Tyrosine-phosphorylated bacterial proteins: Trojan horses for the host cell.

For over a century, since the discovery by Roux and Yersin that sterile culture supernatants of Corynebacterium diphtheriae contained a potent toxin able to reproduce the lesions caused by diphtheria 1, most pathogenic bacteria have been considered to be microorganisms able to intoxicate local and distant tissues by secreting toxins in the extracellular medium. Recently, it has been shown that some bacteria inject toxic proteins directly into the cytoplasm of host cells using a specialized, needle-like 2 secretion apparatus (molecular syringe) known as a type III secretion system 3 4. A report in this issue by Ashai et al. 5 and reports from Haas et al. (Haas, R., personal communication), Segal et al. 6, and our own laboratory 7 now provide definitive evidence for the existence in Helicobacter pylori of a second type of molecular syringe (type IV secretion system) that is also able to inject toxic proteins into eukaryotic cells. These papers describe the type IV–mediated delivery into eukaryotic cells and the subsequent tyrosine phosphorylation of CagA, an immunodominant protein of H. pylori, encoded by a 40-kb pathogenicity island (cag). The need for a functional type IV secretion system for CagA translocation into the membrane fraction of host cells and its tyrosine phosphorylation is shown at several levels using a variety of isogenic mutants in the cag region 7, biochemical fractionation of the host cell compartments 7, and confocal microscopy 6. In addition, it has been shown that CagA present in culture supernatants or bacterial cell lysates is unable to enter eukaryotic cells, indicating that this is not a function that the soluble protein can perform on its own 5 7. Although the correlation between expression of CagA and H. pylori virulence was described a long time ago 8 9, the last seven years have been marked by the frustration of not finding any role for the CagA protein. Finally, the concomitant report by four independent laboratories of a role for CagA sends important messages to the scientific community. They are summarized below.

Type IV and Type III Secretion Systems Are Functionally Equivalent Molecular Syringes.

Gram-negative bacteria build and anchor different extracellular organelles, such as flagella and conjugative pili, using specialized supramolecular structures. These molecular engines transport the building blocks of flagella and pili across both the inner and outer membranes and polymerize the external filamentous structures by adding new monomers from the inside. In their evolution, bacteria found it convenient to duplicate the ancestral cluster of genes and use one set for further specialization as secretion apparatuses to translocate proteins or protein complexes into host cells (Fig. 1). In other terms, we may consider the type III and type IV secretion systems as spin-offs of flagella and conjugative pili, respectively. The two secretion systems have been extensively reviewed 4 10; a list of the bacteria known to contain them is printed in Table . Here we will just mention that although both systems are functionally equivalent and are used to translocate proteins into mammalian or plant cells, a number of properties differentiate them: (a) Type III apparatus and flagella 4 are encoded by 15–35 genes, at least 8 of which are well conserved in most systems. Type IV system and conjugative pili 10 are encoded by 11–31 genes, at least 6 of which are well conserved in most of them. No homology is present between the genes present in the two systems; however, both encode proteins with predicted ATPase activity. (b) The type III system secretes monomeric proteins with no apparent cleavable sec-dependent signal sequence. The type IV system may secrete assembled multimeric proteins such as pertussis toxin that are composed of different monomers, each having a typical sec-dependent signal peptide. This suggests that in some cases the proteins may enter the export machinery after being exported across the inner membrane by the general secretion system. Type IV secretion may, however, also export across both the inner and outer membranes nucleoproteins containing proteins and DNA (as in the case of the Ti plasmid T-DNA of Agrobacterium tumefaciens). In the case of the CagA protein reported in references 5–7, it is not known whether this is exported as a monomer or a more complex structure; however, the absence of a typical sec-dependent signal sequence in this protein suggests that the type IV system assists the translocation of CagA across both bacterial membranes and also across the host cell membrane. Finally, both type III and type IV secretion systems are encoded by genes that are clustered, often present in pathogenicity islands with a GC content different from the rest of the chromosome and that most likely have been acquired by horizontal transfer.

Figure 1

Schematic representation of a bacterial flagellum (a) and a conjugative pilus (b) showing their similarity to type III and type IV secretion systems, respectively. In a, the structural similarities between the core structures of the flagellar apparatus and the type III secretion system are sketched. In b, a hypothetical conjugative apparatus and a type IV system are reduced to an artistic impression; the proteic subunits forming the structure are not represented.

Table 1

Bacterial Species with a Fully Annotated Type III or Type IV Secretion System

Bacterial species	Type III	Type IV
Actinobacillus actinomycetemcomitans	—	+
Agrobacterium tumefaciens	—	+
Bordetella bronchiseptica	+	+
Bordetella pertussis	+	+
Brucella suis	—	+
Chlamydia spp.	+	—
Citrobacter rodentium	+	—
E. coli	+	+
Enteropathogenic E. coli	+	—
Enterohemorrhagic E. coli	+	—
Erwinia amylovora	+	—
Erwinia chrysanthemi	+	—
Erwinia herbicola pv. gysophila	+	—
Erwinia stewartii	+	—
Hafnia alveii	+	—
H. pylori	—	+
Legionella pneumophila	—	+
Pseudomonas aeruginosa	+	—
Pseudomonas syringae	+	—
Ralstonia solanacearum	+	—
Rickettsia prowazekii	—	+
Rhizobium spp.	+	—
Salmonella enterica	+	—
Shigella spp.	+	—
Xantomonas spp.	+	—
Yersinia spp.	+	—

cag Delivers Multiple Signals.

The presence of the cag pathogenicity island in H. pylori correlates with increased virulence and disease severity. The findings reported by Asahi et al. 5, Haas et al. (personal communication), Segal et al. 6, and Stein et al. 7 begin to unravel the molecular mechanisms that may link the presence of cag to disease. Two events are known to happen after cag-mediated bacterium–cell contact: induction of the proinflammatory lymphokine IL-8 and CagA tyrosine phosphorylation. The latter triggers host cell morphological changes such as cell elongation and spreading, a phenotype similar to that induced in AGS cells by hepatocyte growth factor or bacterial toxins such as cytotoxin necrotizing factor 1, which activate the small GTP-binding proteins Rho, Rac, and Cdc42 6. Data obtained by mutagenesis of the genes in the cag pathogenicity island (Table ) show that although both signals require an intact type IV secretion system, they are delivered by independent effectors. In fact, one of the mutants (cagF) is able to activate nuclear factor (NF)-κB but unable to induce IL-8 and CagA tyrosine phosphorylation. A schematic representation of the possible signaling mechanisms is shown in Fig. 2. Here the cag type IV secretion system is shown to translocate into the cell an unknown factor (in the figure it is named “?” for convenience) that activates the transcription factor NF-κB to induce IL-8 mRNA. As proposed by Naumann et al. 11, this mechanism is likely to involve activation of mitogen-activated protein (MAP) kinases and the transcription factor activator protein (AP)-1. It is not yet certain whether this pathway needs a yet unknown effector or whether it is simply activated by the type IV system itself, which perturbs the membrane. The second pathway activated by the cag system involves the protein CagA. This protein, as reported in references 5–7, is translocated into the eukaryotic cells, where it is tyrosine phosphorylated by a eukaryotic cell kinase (in vitro c-src and epidermal growth factor receptor protein kinases have been shown to be able to phosphorylate CagA 5; however, we do not yet know which kinase is precisely involved in vivo). Once phosphorylated, CagA is likely to bind an Src homology (SH)2-containing protein (SHC, phosphatidylinositol 3 kinase, and Nck are possible candidates). This complex can activate multiple pathways: (a) it may bind directly to N-WASP (neural Wiskott-Aldrich syndrome protein) and activate it to bind the Arp2/3 actin nucleator, thus stimulating actin polymerization and pedestal formation, as is the case of the Shigella flexneri IcsA 12. (b) Alternatively, the CagA–P-SH2 protein complex may activate the Rho family of small, GTP-binding proteins (Cdc42, Rac, or Chp), which control the organization of the actin cytoskeleton. This pathway could also cause actin polymerization and pedestal formation by activating N-WASP 13. (c) The CagA–P-SH2 protein complex may trigger a signaling cascade, possibly via the MAP pathway, which may induce transcription of nuclear genes. Although there is plenty of evidence that actin polymerization occurs and therefore one or both of the mechanisms mentioned in (a) or (b) must be activated by CagA–P, there is not yet any evidence for the pathway mentioned in (c). However, the increased frequency of gastric cancer in patients infected by cag + H. pylori strains and the H. pylori–dependent MALT (mucosal-associated lymphoid tissue) lymphoma suggest that nuclear signaling is a possible mechanism of pathogenesis. Whether the delivery of CagA into host cells facilitates antigen presentation, thus explaining why CagA is the immunodominant antigen of H. pylori, is an intriguing possibility that deserves further investigation.

Table 2

List of Genes and Crossed Nomenclature of the cag Pathogenicity Island of H. pylori

Nomenclature of cag genes according to			Properties of single gene inactivation in the cag region
Tomb et al.	Akopyants et al.	Censini et al.	IL-8 secretion	NF-κB activation	CagA–Ptyr translocation
ND	ND	G27wt,cag+	+	+	+
ND	ND	G27Δcag	−	−	−
HP0520 cag1	ORF6	cagζ	ND	ND	ND
HP0521 cag2	ORF7	cagε	−	ND	ND
HP0522 cag3	ORF8	cagδ	ND	ND	ND
HP0523 cag4	ORF9	cagγ	ND	ND	ND
HP0524 cag5	ORF10	cagβ(virD4)	++	ND	−
HP0525	ORF11	cagα(virB11)	−	ND	−
HP0526 cag6	ORF12	cagZ	ND	ND	ND
HP0527 cag7	ORF13	cagY(virB10)	−	ND	−
HP0527 cag7	ORF14	ND	−	ND	ND
HP0528 cag8	ORF15	cagX(virB9)	−	ND	−
HP0529 cag9	ORF16	cagW(virB8)	−	ND	ND
HP0530 cag10	ORF17	cagV	ND	ND	ND
HP0531 cag11	ORF18	cagU	−	ND	ND
HP0532 cag12	ORF19	cagT(virB7)	−	ND	ND
HP0533	ND	ND	ND	ND	ND
HP0534 cag13	ORF20	cagS	−	ND	ND
ND	ORF21	tnpA	ND	ND	ND
ND	ORF22	tnpB	ND	ND	ND
HP0535 cag14	ND	cagQ	ND	ND	ND
ND	ND	cagR	ND	ND	ND
HP0536 cag15	ND	cagP	ND	ND	ND
ND	ND	cagO	ND	ND	ND
HP0537 cag16	ND	cagM	−	−	−
HP0538 cag17	ND	cagN	+	+	+
HP0539 cag18	ND	cagL	−	−	−
HP0540 cag19	ND	cagI	−	−	−
HP0541 cag20	ND	cagH	−	−	−
HP0542 cag21	ND	cagG	−	−	−
HP0543 cag22	ND	cagF	−	+	−
HP0544 cag23	ND	cagE(virB4), picB	−	−	−
HP0545 cag24	ND	cagD	−	ND	−
HP0546 cag25	ND	cagC	−	ND	−
ND	ND	cagB	−	ND	−
HP0547 cag26	ND	cagA	+	+	−
HP0548	ND	cagΩ	ND	ND	ND
HP0549 glr	ND	glr	ND	ND	ND

List of genes and crossed nomenclature of the cag pathogenicity island of H. pylori and of the effects on IL-8 secretion, NF-κB activation, and CagA translocation and tyrosine phosphorylation after single gene inactivation. Type IV homologues are indicated in bold type in parentheses. The CagE gene was also independently identified by Tummuru et al. as picB (reference 20). The wild-type G27 strain and its mutant derivative Δcag are positive and negative controls.

Figure 2

After type IV contact and CagA translocation, two independent signaling pathways are induced in the host cell.

Tyrosine Phosphorylation of Injected Bacterial Proteins Is an Emerging Signaling Mechanism.

Before CagA, another protein was described as injected into mammalian cells and tyrosine phosphorylated by eukaryotic cell kinases. This is the translocated intimin receptor (Tir) of enteropathogenic Escherichia coli that, after injection and tyrosine phosphorylation, serves as a receptor for “intimin” adhesin (Fig. 3; reference 14). CagA is the second example of a bacterial protein that is shown to be injected into eukaryotic cells and then tyrosine phosphorylated by eukaryotic cell kinases. During Chlamydia infection, it has been found that eukaryotic cells can also phosphorylate a bacterial protein at serine/threonine residues 15. The discovery in different bacteria of this new type of cell intoxication mechanism suggests that many more examples will emerge in the near future.

Figure 3

Type III and type IV secretion systems functionally converge. Enteropathogenic E. coli and H. pylori (a) inject tyrosine-phosphorylated effector proteins by type III or type IV engines, respectively. (b) Tir is the EPEC receptor for intimin (binary products of the LEE [locus of enterocyte effacement] pathogenicity island). CagA is a Helicobacter-translocated molecule. (c) After phosphorylation on a tyrosine residue by a host cell kinase, cortical actin polymerization and pedestal protrusion are induced. Both microorganisms promote similar cellular responses.

The use of phosphotyrosines by bacterial pathogens is not surprising; pathogenic bacteria are known to target most crucial regulatory circuits of the eukaryotic cell. Large and small GTP-binding proteins that act on most of the cellular signal transduction pathways are targets of many classical toxins and toxins injected by the type III secretion system. Now that bacterial virulence factors have been shown to also target tyrosine phosphorylation, the other major cellular signaling mechanism, we can probably conclude that bacteria manage to target virtually all key regulatory circuits of eukaryotic cells. It is interesting that in this case the targets are not the key enzymes such as kinases or phosphatases, which would have a more pleiotropic effect. Rather, the action is very precise: bacteria inject molecules that appear as host proteins but that act as Trojan horses containing a bacterial hidden core message that allows the microorganism to take control over the host cell. Surprisingly, vaccinia virus exploits similar mechanisms of forced actin polymerization—mediated by tyrosine phosphorylation of the viral protein A36R—to spread from cell to cell 16.