Presence of virulence genes, adhesion and invasion of Arcobacter butzleri

Authors


Correspondence

Gretal Gölz, Freie Universität Berlin,

Institute of Food Hygiene, Königsweg 69, 14163 Berlin, Germany.

E-mail: greta.goelz@fu-berlin.de

Abstract

Aims

The pathogenic potential of Arcobacter butzleri isolates was investigated by detecting the presence of putative virulence genes and analysing the adhesive and invasive capabilities in cell cultures of human cell lines.

Methods and Results

The presence of ten putative virulence genes in 52 A. butzleri isolates was determined by PCR. The genes ciaB, mviN, pldA, tlyA, cj1349 and cadF were detected in all, whilst irgA (15%), iroE (60%), hecB (44%) and hecA (13%) were detected only in few A. butzleri isolates. On HT-29 cells, four of six isolates adhered to and three of them were able to invade, whilst all six isolates adhered to and invaded Caco-2 cells with higher degrees. The genes ciaB, cadF and cj1349 of all six isolates were sequenced, but no considerable changes of the amino acids in putative functional domains were observed.

Conclusion

Selected A. butzleri isolates adhere to and invade HT-29 and Caco-2 cells, which emphasize their human pathogenic potential. The efficiency of invasion depends on the eukaryotic cell line and individual bacterial strain used. We could not show any functional correlation between the amino acid sequence of CadF, CiaB or Cj1349 and the adhesive and invasive phenotype.

Significance and Impact of the Study

We have shown that some A. butzleri strains invade various cell lines. This underlines their pathogenic potential and hints at their relevance in human disease.

Introduction

Arcobacter spp. are gram-negative, motile and spiral-shaped bacteria belonging to the family of Campylobacteraceae. They are increasingly isolated from a wide range of food of animal origin and water. The highest prevalence in food is reported for poultry meat, followed by pork and beef (Rivas et al. 2004; Van Driessche and Houf 2007a,b). Arcobacter spp. have been rated a serious hazard to human health by the International Commission on Microbiological Specifications for Foods in 2002 (ICMSF, 2002). At present, 17 Arcobacter species have been characterized with A. butzleri as the most important and predominant species associated with human disease (Vandenberg et al. 2004). A. butzleri infections may result in abdominal pain with acute diarrhoea or prolonged watery diarrhoea for up to 2 months (Vandenberg et al. 2004). Little is known about the pathogenic mechanisms and putative virulence genes of A. butzleri. The genomic sequence of A. butzleri RM 4018 revealed the presence of 10 putative virulence genes: cadF, mviN, pldA, tlyA, cj1349, hecB, irgA, hecA, ciaB and iroE (Miller et al. 2007), but it is still unknown whether these putative virulence factors have similar functions as described for their homologues in other bacterial species. MviN is an essential protein required for peptidoglycan biosynthesis in E. coli (Ruiz 2008), but there is no direct evidence for the involvement of MviN in virulence of Salmonella typhimurium or E. coli (Inoue et al. 2008). In Campylobacter, cadF and cj1349 encode fibronectin-binding proteins, which promote the binding of bacteria to intestinal cells (Konkel et al. 1999; Flanagan et al. 2009), whilst Campylobacter invasive antigen B (CiaB) contributes to host cell invasion. HecA is a member of the filamentous hemagglutinin family and involved in attachment, aggregation and epidermal cell killing of Erwinia chrysanthemi (Rojas et al. 2002). The hecB encodes a haemolysin activation protein (Miller et al. 2007) and tlyA a haemolysin, which is also present in Mycobacterium tuberculosis and Serpulina hyodysenteriae (Wren et al. 1998). The outer membrane phospholipase PldA is associated with lysis of erythrocytes (Grant et al. 1997). The iron-regulated outer membrane protein IrgA and the periplasmic enzyme IroE are functional components for the iron acquisition in E. coli and therefore required for establishing and maintaining infections (Mey et al. 2002; Zhu et al. 2005; Rashid et al. 2006).

The potential virulence of A. butzleri isolates on CHO, Vero, HeLa and INT 407 cell lines was investigated by several authors. A. butzleri isolates showed in vitro cytotoxic effects in CHO cells and Vero cells, but any toxin genes were not identified yet. Adhesion of A. butzleri was observed in several cell line models (Musmanno et al. 1997; Johnson and Murano 2002; Carbone et al. 2003; Ho et al. 2007; Gugliandolo et al. 2008), but data on its ability to invade are sparse (Vandenberg et al. 2005).

The aim of this study was to examine the presence of putative virulence genes in A. butzleri strains isolated from different sources, to investigate the capability of A. butzleri to adhere to and invade into the cell lines HT-29 and Caco-2 and to analyse the sequence of putative invasive and adhesion genes.

Materials and Methods

Bacterial strains

All strains used in this study are listed in Table 1. A. butzleri were isolated from fresh meat (chicken meat, n = 23; pork, n = 12; mixed minced meat of pork and beef, n = 2), from water (n = 9) and from humans (n = 6). A. butzleri CCUG 30485 was included as reference strain. All isolates were grown on Mueller–Hinton blood agar (MHB; Oxoid, Wesel, Germany) or in Brucella Broth (BB; BD, Heidelberg, Germany) at 30°C in a micro-aerobic atmosphere (5 % O2, 10 % CO2) generated by the Anoxomat system (Mart, Drachten, the Netherlands). For the adhesion and invasion assays, the A. butzleri isolates were precultured overnight in BB at 30°C. The optical density of each preculture was measured at 600 nm and 10 ml BB inoculated obtaining an OD = 0·001. These cultures were further incubated for 24 h at 30°C. A total OD = 1 was centrifuged and resuspended in 450 μl PBS. Each well in the adhesion and invasion assay was inoculated with 50 μl of this bacterial suspension containing approx. 5 × 108 CFU.

Table 1. Bacterial isolates used in this study
StrainSourceSupplierStrain designation
  1. CCUG, Culture Collection University of Göteborg, Sweden; NRZ Helicobacter, National Reference Centre for Helicobacter pylori; University Medical Center Freiburg, Germany; LGL Oberschleißheim, Bavarian Health and Food Safety Authority, Oberschleißheim, Germany.

  2. a

    Pork and beef.

CCUG 30485HumanCCUGH1
FR1HumanNRZ HelicobacterH2
FR2HumanNRZ HelicobacterH3
FR3HumanNRZ Helicobacter 
FR4HumanNRZ Helicobacter 
FR5HumanNRZ Helicobacter 
Ab 47WaterLGL Oberschleißheim 
Ab 48WaterLGL Oberschleißheim 
Ab 49WaterLGL Oberschleißheim 
Ab 50WaterLGL Oberschleißheim 
Ab 51WaterLGL Oberschleißheim 
Ab 52WaterLGL Oberschleißheim 
Ab 53WaterLGL Oberschleißheim 
Ab 54WaterLGL Oberschleißheim 
Ab 55Chicken meatLGL Oberschleißheim 
Ab 56Chicken meatLGL Oberschleißheim 
Ab 57Chicken meatLGL Oberschleißheim 
Ab 60Chicken meatLGL Oberschleißheim 
Ab 61Chicken meatLGL Oberschleißheim 
Ab 62Chicken meatLGL Oberschleißheim 
Ab 63Chicken meatLGL Oberschleißheim 
Ab 64Chicken meatLGL Oberschleißheim 
Ab 65Chicken meatLGL Oberschleißheim 
Ab 66Chicken meatLGL Oberschleißheim 
Ab 68Chicken meatLGL Oberschleißheim 
Ab 69Chicken meatLGL Oberschleißheim 
Ab 70Mixed minced meataLGL Oberschleißheim 
Ab 73Mixed minced meataLGL Oberschleißheim 
Ab 74Chicken meatLGL Oberschleißheim 
Ab 75PorkLGL Oberschleißheim 
Ab 84WaterLGL Oberschleißheim 
Ab 86Chicken meatLGL Oberschleißheim 
Ab 87Chicken meatLGL Oberschleißheim 
Ab 88Chicken meatLGL OberschleißheimC1
Ab 89Chicken meatLGL OberschleißheimC2
Ab 91Chicken meatLGL Oberschleißheim 
Ab 92Chicken meatLGL Oberschleißheim 
Ab 93Chicken meatLGL Oberschleißheim 
Ab 94Chicken meatLGL OberschleißheimC3
Ab 96Chicken meatLGL Oberschleißheim 
Ab 97PorkLGL Oberschleißheim 
Ab 98PorkLGL Oberschleißheim 
Ab 99PorkLGL Oberschleißheim 
Ab 100PorkLGL Oberschleißheim 
Ab 101PorkLGL Oberschleißheim 
Ab 102PorkLGL Oberschleißheim 
Ab 103PorkLGL Oberschleißheim 
Ab 104PorkLGL Oberschleißheim 
Ab 106PorkLGL Oberschleißheim 
Ab 109PorkLGL Oberschleißheim 
Ab 110PorkLGL Oberschleißheim 
Ab 111Chicken meatLGL Oberschleißheim 

Cell culture

The human colon adenocarcinoma cells HT-29 (DSMZ ACC 299) were grown in 25-cm2 tissue culture flasks in RPMI1640 medium (Biochrom, Berlin, Germany) supplemented with 10% foetal calf serum superior (Biochrom) at 37°C in a 5% CO2 humidified atmosphere until a confluence of approx. 80% was reached. For adhesion and invasion assays, each well of a 24-well plate was seeded with 2 × 105 HT-29 cells and incubated for 24 h.

The human colon adenocarcinoma cells Caco-2 (DSMZ ACC 169) were grown in 75-cm2 tissue culture flasks in DMEM medium (Biochrom) supplemented with 10% foetal calf serum superior (Biochrom), 1% nonessential amino acids (Biochrom) and 5 μg ml−1 gentamicin (Roth, Karlsruhe, Germany) at 37°C in a 5% CO2 humidified atmosphere until a confluence of approx. 90% was reached. For adhesion and invasion assays, each well of a 24-well plate was seeded with 5 × 104 Caco-2 cells and incubated for 21 days with media changes every third day but without gentamicin.

DNA extraction

Bacteria were suspended in 500 μl 0·1 × TE buffer (1 mmol l−1 Tris/HCl, pH 8·0, 100 μmol l−1 EDTA; Roth) and centrifuged for 5 min at 16 000 g. The pellets were resuspended in 250 μl 5% Chelex (Bio-Rad, Munich, Germany) and incubated for 1 h at 56°C followed by 15 min at 95°C. The supernatants obtained after centrifugation were stored at 4°C until used for PCR.

PCR assays

Detection of putative virulence genes was performed by PCR. Primers and PCR protocols for partial amplification of cadF, pldA, irgA, hecA and hecB were used according to Douidah et al. (2012). Primers for partial amplification of cj1349, ciaB, mviN, tlyA and iroE were designed with Primer3 version 0.4.0 (Rozen and Skaletsky 2000) based on the published sequence of the whole-genome A. butzleri RM 4018 (ATCC 49616). All primers used in this study are listed in Table S1. PCRs were carried out in a total volume of 25 μl containing 1 × PCR buffer (Qiagen, Hilden, Germany), 200 μmol l−1 of each dNTP (Fermentas, St. Leon-Rot, Germany), 12·5 μmol l−1 of each primer, 0·5 U Taq-Polymerase (Qiagen) and 2 μl DNA. An initial denaturation step at 95°C for 4 min was followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 50°C for 30 s and elongation at 72°C for 30 s and a final elongation step for 6 min at 72°C.

To investigate whether adhesive and invasive phenotypes depend on alterations in the amino acid sequence of CadF, Cj1349 or CiaB, the genes of these proteins were sequenced and in silico-translated. The complete genes cj1349, ciaB and cadF were amplified in a total volume of 50 μl containing 1× PCR buffer, 200 μmol l−1 of each dNTP, 12·5 μmol l−1 of each primer, 0·5 U Taq-Polymerase and 5 μl DNA. An initial denaturation step at 95°C for 5 min was followed by 36 cycles of denaturation (94°C, 60 s), primer annealing (the temperatures are indicated in Table S1) for 60 s and elongation at 72°C for 60 s followed by final elongation for 5 min at 72°C.

After gel electrophoresis, PCR products were visualized with GR Green (Labgene, St. Ingbert, Germany) under UV light. PCR products were purified by GeneJet PCR-Purification Kit (Fermentas) and sequenced (GATC Biotech, Konstanz, Germany) with primers shown in Table S1. Sequences were trimmed and cds-translated with standard genetic code by EditSeq (DNASTAR Lasergene version 7, Madison, WI, USA). Alignment of the amino acid (aa) sequences was performed with Multalin version 5.4.1 (Corpet 1988).

Adhesion and invasion assay

The adhesion and invasion assays with HT-29 and Caco-2 cells were performed with six randomly selected A. butzleri isolates (three chicken meat isolates, C1–C3; three human isolates, H1–H3). HT-29 cells and Caco-2 cells were infected with approx. 5 × 108 bacteria. This corresponds to an MOI 1000 for HT-29 cells. The MOI for Caco-2 cells could not be calculated as cells were only enumerated before seeding. For adhesion assay, infected monolayers were incubated for 1 h at 37°C and rinsed three times with phosphate-buffered saline to remove unbound A. butzleri. To count adherent A. butzleri, HT-29 and Caco-2 cells were lysed with 500 μl 1 % SDS for 10 min.

To determine the invasive abilities of A. butzleri, monolayers were infected for 3 h, and extracellular bacteria were killed by 2-h incubation at 37°C with 300 μg ml−1 gentamicin (Roth). Cell monolayers were rinsed three times with PBS and lysed with 500 μl 1 % SDS for 10 min.

As control, the Campylobacter (Camp.) jejuni 81-176 strain was included in the adhesion and invasion assays. For Camp. jejuni 81-176, cells were lysed with 1 % Triton X-100 (Roth). Camp. jejuni 81-176 was cultivated as described for A. butzleri except at 37°C.

Total numbers of adherent or invasive bacteria were determined by plating serial dilutions of respective lysates on MHB agar, which were incubated for 48 h at 30°C. Each experiment was performed in triplicates, and average cell numbers were calculated out of three infected wells in each approach. Adhesion index and invasion index are calculated as ratio of adhering and invading bacteria of the inoculum, respectively.

Statistical analysis

P-values were calculated by nonparametric Mann–Whitney U-test (GraphPad Prism version 5, La Jolla, CA, USA). Results were considered significant at P < 0·05. Data were shown as median with interquartile range (IQR).

Results

Presence of putative virulence genes in Arcobacter butzleri

The presence of ten putative virulence genes in the 52 investigated A. butzleri strains is summarized in Table 2. Thirteen per cent (7 of 52) of the strains possessed all ten genes, whilst 31% of isolates carried only six genes (iroE, hecB, irgA and hecA were not detected). In 25% of the isolates, hecB, irgA and hecA were missing; in 10 %, iroE, irgA and hecA were absent; in 17 %, irgA and hecA were missing; and in 4 %, only hecA was missing (Fig. S1).

Table 2. Presence of putative virulence genes in Arcobacter butzleri isolates
Source n mviN cadF cj1349 ciaB hecA hecB iroE irgA tlyA pldA
  1. The presence of putative virulence genes is represented as percentage of analysed isolates for each source of origin.

Human610010010010017335017100100
Chicken meat23100100100100943659100100
Pork121001001001000425017100100
Minced meat210010010010000500100100
Water910010010010044676744100100

All investigated A. butzleri isolates carried the genes mviN, cadF, cj1349, ciaB, tlyA and pldA (Table 2). The gene iroE was detected in 50% of the human (3 of 6), pork (6 of 12) and minced meat (1 of 2) isolates and in 65–67% of the chicken meat (10 of 23) and water (6 of 9) isolates. None of the two isolates from minced meat encoded hecB, but 33 % of the human isolates, 42–43% of the pork and chicken meat isolates and 67% of the water isolates. The genes irgA and hecA were detectable in 17% of the human isolates, 9% of chicken meat isolates, 44% of water isolates, but in none of the minced meat isolates. In isolates recovered from pork meat, only irgA (17%) but not hecA was detected.

Adhesive and invasive abilities of Arcobacter butzleri

Three A. butzleri isolates recovered from chicken meat (C1-3) and three from humans (H1-3) were tested for their adhesive and invasive properties in HT-29 cells (Fig. 1). Two human and two chicken meat A. butzleri isolates adhered to HT-29 cells (Fig. 1a). Highest adherence was observed with the isolate C2 (1·14 × 100 fold [3·02 × 100] of the inoculum), followed by H1 (5·4 × 10−1 fold [5·76 × 10−1]), H3 (1·49 × 10−1 fold [2·12 × 100]) and C1 (1·3 × 10−1 fold [9·22 × 10−1]). Both chicken isolates showed invasion in HT-29 cells (C1 with 4·26 × 10−5 fold [1·54 × 10−4] and C2 with 5·05 × 10−5 fold [2·73 × 10−4] of the inoculum; Fig. 1b). Of the two adherent human isolates, only the human isolate H1 was invasive (2·86 × 10−5 fold [3·8 × 10−5]). The chicken isolate C3 showed adhesion and invasion in one of three approaches only, but to a far lesser extent (1·0 × 10−5 fold [2·65 × 10−5] and 5·0 × 10−7 fold [9·76 × 10−7]). The differences between the adhesive and nonadhesive or between invasive and noninvasive A. butzleri isolates, respectively, were statistically significant. As control, the Camp. jejuni 81-176 strain was investigated. This strain showed 5·64 × 10−3 fold [3 × 10−3] adhesion and 2·93 × 10−4 fold [2·7 × 10−4] invasion on HT-29 cells. This result shows differences in the abilities of A. butzleri strains to adhere to and to invade on HT-29 cells, independent of their origin.

Figure 1.

Adhesion index and invasion index of Arcobacter butzleri with the intestinal cell line HT-29. Adhesion of six A. butzleri isolates to HT-29 cells was detected after 1 h of incubation. To investigate the invasion ability, HT-29 cells were incubated for 3 h followed by 2-h gentamicin treatment, with six A. butzleri isolates. Adhesion index (a) and invasion (b) index are calculated as ratio of the inoculum. Expressed are the medians + interquartile range (IQR) (n = 3). H = human isolates, C = chicken meat isolates.

All isolates investigated were adhesive on Caco-2 cells (Fig. 2a). Highest adherence was observed for the human isolate H1 (3·2 × 100 fold [1·91 × 100]) followed by the chicken isolates C2 and C1 (1·37 × 100 fold [2·18 × 100] and 1·07 × 100 fold [1·35 × 100]). The isolates H3, C3 and H2 showed lower adherence with 7·55 × 10−2 fold [3 × 10−2], 5·71 × 10−3 fold [1·2 × 10−3] and 1·25 × 10−4 fold [2·45 × 10−3]. These values were all significantly different (P < 0·05), except for C2 and C1. Further, all isolates were invasive on Caco-2 cells (Fig. 2b). The isolates H2, H3 and C3 showed lesser invasion (1·88 × 10−5 fold [6 × 10−5], 1·43 × 10−4 fold [1 × 10−4] and 6·47 × 10−4 fold [7 × 10−3]) than the isolates C2 and H1 (1·52 × 10−3 and 1·47 × 10−3 fold [2 × 10−3 and 2·5 × 10−3]) and C1 (5·0 × 10−3 fold [1·5 × 10−2]). On Caco-2 cells, 5·41 × 10−3 fold [2·5 × 10−3] of the control strain Camp. jejuni 81-176 adhered to and 5·88 × 10−3 fold [1·5 × 10−2] invaded.

Figure 2.

Adhesion index and invasion index of Arcobacter butzleri with the intestinal cell line Caco-2. Adhesion of six A. butzleri isolates to Caco-2 cells was detected after 1 h of incubation. To investigate the invasion ability, Caco-2 cells were incubated for 3 h followed by 2-h gentamicin treatment, with six A. butzleri isolates. Adhesion index (a) and invasion (b) index are calculated as ratio of the inoculum. Expressed are the medians + interquartile range (IQR) (n = 3). H = human isolates, C = chicken meat isolates.

Comparison of CadF, CiaB and Cj1349 amino acid (aa) sequences

To elucidate whether the different adhesive and invasive phenotypes depend on alterations in CadF, Cj1349 or CiaB, the genes of these proteins were sequenced and in silico-translated employing the standard genetic code in the six investigated A. butzleri isolates. The Cj1349 aa sequences of the H2, H3, C1, C2 and C3 isolates showed only 0·9–1·6% substitutions compared with the reference strain A. butzleri CCUG 30485 (H1), but all isolates showed a conserved motif D/E-X-W/Y-X-H (Fig. S2A; positions 367-371). This motif is part of a conserved domain of the DUF814 super family, occurring in proteins annotated as fibrinogen-/fibronectin-binding proteins, and might be functionally important (box in Fig. S2A). When comparing the CadF sequences of isolates investigated with the reference strain A. butzleri CCUG 30485, aa alterations varied from 1·5 to 3·25%. These alterations were often located at the same position (Fig. S2B). The fibronectin-binding site of Campjejuni (FRLS) described by Konkel et al. (2005) is substituted to YNLA in A. butzleri, but this site is conserved within the analysed isolates (box in Fig. S2B). Pronounced variations could be observed within the CiaB sequences (Fig. S2C). The aa sequences of H2 and H3 had the same lengths (630 aa) as the reference strain A. butzleri CCUG 30485, whilst CiaB in C2 is composed of 633 aa, in C3 of 623 aa and in C1 of 523 aa. The CiaB sequences of C1 and C2 showed a three-aa insert (AKS) in between positions 332 and 333. However, all isolates showed aa substitutions of 3–5 % compared with the reference strain. But they all shared the conserved zinc-binding domain TIGHEYGHIL at positions 441-450 (Onozato et al. 2009). In the isolates C1 and C3, isoleucine is substituted by valine I442V (box in Fig. S2C).

Discussion

Arcobacter butzleri is considered a potential foodborne pathogen (summarized by Collado and Figueras 2011), but little is known about the pathogenic mechanisms and putative virulence factors. Investigations on the pathogenicity of A. butzleri have been primarily based on the knowledge on Campylobacter (Musmanno et al. 1997; Johnson and Murano 2002). However, a study by Miller et al. (2007) indicated that the proteome of A. butzleri shares greater phylogenetic similarities with members of the Helicobacteraceae like Sulfurimonas denitrificans and Wolinella succinogenes and those of the deep-sea vent Epsilonproteobacteria Sulfurovum and Nitratiruptor than with Campylobacter.

In this study, we investigated the presence and distribution of putative virulence genes in A. butzleri isolates recovered from different sources (n = 52, including the reference strain A. butzleri CCUG 30485). Only the genes mviN, cadF, cj1349, ciaB, pldA and tlyA, which share homologies to genes in Campylobacter, were identified in all A. butzleri isolates. This is in agreement with Douidah et al. (2012) who detected these genes in all 182 investigated A. butzleri isolates originating from different sources. Furthermore, these authors identified the presence of nine virulence genes (iroE was not investigated) in 15 % of analysed strains, which is comparable to our results (13 %). In contrast to Douidah et al. (2012), we could confirm neither the similar occurrence of hecA in human, pork and chicken meat isolates nor the higher prevalence of hecB in human isolates compared with pork and chicken meat isolates. Furthermore, we could not detect a lower occurrence of irgA in pork isolates compared with human and chicken meat isolates. However, these conflicting results might be due to the lower number of isolates included in our study.

Previous studies demonstrated the capacity of A. butzleri to adhere to different cell lines such as HeLa, Hep-2, INT 407, Caco-2 and IPI-2I (Musmanno et al. 1997; Ho et al. 2007). Only few studies examined the ability of A. butzleri to invade colon and larynx cells (Musmanno et al. 1997; Vandenberg et al. 2005; Ho et al. 2007). Amongst these studies, only Vandenberg et al. (2005) described an ability of A. butzleri to invade cell lines. These authors showed that 33% of the strains (4/12) were invasive in the human larynx carcinoma cell line Hep-2.

We also investigated the adhesive and invasive properties of three human and three chicken meat A. butzleri isolates on HT-29 and Caco-2 cells. Whilst two human (H1 and H3) and two chicken meat (C1 and C2) isolates adhered to HT-29 cells, only one human (H1, CCUG 30485) but both chicken meat isolates also invaded. On Caco-2 cells, all investigated isolates adhered to and invaded. The two isolates H2 and C3, which showed no adhesion on HT-29 cells, showed also significantly lower adhesion on Caco-2 cells compared with the four isolates H1, H3, C1 and C3. Coincident with this, the three isolates H2, H3 and C3, noninvasive on HT-29 cells, showed lesser invasion in the Caco-2 cells. Therefore, it seems that the adhesive and invasive phenotype of A. butzleri depends on the isolate as well as the cell line used.

That A. butzleri isolates investigated by others did not show invasive potential on the cell lines HeLa, INT 407, Caco-2 and IPI-2 could be explained by the different isolates, different cell lines or protocols used (Musmanno et al. 1997; Ho et al. 2007).

So far, it is still unknown whether A. butzleri adhere to and invade in vivo and which part of the human intestine might be affected. Based on the invasion data obtained by our assays, it can be speculated that A. butzleri invade poorly differentiated cryptic cells (HT-29: invasion index from 2·86 × 10−5 to 5·05 × 10−5 fold) with lower efficiency than the better differentiated colonocytes (Caco-2: invasion index from 1·88 × 10−5 fold to 5·0 × 10−3 fold). In addition, Bücker et al. (2009) demonstrated that A. butzleri CCUG 30485 is able to induce barrier dysfunctions in HT-29/B6 cells, thereby facilitating translocation of bacteria and probably inducing diarrhoea.

To elucidate which virulence genes might be responsible for adhesion and invasion, A. butzleri isolates with different virulence gene patterns were used for in vitro cell culture assays. The isolates H1 and C1 possessed all ten putative virulence genes, whilst in the isolates H2 and C2 the genes irgA, hecA and hecB and in H3 and C3 additionally iroE were missing. Based on these groupings, we could not detect any correlation between virulence gene patterns and adhesive or invasive capabilities. Because only two isolates per virulence gene pattern group were used, further strains need to be tested to confirm our observations. As all isolates possessed the adhesion genes cadF and cj1349 and the invasion gene ciaB but had different adhesive and invasive phenotypes, we compared the amino acid sequences of these putative virulence factors. Although the amino acid sequences of the three genes showed alterations from 0·9 to 5 % when compared with the reference strain A. butzleri CCUG 30485, the putative functional domains were conserved for all three genes in the six isolates investigated.

The CadF fibronectin-binding site of Camp. jejuni (FRLS) revealed substitution of similar amino acids within A. butzleri (YNLA). F/Y are both hydrophobic and aromatic, R/N are both polar, and S/A are both small and less hydrophobic amino acids. Therefore, this domain might represent also the fibronectin-binding site in A. butzleri. Similarly, both I and V (substitution in the zinc-binding domain of CiaB) belong to hydrophobic and aliphatic amino acids (Livingstone and Barton 1993). In addition, the C-terminus of CiaB does not seem to be important for efficient invasion of A. butzleri as isolate C1 showed high invasiveness in HT-29 and Caco-2 cells although the last 100 C-terminal amino acids of CiaB are missing. Also the insertion of the three-aa AKS in CiaB of the isolates C1 and C2 does not seem to be functionally relevant. In addition to these two strains showing invasion in HT-29 cells, strain H1 showed invasion on HT-29 cells, even though the three amino acids were absent in this isolate. The reasons for these differences in the amino acid sequence are not known. However, no correlation between adhesive and/or invasive phenotypes and the amino acid sequence of Cj1349, CiaB and CadF factors was shown.

We have shown here that A. butzleri isolates have the capability to adhere to and to invade cells as classical gastrointestinal pathogens do. However, further in vitro and in vivo studies are required to clarify the role of these virulence factors in the pathogenesis of A. butzleri infections.

Acknowledgements

The work of S.S. and T.A. is supported by the German Research Foundation (SFB852). The work of G.G. is supported by the Federal Office for Agriculture and Food (FKZ 281680141, Innostep).

Ancillary