A genomic update on clostridial phylogeny: Gram-negative spore formers and other misplaced clostridia

Authors

  • Natalya Yutin,

    1. National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author
  • Michael Y. Galperin

    Corresponding author
    • National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author

For correspondence. E-mail galperin@ncbi.nlm.nih.gov; Tel. (+1) 301 435 5910; Fax (+1) 301 435 7793.

Summary

The class Clostridia in the phylum Firmicutes (formerly low-G+C Gram-positive bacteria) includes diverse bacteria of medical, environmental and biotechnological importance. The Selenomonas-Megasphaera-Sporomusa branch, which unifies members of the Firmicutes with Gram-negative-type cell envelopes, was recently moved from Clostridia to a separate class Negativicutes. However, draft genome sequences of the spore-forming members of the Negativicutes revealed typically clostridial sets of sporulation genes. To address this and other questions in clostridial phylogeny, we have compared a phylogenetic tree for a concatenated set of 50 widespread ribosomal proteins with the trees for beta subunits of the RNA polymerase (RpoB) and DNA gyrase (GyrB) and with the 16S rRNA-based phylogeny. The results obtained by these methods showed remarkable consistency, suggesting that they reflect the true evolutionary history of these bacteria. These data put the Selenomonas-Megasphaera-Sporomusa group back within the Clostridia. They also support placement of Clostridium difficile and its close relatives within the family Peptostreptococcaceae; we suggest resolving the long-standing naming conundrum by renaming it Peptoclostridium difficile. These data also indicate the existence of a group of cellulolytic clostridia that belong to the family Ruminococcaceae. As a tentative solution to resolve the current taxonomical problems, we propose assigning 78 validly described Clostridium species that clearly fall outside the family Clostridiaceae to six new genera: Peptoclostridium, Lachnoclostridium, Ruminiclostridium, Erysipelatoclostridium, Gottschalkia and Tyzzerella. This work reaffirms that 16S rRNA and ribosomal protein sequences are better indicators of evolutionary proximity than phenotypic traits, even such key ones as the structure of the cell envelope and Gram-staining pattern.

History of the Gram-negative members of the Firmicutes

Gram-negative, low-G+C Gram-positive bacteria may sound like an oxymoron. Nevertheless, the phylum Firmicutes (low-G+C Gram-positive bacteria) includes a number of organisms whose cells are surrounded by two membranes, which are separated by a relatively thin layer of peptidoglycan, and, accordingly, stain Gram-negative. Historically, a representative of this group may have been one of the first bacteria observed by Antonie van Leeuwenhoek in his own sputum back in 1683 (Kingsley and Hoeniger, 1973). Two hundred years later, it was named Spirillum sputigenum, and later renamed Selenomonas sputigena. While S. sputigena and the closely related Selenomonas ruminantium could be easily identified based on the characteristic curved rod cell shape and flagellar tuft protruding from the concave side, their taxonomic position long remained obscure, being assigned to spirilla, vibrios and even protozoa. With the advent of the 16S rRNA sequencing, selenomonads were shown to belong to a large group of bacteria, whose representatives had been initially assigned to various Gram-negative bacterial lineages, primarily to Bacteroidaceae (Dialister pneumosintes, Megamonas hypermegale, Pectinatus cerevisiiphilus, Propionispira arboris and others), but also to Neisseriaceae (Veillonella parvula, Acidaminococcus fermentans), or remained without a clear assignment (Quin's oval, later renamed Quinella ovalis). The overall phylogenetic position of Selenomonas-related bacteria remained unresolved until 1984, when scientists from the Deutsche Sammlung von Mikroorganismen (Braunschweig) and the University of Göttingen described a very unusual genus, Sporomusa. Its two members, Sporomusa sphaeroides and Sporomusa ovata, stained Gram-negative but were able to form typical round- or oval-shaped endospores that could survive heating to 80°C for 10 min (Möller et al., 1984), identifying them as legitimate members of the Firmicutes. This observation prompted a detailed analysis of their phylogenetic position using the 16S rRNA oligonucleotide catalogues and the assignment of the entire Selenomonas-Megasphaera-Sporomusa group to the ‘Clostridium-Bacillus cluster’ (Stackebrandt et al., 1985), i.e. to the current phylum Firmicutes. Their strictly anaerobic lifestyle, as well as 16S rRNA sequence data, supported assignment of these bacteria to the class Clostridia (Willems and Collins, 1995a,b).

In the 2nd edition of Bergey's Manual of Systematic Bacteriology (Ludwig et al., 2009), all 26 genera of the Selenomonas-Megasphaera-Sporomusa group were assigned to a single family Veillonellaceae within the class Clostridia. However, a year later, based on the Gram-negative type of cell wall and a revised 16S rRNA tree, Marchandin et al. (2010) suggested assigning this group to the new order Selenomonadales in the new class Negativicutes and dividing it into two families, Acidaminococcaceae and the emended Veillonellaceae. In addition, the family designation for many genera of the new order were unassigned, or Selenomonadales Incertae Sedis. This new taxonomy has been adopted by such key resources as the List of Prokaryotic Names with Standing in Nomenclature [LPSN, http://www.bacterio.net/ (Euzéby, 1997)], the Ribosomal Database Project [RDP, http://rdp.cme.msu.edu/ (Cole et al., 2009)] and the NCBI Taxonomy database [http://www.ncbi.nlm.nih.gov/taxonomy (Federhen, 2012)] and used for taxonomic assignments by the International Nucleotide Sequence Database Collaboration which includes GenBank, the European Nucleotide Archive and the DNA Data Bank of Japan (Nakamura et al., 2013). In contrast, SILVA rRNA gene database [http://www.arb-silva.de/ (Quast et al., 2013)] and Greengenes database [http://greengenes.lbl.gov/ (McDonald et al., 2012)] still list these organisms as members of the family Veillonellaceae in the order Clostridiales, class Clostridia.

In the past years, the Gram-negative cell walls of these bacteria have been characterized in some detail. The integrity of the cell envelope, with its relatively thin layer of peptidoglycan, has been attributed to the presence of cadaverine, which provides links between peptidoglycan and outer membrane proteins (Kojima et al., 2011). A recent detailed study of Acetonema longum membranes using electron cryo-tomographic imaging showed that in the course of sporulation, the inner membrane of the mother cell engulfs the prespore, becomes included into the growing spore and, upon its germination, forms the outer membrane of the daughter cell (Tocheva et al., 2011). This work put forward an intriguing hypothesis that the same process could have been responsible for the rise of the bacterial outer membrane and, more generally, for the origin of Gram-negative bacteria from Gram-positive ones, whereby the loss of sporulation in the relatives of Acetonema resulted in permanently Gram-negative bacteria, such as S. sputigena. While similar ideas on the evolutionary primacy of Gram-positive (single-membrane or ‘monoderm’) bacteria (and/or archaea) giving rise to the Gram-negative (two-membrane or ‘diderm’) bacteria have been proposed previously (Gupta, 2000; Sutcliffe, 2010), the work by Tocheva et al. (2011) provided the first experimental observation in support of that concept and proposed a realistic mechanism of how that could have happened.

Sporulation and phylogeny of the Negativicutes

The 16S rRNA tree constructed by Tocheva et al. (2011) put A. longum squarely within the clostridial clade, in accordance with the earlier data (Willems and Collins, 1995a,b; Ludwig et al., 2009). The tree built from the SpoIVA sequences had essentially the same topology (see figs 6 and S7 in Tocheva et al., 2011). As of 1 April, 2013, there were no complete genome sequences for any spore-forming members of the Negativicutes, which is why these bacteria were outside the scope of our recent comparative genomics analysis of the sporulation proteins in Bacilli and Clostridia (Galperin et al., 2012). However, draft genome sequences have recently become available for A. longum and for six different strains of Pelosinus fermentans, another spore-forming member of Negativicutes (Tables 1 and S1). One more member of that group, Thermosinus carboxydivorans, has been reported not to form spores (Sokolova et al., 2004), but its draft genome encoded a substantial number of sporulation genes. These sequence data offered a possibility of analysing sporulation genes in eight different genomes from three species of Negativicutes.

Table 1. Genome sequences of the Gram-negative members of the Firmicutesa
Organism nameSpore formerComplete genomesbGenome size, MbcProteinscGenBank accessiondProposed family assignment
  1. aAs of 1 April 2013. Only those genomes used in the phylogenetic trees (Figs 1, S1–S3) are listed here. The complete genome list is available as Table S1 and proposed family assignments as Table S6 in Supporting Information.
  2. bThe number in parentheses shows the number of unfinished genome projects that are registered with the NCBI's RefSeq database (Pruitt et al., 2012).
  3. cExact or rounded numbers for complete genomes, approximate numbers for draft genomes.
  4. dSome of these genomes have been described (Bowen De León et al., 2012; Brown et al., 2012; Chang et al., 2010; Gronow et al., 2010; Marx et al., 2011).
Acetonema longumYes0 (1)4.324284 AFGF00000000 Sporomusaceae
Acidaminococcus fermentansNo12.332026 CP001859 Acidaminococcaceae
Centipeda periodontiiNo0 (1)2.722631 AFHQ00000000 Selenomonadaceae
Megasphaera elsdeniiNo0 (1)2.472219 HE576794 Veillonellaceae
Pelosinus fermentans     Sporomusaceae
Strain A11Yes0 (1)5.064754 AKVM00000000  
Strain A12Yes0 (1)4.855138 AKVL00000000  
Strain B3Yes0 (1)4.885140 AKVK00000000  
Strain B4Yes0 (1)5.044691 AKVJ00000000  
Strain DSM 17108Yes0 (1)4.934593 AKVN00000000  
Strain JBW45Yes0 (1)5.284762 AKVO00000000  
Phascolarctobacterium succinatutensNo0 (1)2.122150 AEVN00000000 Acidaminococcaceae
Selenomonas ruminantiumNo13.633512 AP012292 Selenomonadaceae
Thermosinus carboxydivoransNo0 (1)2.892750 AAWL00000000 Sporomusaceae
Veillonella parvulaNo12.131844 CP001820 Veillonellaceae

Previous studies have shown that Bacilli and Clostridia share a common set of key sporulation genes but exhibit substantial differences in the regulation of the onset of sporulation, the engulfment process, and the assembly and protein content of the spore coat (Stragier, 2002; Onyenwoke et al., 2004; de Hoon et al., 2010; Galperin et al., 2012). A recent analysis identified ∼ 60 sporulation genes that were found in all spore former genomes and ∼ 60 genes that were found in all spore-forming bacilli but absent in all spore-forming clostridia (Galperin et al., 2012). A comparison of these gene sets against draft genomes of A. longum, T. carboxydivorans and six strains of P. fermentans revealed a surprisingly consistent pattern. Of 56 genes that were present in all bacillar and clostridial spore formers, 52 were also present in all eight spore-forming members of the Negativicutes, and three more (all except for spoVG) were only absent in T. carboxydivorans (Tables S2 and S3). This result was consistent with the classification of these bacteria within the Firmicutes. Further, out of 61 bacillar sporulation genes that were never found in clostridia, at least 57 were missing in all eight Negativicutes genomes (Table S4). Only few clostridia-specific sporulation genes have been identified so far (Lawley et al., 2009), and several of them had orthologs in A. longum, P. fermentans and/or T. carboxydivorans (Table S5). Thus, genomes of spore-forming members of the Negativicutes displayed essentially the same distribution of sporulation genes as genomes of spore-forming members of the Clostridia. Given the previous 16S rRNA-based assignment of these bacteria (and the entire Sporomusa-Selenomonas-Megasphaera group) to the class Clostridia (Stackebrandt et al., 1985; Willems and Collins, 1995a,b; Ludwig et al., 2009), we have decided to re-examine their systematic position using protein-based phylogenetic trees.

The availability of complete or draft genomes for several members of the Negativicutes allowed us to collect the necessary protein sequences and construct phylogenetic trees from (i) a concatenated alignment of 50 ribosomal proteins and alignments of (ii) β-subunit of the DNA-directed RNA polymerase (RpoB) and (iii) β-subunit of the DNA gyrase (GyrB) (see Fig. 1). Analysis of the phylogenetic trees for three spore-forming and six non-spore-forming species of the Negativicutes showed that they formed a well-defined separate group within the clostridial lineage (Fig. 1A). The pairwise groupings of Acetonema with Thermosinus, Acidaminococcus with Phascolarctobacterium, Centipeda with Selenomonas and Veillonella with Megasphaera were fully consistent with the 16S rRNA-based trees (Ludwig et al., 2009; Rainey, 2009; Marchandin et al., 2010) However, protein-based trees did not fully support the division of this group proposed by Marchandin et al. (2010), as Acetonema, Pelosinus and Thermosinus clearly grouped together and separately from both Acidaminococcus/Phascolarctobacterium and Veillonella/Megasphaera branches (Fig. 1A). Therefore, if the original family Veillonellaceae (Selenomonas-Megasphaera-Sporomusa group), with all 26 genera that are listed by Ludwig et al. (2009) and Rainey (2009) and up to eight recently described genera (see Table S6), is to be divided, A. longum, P. fermentans and T. carboxydivorans should be put into a separate family. Thus, Veillonellaceae would need to be split into at least four different families, including the Acidaminococcaceae with its four genera and Veillonellaceae sensu stricto with six genera as emended by Marchandin et al. (2010). The remaining members could be divided between families Selenomonadaceae (up to 10 genera) and Sporomusaceae (up to 14 genera of mostly spore-forming bacteria), in accordance with the 16S rRNA trees (Ludwig et al., 2009; Marchandin et al., 2010). The respective proposals are summarized in Table S6. Summing up, protein-based phylogenetic trees fully support the 16S rRNA-based trees in Ludwig et al. (2009) and Tocheva et al. (2011) and the well-supported branches of the trees in Rainey (2009) and Marchandin et al. (2010), as well as the taxonomic assignments for the Veillonellaceae in Bergey's and in the SILVA and Greengenes databases. Of the changes proposed by Marchandin et al. (2010), creation of the class Negativicutes does not seem to be justified, at least from the phylogenetic point of view. The name ‘Negativicutes’ could be retained as a mnemonic synonym to the Selenomonas-Megasphaera-Sporomusa group (i.e. the order Selenomonadales) within the class Clostridia. The family Acidaminococcaceae is only warranted if there are two other new families, Sporomusaceae and Selenomonadaceae; creation of the latter family is also needed to justify the order Selenomonadales. That said, a formal adoption of this classification should probably await completion of the genomes of Sporomusa ovata, Anaeroarcus burkinensis and Anaeromusa acidaminophila (currently in progress), which would also help in resolving the weak affiliation of the latter two organisms with the Sporomusa group.

Figure 1.

A ribosomal proteins-based phylogenetic tree of the Firmicutes that shows the positions of (A) Gram-negative members of the Firmicutes and (B) misassigned Clostridium spp. The maximum-likelihood tree was built from a concatenated alignment of 50 ribosomal proteins from 70 organisms with a total of 6164 unambiguously aligned positions, essentially as described in Yutin et al. (2012). The tree was rooted using two members of the Fusobacteria, Fusobacterium nucleatum and Leptotrichia buccalis, as an out-group. The numbers on the branches show TreeFinder confidence values. Those branches shared with the RpoB tree are indicated with thick lines; bootstrap values of those branches that are shared with the GyrB tree are shown in bold. The Roman numerals on the right correspond to the clusters of Collins et al. (1994). Grey triangles indicate clusters that are shown in more details on the other panel (see Supporting Information for details, original trees and Table S9 for the full list of organisms).

Phylogeny of the bacteria misassigned to the genus Clostridium

While members of Veillonellaceae stain Gram-negative owing to the structure of their cell envelopes, many Clostridium spp. have been reported to stain Gram-negative despite having a typical Gram-positive cell wall (e.g. Freier et al., 1988). Just 10 years ago, Sydney Finegold described the state of clostridial classification this way: ‘The genus Clostridium defies all the simple rules we learnt years ago when it was thought to consist of Gram-positive, spore-forming, anaerobic rods. Now the genus includes Gram-negatives, non-spore formers, cocci, and non-anaerobes’ (Finegold et al., 2002). Since that time, taxonomy of the genus Clostridium sensu stricto (cluster I in the classification of Collins et al., 1994) has been mostly resolved through concerted efforts of several different groups, assisted by the rapidly growing amount of genome sequence data (Stackebrandt et al., 1999; Gupta and Gao, 2009; Rainey et al., 2009). An extensive update of clostridial classification has been performed in the latest edition of Bergey's (Ludwig et al., 2009). In the course of that update, more than 50 bacteria previously placed in the genus Clostridium (Garrity et al., 2007) have been reassigned to other taxonomic groups, based on their 16S rRNA sequences and some other features. However, despite this reassignment, many organisms still retained the Clostridium name (Ludwig et al., 2009; Rainey et al., 2009), causing a major confusion in the clostridial taxonomy (see, e.g. McDonald et al., 2012). Thus, these bacterial species remain listed as Clostridium spp. in the LPSN and in GenBank\ENA\DDBJ, which creates a false impression that they are legitimate members of the Clostridium genus in the family Clostridiaceae. As a way to resolve this conundrum, the NCBI Taxonomy database and SILVA database currently display some of these organisms as [Clostridium], but this name has its own problems and can hardly be considered a permanent solution. We have compared the positions of several questionable Clostridium spp. on the ribosomal proteins-based tree (Fig. 1B) with the taxonomic assignments in the latest edition of Bergey's and in SILVA, RDP, Greengenes and GenBank databases. We hope that this analysis and the suggestions listed later (Table 2; see Tables S7 and S8 for details) would help in removing these stumbling blocks, or will at least stimulate a discussion on how to properly do that.

Table 2. Proposed genus assignments for former Clostridium spp
Organism nameaProposed family and genus name
  1. aThe list includes only those Clostridium spp. that (i) have been validly described and listed in the List of Prokaryotic names with Standing in Nomenclature (http://www.bacterio.net/, Euzéby, 1997), (ii) do not belong to the family Clostridiaceae and (iii) fall within the suggested new genera. Proposed type species are listed first and indicated in bold typeface.
  2. bThese species designations have been changed: Gottschalkia purinilytica, Gottschalkia angusta, Tyzzerella nexilis, Tyzzerella colina, Tyzzerella neopropionica, Tyzzerella piliformis, Tyzzerella propionica.
Clostridium ramosum, C. cocleatum, C. innocuum, C. saccharogumia, C. spiroformeErysipelotrichaceae, Erysipelatoclostridium
Clostridium acidurici, C. purinilyticumb, Eubacterium angustumbUnassigned, Gottschalkia
Clostridium phytofermentans, C. aerotolerans, C. aldenense, C. algidixylanolyticum, C. aminophilum, C. aminovalericum, C. amygdalinum, C. asparagiforme, C. bolteae, C. celerecrescens, C. citroniae, C. clostridioforme, C. fimetarium, C. glycyrrhizinilyticum, C. hathewayi, C. herbivorans, C. hylemonae, C. indolis, C. jejuense, C. lavalense, C. methoxybenzovorans, C. oroticum, C. polysaccharolyticum, C. populeti, C. saccharolyticum, C. scindens, C. sphenoides, C. symbiosum, C. xylanolyticum, C. xylanovorans, Desulfotomaculum guttoideum, Eubacterium contortum, Eubacterium fissicatenaLachnospiraceae, Lachnoclostridium
Clostridium difficile, C. bartlettii, C. bifermentas, C. ghonii, C. glycolicum, C. hiranonis, C. irregulare, C. litorale, C. lituseburense, C. mangenotii, C. mayombei, C. paradoxum, C. sordelii, C. sticklandii, C. thermoalcaliphilum, Eubacterium tenue, Eubacterium yuriiPeptostreptococcaceae, Peptoclostridium
Clostridium thermocellum, C. aldrichii, C. alkalicellulosi, C. caenicola, C. cellobioparum, C. cellulolyticum, C. cellulosi, C. clariflavum, C. hungatei, C. josui, C. leptum, C. methylpentosum, C. papyrosolvens, C. sporosphaeroides, C. stercorarium, C. straminisolvens, C. sufflavum, C. termitidis, C. thermosuccinogenes, C. viride, Bacteroides cellulosolvens, Eubacterium siraeumRuminococcaceae, Ruminiclostridium
Clostridium nexileb, C. colinumb, C. lactatifermentans, C. neopropionicumb, C. piliformeb, C. propionicumbLachnospiraceae, Tyzzerella

Clostridium difficile

In their classical paper 20 years ago, Collins et al. (1994) already noted that C. difficile and its close relatives C. paradoxum and C. sticklandii, along with Peptostreptococcus anaerobius, belong to a distinct family-level group (cluster XI or Family 13). Accordingly, the recent edition of Bergey's reclassified these 13 species into the family Peptostreptococcaceae (Ludwig et al., 2009). Unfortunately, the names of these organisms have not been changed. As a result, many biologists remain unaware that C. difficile is substantially distinct from Clostridium butyricum (the type species of the genus) and its close relatives, such as C. botulinum, C. perfringens and C. tetani. This distance is manifested, for example in the sporulation gene patterns: most C. difficile genomes lack such genes as spoIVFB, bofA, cotS, cotM, ydhD, gerA and gerC, which are widespread among Clostridium sensu stricto (Xiao et al., 2011; Galperin et al., 2012), but encode certain proteins that are not found in other clostridia (Lawley et al., 2009). The current versions of the NCBI Taxonomy database and SILVA database display C. difficile and its relatives as [Clostridium] spp., which is somewhat better but hardly resolves the confusion. We suggest renaming these organisms Peptoclostridium spp. Complete genomes of numerous isolates of C. difficile (Peptoclostridium difficile) and of C. sticklandii (Peptoclostridium sticklandii) strain DSM 519 are already available, and sequencing of other representatives of this group is currently under way. Analysis of their genome sequences may suggest that these organisms represent as many as four genera as has been suggested by Collins et al. (1994).

Clostridium acidurici

Protein trees agree with the 16S rRNA-based phylogeny (Hartwich et al., 2012), placing C. acidurici (cluster XII) as a sister group of Anaerococcus prevotii and Finegoldia magna (cluster XIII) that are currently assigned to Clostridiales Family XI Incertae Sedis. The complete genome sequences of these three organisms, as well as draft genomes of several other members of the Family XI Incertae Sedis, are already available, which should allow a better characterization of this family. Meanwhile, we propose assigning C. acidurici and its close relative C. purinilyticum to the new genus Gottschalkia as Gottschalkia acidurici and Gottschalkia purinilytica respectively.

Clostridium cellulolyticum

Clostridium thermocellum (both cluster III) and C. leptum (cluster IV), along with 13 other Clostridium spp., have been reassigned by Ludwig et al. (2009) to the family Ruminococcaceae. In protein-based trees, these three bacteria confidently clustered with Ruminococcus albus, confirming this placement. This group includes three other Clostridium spp. with completely sequenced genomes, C. clariflavum, C. stercorarium and Clostridium sp. BNL1100. These organisms are being intensively studied owing to their ability to metabolize cellulose, a common trait of all cluster III organisms described so far. Obviously, keeping these bacteria under the name Clostridium is counterproductive, as they are often being confused with non-cellulolytic clostridia. As a tentative solution, based on the relatively high rRNA similarity levels (Izquierdo et al., 2012), we suggest assigning cluster III members to the new genus Ruminiclostridium. In the future, some of these organisms might need to be moved to the genus Acetivibrio whose type species A. cellulolyticus is closely related to C. clariflavum.

Clostridium phytofermentans

In the 2nd edition of Bergey's (Ludwig et al., 2009), Clostridium symbiosum and 31 other Clostridium species that fall within clusters XIVa and XIVb of Collins et al. (1994) have been transferred to the family Lachnospiraceae, in full agreement with the assignments of SILVA and RDP databases. One of these species, Clostridium lentocellum, has been renamed Cellulosilyticum lentocellum (Cai and Dong, 2010), whereas, as far as we could see, the rest still retain the Clostridium name. Several well-known species from clusters XIVa and XIVb, such as C. sphenoides or C. piliforme, have not been mentioned by Ludwig et al. (2009) or Rainey et al. (2009), creating further confusion as to which Clostridium spp. belong to the family Lachnospiraceae and which should stay in Clostridiaceae. Thus, the web site http://www.broadinstitute.org/annotation/genome/clostridium_group/ of the Broad Institute, which has obtained draft genomic sequences of several organisms from this group (C. aldenense, C. bolteae, C. citroniae, C. clostridioforme, C. hathewayi and C. symbiosum), lists them all as Clostridium spp. Again, protein trees supported 16S rRNA-based assignments, placing C. phytofermentans and C. symbiosum in a tight cluster with Butyrivibrio proteoclasticus, Roseburia hominis and other members of Lachnospiraceae. We propose tentatively assigning all cluster XIVa organisms that are still listed as Clostridium spp. to the new genus Lachnoclostridium. For the cluster XIVb organisms, which include C. piliforme, the causative agent of Tyzzer's disease, we propose the new genus Tyzzerella (see below).

Clostridium ramosum

Clostridium ramosum and C. spiroforme, members of the cluster XVIII (Collins et al. (1994) along with two other Clostridium species, have been transferred by Ludwig et al. (2009) to the family Erysipelotrichaceae in the class Erysipelotrichi. Our data confirm their close relationship to Erysipelothrix rhusiopathiae, as well as clustering with mollicutes (Marchandin et al., 2010; Ogawa et al., 2011), which are currently assigned to the separate phylum, the Tenericutes. We propose assigning C. ramosum, C. spiroforme and three related species, to the new genus Erysipelatoclostridium.

Proposed new genera

The above suggestions are based primarily on the results from ribosomal protein-based phylogenetic trees (Fig. 1) and therefore miss those Clostridium spp. for which sequence information has been unavailable or insufficient. However, the excellent agreement of our protein trees with 16S rRNA-based classification presented in the SILVA and RDP databases indicates that, at least in the case of Clostridia, the assignments of these databases could be used to build a fairly reliable phylogeny-based taxonomy. Hence, based on the assignments in Bergey's, RDP and SILVA, as well as results of 16S rRNA similarity searches, we propose the following new genera (Table 2; see Table S8).

Description of Peptoclostridium gen. nov.

Peptoclostridium [Pep.to.clos.tri'di.um. Gr. v. peptô, digest; N.L. neut. dim. n. Clostridium, a bacterial genus name (from Gr. n. klôstêr, a spindle); N.L. neut. dim. n. Peptoclostridium, the digesting clostridium].

Gram-staining-positive, motile, spore-forming rods 0.3–1.5 μm × 1.5–20 μm. Obligate anaerobes, no microaerophilic or aerobic growth. Strains are mesophilic or thermophilic (temperature range from 20°C to 63°C) and grow in neutral to alkaline pH (some strains up to pH 11). Chemoorganotrophs. Oxidase and catalase negative. Peptone may serve as nitrogen source. Yeast extract can be used as the sole carbon and energy source. Several members require 1.5% NaCl for growth. Some mono- and disaccharides can be fermented; acetate is produced as a major end product. Sulfate is not reduced. The G+C content of the genomic DNA ranges from 25 to 32 mol%. The type species is Peptoclostridium difficile (formerly C. difficile); the type strain is ATCC 9689 = DSM 1296.

The newly proposed genus Peptoclostridium is equivalent to genus Clostridium XI in the RDP and the Peptostreptococcaceae genus Incertae Sedis in SILVA [see the 16S rRNA trees in Song et al. 2004) and Pikuta et al. (2009)]. It includes 11 validly described species that have been transferred to the family Peptostreptococcaceae in the recent edition of Bergey's, as well as C. mayombei and C. thermoalcaliphilum (Table 2). In addition, we propose that the genus include Eubacterium tenue, Eubacterium yurii and the following species that have not been validly described but whose 16S rRNA sequences are available in GenBank: C. maritimum (GenBank accession number EU089965), C. metallolevans (DQ133569), C. ruminantium (EU089964), C. venationis (EU089966) and the misnamed C. hungatei strain mc (JX073559; other C. hungatei strains go to Ruminiclostridium, see below). Two more members of the family Peptostreptococcaceae, Clostridium sticklandii and C. litorale, have been tentatively assigned to the genus Peptoclostridium to resolve the naming conundrum but might deserve to be put into a separate genus (or genera) [see Fig. 1B and Pikuta et al. (2009)]. Sporacetigenium mesophilum falls within the diversity of the new genus but is left as is because of its unusual metabolic properties (Chen et al., 2006).

Description of Gottschalkia gen. nov.

Gottschalkia (Gott.shal'ki.a. N.L. fem. n., named after Gerhard Gottschalk, in recognition of his contributions to the studies of various anaerobic bacteria, including clostridia).

Obligately anaerobic purinolytic spore-forming rods that, in the presence of 0.1% yeast extract, are capable of utilizing uric acid as sole carbon and energy source. Gram-staining is variable; motility is by peritrichous flagella. Optimal growth is at 19–37°C and pH 7.3–8.1. No utilization of carbohydrates, no reduction of nitrate and no production of H2. The DNA G+C content is 28–29 mol%.

The proposed genus has been first suggested by Collins et al. (1994); it is equivalent to Clostridiales Family XI Incertae Sedis genus Incertae Sedis in SILVA and includes two validly described organisms: Clostridium acidurici and C. purinilyticum [see Hartwich et al. (2012) and references therein]. Based on the similar 16S rRNA sequence and metabolic properties, Eubacterium angustum could be assigned to the same genus, despite its inability to form spores and higher G+C content (Beuscher and Andreesen, 1984). The type species is Gottschalkia acidurici (formerly Clostridium acidurici); the type strain is ATCC 7906 = DSM 604.

Description of Ruminiclostridium gen. nov.

Ruminiclostridium [Ru.mi.ni.clos.tri'di.um. L. n. rumen -inis, the rumen; N.L. neut. dim. n Clostridium, a bacterial genus name (from Gr. n. klôstêr, a spindle); N.L. neut. dim. n. Ruminiclostridium, clostridia-like bacteria in the family Ruminococcaceae].

Obligately anaerobic, mesophilic or moderately thermophilic, spore-forming, straight or slightly curved rods 0.5–1.5 μm × 1.5–8 μm. The cells have a typical Gram-positive cell wall, although often stain Gram-negative. Produce spherical or oblong terminal spores, which results in swollen cells. Most species are motile and have polar, subpolar or peritrichous flagella. The temperature range for various species is from 20°C to 70°C with Topt between 33 and 65°C. Optimal pH values are between 7 and 9 (some members can grow at pH as low as 5.9 or as high as 10.2). Oxidase and catalase are not produced. Yeast extract or vitamins are usually required for anabolic purposes. All known members can use cellulose, xylan and/or cellobiose as substrates, fermenting them primarily to acetate, ethanol, H2 and CO2, as well as lactate, propionate, butyrate or other end products. The ability to ferment other carbohydrates varies between species. Several species are capable of fixing N2. Sulfate is not reduced. The G+C content of the genomic DNA is typically 39–41.5%, but ranges from 27 to 51 mol% [while two species, C. alkalicellulosi and C. papyrosolvens, have been initially reported to have the G+C content of 29.9–30.0%, the genomic sequence of C. papyrosolvens DSM 2782 showed G+C content of 36.9% (Hemme et al., 2010)]. The type species is Ruminiclostridium thermocellum (formerly Clostridium thermocellum); the type strain is ATCC 27405 = DSM 1237.

The proposed genus Ruminiclostridium includes organisms from clostridial cluster III of Collins et al. (1994) and is equivalent to the genus Clostridium III in the RDP and the Ruminococcaceae genus Incertae Sedis in SILVA [see the 16S rRNA trees in Shiratori et al. (2009) and Izquierdo et al. (2012)]. It includes 15 validly described species, 12 of which have been transferred to the family Ruminococcaceae in the recent edition of Bergey's, as well as C. caenicola, C. clariflavum and C. sufflavum (Table 2). The Clostridium strain Rt51.B1 (GenBank: L09175), misnamed as C. sporogenes, and Clostridium sp. BNL1100 also belong to this genus.

Several Clostridium spp. that fall within the family Ruminococcaceae have been tentatively assigned to the genus Ruminiclostridium but will have to be reclassified and renamed after their phylogenetic status is better resolved. These include five members of the Collins et al. (1994) cluster IV (and genus Clostridium IV in the RDP): Clostridium leptum, C. cellulosi, C. methylpentosum, C. sporosphaeroides and C. viride. One more member of Ruminococcaceae, Clostridium orbiscindens, has been recently reclassified as Flavonifractor plautii (Carlier et al., 2010). In addition to former Clostridium spp., Bacteroides cellulosolvens, Eubacterium desmolans and Eubacterium siraeum fall within the proposed new genus. However, in future, some of its members might have to be reassigned to Acetanaerobacterium, Acetivibrio, Flavonifractor, Oscillibacter, Ruminococcus and/or new genera of Ruminococcaceae.

Description of Lachnoclostridium gen. nov.

Lachnoclostridium [Lach.no.clos.tri'di.um. Gr. n. lachnos, wool; N.L. neut. dim. n. Clostridium, a bacterial genus name (from Gr. n. klôstêr, a spindle); N.L. neut. dim. n. Lachnoclostridium, the clostridia within the family Lachnospiraceae].

Gram-positive, motile, obligately anaerobic spore-forming rods 0.3–1.5 μm × 1.5–20 μm. Strains are mesophilic or thermophilic (temperature range from 20°C to 63°C) and grow in neutral to alkaline pH (some up to pH 11). Chemoorganotrophs. Oxidase and catalase are not produced. Some mono- and disaccharides can be fermented; acetate is produced as a major end product. Sulfate is not reduced. The G+C content of the genomic DNA ranges from 25.6 to 32 mol%. The type species is Lachnoclostridium phytofermentans (formerly C. phytofermentans); the type strain is ATCC 700394 = DSM 18823.

The proposed genus Lachnoclostridium includes organisms from clostridial cluster XIVa of Collins et al. (Collins et al., 1994), the genus Clostridium XIVa in the RDP and the Lachnospiraceae genus Incertae Sedis in SILVA [see the 16S rRNA trees in Warren et al. (2006) and Domingo et al. (2009)]. It includes 30 validly described species, most of which have been assigned to the family Lachnospiraceae in the recent edition of Bergey's (Table 2). It also includes the following species that have not been validly described but whose 16S rRNA sequences are available in GenBank: C. boliviensis (AY943862), C. fusiformis (AB702934), C. sulfatireducens (AY943861) and a misnamed strain of C. leptum (AF262239; C. leptum type strain DSM 753 goes to Ruminiclostridium; see Fig. 1B). Desulfotomaculum guttoideum, Eubacterium contortum, Eubacterium fissicatena and Ruminococcus torques also belong to this genus.

Description of Tyzzerella gen. nov.

Tyzzerella [Ti.ze.rel'la] N.L. fem. n. Tyzzerella, named after Ernest Tyzzer, an American pathologist who isolated and described ‘Bacillus piliformis’, the causative agent of Tyzzer's disease.

A closely related cluster of organisms in the family Lachnospiraceae includes six Clostridium spp. that warrant assignment to a separate genus. The description of the new genus is essentially the same as that of Lachnoclostridium (see earlier), although some of its members are non-motile and non-spore-forming and have higher G+C contents, from 40% in C. nexile to 46.8% in C. colinum. The genus is named after Ernest Edward Tyzzer (1875–1965), who in 1917 characterized ‘Bacillus piliformis’, the causative agent of an infectious diarrhoea of laboratory mice, which was later found in a variety of animals and became known as ‘Tyzzer's disease’. Unfortunately, no Tyzzerella piliformis (formerly Clostridium piliforme) strains have been deposited in public culture collections so far (deposition is currently in progress). Accordingly, Tyzzerella nexilis [formerly Clostridium nexile (Holdeman and Moore, 1974)] is selected as the type species and ATCC 27757 = DSM 1787 as the type strain.

Description of Erysipelatoclostridium gen. nov.

Erysipelatoclostridium [E.ri.si.pe.la.to.clos.tri'di.um]. Gr. n. erusipelas -pelatos, erysipelas; N.L. masc. n. Clostridium, a bacterial genus name (from Gr. n. klôstêr, a spindle); N.L. neut. dim. n. Erysipelatoclostridium, Clostridium-like members of the order Erysipelotrichales.

Gram-positively staining, non-motile, obligately anaerobic straight or helically curved rods 0.3–1.0 μm × 2–4 μm. Spore formation is rare or absent. The G+C content of the genomic DNA is 27–33 mol%. Ferment glucose, fructose and sucrose [see Kaneuchi et al. (1979) for a detailed comparison]. The type species is Erysipelatoclostridium ramosum (formerly Clostridium ramosum); the type strain is ATCC 25582 = DSM 1402.

The newly proposed genus Erysipelatoclostridium is equivalent to the Clostridium XVIII genus in RDP and Erysipelotrichaceae genus Incertae Sedis in SILVA [see the 16S rRNA trees in Clavel et al. (2007) and Ogawa et al. (2011)]. It includes four validly described species: Clostridium cocleatum, C. ramosum, C. saccharogumia and C. spiroforme. In addition, Clostridium innocuum, which is more distantly related to the rest of the group and has G+C content of 43–44%, is tentatively assigned to this species but might have to be reclassified in the near future. The ability of C. cocleatum, C. ramosum and C. spiroforme to form spores contradicts the current description of the family Erysipelotrichaceae, which is why the proposed genus Erysipelatoclostridium should either be placed in the order Erysipelotrichales outside the family Erysipelotrichaceae or the description of the family be emended.

Protein-based and rRNA-based phylogeny versus cell wall structure

This study, in agreement with many earlier ones, demonstrated a high degree of coherence between 16S rRNA-based and protein-based trees for various members of the Firmicutes. While the congruity between 16S rRNA and ribosomal proteins S2-S20 that bind to this rRNA is hardly surprising, it must be noted that the concatenated alignment of 50 ribosomal proteins used in this work included 6164 unambiguously aligned positions, of which only 2367 (or 38%) were provided by 20 small subunit proteins, while the rest came from the large subunit of the ribosome (see the Supporting Information). Sequences of two other proteins, RpoB and GyrB (1154 and 631 positions respectively) provided an additional, independent measure of the evolutionary proximity of the studied organisms.

In general, the results of this work reaffirm that protein sequences deduced from selected groups of informational genes, as defined by Rivera et al. (1998), provide a valid tool for phylogenetic analysis of distant bacterial species (see, e.g. Wolf et al., 2001; Ciccarelli et al., 2006; Gupta and Gao, 2009; Yutin et al., 2012) and could successfully complement 16S rRNA-based trees in building the ultimate genome-based classification of Bacteria and Archaea (Klenk and Göker, 2010). This conclusion becomes particularly important when the results of phylogenetic analyses contradict taxonomic assignments that are based on phenotypic traits, such as the structure (or even presence) of the bacterial cell wall, Gram-staining pattern, motility or metabolic properties. Thus, despite the Gram-negative structure of their cellular envelopes, members of the family Veillonellaceae clearly belong within the Clostridia, and do not deserve placement into a separate class of Firmicutes.

Acknowledgements

We thank Drs. Boris Belitsky, Elizaveta Bonch-Osmolovskaya, Ilya Borovok, Jean Paul Euzéby, William B. Whitman and Juergen Wiegel for critically reading the manuscript and many helpful comments and Drs. Yuri Wolf and Eugene Koonin for advice on the phylogenetic trees. This work was supported by the NIH Intramural Research Program at the National Library of Medicine.

Ancillary