Differences in the Carboxy-Terminal (Putative Phospholipid Binding) Domains of Clostridium perfringens and Clostridium bifermentans Phospholipases C Influence the Hemolytic and Lethal Properties of These Enzymes

Abstract

The phospholipases C of C. perfringens (alpha-toxin) and C. bifermentans (Cbp) show >50% amino acid homology but differ in their hemolytic and toxic properties. We report here the purification and characterisation of alpha-toxin and Cbp. The phospholipase C activity of alpha-toxin and Cbp was similar when tested with phosphatidylcholine in egg yolk or in liposomes. However, the hemolytic activity of alpha-toxin was more than 100-fold that of Cbp. To investigate whether differences in the carboxy-terminal domains of these proteins were responsible for differences in the hemolytic and toxic properties, a hybrid protein (N_biC_α) was constructed comprising the N domain of Cbp and the C domain of alpha-toxin. The hemolytic activity of N_biC_α was 10-fold that of Cbp, and the hybrid enzyme was toxic. These results confirm that the C-terminal domain of these proteins confers different properties on the enzymatically active N-terminal domain of these proteins.

In 1941 MacFarlane and Knight (9) reported a breakthrough in the understanding of the biochemical basis of bacterial disease by showing that the alpha-toxin of Clostridium perfringens was a phospholipase C enzyme. This finding prompted a number of studies to characterize the phospholipases C from other species of clostridia in the anticipation that they, too, would possess toxic properties. However, this possibility was soon disproved by Miles and Miles (11), who showed that the Clostridium bifermentans phospholipase C (Cbp) was, by their criteria, nontoxic. Attempts to explain this difference focused on the lower phospholipase C and hemolytic activities of the C. bifermentans enzyme, and MacFarlane (10) proposed that the toxicity of the C. perfringens and C. bifermentans enzymes correlated with hemolytic activity rather than with phospholipase C activity. The molecular basis for these differences have been speculated upon but not examined in detail until more recently. The alpha-toxin is now known to be composed of two domains (16). The isolated N-terminal domain possessed the phosphatidylcholine phospholipase C activity of the holotoxin, but was devoid of hemolytic and lethal activities, and showed reduced sphingomyelinase activity (24). The isolated C-terminal domain was devoid of any detectable biological activity but appeared to confer hemolytic, lethal, and sphingomyelinase activities on the toxin (22, 24). Studies with site-directed mutants of the toxin have confirmed that N-domain mutants with abolished phosphatidylcholine phospholipase C activity also show loss of hemolytic and lethal activities (5, 13), confirming that the phospholipase C activity of the toxin is necessary for all of the biological activities.

The crystal structure of alpha-toxin (16) reveals that the C domain is organized into a β-sandwich, termed a C2 domain. The C domain of alpha-toxin has a topology similar to that of the calcium-dependent phospholipid binding domains of some eukaryotic proteins, namely, pancreatic lipase, synaptotagmin I, human arachidonate 5-lipoxygenase (HA5L), and phosphoinositide-specific phospholipase Cδ1 (15). These eukaryotic proteins share little overall sequence homology with alpha-toxin, but aspartic acid residues involved in calcium binding in the C2 domain of HA5L are conserved in the alpha-toxin C domain (16). Using site-directed mutagenesis, Guillouard et al. (6) showed that Asp₂₆₉ and Asp₃₃₆ (Fig. 1) were essential for alpha-toxin activity and suggested that these residues are involved in calcium binding. Naylor et al. (16) have designated the area coordinated by these key aspartic acid residues and other residues (Gly₂₇₁ and Ala₃₃₇) as a calcium-binding site in the C domain of alpha-toxin.

FIG. 1.

Alignment of the deduced amino acid sequences of C. perfringens alpha-toxin (Cpa) and C. bifermentans phospholipase C (Cbp). Symbols: ▴, region (Ser₂₅₀ and Val₂₅₁) at which a unique AatII restriction site (Asp-Val) was introduced; ∗, aspartic acid residues thought to be involved in calcium binding in the C domain of alpha-toxin.

A comparison of the deduced amino acid sequences of the C. perfringens and C. bifermentans proteins revealed 51% overall homology (26), and the amino acid sequence alignment suggests that Cbp is also a two-domain protein (23, 26). This alignment also reveals differences in key residues involved in calcium and phospholipid binding within the C domains. These differences in the C domain of Cbp could result in diminished calcium binding and therefore reduce the ability of the protein to interact with membranes, resulting in its lowered hemolytic activity. We have set out to compare the properties of the C. perfringens and C. bifermentans phospholipases C and to construct a hybrid form of the C. perfringens and C. bifermentans proteins, with a view towards determining whether the N or C domains of these enzymes are responsible for the differences in activity of these enzymes.

MATERIALS AND METHODS

Chemicals and enzymes.

All chemicals were obtained from Sigma (Poole, United Kingdom) or BDH (Poole, United Kingdom) unless otherwise stated. Materials for protein purification were obtained from Pharmacia Biotech. Materials for DNA purification were obtained from Boehringer Mannheim (Lewes, United Kingdom).

Isolation of DNA.

Escherichia coli containing plasmids pKSα3, pCBPLC3, or pαPROM27 were grown in L broth containing ampicillin at 37°C overnight (20). Plasmid DNA was isolated from the cell pellet by the Qiagen procedure.

Expression of the C. bifermentans phospholipase C gene (cbp).

The C. bifermentans phospholipase C gene was amplified by PCR (20) with plasmid pCBPLC3 (encoding cbp) as the template DNA and oligonucleotides NbifUP (5′-GCTTGGGGCCCAAGATTAGAGGATATTA-3′) and CbiRev (5′-TGCTCTAGATTATTTATTTATGTAATAAGT-3′) for PCR amplification of the DNA fragment. The DNA fragment obtained after 35 cycles of amplification (95°C, 15 s; 50°C, 15 s; 72°C, 30 s; Perkin-Elmer 9600 GeneAmp PCR system) was purified, digested with the appropriate restriction endonucleases, ligated with plasmid vector pαPROM27 containing the alpha-toxin promoter, and transformed into E. coli JM109 cells by electroporation (20). Colonies containing the cloned cbp DNA fragment were identified by PCR with the oligonucleotides NbifUP and CbiRev.

Nucleotide sequencing.

Extension reactions with dithiothreitol chemistry and containing 500 ng of template DNA and 4.8 pmol of primer were prepared by using ABI 800 Molecular Biology Labstation Catalyst overnight at 50°C. Sequence analysis was carried out with an Applied Biosystems 373A sequencer. The nucleotide sequence of the cpb gene, encoding the C. bifermentans phospholipase C, has been deposited in GenBank (accession number AF072123).

Cloning of the N and C domains.

DNA fragments, which encoded the N domain of alpha-toxin or the C domain of Cbp were amplified by PCR (20) with either oligonucleotides NbifUP and NbiREV (5′-CGGGGTACCTATGACGTCTCCTGTATTTCC-3′) or oligonucleotides CaUP (5′-TGCTCTAGACCAGACGTCGAAAGAATGTA-3′) and CaREV (5′-TGCTCTAGATTATTTTATATTATAAGTTGA-3′). The template DNA for these reactions was either plasmid pKSα3 (encoding alpha-toxin) or plasmid pCBPLC3 (encoding Cbp). The DNA fragments obtained after 35 cycles of amplification (95°C, 15 s; 50°C, 15 s; 72°C, 30 s; Perkin-Elmer 9600 GeneAmp PCR System) were purified, digested with the appropriate restriction endonucleases, ligated with plasmid vector pBluescript KS(+) (Stratagene), and transformed into E. coli JM109 cells by electroporation (20). Colonies containing the cloned DNA fragments were identified by PCR with the oligonucleotides described above.

Purification of phospholipase C.

Phospholipase C was extracted from the periplasmic space of E. coli cells as described previously (25). The phospholipase C was purified by using two successive fast protein liquid chromatography (FPLC) ion-exchange chromatography steps. The crude extract was initially applied to a FPLC anion-exchange chromatography column (3.2 by 25 cm; DEAE Sephacel) and eluted with a 0 to 100% (0 to 1 M) sodium chloride gradient at a flow rate of 0.2 ml/m. Fractions (10 ml) were collected, and those containing phospholipase C were concentrated by stirred-cell ultrafiltration (Amicon) under nitrogen pressure through a blocked membrane. The concentrated sample was buffer exchanged into 10 mM Tris-HCl (pH 8.0) and applied to a second anion-exchange chromatography column (MonoQ HR10/10). Protein was eluted with a 0 to 100% (0 to 1 M) NaCl gradient at a flow rate of 0.2 ml/m. For Cbp, the protein was further purified by using a gel filtration column (FPLC; Superose12) and eluted in phosphate-buffered saline (0.1 M sodium phosphate, 0.9% [wt/vol] NaCl; pH 7.2) at a flow rate of 0.2 ml/m. Fractions containing phospholipase C were collected manually, pooled, and concentrated by stirred-cell ultrafiltration. The concentrated sample was buffer exchanged into 10 mM Tris-HCl (pH 8.0), and the final preparations were stored at −20°C. Protein content was determined by using the bicinchoninic acid protein assay. Purified products were analyzed by sodium dodecyl sulfate–10 to 15% polyacrylamide gel electrophoresis by using a PhastSystem and were judged to be >95% pure (Fig. 2). Gels were stained with Coomassie blue R250. The yields of purified alpha-toxin, Cbp, and hybrid N_biC_α were 10, 5, and 15 mg, respectively.

FIG. 2.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of purified phospholipases C. C. bifermentans Cbp (lane 1), hybrid N_biC_α (lane 2), C. perfringens alpha-toxin (lane 3), and molecular mass weight markers in kilodaltons (lane 4) are shown.

CD spectroscopy studies.

Secondary structures of the purified proteins were determined by using circular dichroism spectroscopy (CD) in a Jasco J-720 spectropolarimeter at 25 ± 1°C. The bandwidth was either 1 or 2 nm with a response time of 1 s, and data were collected at between 178 and 260 nm for far-UV by using 0.01-cm demountable and 0.05-cm cylindrical quartz cells. Data were collected at between 250 and 320nm for near-UV by using 1-cm pathlength quartz cuvettes (Hellma GmbH) at 0.5-nm intervals, with averaging for 9, 16, or 25 scans. Varselec 1 software was applied to the CD spectrum between 178 and 260 nm, with a 33-protein basis set, to calculate secondary-structure composition.

Biological activities of the phospholipases C.

Phospholipase C activity was measured by egg yolk assay emulsion in a microtiter assay (25) or by using a modified version of the protocol by Kurioka et al. (8), with p-nitrophenylphosphorylcholine (pNPPC; 40 mM in water). The absorbance was read at 414 nm (pNPPC) or 540nm (egg yolk) on a Titertek Multiscan MCC/340 Platereader (Life Sciences). Liposomes composed of either phosphatidylcholine (Lipoid) or sphingomyelin encapsulating carboxyfluorescein (CF) were prepared according to the method of Nagahama et al. (14). The liposomes were probe sonicated as an additional step after drying and hydration of the liposomes. Phospholipase C (100 μl), at various concentrations, was incubated with liposomes (100 μl) at 37°C for 1 h. Fluorescence intensity was measured with a luminescence spectrometer LS-5B (Perkin-Elmer, excitation at 485 nm, emission at 520 nm). A 100% CF release was determined by incubation of liposomes with 1% Triton X-100 at 37°C for 1 h. Hemolytic activity was determined by using mouse erythrocytes (5% [vol/vol]) in a microtiter tray assay as described previously (25).

Toxicity in mice.

The toxicity of the phospholipases C was determined by intraperitoneal injection of 50 μl of purified toxin (in saline) into groups of four BALB/c mice. The mice were monitored for 18 h after injection.

Complementation of the C domain.

The isolated C domain of alpha-toxin was incubated with mouse erythrocytes which had been preincubated with phospholipase C, and the hemolytic activity was determined as described above.

RESULTS

Cloning and expression of C. perfringens alpha-toxin and C. bifermentans phospholipase C.

C. perfringens NCTC 8237 alpha-toxin was expressed in E. coli as described previously (25). The plasmid-cloned gene encoding the C. bifermentans phospholipase (26) (cbp) was nucleotide sequenced, and seven differences between the previously reported sequence (26) were found. These differences resulted in six amino acid sequence changes (Tyr₃₁→Ser₃₁, Ser₉₄→Ala₉₄, Arg₉₅→Glu₉₅, Asn₉₆→Thr₉₆, Ser₉₇→Gln₉₇, and Ile₂₅₂→Asp₂₅₂). The level of expression of the cloned C. bifermentans phospholipase C gene (cbp) from the native promoter in E. coli was low. To increase the level of expression, the cbp open reading frame was amplified by PCR and cloned downstream of the C. perfringens alpha-toxin promoter. Nucleotide sequencing of the cloned fragment confirmed the authenticity of the cbp gene.

Construction of a C. perfringens-C. bifermentans hybrid phospholipase C.

Previous studies have shown that the C. perfringens alpha-toxin is composed of two domains (N and C domains) linked by a potentially flexible region (16). An alignment of the deduced amino acid sequences of the alpha-toxin and Cbp (Fig. 1) suggested a similar organization of these proteins, and we considered that by modifying the linker region (Ser₂₅₀-Val₂₅₁ in alpha-toxin or Asp₂₅₀-Asn₂₅₁ in Cbp) to encode Asp-Val in both proteins we would be able to introduce a unique AatII restriction site into the coding DNA. To introduce this site, the regions of DNA encoding the N domain of Cbp or the C domain of alpha-toxin were amplified by PCR with appropriately tailed PCR primers. The changes in the amino acid sequence did not affect the phospholipase C or hemolytic activities of alpha-toxin or Cbp (data not shown). To construct the hybrid phospholipase C, the DNA fragment encoding the N domain of Cbp was fused with the C. perfringens cpa promoter sequence. The DNA fragments were ligated at the AatII site to generate a gene encoding a hybrid C. perfringens-C. bifermentans phospholipase C (hybrid N_biC_α).

CD spectroscopy study of the enzymes.

Purified alpha-toxin, Cpb, or hybrid N_biC_α were analyzed by using circular dichroism spectroscopy. Secondary-structure composition was determined with Varselec 1 software applied to the CD spectrum at between 178 and 260 nm with a 33-protein basis set. The alpha-toxin was shown to consist of 43% α-helix, 12% β-sheet, and 16% turn, which is consistent with the crystal structure data (44% α-helix, 18% β-sheet, and 4% turn [15a]). Cbp contained 40% α-helix, 15% β-sheet, and 15% turn. Hybrid N_biC_α consisted of 43% α-helix, 15% β-sheet, and 15% turn.

Phospholipase C: hemolytic and lethal activity of the alpha-toxin and Cpb.

The specific activities of alpha-toxin or Cbp were determined by using a variety of assays. The activity of the alpha-toxin or Cbp towards sphingomyelin liposomes was increased 127- and 146-fold, respectively (mean of three determinations), when the assays were carried out in the presence of 2.5 mM Ca²⁺. Enhancement of activity towards murine erythrocytes in the presence of 2.5 mM Ca²⁺ was also observed (data not shown). All subsequent assays were carried out in the presence of 2.5 mM Ca²⁺ ions.

The alpha-toxin was more active in all of the assays compared with the other enzymes (Table 1). However, the ratio of alpha-toxin activity to Cbp activity depended on the form of the substrate. When phosphatidylcholine in egg yolk or in liposomes was used, alpha-toxin was twice as active as Cbp. When tested with sphingomyelin liposomes, alpha-toxin was 10 times as active as Cbp. When a synthetic substrate with the phosphorylcholine head group linked to p-nitrophenyl (pNPPC) was used, alpha-toxin was 24 times as active as Cbp. The hemolytic activity of alpha-toxin was 132 times that of Cbp when tested with mouse erythrocytes. When administered intraperitoneally into mice, 10 μg of Cbp did not cause any deaths over a period of 18 h, whereas 1 μg of alpha-toxin resulted in the death of all of the mice within 2.5 h (Table 2).

TABLE 1.

Comparison of the phospholipase C activities toward different substrates and the hemolytic activities of C. perfringens alpha-toxin, C. bifermentans Cbp, and hybrid N_biC_α^a

Enzyme	Sp act (U/mg/60 m)a
	Egg yolk emulsion	pNPPC	Hemolysis	SPH liposomes	PC liposomes
Alpha-toxin	64,445	238,000	256,000	290,000	119,400
Cbp	42,552	10,000	1,933	28,901	57,803
Hybrid NbiCα	35,705	45,600	21,560	93,586	101,022

^aThe results are the mean of five determinations (standard deviations ± 0.1%. SPH, sphingomyelin; PC, phosphatidylcholine. 

TABLE 2.

The toxicity of C. perfringens alpha-toxin, C. bifermentans Cbp, or hybrid N_biC_α when administered intraperitoneally to BALB/c mice and monitored for 18 h

Enzyme	Dose administered (μg)	No. of deaths/total no. of mice
Alpha-toxin	1	6/6
Cbp	1	0/6
	10	0/6
Hybrid NbiCα	1	0/6
	10	6/6

Phospholipase C: hemolytic and lethal activity of hybrid N_biC_α.

The replacement of the C domain of Cbp with the corresponding region from alpha-toxin yielded an enzymatically active protein (N_biC_α) which, on the basis of CD analysis, was correctly folded. The hybrid phospholipase C had properties that were intermediate between the two parent enzymes (Table 1). The specific activity of hybrid N_biC_α was broadly similar to alpha-toxin and Cbp when tested with egg yolk phosphatidylcholine or with liposomes composed of phosphatidylcholine. The activity of hybrid N_biC_α was greater than the activity of Cbp when assayed with liposomes composed of sphingomyelin. The hemolytic activity of hybrid N_biC_α was increased 10-fold compared to Cbp but was lower than the activity of the alpha-toxin. When administered intraperitoneally into mice, hybrid N_biC_α phospholipase was intermediate in toxicity between Cbp and alpha-toxin (Table 2).

Complementation of the C domain.

Guillouard et al. (6) reported that mutation of residue Thr₂₇₂ to Pro₂₇₂ in alpha-toxin results in the abolition of phospholipase C and hemolytic activity. They further reported that hemolytic activity was restored when the isolated C domain of alpha-toxin was added to a solution of this mutated protein. When purified C domain of the alpha-toxin was added to a solution of alpha-toxin (78 hemolytic units [HU]) or Cbp (14 HU), there was no significant increase in the hemolytic activity of either of these enzymes (86 and 13 HU, respectively; all data are the mean of two determinations).

DISCUSSION

Phospholipases produced by pathogenic bacteria have been shown to play a variety of roles in the pathogenesis of disease (12, 21, 23). In the case of alpha-toxin, the enzyme has been shown to activate the arachidonic acid cascade and protein kinase C in eukaryotic cells (4, 7, 18, 19), and the toxic effects (1) might be a result of this perturbation of host cell metabolism. These changes in host cell metabolism are a result of the accumulation of diacylglycerol within the host cell, and the ability of bacterial phospholipases C to hydrolyze membrane phospholipids, yielding diacylglycerol, provides an obvious role for these enzymes in this process. Within cell membranes the phospholipid head groups are packed together, shielding the ester bond which is the target of the alpha-toxin (2). Therefore, it has been suggested that alpha-toxin is able to insert into the cell membrane (16). Hemolysis is often used as an indicator of activity towards membrane phospholipids.

The C. bifermentans Cbp has been shown to be nontoxic and weakly hemolytic when compared to alpha-toxin (26), and previous studies have suggested that these properties can be explained by the correspondingly reduced phospholipase C activity (23, 26). Our findings indicate that the phospholipase C activity of Cbp towards some substrates, such as egg yolk phosphatidylcholine or phosphatidylcholine liposomes, is close to the activity of alpha-toxin. Several studies have suggested that the ability of some bacterial phospholipases to cause hemolysis is due to their ability to hydrolyze both phosphatidylcholine and sphingomyelin (17, 23, 24), and it may be significant that Cbp was 10-fold less active than alpha-toxin toward sphingomyelin in liposomes. However, this difference in substrate preference does not provide a full explanation for the >100-fold difference in the hemolytic activity of Cbp and alpha-toxin. Our finding that purified alpha-toxin was at least 10 times as toxic to mice as Cbp lends support to the suggestion that hemolytic activity provides an in vitro indicator of toxicity.

Crystallographic studies show that alpha-toxin is composed of an active site domain (N domain) and a calcium-binding domain (C domain), which are linked by a flexible peptide (16). Removal of the C domain from alpha-toxin abolished hemolytic activity (but not the phospholipase C activity), and antisera against the C domain neutralized hemolytic activity (27), cytotoxicity towards UDP-glucose-deficient cells (3), and lethality (27). It is possible that this domain plays a role in the recognition of membrane phospholipids. Naylor et al. (16) have constructed a model indicating how surface-exposed hydrophobic side chains (Tyr₃₃₁ and Phe₃₃₄) and calcium ions coordinated by the phospholipid phosphate group and the C domain allow alpha-toxin to become inserted into membranes. Our finding that the activity of alpha-toxin and Cbp were similarly stimulated by Ca²⁺ ions and that, with the exception of Asp₂₆₉ (Tyr in Cbp), Ca²⁺-binding ligands are conserved in both alpha-toxin and Cbp indicates that the differences in cytolytic properties of these proteins are not due to differences in Ca²⁺-mediated recognition of phospholipids. However, Tyr₃₃₁ and Phe₃₃₄ in the C domain of alpha-toxin are replaced in Cbp by residues, which are unlikely to play similar roles (Leu and Ile, respectively).

The data we have reported here provide the first evidence that the function of the C domains of alpha-toxin and Cbp are not identical. Replacement of the Cbp C domain with the C domain from alpha-toxin enhanced the hemolytic activity of the enzyme 10-fold, which would be expected if the C domain plays a key role in membrane phospholipid recognition. However, structural motifs in the N domain are also thought to play a role in the recognition of membrane phospholipids (Trp₂₁₄). This would explain why replacement of the Cbp C domain with the alpha-toxin C domain did not fully restore hemolytic activity. Although Guillouard et al. have shown that the low hemolytic activity of mutated forms of alpha-toxin could be complemented by adding the wild-type C domain of alpha-toxin, we were not able to demonstrate a similar effect by adding the alpha-toxin C domain to Cbp.

The finding that replacement of the Cbp C domain with the C domain from alpha-toxin increased its activity toward pNPPC was unexpected because this substrate lacks a hydrocarbon tail group. In previous studies with the alpha-toxin it was shown that removal of the C domain increased the activity of the enzyme toward pNPPC, a result which was attributed to the enhanced accessibility of the active site (24). Although the overall conformation of Cbp appeared to be similar to that of alpha-toxin, it is possible that the different C domains modify the accessibility of the active site to pNPPC. The determination of the structure of the Cbp might answer this question.

The findings reported here suggest a similar phospholipid-recognizing role for the C domain in other hemolytic zinc-metallophospholipases C, such as the C. novyi gamma-toxin. Whether other bacterial phospholipases such as those produced by Mycobacterium tuberculosis and Pseudomonas aeruginosa possess similar phospholipid-binding domains awaits investigation. However, the overall amino acid sequence homology between the alpha-toxin C domain and the calcium-dependent phospholipid-binding domains of synaptotagmin, phosphatidylinositol phospholipase C, and pancreatic lipase is low (15, 16), and the identification of domains with similar functions in other hemolytic bacterial phospholipases C may not be possible on the basis of sequence homology alone. Nevertheless, it seems likely that these hemolytic enzymes do possess specialized features which allow membrane phospholipid recognition.