Helicobacter pylori: an invading microorganism? A review

Abstract

In this review we evaluate the pros and cons of Helicobacter pylori invasion of epithelial cells as part of the natural history of H. pylori infection. H. pylori is generally considered an extracellular microorganism. However, a growing body of evidence supports the controversial hypothesis that at least a subset of H. pylori microorganisms has an intracellular (intraepithelial) location. Most significant is the fact that H. pylori invades cultured epithelial cells with invasion frequencies similar to Yersinia enterocolitica and better than Shigella flexneri; furthermore, studies of invasion mechanisms suggest that H. pylori invasion of and survival within epithelial cells is not merely a passive event, but requires active participation of the microorganism. Although many studies of human gastric biopsy specimens have failed to demonstrate any intracellular H. pylori, some studies have revealed a minor fraction of H. pylori inside gastric epithelial cells, with possible linkage to peptic ulceration and epithelial cell damage. In conclusion, these data encourage further research to establish whether intracellular H. pylori does play a role in H. pylori colonization of the human stomach and in peptic ulcer pathogenesis.

Helicobacter pylori, Invasion, Epithelial cell, Intracellular survival

1 Introduction

Helicobacter pylori colonization of the human stomach was first associated with human disease when Marshall and Warren succeeded in culturing the bacterium in 1983 [1,2]. Although a majority of H. pylori-infected individuals will be asymptomatic, H. pylori is now accepted as the major cause of duodenal and gastric ulcers [3,4]. In addition, a relationship to adenocarcinoma and low-grade B-cell mucosa-associated lymphoid tissue lymphoma of the stomach has been suggested [5–7].

Although H. pylori infection can be treated, the organism still infects approximately one half of the world's population [8]. The treatment of H. pylori is complicated, requiring a minimum of two different antibiotics plus gastric acid suppression for successful H. pylori eradication [9]. Since it is not possible, or necessary, to give half of the world's population appropriate antibiotics for H. pylori infection, a possible separation of H. pylori into pathogenic and non-pathogenic subspecies would be of great importance.

H. pylori leads to a chronic infection associated with gastric inflammation and formation of circulating antibodies. The immune response, however, fails to eliminate the organisms from the surface of the gastric mucosa. The bacterium has chosen a niche where very few bacteria dare to reside. Apparently, this niche may also render the organism safe from the host's immune defense and, to some extent, safe from antibiotics.

H. pylori adhesion to the gastric mucosa has been described in both cell lines and animal models, as well as by ultrastructural examination of human gastric biopsy specimens [10]. H. pylori adheres intimately to gastric epithelial cells with specific tissue tropism [11]. The adhesion of H. pylori has been compared to the adherence of attaching and effacing (eae-positive) Escherichia coli to epithelial cells [12,13], although genes corresponding to the E. coli eae gene have not been found in H. pylori[14].

For many other enteric bacteria and chronic bacterial infections, invasion of epithelial cells is of major significance in the survival and multiplication of the microorganisms [15]. Controversy exists whether H. pylori is capable of invasion into epithelial cells [16]. In this review we evaluate the pros and cons of H. pylori invasion of epithelial cells as part of the natural history of H. pylori infection and as part of the pathogenesis of H. pylori-induced gastritis.

2 H. pylori in human gastric epithelial cells in vivo

Although several ultrastructural studies have described H. pylori inside gastric epithelial cells in human gastric biopsy specimens, only few of these studies have evaluated the frequency of H. pylori invasion (Table 1). In a study of 27 human gastric biopsy specimens, it was found that H. pylori often associated with intercellular tight junctions [17]. In addition, deep penetration of epithelial cells and several examples of genuine intracytoplasmic penetration by H. pylori were described. Most bacteria were found free in the gastric mucus, but among all bacteria observed with intimate contact to epithelial cells (n=231), 3% were found intracellularly. A majority of adhering bacteria abutted to the cell membrane, 8% adhered to adhesion pedestals, and 11% were associated with depressions in the cell membrane, some of the latter often giving the impression of beginning an intracytoplasmic penetration [17].

Table 1

H. pylori invasion into gastric mucosa cells in human biopsy specimens

Cell type	Invasion frequency	Intracellular location	References
Epithelial	3%	Epithelial cell cytoplasm and examples of apparent engulfment of H. pylori by epithelial cells	[17]
Epithelial, parietal and chief cells	<1% of epithelial cells observed	Within vacuoles and cytoplasm	[18]
Metaplastic surface cells	10% of patients with active duodenal ulcer	Within vacuoles and cytoplasm	[19,20]
Epithelial	Invaded H. pylori was seen in four of 12 patients, from whom several biopsies were taken from different locations	Within the cytoplasm of the gastric cells, without association with lysosomes or mucous granules	[21]
Mucus-secreting cell and a stromal cell in the lamina propria	In two of 100 antral gastric biopsy specimens	Cytoplasm	[22]
Epithelial	“H. pylori are capable of entering epithelial cells”, no quantification of the event	In mucus within an epithelial cell	[23]
Epithelial	Rarely inside cells	Not described	[24]
Epithelial	Only one H. pylori observed within an epithelial cell	Within a lysosome	[25]
Epithelial	“Some”	Within vacuoles	[26]
Epithelial	Occasionally	Within vacuoles and cytoplasm	[27]
Epithelial	Occasionally	Not described	[28,29]
Epithelial	5.6–100% of samples	Apparently within vacuoles	[30]

All studies were based on electron microscopic evaluation except study [30] which was based on light microscopy and interference contrast microscopy.

In another study, which included patients with either gastric ulcer or chronic gastritis, H. pylori invasion of epithelial cells was found to be an uncommon, but still repeatable, observation. Thus, intracellular H. pylori was found in four of eight examined biopsy specimens [18], although invading H. pylori in these specimens was found in only 1% of observed epithelial cells. The intracellular H. pylori was seen both in surface epithelial cells, in parietal cells and within chief cells [18].

Intracellular H. pylori has been observed in 10% of patients with active duodenal ulcer [19,20]; H. pylori was found in the cytoplasm of metaplastic surface mucous cells of the duodenum, between mucous granules, attached to vacuoles and close to lysosome-like structures. These intracellular bacteria were only seen in cells at the edge of duodenal ulcers.

In four out of 12 cases where gastric biopsy specimens were examined with electron microscopy (EM), viable H. pylori was found within the cytoplasm of gastric mucous cells [21]. In a study of 100 antral gastric biopsy specimens, where gastric biopsies were fixed and stained with specific H. pylori antibodies, intracellular H. pylori was found by EM in two specimens. In one case, H. pylori was found in a gastric epithelial mucus-producing cell and in the other case, a few H. pylori cells were found in a stromal cell in the lamina propria [22]. Intracellular H. pylori has been described in several other EM studies of epithelial cells in gastric biopsy specimens, without a more precise quantification of the phenomenon [23–27] (Table 1).

It has been found that H. pylori eradication rate was significantly lower in patients with H. pylori bacteria deeply embedded in the gastric mucosa than it was in patients without EM features of infiltration. Depth of bacterial infection was defined by the number of bacteria found between, and occasionally within and beyond, gastric mucosal cells. In addition, a correlation was found between depth of mucosal invasion of H. pylori and the existence of a peptic ulcer in patients [28,29].

In a light and differential interference contrast microscopic study of 144 gastric biopsies, several examples were found of H. pylori apparently lying within epithelial cells below the apical cell border. Furthermore, a correlation was found between the incidence of intercellular and intracellular colonization and the epithelial damage. Biopsies were divided into seven groups based on severity of epithelial cell damage and examples of intracellular H. pylori were found in a proportion of samples ranging from 5.6% to 100%, increasing with an increase in cell damage grade. In 7/144 (4.9%) of the examined biopsies, intracellular location was the predominant mode of adhesion compared to free-in-mucus, surface adhesion and H. pylori between cells [30].

Indirect proof for significance of H. pylori invasion of epithelial cells was found by immunohistochemical staining of H. pylori heat shock protein (HSP60) in gastric biopsy specimens. Bacterial HSP60 was found both on the mucosal surface and within epithelial cells. Control experiments showed no such expression of human HSP60 [31]. Despite the disappearance of H. pylori from the surface of the mucus cells immediately after H. pylori eradication therapy had been completed, bacterial HSP60 was found within epithelial cells in all of 10 H. pylori-positive patients who had received treatment. In five of these patients, H. pylori reappeared on the surface of epithelial cells when gastric biopsy specimens were examined 4 weeks after treatment [31].

Nevertheless, it is necessary to emphasize that other ultrastructural studies of clinical biopsies and primary cultures of human antral gastric epithelial cells did not reveal any intracellular H. pylori[32–41]. In six studies, however, H. pylori was found deep in the gastric glands and within the parietal cell canaliculi [34,38–40,42,43]. In addition, the EM study of duodenal ulcer margins by Ogata [41] did reveal disrupted cell membranes and numerous apical blebs in areas of the epithelial cells where H. pylori attached. Such disruption of cell membranes could be the result of intracytoplasmic penetration of great numbers of H. pylori. This has been suggested based on ultrastructural examination of 32 endoscopic biopsies, in which H. pylori was found to penetrate deeply into gastric cells [44].

The presence of H. pylori beyond the epithelial cell barrier, i.e. within the lamina propria of the gastric mucosa, has been described. Whole H. pylori and H. pylori-immunopositive material were found in the lamina propria in 13 out of 34 H. pylori-positive patients, where biopsies were taken for histopathologic evaluation [45], and H. pylori surface proteins were found within the lamina propria of gastric antrum from patients with H. pylori-associated gastritis [46]. In a study of patients with non-ulcer dyspepsia, the presence of whole bacterial cells, mainly of coccoidal form, was described within the tunica propria of the gastric antrum [47]. Other investigators have emphasized the absence of H. pylori within the lamina propria [19,26,48].

In conclusion, H. pylori seems to have the ability to invade human gastric epithelial cells, although only a minor proportion of the bacteria are found intracellularly.

3 Cell cultures

A convenient model for studies of interactions between H. pylori and eukaryotic cells employs cultured human cell lines. Many investigators have described internalization of H. pylori into immortalized cancer cell lines [12,13,49–54] (Table 2).

Table 2

H. pylori invasion in epithelial cell cultures

Method used	Cell type	Invasion frequency	Intracellular location	References
EM	Kato III cells	Occasional	Within vacuoles	[49]
EM	AGS cells	Frequently	Usually in membrane-defined vacuoles	[12]
EM	AGS cells	H. pylori engulfed by several of the AGS cells	Within vacuoles	[13]
Gentamicin invasion assay and EM	AGS cells	12% of cell-associated H. pylori after 3 h incubation	Within vacuoles	[54]
Gentamicin invasion assay and EM	AGS cells	0.6–2.0% of the inoculum (9–17% of the inoculum adhered to AGS cells in these experiments)	Within vacuoles, taken in by endocytosis	[52]
Gentamicin invasion assay	Hep-2 cells	0.1% of the inoculum (number of adhering H. pylori not defined)	Within vacuoles	[51]
Gentamicin invasion assay	Hep-2 cells	<0.0019% of the inoculum (number of adhering H. pylori not defined)	Within vacuoles and free in cytoplasm	[53]
Gentamicin invasion assay	HeLa	One strain 3.4% of cell-associated H. pylori, other strains <1%	Not described	[50]

In a comparison between E. coli, Campylobacter jejuni and H. pylori, only H. pylori was found to have tissue specificity for Kato III cells (gastric adenocarcinoma), and in these experiments H. pylori was occasionally found within Kato III cells [49]. When AGS cells (gastric adenocarcinoma) were used for adhesion assays, intracellular H. pylori has frequently been observed [12,13]. Numerous steps of internalization of H. pylori coccoid forms by the epithelial cells were seen, including bacteria that were totally engulfed and appeared to be enclosed within a cytoplasmic vacuole. Intracellular spiral forms were also observed, usually located within defined membrane-bound vacuoles [12]. In one EM study, using AGS cells, bacteria were found to be in coated pits, to be taken into the host cell by endocytosis, and inside AGS cells within vacuoles [52]. In a gentamicin assay with AGS cells, 0.6–2.0% of the initial inoculum was found intracellularly, whereas 9–17% of the initial inoculum attached to the epithelial cells [52]. Similarly, in another study, 12% of adhering H. pylori were found to have invaded AGS cells [54]. H. pylori was able to pass through a bilayer of AGS cells and an endothelial layer [52], supporting in vivo observations from human biopsies of H. pylori antigens and whole H. pylori in the lamina propria of the human stomach [45–47]. Another frequently used cancer cell line (HeLa cells) did not appear to be invaded by H. pylori at a significant level [50]. On this basis, the authors concluded that invasion of epithelial cells by H. pylori does not play an important role in the pathogenesis of infections caused by H. pylori. Nevertheless, as stated by the authors, HeLa cells, derived from a cervical adenocarcinoma, are probably not optimal in testing H. pylori adhesion and invasion. Yet, the authors did find an invasion percentage in gentamicin invasion assays for a least one strain of 3.4% ((intracellular/total cell-associated bacteria)×100%), whereas the remaining strains of a total of 10 strains all had an invasion percentage below 1% [50]. In a study of H. pylori invasion of Hep-2 cells (laryngeal adenocarcinoma), the level of internalization was 0.1% of the bacterial inoculum (10⁷ organisms per well) according to a gentamicin invasion assay [51]; similar percentages have been observed for Shigella spp. by the same method. Others have found lower H. pylori internalization frequencies in Hep-2 cells varying from 0.0006 to 0.0019% of the initial inoculum (10⁸ organisms per well), perhaps explained by both saturation of the invasion mechanism and greater toxicity of a larger amount of H. pylori used [53].

Of course, few cultured cell lines closely mimic the original tissue from which they were isolated. Polarized epithelial cell lines, with apical and basolateral cell surfaces separated by tight junctions and well-defined microvilli, have been used to better approximate epithelial barriers in vivo [55]. In one study, polarized cell lines, T84 (human intestinal cell monolayer) and MDCK (dog kidney), have been used, and a gentamicin assay would suggest that H. pylori does reach an intracellular compartment even in polarized cells, since approximately 10% of cell-associated H. pylori were gentamicin-tolerant after 24 h of incubation. But although EM revealed H. pylori in deep invaginations of the cell membrane, it was not possible to visualize H. pylori within vacuoles in the T84 cell monolayer [56]. In conclusion, H. pylori has the capacity to invade epithelial cells in epithelial cell cultures and, interestingly, the invasion capacity seems to be more pronounced in gastric adenocarcinoma cell lines compared to cancer cell lines of non-gastric origin.

4 Studies of H. pylori invasion mechanisms

In a gentamicin invasion assay, it was found that at least two mechanisms could be involved in the internalization of H. pylori into AGS cells [52] (Table 3): methylamine, an inhibitor of pinocytosis, reduced invasion from 0.6–2.0% of the inoculum to 0.04%; cytochalasin D, an inhibitor of actin polymerization, reduced the invasion to 0.07% of the inoculum. The combination of methylamine and cytochalasin D in the same study reduced the invasion percentage to 0.002%, as determined by a gentamicin invasion assay [52]. In another study, few H. pylori were found intracellularly in epithelial cell lines deficient in αβ1-integrins, but the number of invaded H. pylori rose when cells were transfected with a β1-expressing cDNA clone; therefore, it could be suspected that H. pylori invasion is integrin-mediated [57]. In addition, H. pylori invasion was inhibited by cytochalasin D in this study.

Table 3

Proposed H. pylori invasion mechanisms

Cell type	Assay modifications	Proposed mechanism	References
AGS cells	Cytochalasin D and methylamine both inhibited invasion	Invasion based on a combination of pinocytosis and actin polymerization	[52]
AGS cells	Cytochalasin D and genistein both inhibited H. pylori invasion	Invasion requires tyrosine kinase activity and actin polymerization	[57]
AGS cells	FCS was found to promote H. pylori invasion, and proteinase K treatment of FCS abolished this effect	Invasion mediated by mammalian proteins	[54]
Hep-2 cells	Formalin, ammonium chloride, chloroquine, dansylcadaverine	Invasion depends on live bacteria, on receptor clustering and receptor-mediated endocytosis, but not actin polymerization, since cytochalasin B did not inhibit internalization	[51]

Intracellular uptake of H. pylori by Hep-2 cells seems to require a factor present only in living bacteria, since formalin-killed bacteria lost the ability to invade Hep-2 cells [51]. In contrast, formalin-killed bacteria were still able to adhere to Hep-2 cells, probably through the neuraminyllactose-binding adhesin, since neuraminidase treatment of Hep-2 cells inhibited the binding [51]. This experiment suggests that the H. pylori hemagglutinin is important for adherence, but another as yet unknown factor is required to initiate internalization.

Uptake of H. pylori by Hep-2 cells was decreased 100-fold, when cells were treated with ammonium chloride and chloroquine, inhibiting receptor-mediated endocytosis, at concentrations which did not affect adherence [51]. Dansylcadaverine, an inhibitor of receptor clustering and internalization, also decreased uptake 100-fold, but not adherence of H. pylori, and incubation at 4°C completely blocked the uptake of H. pylori. Cytochalasin B, an inhibitor of phagocytosis, did not inhibit uptake in Hep-2 cells. It was concluded that H. pylori was internalized into Hep-2 cells either by receptor-mediated endocytosis or by a closely related pathway, and that actin polymerization was not required for uptake of these bacteria by the Hep-2 cells [51].

Controversy thus exists between results obtained with AGS and Hep-2 cells concerning the involvement of actin in H. pylori invasion. It is, however, worth noting that AGS cells, unlike Hep-2 cells, are of gastric origin, and that in the Hep-2 cell assay cytochalasin B was used instead of cytochalasin D, which is a more potent and specific depolymerizer of microfilaments [58]. Host cell cytoskeletal actin provides the machinery for the internalization of many invading enteric bacteria. For some of these invading bacteria, the cytoskeletal changes will be moderate and disappear as soon as the bacteria have invaded the host cell, as seen for Yersinia enterocolitica[59] and Salmonella spp. [60]. We found the changes in host cytoskeletal structures in AGS cells induced by H. pylori to be very moderate and the presence of these cytoskeletal changes was inversely related to the amount of bacteria having invaded AGS cells after 3 h incubation [54].

In vivo and in vitro invasion frequencies seem to be quite different. Such a paradox is known from other bacteria, which invade only the basolateral (bottom) surface but not the apical surface of epithelial cells, such as Shigella spp. and Yersinia spp. Before these bacteria are able to invade epithelial cells, they will have to pass through M-cells or let recruited polymorphonuclear cells (PMN) breach the epithelial cell tight junctions and, thereby, open a route for bacterial contact with the basolateral surface of epithelial cells [15]. Something similar could be the case for H. pylori, since H. pylori infection of gastric epithelial cells does lead to increased secretion of polymorphonuclear chemotactic substances [16]. Whether the gastric mucosa contains M-cells is not known. A potential contact between H. pylori and epithelial cell basolateral areas could be reached at the edge of gastric/duodenal ulcers, where H. pylori, as described above, has been found within epithelial cells [19,20].

5 Comparison to other enteric invading microorganisms

Some H. pylori strains, particularly those isolated from patients with peptic ulcer, invaded Hep-2 cells at frequencies similar to a Y. enterocolitica strain and at higher frequencies than Shigella flexneri which were used for comparison [53].

A level of H. pylori invasion of Hep-2 cells of approximately 0.1% of the inoculum after 4 h of incubation [58] was comparable to that previously described with several different species of Shigella spp. and entero-invasive E. coli[61,62].

It has been shown that Salmonella spp. and Yersinia spp. are internalized by a mechanism similar to receptor-mediated endocytosis and that uptake of these organisms is blocked by amines such as chloroquine and dansylcadaverine, chemicals inhibiting receptor recycling and clustering [63]. Similar effects of these substances have been found for H. pylori[51]. Uptake of Yersinia spp. and Shigella spp. is also inhibited by blocking actin polymerization, a process required for phagocytosis, with cytochalasin B [61,63,64]. On the other hand, this substance does not inhibit Salmonella spp. invasion [65]. Therefore, it is possible that H. pylori, like Yersinia, will use both receptor-mediated endocytosis and active participation of cell cytoskeletal components in invasion, as discussed previously [52].

6 Relation to other ultrastructural adhesion modes

On the basis of morphological appearances, the different types of interaction of H. pylori with gastric epithelial cells have been categorized as (a) binding through filamentous strands, (b) adhering to microvilli, (c) abutting to the host cell membrane, (d) binding to adhesion pedestals, (e) occupying depressions in the host cell membrane, and (f) invading host cells [17]. That attachment occurs in this succession has been proposed previously [44]; that engulfment of H. pylori rather than adhesion pedestal formation is the end of the attachment process in cell cultures is supported by in vitro observations, where ultrastructural attachment modes after 4 and 24 h incubation of H. pylori with AGS cells were described [66].

In vivo, most H. pylori will be scattered in the mucus layer without membrane-to-membrane attachment with the epithelial cells [17,37,67]. Overall, most H. pylori adhering to gastric cells are abutting or binding to microvilli [17,37,67]. Whether the different morphological features of adhesion really represent different stages of attachment, which will ultimately end in intracytoplasmic penetration or engulfment, still needs to be proved.

Involvement of E. coli eae genes in both adhesion pedestal formation and invasion has been described [68]. It has been proposed that combined action of eae genes and a plasmid-determined factor will lead to adhesion pedestal formation and localized adherence and no significant invasion of HeLa cells, while a plasmid-cured eae-positive E. coli will be internalized efficiently [69]. A similar correlation between H. pylori adhesion pedestals and H. pylori invasion could be speculated.

7 Circumstantial evidence for H. pylori invasion as part of H. pylori pathogenesis

7.1 Putative involvement of H. pylori genes in epithelial cell invasion

An important feature of intracellular microbes is that they have enzymes to detoxify oxygen metabolites formed e.g. during the respiratory burst of immune cells [70]. H. pylori has a superoxide dismutase (SOD) and sequencing and alignment revealed striking homology to the following facultative intracellular human pathogens: Listeria ivanovii, Listeria monocytogenes, Coxiella burnetii, Porphyromonas gingivalis, Legionella pneumophila and Entamoeba histolytica[71]. The H. pylori SOD gene further resembles a SOD gene found in Campylobacter jejuni and it has been shown that the absence of the SOD gene impairs the ability of C. jejuni to survive within epithelial cells (INT407) (12-fold reduction in viability of intracellular organisms) [72]. A similar mechanism could be suspected in H. pylori.

Bacterial invasion mechanisms often target the invading bacteria to a different vesicular pathway than that normally used during phagocytosis [15], for instance, inhibiting phagosome–lysosome fusion, inhibiting the normal acidification of intracellular vacuoles. Mammalian cells have a series of specialized proteins dedicated to budding, fusing, docking, and routing endogenous vesicles appropriately within a cell. Finlay and Falkow [15] have speculated that this group of proteins could be a target for a toxin produced by an intracellular pathogen, and it appears that the vacuolating cytotoxin of H. pylori causes large amounts of such proteins to accumulate inappropriately in the large vacuoles triggered by this pathogen [73]. Increased internalization in Hep-2 cells and increased survival in AGS cells of H. pylori strains compared to their isogenic VacA knockout mutants supports such a role of VacA [66,74]. Parasites such as Toxoplasma gondii and Mycobacterium spp. are able to inhibit fusion of phagosomes with lysosomes (comparable to the action of H. pylori VacA), thus escaping the potentially harmful action of lysosomal hydrolases [70].

7.2 H. pylori invasion of immune cells

Mechanisms used to enter non-phagocytic cells will also often enhance or mediate entry into phagocytic cells. For example, Salmonella Typhimurium mutants with a decreased capacity of epithelial cell invasion also have a decreased capacity for uptake into phagocytic cell and Yersinia spp. bind preferentially to β1-integrins of phagocytic cells rather than to the normal host cell receptors of phagocytes [73]. Thus, studies of H. pylori interaction with phagocytes are also interesting when trying to determine the ability of H. pylori to invade epithelial cells.

Without opsonization of H. pylori by serum, the ultrastructure of H. pylori ingested by PMN or monocytes showed no sign of degradation [74]. With more than one H. pylori per immune cell, neither PMN nor monocytes were able to kill H. pylori efficiently within 60 min, even after opsonization [75]. In contrast, other bacterial species will be killed rapidly by PMN even with an excess of bacteria present [76].

In a recent paper, it has been suggested that virulent strains of H. pylori (type I strains) survive better within macrophages than do type II strains [77], explained by disruption of phagosome maturation in macrophages by the virulent strains. These results may indicate that some H. pylori seem to have ways of securing intracellular survival, and that PMN and monocytes will only have a limited effect on H. pylori in the gastric mucosa.

8 Intracellular survival and multiplication

Following cell entry, most invading bacteria will immediately be localized within a membrane-bound vacuole inside the host cell, and then the organisms may or may not escape this vacuole, depending on the pathogen and its strategy for survival [15]. H. pylori has been observed both within vacuoles and freely in the cytoplasm both in vivo and in vitro [20,27,51,53].

So far, no evidence has been presented regarding any possible intracellular growth of H. pylori, but survival has been reported on several occasions. H. pylori internalized into Hep-2 cells were found to be alive for at least 6 h [53]. A substantial number of H. pylori will enter Hep-2 cells within the first hour of co-incubation of cells and bacteria. The number of intracellular bacteria further increased at the 4- and 24-h intervals, whereupon the number decreased dramatically. At 48 h, Hep-2 cells rarely contained intact bacteria, and those seen were in the process of degradation inside vacuoles [51].

Others have found that H. pylori can survive intracellularly in Hep-2 cells for 40 h in vitro [78]. Long survival within gastric epithelial cells would not be required in vivo, since gastric epithelial cells have a turnover time of 2 days. Despite this, it was found that coccoidal forms of H. pylori may persist in the Hep-2 cells after several transfers of trypsinized monolayers. In addition, it was shown that eight of 10 Hep-2 cells prepared after four consecutive transfers of trypsinized monolayers of the original infected cells still contained numerous coccoidal forms of H. pylori[51].

Occasionally, H. pylori invasion of Kato III has also been observed and H. pylori invasion seemed to be a commensalism since it did not affect the survival of Kato III cells substantially, in contrast to C. jejuni and E. coli which both had a marked toxic effect on Kato III cells [49]. Furthermore, vacA mutants were found to have reduced survival inside AGS cells, when compared to wild-type strains, suggesting a specific role of the vacuolating cytotoxin in intracellular survival of H. pylori in epithelial cells [66]. As previously mentioned, type I strains were found to survive better inside macrophages than type II strains [77], suggesting that either VacA or genes within the cag pathogenicity islands (PAI) is involved in intracellular survival of H. pylori. Intriguingly, a recent study with deletion of the PAI in a type I H. pylori strain could not demonstrate any difference in survival rate within professional phagocytes between wild-type and mutant strains [79], whereas no similar experiments have been performed with H. pylori vacA mutants and professional phagocytes.

9 Facultative and serum-dependent Helicobacter invasion

Although H. pylori is not an obligate intracellular microorganism, it is still possible that invasion of epithelial cells could be a facultative event. In one study, it was shown that especially at the edge of duodenal ulcers, H. pylori occupied the cytoplasm of epithelial cells, whereas fewer bacteria were seen intracellularly in other gastric locations [19]. In addition, H. pylori cells residing in the antrum were larger than H. pylori cells residing at the edge of the duodenal ulcers [19]. This could indicate that both the maturity of the epithelial cells and characteristics of H. pylori are involved in determining at which rate H. pylori will be an invasive microorganism.

Increased invasion of H. pylori into AGS cells was seen with increasing amounts of fetal calf serum (FCS) added to the cell culture medium, and a maximum was reached with approximately 10% FCS added [54]. The outcome of H. pylori adhesion to AGS cells thus seemed to depend on specific attachment factor(s) in FCS, with a saturable binding to H. pylori and/or AGS cells. It may be speculated that in areas such as ulcerated epithelium, where the H. pylori would be bathed in serum factors, the bacterium would become increasingly invasive. The significance of adding FCS to the test medium when studying bacterial invasion of epithelial cells has been described previously for other bacteria. FCS increased the internalization of Neisseria gonorrhoeae by Chinese hamster ovary cells and HeLa cells [80,81], and invasion of human epithelial cell cultures by a strain of Streptococcus pyogenes serotype N1 was shown to multiply over 50-fold when FCS was added to the test medium [82].

The component in FCS mediating H. pylori invasion appears to be a protein, since proteinase K-treated FCS lost the ability to mediate this invasion [54]. The serum factors mediating N. gonorrhoeae and S. pyogenes invasion of epithelial cells were also found to be proteins (vitronectin and fibrinogen, respectively) [81,82]. The mechanism by which neisseriae and streptococci bind mammalian host proteins such as vitronectin and fibronectin has been demonstrated to be through primary binding of sulfated polysaccharides, and in the same study it was demonstrated that H. pylori has similar capacities to bind such polysaccharides and subsequently bind mammalian proteins [83]. In this context, it would not be surprising if H. pylori, like Neisseria spp. and Streptococcus spp., are able to exploit human serum proteins for epithelial cell invasion.

10 Conclusions and perspective

An intracellular location of H. pylori in human gastric epithelial cells is probably not common in H. pylori colonization. Nevertheless, H. pylori has been described within gastric epithelial cells in several studies of human gastric biopsies. Systematic search for the evidence of H. pylori within epithelial cells is definitely needed and this search should include various gastric/duodenal locations and various cell types. In addition, a possible relation to gastric and duodenal ulcers of intracellular H. pylori must be evaluated. Furthermore, host factors such as age and blood type (secretor/non-secretor) should be taken into account as well as H. pylori subtypes.

H. pylori has the ability to invade cultured epithelial cells and it seems that H. pylori invasion involves a combination of receptor-mediated endocytosis and exploitation of epithelial cell actin. The relationship between H. pylori invasion and different H. pylori virulence genes should be addressed in future studies, especially the possibility of an influence of already suspected virulence genes such as the vacA and genes in the cag PAI. The fact that the vacuolating cytotoxin improves intracellular survival of H. pylori in AGS cells, as well as a better survival of type I strains than of type II strains within macrophages, supports the theory of H. pylori as a facultative intracellular microorganism.

Comparison with the mechanisms used by other bacteria such as Neisseria spp. and Streptococcus spp. in exploiting mammalian proteins in invasion suggests that specific environmental exposures of H. pylori also make this bacterium capable of epithelial cell invasion.

If H. pylori is occasionally protected in an intracellular niche, this could explain some of the difficulties in H. pylori eradication using standard regimes and therefore new antibiotics may have to be considered as well as anti-adhesion therapy directed against possible adhesins/invasins involved in this process.

At present, only a few limited and non-conclusive studies have dealt with the clinical significance of H. pylori invasion capabilities, as determined in cell culture assays [51,53]. A number of questions regarding H. pylori involvement in human disease remain unanswered and new pathogenic mechanisms of H. pylori infection must be sought. In this context, it is very relevant to look at the subpopulation of H. pylori lying within gastric epithelial cells and its possible involvement in determining the clinical outcome of H. pylori infection.

Acknowledgements

The authors thank Leif Percival Andersen, MD, Vibeke Binder, DMSci, and Ole Østergaard Thomsen, DMSci, for critical revision of the manuscript.