Background

Bacterial identification becomes a challenge particularly in case they are either involved in an industrial process with heavy investments at risk or are a serious threat to human beings. Sequencing of the rrs (16S rDNA) of bacteria is vigorously pursued for correct identification and classification [1–3]. It has led to a large database of 1,237,963 entries http://rdp.cme.msu.edu/. The key questions which we are addressing: Whether there are certain latent and as yet un-explored features in the nucleotide sequences of the rrs, which can be used to define the phylogenetic limits of a particular genus or species? Are there certain unique patterns of nucleotide strings (length and order) and signatures within them, which can enable tracking the identity of an organism within the phylogenetic framework? In fact, it has been stated that novel biological findings are being driven by comparison and integration of massive data sets [4]. They also predict that numerous tools will be designed to use such large and well organized data. Comparative analyses of rrs sequences of Clostridium spp. can be exploited to develop molecular tools for defining the genetic variability and tracking the evolutionary events.

Clostridium represents a large bacterial genus of significant biotechnological as well as medical importance. The diversity of its economic importance varies from production of solvents - Clostridium beijerinckii, C. acetobutylicum, C. saccharoperbutylacetonicum and C. sachharobutylicum[5], biofuels such as ethanol and hydrogen - Clostridium thermocellum and C. acetobutylicum[6–9], indigo dye and flax retting to enzymes as therapeutic and cosmetic agents [10]. The potential of Clostridium butyricum to naturally produce 1,3-propanediol holds promise as base material for fiber - poly(propylene terapthalate) industry [11]. Cheese (Gonda and Grana) industry is invariably plagued by defects caused by Clostridium tyrobutyricum, C. beijerinckii, C. butyricum and C. sporogenes[12, 13]. In contrast are the Clostridium strains with medical relevance particularly those with abilities to produce some of the deadliest toxins, leading to devastating clinical conditions [14–16]. Production of potent extracellular toxins by Clostridium spp. especially those due to C. botulinum, C. perfringens, C. tetani and C. difficile result in diseases such as botulism, gas gangrene, tetanus, pseudomembranous colitis and food-borne illness in man and animals [17–23]. To summarize, members of the Clostridium spp. are responsible for economic and medical problems [24].

Multiple methods have been employed for the identification of strains and/or samples of Clostridium and its species, such as PCR-ribotyping combined with DNA-DNA reassociation [25, 26], Amplified fragment length polymorphism (AFLP) [17], Randomly amplified polymorphic DNA [27], Multilocus sequence typing [28], Restriction enzyme (RE) analysis [29, 30] and Microarray [31].

The heterogeneity of Clostridium species

C. botulinum

The heterogeneity of C. botulinum is evident from the variability of neurotoxins produced by them. Genetic studies reveal complex relationship between their taxonomy and toxin types, which needs a thorough re-examination. Certain non-toxigenic strains of C. botulinum, which have been genotypically shown to be authentic members of this species possess cryptic/silent genes or sequences coding for BoNT [32] but are functionally categorized as "non-producers". The complexity of the identification process is further enhanced by strains which phenotypically resemble Clostridium baratii and C. butyricum yet produce BoNT types F and E, respectively [33]. So these strains are referred to as C. botulinum in spite of the observations made on their DNA-DNA pairing hybridization, rrs sequencing and BoNT-encoding plasmids, etc. [16].

C. acetobutylicum

Different studies have provided evidences to support high genetic variability within C. acetobutylicum[34]. C. acetobutylicum is phylogenetically close to C. felsineum but DNA-DNA hybridization studies classify them as distinct species [34, 35]. A genome wide comparison of C. acetobutylicum with Bacillus subtilis reveals significant local conservation of gene order [36]. Such high homology between proteins from other taxa emphasizes the potential of horizontal gene transfer, which leads to microbial evolution [16, 36–38]. Interestingly, B. subtilis itself has been shown to be very heterogeneous [3, 39]. Clostridial and Bacillus binary toxins share 80-85% identity within the i toxin family [40]. Phylogeny and evolutionary relationship within Clostridium are poorly understood even today [28].

The needs of the genus Clostridium

Clostridium are phylogenetically extremely heterogeneous bacteria [35, 41], with many non-spore formers grouped along with spore forming Clostridium spp. [42]. Clostridium contains both Gram-positive and Gram-negative species - C. phytofermentans shows a Gram-negative reaction in spite of possessing a Gram-positive cell wall ultrastructure [43]. In view of multiple hindrances encountered in accurately identifying these organisms it becomes difficult to establish a good co-relation between their pathogenic potential and disease manifestation [44]. The need is to have a reliable identification system [17] and to recognize distinct genetic-based clusters, which would help to establish a phenotype to genotype co-relation and lead to a clear taxonomic classification [45]. Since symptoms can be confusing, the need is to develop molecular genetics tools for improved diagnosis of Clostridium species [46]. Another, equaling challenging scenario is encountered due to their GC content, which varies from 24 mol% (C. perfringens) [47] to 58 mol% (Clostridium barkeri) [48]. Such a wide range of GC content is perhaps too great for a single genus [42] and may even demand re-classification.

A new dimension to establish evolutionary relationship among bacteria in general and specifically Clostridium species has been added by the availability of a large reservoir of rrs sequencing data http://rdp.cme.msu.edu/. Another important development has been the 1173 sequenced genomes http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi including 224 Firmicutes consisting of 24 Clostridium spp. http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi. All these put together are expected to prove helpful in redefining microbial taxonomy and phylogeny [3, 4]. It may be remarked that studies based on rrs sequences need to be supplemented with information on other genes for authentic segregation of isolates [3]. Recent studies based on certain 'latent' features (unique signatures and in silico RE digestion patterns) of rrs alone could very clearly distinguish the two subgroups of B. subtilis[3], where as others have resorted to gyrA gene to elucidate this genetic variability [39]. It also proved instrumental in evaluating the biodiversity of Stenotrophomonas spp. [49]. The sequence of the 16S rDNA contains potential opportunities to characterize and differentiate species.

In this study, we have used rrs sequences to identify molecular markers, which may be sufficient enough to segregate Clostridium spp. Here, we have carried out comparative analyses of rrs sequences belonging to 15 Clostridium species as reference sets for generating species specific: (i) phylogenetic frameworks, (ii) signatures (30 nucleotides, nts) and (iii) in silico RE digestion patterns with 14 Type II REs. These tools were then used for identifying: (i) Clostridium sp. up to species level, (ii) identifying potential novel species, and (iii) tracking phylogenetic relationships among the other Clostridium species, which have fewer members at present.

Results

The genus Clostridium has been reported (from RDP/NCBI sites: http://rdp.cme.msu.edu/; http://www.ncbi.nlm.nih.gov/ to consist of 110 species. For this study, the selected 765 rrs sequences (> 1200 nts) belonging to these 110 Clostridium species have been grouped into three categories (Table 1, Additional file 1: Table S1): (i) 15 Clostridium spp. with relatively higher number (8-128) of identified organisms, (ii) 94 Clostridium spp. (including 9 un-classified species and 5 species of Eubacterium) with relatively low frequency (1-15) of identified organisms and (iii) 179 organisms identified only up to genus level - Clostridium sp. Out of a total of 765 rrs sequences of different Clostridium species, we chose 404 for generating phylogenetic framework for the following 15 species belonging to the first group (Table 1): C. botulinum - 128, C. perfringens - 92, C. butyricum - 32, C. acetobutylicum - 24, C. beijerinckii - 23, C. novyi - 17, C. kluyveri - 14, C. pasteurianum - 13, C. sporogenes - 11, C. colicanis and C. sardiniense - 9 each, C. baratii, C. chauvoei, C. subterminale and C. tetani - 8 each. The species-specific phylogenetic framework developed here was used as a tool for (i) identifying 179 organisms Clostridium sp. up to genus level, and (ii) establishing the phylogenetic relationships among 182 organisms reported to be occur in low frequency (Additional file 1: Table S1).

Table 1 16S rDNA sequences of Clostridium species which occurred with high frequency and number of sequences used in this study http://rdp.cme.msu.edu/
Full size table

Phylogenetic framework for Clostridium species

C. botulinum

Phylogenetic tree based on the rrs sequences of 128 strains of C. botulinum revealed their segregation into 4 major clusters (CBoI-CBoIV) (Additional file 2: Figure S1). These different clusters were represented by 3-83 strains. These four clusters may be representing 4 distinct lineages (I-IV) of C. botulinum, which have been segregated on the basis of neurotoxin (BoNTs) produced by them [20, 50]. This segregation at this stage was supported by varied Bootstrap Value (BV): CBoI showed a BV in the range of 4-536 and exceptionally a value of 1000, indicating very high genetic variability among its members. Members of the cluster CBoII had BV in the range of 242-1000. A relatively low genetic heterogeneity was recorded among members of clusters CBoIII and CBoIV, where BV varied from 495-1000. In view of the high genetic variability among the different isolates, we selected four rrs sequences to represent CBoI, whereas two sequences each were selected for representing the rest three clusters: CBoII-CBoIV. In all, the 10 representative sequences including the type strains S000414699 and S000260209 were short listed. These rrs sequences were distributed all along the phylogenetic tree of C. botulinum. Further analyses based on nucleotide signatures and in silico RE digestion patterns supported the selection of these representatives.

C. perfringens

A phylogenetic tree based on the rrs sequences of 92 strains of C. perfringens showed 9 distinct clusters (CPeI-CPeIX) (Additional file 2: Figure S2) each consisting of between 3-20 strains. Four out of these 92 strains did not fell into any of these clusters. High genetic diversity among the different strains of C. perfringens was supported by low BV in each of the clusters. A total of ten representative sequences, including the one of the type strain (S000721508) were selected from 6 clusters (CPeI-CpeIV, CPeVI-CPeVII) for developing phylogenetic framework.

C. butyricum

The 32 different rrs sequences of C. butyricum were distributed largely on 6 different clusters (CBuI-CBuVI) (Additional file 2: Figure S3) of the phylogenetic tree. These different clusters were represented by 2-8 strains each. The low BV in different clades is a clear reflection on the high genetic heterogeneity within this Clostridium species. In order to represent the genetic heterogeneity of this Clostridium species in the phylogenetic framework, 3 rrs sequences were short listed (including one type strain (S000116309) from different clades - CBuII, CBuV and CBuVI.

C. acetobutylicum

The 24 strains of C. acetobutylicum were very heterogeneously distributed on their rrs sequence based phylogenetic tree. The two major clades (CAcI-CAcII) (Additional file 2: Figure S4) consisted of 9 and 5 sequences each, with a BV in the range of 393-1000 and 338-698, respectively. The rest 10 sequences were segregated into 9 different branches with a BV ranging from 7-369 and exceptionally it was 1000. The selection of five rrs sequences to develop a phylogenetic frame work was done in order to cover maximum range of the genetic diversity. Two of the selected strains (S000437209 and S000000299) represented the extreme ends of the phylogenetic tree, whereas the type strain S000628106 was located almost in its middle. A support to high genetic variability within C. acetobutylicum has been reported [34], with high potential for horizontal gene transfers [36].

C. beijerinckii

The 23 sequences of rrs of C. beijerinckii further supported the genetic heterogeneity within Clostridium. Four distinct clusters (CBeI-CBeIV) (Additional file 2: Figure S5) each containing 3-8 strains were represented by highly variable BVs: 197-1000, 95-135, 86-824, and 160-1000, respectively. Three sequences were selected to represent the diversity of the evolutionary tree: 2 from CBeI and the type strain (S000014607) from CBeIV.

C. novyi

Phylogenetic tree based on the rrs sequences from 17 strains of C. novyi showed 4 distinct clusters (CNoI-CNoIV) (Additional file 2: Figure S6) with 1-6 strains each. Only one strain (S000750034) was on an extreme end of these clusters. Four strains, including the type strain (S000016169) spanning the whole phylogenetic tree were taken into consideration while developing the framework.

C. kluyveri

A phylogenetic tree based on the rrs sequences of 14 strains of C. kluyveri formed 4 distinct clusters (CKlI-CKlIV) (Additional file 2: Figure S7) each containing at least 2 strains and a maximum of 6 strains. Two of the 14 strains, present almost at the extreme ends of the phylogenetic tree were selected for further use as framework sequences.

C. pasteurianum

The low genetic diversity among the 13 strains of C. pasteurianum was evident in their rrs sequence based phylogenetic tree. It showed a major cluster (CPaI) (Additional file 2: Figure S8) (containing 4 strains), whereas the rest of the sequences were distributed on independent branches with a maximum of two in a group. A total of three strains, including the type strain (S001792892), were selected for framework sequences.

C. sporogenes

Phylogenetic tree based on the rrs sequences from 11 strains of C. sporogenes formed a single group (Additional file 2: Figure S9). Two strains, including type strain (S000260539) and another from the extreme end of the tree were selected for framework sequences were taken to represent maximum diversity within this species.

C. colicanis

Phylogenetic tree based on the rrs sequences from 9 strains of C. colicanis showed no clear cut grouping as evident by the low BV (Additional file 2: Figure S10). Two strains, including type strain (S000366397), representing the two extreme ends of the phylogenetic tree were selected as framework sequences.

C. sardiniense

High genetic diversity in the phylogenetic tree based on the rrs sequences from 9 strains of C. sardiniense was supported by low BV in most of the branches (Additional file 2: Figure S11). Three strains, including type strain (S000539075), were selected as framework sequences to represent maximum genetic diversity within the species.

C. baratii

Phylogenetic tree based on the rrs sequences from 8 strains of C. baratii showed a single major group with high BV among most of the strains (Additional file 2: Figure S12). Three strains, including type strain (S000009597), were selected for framework sequences.

C. chauvoei

Phylogenetic tree based on the rrs sequences from 8 strains of C. chauvoei showed no clear cut grouping as reflected by the wide range of BV, which varied from 246-1000 (Additional file 2: Figure S13). Two strains, including type strain (S000437764), one from either end of the tree were selected for developing the phylogenetic framework.

C. subterminale

The genetic diversity within the rrs sequences of 8 strains of C. subterminale was quite low, as evident from the phylogenetic tree (CSuI-CSuII) (Additional file 2: Figure S14). Two representatives, one from each cluster, including the type strain (S000389872) were found to be sufficient for developing the phylogenetic frame work.

C. tetani

Phylogenetic tree based on the rrs sequences from 8 strains of C. tetani species showed three small groups (CTeI-CTeIII) (Additional file 2: Figure S15). Two strains, including type strain (S000260778), one from each end of the tree were selected for framework sequences to ensure inclusion of maximum diversity within the species in the study.

A total of 56 rrs sequences were short listed to represent the genetic diversity in 15 different Clostridium species (Table 2). A reference phylogenetic tree based on framework sequences (Figure 1) shows clear cut segregation of rrs sequences into separate clades, each consisting of members of C. butyricum, C. chauvoei, C. colicanis, C. kluyveri, C. perfringens and C. tetani. In contrast, C. acetobutylicum was found to be segregated into two groups (i) one close to C. pasteurianum (BV 874) and (ii) another showing high similarity with C. beijerinckii and C. butyricum. High heterogeneity was observed between C. baratii and C. sardiniense. The most heterogeneous group was represented by the strains of C. botulinum. Four different associations were recorded between C. botulinum and (i) C. sporogenes, (ii) C. subterminale, (iii) C. novyi and (iv) a group of C. acetobutylicum and C. butyricum The most interesting feature of the reference phylogenetic framework tree were (i) completely independent placing of C. chauvoei, C. kluyveri, C. tetani and (ii) close relationships among (a) C. acetobutylicum, C. beijerinckii, C. botulinum and C. butyricum (BV 982-1000), (b) C. baratii, C. colicanis and C. sardiniense (BV 840-1000), (c) C. botulinum and C. novyi (BV 703-1000), and (d) C. acetobutylicum, C. botulinum, C. pasteurianum and C. subterminale (BV 874-926).

Table 2 Accession numbers of 16S rDNA sequences of Clostridium species used for generating phylogenetic framework (http://www.ncbi.nlm.nih.gov/ and http://rdp.cme.msu.edu/)
Full size table
Figure 1
figure 1

Phylogenetic tree of 56 16S rDNA framework sequences of different Clostridium spp. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences belonging to C. acetobutylicum, C. baratii, C. beijerinckii, Clostridium botulinum, C. butyricum, C. chauvoei, C. colicanis, C. kluyveri, C. novyi, C. pasteurianum, C. perfringens, C. sardiniense, C. sporogenes, C. subterminale, and C. tetani. Bootstrap values are given at nodes. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image

Validation of species-specific phylogenetic framework sequences

In our quest to validate the phylogenetic framework sequences, different phylogenetic trees were drawn on the basis of 404 rrs sequences belonging to 15 different Clostridium species along with 56 reference sequences. Seven different phylogenetic trees (Figures 2, 3, 4, 5, 6, 7 and 8) very clearly showed that the Clostridium isolates segregated well within their respective phylogenetic framework sequences except C. acetobutylicum (Figure 2). In the case of C. acetobutylicum eighteen rrs sequences segregated well with the type strain (S000628106) and 2 framework sequences and the rest six (including two frame work sequences) formed an independent group, which was quite close to C. beijerinckii along with C. butyricum.

Figure 2
figure 2

Phylogenetic tree of 16S rDNA of Clostridium acetobutylicum and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. acetobutylicum - 24 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Figure 3
figure 3

Phylogenetic tree of 16S rDNA of Clostridium baratii, C. sardiniense and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. baratii - 8, C. sardiniense - 9 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Figure 4
figure 4

Phylogenetic tree of 16S rDNA of Clostridium botulinum, C. sporogenes and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. botulinum - 83, C. sporogenes - 11 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Figure 5
figure 5

Phylogenetic tree of 16S rDNA of 5 Clostridium species and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. botulinum - 45, C. butyricum - 32, C. novyi, - 17, C. beijerinckii - 23, C. subterminale - 8 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Figure 6
figure 6

Phylogenetic tree of 16S rDNA of Clostridium perfringens and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. perfringens - 46 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/). Note: Out of a total of 92 C. perfringens rrs sequences, 46 have been presented here to achieve clarity of presentation. The rest 46 rrs sequences have been presented in Figure 7.

Full size image
Figure 7
figure 7

Phylogenetic tree of 16S rDNA of rest Clostridium perfringens and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of C. perfringens - 46 (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/). Note: Out of a total of 92 C. perfringens rrs sequences, 46 have been presented here to achieve clarity of presentation. The first 46 rrs sequences have been presented in Figure 6.

Full size image
Figure 8
figure 8

Phylogenetic tree of 16S rDNA of 5 different Clostridium species and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of Clostridium kluyveri - 14, C. pasteurianum - 13, C. colicanis - 9, C. chauvoei and C. tetani - 8 each (shown in red, except those used as framework sequences) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image

(i) Phylogenetic relationships of Clostridium sp

The phylogenetic framework of rrs sequences described above was used to check if 179 isolates identified as Clostridium sp. can be classified among these Clostridium species reported to occur with high frequency (Figures 9, 10 and 11). Out of these 179 Clostridium sp., 95 were found to segregate among 15 Clostridium species with a high BV (Figures 9 and 10, Table 3): C. acetobutylicum - 29, C. subterminale - 20 (Figure 9), C. tetani - 7, C. baratii and C. botulinum - 5 each, C. sporogenes/C. botulinum and C. chauvoei - 4 each, C. pasteurianum and C. kluyveri - 3 each, C. botulinum/C. butyricum - 2 and C. perfringens - 1 (Figure 10). A unique case of 11 Clostridium sp. was their association with C. acetobutylicum and C. beijerinckii albeit with low BV of 350-460 (Figure 10). However, subsequent analyses (signatures and in silico RE digestion pattern) revealed that these 11 Clostridium species can be classified as C. beijerinckii. Incidentally, none of the Clostridium sp. was found to belong to C. sardiniense, C. novyi and C. colicanis. A group of 84 rrs sequences of Clostridium sp. could not be grouped among the 15 known Clostridium species (Figure 11). Since these 84 isolates have the potential to be classified as novel Clostridium species, these were also checked for their phylogenetic relationships with Clostridium spp. with low frequency of occurrence.

Figure 9
figure 9

Phylogenetic tree of 16S rDNA of Clostridium sp. and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of Clostridium sp. - 49 (the different isolates could be segregated as C. acetobutylicum - 29 and C. subterminale - 20) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/). Note: Out of a total of 95 Clostridium sp. rrs sequences, 49 have been presented here to achieve clarity of presentation and are shown in red. The rest 46 rrs sequences have been presented in Figure 10.

Full size image
Figure 10
figure 10

Phylogenetic tree of 16S rDNA of rest Clostridium sp. and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of Clostridium sp. - 46 (the different that isolates could be segregated as C. tetani - 7, C. baratii - 5, C. chauvoei and C. pasteurianum - 4 each, C. kluyveri - 3, C. perfringens - 1, C. acetobutylicum/C. beijerinckii - 13 and C. botulinum/C. novyi/C. sporogenes - 9) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/). Note: Out of a total of 95 Clostridium sp. rrs sequences, 46 have been presented here to achieve clarity of presentation and are shown in red. The first 49 rrs sequences have been presented in Figure 9.

Full size image
Figure 11
figure 11

Phylogenetic tree of 16S rDNA of novel Clostridium sp. and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of novel Clostridium sp. - 84 (shown in red) (as these could not be segregated among the 15 Clostridium sp. known to occur at high frequency and rest of the Clostridium spp. known to occur with low frequency (Additional file 6: Figures S18 and S19) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Table 3 Accession numbers of 16S rDNA sequences of Clostridium spp. identified up to species level. http://rdp.cme.msu.edu/
Full size table

(ii) Phylogenetic relationships of Clostridium species with low frequency occurrence

A total of 182 rrs sequences of Clostridium isolates belonging to 80 distinct species and 7 groups which have been classified as swine manure, intestinal bacterium, Eubacterium, etc. (Additional file 1: Table S1) were found to form distinct phylogenetic groups (Figures 12 and 13, Additional file 3: Figure S16). The evolutionary relationship among the 68 rrs sequences of Clostridium spp. (Table 4) which have been reported so far to occur with a low frequency formed a few very distinct and independent groups along with those occurring with relatively high frequency, for example - C. subterminale and C. estertheticum. Although these appear to be close on a single clade but are phylogenetically quite distant as evident by low BV 385. Another clade, which harbored C. subterminale was associated quite closely (BV 703) with Clostridium tunisiense, and red sea bacterium on one clade, Clostridium sulfidigenes, C. thiosulfatireducens and bacterium isolates (BV 852-1000) on a second clade and yet another branch carried swine manure bacterium (BV 961). The two species - C. acetobutylicum and C. beijerinckii reported to occur with high frequency were weakly grouped along with Clostridium diolis (BV 219), C. chromoreductans (BV 398), and C. corinoforum and C. saccharoperbutylacetonicum (BV 643), but showed strong association with C. roseum (BV 998), C. puniceum and C. favososporum (BV 823), (Figure 12). These diverse Clostridium species are quite heterogeneous and reflect their individual identities. However, their grouping together indicates the possibility of their common origin, which was quite high in the case of C. kluyveri branching with C. tyrobutyricum (BV 1000) but moderate in the case of Clostridium ljungdahlii (BV 548). C. botulinum, which is known to be genetically quite heterogeneous, showed close relationship with Eubacterium and rumen bacterium (BV 664-932) on one hand and with C. neonatale on the other (BV 996). Similarly, C. acetobutylicum showed strong association with Clostridium aurantibutyricum, C. felsineum and C. roseum (BV 998-1000). It also branched along with Clostridium collagenovorans albeit with a relatively low BV of 615. C. botulinum and C. novyi shared a branch on the phylogenetic tree along with C. haemolyticum (BV 932-1000). C. tetani is a unique Clostridium sp. which appears to be of independent origin; however, it showed closeness to Clostridium cochlearium and C. tetanomorphum (BV 982-1000). C. chauvoei was found to be close to Clostridium carnis and C. septicum (BV 995-1000), whereas C. pasteurianum showed high homology with C. acidisoli and C. akagii (BV 1000) and less homology with Clostridiales in general (BV 670) (Figure 12). Thirty one rrs sequences of Clostridium isolates belonging to swine fecal, unidentified eubacterium and clostrideaceae bacterium (Additional file 1: Table S1), which did not cluster with Clostridium spp., were found to have an overall low phylogenetic relationship (Additional file 3: Figure S16). The evolutionary relationship among the rest 83 of 182 rrs sequences of Clostridium spp. which have been reported so far to occur with a low frequency formed a few very distinct and independent groups (Figure 13).

Figure 12
figure 12

Phylogenetic tree of 16S rDNA of low frequency Clostridium sp. (segregated) and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of low frequency Clostridium sp. - 68 (shown in red) (which could be segregated among the 15 known Clostridium sp. known to occur at high frequency) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Figure 13
figure 13

Phylogenetic tree of 16S rDNA of low frequency Clostridium sp. (unsegregated) and Framework sequences. A neighbor - joining analysis with Jukes-Cantor correction and bootstrap support was performed on the rrs sequences of low frequency Clostridium sp. - 83 (shown in red) (which could not be segregated among the 15 known Clostridium sp. known to occur at high frequency) along with 56 of phylogenetic framework (Figure 1). Bootstrap values are given at nodes. Sequences marked by filled square are the ones considered as framework in the study whereas type strains are indicated by 'T' as superscript. Values in parentheses are accession numbers (RDP and NCBI) (http://rdp.cme.msu.edu/ and http://www.ncbi.nlm.nih.gov/).

Full size image
Table 4 Phylogenetic relationships of Clostridium spp. (low frequency) with Frame work organisms supported by RE patterns and signaturesa
Full size table

(iii) Phylogenetic relationships of novel Clostridium species

Out of 179 isolates of Clostridium sp., 84 rrs sequences which have the potential to be classified as novel were found to be represented by 56 groups (Additional file 4: Table S2). A quite low BVs among them implies high genetic heterogeneity (Additional file 5: Figure S17). To further verify if these 84 Clostridium sp. are of independent origin, their phylogenetic relationships were checked with Clostridium spp. reported to occur with low frequency (Additional file 4: Table S3). It can be seen from the phylogenetic trees (Additional file 6: Figures S18 and S19) that they are quite independent as supported by low BV. The possibility of assigning these 84 rrs sequences to novel Clostridium species was further supported by subsequent analyses - nucleotide signatures and in silico RE digestion patterns.

In silico restriction enzyme activity

Our recent studies [3, 49] have revealed a high level of latent characteristics within rrs sequences based on in silico digestion with 14 Type II REs. These in silico digestions have resulted in predicting clear cut fragment lengths in the rrs with six REs: AluI, BfaI, DpnII, HaeIII, RsaI and Tru9I [3, 49]. In the present study, 7 REs - AluI, BfaI, DpnII, HaeIII, RsaI, SmaI and Tru9I were found effective in drawing meaningful conclusions (Figures 14, 15, 16, 17, 18, 19 and 20). RE sites for BamHI, NotI, PstI and SacI were almost completely absent from the rrs sequences of all the strains of 15 Clostridium spp. and thus proved to be ''non"-cutters (Additional file 7: Table S4). Another group, which did not permit clear cut conclusions to be drawn was composed of the following three REs: EcoRI, HindIII, and NruI, with one RE site each (leading to two fragments of varied lengths) in most of the Clostridium strains.

Figure 14
figure 14

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: AluI. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences.

Full size image
Figure 15
figure 15

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: BfaI. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences.

Full size image
Figure 16
figure 16

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: DpnII. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences. NA indicates, not applicable.

Full size image
Figure 17
figure 17

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: HaeIII. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences. NA indicates, not applicable.

Full size image
Figure 18
figure 18

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: RsaI. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences.

Full size image
Figure 19
figure 19

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: SmaI. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences. NA indicates, not applicable.

Full size image
Figure 20
figure 20

In silico Restriction Enzyme activity in 16S rDNA sequences of Clostridium spp.: Tru9I. aFrequency of organisms showing this RE digestion pattern. Symbol (Filled square) indicates RE site in the rrs sequences.

Full size image

AluI

In silico digestion of rrs sequences of 15 different Clostridium spp. with RE - AluI resulted in 4-8 fragments. The segregation of Clostridium species was evident at two levels - (i) intra-species and (ii) inter-species. Intra-species variation was observed among C. pasteurianum - rrs sequences, which were found to possess 4-7 RE - AluI digestion sites. Two fragments of 186-610 nts were found in all the four groups. At the 5' end, upstream of the 186 nts fragments, the RE sites were not detectable such that the two fragments of 12 and 49 nts seem to have merged together to generate a 61/62 nts long fragment. Similarly on the 3' end, the merger of the two fragments of 207 and 142 nts resulted in a larger nucleotide fragment of 349 nts (Figure 14).

Intra-species level segregation was also evident in case of C. botulinum. Here, the 120/128 rrs sequences were found to be distributed into 5 groups of 5, 10, 21, 22 and 62 sequences each. The largest group of 62 sequences of C. botulinum had 5 RE - AluI digestion sites resulting in the following fragments between them: 170-795-265-85 nts. The next group of 21 sequences of C. botulinum was quite close to the former. However, it lacked a RE site at the 5' end, such that one could not predict the presence of 170 nts sized fragment. The two smaller groups of 10 and 22 rrs sequences of C. botulinum had 6-7 RE - AluI sites each. These two groups shared the following fragment sizes and the order of their occurrence: 566-228/230-205/207-57-109 nts. The later group of 22 of C. botulinum sequences had an additional RE site which led to the prediction of a 170 nts fragment at the 5' end of this gene. It may be remarked that these four groups differed because of the absence/presence of RE sites (Figure 14).

Variability in the RE - AluI sites among rrs sequences of C. acetobutylicum was also quite high. Out of 24 rrs sequences, 17 of them showed two RE - AluI sites resulting in clear cut fragment of 797 nts. On the other hand, a small group of 3 sequences had additional RE sites, which resulted in additional fragments of 170 nts at 5' end and 266 nts at 3' end. C. butyricum 3/32 and 3/24 C. acetobutylicum rrs sequences show high homology in their RE - AluI digestion pattern. C. baratii and C. sardiniense had similar RE patterns leading to the following order of fragments - 186-604/605-206-57-109 nts. The situation of C. colicanis was unique as it resembled C. baratii and C. sardiniense except in the absence of a RE - AluI site resulting in the merger of 206-57 nts fragments into a single fragment of 264 nts. On the other hand, C. colicanis showed high resemblance in its in silico RE digestion pattern to most C. perfringens (66/92 sequences) (Figure 14).

A perusal of all the numbers and the order of occurrence of the fragments predicted to be generated by RE -AluI in silico digestion enabled us to segregate the following: (i) C. acetobutylicum (17/24 sequences), (ii) C. chauvoei, (iii) C. colicanis, (iv) C. kluyveri, (v) C. pasteurianum, (vi) C. perfringens, and (vii) C. subterminale (Additional file 4: Table S2). C. sporogenes/C. botulinum (62/128 sequences) are indistinguishable here but can be distinguished on the basis of other REs (Figure 14).

BfaI

With RE - BfaI in silico digestion of rrs sequences of 15 Clostridium species predicted the presence of 5-8 fragments of distinct nucleotide sizes. Intra-species level variability among RE - BfaI generated fragments was evident in C. acetobutylicum and C. botulinum. RE digestion of C. acetobutylicum rrs sequences resulted in segregating them into two populations - (i) 16/24 sequences and (ii) 4/24 sequences. These two populations of C. acetobutylicum showed high similarity in RE pattern towards their 3' end: 146-159/160-161/164-68-126 nts fragments. They differed by the presence of two additional RE - BfaI sites, which resulted in two more fragments of 378-29 nts towards the 5'end of the 4/24 rrs sequences of C. acetobutylicum (Figure 15).

On the basis of RE - BfaI in silico digestion of rrs of C. botulinum, high genetic heterogeneity was evident among the 128 isolates. The four groups of C. botulinum were composed of as follows: (i) 57, (ii) 27, (iii) 9, and (iv) 34/128 sequences. All the four groups of C. botulinum could be easily distinguished on the basis of their RE - BfaI digestion fragment lengths. The first group of 57/128 C. botulinum had a unique RE - BfaI digestion pattern: 195-20-163-29-146-185-331 nts in their rrs. Second group of 27/128 C. botulinum had similarity only with C. sporogenes for their RE - BfaI digestion pattern: 378-29-146-185-331 nts in their rrs. It differed from the first group due to merger of three fragments 195-20-163 at the 5' end in to a single fragment of 378 nts in the latter. The third group of 9/128 C. botulinum were found to possess an in silico RE - BfaI digestion pattern: 378-29-146-158-231-126 nts in their rrs sequences. It was similar to that observed in C. novyi and C. tetani. The fourth cluster of C. botulinum (34/128 sequences), did not show any genetic variability with respect to RE - BfaI in silico activities with respect to those observed in C. beijerinckii, C. butyricum, C. sardiniense, C. subterminale, and C. acetobutylicum (4/24 sequences) (Figure 15). This feature implies that these species might have a common phylogenetic origin.

On the basis of inter-species variation in the RE - BfaI sites on the rrs the following Clostridium species can be distinguished from the rest of the species: C. acetobutylicum (16/24 sequences), C. baratii, C. botulinum (57/128 sequences), C. chauvoei, C. colicanis, C. kluyveri, C. pasteurianum and C. perfringens (Additional file 4: Table S2).

DpnII

In the case of in silico digestion of rrs sequences with RE - DpnII, a large intra-species variation was evident in the cases of 9 different Clostridium species: (i) C. acetobutylicum, (ii) C. baratii, (iii) C. beijerinckii, (iv) C. botulinum, (v) C. butyricum, (vi) C. kluyveri, (vii) C. novyi, (viii) C. pasteurianum and (ix) C. perfringens. On the other hand, interspecies variation was evident in the cases of C. chauvoei, C. colicanis, C. sardiniense, C. subterminale and C. tetani. In spite of high homology among different Clostridium species in terms of the order, number and length of the fragments generated by RE - DpnII, the following seven Clostridium species: C. butyricum, C. chauvoei, C. colicanis, C. kluyveri, C. subterminale and C. tetani could still be easily distinguished (Figure 16, Additional file 4: Table S2). On the other hand, in spite of certain unique features in C. acetobutylicum and C. perfringens, intra-species variation did not allow complete segregation and may need supplementary information from other REs.

HaeIII, RsaI, SmaI and Tru9I

In silico RE - HaeIII, RsaI, SmaI and Tru9I digestion of rrs of 15 Clostridium species resulted in predicting the presence of 1-12 sites. Due to high variation in the characteristics of the nucleotide fragments within and among Clostridium species, a unique RE digestion pattern was observed in in the following cases: (i) C. chauvoei, C. colicanis, C. kluyveri and C. tetani with HaeIII (Figure 17, Table 5), (ii) C. novyi with RsaI (Figure 18, Table 5), (iii) C. colicanis with SmaI (Figure 19, Table 5) and (iv) C. baratii, C. novyi and C. sardiniense with Tru9I (Figure 20, Table 5). It implies limited scope for exploiting these REs for identification purposes.

Table 5 Restriction enzymes with unique in silico digestion pattern of 16S rDNA of different Clostridium species
Full size table

In silico RE digestion of rrs of 15 Clostridium species enabled us to identify unique features (order of occurrence and length, nts) of this gene which can be used as bio-markers (Table 5). C. butyricum, C. perfringens and C. sardiniense could be distinguished from all other Clostridium spp. on the basis of the in silico digestion with RE - DpnII, BfaI and Tru9I, respectively. C. baratii, C. novyi, C. pasteurianum, C. subterminale and C. tetani could be distinguished from the rest of the species on the basis of in silico digestion pattern of their rrs with all REs except SmaI. The most interesting features seem to reside in the rrs sequences of C. chauvoei, C. colicanis and C. kluyveri, which possessed unique pattern of active sites for 4-5 different REs such that any one of these was sufficient to identify them. In the cases of C. acetobutylicum, C. beijerinckii, C. botulinum and C. sporogenes, no unique pattern could be detected. For the groups comprised of (i) C. botulinum - C. sporogenes and (ii) C. acetobutylicum - C. beijerinckii, we could not detect any unique RE digestion patterns and may need to be compared with a combination of REs. The close phylogenetic relationship between C. botulinum and C. sporogenes has been reported previously also [51]. Since they differ on the basis of their phylogenetic relationship (Figures 1, 4 and 5), it appears that C. beijerinckii is a part of population comprised of C. acetobutylicum and similarly C. sporogenes is a part of isolates belonging to C. botulinum. Alternatively, the issue can be resolved by creating new species by subdividing C. acetobutylicum and C. botulinum.

Validation of framework sequences by in silico RE activity on 16S rDNA sequences of organisms identified as Clostridium sp

The second level of investigation after segregating Clostridium sp. (95 isolates) (Table 3) on the basis of phylogenetic frame work was to validate them on the basis of unique RE digestion patterns deduced from the rrs sequences of 15 known Clostridium spp.

C. acetobutylicum

In the case of C. acetobutylicum, two distinct populations were observed in the initial phylogenetic tree which was based on rrs sequences of 24 isolates of this species. A similar trend of two major groups was very clearly evident also on the basis of RE - AluI in silico digestion patterns: (i) 186-611-264 nts and (ii) 798 nts. The 29 Clostridium sp., which we could classify as C. acetobutylicum on the basis of phylogenetic framework (Table 3) were further supported by RE digestion patterns of their rrs sequences. Here also we could observe that the total population was composed of 2 different groups, consisting of 12 and 17 organisms each. The two groups segregated due to single nucleotide polymorphism (SNPs) at RE sites. Modifications in RE sites can arise primarily through single nucleotide insertion, deletion and/or substitution [52].

C. beijerinckii

High homology between C. beijerinckii and C. acetobutylicum was evident due to similarities in in silico digestion pattern generated by the action of most of the REs employed in this study especially - AluI, BfaI, HaeIII, RsaI, SmaI and Tru9I. However, it was "unique" in its RE pattern with respect to DpnII. The rrs sequences of C. beijerinckii on digestion with RE - Tru9I showed a pattern of 114-40-363-26-251-34-52-56-76-76-117 nts, which is quite similar to that recorded with a proportion of C. acetobutylicum population. However, these could be distinguished on the basis of RE - DpnII in silico digestion fragments 287 or 287-1196 nts found in rrs sequences of C. acetobutylicum, but were absent in these 13 newly identified C. beijerinckii isolates. Thus, we recommend the usage of a combination of DpnII and Tru9I to validate the identification of 13 Clostridium sp. as C. beijerinckii (Table 3).

C. subterminale

The phylogenetic framework based on 8 rrs sequences of C. subterminale allowed identification of 2 unique in silico RE digestion patterns with AluI and DpnII. The validation of 20 Clostridium sp. which could be classified as C. subterminale through phylogenetic framework (Table 3) was done through in silico RE digestion of their 16S rDNA with both the REs. Thus in silico digestions supported the segregation of these Clostridium sp. as C. subterminale. Two of three populations were distinct on the basis of the uniqueness of their fragment characteristics - number, length and order of occurrence (i) RE - AluI a) 345-450-347 nts and b) 170-796-207-143 nts and (ii) RE - DpnII a) 285-565-632 and b) 273-565 nts. However, the third group of isolates identified as C. subterminale was intermediate to the other two populations with respect to the patterns obtained with both the REs - AluI and DpnII.

C. botulinum and C. sporogenes

C. botulinum sp. produces seven serotypes (A through G) of botulinum neurotoxins (BoNTs) [16, 20]. It shows 4 distinct lineages (I-IV) [33, 50], where good correlations have been found between their phenotypes and phylogenetic classification: (i) Group I - proteolytic C. botulinum types A, B and F which are close to C. sporogenes, (ii) Group II consists of saccharolytic types B, E and F, (iii) Group III includes types C and D and C. novyi type A and (iv) Group IV has Clostridium argentinense (C. botulinum type G) and C. subterminale[32]. Based on DNA S1 nuclease analysis, C. botulinum was found to be genetically close to C. sporogenes at the species level, however, the later was proposed to be conserved for non-toxigenic strains [51].

In a phylogenetic tree based only on the framework sequences, C. botulinum and C. sporogenes were found to occur together on a single clade. A comparison of the different in silico RE digestion patterns of rrs of C. sporogenes and C. botulinum showed exact homology in the cases of REs - AluI, BfaI, HaeIII, RsaI, Tru9I and SmaI except DpnII (Figures 14, 15, 16, 17, 18, 19 and 20). In the case of DpnII, C. sporogenes was segregated into 2 populations, where one of them showed exact homology with C. botulinum and the other one was different and unique in comparison to all other Clostridium species. It may not be too inappropriate to suggest that C. sporogenes is perhaps a sub-species of C. botulinum or it may find its appropriate place if C. botulinum can be reclassified as 4 different sub-species. This proposal finds support from the RE pattern of sequences of Clostridium species (S000357670 and S000357671) which had been classified as C. sporogenes with the help of phylogenetic frame work. Nine Clostridium sp. which appeared close to C. botulinum, C. sporogenes and C. novyi (Table 3) were found to be similar to each other with respect to the in silico activities of one or more REs - Tru9I, HaeIII, AluI and BfaI.

Other Clostridium spp

The rest 24 Clostridium sp. identified up to species level on the basis of phylogenetic framework (Table 3), could be supported also by high similarity in the in silico RE digestion pattern of their rrs: (i) C. tetani - HaeIII, (ii) C. chauvoei - AluI and BfaI, (iii) C. kluyveri - AluI, (iv) C. pasteurianum and C. perfringens - BfaI and (v) C. baratii - Tru9I and BfaI (Figures 14, 15, 17 and 20).

The phylogenetic relationships observed among Clostridium spp. reported to occur with low frequency formed quite distinct clusters. Sixty eight rrs sequences belonging to 37 species were found to clusters along with 14 Clostridium spp. used for framework analysis. The close relationship among the different Clostridium spp. was validated by comparing their in silico analysis RE digestion pattern of their rrs (Table 4), which matched invariably with the unique RE pattern of known Clostridium spp. It is perhaps the first report where in 37 different Clostridium species have been segregated among 14 groups based on phylogeny and in silico RE digestion pattern of their rrs.

Nucleotide signature analysis - frequency and distribution pattern

To further validate the segregation of Clostridium sp. done on the basis of phylogenetic frame work, we investigated them for the presence of nucleotide signatures (30 nts) deduced from isolates of the 15 known Clostridium spp. The validity of the nucleotide signatures was done through the frequency of occurrence of the reference organism among the top 10 hits (BLAST).

The sequences of 8-128 data sets submitted group-wise to MEME (Multiple EM for Motif Elicitation) program http://meme.sdsc.edu/meme4_5_0/cgi-bin/meme.cgi revealed ten signatures (30 nts) for each species. Regular expression diagram of signatures (nts) (Additional file 8: Figures S20,S21,S22,S23,S24,S25,S26,S27,S28,S29, S30,S31,S32,S33,S34,S35 and S36) and the frequency of occurrence of motifs obtained through MEME for different taxonomic groups have been presented with reference to 15 Clostridium spp. (Additional file 9: Tables S5-S20).

Out of the 10 signatures (30 nts), identified in each of the 15 different Clostridium spp., unique signatures in the rrs sequences (Table 6) were found in the following cases: C. beijerinckii and C. sporogenes - 1 each, C. butyricum - 2, C. colicanis - 4, C. kluyveri, C. pasteurianum and C. perfringens - 5 each, C. chauvoei - 6 and C. tetani - 9. However, using similar strategy, no unique signatures could be identified in the cases of (i) C. baratii and C. sardiniense, (ii) C. botulinum, C. novyi and C. subterminale and (iii) C. acetobutyilicum. These latter 6 Clostridium spp. have been dealt with separately as detailed below.

Table 6 Characteristics of "unique" nucleotide signatures for rrs sequences of 15 different Clostridium spp
Full size table

C. baratii and C. sardiniense

Out of 10 signatures deduced among the rrs sequences of 8 isolates of C. baratii only two - M3 and M6 could be designated as "unique". In fact, these two signatures were absent from all other Clostridium spp. except C. sardiniense. Similarly, in C. sardiniense 6 out of 10 nucleotide signatures were found to be "unique" to it but were present in the overall rrs sequence of C. baratii. We may conclude that although these two Clostridium spp. did not show any exact resemblance in their signatures, however, they shared 19 out of 20 signatures in their overall rrs sequence. Thus it is difficult to use them for segregating the two Clostridium species. This observation of high homology is further supported by the fact that C. baratii and C. sardiniense are phylogenetically present on a single clade and had high BV in the range of 654-999 (Figure 1). In spite of such high similarity they could be separated on the basis of nucleotide signature (M2) of C. baratii - 5-TATTGTTAGTTGCTACCATTTAGTTGAGCA-3' (which incidentally was present in most other Clostridium species).

C. botulinum and C. acetobutylicum

Organisms belonging to C. botulinum were segregated phylogenetically into 4 groups. Distinct nucleotide signatures could be detected on segregating 128 rrs sequences of C. botulinum isolates into two groups - i) 83 equivalent to CBoI of the phylogenetic tree (Additional file 2: Figure S1) and ii) 45 representing (CBoII-CBoIV) (Additional file 2: Figure S1). The first group of 83 isolates had 5 unique nucleotide signatures, which could be used to distinguish them from all other Clostridium species except C. sporogenes isolates. On the other hand, 45 rrs sequences of isolates belonging to CBoII-CBoIV could not be distinguished on the basis of the 10 nucleotide signatures deduced from their rrs sequences. Previous reports of screening with four REs and 30 primers combinations revealed that HindIII and HpyCH4IV are effective in demonstrating extensive diversity among C. botulinum type E, whereas C. botulinum Group I indicted low genetic diversity [50, 53, 54]. A similar scenario was evident also in the case of C. acetobutylicum. It prompted us to conclude that C. acetobutylicum and C. botulinum perhaps represent ancestral Clostridium species.

Validation of framework sequences for Clostridium sp: nucleotide signatures

To validate the categorization and classification of 95 rrs sequences of Clostridium sp. belonging to 13 Clostridium spp. (Additional file 9: Table S21), we looked up for 10 nucleotide signatures of the reference framework sequences in them. Of these 10 signatures, the emphasis has been laid on i) the signatures unique to a particular Clostridium sp. and ii) presence of rest of the signatures. On the basis of these two criteria, we may conclude that Clostridium spp.: C. subterminale - 16/20 sequences, C. tetani - 2/7 sequences, C. perfringens - 1/1 sequence, C. pasteurianum - 4/4 sequences, C. baratii and C. sardiniense - 5/5 sequences showed quite close match with the reference framework sequences. On the other hand, C. beijerinckii had only one unique signature which was absent in 13/13 sequences of Clostridium sp. but due to the presence of all other signatures we may conclude that they are quite close to the reference framework sequences. Similarly, C. acetobutylicum showed that 29/29 Clostridium sp. were quite similar to the reference sequences. This implies that in these cases signature analysis supports the framework sequence based analysis. However, in the following Clostridium spp., the identified Clostridium sp. did not show much homology with respect to the signatures found in the reference framework sequences: (i) C. botulinum, (ii) C. chauvoei, C. kluyveri, C. novyi and C. sporogenes.

Discussion

Molecular techniques have added new dimensions and sensitivity to the phylogeny and taxonomy of organisms. The most remarkable and widely acknowledged are the unique features of highly conserved gene - rrs[55]. Based on rrs sequencing Collins et al. revealed the polyphyletic nature of Clostridium sensu stricto[35]. Since then there have been many additions and rectifications in the classification of Clostridium[56, 57]. There are two major issues with respect to Clostridial taxonomy and phylogeny. First is the high genetic diversity - GC content, which varies from 24 mol% (C. perfringens) [47] to 58 mol% (C. barkeri) [48]. It is perhaps too wide a range to encompass a single genus [42] and may need revision. Secondly, Clostridium comprises a very heterogeneous group of bacteria representing more than 110 spp., with a large proportion having only 1-5 strains. The question is whether we should split the genus Clostridium (especially C. acetobutylicum and C. botulinum) and/or we may merge quite a few of its species (small sized). In certain cases of recent speciation events, where rrs is not sufficient to discriminate phylogenetically close relatives, one resorts to DNA-DNA reassociation technique [58]. In spite of an ever increasing knowledge about various conserved genes for distinguishing microbes [3], rrs continues to be quite effective in segregating and even in reclassifying quite a few microbial isolates as new genera and species. A large number of organisms reported initially as Bacillus species [59] have been reclassified as genus Geobacillus[60], Sporosarcina[61], or Marinibacillus marinus[62]. In fact, many species of Clostridium[35] have also been reclassified as new genera, such as Sedimentibacter[63]. The three Clostridium spp. - C. carboxidivorans, C. drakei, and C. scatologenes, which showed high (99.7-99.8%) sequence similarity for their rrs sequences [64] were later found to be distinct species based on their DNA-DNA reassociation values [58]. In spite of such high volume of data available on rrshttp://rdp.cme.msu.edu/, the fact remains that it needs support from other genes to arrive at authentic conclusions. Does it imply that rrs is already at its dead end? or Do we need to look deeper into the latent and as yet un-explored features of this wonderful gene?

Our recent studies on the diversity of Bacillus spp. [3] and Stenotrophomonas isolates [49] revealed certain very interesting 'latent' features of rrs. We could designate rrs of certain isolates as phylogenetic framework sequences. These were supported by species specific unique signatures and in silico RE patterns. Together, all these features of rrs allowed us to (i) classify Bacillus sp., reveal novel lineages and distinguish the two subgroups of B. subtilis[3] and (ii) evaluate the biodiversity of Stenotrophomonas isolates [49]. These studies indicate towards unexplored features within the rrs to define and circumscribe the genus Clostridium sensu stricto[52].

The species-specific phylogenetic framework developed here proved instrumental as a tool to deducing the limits of genetic diversity within 15 Clostridium spp. The validity of the phylogenetic framework sequences was established by clear cut segregation of members of a given Clostridium sp. against a total of 56 rrs sequences (Table 2, Figure 1). High genetic heterogeneity was recorded among members of C. acetobutylicum and C. botulinum in our study. It has been previously supported by significant differences in their physiological and genetic characteristics [65, 66]. Within the 56 framework sequences of 16S rDNA of 15 Clostridium spp. based phylogenetic tree, we observed the 5 framework sequences of C. acetobutylicum to segregate in to two groups: (i) 2 sequences were close to C. beijerinckii and C. butyricum and (ii) 3 were close to C. pasteurianum (Figure 1). The findings in our study find support from previous phylogenetic studies based on concatenated sequences of 37 proteins, which showed that C. acetobutylicum clusters along with C. beijerinckii in addition to other species such as C. botulinum, C. kluyveri, C. novyi, C. perfringens, and C. tetani with 100% BV [52]. In fact, 24 strains belonging to C. acetobutylicum were reclassified as either C. beijerinckii (Sub group 3) or C. saccharobutylicum[25, 46]. Here, we observed that some strains of C. acetobutylicum are phylogenetically close to C. felsineum, however DNA-DNA hybridization study [34] shows them to be distinct species.

Similarly, C. botulinum is well known to for its segregation on the basis of a wide range of neurotoxins produced by them [16, 28, 67]. Here in our study, C. botulinum 16S rDNA sequences showed 4 clear grouping in accordance with those reported in literature [32]: (i) C. sporogenes, (ii) a group of C. butyricum and C. acetobutylicum, (iii) C. novyi and (iv) C. subterminale. This close relationship among the species of Clostridium cluster I (C. botulinum, C. acetobutylicum and C. novyi) finds support from the presence of 4 amino acid insert in a highly conserved region of Gyrase A protein, which could not be traced in any other Clostridia or other group of bacteria [52]. High phylogenetic closeness based on rrs has been reported between C. botulinum and C. sporogenes[18, 68] and C. botulinum type G and C. subterminale[32]. Three distinct lineages of C. botulinum were confirmed by AFLP analysis of their rrs sequences [32]. In the present work, other close relationships were observed among (i) C. baratii, C. colicanis and C. sardiniense (BV 840-1000) and (ii) C. botulinum and C. novyi (BV 703-1000). A complete molecular typing of C. botulinum is needed for correct classification of this species [50]. In depth analysis of rrs reported here can in fact provide enough supplementary information to counter multilocus sequence typing recommended for elucidating phylogeny of C. botulinum[28]. On the basis of the evidences provided here (Additional file 2: Figure S1, Figures 1, 14, 15, 17, 18, 19 and 20), we further propose here that C. botulinum needs to be reclassified in to 4 different (sub)-species, as is generally believed by many researchers [16]. The validity of Framework sequences was further confirmed in the case of the classification of 95 Clostridium sp. up to species level, by in silico RE digestions especially - AluI, BfaI and DpnII (Figures 14, 15, 16, 17, 18, 19 and 20).

Our present analysis which shows a close phylogenetic relationship between C. subterminale and certain Clostridium spp. such as C. argentinense, C. estertheticum, and C. thiosulfatireducens (Table 4), finds support from the previous reports of Spring et al., [69]. Although C. tetanomorphum has been observed to be close to C. subterminale and C. thiosulfatireducens with a 97% BV [70] but we found it to be close to C. tetani (BV 982). Unlike previous studies [17, 42, 71], we could demonstrate closer relationships among (i) C. puniceum, C. beijerinckii and C. butyricum and (ii) C. haemolyticum and C. novyi with a much smaller number of rrs frame work sequences.

Quite a few species of other Clostridial genera in Cluster I belong to Eubacterium moniliforme, E. terantellae, Ancerobacter polyendosporus, Sarcina ventriculi and S. maxima[42]. As far as Eubacterium is concerned, several Clostridium spp. have been reclassified within newly created genera: Moorella and Oxobacter[45] or within Eubacterium itself [35]. C. barkeri, a member of Cluster XV was found close to Eubacterium limosum with a level of similarity as high as 95% [35]. In Cluster XVI: phylogenetically significant association was observed among the 3 species - Eubacterium biforme, C. innocuum, and Streptococcus pleomorphus[35]. In the present study, Eubacterium combesii, E. moniliforme and E. budayi were found to cluster with C. botulinum, C. baratii and C. sardiniense, respectively.

High similarity in the nucleotide signatures within rrs sequences of C. acetobutylicum and C. botulinum prompt us to speculate that C. acetobutylicum and C. botulinum might have common ancestor. In fact, these two Clostridium spp. shared 34-35% of the total rrs sequences. Similarly, we have observed that certain signatures are common to almost all the 15 Clostridium spp. which were used for developing framework sequences. It may be remarked that at least 18 common signatures (30 nts) were distributed along the length of the rrs (Additional file 10: Table S22). The most unique features were the positions of signatures with maximum frequency, which varied from 394-1170, implying higher susceptibility of the flanking regions to genetic modifications. In fact, such large core regions were also observed in Bacillus sp. [3]. It can be concluded that about 540 nts were shared by different Clostridium spp. and may represent those sequences which are highly conserved during evolution and may be designated as representatives of the genus Clostridium.

Conclusions

rrs sequences have certain features such as nucleotide signatures (25 to 30 nts long) and unique RE digestion patterns which can be exploited for (re)defining bacterial taxonomy and phylogeny. This approach enabled us to develop species specific phylogenetic framework. These genetic tools allowed us to categorize Clostridium isolates which have been classified presently up to genus level into (i) well defined Clostridium spp. and (ii) as novel species. It also holds promise to reduce the number of Clostridium species represented by small populations. This integrated approach is quite sensitive and can be easily extended as a molecular tool for diagnosis of microbes used in food industries and health departments.

It appears as if we are reiterating (today in 2010) the statement made by Prof. C.L. Oakley, almost 6 decades ago: "Little is gained by multiplying genera to include rare or imperfectly-described forms, though increase in our knowledge may justify subdivision of the genus (Clostridium) in the future." [72].

Methods

Sequence data

The rrs sequences (> 1200 nts) chosen for this study corresponds to a total of 765 isolates belonging to the genus Clostridium (from RDP/NCBI sites: http://rdp.cme.msu.edu/; http://www.ncbi.nlm.nih.gov/. These included sequences belonging to isolates of C. botulinum - 128, C. perfringens - 92, C. butyricum - 32, C. acetobutylicum - 24, C. beijerinckii - 23, C. novyi - 17, C. kluyveri - 14, C. pasteurianum - 13, C. sporogenes - 11, C. colicanis and C. sardiniense - 9 each, C. baratii, C. chauvoei, C. subterminale and C. tetani - 8 each, Clostridium sp. - 179 and Clostridium spp. having members with low frequency (Table 1). The first fifteen sets of Clostridium species consisting of 404 rrs sequences were used as the reference species set for generating species specific: (i) phylogenetic framework, (ii) signatures and (iii) in silico RE digestion pattern. These tools were then used for identifying (i) Clostridium sp. up to species level and (ii) developing phylogenetic relationships among the other Clostridium species, which are represented by relatively smaller number of isolates.

Phylogenetic Analyses

For phylogenetic analyses, rrs sequences of each of these 15 species were assembled and aligned using the multiple alignment program - ClustalX version 2.0.12 [73]. To estimate evolutionary distance, pairwise distances between all species were calculated with the DNADIST of the PHYLIP 3.69 package [3, 74]. The resultant distance matrix was then used to draw a neighbour joining tree with the program NEIGHBOR. The program SEQBOOT [74] was used for statistical testing of the trees by resampling the dataset 1000 times. The trees were viewed through TreeView Version 1.6.6 [75], Phylodraw [76], MEGA [77] (Additional file 2: Figures S1,S2,S3,S4,S5,S6,S7,S8,S9,S1,S11,S12, S13,S14 and S15). For each of these 15 data sets, a 'guide' tree was made and 2-10 representative sequences were selected to develop framework sequences for the rest of the study. Thus a reference set of 56 rrs sequences was selected and regarded these as likely candidates that could give information about the organismal phylogeny (Figure 1). These framework sequences were validated by using 404 rrs sequences from 15 Clostridium species (Figures 2, 3, 4, 5, 6, 7 and 8). Subsequently, these 56 framework sequences were used to reclassify the Clostridium sp. (Figures 9, 10 and 11) and show phylogenetic relationships among other Clostridium spp. (with small populations) (Figures 12 and 13, Additional file 3: Figure S16).

In order to establish whether the 84 rrs sequences of Clostridium sp. which could not be clearly segregated among 15 Clostridium species (Figure 11), they were checked for their phylogenetic relationship with 80 Clostridium spp. (consisting of small populations) in the following manner. Initially, these 84 sequences (Figure 11) were reduced to 56 representative sequences on the basis of their distribution on the phylogenetic tree (Additional file 5: Figure S17). Two phylogenetic trees of 28 sequences each were drawn (Additional file 6: Figures S18 and S19) along with 83 sequences of 80 different Clostridium sp. (Additional file 4: Table S3).

Restriction Enzyme Analyses

A total of 14 Type II REs http://rebase.neb.com/rebase/rebase.html were considered for these analyses: (i) 4 nts cutters - AluI (AG↓CT), BfaI (C↓TAG), DpnII (↓GATC), HaeIII (GG↓CC), RsaI (GT↓AC), Tru9I (T↓TAA), (ii) 6 nts cutters - BamHI (G↓GATCC), EcoRI (G↓AATTC), HindIII (A↓AGCTT), NruI (TCG↓CGA), SacI (GAGCT↓C), SmaI (CCC↓GGG), PstI (CTGCA↓G) and (iii) 8 nts cutter - NotI (GC↓GGCCGC) [3]. All the 15 Clostridium species considered for developing the phylogenetic framework were checked for all the 14 Type II RE using the online software: Restriction Mapper Version 3 http://www.restrictionmapper.org/. Sequences (one at a time) were entered in the restriction mapper site by removing all non base characters, results obtained were analyzed and consensus pattern was determined for each species depending upon its frequency of occurrence and the nucleotide fragment lengths of the sequences. The different Clostridium sp. and Clostridium spp. with low frequency of occurrence were checked for these consensus RE patterns.

Clostridium Species-Specific Signatures

MEME has been used here for searching novel motifs or signatures in sets of biological sequences. MEME works by searching for repeated, ungapped sequence patterns that occur in the DNA or protein sequences [78, 79]. MEME searches can be performed via the web server http://meme.sdsc.edu/meme4_4_0/intro.html and its mirror sites [79]. To successfully discover motifs with MEME, it is necessary to choose and prepare the input sequences carefully. Ideally, the sequences should be, 1000 base pairs long [80]. In our analysis, sequences of each of the 15 data sets in FASTA format were submitted group wise in MEME program Version 4.4.0 http://meme.sdsc.edu/meme4_4_0/intro.html. In order to obtain maximum number of motifs in our sequences, we modified default settings from 3-10 motifs. The default value of motif widths, set between 6 (minimum) and 50 (maximum) were modified and re-set between 25 and 30, respectively. We used default setting zero or one motif per sequence to get the occurrence of single motif which is distributed among the sequences. MEME search stops when this number of motifs has been found, or when none can be found with E-value less than 10000 http://meme.sdsc.edu/meme4_4_0/intro.html. Since, it posed a restriction of using 60000 characters at a time, we also used its stand alone format (MEME program Version 4.3.0) for our study for generating nucleotide signatures for data sets having greater than 60000 characters (for example in the cases of C. botulinum and C. perfringens, (Additional file 8: Figures S20,S21,S22 and S23, Additional file 11: Tables S23,S24,S25 and S26), where a large data was compressed for easy comprehension. Each of the 10 signatures (30 nts) (Additional file 9: Tables S5-S20) was checked through BioEdit [81] for its frequency of occurrence among all the sequences of all the 15 Clostridium spp. and the ones with highest frequency that did not appear in other Clostridium spp. were considered as unique to this species. In the case of 128 rrs sequences of C. botulinum, we could not detect any unique signatures. We thus divided the C. botulinum isolates into two groups of 83 and 45 on the basis of their distribution on the phylogenetic (guide) tree (Additional file 2: Figure S1). The rrs sequences of these two groups were then analyzed through MEME for identifying unique nucleotide signatures. These unique motifs were used as query sequence to validate the Clostridium spp. segregated through phylogenetic framework. The MEME result gives a graphic representation and Motifs in regular expression format.