The strain Escherichia coli М-17, introduced by Peretts (1955), is used for producing coli-containing probiotics in Russia. This strain is characterized by a broad spectrum of useful properties and has been used in biotechnology for over seventy years; however, its technological efficiency, is relatively low due to a low biomass accumulation rate. Search for alternative industrially promising strains of this species suitable to be used for technological purposes is therefore an urgent issue. This work represents the results of the study of the genomes of two Escherichia strains, M-17 conventionally used for production of probiotics and the new strain E. coli BM isolated from the intestines of a healthy human and characterized by good technological characteristics.

According to the present-day normative documents, the safety of probiotic producer strains must be confirmed by the absence in them of transmissive genes of antibiotic resistance, virulence and pathogenicity genes, as well as of incorporated plasmids and transposons (Metodicheskie ukazaniyano. 2.3.2.2789-10, 2010; Metodicheskie ukazaniya no. 4.2.2602-10, 2011). The biological properties and technological characteristics of both strains were studied earlier; however, no in-depth study of their genomes was conducted (Gaiderov, 2007; Ladygina et al., 2009). Within the framework of this work, we studied whole-genome sequences of the strains using the following resources available online: the Warwick University EnteroBase International Escherichia and Shigella genomic sequence database (Warwick Medical School, UK, http://enterobase.warwick.ac.uk) and the services available at the Center for Genomic Epidemiology website (Denmark, http://www. genomicepidemiology.org/). Thus, analysis of genomes for the presence of the pathogenicity, antibiotic resistance, and virulence determinants was carried out, as well as their molecular typing using the multilocus sequence typing (MLST).

The goal of the present work was to study the genomes of the strains E. coli M-17 and E. coli BM for the presence of the antibiotic resistance, pathogenicity, and virulence determinants, as well as of incorporated plasmids and transposons.

MATERIALS AND METHODS

Study subjects. The work used the strains Esc-herichia coli M-17 and E. coli BM. The strain E. coli M-17 was obtained from the NPO Mikrogen Stock Cultures Collection, Ministry of Health of the Russian Federation, Nizhny Novgorod, ImBio Enterprise for Manufacture of Bacterial Preparations. The strain is registered in the Collection of the Scientific Center of Expert Evaluation of Medical Products (SCPM State Collection of Pathogenic Microorganisms, no. 240418). The strain E. coli BM obtained from the collection of the Laboratory for Human Microbiome and the Means of Its Correction, Blokhina Nizhny Novgorod Research Institute of Epidemiology and Microbiology, Federal Service for Supervision of Consumer Rights Protection and Human Wellbeing, was isolated from intestine of a healthy human and was characterized as a high-technology one. The strains were obtained in the liophilically dry state.

Strain recovery was performed using the HiMedia nutrient broth (India). The grown culture was titrated in the 10–1–10–7 dilution range, and Endo medium (nutrient medium for isolation of enterobacteria–GRM Endo agar, Obolensk, Russia) was inoculated with 0.05 mL of the culture; the inocula were incubated at 37 ± 1°С for 24 h. The colonies of grown microorganisms were assessed by morphology and selected for extensive study of the biochemical features using Diagnostic Systems Biochemical Enterobacterium-Differentiating Plates (BEDP) (Russia), Paper Indicator Systems (PIS) for identification of microorganisms—kit no. 2 for intergeneric and species differentiation of enterobacteria (NPO Mikrogen, Ministry of Health of the Russian Federation, Russia), as well as HiMedia media (India).

Genomic DNA was isolated using the commercial QIAamp DNA Mini Kit (Qiagen, Germany); fragmentation was carried out using the Covaris E210 ultrasonic fragmentation system (Applied Biosystems, United States) according to the manufacturer’s instructions. The mixture was purified and 300‒500-bp fragments were selected using Agencourt AMPure magnetic beads (Beckman Coulter, United States). The TrueSeq Kit (Illumina, United States) was used for library preparation; sequencing was performed on the MiSeq platform (Illumina, United States). Short reads were collected into contigs using the CLC Bio v. 8.0 software package (Aarhus, Denmark).

Genome annotation was performed using the Prokka v. 1.11 utility (Seemann, 2014) and the RAST genomic server (http://rast.nmpdr.org). For detailed genome studies, specialized software products available at the Center for Genomic Epidemiology website (https://cge.cbs.dtu.dk/services/) were used. The study of the CRISPR region was carried out with the CrisprFinder software package (Grissa et al., 2017); search for the antibiotic resistance and pathogenicity determinants, with ResFinder 2.1 and PathogenFinder (Cosentino et al., 2013; Zankari et al., 2012). The VirulenceFinder package was used to search for virulence determinants; PlasmidFinder 1.3, to detect incorporated plasmids; SerotypeFinder 1.1 was used for serotype determination; the FimHTyper service, to study the genovariant of FimH adhesin (Joensen et al., 2014; Carattoli et al., 2014; Joensen et al., 2015; Roer et al., 2017).

MLST typing was carried out using the MLST-1.8 Server software package (Larsen et al., 2012) and the Warwick University MLST database (Mark Achtman Database, http://mlst.warwick.ac.uk/mlst/dbs/ Ecoli/) by analyzing the sequences of seven constitutive housekeeping genes (Alikhan et al., 2018).

RESULTS AND DISCUSSION

Investigation of the biochemical properties of E. coli strains using the standard BEDP and PIS test-systems, as well as additional tests (xylose, rhamnose, dulcitol, and gelatinase activity) using HiMedia media, revealed differences in the biochemical profiles of the strains. E. coli M-17 was capable of fermenting sucrose, maltose, sorbitol, glucose, lactose, arabinose, mannitol, xylose, and rhamnose; it formed indole, possessed β-galactosidase activity, utilized sodium citrate with glucose, and possessed lysine and ornithine decarboxylases. The strain was unable to liquefy gelatin, utilize citrate and sodium malonate, did not have arginine dehydrolase and phenylalanine deaminase, did not form acetylmethylcarbinol, did not ferment inositol and dulcitol, did not hydrolyze urea and did not form hydrogen sulfide. It is this biochemical profile that is characteristic of the strain E. coli M-17 proposed by Peretts in 1930 and declared in the pharmacopoeial monograph for dry colibacterin (Peretts, 1955). The strain E. coli BM differed in that it was incapable of fermenting sucrose, malose, and sorbate and of decarboxylating ornithine.

When whole-genome sequencing of E. coli M-17 and E. coli BM was performed, 42 (N50 = 465.806, mean coverage 250) and 59 (N50 = 205.755, mean coverage 250) contigs, respectively, were collected. Further analysis allowed us to determine the main characteristics of the strain genomes (Table 1).

Table 1.   Phenotypic and genetic characteristics of E. coli strains M-17 and BM
Full size table

It is evident that both strains possessed similar genome sizes and both genomes include colicin V synthesis operons, while E. coli M-17 additionally contained colicin B synthesis genes.

CRISPR regions were present in the genomes of both strains, E. coli M-17 CRISPR locus was within contig 57 (LBDD01000002); that of E. coli ВМ was within contig 68 (LBDC01000001). In terms of structural organization, these determinants of both strains can be assigned to type I-E, which is typical of members of the genus Escherichia (Makarova et al., 2015). A distinctive feature of this type is the presence of the cas3 gene, which encodes single-stranded DNA stimulating helicase that untwists double-stranded DNA (Fig. 1).

Fig. 1.
figure 1

Organization of CRISPR cassettes of the two E. coli strains. The arrows designate the corresponding Cas proteins, the diamonds, repeats; the triangles, unique spacers.

Full size image

In a number of cases, the CRISPR cassettes of microorganisms are used for strain typing as this genomic region does not correlate with phylogeny and its structure may differ even in different strains of the same species (Makarova et al., 2015). However, we established in the course of the work that, despite the fact that the CRISPR loci of both strains differed in the arrangement of the Cas proteins and had different sequences of spacers and repeats (Fig. 1), their spacers were not strain-specific and had 100% identity with the sequences of a number of other E. coli strains represented in the GenBank database.

The genomic context of strain metabolism was studied with RAST. Several pathways of central carbohydrate metabolism were found to be determined in the genomes: glycolysis, the Entner–Doudoroff pathway, and the pentose phosphate pathway. Using the KEGG map algorithm (http://rast.nmpdr.org/seedviewer.cgi), it was shown that the metabolic pathways of the strains had no specific features. However, the genome of E. coli M-17 had the determinants responsible for sucrose utilization: the sucrose-6-phosphate hydrolase (scrB) and phosphofructokinase (fruK) genes, and the corresponding transportation system (PTS scrA). At the same time, E. coli ВM did not possess these determinants, which was consistent with the specific features of its biochemical profile—the strain was unable to ferment sucrose. Moreover, E. coli ВM did not decarboxylate ornithine and did not ferment maltose and sorbitol, this being its distinctive feature, because over 90% of E. coli strains utilize maltose and sorbitol, and the ability to utilize sucrose and ornithine is variable within the species (De Vos et al., 2009). It was noted that the determinants of maltose, sorbitol, and ornithine metabolism were present in the strain genome, but for some reason, they did not reveal themselves phenotypically. Such a loss of individual determinants or their inactivation may be associated with the adaptation of a microorganism to a specialized ecological niche—human intestine (Ravin and Shestakov, 2013). On the whole, both strains possessed a marked metabolic potential, being capable of synthesizing colicins and utilizing a broad spectrum of substrates. The whole-genome sequences of the strains were analyzed for the presence of the antibiotic resistance, pathogenicity, and virulence determinants, as well as for the presence of incorporated plasmids and transposons. In the process, it was established that both genomes shared the transporter gene cluster (mdtABCD) responsible for resistance to novobiocin and doxycholate, the MAR locus genes responsible for the system of efflux of pharmaceutical preparations, including antibacterials, as well as other forms of molecular efflux pumps: cmeA, cmeB, tolC, acrR, macA, macB, the MATE and MFS families. All the above determinants have chromosomal localization, are typical of this genus, and do not pose threat in terms of transmissive spread (Baranova and Nikaido, 2002; Alekshun and Levy, 2004).

Investigation of E. coli M-17 did not reveal the antibiotic resistance determinants capable of transmissive transfer, whereas the sul2 sulfanilamide resitance gene in the genome of E. coli ВM was found to encode sulfanilamide-resistant dehydropteroate synthase, as well as the aminoglycoside resistance genes: the aadA1 gene encoding aminoglycoside-adenylyl transferase, the genes strB and strA encoding aminoglycoside phosphotransferases, and the dfrA1 trimethoprim resitance determinant. In the studied genome, the determinants sul2, strB, and strA found in the twenty-second contig were flanked by proteins from the family of transposases (WP_001120891.1), and the genes aadA1 and dfrA1 located within the sixty-first contig were the components of transposon Tn7 that plays an important role in genetic information transfer and the spread of antibiotic resistance genes (Petrova, 2008).

When studying the virulence and pathogenicity determinants, we did not reveal in the genomes of either strain the determinants encoding enterotoxins Stx1 and Stx2, thermostable toxin St, intimins Eae, and enterohemolysins Ehx, which are linked to the clinically significant diarrheagenic strains of E. coli (Kartsev et al., 2018). However, the genome of E. coli BM was found to have other significant determinants: the tieB gene of plasmid origin also encoding enterotoxin, the bifunctional adhesion gene iha, the glutamate decarboxylase gene gad, and the gene encoding the Exc1 mobile transfer protein (Table 1). According to the PathogenFinder service, E. coli BM was assigned to the strains pathogenic for humans (score = 0.868). No pathogenicity and virulence determinants were revealed in the genome of E. coli M-17; the strain was characterized as nonpathogenic (score = 2.0).

The strain genomes were tested for the presence of incorporated plasmids and their fragments using the PlasmidFinder-1.3 software. It was found that the genome of E. coli M-17 did not contain incorporated plasmids, while the genome of E. coli ВM contained the fragments of two plasmids—IncFIB typical of uropathogenic Escherichia strains and Col156. The genes of plasmid origin were located within contigs 46 and 73 and encoded a complex of the conjugative transfer proteins TraA, TraM, TraY, etc. (Table 1).

The presence and genovariants of the fimH gene determining type I mannose-sensitive pili responsible for adhesion of E. coli cells (Markova et al., 2018) were then studied. It was established that both strains carried this gene, E. coli M-17 fimH gene being identified with type 14; fimH of E. coli ВM belonged to type 215. Some researchers consider the presence of fimH as a property determining potential virulence of a microorganism due to high adhesion values and ability to form biofilms and, hence, unsuitability for industrial purposes (Nechaeva et al., 2016; Slukin et al., 2017). However, according to the present-day data, the presence of type I pili and of adhesin fimH is not always associated with virulence and occurs in E. coli strains not possessing the adhesive–invasive phenotype. Note that the authors link an increased capacity for adhesion and invasion to point mutations of the fimH gene at hot points (Dreux et al., 2013). No mutations determining the G73A/E/R/W and T158A/P substitutions and contributing to increased adhesion of a bacterial cell on intestinal epithelial cells, which is characteristic of adhesive–invasive E. coli (AIEC), were found in the fimH genes of the strains studied. Fim-types 14 and 215, the identity with which was revealed in the course of the study, are not inherent in pathogenic strains and occur rather rarely among the population of E. coli (Roer et al., 2017).

Using the SerotypeFinder-1.1 software product, we analyzed the determinants of E. coli antigens: the genes encoding the structural protein flagellin FliC (KLD44527.1; KLD52182.1) and the O-antigen proteins: Wzx (KLD49178.1; KLD50213.1) and Wzy (KLD49266.1; KLD50216.1). It was established that E. coli M-17 and E. coli ВM belonged to serotypes О2:Н6 and О21:Н4, respectively.

In the framework of the study, MLST typing of the strains was carried out using the scheme proposed by Worwick University. This scheme uses the fragments of seven housekeeping genes: the adenylate kinase (adk), fumarate hydratase (fumC), DNA gyrase (gyrB), isocitrate dehydrogenase (icd), malate dehydrogenase (mdh), adenylosuccinate dehydrogenase (purA), and recombinase А (recA) genes. It was established that the strain E. coli M-17 belonged to sequence type 141 (ST-141), and E. coli ВM was attributed to sequence type 10 (ST-10) (Table 2).

Table 2.   Allelic profiles of the studied Escherichia strains
Full size table

According to the EnteroBase International Database information (accession date April 28, 2018), 12 out of 24 strains with the О2:Н6 serotype were identified with sequence type 141; they are characterized as nonpathogenic and are not related to the clinical manifestations of diseases.

The strain E. coli ВM had the serotype О21:Н4 and was assigned to sequence type 10 (ST-10). At the moment of accession (April 28, 2018), four strains with serotype О21:Н4 were deposited to the EnteroBase database, two of them belonging to sequence type 10. Importantly, one of these strains, which is of human origin and is characterized as enteroaggregative, is associated with the diarrheal syndrome. Moreover, the heteropathogenic E. coli strains known to act as entero- and uropathogens belong to the same ST-complex (ST complex 10) (Kartsev et al., 2018).

Thus, it was shown in the course of the work that E. coli M-17 does not contain the antibiotic resistance and virulence determinants or incorporated plasmids; in terms of the pooled data of analysis of the whole-genome sequence and molecular typing, the strain is not pathogenic for humans and can be used for the manufacture of probiotics and biologically active food supplements.

In turn, the strain E. coli ВM possesses a complex of antibiotic resistance and virulence genes, carries the mobile element genes, incorporated plasmids, and is found to be pathogenic for humans. The results of molecular typing enabled us to establish its sequence type–ST-10, and confirmed its identity with the epidemically significant clone. The results obtained rule out the use of this strain as a probiotic according to the present-day regulatory documents.

Our findings prove that the use of whole-genome sequencing that allows researchers to detect and analyze all determinants of interest and to perform molecular typing of the strain is a priority in the work with industrially promising strains.