OUP user menu

Helicobacter pylori vacuolating cytotoxin binding to a putative cell surface receptor, heparan sulfate, studied by surface plasmon resonance

Meeme Utt, Bengt Danielsson, Torkel Wadström
DOI: http://dx.doi.org/10.1111/j.1574-695X.2001.tb01557.x 109-113 First published online: 1 March 2001


The Helicobacter pylori vacuolating cytotoxin or VacA toxin is a major virulence factor in H. pylori infection and type B gastritis. We predicted heparin/heparan sulfate (H/HS) binding properties of the 58-kDa subunit of VacA cytotoxin using bioinformatics tools and showed this by surface plasmon resonance (SPR)-based biosensor studies. Putative H/HS binding peptides were synthesized and binding to HS was shown by SPR in the absence or presence of trifluoroethanol. We found that a recombinant cytotoxin VacA polypeptide binds to surface-immobilized HS and propose that HS might be a receptor/co-receptor for H. pylori VacA cytotoxin.

  • VacA cytotoxin
  • Heparan sulfate
  • Surface plasmon resonance
  • Bioinformatics
  • Synthetic peptide
  • Helicobacter pylori

1 Introduction

Vacuolating cytotoxin or VacA toxin is a major virulence factor in Helicobacter pylori infection associated with acute and chronic type B gastritis [1]. The precursor of VacA toxin is integrated into the outer membrane of H. pylori and after cleavage of the C-terminal membrane domain by protease, a 103-kDa polypeptide is secreted into the extracellular space [2] to form oligomers of 12 monomers [3]. It is known that native, acid non-activated toxin monomers can form non-covalently bound smaller subunits (37 kDa and 58 kDa) by enzymatic cleavage at neutral pH during prolonged storage at +4°C [3]. Below pH 5, the native polymeric cytotoxin dissociates into monomers, but can re-associate at neutral pH into polymeric structures with a different conformation [4]. This ‘acid-activated’ molecule exhibits protease and acid stability with enhanced cytotoxic effects to tissue culture cells. All H. pylori isolates contain the VacA toxin gene, but it is not always expressed. The toxicity has been associated with mosaicism of the vacA gene, three different types of signal sequences (s1a, s1b, s2) and two types of mid-region (m1, m2) have been described [5]. Isolates with the m1 type were toxic to HeLa cells whereas the m2 type was not. Both types were toxic to primary cultured human gastric cells and to the rabbit kidney epithelial cell line RK-13 [6].

How H. pylori cytotoxin enters the host cell is unknown, but it seems to bind to the epidermal growth factor receptor [7], the 140-kDa membrane protein [8], and the 250-kDa protein tyrosine phosphatase [9].

Heparan sulfate (HS) is a glycosaminoglycan present on the surface of most eukaryotic cells and is involved in multiple biological functions. Several viral and bacterial proteins have been identified to interact with HS [10]. Proteins and peptides able to bind heparin/heparan sulfate (H/HS) should have clusters of the positively charged amino acids lysine and arginine. It has been shown that secondary structure elements, α-helix or β-strand, are present in these positive clusters [11,12].

In this study we demonstrate that modern bioinformatics tools can be used to predict these structures. We also demonstrate the presence of H/HS binding sequences along the VacA 58-kDa subunit and show the binding of predicted H/HS binding peptides and the recombinant VacA polypeptide to the surface-immobilized HS by surface plasmon resonance.

2 Materials and methods

2.1 Materials

HS prepared from bovine trachea was a kind gift from Lars-Åke Fransson and Anders Malmström, Center for Chemistry and Engineering of Lund University (Lund, Sweden). Biotinamidocaproate N-hydroxysuccineimide ester was obtained from Sigma (St. Louis, MO, USA). End point biotin-labelled heparan sulfate (B-HS) was prepared by biotinylation of free amino groups on HS with biotinamidocaproate N-hydroxysuccineimide ester in phosphate-buffered saline (PBS) and purified on a Fast Desalt® column (Amersham Pharmacia Biotech, Uppsala, Sweden). The recombinant rVacA 95-kDa polypeptide was a kind gift of Dr. D. Burroni (Siena, Italy). Synthetic peptides corresponding to VacA precursor HP0887 sequences aa 715–736 (M530) and aa 801–823 (M531) were purchased from Vulpes Ltd. (Tallinn, Estonia).

2.2 Prediction of potential H/HS binding sequences

The molecular mass (MW) and isoelectric point (pI) were calculated for the VacA toxin precursor (HP0887 of the H. pylori genome published at TIGR, http://www.tigr.org) and synthetic peptides using the pI/MW compute tool at the ExPASy server, http://www.expasy.ch/tools/pi_tool.html. To predict the putative H/HS binding sequences, the primary sequence of the 58–kDa subunit of VacA cytotoxin precursor (derived from precursor HP0887 between aa 370 and 853) was analyzed as follows: firstly, the secondary structure of the 58-kDa VacA cytotoxin subunit was predicted using the software package PredictProtein (http://dodo.cpmc.columbia.edu/predictprotein/), secondly, clusters of Lys and Arg were identified along the primary sequence, and finally, the MW and pI were calculated for peptides with Arg/Lys clusters. The helical contents of synthetic peptides were predicted using AGADIR software at EMBL (http://www.embl-heidelberg.de/Services/serrano/agadir/agadir-start.html) [13]. Secondary structures of the 58-kDa domain of m1 type strain 11638 (GenBank S72494) , m2 type strain 95-54 (GenBank U95971) and chimeric type ch2 (sequence from [14]) were performed and pIs were calculated as described.

2.3 Surface plasmon resonance (SPR)

BiaCore chips, pre-coated with streptavidin (SA-chips) from BiaCore AB (Uppsala, Sweden) were used. The B-HS binds to the surface an amount corresponding to a shift of 400–600 in RU. All experiments were performed in 5 mM PBS supplemented with 75 mM NaCl and 0.005% Tween 20 at pH 6.5. The following peptide concentrations were used: M530 39.4 µM, M531 39.1 µM, and rVacA 168 nM. Flow rates of 20 µl min−1 for peptides and 30 µl min−1 for rVacA were used. Experiments with synthetic peptides were carried out in the absence or presence of 9% (v/v) trifluoroethanol (TFE) in running buffer at the flow rate of 20 µl min−1. HS-coated SA-chips were regenerated using heparin (1 mg ml−1), 1 M NaCl, 2 M NaCl or 1 M NaCl/50 mM NaOH. A channel without B-HS was used to monitor non-specific binding of analytes. All sensorgrams were recorded as duplicates on a BiaCore 1000 Upgrade instrument and evaluated using BIAevaluation software 3.0, both from BiaCore AB.

3 Results

3.1 Prediction of potential H/HS binding sequences

The pI/MW values calculated for different types of the VacA 58-kDa subunit are shown in Table 1 and predicted secondary structures of the VacA 58-kDa subunit are aligned in Table 2. The primary sequence contains several clusters of Lys and Arg residues. We selected the following two peptide sequences for further binding studies: one helical, 715MKINSAQDLIKNKEHVLLKAKI736 (M530), and one non-helical, 801YKYLIGKAWKNIGISKTANGSKI823 (M351), according to PHD prediction (Table 2 in bold). The calculated MW and pI values of these peptides are shown in Table 3. The predicted helical contents of small synthetic peptides at pH 6.5 and ionic strength at 0.1 by AGADIR prediction in water solution were 1.12% (M530) and 0.36% (M531). We found that an inserted short sequence of 21 amino acids in the m2 type has a pI of 9.53 and a MW of 2297.56 Da. Surprisingly, the first 148 amino acids in the mid-region between 501 and 647 of the m1 type sequence (S72494), which has been proposed as a toxicity determinant for HeLa cells [14], has a pI of 9.16 and a MW of 16 534 Da, whereas the corresponding region of the m2 type sequence (U95971) between amino acids 495 and 666 with an insert of 21 amino acids has a pI of 9.47 and a MW of 19 024 Da. Naturally the chimeric mid-region m1–m2 in VacA toxin shows a surprisingly high pI of 9.49 and a MW of 16 551.62 Da. The corresponding region in HP0887 has a pI of 9.36 and a MW of 16 596, although it belongs to the m1 type.

View this table:
Table 1

Calculated pI values for different types of the 58-kDa subunit

VacA typepI/MW 58-kDa subunitpI/MW variable regionaa
m18.99/51 6799.16/16 534RFV.KLM
ch8.99/52 9919.49/16 552RFV.KLM
m29.21/55 5079.47/19 024KAV.KLM
HP00879.29/52 1109.36/16 596RFV.KLM
View this table:
Table 2

Aligned secondary structure predictions of VacA 58-kDa toxin domains from different VacA types

View this table:
Table 3

The calculated pI/MW values for synthetic peptides

Theoretical pI/MW (Da)MS (Da)
  • The last column presents the peptide masses from mass spectroscopic analysis by Vulpes Ltd.

3.2 SPR analysis

Binding studies by SPR analysis with synthetic peptides confirmed our prediction that the selected peptides should bind HS (Fig. 1). We found that a predicted H/HS binding region between aa 715 and 736 (M530) in the precursor sequence of HP0887 by PHD prediction is mainly helical. We observed binding of this peptide to immobilized HS in the presence of 9% (v/v) TFE, but not in the absence of a helical structure stabilizer (Fig. 1A). This is in agreement with the AGADIR prediction which showed a low 1.12% helical content of this peptide in water solution. The presence of TFE probably induces helical structures in these synthetic peptides which are needed for binding to HS. A second peptide (M531) bound to HS in the absence as well as in the presence of TFE with slightly different affinity (Table 4), although the residual binding represented by ΔRU was increased (Fig. 1B). These data show that additional helical structures were formed in the presence of TFE. Both peptides bound to HS with moderate affinity calculated from Langmuir dissociation and association kinetic parameters (Table 4). Injection of a rVacA polypeptide over a HS-coated chip at pH 6.5 showed residual binding to the surface (Fig. 2). No binding was observed on a non-coated chip. The best fit (χ2=0.674) to the experimental curve was achieved with a model considering a bivalent analyte. The affinity constant calculated from this model was KA=4.89×105 M−1 at the experimental conditions used. The binding models considering a conformational change showed a relatively good fit to the experimental curve (χ2=1.09) and higher binding affinity KA=1.19×106 M−1.

Figure 1

Sensorgrams of synthetic peptides at a flow rate of 20 µl min−1 and 25°C, pH 6.5. A: M530, 39.4 µM. Curve 1: binding to HS-modified surface; curve 2: binding in the presence of 9% TFE to non-modified surface; curve 3: binding to HS-modified surface in the presence of 9% TFE. B: M531, 39.1 µM. Curve 1: binding to non-modified surface; curve 2: binding in the presence of 9% TFE to non-modified surface; curve 3: binding to HS-modified surface; curve 4: binding to HS-modified surface in the presence of 9% TFE.

View this table:
Table 4

KA values for HS-synthetic peptide interaction and rVacA in the presence and absence of 9% TFE

Peptide9% TFEKA (M−1)
M530no binding
Figure 2

rVacA (168 nM) binding to HS. SPR sensorgram recorded at 30 µl min−1 and 25°C, pH 6.5.

4 Discussion

After entering epithelial cells, the H. pylori VacA cytotoxin causes vacuolation and finally cell death. The toxicification mechanism and how the internalization of VacA toxin occurs are not known in detail. It is proposed that the C-terminal 58-kDa domain of the mature toxin is involved in binding to a cell receptor whereas the N-terminal 37-kDa domain exhibits the toxic activity [2,15]. We predicted a putative H/HS binding sequence also in the C-terminal part of the VacA monomer. We found that close to the C-terminal end of the VacA precursor, the net charge of the predicted H/HS binding peptides exceeds the precursor pI value approximately by 2.5 pH units. We assume that the C-terminus of the native VacA cytotoxin possesses putative H/HS binding sites with a higher net charge than the N-terminus. We also observed differences in predicted secondary structures of heparin binding sequences corresponding to peptide M530 between the m1 and m2 VacA types which could lead to different binding to different cells.

This SPR study with synthetic VacA toxin peptides confirms the location of H/HS binding sites in this region (see Tables 14). The analysis of the experimental data obtained from binding of rVacA to HS is in good agreement with the prediction of multiple H/HS binding sequences in the VacA molecule. This supports earlier reported findings that the binding sites of toxin receptors are located in the 58-kDa C-terminal domain [16]. Although the observed binding has moderate affinity, the multimeric form of the VacA cytotoxin may bind to HS involving several domains and hence the final binding may be much stronger. More recently, Yahiro et al. [9] stated that activation of the vacuolation activity at high pH (11.5) is as effective as previously reported for toxin activation at low pH [5]. Under both conditions, a conformational change occurs and a higher toxic activity is obtained via pH-induced conformational changes. The same effect could probably be achieved via binding to a ligand, for example to HS. Such conformational changes induced by binding to H/HS have been reported for fibroblast growth factors and are needed for binding to their high-affinity receptors [17]. We found by SPR analysis of rVacA binding to HS that conformational changes should be considered.

Secondary structure elements of small peptides are less stable in water solution. Helical structures in water solutions can, however, be stabilized in the presence of TFE. This effect was recently described for heparin binding peptides by Lookene et al. [18]. One putative H/HS peptide, M530, demonstrates a similar behavior in our experiments.

In conclusion, we showed that putative H/HS binding polypeptides with a high net charge and heparin binding supporting secondary structure elements could be predicted using bioinformatics tools. We showed the binding of recombinant H. pylori cytotoxin VacA to surface-immobilized HS and propose HS as a new low-affinity receptor or co-receptor for this cytotoxin similar to fibroblast growth factor. Our bioinformatics- and biochemical study-based conclusion is supported by cell culture studies by Sommi et al. [19], who showed inhibition of cell vacuolation by heparin, but did not describe the mechanism. How H. pylori cytotoxin enters the host cells is not known. As HS is a component of most eukaryotic cell surfaces, including epithelial cells and their extracellular matrix, the HS-dependent endocytosis pathway could be one possible route of cell entry. We propose that H. pylori VacA cytotoxin may be internalized in host cells via a HS proteoglycan-dependent endocytosis pathway.


This study was supported by grants from the Swedish Medical Research Council (16x04723) and Lund University Hospital to T.W. and Swedish Medical Research Council Grant K98-04XS-12713 to B.D.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
View Abstract