Helicobacter pylori type IV secretion apparatus exploits beta1 integrin in a novel RGD-independent manner

Abstract

Translocation of the Helicobacter pylori (Hp) cytotoxin-associated gene A (CagA) effector protein via the cag-Type IV Secretion System (T4SS) into host cells is a major risk factor for severe gastric diseases, including gastric cancer. However, the mechanism of translocation and the requirements from the host cell for that event are not well understood. The T4SS consists of inner- and outer membrane-spanning Cag protein complexes and a surface-located pilus. Previously an arginine-glycine-aspartate (RGD)-dependent typical integrin/ligand type interaction of CagL with alpha5beta1 integrin was reported to be essential for CagA translocation. Here we report a specific binding of the T4SS-pilus-associated components CagY and the effector protein CagA to the host cell beta1 Integrin receptor. Surface plasmon resonance measurements revealed that CagA binding to alpha5beta1 integrin is rather strong (dissociation constant, K(D) of 0.15 nM), in comparison to the reported RGD-dependent integrin/fibronectin interaction (K(D) of 15 nM). For CagA translocation the extracellular part of the beta1 integrin subunit is necessary, but not its cytoplasmic domain, nor downstream signalling via integrin-linked kinase. A set of beta1 integrin-specific monoclonal antibodies directed against various defined beta1 integrin epitopes, such as the PSI, the I-like, the EGF or the beta-tail domain, were unable to interfere with CagA translocation. However, a specific antibody (9EG7), which stabilises the open active conformation of beta1 integrin heterodimers, efficiently blocked CagA translocation. Our data support a novel model in which the cag-T4SS exploits the beta1 integrin receptor by an RGD-independent interaction that involves a conformational switch from the open (extended) to the closed (bent) conformation, to initiate effector protein translocation.

Figure 1. Host cell β1 integrin is essential for Hp to translocate and tyrosine-phosphorylate CagA in different cell lines.

(A) Immunoblots to determine translocation of CagA in HL60 versus dHL60 cells. (B) Immunoblots of β1 integrin knockout fibroblasts (GD25), epithelial cells (GE11) or β1 gene-complemented cells (GD25β, GE11β) infected with Hp strains or media (control). The upper panel shows CagA translocation using a phosphotyrosine-specific antibody (mAb PY99), the lower panel shows CagA production by Hp. Bands corresponding to Phospho-CagA (CagA-P) or CagA are indicated by arrows or open arrowheads, respectively. (C) FACS analysis quantifying β1 and β2 integrin surface localization on HL60 and dHL60 cells using antibody CD29 FITC (β1 integrin) or CD18 PE (β2 integrin). (D) Quantification of co-localization events of Hp wt, ΔcagA or ΔPAI bacteria and β1-integrin by laser scanning confocal microscopy. (E) Confocal micrographs, showing binding of Hp P12-gfp wt (upper panel) or P12ΔPAI (lower panel) and β1-integrin-labelled AGS cells (4B7-AlexaFluor₅₆₈) and areas of co-localization of Hp and β1 integrin (arrows).

Figure 2. Identification and verification of Hp T4SS proteins interacting with human β1 integrin.

(A) Domain organization of the β1 integrin clone used as bait in a YTH screen and identification of interacting proteins encoded by the cag-PAI used as preys. (+, interaction; −, no interaction), tenfold dilution steps still allowing growth of yeast on selective media are indicated by number of “+” signs (see Figure S1). (B) For pulldown assays, magnetic beads (SiMAG) were loaded with purified functional α1β1 or α5β1 integrin or Tris buffer (control) and incubated with processed Hp P12 wt or defined mutant strains, as indicated (fraction Soluble II, Figure S2A). Beads were recovered by magnetic forces, washed, boiled and run for SDS-PAGE. Immunoblotting with α-CagA or α-CagY antibody detects precipitated proteins (arrowheads). (C) Quantification of binding of purified GST-CagA, GST-CagYc, GST-CagI or GST-Inv397 (30µg/1×10⁶ cells) to integrin-deficient GE11 versus integrin-proficient GE11β cells by flow cytometry. Binding is analysed by α-GST antibody (fold increase in binding versus GST, the values for binding of GST alone has been set to 1). RPI, repeat region I, RPII, repeat region II, (*, p<0.05; **, p<0.01; ***, p<0.001; students T-test). MF, mean fluorescence, (n = 20). Oligonucleotides for construction of GST fusion proteins see Table S2.

Figure 3. Construction and functional characterization of Hp P12 strains producing CagL proteins with defined amino acid exchanges in the RGD motif and CagA binding to β1 integrin.

(A) Immunoblot of P12 wt and P12ΔcagL strains genetically complemented by a cagL wt or a mutated gene carrying a RAD or a RGA exchange, or an RGD empty site (ΔRGD). (B) Sandwich ELISA to determine the IL-8 secretion of AGS cells infected by P12 wt or specific cagL mutant strains. (C) Surface plasmon resonance sensograms of the interaction of immobilized CagA (stable 100 kDa N-terminal fragment) with integrin α5β1 (left) and αVβ3 (right) (0.07–1.12 ng/µl) shown in resonance units (RU). The following concentrations of integrin were used: magenta, 0.07 ng/µl; green, 0.14 ng/µl; red, 0.28 ng/µl; grey, 0.56 ng/µl; blue, 1.12 ng/µl. (n = 10 individual measurements and K_D calculations).

Figure 4. β1 Integrin signalling is dispensable for CagA translocation.

(A) CagA translocation assay using Hp strains P12 and P217 and CHO cells stably expressing either no (CHO K1), the β1A (complete gene), the β1COM (only common region of cytoplasmic tail) or the β1TR (no cytoplasmic tail) version of the human β1 integrin gene. For a full description of the truncated β1 gene constructs please refer to . Filled arrowheads depict CagA-P, open arrowheads CagA. ext, integrin external region, TM, transmembrane region. (B) Lysates of AGS epithelial integrin linked kinase (ILK) knockdown cells (siRNA-ILK) and cells transfected with GC-matched oligonucleotides (siRNA-Control) or lipofectamine-transfection (Mock-Transfection) were immunoblotted with α-ILK (Sigma) and anti-tubulin antibodies (loading control). The level of ILK knockdown was determined to be 86% by densitometry. AGS wt, ILK knockdown or control cells were infected with Hp strains, as indicated. Cell lysates were immunoblotted with α-phospho-tyrosine (PY99) or α-CagA antibodies (AK257). Bands representing CagA-P or CagA are marked by filled arrowheads, open arrowheads depict tyrosine-phosphorylated host cell proteins.

Figure 5. Interference with CagA translocation using β1 integrin-specific monoclonal antibodies.

(A) Mapped binding sites of anti-β1 integrin mAbs and their capacity to block CagA translocation are indicated (see also Table 1). (B) Pre-treatment of synchronized AGS epithelial cells by β1-specific mAbs (30µg, 1h or 3h), or (C) Pre-treatment of synchronized AGS epithelial cells by mAb 9EG7 or its Fab fragments generated by papain digestion (15µg, 1h or 3h). After Hp infection CagA translocation was determined (CagA-P, tyrosine-phosphorylated CagA, PY99), CagA and β1 Integrin were used as loading controls. (D) Verification of β1 integrin antibody binding to AGS cells, as determined by FACS (MF, mean fluorescence (α-mouse and α-rat AlexaFluor₄₈₈). (E) Quantification of 9EG7 binding to AGS cells with or without Mn²⁺ treatment by flow cytometry. (F) Quantification of 9EG7 Fab fragment binding to AGS cells with or without Mn²⁺ treatment by flow cytometry. (G) Quantification of binding of GST-CagA, GST-CagYc or GST-CagI to AGS cells treated with Mn²⁺ or Mn²⁺ and mAb 9EG7 by FACS analysis, to determine a possible competition in binding of GST fusion protein and 9EG7. # indicates no significant difference.

Figure 6. HeLa cells are inefficient for CagA translocation and produce constitutively active β1 integrin.

(A) Human gastric AGS and HeLa cells were infected with Hp strains P12 P217 and P145 and CagA translocation was determined by immunoblotting with antibody PY99 (CagA-P). Equal loading of bacteria and cells was determined by CagA (AK257) and β1 integrin detection (LM534). (B) Quantification of binding of mAb 9EG7 to AGS or HeLa cells with or without Mn²⁺ treatment by flow cytometry.

Figure 7. Proposed working model for β1 integrin acting as a receptor of the cag-T4SS pilus for translocation of CagA protein.

Hp T4SS pili (consisting of the pilus subunit CagC) are decorated by proteins CagA, CagY, CagL and probably CagI. The pilus associated proteins contact β1 integrin in the open extended conformation of the head, leading to the clustering of the integrin heterodimers. These binding events may induce a change at the legs of the integrin α/β chains, moving both legs together, which results in a closing and downward movement of the integrin head domains (bent, closed conformation) , which seems to be essential for CagA delivery. The mAb 9EG7 binds to the EGF-like domains of the β1 legs and stabilizes the open conformation by preventing a close interaction between the α and β integrin legs (disulfide bridges, indicated by red bars), which is suggested to prevent the downward movement and therefore a possible membrane insertion of the T4SS pilus and thus CagA delivery. α, alpha integrin subunit; β, beta integrin subunit; ILK, integrin-linked kinase; A, CagA; Y, CagY; L, CagL protein; E, EGF-like1-4; β, β-tail domain; H, Hybrid; I, I-like; P, PSI domain, 9EG7, mAb 9EG7.