Deciphering Clostridium tyrobutyricum Metabolism Based on the Whole-Genome Sequence and Proteome Analyses

ABSTRACT Clostridium tyrobutyricum is a Gram-positive anaerobic bacterium that efficiently produces butyric acid and is considered a promising host for anaerobic production of bulk chemicals. Due to limited knowledge on the genetic and metabolic characteristics of this strain, however, little progress has been made in metabolic engineering of this strain. Here we report the complete genome sequence of C. tyrobutyricum KCTC 5387 (ATCC 25755), which consists of a 3.07-Mbp chromosome and a 63-kbp plasmid. The results of genomic analyses suggested that C. tyrobutyricum produces butyrate from butyryl-coenzyme A (butyryl-CoA) through acetate reassimilation by CoA transferase, differently from Clostridium acetobutylicum, which uses the phosphotransbutyrylase-butyrate kinase pathway; this was validated by reverse transcription-PCR (RT-PCR) of related genes, protein expression levels, in vitro CoA transferase assay, and fed-batch fermentation. In addition, the changes in protein expression levels during the course of batch fermentations on glucose were examined by shotgun proteomics. Unlike C. acetobutylicum, the expression levels of proteins involved in glycolytic and fermentative pathways in C. tyrobutyricum did not decrease even at the stationary phase. Proteins related to energy conservation mechanisms, including Rnf complex, NfnAB, and pyruvate-phosphate dikinase that are absent in C. acetobutylicum, were identified. Such features explain why this organism can produce butyric acid to a much higher titer and better tolerate toxic metabolites. This study presenting the complete genome sequence, global protein expression profiles, and genome-based metabolic characteristics during the batch fermentation of C. tyrobutyricum will be valuable in designing strategies for metabolic engineering of this strain.

IMPORTANCE Bio-based production of chemicals from renewable biomass has become increasingly important due to our concerns on climate change and other environmental problems. C. tyrobutyricum has been used for efficient butyric acid production. In order to further increase the performance and expand the capabilities of this strain toward production of other chemicals, metabolic engineering needs to be performed. For this, better understanding on the metabolic and physiological characteristics of this bacterium at the genome level is needed. This work reporting the results of complete genomic and proteomic analyses together with new insights on butyric acid biosynthetic pathway and energy conservation will allow development of strategies for metabolic engineering of C. tyrobutyricum for the bio-based production of various chemicals in addition to butyric acid.
inferior cheese quality mainly due to the undesirable flavor of butyric acid (3)(4)(5). C. tyrobutyricum can be employed as a great microbial cell factory for the production of various chemicals due to several advantages including its ability to grow on a minimal medium, high metabolic flux toward butyric acid, and tolerance to the product at high concentrations (6). Despite such potential, C. tyrobutyricum has received less attention as an industrial host strain compared to other clostridial species, mainly due to our limited understanding on its genotypic and metabolic characteristics at the genome level or scale. Most studies of this bacterium so far have been focused on the optimization of fermentation processes (7)(8)(9)(10). To date, only a few metabolic genes of C. tyrobutyricum have been knocked out (11)(12)(13). It has been reported that deletion of the pta gene encoding phosphotransacetylase (PTA) resulted in reduction of PTA activity up to 60%, and consequently increased butyric acid titer by 68% (12). Deletion of the ack gene encoding acetate kinase (AK) led to a similar result, increasing the butyric acid titer and yield (14). More recently, deletion of the ptb gene encoding phosphotransbutyrylase (PTB) has been demonstrated, but the butyric acid yield of the ptb mutant was similar to that of the wild-type strain despite its reduced PTB activity (13).
In addition to butyric acid, C. tyrobutyricum has great potential to be employed for the production of other chemicals. A recent study on the metabolic engineering of C. tyrobutyricum suggested that overexpression of C. acetobutylicum adhE2 in an ack-deleted C. tyrobutyricum strain resulted in production of 10 g liter Ϫ1 of 1-butanol from glucose (15). It was demonstrated that C. tyrobutyricum could tolerate 1-butanol at the initial concentration of 15 g liter Ϫ1 , at which a native 1-butanol producer Clostridium beijerinckii ATCC 55025 was not able to grow (15). More recently, optimizing the expression of Clostridium acetobutylicum adhE2 in C. tyrobutyricum led to the production of 20 g liter Ϫ1 of 1-butanol from mannitol without manipulating the competing pathways (16); this 1-butanol concentration is higher than those obtained by native 1-butanol producers (17,18). Even though mannitol is a more-reduced substrate than glucose, this result suggests that further improvement could be achieved in C. tyrobutyricum through systems metabolic engineering (19,20). More recently, a draft genome sequence of C. tyrobutyricum has been reported (21,22). Based on the draft genome sequence, proteomic studies of the wild-type and mutant strains of C. tyrobutyricum have been performed (23). However, the lack of complete genome sequence and transcriptomic and/or proteomic studies during the course of fermentation has limited our understanding of the physiological characteristics of C. tyrobutyricum for subsequent metabolic engineering at the systems level.
In this study, we report the complete genome sequence of C. tyrobutyricum KCTC 5387 (ATCC 25755) and its proteome profiles during the course of batch fermentation. Strikingly, the complete genome sequence suggested that this bacterium does not use PTB and butyrate kinase (BK) for butyric acid production; instead, a coenzyme A (CoA) transferase (CoAT) mediates butyric acid production from butyryl-CoA by reassimilation of acetic acid, which is produced by PTA and AK. Also, several energy conservation mechanisms in C. tyrobutyricum, which are responsible for its capability to efficiently produce butyric acid and tolerate toxic products, were identified by proteomic analyses during the course of fermentation.

RESULTS
Features of the complete C. tyrobutyricum genome. The general features of the C. tyrobutyricum genome are summarized in Table 1. The important statistics are generally in agreement with the draft sequences reported recently (21,22). The genome of C. tyrobutyricum KCTC 5387 contains one circular 3.07-Mbp chromosome and one circular 63-kbp plasmid (designated pCTK01; Fig. 1A). The strand bias of the chromosomal genes is one important characteristic in Firmicutes (24), and the origin and terminus of replication of the chromosome are located at nearly 0°and 180°, respectively; about 80% of the chromosomal genes are carried on the leading strand. The locations of the chromosomal replication origin and terminus are consistent with those predicted by GC skew analysis (Fig. 1). There are two 52-kbp duplicated regions at 116°and 122°on the chromosome; the majority of the genes in these clusters are those with unknown functions, and some of them encode phage-related proteins.
Like pCKL555A in Clostridium kluyveri (25), a species taxonomically close to C. tyrobutyricum, the genes on the pCTK01 plasmid have no clearly known functions, except for those encoding a DNA topoisomerase (CTK_P00130), a sigma factor (CTK_P00210), an arsenite transporter (CTK_P00580-610), and those encoding putative proteins for conjugal plasmid transfer (CTK_P00290-410). Plasmid pCTK01 shows neither the strand bias of the genes nor a clear transition of GC skew, suggesting that this plasmid might replicate via rolling-circle mechanism rather than theta-type mechanism (26). However, we could not find those known genes in pCTK01 for its replication, and thus, further studies are needed to understand how this plasmid replicates.
The genes in C. tyrobutyricum were functionally annotated using the Kyoto Encyclopedia of Genes and Genomes (KEGG) automatic annotation server (27), and the annotation results were compared with those for C. kluyveri and Clostridium ljungdahlii. Among the 3,138 protein-coding genes, 49.4% of them (1,552 genes) could be assigned with KEGG Orthology (KO) numbers. This result is slightly better than those of C. kluyveri (42.2%) and C. ljungdahlii (45.7%). Despite its smaller genome size, the number of KO categories assigned to C. tyrobutyricum genes was 1,257, which was comparable to those in C. kluyveri (1,248 KOs) and C. ljungdahlii (1,370 KOs).
Shotgun proteomic analyses of C. tyrobutyricum. In order to better understand metabolic characteristics of C. tyrobutyricum, protein expression profiles were examined during batch fermentation on glucose. Samples from duplicated batch fermentations of C. tyrobutyricum were subjected to shotgun proteomic analyses ( Fig. 2A). About 100 g liter Ϫ1 of glucose was completely consumed within 30 h, and the maximum optical density at 600 nm (OD 600 ) reached was about 38. The major fermentation products were acetic acid (11.1 g liter Ϫ1 ) and butyric acid (32.9 g liter Ϫ1 ), and larger or higher-carbon-numbered (ϾC 5 ) fatty acids were not produced. While the butyric acid yield (0.32 g g Ϫ1 glucose) was similar to that reported in a previous study (23), the acetic acid yield was slightly higher in our result (0.11 versus 0.074 g g Ϫ1 ), possibly due to the different composition of the culture media. As expected from its genome sequence, acetone and butanol were not detected throughout the fermentation, but small amounts of formic acid (3.1 g liter Ϫ1 ) and ethanol (0.7 g liter Ϫ1 ) were produced. Samples for proteome profiling were taken at early exponential (C2), late exponential (C4), and stationary (C6) phases ( Fig. 2A). Proteins isolated from two independent samples at the same phase were mixed and fragmented using trypsin. The resulting peptide fragments were analyzed by high-performance liquid chromatography coupled to tandem mass spectrometry (HPLC-MS/MS). All unique peptide sequences identified were compared to the predicted amino acid sequences of C. tyrobutyricum genes. Estimated false-positive rates of peptides were 0.3% for C2, 0.7% for C4, and 1.0% for C6. A total of 1,263 proteins were identified during the course of C. tyrobutyricum fermentation, and of these proteins, 774 proteins were identified at all phases ( Fig. 2B; see Data Set S1 in the supplemental material). The numbers of identified proteins at C2, C4, and C6 were 1,046, 1,082, and 963, respectively, which were categorized based on KO assignment in Table 2.
From the label-free proteome data obtained, the relative abundance of the proteins were calculated using the normalized spectral abundance factors (NSAF; Data Set S1). Compared to C2, about 85% and 66% of proteins at C4 and C6, respectively, showed changes in their expression levels within twofold. Highly abundant proteins in all phases included 50S ribosomal protein L7/L12 (CTK_28970), 60-kDa chaperonin GroEL (CTK_C26260), elongation factor Tu (CTK_C28900 and CTK_C29040), thiolase (CTK_01450), and 3-hydroxybutyryl-CoA dehydrogenase (CTK_26330). These results are similar to those obtained from the shotgun proteomic analyses of C. acetobutylicum (28). One exceptional case was flavodoxin (CTK_C21890), which was only the 465th most abundant at C2, but the most abundant at C4 and the 5th most abundant at C6.
For more accurate determination of protein fold changes between the samples, we additionally performed relative quantification of the peptides labeled by Tandem Mass Tag (TMT) isobaric tags in duplicates for each sampling point ( Fig. 2C; see Data Set S2 in the supplemental material). In this case, 587 proteins were identified. Proteins with significant fold changes (P value of 0.05) are shown in Table 3. Significantly upregulated genes include the genes involved in flagellar formation, iron transport, cofactor biosynthesis, stress response, and sporulation, whereas downregulated genes include genes encoding a transcription elongation factor, an ABC transporter, enzymes involved in fructose metabolism, and carbon monoxide dehydrogenase. Central metabolism of C. tyrobutyricum. An overview of the metabolic pathway in C. tyrobutyricum and changes in the expression levels of proteins involved in key metabolism are shown in Fig. 3. Shotgun proteomic data suggested that the phosphotransferase system (PTS) is mainly used in C. tyrobutyricum for glucose uptake. During the batch fermentation of C. tyrobutyricum on glucose, PTS enzyme I (CTK_C28440) and phosphocarrier protein Hpr (CTK_C17590) were identified at all phases. The results of genome sequence analysis suggest that there are two putative glucose-specific PTS enzyme II genes in C. tyrobutyricum: CTK_C27920 encoding only the IIA domain and CTK_C20580 encoding the entire set of IIA, IIB, and IIC domains. CTK_C27920 expression was not observed at any phase, while CTK_C20580 expression was observed in all three phases. Thus, the latter might be the major PTS II enzyme for glucose uptake. The expression levels of PTS II enzyme and HPr increased at C4 and C6, whereas PTS I decreased gradually.
Besides the PTS genes, three genes (CTK_C04350, -C07580, and -C21040) were predicted to encode sugar:H ϩ symporters, which might be used as alternative routes for glucose transport as in C. beijerinckii (29). However, only CTK_C04350 was weakly expressed at the stationary phase, suggesting that the PTS system described above is the major route for glucose transport in C. tyrobutyricum. In addition to glucose transporters, fructose-specific PTS (CTK_C06320; enzyme IIABC [EIIABC]) was abundantly expressed at the early exponential phase (C2), but its expression level decreased dramatically at the later phases (Table 3). This transient expression of the fru operon genes might be independent of catabolite repression in that the expression levels of catabolite control proteins, CcpA and histidine protein (HPr) kinase, were decreased at C6.
C. tyrobutyricum is found to catabolize glucose into pyruvate mainly via glycolysis. Genes related to the Entner-Doudoroff pathway and the oxidative branch of the pentose phosphate (PP) pathway are not present. Interestingly, a cluster (CTK_C23730 to -C23750) consisting of the rpi (ribose 5-phosphate isomerase), tal (transaldolase), and tkt (transketolase) genes was found in the genome, but none of their products were identified in the proteome. Additional rpi (CTK_C01400) and tkt (CTK_C09210) genes were found to be actively expressed, but no additional tal gene was found in the C. tyrobutyricum genome. As in other clostridial species, C. tyrobutyricum converts pyruvate into acetyl-CoA and CO 2 using pyruvate:ferredoxin oxidoreductase. Of the two genes encoding pyruvate:ferredoxin oxidoreductase (CTK_C21650 and CTK_C25530), CTK_C21650 was constitutively expressed (20th, 24th, and 30th most abundant at C2, C4, and C6 phases, respectively), while CTK_C25530 was weakly expressed only at the C2 phase. Formic acid is produced by pyruvate: formate lyase (PFL); the genes encoding PFL (CTK_C22190) and PFL-activating enzyme (CTK_C22180) were expressed at all of the sampling points, but its expression level decreased as cells grew.
Fermentative pathways in C. tyrobutyricum. Previously, the pta (12), ack (11), and ptb (13) knockout mutant strains of C. tyrobutyricum ATCC 25755 have been reported. Based on the complete genome sequence of the KCTC 5387 (same as ATCC 25755) strain determined in this study, the fermentative pathways in C. tyrobutyricum were revisited. As expected, acetic acid is produced mainly through PTA and AK; the pta (CTK_C18230) and ack (CTK_C18220) genes are present in the C. tyrobutyricum genome. As in C. acetobutylicum, the expression of pta and ack was constitutive (30,31). PTA was not identified after TMT labeling, but AK gradually increased from C2, reaching 1.5-fold increase at C6 (Fig. 3).
Our biggest surprise was that the genes encoding PTB and BK are missing in the whole genome of the KCTC 5387 strain and in the draft genome sequences of the ATCC 25755 strain that were previously reported (21,22). Consistent with this finding, our numerous attempts to amplify the partial ptb fragment according to the previous study (13) were not successful (see Fig. S4 in the supplemental material). This result suggests that the ptb and buk (C) Histograms that present relative protein fold changes at C4 and C6 compared to C2 based on TMT-tagged proteome samples. Lee et al. genes are indeed absent in the C. tyrobutyricum genome, and butyric acid is produced by C. tyrobutyricum using a pathway that is different from PTB-BK pathway used by C. acetobutylicum and C. beijerinckii (Fig. 4A). To further confirm this, the enzyme activities of PTB and BK were measured in the crude extracts of C. tyrobutyricum and C. acetobutylicum ATCC 824 (as a positive control). The crude extract of C. tyrobutyricum showed much lower activities of PTB and BK compared to those of C. acetobutylicum (Fig. 4B), strongly supporting the absence of PTB and BK in C. tyrobutyricum. The PTB and BK activities in the C. tyrobutyricum extract were 0.144 and 0.061 U/mg protein, which were only 2.7% and 1.6% of those in the C. acetobutylicum extract, respectively; the residual weak PTB/BK activities seem to come from the cross-activities of PTA/AK. It was suggested that butyrate:acetate CoAT is required for butyrate biosynthesis in C. kluyveri (25). There are four genes encoding different CoA transferases in the C. tyrobutyricum genome. Products of two genes (CTK_C06520 and CTK_C07660) are homologous to C. kluyveri CoAT (25), and the other two genes (CTK_C16610 and CTK_C28060) share homology with the propionate:acetate CoAT from C. propionicum (32). There were no orthlogs of C. acetobutylicum CtfAB in C. tyrobutyricum as previously suggested (23). As the cell extract of C. tyrobutyricum clearly showed CoAT activity (Fig. 4C), we aimed to identify the key CoAT responsible for butyrate formation. Based on both reverse transcription-PCR (RT-PCR) (Fig. 4D) and proteome analyses, it was concluded that the cat1 (CTK_C06520) gene is the only gene responsible for butyrate biosynthesis. To see whether CoAT encoded by the cat1 gene is truly responsible for butyrate formation, this gene was knocked out using a mobile group II intron. However, insertion of a mobile group II intron into the coding sequence (CDS) in the sense direction (cat1::intS strain) did not decrease butyric acid production during batch fermentation (data not shown). RT-PCR of the cat1::intS strain suggested that the other putative CoAT genes were not turned on in consequence of cat1 inactivation either (Fig. 4D); in fact, only the wild-type mRNA of the cat1 gene was observed, suggesting that the active Cat1 was still produced after the resplicing of the mutant mRNA due to the presence of the knockout plasmid. To solve this problem, we tried plasmid curing in the cat1::intS strain and inserting the intron in the antisense direction, both of which were not successful. Insertion of the intron in the antisense direction into the pta and ack genes was also not successful; as well-known in the field, knocking out certain genes in clostridial strains can be extremely difficult, which was the case in this study.
Energy conservation. There are three energy conservation mechanisms found in the C. tyrobutyricum genome (Fig. 3): Rnf complex (CTK_C13030-CTK_C13080), NfnAB (CTK_C26320 and CTK_C26310), and pyruvate-phosphate dikinase (CTK_C24060). The Rnf complex generates NADH and proton gradient using reduced ferredoxin (25,33). The proton gradient generated can be used to generate ATP via F o F 1 AT-Pase. The NfnAB complex was first characterized in C. kluyveri, and it generates two NADPH molecules by the oxidation of one reduced ferredoxin and one NADH (34). Pyruvate-phosphate dikinase (PPDK) can generate ATP using phosphoenolpyruvate, AMP, and pyrophosphate (35).
Based on the proteomic analysis, the major energy conservation mechanisms when using glucose involve NfnAB and PPDK. The expression of the Rnf complex was apparently weak; only RnfC was identified in the label-free shotgun proteome. Both NfnA and NfnB were strongly expressed at early exponential phase; NfnA and NfnB were the 82nd and the 38th most abundant proteins, respectively, in the label-free proteome at C2. However, expression of NfnAB gradually decreased as cells grew; both NfnA and NfnB in the TMT-labeled proteome decreased by about 50% at C6 compared to C2. Expression of PPDK was not strong at C2 (the 741st most abundant); it was upregulated at C4 (1.6-fold compared to C2) and C6 (2.2-fold increase compared to C2). Sporulation. Sporulation is a key characteristic of Bacillus and Clostridium species (30,36). Even though Clostridium shares many proteins involved in the sporulation process with Bacillus, the mechanism of sporulation initiation in Clostridium is known to be different from that in Bacillus. The clostridial sporulation program has recently been reviewed thoroughly (36), but it might be different even among clostridial species. Since C. tyrobutyricum is able to undergo the sporulation process as well, the expression patterns of the sporulation genes were compared with those in other clostridial species.
Several genes for the initiation of sporulation were found in the genome, and the changes in the expression levels of some of these genes during the fermentation are shown in Fig. S3 in the supplemental material. The initiation of sporulation appears to be similar to those in other clostridial species. The expression level of Spo0A (CTK_C1943), the master regulator of sporulation, increased as fermentation proceeded. The expression level of AbrB protein (CTK_00640), which represses the sigH gene (CTK_29050) encoding H and is repressed by phosphorylated Spo0A, decreased. This further promotes increased expression of Spo0A. It has been reported that orphan histidine kinases (HKs) directly phosphorylate Spo0A in clostridia (14,37,38). Two genes encoding putative orphan HKs (CTK_02350 and CTK_C04290) are present in the genome. Also, there is evidence in clostridial species that K is involved at the early sporulation cascade (39,40).
Once the forespore is developed, expression of the sporulation genes is spatially differentiated (see Fig. S3 in the supplemental material). While the early sporulation proteins could be identified in the proteome, those involved in the later stages seemed to be expressed rather weakly. It is known that in the forespore, spoI-IAA (CTK_C08840), spoIIAB (CTK_C08850), and spoIIE (CTK_C29350) are involved in the control of F (CTK_C08860) activity. The abundance of these proteins was much lower than that of Spo0A. Even though SpoIIAA and SpoIIE were not identified at some or all phases in nonlabeled experiments, it was possible to observe their relative changes after TMT tagging. SpoIIAB was induced at C4 and showed strongest expression at C6, whereas SpoIIE slightly decreased at C4 and increased at C6 compared to C2. This pattern is similar to previous cDNA microarray studies in C. acetobutylicum (31,41). The expression levels of SpoIIAA and F in the total protein pool were much lower than SpoIIAB (Fig. S3 in the supplemental material), which is unexpected as spoIIAA, spoIIAB, and sigF constitute an operon in C. tyrobutyricum as in other species. Both E and F were identified only at C6.

DISCUSSION
C. tyrobutyricum is a potentially versatile microbial cell factory for the production of various chemicals due to its proven ability to produce butyric acid to a high titer with high-level product tolerance. It is however difficult to perform metabolic engineering of this strain due to the lack of understanding of its genotypic and metabolic characteristics at the genome level. This is why we de-cided to determine the complete genome sequence beyond an available draft genome sequence and perform genome-wide proteomic studies during the course of batch fermentation. One interesting feature of the C. tyrobutyricum genome is its compactness; its genome size is about 78% and 65% of that of its closest relatives C. kluyveri and C. ljungdahlii, respectively ( Table 1). The C. tyrobutyricum genome contains 20 multiplicated genes encod- products were identified are indicated in blue. Genes indicated in black are those genes whose products were not identified in TMT-labeled samples but were identified in label-free samples. Genes indicated in red are those genes whose products were not identified both in TMT-labeled and label-free peptide samples. Glucose-6-P, glucose-6-phosphate; Fd ox , oxidized ferredoxin.
ing putative transposases, whereas at most, 5 genes are present in the C. ljungdahlii genome. This large number of insertion sequence (IS) elements might contribute to the loss of genes in C. tyrobutyricum via homologous recombination (42). A region with inverted GC skew located from 318°to 322°of the chromosome strongly supports our hypothesis (Fig. 1). This inverted region is flanked by two duplicated transposase genes (CTK_C27190 and CTK_C27510) oriented in the opposite direction, indicating that this inversion is due to homologous recombination. From the whole-genome and shotgun proteome data, most genes involved in central metabolism could be identified. On the basis of the genome information, C. tyrobutyricum depends on the nonoxidative PP pathway to synthesize precursors of nucleotides and amino acids. However, transaldolase was not identified in the shotgun proteome; additional rpi and tkt genes, distantly located from the rpi-tal-tkt operon, were actively expressed (Fig. 3). It was suggested in Clostridium stercorarium, which does not have a transaldolase gene but utilizes xylose as the sole carbon source, that 3-deoxy-D-arabino-heptulosonate-7-phosphate synthase might replace the role of transaldolase (43). In C. tyrobutyricum, this gene (CTK_C27090) was strongly expressed (see Data Set S1 in the supplemental material), supporting this hypothesis.
As clostridial species do not have an oxidative phosphorylation mechanism, a major role of tricarboxylic acid (TCA) cycle might be the generation of precursors for various building blocks including amino acids. It is suggested that C. acetobutylicum uses a bi-furcated TCA cycle (44,45). Several key enzymes, including (Re)citrate synthase, pyruvate carboxylase, NAD-dependent malic enzyme and malate dehydrogenase, were identified in the shotgun proteome of C. tyrobutyricum. Unlike C. acetobutylicum, however, ␣-ketoglutarate:ferredoxin oxidoreductase, which is required for the synthesis of succinyl-CoA from ␣-ketoglutarate, is absent in the C. tyrobutyricum genome and also in the genomes of C. kluyveri and C. ljungdahlii. Since these strains have only homoserine O-succinyltransferase, how succinyl-CoA is synthesized for methionine biosynthesis remains a question.
In many clostridial species, ferredoxin mediates the electron transfer of several enzymes including hydrogenase and pyruvate: ferredoxin oxidoreductase. Pyruvate:ferredoxin oxidoreductase uses ferredoxin as the major electron acceptor under high-iron conditions, but flavodoxin is an alternative electron acceptor when iron is limited (46). The expression of one flavodoxin protein (CTK_C2203) dramatically increased from the late exponential phase (ca. 10-fold); in fact, estimated abundance of the flavodoxin protein based on its NASF value indicated that this protein constituted ca. 3% of the total proteins identified. This result is in agreement with the previous findings using Clostridium formicoaceticum (46).
One of the most striking findings was that C. tyrobutyricum uses the CoAT pathway for butyric acid production, rather than the PTB-BK pathway. The strain employed for genome sequencing in this study was C. tyrobutyricum KCTC 5387, which is the same strain as ATCC 25755 that has been draft sequenced before. The partial sequence of the pta gene reported for ATCC 25755 (12) is almost identical to that of the KCTC 5387 strain (see Fig. S1 in the supplemental material for amino acid sequence alignment). In the case of ack (11), however, neither nucleotide nor amino acid sequence is similar to that of the KCTC 5387 strain (see Fig. S2 for amino acid sequence alignment). In fact, the partial AK sequence reported by X. Liu et al. (11) is exactly the same as that of the Acinetobacter sp. strain AK. Y. Zhang et al. (13) have reported inactivation of the ptb gene in C. tyrobutyricum ATCC 25755. However, there were neither ptb nor buk in the complete genome of the C. tyrobutyricum KCTC 5387 strain, which was further supported by in vitro enzyme assay (Fig. 4B). Just in case the KCTC 5387 strain deposited at the Korean Collection for Type Cultures (KCTC) is different from ATCC 25755, we ordered a new ATCC 25755 strain from ATCC and examined the newly obtained strain for the presence of ptb; the ptb fragment could be amplified neither from KCTC 5387 nor from ATCC 25755 and DSM 2637 (Fig. S4). Thus, the previous results reported on the presence and knockout of ptb (13) need to be validated.
Despite the absence of ptb and buk, C. tyrobutyricum is still capable of producing a large amount of butyric acid using butyrate:acetate CoAT ( Fig. 4C; see Fig. S5 in the supplemental material). Based on our results of homology-based search, RT-PCR, and proteome analyses, the cat1 gene was found to be responsible for this activity (Fig. 3 and 4D). We tried disrupting the pta, ack, and cat1 genes using a mobile group II intron, but all of these attempts were not successful. On the basis of our hypothesis (Fig. 4A), elimination of acetate biosynthesis would lead to no fermentation end product; thus, the pta and ack genes are essential. It is known that electron transfer from NADH to ferredoxin is thermodynamically infeasible in clostridial species, and butyryl-CoA dehydrogenase bifurcates electrons from two NADHs to crotonyl-CoA and ferredoxin (47). Reduced or lack of butyric acid production seems to affect the electron bifurcation, which makes the cells unable to oxidize NADH; this is most likely why the cat1 gene could not be knocked out. The ATP yield obtainable through the use of butyrate:acetate CoAT is equivalent to that obtainable with PTB-BK because butyrate formation by CoAT is coupled with acetate formation during which ATP is generated.
Endospore formation is one of the important characteristics of clostridia. For the initiation of sporulation, clostridial strains directly phosphorylate Spo0A by orphan HKs. As in other clostridial species, spo0F and spo0B are not present in the C. tyrobutyricum genome, suggesting that the Spo0A protein might be directly phosphorylated by orphan HKs. However, even though two genes encoding putative orphan HKs (HKs; CTK_02350 and CTK_C04290) are found in the C. tyrobutyricum genome, the products of these two genes were not identified in the proteome samples. The current clostridial sporulation model is based on the transcriptional changes, and our result does not necessarily suggest that C. tyrobutyricum has a different mechanism for sporulation initiation. This should be addressed in the future by construction of various mutant strains and comparison of transcriptional and translational level changes in these mutants.
In this paper, we reported the complete genome of C. tyrobutyricum KCTC 5387 (ATCC 25755). On the basis of genomic and accompanying proteomic analyses during the course of batch fermentation, we were able to provide new insights on its butyric acid production pathway, energy conservation mechanisms, sporula-tion processes, and other metabolic characteristics. The results and data presented in this paper will be valuable in better understanding the physiological characteristics of C. tyrobutyricum and developing metabolic engineering strategies for constructing high-performance strains capable of producing various chemicals of interest. As stated above, it will also be important to validate previous results of butyric acid production through the reported PTB-BK pathway (13), as our results suggest that this pathway does not exist in the same strain examined (KCTC 5387 or ATCC 25755); instead, butyric acid is produced through the CoAT pathway in C. tyrobutyricum.

MATERIALS AND METHODS
Batch fermentation of C. tyrobutyricum. C. tyrobutyricum spores suspended in 2ϫ YTG medium (glucose, 5 g liter Ϫ1 ; NaCl, 4 g liter Ϫ1 ; Bacto yeast extract, 10 g liter Ϫ1 ; Bacto tryptone, 16 g liter Ϫ1 ) supplemented with 15% (wt/vol) glycerol were inoculated in 10 ml of 2ϫ YTG medium and heat shocked at 80°C for 10 min to kill the vegetative cells remaining in the suspension. After the heat shock, the test tube was incubated at 37°C in an anaerobic cabinet (Coy Laboratory Products, MI). A 500-ml capped flask containing 200 ml of the SP2 medium was inoculated with 5 ml of the 2ϫ YTG culture grown overnight and incubated at 37°C until the OD 600 reached to 1.0 to 2.0. This flask culture was used to inoculate a 5-liter bioreactor (LiFlus GX; Biotron, Gyeonggi-do, South Korea) containing 1.8 liters of SP2 medium. The SP2 medium contains the following ingredients (all per liter): glucose, 100 g; KH 2 PO 4 , 0.5 g; K 2 HPO 4 , 0.5 g; MgSO 4 ·7H 2 O, 0.2 g; MnSO 4 ·5H 2 O, 0.014 g; FeSO 4 ·7H 2 O, 0.01 g; (NH 4 ) 2 SO 4 , 2 g; Bacto yeast extract, 5 g; CH 3 COONa·3H 2 O, 5 g. During the fermentation, agitation speed and temperature were controlled at 200 rpm and 37°C, respectively. The pH of the fermentation medium was maintained at 6.0 by automatic addition of 7.5 M ammonia solution.
Isolation of genomic DNA and genome sequencing. For genomic DNA isolation, C. tyrobutyricum KCTC 5387 grown in 2ϫ YTG medium was harvested at late exponential phase. Total DNA from C. tyrobutyricum KCTC 5387 used for the genome sequencing was extracted using a G-Spin total DNA extraction minikit (Intron Biotechnology, Seongnam, South Korea) according to the manufacturer's protocol. The DNA prepared was fragmented to construct the library which was read using an Ion Torrent PGM 316 Chip sequencer at Genotech (Daejeon, South Korea). A total of 3,024,966 reads with the average length of ca. 230 bp were obtained and assembled into 96 contigs by a CLC Genomics Workbench 5.5.2 software (CLC Bio, Denmark). The maximum and average sizes of the contigs were ca. 190 kb and 31 kb, respectively. The complete genome sequence was assembled by comparing our contig sequences to other shotgun genome sequences of C. tyrobutyricum DSM 2637 and ATCC 25755 (21,22) and performing PCR analysis. The ambiguous sequences compared to other draft genomes were further confirmed by PCR and Sanger sequencing.
Sample preparation for proteome analysis. C. tyrobutyricum cells (about 10 OD 600 /ml) were homogenized in 1 ml of lysis buffer (10 mM Tris-HCl [pH 8.0], 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol [DTT], 0.5 mM Pefabloc SC, and 0.1% SDS) using Precellys 24 (Bertin Technologies, France) three times for 20 s each time at 6,500 rpm with a 5-min pause between homogenization cycles to prevent overheating. The supernatant was separated by centrifugation at 18,000 rpm for 30 min at 4°C, the protein concentration was determined using a Bradford protein assay (Bio-Rad), and aliquots were stored at Ϫ80°C. Each protein extract (100 g) was reduced with 500 mM tris(2-carboxyethyl)phosphine (TCEP) at room temperature for 60 min and then alkylated with 500 mM iodoacetamide (IAA) at room temperature in the dark for 60 min. The samples were desalted using a membrane filter of 10 kDa cutoff and dissolved in 200 mM triethylammonium bicarbonate (TEAB) buffer to a final concentration of 1 g/l. Sequencing-grade trypsin (Promega, Madison, WI) was added at 1:20 (wt/wt) into proteins in TEAB buffer and incubated overnight at 37°C. TMT labeling. Each peptide sample of individual proteins extracted from three C2, C4, and C6 sample points of Escherichia coli was divided into two fractions and was individually labeled using a TMTSixplex (TMT stands for tandem mass tag) reagents kit (catalog no. 90064, Thermo Scientific, MA) according to the manufacturer's protocol. TMT-126 and TMT-129 were used to label C2, TMT-128 and TMT-131 were used to label C4, and TMT-127 and TMT-130 were used to label C6. Aqueous hydroxylamine solution (5%, wt/vol) was added to quench the reaction. The six samples were then combined, speed vacuum dried, and then dissolved in 50 l of Milli-QϪwater containing 0.1% formic acid for liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) analysis.
2D-LC-MS/MS analysis. The TMT-labeled or label-free samples were analyzed using a two-dimensional LC-MS/MS (2D-LC-MS/MS) system consisting of a nanoACQUITY ultraperformance LC system (Waters, USA) and an LTQ Orbitrap Elite mass spectrometer (Thermo Scientific, USA) equipped with a nanoelectrospray source. The 2D-LC-MS/MS analysis was performed as previously reported by M. P. Washburn et al. (48) with slight modifications. Briefly, a strong cation exchange (SCX) (5 m, 3 cm) column was placed just before the C 18 trap column (inner diameter [i.d.], 180 m; length, 20 mm; particle size, 5 m; Waters). Peptide solutions were loaded in 5-l aliquots for each run. Peptides were displaced from the SCX phase to the C 18 phase by a salt gradient that was introduced through an autosampler loop and desalted for 10 min at a flow rate of 4 l/min. The trapped peptides were then separated on a 200-mm homemade microcapillary column consisting of C 18 (Aqua; particle size, 3 m) packed into 100-m silica tubing with an orifice i.d. of 5 m. A 10-step salt gradient was performed using 3 l of 0, 25, 50, 100, 250, and 500 mM ammonium acetate (0.1% formic acid in 5% acetonitrile [CAN]) and 4, 5, 9, and additional 9 l and 500 mM CAN (0.1% formic acid in 30% ACN). The mobile phases A and B were composed of 0 and 100% acetonitrile, respectively, and each contained 0.1% formic acid. The LC gradient began with 5% mobile phase B for 1 min and was ramped to 20% mobile phase B over 5 min, to 50% mobile phase B over 80 min, to 95% mobile phase B over 4 min, and remained at 95% mobile phase B over 10 min and 5% mobile phase B for another 5 min. The column was reequilibrated with 5% mobile phase B for 15 min before the next run. The voltage applied to produce an electrospray was 2.0 kV. During the chromatographic separation, the LTQ Orbitrap Elite mass spectrometer was operated in a datadependent mode. The MS data were acquired using the following parameters: five data-dependent collision-induced dissociation-high energy collision dissociation (CID-HCD) dual MS/MS scans per full scan; CID scans acquired in linear trap quadrupole (LTQ) with two-microscan averaging; full scans and HCD scans were acquired in Orbitrap at a resolution of 60,000 and 15,000, respectively, with two-microscan averaging; 35% normalized collision energy (NCE) in CID and 45% NCE in HCD; Ϯ1-Da isolation window. Previously fragmented ions were excluded for 60 s. In CID-HCD dual scan, each selected parent ion was first fragmented by CID and then by HCD.
Protein identification and quantification. MS/MS spectra were analyzed using the following software analysis protocols with the protein sequence set of all C. tyrobutyricum proteins. The reversed sequences of all proteins were appended into the database for calculation of the falsediscovery rate (FDR). ProLucid (49) was used to identify the peptides, with a precursor mass error of 25 ppm and a fragment ion mass error of 600 ppm. Trypsin was selected as the enzyme, with three potential missed cleavages. TMT modification (ϩ 229.1629) at the N terminus and lysine residue by the labeling reagent and carbamidomethylation at cysteine were chosen as static modifications. Oxidation at methionine was chosen as a variable modification. The CID and HCD tandem MS spectra from the same precursor ion are often combined by software to allow better peptide identification and quantification (50). We used a homemade software program where reporter ions from the HCD spectrum were inserted into the CID spectrum with the same precursor ion at the previous scan. Reporter ions were extracted from small windows (Ϯ20 ppm) around their expected m/z in the HCD spectrum. The output data files were filtered and sorted to compose the protein list using the DTASelect (51) with two and more peptide assignments for a protein identification and a falsepositive rate of less than 0.01.
A quantitative analysis was conducted using Census in the IP2 pipeline (Integrated Proteomics, USA). The intensity at a reporter ion channel for a protein was calculated as the average of this reporter ion's intensities from all constituent peptides from the identified protein (52). The measured intensity ratios of proteins were transformed to log 2 scale. The resulting ratios were log transformed (base 2) in order to achieve a normal distribution.
In order to control quality in protein quantification, we labeled the same sample with two different TMT reagents. Since every protein was expected to have an abundance ratio of 1:1, in comparisons of different TMT-labeled samples for the same sample, data with more than 30% variation was excluded (for example, TMT-126 and TMT-129 for the C2 sample). TMT ratios for proteins were calculated over average reporter ion intensities over all peptides assigned to a protein subgroup. Proteins with P Ͻ 0.05 were defined as significantly changed in abundance. The false-discovery rate was performed by selecting a P value from the two biological replicates and ensuring that the fold change was significant in the other two conditions.
Inactivation of the cat1 gene using a mobile group II intron. A retargeted intron fragment was amplified using primers in Table S1 in the supplemental material and cloned into pCACYS3 according to the previous studies (17,53). After verification of the intron sequence, the retargeted pCACYS3 plasmid was digested with XmaI, and the 3.9-kb fragment containing the whole L1.LtrA cassette was subcloned into the XmaIdigested pMTL82251 plasmid. The resulting plasmid was transferred to E. coli CA434 (donor strain) and finally transferred to C. tyrobutyricum by conjugation according to the previous reports (15,54). After 12-h mating, transconjugants were selected onto 2ϫ YTG agar containing 240 g ml Ϫ1 of D-cycloserine and 50 g ml Ϫ1 of erythromycin. Transconjugant colonies were streaked again onto selective agar to exclude donor E. coli cells, and gene disruption was confirmed by colony PCR.
Preparation of cell-free extracts. Either C. tyrobutyricum or C. acetobutylicum cells were grown in 200 ml of the SP2 medium at 37°C. When OD 600 reached about 1.0, the cells were harvested by centrifugation of 40 ml of the culture at 2,200 ϫ g and 4°C for 15 min. The cells were washed once using distilled water, resuspended in 1 ml of 0.1 M potassium phosphate buffer solution (pH 7.4), and finally homogenized using a VC 750 sonicator (Sonics & Materials, CT). Cell debris was removed by centrifugation at 16,100 ϫ g and 4°C for 10 min, and clear supernatants were used for enzyme assay. Total protein concentration was determined using the Bradford assay.
Enzyme assay. Phosphotransbutyrylase activity was determined by measuring liberation of free CoA from butyryl-CoA (55). Butyrate kinase activity was determined in the butyryl-phosphate-forming direction by measuring butyrohydroxamic acid (56,57). CoAT activity was measured in the acetyl-CoA-forming direction by the method of W. Buckel et al. (58), using butyryl-CoA and acetate as the substrates; formation of acetyl-CoA was determined by liberation of free CoA by a coupled citrate synthase reaction (58). An Ultrospec 3100 Pro spectrophotometer (GE Healthcare, United Kingdom) was used for the enzyme assay. One unit of the enzyme was defined as conversion of 1 M substrates per min.
RNA isolation and RT-PCR. For RNA isolation for reverse transcription-PCR (RT-PCR), all strains were cultivated in the anaerobic chamber at 37°C. An overnight 2ϫ YTG culture of each C. tyrobutyricum strain was diluted 50 times into 200 ml of the SP2 medium (100 g liter Ϫ1 of initial glucose) contained in a 500-ml screw-cap flask and kept at 37°C until the OD 600 reached 0.9 to 1.1. One milliliter of each SP2 culture was collected, and cells were harvested for 5 min at 4°C and 8,000 ϫ g. RNA was isolated according to a hybrid protocol of an acid guanidinium thiocyanate-phenol-chloroform method using Easy-Red BYF (IntronBio, Gyeonggi-Do, South Korea) and a spin column purification method using RNeasy mini (Qiagen, Valencia, CA). Reverse transcription was performed using RocketScript RT premix (Bioneer, Daejeon, South Korea) according to the manufacturer's protocol; 1 g of total RNA and 100 pmol of random nonadeoxynucleotides (dN 9 ) were subjected to a 20-l RT reaction. When using gene-specific primers, 20 pmol of the reverse primer, which can bind to the target mRNA, was used instead of dN 9 . If required, DNase I was used to treat total RNA prior to RT reactions to destroy residual genomic DNA fragments. PCR was performed using the AccuPower PCR premix (Bioneer; 20-l total reaction mixture volume). A T100 thermocycler (Bio-Rad, Hercules, CA) was used for all RT and PCRs throughout the study. The thiolase gene was chosen as an internal control of RT-PCR based on our shotgun proteome data (see Data Set S1 in the supplemental material). Primer sequences for PCR are described in Table S2 in the supplemental material.
Nucleotide sequence accession numbers. The genome sequence for C. tyrobutyricum has been deposited at GenBank under the following accession numbers: CP014170 (chromosome) and CP014171 (plasmid).