- Email: [email protected]
The phylogeny of purple bacteria: The alpha subdivision
System. Appl. Microbiol. 5, 315-326 (1984) Department of Genetics and Development, University of Illinois, Urbana, IL 61801, U. S.A. Department of Dairy Science, University of Illinois, Urbana, IL 61801, U. S.A. 3 Lehrstuhl flir Mikrobiologie, Technische Universitiit Mlinchen, 8000 Mlinchen, Federal Republic of Germany 4 Department of Microbiology, Southern Illinois University, Carbondale, IL 62901, U.S.A • Scripps Institute of Oceanography, La Jolla, CA 92093, U.S.A. 6 Department of Biochemical and Biophysical Sciences, University of Houston, University Park, Cullen Blvd., Houston, TX 77004, U.S.A. 1
2
The Phylogeny of Purple Bacteria: The Alpha Subdivision C. R. WOESE1':', E. STACKEBRANDT3, W. G. WEISBURG!, B. J. PASTER1,2, M. T. MADIGAN 4 , V. J. FOWLER 3, C. M. HAHN!, P. BLANZ!, R. GUPTA!, K. H. NEALSON 5 , and G. E. FOX 6
Received April 24. 1984
Summary The technique of oligonucletide cataloging shows the purple photosynthetic eubacteria to comprise three major subdivisions, temporarily called alpha, beta, and gamma - previously designated groups I-Ill by Gibson et al. (1979). Each subdivision contains a number of non-photosynthetic genera in addition to the photosynthetic ones. The alpha subdivision, the subject 'of the present report, contains most but not all of the species that fall into the classically defined genera Rhodospirillum, Rhodopseudomonas and Rhodomicrobium. Intermingled with these are a variety of non-photosynthetic species from genera such as Agrobacterium, Rhizobium, Azospirillum, Nitrobacter, Erythrobacter, Phenylobacterium, Aquaspirillum, and Paracoccus. The phylogenetic substructure of the alpha subdivision is presented and the evolutionary significance of the admixture of biochemical phenotypes is discussed.
Key words: Alpha purple bacteria - Evolution - Phylogeny - rRNA catalog
Introduction Partial sequencing of ribosomal RNAs (by the oligonucleotide cataloging method - Fox et aI., 1977) has let us see for the first time the panorama of microbial evolu-
tion. Two vastly different classes of bacteria inhabit the microbial world, the true bacteria (or eubacteria) and the archaebacteria - each no more like the other than it is like the eucaryotes (Woese and Fox, 1977). In a real sense the two are alike " To whom requests for reprints should be sent.
316
C. R. Woese et aI.
in name only; they are that distinct at the molecular level. The eubacteria are the more thoroughly characterized group - with over 300 species cataloged to date. These catalogs permit us to define most if not all the major phylogenetic groups of eubacteria, of which there are approximately ten (Woese et aI., in preparation). Each such group is equivalent phylogenetically to a eucaryotic Phylum if not a Kingdom, and, interestingly, they share with eucaryotic phyla what seems to be a basic evolutionary characteristic - i. e. all appear to branch almost simultaneously from their common ancestor point. Similarly, as with the Classes of animals, the main subdivisions within each bacterial "phylum" again appear to spring in close succession (if not coincidently) from the common ancestor of the group. For this apparent reason it has not yet been possible to ascertain the exact order of branching either between the bacterial "phyla" or among major subdivisions within any "phylum" (Fox et aI., 1980). As the phylogeny of the eubacteria unfolds, it becomes increasingly apparent that our prejudices as to the evolutionary relationships among bacteria (implied by the present system of classification) are for the most part wrong. Perhaps no bacterial group exemplifies the prejudices and problems of the existing system of bacterial classification more than do the purple photosynthetic bacteria. These organisms have played a major role in the history of microbiology, and their consequent revered status may contribute to their being singled out for high taxonomic distinction (Truper and Pfennig, 1981). In any case, their classically defined taxonomy is now known to be in considerable disagreement with their phylogenetic relationships, as defined by the genotypic criterion of rRNA sequence comparisons. One major discrepancy is the failure of the classically defined categories to reflect the close relationships among various of the purple photosynthetic bacteria and certain non-photosynthetic species; for example, the specific relationships between Rhodopseudomonas sphaeroides and Paracoccus denitri{icans (Gibson et aI., 1979), between R. palustris and the rhizobacteria, and between Rhodospirillum rubrum and Azospirillum brasilense (Woese et aI., 1982). While the classical split between non-sulfur (prefix Rhodo-) and sulfur purple bacteria, which defines the two major subdivisions of the purple bacteria, is phylogenetically valid in one sense, it is not in another. Within the purple non-sulfur group itself there occurs a major phylogenetic split no less profound than the classically defined split
between the sulfur and non-sulfur types (Gibson et aI., 1979). Although the newly recognized division was defined in terms of ribosomal RNA sequence homologies, it can also be rationalized in terms of basic properties such as cytochrome c subunit type (Dickerson, 1980) and the structure of the photo-reaction center (Clayton and Clayton, 1978). In the present communication we will define the purple photosynthetic bacterial group in terms of its oligonucleotide signature (Woese et aI., 1980a), and explore the phylogenetic structure of one of its (three) major subdivisions, alpha, which contains all but a few of the purple non-sulfur bacteria - such as Rps. gelatinosa and R. tenue (Gibson et aI., 1979). Materials and Methods The organisms used in this study are these: Rhodospirillum rubrum ATCC 11170, Rhodopseudomonas palustris ATCC 11168, Rps. capsulata strain St. Louis, Rps. viridis
The Phylogeny of Purple Bacteria: The Alpha Subdivision
317
strain NTH, Rps. sphaeroides strain NCIB 8253, and Rhodomicrobium vannielii, all reported in Gibson et ai. (1979); Aquaspirillum itersonii ATCC 12639, Azospirillum brasilense ATCC 29145, as reported in Woese et ai. (1982); Pseudomonas diminuta ATCC 11508, and Rhizobium leguminosarum ATCC 10004 (Woese et aI., 1984); Nitrobacter winogradskyi (Seewaldt et aI., 1982); Phenylobacterium immobile DSM 1986 (Ludwig et aI., 1984). [A second species of Nitrobacter, N. hamburgensis (Bock et aI., 1983) is sufficiently similar in its rRNA catalog to N. winogradskyi that the former will be used throughout to represent both.] Not reported previously are R. photometricum ATCC 27871, Rps. globiformis ATCC 7950, R. molischianum ATCC 14031, Rps. acidophila ATCC 7050, Paracoccus denitrificans ATCC 17741, Agrobacterium tumefacians DSM 30150, R. salexigens DSM 2132 (Drews, 1981), Erythrobacter longus IFO 14126 (Shiba and Simidu, 1982) (a gift of Prof. Shiba), and the manganese oxidizing isolates #36 and #63 (K.H.N. unpublished) . Growth and labelling of the organisms was as described in the appropriate references cited above. Special growth conditions for various of the species are described in Madigan and Cox (1982), Madigan et ai. (1984), Drews (1981) and Pfennig (1969). Extraction of 32P labelled (or unlabelled) 16S rRNA and generation of the oligonucleotide catalogs followed either of two standard procedures (Uchida et aI., 1974; Woese et aI., 1976; Stackebrandt et aI., 1982; Fowler et aI., 1984). The numeric analysis of data, in terms of binary association coefficients, or Sab values, is described in Fox et ai. (1977). Results and Discussion The cataloging method for 165 ribosomal RNA is generally very reliable for determining the phylogenetic relationships among bacteria. It appears to be more reliable than the methods based upon full sequencing of 55 ribosomal RNA, cytochrome c, and so on. The main reason for this lies in the fact that the latter molecules are small, while the former is large, i.e. the latter comprise a small number of functionally defined domains, while the former comprises perhaps an order of magnitude more such domains, which for the 165 rRNA means perhaps fifty (Woese et aI., 1983). Rarely, one such "domain" becomes redesigned evolutionarily, which results in a number of changes in sequence that are not selectively neutral. These bursts of sequence change distort the molecular distance measure when the extent of the perturbation is significant in terms of the whole sequence - which is the case for the small molecules, but is not for the large ones. It has been customary to analyse rRNA catalog data in terms of binary association coefficients, or 5ab values (Fox et aI., 1977), and from a table of these to construct a dendrogram by the unweighted pair group method (Fox et aI., 1977). This approach is not the optimal analysis, for it does not make full use of the data. The method is reliable when all of the species considered represent lines of descent that evolve at the same rate. Catalogs representing rapidly evolving lines of descent are poorly ordered phylogenetically by this approach (Woese et aI., 1980). [Fortunately, most bacterial lines appear to be evolving at about the same rate (Stackebrandt and Woese, 1981).] The use of binary association coefficients also does not take advantage of the fact that different positions in the (165) rRNA sequence have different phylogenetic significance. [Positions can differ by well over an order of magnitude in the frequency with which they change over evolutionary time. And, some positions exhibit characteristic, recognizable patterns of change. Also some positions appear to change during the formation of major phylogenetic groups, but remain constant once lines
318
C.R.Woeseetal.
Table 1. Sab values (Fox et a!', 1977) for various purple bacteria with one another and with outgroup organisms, showing the coherence of the purple group and its major subdivisions. Numbers correspond to the following organisms: 1 - R. rubrum, 2 - R. molischianum, 3 - Rps. capsulata, 4 - Rps. palustris, 5 - Rh. leguminosarum, 6 - Rm. vannietii,
7 - E. longus, 8 - Rps. gelatinosa, 9 - Ps. acidovorans, 10 - Thiobacillus intermedius, 11 R. tenue, 12 - Aquaspirillum serpens, 12 - Ps. cepacia, 14 - Spirillum volutans, 15 - Escherichia coli, 16 - Ps. fluorescens, 17 - Oceanospirillum maris, 18 - Beggiatoa leptomitiformis, 19 - Chromatium vinosum, 20 - Ectothiorhodospira strain 7, 21 - Legionella pneumophilia, 22 - Bacillus subtilus, 23 - Aphanocapsa str 6714, 24 - Flavobacterium heparinum. (Fox et a!', 1980; Gibson et a!', 1979; Woese et a!', 1982; Woese et a!., 1984; and unpub-
lished catalogs)
alpha group 1 234 567 1 R rub
beta group 8 9 10 11 12 13 14
2 3 4 5 6 7
R mol Reap R pal Rh Ie R vir E Ion
5640 43 36 44 47 55 41 48 41 42
-43 53 40 43
-61 -50 56 -40 48 37 --
8 9 10 11 12 13 14
R gel Ps ac
28 28 32 30 26 29 24
33 35 36 34 31 31 29
32 32 31 31 30 29 28
26 29 28 26 31 27 25
32 32 34 31 32 30 28
32 30 34 29 31 28 26
33 34 35 33 33 30 31
68 64 60 51 57 49
-52 53 52 50 45
-58 51 51 41
59 -61 54 -53 51 54 --
34 34 34 26 41 35 32
33 34 31 30 41 35 34
34 30 29 24 34 30 36
28 34 26 27 38 26 33
31 34 28 29 36 31 37
28 31 26 28 35 32 35
33 36 28 26 37 36 36
32 35 33 30 36 39 34
33 36 28 30 34 33 32
29 33 28 30 37 36 33
32 32 24 26 37 31 33
15 16 17 18 19 20 21
Th in
R ten Aq se Ps ce Sp vo
E col Ps f1 Oc ma Be Ie
Cr vi Ec 7
Le pn
22 B sub 23 Aph 24 Fl he
33 32 27 28 33 32 31 30 25 28 20 22 27 26 26 26 26 24 24 24 25
31 35 27 28 34 31 33
31 37 30 31 31 35 33
29 31 26 26 31 30 32
29 31 29 27 31 27 24 29 25 21 19 23 22 20 28 28 23 19 22 18 17
galllDa group 15 16 17 18 19 20 21
44 44 32 45 36 38
-49 33 42 41 42
30 38 38 40
other 22 23
-31 -37 48 35 41 41 --
26 32 25 20 32 32 25 28 30 27 17 30 27 23 27 21 26 20 29 24 25
29 -22 21
of descent have become established (Woese, 1984).] By developing families of the oligonucleotides that cover (some of) the phylogenetically more significant positions in the molecule, one can refine considerably and extend the customary analysis, based upon Sab values (Fox et aI., 1977). Table 1 is a set of Sab values that distinguishes the greater group of purple bacteria and their relatives from the other major eubacterial groups and defines the main subgroups of the former. It is not our intention to provide formal names for the main group or its subgroups (though this naming will become necessary), so the tentative designations "purple bacteria" or "purple group" will be used when referring to the entire collection, and the prefixes alpha, beta, and gamma added when referring to the three main subdivisions; within each of these the suffixes - 1, - 2, etc. will be used to denote further substructure. Table 2 defines the purple bacteria in terms of an oligonucleotide signature (Woese et aI., 1980a). The sequences shown are found in many to most catalogs
The Phylogeny of Purple Bacteria: The Alpha Subdivision
319
Table 2. Oligonucleotide signature distinguishing the purple group from other major bacterial groups. For each group the number of organisms included is shown below the column designation. The "all" column refers to the sum of the three purple bacterial subdivisions and the numbers in this column are fraction of total rather than number of organisms, as in the other columns. The "other occurrences" column lists the total number of occurrences of the sequence in question among eubacteria outside of the purple group. Bases in parenthesis - e.g. (A, Y) - are alternatives to one another. Abbreviations: Y-pyrimidine, R-purine; G + - Gram positive group; CFB - cytophaga-flavobacter-bacteroides group; Myx - myxobacteria group; Gr - green photosynthetic bacteria. sequence
alpha (23)
beta (27)
ganma (53)
25 27 4 27 16 22 0 25 12
31 32 25 47 16 2 41 19 13
other occurrences (240)
all ( 103)
-----------------------------------------------------------------AUCCAG ACAAUG AnUUCG UUAAUCG CCCCCUG CCAACCCG CCCUUACG AAAUye Y( A. Y)G CUYACCAC(G,U)UUG
13 19 2 0 0 1
17 9
0
10 [6G+; 4CFB] 15 [9G+; 4CFB] 1 1 0 8 [5G+; 3Hyx] 20 [ 14CFB; 5Gr] 3 4
0.64 0.72 0.29 0.68 0.31 0.24 0.56 0.51 0.24
of organisms within the purple group but rarely if at all in catalogs representing organisms outside the group. [Oligonucleotides that are found in only one of the three subdivisions of the purple group are not included in this signature.] Table 3 distinguishes the alpha purple bacterial subdivision from the other two subdivisions. Note that by signature as well as by the Sab measure, the two groups of purple non-sulfur bacteria are quite distinct from one another. The first of these groups, alpha, concerns us in detail here; the other groups and isolated individual members of the greater purple group are the subjects of subsequent publications. Table 3. Oligonucleotide signature distinguishing the alpha subdivision of purple bacteria from the beta and gamma subdivisions. "Other" refers to occurrences among eubacteria outside of the purple group. Abbreviations: A-1, -2, or -3 refer to the three main subgroups of the alpha subdivision; £1 - Erythrobacter longus; Band G refer to the beta and gamma subdivisions of the purple bacteria; other abbreviations as in Table 2 caption. sequence
A-1 (7)
A-2 (10)
A-3 (5)
El (1)
B (27)
G (53)
0 2 1 0 0 0 0 0 0 2 0
0 4 0 0 2 0 0 1 0 0
other (240)
-------------------------------------------------------------AUAAUG CCUUUG CUAACCG ceCUUCG CUCACUG AAAUUCG AUAUUCG AUCCAUAG AUUUAUCG AAAUCCCAG CAACCC( A, U) CR ••
0 2 5 2 0 7 1 2 3 1 6
9 10
5 2 7 10 9
4
10 3 9
5 3 0 3 5
5
5 3
4 4
5
+ + 0 0 0 + + 0 + 0
1
24 18 11 27 5 12 4 0 18 0 many
The alpha subdivision can be substructured by signature into three clusters plus an isolated species, E.longus, which represents a fourth cluster; see Table 4. [A meaningful signature cannot be defined for this fourth cluster on the basis of one
320
C. R. Woese et al.
Table 4. Oligonucleotide signature substructuring the alpha subdivision of purple bacteria. The abbreviations for species names are as follows: R. rubrum - Rr, R. photometricum Rp, Aq. itersonii - Ai, Az. brasilense - Ab, Rps. globiformis - Rg, Rm. molischianum Rrn, R. salexigens - Rs, Rps. acidophila - Ra, Rps. viridis - Rv, Rm. vannielii - Rn, Rps. palustris - Rp, N. winogradskyi - Nw, Rh.leguminosarum - RI, Ag. tumifaciens - At, Ps. diminuta - Pd, Ph. immobile - Pi, Rps. sphaeroides - Ro, Rps. capsulata - Rc, P. denitrificans - Pn, the two manganese oxidizing isolates - 36 and 63, and E. longus - El. "+" indicates the presence, "." the absence of an oligonucleotide.
sequence
R R A A R R R r p i b g ms
R R R R N RAP P a v n p w1 t d i
RRP 3 6 o c n 6 3
E 1
----------------------------------------------------------------? ACAAG AU AU UG CACUCCG UAACACG AAUUCCCAG UCAcACCAUG
UACCCG CAAUACG AACUCCG AAACUUG AAUCUUG CCACAUUG ACCCUUCG AUUUACUG CCCUUACAG UUUUACCCG CUACCAUUUAG CCAUCAUUUAG CCCAAACYCCYACG AUCUAG AUUAAG UAAUACCG AAUCUUAG ACACACCUAACG CAACCCACA ••• UUAAUCCC •••• AACCUUACCAACCCUUG
+ + + + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + + +
+
+
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+
+ +
+ + +
+
+ + + + + + + + + + + + + +
+ + + + +
+ + + + + + + + + + +~ + +
+
+ + +
+ + + + +
species alone. ] Table 5 shows the internal structure in the alpha subdivision as defined by 5ab values, and Table 6 provides additional phylogenetic fine structure as defined by signature. The groupings defined on the basis of 165 rRNA catalogs, a genotypic measure, are readily rationalized in terms of phenotypic properties as well. The first subgroup, alpha-I, comprising R. rubrum, R. photometricum, Rps. globiformis, R. molischianum, R. salexigens, Aq. itersonii, and Az. brasilense, is notable for the spiral shape of most of its representatives. Two types of photosynthetic membranes are characteristic of the group, the stacked configuration and the vesicular configuration (Truper and Pfennig, 1981) (the latter also being found among members of alpha-3 - see below).
The Phylogeny of Purple Bacteria: The Alpha Subdivision
321
Table 5. Set of Sab values (Fox et aL, 1977) substructuring the alpha subdivision of the purple bacteria. Organism name abbreviations are as in Table 4. alpha-l
1 234
567
alpha-2
8 9 10 11 12 13 14 15 16
1 Rr 2 Rp 3 Ai 4 Ab 5Rm 6 Rs 7 Rg
74 54 51 56 48 44
-49 51 49 45 45
-47 52 47 -46 51 40 -48 50 48 47 --
8 Nw 9 Rp 10 Rl 11 At 12 Rv l3Rn 14 Pd 15 Pi 16 Ra
35 36 47 46 50 41 43 40 37
34 38 42 39 48 39 39 41 36
44 46 53 48 52 46 48 48 47
34 36 40 39 47 39 42 38 43
42 44 55 51 55 48 40 43 44
37 36 43 40 46 45 41 38 37
38 41 47 42 48 42 41 36 40
7760 61 50 51 52 54 44 50 40 45 41 41 53 56
-78 59 56 48 52 57
57 51 43 44 47
-58 51 50 54
-43 -48 54 -43 49 44 --
17 18 19 20 21
40 41 37 41 38
36 37 35 38 38
44 46 44 42 39
45 43 43 43 43
43 41 40 41 37
40 44 43 41 38
34 36 34 33 32
41 44 42 36 41
53 54 51 47 52
50 48 46 43 46
45 47 45 47 46
40 41 42 36 39
Rc Ro Pn 36 63
22 El
41 40 46 39 42 41 44
43 42 44 37 43
45 51 44 46 46
49 40 45 40 43
alpha-3
17 18 19 20 21
42 47 43 40 42
7171 67 61 66 58 61 63 61 75 --
38 40 48 46 47 37 41 39 44
43 45 42 39 44
22
Members of the alpha-2 subgroup, represented by Rhodopseudomonas palustris, Rps. viridis, Rps. acidophila, Rm. vannielii, Ps. diminuta, Rh.leguminosarum, and others, have two gross morphological features in common. They generally divide in an asymmetric fashion (or actually bud), and they have intracytoplasmic membranes that are adjacent and parallel to the cytoplasmic membrane (Truper and Pfennig, 1981). These properties pertain, at least, to the photosynthetic members and to the two N itrobacter species (Watson, 1971; Seewaldt et aL, 1982). This subgroup contains organisms of interest because of their intimate intracellular association with plant cells - i.e. the rhizobacteria and agrobacteria. For those members of the subgroup that have so far been characterized, the sugar 2,3-diamino2,3-dideoxy D-glucose has been found in the lipid A moiety of the lipopolysaccharide (see Ludwig et aL, 1984). There is an indication from cytochrome c sequence comparisons that most mitochondria may have come from this particular subgroup as well (Swartz and Dayhoff, 1978). The alpha-3 subgroup, is represented by Rhodopseudomonas sphaeroides, Rps. capsulata, Paracoccus denitrificans and the two manganese oxidizing strains of Nealson. All share the unusual property of having 235 rRNAs that are post-transcriptionally cleaved into two nearly equal halves (Marrs and Kaplan, 1970; Gibson et aL, 1979; Woese et aL, unpublished observations). [We have not encountered this property in any other bacteria in the cataloging study except for Flavobacterium capsulatum, an organism whose catalog is incomplete, but which can nevertheless be placed in the alpha subdivision, though not within any of its three subgroups, by signature analysis (Blanz and Weisburg, unpublished).] Within this subgroup the two photosynthetic strains and P. denitrificans form a cluster distinct from the cluster formed by the two manganese oxidizing isolates; see Tables 5 and 6.
322
C. R. Woese et al.
Table 6. Oligonucleotide signatures supporting some specific subclusters within the various subgroups of the alpha purple bacteria. The sequences shown are found within the alpha subdivision only in the species under which they are listed. The number of their occurrences in the beta and gamma subdivisions, and elsewhere, are listed in the appropriate columns. sequence
beta
A. !. rubrum-!.photOllletricum (alpha-I) 0 CCCAAG I AAUCCG 0 UACAUG 0 CCACCUG 0 UACCUCG 0 UACCUUG CAUCCUUCG 0 ACUAAUACCG 0 B. Rhizobium-Agrobacterium (alpha-2) 0 UAUCCG 0 UAUACUG 1 CUAACCACG 0 UCCAUUACUG 0 CUCUUUCACCG
ga1llDa
other
2 0 0
10
5
0 0 0 0 0 0 0 0 0
C• .!£!. palustris-!. winogradskyi (alpha-2) 0 CUCUUUUG 0 0 0 CAACCCCCG 0 0 UCCAACUUG 0 UACCUUUUG 0 0 UUUACUCACUAG 0 0 0 ACCUUCUCUUCG diminuta-Ph. immobile (alpha-2) 0 UCUUAG CCAUUAG 8 (A,C)CACUCUAAUG 13 UAAAAU(A,U)CUUUCACCG 0 AACCUUACCACCUUUUG 0
D. ~.
2 0 I 0
0
11
I
8
2
14
0 2
15
0 0 0 0 0 0 0 0 0 0
8
3 2 0 0
E• .!£!. sphaeroides-.!£!. capaulata-f!. denitrificans (alpha-3) UAAUACCG 0 0 8 UUCAACUUG 0 0 0
F.
Mn-ox #36-Mn-ox #63 (alpha-3) 0 UUUUAG AAACCCCG 0 CUAACCUUCG 0 0 CAACCCACAUCCUUAG 0 UUAAUCCC(C,A)MAAACUG 0 AAACCCCG
0
0 0 0
0 0
3 1
4
0 0 I
It is of interest that most of the phenotypic properties of this group of organisms do not fall strictly along the phylogenetically defined lines. Photosynthetic capacity itself is, of course, the most obvious one. Throughout the alpha subdivision photosynthetic and non-photosynthetic phenotypes are completely intermixed. There is little doubt that the non-photosynthetic members have arisen from photosynthetic ancestors rather than the reverse. E. longus, which contains bacteriochlorophyll a, but is an aerobe (Shiba and Simidu, 1982), could represent a transition phase. Cytochrome c subunit type correlates only to a first approximation with these
The Phylogeny of Purple Bacteria: The Alpha Subdivision
323
Sob
)
Ps.
A. diminula lumefac/(ms
Ph. immobile
Rh. lequminasarum
R
N. It'/noqradskyi
Rps paluslris
molischianum Rps. viridis Rm vannlelii
B
7
6
5
4
y -group
13 - group
3 Fig. 1. Schematiclphylogenetic tree representing the relationships shown by Sab analysis as modified by signature analysis. The order of branching is that revealed by the combined analysis. The branching points are drawn in approximate agreement with the Sab analysis.
groupings (Woese et aI., 1980b). Those cytochromes c from the alpha-3 subgroup which have been studied, are all of the large subunit type. However, within the other subgroups both large and medium sized cytochrome c types are encountered (Dickerson, 1980). Furthermore, the subgroups as defined by rRNA catalog comparisons are in imperfect agreement with the groupings defined by cytochrome c sequence comparisons (Dickerson, 1980). Both approaches define the same alpha-3 subgroup, as well as the Rm. vannielii-Rps. viridis-Rps. acidophila subcluster of alpha-2, and the R. rubrum-R. photometricum subcluster in alpha-I. And Rps. palustris, despite the size difference of its cytochrome c, does fall with its relatives in alpha-2 by that measure. However, a clustering of R. molischianum or Rps. globiformis with the other members of alpha-1 (or with one another) does not occur by cytochrome c analysis; both cluster peripherally with the alpha-2 subgroup. Some, if not all of the lack of agreement here is due to the fact that for the deep phylogenetic relationships represented by the bacteria the cytochrome c molecule
324
C.R. Woese et al.
ceases to be a good molecular chronometer. Sequence comparisons between cytochromes c of different size undoubtedly yield less resemblance than would comparisons between cytochromes c of the same size, all other factors being equal. If one of the two size classes has evolved more than once from the other (which would seem to be the situation here), then further non-chronometric changes (either convergences or divergences) can have been introduced. For these reasons we do not consider the cytochrome c molecule to be as reliable a chronometer for this group of organisms as is the ribosomal RNA. There seems reasonable agreement between the cytoplasmic membrane types and the present genotypic grouping - though again, imperfect (Truper and Pfennig, 1981). The organisms in alpha-2, as stated, exhibit a "parallel" membrane orientation; the membranes for alpha-3 organisms are of the vesicular type, as are those of R. rubrum and Rps. globiformis, from group alpha-l, but those of R. photometricum and R. molischianum (also in alpha-i) exhibit a stacked configuration. [R. rubrum and R. photometricum are specific relatives of one another by both the rRNA and the cytochrome c criteria.] Neither the isoprenoid composition (Collins and fones, 1981) nor the predominant carotenoids (Truper and Pfennig, 1981) can be used to distinguish phylogenetically among the subgroups of the alpha subdivision. What these various disagreements point out is not that the rRNA genotypic measure of relatedness is questionable, for it is certainly the most consistent measure yet developed for bacterial phylogenetic classification, and it yields a more consistent picture of phylogeny of the present group than does any other criterion or combination thereof. What is emerging from these comparisons is a picture of a bacterial evolution that is highly varied in biochemical and morphological aspects (as seen in other major groups as well- Stackebrandt and Woese, 1981). Photosynthetic membrane types can change (compare R. rubrum and its close relative R. photometricum); cytochrome c size can change (compare the Rps. palustris and Rps. viridis examples, which differ in size but retain related sequences); apparent mode of growth and division can change, and so on. While it is fashionable to invoke inter-species gene transfer to explain these sorts of phenomena (Ambler et aI., 1979), we would be loathe to do so without far better evidence. Rather, the group under consideration is an enormous one from an evolutionary perspective; it would seem to be broader, older than all eucaryotes taken together - as judged by cytochrome c trees, for example. From this perspective, remarkable shifts in various phenotypic characteristics become less remarkable. Acknowledgements. CRW was supported by a grant from the National Science Foundation; ES was financed by the Gesellschaft fur Biotechnologische Forschung to support the German Collection of Microorganisms.
References Ambler, R.P., Daniel, M., Hermosa, j., Meger, T.E., Bartsch, R.C., Kamen, M.D.: Cyto-
chrome c2 sequence variation among the recognised species of purple nonsulphur photosynthetic bacteria. Nature 278, 659-660 (1979) Bock, E., Sundermeyer-Klinger, H., Stackebrandt, E.: Characterization of a novel facultative lithoautotrophic nitrite-oxidizing bacterium. Arch. Microbiol. 136,281-284 (1983)
The Phylogeny of Purple Bacteria: The Alpha Subdivision
325
Clayton, R. f., Clayton, R. K.: Properties of photochemical reaction centers purified from Rhodopseudomonas gelatinosa. Biochim. Biophys. Acta 501, 470-477 (1978) Collins, M. D., fones, D.: Distribution of isoprenoid quinone structural types in bacteria and their taxonomic implications. Microbiol. Rev. 45,316-354 (1981) Dickerson, R. E.: Evolution and gene transfer in purple photosynthetic bacteria. Nature 283, 210-212 (1980) Drews, G.: Rhodospirillum salexigens, sp. nov. an obligatory halophilic phototrophic bacterium. Arch. Microbiol. 130, 325-327 (1981) Fowler, V. f., Ludwig, W., Stackebrandt, E.: Ribosomal ribonucleic acid cataloging in bacterial systematics: the phylogeny of Actinomadura. In: M. Goodfellow, D. E. Minnikin (eds.) The use of chemotaxonomic methods for bacteria. New York Academic Press (in press) Fox, G.E., Pechman, K.]., Woese, C.R.: Comparative cataloging of 16S ribosomal ribonucleic acid: molecular approach to procaryotic systematics. Int. J. system. Bact. 27, 44-57 (1977) Fox, G.E., Stackebrandt, E., Hespell, R.B., Gibson, f., Maniloff, f., Dyer, T.A., Wolfe, R.S., Balch, W.E., Tanner, R., Magrum, L.]., Zablen, L.B., Blakemore, R., Gupta, R., Bonen, L., Lewis, B.]., Stahl, D. A., Luehrsen, K. R., Chen, K. N., Woese, C. R.: The phylogeny of prokaryotes. Science 209, 457-463 (1980) Gibson, ]., Stackebrandt, E., Zablen, L. B., Gupta, R., Woese, C. R.: A phylogenetic analysis of the purple photosynthetic bacteris. Curr. Microbiol. 3, 59-64 (1979) Ludwig, W., Eberspacher, ]., Lingens, F., Stackebrandt, E.: 16S ribosomal RNA studies on the relationship of a chloridazon-degrading Gram negative eubacterium. System. Appl. Microbiol. 5, 241-246 (1984) Madigan, M. T., Cox, S. S.: Nitrogen metabolism in Rhodospeudomonas globiformis. Arch. Microbiol. 133, 6-10 (1982) Madigan, M. T., Cox, S. S., Stegeman, R. A.: Nitrogen fixation and nitrogenase activities in members of the family Rhodospirillaceae. J. Bact. 157, 73-78 (1984) Marrs, B., Kaplan, S.: 23S precursor ribosomal RNA of Rhodopseudomonas sphaeroides. J. molec. BioI. 49, 293-317 (1970) Pfennig, N.: Rhodopseudomonas acidophila sp. n. A new species of the budding nonsulfur bacteria. J. Bact. 99, 597-602 (1969) Seewaldt, E., Schleifer, K. H., Bock, E., Stackebrandt, E.: The close phylogenetic relationship of Nitrobacter and Rhodopseudomonas palustris. Arch. Microbiol. 131, 287-290 (1982) Shiba, T., Simidu, U.: Erythrobacteri longus gen. nov. sp. nov. an aerobic bacterium which contains bacteriochlorophyll a. Int. J. system. Bact. 32, 211-217 (1982) Stackebrandt, E., Woese, C. R.: The evolution of prokaryotes. In: Molecular and Cellular Aspects of Microbial Evolution. Eds. M.]. Carlile, ].R. Collins, B. E. B. Moseley, pp. 1-31. Cambridge University Press 1981 Stackebrandt, E., Seewaldt, E., Ludwig, W., Schleifer, K.-H., Huser, B. A.: The phylogenetic position of Methanothrix soehngenii Elucidated by a modified technique of sequencing oligonucleotides from 16S rRNA. Zbl. Bakt. Hyg. I. Abr. Orig. C 3 90-100 (1982) Schwartz, R. M., Dayhoff, M. 0.: Origins of prokaryotes, eukaryotes, mitochondria and chloroplasts. Science 199, 395-403 (1978) Triiper, H. G., Pfennig, N.: Characterization and identification of the anoxygenic phototrophic bacteria. In; The Prokaryotes eds. M. P. Starr, H. Stolp, H. G. Triiper, A. Balows, and H. G. Schlegel, pp. 299-312. Berlin - Heidelberg - New York Springer-Verlag 1981 Uchida, T., Bonen, L., Schaup, H. W., Lewis, B.f., Zablen, L.B., Woese, C.R.: The use of ribonuclease U2 in RNA sequence determination. Some corrections in the catalog of oligomers produced by ribonuclease T1 digestion of Escherichia coli 16S ribosomal RNA. J. Molec. Evol. 3, 63-77 (1974)
326
CR. Woese et a!.
Watson, S.: Taxonomic considerations of the family Nitrobacteraceae Buchanan. Int. J. system. Bact. 21, 254-270 (1971) Woese, CR.: Molecular perspectives on the relationship between tempo and mode in evolution. Science submitted (1984) Woese, C.R., Sogin, M., Stahl, D.A., Lewis, B.]., Bonen, L.: A comparison of the 165 ribosomal RNAs from mesophilic and thermophilic bacilli: Some modifications in the Sanger method for RNA sequencing. J. Molec. Evo!. 7, 197-213 (1976) Woese, CR., Fox, G.E.: The phylogenetic structure of the procaryotic domain: The primary kingdoms. Proc. nat. Acad. Sci. (Wash.) 74,5088-5090 (1977) Woese, CR., Maniloff, ]., Zablen, L. B.: Phylogenetic analysis of the mycoplasmas. Proc. nat. Acad. Sci. (Wash.) 77 494-498 (1980a) Woese, CR., Gibson, ]., Fox, G.E.: Do genealogical patterns in purple photosynthetic bacteria reflect interspecific gene transfer? Nature 283, 212-214 (1980 b) Woese, CR., Blanz, P., Hespell, R.B., Hahn, C.M.: Phylogenetic relationships among various helical bacteria. Curro Microbio!. 7, 127-132 (1982) Woese, C. R., Gutell, R.R., Gupta, R., Noller, H.R.: A detailed analysis of the higher order structure of 16S-like ribosomal RNAs. Microbiol. Rev. 47,621-669 (1983) Woese, CR., Blanz, P., Hahn, C.M.: What isn't a pseudomonad: the importance of nomenclature in bacterial classification. System. App!. Microbiol., 5, 179-195 (1984) Dr. Carl R. Woese, Dept. of Genetics and Development, University of Illinois, 515 Morrill Hall, 505 South Goodwin Avenue, Urbana, Illinois 61801, U. S. A.
2
The Phylogeny of Purple Bacteria: The Alpha Subdivision C. R. WOESE1':', E. STACKEBRANDT3, W. G. WEISBURG!, B. J. PASTER1,2, M. T. MADIGAN 4 , V. J. FOWLER 3, C. M. HAHN!, P. BLANZ!, R. GUPTA!, K. H. NEALSON 5 , and G. E. FOX 6
Received April 24. 1984
Summary The technique of oligonucletide cataloging shows the purple photosynthetic eubacteria to comprise three major subdivisions, temporarily called alpha, beta, and gamma - previously designated groups I-Ill by Gibson et al. (1979). Each subdivision contains a number of non-photosynthetic genera in addition to the photosynthetic ones. The alpha subdivision, the subject 'of the present report, contains most but not all of the species that fall into the classically defined genera Rhodospirillum, Rhodopseudomonas and Rhodomicrobium. Intermingled with these are a variety of non-photosynthetic species from genera such as Agrobacterium, Rhizobium, Azospirillum, Nitrobacter, Erythrobacter, Phenylobacterium, Aquaspirillum, and Paracoccus. The phylogenetic substructure of the alpha subdivision is presented and the evolutionary significance of the admixture of biochemical phenotypes is discussed.
Key words: Alpha purple bacteria - Evolution - Phylogeny - rRNA catalog
Introduction Partial sequencing of ribosomal RNAs (by the oligonucleotide cataloging method - Fox et aI., 1977) has let us see for the first time the panorama of microbial evolu-
tion. Two vastly different classes of bacteria inhabit the microbial world, the true bacteria (or eubacteria) and the archaebacteria - each no more like the other than it is like the eucaryotes (Woese and Fox, 1977). In a real sense the two are alike " To whom requests for reprints should be sent.
316
C. R. Woese et aI.
in name only; they are that distinct at the molecular level. The eubacteria are the more thoroughly characterized group - with over 300 species cataloged to date. These catalogs permit us to define most if not all the major phylogenetic groups of eubacteria, of which there are approximately ten (Woese et aI., in preparation). Each such group is equivalent phylogenetically to a eucaryotic Phylum if not a Kingdom, and, interestingly, they share with eucaryotic phyla what seems to be a basic evolutionary characteristic - i. e. all appear to branch almost simultaneously from their common ancestor point. Similarly, as with the Classes of animals, the main subdivisions within each bacterial "phylum" again appear to spring in close succession (if not coincidently) from the common ancestor of the group. For this apparent reason it has not yet been possible to ascertain the exact order of branching either between the bacterial "phyla" or among major subdivisions within any "phylum" (Fox et aI., 1980). As the phylogeny of the eubacteria unfolds, it becomes increasingly apparent that our prejudices as to the evolutionary relationships among bacteria (implied by the present system of classification) are for the most part wrong. Perhaps no bacterial group exemplifies the prejudices and problems of the existing system of bacterial classification more than do the purple photosynthetic bacteria. These organisms have played a major role in the history of microbiology, and their consequent revered status may contribute to their being singled out for high taxonomic distinction (Truper and Pfennig, 1981). In any case, their classically defined taxonomy is now known to be in considerable disagreement with their phylogenetic relationships, as defined by the genotypic criterion of rRNA sequence comparisons. One major discrepancy is the failure of the classically defined categories to reflect the close relationships among various of the purple photosynthetic bacteria and certain non-photosynthetic species; for example, the specific relationships between Rhodopseudomonas sphaeroides and Paracoccus denitri{icans (Gibson et aI., 1979), between R. palustris and the rhizobacteria, and between Rhodospirillum rubrum and Azospirillum brasilense (Woese et aI., 1982). While the classical split between non-sulfur (prefix Rhodo-) and sulfur purple bacteria, which defines the two major subdivisions of the purple bacteria, is phylogenetically valid in one sense, it is not in another. Within the purple non-sulfur group itself there occurs a major phylogenetic split no less profound than the classically defined split
between the sulfur and non-sulfur types (Gibson et aI., 1979). Although the newly recognized division was defined in terms of ribosomal RNA sequence homologies, it can also be rationalized in terms of basic properties such as cytochrome c subunit type (Dickerson, 1980) and the structure of the photo-reaction center (Clayton and Clayton, 1978). In the present communication we will define the purple photosynthetic bacterial group in terms of its oligonucleotide signature (Woese et aI., 1980a), and explore the phylogenetic structure of one of its (three) major subdivisions, alpha, which contains all but a few of the purple non-sulfur bacteria - such as Rps. gelatinosa and R. tenue (Gibson et aI., 1979). Materials and Methods The organisms used in this study are these: Rhodospirillum rubrum ATCC 11170, Rhodopseudomonas palustris ATCC 11168, Rps. capsulata strain St. Louis, Rps. viridis
The Phylogeny of Purple Bacteria: The Alpha Subdivision
317
strain NTH, Rps. sphaeroides strain NCIB 8253, and Rhodomicrobium vannielii, all reported in Gibson et ai. (1979); Aquaspirillum itersonii ATCC 12639, Azospirillum brasilense ATCC 29145, as reported in Woese et ai. (1982); Pseudomonas diminuta ATCC 11508, and Rhizobium leguminosarum ATCC 10004 (Woese et aI., 1984); Nitrobacter winogradskyi (Seewaldt et aI., 1982); Phenylobacterium immobile DSM 1986 (Ludwig et aI., 1984). [A second species of Nitrobacter, N. hamburgensis (Bock et aI., 1983) is sufficiently similar in its rRNA catalog to N. winogradskyi that the former will be used throughout to represent both.] Not reported previously are R. photometricum ATCC 27871, Rps. globiformis ATCC 7950, R. molischianum ATCC 14031, Rps. acidophila ATCC 7050, Paracoccus denitrificans ATCC 17741, Agrobacterium tumefacians DSM 30150, R. salexigens DSM 2132 (Drews, 1981), Erythrobacter longus IFO 14126 (Shiba and Simidu, 1982) (a gift of Prof. Shiba), and the manganese oxidizing isolates #36 and #63 (K.H.N. unpublished) . Growth and labelling of the organisms was as described in the appropriate references cited above. Special growth conditions for various of the species are described in Madigan and Cox (1982), Madigan et ai. (1984), Drews (1981) and Pfennig (1969). Extraction of 32P labelled (or unlabelled) 16S rRNA and generation of the oligonucleotide catalogs followed either of two standard procedures (Uchida et aI., 1974; Woese et aI., 1976; Stackebrandt et aI., 1982; Fowler et aI., 1984). The numeric analysis of data, in terms of binary association coefficients, or Sab values, is described in Fox et ai. (1977). Results and Discussion The cataloging method for 165 ribosomal RNA is generally very reliable for determining the phylogenetic relationships among bacteria. It appears to be more reliable than the methods based upon full sequencing of 55 ribosomal RNA, cytochrome c, and so on. The main reason for this lies in the fact that the latter molecules are small, while the former is large, i.e. the latter comprise a small number of functionally defined domains, while the former comprises perhaps an order of magnitude more such domains, which for the 165 rRNA means perhaps fifty (Woese et aI., 1983). Rarely, one such "domain" becomes redesigned evolutionarily, which results in a number of changes in sequence that are not selectively neutral. These bursts of sequence change distort the molecular distance measure when the extent of the perturbation is significant in terms of the whole sequence - which is the case for the small molecules, but is not for the large ones. It has been customary to analyse rRNA catalog data in terms of binary association coefficients, or 5ab values (Fox et aI., 1977), and from a table of these to construct a dendrogram by the unweighted pair group method (Fox et aI., 1977). This approach is not the optimal analysis, for it does not make full use of the data. The method is reliable when all of the species considered represent lines of descent that evolve at the same rate. Catalogs representing rapidly evolving lines of descent are poorly ordered phylogenetically by this approach (Woese et aI., 1980). [Fortunately, most bacterial lines appear to be evolving at about the same rate (Stackebrandt and Woese, 1981).] The use of binary association coefficients also does not take advantage of the fact that different positions in the (165) rRNA sequence have different phylogenetic significance. [Positions can differ by well over an order of magnitude in the frequency with which they change over evolutionary time. And, some positions exhibit characteristic, recognizable patterns of change. Also some positions appear to change during the formation of major phylogenetic groups, but remain constant once lines
318
C.R.Woeseetal.
Table 1. Sab values (Fox et a!', 1977) for various purple bacteria with one another and with outgroup organisms, showing the coherence of the purple group and its major subdivisions. Numbers correspond to the following organisms: 1 - R. rubrum, 2 - R. molischianum, 3 - Rps. capsulata, 4 - Rps. palustris, 5 - Rh. leguminosarum, 6 - Rm. vannietii,
7 - E. longus, 8 - Rps. gelatinosa, 9 - Ps. acidovorans, 10 - Thiobacillus intermedius, 11 R. tenue, 12 - Aquaspirillum serpens, 12 - Ps. cepacia, 14 - Spirillum volutans, 15 - Escherichia coli, 16 - Ps. fluorescens, 17 - Oceanospirillum maris, 18 - Beggiatoa leptomitiformis, 19 - Chromatium vinosum, 20 - Ectothiorhodospira strain 7, 21 - Legionella pneumophilia, 22 - Bacillus subtilus, 23 - Aphanocapsa str 6714, 24 - Flavobacterium heparinum. (Fox et a!', 1980; Gibson et a!', 1979; Woese et a!', 1982; Woese et a!., 1984; and unpub-
lished catalogs)
alpha group 1 234 567 1 R rub
beta group 8 9 10 11 12 13 14
2 3 4 5 6 7
R mol Reap R pal Rh Ie R vir E Ion
5640 43 36 44 47 55 41 48 41 42
-43 53 40 43
-61 -50 56 -40 48 37 --
8 9 10 11 12 13 14
R gel Ps ac
28 28 32 30 26 29 24
33 35 36 34 31 31 29
32 32 31 31 30 29 28
26 29 28 26 31 27 25
32 32 34 31 32 30 28
32 30 34 29 31 28 26
33 34 35 33 33 30 31
68 64 60 51 57 49
-52 53 52 50 45
-58 51 51 41
59 -61 54 -53 51 54 --
34 34 34 26 41 35 32
33 34 31 30 41 35 34
34 30 29 24 34 30 36
28 34 26 27 38 26 33
31 34 28 29 36 31 37
28 31 26 28 35 32 35
33 36 28 26 37 36 36
32 35 33 30 36 39 34
33 36 28 30 34 33 32
29 33 28 30 37 36 33
32 32 24 26 37 31 33
15 16 17 18 19 20 21
Th in
R ten Aq se Ps ce Sp vo
E col Ps f1 Oc ma Be Ie
Cr vi Ec 7
Le pn
22 B sub 23 Aph 24 Fl he
33 32 27 28 33 32 31 30 25 28 20 22 27 26 26 26 26 24 24 24 25
31 35 27 28 34 31 33
31 37 30 31 31 35 33
29 31 26 26 31 30 32
29 31 29 27 31 27 24 29 25 21 19 23 22 20 28 28 23 19 22 18 17
galllDa group 15 16 17 18 19 20 21
44 44 32 45 36 38
-49 33 42 41 42
30 38 38 40
other 22 23
-31 -37 48 35 41 41 --
26 32 25 20 32 32 25 28 30 27 17 30 27 23 27 21 26 20 29 24 25
29 -22 21
of descent have become established (Woese, 1984).] By developing families of the oligonucleotides that cover (some of) the phylogenetically more significant positions in the molecule, one can refine considerably and extend the customary analysis, based upon Sab values (Fox et aI., 1977). Table 1 is a set of Sab values that distinguishes the greater group of purple bacteria and their relatives from the other major eubacterial groups and defines the main subgroups of the former. It is not our intention to provide formal names for the main group or its subgroups (though this naming will become necessary), so the tentative designations "purple bacteria" or "purple group" will be used when referring to the entire collection, and the prefixes alpha, beta, and gamma added when referring to the three main subdivisions; within each of these the suffixes - 1, - 2, etc. will be used to denote further substructure. Table 2 defines the purple bacteria in terms of an oligonucleotide signature (Woese et aI., 1980a). The sequences shown are found in many to most catalogs
The Phylogeny of Purple Bacteria: The Alpha Subdivision
319
Table 2. Oligonucleotide signature distinguishing the purple group from other major bacterial groups. For each group the number of organisms included is shown below the column designation. The "all" column refers to the sum of the three purple bacterial subdivisions and the numbers in this column are fraction of total rather than number of organisms, as in the other columns. The "other occurrences" column lists the total number of occurrences of the sequence in question among eubacteria outside of the purple group. Bases in parenthesis - e.g. (A, Y) - are alternatives to one another. Abbreviations: Y-pyrimidine, R-purine; G + - Gram positive group; CFB - cytophaga-flavobacter-bacteroides group; Myx - myxobacteria group; Gr - green photosynthetic bacteria. sequence
alpha (23)
beta (27)
ganma (53)
25 27 4 27 16 22 0 25 12
31 32 25 47 16 2 41 19 13
other occurrences (240)
all ( 103)
-----------------------------------------------------------------AUCCAG ACAAUG AnUUCG UUAAUCG CCCCCUG CCAACCCG CCCUUACG AAAUye Y( A. Y)G CUYACCAC(G,U)UUG
13 19 2 0 0 1
17 9
0
10 [6G+; 4CFB] 15 [9G+; 4CFB] 1 1 0 8 [5G+; 3Hyx] 20 [ 14CFB; 5Gr] 3 4
0.64 0.72 0.29 0.68 0.31 0.24 0.56 0.51 0.24
of organisms within the purple group but rarely if at all in catalogs representing organisms outside the group. [Oligonucleotides that are found in only one of the three subdivisions of the purple group are not included in this signature.] Table 3 distinguishes the alpha purple bacterial subdivision from the other two subdivisions. Note that by signature as well as by the Sab measure, the two groups of purple non-sulfur bacteria are quite distinct from one another. The first of these groups, alpha, concerns us in detail here; the other groups and isolated individual members of the greater purple group are the subjects of subsequent publications. Table 3. Oligonucleotide signature distinguishing the alpha subdivision of purple bacteria from the beta and gamma subdivisions. "Other" refers to occurrences among eubacteria outside of the purple group. Abbreviations: A-1, -2, or -3 refer to the three main subgroups of the alpha subdivision; £1 - Erythrobacter longus; Band G refer to the beta and gamma subdivisions of the purple bacteria; other abbreviations as in Table 2 caption. sequence
A-1 (7)
A-2 (10)
A-3 (5)
El (1)
B (27)
G (53)
0 2 1 0 0 0 0 0 0 2 0
0 4 0 0 2 0 0 1 0 0
other (240)
-------------------------------------------------------------AUAAUG CCUUUG CUAACCG ceCUUCG CUCACUG AAAUUCG AUAUUCG AUCCAUAG AUUUAUCG AAAUCCCAG CAACCC( A, U) CR ••
0 2 5 2 0 7 1 2 3 1 6
9 10
5 2 7 10 9
4
10 3 9
5 3 0 3 5
5
5 3
4 4
5
+ + 0 0 0 + + 0 + 0
1
24 18 11 27 5 12 4 0 18 0 many
The alpha subdivision can be substructured by signature into three clusters plus an isolated species, E.longus, which represents a fourth cluster; see Table 4. [A meaningful signature cannot be defined for this fourth cluster on the basis of one
320
C. R. Woese et al.
Table 4. Oligonucleotide signature substructuring the alpha subdivision of purple bacteria. The abbreviations for species names are as follows: R. rubrum - Rr, R. photometricum Rp, Aq. itersonii - Ai, Az. brasilense - Ab, Rps. globiformis - Rg, Rm. molischianum Rrn, R. salexigens - Rs, Rps. acidophila - Ra, Rps. viridis - Rv, Rm. vannielii - Rn, Rps. palustris - Rp, N. winogradskyi - Nw, Rh.leguminosarum - RI, Ag. tumifaciens - At, Ps. diminuta - Pd, Ph. immobile - Pi, Rps. sphaeroides - Ro, Rps. capsulata - Rc, P. denitrificans - Pn, the two manganese oxidizing isolates - 36 and 63, and E. longus - El. "+" indicates the presence, "." the absence of an oligonucleotide.
sequence
R R A A R R R r p i b g ms
R R R R N RAP P a v n p w1 t d i
RRP 3 6 o c n 6 3
E 1
----------------------------------------------------------------? ACAAG AU AU UG CACUCCG UAACACG AAUUCCCAG UCAcACCAUG
UACCCG CAAUACG AACUCCG AAACUUG AAUCUUG CCACAUUG ACCCUUCG AUUUACUG CCCUUACAG UUUUACCCG CUACCAUUUAG CCAUCAUUUAG CCCAAACYCCYACG AUCUAG AUUAAG UAAUACCG AAUCUUAG ACACACCUAACG CAACCCACA ••• UUAAUCCC •••• AACCUUACCAACCCUUG
+ + + + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + + +
+
+
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+
+ +
+ + +
+
+ + + + + + + + + + + + + +
+ + + + +
+ + + + + + + + + + +~ + +
+
+ + +
+ + + + +
species alone. ] Table 5 shows the internal structure in the alpha subdivision as defined by 5ab values, and Table 6 provides additional phylogenetic fine structure as defined by signature. The groupings defined on the basis of 165 rRNA catalogs, a genotypic measure, are readily rationalized in terms of phenotypic properties as well. The first subgroup, alpha-I, comprising R. rubrum, R. photometricum, Rps. globiformis, R. molischianum, R. salexigens, Aq. itersonii, and Az. brasilense, is notable for the spiral shape of most of its representatives. Two types of photosynthetic membranes are characteristic of the group, the stacked configuration and the vesicular configuration (Truper and Pfennig, 1981) (the latter also being found among members of alpha-3 - see below).
The Phylogeny of Purple Bacteria: The Alpha Subdivision
321
Table 5. Set of Sab values (Fox et aL, 1977) substructuring the alpha subdivision of the purple bacteria. Organism name abbreviations are as in Table 4. alpha-l
1 234
567
alpha-2
8 9 10 11 12 13 14 15 16
1 Rr 2 Rp 3 Ai 4 Ab 5Rm 6 Rs 7 Rg
74 54 51 56 48 44
-49 51 49 45 45
-47 52 47 -46 51 40 -48 50 48 47 --
8 Nw 9 Rp 10 Rl 11 At 12 Rv l3Rn 14 Pd 15 Pi 16 Ra
35 36 47 46 50 41 43 40 37
34 38 42 39 48 39 39 41 36
44 46 53 48 52 46 48 48 47
34 36 40 39 47 39 42 38 43
42 44 55 51 55 48 40 43 44
37 36 43 40 46 45 41 38 37
38 41 47 42 48 42 41 36 40
7760 61 50 51 52 54 44 50 40 45 41 41 53 56
-78 59 56 48 52 57
57 51 43 44 47
-58 51 50 54
-43 -48 54 -43 49 44 --
17 18 19 20 21
40 41 37 41 38
36 37 35 38 38
44 46 44 42 39
45 43 43 43 43
43 41 40 41 37
40 44 43 41 38
34 36 34 33 32
41 44 42 36 41
53 54 51 47 52
50 48 46 43 46
45 47 45 47 46
40 41 42 36 39
Rc Ro Pn 36 63
22 El
41 40 46 39 42 41 44
43 42 44 37 43
45 51 44 46 46
49 40 45 40 43
alpha-3
17 18 19 20 21
42 47 43 40 42
7171 67 61 66 58 61 63 61 75 --
38 40 48 46 47 37 41 39 44
43 45 42 39 44
22
Members of the alpha-2 subgroup, represented by Rhodopseudomonas palustris, Rps. viridis, Rps. acidophila, Rm. vannielii, Ps. diminuta, Rh.leguminosarum, and others, have two gross morphological features in common. They generally divide in an asymmetric fashion (or actually bud), and they have intracytoplasmic membranes that are adjacent and parallel to the cytoplasmic membrane (Truper and Pfennig, 1981). These properties pertain, at least, to the photosynthetic members and to the two N itrobacter species (Watson, 1971; Seewaldt et aL, 1982). This subgroup contains organisms of interest because of their intimate intracellular association with plant cells - i.e. the rhizobacteria and agrobacteria. For those members of the subgroup that have so far been characterized, the sugar 2,3-diamino2,3-dideoxy D-glucose has been found in the lipid A moiety of the lipopolysaccharide (see Ludwig et aL, 1984). There is an indication from cytochrome c sequence comparisons that most mitochondria may have come from this particular subgroup as well (Swartz and Dayhoff, 1978). The alpha-3 subgroup, is represented by Rhodopseudomonas sphaeroides, Rps. capsulata, Paracoccus denitrificans and the two manganese oxidizing strains of Nealson. All share the unusual property of having 235 rRNAs that are post-transcriptionally cleaved into two nearly equal halves (Marrs and Kaplan, 1970; Gibson et aL, 1979; Woese et aL, unpublished observations). [We have not encountered this property in any other bacteria in the cataloging study except for Flavobacterium capsulatum, an organism whose catalog is incomplete, but which can nevertheless be placed in the alpha subdivision, though not within any of its three subgroups, by signature analysis (Blanz and Weisburg, unpublished).] Within this subgroup the two photosynthetic strains and P. denitrificans form a cluster distinct from the cluster formed by the two manganese oxidizing isolates; see Tables 5 and 6.
322
C. R. Woese et al.
Table 6. Oligonucleotide signatures supporting some specific subclusters within the various subgroups of the alpha purple bacteria. The sequences shown are found within the alpha subdivision only in the species under which they are listed. The number of their occurrences in the beta and gamma subdivisions, and elsewhere, are listed in the appropriate columns. sequence
beta
A. !. rubrum-!.photOllletricum (alpha-I) 0 CCCAAG I AAUCCG 0 UACAUG 0 CCACCUG 0 UACCUCG 0 UACCUUG CAUCCUUCG 0 ACUAAUACCG 0 B. Rhizobium-Agrobacterium (alpha-2) 0 UAUCCG 0 UAUACUG 1 CUAACCACG 0 UCCAUUACUG 0 CUCUUUCACCG
ga1llDa
other
2 0 0
10
5
0 0 0 0 0 0 0 0 0
C• .!£!. palustris-!. winogradskyi (alpha-2) 0 CUCUUUUG 0 0 0 CAACCCCCG 0 0 UCCAACUUG 0 UACCUUUUG 0 0 UUUACUCACUAG 0 0 0 ACCUUCUCUUCG diminuta-Ph. immobile (alpha-2) 0 UCUUAG CCAUUAG 8 (A,C)CACUCUAAUG 13 UAAAAU(A,U)CUUUCACCG 0 AACCUUACCACCUUUUG 0
D. ~.
2 0 I 0
0
11
I
8
2
14
0 2
15
0 0 0 0 0 0 0 0 0 0
8
3 2 0 0
E• .!£!. sphaeroides-.!£!. capaulata-f!. denitrificans (alpha-3) UAAUACCG 0 0 8 UUCAACUUG 0 0 0
F.
Mn-ox #36-Mn-ox #63 (alpha-3) 0 UUUUAG AAACCCCG 0 CUAACCUUCG 0 0 CAACCCACAUCCUUAG 0 UUAAUCCC(C,A)MAAACUG 0 AAACCCCG
0
0 0 0
0 0
3 1
4
0 0 I
It is of interest that most of the phenotypic properties of this group of organisms do not fall strictly along the phylogenetically defined lines. Photosynthetic capacity itself is, of course, the most obvious one. Throughout the alpha subdivision photosynthetic and non-photosynthetic phenotypes are completely intermixed. There is little doubt that the non-photosynthetic members have arisen from photosynthetic ancestors rather than the reverse. E. longus, which contains bacteriochlorophyll a, but is an aerobe (Shiba and Simidu, 1982), could represent a transition phase. Cytochrome c subunit type correlates only to a first approximation with these
The Phylogeny of Purple Bacteria: The Alpha Subdivision
323
Sob
)
Ps.
A. diminula lumefac/(ms
Ph. immobile
Rh. lequminasarum
R
N. It'/noqradskyi
Rps paluslris
molischianum Rps. viridis Rm vannlelii
B
7
6
5
4
y -group
13 - group
3 Fig. 1. Schematiclphylogenetic tree representing the relationships shown by Sab analysis as modified by signature analysis. The order of branching is that revealed by the combined analysis. The branching points are drawn in approximate agreement with the Sab analysis.
groupings (Woese et aI., 1980b). Those cytochromes c from the alpha-3 subgroup which have been studied, are all of the large subunit type. However, within the other subgroups both large and medium sized cytochrome c types are encountered (Dickerson, 1980). Furthermore, the subgroups as defined by rRNA catalog comparisons are in imperfect agreement with the groupings defined by cytochrome c sequence comparisons (Dickerson, 1980). Both approaches define the same alpha-3 subgroup, as well as the Rm. vannielii-Rps. viridis-Rps. acidophila subcluster of alpha-2, and the R. rubrum-R. photometricum subcluster in alpha-I. And Rps. palustris, despite the size difference of its cytochrome c, does fall with its relatives in alpha-2 by that measure. However, a clustering of R. molischianum or Rps. globiformis with the other members of alpha-1 (or with one another) does not occur by cytochrome c analysis; both cluster peripherally with the alpha-2 subgroup. Some, if not all of the lack of agreement here is due to the fact that for the deep phylogenetic relationships represented by the bacteria the cytochrome c molecule
324
C.R. Woese et al.
ceases to be a good molecular chronometer. Sequence comparisons between cytochromes c of different size undoubtedly yield less resemblance than would comparisons between cytochromes c of the same size, all other factors being equal. If one of the two size classes has evolved more than once from the other (which would seem to be the situation here), then further non-chronometric changes (either convergences or divergences) can have been introduced. For these reasons we do not consider the cytochrome c molecule to be as reliable a chronometer for this group of organisms as is the ribosomal RNA. There seems reasonable agreement between the cytoplasmic membrane types and the present genotypic grouping - though again, imperfect (Truper and Pfennig, 1981). The organisms in alpha-2, as stated, exhibit a "parallel" membrane orientation; the membranes for alpha-3 organisms are of the vesicular type, as are those of R. rubrum and Rps. globiformis, from group alpha-l, but those of R. photometricum and R. molischianum (also in alpha-i) exhibit a stacked configuration. [R. rubrum and R. photometricum are specific relatives of one another by both the rRNA and the cytochrome c criteria.] Neither the isoprenoid composition (Collins and fones, 1981) nor the predominant carotenoids (Truper and Pfennig, 1981) can be used to distinguish phylogenetically among the subgroups of the alpha subdivision. What these various disagreements point out is not that the rRNA genotypic measure of relatedness is questionable, for it is certainly the most consistent measure yet developed for bacterial phylogenetic classification, and it yields a more consistent picture of phylogeny of the present group than does any other criterion or combination thereof. What is emerging from these comparisons is a picture of a bacterial evolution that is highly varied in biochemical and morphological aspects (as seen in other major groups as well- Stackebrandt and Woese, 1981). Photosynthetic membrane types can change (compare R. rubrum and its close relative R. photometricum); cytochrome c size can change (compare the Rps. palustris and Rps. viridis examples, which differ in size but retain related sequences); apparent mode of growth and division can change, and so on. While it is fashionable to invoke inter-species gene transfer to explain these sorts of phenomena (Ambler et aI., 1979), we would be loathe to do so without far better evidence. Rather, the group under consideration is an enormous one from an evolutionary perspective; it would seem to be broader, older than all eucaryotes taken together - as judged by cytochrome c trees, for example. From this perspective, remarkable shifts in various phenotypic characteristics become less remarkable. Acknowledgements. CRW was supported by a grant from the National Science Foundation; ES was financed by the Gesellschaft fur Biotechnologische Forschung to support the German Collection of Microorganisms.
References Ambler, R.P., Daniel, M., Hermosa, j., Meger, T.E., Bartsch, R.C., Kamen, M.D.: Cyto-
chrome c2 sequence variation among the recognised species of purple nonsulphur photosynthetic bacteria. Nature 278, 659-660 (1979) Bock, E., Sundermeyer-Klinger, H., Stackebrandt, E.: Characterization of a novel facultative lithoautotrophic nitrite-oxidizing bacterium. Arch. Microbiol. 136,281-284 (1983)
The Phylogeny of Purple Bacteria: The Alpha Subdivision
325
Clayton, R. f., Clayton, R. K.: Properties of photochemical reaction centers purified from Rhodopseudomonas gelatinosa. Biochim. Biophys. Acta 501, 470-477 (1978) Collins, M. D., fones, D.: Distribution of isoprenoid quinone structural types in bacteria and their taxonomic implications. Microbiol. Rev. 45,316-354 (1981) Dickerson, R. E.: Evolution and gene transfer in purple photosynthetic bacteria. Nature 283, 210-212 (1980) Drews, G.: Rhodospirillum salexigens, sp. nov. an obligatory halophilic phototrophic bacterium. Arch. Microbiol. 130, 325-327 (1981) Fowler, V. f., Ludwig, W., Stackebrandt, E.: Ribosomal ribonucleic acid cataloging in bacterial systematics: the phylogeny of Actinomadura. In: M. Goodfellow, D. E. Minnikin (eds.) The use of chemotaxonomic methods for bacteria. New York Academic Press (in press) Fox, G.E., Pechman, K.]., Woese, C.R.: Comparative cataloging of 16S ribosomal ribonucleic acid: molecular approach to procaryotic systematics. Int. J. system. Bact. 27, 44-57 (1977) Fox, G.E., Stackebrandt, E., Hespell, R.B., Gibson, f., Maniloff, f., Dyer, T.A., Wolfe, R.S., Balch, W.E., Tanner, R., Magrum, L.]., Zablen, L.B., Blakemore, R., Gupta, R., Bonen, L., Lewis, B.]., Stahl, D. A., Luehrsen, K. R., Chen, K. N., Woese, C. R.: The phylogeny of prokaryotes. Science 209, 457-463 (1980) Gibson, ]., Stackebrandt, E., Zablen, L. B., Gupta, R., Woese, C. R.: A phylogenetic analysis of the purple photosynthetic bacteris. Curr. Microbiol. 3, 59-64 (1979) Ludwig, W., Eberspacher, ]., Lingens, F., Stackebrandt, E.: 16S ribosomal RNA studies on the relationship of a chloridazon-degrading Gram negative eubacterium. System. Appl. Microbiol. 5, 241-246 (1984) Madigan, M. T., Cox, S. S.: Nitrogen metabolism in Rhodospeudomonas globiformis. Arch. Microbiol. 133, 6-10 (1982) Madigan, M. T., Cox, S. S., Stegeman, R. A.: Nitrogen fixation and nitrogenase activities in members of the family Rhodospirillaceae. J. Bact. 157, 73-78 (1984) Marrs, B., Kaplan, S.: 23S precursor ribosomal RNA of Rhodopseudomonas sphaeroides. J. molec. BioI. 49, 293-317 (1970) Pfennig, N.: Rhodopseudomonas acidophila sp. n. A new species of the budding nonsulfur bacteria. J. Bact. 99, 597-602 (1969) Seewaldt, E., Schleifer, K. H., Bock, E., Stackebrandt, E.: The close phylogenetic relationship of Nitrobacter and Rhodopseudomonas palustris. Arch. Microbiol. 131, 287-290 (1982) Shiba, T., Simidu, U.: Erythrobacteri longus gen. nov. sp. nov. an aerobic bacterium which contains bacteriochlorophyll a. Int. J. system. Bact. 32, 211-217 (1982) Stackebrandt, E., Woese, C. R.: The evolution of prokaryotes. In: Molecular and Cellular Aspects of Microbial Evolution. Eds. M.]. Carlile, ].R. Collins, B. E. B. Moseley, pp. 1-31. Cambridge University Press 1981 Stackebrandt, E., Seewaldt, E., Ludwig, W., Schleifer, K.-H., Huser, B. A.: The phylogenetic position of Methanothrix soehngenii Elucidated by a modified technique of sequencing oligonucleotides from 16S rRNA. Zbl. Bakt. Hyg. I. Abr. Orig. C 3 90-100 (1982) Schwartz, R. M., Dayhoff, M. 0.: Origins of prokaryotes, eukaryotes, mitochondria and chloroplasts. Science 199, 395-403 (1978) Triiper, H. G., Pfennig, N.: Characterization and identification of the anoxygenic phototrophic bacteria. In; The Prokaryotes eds. M. P. Starr, H. Stolp, H. G. Triiper, A. Balows, and H. G. Schlegel, pp. 299-312. Berlin - Heidelberg - New York Springer-Verlag 1981 Uchida, T., Bonen, L., Schaup, H. W., Lewis, B.f., Zablen, L.B., Woese, C.R.: The use of ribonuclease U2 in RNA sequence determination. Some corrections in the catalog of oligomers produced by ribonuclease T1 digestion of Escherichia coli 16S ribosomal RNA. J. Molec. Evol. 3, 63-77 (1974)
326
CR. Woese et a!.
Watson, S.: Taxonomic considerations of the family Nitrobacteraceae Buchanan. Int. J. system. Bact. 21, 254-270 (1971) Woese, CR.: Molecular perspectives on the relationship between tempo and mode in evolution. Science submitted (1984) Woese, C.R., Sogin, M., Stahl, D.A., Lewis, B.]., Bonen, L.: A comparison of the 165 ribosomal RNAs from mesophilic and thermophilic bacilli: Some modifications in the Sanger method for RNA sequencing. J. Molec. Evo!. 7, 197-213 (1976) Woese, CR., Fox, G.E.: The phylogenetic structure of the procaryotic domain: The primary kingdoms. Proc. nat. Acad. Sci. (Wash.) 74,5088-5090 (1977) Woese, CR., Maniloff, ]., Zablen, L. B.: Phylogenetic analysis of the mycoplasmas. Proc. nat. Acad. Sci. (Wash.) 77 494-498 (1980a) Woese, CR., Gibson, ]., Fox, G.E.: Do genealogical patterns in purple photosynthetic bacteria reflect interspecific gene transfer? Nature 283, 212-214 (1980 b) Woese, CR., Blanz, P., Hespell, R.B., Hahn, C.M.: Phylogenetic relationships among various helical bacteria. Curro Microbio!. 7, 127-132 (1982) Woese, C. R., Gutell, R.R., Gupta, R., Noller, H.R.: A detailed analysis of the higher order structure of 16S-like ribosomal RNAs. Microbiol. Rev. 47,621-669 (1983) Woese, CR., Blanz, P., Hahn, C.M.: What isn't a pseudomonad: the importance of nomenclature in bacterial classification. System. App!. Microbiol., 5, 179-195 (1984) Dr. Carl R. Woese, Dept. of Genetics and Development, University of Illinois, 515 Morrill Hall, 505 South Goodwin Avenue, Urbana, Illinois 61801, U. S. A.