The Evasive Enemy: Insights into the Virulence and Epidemiology of the Emerging Attaching and Effacing Pathogen Escherichia albertii

The diarrheic attaching and effacing (A/E) pathogen Escherichia albertii was first isolated from infants in Bangladesh in 1991, although the bacterium was initially classified as Hafnia alvei. Subsequent genetic and biochemical interrogation of these isolates raised concerns about their initial taxonomic placement.

ABSTRACT

The diarrheic attaching and effacing (A/E) pathogen Escherichia albertii was first isolated from infants in Bangladesh in 1991, although the bacterium was initially classified as Hafnia alvei. Subsequent genetic and biochemical interrogation of these isolates raised concerns about their initial taxonomic placement. It was not until 2003 that these isolates were reassigned to the novel taxon Escherichia albertii because they were genetically more closely related to E. coli, although they had diverged sufficiently to warrant a novel species name. Unfortunately, new isolates continue to be mistyped as enteropathogenic E. coli (EPEC) or enterohemorrhagic E. coli (EHEC) owing to shared traits, most notably the ability to form A/E lesions. Consequently, E. albertii remains an underappreciated A/E pathogen, despite multiple reports demonstrating that many provisional EPEC and EHEC isolates incriminated in disease outbreaks are actually E. albertii. Metagenomic studies on dozens of E. albertii isolates reveal a genetic architecture that boasts an arsenal of candidate virulence factors to rival that of its better-characterized cousins, EPEC and EHEC. Beyond these computational comparisons, studies addressing the regulation, structure, function, and mechanism of action of its repertoire of virulence factors are lacking. Thus, the paucity of knowledge about the epidemiology, virulence, and antibiotic resistance of E. albertii, coupled with its misclassification and its ability to develop multidrug resistance in a single step, highlights the challenges in combating this emerging pathogen. This review seeks to synthesize our current but incomplete understanding of the biology of E. albertii.

IDENTIFICATION OF E. ALBERTII—CLASSIFICATION AND RECLASSIFICATION

Escherichia albertii is a Gram-negative bacterium that was first isolated in 1991. This original report identified a diarrheic isolate from the feces of a 9-month-old febrile Bangladeshi girl, exhibiting symptoms of watery diarrhea, vomiting, mild dehydration, and abdominal distension (1). The isolate was originally designated to be Hafnia alvei strain 19982, on the basis of a panel of biochemical reactions with the API 20E strip, as well as by Edwards and Ewing’s criteria for the classification of Enterobacteriaceae (1).

Subsequent histological and electron microscopic examination of the intestines of rabbits infected with this isolate revealed that the bacterium was capable of forming attaching and effacing (A/E) lesions, pathognomonic organelles characteristic of A/E pathogens such as enteropathogenic E. coli (EPEC) and enterohemorrhagic E. coli (EHEC) (1). Consistent with this phenotype, the isolate contained the eae gene (E. coli attaching and effacing) that encodes intimin (1–3). Intimin, located on the outer bacterial membrane, binds its receptor Tir on the host cell membrane (2–4). Tir-intimin interactions lead to the polymerization of actin and the formation of pedestal-like structures beneath adherent bacteria (2, 4–11). Notably, the eae positivity and the capacity of H. alvei 19982 to form A/E lesions had not previously been reported among H. alvei isolates (1). Furthermore, the 19982 isolate did not express other virulence determinants common to E. coli pathovars such as heat-stable enterotoxin (ST), heat-labile enterotoxin (LT), Shiga toxins, or the ability to invade HeLa cells (1, 12). Subsequently, six additional isolates with phenotypes similar to those of the original H. alvei isolate were reported (13).

As early as 1995, the initial taxonomic classification of strain 19982 and the other diarrheic isolates as H. alvei came into question (14). Ridell et al. noted that these eae-positive isolates exhibited profound genotypic and phenotypic differences from their eae-negative H. alvei counterparts (14). For example, all the eae-positive isolates were able to assimilate 3-hydroxybenzoate but not 2-ketogluconate (Table 1). Moreover, the isolates could not be reproducibly biotyped as H. alvei by different diagnostic methods (14). In addition, profiling by random amplification of polymorphic DNA (RAPD) PCR revealed that these strains shared lower identity (92%) at the 5′ end of the 16S rRNA gene with the eae-negative H. alvei isolates but shared much greater identity (98%) with the 16S rRNA gene of enteropathogenic E. coli strain E2348/69, a prototypical A/E pathogen (14). Similarly, 11 additional eae-negative H. alvei clinical isolates from Canada, obtained from diarrheal stool samples of infected children, exhibited a biochemical and phenotypic signature different than that of original eae-positive Bangladeshi isolate 19982 (15). These strains had a different outer membrane protein (OMP) composition and had different patterns of DNA fragmentation by macrorestriction digest than the reference strain 19982; moreover, they did not form A/E lesions (15). These data suggested that the ICDDR-B isolates did not fit the taxonomic criteria for membership in the H. alvei taxon and that reclassification was warranted.

TABLE 1.

Biochemical/phenotypic traits of E. albertii, E. coli, and H. alvei^a

Biochemical test	% Positive or negative result for:
	E. albertii	E. coli	H. alvei
Methyl red positive	100 (12, 23)	100	40
Nitrate reduction	100 (23, 51)	100	100
Glucose utilization	100 (12, 23, 25, 38, 51)	100	100
d-Mannose utilization	100 (23, 51)	98	100
Galactose utilization	100 (23, 51)	100 (23)	100 (125)
Mannitol utilization	100 (12, 23, 25, 38, 51)	98	99
3-Hydroxybenzoate assimilation	100 (14, 16, 38)	0 (123)	0 (14)
l-Arabinose utilization	>98 (12, 23, 25, 38, 51)	99	95
Lysine decarboxylation	>95 (12, 23, 25, 38, 51)	90	100
Ornithine decarboxylation	>95 (12, 23, 25, 38, 51)	65	98
Trehalose	>94 (12, 23, 25, 51)	98	99
ONPG breakdown	>94 (12, 23, 25, 51)	95	90
d-Galacturonate	>92 (23)	98 (23)	>90 (126)
Glycerol utilization	>82 (23, 51)	75	95
Acetate utilization	>80 (12, 23, 25, 51)	90	15
Raffinose utilization	<16 (23, 51)	50	2
Sucrose utilization	<13 (23, 25, 38, 51)	50	10
Methylumbelliferyl glucuronide (MUG)	<11 (12, 23, 25, 51)	>93 (23)	0 (12)
Xylose utilization	<9 (23, 25, 38, 51)	95	98
Rhamnose utilization	<5 (12, 23, 25, 38, 51)	80	97
Citrate utilization	<1 (23, 25, 38, 51)	1	10
Lactose utilization	<1 (12, 23, 25, 51)	95	5
Oxidase positive	0 (12, 16, 38, 51)	0	0
Voges-Proskauer positive	0 (12, 16, 23, 25, 51)	0	85
H2S production	0 (23, 25, 38, 51)	1	0
Cellobiose utilization	0 (16, 23, 25, 51)	2	15
Adonitol utilization	0 (23, 25, 51)	5	0
Myoinositol utilization	0 (23, 25, 51)	1	0
Growth on KCN broth	0 (12, 16, 23, 51)	3	95
Motility	0 (12, 16, 23, 25, 38, 51)	95	85
2-Ketogluconate assimilation	0 (14, 16)	0 (124)	100 (14)

^aThe sum total of all E. albertii isolates that were screened for a specific trait across all the studies was used to determine the percentage that tested positive or negative for the specific biochemical/phenotypic trait in question. For E. coli andH. alvei, the presented results are from the study conducted by Farmer et al. (122), unless otherwise stated, in which case the result is from the study or studies indicated in parentheses.

The idea that these isolates were misclassified was reinforced in 1999 when it was found that the original ICDDR-B strains differed phenotypically from H. alvei and resembled E. coli (12). Each of the original strains was methyl red positive, Voges-Proskauer test negative, utilized acetate, failed to grow on KCN medium (HiMedia Laboratories), and was resistant to a Hafnia-specific phage, all characteristic of E. coli (12) (Table 1). However, the classification was not a perfect match; unlike typical E. coli strains, the ICDDR-B isolates were indole negative, nonmotile, and did not ferment lactose (Table 1), with the last two traits also being reminiscent of most enteroinvasive E. coli (EIEC) isolates. On the basis of these results, Janda et al. proposed the reclassification of the eae-positive ICDDR-B isolates as unusual biotypes belonging to the genus Escherichia (12). In 2003, DNA-DNA hybridization studies demonstrated that the genomes of these eae-positive ICDDR-B strains were more closely related to E. coli (relatedness value of 55 to 64%) than H. alvei (relatedness value of 9 to 17%) (16). Sequencing of the complete 16S rRNA gene indicated a similarity value of 98.3% with E. coli but only 93.5% with H. alvei (16). Collectively, these data led to the formal reclassification of these isolates into the novel taxon Escherichia albertii, in honor of M. John Albert, who first isolated the bacterium (16).

E. ALBERTII AS AN INFECTIOUS ENTERIC PATHOGEN: MISTAKEN IDENTITY, CHARACTERISTIC OF DISEASE, AND HOST RANGE

Although the bacterium was first isolated from infants, E. albertii infections have since been reported in children, as well as in middle-aged and elderly adults (17–19). Humans infected with E. albertii typically exhibit symptoms associated with gastroenteritis, including watery diarrhea, dehydration, abdominal distension, vomiting, and, in some instances, fever (1). A small subset of E. albertii isolates harbor the stx_2a allele and can induce bloody diarrhea (18). Moreover, there has been one reported case of a patient developing bacteremia after becoming infected with E. albertii (17). Beyond these sporadic observations, the epidemiology, transmissibility, prevalence, and incidence of E. albertii infections remain unexplored (20, 21).

Frequent mistyping of E. albertii has contributed to our limited understanding of this organism and has also limited the application of Koch’s postulates to link the bacterium to disease (20, 21). Misidentification of E. albertii stems from the genotypic and phenotypic resemblance of the bacterium to enterovirulent species of Escherichia, especially atypical strains of enteropathogenic E. coli (20–24). This difficulty in distinguishing E. albertii from other E. coli strains adds to the urgency to identify unique traits of E. albertii in order to reliably distinguish the bacterium from other members of Enterobacteriaceae (20, 24).

Recent studies have demonstrated a strong correlation between E. albertii and disease outbreaks (19, 24–27), some of which had previously been incorrectly attributed to EPEC and EHEC (19, 22, 25). For instance, in 2012, a retrospective study reexamined 179 eae-positive bacterial strains that had been typed as EPEC or EHEC using available commercial diagnostic kits. Many of these isolates were obtained from disease outbreaks in humans and other animals, including birds (25). Multilocus sequence typing (MLST) demonstrated that 26 of these isolates (14.5%) were actually E. albertii (25). More importantly, 14 of the E. albertii isolates were obtained from humans, and 13 of these were from symptomatic patients.

Similarly, EPEC was wrongly diagnosed as the etiologic agent in multiple diarrheal outbreaks (19, 22, 26–28). In 2003, a diarrheal outbreak in the city of Fukuoka, Japan, was attributed to boxed lunches (19). Of the 31 individuals that consumed these lunches, 20 developed diarrhea, abdominal pain, and high fever. The bacterial strain that was isolated from this outbreak was typed as an atypical EPEC strain that was a lactose nonfermenter and had the locus of enterocyte effacement (LEE) eae gene (19). A rigorous reexamination of this isolate by allele-specific PCR and a library of biochemical assays revealed that the isolate was in fact E. albertii (19). Likewise, a second outbreak occurred in 2011 in Kumamoto, Japan. Ninety-four individuals dined at a Japanese restaurant, and 48 of them became ill, with symptoms including diarrhea, abdominal pain, fever, and nausea. Originally, this outbreak was attributed to atypical EPEC OUT:HNM, a lactose nonfermenter that possessed eae (28). However, further genetic and biochemical analysis of the archived isolates revealed that the bacterium more closely resembled E. albertii phylogenetically and phenotypically (26).

E. albertii has also often been misclassified as EHEC, because it is not uncommon to discover isolates harboring the stx genes that encode Shiga toxin (18, 25). EHEC, a subtype of Shiga toxin-producing E. coli (STEC), is responsible for the deadly clinical illness in humans that is characterized by bloody diarrhea, thrombocytopenia, and hemolytic uremic syndrome (HUS) (29). For a long time, the presence of stx and eae genes was used as a diagnostic hallmark of A/E EHEC isolates (30–32). Thus, the isolation of E. albertii strains possessing these two genetic markers often resulted in their incorrect classification as EHEC (25). Recently, an E. albertii strain harboring the stx_2a allele was isolated from a Norwegian patient who suffered from bloody diarrhea (18). The stx_2a allele encodes the Shiga toxin 2a isoform, one of the most potent and toxinogenic forms of the toxin (33–36). Bacterial strains harboring this allele are considered to be highly virulent because they can cause bloody diarrhea, HUS, and thrombocytopenia (33–35). Some E. albertii strains harbor the stx_2f allele that is associated with a milder form of disease (see “Shiga toxin,” below) (18, 37). The presence of stx genes in E. albertii strains calls for a systematic reexamination of all stx-expressing E. coli isolates to rigorously validate their classification to ensure that they are not misidentified E. albertii strains.

Besides localized gastrointestinal infections, some reports suggest that E. albertii can cause bacteremia (17). The bacterium was recently isolated from the blood of a 76-year-old woman who exhibited multiple comorbidities, including high fever, pelvic fracture, hypothyroidism, and hypertension (17). Extensive biochemical and genetic testing, including 16S rRNA analysis, confirmed the isolate to be E. albertii, although this unusual isolate did not conform to the genotypic and phenotypic traits characteristic of quintessential E. albertii strains. For instance, the isolate was a slow lactose fermenter (17). Moreover, and unlike typical E. albertii infections, the patient had no immediate history of gastrointestinal discomfort and/or diarrhea and did not have any recent contact with children, birds, or adults, all previously demonstrated reservoirs of E. albertii. This suggests that the bacterium has alternative portals of entry. The patient recovered from the infection without any extenuating complications (17).

In addition to humans, E. albertii has been implicated as an etiologic agent of disease in multiple species of birds (38). In 1994, dead finches were reported at numerous feeding stations in Inverness, Scotland (39, 40). Postmortem examination of the carcasses revealed the causative agent to be E. coli belonging to the serotype O86:K61 (39, 40). This serotype was previously associated with diarrhea in calves, pigs, and horses and with cellulitis in broiler chickens (41). However, a subsequent study that reexamined those isolates by MLST and by a larger panel of phenotypic and biochemical tests concluded that they were, indeed, E. albertii (38, 42). Although all 34 of the isolates harbored the LEE-carried eae gene, A/E lesions were not observed in the deceased birds (39–41). In 2004, more than 100 dead redpoll finches were reported in Fairbanks, Alaska (38). Bacteriologic and biochemical examination of tissue biopsy specimens isolated E. albertii (38), but again no A/E lesions were observed despite histopathology consistent with enteritis (38). Interestingly, E. albertii was also isolated from subclinically colonized birds, an observation that is reminiscent of the epizoology of Salmonella enterica serovar Typhimurium in birds (38). These studies collectively suggest that E. albertii is virulent in certain bird species and accounts for some of the bird epidemics with unresolved etiology (42). Moreover, the pathology associated with E. albertii infections may extend beyond the classical A/E histopathology that is characteristic of members of this morphotype. Unfortunately, beyond these superficial observations, systematic epidemiological or controlled clinical studies on E. albertii infections are severely lacking.

Birds represent a primary reservoir for E. albertii. In Australia, two major surveys of enteric bacteria in vertebrates have shed light on the prevalence and distribution of E. albertii (42, 43). The first survey, which spanned a period of ∼10 years, from 1994 to 2004, did not detect E. albertii in fish, frogs, snakes, lizards, and crocodile. However, E. albertii was detected in birds at various frequencies, with chickens (33%) and magpies (18%) representing the major reservoirs (43). The second survey, conducted between 2010 and 2011 and restricted to avian species, also identified chickens (6.7%) and magpies (14.3%) as important reservoirs (42). In fact, multiple independent investigations conducted in the United States, Japan, and China recovered E. albertii from processed chicken, suggesting that chicken is one of the key vehicles for the transmissibility of E. albertii to humans (44–46).

Besides chicken, E. albertii has been isolated from other processed meat-based food products, including duck, mutton, and pork (46). In Norway, a presumed strain of H. alvei that possessed the eae gene was isolated from minced meat (47). Subsequent studies suggested that this presumptive H. alvei strain belongs to the E. albertii lineage (48). Thus, there are considerable data to implicate E. albertii as a food-borne pathogen. A few epidemiological and molecular investigations have also implicated contaminated water as a vehicle for the transmission of E. albertii. In 2005, E. albertii isolates belonging to the serotypes OUT:H- and O168:H- were cultured from a diarrheal outbreak among campers who complained of abdominal pain, vomiting, and fever (19, 22). Samples collected from the campsite water sources also tested positive for the same isolates, suggesting that this outbreak occurred as a result of contaminated water (22). E. albertii has also been isolated from water distribution systems in other geographic areas of the world, including Quebec City, Canada, and Budapest, Hungary (49). Thus, E. albertii is potentially both a food- and waterborne enteric pathogen.

E. albertii outbreaks continue to be underreported and misclassified, and studies on its epidemiology and virulence are limited. As a consequence, an accurate assessment of the pathogen’s true impact on human health is currently lacking. Comprehensive characterization of E. albertii virulence determinants may provide a basis to differentiate E. albertii from other A/E pathogens and also aid in the development of prophylactic and therapeutic treatments against this pathogen.

Inclusive and exclusive traits of E. albertii.

Recent molecular work has made important inroads into identifying genetic and biochemical markers unique to E. albertii (20, 21, 24, 50). All E. albertii isolates share several key biochemical traits/markers (Table 1). All isolates identified to date are positive for methyl red, nitrate reduction, 3-hydroxybenzoate assimilation, utilization of glucose, d-mannose, galactose, and mannitol (Table 1). E. albertii isolates are negative for oxidase, Voges-Proskauer test, H₂S production, and utilization of cellobiose, adonitol, myoinositol, and 2-ketogluconate; moreover, E. albertii strains are nonmotile and do not grow on KCN medium (Table 1) (12, 16, 23, 25, 51). In addition, the vast majority of the isolates are positive for the breakdown of arabinose (>98%), lysine (>95%), ornithine (>95%), o-nitrophenyl-β-d-galactopyranoside (ONPG) (>94%), trehalose (>94%), d-galactouronate (>92%), glycerol (>82%), and acetate (>80%) and negative for import and/or breakdown of lactose (>99%), citrate (>99%), rhamnose (>95%), xylose (>91%), methylumbelliferyl glucuronide (MUG) (>89%), sucrose (>87%), and raffinose (>84%) (Table 1) (12, 16, 23, 25, 51). Interestingly, whereas the original isolates from Bangladesh, 19982, 9194, 10457, 10790, and 12502, could not synthesize indole or utilize sorbitol (16, 51), subsequent studies have shown that there is considerable heterogeneity among E. albertii biotypes for these traits (23). An additional and important commonality among E. albertii isolates is the ability to form A/E lesions (1, 15, 52).

Besides phenotyping studies, complementary whole-genome sequencing studies and MLST of dozens of E. albertii isolates have identified genetic markers that could serve as diagnostic signatures for the bacterium (24, 53). These studies have also illuminated the potential repertoire of its candidate virulence factors and provided computational insight into its ecology, evolution, and molecular epidemiology. In one study, 29 E. albertii strains were sequenced (24). The complete genome sequences of three strains (NIAH_Bird_3, EC06-170, and CB9786) and the draft sequences of 26 additional strains, isolated from different sources, were determined. These genomes, along with the genomes of 5 previously sequenced E. albertii strains, were used for intraspecies and intragenus comparisons to identify lineage-specific as well as broadly conserved genetic markers. Their analyses revealed that the median E. albertii genome size of ∼4,777 kb is smaller than that of E. coli (∼5,132 kb) (24). Furthermore, the average nucleotide identity (ANI), a measure of the phylogenetic relatedness, exceeded 98% across E. albertii isolates, solidifying its classification as a clade. In contrast, the ANI between E. albertii and E. coli was 89.2 to 90.1%, consistent with the rectified classification of the former as a novel taxon within the genus Escherichia (24). Genomic studies also provided genetic evidence to support previously observed biochemical characteristics. For instance, the sequenced E. albertii strains lacked the following genes/operons involved in important metabolic pathways: lacY (lactose metabolism), xylBAFGHR locus (xylose metabolism), melRAB (raffinose metabolism), uidCBAR (β-glucuronidase synthesis), and rha locus (rhamnose metabolism) (24). A curious phenotype exhibited by most E. albertii isolates is that, despite having a Lac⁻ phenotype, they metabolize the β-galactosidase substrate ONPG, suggesting that E. albertii isolates encode a functional lacZ allele (16, 24, 51) (Table 1). Consistent with this prediction, all of the sequenced isolates discussed above possess an intact lacZ allele (24).

Genomic studies have also been informative about morphogenetic pathways, such as those involved in the synthesis and assembly of flagella and the type 3 secretion system (T3SS). Even though the bacterium is naturally nonmotile (16, 23, 24, 51) (Table 1), gene clusters involved in flagellar morphogenesis are well conserved and expressed in different E. albertii isolates (24). In the aforementioned study by Ooka et al., 26 of the 34 strains tested (∼76.5%) contained intact operons and islets involved in flagellar biogenesis (24). Furthermore, many of the flagellar regulatory and structural genes were expressed and displayed remarkable degrees of sequence conservation (>90% identity) among the different E. albertii isolates (24). In contrast, the sequence of fliC, encoding flagellin monomers that form the flagellar filament, had diverged substantially (65 to 85% identity) among the different E. albertii isolates (24). Consistent with the sequence divergence, flagella were undetectable by microscopy on the bacterial surface, despite the expression of many of the genes involved in the flagellar regulatory cascade (24). It has been suggested that the loss of flagella stems from immunological counterselection in the host (24), but the retention and expression of the other flagellum-related genes may be driven by their integration into alternative regulatory circuits. Alternatively, the core architectural secretory apparatus of the flagella may have evolved to secrete a different set of effectors than the ones that bestow flagellum-dependent motility to related Enterobacteriaceae. In contrast to the flagellar structures, which do not appear to be made in E. albertii, a surface-associated T3SS that is essential for A/E lesion formation is detectable on many E. albertii isolates (see “T3SS,” below) (1, 15, 52).

The availability of these recently mined phenotypic and genomic data sets has identified biomarkers for the reliable and reproducible typing of E. albertii (24). E. albertii strains can be distinguished by both phenotypes and genetics from Hafnia alvei as described above (see “Identification of E. albertii—classification and reclassification,” above, and Table 1). Distinguishing E. albertii from EPEC or EHEC is more difficult and requires a combination of genetic and/or biochemical markers for accurate discrimination. In contrast to EPEC and EHEC, E. albertii strains are naturally nonmotile (100%) and do not hydrolyze lactose (>99%) (12, 16, 23–25, 51). Moreover, the vast majority of E. albertii isolates are unable to metabolize xylose (>91%) and MUG (>89%). Another distinguishing feature of E. albertii, revealed by sequencing of the eae gene, is that isolates typically harbor intimin subtypes that are rare or have yet to be described in EPEC/EHEC strains (25). Recent whole-genome sequencing studies have shed light on genetic regions unique to E. albertii. In one study, 118 genetic loci were identified that ranged in size from 100 bp to ∼4 kb and were specific to E. albertii, with no discernible orthologs in other species of Escherichia (24). Using this metagenomic information, nested PCR assays were developed to amplify an E. albertii-specific locus that harbors genes encoding a fumarate reductase and a putative cytochrome c-type protein. Predictably, amplicons were generated only when E. albertii, but not E. coli, strains provided the template (24). These biochemical and genetic signatures are invaluable in the correct typing of E. albertii strains, especially clinical isolates, for accurately distinguishing them from EPEC and EHEC.

E. albertii possesses an array of candidate virulence factors, some of which are shared with other pathovars of E. coli (1, 13, 20, 21, 24, 25). However, beyond computational comparisons and limited expression-based studies, the virulence landscape of E. albertii remains virtually unmapped, despite multiple reports linking the bacterium to outbreaks worldwide (19, 26, 27). Thus, there is a need to identify its arsenal of virulence factors, characterize their environmental and regulatory controls, perform structure-function analysis, and investigate their roles in the pathobiology of E. albertii. Below we highlight the virulence factors whose expression and/or activity have been experimentally verified in E. albertii (23, 24, 50, 54).

(i) T3SS. A/E pathogens form a diagnostic lesion, termed a pedestal, that protrudes from the surface of an infected intestinal epithelial cell and is crowned by the infecting bacterium (55–59). Pedestal formation depends on a functional T3SS, the genes of which are located in an ∼35- to 37-kb pathogenicity island (PAI) called LEE (57–67) (Fig. 1). Metagenomic analysis reveals that the LEE is conserved in all E. albertii strains (1, 16, 24, 52). The T3SS is expressed in many E. albertii strains, and these strains can form A/E lesions both in vivo and in vitro (1, 15, 50, 52). However, in a subset of E. albertii isolates, A/E lesion formation has not been observed despite the strict presence of an intact LEE (38, 41).

FIG 1.

Genetic architecture of the LEE and the degree of conservation of the LEE-encoded proteins between EPEC, EHEC, and E. albertii. (A) The LEE is a 35- to 37-kb pathogenicity island that is composed of five polycistronic operons (LEE1-5), two bicistronic operons (grlRA and LEE6), and numerous monocistronic transcription units. The different LEE-encoded proteins have been color coded to depict their specific role in type 3 secretion (T3S). (B) The primary structure of every predicted LEE-encoded protein from the E. albertii clinical isolate CB786 was aligned to its corresponding homolog from the EPEC strain E2348/69 and the EHEC strain TW14359 to determine the extent of protein conservation across the three strains. The percent identity between the specific LEE-encoded protein of E. albertii and its corresponding homolog in EPEC or EHEC is depicted by empty or filled circles, respectively.

The LEE is a horizontally acquired genomic island that is remarkably colinear between EPEC, EHEC, and E. albertii (63, 68–70) (Fig. 1). EPEC and EHEC appear to have acquired the LEE on multiple occasions during the course of their evolutionary history (71). This is evident from the molecular observation that in the EPEC1/EHEC1 cluster the LEE is inserted into the selC tRNA locus, whereas in the EPEC2/EHEC2 cluster the integration site of the LEE is the pheU tRNA locus (71–74). Besides selC and pheU, A/E E. coli isolates harboring the LEE at a third locus, pheV, have been isolated (75). In striking contrast, the LEE appears to be strictly present at the pheU locus in all E. albertii strains that have been sequenced to date (24, 25), suggesting that the LEE has, thus far, been acquired on a single occasion by the bacterium. Moreover, because the E. albertii clade radiated before the divergence of the E. coli-Shigella group (54), the bacterium presumably acquired the LEE independently of the acquisition of this PAI by EPEC and EHEC.

Among A/E pathogens, the LEE has been best characterized in EPEC and EHEC, where its regulation has been systematically dissected (57–59). The LEE is genetically arranged into five multicistronic operons (LEE1-5), a bicistronic operon (grlRA), and multiple monocistronic transcription units (57–60) (Fig. 1). In response to permissive conditions, such as those encountered in the intestinal environment, genes of the LEE are induced and the structural proteins of the T3SS assemble extracytoplasmically to form a secretory channel that links the bacterial cytosol to the extracellular milieu (57–59, 76, 77). Upon contact with a eukaryotic cell, the T3SS pierces the cell membrane to form a passageway linking the bacterial and host cytoplasms (78–84). Bacterial effectors are translocated via this tube directly into the host intestinal cells, where they subvert host regulatory and signal transduction pathways that lead to the destruction of the local microvilli due to depolymerization of actin (57–59). These events are followed by the subsequent recruitment and repolymerization of actin monomers beneath the adherent bacterium to form a filamentous membrane-enclosed protrusion from the infected cell, akin to a pedestal, upon which the bacterium is perched (66, 85, 86).

Despite the preservation of the genetic organization of the LEE, the proteins encoded in the LEE exhibit varying degrees of conservation between EPEC, EHEC, and E. albertii (Fig. 1). For instance, the major transcription factors Ler, GrlR, and GrlA exhibit >90% identity across the three A/E pathogens (Fig. 1). Similarly, the primary structures of the core architectural components of the T3SS, which are mainly distributed between the LEE1, LEE2, and LEE3 operons, exhibit a very high degree of conservation (69) (Fig. 1). In striking contrast, many of the effectors, such as Tir, EspZ, EspH, EspG, Map, and EspF, as well as the translocators EspA, EspB, and EspD, have substantially diverged in E. albertii from their structure in EPEC/EHEC (69) (Fig. 1). However, beyond these observations, studies interrogating the extrinsic and intrinsic regulation of the LEE, structure-function analysis of the LEE-encoded gene products, and mechanism(s) of action have not been conducted in E. albertii (20). In fact, not a single regulator of the LEE has been identified in E. albertii. This contrasts with the observed regulation of the LEE in EPEC and EHEC, where collectively more than 40 regulatory factors have been identified that fine-tune LEE gene expression by operating at the transcriptional, posttranscriptional, posttranslational, and epigenetic levels (57–59, 87, 88).

Some isolates of E. albertii possess a second T3SS, E. coli type 3 secretion system 2 (ETT2), that is inserted at the tRNA-glyU locus (24). ETT2, which bears genetic resemblance to the T3SS encoded in the Salmonella pathogenicity island-1 (SPI-1), is present in the E. coli/Shigella lineage but has undergone extensive genetic degradation and is not expressed. Remarkably, at least 16 of the 34 sequenced E. albertii strains possess an intact ETT2, with multiple genes from the PAI being expressed (24). Moreover, 6 of the remaining strains have a single-nucleotide mutation in the eivJ gene and could yield a functional EivJ protein by programmed ribosomal frameshifting or transcriptional realignment, resulting in the expression of the second T3SS (24). However, similar to the paucity of information on the LEE, the regulation and role of ETT2, as well as potential cross talk with the LEE, remain understudied.

The repertoire of effectors translocated by the T3SS of E. albertii remains uncharacterized. Approximately 15 loci appear to encode putative T3S effectors, and many loci encode multiple effectors (24). The core LEE-encoded effectors, Tir, Map, EspH, EspF, EspG, and EspZ, conserved among other A/E pathogens, were also present in 100% of the sequenced E. albertii isolates (24, 89) (Fig. 2). The effector protein Ibe, encoded by a variable segment of the LEE that is more frequently observed in EHEC (present in the isolate TW14359) than EPEC (absent in E2348/69) isolates (90, 91), was present in all but one E. albertii isolate (Fig. 2). In contrast to the LEE-encoded effectors, non-LEE-encoded effectors were present at various degrees in E. albertii isolates. Some of them, such as EspX (100%), EspM (100%), NleG (100%), NleF (97%), NleH (97%), EspJ (91%), EspO (91%), and EspY (88%), were present in >85% of isolates, while others, including NleE (47%), NleD (35%), TccP (26%), EspL (26%), EspW (24%), EspN (6%), Cif (6%), and OspG (6%), were present in <50% of sequenced isolates (24) (Fig. 2).

FIG 2.

Prevalence of T3SS-dependent effectors among E. albertii strains. The T3 pansecretome (full set of the T3SS-dependent effector molecules that are predicted to be encoded across the genomes of all sequenced isolates) of E. albertii was predicted and the percent prevalence in the E. albertii clade was determined. Filled circles indicate that the putative T3SS-dependent effector of E. albertii is also present in both the EPEC strain E2348/69 and the EHEC strain TW14359, whereas half-filled circles indicate that the E. albertii effector is present only in EPEC E2348/69 (circles with the left half filled) or in the EHEC strain TW14359 (circles with the right half filled).

Despite the diversity of effector repertoire in E. albertii, the only effector whose expression and/or functionality has been experimentally validated is Tir. Interestingly, Tir and its cognate ligand, intimin, essential for the formation of A/E lesions, are also required for the invasion of epithelial cells, although the invasive phenotype has only been documented in a single E. albertii isolate (50, 92, 93). This intriguing invasive phenotype has previously been reported for EPEC, where diverse T3S effectors, including Tir, Map, or EspT, independently promote invasion of host cells (94–96). However, whether the same mechanism is operational to promote the invasion of E. albertii remains to be determined (50).

(ii) Shiga toxin. A subset of E. albertii strains that express Shiga toxin have also been identified. The stx genes are located on genetic regions that are reminiscent of bacteriophage-like elements (18). Shiga toxins, first identified in Shigella dysenteriae at the beginning of the 20th century, belong to the AB family of toxins that consists of a single enzymatic A subunit and 5 structural B subunits (97). The toxin inactivates the eukaryotic ribosome, resulting in the eventual cessation of protein synthesis (97, 98). stx genes have been identified in a range of pathogens, including different pathovars of E. coli and, to a lesser extent, in Citrobacter freundii, Enterobacter cloacae, and Shigella flexneri (99–101). Shiga toxins can be classified into two types in E. coli: Stx1 and Stx2. Stx1 is the homologue of the Shigella Stx and was the first of the two types to be identified. Stx2 was isolated several years later and is mechanistically similar to but antigenically distinct from Stx1 (97). Since the initial discovery of the Stx1 and Stx2 prototypes, multiple variants of the two toxins have been identified.

The first report documenting the expression of Shiga toxins in E. albertii was published in 2012 (25). In this report, a set of bacterial strains that had previously been classified as EPEC or EHEC were reexamined, and it was determined that 15% of the isolates were actually E. albertii, with two of them carrying the stx_2f allele. One strain, HIPH08472, was isolated from a symptomatic human, whereas the other isolate, E2675, was obtained from a healthy bird (25). Other studies identified more clinical isolates expressing Stx2f (102, 103). In one such study in 2015, as many as 10% of E. albertii isolates harboring the stx_2f allele were identified (18). To date, the majority of E. albertii isolates that express Shiga toxin contain the stx_2f subtype (18). The stx_2a-expressing isolate was cultured from a 48-year-old patient with bloody diarrhea. This isolate is the only reported observation of Stx2a in E. albertii (18).

Patients infected with stx_2f-expressing E. albertii typically exhibit milder clinical symptoms, such as watery diarrhea, abdominal pain, headaches, and mild fever, while infections by stx_2a-expressing isolates are more severe and include bloody diarrhea (104–107). Clinical and epidemiologic studies investigating outbreaks caused by stx_2a-expressing E. coli reveal that in ∼5 to 15% of individuals, especially children and the elderly, the episodes of bloody diarrhea are a prodrome for the life-threatening complication HUS (33, 108, 109). HUS is characterized by premature destruction of red blood cells (hemolytic anemia), decreased platelet count (thrombocytopenia), and renal failure (109). Although stx_2a-expressing E. albertii strains have yet to be directly implicated in HUS, it would not be unreasonable to suspect that a subset of the reported cases of HUS that are currently attributed to stx_2a-expressing E. coli are indeed caused by mischaracterized E. albertii isolates.

Studies on the purified Stx2a and Stx2f subtypes from E. coli reveal that although they are mechanistically similar, there are key differences in their biochemical and toxicological properties, which may contribute to the differences in the accompanying pathophysiology (110). The Stx2f A subunit exhibits 71% and the B subunit 82% identity to the respective subunits of Stx2a, with much of the active site being identical, accounting for the similarities in catalytic activity. However, Stx2f is less toxic than Stx2a, with the former requiring an almost 3- to 5-fold higher dose to induce cytotoxicity in Vero cells (110). Thus, many factors may contribute to the differential toxicity and pathology associated with the two variants, including differences in receptor preference and/or affinity, pattern of toxin distribution on the surface of infected cells, and kinetics of internalization and/or cytoplasmic release (110). However, other than the few epidemiologic reports and studies genotyping the toxin subtypes and variants in E. albertii, there has not yet been a study on the regulatory controls of the stx genes or the biochemical and functional characteristics of the toxin in the bacterium (16, 51, 54).

(iii) CDT. The cytolethal distending toxin (CDT) is a genotoxin that is synthesized by several gastrointestinal bacteria, including strains of EPEC, EHEC, and E. albertii (16, 27, 54, 102, 111–113). The CDT holotoxin also belongs to the AB family, although it is composed of one subunit each of CdtA, CdtB, and CdtC (114–116), with CdtA and CdtC constituting the B subunit that promotes receptor-mediated endocytosis of CdtB (117, 118). CdtB functions as the enzymatic A subunit that fragments DNA by the introduction of double-stranded breaks that lead to irreversible cell cycle arrest and apoptosis (54, 117, 118). Previously, E. coli was recognized as the sole outlier that expressed different isoforms of CdtB, with at least five recognized variants, designated CdtB I to V (119). Recent molecular genetic analysis has revealed isolates of E. albertii that express diverse cdtB alleles, including cdtB-I, cdtB-II, cdtB-III, and cdtB-V, with some harboring multiple alleles (54, 102). More recently, a reexamination of 20 cdtB-II-positive diarrheic E. coli strains, using an array of genetic and biochemical tests, revealed that all isolates were E. albertii and had been misclassified as E. coli. This finding suggests that the presence of the cdtB-II subtype is a more reliable biomarker for E. albertii and not E. coli (113). The functional role of CDT in the molecular pathogenesis of E. albertii has been limited to in vitro investigations. HeLa cells exposed to protein extracts of E. albertii exhibited cytoplasmic distension and nuclear fragmentation, symptoms of genotoxicity that are typically associated with CDT (54). However, whether CDT contributes to the virulence of E. albertii in vivo remains to be addressed. Moreover, and as is the case with each of the other virulence factors of E. albertii, studies delving deeper into the regulatory controls on the cdt locus and the molecular mechanism of action of CDT have not been conducted.

In summary, very few studies have investigated the molecular mechanisms of virulence of E. albertii; consequently, our understanding of the bacterium’s pathobiology is still rudimentary. The paucity of knowledge in this area presents a particular challenge to the medical and scientific community when dealing with clinical isolates of E. albertii.

PERSPECTIVE

The newest member of the A/E family of bacterial pathogens, E. albertii, remains grossly undercharacterized, despite strong evidence implicating the bacterium in multiple outbreaks globally. E. albertii has been, and continues to be, mischaracterized as EPEC, EHEC, or other pathogenic bacteria. Although the genomes of multiple isolates have been sequenced, annotated, and compared, genome-wide expression-based transcriptomic, proteomic, metabolomic, and phenotypic studies interrogating its virulence arsenal are lacking. To date, only a few studies have superficially explored the regulation and function of individual virulence determinants of E. albertii (50, 120). The pathogenic potential of E. albertii is exacerbated by the recent identification of multidrug-resistant strains. E. albertii strains that are resistant to at least 12 structurally and functionally diverse antibiotics have recently been isolated, with most of the resistance markers being present on a single plasmid (121). The natural emergence of multidrug resistance, via a single evolutionary step, in a bacterium with a cryptic and ill-defined virulome heightens the urgency with which research must be undertaken to understand the biology of this formidable adversary.