Helicobacter pylori CagA is the first bacterial oncoprotein to be identified in relation to human cancer. CagA is delivered into gastric epithelial cells through a bacterial type IV secretion system and localizes to the plasma membrane, where it undergoes tyrosine phosphorylation by host cell kinases. Membrane-localized CagA then mimics mammalian scaffold proteins and perturbs a number of host signaling pathways in both tyrosine phosphorylation-dependent and -independent manners, thereby promoting transformation of gastric epithelial cells. Helicobacter pylori CagA is noted for structural diversity in its C-terminal region, with which CagA interacts with numerous host cell proteins. This CagA polymorphism is primarily due to differential combination and alignment of the four distinct EPIYA segments and the two different CagA-multimerization sequences in making the C-terminal region. The structural diversity substantially influences the pathophysiological action of CagA. This review focuses on the molecular basis for the structural polymorphism that determines the degrees of virulence and oncogenic potential of individual CagA. The pylogeographic distribution of differential CagA isoforms is also discussed in the context of human migration history, which may underlie large geographical variations in the incidence of gastric cancer in different parts of the world. (Cancer Sci 2011; 102: 36–43)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Gastric carcinoma is the second most common cause of cancer-related deaths (700 000 deaths/year), accounting for approximately 10% of total cancer deaths.(1) Recent studies have uncovered an etiologic relationship of Helicobacter pylori, which inhabits the stomach of approximately half of the world’s human population, to gastric carcinoma.(2,3) Much attention has been focused on the oncogenic role of H. pylori CagA (cytotoxic antigen gene A), a highly immunogenic 120∼145-kDa protein encoded by the cagA gene.(4–6) Despite its role in H. pylori virulence, cagA on its own is not an essential gene and is thought to have been introduced into the H. pylori genome as a component of the cag pathogenicity island (cag PAI) through horizontal transfer.(7) Consistently, H. pylori can be subdivided into cagA-positive and cagA-negative strains: in East Asian countries, virtually all of the H. pylori isolates are cagA-positive, whereas approximately 30–40% of H. pylori isolates in Western countries are cagA-negative.
Several genes in the cag PAI encode a bacterial microsyringe structurally related to the Agrobacterium tumefaciens type IV secretion system (TFSS).(7) Like the A. tumefaciens TFSS, which mediates translocation of the oncogenic T-DNA and associated proteins into plant cells, H. pylori TFSS delivers the CagA protein into gastric epithelial cells.(8–13) Although the detailed mechanism underlying CagA delivery remains largely unknown, a recent study has shown that the interaction of CagL, a pilus component of H. pylori TFSS, with integrins α5/β1 in an Arg-Gly-Asp (RGD) sequence-dependent manner, is crucial for the translocation of CagA into host epithelial cells.(14) Another study indicated that the binding of H. pylori CagY and CagA to β1-integrin in an RGD-independent manner causes structural changes in integrin and that these structural changes play an important role in the delivery of CagA into host cells.(15) More recently, direct contact of H. pylori with epithelial cells has been reported to provoke rapid and transient exposure of phosphatidylserine (PS) to the outer leaflet of the plasma membrane. The surface-exposed PS specifically interacts with CagA, which is placed on top of the bacterial TFSS pilus. The CagA–PS interaction triggers the process of CagA delivery into cells, which also requires active host machineries.(16)
Following delivery into gastric epithelial cells, CagA interacts with a number of host cell proteins and perturbs multiple signaling pathways, thereby contributing to the transformation of cells.(5,6) Indeed, the oncogenic potential of CagA was directly proven by the observation that systemic expression of CagA in mice induces gastrointestinal and hematological malignancies.(17,18)
EPIYA Motif and EPIYA Segment
Once inside, CagA is localized to the plasma membrane, where it undergoes tyrosine phosphorylation initially by Src family kinases and then by c-Abl.(19) Tyrosine phosphorylation sites of CagA are characterized by the unique 5-amino-acid sequence (Glu-Pro-Ile-Tyr-Ala), termed the EPIYA motif, which is present in variable numbers in the C-terminal CagA region (hereafter denoted as the EPIYA-repeat region) (Fig. 1).(20) The majority of the CagA isolates contain three EPIYA motifs, although the number can vary from one to seven among CagA species.
Based on amino acid sequences flanking the EPIYA motif, four distinct peptide segments, EPIYA-A, EPIYA-B, EPIYA-C and EPIYA-D, have been defined (Fig. 1).(5) The boundaries that separate each of the EPIYA segments have been determined through specific recombination sites in the cagA sequence that encodes the EPIYA segments.(21,22) Among these, the EPIYA-A and EPIYA-B segments are present in almost all CagA isolates. In contrast, the EPIYA-C segment is characteristic of CagA of H. pylori distributed in Europe, North America, and Australia. Thus, CagA species containing the EPIYA-C segment is denoted as “Western CagA”. Likewise, the EPIYA-D segment is specific to CagA of H. pylori endemically circulating in East Asian countries. Accordingly, CagA with the EPIYA-D segment is termed “East Asian CagA”. There is no report of the co-existence of EPIYA-C and EPIYA-D segments in a single CagA protein. The EPIYA motif is highly preserved and only a small fraction of CagA isolates bear modified EPIYA sequences such as EPIYT (5.1% of 1796 EPIYA motifs investigated), ESIYA (1.3%), and ESIYT (0.4%).(23) Of note, most of the changes in the EPIYA motif are found in the EPIYA-B segment of Western CagA.
Configuration of the EPIYA-repeat Region
The EPIYA-repeat region of CagA is made of tandem alignment of the four distinct EPIYA segments (A, B, C, D) in various combinations (Fig. 1).(5,24) In typical Western CagA, the region is composed of EPIYA-A, EPIYA-B, and EPIYA-C segments one after another (EPIYA-ABC), whereas in the case of prevalent East Asian CagA the region is composed of EPIYA-A, EPIYA-B, and EPIYA-D segments in that order (EPIYA-ABD). Interestingly, the EPIYA-C segment variably duplicates among distinct Western CagA isolates (Fig. 2).(5,21) More than 60% of Western CagA proteins shows the EPIYA-ABC configuration (EPIYA-ABC; 66.5%), followed by the EPIYA-ABCC configuration (EPIYA-ABCC; 20.3%), then the EPIYA-ABCCC configuration (EPIYA-ABCCC; 4.0%).(23) In contrast to the EPIYA-C segment, the EPIYA-D segment rarely duplicates and thus most, if not all, of the East Asian CagA isolates contain a single EPIYA-D segment (98.8%), which in many cases is present in the EPIYA-ABD configuration (83.6%).(23) A small fraction of Western and East Asian CagA isolates display more complicated mosaic structures for alignment of the EPIYA segments (Fig. 2).(24) Mechanistically, the structural diversity in the EPIYA-repeat region was most probably generated through an intrachromosomal homologous recombination in the cagA gene or homologous recombination between two cagA genes on sister chromosomes.
In contrast to the highly divergent nature of the C-terminal EPIYA region, N-terminal CagA is well conserved among CagA proteins. The N-terminal CagA region is required for the membrane association of CagA in polarized epithelial cells.(25) This is in contrast with the observation that the C-terminal EPIYA segment plays a role in membrane association of CagA in non-polarized epithelial cells.(22) As noted above, CagA–PS interaction is critically involved in the delivery of CagA and the N-terminal region contains a sequence that specifically interacts with membrane PS.(16) More specifically, the CagA–PS interaction requires highly conserved and positively charged Arg residues (RxR) in the middle portion of CagA (R619 and R621 in the case of NCTC11637-derived CagA), which may directly bind to the negatively charged PS (Fig. 3). This CagA-PS interaction also plays an important role in membrane localization of the delivered CagA in polarized epithelial cells.(16)
Recognition of CagA by H. pylori TFSS requires the Arg-rich CagA-secretion signal sequence (20 amino acids) at the C-terminus, which shares a weak homology with the C-terminal sequences of other type IV-secreted proteins such as RalF, VirE2, andVirD2.(26) It has also been reported that the C-terminal region (∼100 amino acid stretch) adjacent to the CagA-secretion signal sequence interacts with a secretion chaperone-like protein, CagF, and that the CagA–CagF interaction allows the type IV secretion system to recognize CagA as an effector (Fig. 3).(27,28) However, the C-terminal tail is not sufficient for delivery of CagA because a small deletion of the N-terminal CagA also abrogates the type IV-mediated CagA translocation into the host cells.(26) Thus, a tertiary structure of CagA comprising the N-terminal and C-terminal regions may be important for the CagA recognition by TFSS. Alternatively, CagA might be recognized by TFSS through a two-step process, which sequentially requires the N-terminal and C-terminal CagA sequences.
Functional Roles of EPIYA Segments
A decisive role of tyrosine phosphorylation in the oncogenic potential of CagA was confirmed by results showing that transgenic mice expressing phospho-resistant CagA never developed tumors.(17) All of the EPIYA motifs in the four distinct EPIYA segments serve as tyrosine phosphorylation sites of CagA in host cells. Upon tyrosine phosphorylation, the EPIYA-C or EPIYA-D segment acquires the ability to specifically interact with SHP2 phosphatase, a bona fide oncoprotein, gain-of-function mutation of which is associated with a variety of human malignancies (Fig. 3).(5,20,29–31) SHP2 is required for full activation of the Erk MAPK pathway, which conveys a potent mitogenic signal.(32) SHP2 also potentiates cell motility by dephosphorylating focal adhesion kinase. Hence, CagA-deregulated SHP2 is involved in the induction of an extremely elongated cell shape known as the hummingbird phenotype.(33,34) Intriguingly, East Asian CagA binds more strongly to SHP2 and more potently induces the hummingbird phenotype than does Western CagA. Among Western CagA species, those with a greater number of EPIYA-C segments display a stronger ability to bind to SHP2 and thereby more potently induce the hummingbird phenotype than do those with less EPIYA-C segments.(24)
The tyrosine-phosphorylated EPIYA-A or EPIYA-B segment acquires the ability to interact with the C-terminal Src kinase (Csk) (Fig. 3).(35) Interaction of CagA with Csk stimulates the kinase activity of Csk, which in turn inhibits Src, the kinase that phosphorylates CagA. The CagA–Csk interaction is therefore considered to create a negative feedback loop that constrains the phosphorylation-dependent CagA activity below a certain threshold to ensure long-term colonization of H. pylori in the stomach without causing fatal damages.(24,35) The CagA proteins with more EPIYA-A and EPIYA-B segments more strongly inhibit Src and thus more efficiently attenuate tyrosine phosphorylation-dependent CagA activities than do those with less EPIYA-A and EPIYA-B segments.(24) A recent proteomic screening showed that, in addition to SHP2 and Csk, the CagA EPIYA segments can interact with SHP1, Grb2, Grb7, PI-3 kinase, and Ras-GAP1 in a tyrosine phosphorylation-dependent manner.(36) The potential of the EPIYA segments to interact with a variety of distinct SH2-containing proteins suggests that CagA behaves quite differently from mammalian SH2 domain-interacting phosphoproteins, which usually display a high selectivity toward a single SH2 domain. The notion led to the hypothesis that H. pylori CagA acts as a pathogenic “master key” that manipulates multiple signaling pathways in mammalian cells.(37) Nevertheless, pathophysiological relevance for the interaction of CagA with these proteins needs further investigation. For instance, Grb2 has been reported to interact with CagA in a phosphorylation-independent manner and thereby cause activation of Ras-Erk signaling.(38) CagA has also been reported to bind to Crk adaptor proteins in an EPIYA phosphorylation-dependent manner, although the proteomic screening failed to identify Crk.(39) The CagA–Crk interaction is involved in the morphogenetic activity of CagA as well as disruption of adherens junctions.
In non-polarized cells, the EPIYA segment also plays an important role in the plasma membrane association of CagA.(22) The CagA membrane association requires at least one EPIYA segment (any of EPIYA-A, -B, -C, or -D) but is independent of EPIYA tyrosine phosphorylation. Furthermore, whereas the EPIYA motif is indispensable, it is not sufficient for the membrane localization of CagA. This fact indicates that sequences flanking the EPIYA motif also contribute to the interaction between CagA and the plasma membrane in non-polarized cells.
CagA-multimerization (CM) Sequence
The CM sequence is a 16-amino acid stretch located immediately distal to the EPIYA-C or EPIYA-D segment (Fig. 3). The sequence was originally identified as the sequence that mediates CagA multimerization (dimerization) in host cells.(40) CagA dimerization is important for stable interaction with SHP2 and is also crucial for efficient induction of the hummingbird phenotype. The CM sequence was subsequently found to coincide with the sequence that mediates the interaction of CagA with partitioning-defective 1 (PAR1)/microtubule affinity-regulating kinase (MARK), an evolutionally conserved serine/threonine kinase that regulates cell polarity by controlling microtubule stability.(41) In mammals, PAR1/MARK comprises four homologues (PAR1a/MARK3, PAR1b/MARK2, PAR1c/MARK1, and PAR1d/MARK4), which redundantly act to establish and maintain apical-basolateral polarity in epithelial cells.(42) Interaction of CagA with these PAR1 members inhibits PAR1 kinase activity and thereby causes junctional and polarity defects.(41,43–45) Through determination of the crystal structure of PAR1b/MARK2 in complex with a CagA peptide (120 amino acids) that spans the EPIYA-repeat region as well as the CM sequence, the first 14-residue peptide of CM (termed MKI for MARK2 kinase inhibitor) has been shown to directly interact with the kinase catalytic domain of MARK2.(46) Why then was the PAR1 binding sequence first identified as a sequence required for CagA multimerization (dimerization)? As PAR1 exists as a homodimer, it is most likely that two CagA molecules interact with a PAR1 dimer through the CM sequence and thereby passively dimerize in the presence of the PAR1 dimer. Recently, disruption of epithelial polarity by CagA-PS interaction was found to activate RhoA, which suppresses accumulation of p21 cyclin-dependent kinase inhibitor.(47) This RhoA activation enables CagA to provoke unconstrained mitogenesis by aberrantly activating Erk signaling, which, in the absence of active RhoA, triggers oncogenic stress that induces p21 accumulation and subsequent cell senescence/apoptosis.(47) Because PAR1 is involved in the regulation of microtubule stability, CagA–PAR1 interaction also impairs the function of microtubules in the mitotic spindle.(48) Consequently, CagA provokes a delay in prophase–anaphase transition during mitosis and thereby induces chromosomal instability, which potently increases the chance of cell transformation. CagA has also been reported to disrupt the E-cadherin/β-catenin complex in a CM sequence-dependent manner.(49,50) This finding suggests that the CagA-PAR1 complex directly or indirectly associates with E-cadherin (possibly, as constituents of a super protein complex) and thereby destabilizes the physical interaction of E-cadherin with β-catenin. Consistent with this, CagA has recently been reported to form a multiprotein complex with E-cadherin and p120 catenin by way of c-Met hepatocyte growth factor receptor.(51) It has also been reported that the CM sequence (termed “CRPIA” in the report, for conserved repeat responsible for phosphorylation-independent activity) interacts with c-Met and mediates activation of PI3K-Akt signaling that leads to upregulation of Wnt/β-catenin- and NF-κB-dependent transcriptions. It would be interesting to know whether CagA directly interacts with c-Met through CM, or whether the interaction is mediated through PAR1/MARK.(52) The CagA–c-Met interaction also contributes to the elevated cell motility by CagA, although H. pylori-induced motogenic response does not necessarily require c-Met.(53,54)
The CM sequence of Western CagA (W-CM) shows approximately 70% identity with that of East Asian CagA (E-CM).(43) Surprisingly, the N-terminal 16-amino acid sequence of the EPIYA-C segment was found to be exactly the same as the W-CM sequence (Fig. 4a).(40) Hence, Western CagA proteins carry at least two W-CM sequences, the number of which increases in parallel with the number of EPIYA-C segments. Consequently, the ability of Western CagA to bind PAR1 is proportional to the number of W-CM sequences. Meanwhile, a single E-CM sequence is almost twice as strong as a single W-CM sequence in PAR1 binding. Accordingly, East Asian CagA carrying one E-CM (CagA-ABD) and Western CagA carrying two W-CMs (CagA-ABC) display comparable levels of PAR1 binding activity.(44) Several CagA proteins show variations in CM sequences (Fig. 4b). For instance, two Western CagA species isolated in Central America contain two E-CM sequences.(55) An East Asian CagA bears a combination of one W-CM sequence and two E-CM sequences.(56) Conversely, several Colombian H. pylori strains carry Western CagA with a combination of one W-CM sequence and one E-CM sequence.(57)
Geographic Variation in CagA Structure
Transmission of H. pylori occurs predominantly within families, mostly from mother to infants. In the long run, the quasi-vertical transmission leads to colonization by genetically distinct H. pylori subpopulations in different human populations. In terms of CagA polymorphism, the vast majority of H. pylori isolates in East Asian countries bear East Asian CagA. In contrast, most if not all H. pylori cagA-positive strains isolated in non-East Asian countries carry Western CagA (Fig. 5). A clear exception is Southeast Asia, where H. pylori strains carrying East Asian CagA and Western CagA co-exist with various ratios in different areas/countries.(58–61) Notably, the incidence of H. pylori infection is extremely low in Java and Malaysia compared to the incidence in the rest of the world.(62,63) Based on this finding, Graham et al.(64) postulated that ancestral Javanese and Malaysian passed through a population bottleneck in which H. pylori infection/transmission was effectively prevented or a significant percentage of individuals infected with H. pylori were negatively selected. After the bottleneck, East Asian CagA-carrying H. pylori and Western CagA-carrying H. pylori may have been introduced into Southeast Asian people through migrations of ethnic Chinese and ethnic Indian people, respectively. Another interesting observation is that a substantial fraction of H. pylori strains isolated in Okinawa, the southmost prefecture of Japan consisting of more than 100 small islands, carry Western CagA with the EPIYA-C segment.(65,66) Geographically, Southeast Asia is much closer to Okinawa than other areas of Japan. Also, Okinawa was under administration by the USA for 27 years after the end of World War II until 1972, and there are still large US military bases in Okinawa. This raises the idea that Western H. pylori strains circulating in Okinawa originated from Southeast Asia or were introduced by recent Western influence. However, the Western CagA isolates in Okinawa are phylogenetically separable from the major Western CagA cluster.(66) As all of the Western CagA proteins isolated in Southeast Asian countries such as Vietnam, Thailand, and the Philippines belong to the major Western CagA cluster, the CagA variant termed “J-Western CagA” appears to be peculiar to Okinawa.(59,61) This finding argues against the idea that Western CagA in Okinawa was derived from Southeast Asia or Western countries. Drawing an analogy with the model proposed by Ishida and Hinuma(67) that explains the distribution of human T lymphotropic virus type 1 (HTLV-1) carriers in Japan, it is tempting to hypothesize that H. pylori strains carrying J-Western CagA had been accompanied by Japanese aboriginal people, known as Jomon people, who were the earliest inhabitants of the islands of Japan more than 10 000 years ago. Approximately 2500–1500 years ago, Yayoi people, who were the ancestors for most of the present Japanese population, moved from northern Asia through the Korean Peninsula to the central part of Japan (Fig. 6a). This movement pushed Jomon people away to the northern or southern periphery of Japan such as Hokkaido and Okinawa. As Yayoi people, who originated from northern Asia, were most likely to have carried H. pylori with East Asian CagA, the prehistoric racial movement could explain the enigmatic distribution of two distinct CagA species, East Asian CagA and J-Western CagA, in Japan. Notably, the incidence of gastric carcinoma is significantly lower in HTLV-1 carriers than in HTLV-1 non-carriers in Japan.(68) It would be interesting to know if H. pylori carrying J-Western CagA is also found in Ainu people, northern descendents of the Jomon people inhabiting Hokkaido, who share several physical and genetic traits in common with ethnic Okinawa people.
It is thought that H. pylori accompanied humans when they crossed the Bering Strait from Asia to the New World.(69) Consistent with this hypothesis, several H. pylori isolates from indigenous Colombian people carry East Asian-type CagA.(70) Curiously, however, the vast majority of H. pylori isolated from indigenous Americans carry Western CagA, even though they were living in areas where the influence of modern Western culture was almost negligible.(70,71) The seemingly controversial observation is currently explained by rapid extinction of H. pylori carrying East Asian CagA as a result of expansion of those carrying Western CagA that came with European and African immigrants after the 15th century. However, the possibility that a fraction of the ancient East Asian aboriginal peoples (such as Jomon people) carried H. pylori strains with Western CagA raises an alternative model (Fig. 6b). In this model, prehistoric Asian peoples carrying H. pylori with Western CagA and those carrying H. pylori with East Asian CagA migrated to North America and then to South America by turns. Subsequent founder effects, geographic separation, and differential efficiency of colonization in the human stomach may have resulted in predominant inhabitance of H. pylori with Western CagA in North America and South America, long before the settlement of European and African people carrying H. pylori with Western CagA, after the 15th century.
Pathophysiological Relevance of CagA Diversity
The potential of H. pylori CagA to perturb host cell functions is determined by the degree of CagA interaction with target molecules represented by SHP2 and PAR1. This in turn indicates that the diversity in the combination of EPIYA segments influences the strength of the pathological activity of individual CagA. Variation in the CM sequence also contributes to the degree of CagA interaction with cellular targets. Consequently, structural polymorphism in the EPIYA-repeat region of CagA is an important determinant of the virulence of cagA-positive H. pylori. In fact, the incidence of gastric carcinoma co-relates quite well with the geographic distribution of H. pylori carrying East Asian CagA, making Japan, Korea, and China the top three in ranking.(72) Consistently, transgenic mice expressing East Asian CagA developed tumors more efficiently than did those expressing Western CagA.(18) Several clinical studies have also shown that H. pylori strains with East Asian CagA are more virulent and are more intimately associated with gastric carcinoma than are those carrying Western CagA.(65,73) Among Western CagA-carrying H. pylori strains, those possessing CagA with multiple EPIYA-C segments are more likely to cause gastric carcinoma than are those having CagA with a single EPIYA-C segment, the number of EPIYA-C segments thus being an important risk factor for gastric carcinoma in Western populations.(74,75)
A number of studies have addressed the connection of a particular CagA isoform to a specific disease such as gastric carcinoma. However, it should be noted that CagA polymorphism contributes to the level of perturbation by individual CagA of host cell signaling, malfunctioning of which can cause broad clinical manifestations ranging from gastritis to peptic ulceration to gastric carcinoma. This argues against the simple view that a certain CagA subtype specifies a particular disease. Development of gastric carcinoma involves multiple factors, including degree of H. pylori virulence, host genetic traits, and environmental factors. Notably, however, some of those factors may synergize with H. pylori in gastric carcinogenesis. For instance, high salt intake destroys the gastric mucous layer, which acts as a physical defense protecting gastric epithelial cells against direct attack by H. pylori.(76) Hence, a high salt diet, a known risk factor for gastric carcinoma, may promote cell transformation by increasing the opportunity for H. pylori to deliver CagA.
It seems that humans had already been infected with H. pylori long before their migration out of Africa.(77) Until modern civilization, however, gastric carcinoma must have been very rare among H. pylori carriers because ancient people mostly died from acute infections or injuries at young ages. This means that the primary purpose for H. pylori to have CagA is not to give rise to gastric carcinoma. Instead, the bacterium may have developed a symbiotic relationship with humans by making use of CagA. Probably, a key observation is that CagA inhibits PAR1 to elicit junctional and polarity defects and at the same time deregulates Erk signaling that elicits oncogenic stress, thereby causing p21 accumulation and subsequent premature senescence/apoptosis in non-polarized epithelial cells.(47) Consequently, infection with cagA-positive H. pylori more or less suppresses gastric acid secretion by reducing the number of gastric epithelial cells, which is advantageous for long-term colonization of H. pylori in the stomach. This H. pylori activity could have helped ancient people by functioning as natural “antacids”. In particular, East Asian ancestral people, who traditionally had an agricultural life, might have benefited from infection with more virulent H. pylori because excess acid secretion was unfavorable in an agrarian society. In contrast, Western ancestral people, who were hunters, might have favored less virulent H. pylori because their eating habits needed a certain level of gastric acid secretion.
Given the above-described scenario, oncogenic potential of CagA may be a secondary effect that has become evident with expansion of the human lifespan. Chronic infection with H. pylori cagA-positive strains hyper-stimulates gastric epithelial turnover by constitutively exposing cells to oncogenic stress.(47) Long-term sustenance of such a situation substantially increases the chance of epithelial cells to acquire genetic/epigenetic defects in signaling pathways including those involved in senescence/apoptosis, the malfunctioning of which is an important hallmark of cancer.(78)Helicobacter pylori-mediated delivery of CagA into cells with such abnormalities will selectively propagate a pool of cancer-prone descendant cells. Multiple rounds of expansion of more cancer-predisposed cells by CagA should greatly facilitate multistep gastric carcinogenesis.