Structural Studies of the Gas-gangrene Toxin

 

School of Crystallography, Birkbeck College, University of London, UK

 

Introduction

 

The C. perfringens a-toxin is the main pathogenic agent of gas-gangrene in humans It is a 42.5KDa

protein, composed of 370-residues and is a zinc metallophospholipase C consisting of two domains. The crystal structure of the non-toxic B. cereus PC-PLC, which does not possess the C-terminal amino acids is known and residues involved in the co-ordination of catalytically essential zinc ions in PC-PLC are conserved in a-toxin. Both enzymes are able to hydrolyze monodispersed phospholipids and the toxicity of C. perfringens a-toxin is thought to be related to the ability of this enzyme also to interact with phospholipids in eukaryotic cell membranes. Membrane phospholipid hydrolysis results in the perturbation of cell metabolism leading to activation of the arachidonic acid cascade and protein kinase C. Increasing evidence indicates that other bacterial phospholipases C which play important roles in disease, also affect host cell metabolism in this way.

 

Strains of C. perfringens

 

Cloning, expression and purification of C. perfringens carried out during this contract has led to the solution of the crystal structures of a-toxin from three strains of C. perfringens by X-ray crystallography. These strains and their sources are CER89L43 (gut of dead calf), NCTC8237 (human gangrenous wound) and SWCP (gut of dead swan). The toxins from the two mammalian sources, although antigenically different, have protein sequences that differ by less than two per cent. However, the swan toxin sequence has only 84% identity, differing by about 60 residues from the other two proteins.

 

X-ray data collection and structure solution

 

X-ray data collection used both our in-house facilities and the CLRC Daresbury laboratory. The phase problem was not easy to solve but it finally yielded to a combination of heavy-atom derivative work and cross-crystal averaging. Some data for the crystal structures is summarised in the Table below.

 

Structure

Strain

Divalent ions in buffer

Resolution

()

Space

Group

Cell Dimensions

()

R, Rfree

(%)

Conform-ation

1

NCTC8237

None

2.9

R32

153.0, 190.2

22.0, 28.0

Closed

2

CER89L43

Cd2+

1.90

C2221

61.3, 177.3, 79.1

20.8, 25.5

Open

3

CER89L43

None

1.95

R32

151.4, 195.5

18.8, 23.3

Closed

4

NCTC8237

Cd2+

2.2

C2221

60.5, 175.7, 80.2

20.6, 27.2

Open

5

NCTC8237

None

2.4

R32

149.6, 193.8

22.0,24.7

Closed

6

NCTC8237

Ca2+

2.5

R32

149.9, 192.9

18.3,25.7

Closed

7

SWCP

None

2.4

P4122

117.2, 220.6

20.6, 24.2

Open


 

Overall fold

 

The molecule has a face characterised by hydrophobic residues on the surface. This face spans both domains and we have proposed that the residues on this face interact with cell membranes as shown in Figure 1.

 

The N-terminal domain is a-helical and is structural similarity to Bacillus cereus PLC structure and also to the structure of the P1 nuclease from Penicillium citrinum. The C-terminal domain. This domain contains the active site, binds the zinc ions and consists of 245 residues organised into nine a-helices. Th C-terminal domain consists of a sandwich of two four-stranded b-sheets and contains calcium binding sites (fig 2a and ref 5).

 

The closed structures (see Table above) are significantly different from open structures. In the open structures the active site is available for substrate binding whereas in the closed structures the active site is closed off by the movement of two loops formed by residues 63-90, 132-149. The open and closed forms of the a-toxin structures from mammalian sources are shown in Figure 2. Figure 3 shows the superposition of the SWCP toxin and the open form of CER89L43 toxin.

 

Active site

 

Zinc ions had previously been identified as essential for catalytic activity and at the end of refinement the presence of metal ions in the active site was indicated by difference Fourier density. Structures 2, 4 and 7 all have three cations at the active site. Whilst in 1 and 3 the co-ordination geometry indicates one Zn2+ and two Cd2+ ions to be present, in the cadmium-free crystallisation conditions of structure 7 all three cations are Zn2+ (fig 3b).

 

In the closed form of the structure however, only two Zn2+ ions are present in the active site and one histidine ligand that would co-ordinate the third zinc has moved over 8 from the active site. Figure 2b shows the superposition of open and closed structures viewed down the proposed membrane binding surface.

 

C-terminal domain

 

The C-terminal domain, which has no sequence homology with any proteins of known structure, was shown to have a strong structural analogy with eukaryotic C2 domains. This domain is a Ca2+-dependent phospholipid binding domain found in eukaryotic second messenger proteins, but not previously recognised in prokaryotes. Its identification in a prokaryote which indirectly causes changes in the activation state of eukaryotic proteins possessing C2 domains is interesting.

 

Earlier work had shown that removal of this domain from the enzyme leaves a molecule that is catalytically active against monodisperse substrates but has no haemolytic and toxic properties. It was also known that the latter properties also depended on the presence of Ca2+ ions. Our a-toxin structures have revealed highly conserved aspartate residues associated with Cd2+ binding in structures 1 and 3 and in structure 6 we have identified three calcium binding sites that are shown in Figure 2a.


 

Inhibitor work

 

We tried soaking and co-crystallising a number of available substrate analogues. Difference maps usually revealed little interpretable electron density. However a-toxin complexed with the substrate paranitrophenyl-phophatidylcholine did reveal interpretable difference density with the cleaved substrate left in the active site. This complex is consistent with the hypothesis that phospholipid hydrolysis takes place by in-line nucleophilic attack by water on the phosphorus atom and the structure suggests which of the zinc ions withdraws electrons from the phosphate. The exact role of the other zinc ions still requires investigation.

 

Mutant studies

 

We have made ten mutant enzymes based on our structural work. These experiments have been designed to probe the role of residues in calcium binding, toxin-membrane interactions and inter-domain interactions. We are currently characterising the mutants in terms of egg yolk activity, haemolytic activity and their activity towards phosphatidylcholine and sphingomyelin liposomes.

 

Conclusions

 

The experimental work under this contract has enabled us to

        propose a model of toxin-membrane interaction

        establish the existence of a C2-like domain in a prokaryotic protein

        understand the role of calcium at a molecular level. In particular it appears from the crystal structures solved at different pH both with and without divalent ions, that both neutral pH or the presence of divalent cations are necessary for a catalytically active form of the enzyme.

        partially understand the role of zinc in the enzyme mechanism

 

Further work

 

We have initiated microfluorimetry work at Daresbury Laboratory for which the Biotechnology and Biological Sciences Research Council (BBSRC) has awarded a committee studentship. This work is designed to examine the role of the a-toxin in disrupting secondary messenger pathways in cell signalling.

 

To make further progress on the inhibitor work requires careful chemical synthesis and testing. This work began in February 2001 with the aid of project grant funding from BBSRC and MOD for these studies.

 

Acknowledgement

 

The structural aspects of the work carried out under this contract were also supported by a project grant from the BBSRC and financial assistance from the MOD.

 

David S Moss

Professor of Biomolecular Structure

22 May 2001
X-ray Structural Studies of Bacterial
a-Toxin and their implication in gas-gangrene

 

 

Publications

 

 

1.      Structure function studies of Clostridium perfringens alpha toxin; a gas-gangrene causing protein. N. Justin, J. Eaton, C. Naylor, A. Howells, D. Moss, R. W. Titball. & A. K. Basak. Proceedings of the IVth. International symposium of protein structure and function relationships, 1997.

 

2.      Crystallisation and preliminary X-ray diffraction studies of alpha-toxin from two different strains (NCTC-8237 & CER89L43) of Clostridium perfringens. A. K. Basak, A. Howells, J. T. Eaton, D. S. Moss, C. E. Naylor, J. Miller. & R. W. Titball. Acta Crystallographica, (1998), Sect-D, Sect-D, 54, 1425-1428.

 

3.      Structure of the key toxin in gas-gangrene. C. E. Naylor, J. T. Eaton, A. Howells, N. Justin, D. S. Moss, R. W. Titball. & A. K. Basak. Nature Structural Biology, (1998), Vol.5, No.8. Pp 738-746 (Front cover page and News & Views article).

 

4.      The Clostridium perfringens a-toxin. (Review Article) Richard W. Titball, Claire E. Naylor and Ajit K. Basak. Anaerobe, 1999, 5, 51-64.

 

5.      Characterization of the calcium binding C-terminal domain of Clostridium perfringens alpha-toxin C. E. Naylor, M. Jepson, D. T. Crane, R. W. Titball, J. Miller, A. K. Basak and B. Bolgiano. Journal of Molecular Biology, 1999, 294, (3), 757-770.

 

 

6.      Opening of the active site of Clostridium perfringens alpha-toxin may be triggered by membrane binding, Titball R W, Naylor C E, Miller J, Moss D S, and Basak A K, Int. J. Med. Microbiol., (2000), 290, 357-361.

 

7.      Mapping critical residues for toxicity in Clostridium perfringens phospholipase C, the key toxin in Gas gangrene: Walker N, Holley J, Naylor C, Flores-Diaz M, Alape-Giron A, Thelestam M, Moss D, Basak , Miller J and Titball, Arch. Biochem. Biophys., (2001), 384, 24-30.

 

8.      Tyrosine 331 and phenylanaline 334 in Clostridium perfringens a-toxin are essential for cytotoxic activity, Jepson M, Bullifent, H L, Crane D, Flores-Diaz, M, Alape-Giron, A, Jayasekeera, P, Lingard, B, Moss, D and Titball R W, FEBS Lett., (2001), 495(3), 172-7.

 

 

 

Toxin Group Home