PAR1b takes the stage in the morphogenetic and motogenetic activity of Helicobacter pylori CagA oncoprotein

Abstract

Helicobacter pylori CagA oncoprotein is critically involved in gastric carcinogenesis. Upon delivery into gastric epithelial cells via type IV secretion, CagA induces an extremely elongated cell-shape known as the hummingbird phenotype, which is associated with massive changes in actin cytoskeleton and elevated motility. With the notion that the hummingbird phenotype reflects pathogenic/oncogenic activity of CagA, many studies have focused on the mechanism through which CagA induces the morphological change. Once delivered, CagA interacts with host proteins such as oncogenic phosphatase SHP2 and polarity-regulating kinase PAR1b. Whereas the essential role of the CagA-SHP2 interaction in inducing the hummingbird phenotype has been extensively investigated, involvement of the CagA-PAR1b interaction in the morphological change has remained uncertain. Recently, we found that the CagA-PAR1b interaction, which inhibits PAR1b kinase activity, influences the actin cytoskeletal system and potentiates the magnitude of the hummingbird phenotype. We also found that PAR1b inactivates a RhoA-specific GEF, GEF-H1, via phosphorylation and thereby inhibits cortical actin and stress fiber formation. Collectively, these findings indicate that CagA-mediated inhibition of PAR1b promotes RhoA-dependent actin-cytoskeletal rearrangement and thereby strengthens the hummingbird phenotype induced by CagA-stimulated SHP2 during infection with H. pylori cagA-positive strains.

Introduction

Helicobacter pylori is a micro-aerophilic spiral-shaped bacterium that colonizes in the human stomach, infecting about half of the world’s population.¹ Its chronic infection is known to cause chronic atrophic gastritis and peptic ulcer. Furthermore, epidemiological studies as well as animal infection studies with Mongolian gerbils have shown that its long-term infection causes gastric adenocarcinoma.^2,3 Since 1994, the International Agency for Research on Cancer (IARC) of the World Health Organization (WHO) has classified H. pylori as a definite (Group I) carcinogen.

Depending on the presence or absence of a gene called cagA, H. pylori strains are subdivided into two groups, cagA-positive strains and cagA-negative strains. The cagA gene is located at one end of the cag pathogenicity island (cag PAI), a ~40 kb DNA segment that contains a set of genes encoding components of the bacterial type IV secretion system.⁴ Upon infection with H. pylori cagA-positive strains, the cagA-encoded ~120–135 kDa CagA protein is delivered into gastric epithelial cells via the type IV secretion system (Fig. 1). Clinico-epidemiological studies have shown that infection with H. pylori cagA-positive but not cagA-negative strains markedly increases the risk of developing gastric carcinoma.⁵ Furthermore, transgenic mice systemically expressing CagA develop gastrointestinal and hematological malignancies spontaneously.⁶ A number of biochemical and cell biological studies have also shown that delivered CagA acts as a pathogenic scaffold/hub that promiscuously interacts with multiple host cell proteins regulating proliferation, motility, polarity and differentiation of cells.⁷ Those CagA-interacting proteins include SH2 domain-containing protein tyrosine phosphatase 2 (SHP2), a bona fide oncoprotein, and the cell polarity-regulating serine/threonine kinase Partitioning defective-1 (PAR1) (Fig. 1).^8,9

Figure 1.H. pylori CagA: host cell interaction. Upon delivery into gastric epithelial cells via the bacterial type IV secretion system, H. pylori CagA localizes to the inner face of the plasma membrane where it undergoes tyrosine phosphorylation at the C-terminal EPIYA segments termed EPIYA-A, EPIYA-B and EPIYA-C or EPIYA-D (EPIYA-C in the case of Western CagA and EPIYA-D in the case of East Asian CagA). Delivered CagA then interacts with SHP2 tyrosine phosphatase through EPIYA-C or EPIYA-D segment (EPIYA-C/D) in a tyrosine phosphorylation-dependent manner. The CagA-SHP2 interaction requires both N-SH2 and C-SH2 domains of SHP2. CagA also interacts with the PAR1/MARK family of polarity regulating serine/threonine kinases, comprising PAR1a/MARK3, PAR1b/MARK2, PAR1c/MARK1 and PAR1d/MARK4, via a C-terminal 16-amino acid stretch termed (C)agA-(M)ultimerization (CM) sequence. Of these, PAR1b is a major PAR1 isoform in epithelial cells. CagA directly binds to the catalytic domain of PAR1, thereby suppressing PAR1 kinase activity. Both CagA-SHP2 and CagA-PAR1 interactions have been reported to be involved in induction of the hummingbird phenotype by CagA. Arrows in the inlet indicate cells showing an extremely elongated cell-shape, termed the hummingbird phenotype.

Induction of Hummingbird Phenotype by CagA-SHP2 Interaction

Infection of cultured gastric epithelial cells such as AGS cells with H. pylori cagA-positive strains induces an extremely elongated cell-shape with elevated motility, termed the hummingbird phenotype (Fig. 1).¹⁰ Treatment of cells with hepatocyte growth factor (HGF) also induces a similar morphological change,¹¹ indicating that cagA-positive H. pylori infection triggers a growth factor-like response with intense cytoskeletal rearrangements. Because transfected CagA reconstituted the hummingbird phenotype in gastric cells, the bacterial protein was concluded to be responsible for the morphological change.⁸

Since then, many studies have been performed to elucidate molecular mechanisms underlying the CagA-mediated induction of the hummingbird phenotype as it appears to reflect in vivo CagA virulence. The discovery that SHP2, a CagA target, is critically involved in the HGF-induced cell-morphological change raised the possibility of involvement of SHP2 in the morphogenetic activity of CagA.¹² In gastric epithelial cells, delivered CagA undergoes tyrosine phosphorylation by Src family kinases and/or Abl kinase at the Glu-Pro-Ile-Tyr-Ala (EPIYA) motif that is located in variable numbers in the C-terminal region of CagA.^8,13 Upon tyrosine phosphorylation, CagA acquires the ability to interact with SHP2 and thus deregulates SHP2 phosphatase activity. CagA-activated SHP2 potentiates the magnitude of Erk MAP kinase signaling in both Ras-dependent and Ras-independent manners.¹⁴ CagA-stimulated SHP2 also dephosphorylates focal adhesion kinase (FAK) and thereby inhibits FAK kinase activity that mediates cell-extracellular matrix interaction.¹⁵ Both Erk activation and FAK inhibition by CagA-deregulated SHP2 are required for induction of the hummingbird phenotype (Fig. 2).^14,15

Figure 2. CagA-deregulated signals that mediate hummingbird phenotype. In gastric epithelial cells, CagA-deregulated SHP2 aberrantly activates the ERK MAP kinase signaling pathway via both Ras-dependent and -independent mechanisms. CagA-deregulated SHP2 also inactivates focal adhesion kinase (FAK) by dephosphorylating activating phosphorylation residues, Y576 and Y577, and thereby impairs focal adhesion turnover. The CagA-PAR1b interaction prevents PAR1-dependent inhibitory phosphorylation of GEF-H1, a RhoA-specific GEF, on S885 and S959. As a result, CagA potentiates GEF-H1-dependent RhoA activation, which in turn stimulates RhoA-dependent stress fiber formation.

Physical Complex Formation of CagA with PAR1b

When expressed in polarized epithelial cells mimicking the gastric mucosa, CagA disrupts tight junctions and causes loss of apical-basal polarity. This CagA activity is attributed to the interaction of CagA with PAR1 [also known as microtubule affinity-regulating kinase (MARK)],⁹ anevolutionally conserved serine/threonine kinase playing a critical role in both establishment and maintenance of cell polarity by regulating microtubule stability.¹⁶ In mammals, PAR1 comprises four isoforms, PAR1a/MARK3, PAR1b/MARK2, PAR1c/MARK1 and PAR1d/MARK4. Among these, PAR1b is thought to be a major isoform in gastric epithelial cells. CagA directly binds to the kinase catalytic domain of PAR1 in a tyrosine phosphorylation-independent manner and thereby directly inhibits PAR1 kinase activity.⁹ PAR1 has been thought to regulate cell polarity at least in part via phosphorylation of microtubule-associated proteins (MAPs), which influences microtubule dynamics involved in asymmetric protein distributions.^16,17 The CagA sequence mediating interaction with PAR1 is a C-terminal 16-amino-acid stretch, which was originally identified as a CagA-Multimerization (CM) sequence (Fig. 1).

Involvement of PAR1 Inhibition in the Morphogenetic and Motogenetic Activity of CagA

More recently, PAR1b has come into the spotlight as yet-another player in the CagA-induced cell-morphological change. Based on the results of a series of transfection experiments showing that both SHP2 and PAR1b are required for full-blown induction of the hummingbird phenotype by CagA, Kikuchi et al. investigated the contribution of CagA-SHP2 and/or CagA-PAR1 interaction in the pathophysiological activity of cagA-positive H. pylori by using an in vitro bacterial infection system.¹⁸ To this end, H. pylori isogenic strains carrying a cagA mutant encoding a phosphorylation-resistant CagA (PR-CagA), which does not bind to SHP2, and a cagA mutant encoding CagA lacking the PAR1-binding CM (CagA-ΔCM) sequence were generated.

In gastric epithelial cells infected with H. pylori, wild-type CagA interacted with PAR1b via the CM sequence in a tyrosine phosphorylation-independent manner and with SHP2 in a tyrosine phosphorylation-dependent manner, as previously reported.^8,9 In contrast, there was little complex formation of CagA-ΔCM with SHP2 despite the fact that CagA-ΔCM was efficiently tyrosine-phosphorylated.¹⁸ This observation is consistent with results of previous studies with transfected CagA showing that the CagA-PAR1b interaction is important for stable CagA-SHP2 complex formation.⁹ Given that PAR1 is present as a dimer in cells, dimerization of CagA via the PAR1 dimer may stabilize the CagA-SHP2 complex.

The effect of cagA-positive H. pylori on the motility of gastric epithelial cells was next examined by a wound healing assay. In this assay, AGS cells were grown to confluence and scratch-wounded using a pipette tip to create a gap zone, and the area of wound closure was measured 12 h after H. pylori infection to quantify cell motility. Following infection with a wild-type cagA-positive strain, the area of wound closure significantly increased compared with that of cells infected with a cagA-deficient strain, indicating that bacterially delivered CagA enhances cell motility. However, an H. pylori strain producing either PR-CagA or CagA-ΔCM did not enhance cell motility as effectively as did an H. pylori strain producing wild-type CagA. The results indicated that both CagA-PAR1b interaction and CagA-SHP2 interaction substantially contribute to the elevated cell motility upon cagA-positive H. pylori infection.

Since infection with cagA-positive H. pylori induces the hummingbird phenotype in gastric epithelial cells,¹⁰ the effect of isogenic H. pylori infection on cell-morphological change was also investigated. Following infection with wild-type cagA-positive H. pylori, migrating cells displayed morphology characteristic of the hummingbird phenotype. In contrast, infection with H. pylori strains producing PR-CagA or CagA-ΔCM hardly induced cells with the hummingbird phenotype.¹⁸ The results indicated that both CagA-SHP2 and CagA-PAR1b interactions are required for cell elongation, which underlies the enhanced cell motility upon infection with cagA-positive H. pylori (Fig. 2). The above-described observations collectively lead to the conclusion that CagA-mediated inhibition of PAR1b potentiates the morphogenetic and motogenetic activity of CagA-deregulated SHP2.

PAR1b Inhibits RhoA through GEF-H1 Phosphorylation

PAR1b has recently been reported to influence actin cytoskeletal reorganization by inhibiting cortical actin formation along the cell periphery.¹⁹ This observation suggested that PAR1b has a functional link with the RhoA family small GTPases, key regulators of the actin-cytoskeletal system. Since RhoA is critically involved in cortical actin formation, functional interplay between PAR1b and RhoA was first investigated by treating cells with PAR1b-specific siRNA.²⁰ Whereas cells with no PAR1b-siRNA treatment only showed cortical actin formation, PAR1b-knockdown cells displayed not only cortical actin but also stress fiber formation in the cytoplasm. The increase in stress fiber formation was reverted by simultaneous expression of an siRNA-resistant PAR1b mutant or by treatment of cells with Y-27632, an inhibitor of the RhoA target ROCK. The results indicated that PAR1b prevents stress fiber formation by inhibiting RhoA activity.²⁰

PAR1 family members form a physical complex with RhoA-specific guanine nucleotide exchange factor GEF-H1,^21,22 suggesting that GEF-H1 is involved in the PAR1b-mediated inhibition of RhoA. To test this idea, gastric epithelial cells were transfected with a GEF-H1 vector together with a wild-type PAR1b or kinase-dead PAR1b vector, and the number of stress fiber-positive cells was counted. Co-expression of wild-type PAR1b, but not a kinase-dead PAR1b, inhibited stress fiber formation in gastric epithelial cells, indicating that PAR1b kinase activity is required for the inhibition of GEF-H1/RhoA-mediated stress fiber formation.

The above-described observations suggested that PAR1b directly phosphorylates GEF-H1. According to PhosphoSite, a protein phosphorylation database, there are 36 registered phosphorylation sites in human GEF-H1, among which simultaneous phosphorylation of serine 885 (S885) and serine 959 (S959) was reported to inhibit the RhoA-GEF activity of GEF-H1.²³ To test if S885 is phosphorylated by PAR1b in cells, a 14-3-3 binding assay was performed because 14-3-3ζ specifically binds to GEF-H1 phosphorylated on S885.²⁴ The results of the experiment revealed that wild-type PAR1b potentiates GEF-H1/14-3-3ζ interaction, while kinase-dead PAR1b does not, indicating that PAR1b phosphorylates GEF-H1 on S885 in cells.²⁰ An in vitro kinase assay provided additional evidence that PAR1b directly phosphorylates S885 of GEF-H1.

To confirm phosphorylation-dependent regulation of RhoA-GEF activity, a RhoA G17A binding affinity assay was performed. RhoA G17A has high affinity toward the active form of GEF-H1 and is used to measure RhoA-GEF activity of GEF-H1.²⁵ As expected, interaction of GEF-H1 with RhoA G17A was enhanced in the case of a GEF-H1-S885A/S959A mutant.²⁰ In contrast, a phosphomimic GEF-H1-S885D/S959D mutant did not bind RhoA G17A, confirming the requirement of phosphorylation of both S885 and S959 on GEF-H1 for suppressing RhoA-GEF activity. The effect of GEF-H1 phosphorylation on stress fiber formation was also examined by comparing the intensity of stress fibers in cells expressing wild-type GEF-H1 or GEF-H1-S885A/S959A mutant.²⁰ Expression of GEF-H1-S885A/S959A induced more extensive stress fiber formation than did wild-type GEF-H1. Thus, phosphorylation of GEF-H1 on both serine residues by PAR1b suppresses its RhoA-GEF activity and thereby inhibits RhoA-dependent stress fiber formation (Fig. 2). Since PAR1b is a major target of H. pylori CagA, the above-described observation indicates that CagA-mediated inhibition of PAR1 results in the activation of RhoA and RhoA-dependent actin cytoskeletal rearrangement.

Conclusions: Synthesis of the Two Observations

The work by Kikuchi et al.¹⁸ demonstrated that both CagA-SHP2 interaction and CagA-PAR1b interaction are involved in the morphogenetic/motogenetic activity of CagA. In the meantime, Yamahashi et al.²⁰ provided evidence that PAR1b regulates the actin cytoskeletal system by inhibiting GEF-H1-dependent RhoA activation. Now, these two observations collectively provide a molecular link between the CagA-PAR1 interaction and the morphogenetic/motogenetic activity of CagA. CagA-deregulated SHP2 plays a primary role in induction of the hummingbird phenotype by (1) activating Erk MAP kinase signaling and at the same time (2) inhibiting FAK-mediated cell-matrix interaction.^14,15 The elongation and motile phenotype triggered by the CagA-SHP2 interaction is further strengthened upon CagA-mediated inhibition of PAR1 that stimulates the GEF-H1/RhoA signaling. Since GEF-H1 is reported to be specifically localized to the rear of migrating cells,²⁶ CagA-mediated RhoA activation may be confined to the rear part. How, then, can CagA-activated RhoA potentiate the hummingbird phenotype? It has been shown that the hummingbird phenotype is characterized by rear retraction defect.²⁷ Considering that RhoA enhances focal adhesion assembly that facilitates cell attachment to the extracellular matrix, a transient nature of RhoA activation at the rear end must be crucial for the tail detachment during cell migration.²⁸ In cells expressing CagA, however, CagA-mediated inhibition of PAR1 renders sustained RhoA activation at the rear end that prevents tail retraction, resulting in the generation of a long protrusion as the cell moves.

These works also raise an interesting possibility as to the relationship between apical-basal polarity and front-rear polarity. Establishment of epithelial apical-basal polarity requires asymmetric distribution of the atypical protein kinase C (aPKC)/PAR3/PAR6 complex (aPKC complex) and PAR1b;PAR1b is localized to the lateral/basal membrane below the tight junction whereas the aPKC complex is distributed to and above the tight junction (apical membrane) (Fig. 3A, left).^16,29 The disruption of the tight junction causes loss of apical-basal polarity. Epithelial cells that have lost apical-basal polarity extrude from the polarized epithelial layer and display a mesenchymal-like cell morphology with elevated motility, which may be associated with the development of front-rear polarity along the direction of migration (Fig. 3A, middle and right). Although subcellular localization of PAR1b in migrating cells remains currently unknown, it is intriguing to hypothesize that, like apical-basal polarity, establishment and maintenance of front-rear polarity requires asymmetric distribution of the aPKC complex and PAR1b to the front and rear, respectively (Fig. 3A, right). Given that the hummingbird phenotype is characterized by a rear retraction defect,²⁷ it is also tempting to speculate that the loss of front-rear polarity reflects the CagA-induced hummingbird phonotype. If this hypothesis turned out to be true, then CagA-PAR1b interaction disrupts asymmetric distribution of PAR1b and the aPKC complex not only in epithelial cells with apical-basal polarity but also in migrating cells with front-rear polarity (Fig. 3B). At the molecular level, the impact of CagA on cell polarity may be achieved by CagA-mediated inhibition of PAR1b, which perturbs both the actin- and microtubule-dependent cytoskeletal systems. A key question that warrants future investigation is the PAR1-mediated spatiotemporal coordination of the actin cytoskeleton and microtubules during establishment and maintenance of apical-basal polarity (and possibly front-rear polarity as well), which is targeted by the H. pylori CagA oncoprotein. Another important question is the actual in vivo role of the morphogenetic and motogenetic activity of CagA in promoting gastric carcinogenesis in vivo.

Figure 3. A model that links polarity regulation and hummingbird phenotype. (A) In polarized epithelial cells, asymmetric distribution of PAR1b and aPKC complex is critical in the establishment and maintenance of apical-basal polarity; PAR1b (red line) localizes to the basal and lateral membrane, while aPKC/PAR3/PAR6 complex (aPKC complex) (green line) localizes to the apical membrane (left). We hypothesize that loss of apical-basal polarity, which is caused by disruption of the tight junction, elicits relocalization of PAR1b and aPKC that promotes the establishment of front-rear polarity. In this model, asymmetric distribution of aPKC complex and PAR1b should be conserved when the cell undergoes conversion of apical-basal polarity to front-rear polarity (middle, shown in solid and dotted lines for aPKC and PAR1b, respectively). As a consequence, PAR1b localizes to the cell rear in a reciprocal manner to the aPKC complex in migrating cells. (B) According to the model proposed in (A), the hummingbird phenotype may represent the loss of front-rear polarity in cells expressing CagA. Molecularly, perturbation of asymmetric distribution of PAR1b and the aPKC complex could be achieved by the inhibition of PAR1b kinase activity upon complex formation with CagA, which not only perturbs the microtubule-dependent cytoskeletal system but also deregulates the actin-dependent cytoskeletal system by inhibiting the PAR1-b-GEF-H1-RhoA signaling axis.

Kikuchi K, Murata-Kamiya N, Kondo S, Hatakeyama M. Helicobacter pylori stimulates epithelial cell migration via CagA-mediated perturbation of host cell signaling. Microbes Infect. 2012;14:470–6. doi: 10.1016/j.micinf.2011.12.003.

Yamahashi Y, Saito Y, Murata-Kamiya N, Hatakeyama M. Polarity-regulating kinase partitioning-defective 1b (PAR1b) phosphorylates guanine nucleotide exchange factor H1 (GEF-H1) to regulate RhoA-dependent actin cytoskeletal reorganization. J Biol Chem. 2011;286:44576–84. doi: 10.1074/jbc.M111.267021.