EtfA catalyses the formation of dipicolinic acid in Clostridium perfringens


*E-mail; Tel. (+1) 540 2312529; Fax (+1) 540 2319307.


Dipicolinic acid (DPA) is a major component of bacterial endospores, comprising 5–15% of the spore dry weight, and is important for spore stability and resistance properties. The biosynthetic precursor to DPA, dihydro-dipicolinic acid (DHDPA), is produced by DHDPA synthase within the lysine biosynthesis pathway. In Bacillus subtilis, and most other bacilli and clostridia, DHDPA is oxidized to DPA by the products of the spoVF operon. Analysis of the genomes of the clostridia in Cluster I, including the pathogens Clostridium perfringens, Clostridium botulinum and Clostridium tetani, has shown that no spoVF orthologues exist in these organisms. DPA synthase was purified from extracts of sporulating C. perfringens cells. Peptide sequencing identified an electron transfer flavoprotein, EtfA, in this purified protein fraction. A C. perfringens strain with etfA inactivated is blocked in late stage sporulation and produces ≤ 11% of wild-type DPA levels. C. perfringens EtfA was expressed in and purified from Escherichia coli, and this protein catalysed DPA formation in vitro. The sequential production of DHDPA and DPA in C. perfringens appears to be catalysed by DHDPA synthase followed by EtfA. Genome sequence data and the taxonomy of spore-forming species suggest that this may be the ancestral mechanism for DPA synthesis.


Dipicolinic acid (DPA) is a major component of all bacterial endospores, comprising 5–15% of the spore dry weight (Paidhungat et al., 2000). Mutant strains that are deficient in production of DPA or its accumulation into the developing spore produce spores with greatly decreased stability and/or resistance to wet heat (Paidhungat et al., 2000; Magge et al., 2008; Paredes-Sabja et al., 2008). In Bacillus subtilis, DPA is produced during late stage sporulation via a branch off of the lysine biosynthetic pathway. One step in this pathway is catalysed by dihydro-dipicolinate synthase (DHDPA synthase), which is encoded by dapA, and disruption of this gene results in the formation of spores deficient in DPA (Szulmajster et al., 1970; Daniel and Errington, 1993). DHDPA synthase has long been thought to condense aspartate semialdehyde (L-ASA) and pyruvate to produce dihydrodipicolinic acid (DHDPA) (Yugari and Gilvarg, 1965) (Fig. 1). Analysis of the reaction catalysed by DHDPA synthase using NMR spectroscopy has suggested that the product of this enzyme is not DHDPA, but rather 4-hydroxy-tetrahydro-dipicolinic acid, which undergoes spontaneous dehydration to DHDPA (Boughton et al., 2008). Production of DPA from L-ASA and pyruvate in vitro has been accomplished with a sporulating Bacillus megaterium cell extract (Bach and Gilvarg, 1966), apparently requiring the presence of both a DHDPA synthase and a DPA synthase. The latter enzyme is the product of the spoVF operon, which produces two products; spoVFA encodes a putative dehydrogenase while the spoVFB product appears to be a flavoprotein. Inactivation of either gene in B. subtilis resulted in loss of DPA production, and expression of both genes in Escherichia coli under conditions that induce lysine biosynthesis resulted in DPA synthesis (Daniel and Errington, 1993).

Figure 1.

Pathway of DPA formation. L-ASA, HTPA and DHDPA are intermediates of the lysine biosynthesis pathway. Reaction 1, the condensation of L-ASA and pyruvate, is catalysed by DapA (DHDPA synthase). The conversion of HTPA to DHDPA is thought to occur by a spontaneous dehydration reaction and is represented with the dotted arrow. Reaction 2, the conversion of DHDPA to DPA is catalysed by SpoVF in B. subtilis and by EtfA in C. perfringens.

Genome sequence analyses have shown that members of the Cluster I clostridia lack genes with significant homology to spoVF (Stragier, 2002; Onyenwoke et al., 2004) yet produce DPA during sporulation (Paredes et al., 2005). This Cluster is of high significance due to the inclusion of the human pathogens Clostridium perfringens, Clostridium botulinum and Clostridium tetani, and the industrially important Clostridium acetobutylicum and Clostridium beijerinckii. The lack of this enzyme is not a general characteristic of the clostridia, as genome sequence analyses indicate that some species in other clusters possess clear homologues of spoVF. To identify the enzyme responsible for the formation of DPA in C. perfringens we developed a modified version of the Bach and Gilvarg assay system (Bach and Gilvarg, 1966) to purify DPA synthase activity and identified proteins in the purified fraction. We have determined that CPR_2284, which encodes an electron transfer flavoprotein α-chain (EtfA), is directly involved in DPA synthesis in C. perfringens, is likely to be important in this process in all of the Cluster I clostridia, and may represent the ancestral enzyme for DPA production.


Purification of the C. perfringens DPA synthase

The C. perfringens and other Cluster I clostridia do not possess homologues of the characterized DPA synthase of the bacilli (Table S1), despite the fact that the spores produced by these organisms contain DPA. While TBLASN (Altschul et al., 1990) searches reveal homologues of B. subtilis SpoVFA and SpoVFB encoded in the genomes of numerous clostridia and related species (Table 1), similarities to proteins encoded in the genomes of Cluster I clostridia are minimal and extend over only short regions of the protein sequences. All of the species possessed strong homologues of the B. subtilis spoIVB gene (Table S1), which was previously found to be a good marker gene for sporulating species (Onyenwoke et al., 2004).

Table 1.  Purification of C. perfringens DPA synthase.
FractionTotal protein (mg)Specific activity (μg DPA/mg protein)Yield (%)Fold purification
Crude extract9010.081001.0
55–70% (NH4)2SO41720.3885.84.5
HiTrap QXL131.524.617.0
Superdex 2001.88.620.3102

To determine how DPA is made in the Cluster I organisms we developed a biochemical assay for the DPA synthase of C. perfringens. The assay included L-ASA, pyruvate and a source of DHDPA synthase activity. Previous in vitro studies in which DPA was detected by absorbance at 269 nm suggested that DPA might form spontaneously from DHDPA at high pH (Kimura, 1974; Kimura and Sasakawa, 1975). Analysis of the products of our activity assays with multiple reaction monitoring (MRM) mass spectrometry demonstrated that negligible DPA was formed in the presence of a B. subtilis strain FB106 (spoVF::tet) (Paidhungat et al., 2001) cell extract containing DHDPA synthase alone. While an increase in absorbance at 269 nm was observed, it was significantly greater than that expected from the amount of DPA produced. These results strongly suggest that a DPA synthase is required for rapid DPA production at the physiologically relevant conditions present in our assay.

Analyses of whole-cell extracts of C. perfringens NCTC 8679, a strain shown to produce spores with high DPA content (Orsburn et al., 2008), indicated that DPA synthase activity in this strain was greatest 11 h after inoculation into DSSM medium. A summary of the purification of DPA synthase from cells at this stage is shown in Table 1. The whole-cell extract was first separated into five fractions by ammonium sulphate [(NH4)2SO4] precipitation. Unlike the whole-cell extract, none of the individual fractions were capable of producing DPA. Only the 55–70% (NH4)2SO4 fraction was capable of producing DPA when added to a crude extract of B. subtilis strain FB106 (spoVF::tet) (Paidhungat et al., 2001) as the source of DHDPA synthase. To determine which (NH4)2SO4 fraction contained the C. perfringens DHDPA synthase, each fraction was mixed with the others for assay. The combination of the 40–55% and 55–70% (NH4)2SO4 fractions resulted in DPA formation, indicating that the 40–55% (NH4)2SO4 fraction contained the DHDPA synthase (data not shown). Further fractionation using ion exchange and size exclusion chromatography was performed on the 55–70% (NH4)2SO4 fraction, and the in vitro assay was performed using the 40–55% (NH4)2SO4 fraction as a source of enriched DHDPA synthase. The final, most active DPA synthase fraction was separated using both native and denaturing SDS-PAGE (Fig. 2), revealing the presence of > 25 protein species. Slices were excised from the native gel and subjected to the assay system. The proteins in the gel slice containing DPA synthase activity, which migrated on the native gel with an apparent mass of approximately 75 kDa, were subjected to in-gel tryptic digestion and identified by mass spectrometry. A sample of the native gel slice containing DPA synthase activity was also further separated using SDS-PAGE, revealing the presence of ≤ 5 abundant proteins of various sizes (Fig. 2).

Figure 2.

Fractionation of C. perfringens extract containing DPA synthase activity. Protein fractions were separated using SDS-PAGE and stained with Coomassie blue (except lane 8 that was silver-stained). Lanes 1 and 6 contain protein MW markers (sizes indicated in kDa). Lanes 2–5 contain successive fractions eluted from a Superdex 200 column. The fractions in lanes 3 and 4 contained DPA synthase activity. The fraction in lane 4 was used for native gel electrophoresis (not shown) and subsequent detection of activity in the separated proteins. Lane 7 contains proteins eluted from the native gel slice that contained DPA synthase activity. Lane 8 is the exact same gel region as lane 7, but following silver staining. The white arrow in lane 8 indicates a protein band in the region expected for migration of the 36.1 kDa EtfA protein.

Identification of the C. perfringens DPA synthase

MALDI-TOF/TOF peptide sequencing identified five proteins that were present in the native-PAGE gel slices exhibiting DPA synthase activity. These proteins were the products of genes CPR_2403, CPR_0275, CPR_0253, CPR_2284 and CPR_2342 in strain SM101 (NCBI Database NC_008262). Respectively, these gene products are annotated as putative: elongation factor G, metallo-beta lactamase flavodoxin, peptidyl-prolyl isomerase, electron transfer flavoprotein α chain (EtfA) and butyrate kinase.

Published microarray analyses of stationary phase transcription in C. acetobutylicum (Jones et al., 2008) were analysed for homologues of the genes identified above. Of the five genes, only CPR_2284 (etfA) was shown to be upregulated during the sporulation process. It is thought that EtfA, EtfB and butyryl-CoA dehydrogenase (Bcd) form a complex that converts crotonyl-CoA to butyryl-CoA (Boynton et al., 1996). The bcd, etfB and etfA genes are adjacent to one another on the chromosome, and they had previously been predicted to exist within a single operon (Boynton et al., 1996). However, the transcriptome analysis of C. acetobutylicum indicates that EtfA is differentially regulated and is most strongly expressed during late sporulation (Jones et al., 2008), consistent with a role in DPA formation.

C. perfringens EtfA participates in DPA synthesis in vivo

Mutant strain Etf1 was constructed by single crossover insertion of plasmid pSMEtf1 into the etfA locus on the chromosome. The chromosomal insertion was verified by PCR amplification of the region (data not shown). Strain Etf1 demonstrated no change in vegetative growth, measured as change in optical density at 600 nm (OD600), or in early sporulation, as observed by phase contrast microscopy, relative to the wild-type strain SM101 (data not shown). Sporulating cultures of SM101, Etf1 and an etfA-complemented strain, Etf2, were analysed for DPA content during sporulation (Fig. 3). Etf1 produced significantly less DPA than the other strains (P ≤ 0.005, unpaired t-test) at 2 h into stationary phase and later time points, while DPA accumulation in Etf2 was not significantly different from that seen in the wild-type strain SM101. These results are similar to those obtained for the B. subtilis spoVF mutant FB106, which was shown to produce < 5% of the wild-type levels of DPA (Paidhungat et al., 2000).

Figure 3.

Production of DPA during sporulation of C. perfringens strains. SM101 (wild type, □), Etf1 (etfA, ◆) and Etf2 (etfA, pSMnk2/etfA+, ○) were inoculated into DSSM media. Culture samples were assayed for OD600 and for DPA concentration by MRM. Points are means and error bars are standard deviations for samples from 3 to 6 independent cultures.

The effect of the mutation in etfA on heat-resistant spore production was dramatic. Mature spores in 72 h cultures of SM101 easily survived the 70°C heat shock necessary to induce efficient germination (Harry et al., 2009) and then exhibited a heat-resistant titer of > 3 × 106 cfu ml−1. For Etf1, no colonies were obtained following the 70°C heat shock alone, indicating that either no spores were produced or that the spores were heat-sensitive at the 72 h time point. The complemented strain, Etf2, produced > 1 × 105 heat-resistant cfu ml−1. The spores produced by SM101 and Etf2 exhibited similar D-values at 90°C of 22 ± 2 and 19 ± 3 min respectively. The D-value is the amount of time necessary to reduce the viable cell count by 1 log value at a given temperature.

Light microscopy revealed that the numbers of sporulating cells appeared similar during early sporulation of Etf1 and SM101; however, Etf1 was defective in late stage sporulation (Fig. 4). Phase bright spores within mother cells could be observed to develop in both strains. While it is difficult to quantify phase brightness, our general observation was that the Etf1 spores were not as phase bright as those of the wild type. By 8 h post inoculation, phase bright SM101 spores were being released by mother cell lysis, but no phase bright spores were visible in the Etf1 culture. The same was true of the Etf1 culture at 24 h post inoculation, although the number of cells had dropped by > 90%.

Figure 4.

Phase contrast images of SM101 and Etf1 during sporulation. SM101 (column A) and Etf1 (column B) were inoculated into DSSM media and examined by phase contrast microscopy at 6, 8 and 24 h post inoculation. The scale for all images matches the scale bar in A1. Arrows in B3 indicate phase dark spores within partially lysed mother cells in the Etf1 strain.

The B. subtilis spoVF mutant strains can take up exogenous DPA and use it to form stable spores (Paidhungat et al., 2000). However, supplementing Etf1 sporulating culture with 100 μg ml−1 exogenous DPA did not restore spore stability or heat resistance. SM101 cultivated in the presence or absence of DPA produced > 3 × 106 heat-resistant spores per ml, while Etf1 grown under both conditions produced < 3 × 103 heat-resistant spores per ml. The reason for this is unclear, but a likely explanation is failure to transport DPA into the mother cell.

These observations suggest that Etf1 attempts to produce a stable spore, but the lack of DPA production disrupts the maturation of the spore. When the mother cell lyses, it appears that the incomplete spore is incapable of surviving in the surrounding medium. We were able to observe this phenomenon using time-lapse microscopy, during which we speeded the lysis of the mother cell by the addition of lysozyme. It was extremely challenging to observe mother cell lysis, which appeared to us as a rapid change in the phase density of the cytoplasm, as among a population it took place either very rapidly or quite asynchronously over a long period. However, in several Etf1 cells, we observed that a change in mother cell phase density was followed immediately by loss of phase brightness of the included spore (Movies S1 and 2), while in the wild-type strain the spores retained phase brightness (Movies S3 and 4).

DPA production by purified DapA and EtfA

To directly demonstrate the roles of DapA and EtfA in our in vitro system, we overexpressed and purified C. botulinum DapA (Dobson et al., 2008) and C. perfringens EtfA. The in vitro assay demonstrated some spontaneous DPA formation in the presence of high concentrations of DapA, as quantified by MRM mass spectrometry (Table 2). The same amount of DapA with the addition of EtfA resulted in a 3.4-fold increase in DPA production. EtfA alone, in the absence of DapA, produced no DPA. In order to determine whether spontaneous DPA formation in the absence of EtfA was due to oxidation of the DapA product by dissolved O2, the in vitro assay was performed simultaneously under aerobic and anoxic conditions. Under anoxic conditions, spontaneous formation of DPA was reduced by > 99%, and addition of EtfA produced a > 1000-fold increase in DPA synthesis (Table 2).

Table 2.  Assay of EtfA activity in DPA synthesis.
Components present in activity assayDPA produceda (μg DPA/mg EtfA min−1)
  • a. 

    Values are the means and standard deviations of two or three repetitions.

  • b. 

    These values were calculated as if the reactions contained the same amount of EtfA as that present in the complete reactions.

DapA aerobic10.4 ± 0.6b
EtfA aerobicNot detectable
DapA anaerobic0.01b
DapA + EtfA aerobic36.5 ± 13.9
DapA + EtfA anaerobic12.4 ± 3.6
DapA, EtfA separated by dialysis membrane, aerobic30.6 ± 3.7
DapA + FAD-bound EtfA, aerobic103 ± 35.3

The in vitro assay was modified to determine whether a direct interaction of DapA and EtfA was necessary for DPA formation by separating the two proteins with a dialysis membrane within the reaction cell. Following incubation, the reaction resulted in 84% of the level of DPA production in the identical reaction lacking the membrane (Table 2). EtfA appears able to act alone on the product of DapA; the reaction does not require EtfA to be present at the site of DapA catalysis.

In order to determine if cofactors are involved in the production of DPA, the in vitro assay was performed with DapA and EtfA with the addition of NAD, NADP, FAD or FMN. Neither NAD nor NADP produced any increase in synthesis of DPA (data not shown). Addition of FMN resulted in a threefold increase in DPA production, and FAD produced a fivefold increase. EtfA contains a putative FAD-binding domain (Tsai and Saier, 1995). In the absence of added FAD, the purified EtfA protein had only minimal absorbance at the wavelengths characteristic of flavin (Sato et al., 2003) (Fig. S1). Following incubation with FAD and repeated washing steps, the protein demonstrated significantly greater absorbance peaks at 266, 377 and 450 nm, characteristic of FAD (Fig. S1). Based on the predicted extinction coefficient of EtfA and the increase in flavin-specific absorbance, we estimate that between 0.3 and 0.5 mol of FAD were bound per mole of EtfA. The FAD-bound EtfA catalysed a nearly threefold increase in DPA production compared with EtfA washed in the same manner without the addition of FAD (Table 2). The production of DPA by EtfA is a catalytic process as each molecule of EtfA was responsible for the production of 18.5 molecules of DPA min−1.


Dipicolinic acid is essential for the formation of stable endospores. The C. perfringens EtfA protein catalysed the conversion of DHDPA to DPA in vitro, and an etfA mutant strain was deficient in DPA production during spore formation. This appears to represent an alternate method for DPA production among the Class I Clostridia, which lack the SpoVF proteins that catalyse this reaction in other spore-forming species. C. perfringens cells lacking EtfA produce endospores that initially achieve a phase bright appearance, suggesting that other mechanisms allow the endospore to achieve a certain level of dehydration while protected by the mother cell. Following mother cell lysis, the defective spores are unstable. B. subtilis spoVF mutant strains can take up exogenous DPA and use it to form stable spores (Paidhungat et al., 2000). The reason that a C. perfringens etfA mutant is unable to produce stable spores when provided with DPA is unclear, but a likely explanation is failure to transport DPA into the mother cell.

Recently, a mutant strain of C. perfringens was reported in which the spoVA locus was inactivated (Paredes-Sabja et al., 2008). This strain is incapable of transporting DPA that is synthesized in the mother cell into the forespore. Unlike Etf1, the spoVA mutant could successfully complete the sporulation process and produce relatively stable spores, but the spores produced were still highly sensitive to heat. The fact that spoVA DPA-less spores are stable may result from their loss of a proposed function of SpoVA in ion release during spore germination (Vepachedu and Setlow, 2004; 2007). The presence of such an ion channel in the Etf1 DPA-less spores may contribute to their instability in the absence of the osmotic protection provided by the cytoplasm of the mother cell.

Analysis of the genomes of the sequenced clostridia revealed multiple homologues of etfA in each organism. C. perfringens has two, CPR_2284 (etfA) and CPR_0304 in strain SM101, which encode products of 335 and 387 amino acids respectively. The primary difference is that CPR_0304 contains a putative ferredoxin domain at its N-terminus that is not present in CPR_2284. The sequenced clostridia all have at least one homologue of both the shorter and the longer etfA genes, while many species appear to have multiple copies of one or both. It is unclear what the function of these homologues may be, although to date etfA in various genera of bacteria has been implicated in calcium mineralization (Barabesi et al., 2007), caffeate respiration (Dilling et al., 2007), butyrate production (Steen et al., 2008), carnitine reduction (Walt and Kahn, 2002) and nitrogen fixation (Earl et al., 1987). In the majority of reactions utilizing these proteins, EtfA is thought to form a heterodimer with the smaller EtfB molecule (Sato et al., 2003). Our results appear to be the first account of EtfA functioning without an EtfB counterpart. We cannot rule out the possibility that EtfA is complexed with another protein in vivo. EtfA was identified as a protein with mobility equivalent to 75 kDa on a native gel. The 36 kDa EtfA may therefore have been present as a homodimer or in a complex with another protein such as DHDPA synthase (32 kDa) or another electron acceptor.

Our results demonstrate that EtfA has the ability to catalyse the formation of DPA from the product of DHDPA synthase. This activity occurs without direct contact of the proteins and likely utilizes FAD as a cofactor. C. perfringens EtfA was expressed in E. coli cells grown aerobically. It is likely that these cells did not contain enough FAD to saturate the binding domain of the overexpressed EtfA; therefore, the addition of FAD to the purified protein increased its catalytic activity. The ability of EtfA to convert DHDPA to DPA in the absence of added FAD may be due to the fraction of EtfA molecules that were purified with bound FAD. The conversion of DHDPA to DPA requires the removal of two hydrogen atoms. We propose that these are passed to the FAD bound to the EtfA molecule, resulting in reduction to FADH2. The oxidation of FADH2 to FADH is known to occur spontaneously in the presence of O2 (Iyanagi et al., 1974). In our aerobic activity assay, O2 could easily function as the terminal electron acceptor, as the released electrons would result in H2O production. This would explain the 66% reduction in DPA production we observed when the reaction was performed under O2-limiting conditions. We note that some residual O2 was likely present in our assays performed in an anaerobic chamber. Our enzyme and substrate solutions had only minimal pre-incubation under anaerobic conditions and thus contained some dissolved O2. In addition, some H2O2 produced during synthesis of our ASA substrate could decompose to produce O2. Residual O2 likely allowed FAD oxidation and reaction turnover in our purified system. In the anaerobic conditions under which C. perfringens grows, FADH2 would have to be recycled to FAD by reduction of a protein or cofactor with a higher redox potential, such as rubredoxin, which is known to be present at high concentrations in the cytoplasm (Jean et al., 2004).

The fact that many sporulation-specific genes are highly conserved across all spore formers, including all clusters of the clostridia as well as the bacilli, indicates that endospore formation was developed prior to the divergence of these groups. A survey of 52 species of Firmicutes by PCR amplification of sporulation genes (Onyenwoke et al., 2004), and our searches of available genome sequences, indicate that spoVF is present in the genomes of the majority of spore formers outside of the Cluster I clostridia. The only other spore-forming species that lacked these genes are members of the genus Thermoanaerobacter. Interestingly, this genus has been identified as the most deeply rooted member on the branch of the tree of life leading to all the Eubacteria (Ciccarelli et al., 2006). Using the principle of parsimony, this suggests that the earliest spore formers did not possess spoVF, and that a DPA synthesis pathway utilizing EtfA, which is present in all Firmicutes, may be the ancestral pathway.

Experimental procedures

Growth, sporulation, an phenotypic assays

Vegetative growth rates were determined by monitoring OD of anaerobic cultures at 37°C in Duncan Strong sporulation medium (Duncan and Strong, 1968; Sacks and Thompson, 1978). To induce sporulation, C. perfringens was cultivated overnight in fluid thioglycollate medium prior to inoculation into Duncan Strong sporulation medium. Cultures were sampled for the determination of heat-resistant spore formation at 72 h post inoculation. Samples were heated at 70°C for 10 min to kill vegetative cells, serially diluted in H2O, plated on PGY medium and incubated anaerobically at 37°C for 24 h to determine heat-resistant cfu. To quantify relative heat resistance, samples were divided into aliquots and heated at 90°C in a mineral oil bath. At 5 min intervals, aliquots were serially diluted in H2O, and plated for cfu determination. The D-value, or the amount of time necessary to reduce viable counts by 1 log value, for each time point was determined using the formula D = U/(log b − log a), where U is the total time of heating, b the initial cfu count and a the number of surviving spores at that time point (Weiss and Strong, 1967). Each sample was followed for a time equal to a 3 log drop in the initial cfu.

Synthesis of L-ASA

The L-ASA was synthesized by the method of Black and Wright (1955). Samples were removed throughout the ozonolysis procedure, neutralized with 1 M NaHCO3 and subjected to mass analysis with an ABI-3200 mass spectrometer. The reaction was terminated when the peak corresponding to the deprotonated mass (M-H+) of L-allylglycine, 114.1, was completely replaced by the appearance of the peak corresponding to that of L-ASA, 116.1. Synthesis of L-ASA was confirmed by the production of DPA by a crude cell extract of sporulating B. subtilis wild-type strain 168. The L-ASA solution was aliquoted and stored at −80°C.

In vitro assay for DPA synthase activity

The system for determining DPA synthase activity was modified from that of Bach and Gilvarg (1966). The total reaction volume of 2 ml contained ∼3 mg of total protein and 10 mM sodium pyruvate in 20 mM Tris HCl, pH 8.0, and was kept on ice. A freshly thawed tube of L-ASA solution was neutralized with cold 1 M NaHCO3 and quickly added to the reaction mixture to an approximate concentration of 10 mM L-ASA. To establish a baseline measurement, 700 μl of the reaction mixture was immediately removed and mixed with 58 μl of 16 M H2SO4, vortexed thoroughly and centrifuged at 13 000 g for 3 min. The remaining reaction mixture was placed in a 37°C water bath for 20 min. Following this incubation, 700 μl of the mixture was removed, acidified and centrifuged in the same manner. The supernatant of both the ‘pre’ and ‘post’ reaction samples were extracted with 3 ml diethyl ether. Two 1 ml aliquots of the ether layer were removed. One aliquot was immediately placed into a 1 ml quartz cuvette and the absorbance at 269 nm was determined for a rapid estimate of DPA production. The second aliquot was dried at room temperature overnight in a fume hood. The dried sample was then resuspended in 150 μl 0.01% formic acid, and DPA was quantified by mass spectrometry using MRM. In order to assay for the activity of the DPA synthase in C. perfringens NCTC 8679 protein fractions, the reaction mix contained 2 mg B. subtilis FB106 (spoVF::tet) (Paidhungat et al., 2001) sporulating crude cell extract protein as a source of DHDPA synthase or crudely enriched DHDPA synthase [the 40–55% (NH4)2SO4 fraction of a cell extract] from NCTC 8679.

The assay was also performed under anoxic conditions in an anaerobic chamber (Coy Laboratories). Oxygen was allowed to diffuse out of freshly autoclaved water and Tris buffer in the chamber for 2 days. All reactants were mixed in the chamber with the exception of L-ASA, which was neutralized outside of the chamber, put on ice with the thawed protein samples, and placed into the chamber. The protein samples and L-ASA were added to the reaction mixture in the chamber and the reaction was performed as described above. The ‘pre’ and ‘post’ samples were removed and acidified in the chamber. The samples were removed from the chamber for extraction of DPA.

In order to determine if a direct interaction between DapA and EtfA was necessary for DPA formation in vitro, a 1.5 ml reaction mix containing DapA, L-ASA and pyruvate was placed into a dialysis chamber. Snake Skin dialysis tubing (Thermo-Scientific) with a 3.5 kDa molecular weight cut-off separated this reaction from a 1.5 ml chamber containing only EtfA. Both chambers contained 20 mM Tris buffer at pH 8.0. To facilitate the diffusion of molecules across the membrane, the reaction time was lengthened to 45 min.

Determination of DPA concentration

All values for DPA reported in this paper were determined by LC-MS/MS using an ABI 3200 QTrap system in MRM mode. Only ions with an M + H+ of 168.1 ± 0.2 that produced daughter ions with an M + H+ of 127.1 ± 0.2 were used for quantization. A standard curve was developed using pure DPA (Sigma) extracted in ether as described above.

Enrichment of the DPA synthase from NCTC 8679

Cultures of sporulating cells grown in 500 ml of DSSM were harvested 11.5 h post inoculation, centrifuged for 15 min at 4°C and 6000 g, washed twice with cold deionized water, and resuspended in 50 ml cold 20 mM Tris, pH 8.0. The cells were then ruptured by sonication on ice using a Fisher Scientific sonicator at 15% maximum power. Total sonication time was 20 min with 5 s on and 15 s off cycle times. The suspension was centrifuged at 60 000 g for 30 min, and the supernatant was collected to obtain the crude cell extract. This extract was fractionated by successive precipitation with (NH4)2SO4 at concentrations of: 20%, 40%, 55%, 70% and 80% saturation at 4°C. The fractions were desalted by dialysis and assayed for DPA synthase activity with DHDPA synthase provided by the FB106 crude cell extract.

The 55–70% (NH4)2SO4 fraction was separated by cation exchange chromatography with a 1 ml HiTrap QXL column (GE Healthcare). Five fractions were assayed by the in vitro assay with the 40–55% (NH4)2SO4 fraction serving as the source of DHDPA synthase. The single fraction with DPA synthase activity was desalted with a HiTrap Desalting column and separated by size exclusion chromatography on a 1.6 cm × 60 cm Superdex 200 column (GE Healthcare).

Native-PAGE fractionation of DPA synthase activity

The fraction with DPA synthase activity collected from size exclusion chromatography was concentrated 10-fold using a Centricon filtration device (Millipore) with a 10 kDa cut-off. The fraction was then separated by native-PAGE with an 8% acrylamide gel at 4°C. After electrophoresis, the gel was briefly washed in cold 20 mM Tris HCl, pH 8.0 to remove the glycine from the PAGE buffer. The gel was then cut down the centre with a razor blade. One half was rapidly stained with Coomassie brilliant blue. The other half of the gel was placed on a glass plate over a grid and 2 mm sections of the sample lane were excised. The stained gel as well as a pre-stained MW marker served as guides for the excision process. One mm sections were cut off of each end of each gel slice and immediately stored at −80°C. The remainder of each gel slice was placed into the in vitro activity assay. The reaction was allowed to proceed for 30 min to facilitate the diffusion of the reactants into, and the products out of, the gel slice. Two adjacent gel slices that were found to have activity were removed from the reaction mixture, washed twice with 20 mM Tris, pH 8.0 and were stored at −80°C. One half of each slice was removed for analysis by SDS-PAGE, the other half, in addition to the gel fragments removed prior to the assay, were prepared for peptide identification. In-gel trypsin digestion was performed (Jeno et al., 1995), and the peptides were analysed using an ABI-4800 MALDI-TOF/TOF. The peptide fragmentation data were analysed with MASCOT (Matrix Science) using the NCBI nr database.

Construction of mutant strains

The construction of mutant and complemented strains in C. perfringens has been previously described (O'Brien and Melville, 2004; Orsburn et al., 2009). Primer sequences are listed in Table S1. Briefly, primers Mnk5 and Mnk6 were used to PCR-amplify a 482 bp internal region of etfA. This product was inserted into the suicide vector pSM300 (O'Brien and Melville, 2004) to form pSMnk1. Strain SM101 was transformed by electroporation with pSMnk1, and strain Etf1 was selected by growth on Brain Heart Infusion (Difco) with 30 μg ml−1 erythromycin. Insertion of pSMnk1 into etfA was verified by PCR using primers Mnk9 and OSM214. To complement etfA, primers Mnk9 and Mnk8 were used to PCR-amplify etfA plus the upstream 150 bp. This product was inserted into pJIR750 (Bannam and Rood, 1993) to form pJIRMnk2, which was electroporated into Etf1. A transformant, Etf2, was identified by growth on Brain Heart Infusion containing 30 μg ml−1 erythromycin and 20 μg ml−1 chloramphenicol.

Purification of DapA and EtfA

The overexpression of C. botulinum DapA was achieved by use of pETSA1 (Dobson et al., 2008), which was a gift from Renwick Dobson, and purification of this protein was performed as previously described (Dobson et al., 2008; Atkinson et al., 2009). The SM101 etfA gene was PCR-amplified with primers Mnk10 and Mnk12 and was ligated between the BamHI and NotI sites of pET21a (Novagen) in frame with an N-terminal T7 tag and a C-terminal 6-histidine tag. The resulting plasmid, pETFa2 was transformed into BL21-CodonPlus-RIL E. coli (Stratagene). The cells were grown at 37°C to an OD600 of 0.8 and etfA expression induced with 1 mM IPTG for 3 h. Cells were pelleted at 4000 g for 20 min, resuspended in lysis buffer containing 30 mM imidazole and 20 mM Tris-HCl, pH 8.0 and lysed by sonication on ice. The lysate was centrifuged at 50 000 g for 30 min, and the supernatant was retained as the crude cell extract. The extract was passed over a 5 ml His Trap FF Column (GE Healthcare) and washed with 20 ml lysis buffer. The His-tagged EtfA was obtained by elution with 300 mM imidazole, 20 mM Tris, pH 8.0. The imidazole was removed by dialysis against 20 mM Tris, with three changes of buffer over 24 h. Approximately 3.9 mg EtfA-His6 was purified from 125 ml culture. The purity of EtfA-His6 and DapA were confirmed by SDS-PAGE (Fig. S2).


This project was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, Grant Number 2004-04069, and by a Virginia Tech 2010 Ph.D. Fellowship. We thank Matt Perugini and Renwick Dobson for the DapA overexpression vector and Keith Ray of the Virginia Tech Mass Spectrometry incubator for assistance in protein identification.