Escherichia albertii as a Potential Enteropathogen in the Light of Epidemiological and Genomic Studies

Escherichia albertii is a new enteropathogen of humans and animals. The aim of the study was to assess the prevalence and pathogenicity of E. albertii strains isolated in northeastern Poland using epidemiological and genomic studies. In 2015–2018, a total of 1154 fecal samples from children and adults, 497 bird droppings, 212 food samples, 92 water samples, and 500 lactose-negative E. coli strains were tested. A total of 42 E. albertii strains were isolated. The PCR method was suitable for their rapid identification. In total, 33.3% of E. albertii isolates were resistant to one antibiotic, and 16.7% to two. Isolates were sensitive to cefepime, imipenem, levofloxacin, gentamicin, trimethoprim/sulfamethoxazole, and did not produce ESBL β-lactamases. High genetic variability of E. albertii has been demonstrated. In the PFGE method, 90.5% of the strains had distinct pulsotypes. In MLST typing, 85.7% of strains were assigned distinct sequence types (STs), of which 64% were novel ST types. Cytolethal distending toxin (CDT) and Paa toxin genes were found in 100% of E. albertii isolates. Genes encoding toxins, IbeA, CdtB type 2, Tsh and Shiga (Stx2f), were found in 26.2%, 9.7%, 1.7%, and 0.4% of E. albertii isolates, respectively. The chromosome size of the tested strains ranged from 4,573,338 to 5,141,010 bp (average 4,784,003 bp), and at least one plasmid was present in all strains. The study contributes to a more accurate assessment of the genetic diversity of E. albertii and the potential threat it poses to public health.


Introduction
Escherichia albertii (E. albertii) is the second-after Escherichia coli (E. coli)-pathogenic species in the Escherichia genus, and is regarded as an emerging human and animal enteropathogen [1]. E. albertii, similarly to enteropathogenic E. coli (EPEC) and enterohemorrhagic E. coli (EHEC) strains, possesses a specific intestinal cell invasion mechanism known as A/E lesions (attaching and effacing lesions), characterized by intimate attachment of bacteria to the residual apical enterocyte membrane and localized destruction of brush border microvilli [2,3]. To date, only a limited number of publications are available about this bacterium due to late recognition of E. albertii as a new species (in 2003) and lack of standard identification methods [4]. Although this bacterium was isolated in the early 1990s and recognized as the cause of diarrhea in five Bangladeshi children, originally it was misidentified as Hafnia alvei [5]. However, further analyses of the isolates reveled substantial phenotypic and genomic differences in comparison to the reference H. alvei of 500 lactose-negative E. coli strains (a trait of E. albertii) from the collection of bacterial strains at the Department of Microbiology of the Medical University of Bialystok, isolated from the feces of patients (children and adults) hospitalized at the University Pediatric Hospital in Bialystok, the Municipal Hospital in Bialystok, and the Psychiatric Hospital in Choroszcz.
Human fecal samples (20-80 g) were collected in sterile containers, while sterile swabs with a transport medium were used to collect bird droppings. In the next step, all stool samples were plated on MacConkey, Hektoen, CHROMagar E. coli selective agar media (bioMerieux, France), blood agar, BHI broth (Oxoid, United Kingdom), and incubated at 37 • C for 48 h. In addition, fecal samples from human diarrhea were tested for common bacterial enteropathogens using culture methods on additional selective media: (i) SS and SF agar broth (bioMerieux, France) for Salmonella spp. and Shigella spp., (ii) selective Campylobacter agar (bioMérieux, France), and (iii) selective Yersinia agar (Merck, Germany). Commercial enzyme immunoassays or immunochromatography were used to identify Clostridioides difficile, norovirus and rotavirus/adenovirus infections. The purpose of these additional tests was to determine, in the case of E. albertii positive samples, whether this was the only etiological factor of the diarrhea.
Water samples (100 mL) were taken into sterile glass bottles and analyzed in accordance with the PN-EN ISO 9308-1:2004 procedure, based on the enumeration of E. coli and coliform bacteria in water by membrane filtration method and culture on selective Chromogenic Coliform Agar (CCA). Food samples were analyzed in accordance with PN-ISO 4832: 2007, i.e., the enumeration of E. coli and coliform bacteria using colonycount technique.

Phenotypic Features of E. albertii Biochemical Profiles
Biochemical features were determined based on API 50 CHE, API 20E and ID32E (bioMérieux, Marcy-l'Étoile, France) (also in order to determine the most common species identified by these commercial tests, which do not include E. albertii) using ATB Expression (bioMérieux). The tests were performed according to the manufacturer's recommendations.

Antimicrobial Susceptibility Patterns
The susceptibility of each isolate was tested with disc diffusion against a panel of 10 antibiotic agents used in E. coli infections in human: ampicillin 10 µg/disc, amoxicillinclavulanic acid 20-10 µg/disc, cefuroxime 30 µg/disc, cefepime 30 µg/disc, imipenem 10 µg/disc, chloramphenicol 30 µg/disc, ciprofloxacin 5 µg/disc, levofloxacin 5 µg/disc, gentamicin 10 µg/disc and trimethoprim/sulfamethoxazole 1.25-23.75 µg/disc (all antibiotics from Oxoid Ltd., Cambridge, UK) ( Table 1). Susceptibility testing was conducted using clinical breakpoints on Mueller-Hinton agar according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) recommendations, including the E. coli ATCC 25922 as a reference strain [25]. When resistance of the strains was detected, the MIC values were determined using E-tests (bioMérieux). The production of E. albertii extended-spectrum β-lactamases (ESBLs) was determined using the double-disk synergy test (DDST) [26]. Specifically, this was performed with cefotaxime (30 µg) and ceftazidime (30 µg) disks placed at a distance of 20 mm (center to center) from the amoxicillin-clavulanic acid disk (20/10 µg). A cefepime (30 µg) disk was placed in the same culture medium in order to improve the detection of ESBL during the simultaneous stable hyperproduction of an AmpC β-lactamase. The test result was considered positive when an enhancement of the inhibition zone around at least one of the antibiotic disks (cefotaxime, ceftazidime, or cefepime) toward the clavulanic acid disk was observed. The control strains Klebsiella pneumoniae ATCC 700603 (ESBL positive) and E. coli ATCC 25922 (ESBL negative) were used for quality control.

Restriction Fragment Length Polymorphism (RFLP)
All E. albertii isolates were typed by pulse field gel electrophoresis (PFGE) with XbaI endonuclease according to the PulseNet protocol (http://www.cdc.gov/pulsenet/pdf/ ecoli-shigella-salmonella-pfge-protocol-508c.pdf) using CHEF Mapper (Bio-Rad, Cressier, Switzerland). PFGE gels after staining with ethidium bromide were captured in the GelDoc 2000 system (Bio-Rad, Hercules, CA, USA) and analyzed with BioNumerics software ver. 7 (Applied Maths). PFGE patterns were compared using the Dice coefficient and unweighted pair group method using arithmetic averages (UPGMA) clustering, with a 1% band position tolerance window and 1% optimization. Isolates were defined as having a clonal relationship if they possessed 85% similarity to the PFGE patterns. Further analyses (except for plasmid profiling) were performed only for nonclonal strains.

Plasmid Profiles
Plasmid profiling assays are necessary from the perspective of virulence determination, strain typing and assembly of whole genome sequences. Therefore, the number and the size of E. albertii plasmids < 100 kb were determined using a commercial plasmid isolation kit (e.g., Qiagen Plasmid Plus Midi Kit; Qiagen) and regular gel electrophoresis. On the other hand, larger plasmids (>100 kb) were detected by means of the PFGE method after plasmid linearization by nuclease S1 [27].

Whole Genome Sequencing of E. Albertii Using Next-Generation Sequencing (NGS)
In total, genomes of 25 nonclonal E. albertii isolates were sequenced using short-and long-read sequencing with MiSeq (Illumina, San Diego, CA, USA) and MinION (Oxford Nanopore Technologies, Oxford, UK) instruments, respectively. In detail, the total genomic DNA of bacteria was extracted using the CTAB/lysozyme method [28], and for the shortread sequencing, Paired-End TruSeq-like libraries were constructed using KAPA Library preparation kit (KAPA/Roche, Basel, Switzerland) according to the manufacturer's instructions, followed by sequencing in paired-end mode with MiSeq Reagent Kit v3 (600-cycle). MinION sequencing was performed using SQK-LSK108 chemistry and R9.4/R9.5 ONT flowcells. Raw nanopore data were basecalled using Albacore v2.0.2 (Oxford Nanopore Technologies, Oxford, UK). The obtained short and long reads were filtered by quality using FastX toolkit ver. 0.0.14 (http://hannonlab.cshl.edu/fastx_toolkit/) and Porechop ver. 0.2.4 (https://github.com/rrwick/Porechop/), respectively. Finally, de novo assembly of the genomes was performed using Canu ver. 1.6 [29], and the Canu assembled contigs were further polished using Illumina sequencing reads and Oxford Nanopore ont-assembly-polish pipeline (https://github.com/nanoporetech/ont-assembly-polish/; accessed on 21 April 2021). The remaining gaps in the genome assemblies were closed by the PCR amplification of DNA fragments, followed by Sanger sequencing with an ABI3730xl Genetic Analyzer (Life Technologies, Carlsbad, CA, USA) using BigDye Terminator Mix v. 3.1 chemistry (Life Technologies). All of the sequence errors and missassemblies were further corrected using Seqman software ver. 7 (DNAStar, Madison, WI, USA) to obtain complete nucleotide sequences of bacterial genomes. The complete sequences of chromosomes and plasmids have been deposited in the GenBank database (BioProject ID: PRJNA931640).

Pangenome Estimation
Roary software [31] was used to estimate the size of E. albertii pangenome and identify the core genes.

Statistical Analysis
The results from biochemical tests, PFGE, MLST, antimicrobial susceptibility patterns, and other binary or categorical variables were analyzed using statistical methods, e.g., cluster analysis with BioNumerics ver. 7 (Applied Maths), Statistica (StatSoft, Tulsa, OK, USA) and/or Stata (StataCorp, College Station, TX, USA) software.

Results
From the tested materials-(i) 497 samples of bird droppings, (ii) 1154 fecal samples from children and adults, (iii) 212 food samples, (iv) 92 samples of water from natural reservoirs and tap water-a total of 42 strains of E. albertii were isolated. E. albertii could not be isolated from the collected human stool samples, indicating that its distribution is mainly restricted to birds and the environment.
In the present study, diagnostic PCR proved useful for rapid identification of the E. albertii strains, since different biochemical profiles were found in 38.1% of E. albertii isolates, which influenced their identification ( Figure 1). Briefly, the studied strains of E. albertii were identified mainly as Escherichia coli type 1 or 2 and Hafnia alvei, and occasionally as Escherichia fergusonii, Edwardsiella trada, Yersinia ruckei, Burkholderia cepacia, Vibrio parahaemolyticus, and Morganella morganii. Unlike many E. coli strains, E. albertii strains failed to ferment a number of sugars, including lactose, dulcitol, Lrhamnose, and melibiose. The ability to ferment D-sorbitol, a trait almost exclusively associated with E. coli among Escherichia species, clearly helps separate this group (Dsorbitol negative) from the type species of the genus.
They can be biochemically distinguished from other Escherichia species by a number of tests, including the inability to utilize malonate and the lack of acid production from D-xylose, D-arabitol, melibiose, and cellobiose. We also found that the weak to moderate L-prolinoaminopeptidase activity noted in E. albertii strains may be a useful auxiliary test to separate this group from H. alvei. Because the species E. albertii was recently discovered, commercial systems do not currently include the species in their database or do not take into account the differences in their biochemical profiles. The results of testing 42 strains of E. albertii on various commercial systems suggest that a majority (80%) of the obtained biochemical profiles are matched with other species included in their databases with an unacceptable probability (around 60%) or are classified as unidentified. When identifying bacteria, the possible presence of E. albertii may be indicated by a result identifying the isolate as H. alvei, which is both L-rhamnose negative and D-xylose negative. Such results should trigger a more in-depth analysis of the test strain with additional testing. However, 45% of E. albertii isolates generated acceptable (92%) to perfect (97 to 99%) identifications as H. alvei or E. coli. Therefore, E. albertii strains may not be identified in routine diagnostics. They can be biochemically distinguished from other Escherichia species by a number of tests, including the inability to utilize malonate and the lack of acid production from Dxylose, D-arabitol, melibiose, and cellobiose. We also found that the weak to moderate Lprolinoaminopeptidase activity noted in E. albertii strains may be a useful auxiliary test to separate this group from H. alvei. Because the species E. albertii was recently discovered, commercial systems do not currently include the species in their database or do not take into account the differences in their biochemical profiles. The results of testing 42 strains of E. albertii on various commercial systems suggest that a majority (80%) of the obtained biochemical profiles are matched with other species included in their databases with an unacceptable probability (around 60%) or are classified as unidentified. When identifying bacteria, the possible presence of E. albertii may be indicated by a result identifying the isolate as H. alvei, which is both L-rhamnose negative and D-xylose negative. Such results should trigger a more in-depth analysis of the test strain with additional testing. However, 45% of E. albertii isolates generated acceptable (92%) to perfect (97 to 99%) identifications When assessing drug susceptibility, we showed that, in total, 33.3% of the 42 tested E. albertii isolates were resistant to one antibiotic, and 16.7% of isolates showed resistance to two antibiotics. Most often, E. albertii was resistant to ampicillin (45.0%), followed by amoxicillin with clavulanic acid (9.0%), and very uncommonly to chloramphenicol (2.0%), ciprofloxacin and cefuroxime (5.0% each). However, none of the isolates were resistant to cefepime, imipenem, levofloxacin, gentamicin, or trimethoprim/sulfamethoxazole ( Table 1). The tested E. albertii isolates did not produce extended-spectrum β-lactamases.
Of the 42 isolated strains of E. albertii, we were able to obtain complete sequences of the genomes (chromosomes and plasmids) of 25 by sequencing using the new generation NGS. The characteristics of the sequenced strains are presented in Table 2. Similarly, genetic typing showed high variability in isolates. In the PFGE method, 90.5% of the strains had distinct pulsotypes (Figure 2). In contrast, in MLST typing, 85.7% of the strains were assigned distinct sequence types (STs), of which 64% were new ST types ( Table 2). 1). The tested E. albertii isolates did not produce extended-spectrum β-lactamases.
Of the 42 isolated strains of E. albertii, we were able to obtain complete sequences of the genomes (chromosomes and plasmids) of 25 by sequencing using the new generation NGS. The characteristics of the sequenced strains are presented in Table 2. Similarly, genetic typing showed high variability in isolates. In the PFGE method, 90.5% of the strains had distinct pulsotypes (Figure 2). In contrast, in MLST typing, 85.7% of the strains were assigned distinct sequence types (STs), of which 64% were new ST types (Table 2).  In the case of typing virulence factors, the genes encoding cytolethal distending toxin (CDT) and porcine attaching and effacing-associated (Paa) toxins were demonstrated in 100% of the tested E. albertii isolates. On the other hand, the presence of genes encoding IbeA toxin (brain microvascular endothelial cell invasion toxin), CdtB type 2 toxin, Tsh toxin (temperature-sensitive hemagglutinin), and Shiga toxin (Stx2f) was found in 26.2%, 9.7%, 1.7% and 0.4% of E. albertii isolates, respectively. A comparison of the virulence profiles of the sequenced E. albertii strains based on bioinformatic database analysis virulence genes (E. coli_vf n = 2710 and vfdb n = 3688) is shown in Figure 3.
The chromosome size of the tested strains ranged from 4,573,338 to 5,141,010 bp (average 4,784,003 bp). In each strain, the presence of at least one plasmid (range 1-8, average 3.8) was found in the range 1589-177,310 bp (average 58,791.6 bp). Among the plasmids, 20 different types were detected replicons, which were dominated by IncF, IncI, IncX, IncY (46.9%), and Col (37.5%) replicons. The genome characteristics are presented in Table 3.
In the case of typing virulence factors, the genes encoding cytolethal d (CDT) and porcine attaching and effacing-associated (Paa) toxins were d 100% of the tested E. albertii isolates. On the other hand, the presence of IbeA toxin (brain microvascular endothelial cell invasion toxin), CdtB ty toxin (temperature-sensitive hemagglutinin), and Shiga toxin (Stx2f) was 9.7%, 1.7% and 0.4% of E. albertii isolates, respectively. A comparison of th files of the sequenced E. albertii strains based on bioinformatic database an genes (E. coli_vf n = 2710 and vfdb n = 3688) is shown in Figure 3.   The results of the pangenome analysis of the studied E. albertii isolates, taking into account the earlier isolate of E. albertii z of our center (DOI: 10.1128/genomeA.00004-14) and the 94 strains of E. albertii available in GenBank database (119 strains in total), are presented in Table 4 and Figure 4. Their phylogenetic relationship is shown in Figure 5.

Discussion
According to WHO estimates, diarrheal diseases are responsible for two billion deaths worldwide each year. In addition, infectious diarrhea is the leading cause of death among children under age 5 in developing countries [32], and in developed countries it is mainly the problem of morbidity and socio-economic costs [33]. Infections caused by enteropathogens can also result in serious complications, such as neurological or autoimmune diseases (Guillain-Barré syndrome). Unfortunately, a number of factors (demographic changes, centralization of food production, changes in eating habits, globalization, etc.), combined with the rapid adaptation of bacteria to new conditions, result in the formation of new enteropathogens [34].
E. albertii is considered a novel enteropathogen for humans and animals. Reports of the isolation of E. albertii from human and animals, the environment and food samples come from different geographical areas of the world, confirming its globalization. The clinical significance of E. albertii is not yet fully understood, partially because it is difficult to distinguish it from other Enterobacteriaceae using routine identification protocols. E. albertii was first isolated from infants in Bangladesh, and in subsequent years its involvement in infections in children and adults was confirmed [4,5,35]. In 2010, we also isolated E. albertii from the diarrheal feces of a child, but to date this has been a single case [23].
People infected with E. albertii usually show symptoms related to gastroenteritis, including watery diarrhea, dehydration, abdominal distension, vomiting, and in some cases fever. In most cases, these infections are self-limiting, with patients often recovering with little to no treatment [5,33,35]. However, a small subset of E. albertii isolates carry the stx2a allele and can cause bloody diarrhea [19]. Incubation periods of E. albertii infections that present as diarrhea are relatively short (average 12-24 h), while mortality rates are unknown. However, morbidity rates appear to be relatively high, i.e., >50% of the exposed population [36,37].
Other researchers, Afshin et al. [38] and Zaki et al. [39], also described cases of urinary tract infections with E. albertii as the etiological agent. In addition, a single case of E. albertii bacteremia was reported in an elderly woman with multiple comorbidities, possibly due to its ability to move from the intestinal lumen to other extraintestinal sites [35,40]. It may seem that these are relatively rare cases of E. albertii involvement in human infections; however, this may be due to misidentification.
The problem of increasing antimicrobial resistance among many bacteria should also be taken into account. E. albertii strains resistant to important antibiotics, including ampicillin, gentamicin, ciprofloxacin, norfloxacin, trimethoprim/sulfamethoxazole, rifampicin, meropenem and imipenem, were found [21,[41][42][43]. In addition, multidrug-resistant strains of E. albertii [42] that showed resistance or coded for resistance to antibiotics from at least eleven classes [44,45] have been described. Combined with the virulence potential of this pathogen, these strains pose a serious threat to public health.
The data obtained from the conducted studies suggest that E. albertii may occur in animals or humans as a commensal or pathogen. E. albertii has been recognized as the etiological agent of diseases in many bird species [7]. In 1994, numerous dead finches were recorded in Scotland [46,47]. The necropsy showed that the causative agent was E. coli belonging to the O86:K61 serotype [46,47], which was previously associated with diarrhea in calves, pigs and horses and with cellulitis in broiler chickens [47]. However, a later study that re-examined these isolates with MLST and a larger panel of phenotypic and biochemical tests showed that it was in fact the E. albertii species [1,7]. In Australia, studies of vertebrate gut bacteria have also shown that birds (chickens and magpies) were the main reservoir of E. albertii [1,48,49]. Our research conducted in northeastern Poland also confirmed the relationship between the occurrence of E. albertii and migratory birds and the water bodies at which they stop. We isolated our strains mainly from the feces of cranes, starlings and geese, and from the water near their habitats. Pigeons were carriers of E. albertii in the urban environment. Our research confirmed the results of other researchers [50][51][52][53] that E. albertii is present in the wild bird population, which may be the potential source of this pathogen for humans, other animals, and the environment.

Conclusions
The isolates tested in our center do not currently constitute an important reservoir of resistance genes. We also found them to be highly sensitive to the most commonly used antibacterial drugs. However, in terms of virulence genes, the pathogenic potential of the studied E. albertii strains may range from watery diarrhea typical of enteropathogenic E. coli strains to bloody diarrhea typical of EHEC strains, due to the presence of genes encoding Shiga toxin. An issue requiring further research is the potential of E. albertii strains containing the IbeA toxin (characteristic of NMEC strains; E. coli neonatal meningitides) to cause infections of the central nervous system. In addition, their pathogenicity to animals-pigs-is also noteworthy due to the presence of the Paa toxin typical of porcine enteropathogenic Escherichia coli (PEPEC) strains. Further understanding of the pathobiology of E. albertii, as well as the mechanisms of colonization, survival, and dissemination within and between hosts, is still limited and requires further investigation. This will only be possible if simple and effective diagnostic tools are developed to enable its effective isolation and identification.
The study provides several important data regarding the epidemiology of E. albertii; for instance, we noted a high genetic similarity between our human isolate-Ea_BIA_KF1 [22] and strains isolated from the environment (Ea_BIA_47) and wild birds (Ea_BIA_22) as well as those described in other countries, e.g., Japan [36] (Figure 5). As a result, it helps for a better estimation of a risk that E. albertii may pose for public health. In the conducted studies, we confirmed the high genetic diversity of E. albertii strains, as is the case with E. coli strains, and the open nature of the pangenome of this bacterium, in which plasmids are the main mobile genetic factors responsible for its plasticity. They may play an important role in the horizontal exchange of virulence genes with E. coli strains pathogenic to animals and humans.
Author Contributions: All authors had full access to all of the data in the study and take responsibility for their integrity and the accuracy of the analysis. K.L. was responsible for coordination of all tasks and the logistic planning of the project. K.L. was responsible for the study concept, design, data acquisition, underlying data verification, statistical analysis and manuscript drafting. I.Ś. was responsible for genetic research and participation in bioinformatic elaboration of genomic data. T.D. she was responsible for bacterial culture, identification and determination of bacterial susceptibility, statistical analysis and manuscript drafting.  The application submitted to the Bioethical Committee of the Medical University of Bialystok, apart from the description of the research and the samples of materials used in them, also contained a template for patients entitled "Consent to perform microbiological tests" with information on the nature and purpose of the tests.

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper.