Differences in Extended-Spectrum Beta-Lactamase Producing Escherichia coli Virulence Factor Genes in the Baltic Sea Region

The aim of this study was to compare the prevalence of different virulence factor (VF) genes in extended-spectrum beta-lactamase (ESBL) producing Escherichia coli strains isolated from the Baltic Sea region. A total of 432 strains of phenotypically ESBL positive E. coli were collected from 20 institutions located in Estonia, Latvia, Lithuania, and the region of St. Petersburg in Russia from January to May 2012 and analyzed for phylogenetic group and prevalence of 23 VF genes. The strains were collected from clinical material (urine, blood, wound, and respiratory tract). Bacterial isolates were compared according to phylogenetic group, clinical material, and geographical origin. Most of the VF genes were concentrated within phylogenetic group B2 and/or D. When comparing strains isolated from different countries, it was found that strains originating from Estonia and Latvia belonged mainly to group B2 and strains from Lithuania and Russia mainly to groups B2 and D. The P-fimbrial adhesin gene papEF was more prevalent in Russian strains, colicin gene cvaC in Lithuanian strains, and capsular gene kpsMTII in Latvian strains; serum resistant gene traT was less prevalent in Estonian strains. The regional differences of VF genes remained statistically significant after taking into account the phylogenetic distribution in the countries.

Pathogenic E. coli clones have acquired specific virulence factors (VF), which confer an increased ability to adapt to new niches and allow them to cause a broad spectrum of diseases [3]. ExPEC isolates have functionally similar VF profiles and clonal background, and they are distinct from commensal and intestinal pathogenic E. coli strains [4]. VF-s of ExPEC include different adhesins, toxins, capsules, siderophores, invasins, and antibiotic resistance. These VFs contribute to colonization and invasion into host tissues, avoidance to immune responses, and antimicrobial drugs and acquiring nutrients from the host [5,6]. The management of infections caused by E. coli is complicated due to the increasing resistance to antibiotics. Extended-spectrum -lactamase (ESBL) production is a common mechanism of resistance to 2 BioMed Research International third-generation cephalosporins in E. coli, associated with the frequent use of -lactam antibiotics in treatment of serious E. coli infections [5].
E. coli strains can be divided into four main phylogenetic groups: A, B1, B2, and D [7]. The strains causing extraintestinal infections belonging preferentially to group B2 and to a lesser extent to group D. Commensal strains largely belong to groups A and B1 [8,9]. However, the distribution of phylogenetic groups may vary in different geographic regions. It depends on the climatic zone and environmental factors [10].
The aim of this study was to compare the phylogenetic distribution and prevalence of different VF-s in extended-spectrum -lactamase producing extraintestinal E. coli strains isolated from the Baltic Sea region.
Some information presented in this paper was previously demonstrated at 23rd ECCMID held on 27-30, April, in Berlin, Germany [11].

Strains.
A total of 423 strains of phenotypically ESBL positive Escherichia coli were investigated. All consecutive ESBL positive strains were collected from patients from 20 institutions located in Estonia ( = 5), Latvia ( = 4), Lithuania ( = 3), and the region of St. Petersburg in Russia (further referred to as Russian strains) ( = 8) from January to May 2012. The strains were isolated from different clinical materials (Table 1), identified as E. coli by Matrix Assisted Laser Desorption/Ionization Time of Flight instrument (MALDI-TOF; Bruker Daltonik GmbH, Germany) and ESBL production was confirmed by ROSCO ESBL kit (Rosco Diagnostica A/S, Denmark) [12].

Phylogenetic Analysis.
All bacterial strains were assigned to one of the four main E. coli phylogenetic groups (A, B1, B2, and D) according to PCR-based method published by Clermont et al. [7].
PCR conditions were as follows: 4 min at 94 ∘ C, followed by 30 cycles of 30 s at 94 ∘ C, 30 s at 55 ∘ C for primer set (i) and 30 s at 63 ∘ C for primer sets (ii)-(iv), 1.30 min at 72 ∘ C, and finally 15 min incubation at 72 ∘ C. The PCR products were analyzed by electrophoresis (150 V, 400 mA for 1.5 h) in 2% agarose gel prepared in 1xTris-acetate-EDTA (TAE) buffer, stained with ethidium bromide (0.5 g/mL).
For each of the detected genes, one PCR product was sequenced and compared with E. coli DNA sequences on NCBI BLAST (http://blast.ncbi.nlm.nih.gov/) in order to control whether primers bind on correct region in bacterial DNA. Controlled DNA samples were further used as positive controls.

Statistical Analysis.
Fisher's exact test was used to compare the prevalence of the 23 individual VF genes between strains isolated from different countries, materials, and strains belonging to different phylogenetic groups. Comparisons of VF scores were assessed using Mann-Whitney test. VF score was calculated as sum of virulence genes detected, adjusted for multiple detection of the pap, sfa, foc, and kpsMTII operon (papAH, papEF, papC, kpsMT K1, and focG were not taken into account). In order to assess the impact of phylogenetic group and country of origin to the VF score while controlling for the possible contribution from the other variable in the model, the mutually adjusted odds ratios with 95% confidence intervals were calculated using fractional logit models [15]. The models were fitted for (1) all VF genes (adjusted for multiple detection of the pap, sfa, foc, and kpsMTII operons), (2) adhesin genes (adjusted for multiple detection of pap, sfa, and foc operons), (3) capsule synthesis genes (adjusted for multiple detection of kpsMTII operon), (4) toxin genes, and (5) genes coding for siderophores. The group least likely to carry VF gene was chosen as reference group. The criterion for significance was taken as < 0.05.
The different phylogenetic groups exhibited disparate VF scores (mean ± standard deviation): group A 3.9 ± 1.6, group B1 4.8 ± 1.5, group B2 6.0 ± 1.3, and group D 5.3 ± 1.3. Strains  belonging to group B2 were found to carry significantly more VF genes than strains belonging to A, B1, and D ( < 0.001; < 0.001; < 0.001). Strains belonging to group A were found to carry significantly less ( < 0.001) VF genes than strains belonging to group D.
Most of the genes were found to be more prevalent in groups B2 and/or D (Table 3). In group B2, capsular gene kpsMTII, siderophore gene fyuA and pathogenicity island marker PAI were more prevalent and at the same time toxin gene cvaC was less prevalent as compared to the other three groups. P-fimbrial adhesin gene papEF was more prevalent in group D when compared to the other groups.

Comparison of VF Genes in Strains according to Clinical Material.
When comparing E. coli strains according to clinical material, no statistical differences in phylogenetic distribution were found. There were differences in prevalence of 3 VF genes: P-fimbrial adhesin gene papGII was found more frequently in strains isolated from respiratory tract than in strains from urine (34.2% versus 14.3%; = 0.004), and capsular gene kpsMTII was also found more frequently in strains isolated from respiratory tract than in strains from blood (60.5% versus 29.6%; = 0.02). In strains isolated from wound, siderophore gene iutA was found more frequently than in strains isolated from urine and blood (90.2% versus 77.8% and 70.4%; = 0.01 and = 0.02).

Comparison of VF Genes in Strains according to Geographical Origin.
The ESBL-producing E. coli strains in Estonia and Latvia belonged mostly to phylogroup B2 and in Lithuania and Russia to groups B2 and D (Table 4).
There were differences in prevalence of 15 VF genes. Compared to the other 3 countries, P-fimbrial adhesin gene papEF was more prevalent in Russian strains, toxin gene cvcC was more prevalent in Lithuanian strains, capsular gene kpsMTII was more prevalent in Latvian strains, and serum resistance gene traT was less prevalent in Estonian strains (Table 5).

What Affects the Existence of VF Genes: Phylogenetic
Group or Origin of Strains? In fractional logit models, some regional differences remained statistically significant after taking into account the phylogenetic distribution in the countries ( Table 6). The odds ratio (OR) of all appointed VF genes was higher in strains isolated from Latvia (OR 1.1, 95% CI: 1.0-1.2) and the OR of adhesin genes was higher in strains isolated from Russia (OR 1.2, 95% CI: 1.0-1.3) compared to strains isolated from Estonia. The capsule synthesis genes were least represented in Russian strains; strains isolated from Latvia (OR 1.8, 95% CI: 1.4-2.3) and Lithuania (OR 1.4, 95% CI 1.0-1.9) were carrying significantly more capsule synthesis genes compared to Russian strains. Siderophores were more than twice as likely to be represented in Russian (OR 2.7, 95% CI 1.4-5.4) than in Lithuanian strains.
The VF scores were also associated with the phylogenetic group after adjusting for country of origin. Strains belonging to the phylogenetic group A were least likely to carry VF genes (   (OR 1.7; 95% Cl: 1.2-2.5) and siderophores (OR 3.5; 95% Cl: 1.7-7.1) was higher than in group A. In phylogenetic groups B2 and D, siderophore genes were twice as likely (OR 3.5, 95% Cl 1.7-7.1 and OR 2.4, 95% Cl 1.1-5.3) as in groups B1 or A.

Discussion
In this study, we characterized the collection of 423 extraintestinal phenotypically ESBL positive E. coli strains with respect to phylogenetic groups and 23 VF genes. To our knowledge, the present study is the first to assess the phylogenetic distribution and prevalence of VF genes within the E. coli strains isolated from the Baltic Sea region.
ESBL positive E. coli strains belonged mostly to phylogenetic groups B2 and D, which contained more VF genes compared to groups A and B1. E. coli strains isolated from Estonia and Latvia belonged mostly to phylogenetic group B2 and strains isolated from Lithuania and Russia mainly to groups B2 and D. Phylogenetic group and country of origin were associated with prevalence of VF genes, whereas clinical materials from which the strains were isolated were not.
Most of the studied ESBL positive E. coli strains belonged to phylogenetic group B2 (71%) and group D was the second most common phylogenetic group (19%). Our results support some previous results obtained with ESBL positive E. coli strains isolated from different clinical material (blood, urine, wound, and sputum) [16,17] [18,19]. They explain the scarce prevalence of group B2 in their studies with higher antibiotic resistance of strains belonging to other phylogroups than B2. Also, the occurrence of resistance encoding integrons has been found more frequent in E. coli strains of phylogenetic group B2 compared to non-B2 strains [20]. Strains belonging to   phylogroups A, B1, and D express significantly less VF genes and invade more commonly compromised hosts; hence, less VF-s would be required to cause infections in such patients. Antibiotic resistance gives such strains an advantage to cause infections; previous antibiotic treatment was common, which would have been selected for ESBL E. coli [18,19].
The distribution of VF genes in different phylogenetic groups was not even. As reported previously, most of the genes, in which case differences in distribution were observed, were more prevalent in phylogenetic groups B2 and/or D [19,[21][22][23]. E. coli strains belonging to group B2 showed the highest virulence score, which is concordant with previous studies [19,21,23]. In group B2, siderophore gene fyuA, capsular gene kpsMTII, and pathogenicity island marker PAI were more prevalent and colicin gene cvaC was less prevalent than in other phylogenetic groups. Carattoli et al. found pathogenicity island in all ESBL positive E. coli strains, whereas the great majority of the strains belonged to phylogenetic group B2 [24]. In strains belonging to phylogenetic group D, P-fimbrial adhesin gene papEF was more prevalent than in strains from other groups.
We found the virulence gene profile to be significantly associated with the geographical origin of the strain. There were differences in prevalence of 15 out of 23 VF studied genes among E. coli strains isolated from different countries. An explanation for this may be because of the different phylogenetic distribution of E. coli strains originating from different countries in the Baltic Sea region; as in the current study and previous studies by other authors, the distribution of VF genes has been found to differ between phylogroups [16,17,19]. E. coli strains originating from Estonia and Latvia showed a similar distribution to phylogenetic groups: more strains belonged to group B2 and less to groups B1 and D in comparison with Lithuanian and Russian strains. Strains isolated from Lithuania and Russia belonged to groups D and B2.
Some differences in prevalence of VF genes seemed to be explainable by differences among phylogenetic distribution: P-fimbrial adhesin (pap) genes occurred more frequently in Russian strains. The difference in prevalence of papgenes could be explained by differences in the phylogenetic distribution of strains, because half of the Russian E. coli strains belonged to group D and we found that pap-genes are associated with group D. However, after adjusting for phylogenetic distribution, we found that Russian strains showed higher statistical probability of containing adhesin genes, which indicates that there could be also other explanations for higher prevalence of pap-genes in Russian strains. The same could be said about capsular gene kpsMTII, which was found to be associated with group B2 and occurred more frequently in Latvian strains, where prevalence of group B2 was the highest. But again we found that the odds ratio of capsular genes in Latvian strains was the highest, so higher prevalence of kpsMTII might not be explainable by differences in phylogenetic distribution. Although comparing prevalence of siderophore genes fyuA and iutA did not show higher proportion of these genes in Russian strains, we found that when differences in phylogenetic distribution were taken into account, siderophore genes were more than twice as likely to be represented in Russian strains.
To our knowledge, there are only a few studies that compare the prevalence of virulence genes in ExPEC strains isolated from different countries [25,26]. Grude et al. compared E. coli strains that were isolated from patients with bacteriuria from Russia and Norway. They found differences in phylogenetic distribution and virulence gene profile: Russian isolates belonged mainly to phylogenetic group A, while phylogroups B2 and D were predominant among the Norwegian isolates. Norwegian isolates also had a significantly higher number of virulence genes compared to isolates from Russia [25]. Differences observed could be due to geographic and climatic factors, as it has been found that they play an important role in structuring E. coli population worldwide, including also commensal populations [10,27]. As the geographic region observed in the current study is small (Baltic countries and region of St. Petersburg in Russia) and countries involved have a similar climate, then differences found are probably caused by other factors.
Our study revealed no significant relatedness between infection site and phylogenetic distribution or virulence gene profile of studied ESBL producing E. coli strains. This indicates the importance of host factors in the process of infection development. We only found significant differences in prevalence of 3 VF genes (papGII, kpsMTII, and iutA) and no significant differences in phylogenetic distribution. Lee et al. had similar results: they found that distribution of phylogenetic groups was similar between isolates from different clinical materials (blood and urine), but in contrast to the current study, they observed differences in prevalence of other genes (fyuA, traT, and PAI), which were more prevalent in strains isolated from blood than in strains from urine [23]. There is not enough data about VF genes of ExPEC strains isolated from other infections than urinary tract infections, bacteremia, and neonatal meningitis, but some authors have described that strains which colonize respiratory tract and other body locations could be similar to strains isolated from blood and urine. This could explain the scarceness of differences among strains isolated from different clinical materials [28,29].
In conclusion, our study indicates that the prevalence of particular virulence factors in extraintestinal extendedspectrum beta-lactamase producing E. coli strains is associated with phylogenetic group and geographical origin of strains rather than infection site. The regional differences of VF genes in ESBL positive E. coli strains remained statistically significant after taking into account the phylogenetic distribution in the countries.