Translocation of Helicobacter pylori CagA into Gastric Epithelial Cells by Type IV Secretion

See allHide authors and affiliations

Science  25 Feb 2000:
Vol. 287, Issue 5457, pp. 1497-1500
DOI: 10.1126/science.287.5457.1497


The Gram-negative bacterium Helicobacter pylori is a causative agent of gastritis and peptic ulcer disease in humans. Strains producing the CagA antigen (cagA +) induce strong gastric inflammation and are strongly associated with gastric adenocarcinoma and MALT lymphoma. We show here that such strains translocate the bacterial protein CagA into gastric epithelial cells by a type IV secretion system, encoded by the cagpathogenicity island. CagA is tyrosine-phosphorylated and induces changes in the tyrosine phosphorylation state of distinct cellular proteins. Modulation of host cells by bacterial protein translocation adds a new dimension to the chronicHelicobacter infection with yet unknown consequences.

The Gram-negative bacterium Helicobacter pylori(Hp) colonizes the human gastric epithelium and is strongly associated with peptic ulceration, MALT-lymphoma, and adenocarcinoma of the stomach (1). The cagA gene is a genetic marker for a 40-kb pathogenicity island (cag-PAI) ofHp, present in certain strains (cag+ , type I), but not in others (cag , type II) (2). The function of the immunodominant CagA protein is unknown. Attachment of Hp to epithelial cells in vitro induces tyrosine phosphorylation of a 145-kD host cell protein (3). Mutations in several genes of thecag-PAI interfere with tyrosine phosphorylation and secretion of the chemokine interleukin-8 (4,5). The cag-PAI carries a set of genes with homology to so-called type IV secretion systems (6), the prototype of which is the Agrobacterium tumefaciens viroperon, which mediates transfer of T-DNA into the nucleus of plant cells (7). Several large conjugative plasmids harbor a homologous system, involved in transfer of plasmid DNA into recipient cells during conjugation (8). Bordetella pertussis uses a similar secretion system for the export of its major proteinaceous virulence factor, the pertussis toxin (8). For the postulated cag secretion system ofHp it was unclear whether DNA or proteins are the target for secretion. Here we provide evidence that the cag system is involved in protein secretion.

In agreement with published data (5), we show that certain cag+ Hp strains (P1, P2, P3, P12, G27, and ATCC43526) induce a de novo tyrosine phosphorylation of a protein in AGS epithelial cells (9, 10) (Fig. 1A). The apparent size of this tyrosine-phosphorylated protein varied significantly between different Hp isolates (125 to 140 kD), suggesting that it is a size-variable bacterial protein, rather than a host cell protein. Because the size variation of the bacterial protein CagA is in the same range (Fig. 1B), we constructed and tested isogeniccagA knockout mutants of Hp strains P1 and P12 (11). Both mutant strains were unable to induce tyrosine phosphorylation, whereas the wild-type (WT) strains did (Fig. 1, C and D). Knockout mutants in cagE, which have a defect in the secretion apparatus, still produced CagA, but lost the tyrosine phosphorylation capability (Fig. 1, C and D). Thus, CagA might be the observed tyrosine-phosphorylated protein translocated into AGS cells by the type IV secretion system.

Figure 1

Tyrosine phosphorylation ofHp CagA. (A to D) Immunoblots showing tyrosine phosphorylation of CagA upon attachment ofHp WT (A and B) or isogenic knockout mutant strains (C and D) to AGS epithelial cells. Blots were reacted with anti-phosphotyrosine antibody PY99 (A and C) or anti-CagA (AK257) (B and D). (E to G) Immunoprecipitation of CagA from a lysate of AGS cells alone (lanes 1 and 2) or AGS cells infected withHp P12 (lanes 3 and 4). AK257 was used for precipitation (lanes 2 and 4); controls without antibody (lanes 1 and 3) (13). (E) Silver-stained gel of the precipitate, (F) immunoblot with PY99, and (G) immunoblot with AK257. (H andI) Protein kinase inhibitor genistein but not staurosporine blocks tyrosine phosphorylation of CagA, and YopH (10) dephosphorylates CagAP-Tyr, as demonstrated by PY99 (H) or AK257 (I) in the immunoblot. The P12 lysate was prepared without contact of Hp to AGS cells, and the faint band in this lane is not sensitive to genistein treatment. (J) Genetic complementation of the P12ΔcagA mutant with the Hp 26695 cagA gene in trans (lanes 1, 2, and 3 are independent clones). Arrows indicate full-length CagA. Asterisks mark the position of cellular proteins p120-130 and p85 dephosphorylated upon translocation of CagA.

To verify this hypothesis, we subjected a lysate of AGS cells infected with Hp P12 to immunoprecipitation with antiserum to CagA (anti-CagA, AK257) (12, 13). CagA was precipitated when strong denaturing conditions, disrupting membrane structures, were applied (RIPA-buffer) (12). The mild detergent Triton X-100, however, resulted in only tiny amounts of immunoprecipitated CagA, arguing for a tight association of a defined CagA pool with subcellular structures of AGS cells. The precipitated protein reacted with both anti-CagA and anti-phosphotyrosine (Fig. 1, E to G, lane 4). Thus, CagA is the tyrosine-phosphorylated protein that is induced by Hp attachment, named CagAP-Tyr. In gastric (KatoIII, St 3051) and nongastric (ME180, but not HeLa and Chang) human epithelial cell lines (9), CagAP-Tyr was identified upon attachment ofHp P1 and P12.

Next, we examined whether CagA phosphorylation can be inhibited. Staurosporine, a serine-threonine kinase inhibitor, did not have a detectable effect, but genistein, a phosphotyrosine kinase (PTK) inhibitor, blocked CagA phosphorylation completely (10) (Fig. 1, H and I), indicating that a eukaryotic-type PTK activity phosphorylates CagA. The phosphotyrosine-specific YopH phosphatase ofYersinia completely removed the tyrosine phosphorylation from CagAP-Tyr, demonstrating that phosphorylation of CagA is reversible and tyrosine-specific (Fig. 1, H and I).

To obtain further, direct biochemical evidence for the residence of CagAP-Tyr in the eukaryotic cell, we established a lysis procedure, based on the detergent saponin, that destroys AGS cells, but leaves Hp intact (14). Selective lysis of AGS cells released high amounts of CagAP-Tyr whencagA+ WT strains attached, but no CagAP-Tyr was released from the P12(cagE) knockout mutant or the P12 strain without epithelial cell contact (Fig. 2, A and C). The P12(cagE) mutant strain occasionally released tiny amounts of nonphosphorylated CagA into the supernatant. The soluble cytoplasmic Hp RecA protein served as a control, demonstrating that the saponin lysis procedure did not release cytoplasmic proteins from Hp (Fig. 2, E and F). The assay shows that a functional type IV secretion system is necessary for translocation of CagA into AGS cells, where it is converted into CagAP-Tyr.

Figure 2

Selective lysis of AGS cells and release of translocated CagAP-Tyr. Supernatant- (A, C, and E) and pellet (B, D, and F) fractions obtained after selective lysis of AGS cells infected with Hp(10) were analyzed by immunoblotting with anti-phosphotyrosine mAb PY99 (A and B), anti-CagA (AK257) (C and D), and anti-RecA (AK263) (E and F). (#) P12 without AGS cells.

A second, independent approach that we used to demonstrate the presence of CagA in AGS cells was immunofluorescence (IF) (10). Anti-CagA (AK257) did not specifically recognize native CagA, and therefore a P12 strain with a precise deletion ofcagA (P12ΔcagA) was constructed and complemented with a COOH-terminally FLAG-tagged version of the 26695cagA gene (cagA-FLAG), expressed from the shuttle plasmid pHel2 (11). Successful translocation of CagA-FLAG into AGS cells was verified by selective lysis. IF of Hpattached to AGS cells revealed staining of Hp P12 with anti-Hp (AK175, green) and of CagA-FLAG with anti-FLAG M2 monoclonal antibody (mAb) (red) (Fig. 3A). Attached Hp translocate CagA-FLAG, which accumulates in the cell close to the attachment site of Hp. Hp-associated CagA-FLAG was detected only at zones of intimate contact between Hp and AGS cells (Fig. 3A, arrows), but not in intact Hp without AGS cell contact or the P12 WT strain (Fig. 3B).

Figure 3

Immunofluorescence of HpP12ΔcagA (cagA-FLAG) and HpP12 WT attached to AGS cells. (A) Detection of CagA-FLAG by anti-FLAG mAb M2 (red), specific for the FLAG epitope. AK175, directed against Hp, reacts with the bacteria (green). CagA-FLAG is detected only at positions of close contact between Hp and AGS cells (arrows). (B) P12 WT strain attaching to AGS cells reacts with AK175 (green) but not with M2 (red). Bar, 2 μm.

Under standard growth conditions, several proteins are tyrosine-phosphorylated in AGS cells (9). These include a prominent but diffuse band between 120 and 130 kD (p120-130) and in the 85-kD range (p85), as well as several smaller proteins (Fig. 1C, lane AGS). Attachment of Hp strains causes a reduction in the intensity, or even a complete loss, of p120-130 and p85, concomitantly with the appearance of CagAP-Tyr (Fig. 1, C and J). The rapid (5 to 10 min) disappearance of both protein bands in a time-course experiment suggests a dephosphorylation of these proteins, rather than a blocking of de novo phosphorylation. This effect was not observed whencagE or cagA mutants were used for attachment (Fig. 1C), indicating that translocation of CagA and its conversion into CagAP-Tyr is the causative step for the dephosphorylation of cellular proteins.

To examine why Hp 26695, carrying two putative tyrosine phosphorylation sites in CagA (Fig. 4), did not induce CagAP-Tyr, we complemented the P12ΔcagA strain by the intactcagA gene of Hp 26695 in trans, using the pHel2 shuttle vector (15). The P12ΔcagA deletion mutant induced neither tyrosine phosphorylation of CagA, nor dephosphorylation of host proteins, but the complemented strain restored both activities (Fig. 1J). Thus, the genetic background of P12 allows translocation of theHp 26695 CagA, suggesting that the type IV secretion apparatus in the Hp 26695 isolate might be switched off or defective.

Figure 4

Putative tyrosine phosphorylation motifs in CagA and phosphorylation intensity of variant CagA proteins. Schematic representation of the Hp 26695 CagA sequence and three putative tyrosine phosphorylation sites A, B, and C in different sequences. The eukaryotic tyrosine phosphorylation consensus sequence and putative motifs in the CagA sequences are boxed, conserved amino acids are shown in outline. Y denotes the phosphorylated tyrosine residue. aa, amino acid position. (+) low, (++) medium, (+++) strong, (−) no tyrosine phosphorylation of CagA. Abbreviations for the amino acid residues are as follows: D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; and Y, Tyr.

A MOTIF databank search (16) identified several putative tyrosine phosphorylation motifs in the CagA protein sequence of independent strains (Fig. 4) (17). The CagA protein of strain J99 without a motif (18) was not tyrosine-phosphorylated (Figs. 1A and 4). The two availableHp genome sequences (18, 19) do not contain genes homologous to bacterial or eukaryotic tyrosine kinases, as is the case in other bacteria (20, 21), which supports our hypothesis that CagA is phosphorylated in the eukaryotic cell. Whether translocated CagA induces a cellular protein tyrosine phosphatase (PTP), or whether CagAP-Tyr itself is a type of eukaryotic PTP is unclear. The identification of CagA as a protein that is translocated into human gastric cells by the type IV secretion apparatus of Hp may have important biological consequences for the chronically persisting pathogen, as well as for the host.

  • * To whom correspondence should be addressed at Max von Pettenkofer Institute, Pettenkoferstrasse 9a, D-80336 Munich, Germany. E-mail: haas{at}


View Abstract

Navigate This Article