Brefeldin A enhances Helicobacter pylori vacuolating cytotoxin-induced vacuolation of epithelial cells

Abstract

Intracellular VacA localises to the vacuolar (late endosome/lysosome) membrane, but little is known about the trafficking of the toxin beyond this region. We show that the Golgi-disturbing agent brefeldin A (BFA) enhances VacA-induced vacuolation of epithelial cells by Helicobacter pylori co-culture and, importantly, BFA treatment induces vacuolation by less toxic forms of VacA. The effect is BFA dose-dependent and occurs within 2.5 h. These data suggest that VacA may be routed deeper within the cell than the vacuole, and that vacuolation is minimised when this occurs efficiently. This may explain why some forms of VacA do not cause vacuolation and why vacuolation is minimal at the low bacteria:cell ratios observed in vivo.

Helicobacter pylori, Vacuolating cytotoxin, VacA, Brefeldin A

1 Introduction

Helicobacter pylori colonises the stomachs of approximately half the world's population. It is the main cause of peptic ulcer disease, distal gastric adenocarcinoma and primary gastric lymphoma, although the majority of infected people develop none of these conditions. The association of H. pylori with disease has been linked to a number of virulence factors, including possession of an active form of the vacuolating cytotoxin (VacA) [1]. Active forms of VacA induce marked cytoplasmic vacuolation in epithelial cell lines. However, although most H. pylori strains possess the toxin gene, vacA, this is polymorphic [2]. Variation between alleles is most marked in two regions: the signal region, encoding the second half of the signal peptide and the N-terminus of VacA (which may be type s1 or s2) and the mid-region, encoding part of the p58 cell binding domain (which may be m1 or m2). The s1 signal region codes for active VacA whereas the s2 form codes for VacA with a 12 residue hydrophilic extension on the N-terminus which blocks vacuolating activity [3]. The mature toxin has two domains, an N-terminal p37 and a C-terminal p58 domain, linked via a flexible loop region. VacA interacts with epithelial cells via the p58 domain, part of which is encoded by the variable vacA mid-region [4,5]. The m1-form of VacA appears capable of binding to receptors on a wide range of epithelial cells, whereas the m2-form has limited cell/receptor specificity [6]. Upon internalisation into epithelial cells, VacA localises to the membranes of late endosomes/lysosomes [7] that form vacuoles following heterotypic fusion between the two compartments [8,9]. VacA forms anion-selective hexameric pores in artificial lipid membranes and in cell surface membranes [10] and is assumed to do the same in endosomal membranes. The influx of anions into late endosomes through the VacA channel is thought to create an osmotic potential that causes swelling and vacuolation [11]. Although VacA has been shown to persist within vacuoles for over 72 h with little degradation of the toxin [9], the fate of VacA after reaching endosomes is not known.

Epithelial cell vacuolation is the most obvious effect of active VacA in vitro, but it is rarely marked in vivo in the human stomach. Nor is VacA-induced vacuolation obvious during co-culture of many strains of H. pylori with epithelial cells, especially at low bacteria:cell ratios similar to those seen in the stomach. In this study, we address the hypothesis that lack of vacuolation in these situations is due to efficient trafficking of VacA beyond the vacuole. Previously, Garner and Cover [4] showed that treatment of cells with the Golgi-disturbing agent brefeldin A (BFA), a macrocyclic lactone which causes the relocation of the Golgi apparatus to the endoplasmic reticulum by interfering with the binding of COPI coat protein by targeting the Sec7 guanine nucleotide exchange factor of Arf1 thereby blocking Arf1 activation and assembly of COPI coats, did not prevent vacuolation of HeLa cells by a highly vacuolating s1/m1 VacA, suggesting that transport through the Golgi apparatus was not required for vacuolation to occur. We now look again at the effect of BFA upon the intracellular trafficking of VacA, in order to determine whether differences in vacuolating capacity might be due to differences in the processing of the different forms of VacA through the cell.

2 Materials and methods

2.1 Materials

Brefeldin A, β-cyclodextrin, chymostatin, leupeptin, N-(trans-epoxysuccinyl)-l-leucine 4-guanidinobutylamide (E-64), pepstatin A, and other general laboratory chemicals were supplied by Sigma–Aldrich (Poole, UK). Tissue culture media was supplied by Invitrogen (Paisley, UK). Blood-agar plates and H. pylori growth medium reagents were supplied by Oxoid (Basingstoke, UK). CampyPaks were supplied by Becton Dickinson (Oxford, UK). ECL Western blotting detection reagents and Hyperfilm ECL were supplied by Amersham Biosciences (Little Chalfont, UK).

2.1.1 Culturing of strains and preparation of culture supernatants

Helicobacter pylori strains 60190 (ATCC 49503), Tx30a (ATCC 51932), 93–67 [12], J99 (ATCC 700824), G27 [13], and J123 [12], were grown on blood-agar plates in a microaerobic environment within a MACS-VA500 workstation (Don Whitley Scientific, Shipley, UK). The strains were sub-cultured every 1–2 days. For preparation of culture supernatants, strains were swabbed from 24 h growth plates into 12 ml sulphite-free brucella broth containing 0.2% (w/v) β-cyclodextrin [14], and grown in a gas jar shaking in a microaerobic environment generated by a CampyPak for 2–3 days at 37 °C. After incubation, cultures were harvested by centrifugation at 5000g for 10 min, the supernatant filtered and stored at 4 °C.

2.1.2 Vacuolation assays

RK13 (rabbit kidney) epithelial cells and human gastric epithelial AGS and MKN28 cells were seeded into 96 well plates at a density of 1–3 × 10⁴ cells/well and grown at 37 °C in a 5% CO₂ air humidified atmosphere overnight to produce plates with sparse to fully confluent cell density. H. pylori strains were swabbed from 24 h growth plates into nutrient mixture F-12 Ham and the optical density at 550 nm determined. An OD₅₅₀ of 0.1 corresponded to an average multiplicity of infection (MOI) of 300. Bacteria were diluted to a starting OD₅₅₀ of 0.1–0.01 into minimal essential medium alone or medium containing BFA, which was then serially diluted to generate a range of MOIs. After incubation for 2.5–24 h at 37 °C in a 5% CO₂ air humidified atmosphere, the degree of epithelial cell vacuolation was determined by the neutral red uptake assay [15] and/or microscopic examination. Alternatively, cells were pre-treated for 1 h at 37 °C with BFA (or medium alone) before the addition of culture supernatant and incubation as above. In experiments using protease inhibitors, cells were co-cultured with H. pylori at OD₅₅₀ 0.1–0.001 in the presence or absence of 0.1 μg/ml BFA and/or 33 μM chymostatin (an inhibitor of cathepsins B, D, G, H and L), 10 μM E-64, 20 μM leupeptin (inhibitors of cathepsins B, H, and L), and 1 μM pepstatin A (an inhibitor of cathepsin D), or triple these concentrations, before incubation as above. Statistical analysis was performed using a two-tailed Student's t-test.

3 Results and discussion

3.1 BFA increases VacA-induced vacuolation of epithelial cells

Vacuolation induced by VacA does not appear to interfere with intracellular trafficking from endosomes to the trans-Golgi network and beyond since vacuolated cells are still sensitive to the toxin ricin [16] which is transported to the lumen of the endoplasmic reticulum via endosomes [17]. Our primary hypothesis was that the lack of vacuolation at low bacterial:cell ratios was due to efficient trafficking of the toxin through this pathway. To address this, we aimed to block transport to the Golgi apparatus using BFA and to assess the effect on vacuolation. In order to show that the Golgi apparatus was disrupted upon treatment with BFA, we incubated RK13 cells with 0.01–10 μg/ml BFA for 1 h at 37 °C before the cells were fixed, permeabilised and the Golgi apparatus stained with anti-Golgin 97 antibodies (Molecular Probes, Eugene, OR) and a fluorescently labelled secondary antibody. The perinuclear Golgi staining present in untreated cells was abolished when RK13 cells were treated with all concentrations of BFA, clearly indicating that BFA disrupts the Golgi cisternae in this cell line (data not shown). We then treated RK13 cells with 0.0001 to 10 μg/ml BFA during co-culture with H. pylori strain 60190 (which produces a highly active s1/m1 form of VacA) and quantified VacA-induced vacuolation using the neutral red uptake assay. We found that vacuolation was enhanced markedly (Fig. 1(a)). To prove that this H. pylori-induced vacuolation was caused by VacA, we performed parallel experiments with a VacA null isogenic mutant of strain 60190 [18] and as expected this did not induce vacuolation. The BFA-induced enhancement in VacA-induced vacuolation was not confined to RK13 cells, as we also observed an enhancement in MKN28 cells (Fig. 1(b)) and AGS cells (data not shown). These findings suggest the possibility that VacA could be transported to the Golgi apparatus, which is supported by the recent finding of Kuo and Wang [19] who showed by confocal microscopy the co-localisation of purified VacA from strain 60190 with the Golgi marker wheat germ agglutinin in CHO cells.

1

Brefeldin A increases VacA-induced vacuolation of epithelial cells in a dose-dependent manner. (a) RK13 cells were co-cultured with H. pylori strain 60190 at a range of MOIs (OD₅₅₀ 0.0003–0.005; MOI 1–15) in quadruplicate for 16 h at 37 °C in the presence of 0 (♦), 0.0001 (●), 0.001 (×), 0.01 (⋄), 0.1 (^), 1 (▲), or 10 (Δ) μg/ml BFA, before vacuolation was assessed by the neutral red assay, and expressed as a percentage relative to the untreated control at OD₅₅₀ 0.005. The error bars are not shown for clarity, but were within 13% of the mean. At OD₅₅₀ 0.005 there was a significant increase in vacuolation in the presence of 0.01–1 μg/ml BFA (p<0.05), but not in the presence of 0.0001–0.001 (p>0.1) or 10 μg/ml BFA (p>0.05) compared to the untreated control. (b) MKN28 cells were co-cultured with H. pylori strain 60190 at OD₅₅₀ 0.01 (MOI 30) in quadruplicate for 16 h at 37 °C in the presence of 0, 0.001, 0.01, or 0.1 μg/ml BFA, before vacuolation was determined by the neutral red uptake assay. There was a significant increase in vacuolation with 0.1 (p<0.05) or 0.01 μg/ml BFA (p<0.01), but not with 0.001 μg/ml BFA (p>0.1), compared to the untreated control.

To support a specific action of BFA, we checked that it was acting in a dose-dependent manner on RK13 cells. We showed that vacuolation induced by VacA progressively increased with increasing doses of BFA (Fig. 1). At the highest concentration of BFA used, we observed a decrease in vacuolation, which we speculate may be the result of direct BFA cell toxicity. BFA at high dose has been shown to affect the cytoskeleton of normal rat kidney (NRK) cells [20], and to cause apoptosis within cultured cells [21]. The three epithelial cell lines, when treated with BFA alone, did not vacuolate at any of the concentrations used. Also, the large BFA-induced enhancement in VacA-induced vacuolation occurred independently of the level of epithelial cell density (whether plated sparse or fully confluent).

3.2 BFA does not damage cells non-specifically at moderate concentrations

Because high doses of BFA appeared toxic to cells, we checked that significant toxicity did not occur at moderate concentrations and so contribute to its effects upon VacA distribution and vacuolation. The specific Golgi-disrupting effects of BFA on mammalian cells have previously been shown to be reversible [22]. To show that this was true in our system, we treated cells with 0.1 μg/ml BFA for 1 h at 37 °C or left cells untreated, washed to remove BFA (or left BFA on the cells) then incubated for 2 h, before co-culture with H. pylori strain 60190 for 16 h. We then assessed vacuolation by neutral red uptake. As expected, the removal of BFA reduced vacuolation to the same level as control cells co-cultured with H. pylori alone. This shows that washing-out the BFA restores normal intracellular trafficking pathways within epithelial cells, thus preventing the enhancement in VacA-induced vacuolation. To further confirm that BFA was acting in the expected manner and not having other non-specific effects on VacA-induced vacuolation, we assessed the time course of the BFA effect. The disruption of the Golgi apparatus upon BFA treatment begins within minutes and is complete after 30–60 min [23]. Since the uptake of VacA is slow [4], we expected the increase in VacA-induced vacuolation to occur within a few hours and to increase over time after addition of VacA to BFA-treated cells. We pre-treated RK13 cells with 0, 0.01, or 0.1 μg/ml BFA for 1 h at 37 °C prior to the addition of culture supernatant from H. pylori strain 60190 and further incubation at 37 °C in the presence of 0–0.1 μg/ml BFA. As expected, more cells became vacuolated in the presence of BFA and this increase began within 2.5 h, and became more pronounced at 5.5 h (Fig. 2).

2

Brefeldin A increases VacA-induced vacuolation within 3h of addition. RK13 cells were pre-treated with 0.01 or 0.1 μg/ml BFA, or left untreated, for 1 h at 37 °C before the addition of VacA-containing culture supernatant (1:2 dilution) prepared from H. pylori strain 60190 with or without BFA. The cells were then incubated at 37 °C and examined for vacuolation (arrowed) by microscopy of randomly chosen fields from 2.5 to 5.5 h. Vacuolation is shown to increase over time under all conditions, and this effect is potentiated with the addition of BFA.

3.3 At low bacteria:cell ratios, vacuolation can be induced by BFA treatment

At low MOI most closely mimicking the situation in the human stomach H. pylori causes little, if any, epithelial cell vacuolation. A similar paucity of vacuolation is observed in H. pylori infection in vivo. We hypothesised that this lack of vacuolation could be because VacA might be being trafficked deeper within the cell, beyond the late endosomal compartment. To address this, we reassessed the effect of BFA on H. pylori-induced vacuolation of RK13 cells using a range of MOIs of H. pylori, and three concentrations of BFA (0.1, 1, and 10 μg/ml). As before, BFA increased VacA-induced vacuolation (as assessed by neutral red uptake): at an OD₅₅₀ of 0.01, by 32% with 10 μg/ml BFA and by 51% with 1 μg/ml BFA (Fig. 3(a)). The neutral red uptake assay only quantifies vacuolation when there is a sufficient incorporation of the dye into the cells, and this only occurs at high levels of vacuolation. Therefore, to assess vacuolation at lower MOIs of H. pylori, we quantified numbers of vacuolated cells in multiple randomly selected fields by direct microscopy. At low MOI (OD₅₅₀ 0.0016) BFA markedly increased the number of vacuolated cells at all three concentrations used (Fig. 3(b)–(e)). Thus, BFA induces vacuolation in cells exposed to H. pylori that do not exhibit it, as well as enhancing vacuolation in vacuolated cells. This suggests that at low MOIs (which may be more physiological) VacA might be transported elsewhere within the cell, and that vacuolation only occurs if this trafficking is blocked, or at higher VacA concentrations.

3

Brefeldin A increases VacA-induced vacuolation at low bacteria:cell ratios. RK13 cells were co-cultured with H. pylori strain 60190 at a range of MOIs (OD₅₅₀ 0.04–0.00006; MOI 0.2–120) in quadruplicate for 16 h at 37 °C in the presence of 0 (♦), 0.1 (Δ), 1 (^), or 10 (⋄) μg/ml BFA, before vacuolation was assessed by the neutral red uptake assay (a) and microscopy of randomly chosen fields at OD₅₅₀ 0.0016 (b–e). Neutral red uptake is expressed as a percentage of the co-cultured cells not treated with BFA at OD₅₅₀ 0.04. At OD₅₅₀ 0.008 there was a significant increase in vacuolation (p≤0.02) in the presence of all three concentrations of BFA, compared to the untreated control. Whereas there is little vacuolation in the absence of BFA (b), there is a clear increase in the number of vacuolated cells in the presence of BFA (c–e). Error bars are omitted from the graph for clarity, but were within 15% of the mean.

3.4 BFA enhances the vacuolating potential of less toxic forms of VacA

Essentially all strains of H. pylori express VacA, but most do not cause epithelial cell vacuolation following co-culture. We hypothesised that this was in part because some forms of VacA were trafficked efficiently beyond the vacuole. To address this, we co-cultured RK13 cells with various H. pylori strains expressing different forms of VacA. The s1/m2 type of VacA causes vacuolation of RK13 cells (but not of other cultured epithelial cell lines) when the purified toxin or crude toxin in culture supernatants is added at high concentrations [6]. However, following co-culture with low MOIs, vacA s1/m2 strains exhibit heterologous effects. For example, strain 93–67 causes appreciable vacuolation of RK13 cells, but strain J123 does not cause obvious vacuolation. We performed co-cultures of RK13 cells with these strains at low MOIs (OD₅₅₀≤ 0.025) with and without 0.1 μg/ml BFA. BFA markedly increased vacuolation by strain 93–67 and induced vacuolation by strain J123 (Fig. 4). Addition of H. pylori strain Tx30a (type s2/m2 VacA) to RK13 cells in the presence of BFA did not lead to vacuolation, showing that the non-vacuolating phenotype of s2/m2 VacA is not due to trafficking of this form of VacA elsewhere within the cell. Although most H. pylori strains with type s1/m1 VacA induce high levels of vacuolation during epithelial cell co-culture, some do not, including the genome sequence strain J99 and strain G27. We co-cultured these two strains with RK13 cells with and without BFA, and showed that as for VacA s1/m2 strains, BFA enhanced vacuolation (Table 1). Taken together, these experiments show that one reason why many strains of H. pylori induce little or no vacuolation is that VacA is processed efficiently beyond the endosomal system.

4

Brefeldin A enhances VacA-induced vacuolation of the less toxic s1/m2 forms of the toxin. RK13 cells were co-cultured with H. pylori strains 93–67 and J123 (both of which express s1/m2 VacA) at OD₅₅₀ 0.006 (MOI 20) (strain 93–67) or 0.025 (MOI 75) (strain J123, which can be seen to clump together) for 16 h at 37 °C in the absence (upper panels) or presence (lower panels) of 0.1 μg/ml BFA. Cells were examined for vacuolation by microscopy of randomly chosen fields.

1

Effect of BFA upon VacA-induced vacuolation

Strain	VacA type	RK13 cell vacuolation	Enhancement with BFA
60190	s1/m1	Yes	Yes
J99	s1/m1	Yes (weak)	Yes
G27	s1/m1	No	Yes
93–67	s1/m2	Yes	Yes
J123	s1/m2	No	Yes
Tx30a	s2/m2	No	No

3.5 BFA treatment may increase lysosomal degradation of VacA

Although we have shown that BFA appears to be acting specifically, our results led us to question in more detail whether BFA could be routing VacA to other compartments within the cell. There are data to support this idea: although the predominant effect of BFA upon mammalian cells is the disruption of the Golgi apparatus and redistribution of the Golgi to the endoplasmic reticulum, BFA at 5 μg/ml has also been reported to affect trafficking to lysosomes within NRK cells [24]. It is possible, therefore, that BFA might also block delivery of VacA to lysosomes, preventing degradation, which may contribute to an increase in the amount of endosome-associated VacA and increased levels of vacuolation. To investigate whether lysosomal degradation occurs in BFA-treated cells we used a range of protease inhibitors known to inhibit the action of lysosomal cathepsins. We co-cultured RK13 cells with H. pylori strain 60190 with or without protease inhibitors (chymostatin, E-64, leupeptin, and pepstatin A) and/or 0.1 μg/ml BFA. The addition of protease inhibitors alone caused a minor enhancement in VacA-induced vacuolation of RK13 cells (Fig. 5), but this increase was not significant (p>0.1). Tripling the concentrations of these inhibitors produced no additional increase in vacuolation. This data agrees with previous results showing that VacA persisted within MKN28 cells with little degradation of the toxin [9]. The addition of both BFA and protease inhibitors was additive, increasing vacuolation further (Fig. 5). This was found to be a significant finding in Fig. 5(b) when vacuolation was assessed by direct counting of vacuolated cells (p<0.01), but not significant in Fig. 5(a) where vacuolation was quantified by neutral red uptake (p>0.1). The BFA-induced increase in VacA-induced vacuolation of RK13 cells was found to be significant in all cases (p<0.01). This suggests that BFA treatment may prevent the degradation of some internalised VacA, although the additional enhancement in VacA-induced vacuolation in the presence of BFA and protease inhibitors is small compared to the increase induced by BFA alone. This also indicates that the major effect of BFA is not through lysosomal routing.

5

Cathepsin inhibitors do not cause a significant enhancement of VacA-induced vacuolation of RK13 cells. RK13 cells were co-cultured with H. pylori strain 60190 at OD₅₅₀ 0.01 (MOI 30) for 16 h at 37 °C in quadruplicate in the absence or presence of 0.1 μg/ml BFA, protease inhibitors, or BFA and protease inhibitors together, before vacuolation was determined by the neutral red uptake assay (a) and by microscopy of randomly chosen fields (b–f). (c) Co-cultured cells without BFA or protease inhibitors; (d) co-cultured cells with BFA; (e) co-cultured cells with protease inhibitors; (f) co-cultured cells with BFA and protease inhibitors.

In summary, we have shown that BFA increases VacA-induced vacuolation of epithelial cells potentially by preventing transport of the toxin deeper within the cell, thereby causing an accumulation of VacA within the endosomal system. At low bacterial:cell ratios (which most closely mimic the situation in the stomach) efficient trafficking appears to be one reason why many forms of VacA are non-vacuolating. We have shown that BFA disrupts the Golgi apparatus within RK13 cells, that the effects are reversible, and that the BFA-induced enhancement in VacA-induced vacuolation occurs over time in a dose-dependent manner. This suggests that VacA may be transported to the Golgi cisternae, as has also been shown by immunofluorescence in CHO cells [19]. However, as BFA has also been shown to have other effects upon cells, including cytoskeletal alterations [20] and tubulation of endosomes and the trans-Golgi network in NRK cells at high concentration [24], the possibility remains that BFA may enhance VacA-induced vacuolation by a process independent of Golgi disruption. Should tubulation of the endosomal system occur with BFA treatment in the epithelial cell lines used here, then this may also lead to the accumulation of VacA, and an increase in vacuolation, and also implies that BFA treatment prevents the intracellular transport of the toxin.

Our data suggest that VacA from different strains of H. pylori is trafficked with varying efficiency by epithelial cells, and this leads to differences in the extent of vacuolation and even to whether or not it occurs. As vacuolation correlates with disease expression, we speculate that differences in efficiency of processing VacA from different strains may be an important determinant of differences in pathogenicity. Recently, purified VacA has been shown to target mitochondria [25] leading to the release of cytochrome c and activation of caspase 3 [26] leading to apoptosis. As this process would require the cytosolic delivery of VacA, one possibility is that VacA is trafficked elsewhere in the cell, possibly through the Golgi apparatus, en route to the cytosol.

Acknowledgements

This work was supported by a grant from the Medical Research Council (UK), and in part by a grant from Cancer Research UK. John Atherton is funded by a Senior Clinical Fellowship from the Medical Research Council.