The genomic basis for the evolution of a novel form of cellular reproduction in the bacterium Epulopiscium.

BACKGROUND: Epulopiscium sp. type B, a large intestinal bacterial symbiont of the surgeonfish Naso tonganus, does not reproduce by binary fission. Instead, it forms multiple intracellular offspring using a process with morphological features similar to the survival strategy of endospore formation in other Firmicutes. We hypothesize that intracellular offspring formation in Epulopiscium evolved from endospore formation and these two developmental programs share molecular mechanisms that are responsible for the observed morphological similarities.
RESULTS: To test this, we sequenced the genome of Epulopiscium sp. type B to draft quality. Comparative analysis with the complete genome of its close, endospore-forming relative, Cellulosilyticum lentocellum, identified homologs of well-known sporulation genes characterized in Bacillus subtilis. Of the 147 highly conserved B. subtilis sporulation genes used in this analysis, we found 57 homologs in the Epulopiscium genome and 87 homologs in the C. lentocellum genome.
CONCLUSIONS: Genes coding for components of the central regulatory network which govern the expression of forespore and mother-cell-specific sporulation genes and the machinery used for engulfment appear best conserved. Low conservation of genes expressed late in endospore formation, particularly those that confer resistance properties and encode germinant receptors, suggest that Epulopiscium has lost the ability to form a mature spore. Our findings provide a framework for understanding the evolution of a novel form of cellular reproduction.

Background

Endospore formation is an ancient and complex developmental process exclusive to certain bacteria within the Firmicutes [1,2]. Endospores endure environmental conditions that would kill most other bacterial cells, including prolonged periods of insufficient nutrients, moderate levels of organic solvents, exposure to phage, extremes in pH, proteases and cell wall degrading enzymes, freezing, desiccation and excessive heat or radiation [3,4]. This form of sporulation preserves the genome in a remarkably dispersible and dormant cell type that can resume vegetative growth when the environment improves. While most sporulating species of Firmicutes produce a single endospore, some have the ability to produce multiple endospores [5]. For example, Clostridium oceanicum regularly forms two endospores, one at each end of the mother cell [6]. Others include the Segmented Filamentous Bacteria, a group of uncultivated inhabitants of the intestinal tract of animals [7]. These multicellular filaments live attached to the lining of the small intestine, and to disperse or reposition itself in the gut, each cell in a filament forms either an endospore containing two cells or two non-dormant intracellular offspring [7,8].

Other lineages within the Firmicutes use multiple endospore formation as a reproductive strategy. The guinea pig intestinal symbiont Metabacterium polyspora may undergo binary fission but the regular formation of multiple endospores, up to nine from a single mother cell, is a significant form of reproduction [9,10]. The life history of M. polyspora may be selecting for this unusual mode of reproduction, which could improve survival as the bacteria cycle in and out of the host gastrointestinal tract [10]. A large and diverse group of surgeonfish intestinal symbionts related to M. polyspora display an array of reproductive modes that involve binary fission and/or sporulation [11]. Like M. polyspora, the identified morphotypes and phylotypes of these surgeonfish symbionts show host-specific distributions, and an individual fish acquires the symbionts by the ingestion of feces or detritus [11,12]. The type C Epulopiscium-like fish intestinal symbionts, rely solely on the formation of two endospores for reproduction and appear to have abandoned binary fission altogether [12]. The largest members of this group of symbionts, Epulopiscium spp. type A and type B, are phylogenetically distinct from the smaller endospore-forming C morphotypes. Epulopiscium spp. type A and B appear to have taken the developmental process one step further and reproduce by the daily production of two or more intracellular offspring that are not dormant [11,13]. The phylogenetic relationship between multiple endospore formers and lineages that produce non-dormant intracellular offspring, and the morphological changes shared between these processes, suggest that the latter developmental process is related to endospore formation [13].

The stages of endospore formation (Figure 1) are described in the model organism Bacillus subtilis[14,15] and these morphological transitions appear conserved in other endospore formers [16]. Cells that exhibit no overt signs of sporulation are defined as stage 0. After initiation of sporulation, the chromosome replicates and replication origins become tethered to opposite poles of the cell. This unusual nucleoid conformation is called the axial filament and these cells are said to be in stage I [17]. Instead of dividing at the midcell, the sporulating cell divides near one pole, producing the forespore and larger mother cell, which marks stage II. Division traps approximately one-third of one of the chromosomes in the forespore [18]. The rest of the chromosome, still within the mother cell, is translocated into the forespore so that the spore contains a complete genome [19]. Enzymatic degradation of peptidoglycan between the mother cell and forespore results in curvature of the septum [20,21]. The mother-cell membrane then wraps around the forespore to completely engulf the forespore, which marks stage III [22]. In stage IV, a modified peptidoglycan called the cortex is synthesized in the space between the mother-cell and forespore membranes. In stage V, a complex proteinaceous coat is applied to the developing spore [23,24]. Stage VI is defined as endospore maturation, when the spore gains many of its resistance traits [3]. Lastly, the mother cell lyses to release the mature spore, in stage VII.

Figure 1

The life cycles of and sp. type B. A) In a favorable environment, B. subtilis undergoes growth and division. B) When nutrient limitations become critical, the cell may develop an endospore. Shown here are the morphological stages described for sporulation. The temporal and spatial activation of Spo0A and the four sporulation-specific sigma factors are shown. The grey circle around the forespore indicates the cortex. The thick black circle around the forespore at stage V and beyond represents the spore coat. C) Earliest stages of offspring development in Epulopiscium sp. type B are based on the similar morphological transitions described for sporulation in B. subtilis. See text for a detailed explanation of the process. Offspring frequently initiate the next round of reproduction prior to exiting the mother cell. Those stages that are seen in offspring still within their mother cell are highlighted with grey boxes. For all diagrams, DNA is shown in blue.

The sequential activation of stage-specific transcription factors controls the proper timing and location of gene expression to ensure the progression of cell-specific developmental events. In B. subtilis, a network of kinases and phosphatases conveys information about intracellular and extracellular conditions to the phosphorelay, which ultimately determines the phosphorylation state of the transcription factor Spo0A [25-27]. While Spo0A is considered the master regulator of sporulation, it also regulates a number of alternative cellular reactions to environmental change [28]. In its active form, Spo0A ~ P either directly or indirectly affects the transcription of more than 500 genes [29]. Spo0A activation is essential for entry into sporulation.

After asymmetric division, gene expression is regulated by four sporulation-specific sigma factors: σF, σE, σG, and σK (Figure 1) [30,31]. σF and σG are activated only in the forespore while σE and σK are activated only in the mother cell. Both σF and σE are expressed prior to asymmetric division [32], but σF is held inactive in a complex with two peptides of its anti-sigma factor SpoIIAB [33] and σE is synthesized as an inactive pro-peptide [34]. The forespore sigma, σF, is the first to be activated. SpoIIE phosphorylates SpoIIAA (the σF anti-anti-sigma factor), which binds SpoIIAB leading to the release of σF[35-37]. SpoIIR, part of the σF regulon [38], is produced in the forespore and inserted into the sporulation septum, where it activates SpoIIGA in the mother cell [39]. SpoIIGA then cleaves pro-σE thus releasing mature σE to the cytoplasm [40]. Likewise, the late-sporulation sigma factors, σG and σK, are not immediately functional when expressed and activation of each entails factor-specific intracellular signaling cascades and release mechanisms [41-48].

Tighter control of particular genes in each regulon is provided by additional transcription factors [49-53] that form both coherent and incoherent feed-forward loops with their associated sigma factor [1]. Coherent feed-forward loops occur when a sigma factor regulates expression of a gene and then combines with that gene product to up-regulate more genes. Incoherent feed-forward loops come about when such a combination leads to the down-regulation of additional genes [1,53]. For example, rsfA and spoIIR are both expressed from σF promoters [38,49]. However, RsfA combined with σF turns off transcription of spoIIR, so only a brief burst of spoIIR expression is seen immediately following asymmetric division. Such feed-forward loops allow the cell to modulate the timing, duration and location of expression of subsets of genes within a regulon. The combination of all central transcriptional regulatory mechanisms modulates the expression of more than 700 genes during sporulation.

The B. subtilis model can serve as a foundation for exploring mechanistic modifications required to support the formation of multiple endospores or intracellular offspring [13]. For example, Epulopiscium sp. type B are intestinal symbionts of the unicornfish Naso tonganus and can reach lengths of 200–300 μm and widths of 50–60 μm [11,54]. While the production of endospores has been observed in related morphotypes [12], Epulopiscium sp. type B does not produce endospores and does not reproduce by binary fission (Figure 1). Instead, each cell forms two or more intracellular offspring in a process that repeats daily [54-56].

The formation of these offspring has been described in stages that parallel stages of endospore formation [54-56]. Stage 0 mother cells contain large offspring (daughter cells) that show no signs of the initiation of the next generation of offspring (i.e. granddaughter cells). Stage I is defined as offspring cells that have coalesced DNA at the poles. Stage II cells have straight polar septa, but not all polar DNA is inside the newly formed offspring. Stage II-III includes cells with curved polar septa, indicating the start of polar cell engulfment, and all polar DNA has been translocated into the offspring. Stage III cells contain small, fully engulfed offspring, with a length-to-width ratio of less than 2:1 while the offspring in stage IV* cells have a ratio greater than 2:1. After engulfment, the two processes diverge and changes occurring in Epulopiscium are not well understood so later stages are not indicated in this model. Offspring continue to grow until they fill the mother-cell cytoplasm. In time, the mother-cell envelope splits open and the offspring are released.

Previous work has shown that the division protein FtsZ localizes to the poles of an Epulopiscium cell in a similar way to FtsZ in B. subtilis during endospore formation [54]. In B. subtilis, both potential division sites appear fully functional and a second, partial polar septum occasionally forms at the forespore-distal pole although the second septum eventually regresses [57,58]. With Epulopiscium, rapid sequential or simultaneous bipolar division occurs [54]. Putative homologs for SpoIIE, SpoIIAA, SpoIIAB and σF also have been identified in the Epulopiscium genome [55]. Moreover, the expression pattern of spoIIE in Epulopiscium during offspring formation [55] is very similar to that seen in sporulating B. subtilis[59]. These results suggest that Epulopiscium uses cell-specific activation of alternative sigma factors in offspring development.

Here we sequenced and examined the draft genome of Epulopiscium sp. type B to determine the extent of conservation of the sporulation genetic program. A list of conserved “core” sporulation genes was assembled and used to determine which of the core sporulation genes are conserved in the Epulopiscium genome and its spore-forming relative Cellulosilyticum lentocellum DSM 5427. As predicted, we found a number of homologs to genes with sporulation-specific functions in Epulopiscium and even more of the core genes conserved in C. lentocellum. These results begin to define the genetic mechanisms that may be used for offspring production and development in Epulopiscium.

Results and discussion

Epulopiscium sp. type B draft genome

The draft genome of Epulopiscium sp. type B was used to assess the conservation of sporulation gene homologs in this bacterium. Epulopiscium sp. type B has become our model for genome studies because N. tonganus harbors morphologically and genetically homogenous populations of these Epulopiscium cells. Clone libraries of 16S rRNA gene fragments generated from amplified DNA using primers (27F and 1492R) to conserved regions of the gene consistently yield a single phylotype [60,61]. Although the genome was not assembled completely, the project produced approximately 2.7 Mbp of sequence in 92 large contigs ranging from 12 – 119 kb in length. The remaining data, consisting of small contigs less than 12 kb and well-represented single reads (singletons), comprise another 13.7 Mbp. This data set likely represents the Epulopiscium genome and may contain some highly repetitive DNA from the fish host or other abundant inhabitants of the N. tonganus intestinal tract that may have been carried along during Epulopiscium cell isolation.

To determine the completeness of the Epulopiscium draft genome, we analyzed the 92 large contigs using two approaches. First, we identified the total number of rRNAs and tRNAs encoded by the genome. We found four 5S, four 23S, and three 16S rRNA genes in addition to 31 tRNAs covering all amino acids except histidine, phenylalanine, and serine. These values are about one-third less the rRNAs and tRNAs encoded by C. lentocellum DSM 5427, the closest relative with a complete genome sequence [62]. We also analyzed the small contigs and singletons in the Epulopiscium draft genome and found additional functional RNAs. This resulted in a grand total of nine 5S, five 16S and five 23S rRNAs, as well as 66 tRNAs covering all amino acids for the Epulopiscium draft genome. These numbers should be viewed with caution as some copies may represent overlapping contigs that did not assemble. Second, we performed a clusters of orthologous genes (COG) analysis using the predicted proteome from the 92 large contigs of the Epulopiscium draft genome and compared the percent proteins encoded in each category against a similar analysis of all completely sequenced genomes in either the phylum Firmicutes or the genus Clostridium (Table 1). We found that the percentage of proteins in each COG category in the Epulopiscium draft genome was comparable to the percentages of either the Firmicutes or the Clostridia. One noteworthy difference was seen in the category of carbohydrate transport and metabolism, which was exceptionally high in Epulopiscium, compared to the other Firmicutes groups. Taken together, these data indicate that this draft is somewhat incomplete, although it is difficult to say with absolute certainty the level of completeness.

Table 1

COG analysis of the sp. type B genome

COG Category	Code	Epulopisciumsp. type B	Clostridium	Firmicutes
Information storage and processing
Translation, ribosomal structure and biogenesis	J	7.57%	5.91%	6.95%
RNA processing and modification	A	0.00%	0.00%	0.00%
Transcription	K	5.35%	9.19%	8.05%
Replication, recombination and repair	L	4.17%	5.43%	6.20%
Chromatin structure and dynamics	B	0.00%	0.03%	0.03%
Cellular processes and signaling
Cell cycle control, cell division, chromosome partitioning	D	1.94%	1.24%	1.29%
Nuclear structure	Y	0.00%	0.00%	0.00%
Defense mechanisms	V	1.60%	2.81%	2.39%
Signal transduction mechanisms	T	5.42%	6.92%	4.89%
Cell wall/membrane/envelope biogenesis	M	5.49%	5.60%	5.36%
Cell motility	N	4.31%	2.73%	1.51%
Cytoskeleton	Z	0.00%	0.01%	0.01%
Extracellular structures	W	0.00%	0.00%	0.00%
Intracellular trafficking, secretion, and vesicular transport	U	1.74%	1.44%	1.53%
Posttranslational modification, protein turnover, chaperones	O	3.13%	2.76%	2.99%
Metabolism
Energy production and conversion	C	5.97%	5.94%	5.17%
Carbohydrate transport and metabolism	G	16.60%	7.26%	8.09%
Amino acid transport and metabolism	E	10.56%	8.42%	8.86%
Nucleotide transport and metabolism	F	2.78%	2.84%	3.33%
Coenzyme transport and metabolism	H	5.00%	4.13%	4.10%
Lipid transport and metabolism	I	2.22%	1.97%	2.36%
Inorganic ion transport and metabolism	P	5.28%	4.87%	5.22%
Secondary metabolites biosynthesis, transport and catabolism	Q	0.49%	1.03%	1.10%
Poorly characterized
General function prediction only	R	11.46%	11.09%	11.20%
Function unknown	S	6.53%	8.36%	9.39%
Total Genomes Analyzed		1	32	300

For the identification of potential sporulation gene homologs in Epulopiscium, contigs and singletons were concatenated into a single pseudomolecule. The position of each junction between adjacent fragments was noted so that any chimeric open reading frames formed during pseudomolecule assembly could be identified.

The core sporulation gene list

As a starting point, a comprehensive list of sporulation genes was compiled based on previous studies of B. subtilis 168 and its derivatives, since this is by far the most thoroughly characterized sporulation program for any member of the Firmicutes. The initial list of 732 genes included those described using classic genetic approaches as well as putative sporulation genes uncovered in transcriptional array analyses of sporulation regulons [29,53,63-65]. BLAST searches refined the list, narrowing it to genes that are conserved in both endospore-forming Bacilli and Clostridia. Further refinement of the list eliminated genes with homologs in non-spore-forming Firmicutes.

Several notable genes that were eliminated are essential in the B. subtilis model of sporulation, including the phosphorelay protein Spo0B, and associated histidine kinases, transcription factors RsfA, GerR and GerE, and the intercellular signal transduction proteins SpoIIQ and SpoIVFA. Our inability to identify a SpoIIQ homolog agrees with the results of a previous study [1]. Likewise, the limited distribution of B. subtilis phosphorelay homologs among endospore-forming bacteria has been noted previously [66,67] and it is likely that clostridia use a different set of kinases for activation of Spo0A [68,69]. For example, recent genetic studies demonstrated phosphorylation of Spo0A in C. acetobutylicum by novel orphan histidine kinases [68,69].

A few key early sporulation proteins (Spo0A, Spo0J, Soj, SpoIIIE and SpoIIIJ), known to function during normal growth in B. subtilis, were placed back on the core list. All of these are part of the σA regulon and were first identified as essential for efficient sporulation. Spo0A determines entry into sporulation. Soj and Spo0J are members of the ParA/B family of partitioning proteins that regulate transcription of early sporulation genes and assist in maintaining chromosome architecture [31]. The FtsK homolog SpoIIIE aids in chromosome separation during binary fission [70], and during sporulation it is responsible for chromosome translocation into the developing forespore [18,19]. The location of genes on the chromosome and timing of translocation affects cell-specific expression of genes, which impacts key molecular events such as the activation of sporulation sigma factors [71-73]. SpoIIIJ is a membrane protein translocase that, in addition to its vegetative growth function, appears to play a key role in the activation of σG during sporulation [41,74]. In the end, our core list included 147 genes (Table 2).

Table 2

Core sporulation genes identified in the168 genome

GENE	REG	FUNCTION
aprX	K	alkaline serine protease
bofA	E	inhibitor of the pro-σK processing machinery
cotI	K	spore coat protein
cotJB	E	component of the inner spore coat
cotJC	E	component of the inner spore coat
cotS	K	spore coat protein
csfB	F	anti-σG factor
cwlD	E, G	N-acetylmuramoyl-L-alanine amidase/germination
cwlJ	E	cell wall hydrolase/germination, cortex lytic
dacB	E	D-alanyl-D-alanine carboxypeptidase (penicillin-binding protein 5)
dacF	F, G	D-alanyl-D-alanine carboxypeptidase (penicilin-binding protein)
gerAA	F, G	component of the germination receptor GerA
gerAB	F, G	component of the germination receptor GerA
gerAC	F, G	component of the germination receptor GerA
gerBA	G	component of germinant receptor B
gerBB	G	component of germinant receptor B
gerBC	G	lipoprotein component of the germination receptor B
gerKA	G	spore germination receptor subunit
gerKB	G	spore germination receptor subunit
gerKC	G	spore germination receptor subunit
gpr	F, G	germination protease
jag	A	SpoIIIJ-associated RNA/ssDNA-binding protein
kamA	E	lysine 2,3-aminomutase
kapD	E,K	inhibitor of the KinA pathway to sporulation
lonB	F	ATP-dependent protease/forespore-specific
mmgC	E	short chain acyl-CoA dehydrogenase
ntdA	E	biosynthesis of neotrehalosadiamine (amino-sugar antibiotic)/aminotransferase
pbpG	F, G	penicillin-binding protein (also known as ywhE)
pbpI	E, F	penicillin-binding protein PBP4B/mother cell specific
pdaA	G	exported N-acetylmuramic acid deacetylase/cortex lysis
prkA	E	serine protein kinase/not well characterized
sigE	A, 0A	sporulation sigma factor/mother cell only
sigF	H, 0A	sporulation sigma factor/forespore only
sigG	F, G	sporulation sigma factor/forespore only
sigK	E, K	sporulation sigma factor/mother cell only
sleB	G	spore cortex-lytic enzyme
soj	A, 0A	chromosome partitioning protein/transcriptional regulator/negative regulation of sporulation initiation
splB	G	spore photoproduct (thymine dimer) lyase
spmA	E	spore maturation protein/spore dehydration
spmB	E	spore maturation protein/spore dehydration
spo0A	A, H, 0A	two-component response regulator central for the initiation of sporulation/"master regulator"
spo0F	H, 0A	two-component response regulator involved pathway leading to phosphorylation of Spo0A
spo0J	A	site-specific DNA-binding protein/chromosome positioning near the pole and transport through the polar septum/antagonist of Soj-dependent inhibition of sporulation initiation
spoIIAA	H, 0A	anti-anti-sigma factor (antagonist of SpoIIAB)
spoIIAB	H, 0A	anti-σF factor
spoIID	E	autolysin required for complete dissolution of the sporulation septum
spoIIE	A, 0A	serine phosphatase (σF activation)/polar septum formation
spoIIGA	A, 0A	protease processing pro-σE
spoIIIAA	E	ATP-binding stage III sporulation protein/mother cell signalling for σG activation
spoIIIAB	E	stage III sporulation protein/mother cell signalling for σG activation
spoIIIAC	E	stage III sporulation protein/mother cell signalling for σG activation
spoIIIAD	E	stage III sporulation protein/mother cell signalling for σG activation
spoIIIAE	E	stage III sporulation protein/mother cell signalling for σG activation
spoIIIAF	E	stage III sporulation protein/mother cell signalling for σG activation
spoIIIAG	E	stage III sporulation engulfment assembly protein/mother cell signalling for σG activation
spoIIIAH	E	stage III sporulation ratchet engulfment protein/mother cell signalling for σG activation
spoIIID	E	transcriptional regulator of σE and σK-dependent genes
spoIIIE	A	spore DNA translocase
spoIIIJ	A	protein translocase/essential for activation of σG
spoIIM	E	autolysin for dissolution of the septal cell wall
spoIIP	E, F, G	spore autolysin
spoIIR	F	endopeptidase/activation of σE
spoIVA	E	morphogenetic protein required for proper spore cortex formation and coat assembly
spoIVB	F, G, 0A	regulatory membrane-associated serine protease/intercompartmental signalling of pro-σK processing and activation in the mother-cell
spoIVFB	E	membrane metalloprotease required for the processing of pro-σK
spoVAA	G	dipicolinic acid uptake by the developing spore
spoVAB	G	dipicolinic acid uptake by the developing spore
spoVAC	G	dipicolinic acid uptake by the developing spore
spoVAD	G	dipicolinic acid uptake by the developing spore
spoVAEB	G	dipicolinic acid uptake by the developing spore
spoVAF	G	dipicolinic acid uptake by the developing spore
spoVB	E	spore cortex synthesis
spoVD	E	penicillin-binding protein
spoVE	E	spore cortex peptidoglycan synthesis
spoVFA	K	spore dipicolinate synthase subunit A
spoVFB	K	spore dipicolinate synthase subunit B
spoVR	E	spore cortex synthesis
spoVS	H	regulator required for dehydration of the spore core and assembly of the coat
spoVT	F, G	transcriptional regulator of σG-dependent genes
spsF	K	putative glycosyltransferase/spore coat polysaccharide synthesis
spsG	E, K	putative glycosyltransferase/spore coat polysaccharide synthesis
spsJ	E, K	putative dTDP-glucose 4,6-dehydratase/spore coat polysaccharide synthesis
sspA	G	small acid-soluble spore protein (alpha-type SASP)
sspB	G	small acid-soluble spore protein (beta-type SASP)
sspC	G	small acid-soluble spore protein (alpha/beta-type SASP)/SPβ phage protein
sspD	G	small acid-soluble spore protein (alpha/beta-type SASP)
sspF	G	small acid-soluble spore protein (alpha/beta-type SASP)
tlp	F, G	small acid-soluble spore protein
yaaH	E	spore peptidoglycan hydrolase
yabG	K	sporulation-specific protease
yabP	E	spore protein involved in the shaping of the spore coat
yabQ	E	membrane protein of the forespore/essential for spore cortex
ybaN	E	polysaccharide deacetylase involved in sporulation
ybdM	G	putative protein kinase
ycgF	E	putative amino acid export permease
ydfS	K, G	hypothetical protein
ydhD	E	spore cortex lytic enzyme
yerB	0A	putative lipoprotein
yfkQ	G	putative spore germination protein
yfkR	G	putative spore germination protein
yfkT	G	putative spore germination integral inner membrane protein
yfnG	K	putative sugar-phosphate cytidylyltransferase
yfnH	K	putative FAD-dependent oxido-reductase
ygaK	K	putative FAD-dependent oxido-reductase
yhbH	E	hypothetical protein
yhcB	G	putative oxidoreductase associated
yhcV	G	putative oxidoreductase
yhfW	F	putative Rieske [2Fe-2S] oxygenase
yisY	E	putative hydrolase
ykuD	K	murein transglycosylase
ykuS	F	hypothetical protein
ykvU	E	spore membrane protein involved in germination
ylaK	E	putative phosphate starvation inducible protein
ylbB	F	putative oxidoreductase
ylbJ	E	putative factor required for spore cortex formation
ymxH	E	hypothetical protein
yndD	G	putative spore germination protein
yndE	G	putative spore germination integral inner membrane protein
yndF	G	putative spore germination lipoprotein
yngE	E	putative propionyl-CoA carboxylase
yngI	E	AMP-binding domain protein
yngJ	E	acyl-CoA dehydrogenase, short-chain specific
yoaR	G	putative factor for cell wall maintenance or synthesis
yobN	E	putative amine oxidase
ypeB	G	spore membrane component
yqfC	E	hypothetical protein
yqfD	E	stage IV sporulation protein
yqgT	E	putative gamma-D-glutamyl-L-diamino acid endopeptidase
yqhO	E	hypothetical protein
yraD	G	putative spore coat protein
yrbG	E, G	hypothetical protein
yrkC	K	putative dioxygenase; cupin family
ytcA	K	putative UDP-glucose dehydrogenase
ytcC	K	putative glucosyltransferase
ytfJ	F	hypothetical protein
ytlA	K	putative ABC transporter component
ytlC	K	putative ABC transporter component, ATP-binding
ytlD	K	putative permease of ABC transporter
ytvI	E	putative permease
ytxC	E	hypothetical protein
yunB	E, K	putative protein involved in spore formation
yutH	F	spore coat-associated protein
ywjD	G	putative UV damage endonuclease
yyaC	F	hypothetical protein
yyaE	E	putative oxidoreductase
yyaO	K	hypothetical protein
yyBI	E	inner spore coat protein

An overview of the core sporulation genes found in Epulopiscium and C. lentocellum

We were concerned that the phylogenetic distance between B. subtilis and Epulopiscium may bias the distribution of conserved genes we would be able to identify. To gauge how much evolutionary divergence is impacting the recovery of homologs in Epulopiscium, we compared our core list of sporulation genes to the C. lentocellum genome as well. Cellulosilyticum lentocellum DSM 5427 is the closest endospore-forming relative of Epulopiscium with an available completely sequenced genome [62]. Epulopiscium sp. type B and C. lentocellum are members of the family Lachnospiraceae and the two share 91% 16S rRNA sequence identity. Of the 4,185 predicted protein-coding genes in the C. lentocellum genome, an Epulopiscium sp. type B gene is the top BLAST hit for 546 of these genes (data not shown).

Of the 147 core sporulation genes, we found 87 homologs in the C. lentocellum genome and 57 putative homologs in the Epulopiscium sp. type B genome (Table 3). All of the sporulation genes found in the Epulopiscium genome were also found in the C. lentocellum genome. In the list of conserved genes, two code for members of a family of functionally redundant small acid-soluble proteins (SASPs) that bind to and protect spore DNA from damage [4,75]. One of the SASPs is similar to two paralogs of this family from B. subtilis and is represented as sspC/F. In both Epulopiscium and C. lentocellum, at least five genes are represented from each of the four sporulation sigma factor regulons, as well as genes in the Spo0A regulon with σH and σA promoters. Of the C. lentocellum homologs not seen in the Epulopiscium genome, 14 code for “y” genes that are expressed during sporulation, although their roles have yet to be characterized. These genes are equally distributed between pre- and post-engulfment regulons, but only two are expressed in the forespore in B. subtilis.

Table 3

Genes conserved inand

GENE	PROMOTER(S)	Epulopiscium	C. lentocellum
σAregulon
jag			X
sigE		X	X
soj		X	X
spo0A		X	X
spo0J		X	X
spoIIE		X	X
spoIIGA		X	X
spoIIIE	σA	X	X
spoIIIJ		X	X
σHregulon
sigF		X	X
spoIIAA		X	X
spoIIAB		X	X
σFregulon
dacF	σG	X	X
gpr	σG	X	X
lonB		X	X
sigG	σF, σG	X	X
spoIIR		X	X
spoIVB	σF, σG	X	X
spoVT	σG	X	X
yhfW			X
ytfJ			X
yyaC		X	X
σEregulon
bofA			X
cotJB			X
cotJC			X
cwlD	σG	X	X
dacB		X	X
kamA			X
prkA		X	X
sigK		X	X
spmA		X	X
spmB		X	X
spoIID	σE	X	X
spoIIIAA	σE	X	X
spoIIIAB	σE	X	X
spoIIIAC	σE	X	X
spoIIIAD	σE	X	X
spoIIIAE	σE	X	X
spoIIIAF	σE	X	X
spoIIIAG	σE	X	X
spoIIIAH	σE	X	X
spoIIID	σE	X	X
spoIIP	σG	X	X
spoIVA	σE	X	X
spoVB		X	X
spoVD		X	X
spoVE		X	X
spoVR		X	X
spsJ			X
yaaH			X
yabP	σE	X	X
yabQ	σE	X	X
yhbH		X	X
ylbJ			X
yngI			X
yqfC	σE	X	X
yqfD	σE	X	X
yqgT			X
yqhO			X
ytvI		X	X
σGregulon
gerKA			X
gerKB			X
gerKC			X
pdaA		X	X
splB			X
spoVAA			X
spoVAB			X
spoVAC			X
spoVAD			X
spoVAEB			X
spoVAF		X	X
sspB		X	X
sspC/F		X	X
ybdM		X	X
ydfS			X
σKregulon
cotS		X	X
spoVFA			X
spoVFB	σK	X	X
yabG		X	X
yfnG		X	X
yfnH		X	X
ygaK			X
yrkC			X
ytcC			X
ytlA			X
ytlC			X
ytlD			X

Conservation of the master regulator Spo0A in Epulopiscium

Although clostridia do not have a phosphorelay system homologous to that used by B. subtilis, Spo0A is conserved in the genomes of all known endospore-forming bacteria. The Epulopiscium sp. type B genome contains an unambiguous Spo0A homolog (Figure 2). The Epulopiscium Spo0A homolog is 51% identical to Spo0A of B. subtilis. Conserved residues in both phospho-acceptor and DNA binding domains [76] are located in the putative Epulopiscium Spo0A homolog as well. This suggests that although offspring production seems hard-wired in Epulopiscium and there is no obvious need to adapt the timing of entry into offspring production to environmental conditions, Spo0A is still important for initiating this process or possibly other metabolic transitions. Bacillus spp. use the phosphorylation state and abundance of Spo0A to engage in activities, such as cannibalism, to maintain a growing population for as long as possible prior to resorting to sporulation [26]. In clostridia, Spo0A also dictates metabolic transitions [77]. For example, in Clostridium acetobutylicum, Spo0A activation shifts the cell from acid production during exponential phase growth to solvent production during stationary phase in addition to entry into sporulation [78]. We searched the Epulopiscium genome for homologs of the newly discovered C. acetobutylicum Spo0A kinases that are important for developmental and metabolic transitions [69], however none of these appear to be conserved in the Epulopiscium genome. The future identification and characterization of kinases or phosphatases in Epulopiscium that interact with Spo0A may provide information about the environmental and cellular cues that regulate population level developmental or physiological transitions.

Figure 2

An alignment of Spo0A homologs. The predicted amino acid sequences of Spo0A from B. subtilis 168, B. anthracis Ames, C. acetobutylicum ATCC 824, C. botulinum ATCC 3502, Cellulosilyticum lentocellum DSM 5427 and Epulopiscium sp. type B were aligned using CLUSTALΩ. The conserved phosphorylation site (highlighted in yellow), the conformational switch (in green) and the DNA recognition helix (light blue) are found in all homologs. The connector segment (outlined in black) links the upstream phospho-acceptor and downstream effector domains. Shaded bars below the effector domain indicate the helix-turn-helix (HTH) DNA binding motif.

The central regulatory network conserved in endospore-forming bacteria appears conserved in Epulopiscium

In B. subtilis, the sequential activation of cell-specific sigma factors directs compartmentalized gene expression that determines the different fates of the mother cell and forespore [1]. The presence of homologous genes suggests a similar system likely also functions in Epulopiscium developmental progression (Figure 3). In a previous publication, we described the σF homolog of Epulopiscium and a structural analysis of the proteins required for σF activation: SpoIIE, SpoIIAA, and SpoIIAB [55]. In the present study, we uncovered genes coding for all additional sporulation sigma factors (σE, σG and σK) and homologs for the intercellular signal transduction proteins responsible for the activation of σE (SpoIIR and SpoIIGA) and σG (SpoIIIAA-AH and SpoIIIJ). In B. subtilis, SpoIIQ interacts with SpoIIIAA-AH in an intracellular signaling system required for the activation of σG[45]. The gene coding for SpoIIQ is absent in clostridia [1] therefore involvement of SpoIIQ is a Bacillus-specific innovation that likely does not represent the ancestral mode of cell-cell communication used to trigger late forespore sigma factor activation.

Figure 3

Conservation of the sporulation regulatory cascade in sp. type B. Sporulation-specific sigma factors (circles), associated transcription factors (diamonds) and other signal transduction or regulatory proteins involved in sigma activation (rectangles) are shown. Colors of the proteins indicate the gene presence in Epulopiscium (green), on the core list but not in Epulopiscium (red), and absence from the core list (blue). Control of gene expression is indicated by dotted lines and arrows. Signaling pathways and other protein interactions are denoted with solid lines and arrows. Temporal transcriptional progression through the cascade is shown by the position on the diagram with earlier stages near the top. A detailed explanation of the regulatory cascade as it occurs in B. subtilis is provided in the text. Figure is adapted from de Hoon et al. (2010).

Mechanisms to activate the late mother-cell sigma factor, σK, are apparently the most specialized of the sporulation sigma factors. In B. subtilis, the gene encoding σK contains a 48 kb insertion sequence called the skin (sigma K intervening) element that must first be excised from the mother-cell chromosome by the recombinase SpoIVCA [79,80]. The reconstituted gene produces an inactive pro-protein which must be processed by SpoIVFB but BofA and SpoIVFA inhibit SpoIVFB activity [47,48]. These three proteins are transcribed by σE in the mother cell and localize to the outer forespore membrane. SpoIVB and CtpB, regulated by σG in the forespore, inhibit the actions of BofA and SpoIVFA allowing cleavage of pro-σK[44,81]. In this way, σK operates in the mother cell only after σG has been activated in the forespore. Only homologs for spoIVB and ctpB were found in Epulopiscium. Note that ctpB homologs are found in many non-spore-forming bacteria, which is why it is not included in the core list. The gene coding for SpoIVFA is absent in clostridia and although spoIVFB and bofA met the requirements to be included on the core list, neither is conserved in many clostridia and only bofA was found in the C. lentocellum genome. The weak conservation of these proteins outside of the Bacilli indicates that the intricate mechanism of σK activation described in B. subtilis is a recent innovation.

In addition to conservation of sporulation sigma factors and many of the proteins responsible for regulating their activation, we found homologs of transcription factors SpoIIID and SpoVT. These proteins work together with their associated sigma factor (σE and σG, respectively) to modulate transcription of genes downstream in the process. Other transcription factors that function during sporulation in B. subtilis (RsfA, GerR and GerE) are not conserved in the clostridia. These feed-forward loops appear to be a Bacilli-specific amendment. Based on these results, we conclude that gene regulation by sporulation-specific sigma factors and their associated transcription factors is highly conserved in Epulopiscium. With the exception of bofA, all of the core sporulation genes coding for components of the central regulatory cascade that are conserved in the C. lentocellum genome are also conserved in Epulopiscium.

Engulfment genes are highly conserved in Epulopiscium

SpoIID, SpoIIP and SpoIIM are required for degradation of septal peptidoglycan and progression toward forespore engulfment in B. subtilis[21]. We found Epulopiscium genes that code for homologs of SpoIID and SpoIIP but could not identify a SpoIIM homolog. SpoIIM appears highly conserved in spore-forming bacteria but is curiously absent from the Lachnospiraceae, as the genome of C. lentocellum also lacks an apparent homolog of spoIIM (Table 3). The Epulopiscium spoIIP initially retrieved from the pseudomolecule did not meet the requirements to be considered a homolog in this analysis, as it appeared to be missing its 5´ end. In B. subtilisspoIIP has its own σE promoter [82] but is also the second gene in an operon with gpr[63], and this operon structure is widely conserved in Bacillus and Clostridium spp. A truncated gpr homolog was identified in the Epulopiscium draft genome. Based on this information, PCR and sequence analysis was used to recover the complete Epulopiscium gpr - spoIIP operon.

Other proteins that function during engulfment in B. subtilis are SpoIIB, SpoIIIE, SpoIIIAH and SpoIIQ. SpoIIB likely plays a role in the localization of the SpoIIDMP complex in B. subtilis[20] but it is not conserved in the clostridia. Null mutants of spoIIB in B. subtilis are oligosporogenous, and this sporulation defect becomes more pronounced when combined with a spoVG mutation [83]. In addition to its DNA translocase activity, SpoIIIE has been implicated in the fusion of the leading edge of the mother-cell membrane after migrating around the forespore [84,85]. We identified homologs of SpoIIIE in both Epulopiscium and C. lentocellum. Beyond the role of the SpoIIIAH - SpoIIQ complex in σG activation, these two proteins may perform an important role during engulfment. It has been suggested that as the mother-cell membrane migrates around the forespore, these proteins bind to one another near the leading edge of the mother-cell membrane to prevent it from retreating back toward the midcell [22]. Homologs of the spoIIIA operon were identified in C. lentocellum and Epulopiscium. With the exception of SpoIIM and SpoIIQ, all of the known proteins required for engulfment in B. subtilis are coded for in the Epulopiscium genome (Figure 4). Neither SpoIIM nor SpoIIQ are coded for in the C. lentocellum genome.

Figure 4

engulfment model. The Epulopiscium genome codes for all of the genes known to be essential for engulfment in B. subtilis, except spoIIM and spoIIQ which are also absent in C. lentocellum. A) SpoIID and SpoIIP assemble into a complex (red ovals) at the division septum and degrade the septal peptidoglycan. B) As the mother-cell membrane wraps around the offspring, the IIDP complex tracks along the leading edge where it is involved in interactions with the mother-cell peptidoglycan and synthesis of offspring cell wall. C) When it reaches the cell tip, membrane fusion is mediated by SpoIIIE (yellow circle). During engulfment, SpoIIIAH (green rectangles) produced in the mother cell and a hypothetical protein (blue rectangles) from the offspring cell bind and prevent backward movement of the mother-cell membrane. In this diagram, black lines indicate membranes and grey peptidoglycan.

Late stage sporulation proteins in the Epulopiscium genome

A small number of late sporulation genes were identified in the Epulopiscium sp. type B genome. This may be an underestimate as many of the proteins in this category are very small and functionally redundant. For example, it is possible that some coat proteins have diverged to a degree that they could not to be detected by our methods. Regardless, B. subtilis uses more than 75 proteins for cortex and coat formation, spore maturation and germination of which Epulopiscium has retained 17 possible homologs. In comparison, the C. lentocellum genome has 31. The genes retained in Epulopiscium may perform novel functions, or are relics of its spore-forming ancestry. Previous studies have identified sporulation homologs in the genomes of non-sporulating members of the Firmicutes [2]. Additionally, many surgeonfish intestinal symbionts that are close relatives of Epulopiscium sp. type B form endospores [12], and would likely need late sporulation proteins similar to those in B. subtilis and the clostridia for the formation of cortex and coat as well as spore maturation and germination.

Based on their functions in B. subtilis, we classified the late sporulation genes located in C. lentocellum and/or Epulopiscium. With the notable exception of genes coding for SASPs (sspBsspC/F) or SASP degradation (gpr), genes involved in resistance properties of a mature spore and germination signal receptors (gerK operon) are more highly conserved in C. lentocellum. Specifically, C. lentocellum has the nine genes required for the synthesis and forespore-uptake of dipicolinic acid (DPA) while Epulopiscium has only one of these genes. Likewise, only C. lentocellum has splB, which codes for the DNA repair enzyme spore-photoproduct lyase. Surprisingly, all of the ten genes on the list involved in cortex biosynthesis or other cortex-associated properties that are conserved in C. lentocellum are also found in Epulopiscium. In addition, spoIVAyabP and yabQ, which encode scaffolding proteins important for cortex and coat morphogenesis [86-88], were found in both C. lentocellum and Epulopiscium. We speculate that the synthesis of a cortex-like peptidoglycan may accommodate several Epulopiscium characteristics including rapid offspring growth, emergence of offspring from the mother cell or other functions that support a large bacterial cell. Four coat protein genes were found in C. lentocellum but only one of these appears conserved in Epulopiscium. The biased distribution of functional categories represented in Epulopiscium suggests that cortex biosynthetic machinery has been retained, as well as the DNA-protective SASPs, and these proteins may still serve some function. Overall, the pattern of conserved genes in C. lentocellum compared with Epulopiscium (Figure 5) supports our hypothesis that there is a strong selection to retain early stage sporulation gene homologs in Epulopiscium such as those necessary for compartmentalized gene expression, engulfment and developmental progression.

Figure 5

Distribution of core sporulation genes conserved in the and genomes by regulon. Venn diagrams represent the conservation of genes in the four sporulation-specific sigma factor regulons. Circle size corresponds to the number genes on the core list for B. subtilis (outermost circle), C. lentocellum (middle circle) and Epulopiscium (inner circle). The numbers below each diagram indicate the total number of genes from B. subtilis in each regulon. Some genes were counted more than once if they are members of multiple regulons as indicated in Table 2

Promoter analysis of putative developmentally regulated genes

For each of the sporulation genes found in the Epulopiscium genome, the region immediately upstream of the predicted start codon (~300 bp) was visually scanned for potential promoter sequences (Table 4). While this analysis is not definitive, we reasoned that identifiable promoters that match their predicted B. subtilis counterparts would reinforce the predictions placing genes in particular regulons. Consensus of the −10 and −35 promoter sequences from B. subtilis promoters were used [53,63,89]. We were able to identify σF promoters upstream of sigG and spoIVB, and σE promoters upstream of spoIID, the spoIIIA operon, spoIIID, the yabPQ operon and the yqfCD operon. σG promoters were found upstream of dacFcwlDsigGspoIVBspoVT and the gpr – spoIIP operon. Only one gene, spoVFB, was found to have a σK promoter. Our analysis of the putative dacF homolog indicated that approximately 50% of the 5´ end of the gene was present in the pseudomolecule, however, we were unable to obtain a full-length gene by PCR. The gene spoVFB codes for the ß-subunit of DPA synthase and is usually located in a bicistonic operon downstream of spoVFA, the gene for the α-subunit of the DPA synthase [90]. DPA is found in abundance only in mature endospores [91] and is a distinct feature of the phase-bright spores of the Epulopiscium-like symbionts of Naso lituratus[12]. Although we were unable to recover spoVFA from Epulopiscium, we did find homologs of spoVFA and spoVFB adjacent to each other in the C. lentocellum genome.

Table 4

Promoters found upstream of sporulation gene homologs in thegenome

GENE/OPERON	PROMOTER
	σF		σE		σG		σK
	−35	−10	−35	−10	−35	−10	−35	−10
sigG	GTATA	GGGTATCCTA			GTATA	TATCCTA
spoIVB	GTATA	GGCAATTTTA			GTATA	AATTTTA
spoIID			TTATAAGT	TCATATAATT
spoIIIA			TAATGATT	GCATATACTG
spoIIID			TAATATAT	GCATATTATT
spoIVA			TCATATCC	ACATATAGTT
yabPQ			GAATAATT	AAATATAAAT
yqfCD			GAATACTT	GCATAATATG
dacF					GTATA	AATAATA
cwlD					GAATT	GATAATA
gpr-spoIIP					GATTA	TATATTA
spoVT					GCATA	CATAATA
spoVFB							TCACA	TCATATTATA
B. subtilis consensusa	GTATA	GGNAANAMTR	TYATATTT	KCATANANTN	GNATA	CAWAMTA	KCACM	GCATANNNTA

Consensus sequences from B. subtilis were based on Wang et al. (2006) and Eichenberger et al. (2004). Bold indicates a highly conserved base. N = A, T, G, C; R = A, G; Y = C, T; K = T, G; M = A, C; W = A, T.

Conclusions

The comparative analysis of the draft Epulopiscium sp. type B genome with the complete C. lentocellum genome substantiates our hypothesis that the production of intracellular offspring in Epulopiscium evolved from endospore formation. All of the genes identified in C. lentocellum that function in engulfment as well as the core transcriptional regulatory cascade, and the associated intracellular communication network that coordinates sigma factor activation, were found in the Epulopiscium genome. While we could identify homologs of late sporulation genes, a large proportion of these were not recovered from Epulopiscium. Since we used a draft genome to explore the conservation of sporulation genes in Epulopiscium, it is possible that we did not recover all of the sporulation genes retained in this genome, however, we would not expect a functional bias in the distribution of genes recovered. Therefore the fewer late genes recovered in Epulopiscium probably reflect the reduction of this class of genes in the evolution of its genome.

Although Epulopiscium sp. type B is closely related to surgeonfish intestinal symbionts that form multiple endospores to reproduce [12], it appears that type B cells may no longer have the genetic capacity to form a dormant and fully resistant endospore. Many intestinal anaerobes, from harmful pathogens to benign commensals, use endospores for effective dispersal between vertebrate hosts [8,10,92]. The ability of an intestinal bacterium to produce an endospore should be valuable to survival. Why then would Epulopiscium sp. type B lose this trait? We speculate that the perpetuation of large cell size and the ability to maintain a longer residence in an individual host may be factors that contributed to the loss of dormancy and associated resistance traits in offspring development in this lineage. The large size of Epulopiscium sp. type B cells may be important to maintain their position in the gut and to avoid predation by the ciliate predators that cohabitate the N. tonganus intestinal tract [61]. The growth and development of offspring within a metabolically active mother cell may be essential for maintaining these benefits throughout the life cycle of an individual. It is also possible that the manifestation of some resistance traits may simply be impossible for a cell as large as an Epulopiscium sp. type B offspring. For example, the formation of a flawless spore coat may be physically impossible for this size of cell.

In the study outlined here, we generated the first genome sequence for any Epulopiscium species and use it to provide insight into the evolution of a novel form of cellular reproduction in the Epulopiscium lineage. Based on its phylogenetic position among endospore-forming lineages, we reason that formation of active intracellular offspring in Epulopiscium sp. type B is a recent modification of the sporulation program. Given the number of sporulation gene homologs found in the Epulopiscium genome, we consider it highly unlikely that the developmental program was assembled through multiple horizontal transfer events. Our comparative study reveals that genes essential for physical (e.g. engulfment of the offspring cell) and regulatory mechanisms (e.g. alternative sigma factors and intracellular communication) have been maintained in a live-offspring-bearing cell. We also found that genes involved in the synthesis of a modified form of peptidoglycan have been conserved. These may be important for offspring growth and intracellular development. Clearly further functional studies are required to test the above models and identify additional mechanisms acquired or modified during evolution of offspring formation in Epulopiscium. The findings presented here will provide a framework to begin to assess genetic programs expressed during development in Epulopiscium.

Methods

Epulopiscium sample collection and DNA extraction

Naso tonganus were collected by spear on outer reefs in the vicinity of Lizard Island, Great Barrier Reef, Australia. Sections of the gut were removed and the contents were fixed in 80% ethanol. Samples were stored at −20 °C upon arrival at the laboratory.

Epulopiscium cells were manually selected from fixed intestinal contents, using a standard Gilson pipettor and a Nikon SMZ-U dissecting microscope. Cell lysis and DNA extraction were performed as previously described [61]. Briefly, Epulopiscium type B cells were incubated with 100 μg/ml Proteinase K for one hour at 50 °C. DNA was extracted with phenol:chloroform and precipitated. After rinsing with 70% ethanol, the DNA was resuspended in TE (10 mM Tris, 1 mM EDTA, pH 8) buffer.

Genome sequencing and analysis

A draft genome sequence was generated using a random shotgun approach and paired-end Sanger sequencing. Sequence reads, providing approximately 8-fold coverage of the estimated 4 Mbp genome were assembled using a combination of the Celera Assembler [93] and TIGR Assembler [94]. To improve assembly, reads were first sorted based on G + C content and only reads that had 23 – 53% G + C were retained. The resulting reads were assembled again and open reading frames (ORFs) predicted using GLIMMER [95]. A BLAST [96] analysis of predicted ORFs against the National Center for Biotechnological Information (NCBI) non-redundant protein (nr) database was used to remove non-bacterial sequences. This strategy yielded 92 large contigs 12 kb to 119 kb in length. These contigs are deposited in GenBank under the accession number NZ_ABEQ01000000.

Draft genome assessment of the 92 large contigs was performed using two approaches. First, the total number of rRNA and tRNA genes were predicted using RNAmmer [97] and tRNAscan-SE [98], respectively. This analysis was also performed on the small contigs and singletons in the Epulopiscium draft genome. Second, a clusters of orthologous genes (COG) analysis [99] was performed using the Epulopiscium predicted proteome. Each protein was compared against a local COG database obtained from NCBI (ftp://ftp.ncbi.nih.gov/pub/mmdb/cdd/, accessed: 06/23/2011) using RPSBLAST [100] and the total number of proteins in each COG category was tabulated and represented as a percent of the number of COG-annotated proteins in the genome. This analysis was also performed for all sequenced genomes belonging to the phylum Firmicutes and genus Clostridium using the complete sequenced microbial genome collection from the NCBI (http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi, accessed: 06/23/2011).

Generation of a core sporulation gene list

An initial list of sporulation genes was created based on four studies describing the regulons of the four sporulation-specific sigma factors and the transcription factor Spo0A from B. subtilis[29,53,63,64]. Other genes reported in the GenoList Comparative Microbial Genome Browser [65] with putative sporulation functions in B. subtilis 168 were added to this list. Any gene coding for a protein with a known vegetative growth function was removed. This list was refined further by BLASTP (http://blast.ncbi.nlm.nih.gov/Blast.cgi) searches to determine the distribution of similar proteins in the NCBI nr database. Multiple searches were performed and limited to the phylum Firmicutes, class Clostridia, and if necessary Clostridium spp. and Bacillus spp. Proteins represented in 15 strains of endospore-forming bacteria in both class Bacilli and class Clostridia with greater than or equal to 70% query coverage and 20% identity were retained on the list. Proteins with top hits for putative homologs in two or more species within the non-spore-forming genera ListeriaStreptococcusLactobacillusStaphylococcusEnterococcus, and Peptostreptococcus, indicating a vegetative function, were removed. Several key proteins that were first identified as essential to sporulation (Spo0A, Spo0J, Soj, SpoIIIE and SpoIIIJ) were added back to the list.

Search for core proteins coded for in the Epulopiscium and C. lentocellum genomes

All contigs and well-represented single reads from the Epulopiscium sp. type B draft genome were concatenated to form a pseudomolecule. Coordinates for all junctions between contigs and single reads were entered into an Excel file for easy reference. The pseudomolecule and the complete C. lentocellum DSM 5427 genome [62] were searched for homologs of core proteins using TBLASTN and BLASTP, respectively. For each query, the top hits in the Epulopiscium or C. lentocellum genome were compared to Bacillus and Clostridium sequences in the NCBI nr database using BLASTP. If reverse-BLAST searches recovered a top hit that was different from the B. subtilis core protein used in the initial BLAST analysis, the protein was eliminated from the list. By this method, additional putative homologs with weak hits (below the original cutoff values) were identified. Operon structure was an additional criterion used to identify homologs. Core sporulation protein homologs identified in C. lentocellum were then used to search the Epulopiscium genome using TBLASTN. Multiple sequence alignments were performed using ClustalΩ [101].

Confirmation of dacF, gpr-spoIIP, spoVT, sspC/F, ybdM and yyaC

Primers were developed to amplify sporulation genes identified in the Epulopiscium genome that were located in small contigs or singletons or to link genes in operons predicted from synteny in the genomes of other spore formers (Table 5). PCR was performed using standard reaction conditions, and products were cloned into the pCR 2.1 TOPO vector (Invitrogen) following the protocol provided by the manufacturer. Sequences of the clones were determined using Big Dye Terminator chemistry and an Applied Biosystems Automated 3730 DNA analyzer, performed at the Cornell Life Sciences Core Laboratories Center and analyzed with Geneious Pro 5.4.2 (Biomatters). The sequences for the gpr-spoIIP operon [accession number JN402987], dacF [JN402985], spoVT [JN402986], sspC/F [JN402984], ybdM [JN402982], and yyaC [JN402983] are available from GenBank.

Table 5

Primers used in this study

DESIGNATION	SEQUENCE (5′ TO 3′)
GprendF	ATAGACGCATTAGGAGCACG
SpoIIPbegR	GCTTAGCGGACTTTGTATCACC
EpuloGprF	GAGAACATTGGTATTACAGGCG
EpuloGprR	GCTTGCATATATCACCTCCTTG
EpuloSpoIIPF	CATTGCTGTTCACCCAGGTA
EpuloSpoIIP859R	GCAGTAACCTTAGACGCA
EpuloSpoIIP684F	CAAAGTATGGGCTAAATGTATTGC
EpuloSpoIIPR	CAAAACAACAGACATCACCG
EpuloDacFF	GAGCCCCTGATTGTAACATTT
EpuloDacFR1	GGTTAATCCAAATTCACTTTCGCC
EpuloSpoVTF	TTTAGTATTATCAAGAGAAAAACAGCAT
EpuloSpoVTR	TGAACATTTGTCAAGATATAAATGCAA
EpuloSspC/FF	CTCCAAAATAATTTAGGAATATTGTCC
EpuloSspC/FR	TACACAGAAGTACCCCTTTGC
EpuloYbdMF	GAGTGGTTCTTTTAGTGGCTCTG
EpuloYbdMR	AACATGACCTCACACTGGCA
EpuloYyaCF	GCGGTGTTTGTTGTAAGTGC
EpuloYyaCR	TACACCCGAAGAATTAAGCA

Competing interests

The author(s) declare that they have no competing interests.

Authors’ contributions

ERA and DAM designed the study and wrote the initial manuscript draft. KDC and ERA collected samples. DAM and GS designed and performed the computational analyses. DAM and ERA interpreted results. All authors contributed to writing the manuscript and approved the final draft.