medical importance of the pathogen h pylori

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.


Although much has been learned about specific H. pylori virulence factors, little is currently understood about why some H. pylori infected individuals progress to develop gastric cancer while others remain asymptomatic. The goal of this thesis was to better understand the association between different polymorphisms in CagA and VacA and disease outcome. Specifically, we showed that East Asian CagA (EPIYA-ABD) was linked to progression to gastric cancer in a South Korean population (39). In fact, all H. pylori strains from cancer patients expressed and delivered phosphorylatable CagA to host cells, whereas the presence of the cagA gene did not strictly correlate to expression and delivery of CagA with non-cancer strains (39). Our study was the first to statistically link a specific cagA allele to gastric cancer development. We next examined the role of VacA polymorphisms within that population, and found that while the distribution of vacA alleles was not directly associated with disease state, it was associated with the distribution of cagA alleles. Furthermore, the vacA allele was associated with the cagA allele and disease state. Next, we were able to analyze the contribution of the newly described i region of VacA to disease development. To this end, we identified an amino acid (196) that was important for development of gastric cancer. We were also able to identify some associations that were CagA-dependent, such as the association of VacA and disease state in the EPIYA-ABD population as well as amino acid distribution at position 231 and disease state in the non EPIYA-ABD population. In addition, in the process of completing this thesis, we were able to optimize techniques that will ultimately be used to characterize CagA isogenic strains. Those future studies will help to elucidate the role of the EPIYA motifs in H. pylori-induced host cell damage both in vitro and in vivo. En masse, the data presented herein add to what we know about the complexity of H. pylori-induced pathogenesis. Overall, it is becoming increasingly more evident that polymorphisms within CagA and VacA, alone and in concert, affect H. pylori-induced disease. However, the reason why only a portion of the population develops gastric cancer still remains unclear. Other bacterial virulence factors, as well as multiple host, dietary, and environmental factors have been indicated as participants in H. pylori-induced disease. Clearly, further study is required to determine which factors are involved and what role they have in the development of H. pylori-induced gastric cancer.

Unanswered Questions Stemming from the Epidemiological Studies


A key question that should be addressed in the future is the difference in the degree of CagA variation between Western and East Asian strains. Western isolates vary widely in the number of EPIYA-C motifs that are present (7, 8), whereas there is a distinct lack of variation in East Asian strains. In fact, one study examining Gen Bank sequences of 500 East Asian strains, found that 441 (88.2%) contained a canonical EPIYA-ABD motif (7). Indeed, additional studies confirmed this conservation among East Asian strains; greater than 84% of the examined strains across all three studies contained an EPIYA-ABD motif (7, 12, 39). Moreover, in our molecular epidemiologic CagA study, the majority of strains (87.5%) contained an EPIYA-D motif. Interestingly, those isolates that contained a nonstandard EPIYA-ABD motif were associated with development of gastritis (39). This suggests that variation in East Asian cagA is not as favorable as in Western isolates and that variation may affect disease progression.

The reasons for strict conservation of the EPIYA-ABD motif are unknown, but may be explained by several possible theories. One theory to explain this conservation could be that there is a difference in the degree of selective pressure for variation imposed on Western and East Asian strains. Western CagA shows a lower affinity for SHP-2 and is associated with less severe inflammation, host cell morphological changes, and disease as compared to East Asian strains (30, 31). Moreover, it has also been demonstrated that there is a dose response in the number of EPIYA-C motifs and the levels of tyrosine phosphorylation, SHP-2 binding, host cell morphological changes, and inflammation induced by Western CagA (32, 39, 87). Perhaps, increased inflammation enhances colonization, and therefore may act as a positive selective pressure to increase the number of EPIYA-C motifs. This pressure would not be experienced by East Asian strains since the EPIYA-D motif is already so biologically active. However, if increased inflammation is important for colonization of H. pylori, then there may be a selective pressure to keep a canonical EPIYA-ABD motif. For instance, perhaps extra EPIYA-A or -B motifs in association with an EPIYA-D motif more strongly activate the negative feedback loop that results from EPIYA-A or -B binding to Csk, thereby decreasing inflammation (7, 82). Additionally, a single EPIYA-D motif may be optimal for SHP-2 binding, and extra EPIYA-D motifs may contort CagA's conformation, thereby destabilizing this interaction, again resulting in a decrease in inflammation. The true reason for this conservation among East Asian strains could help elucidate the impact of these motifs.

Work on the EPIYA motif region of CagA has primarily focused on differences in phosphorylation and subsequent modulation of phosphorylation-dependent host signaling pathways (reviewed in (40). Recently, however, a CagA multimerization domain was described that is located within the EPIYA region, and therefore varies as the EPIYA motifs vary. Some studies suggest that this domain is responsible for the differential modulation of some phosphorylation-independent host signaling pathways (Fig. 2; (59, 74, 77, 83). However, since the existence of this domain is a very recent discovery, more work is needed to clearly define the role of the multimerization domain on H. pylori-induced changes in host signaling pathways. Additionally, changes in this domain as a result of changes in the EPIYA motif will need to be investigated.

While we have gained much knowledge about the role of the C-terminus of CagA, not much is currently known about the N-terminus of the protein. Moreover, of the studies that have been completed, conflicting data has arisen; it has been demonstrated that the N-terminus of CagA is responsible for directing CagA to the plasma membrane (14), but other data demonstrated the EPIYA-motifs located in the C-terminus were responsible for proper localization (32, 33). Pelz et al. recently demonstrated that two independent domains, one in the N-terminus and one in the C-terminus, are responsible for directing CagA to the plasma membrane (69). In fact, these authors showed that the first 200 amino acids of CagA actually act as an inhibitory domain that dampens the host response to the C-terminus of CagA. This domain reduces activation of the oncogenic transcription factor β-catenin, reduces the length of the "hummingbird" protrusions, and increases the speed and strength of new cell to cell contact (69). Thus, an interesting question would be to ask if the activity of the inhibitory domain varies in conjunction with varying motifs in the C-terminus. In other words, is the effect of this inhibitory domain different based on the variations of the different cagA alleles? This question could be addressed by deleting the inhibitory domain within the context of isogenic strains that differ only in the EPIYA region of cagA. Differences in induced host cell morphological changes could then be assessed when these strains were used to infect AGS cells, and fold changes in the number of elongated cells could be calculated and compared between the strains and their isogenic strains lacking the inhibitory domain. If the difference in fold change was similar across the different EPIYA isogenic strains, this would suggest that this inhibitory domain acts independently of the EPIYA motif. If this inhibitory domain is influenced by the EPIYA motif that is present, it could also be interesting to determine whether there is also an interaction with the multermization domain. This could be addressed by making phosphorylation resistant mutants within the above mentioned strains; the tyrosine could be changed to an alanine or serine through site specific mutagenesis (5, 21, 32). While we know quite a lot about the contribution of CagA and the EPIYA motifs to H. pylori-induced disease, there are clearly more questions on the molecular level that need to be answered.


Many questions still remain to be addressed regarding the different vacA alleles. Included among these are questions concerning the signal (s) region of the protein. Geographic differences between the s1 region have been reported, and three subtypes have been identified: s1a, s1b, and s1c (11). Are these differences important for how VacA acts on host cells and/or its interplay with other virulence factors? Although this nomenclature is now seldom used, it would be interesting to determine if the different subtypes display any functional differences in activity, since the s region is responsible for most of the toxicity of the toxin (52-54, 72). This could be accomplished by creating isogenic strains that vary only in the s1 subtype or more simply by intoxicating eukaryotic cells with identical concentrations of purified VacA containing one of the three different subtypes. One could assess the ability of the various VacA toxins to cause vacuoles within host cells as well as induce apoptosis, which could be measured through activated TUNEL assays or by measuring the amount of activated caspase 3.

Another question that has arisen from our VacA epidemiology work concerns the overall importance of the middle (m) region of the protein. The first aspect of the m region that should be addressed is the association between the m region and gender. Our work was the first study to observe such an association (37), and begs the question of why this association exists? Females appear more likely to be infected with strains encoding the m2 vacA allele (37). Does this association exist in populations where the m2 allele is more prevalent, such as in regions of China (67) and Poland (49)? If it does exist, then there may be something physiologically different between the gastric environment of males and females. While what causes this difference may be unknown, there are many possibilities. For instance, is there a difference in the pH of the stomach acid that may consequently influence disease state? Is there a difference between the actual gastric epithelium between males and females? Minute differences in the thickness or composition of the mucus layer, which could in turn impact contact of H. pylori with the gastric epithelium, could affect the amount of toxin delivered to host cells. Are there differences in the amount or type of adherence receptors expressed in the gastric epithelium of males versus females? Furthermore, what affect does the endocrine system, more specifically changes in hormone levels, have on this process? Answers to these questions may help to explain the differences in distribution of the m1 vacA allele between men and women. Overall, the increased cellular tropism of the m1 vacA allele (66), the finding that the presence of the m1 region increases the risk for gastric cancer (79), and the fact that patients infected with H. pylori strains encoding for the m2 allele are more likely to be female (37) may explain why males are overall more likely to develop gastric cancer (reviewed in 76).

The second aspect of the m allele that should be examined is its contribution to the association between VacA and CagA, as well as between VacA, CagA, and disease state. Does the association between the m region and the cagA allele occur simply because the vacA regions that are responsible for more severe disease also co-vary among themselves? This does not seem likely in this population, since we only found the s1 allele and were still able to detect this association. Furthermore, is the three-way association between the m region, CagA, and disease state due to the fact that the m2 allele has a narrower cell tropism, thereby affecting the types of cells VacA can intoxicate? If this were the case, one would expect to see a direct correlation between the m region and disease state. However, this correlation is not seen within this South Korean population.

Several questions also still remain regarding the intermediate (i) region of VacA. The major difference between the i1 and i2 alleles is the addition of three polar amino acids (asparagine, histidine, and serine) in Cluster C (37, 38). Since both clusters B and C have been suggested to affect toxin activity, studies directed towards understanding the specific role of these amino acids in vacuolating activity would be interesting. In order to better examine this, an i1 allele could be genetically engineered through the addition of only these three amino acids. Perhaps, it would not be surprising if the addition of these amino acids decreased toxin activity, since it has been demonstrated that additional amino acids near the cleavage site in the s region decrease the ability of the toxin to integrate into the cellular membrane, which in turn decreases toxin activity (11, 53). Such a result might explain why the i region has been suggested to be a better predictor of disease.

Another aspect of the i region worth examining involves toxin evolution and the presence of the i3 allele. Is the presence of the i3 region really a snapshot of the evolution of the i region from one allele to another (i1 or i2)? It would be interesting to infect animals with H. pylori strains containing the i3 allele, and to sequence several recovered isolates after infection to see if the i3 allele has been replaced with either an i1 or i2 allele. This experiment should be performed with an i3 strain where cluster B is from an i1 strain and cluster C is from an i2 strain, as well as an i3 strain where cluster B is from an i2 strain and cluster C is from an i1 strain. Results from these experiments could indicate if there is an overall selective pressure to evolve in vivo, and if the original cluster sequences influences that evolution. In this same vein, it would be interesting to determine if there is a functional difference between the i3 allele and the i1 and i2 alleles. If so, is there a functional difference between i3 strains that contain cluster B from an i1 strain and cluster C from an i2 strain as compared to an i3 strain that contains cluster B from an i2 strain and cluster C from an i1 strain? Alternatively, is toxicity dependent on the sequence of an individual cluster?

Finally, it would be interesting to look at a population containing an increased percentage of i2 alleles in order to assess the distribution of amino acids at position 196 on the vacA allele as well as on disease state. Our work demonstrated that the distribution of amino acids found at this position was linked to more severe disease, specifically gastric cancer (38). In the South Korean population analyzed, all of the i2 alleles encoded for a leucine at amino acid 196 (38). Thus, perhaps it would be interesting to analyze a population where the i2 allele is more prevalent to see if this trend persists. If so, the amino acid found at this position may be partially responsible for a previous study that concluded that the i region was the best predictor of disease (75). Further studies could also investigate variation in the major amino acid differences seen between the i1 and i2 alleles. This might better indicate which amino acids are critical for toxin activity.

Cag/Vac Interaction

Recent studies have identified an association between the cagA allele and the vacA allele that appears to affect H. pylori toxicity and disease severity (37, 84, 89). Infection with H. pylori strains that encode for CagA and s1/m1 VacA result in highly active corpus gastritis (57), which is linked to the development of gastric cancer (55-57). Our study also found an association between the cagA allele and vacA allele, as well as a three way association between the distribution of the cagA and vacA alleles and disease state (37). Indeed, in our South Korean population, the majority of H. pylori strains carry the most toxic form of both VacA and CagA, and this may explain the high rate of severe gastric disease among the South Korean population (37).

Conventionally, one might think that both toxins concomitantly exert drastic effects on the same host cell. However, recent data suggests that the converse is true; the presence of both CagA and VacA may dampen the effect of each protein alone, possibly leading to increased survival of infected host cells (9). In fact, when both toxins are present, there is less VacA-induced apoptosis then when cells are intoxicated with VacA alone (9). Additionally, eukaryotic cells intoxicated with both toxins demonstrate less CagA-induced morphological changes as compared to cells intoxicated with CagA alone (9).

Since these results are still fairly new, there are many questions that remain to be answered. For instance, is there a direct interaction between CagA and VacA, or more likely, is this effect the result of activation of competing pathways by the two toxins (Fig. 2 and 3)? If there is a direct interaction between these two proteins, this could possibly be detected by performing pull-down assays. Does VacA somehow amplify the function of the newly identified inhibitory domain in the N-terminus of CagA (69)? Also, in thinking about the chronology of H. pylori infection, does the bacterium utilize the two toxins to increase the life span of the host cell and thus, to prolong infection? Indeed, it seems plausible that the most severe forms of gastric disease would result from long term infection of cells, and therefore, long term H. pylori-induced inflammation. In terms of CagA and VacA "interaction," does an order of events exist that is important for the resulting effects? For example, since VacA is secreted while CagA is injected by H. pylori, do cells first need to be intoxicated by one or the other toxin to see the protective correlate? This could be directly assessed by intoxicating eukaryotic cells with one toxin and then subsequently with the other toxin at various later time points, followed by analysis of the cells for apoptosis and for morphological changes. Additionally, is the damping effect of the toxins achieved by reaching a threshold of both toxins, or is the mere presence of any amount of the two toxins sufficient? This could be assessed by determining if there is a dose response to the toxins. For instance, eukaryotic cells could be intoxicated with increasing amounts of VacA and a set amount of CagA, or with a set amount of VacA and increasing amounts of CagA. Eukaryotic cells would then be analyzed for levels of apoptosis as well as morphological changes. Additionally, the order of toxin addition could be inversed, based on the findings from the above study assessing the importance of the order of intoxication.

Finally, considering CagA and VacA "interaction" in the context of our epidemiological data, what are the physiological consequences of intoxication with the different CagA or VacA variants, or combination of these different toxins? For instance, if a strain carries EPIYA-ABD CagA and s2/i2/m2 VacA, which typically shows no toxic activity, does it behave similarly to a strain that carries an EPIYA-ABD CagA but no VacA? Alternatively, is the combination of Western CagA and s1/i1/m1 VacA more or less lethal than East Asian CagA and s2/i2/m2 VacA? These and other allele based questions could be addressed by making VacA isogenic strains within the context of the CagA EPIYA isogenic strain background as discussed later in this chapter. Since evidence is increasing that the association between the cagA allele and the vacA allele impact disease development, I strongly believe that the future of pathogenesis studies in H. pylori will have to focus on the effect of combinations of these virulence factors.

A Hierarchy of Virulence Factors

Another emerging theory is that CagA may be the "master" virulence factor, and that other virulence factors or polymorphisms are important only in the context of which cagA allele is present (13, 37, 38). In fact, studies have found that in terms of gastric cancer, the cagA allele carried is the most important bacterial risk factor (15). Conversely, the i region of VacA is the best predictor for duodenal ulcers (15). Indeed, in our epidemiological studies, we found that different associations existed in populations carrying particular cagA alleles and that these associations were not found in populations encoding for a different cagA allele. For instance, when age and gender were taken into account, a two way association between the distribution of the vacA allele and disease state was found only within the EPIYA-ABD CagA population. Furthermore, non-s1/i1/m1 vacA alleles were associated with duodenal ulcers within the population carrying the East Asian EPIYA-ABD CagA, but with gastritis in the population carrying any other genotype of CagA (37). We also found that an association existed between disease state and amino acid 231 of the VacA i region, but only within the non EPIYA-ABD population (38). These findings again suggest that the effect of different virulence factors or polymorphisms within these virulence factors may be masked by which cagA allele is present. Indeed, this fact may help explain the vast amount of conflicting literature concerning the importance of these different virulence factors. Employing the statistical technique of meta-analysis to survey the epidemiological data in different geographic regions might help to shed light on some of these reported differences.

Lessons Learned from the Current Project

Unfortunately, the largest molecular component of my thesis project encountered problems since it was ultimately discovered that the isogenic cagA strains that were constructed actually contained secondary mutations. While we do not understand exactly how these mutations arose, some clues may come from thinking about the inherent genetic variability of the bacterium. For instance, H. pylori strains are naturally competent, which allows for the constant exchange of DNA. In fact, the bacterium has the ability to uptake new DNA in vivo, which creates a constant chance for genetic exchange since a host can be infected with multiple H. pylori strains (24, 25, 36, 43). This phenomenon has been well documented with one of the H. pylori reference strains (J99). When an archived isolate of J99 was compared to isolates from the same patient taken 6 years apart, there was a high level of genetic diversity (36). Collectively, the new isolates had lost up to 2.3% of the open reading frames compared to the archived J99 strain. Additionally these strains had gained DNA that was not found in the original J99 strain (36). Overall, natural competence has been proposed to contribute to the vast allelic diversity of the organism, and to help account for the considerable genetic variability (6-7%) that is seen between strains (4, 26, 50, 80). Furthermore, animal passage of strains has been shown to induce formation of large numbers of fragmented genes and repeated regions (22). H. pylori also has an increased rate of spontaneous mutations as compared to E. coli, with initial studies demonstrating that the spontaneous mutation rate in H. pylori was 10-7-10-8 (28, 35, 85). Again, this rate varies among strains and a rate as high as 3 x 10-4 has been observed (46). In fact, genetic polymorphisms seem to be normal between strains. In a study that examined the genetic sequence of a house-keeping gene (glmM) it was found that the sequence was unique in all the strains examined (44). Furthermore, this microdiversity has been observed in a number of other genes (1, 3, 4, 34, 65, 78), as well as within strains taken from the same patient (36). Additionally, H. pylori's genome contains multiple genes that are phase variable. Indeed, when a single reference strain was sequenced, up to 27 genes were identified that contained nucleotide repeats that could facilitate phase variation (51, 81); two examples of these genes that have been examined in more detail include fliP (41) and oipA (88).

In order to reduce the potential for genetic variation that could affect our experiments, the Merrell group has adapted certain lab protocols. For instance, when we create a mutant strain, we select a single colony of the mutant strain and then never utilize single colonies again. Bacteria are expanded as patches of cells from the freezer stock (-80°C) on antibiotic-supplemented horse blood agar plates for 36-42 hours, which is the minimal amount of time for growth. Bacteria are then expanded as lawns from these patches for about 20 hours on plates. The lawn is then used to inoculate 18 hour liquid starter cultures that are ultimately used to inoculate OD controlled experimental liquid cultures. All of these protocols are performed in an attempt to minimize the number of lab passages of the strain, and to make sure that if genetic variability occurs, it does so in the context of a population of cells. Furthermore, when feasible, we also create an independent biological isolate of all mutant strains.

In devising our isogenic strain study, we first created a ΔEPIYA strain, which was used as the strain background for all subsequent strains. Moreover, we followed the aforementioned lab protocols for expanding bacteria from freezer stocks, transformation, selection, and growth, which were designed to minimize the possibility of variation. Therefore, there was no reason to believe that these strains would contain secondary mutations. However, upon reflection on the project, there were several different points throughout the process when the data suggested that there might be a secondary mutation that would complicate our study. Sequence analysis as well as growth dynamics suggested that the strains were in fact isogenic. However, the first indication that the strains were not behaving as expected came with the quantification of CagA expression. Theoretically, all of the isogenic strains, with the exceptions of the ΔcagA and ΔEPIYA strains, should have expressed CagA at similar levels. However, up to a two-fold difference could be seen between the EPIYA-ABtCCCC and wild type strains (Fig. 16). Conversely, there were no major differences when CagA expression was compared to the restorant strain, which I believed to be a reasonable comparison; the restorant strain should be genetically identical to the 7.13 wild type strain, and had undergone the same genetic manipulation as the other isogenic strains. Minor differences in the restorant suggested that perhaps the genetic manipulation of the strain slightly altered the overall expression level of CagA. At the time, we considered this as no surprise; however, in hindsight, this result may have been the first indication that there were problems with the strains.

While CagA has not been shown to affect the adherence of H. pylori (5), any difference in adherence of strains to host cells would potentially alter the amount of CagA that could be translocated and phosphorylated, thereby changing the deregulation of host cell signaling pathways and potentially affecting development of gastric disease. In our studies, the adherence assay was the first assay that showed marked differences between the isogenic strains. Indeed, the restorant, EPIYA-ABtC, and wild type strains, which should have all adhered at similar levels, did not behave as expected; the restorant strain was 10-fold less adherent than the wild type or EPIYA-ABtC strains. When these preliminary results showed the lack of consistency between the restorant, EPIYA-ABtC, and wild type strains, we sought to rule out the potential confounding effect of slight differences in the growth phase of the different isogenic strains (Fig. 15). Re-examination of the growth curves suggested that at 18 hours some of the isogenic strains may have entered stationary phase, which could adversely affect the adherence of the bacteria to the AGS cells. We therefore repeated the assay using 12 hour liquid cultures to infect the AGS cells. We found that while the number of adherent bacteria was increased, the trends stayed the same, suggesting that there were secondary mutations within the strains (Fig. 18).

We next assessed the ability of the strains to deliver CagA to host cells, where it is phosphorylated by host cell kinases, thereby causing morphological changes within the cells. While there were slight differences in the peak phosphorylation time of CagA between the strains, the trends across the strains were the same. However, assessment of the five hour time point, which was a time point shown in a previous study to allow detection of high levels of phosphorylation (39), presented more evidence that there were secondary mutations in our strains. While increasing numbers of EPIYA-C motifs corresponded to increasing amounts of phosphorylated CagA, the three isogenic strains that contained only one EPIYA motif (EPIYA-ABtC, EPIYA-ABtD, and the restorant) showed similar levels of phosphorylation (Fig. 19). Unfortunately, these levels of phosphorylation were 2.5 times less than the amount of phosphorylated CagA from the wildtype strain (Fig. 19). This fact combined with the fact that the isogenic strains expressed approximately 70% of the amount of CagA that the wild type strain expressed, immediately suggested that there was something different among the isogenic strains. Additionally, the levels of phosphorylation did not translate into the expected changes in host cell morphology. Once again, the the wild type, EPIYA-ABtC, and restorant strains all behaved differently (Fig. 20) suggesting that there was a second site mutation within at least some of the strains.

Based on the long length of the experiments, the large-scale animal study was regretably started prior to the completion of the in vitro characterization of the isogenic strains. The animal study showed that there was a complete lack of differences in disease progression among Mongolian gerbils infected with the different isogenic strains. Moreover, there was a drastic difference in the pathology induced in animals infected with the wild type strain as compared to the EPIYA-ABtC strain. This fact alone suggests that there is something fundamentally different between these two strains that is not CagA-related. Knowing what we know now, ideally a small pilot experiment of eight to ten weeks duration with fewer numbers of animals should have been conducted before proceeding with the large scale animal study.

Additionally, all H. pylori infection groups progressed to gastritis and eventually gastric cancer at the same rate as gerbils infected with the ΔcagA strain (Fig. 22), which has been shown previously to cause no gastric cancer in this model (23). This led to the re-creation of the ΔEPIYA strain followed by the creation of a new restorant strain. This new restorant strain induced pathology similar to the wild type strain in gerbils six weeks after infection; these animals displayed dysplasia and were progressing to gastric cancer (Fig. 23). This result combined with all the inconsistencies in the in vitro work suggests that there was a second site mutation within the original ΔEPIYA strain; therefore, no correlations could be made and there was no reason to further characterize these strains.

Since that time, we have learned that strain 7.13 loses in vivo virulence after low number of lab passages (R. Peek, personal communication). Additionally, other work in our lab has shown drastic in vivo differences for strains that have been idenitically manipulated to delete the same gene. Knowing this now, it might have been prudent to measure the mutation rate of 7.13 prior to beginning our studies in order to determine whether this strain has a higher mutation rate than other strains of H. pylori.

Looking Forward

Since H. pylori strain 7.13 was too genetically unstable to use for these studies, is there a future for this project?  I believe that there is and would propose the following possibilities: 1) use strain G27 for in vitro characterization and 2) use strain PMSS1 for both in vitro and in vivo characterization. G27 is our lab's commonly used reference strain and has been shown to be fairly genetically stable. Currently, all of the CagA isogenic strains have been created in the G27 background, and these strains can now be used to determine the effect of the EPIYA motifs on host cell signaling. To this end, if I was completing these studies, I would again start by assessing growth kinetics of these strains, as well as analyzing expression of CagA via Western blot analysis. The Western blot analysis should identify any differences in CagA expression that could complicate future experiments. Next, I would look at interaction of the strains with host cells by measuring bacterial localization, adherence, internalization, CagA phosphorylation, and induced host cell morphological changes.  Though I would initiate these studies using AGS cells, it might also be interesting to assess other cell lines: MDCK, T84, and HGE, as they are often used to study H. pylori. Using the protocols in chapter five as a starting point, each assay would need to be optimized for the change in H. pylori strain background.

After completion of the basic characterization, modulation of host cell signalling pathways could be assessed. This should initially be assessed through analyzing of changes in host cell morphology, as described in chapter five. However, there are some changes that might be considered. One concern is that we perhaps did not ever achieve the maximal percent of elongated cells in our assay; subsequent smaller experiments showed that the highest percent of elongated cells was observed between 12-18 hours post infection. Additionally, while the percent of elongated cells tapered off, in a single experiment I conducted, the length of the protrusions in elongated cells infected with the EPIYA-ABtD strain still continued to increase even at 18 hours. These results could indicate that the EPIYA-ABtD strain may elicit reactions more slowly, yet its overall effects may be more drastic. Thus, it would be wise to explore additional time points, such as 24 and 36 hours post infection, to assess this possiblity. Of course, if these strains take longer to adhere or if they adhere at much lower levels, then the MOI or time points studied post infection would have to be adjusted accordingly.

Once the basic charaterization of these strains is complete, modulation of specific pathways could be assessed. For example, Erk activation could be assessed via Western blot analysis, as could activation of NF-κB. Localization of β-catenin to the nucleus could be assessed via microscopy or Western blot analysis of nuclear extracts of infected cell lysates. An alternative option to assess the role of the EPIYA motifs in vitro could also be to create these strains in other reference strains, such as J99, 26695, or HPAG-1.

An additional alternative for moving this project forward, could be the use of the parental strain (PMSS1) of the mouse derivative Sydney strain 1 (SS1). PMSS1 has recently been characterized (10, 47), and colonizes mice, but at a lower level than SS1 (10). PMSS1 encodes for a functional CagA protein, whereas SS1 does not express or deliver functional CagA (10, 71). Not only does PMSS1 produce a functional CagA, but this strain has been shown to cause severe pathology in mice, including atrophy, hyperplasia, and metaplasia (10). Since our knowledge about this strain is limited, and it is known to lose the ability to inject CagA into host cells after 1 month in vivo (10), it might be wise to measure the spontaneous mutation rate of PMSS1 before conclusively deciding to use it for isogenic strain construction. Provided preliminary results are satisfactory, new primers to amplify the upstream (5') cagA region and the downstream (3') cagA region would need to be designed in order to create the constructs needed to produce the isogenic strains. Then, in vitro characterization of the newly constructed strains could proceed as described above. Once these assays are completed, a small pilot animal study to assess colonization load, histology, and the timeline for disease progression should be completed. Overall the utilization of this strain could provide another option to explore the role of the EPIYA motifs not only in vitro but also in in vivo assays.

Importance and the Impact of Future Studies

Gastric cancer is still the second most common cause of cancer morbidity and mortality, and this could be reflective of the high incidence of H. pylori infection (20, 60, 68, 86). It could also be a result of the high prevalence of cagA in many H. pylori strains, or due to the presence of certain CagA polymorphisms that predominate in geographic areas that have high rates of gastric cancer (2, 18, 20, 27, 29, 45, 86). Due to the fact that we do not yet thoroughly understand the process of H. pylori induced pathogenesis, including development of gastric cancer, elucidation of virulence factors or virulence factor polymorphisms that impact disease is imperative. Unfortunately, H. pylori is an organism that shows a high rate of genetic variability, which makes traditional epidemiological studies only good indicators of trends. Thus, in order to elucidate the exact role of virulence factors or polymorphisms, it is best to assess differences by creating isogenic strains.

The successful creation of EPIYA isogenic strains will not only answer the question as to what role the EPIYA motifs play in disease manifestations, but will open the door to the assessment of multiple virulence factors. For instance, it has already been demonstrated that CagA is a "master" virulence factor (13, 37, 38), and that there is an association between other virulence factors and disease among the different cagA alleles. This is especially true for different vacA alleles (38, 42). Thus, once the EPIYA isogenic strains are created and characterized, they could then be used as the parental strain background to create isogenic strains containing different polymorphic forms of other virulence factors. This would allow us to not only assess the role of different virulence factors in disease, but also their role in disease development within the context of a particular cagA allele.

If I were to undertake these assays personally, I would focus first on VacA. As discussed earlier, VacA is polymorphic and different vacA alleles impact disease differently (15, 37, 48, 58). Thus, we could create vacA isogenic strains within certain cagA isogenic backgrounds to more conclusively look at CagA and VacA interaction. Specifically, I would alter the vacA alleles within the context of the EPIYA-ABtC, EPIYA-ABtCCC, EPIYA-ABtD and the restorant strains. Experiments with these strains would identify any differences between East Asian (EPIYA-ABtD) and Western (EPIYA-ABtC and EPIYA-ABtCCC) strains. They would also indicate if the number of EPIYA-C motifs influences the effects of the different vacA alleles. The restorant strain would provide an important control for genetic manipulation of our strains. These types of studies would allow our lab to not only examine the role of the different vacA alleles, but to assess the impact of these alleles within the context of the cagA allele. Furthermore, these types of studies may also clarify some of the discrepancies in the literature as to the impact of different vacA regions (37, 75).

Additionally, I would likely focus on two other polymorphic virulence factors that have been implicated in disease development, HomB (19, 42, 64) and OipA (16, 61). The Helicobacter outer membrane (Hom) proteins are complex because H. pylori has two loci that can encode for a Hom protein; strains can encode for either homA, homB, homA/homB, homA/homA, homB/homB, or be negative for both homA and homB (19, 42, 64). The presence of homB has been linked to the development of more severe disease, as compared to the presence of homA (19, 42, 64). Additionally, a dose response has been identified for strains encoding for homB/homB (62, 63). OipA is an outer membrane protein whose expression has been shown to be subject to phase variation due to the number of CT repeats found in the oipA signal sequence (6). OipA "on" is used to designate a strain that expresses a functional protein (16, 61). Again, strains that encode for an OipA protein are associated with more severe disease outcomes (16, 61). Moreover, the OipA "on" phenotype is often found in cagA positive strains (6). The impact of homB and oipA could be assessed individually or in the context of different virulence factors. In other words, besides creating isogenic derivative of these factors within the EPIYA-ABtC, EPIYA-ABtCCC, EPIYA-ABtD and restorant strains to identify difference among the different cagA alleles, they could also be assessed within the EPIYA isogenic strains that carried different vacA alleles. For instance, I would likely first assess the EPIYA-ABtC, EPIYA-ABtCCC, EPIYA-ABtD and the restorant strains that were s1/i1/m1 and s2/i2/m2. This would help limit the number of strains, especially since there are six different hom combinations that could be assessed. This would also seem to be the most likely place to observe differences, since the s1/i1/m1 is the most virulent and the s2/i2/m2 is the least virulent vacA allele. If a difference between these populations was observed, isogenic strains within the different vacA alleles could then be created and assessed. This system would also allow us to systematically examine these virulence factors in the context of the whole bacterium, and to assess their role in disease progression, including progression to gastric cancer. This would also be a better system to analyze the role of these different virulence factors in the context of geographical differences, which could really help expand the field of H. pylori pathogenesis.


H. pylori is a medically important pathogen that has successfully challenged the preconceived idea that bacteria cannot cause gastric disease. However, more than a few questions remain about this process as well as what host, environmental, and bacterial factors are important for the progression to severe disease. This thesis focused on the bacterial toxins, CagA and VacA, and their role in influencing progression to more severe disease. To that end, we were the first to statistically link a specific cagA allele (EPIYA-ABD) to gastric cancer development. We were also able to analyze this large population of clinical isolates for the role of VacA polymorphisms, and found that while the distribution of vacA alleles was not directly associated with disease state, it was associated with the distribution of cagA alleles and in a three-way association that included the vacA allele, the cagA allele and disease state. During the course of this work, we identified an amino acid (196) important for the development of gastric cancer within the intermediate region of VacA. Additionally, we identified some associations that were CagA-dependent, such as the association of VacA and disease state in the EPIYA-ABD population and amino acid distribution at position 231 and disease state in the non EPIYA-ABD population. Finally, we were able to optimize techniques that will be used in future studies aimed at characterizing CagA isogenic strains.

While little is currently understood about why some H. pylori infected individuals develop gastric cancer while others remain asymptomatic, the data gathered during the course of this thesis will help shed some light on the pathogensis of H. pylori­-induced disease. Indeed, the elucidation of bacterial factors that are involved in the pathogensis of H. pylori­-induced disease, such as EPIYA-ABD CagA, is important, because they can serve as a diagnostic marker of infection with a more virulent strain. Understanding any hierarchy of virulence factors is imperative, because more evidence has accumulated that underscores the fact that colonization with H. pylori is protective against other illnesses, including asthma (73), tuberculosis (70) and esophageal cancer (reviewed in (17). Thus, suggesting that treatment should only be used for individuals infected with highly virulent strains. Such treatments that only target patients infected with strains that could cause severe disease would be akin to geographically personalized treatment, which I believe is the future for treating H. pylori-induced disease. Implementing such location-specific treatments could aid in eliminating a majority of gastric cancer mortality and morbidity worldwide without sacrificing the protective effects provided by infection with H. pylori.