH3
15.6 16.5 15.4 13.7 24.9 24.7 26.1 28.1 23.7 24.8 22.4 27.9
82.2 81.0 81.2 83.1
Bf. bi(idum
12.5 11.3 8.2 22.9 23.4 24.4 27.1 23.5 23.4 21.4 25.1
83.3 84.1 81.1 85.2 85.6
Ac. pyognes
10.1 9.0 21.8 23.3 22.9 27.7 23.0 23.9 21.1 26.6
83.9 85.2 84.2 85.2 84.9 88.4
N. simplex
8.0 19.7 20.2 21.3 24.3 21.3 22.5 19.5 23 .9
85.3 86.2 85.7 86.3 85.8 89.4 90.4
coelicolor
S.
20.1 21.1 21.3 25.2 21.8 23.0 18.8 23.5
85.1 85.4 84.3 86.6 87.2 92.1 91.4 92.3
Ab. globi(ormis
17.6 21.2 21.4 19.3 18.9 18.2 21.9
78.6 78.7 77.8 80.1 78.4 79.9 80.7 82.4 82.0
H. chlorum
21.2 22.3 19.4 21.4 17.7 21.3
80.5 80.8 79.8 81.9 78.6 79.5 79.6 82.0 81.3 84.1
Aa. (er(entans
15.9 21.9 18.3 16.9 23.2
78.5 78.8 77.9 78.9 77.6 78.8 79.9 81.1 81.1 81.2 81.2
23.7 20.8 19.4 25.6
77.7 77.4 76.6 78.1 76.2 76.8 76.4 78.8 78 .2 81.0 80.3 85.5
E. L. rhusio- vitupathiae tinum
18.6 17.7 24.2
78.7 79.8 78.9 78.5 77.8 78.1 78.4 79.7 79.4 81.1 81.0 79.0 77.7
C. butyricum
11.4 23.7
79.9 80.1 79.8 80.5 78.6 79.5 79.1 80.1 79.8 82.9 81.0 83.5 81.5 81.6
L.
acidophilus
21.2
78.8 80.2 78.8 82.6 80.3 81.0 81.3 82.5 83.0 83.5 83.9 84.6 82.7 82.3 89.3
B. subtitis
78.8 79.1 79.1 78.4 76.2 77.3 79.1 79.5 80.6 81.1 74.0 77.9 77.4 79.3 78.4 81.2
Synecchococcus
text for the regions compared. Abbreviations: A, Atopobium; Aa, Acidaminococcus; Ab, Arthrobacter; Ac, Actinomyces; B, Bacillus; Bf, Bi(idobacterium; C, Clostridium; E, Erysipelothrix; H, Heliobacterium; L, Lactobacillus; N, Nocardioides; S, Streptomyces
a The values on the lower left are evolutionary distances, which were calculated as described by Jukes and Cantor (1969). The values on the upper right are similarity values. See
A.minutum A.parvulum A.rimae Strain TH3 Bf. bi(idum Ac. pyogenes N.simplex S. coelicolor Ab. globi(ormis H.chlorum Aa. (ermentans E. rhusiopathiae L. vitulinus C. butyricum L. acidophilus B. subtilis Synecchococcus strain
Species/strain
Evolutionary distance or similarity value (%)a
Table 1. Evolutionary distances and similarity values in the regions compared for the 16S rDNAs of Atopobium species and various representatives of Gram-positive bacteria
-l:>
.....
'"
<1>
~
2
(Jl
::l "n·
r:tQ
0
::r '$.
'"
8. en 3 en S·
r:tQ
.,
..,0
<1>
()
::l
<1>
<1> <1>
..,.....
:;0
42
E. Stackebrandt and W. Ludwig
of the actinomycetes-and the Clostridium/Bacillus clusters. It is not important under this aspect to decide whether the two subphyla of Gram-positive bacteria are members of the same phylum (Woese, 1987) or whether they represent two individual phyla (Both et aI., 1992). Fig. 1. demonstrates that A. minutum, A. rimae and A. parvulum form one main subline of descent, equivalent in phylogenetic depth to those lines defined by the iron-oxidizing strain TH3 (Lane et aI., 1992) and the broad cluster of bifibobacteria (another group of "lactic acid bacteria") and the spore-and non-sporeforming actinomyctes. The branching point of A. minutum (in the past investigated as Lactobacillus minutus) is consistent to that published previously (Olsen and Woese, 1993; Rainey and Stackebrandt, 1993; Rainey et aI., 1993; Stackebrandt et aI., 1993). Evidence that members of Atopobium do not belong into the traditionally defined order Actinomycetales but constitute a deep branching line of descent within the actinomycetes subphylum is the low DNA G+C content of 35-46 mol% (Collins and Wallbanks, 1992), the presence of three signature nucleotides as indicated by Woese (1987) as being specific for bacilli and relatives (i. e., pos. 955 an U-residue in A rimae and A. parvulum, and an A-residues in position 1410 in all three Atopobium species) and the lack of a nucleotide in loop 1357-1365 which is present in all actinomycetes but which is also absent in strain TH3 and members of the low DNA G+C Gram-positive subphylum. Data that support the relationship of Atopobium species to the subphylum of actinomycetes are the significantly higher binary similarity values between Atopobium species and the actinomycetes shown in Fig.1 (82.283.3%) than between Atopobium species and members of the Clostridium/Bacillus line of descent shown in Fig. 1 (80.5-74.6). In addition, the 16S rRNA of the three Atopobium species have an A-residue in position 906, present in all actinomycetes but rarely found in members of the second subphylum of Gram-positive bacteria. The phylogenetic definition of the order Actinomycetales (Stackebrandt, 1982) to include Gram-positive organisms with a DNA base composition of more than 55 mol% G+C was described before the membership of Atopobium and strain TH3 to this branch was known. A revision of the description of Actinomycetales might be necessary if one wants to continue to embrace all members of this major subline in one order. Alternatively the two branches that lead to the Atopobium species and to strain TH3 could be considered to constitute two novel orders but at present the organismic sparesy of these lines prevents the neccessity to describe them. It should be noted that the branching pattern of reference organisms among each other may change slightly when sequences from a larger number of species are used to define more accurately the phylogenetic position of one of those genera from which a single species only was used in this study. Likewise, the relationship between reference genera, such as between Actinomyces and Arthrobacter or between Clostridium and Acidaminococcus, may change as well. The past has shown that for the investigation of the phylogenetic position of a novel taxon it is wiser to search for neighbors from a wide range of apparently re-
lated and apparently unrelated reference organisms then to restrict the selection to a large number of apparently related taxa. Secondly, the restriction of phylogenetic analyses to organisms with are defined by a common phenotypic property which itself has no phylogenetic relevance (i. e. lactic acid production) is a relapse into the "pre-phylogeny era". If the reference group would only consist of lactic acid bacteria any organism that produces lactic acid but is unrelated to Lactobacillus and relatives, e. g. a member of the actinomycete genus Bifidobacterium would neccessarily fall into the radiation of lactic acid bacteria and thus imply specific relationship. In order to avoid this dramatic bias of phylogenetic conclusion one should take advantage of the potential to measure phylogenetic relationships objectively by including a broader range of physiologically diverse types. Thirdly, despite the misleading indication of relationships, the conclusion of Collins and Wallbanks (1992) to describe Atopobium for the three misclassified "lactic acid bacteria", is justified. The phylogenetic distance to the nearest known neighbor is so large that there is no doubt that Atopobium represents a novel genus. The phenotypic properties determined as yet do not reflect this novelty and the unique phylogenetic position within the tree but additional epigenetic and genetic characters may be discovered.
References Both, B., Buckel, W., Kroppenstedt, R., Stackebrandt, E.: Phylogenetic and chemotaxonomic characterization of Acidaminococcus fermentans. FEMS Microbiol. Letters 97, 7-12 (1992) Brosius, j., Palmer, j. j., Kennedy, j. P., Noller, H. F.: Complete nucleotide sequence of a 16S ribosomal RNA gene from Escherichia coli. Proc. Nat!' Acad. Sci. USA 75, 4801-4805 (1978) Collins, M. D., Wallbanks, S.: Comparative sequence analysis of the 16S rRNA genes of Lactobacillus minutus, Lactobacillus rimae and Streptococcus parvulus: Proposal for the creation of a new genus Atopobium. FEMS Microbiol. Letters 95, 235-240 (1992) De Rijk, P., Neefs, j.-M., Van de Peer, Y., De Wachter, R.: Compilation of small ribosomal subunit RNA sequences. Nucl. Acids Res. 20, 2075-2089 (1992) De Soete, G.: A least squares algorithm for fitting additive trees to proximity data. Psychometrika 48, 621-626 (1989) Fox, C. R., Stackebrandt, E., Hespell, R. B., Gibson, j., Maniloff, j., Dyer, T. A., Wolfe, R. S., Balch, W. E., Tanner, R., Magrum, L., Hablen, L. B., Blakemore, R., Gupta, R., Bonen, L., Lewis, B. j., Stahl, D. A., Luehrsen, K. R., Chen, K. N., Woese, C. R.: The phylogeny of prokaryotes. Science 209, 457-463 (1980) Jukes, T. H., Cantor, C. R.: Evolution of protein molecules, pp.21-132. In: Mammalian Protein Metabolism, (Munro, H.N., ed.). Academic Press, New York 1969 Lane, D. J., Harrison, A. P., Stahl, D. A., Pace, B., Giovannoni, S. j., Olsen, G. j., Pace, N. R.: Evolutionary relationships among sulfur- and iron-oxidizing bacteria. J. Bacteriol. 174, 269-278 (1992)
Reference Organisms in Phylogenie Studies Lawson, P. A., P. A. Liop-Perez, Hutson, R. A., Hippe, H., Collins, M. D.: Towards a phylogeny of the clostridia based on 165 rRNA sequences. FEMS Microbiol. Lett. 113, 87-92 (1993) Olsen, G. j., Overbeek, R., Larsen, N., Marsh, T. 1., Xing, Y. Q., Woese, C. R.: The ribosomal RNA database project. Nucl. Acids Res. 20, 2199-2200 (1992) Olsen, G. j., Woese, C. R.: Ribosomal RNA: a key to phylogeny. FASEB 7, 113-123 (1993) Rainey, F. A., Stackebrandt, E.: Phylogenetic evidence for the classification of Acidothermus cellulolyticus in the subphylum of actinomycetes. FEMS Microbiol. Letters 108, 27-30 (1993) Rainey, F. A., Schumann, P., Prauser, H. Toalster, R., Stackebrandt, E.: Sporichtya polymorpha represents a novel line of descent within the order Actinomycetales. FEMS Microbiol. Letters 109, 263-268 (1993) Stackebrandt, E., Liesack, W., Goebel, B. M.: Bacterial diversity in a soil sample from a subtropical Australian environment as determined by 16S rDNA analysis. FASEB 7, 232-236 (1993)
43
Stackebrandt, E., Liesack, W., Witt, D.: Ribosomal RNA and ribosimal DNA sequence analysis. Gene 115,255-260 (1991) Stackebrandt, E.: What is an actinomycete? The Actinomycetes 16: 132-138 (1982) Weisburg, W. G., Tully, j. G., Rose, D. 1., Petzel, j. P., Oyaizu, H., Yang, D., Mandelco, 1., Sechrest, j., Lawrence, T. G., Van Etten, j., Maniloff, J., Woese, C. R.: A phylogenetic analysis of the mycoplasmas: basis for their classification. J. Bacteriol. 171,6455-6467 (1989) Woese, C. R., Achenbach, 1., Rouviere, P., 1. Mandelco, 1.: Archaeal phylogeny: reexamination of the phylogenetic position of Archaeoglobus fulgidus in the light of certain composition-induced artifacts. Syst. Appl. Microbiol. 14, 364-37l. (1991) Woese, C. R., Maniloff, j., Zablen, 1. B.: Phylogenetic analysis of the mycoplasmas. Proc. Nat!. Acad. Sci. USA 77, 494-498 (1980) Woese, C. R.: Bacterial evolution. Microbiol. Rev. 51, 221-271 (1987)
Professor Dr. E. Stackebrandt, DSM-Deutsche Sammlung von Mikroorganismen, Mascheroder Weg lB, 38124 Braunschweig, Germany. Tel. 0495312616352, FAX 0595312616418