Hcp and VgrG1 are secreted components of the Helicobacter hepaticus type VI secretion system and VgrG1 increases the bacterial colitogenic potential


For correspondence. E-mail suerbaum.sebastian@mh-hannover.de; Tel. (+49) 511 532 6769; Fax (+49) 511 532 4366; E-mail josenhans.christine@mh-hannover.de; Tel. (+49) 511 532 4348; Fax (+49) 511 532 4355.


The enterohepatic Epsilonproteobacterium Helicobacter hepaticus persistently colonizes the intestine of mice and causes chronic inflammatory symptoms in susceptible mouse strains. The bacterial factors causing intestinal inflammation are poorly characterized. A large genomic pathogenicity island, HHGI1, which encodes components of a type VI secretion system (T6SS), was previously shown to contribute to the colitogenic potential of H. hepaticus. We have now characterized the T6SS components Hcp, VgrG1, VgrG2 and VgrG3, encoded on HHGI1, including the potential impact of the T6SS on intestinal inflammation in a mouse T-cell transfer model. The H. hepaticus T6SS components were expressed during the infection and secreted in a T6SS-dependent manner, when the bacteria were cultured either in the presence or in the absence of mouse intestinal epithelial cells. Mutants deficient in VgrG1 displayed a significantly lower colitogenic potential in T-cell-transferred C57BL/6 Rag2−/− mice, despite an unaltered ability to colonize mice persistently. Intestinal microbiota analyses demonstrated only minor changes in mice infected with wild-typeH. hepaticus as compared with mice infected with VgrG1-deficient isogenic bacteria. In addition, competitive assays between both wild-type and T6SS-deficient H. hepaticus, and between wild-type H. hepaticus and Campylobacter jejuni or Enterobacteriaceae species did not show an effect of the T6SS on interbacterial competitiveness. Therefore, we suggest that microbiota alterations did not play a major role in the changes of pro-inflammatory potential mediated by the T6SS. Cellular innate pro-inflammatory responses were increased by the secreted T6SS proteins VgrG1 and VgrG2. We therefore concluded that the type VI secretion component VgrG1 can modulate and specifically exacerbate the innate pro-inflammatory effect of the chronic H. hepaticus infection.


Helicobacter hepaticus is a chronic murine pathogen of the intestinal tract, which can cause liver disease, intestinal inflammation and malignancies in susceptible mice (Solnick and Schauer, 2001; Fox et al., 2011). Since its genome was fully sequenced (Suerbaum et al., 2003), the bacterium has been intensively studied as a paradigm of chronic intestinal and hepatobiliary inflammatory diseases caused by Helicobacter-like organisms in mice (Cahill et al., 1997; Kullberg et al., 2003; 2006; Chow and Mazmanian, 2010). The H. hepaticus genome is rather small but contains a large genomic island of 77 kilobases, H. hepaticus genomic island 1 (HHGI1; Boutin et al., 2005; Ge et al., 2008). The function and ecophysiological role of this island, which is assumed to have been acquired horizontally from an as yet unknown source, are not clear so far. H. hepaticus isolates that lack the island either completely or in part, and an engineered isogenic clone of the sequenced strain 3B1 (ATCC 51449) lacking part of the island caused diminished pathology in murine liver and intestinal inflammation models (Boutin et al., 2005; Ge et al., 2008). At least one complex bacterial secretion system, with components resembling type 4 (Cascales and Christie, 2003) and type VI secretion systems (T6SS; Pukatzki et al., 2006; Cascales, 2008) is encoded on the HHGI1 genomic island (Suerbaum et al., 2003). The H. hepaticus HHGI1 contains a small contiguous cluster of 11 clearly identifiable genes homologous to core T6SS components (Pukatzki et al., 2009), including vgrG1 (HH0242), hcp (HH0243), vasA (HH0245), vasB (HH0244), vipA, vipB (HH0248, HH0247), icmH, icmF (HH0251, HH0252), fha (HH0253) and vasE (HH0250). Scattered elsewhere on the island, two other, shorter orthologues of vgrG genes, vgrG2 (HH0291) and vgrG3 (HH0237) are located. Orthologues of ppkA and pppA genes (Mougous et al., 2007) were not identified in H. hepaticus.

The aim of the present study was to characterize the bacterial subcellular localization, expression patterns and potential role during the infection of the putative secreted T6SS components VgrG1, VgrG2, VgrG3 and Hcp encoded on the HHGI1 island of H. hepaticus.

We could demonstrate the expression and T6SS-dependent secretion of Hcp and VgrG1 homologues of H. hepaticus. The presence of T6SS component VgrG1 increased the colitogenic potential of H. hepaticus in a murine T-cell transfer colitis model. Cellular pro-inflammatory responses were increased by secreted T6SS proteins. We therefore conclude that type VI secretion is a means that modulates and, in specific settings, can specifically exacerbate the pro-inflammatory effect of the chronic H. hepaticus infection.


Expression of H. hepaticus T6SS components in vitro and in vivo

The first aim of this study was to investigate the expression of the T6SS genes hcp (HH0243), vgrG1 (HH0242), vgrG2 (HH0291) and vgrG3 (HH0237) encoded on the H. hepaticus HHGI1 island (Fig. 1A) and their predicted protein products (Fig. 1B and C) under various bacterial incubation conditions, in vitro and in vivo. For the in vitro conditions, the bacteria were grown in liquid culture and harvested at four different optical densities, corresponding to early, mid- and late log growth phases. Total RNAs were isolated from the bacteria, and cDNAs synthesized. For testing of in vivo transcript amounts, tissue samples from the caecum and colon of infected mice [IL-10−/− (data not shown) and Rag2−/− animals] were subjected to total RNA isolation and cDNA synthesis. Subsequently, both for in vitro and for in vivo samples, RT-PCR amplification of the cDNAs using specific bacterial primers for T6SS genes and other genes of the HHGI1 island and for control genes was performed (Fig. 2). Liquid-grown bacteria expressed mRNAs for all T6SS and island genes at all tested time points. However, for the T6SS genes hcp and vgrG1, vgrG2, vgrG3, the transcript levels oscillated at the different phases of growth, indicating a tightly regulated gene expression regarding the T6SS (Fig. 2A) which appeared to be dependent on the bacterial growth phase. When bacteria were co-incubated in direct contact with murine intestinal epithelial cells, no increase of transcript levels of the T6SS genes was noted (data not shown). This suggested that the genes are expressed under standard growth conditions and do not require the presence of cells or direct cell contact to be induced. For the in vivo samples from H. hepaticus-infected mouse tissues, transcripts were only detected for hcp and vgrG3 both in the caecum (Fig. 2B) and in the colon (not shown), where transcript amounts were lower. No transcripts were detected from the in vivo samples for vgrG2 and vgrG1. We also raised and purified antisera against selected Hcp and VgrG peptides that were designed to react specifically with Hcp and all three VgrG orthologues (Experimental procedures). The antisera were first used to detect Hcp and VgrGs in bacterial lysates of plate-grown and liquid-grown bacteria by Western blot (Fig. 3; total cell fraction ‘T’; Fig. 4A). Hcp and VgrG1 were both readily detected in all bacterial lysates in all phases of bacterial in vitro growth (Fig. 3, and not shown), which indicated a constitutive expression of these two proteins in the absence of mammalian cells. Pre-incubation of the 3B1 wild-type bacteria in contact with murine intestinal epithelial cells (mICcl2; Bens et al., 1996; Sterzenbach et al., 2007) did not lead to an increase in the expression of Hcp and VgrG1 in planktonic bacteria (Fig. S2). In order to characterize the impact of different T6SS proteins on Hcp and VgrG expression and on T6SS function, isogenic mutants in hcp, all three vgrG genes, icmF, the HHGI1 virB4 orthologue (Suerbaum et al., 2003), the regulatory gene fha (Mougous et al., 2007), and an hcpvrgG1 double mutant were generated by allelic exchange. Except for the vgrG1 and hcp mutants, which did not express VgrG1 and Hcp, respectively, and the vrgG1hcp double mutant, all other mutants expressed VgrG1 and Hcp (see total cell fraction ‘T’ in Fig. 3). A slight polarity in the hcp mutant on the expression of VgrG1 was noted (Fig. 3; Fig. S2). VgrG2 and VgrG3 proteins were not detected in H. hepaticus at all under all incubation and cell-co-incubation conditions used and in none of the mutants, although the antiserum raised against two VgrG-specific peptides reacted well with H. hepaticus VgrG2 and VgrG3 produced in Escherichia coli or expressed in eukaryotic cells (data not shown).

Figure 1.

Type VI secretion system homologues in H. hepaticus.

A. Gene organization of the H. hepaticus type VI secretion cluster on the genomic island HHGI1. Orthologous vgrG genes are indicated with the same colour. T4SS-like genes are indicated in dark grey. Genes orthologous to T6SS regulatory genes ppkA and pppA of other bacteria (Mougous et al., 2007) were not identified on the H. hepaticus HHGI1 island or the rest of the genome.

B. Model of Hcp (HH0243) protein structure predicted by PHYRE (Kelley and Sternberg, 2009) and viewed in Yasara (http://www.yasara.org).

C. Model of domain homology of the three H. hepaticus VgrG proteins to bacteriophage tail proteins (gp27 and gp5 of T4 bacteriophage and a 43-kDa-large tail protein of bacteriophage Mu) based on HHpred domain search (Soding et al., 2005). H. hepaticus VgrGs including VgrG1 appear not to possess a C-terminal extension, unlike the VgrGs of some other bacteria (Pukatzki et al., 2007; Pukatzki et al., 2009; see also alignment of Fig. S1 in Supporting information).

Figure 2.

Expression of selected HHGI1 genes during H. hepaticus growth in liquid medium (A) or during mouse infection (B). Transcripts were detected by RT-PCR using primer pairs listed in Table 3. Abundance of H. hepaticus-specific 16S rRNA was tested as a control for RNA amounts. Technical PCR controls: − = H2O, + = H. hepaticus 3B1 genomic DNA.

A. RT-PCR of cDNA samples reverse-transcribed using H. hepaticus 3B1 total RNA samples after growth in serum-free liquid culture (OD600 indicated). PCR cycle numbers are indicated.

B. RT-PCR of cDNA samples reverse-transcribed from total RNAs isolated from infected 129 Rag2−/− mice caeca. Duration of infection and animal identity numbers are indicated. PC reactions were run for 40 repetitive cycles.

Figure 3.

Hcp and VgrG1 localization in H. hepaticus subcellular fractions. Plate-grown bacteria were resuspended in PBS, briefly homogenized by ultrasonication and further disrupted with a FastPrep instrument. The total lysates were then repeatedly centrifuged to separate the fractions into soluble and insoluble (membrane-enriched) fractions. Bacterial subcellular fractions were further analysed by immunoblot using affinity-purified α-Hcp and α-VgrG rabbit antisera. As a fractionation control, α-FlhA-antiserum reacting with the inner membrane-anchored flagellar export channel protein FlhA was used (lower panel; only present in insoluble fractions). T = total cell lysate, IS = insoluble fraction, Sol = soluble fraction. Ten micrograms of protein per sample was used for SDS-PAGE. H. hepaticus strains: 3B1 = wild type; T6SS mutants: vgrG1hcp (HH0242-243),icmF (HH0252),vgrG1 (HH0242), vgrG2 (HH0291), vgrG3 (HH0237), hcp (HH0243),fha (HH0253); other mutants: HhPAId1 = 19 HHGI1 genes (HH0250-HH0268) deleted, including fha and icmF; virB4 (HH0260).

Figure 4.

Hcp and VgrG1 detection in supernatants and pellets of H. hepaticus wt strain and selected HHGI1 mutants.

A. Detection of secreted proteins from liquid-grown bacteria. Bacteria were inoculated from overnight pre-culture into fresh BHI without horse serum and grown to an OD600 of ∼ 0.7 to 0.8, and samples containing equivalent bacterial numbers were harvested. Sterile-filtered supernatants were TCA-precipitated and dissolved in equal volumes of SDS loading buffer; pelleted cells were resuspended in equal amounts of PBS mixed with SDS loading buffer, upper panel – bacterial pellets; lower panel – culture supernatants. Molecular mass markers (kDa) are indicated on the right. The bottom panel shows a fractionation control for the secreted fraction, using α-aconitase B (AcnB) antiserum, which only recognizes cell-bound AcnB.

B. Detection of secreted proteins in supernatants of H. hepaticus co-incubated with cell cultures at an moi of 100 for 2.5–3 h. Serum-free media appropriate for each cell-line were used during the experiment. The supernatants were sterile-filtered, concentrated by TCA precipitation and volumes corresponding to one-fifth of the original medium volumes were used for SDS-PAGE and Western blot analysis. Tested cell lines: mICcl2 – murine intestinal epithelial cells (final numbers: 3.5 × 105 cells ml−1), J774 – murine macrophages (final numbers: 4 × 105 cells ml−1), THP-1 – human monocytes (final numbers: 3 × 105 cells ml−1).

Immunodetection by affinity-purified IgG raised against Hcp peptide (predicted MW = 19 kDa) and affinity-purified anti-VgrG IgG. Only VgrG1 (predicted MW = 103.7 kDa) was specifically detected in H. hepaticus samples. H. hepaticus strain designation as in legend to Fig. 3. 3B1 (T) = control total cell lysate of wild-type H. hepaticus grown on plate.

Hcp and VgrG1 are soluble and membrane-associated components, respectively, of the H. hepaticus T6SS

In several other bacteria which possess T6SS, such as Vibrio cholerae and Pseudomonas aeruginosa, Hcp and VgrG orthologues have been reported to be secreted components of their respective T6SS (Wu et al., 2008; Ma et al., 2009b; Hachani et al., 2011). In some instances, transport of these components into eukaryotic target cells has been demonstrated, and specific modulatory effects of these cell-translocated components towards, e.g. the eukaryotic cytoskeleton, have been characterized (Pukatzki et al., 2007; Ma et al., 2009a; Ma and Mekalanos, 2010; Russell et al., 2011). The proposed structural arrangement of the Hcp and VgrG components within the T6SS apparatus, and their structural relationship with phage tail proteins (Kanamaru, 2009) also suggested that they can be localized on the bacterial surface. The bacterial subcellular localization of the H. hepaticus T6SS homologues has not been studied in detail so far. To test the hypothesis that these specific T6SS components are localized on the bacterial surface in H. hepaticus, we prepared bacterial fractions, in order to separate soluble (cytoplasmic and periplasmic), insoluble (membrane-enriched) and surface-associated bacterial proteins (Experimental procedures) (Fig. 3). We also investigated whether the expression and localization of the Hcp and VgrG components was altered in H. hepaticus isogenic mutants in defined T6SS components that were described to be essential for T6SS in other bacteria (Hachani et al., 2011). H. hepaticus Hcp was detected mainly in the soluble fraction in wild-type bacteria. VgrG1 was detected in both the soluble and insoluble fractions in wild-type bacteria (Fig. 3). In the hcp and, to a lesser extent, in the fha mutant, the overall amounts of VgrG1 were reduced (Fig. 3). Insertion mutations in other T6SS-associated genes had no detectable impact on the expression of Hcp and VgrG1 (Fig. 3). Both VgrG1 and Hcp were not detected on the bacterial surface by a mild shearing procedure (not shown).

Hcp and VgrG1 are secreted by H. hepaticus, and secretion depends on an active T6SS

In order to test whether Hcp and VgrG are actively secreted by H. hepaticus, as has been demonstrated for the orthologues from other bacteria (Mougous et al., 2006; Wu et al., 2008), culture supernatants of bacteria grown in liquid culture in the absence of serum were investigated. The presence of horse serum in the liquid growth medium increased the fraction of lysed bacteria in the samples, indicated by the increased detection of cytoplasmic proteins (such as aconitase B) in the supernatants (not shown), and was therefore not suitable for this analysis. The bacterial supernatants from equal numbers of bacteria collected during different growth phases (Fig. 4A) were concentrated approximately 10- to 20-fold by precipitation with TCA and separated on SDS-PAGE. Both Hcp and VgrG1 were readily detectable by Western blotting in bacterial supernatants, while cytoplasmic control proteins were not detected under the same experimental conditions (Fig. 4A). The amount of the two T6SS-specific proteins in the supernatants was dependent on the growth phases (differences observed in different mutants, an increase starting at mid-log growth was observed; data not shown).

We next asked if secretion of Hcp and VgrG1 was dependent upon the presence of an active T6SS. Inactivation of icmF (T6SS channel component; Ma et al., 2009b) and the proposed secretion regulatory gene fha (Mougous et al., 2007) largely abolished the secretion of Hcp and VgrG1 into the bacterial supernatants (Fig. 4A), while the proteins were readily detectable in all bacterial lysates prepared from the mutants. VgrG1 and Hcp secretion was mutually dependent on the respective other protein being expressed. VgrG2 and VgrG3 were not required for Hcp or VgrG1 secretion (Fig. 4A). VgrG1 was secreted in co-culture with mouse intestinal epithelial cells (mICcl2) while Hcp was not (Fig. 4B).

Reduced intestinal pathology in mice infected with mutants deficient in the secreted component VgrG1 of the H. hepaticus T6SS

Our previous results indicated a role of the H. hepaticus genomic island HHGI1 in mouse infection models of hepatic and intestinal inflammation (Boutin et al., 2005; Ge et al., 2008). Furthermore, the colitogenic potential of H. hepaticus in various mouse models is very well established (Kullberg et al., 1998; Nell et al., 2010). However, in these previous studies, a specific role of the T6SS in the intestinal inflammatory process and intestinal pathologies had not been addressed. We therefore investigated the hypothesis that the H. hepaticus T6SS has an impact on its colitogenic potential. We tested H. hepaticus wild type and isogenic T6SS mutants (in icmF or vgrG1) in a murine T-cell transfer model of bacterial typhlocolitis (Kullberg et al., 2002). Six- to 8-week-old C57BL/6 Rag2−/− mice were first inoculated with three subsequent doses of the respective H. hepaticus strains. One week after the start of the infection, isolated enriched CD4+ T-cell preparations from H. hepaticus-naïve mice were transferred into these recipient mice, which do not produce T- or B-cells. H. hepaticus infections with wild type and all isogenic mutants persisted throughout the whole duration of the experiments as assessed by regular faecal monitoring. Control groups infected with the same bacterial strains were mock-transferred using PBS alone. Five weeks after the T-cell transfer, all groups of mice were sacrificed and gross morphological changes were visually evaluated. Bacteria were enumerated from both the caecum and the colon of all mice (Fig. 5A), and fixed sections of caecum and different locations in the colon of all mice were scored for histopathological lesions (Fig. 5B and C). During the sections, gross pathological differences (not shown) were observed between the groups: the wild type-infected, T-cell-transferred group had clear macroscopic pathology with thickened colon walls and loss of flexibility of the colon; in addition, the caecum sizes were greatly reduced in this group (not shown). The icmF mutant-infected T-cell-transferred group also showed clear pathology; however, the macroscopic observations were less severe in comparison with animals of the wild-type group; slightly thickened colon walls and shrunken caeca were observed in all animals. The vgrG1 mutant-infected T-cell-transferred group showed only slight or no macroscopically visible pathological aspects of caecum and colon. The control groups injected with only PBS also showed only slight or no macroscopical changes.

Figure 5.

Adoptive T-cell transfer colitis model in C57BL/6 Rag2−/− mice demonstrates colitogenic potential of the T6SS component VgrG1. Mice were infected with H. hepaticus 3B1 wild type and isogenic icmF and vgrG1 mutants, followed by intraperitoneal injection of CD4+ T-cells from naïve mice or mock transfer. Panel (A) shows the colony-forming units re-isolated at the time point of sectioning from the caecum of all mice; (B) summarizes the overall histopathology scores (Burich et al., 2001) of all mouse groups added up from all different tested locations of the mouse intestinal tract; (C) shows the results of histopathological scoring separately for caecum, proximal, middle and distal colon for the infected, T-cell-transferred animal groups only. Controls in (C): mock-infected CD4+ T-cell-transferred mouse groups. Statistical significance between different experimental groups (Mann–Whitney-U-test) is indicated by asterisks (*P < 0.05; **P < 0.01). BHI – mock-infected animals were gavaged with BHI medium only. PBS – mock-transferred animals were injected intraperitoneally with PBS only. CD4+ indicates animal groups which were intraperitoneally injected with naïve CD4+ T-cell-enriched fractions.

The numbers of bacteria in the caecum were quite similar in all groups of mice infected with the different strains (Fig. 5A); bacterial cfu in the caecum were slightly increased for the vgrG1 and icmF mutant groups (significant increase in one of two experiments for vgrG1 mutant-infected animals) in comparison with the wild type. This increase might suggest that, in wild-type bacteria, the T6SS contributes to eliciting an effective immune response against H. hepaticus in order to control colonization levels. In terms of histological assessment of tissue pathology, the T-cell-transferred wild type-infected mice and icmF mutant-infected mice had the highest pathology scores (Fig. 5B and C, Fig. S3), which were significantly different from the non-transferred animals (Fig. 5B). For T-cell-transferred animal groups, icmF-infected animals had slightly lower histopathology scores (not significant; Fig. 5B and C, Fig. S3) in comparison with the wild type-infected group. PBS-transferred control mice of all infected groups showed low pathology scores (Fig. 5). Surprisingly, in comparison with wild type-infected animals after T-cell transfer, the overall mean histopathology scores were significantly reduced both in the caecum and in all colon regions for vgrG1 mutant-infected animal groups which had undergone CD4+ T-cell reconstitution (Fig. 5C, Fig. S3). Comparing the data between T-cell-transferred icmF and vgrG1 mutants, the results indicated that not only the presence or activity of the T6SS per se, but specifically the presence of VgrG1, when expressed from HHGI1, contributed strongly to the extent of local innate inflammation caused by the bacterial infection.

Determining subcellular localization, and cytotoxic and pro-inflammatory effects of secreted components of the H. hepaticus T6SS expressed in mammalian cells

Subsequent to the above results, we therefore hypothesized that T6SS components, and specifically VgrG1, might have cell-stimulating and/or cytotoxic effects on host cells. Since we were unable to detect the direct translocation of Hcp or VgrG proteins from H. hepaticus into mammalian cells in co-incubation experiments in vitro (Fig. S2), probably due to very low and strictly localized expression of the proteins in the bacteria, we took the alternative approach of expressing H. hepaticus Hcp and its three VgrG orthologues in mammalian cells (HEK293-T, THP-1, mICcl2 and J774). The bacterial T6SS genes were cloned into mammalian expression vectors able to produce fusion proteins with a C-terminal V5 tag, and cells were transiently transfected with these plasmids in order to induce protein expression. Expression of the proteins was detected by immunofluorescent labelling (Fig. 6) or by Western immunoblotting (Fig. 7A) using anti-V5 epitope tag antibodies or the specific purified antisera against Hcp and VgrG. VgrG3 and Hcp were not detected in cellular lysates of pEF6-transfected cells by Western blot with either antibodies, while VgrG1 and VgrG2 were readily detectable in pEF6 expression plasmid-transfected HEK293 cells, both at 24 h and at 48 h post transfection (Fig. 7A). Using pTRE3G inducible expression plasmids in HEK293 cells, Hcp and VgrG3 were detectable in transfected HEK cell lysates by Western blot at 20 h after induction of the tet promoter (data not shown), but VgrG1 and VgrG2 were not detectable at 6 h or 20 h after the induction. Furthermore, the expression and subcellular localization of Hcp and VgrGs was analysed in transfected cells using specific immunofluorescent labelling and confocal microscopy. In HEK293 cells, VgrG1 localized diffusely to the cytoplasm, whereas VgrG2 appeared to localize to cell membrane patches (data not shown). VgrG3 was detectable in very few single cells (less than 1% in a transfection experiment), while Hcp was not detected in HEK cells using immunofluorescent labelling. In transfected mouse intestinal mICcl2 cells, all four proteins were detectable using immunofluorescent labelling (approximately 1% of cells in each transfection experiment); however, cell morphology was frequently changed reminiscent of proapoptotic alterations (Fig. 6). In human and mouse macrophage-like cell lines after transfection, expression of the proteins was not detectable.

Figure 6.

Expression and immunofluorescence detection of H. hepaticus T6SS proteins VgrG and Hcp in mICcl2 mouse small intestinal epithelial cells. Cells were transiently transfected using pEF6-derived expression plasmids (Table 1). Expressed proteins carrying C-terminal V5 tags (Hcp, VgrG1, VgrG2, VgrG3) were detected by immunofluorescent labelling using α-V5 antibody (mouse monoclonal. Invitrogen; dilution 1:1000) and Alexa488-labelled secondary antibody. peGFP_C3 expression plasmid (right panels) was used as a transfection technical control.

Figure 7.

Increase of innate pro-inflammatory effect by H. hepaticus VgrG1 and VgrG2 upon expression in TLR4-competent HEK293 cells.

A. Expression of VgrG1 (white arrow) and VgrG2 (grey arrow) from pEF6-derived plasmids transfected into HEK293 cells was detected in Western blot using α-V5 monoclonal antibody (mouse, Invitrogen; dilution 1:5000), at 48 h post transfection. Black arrow indicates coexpression of mouse MD2 protein (mMD2), which was coexpressed after co-transfection of mTLR4 and mMD2 expression plasmids together with H. hepaticus T6SS component expression plasmids. Thirty micrograms of total cell protein was loaded for each sample.

B. IL-8 release into cell culture supernatants in the same experimental set-up, where H. hepaticus T6SS components VgrG1 or VgrG2 were coexpressed with mTLR4 and mMD2 in HEK293 cells. Eco-LPS: additional treatment with ultrapure E. coli LPS (TLR4 ligand) 4 h before harvest.

Table 1. Lactate dehydrogenase (LDH) assay of T6SS plasmid-transfected HEK293 cells
Plasmid used for transfectionLoss LDHC compared with lipofectamine control (% of 3.55)Mean LDHSMean LDHCMean cytotoxicity (%)a
  1. aMean cytotoxicity of each plasmid-transfected sample was calculated for each sample as the ratio of mean cellular LDH content of transfected cells against the mean cellular total LDH of lipofectamine-treated cells.
  2. Calculated final values (mean values from two biological replicates that were performed in technical duplicates) for cytotoxicity compared with lipofectamine ‘mock'-treated cells (lipofectamine background was subtracted).
  3. LDHS, LDH in supernatants; LDHC, total cellular LDH content.

The induction of cytotoxic, proapoptotic and pro-inflammatory effects in the transfected human or mouse cells were determined by LDH assay, caspase-1 inflammasome activation testing (Franchi et al., 2012) and Caspase-3/7 assay respectively. The LDH assay revealed slight cytotoxic effects for all T6SS proteins, which were strongest for vgrG3 plasmid-transfected cells (Table 1 ). Caspase-3/7 assay revealed some effector caspase activation for all plasmid-transfected cells (tested in human and mouse cells) including the empty plasmid controls (not shown). Inflammasome and caspase-1 activation was detected most strongly in transfected human THP-1 macrophages, but was observed with any control plasmid transfection, independently of H. hepaticus T6SS proteins (not shown). When expressed in transiently transfected HEK293 cells, VgrG1 and VgrG2 proteins increased pro-inflammatory signalling, visible by significantly enhanced IL-8 release when mouse TLR4 complex was coexpressed and the cells were activated in parallel by the TLR4 agonist E. coli LPS (Fig. 7B).

Effects of H. hepaticus carrying HHGI1 and the T6SS on H. hepaticus strains not possessing the island and on other intestinal bacteria

Interbacterial effects have been proposed to be related to some bacterial T6SS (MacIntyre et al., 2010; Schwarz et al., 2010b). Since the habitat of H. hepaticus, the mouse caecum and colon, is a niche containing numerous different bacterial species, interbacterial interactions would be assumed to occur frequently. We tested a potential role of the H. hepaticus T6SS in interbacterial interaction or competition in different combinations of two bacterial strains. H. hepaticus wild type was co-incubated on a solid medium with H. hepaticus HHPAId1 mutant (Ge et al., 2008), Campylobacter jejuni wild type (strain NCTC 11168; without a T6SS), three different E. coli wild-type strains, and Citrobacter freundii, another member of the Enterobacteriaceae. No competitive effect of the functional T6SS in the H. hepaticus wild type was observed in any of the combinations, when compared with the counts of singly grown strains (Fig. 8). Intestinal microbiota analyses by 16S rDNA amplicon sequencing were performed for the T-cell-transferred mice infected with either H. hepaticus wild type or vgrG1 mutant, and only minor changes in the composition of the microbiota were observed between the two groups (data not shown).

Figure 8.

Interbacterial interaction of H. hepaticus 3B1 wild type or the HHPAId1 mutant with other bacterial species. Different bacterial species as indicated (Enterobacteriaceae) were incubated alone or competitively co-incubated with H. hepaticus [methods; H. hepaticus 3B1 wild type (A) or the isogenic HHPAId1 (island deletion mutant, B)] and enumerated before (time zero control; white bar), after incubation alone (growth control, grey bars), or after co-incubation (competition experiment, dotted bars). No increased competitive effect was observed for the H. hepaticus wild type (A) in comparison with the HHPAId1 mutant (B).


The intestinal persistent pathobiont H. hepaticus is one of the numerous bacteria harbouring the most recently described complex secretion system of Gram-negative bacteria, a T6SS (Mougous et al., 2006; Pukatzki et al., 2007; Cascales and Cambillau, 2012). In comparison with other bacterial T6SS characterized to date, most so far recognized core proteins of the secretion system can be identified, which are located in a cluster of approximately 13 genes on the large pathogenicity island HHGI1 (Suerbaum et al., 2003; Ge et al., 2008). We have focused in this study on the potentially secreted substrates or proposed extracellular tip proteins of this system, Hcp and the three VgrG orthologues, on their subcellular localization in the bacteria, and their potential role as virulence factors during the infection of mice.

Transcripts of all hcp and vgrG genes were detected upon growth of the bacteria under in vitro conditions. vgrG1 and vgrG2 transcript amounts appeared to be exquisitely sensitive to growth phase-dependent fluctuations, while this appeared to be less the case for hcp and vgrG3. In co-culture of bacteria with cells, hcp and all vgrG transcripts were also readily detectable, although cell-adherent bacteria appeared to produce less vgrG1 and vgrG2 transcript (own unpublished data). Interestingly, during mouse infections, only hcp and vgrG3 transcripts were present in detectable amounts, and vgrG3 was only detected at later time points during the infection, whereas the other two vgrG transcripts were not detected in vivo in the mouse caecum at all. This differential detection could be due to detection limitations in the caecum material or to differential regulation in vivo as opposed to laboratory culture. It is tempting to speculate from the current knowledge (Kanamaru, 2009) about the structural contributions of Hcp and VgrG proteins to the T6SS (Cascales, 2008) that Hcp acting as the major structural tube protein of the system would require the highest expression levels, while the VgrG orthologues as proposed tip proteins or secreted effectors of the system would need lower expression levels. The requirement of low protein amounts of the VgrG orthologues in vivo could explain the low transcript levels in vivo. It is not known whether the growth characteristics of H. hepaticus in vitro truthfully reflect the growth and expression levels in the mouse caecum. The bacteria might be dividing more slowly, and metabolic state and expression patterns of the genes might differ from the in vitro settings. We also cannot exclude technical issues as a reason for the lack of detection of vgrG2 and vgrG1 transcripts, since whole mouse tissue served as a starting material for the isolation of combined host and bacterial RNA, likely reducing sensitivity.

The three VgrG orthologues in H. hepaticus are very different in length, as VgrG2 and VgrG3 lack the N-terminal gp27-like domain of the full-length VgrG1. Based on previously reported evolutionary aspects of proposed tip proteins in other secretion systems (Nystedt et al., 2008) and the variable occurrence of single or multiple VgrG orthologues in other bacteria including other Helicobacter species (http://www.ncbi.nlm.nih.gov/), it is likely that the VgrG orthologues might have arisen by gene duplications, and that further genetic differentiations subsequently occurred according to their, yet unknown, functional diversification.

Hcp and VgrG1 proteins were found to be actively secreted by H. hepaticus in the present study, depending on a functional T6SS. Isogenic mutants deficient in the energizing component IcmF (HH0252) (Ma et al., 2009b) and the post-translational regulator Fha (HH0253) (Mougous et al., 2007) did not secrete VgrG and Hcp, in contrast to the wild type. Concerning their own role in secretion, secretion of Hcp and VgrG1 were mutually dependent on each other. This is consistent with reports from T6SS of other bacteria (Pukatzki et al., 2009; Hachani et al., 2011). Hcp or VgrG1 transport were independent of HH0260, a homologue of VirB4, an ATPase-like protein involved in DNA transfer in T4SS (Cascales and Christie, 2004). Nevertheless, the VirB4-like protein may have another role in the H. hepaticus T6SS apart from the protein substrate transport. Interestingly, by examining in vitro transcript levels, we observed a similar oscillating pattern for virD4 transcript as for the transcripts of designated T6SS components vgrG1 and icmF during the different phases of growth, indicating common regulation. The T4SS-like proteins HH0260 (VirB4-like) and HH0275 (VirD4-like) may be remnants of a T4SS cluster or could have been acquired individually for a specific function in the H. hepaticus T6SS.

An attractive hypothesis for a role of T6SS in bacteria chronically colonizing the densely populated intestinal tract of mammals would be to gain an edge in intestinal interbacterial competition (MacIntyre et al., 2010; Schwarz et al., 2010a; Russell et al., 2011). The present study did not yield evidence for a role of the H. hepaticus wild-type T6SS in interbacterial interaction, neither in competitive co-cultivation with a H. hepaticus mutant lacking a functional T6SS nor in competitive co-cultivation with other tested bacteria. Previous reports have suggested that H. hepaticus successfully competes with the resident microbiota, rapidly becoming the dominant component of the intestinal microbiota after experimental infection (Kuehl et al., 2005). However, these early studies were not based on detailed quantitative microbiome analyses and further investigation of the interaction between H. hepaticus and the resident microbiota seems warranted. Bacterial T6SS might also play a more general role in response to environmental conditions, which could be important as well in the intestinal tract. In the light of a recent publication which uncovered a role of a C. jejuni T6SS in resistance to bile acid (deoxycholic acid; Lertpiriyapong et al., 2012) we have also tested an effect of bile and bile acid on H. hepaticus and its isogenic T6SS mutants. We did not observe any difference in H. hepaticus resistance to native bile and deoxycholic acid with regard to the presence or absence of a functional T6SS (own unpublished experiments). Further questions how a functional H. hepaticus T6SS may influence the response to or be regulated by environmental cues should be addressed in future experiments.

We previously reported that a H. hepaticus mutant lacking 19 HHGI1 genes was attenuated in its ability to induce colitis (Ge et al., 2008). To dissect the mechanisms underlying this attenuation further, we have investigated a role of the T6SS proteins IcmF and VgrG1 in pathogenesis and host interaction using a mouse T-cell transfer colitis model. vgrG2 and vgrG3 mutants could not be tested in mouse infections, since all clones of these mutants were reproducibly non-motile (own unpublished data and Fig. S4), and the lack of motility is prohibitive for mouse colonization (Sterzenbach et al., 2008). Inactivation of vgrG1 and icmF did not reduce the colonization potential of the mutants, since bacterial numbers were even slightly increased for the vgrG1 and icmF mutant groups in comparison with the wild type. This increase might suggest that, for wild-type bacteria, the fully functional T6SS contributes to an effective innate immune response of mice against H. hepaticus which could aid the host to control colonization levels. Surprisingly, in T-cell-transferred animals infected with isogenic H. hepaticus icmF mutants deficient in the secretory function of the T6SS, colitis scores were only slightly reduced (P > 0.05). On the other hand, infection with mutants deficient in the proposed T6SS tip component VgrG1 led to significantly reduced colitis scores (P < 0.01), including the diminished potential to induce pro-inflammatory cell influx and histopathological lesions in T-cell-transferred animals (Fig. S3). Since icmF mutants still expressed and even slightly secreted VgrG1 in our hands, a danger signal by VgrG1 might still be provided and sensed by host cells, which could be one possible explanation for the intermediate phenotype of the icmF mutant in our colitis model.

T-cell transfer experiments in Rag−/− mice now performed by us and previously by others (Maloy et al., 2003; Boulard et al., 2012) suggested that H. hepaticus and its components, in this case the T6SS protein VgrG1, can induce colitis by innate immune mechanisms. In order to elucidate a mechanism for a proposed innate pro-inflammatory effect of VrgG1 on cells, we have tested its cytotoxic/cytolytic activity and the hypothesis that it would lead to inflammasome activation as has been proposed previously for other external or tip proteins of bacterial type III secretion systems, which have been shown to act as bacterial Microbe-Associated Molecular Patterns (MAMP) on intracellular pattern recognition receptors (PRR) (Lightfield et al., 2008; Zhao et al., 2011). In the transfected cell systems we used for the testing, VgrG1 and VgrG2 expression was detectable by Western blot, while the abundance of VgrG3 and Hcp proteins was below detection levels. No clear cytolytic effects, inflammasome activation or proapoptotic caspase-3/7 activation appeared to occur when VgrG1 or VgrG2 proteins were detectably expressed in cells. Caspase-1 cleavage as a sign of inflammasome activation was induced in cells transfected with VgrG expression plasmids, but also occurred in the controls. Interestingly, we observed an increase of innate pro-inflammatory signalling and chemokine secretion induced by VgrG1 and VgrG2 expression in HEK293 cells, when mouse TLR4 complex was coexpressed with those proteins and cells were pre-activated by LPS (Fig. 7). This result supports our hypothesis that VgrG proteins may contribute to innate pro-inflammatory signalling in cells, possibly by acting as a danger signal on innate receptors. The ambiguous results obtained in some assays detecting cell activating effects might be due to the very low numbers of expression-positive cells (counted as specific transfection efficiency) for the T6SS proteins in both inducible and non-inducible expression systems after transfection. The reason for the low expression levels of VgrG or Hcp proteins is currently unclear. It will be interesting to test again for pro-inflammatory or inflammasome-inducing effects if stable cell lines that express the proteins can be obtained.

Recently, the group of Mazmanian (Chow and Mazmanian, 2010) also characterized an impact of H. hepaticus T6SS function and components in a T-cell transfer colitis model. In their model, the outcome of infections using mutants deficient in the Hcp and IcmF T6SS components was compared with the isogenic wild-type infection. They detected that pro-inflammatory cytokine transcripts were locally increased by the two T6SS mutants in comparison with the wild type, and they concluded that the presence and activity of the T6SS appeared to inhibit innate pro-inflammatory responses. The histopathology scores in their hands were not significantly elevated in any group or different between the groups. In our infection model, we determined highly significantly elevated pathology scores in wild type-infected T-cell-transferred animals in comparison with non-transferred controls. We also demonstrated a significant reduction in both macroscopical ‘gross’ pathology and microscopical assessment of histopathology scores of the vgrG1 mutant-infected transferred animals in comparison with the isogenic wild type. Interestingly, for the icmF mutant, we also obtained slightly lower gross pathology and histopathology scores in comparison with the wild type, in apparent contrast to the results of Chow and Mazmanian (2010). It is difficult to exactly compare the two experimental settings, since in our hands, bacterial counts in the caecum and colon did not differ significantly between icmF mutant and wild type, while Mazmanian and colleagues determined highly significantly increased tissue counts for both the icmF and the hcp mutants in comparison with the wild type. Since they did not report testing vgrG1 mutants in the T-cell transfer model, it is not possible to comparatively discuss the striking reduction in pathology that we obtained for this specific single gene mutant, which still possesses a functional T6SS, apart from the loss of VgrG1 production and Hcp secretion. It is possible that the anti-inflammatory potential of a functional T6SS in intestinal pathology, as suggested by Mazmanian and colleagues, which could be due to as yet unknown T6SS components or effectors, is distinct from an innate pro-inflammatory danger-sensing effect suggested by us to be specifically induced by a T6SS tip component, VgrG1. The hypothesis that two distinct and separate effects occur is also in keeping with the intermediate histopathology scores that we obtained with the icmF mutant. It is also not excluded that, in different animal facilities, as yet undisclosed effects on the intestinal microbiota in the context of the H. hepaticus infection may contribute to differential results with regard to the colitogenic impact of the T6SS and its components. Our present results suggest that an increase of innate immune pro-inflammatory signalling triggered by the secreted T6SS component VgrG1 might lead to enhanced colitogenic effects in H. hepaticus-infected mice.

Experimental procedures

Bacterial strains and growth conditions

Helicobacter hepaticus ATCC 51449 was cultured at 37°C under microaerobic conditions (10% CO2, 10% H2, 80% N2) in vented jars or in a Scholzen incubator on blood agar plates (Columbia agar base II; Oxoid, Wesel, Germany), supplemented with 10% horse blood and antibiotics (10 mg l−1 vancomycin, 2500 U l−1 polymyxin B, 5 mg l−1 trimethoprim, 4 mg l−1 amphotericin B), or in the complex liquid medium BHI (brain heart infusion broth, Oxoid or Difco) with the addition of 2.5 g l−1 yeast extract (Merck) and the standard antibiotics supplement detailed above. The H. hepaticus mutant strains were grown on plates with standard antibiotics supplement and 20 mg l−1 chloramphenicol (cm); in liquid culture, the same medium was used for the mutants as for the wild-type bacteria. C. jejuni wild-type strain NCTC 11168 for the competitive growth experiments were grown exactly as the H. hepaticus strains. E. coli DH5α was used for the cloning and mutagenesis of H. hepaticus genes, and for some of the bacterial co-incubation assays. E. coli was grown in Luria–Bertani (LB) broth (Difco™ LB Agar, Lennox; BD) or on LB plates containing 1.5% Bacto agar and supplemented with selective antibiotics (ampicillin 200 mg l−1; chloramphenicol 20 mg l−1) as required. E. coli and C. freundii strains (see Table 2) to be tested in competitive assays with H. hepaticus, were pre-cultivated in aerobic atmosphere on horse blood agar plates without antibiotics. Bacterial strains are summarized in Table 2. Bacterial competitive growth assays for H. hepaticus wild type in combination with isogenic H. hepaticus mutants or with other bacterial species were performed as described in the supplementary methods.

Table 2. Bacterial strains and plasmids
Strain/mutant nameMutated ORFAnnotationResistance markerOrigin (reference)
H. hepaticus 3B1ATCC 51449Suerbaum et al. (2003)
H. h. HHPAId1HH250–HH268Part of the HHGI1 islandcmRGe et al. (2008)
H. h. vgrG1hcpHH0242–HH0243Homologues to T6SS componentscmRThis study
H. h. hcpHH0243Homologue to T6SS componentcmRThis study
H. h. vgrG1HH0242Homologue to T6SS componentcmRThis study
H. h. vgrG2HH0291Homologue to T6SS componentcmRThis study
H. h. vgrG3HH0237Homologue to T6SS componentcmRThis study
H. h. icmFHH0252Homologue to T6SS/T4SS componentcmRThis study
H. h. fhaHH0253Homologue to T6SS componentcmRThis study
H. h. virB4HH0260Homologue to T4SS componentcmRThis study
Campylobacter jejuniNCTC 11168 wild typeNCTC
Citrobacter freundiiRing 311, wild type (from human peritonitis patient)Isolated in Switzerland, 2008
Escherichia coliATCC 25922 wild typeATCC
E. coli MC1061F−, araD139 Δ(ara-leu)7696 galE15 galK16 Δ(lac)X74 rpsL (Strr) hsdR2 (rKm+) mcrA mcrB1Casadaban and Cohen (1980)
E. coli DH5αF−, endA1, recA1, hsdR17, Δ(lacZYA-argF)U169, thi-1, supE44, gyrA96, relA1Hanahan (1983)
PlasmidsCloned ORFAnnotationResistance markerOrigin
pCJ1104HH0243H. h. hcp in pEF6-V5AmpR (bla)This study
pCJ1105HH0242H. h. vgrG1 in pEF6-V5AmpR (bla)This study
pCJ1106HH0291H. h. vgrG2 in pEF6-V5AmpR (bla)This study
pCJ1107HH0237H. h. vgrG3 in pEF6-V5AmpR (bla)This study
pSUS2111HH0252H. h. icmF:O:cm in pUC19cmRThis study
pSUS2310HH0260H. h. virB4:O:cm in pUC19cmRThis study

Motility assay

Helicobacter hepaticus and C. jejuni motility was tested in medium containing 0.3% (w/v) agar (Noble agar, BD), 2.8% (w/v) Brucella broth (BD), 10% (v/v) heat-inactivated horse serum (Gibco) and the standard antibiotic supplement (see above), which was cast in six-well cell culture plates. Bacteria grown on standard blood agar plates overnight were inoculated into the centre of the motility agar disc with a toothpick or pipette tip. The motility plates were then cultivated in the Scholzen incubator for 4–6 days (H. hepaticus) or 1–2 days (C. jejuni).

Standard protein and DNA techniques

All DNA methods were performed according to standard protocols (Sambrook and Russell, 2004), using enzymes manufactured by New England Biolabs, Roche or Invitrogen. Plasmids were sequenced using Sanger sequencing technology. SDS-polyacrylamide gel electrophoresis, staining and Western blotting was performed according to standard methods (Laemmli, 1970; Towbin et al., 1979; Sterzenbach et al., 2008). As secondary antisera, goat anti-mouse or goat anti-rabbit conjugated to horseradish peroxidase (HRP) (Jackson Immunoresearch Laboratories) were used in 5% skim milk and 0.1% Tween in TBS (TBS-T). HRP signal was then detected by incubation with chemiluminescent substrate Pico SuperSignalWest (Pierce ThermoScientific).

Construction of H. hepaticus isogenic mutants

Allelic exchange insertion mutants in H. hepaticus wild-type strain 3B1 (ATCC 51449) in several genes coding for T6SS components in the HHGI1 genomic island were constructed, including the hcp (HH0243), vgrG1 (HH0242), vgrG2 (HH0291), vgrG3 (HH0237), virB4 (HH0260), fha (HH0253) and icmF (HH0252) genes. The icmF and virB4 chromosomal mutants were generated using plasmids containing the respective genes with the insertion of a chloramphenicol (cm) resistance cassette (Table 2; cloning primers: Table 3 ). The island mutant HhPAId1 contains a large deletion removing 19 chromosomal genes in HHGI1 (HH0250-HH0268) and was described previously; for details of HHPAId1 mutant construction see Ge et al. (2008).

Table 3. Primer pairs used for PCR, cloning and DNA sequence determination
Amplified genePrimer nameSequence (5′–3′)Annealing temperature (°C)Restriction site

The fha, hcp, vgrG1, hcp-vgrG1 double mutant, fha, vgrG2 and vgrG3 mutants in H. hepaticus were constructed using a PCR-based method (overview of cloning primers in Table 3) as further outlined in the supplementary methods. Polarity of the allelic exchange insertion mutants on the transcription of neighbouring genes was assessed by RT-PCR. No polarity of the mutants was revealed, with the exception of a slight polarity of the insertion mutation in hcp (HH0243) on the neighbouring gene vgrG1 (HH0242) (not shown). vgrG1 mutants had no polar effect on transcripts of downstream genes HH0241 and HH0240 of unknown function.

Generation of antisera against H. hepaticus Hcp and VgrG1

The antisera against selected H. hepaticus proteins (Hcp, VgrG1) were produced by intradermally injecting rabbits with predicted immunogenic peptides (prediction of B-cell immunogenic peptides: Protean software). Based on the predicted properties of the Hcp and VgrG1 amino acid sequences, two immunogenic peptides were synthesized for each protein of interest (Hcp-specific peptides: Hcp peptide No. 1: C-PNAQDTSNNNKTE, Hcp peptide No. 2: C-TVPTDTQSGQPSGQ; VgrG-specific peptides: VgrG peptide No. 1: C-QTIHNNKESQVEGTYN, VgrG peptide No. 2: C-EYVGEDKEVEIGGN; Biosyntan, Berlin, Germany). Both of the two VgrG-specific peptides (located at the C-terminus of the protein) were predicted to be specific for all three H. hepaticus VgrG variants, VgrG1, VgrG2, VgrG3. The crude serum raised against the Hcp peptides reacted against H. hepaticus Hcp, and the serum raised against the two VgrG peptides reacted against all three VgrG variants when they were expressed in E. coli or in transfected eukaryotic cells (not shown). The primary rabbit sera were then further purified by affinity column purification against the immunization peptides of each protein, in order to obtain antisera specific for Hcp and VgrG. In whole cell lysates of H. hepaticus wild type separated on Western immunoblots, the Hcp-specific purified antiserum routinely detected only one band with a molecular mass of c. 26 kDa (corresponding to the predicted mass of Hcp), and the VgrG-specific purified antiserum recognized a protein with a predicted molecular mass of c. 100 kDa (corresponding to the predicted mass of VgrG1 of 103.7 kDa).

Generation of bacterial whole lysates and subcellular fractions

Plate-grown bacteria were suspended in ice-cold PBS containing protease inhibitor cocktail (Roche) and sonicated for 2–5 min at 4°C or lysed in a FastPrep® homogenizer (MP Biomedicals, Irvine CA). The total lysate was then directly analysed, fractionated or stored at −20°C. Alternatively, bacteria grown in liquid culture were pelleted at 4000 g for 20 min (H. hepaticus) or 5 min (E. coli), resuspended in PBS with protease inhibitors and lysed as described above. The full methods to separate and control bacterial subcellular fractions (soluble, insoluble, surface and secreted fractions) are presented in the supplementary methods.

Co-incubation of cells with H. hepaticus

Adherent mouse cells (J774, mICcl2) and suspension THP-1 cells (human monocyte-like cell line) were seeded into 24-well plates at 2 × 105 cells per ml per well 24 h or 3 h prior to the infection respectively. The infection was performed in fresh media with 10% FCS (changed in the adherent cells 2.5 h in advance) after a 1.5 h pre-activation of the cells with 50 ng of E. coli LPS per ml of cell culture. In samples designated for WB detection of H. hepaticus proteins in TCA-precipitated culture supernatant, the serum and LPS addition were omitted. Bacteria grown on standard blood agar plates for 24 h were diluted in PBS so that 100 μl of the suspension contained an equivalent to multiplicity of infection (moi) of 10 or 100 bacteria per cell. For mock samples, 100 μl of PBS was used. To ensure contact with the cells, the bacteria were spun down onto the cells at 300 g for 10 min at RT. A 2–4 h co-incubation step in 5% CO2 at 37°C followed. The cells were harvested in the co-incubation medium, pelleted at 10 000 g for 3 min at 4°C and supernatants and pellets frozen in liquid nitrogen. Samples were stored at −20°C. For Western blot analysis of bacterial proteins in the co-culture supernatants, the medium was harvested, centrifuged at 10 000 g for 3 min, sterile-filtered through 0.22 μm PVDF membrane filters (Millex GV, Millipore) and precipitated with 10% TCA at −20°C. Precipitated proteins were further processed as described in ‘Generation of bacterial subcellular fractions’ in Supplementary methods.

Cell transfection with plasmid DNA and analysis of protein expression

The recombinant expression experiments were performed with all cell types listed. HEK293-T and J774 cells were seeded in 24-well plates at a concentration of 1–2 × 105 cells per well in 1 ml of Dulbecco's MEM medium (Biochrom) containing 10% FCS. After c. 24 h growth at 37°C, the medium was changed for 0.5 ml per well of Optimem medium (Gibco, Karlsruhe, Germany) supplemented with 5% FCS. The cells were further incubated for 1 h and then transfected with a plasmid of interest [prepared by endofree plasmid Midi Kit (Qiagen), 0.4–0.8 μg of plasmid per well for HEK293-T cells or 0.2–0.4 μg for J774 cells] by use of Lipofectamine 2000TM (Invitrogen, Carlsbad, California, USA) according to the manufacturer's instructions. mICcl2 cells were treated in a similar way using DMEM/Ham's F-12 medium with the required supplements (Bens et al., 1996) as a standard growth medium and 0.4 μg of plasmid DNA per well for the transfections. THP-1 cells were grown in RPMI 1640 medium with 10% FCS and were harvested directly before transfection by 10 min centrifugation at 120 g to obtain 106 cells per sample. The pelleted cells were then resuspended in 100 μl of nucleofector solution containing supplement (Amaxa® Cell Line Nucleofector Kit V, Lonza), mixed with 0.25 μg of DNA and transfected in cuvettes using the Nucleofector®Program U-001. Immediately after nucleofection, the cells were mixed with 500 μl of RPMI medium and transferred into a 24-well plate containing 500 μl of pre-warmed medium.

For the expression of H. hepaticus Hcp and VgrG1, VgrG2 and VgrG3 proteins in eukaryotic cells, the corresponding coding sequences were cloned into a pEF6-V5 expression plasmid (modified from a pEF6/V5-His-TOPO plasmid, Invitrogen; C. Josenhans, unpublished), which enables labelling of the expressed proteins with a V5 peptide tag at their C-terminus, or into a pTRE3G_mCherry vector (Clontech Laboratories, Mountain View, CA, USA), which is tetracyclin- or transactivator-inducible. As negative controls, pEF6-V5 or pTRE3G_mCherry empty plasmids were used, and an eGFP expression plasmid served for determining the transfection efficiencies. The transiently transfected cells were further incubated for 24 h or 48 h to allow expression of the proteins, or induced with Tet transactivator mix (Clontech) 6 h before further processing. Subsequently, the cells were analysed for protein expression by Western blotting and fluorescence microscopy, and for cell toxicity by LDH assay, and examined using further tests. For generating cell lysates, the cells grown in 24-well plates were harvested at selected time points after the experimental treatment (transfection, infection). The cell pellet was then directly frozen in liquid nitrogen and stored at −20°C. For protein analysis, the pellet equivalent to the cell content of one well was resuspended in 60 μl of ice-cold PBS with or without proteinase inhibitor cocktail tablets (CompleteTM, Roche).

Cell activation/activation of transfected cells with LPS, ATP or PMA

To stimulate cell activation and differentiation of the macrophages, J774 and THP-1 cells were treated with E. coli LPS, ATP or PMA upon transfection. LPS was added in an amount of 100 ng per well 4–5 h before harvest, after the J774 cells had been provided with 600 μl per well fresh DMEM with 10% FCS. The stock of E. coli LPS (100 μg ml−1, List Biological Laboratories Ultrapure LPS of E. coli K12) was sonicated for 1 min, and 100 ng were subsequently applied to each well. In one experiment, the J774 macrophages were additionally stimulated with a 3 mM ATP pulse 15 min before harvest as suggested by others to induce inflammasome formation (Bryan et al., 2010). The ATP powder (cell culture grade, Sigma) was resuspended directly before use in sterile, HPLC-purified water so that 18 μl per well corresponded to the desired concentration. The phorbol-1,2-myristate-1,3-acetate (PMA; Sigma) was used at a concentration of 50 ng ml−1 to differentiate THP-1 cells into macrophages and make them adherent.

Detection of H. hepaticus proteins expressed in HEK293-T cells

The transfected HEK293-T cells were harvested at different time points after plasmid transfection by suspending them in the culture medium with a pipette and spinning down at 21 000 g for 3 min. Each cell pellet was then resuspended in 60 μl of ice-cold PBS containing proteinase inhibitor cocktail tablets (CompleteTM, Roche, Germany), 10 μl of each sample, which was equivalent to approximately 10 μg of total protein, was mixed with 5× SDS loading buffer (with β-mercapto-ethanol), heat-denatured at 100°C for 10 min and separated on SDS-PAGE (11.5% acrylamide/bisacrylamide in resolving gel; Laemmli, 1970). Western blots were then incubated with murine anti-V5 monoclonal antibody (Invitrogen, Carlsbad, USA) at 4°C overnight and developed as described above.

LDH release cytotoxicity assay

Lactate dehydrogenase (LDH) release from cells was measured to determine cytotoxic and lytic effects of expressed bacterial proteins on cells. LDH was measured in the growth media and in whole cells (control) of the transfected cell cultures using the cytotoxicity Detection KitPLUS (Roche). The protocol provided by the manufacturer was adapted for this specific purpose (see supplementary methods).

Immunofluorescent labelling and immunofluorescence microscopy

For visualization of the proteins expressed in HEK293-T cells by immunofluorescence, the cells were seeded onto coverslips coated with 0.3% gelatin and 24 h after transfection, fixed with 2% paraformaldehyde in 100 mM potassium phosphate buffer (pH = 7) at RT for 2–3 h. The fixing agent was then replaced by quenching buffer (0.1% glycine in PBS), and afterwards the samples washed twice with PBS. After a 30 min blocking step with 1% bovine serum albumin in PBS and 1% FCS (Biochrom) in PBS, the primary antibody (anti-V5 monoclonal, Invitrogen, diluted 1:1000 or 1:500 in blocking buffer) was added and incubated at 4°C overnight. Next, the samples were washed three times with 0.1% BSA in PBS and incubated with the secondary antibody (goat anti-rabbit Alexa Fluor®488, Invitrogen, diluted 1:5000 in blocking buffer) at RT for 1 h. Following the next washing steps, the cells were stained with DAPI (4′,6-diamidino-2-phenylindole, 0.4 μg ml−1, Sigma) and AlexaFluor®546-phalloidin (5 units ml−1, Invitrogen) in PBS for 15 min. Finally, the coverslips were mounted with Mowiol containing 2.5% DABCO (1,4-diazabicyclo-[2.2.2]octane, Biomol) on microscope slides. The samples were then analysed using a Zeiss Apotome microscope.

Quantification of cytokine release by ELISA

Enzyme-linked immunosorbent assays were used to analyse the release of selected cytokines into the supernatants of transfected or infected cells (see supplementary methods).

Isolation of murine T-cells

T Cells were isolated from the spleens of two C57BL/6 mice by negative selection magnetic separation (CD4+ T-Cell Isolation kit, Miltenyi Biotec, Germany), according to the manufacturer's instructions. Briefly, approximately 1 × 107 total cells isolated from homogenized mouse spleens were incubated with non-CD4+ cell-depleting antibodies coupled to magnetic beads. MidiMacs columns (Miltenyi Biotec) were used to deplete the non-CD4+ cells. The flow-through, enriched in untouched CD4-positive cells, was subjected again to a Dead Cell Removal Kit (Miltenyi Biotec) to deplete dead cells. Both purified and non-purified control cells stained with anti-CD4 and anti-CD8 antibodies (CD4+ identification Kit, Miltenyi Biotec) were analysed further in a FACSCalibur (Becton Dickinson). Routinely, this purification procedure yielded approximately 90% pure CD4+ T-cells as quantified by FACS analysis.

Mouse infections and adoptive T-cell transfer experiments

C57BL/6 mice were obtained from Jackson Laboratories. C57BL/6 Rag2−/− mice were bred at the animal facility of Hannover Medical School under Helicobacter-free SPF conditions in IVC isolator cages. All animals were further kept in ventilated isolator cages under SPF conditions and fed sterile water and irradiated chow (SSNIF M-Z feed) ad libitum. The breeding colony was regularly monitored for the presence of common murine pathogens and various Helicobacter species, and all animals were tested again for the presence of Helicobacter species before the start of an experiment. All experiments involving mice were performed under the German laws for animal protection, after an ethical permit was obtained from the government of Lower Saxony (LAVES AZ 33.9-42502-04-08/1456).

For the T-cell transfer infection model, the method established by Kullberg et al. (2002) and Maloy et al. (2003) was employed, with minor modifications: briefly, 6- to 8-week-old SPF C57BL/6 Rag2−/− mice of mixed gender (distributed in gender- and age-matched groups of 16–20 animals) were infected intragastrically with H. hepaticus wild type and selected T6SS mutants at 2 × 108 cfu per mouse, on three subsequent time points, on days 1, 2 and 3. On day 6 or 7 post infection, the infected mouse groups and mock-infected (BHI) control groups (gavaged with brain–heart infusion medium only) were divided each into two matched groups of seven to eight animals and injected intraperitoneally either with PBS or with 4 × 105 cells from the enriched CD4+ T-cell preparation. The mice were weighed at the start of the infection and once a week during each experiment to monitor weight loss. To monitor the infection status throughout the experiment, faecal pellets were collected once a week and the presence or absence of H. hepaticus was assessed by semi-quantitative PCR from total DNA isolated from the stools (Tissue Amp Kit, Qiagen). Usually, no weight loss or any other external signs of disease or ill health upon visual inspection of the animals were observed throughout the observation period. Occasionally, loose stools were observed with the positive experimental groups, in particular in the both wild type-infected and T-cell-transferred animals.

Five weeks after the cell transfer, the mice were euthanized. Tissue samples from both caecum and colon were weighed, homogenized and plated on selective blood plates for H. hepaticus cfu counts per mg tissue. Samples from the caecum and colon were also fixed and embedded for tissue sectioning and pathological scoring. In parallel, tissue samples were stored in RNALater solution (Ambion) and used for total RNA and DNA preparation. In some of the experiments, Sv129 IL-10−/− and C57BL/6 Rag2−/− mice were inoculated with H. hepaticus 3B1 wild-type bacteria according to the same infection scheme as above, but without T-cell transfer.

Mouse pathology

Tissue sections were scored according to the scoring system developed by Burich et al. (2001) by blinded pathologists who are experts in rodent intestinal pathology.

Bacterial gene expression (semiquantitative reverse transcription PCR from bacterial and mouse samples)

Samples for detection of H. hepaticus RNA were collected either from in vitro H. hepaticus grown in liquid culture or from tissue samples isolated from mice chronically infected with H. hepaticus bacteria. Samples of bacterial liquid culture were harvested at selected time points during growth by centrifugation (21 000 g, 4°C, 1 min). The pellets were snap-frozen in liquid N2 and stored at −80°C. For in vivo expression analysis, tissue samples were removed from the mouse caecum and colon and stored in RNA laterTM (Sigma) at −80°C. The isolation of total RNA from bacterial pellets or infected murine tissue was performed using Qiagen RNeasy Mini Spin Kits in combination with ribolyser tubes (lysing matrix B – for bacteria; or matrix D – for tissue; FASTRNA) in a Fastprep Instrument (MP Biomedicals, Irvine CA). The obtained total RNA was tested for bacterial DNA contamination using the B38_hepaticus and B39_hepaticus primer pair amplifying a fragment of the H. hepaticus 16S rDNA gene. One microgram of total RNA of each sample was subsequently treated with DNase I (Roche) if required, and reverse transcribed into cDNA using random hexamer primers and Superscript III reverse transcriptase (Invitrogen). The cDNA (2.5 μl per reaction mixture) served then as a template for PCR with selected specific H. hepaticus primer pairs (Table 4).

Table 4. Primers used for RT-PCR
DesignationLocation in genomeSequenceProduct size (bp)Reference

B38 (= B38633f_hepaticus)

 16S rDNA

957744–957724GCATTTGAAACTGTTACTCTG417Shames et al. (1995)

B39 (= B391047r_hepaticus)

 16S rDNA

HH0291cdna_fw284819–284800TAACCCTTTTACCTCTCCCC290This study
HH0237cdna_fw227568–227549GGGACATTAGACATACAGAG206This study
HH0242cdna_rv233847–233866AGGAGAAGTGAGTGTAAAGG539This study
HH243_1234559–234578TCATCACTTCCGCTCGTTCC471This study
HH252_1243394–243414AAGTGCTTATTGTCGTATGGC967This study
HH275_1270154–270174CTTGGGAGATGTTCCTTAAGC958This study


Sophie Brinkmann and Wiebke Behrens are kindly acknowledged for technical support during a practical laboratory course and a Bachelor thesis. We are grateful to Ines Yang for contributions to the 16S rDNA-based microbiota analysis. Yvonne Speidel is acknowledged for excellent technical assistance. We are very grateful for continuous support by the animal facility of Hannover Medical School led by André Bleich and Hans-J. Hedrich. L.B. was supported by a PhD fellowship of the HBRS ZIB programme at Hannover Medical School. The German Research Foundation is acknowledged for funding in the framework of the centre grant programmes SFB621 ‘Pathobiology of the intestinal mucosa’ to S.S. and SFB900 ‘Chronic infections: microbial persistence and control’ to S.S. and to C.J. Support by the German Center for Infection Research (DZIF) to C.J. and S.S. is gratefully acknowledged.