Background

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 (46) described spiral or curved bacilli in histologic sections from 58 of 100 consecutive biopsies of human gastric antral mucosa. Eleven biopsies were culture positive for a gram-negative, microaerophilic bacterium, whose proper classification generated considerable interest at the Second International Workshop on Campylobacter Infections held in Brussels, Belgium, in September 1983 (57). The organism resembled Campylobacter in several respects, including curved morphology, growth on rich media under microaerophilic conditions, failure to ferment glucose, sensitivity to metronidazole, and a G + C content of 34%. It was therefore first referred to as "pyloric Campylobacter" (pylorus, Greek, gatekeeper, or one who looks both ways) and validated as Campylobacter pyloridis in 1985 (1). The specific epithet was revised to Campylobacter pylori in 1987 to conform to the correct Latin genitive of the noun pylorus (45).

Yet almost from its initial cultivation it was suspected that perhaps C. pylori was not a true Campylobacter. Early electron micrographs showed multiple sheathed flagella at one pole of the bacterium, in contrast to the single bipolar unsheathed flagellum typical of Campylobacter spp. (28). Major protein bands and fatty acids of C. pylori were also markedly different from those of Campylobacter species (28, 56). Subsequent 16S rRNA sequence analysis showed that the distance between C. pylori and the true campylobacters was sufficient to exclude it from the Campylobacter genus (60), and it was renamed Helicobacter pylori, the first member of the new genus Helicobacter (26).

Cellular Morphology and Ultrastructure

Helicobacters are non-spore-forming gram-negative bacteria. The cellular morphology may be curved, spiral, or fusiform, typically 0.2 to 1.2 μm in diameter and 1.5 to 10.0 μm long. The spiral wavelength may vary with the age, the growth conditions, and the species identity of the cells. In old cultures or those exposed to air, cells may become coccoid.

Periplasmic fibers or an electron-dense glycocalyx or capsule-like layer has been observed on the cellular surface of several species (26, 43, 55, 63). Electron-dense granular bodies have been observed in H. pylori (4) and H. rodentium (63). In H. pylori these bodies are known to be aggregates of polyphosphate and may serve as a reserve energy source.

Helicobacter cells are motile, with a rapid cork-screw-like or slower wave-like motion due to flagellar activity. Strains of most species have bundles of multiple sheathed flagella with a polar or bipolar distribution. Other species have only a single polar or bipolar flagellum (Table 1). However, flagellation can be peritrichous (H. mustelae) or nonsheathed (H. pullorum, H. rodentium, and "H. mesocricetorum") as well.

Table 1

Characteristics of cultivated Helicobacter species^a.

Growth Characteristics

In laboratory conditions, strains typically grow under microaerobic conditions at 37°C. No growth is observed in aerobic conditions. H. rodentium strains grow in anaerobic as well as microaerobic conditions (63). Strains of several species do not require atmospheric hydrogen for growth in microaerobic conditions, although there may be a growth-stimulating effect of atmospheric hydrogen on the growth of such strains. However, as for some Campylobacter species (e.g., Campylobacter hyointestinalis), hydrogen requirements for microaerobic growth of Helicobacter may be strain dependent. For instance, the isolates initially described as belonging to a novel anaerobic species, "Helicobacter westmeadii," were subsequently identified as hydrogen-requiring H. cinaedi strains that grow optimally in microaerobic conditions (77). Many laboratories routinely use microaerobic conditions including hydrogen for the routine culture of Helicobacter species and the role of this atmospheric component is therefore not well documented.

Helicobacters will grow at 37°C on a variety of rich agar bases supplemented with 5% whole blood or serum. Many species require fresh media with moist agar surfaces for optimal growth conditions, though this is not usually the case for H. pylori. In such conditions, growth is often not seen as individual colonies but as a thin, sometimes colorful or watery, spreading film. The most fastidious species such as H. bizzozeronii and H. salomonis prefer very moist conditions, and a thin broth layer can be poured on top of the agar surface to stimulate growth. Prolonged incubation up to 1 week may be required.

Biochemical Characteristics

Helicobacters are chemoorganotrophs and show a respiratory type of metabolism. They are asaccharolytic when sugar catabolism is examined by standard methods (neither oxidation nor fermentation is observed). Recent studies have, however, indicated that glucose oxidation occurs in at least H. pylori (6, 72) (see chapter 10). The glycolysis-gluconeogenesis pathway probably comprises the principal means of energy production as well as the starting point for many biosynthetic pathways. The Entner-Doudoroff pathway, the pentose phosphate shunt, and the tricarboxylic acid cycle are at least partially present; the glyoxylate shunt is absent (6, 72).

Gelatin, starch, casein, and tyrosine are not hydrolyzed. Helicobacters are methyl red and Voges-Proskauer negative. Oxidase activity is present in all species. Strains of most species produce catalase. Many species produce urease, alkaline phosphatase, or both. There is no production of pigments.

Other Characteristics

The moles percent G + C range of the DNA ranges from 30 (H. acinonychis) to 48 (H. canis). One species, H. nemestrinae, was reported to have a DNA base ratio of 24% (5). However, repeat analyses of the H. nemestrinae type strain yielded a reproducible value of 39 mol% (P. Vandamme, unpublished observations). It would therefore be appropriate to do a comparative analysis of various subcultures of the H. nemestrinae type strain to verify the strain identity in the different public and private culture collections.

Internal transcribed spacers or intervening sequences have been described in the 16S ribosomal RNA genes of H. canis (44), H. bilis (19), "H. typhlonicus" (24), and in an unnamed helicobacter referred to as Helicobacter sp. cotton-top (61). Similar intervening sequences have also been found in the 23S rRNA gene of H. canis, H. mustelae, and H. muridarum (34). The length of intervening sequences in the 16S rRNA gene ranges from 187 bp to 235 bp, while those found in the 23S rRNA gene range from 93 bp to 377 bp.

The cellular fatty acid components have been studied for a restricted number of Helicobacter species. Major cellular fatty acids reported include tetradecanoic acid; most species contain moderate to high levels of hexadecanoic acid and octadecanoic acid (27, 41).

The isoprenoid quinone content of H. pylori, H. cinaedi, and H. fennelliae has been determined and identifies menaquinone-6 (2-methyl-3-farnesyl-farnesyl-1,4-naphthoquinone) and a second, unidentified quinone as major respiratory quinones (51).

Classical Phenotypic Characteristics

Most routine laboratories apply the same basic biochemical tests for the identification and differentiation of all Campylobacter-like organisms and would fail to identify many Helicobacter species. Although the number of Helicobacter species encountered in human clinical samples is fairly small, the lack of application of highly standardized procedures and the well-known biochemical inertness of Campylobacter-like organisms render biochemical identification of all of these bacteria very difficult. Whereas Arcobacter strains can be differentiated from Campylobacter and Helicobacter strains by their ability to grow in air and at low temperature (79), there are no clear biochemical characteristics to separate the genus Helicobacter from the genus Campylobacter. Theoretically, one has to differentiate over 35 validly named species and subspecies, as well as various unnamed taxa.

An overview of biochemical and other methods to differentiate Campylobacter and Arcobacter species was described earlier (75). A summary of the characteristics of cultivated Helicobacter species shows that discrimination between some species may rely on only one differential feature (Table 1). Moreover, some species, notably H. pylori and H. acinonychis, and H. felis and H. bizzozeronii, cannot be differentiated with conventional phenotypic tests. In addition, it should be noted that H. pullorum bears a close resemblance to certain Campylobacter species (notably C. lari).

DNA-DNA Hybridization

In taxonomic practice, the reference method for the delineation and identification of bacterial species is determination of the level of DNA-DNA hybridization (81). Strains are considered to belong to a single species if their whole genome DNA-DNA hybridization level is about 70% or greater. The fastidious growth characteristics of many Helicobacter species hamper the isolation of sufficient quantities of highly purified high molecular weight DNA required for these hybridization experiments. Yet, a number of DNA-DNA hybridization studies have been performed within and between Helicobacter species. Significant but low levels of DNA-DNA hybridization (below 70%) have been reported between (i) H. pylori and H. mustelae (21); (ii) H. cinaedi, H. fennelliae, and H. canis (68); and (iii) H. felis, H. bizzozeronii, and H. salomonis (38). Most other species have not been included in quantitative DNA-DNA hybridization studies.

Protein Electrophoresis

It is not practical to implement DNA-DNA hybridizations in a routine laboratory or to use it for routine identification in a reference laboratory. The comparison of whole-cell protein patterns obtained by highly standardized sodium dodecyl sulfate-polyacrylamide gel electrophoresis (PAGE) has proven to be extremely reliable to screen and identify large numbers of strains. Numerous studies revealed a correlation between high similarity in whole-cell protein content and level of DNA-DNA hybridization (78). Like for Campylobacter species, PAGE of whole-cell proteins has also been a most successful method for the differentiation of helicobacters (7, 40, 67, 76, 77, 79, 80), and strains of all Helicobacter species are readily distinguished by means of simple numerical analysis. However, this method is not appropriate for routine identification studies because it is laborious, time-consuming, and technically demanding to run the patterns in a sufficiently standardized way.

Cellular Fatty Acid Analysis

The total cellular fatty acid methyl ester composition is a stable parameter provided that highly standardized culture conditions are used. Comparison of fatty acid profiles is of little value if different culture conditions or extraction procedures are used. However, it is a simple, inexpensive, and rapid method that is highly automated. Only a few of the presently known Helicobacter species have been included in published studies (27, 41). H. cinaedi, H. fennelliae, H. pylori, H. mustelae, and H. nemestrinae were readily distinguished (27, 41).

DNA Base Ratio

The DNA G + C ratio of Helicobacter species generally ranges from 30 to 48 mol%. Table 1 lists the known DNA base compositions of the cultivated Helicobacter species.

Ribosomal DNA Restriction Analysis

Restriction profile analysis of PCR amplicons derived from the 23S rRNA gene (34) was shown to differentiate 13 Helicobacter species (including "F. rappini") but cannot unequivocally discriminate H. felis, H. bizzozeronii, and H. salomonis (37).

DNA Probes and PCR Assays

Specific oligonucleotide probes or PCR assays have been described for H. pylori (54) and several other Helicobacter species, including H. pametensis, the Helicobacter Bird-B and Bird-C groups, H. felis, H. hepaticus, H. pullorum, H. bilis, H. trogontum, H. canis, "Candidatus Helicobacter suis," and "Candidatus Helicobacter bovis" (3, 8–10, 19, 25, 48, 68). It should be stressed that, because of the constant developments in the taxonomy of Helicobacter species, none of these probes or PCR assays have been fully evaluated against all species presently described. For instance, the 16S rDNA gene sequences of multiple strains of H. felis, H. bizzozeronii, and H. salomonis have become available during the past few years, revealing a 98 to 100% similarity in 16S rDNA sequence among strains of these species. The specificity of the H. felis PCR assay (25) should be reevaluated in light of these new findings.

Sequence Analysis of the Small Subunit rDNA

Comparison of nearly complete 16S rDNA sequences is probably the most powerful tool to establish the phylogenetic neighborhood of an unknown organism. Sequence analysis of 16S RNA genes has become a popular tool and is accessible to a wide scientific community. Particularly with organisms that are otherwise difficult to identify, such as helicobacters, it has become common practice to determine the 16S rDNA sequence, or a fragment of it, and compare this with sequences available in public databases. The pitfalls of such an identification approach will be discussed below.

Gastric Helicobacter Species

The well-documented host range of gastric Helicobacter species extends from humans and nonhuman primates to a variety of domesticated and feral land mammals. However, it is likely that the extent of gastric Helicobacter species is even broader than this. A preliminary report (22) described a Helicobacter species cultured from the true (glandular) stomach of the white-sided dolphin (Lagenorhynchus acutus). There is also microscopic evidence that reptiles have gastric spiral bacteria, which will likely be one or more Helicobacter species (Karen Terio, personal communication). The defining feature of gastric Helicobacter species is that they all express urease, which appears to be essential for colonization of this niche. Other biochemical and morphologic characteristics are variable (Table 1).

There are currently seven validly named gastric Helicobacter species, two Candidatus species, and one species that has been described but not yet validated (Table 2). Uncultivated bacteria referred to descriptively as gastrospirillum or as "Gastrospirillum hominis" were identified as helicobacters when 16S rRNA analysis showed that they clearly belong to the Helicobacter genus (65). The bacterium was tentatively designated "H. heilmannii," 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. However, 16S rRNA sequences between clones derived from two patients differed significantly, suggesting that they might be different species. We originally referred to these as "H. heilmannii" type 1 and "H. heilmannii" type 2 (65). "Candidatus Helicobacter suis" is nearly identical by 16S rRNA gene sequence to "H. heilmannii" type 1, whereas "H. heilmannii" type 2 clusters with H. felis, H. salomonis, H. bizzozeronii, and others. The taxonomic difficulties raised by these and other morphologically indistinguishable bacteria are discussed below in more detail.

H. nemestrinae, isolated from the pigtailed macaque (Macaca nemestrina), was reported to differ from H. pylori by virtue of its growth at 42°C and cellular fatty acid profile (5). Studies of DNA-DNA hybridization and 16S rDNA also suggested that it was a novel species most closely related to H. pylori. Surprisingly, the G + C content was only 24%, much lower than that of all known Helicobacter species, which prompted a reappraisal. The results suggest that H. nemestrinae (ATCC 49396) may in fact be identical to H. pylori. Repeat determination of the 16S rDNA sequence (Floyd Dewhirst, unpublished observations) showed that it is 99.5% identical to human H. pylori and 100% identical to H. pylori isolated from the rhesus macaque (13). Repeat analysis of DNA base composition indicates that H. nemestrinae (ATCC 49396) is 39% G + C (identical to H. pylori) and that the protein profile is also consistent with H. pylori (Peter Vandamme, unpublished observations). Confirmation must await publication of these results.

Enterohepatic Helicobacter Species

The first enterohepatic Helicobacter species was isolated from conventional Wistar rats and BALB/c mice by Phillips and Lee in 1983 (58), almost 10 years before it would be formally named H. muridarum. Since then, a diverse group of Helicobacter species has been found in the intestinal tract and/or hepatobiliary system of humans, other mammals, and birds. Because of potential implications for a broad range of research with laboratory rodents, mice and rats have been most intensively studied, though the host range of these organisms is clearly broad and probably not yet fully defined.

There are currently 11 validated enterohepatic Helicobacter species and several others that await validation or further description (Table 3). Organisms isolated from feces of terns, gulls, a house sparrow, and swine were identified as three biotypes of Helicobacter, originally denoted Helicobacter sp. Bird-A, -B, and -C. The Helicobacter Bird-A isolated was later classified as H. pametensis (10); the other closely related organisms are not formally named and are still referred to as Helicobacter sp. Bird-B and Bird-C because they are represented by only two isolates for one species and one for the other (10). Similarly, Helicobacter CLO-3 isolated from a rectal culture in a homosexual man (16) appears to represent a unique species (68) but has not been formally named because additional isolates have not been identified.

Unlike helicobacters that colonize the stomach, expression of urease in enterohepatic Helicobacter species is variable (Table 1). While these organisms are not normally found in the stomach, occasionally urease-positive species can exploit an altered gastric microenvironment and displace the gastric spirals. This phenomenon has been observed primarily with H. muridarum (42). Some enterohepatic Helicobacter species superficially resemble Campylobacter species but are longer and have a single polar flagellum at each end. Others have periplasmic fibers that wrap helically around the body of the bacterium and give a cross-hatched appearance in negative stain electron micrographs. This morphologic distinction is descriptively convenient, but it has no phylogenetic basis. Like the large group of organisms with the gastrospirillum morphology, which are phylogenetically diverse, recent data suggest that strains that have been called "Flexispira rappini" represent multiple Helicobacter taxa, including the two named species, H. bilis and H. trogontum.

The Problem of Intraspecies Divergence of rDNA Genes

Like for other fastidious and inert organisms, comparison of partial or nearly complete 16S rDNA sequences has regularly been used to identify novel Helicobacter isolates. Although unsurpassed in its capacity to reveal the phylogenetic neighborhood of an unknown bacterium, comparison of entire 16S rDNA sequences and certainly of partial sequences is not always adequate for species-level identification of strains. There is a lack of knowledge not only of the strain-to-strain variation within a species, but also of the interoperon variation within a single strain. An important pitfall is that 16S rDNA sequences may be too conserved to reveal diversity among species. In Helicobacter species, this is illustrated by comparison of 16S rDNA gene sequences of multiple H. felis, H. bizzozeronii, and H. salomonis strains (38). Although DNA-DNA hybridization studies demonstrated unambiguously that these three taxa represent three distinct species (all three can be distinguished by whole-cell protein electrophoresis as well), all three species have virtually identical 16S rDNA genes (38) (Fig. 1). This failure of 16S rDNA sequence analysis to distinguish closely related species is obviously not restricted to the Helicobacter genus (17, 78).

The lack of sufficient variability in 16S rDNA sequences is not the only rDNA-related problem. Helicobacter isolates have been classified as novel species or putative novel species based on a distinct position in a phylogenetic tree, and on one or a few differential biochemical characteristics, thereby ignoring the putative intraspecies diversity in the 16S rDNA gene sequence or in the biochemical reactivity pattern. The reported intraspecies diversity in 16S rDNA sequences of some Campylobacter and Helicobacter species is exceptionally high (31, 77). Whereas intraspecies diversity in 16S rDNA sequence of up to 3% is generally not unusual, differences up to 4.3% among C. hyointestinalis strains (31) and up to 4.5% among H. cinaedi strains (77) have been reported. Some of these H. cinaedi strains share higher percentages of their 16S rDNA sequence with strains of species like H. canis, H. bilis, or H. fennelliae than with other strains of H. cinaedi. In a phylogenetic tree based on 16S rDNA, such strains may cluster far apart and erroneously suggest that they represent novel species (35, 77).

In such a situation when no additional taxonomic analyses are performed, it is important to realize that there are no criteria to balance the significance of detected phenotypic differences among isolates. "H. westmeadii" strains, although grouping very closely with H. cinaedi reference strains in the 16S rDNA phylogenetic tree, were considered to be distinct species because of their anaerobic growth requirements and supposed biochemical differences from H. cinaedi reference strains (74). The failure to grow the isolates under optimal conditions (a microaerobic atmosphere with hydrogen) probably led to at least a few equivocal biochemical test results. When grown under optimal conditions and compared with the biochemical profile of a large panel of H. cinaedi strains, the "H. westmeadii" isolates conformed to the general biochemical reactivity pattern of H. cinaedi (77). It is essential to use an independent taxonomic tool to balance the significance of such biochemical differences when they occur. In the case of "H. westmeadii," both isolates were readily identified as H. cinaedi by means of whole-cell fatty and whole-cell protein analysis (74, 77), indicating that the biochemical differences revealed intraspecies diversity rather than interspecies diversity.

The value of 16S rDNA sequence analysis as a tool to reveal the taxonomic neighborhood of unclassified isolates is beyond dispute. However, whole-cell protein or fatty acid analysis, extended biochemical testing, or restriction profile analysis of PCR amplicons derived from the 23S rRNA gene (see above) should be considered to endorse tentative identification results obtained by comparison of complete 16S rDNA genes, and to balance the significance of potential biochemical differences. As useful as the 16S rDNA sequence analysis method is, present data clearly indicate that it cannot be regarded as the gold standard for species-level identification of helicobacters and other epsilon Proteobacteria species. It should be clear that further descriptions of putative new species should be made with due care and attention to the present view that interstrain relationships should be described by means of a polyphasic taxonomic analysis.

The Problem of Morphologically Defined Taxa: "Flexispira" and "Gastrospirillum"