Reclassification of Clostridium coccoides, Ruminococcus hansenii, Ruminococcus hydrogenotrophicus, Ruminococcus luti, Ruminococcus productus and Ruminococcus schinkii as Blautia coccoides gen. nov., comb. nov., Blautia hansenii comb. nov., Blautia hydrogenotrophica comb. nov., Blautia luti comb. nov., Blautia producta comb. nov., Blautia schinkii comb. nov. and description of Blautia wexlerae sp. nov., isolated from human faeces

It is becoming increasingly clear that humans have a symbiotic relationship with their intestinal microbial community. This community enjoys a stable nutrient-rich environment with limited host immune responses, and in turn, it facilitates the normal functions of the body. Although this community has been studied using culture techniques, culture-independent investigations based on sequence analysis of rRNA genes suggest that 40 to 80 % of the total 16S rRNA gene sequences retrieved represent as-yet uncultivated species (Suau et al., 1999). It is recognized that 16S rRNA gene sequencing used in conjunction with culturing techniques represents a powerful approach for discerning novel species diversity within the human gut and faeces (Barcenilla et al., 2000; Suau et al., 1999). In the present study, a combination of culture-based and molecular-based methodologies was used to facilitate the identification of 15 isolates of a hitherto unknown species within the Clostridium coccoides rRNA XIVa cluster. On the basis of an investigation using a polyphasic taxonomic, it is proposed that these strains be classified, alongside a number of misclassified ruminococci, in a novel genus.

Fifteen isolates were recovered from stool specimens from control children and children with late onset autism at the Rush Children's Hospital, Chicago, USA (Finegold et al., 2002). The strains were characterized biochemically by using a combination of conventional tests as described previously in the Wadsworth and VPI anaerobic bacteriology manuals (Holdeman et al., 1977; Jousimies-Somer et al., 2002) and the API ZYM and Rapid ID32A systems (bioMérieux) according to the manufacturer's instructions. All biochemical tests were performed in duplicate. Carbohydrate fermentation tests were conducted using pre-reduced, anaerobically sterilized (PRAS) peptone-yeast-carbohydrate broth tubes (Anaerobe Systems). The strains were grown in peptone-yeast (PY) broth and peptone-yeast-glucose (PYG) broth (Anaerobe Systems) for metabolic end product (short-chain volatile and non-volatile fatty acids) analysis by GLC (Jousimies-Somer et al., 2002). Long-chain cellular fatty acids were analysed as previously described (Wexler et al., 1996). The 16S rRNA genes were amplified by PCR using universal primers 8UA (positions 8–28, Escherichia coli numbering) and 1485B (positions 1485–1507) as described previously (Brosius et al., 1978). The amplified product was purified by using QIAamp PCR purification kit (Qiagen, Inc.) and directly sequenced with an ABI 3100 Avant genetic system (Applied Biosystems). The closest known relatives of the new isolates were determined by performing database searches using the fasta program (Pearson & Lipman, 1988). These sequences and those of other known related strains were retrieved from GenBank and aligned with the newly determined sequence using the arb program (Ludwig et al., 2004). Phylogenetic analysis was performed on 1297 shared nucleotides using the maximum-likelihood and maximum-parsimony algorithms and the stability of the groupings was estimated by bootstrap analysis (1000 replications) using the same programs.

The isolates originating from the faecal material were strictly anaerobic, Gram-positive, non-spore forming, 1.0–1.5×1.0–3.0 μm in size and coccobacillus-shaped. After 48 h of anaerobic incubation at 37 °C under a gas phase of N₂ (86 %), H₂ (7 %) and CO₂ (7 %), colonies on Brucella blood agar plates were 1–2 mm in diameter, grey with a white centre, entire edges, umbonate and opaque. The isolates grew well anaerobically, but no growth occurred following subculture in an atmosphere of 2 % or 6 % O_2. All of the isolates were negative for lecithinase, lipase, catalase and indole. They were capable of hydrolysing aesculin, but not gelatin. All of the isolates utilized arabinose, glucose, mannose, ribose, xylose, but not glycogen as substrates when grown in PRAS carbohydrate broths. The utilization of amygdalin, cellobiose, fructose, lactose, maltose, mannitol, melezitose, melibiose, raffinose, rhamnose, salicin, sorbitol, sucrose and trehalose was variable. Analysis of end products by GLC from peptone-yeast-glucose broth revealed acetate and succinate as the major end products of metabolism. The long-chain cellular fatty acids of the isolates were of the straight-chain saturated and monounsaturated types, with C_16 : 0 (major component), C_14 : 0, and C_16 : 0 dimethyl acetal fatty acids predominating. Using the API ZYM and Rapid ID32A systems, all 15 isolates produced a similar profile. Positive reactions were obtained for α-galactosidase, β-galactosidase, α-glucosidase, β-glucosidase and α-fucosidase by both systems. With the Rapid ID32A system, α-arabinosidase activity showed as positive and acid phosphatase was detected by the API ZYM system. No activity was detected for N-acetyl-β-glucosaminidase, alkaline phosphatase, α-chymotrypsin, cystine arylamidase, β-glucuronidase, lipase (C14), leucine arylamidase, α-mannosidase, trypsin or valine arylamidase. Urease results were variable when tested by the Rapid ID32A system and esterase (C4), esterase lipase (C8), and naphthol-AS-Bl-phosphohydrolase were variable as tested by the API ZYM system. The isolates were sensitive to vancomycin (5 μg) and kanamycin (1000 μg), but resistant to colistin (10 μg) identification discs.

To determine the phylogenetic affinities of the novel isolates, their 16S rRNA genes were amplified by PCR and sequenced. The 15 isolates were highly related to each other with 99.5–100 % 16S rRNA gene sequence similarity. Sequence searches of GenBank and Ribosomal Database Project libraries revealed that the isolates were members of the phylum Firmicutes. It was evident from treeing analysis that the novel isolates represented a previously unknown lineage within the clostridial rRNA cluster XIVa subgroup and, in particular, shared a close relationship with Ruminococcus luti and Ruminococcus obeum. Pairwise comparison revealed approximately 5.0 % sequence divergence between the novel isolates and the type strains of the closest recognized species, R. luti and R. obeum, based on the almost full-length 16S rRNA gene sequences (=1400 nt). A phylogenetic tree, constructed by the maximum-parsimony method, depicting the phylogenetic affinity of the novel bacterium as exemplified by strain WAL 14507^T is shown in Fig. 1⇓. Strain WAL 14507^T formed a distinct lineage within a small subcluster of species which included several recognized ruminococcal species (Ruminococcus hansenii, Ruminococcus hydrogenotrophicus, R. luti, R. obeum, Ruminococcus productus, Ruminococcus schinkii) and Clostridium coccoides. The major branching orders were confirmed by using the maximum-likelihood method (see Supplementary Fig. S1 in IJSEM Online). Although there is no precise correlation between the 16S rRNA gene sequence divergence value and species delineation, it is generally recognized that similarity values of 97 % or less are significant (Stackebrandt & Goebel, 1994). However, Stackebrandt & Ebers (2006) have recently made the recommendation that this value should be increased to 98.7–99 % without sacrificing the quality and precision of a ‘species’ description and as an aid to taxonomists. This followed observations that with a threshold value of 98.5 % gene sequence similarity, the corresponding DNA–DNA reassociation values were always lower than 70 %, the accepted value for a species. It is apparent that the 16S rRNA gene, although a powerful tool in phylogenetic reconstructions, has limitations in discriminating recent speciation events. This is reflected in an increasing number of species now being described within the literature with very high rRNA gene sequence similarities to their nearest relatives. For example, Clostridium carboxidivorans displayed 99.7 % and 99.8 % to Clostridium scatologenes and Clostridium drakei, respectively (Liou et al., 2005), with DNA–DNA reassociation values demonstrating that all three strains represented distinct species. When such high 16S rRNA gene sequence similarity values are used for describing novel species, they must always be accompanied by appropriate phenotypic observations that can be used to distinguish the novel taxa. Support for the distinctiveness of the novel bacterium isolated from human faeces from the related bacterial species is also provided from the phenotypic characterization. Although 16S rRNA gene sequence analysis showed the closest phylogenetic relatives to the novel bacteria are R. luti and R. obeum, strain WAL 14507^T could be distinguished easily from R. luti and R. obeum by several features. In contrast to R. luti, the novel strain did not produce β-N-acetylglucosaminidase, alkaline phosphatase or α-mannosidase. Similarly, strain WAL 14507^T could be differentiated from R. obeum by the production of acetate and succinate as major fermentation products, but not ethanol and also by its ability to strongly hydrolyse aesculin and produce significant amounts of α-arabinosidase and α-fucosidase. The characteristics that distinguish the novel strains from their most closely related species are summarized in Table 1⇓.

In several studies, a number of uncultured bacteria represented by sequences derived from faecal 16S rRNA gene clone libraries were found to be closely related to R. obeum (Suau et al., 1999; Zoetendal et al., 1998). Zoetendal et al. (2002) followed up on these studies using a combination of fluorescent in situ hybridization and flow cytometry. Several probes, one of which was designated Urobe63, were specific for these R. obeum-like sequences. Their data showed that approximately 16 % of the total community belonged to the cluster XIVa (C. coccoides group) and of this 2.5 % belonged to R. obeum-like organisms, although this value varied between 1 and 6 % for different individuals. A comparison of the nucleotide sequence of the Urobe63 probe and the 16S rRNA gene sequence of our novel organism showed that the probe sequence was present within the 16S rRNA genes of the novel bacterium. Furthermore, two of the sequences that were used by Zoetendal et al. (2002) to validate and test the probes were found to be 98–99 % related to our novel organism and it is highly likely that these sequences correspond to strains of this hitherto unknown bacterium. Therefore it can be reasonably assumed that some proportion of the organisms enumerated by the Urobe63 probe represented our novel organism and this suggests that the novel bacterium (and other closely related R. obeum-like strains) comprises a significant fraction of the faecal community.

A number of studies have shown that the genus Ruminococcus is not monophyletic and is phylogenetically heterogeneous (Rainey & Janssen, 1995; Willems & Collins, 1995; Rieu-Lesme et al., 1996). The type species, Ruminococcus flavefaciens, is a member of a cluster designated clostridial rRNA cluster IV (Collins et al., 1994) and this cluster is also referred to as the Clostridium leptum group of organisms. Ruminococcus albus, Ruminococcus callidus and Ruminococcus bromii are also members of this suprageneric cluster, although the latter species is somewhat removed from the former species. Phylogenetically, the remainder of the ruminococci are members of the rRNA cluster XIVa (Collins et al., 1994) and therefore should not be considered true phylogenetic members of the genus Ruminococcus sensu stricto. This grouping is often referred to as the Clostridium coccoides group, although this label is misleading due to this large supra-generic cluster embracing taxa that bear a plethora of generic names. Currently the grouping embraces almost 20 genera such as Acetitomaculum, Anaerostipes, Bryantella, Butyrivibrio, Catonella, Clostridium, Coprococcus, Dorea, Eubacterium, Hespellia, Johnsonella, Lachnospira, Lachnobacterium, Pseudobutyrivibrio, Roseburia, Ruminococcus, Shuttleworthia, Sporobacterium and Syntrophococcus and many misclassified clostridial species. It is now recognized that members of this cluster are in need of extensive taxonomic revision with Collins et al. (1994) proposing to retain the true clostridia in cluster I which bears Clostridium butyricum as the type species.

Phylogenetic analysis employing a number of software programs (neighbour-joining, maximum-likelihood and parsimony) using only matched nucleotides demonstrated that the novel faecal organism formed a robust group with a number of misclassified ruminococci (R. hansenii, R. hydrogenotrophicus, R. luti, R. obeum, R. productus, R. schinkii) together with C. coccoides (Fig. 1⇑). It is pertinent to point out that C. coccoides, the only species isolated from animal sources (mice), produces endospores unlike all other species in this clade. It may be argued that this organism should not be included in any new genus that contains these Ruminococcus species. However, C. coccoides and R. productus share a very high 16S rRNA gene sequence similarity (99.7 %) and an examination of the literature demonstrates that the biochemical profiles, products of glucose metabolism and DNA G+C content of these two organisms are almost identical (Kaneuchi et al., 1976; Ezaki et al., 1994). Although some internal structure is present (see Table 2⇓ and Fig. 1⇑), with C. coccoides, R. hansenii, and R. productus forming a somewhat separate subcluster, the biochemical, morphological and phylogenetic characteristics of these organisms are consistent with generic unity. From a practical standpoint, taxonomy should be a practical subject embracing groups of organisms with unifying features. The proliferation of taxa based solely on a few traits should be resisted. In this diverse group of organisms (cluster XIVa), an accurate identification relies to a great extent upon molecular genetic techniques such as 16S rRNA gene sequence comparisons. Once the sole domain of specialized research facilities, these high throughput methodologies are becoming increasingly automated, reducing costs and making them routinely accessible. The availability of these methods facilitates more rapid and accurate identification of hitherto unknown or misclassified taxa. In addition to the above-mentioned taxa, the species Ruminococcus gnavus, Ruminococcus lactaris and Ruminococcus torques are also present in the XIVa suprageneric cluster although these species are phylogenetically removed from the cluster which contains C. coccoides, R. hansenii, R. hydrogenotrophicus, R. luti, R. obeum, R. productus, R. schinkii and the novel faecal isolate. It is not the purpose of this article to reclassify all misclassified ruminococci. We believe that the most appropriate time to reclassify these remaining three species will be when additional novel strains are isolated and can be assigned along with these species to form the nuclei of new genera.

Therefore, based on morphological, phenotypic and phylogenetic considerations, we believe that C. coccoides and a number of misclassified ruminococci merit classification in a novel genus for which we propose the name Blautia gen. nov. The novel coccobacillus-shaped organisms isolated from human faeces are proposed as a distinct species within this genus as Blautia wexlerae sp. nov. R. obeum is clearly a member of the genus Blautia, as currently defined. The rules of the Bacteriological Code currently require that the type strains of all new species and subspecies (including new combinations) be deposited in two different collections in two different countries. The type strain of R. obeum is currently only deposited in the ATCC and a second independent deposit, as required by the Code, has not been possible (requests to the ATCC were made to enable this but could not be accommodated satisfactorily). Consequently, the transfer of this species to the genus Blautia cannot be made because the resulting species name would not conform to the Rules of the Code governing the valid publication of species names and deposit of type material (Rules 27 and 30) and consequently would not be considered to be validly published.