INTRODUCTION

Top Previous Next References

It was long believed that the normal human stomach was sterile or colonized only with small numbers of bacteria. The mechanism of what was referred to as the "gastric bactericidal barrier" was debated in the early part of this century 22, but most authors then, as well as more recently 171, concluded that the predominant effect was due to gastric acid. However, the cultivation of a novel bacterium from gastric mucosa in 1982 marked a turning point in our understanding of gastrointestinal microbial ecology and disease. Marshall and Warren 265 described spiral or curved bacilli in histologic sections from 58 of 100 consecutive biopsy specimens of human gastric antral mucosa, 11 of which were culture positive for a gram-negative, microaerophilic bacterium. The organism was thought originally to be a member of the genus Campylobacter and was named Campylobacter pyloridis, later corrected to Campylobacter pylori. Because subsequent 16S rRNA sequence analysis showed that the distance between the true campylobacters and C. pylori was sufficient to exclude it from the Campylobacter genus 336, it was renamed Helicobacter pylori 180, the first member of the new genus Helicobacter.

The early proposal of Marshall and Warren 265 that the newly described bacterium caused gastritis and peptic ulcer proved correct. Furthermore, there is now overwhelming evidence that H. pylori is linked to gastric adenocarcinoma 208, 301, 317, the second most common cause of cancer morbidity and mortality worldwide, and to the development of gastric non-Hodgkin's lymphoma 318, 435. The clinical significance of this bacterium has recently been emphasized by a National Institutes of Health consensus panel that recommended antibiotic therapy for the large majority of peptic ulcer patients who are infected with H. pylori 12 and by classification of H. pylori as a class I (definite) carcinogen by the World Health Organization 13. The vast literature on H. pylori has been reviewed recently in this journal 93, and the complete genome sequence of two strains is now available 6, 82, 260, 395.

Nevertheless, Marshall and Warren were not the first to detect gastric spiral bacteria. Spiral organisms were first seen in human gastric mucosa beginning early in the 20th century 235 and were subsequently described by several investigators (reviewed in reference 264. The bacteria were often seen in malignant or ulcerated gastric tissue 165, and the possibility of an infectious cause of peptic ulcer disease was considered 19. Some even specifically proposed that a search be made for an organism "thriving in hydrochloric acid medium ... as a possible factor of chronicity, if not an etiological factor, in peptic ulcer" (see the discussion following reference 165. The suggestion that ulcers might be caused by infection was not new at that time, although an authoritative review published in 1950 concluded that the evidence did not support infection as a cause of peptic ulcer in humans 216. Even before the early observations of gastric spiral bacteria in humans, similar organisms were seen in animals. In his 1881 thesis submitted to the Faculty of Medicine, Rappin described spiral bacteria in gastric scrapings from dogs 330. This observation was later confirmed by Bizzozero 30 and Salomon 344, who performed experimental inoculations with gastric scrapings to transmit infection to mice. Gastric spiral bacteria were subsequently seen in cats 255, rhesus macaques 81, and, more recently, in a variety of other animals.

The cultivation of H. pylori and the recognition of its clinical significance served to renew interest in bacteria associated with the gastrointestinal and hepatobiliary tracts of humans and other animals, many of which have now been identified as novel species of Helicobacter. These organisms are of interest both because of their pathogenic role in humans and animals and because of their value as models of human disease. Other bacteria have also been newly identified, or in some cases reclassified, as novel Helicobacter species that infect humans. The purpose of this review is to describe these other Helicobacter species, characterize their role in the pathogenesis of gastrointestinal and enterohepatic diseases, and discuss their implications for our understanding of H. pylori infection in humans. We conclude with a discussion of an ecological perspective on Helicobacter pathogenesis and recommendations for future work.

To date, eight cultivated Helicobacter species have been found in the stomach of humans and other animals, as well as two uncultivated organisms (Table 1). Occasionally a species such as Helicobacter muridarum, which typically colonizes the rodent bowel, is found in the stomach. These organisms are considered later along with other enterohepatic Helicobacter species.

Most of the early observations on gastric spiral bacteria were made in dogs and cats. When the first electron micrograph of these bacteria was published, it was immediately apparent that more than one morphological form could be found 418. Lockard and Boler 257 provided the first high-quality electron micrographs of what are now called Lockard type 1, 2, and 3 bacteria 150, 151; all are now known to represent Helicobacter species. Lockard type 1, which is representative of Helicobacter sp. flexispira, Helicobacter bilis, and others (see "Enterohepatic Helicobacter species" below), has a fusiform to slightly spiral morphology with tapered ends. Multiple periplasmic fibers appear to cover the entire surface of the bacterium (Fig. 1). Lockard type 2 is spiral rather than cylindrical and has periplasmic fibers that are more sparsely distributed and can appear singly or in groups of two, three, and occasionally four (Fig. 2). This organism is the typical morphology of Helicobacter felis. Lockard type 3, which resembles type 2 but is somewhat more tightly coiled and does not have periplasmic fibers, is typical of Helicobacter bizzozeronii and the uncultivated "Helicobacter heilmannii" (Fig. 3). A fourth type, similar to Lockard type 3 but thicker and with fewer coils, was described in the original electron micrographs published by Weber and Schmittdiel 418 but not by Lockard and Boler. This organism may represent Helicobacter salomonis, recently cultivated from dogs 219. The morphology of gastric Helicobacter species isolated from hosts other than dogs and cats is sometimes distinctive (e.g., Helicobacter mustelae [Fig. 4]) and in other cases resembles H. pylori (eg., H. acinonychis).

View larger version (137K): [in this window] [in a new window]   FIG. 1.   (A and B) Transmission electron micrographs of Helicobacter sp. flexispira (Lockard type 1) taken from a pure culture (A) and from an intestinal crypt in a mouse (B). Bipolar flagella and periplasmic fibers are apparent. (C) Scanning electron micrograph of Helicobacter sp. flexispira shows the organisms among intestinal microvilli in a mouse. Bars, 1 µm. Reprinted from reference 346 with permission from the publisher.

View larger version (113K): [in this window] [in a new window]   FIG. 2.   (A) Negatively stained preparation of H. felis (Lockard type 2) isolated from the gastric mucosa of an adult cat shows the multiple bipolar flagella. Bar, 1 µm. (C) Although faintly seen in Panel A, the characteristic periplasmic fibers are better visualized on a scanning electron micrograph. Bar, 1 µm. (B) Transmission electron micrograph of H. felis in the gastric mucosa of a germ-free mouse shows the characteristic spiral morphology. Pairs of periplasmic fibers can be seen en face (arrows). Bar, 0.5 µm. Photos courtesy of Adrian Lee, Jani O'Rourke, and Lucinda Thompson.

View larger version (149K): [in this window] [in a new window]   FIG. 3.   Transmission (A) and scanning (B) electron micrographs of bacteria in the gastric mucosa of a healthy pet cat. The helical morphology without periplasmic fibers (Lockard type 3) is characteristic of "H. heilmannii" and H. bizzozeronii. Bars, 0.5 µm. (A) Reprinted from reference 303 with permission from the American Society for Microbiology. (B) Photo courtesy of Robert Munn.

View larger version (116K): [in this window] [in a new window]   FIG. 4.   (A) Negatively stained preparation of H. mustelae shows the bipolar and lateral flagella (bar, 0.5 µm). (B) Thin section of a ferret antral gastric pit shows the intimate association of the bacterium with the epithelial surface, including the apparent formation of adhesion pedestals, Bar, 1 µm. Reprinted from reference 309 with permission from the publisher.

The ferret (Mustela putorius) stomach has anatomical and physiological similarities to that of humans 135 and is known to experience naturally occurring gastritis and gastric ulcers 191. Shortly after the publication in 1984 of the seminal observations of Marshall and Warren, a Campylobacter-like organism was isolated by Fox et al. from gastric tissue of one ferret with a gastric ulcer and from two others with normal gastric mucosa 148. This observation was quickly confirmed by others 331, 396. The ferret organism was morphologically and biochemically very similar to what was then called C. pylori, and it was originally designated C. pylori subsp. mustelae 139. However, DNA relatedness and 16S rRNA sequence analyses showed that the organism isolated from ferrets was a novel species, which was named C. mustelae 141 and later renamed H. mustelae 180.

Microbiology and phylogeny. Compared with H. pylori (Fig. 5), H. mustelae is a small rod (0.5 by 2 µm), sometimes slightly curved, with multiple sheathed flagella located at both poles as well as laterally (Fig. 4). Like all gastric Helicobacter species, H. mustelae hydrolyzes urea, although it has other distinctive characteristics such as susceptibility to nalidixic acid (Table 2). The fatty acid composition 379 and protein profiles 287 of H. mustelae are also distinct from those of H. pylori. A phylogenetic tree (Fig. 6) based on a 16S rRNA similarity matrix (Table 3) places H. mustelae closer to Helicobacter species that infect the colon or the hepatobiliary system, particularly H. pametensis and H. cholecystus, than to other gastric Helicobacter species. H. mustelae has a preponderance of hexadecanoic fatty acids, which is characteristic of enteric Helicobacter species and unusual among species that infect the stomach 198. The genome size of H. mustelae is approximately 1.7 Mb 387, which is nearly the same as that determined by sequencing the H. pylori genome 395. Interestingly, there appears to be significant genomic conservation among isolates of H. mustelae 286, 387. This is in marked contrast to the heterogeneity seen among isolates of H. pylori, which are nearly always genetically unique unless derived from the same or related persons 3, 174.

View larger version (114K): [in this window] [in a new window]   FIG. 5.   (A) Scanning electron micrograph of H. pylori shows the gently curved morphology, multiple bipolar flagella, and absence of periplasmic fibers. Bar, 1 µm. Photo courtesy of Adrian Lee, Jani O'Rourke, and Lucinda Thompson. (B) Transmission electron micrograph of human gastric epithelium with large numbers of H. pylori intimately attached to the surface. Bar, 1 µm. Reprinted from reference 248a with permission from the publisher.

View larger version (20K): [in this window] [in a new window]   FIG. 6.   Phylogenetic tree for 19 validated Helicobacter species and 9 additional provisional species, based on 16S rRNA sequence. The scale bar represents a 1% difference in nucleotide sequence as determined by measuring the lengths of horizontal lines connecting any two species. The tree was constructed from the differences matrix (Table 3) using TREEVIEW (314a).

Epizootiology. In the original report, 3 (18%) of 17 ferrets aged 9 to 10 months were found to be infected with H. mustelae when examined at necropsy 148. Subsequent studies showed that the prevalence of H. mustelae infection increases with age, from 0% in kits less than 1 month of age to 100% in adults over 1 year 139, 156. This relationship of prevalence to age mimics the seroepidemiology of H. pylori in humans, particularly in developing countries 21, as well as the seroepizootiology of H. pylori in nonhuman primates 89, 369. Like H. pylori, infection with H. mustelae is apparently persistent. H. mustelae is widespread among colonies of laboratory ferrets 180, 331, 396 and has also been seen in ferrets kept as pets 156. Although examination of adult ferrets from one New Zealand pelt farm failed to find any evidence of infection 290, more recently H. mustelae has been isolated from captive and wild ferrets in New Zealand 131. It seems likely that H. mustelae is a member of the resident flora of the ferret stomach that infects virtually all animals by adulthood, much as is true for human H. pylori infection in most of the world.

Transmission. Direct person-to-person transmission of H. pylori is supported by the clustering of cases in families 86, the similarity of H. pylori genotypes that is sometimes found among isolates from related persons 409, and the failure to find evidence of an environmental reservoir, although transmission in developing countries by contaminated food or water remains possible 209, 232. Nevertheless, the mechanism by which H. pylori moves from the stomach of one host to that of another remains an enigma. Fox and colleagues have offered a series of observations which suggest that transmission of H. mustelae may occur by the fecal-oral route and that it is promoted by hypochlorhydria. Fecal cultures from 9-week-old ferrets were positive for H. mustelae in 8 (31%) of 26 animals, but cultures from the same ferrets at 20 weeks of age were negative 158. The authors hypothesized that the animals were naturally infected at 5 to 6 weeks of age and were hyphochlorhydric when sampled at 9 weeks of age, a time course which roughly corresponds to the period of transient hypochlorhydria seen in experimentally inoculated animals 157. The increased gastric pH may have permitted greater numbers of bacteria to exit the stomach and enter the lower gastrointestinal tract, where they could be cultivated from feces and possibly serve as a mechanism for transmission. However, the gastric pH was not measured, nor was the timing of acute infection documented, although subsequent rising titers suggested that it was recent. Furthermore, fecal cultures from three (75%) of four ferrets that were 8 months old, and probably not in the window of transient hypochlorhydria, were also positive for H. mustelae. When adult ferrets chronically infected with H. mustelae were treated with a proton pump inhibitor (omeprazole) to raise their gastric pH, recovery of H. mustelae in fecal cultures was increased compared to that before treatment, although in one animal the fecal cultures were positive even though the gastric pH failed to rise 138. On balance, these experiments suggest a role for raised gastric pH in transmission of H. mustelae by the fecal-oral route, although they do not directly demonstrate fecal-oral transmission. Replication of these results by an alternate method of raising the gastric pH, such as blockade of histamine 2 receptors, might exclude the possibility that the effect of omeprazole was through a mechanism other than alteration of gastric pH. Future studies on transmission may find it useful to exploit the nonhuman primate model of H. pylori 87, 88, 369 to extend the provocative work by Fox et al.

Development of the ferret model. Early attempts to develop a small-animal model of H. pylori were unsuccessful. Although small laboratory animals have subsequently been infected with H. pylori 261, 361, 424, none of these animals is naturally infected. H. mustelae infection of ferrets remains the best-studied animal model of a gastric Helicobacter in its natural host and provides the opportunity to study the relationship between infection and disease.

(i) Gastritis. Ferrets naturally infected with H. mustelae have a predominantly mononuclear gastritis composed of lymphocytes and plasma cells, with only occasional eosinophilic and polymorphonuclear leukocytes 142, 183. The near absence of an active (polymorphonuclear) component to the inflammation distinguishes the gastritis from that often seen in adults infected with H. pylori, although children and most other animals infected with a gastric Helicobacter strain tend also to have a mononuclear infiltrate. In the corpus of the ferret stomach, the gastritis is minimal or only superficial. However, in the antrum, where the bacteria predominate, the inflammatory infiltrate may occupy the full thickness of the mucosa, similar to the diffuse antral gastritis seen in humans infected with H. pylori. Gastric glands may also be affected, with evidence of both atrophy and regeneration. Experimental inoculation of ferrets with H. mustelae produces elevated immunoglobulin G (IgG) titers to H. mustelae and a histologic gastritis 157 which resembles that seen in naturally infected animals 157. These changes are also accompanied by an elevation in meal-stimulated gastrin levels similar to what is seen in humans infected with H. pylori 320. Surprisingly, very minimal gastritis was seen in H. mustelae-infected ferrets in England 396. This observation has not been repeated, and it is unknown whether it was due to differences in the host, the pathogen, or both.

(ii) Ulcers. Unlike gastritis, which is seen in all cases, most patients infected with H. pylori do not develop peptic ulcer disease. While the association between H. pylori and peptic ulcer disease is undisputedthe current standard of care is to use antibiotics to treat patients with peptic ulcers and H. pylori infectionthe evidence for this association is nonetheless indirect. It is based primarily on the demonstration that ulcer recurrence in patients treated with antibiotics, with or without acid suppression, is markedly lower than in patients treated with acid suppression alone 184. Limited case-control data also show that preexisting H. pylori infection increases the risk for subsequent development of duodenal and gastric ulcers 302. In principle, H. mustelae infection in the ferret offers the opportunity to study experimentally the association between Helicobacter infection and peptic ulcer. Ferrets sometimes develop gastric ulcers and hemorrhagic gastric erosions 191, with a prevalence of 35% in a series of 31 ferrets in England 9 but only 1.4% in a large postmortem study of 350 animals performed by the same author 8. Gastric ulcers have also been found in related fur-bearing animals 26 and other laboratory species 269. However, there are no studies which provide convincing evidence that H. mustelae infection in the ferret causes peptic ulcer disease, although H. mustelae has occasionally been seen in ferrets with gastric ulcer. Such a demonstration would require long-term observation of experimentally infected and pathogen-free ferrets, which has not been done. Nevertheless, since H. mustelae-infected ferrets may develop duodenal ulcer or gastric ulcer, the latter of which may be associated with atrophic gastritis, dysplasia, and gastric adenocarcinoma, the model mimics in many respects the relationship between human hosts and infection with H. pylori.

(iii) Gastric malignancy. H. mustelae also provides an opportunity to study Helicobacter infection and gastric tumorigenesis, but, as with ulcer disease, the role of H. mustelae in gastric tumors of the ferret has not yet been clearly demonstrated. Gastric adenocarcinoma 333 and gastric mucosa-associated lymphoid tissue (MALT) lymphoma 110 have been found occasionally in the ferret, sometimes in association with H. mustelae infection 143. However, since ferrets are routinely infected with H. mustelae, infection alone must be insufficient to produce malignancy with any frequency, at least during the first few years of the animal's life, when most studies are performed. Treatment of H. mustelae-infected ferrets with N-methyl-N-nitro-N'-nitrosoguanidine (MNNG), a gastric carcinogen in several animal species, produced adenocarcinoma in 9 of 10 animals 159. Unfortunately, uninfected animals were not studied, and MMNG alone produces gastric adenocarcinoma in many species 166, 381, 382. H. mustelae-infected ferrets do have an increase in gastric epithelial-cell proliferation, especially in the antrum, where colonization is heaviest. This cell proliferation might promote the development of carcinoma in the setting of an appropriate initiation event 434. Although definitive evidence for the role of H. mustelae in gastric malignancy and peptic ulcer disease is lacking, the model remains a valuable tool for the study of gastritis and epithelial proliferation. Future studies will probably clarify the role of H. mustelae in the development of peptic ulcer disease and gastric cancer.

(iv) Therapy. Limited studies of antibiotic therapy for H. mustelae indicate that it is sensitive in vitro to many of the agents effective against H. pylori. The MIC of amoxicillin is 1 to 2 log units higher than for H. pylori 4, 221, 313, and therefore combination treatment containing amoxicillin may be less effective in ferrets than in humans 4, 313. The combination of ranitidine bismuth citrate and clarithromycin is effective therapy for H. mustelae infection in ferrets 263. Both these drugs are often used to treat H. pylori infection in humans, although this combination is not a generally recommended treatment regimen.

(i) Urease. H. mustelae produces a urease, composed of two subunits, that is similar in stoichiometry (hexameric 1:1), molecular mass (564 kDa), K_m (0.45 mM), and percentage of total cellular protein (2%) to that from H. pylori and other gastric Helicobacter species 94, 401, 402. Partial DNA sequencing of the H. mustelae urease genes showed that the predicted proteins were 67 to 68% identical to other Helicobacter ureases for UreA and 79% identical for UreB 370. The ureases of H. pylori, H. felis, and "H. heilmannii" are more closely related to one another than they are to the urease of H. mustelae. This finding is similar to the results of a phylogenetic study based on 16S rRNA sequence as well as G+C content 370, demonstrating that the urease structural genes can be used as a basis for phylogenetic analysis. An isogenic urease-negative mutant of H. mustelae does not colonize the ferret 11, much the same as has been found for H. pylori in the pig and mouse models 98, 101, 400. The mechanism for this failure is unknown, but it probably involves more than buffering of gastric acid, since an isogenic urease-negative H. pylori strain will not colonize the pig even if the gastric pH is raised pharmacologically 100. H. mustelae appears to have reduced survival at pH 6.0 if physiological concentrations of urea are present, which has been attributed to the accumulation of ammonia and its metabolism by the cell 423. However, a similar observation made with H. pylori appears to result from a rise in the pH of the medium rather than from ammonia accumulation 54. The apparent requirement of urease for colonization has led to an investigation of the therapeutic potential of fluorofamide, a potent urease inhibitor. Although fluorofamide markedly inhibited H. mustelae urease in vitro and in vivo 270, 324 and reduced bacterial numbers, it failed to eradicate H. mustelae, even when administered with amoxicillin. This may be due to inadequate drug delivery or to residual intracellular urease. It may also suggest that in vivo some bacteria are not completely dependent on urease for survival 324. H. mustelae has a gene identical to the H. pylori hpn locus, which codes for a protein with 47% histidine residues that binds nickel but is not required for urease activity 172.

(ii) Flagella. Flagellar organization in H. mustelae (and H. pylori) resembles that seen in Campylobacter, where the mature flagella are assembled from two flagellin proteins, FlaA and FlaB. The flaB gene has a sigma 54-type promoter that is functionally active, which suggests that it may be environmentally regulated. FlaA and FlaB of H. mustelae have 80% amino acid identity to the corresponding proteins in H. pylori. Unlike in Campylobacter, the Helicobacter flaA and flaB genes are not linked on the chromosome, nor are they closely related to one another 225, 380. Many other genes have been identified in the H. pylori genome that are probably involved in secretion, regulation, and assembly of flagella 395. One of these, the flagellar hook gene (flgE), has also been identified in H. mustelae 312, and it is likely that others will be found. A flaA isogenic mutant of H. mustelae has markedly diminished motility, while the motility of a flaB mutant is diminished 30 to 40% 225. Single-gene mutations in flaA or flaB reduce the density of colonization, whereas the nonmotile flaA flaB double mutant is unable to colonize the ferret 10. These results clearly indicate that motility is an important virulence factor for colonization in the H. mustelae ferret model.

(iii) Attachment. H. mustelae is found almost exclusively in intimate association with ferret gastric epithelial cells (Fig. 4), with few if any bacteria in the overlying mucus gel 309. Occasionally, bacteria are actually endocytosed into the gastric epithelial cells. Studies of attachment in Helicobacter have sometimes been plagued by a discrepancy between the striking host and tissue specificity seen in vivo and the sometimes nonspecific attachment seen in vitro. When examined by flow cytometry, H. pylori but not H. mustelae attached well to AGS cells, a human gastric carcinoma cell line. This observation suggests some specificity for the appropriate host cell 258, but H. mustelae adhered to explants of stomach from pigs, rats, and rabbits, as well as to pig duodenum and urinary bladder 279. H. pylori attachment to gastric epithelial cell lines causes pedestal formation similar to that seen with enteropathogenic Escherichia coli, accompanied by cytoskeletal rearrangements and tyrosine phosphorylation of host cell proteins 350. Similar adhesion pedestals have been seen with H. mustelae 309, but recruitment of filamentous actin was not apparent when examined in tissue culture by electron microscopy 176. The alpAB locus, which codes for outer membrane proteins that are required for in vitro attachment of H. pylori to gastric epithelium, is not present in H. mustelae or in H. felis 306.

(iv) Lipopolysaccharide. H. pylori lipopolysaccharide (LPS) has relatively low biological activity compared to that from the family Enterobacteriaceae 294, but it is of particular interest because of evidence that it expresses human Lewis (Le) antigens that are also present on the gastric epithelium 358. The relatively inactive LPS, combined with host antigens on its surface, may be a mechanism for H. pylori to down regulate and evade the host inflammatory response and thereby favor long-term colonization 36. Autoantibodies against the bacterial Le antigens may also be important in the pathogenesis of gastroduodenal pathology 15. H. mustelae LPS does not express Le antigens, nor are they expressed on ferret gastric epithelial cells. However, both ferret gastric epithelium and H. mustelae LPS express blood group antigen A, which may be a mechanism of molecular mimicry similar to expression of Le antigens by H. pylori 285, 305.

Spiral bacteria have long been seen on histologic sections of gastric mucosa from cats and dogs. In 1988 Lee et al. reported for the first time the culture of one of these organisms from the cat stomach 246. A similar organism was found in dogs, and both were designated H. felis 319.

Microbiology and phylogeny. H. felis is biochemically similar to other gastric Helicobacter species (Table 2). It has a helical morphology (Fig. 2), rather than the curved or loosely spiral appearance of H. pylori. H. felis is also characterized by periplasmic fibers, usually in pairs, that wind around the organism and have been used to distinguish it microscopically from the morphologically very similar but uncultivated "H. heilmannii." However, these fibers may not be present on all strains of H. felis and may disappear on subculture 96. Their function and genetic basis of expression are unknown. The sequence of the 16S rRNA gene from several cat and dog isolates of H. felis shows that they differ by less than 1% 96 and are most similar to other gastric Helicobacter species (Fig. 6; Table 3). The cat isolates are more closely related to one another, as are the dog isolates, but these differences are subtle. The fatty acid composition of H. felis is typical of other gastric Helicobacter species, with a predominance of 19-carbon cyclopropane fatty acid and tetradecanoic acid 198.

Epizootiology. It is now apparent that the many descriptions of pleomorphic spiral bacteria in the stomachs of dogs and cats 92, 255, 257, 353, 404, 417, 418 represent multiple species that often can be distinguished by 16S rRNA sequence analysis or DNA-DNA hybridization. These include, as well as H. felis, several other organisms discussed below such as H. bizzozeronii, H. salomonis, and "H. heilmannii." Still other isolates that may be novel species remain unnamed 96. The combination of enzyme-linked immunosorbent assay and immunoblotting detected Helicobacter infection with 95.6% sensitivity and 79.8% specificity in dogs 378, but efforts to specifically detect H. felis were less successful 351. Therefore, there are no seroepidemiologic studies which examine the prevalence of H. felis. Nevertheless, despite its name, H. felis is apparently not the most common Helicobacter species in cats and dogs. In six studies that collectively cultured gastric biopsy specimens from 147 cats and 85 dogs, one H. felis strain was isolated from a cat and only four strains were identified in dogs 96, 197, 298, 303, 315, 427. Even with modified culture conditions that appear more effective than those used previously, H. felis was cultivated from only 3 of 22 cats and 8 of 95 dogs 220. Most animals were colonized by organisms that resembled "H. heilmannii" and could not be cultivated. Human infection with H. felis has been reported rarely 168, 237, apparently as a zoonosis, but no environmental or other host reservoir is known.

Clinical significance. Much as was true of the early descriptions of H. pylori infection in humans, the clinical significance of gastric Helicobacter infection in dogs and cats has been difficult to determine. Most studies find that between 50 and 100% of cats and dogs are infected with Helicobacter. There is no convincing evidence that infection is associated with clinical symptoms such as chronic vomiting or inappetence, nor is there typically a clear association between infection and histologic gastritis 96, 116, 169, 197, 200, 202, 203, 298, 303, 314, 315, 321, 353, 377, 417, 418, 427. However, in many studies the animals are incompletely described, there are no controls, and the bacteria are only characterized by urease assay or routine histologic testing. Since animals may be infected with more than one species of Helicobacter that are urease positive and that are indistinguishable on light microscopy, the pathogenic role of H. felis is difficult to determine from these observational studies. Experimental infection of H. felis in 7-day-old gnotobiotic dogs resulted in seroconversion and large numbers of lymphoid nodules throughout the gastric mucosa, as well as a mild diffuse lymphocytic infiltrate 247. Similar nodules described previously as components of the normal microscopic anatomy of the dog stomach may also have been a result of H. felis infection 1. Control animals had no evidence of infection or lymphoid nodules at necropsy. However, when 4-month-old specific-pathogen-free dogs were infected with the same strain of H. felis, the results were very different 363. Mild gastritis and lymphoid follicles were found in both infected and uninfected dogs. There was no correlation between the number of organisms and the intensity of inflammation, nor did infection produce alterations in gastric pH, acid output, or plasma gastrin. The different results in these two studies probably reflect host differences, such as alterations in gastric acidity or immune response with age, or differences between specific-pathogen-free and gnotobiotic animals. Experimental inoculation of specific-pathogen-free cats with H. felis induced lymphoid follicular hyperplasia but only mild gastritis 364. There was no accompanying up regulation of antral mucosal interleukin-1 (IL-1), IL-1, or tumor necrosis factor alpha, nor were there changes in gastric secretory function.

Virulence determinants and other characterized genes. The H. felis urease is composed of A and B subunits, which are 73.5 and 88.2% identical, respectively, to the corresponding polypeptides from H. pylori 117, 181. One presumes that urease is required for H. felis colonization, but efforts at genetic manipulation have been largely unsuccessful. Recently, the flagellin genes from H. felis (flaA and flaB) were cloned and isogenic mutants were generated by electroporation 224. Transformation efficiency was low relative to what is typically seen with H. pylori or H. mustelae, and mutants could be obtained only using plate-grown bacteria. Both flaA and flaB mutants showed truncated flagella and were poorly motile in vitro, and the flaA mutant was unable to colonize gastric mucosa in the mouse model. This result differs from that with H. mustelae, in which mutation in a single flagellin gene reduces but does not abolish colonization. Unlike H. mustelae, H. felis is found exclusively in the mucus layer and is not attached to the gastric epithelium 348; it has been speculated that any impairment of motility may therefore eliminate colonization 224.

Development of the rodent model. H. felis readily colonizes mice, with the same gastric tropism seen with H. pylori in humans 77, 134, 244. Infection typically predominates in the gastric antrum but, interestingly, is accompanied by a largely mononuclear cell inflammatory response that is more prominent in the corpus 134, 281, 341. The difference in the anatomic locations of the infection and the inflammation has led to the suggestion that H. felis in mice may stimulate an autoimmune response to gastric tissue 341. A similar mechanism has been proposed for H. pylori gastritis in humans based on molecular mimicry between bacterial LPS and host Le antigens 15. As inflammation progresses into atrophic gastritis, substantial reduction in bacterial colonization of the gastric antrum may develop 341. This can also be produced, along with increased colonization of the corpus, by treatment of H. felis-infected mice with the proton pump inhibitor omeprazole 64. These findings suggest that local acid production may be important in the distribution of H. felis in the mouse stomach. Unlike H. pylori in humans, H. felis in the mouse does not attach intimately to gastric epithelial cells, nor does it produce a prominent polymorphonuclear infiltrate.

(i) Atrophic gastritis, lymphoma, and gastric cancer. Atrophic gastritis (thought to be the histologic precursor lesion to gastric adenocarcinoma) and gastric MALT lymphoma associated with H. pylori infection in humans 56, 435 have also been observed in mice infected with H. felis 239. In an initial study of conventional mice experimentally infected with H. felis CS1, atrophic changes in the gastric corpus developed 48 weeks after inoculation and progressively increased over the subsequent 24 weeks 241. Uninoculated animals developed less extensive atrophy, but the results were difficult to interpret because animals in both groups developed coinfection with Helicobacter muridarum, an organism that normally colonizes the small and large bowel of rodents. However, subsequent inoculation of specific-pathogen-free mice with H. felis confirmed the development of corpus atrophy without coinfection by H. muridarum 341. Long-term infection of specific-pathogen-free mice with H. felis can also produce lesions that resemble human gastric B-cell MALT lymphoma 109. Antibiotic treatment of H. felis in chronically infected mice reduces the development of gastric MALT lymphoma 108. This suggests that the lymphoma is antigen dependent, at least while it remains low grade, in much the same fashion as H. pylori-associated gastric MALT lymphoma in humans 211, 335. Elegant studies with insulin-gastrin (INS-GAS) transgenic mice also suggest that H. felis infection acts synergistically with chronic hypergastrinemia to produce parietal cell loss and development of gastric cancer 410. Further studies with transgenic mice, as well as the recently described H. pylori gerbil model 383, will probably contribute significantly to our understanding of the relationship between Helicobacter infection and gastric malignancy.

(ii) Therapy. The H. felis mouse model has been promoted as a preclinical tool for evaluation of novel antimicrobial therapy for H. pylori infection. In general, the results obtained by treatment of H. felis in the mouse model mimic the outcome of human clinical trials investigating the treatment of H. pylori 78, 206, 233, 367. However, given the relative ease of performing human clinical studies, hundreds of which have now been completed 403, and the development of the H. pylori mouse model, it seems unlikely that the H. felis mouse model will play an important role in efficacy studies of novel H. pylori therapies.

(iii) Mechanisms of inflammation and atrophy. The nature of the host response is an important factor in Helicobacter-associated diseases. Infection with H. felis in several inbred strains of mice (C57BL/6, C3H/He, and SJL) produces severe gastritis, while in other strains (BALB/c, CBA) the inflammatory response is much less marked 283, 341. F₁ hybrids of the CBA mouse crossed with strains that develop severe gastritis maintained the mild inflammation phenotype of the CBA parent 384. Thus, low inflammation was a dominant response which might in part reflect immune suppression rather than a genetic defect. It has also been proposed that increased proliferation and apoptosis seen in the C57BL/6 mice may be related to the lack of secretory phospholipase A₂ (sPLA₂), which is encoded by the Mom1 locus responsible for variability in the number of polyps in mice with multiple intestinal neoplasia 411. SV129 mice are also sPLA₂^/ and show similarly severe pathologic changes in response to H. felis infection, while BALB/c and C3H/HeJ mice are sPLA₂^+/+ and have less inflammation. The decrease in gastric sPLA₂ levels after infection of C57BL/6 mice with H. felis is also consistent with the role of sPLA₂ in maintaining gastric differentiation 254. However, since these studies were not performed on congenic strains, other genes may also be involved. For example, C3H/HeJ mice have a defect in responsiveness to LPS that also contributes to reduced atrophic gastritis 342. Genes of the major histocompatibility complex probably also contribute to individual differences described in mouse strains 283.

(iv) Vaccine development. The limitations of current H. pylori therapies have prompted interest in the development of vaccines for both treatment and primary prevention. Although natural infection with H. pylori induces specific IgG and IgA that do not prevent initial colonization or subsequent reinfection, the results of considerable work suggest that it may be possible to prevent Helicobacter infection by immunization. Much of this work has utilized the H. felis mouse model. Orogastric immunization of mice with an H. felis sonic extract together with cholera toxin produces a significant antibody response in serum and gastric secretions 62 that is associated with protection after challenge with H. felis, often with 90 to 100% efficacy 33, 51, 61, 118-120, 240, 316, 328, 352. H. pylori sonic extract plus cholera toxin also protects against challenge with H. felis, although less effectively 240, 280. Studies of particular antigens suggest that protection against H. felis infection in the mouse model can be achieved with H. pylori urease holoenzyme or its B subunit 112, 118, 250, 280, 297, 316, catalase 329, and the GroES and GroEL homologs, HspA and HspB, respectively 119. Similar vaccine efficacy has also been described in the H. pylori mouse model by immunization with H. pylori cytotoxin (VacA), a protein associated with cytotoxin expression (CagA), catalase, and urease 58, 170, 179, 261, 262, 329.

Large gastric spiral bacteria, which resemble those seen by early investigators and described most clearly by Lockard and Boler 257, were recently cultivated 195, 219 and named after Bizzozero 30 and Salomon 344. Both organisms were isolated from dogs by using culture techniques that differed only modestly from those used by other investigators, particularly in using brain heart infusion rather than brucella broth, cultivation for up to 12 days, careful attention to keeping plates moist, and cultivation only of biopsy specimens that were rapidly urease positive. H. bizzozeronii, which is indistinguishable morphologically from the Lockard type 3 bacterium and from "H. heilmannii," is typically 5 to 10 µm long and 0.3 µm wide, with bipolar sheathed flagella (Fig. 3). H. salomonis is usually somewhat smaller (5 to 7 by 0.8 to 1.2 µm) and less tightly coiled, a morphology similar to that originally published by Weber and Schmittdiel 418 but not described by Lockard and Boler. The 16S rRNA sequences from these bacteria are approximately 99% similar to one another and to that from the closely related H. felis. Restriction fragment length polymorphism (RFLP) analysis suggests that the 23S rDNA genes are also closely related 218. However, distinct species occasionally have 16S rRNA genes that are virtually identical 133. In this case, DNA-DNA hybridization studies show clearly that H. bizzozeronii and H. salomonis are each genetically homogeneous and distinct from one another, as well as from other canine and feline gastric Helicobacter species. Pulsed-field gel electrophoresis suggests that there is more heterogeneity among strains of H. bizzozeronii than among strains of H. salomonis 196, although this result will require replication with a larger number of strains.

Cultivation of Helicobacter species from dogs and cats has been typically unsuccessful, despite microscopic and DNA evidence of Helicobacter 96, 298, 303, 427. When the culture methods used originally to isolate H. bizzozeronii and H. salomonis were applied to specimens from a group of dogs and cats, the results showed that the seemingly subtle modifications in culture methods were probably important 220. Cultures of specimens from 48 (51%) of 95 dogs and 3 (14%) of 22 cats were positive, which was considerably greater than the 0 to 10% that has recently been reported 96, 298, 303, 427. Approximately half the positive cultures from dogs yielded H. bizzozeronii. The remainder were about evenly divided between H. felis and H. salomonis, with "Helicobacter sp. flexispira" (see "Enterohepatic Helicobacter species" below) cultivated in two dogs. The three positive cultures in cats were all H. felis. H. bizzozeronii, H. salomonis, and H. felis were difficult to distinguish unequivocally by using morphology, bacteriology, routine biochemistry, or 16S rRNA sequence analysis. However, numerical analysis of whole-cell-protein electrophoresis results and calculation of percent similarity yielded clusters that best corresponded to the three different species.

In 1987 a novel gram-negative spiral bacterium was found in gastric biopsy specimens from three patients, although in retrospect the organism may have been the same as that described in humans many years earlier 235. It was easily distinguished from H. pylori by virtue of its larger size, more tightly coiled morphology, and failure to grow in microaerobic culture 72. Although it was recognized early that the organism closely resembled bacteria seen in a variety of other mammals, it was nevertheless tentatively assigned a new genus and species designation, "Gastrospirillum hominis," which reflected its occurrence in humans 271. The initial observation was quickly confirmed by numerous other case reports that described similar organisms in small numbers of patients 2, 5, 38, 85, 95, 121, 122, 124, 204, 213, 284, 289, 307, 385, 426, 428, 430; L. Mazzucchelli, Letter, Dig. Dis. Sci. 40:1463, 1995).

Taxonomy and nomenclature. In vitro cultivation of this large gastric spiral bacterium remains elusive, but inoculation of mice and rats with gastric homogenates from infected humans and nonhuman primates permitted cultures to be maintained in vivo 77, 243. Mouse gastric tissue infected with organisms from two human patients was then used as a DNA template to amplify and sequence bacterial 16S rRNA genes, a technology that is now commonplace for the identification of uncultivated bacteria 347. The results confirmed early speculation based on antibody cross-reactivity 242 that the large gastric spiral organism originally designated "Gastrospirillum hominis" was actually a new species of Helicobacter that is most closely related to H. felis 371. The bacterium was tentatively designated "Helicobacter heilmannii" 371, after Konrad Heilmann, a German histopathologist who at the time had described the largest series of cases and who died prematurely shortly after its publication 201.

Microbiology. The morphology of "H. heilmannii" is similar to that of H. felis, but periplasmic fibers are absent. The organism is 4 to 10 µm in length and 0.5 to 0.8 µm in diameter and has four to eight tight spirals (Fig. 3). There are typically 6 to 10 tufts of bipolar flagella. "H. heilmannii" is well visualized and distinguished from H. pylori by light microscopic examination of paraffin sections stained with hematoxylin and eosin or Warthin-Starry silver stain. Surprisingly, a recent report found that H. pylori can be induced to assume the morphology of "H. heilmannii" by being grown in brucella broth with 1% cyclodextrin rather than on blood agar 115. Since 16S rDNA sequences amplified from tissue infected with an "H. heilmannii"-like organism do not typically resemble H. pylori, it is unlikely that this observation is often relevant in vivo, but further study is warranted.

Host range and epidemiology. The prevalence of infection with "H. heilmannii"-like organisms in humans is less than 0.5% among patients presenting for upper gastrointestinal endoscopy 124, 201, 271, 289, although it is reportedly as high as 6% in China and Thailand 425, 429. This latter observation requires confirmation by studies in other developing countries. In contrast to the low prevalence in humans, infection is very common in dogs 96, 427, cats 298, 303, 427, pigs 20, 326, and nonhuman primates 88, 91, 369. Recent studies also suggest that "H. heilmannii"-like organisms may infect wild urban rats 173 and both small and large exotic felids 104, 217, 227, 228. Thus, unlike the host range restriction of many Helicobacter species, such as H. pylori in humans and other primates, H. felis in cats and dogs, and H. mustelae in ferrets, "H. heilmannii"-like organisms are widely distributed in the animal kingdom. To date, "H. heilmannii" infection has been confirmed by 16S rDNA sequence analysis (rather than morphology alone) only in humans 371, pigs 70, 326, and nonhuman primates (Solnick and O'Rourke, unpublished), although it seems likely that it will be identified in other hosts.

Histopathology. The histopathology of "H. heilmannii"-like infection in humans was described for 39 cases in the first large series reported by Heilmann 201 and more recently in a study that compared 202 patients with "H. heilmannii" infection to 202 matched controls infected with H. pylori 376. Compared with H. pylori, "H. heilmannii" infection in humans is more often focal, with fewer organisms, and is often restricted to the gastric antrum. "H. heilmannii" is typically found in the mucus layer above surface epithelial cells and does not show the intimate adherence and pedestal formation often seen with H. pylori. Organisms may also be found deep in the lumen of the gastric glands and within parietal cells. Gastritis, while present, is much less marked in patients