Interaction with CagF Is Required for Translocation of CagA into the Host via the Helicobacter pylori Type IV Secretion System

PCR and plasmids.All oligonucleotides used for PCR amplification are shown in Table 1. Plasmids used in this work are described in Table 2. Plasmids were constructed to express epitope-tagged versions of Cag proteins in Escherichia coli and H. pylori. The plasmid pBluescript SK(+) (Stratagene) was modified to yield the plasmid pSK+recxorf8. Briefly, a 1,500-base-pair region upstream of the recombinase gene in H. pylori (recA) was amplified using the oligonucleotides recUp+ and recUp− and cloned into the KpnI/XhoI sites of the pBluescript SK(+) vector. The resulting vector was digested with NotI/SacI, and a 750-base-pair recA downstream region was inserted, which was amplified using the oligonucleotides recD+ and recD− and digested with identical enzymes. A kanamycin resistance gene cassette (kan) was introduced between the recA upstream and downstream regions using the BamHI and NotI restriction sites. The cagA promoter was amplified (CagA^Pr+ and CagA^Pr−) and digested by XhoI/NdeI. orf8 was amplified (orf8FL+ and orf8FL−), introducing a FLAG sequence flanked by XbaI and BamHI restriction sites at the 3′ end of the gene. Both fragments, the cagA promoter and orf8-FLAG, were simultaneously inserted between the kan cassette and the recA upstream region (Xho/BamHI). The resulting vector was designated pSK+recxorf8. To create pSK+recxcagA and pSK+recxcagF, the orf8 gene was excised with NdeI/XbaI from the vector and replaced by one of the respective genes, which were amplified excluding the stop codons, using the oligonucleotide pairs cagA+ and cagA− and cagF+ and cagF−.

To express the CagF-FLAG fusion from pACYC184 (New England Biolabs), cagF-FLAG including the cagA promoter was excised by using XhoI/BamHI from pSK+recxcagF and cloned into pACYC184 digested with SalI/BamHI. To express CagA without the FLAG tag, cagA was amplified with its natural promoter using the oligonucleotides cagA^Pr+ and cagAD− and cloned into pBluescript SK(+) using the SalI/NotI sites.

Bacterial strains and growth conditions.All H. pylori strains (Table 2) were grown on brucella broth medium supplemented with 5% fetal bovine serum (FBS), and the following antibiotics were added for selection: trimethoprim (10 μg/ml), vancomycin (5 μg/ml), amphotericin B (8 μg/ml), and cycloheximide (100 μg/ml). Strains G27/8-FL, G27/F-FL, G27/A-FL, and G27ΔcagF/F-FL were grown on the brucella broth plates described above supplemented with kanamycin (20 μg/ml). Incubation of H. pylori strains was performed at 37°C for 48 h in an anaerobic jar containing 5% CO₂, 5% H₂, and 90% N₂. Liquid cultures were grown in brucella broth supplemented with 10% FBS (previous antibiotics and concentrations) in an anaerobic jar containing a Campygen gas mix of 5% O₂, 10% CO₂, and 85% N₂ (Oxoid). Cultures were grown at 37°C overnight, rotating at 160 rpm. E. coli strains (Table 2) were grown on LB plates and medium supplemented with the appropriate antibiotics: kanamycin (50 μg/ml), chloramphenicol (50 μg/ml), and ampicillin (100 μg/ml).

Helicobacter pylori strain G27 has been described previously (57). The G27 cagF-deleted mutant G27ΔcagF/F-FL was generated as described previously by Copass et al. (15a). Briefly, short regions of the cagF 5′ (oligonucleotides FUp+ and FUp−) or 3′ end (oligonucleotides FD+ and FD−) containing several hundred base pairs of flanking sequences were amplified by PCR and successively cloned after digestion into the pBluescript SK(+) vector to create pMA1. A cassette containing a kanamycin resistance gene and the sacB gene was then cloned between the two amplified regions, resulting in vector pMA2. Following transformation of pMS1 into strain G27, the cagF-kan mutant was selected for kanamycin resistance and the corresponding DNA region was amplified by PCR to confirm recombination. To create a nonpolar cagF deletion strain that was lacking artificially introduced genes, the cagF-kan mutant was transformed with pMA2 and mutants selected on brucella broth plates containing 10% sucrose. Colonies were confirmed by PCR and by testing for sensitivity to kanamycin.

H. pylori transformation.To express a FLAG-tagged Orf8, CagA, and CagF protein in G27, the vectors pSK+recxorf8, pSK+recxcagA, and pSK+recxcagF were transformed into the H. pylori strain G27 or the ΔcagF mutant, and recombinants expressing the FLAG-tagged proteins were selected on brucella broth plates containing 20 μg/ml kanamycin. Replacement of the recA gene by the individual FLAG-tagged cag genes via homologous recombination was confirmed by PCR and immunoblotting.

Immunoblotting.Samples were run on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels, transferred to Immobilon-P polyvinylidene difluoride (PVDF) (Millipore) membranes, and blocked in 5% (wt/vol) skim milk. All primary and secondary antibodies were diluted in 1% (wt/vol) bovine serum albumin. FLAG-tagged proteins were detected with a primary monoclonal mouse anti-FLAG antibody (1:5,000) (Sigma). CagA was detected with a polyclonal rabbit anti-CagA serum (1:2,500) (kindly provided by Antonello Covacci). CagF was detected with precleared anti-CagF rabbit polyclonal serum (48). The polyclonal rabbit antibody anti-VirB11 (1:5,000) (kindly provided by Rainer Haas) was used to detect the putative inner membrane protein VirB11 (Hp0525). The mouse monoclonal antibody anti-HopE (1:1,000) (kindly provided by Peter Doig) was used to detect the outer membrane protein HopE. Secondary antibodies were anti-rabbit or anti-mouse conjugated to horseradish peroxidase (Sigma), and membranes were developed using the chemiluminescent substrate ECL (Amersham).

³⁵S metabolic labeling, cross-linking, and immunoprecipitation. Helicobacter pylori strains were grown under microaerophilic conditions overnight in 90% RPMI-based minimal medium (lacking Cys and Met)-10% brucella broth mixture at 37°C shaking at 160 rpm. Ten microliters of Redivue [³⁵S]methionine (Amersham) was added to each culture and incubated for an additional 5 h. Three milliliters of each culture was collected by centrifugation, washed two times, and resuspended in 200 μl 10 mM phosphate-buffered saline. Ten microliters of 10% NP-40 (Calbiochem) and then 15 μl of 25 mM dithiobis (succinimidyl propionate) (DSP) in dimethyl sulfoxide was added to each sample. Samples were incubated on ice for 2 h, and the cross-linking reaction was stopped with 5 μl of 1 M Tris (pH 7.5). For immunoprecipitation, a threefold volume of lysis buffer (150 mM phosphate-buffered saline, 7.15% [wt/vol] sucrose, 1% NP-40, 1 mM phenylmethylsulfonyl fluoride, 1/66 of 15 μg/ml) was added to each sample. Cells were frozen and thawed 5 times and centrifuged for 10 min at 18,000 × g. The supernatant was removed and added to 30 μl anti-FLAG M2 agarose beads (Sigma). Beads were previously equilibrated with three washes of lysis buffer and blocked with 1% (wt/vol) bovine serum albumin for 6 h to eliminate nonspecific binding. The supernatant and beads were incubated at 4°C for 2 h on a rotary shaker. Beads were centrifuged at 16,100 × g, washed in lysis buffer three times, resuspended in 50 μl of 1× Laemmli sample buffer (5% β-merceptoethanol), and boiled for 10 min.

Bacterial fractionation via differential centrifugation and detergent solubility.Bacterial fractionation was performed as described previously by Gauthier et al. (25). H. pylori strains were grown overnight in brucella broth medium and adjusted to an approximate optical density at 600 nm of 1.5. Cells were sonicated five times for 1-min bursts at 35% output (Fisher sonic dismembrator; amplitude, 1.4). N-Lauroyl sarcosine was used at 2% (wt/vol) for solubility of the inner membrane proteins with rotation at room temperature for 30 min. The periplasmic and cytosolic fractions were recovered in 1 ml, the inner membrane fraction in 500 μl, and the outer membrane in 100 μl of each appropriate buffer. Equal percentages of each sample were analyzed by immunoblot analysis.

Infection and fractionation of AGS cells.Gastric epithelial AGS tissue culture cells (ATCC CRL-1739) were grown at 37°C in 5% CO₂-95% air in RPMI medium supplemented with 10% (vol/vol) FBS (Invitrogen). In vitro infection and fractionation of AGS cells yielding cytosolic, membrane, and insoluble/bacterial fractions were performed as described previously by Stein et al. (50). AGS cells were infected at a multiplicity of infection of 100:1.

CagF interacts directly with two molecular mass species of the CagA antigen in E. coli.We sought to screen for cag protein interactions using an activity reconstitution-based bacterial two-hybrid system, because a yeast two-hybrid assay did not identify any interactions between members of the cag PAI (53). We identified several potential interacting partners but were unable to obtain reliable and consistent results (unpublished data). One possible interaction revealed by the bacterial two-hybrid results involved the CagA antigen and a protein of unknown function, CagF. To test this potential interaction, plasmids expressing FLAG-tagged CagF and native CagA were constructed as described and cotransformed into E. coli. Coimmunoprecipitation of CagA together with the FLAG-tagged CagF in E. coli supported the interaction between the two proteins (Fig. 1). CagA was not unspecifically precipitated by the FLAG beads when expressed alone, supporting the specificity of the CagA-CagF interaction. The ability for the two proteins to interact in E. coli suggests that the interaction occurs directly and is not conferred by an intermediary Cag protein. CagA is unstable in E. coli and breaks down into two major fragments, a 100-kDa amino-terminal fragment and a 35-kDa carboxy-terminal species as seen with H. pylori (32, 36, 57). Both the full-length and cleaved 100-kDa fragment of CagA coprecipitated with CagF. These results suggest that the distal carboxy-terminal portion of CagA may affect, but is not absolutely required for, CagF binding.

• Open in new tab
• Download powerpoint

FIG. 1.

CagA coimmunoprecipitates with CagF after expression in E. coli. CagA was coexpressed with CagF in E. coli strain DH5α (DH5αAF) or expressed alone as a negative control (DH5αA). Proteins interacting with CagF were coimmunoprecipitated from cell lysates with monoclonal anti-FLAG antibody coupled to agarose beads. Immunoprecipitates (IP) and supernatants (S) were separated by 9% SDS-PAGE and transferred onto PVDF membranes in duplicate. The membranes were probed with polyclonal CagA antiserum (α-CagA) or with monoclonal anti-FLAG antibody (α-FLAG). Blots were developed with peroxidase-coupled secondary antibodies.

Confirmation of the CagA-CagF interaction in H. pylori.A chromosomal insertion of CagF-FLAG was made in strain G27 as described in Materials and Methods. H. pylori strains G27 and G27/F-FL were metabolically labeled with ³⁵S and cross-linked with DSP, and cell lysates were applied to anti-FLAG agarose beads (Sigma). Samples of the immunoprecipitations and supernatants were subjected to SDS-PAGE and transferred to PVDF membranes. The membranes were developed by autoradiography (Fig. 2). An intense band running in the 35-kDa range was detected on the radiograph. This band comigrated with the specific immunoprecipitated CagF-FLAG (theoretical molecular mass, 32 kDa) and was absent in the control G27 immunoprecipitation. A second band in the G27/F-FL immunoprecipitation was present at roughly the 140-kDa range. This protein we suspected was CagA, and it was also not present in the G27 control precipitation. CagF was positively identified using the specific anti-FLAG antibody. CagA was positively identified using a polyclonal antibody directed against the full-length protein (data not shown). The lack of discernible immunoprecipitation bands with the G27 control strain suggests that the interaction with CagF was specific and that the proteins visualized were in fact coprecipitated. Several other bands were detected by autoradiographic analysis of the immunoprecipitate of G27/F-FL, suggesting that other proteins may be involved in a multiprotein complex. Further investigations of the identities of these proteins are under way in our laboratory.

• Open in new tab
• Download powerpoint

FIG. 2.

Immunoprecipitation (IP) of ³⁵S-labeled proteins that interact with CagF in H. pylori. H. pylori strains G27 (wild type) and G27/F-FL were metabolically labeled with ³⁵S for 5 h. Lysis, fractionation, and immunoprecipitation of both strains were performed as described for Fig. 1. Following 10% SDS-PAGE and transfer onto PVDF membranes, the ³⁵S-labeled proteins in each sample were visualized by exposure of the membrane to X-ray film.

The experiment was then modified to determine if this interaction was strong enough to be detected in the absence of the DSP cross-linker. Figure 3A shows that both CagA and CagF readily interacted in the absence of a cross-linking agent, suggesting that this interaction is specific, strong enough to allow coimmunoprecipitation, and not artificially induced by the cross-linker. CagA was also FLAG tagged and chromosomally inserted to allow for the reciprocal immunoprecipitation to be screened. The reciprocal immunoprecipitation of CagF with CagA-FLAG is illustrated in Fig. 3B. When FLAG-tagged CagA was immunoprecipitated from H. pylori, CagF was detected with a rabbit polyclonal CagF antibody. In order to ensure that the interaction was due to a specific protein interaction and not conferred directly by the FLAG tag fused to the C terminus of either CagF or CagA, Orf8, a cag protein from the distal 3′ region of the cag PAI, was FLAG tagged and used for immunoprecipitation as described above (Fig. 3C.). CagA was not detected, suggesting that the interaction between CagA and CagF was not an artifact caused by the presence of the FLAG tag. The G27 negative control was clear of unspecific binding of any of the proteins probed by immunoblotting (Fig. 3A and B.).

• Open in new tab
• Download powerpoint

FIG. 3.

CagA and CagF coimmunoprecipitate after expression in H. pylori. Lysates of H. pylori strains G27 (A and B), G27/F-FL (A), G27/A-FL (B), and G27/8-FL (C) were used in immunoprecipitation studies as described in the legend to Fig. 1. PVDF membranes were exposed to monoclonal anti-FLAG (α-FLAG), polyclonal anti-CagA (α-CagA), or polyclonal anti-CagF (α-CagF) serum as appropriate. The strains G27 (wild type) and G27/8-FL served as negative controls to test for proteins that might unspecifically bind to the beads or the FLAG tag, respectively.

Subcellular localization of CagF.We next sought to determine the cellular location of CagF by use of the bacterial fractionation protocol of Gauthier et al. (25). This protocol has been well established for enteropathogenic E. coli and yielded a better fractionation than similar protocols used for H. pylori (7, 21, 22, 40, 60). Through prediction software analysis, it was suggested that CagF is a 31-kDa protein with a pI of 4.64 (Compute pI/M_w tool; http://ca.expasy.org/tools/pi_tool.html ), contains no signal peptide sequence (SignalP v3.0; http://www.cbs.dtu.dk/services/SignalP/ ), and has one strongly predicted transmembrane domain proximal to the C terminus of the protein (DAS [dense alignment sequence]; http://www.sbc.su.se/∼miklos/DAS/ ) (16, 19, 24, 35). The CagF protein was predicted to localize to the cytoplasm by multiple topographic prediction software. However, the transmembrane domain would suggest that a membrane localization is more likely for this protein than the cytoplasm (8, 23).

As Fig. 4A shows, the CagF antibody bound to the inner membrane and cytoplasmic fractions of strain G27 of H. pylori. The control inner membrane antibody directed toward the VirB11 ortholog was strongly reactive to the inner membrane fractions (Fig. 4B). Structural analysis suggests an inner membrane localization of VirB11 in H. pylori, which is consistent with models proposed for A. tumefaciens that suggest inner membrane anchoring and cytoplasmic exposure of the majority of the protein (11, 44, 58). A monoclonal antibody directed toward the outer membrane porin HopE was reactive to the outer membrane and the inner membrane (Fig. 4B) (18). These results suggest that the separation of the inner membrane from the outer membrane is not 100% efficient; however, the sarcosine-insoluble outer membrane fraction was free of inner membrane contamination (Fig. 4B). Since the CagF antibody was not reactive against the outer membrane fraction (Fig. 4A), this suggests that CagF is not likely outer membrane localized as previously published. The lack of a signal peptide sequence and the prediction of association with the cytoplasm and inner membrane, combined with our results, all argue against outer membrane localization. It should be noted that the method employed for bacterial fractionation has potential limitations with H. pylori. Results may vary depending on the growth phase and exact protocol used. Multiple popular detergent-based separation protocols were conducted, all yielding similar results.

• Open in new tab
• Download powerpoint

FIG. 4.

Bacterial fractionation. G27 (wild type) and G27/ΔcagF were fractionated into four major fractions: cytoplasm (C), inner membrane (IM), periplasm (P), and outer membrane (OM). The cytoplasm was purified by ultracentrifugation, the periplasm by lysozyme treatment, and inner and outer membranes by differential solubility in 2% (wt/vol) N-lauroyl sarcosyl. Equal percentages of samples were separated by 10% SDS-PAGE and transferred to PVDF membranes. A. Immunodetection of CagF and CagA was conducted by anti-CagF (α-CagF) and anti-CagA (α-CagA) polyclonal sera, respectively. B. Immunoblots were probed with the control antibodies anti-VirB11 (α-VirB11) (inner membrane) and anti-HopE (α-HopE) (outer membrane) to monitor the purity of the fractions.

The localization of CagF for bacteria harvested from an AGS tissue culture infection was identical to that for an overnight liquid culture (data not shown), suggesting that host cell contact had no noticeable effect on the localization of CagF. A virD4 mutant was also used to screen for localization of CagF and CagA. Previous work showed that a mutant of the putative ATP binding and hydrolyzing protein (52) prevents CagA translocation, likely due to a disruption of the T4SS (20). In the absence of the VirD4 protein, CagF and CagA fractionated similarly to the wild type, suggesting that their localization does not depend on the VirD4 protein (data not shown).

CagA localization and stability in a CagF mutant.To investigate a possible function for the CagF-CagA interaction, fractionation experiments were conducted on a ΔcagF mutant of G27. The interaction of CagA and CagF paired with the possible inner membrane localization of CagF suggested that CagF might play a role in delivery of CagA to the membrane for transport through the type IV secretion apparatus. When the fractions of G27 and ΔcagF were probed with a polyclonal CagA antibody, CagA was found located in the inner membrane for both samples (Fig. 4A). These results suggest that localization of CagA to the inner membrane of the bacterium was independent of CagF. However, since CagF is necessary for the successful translocation of CagA, we tested whether CagF stabilized CagA in H. pylori. CagA did not break down in the absence of CagF, suggesting that the stability of CagA in H. pylori is not dependent on the interaction with CagF (Fig. 4A). The interaction between CagA and CagF is therefore likely to exert its function after CagA localizes to the membrane through a mechanism still unknown. However, as is visible in Fig. 1, CagA appeared to be more stable when coexpressed with CagF in E. coli than CagA expressed alone.

CagF translocation was not detected in AGS cells.To test the possibility that CagF may translocate with CagA, the strains G27, G27ΔcagF, G27ΔvirB9, and G27ΔcagF/F-FL were tested in a translocation assay in which bacteria were incubated with AGS cells for 3 h, after which the AGS cells were washed and fractionated into soluble, host membrane, and insoluble fractions by mechanical lysis. This fractionation method was described previously and allows efficient separation of the host cytosol and membrane from the bacteria, which remain in the insoluble fraction. The fractions were analyzed for the presence of CagA, CagF-FLAG, and tyrosine-phosphorylated proteins. In both the wild type and the G27ΔcagF/F-FL strain, CagA translocation into the AGS cell membrane was detected by immunoblotting (Fig. 5A), while the G27ΔcagF and G27ΔvirB9 strains did not allow for CagA translocation. This result was also confirmed with antiphosphotyrosine antibody, which indicated that translocated CagA (136 kDa) was phosphorylated on tyrosine residues. This process was associated with decreased phosphorylation of host cell proteins (70 kDa) (42, 47). The FLAG antibody was able to detect CagF only in the insoluble fraction containing the bacteria but not in the host cell membrane or the cytosol (Fig. 5A and data not shown). These data suggested that CagF was not translocated with CagA. The requirement of both CagF and VirB9 for CagA translocation was previously reported by Fischer et al. (20). However, in that work the cagF mutant was created by insertion of a kanamycin resistance cassette into the cagF sequence, and thus, a polar effect on other genes could not be excluded. Our experiment employed a nonpolar in-frame deletion of most of the open reading frame of CagF, which we could use to verify the importance of CagF for CagA translocation. Reintroduction of the cagF-FLAG gene into the G27ΔcagF strain also demonstrated that CagF-FLAG is functional and able to restore the elongation phenotype of AGS cells (Fig. 5B).

• Open in new tab
• Download powerpoint

FIG. 5.

CagF is required for CagA host cell translocation but is not translocated itself. AGS cells were infected with the wild-type strain G27, its isogenic mutants G27Δhp532 (virB9) and G27ΔcagF, and G27ΔcagF/F-FL for 4 h or left uninfected. (A) Cells were fractionated into membrane (Mb) and insoluble (Insol) fractions as described previously (52). The fractions were separated by 10% SDS-PAGE and transferred onto PVDF membranes. Membranes were probed with anti-CagA (α-CagA), anti-phosphotyrosine (α-PY), or anti-FLAG (α-FLAG) antibodies to screen for CagA translocation, tyrosine phosphorylation, or the CagF-FLAG location, respectively. (B) Cells were visualized by phase-contrast microscopy to verify AGS cell morphology. CagA translocation was indicated by the characteristic elongation phenotype.