ABSTRACT
Escherichia coli is overrepresented in all bloodstream infections (BSIs) series, mostly associated with a few clonal lineages. Its population structure has been analyzed but the dynamics remains to be fully understood. We analyze the dynamics of E. coli-BSIs in a sample of 649 isolates, representing all 7165 E. coli BSI episodes recorded in a tertiary hospital (1996-2016) according to clonal identification (phylogenetic groups/subgroups, STc131 subclades), antibiotic susceptibility (13 antibiotics), and virulence-associated genes (VAGs, 29 genes). Patient data were obtained from the laboratory system and clinical charts. The incidence of BSI-EC doubled from 1996 to 2016 (5.5 to 10.8 BSI episodes/1000 hospitalizations). Intertwined waves of community-acquired (CA) and hospital-acquired isolates (HA) episodes of both B2 and non-B2 phylogroups, occurred until B2 overtook non-B2 BSI episodes. ST131 contributed to increasing the B2 rates, but only transiently altered the population structure. B2 isolates predominates (53%), overrepresented by subgroups B2-I (STc131), B2-II, B2-IX, and B2-VI (25%, 25%, 14%, and 9%). We observed a remarkable increase only for B2-I-STc131 (C1/C2 subclades), a decreasing trend for phylogroup D, and oscillations for other B2 subgroups throughout the years. According to VAG patterns, B2 strains exhibit a population structure compatible with the niche specialization theory. A reservoir of B2 and non-B2 strains represented in human microbiota, flows from the community to the hospital and vice-versa, where they can either be selected or coexist. The increase of BSI is determined by waves of CA that predate the amplification of HA episodes of both B2 and non-B2 phylogroups in various time periods, influenced by FQR and microbiota composition.
IMPORTANCE The gut microbiota is an important reservoir for bacteria that cause extraintestinal infections including sepsis, which is the third cause of mortality in Western countries and one of the Global Health threads recognized by the WHO since 2017. Most of the bloodstream infections (BSI) and UTIs due to Escherichia coli strains originate in the gut microbiota and belong to the phylogenetic group B2. Most B2 strains recovered from BSI infections are clonal lineages predominant in fecal isolates, often associated with outbreaks. Our study analyzes the long-term dynamics of B2 E. coli subtypes and reveals waves of different E.coli lineages including pandemic clones that emerge periodically and are established in the intestinal microbiota afterward. It also reflects that clonal amplifications in the community predates the clonal increases in hospitals which may favor the acquisition of antibiotic resistance in health centers and further dissemination of well adapted clones that become multidrug resistant.
INTRODUCTION
Escherichia coli is ubiquitously present in the gut microbiota of humans and other vertebrates, and also represents the primary cause of bloodstream infections (BSIs) and urinary tract infections (UTIs) (1, 2). The increasing and progressive annual rate of BSIs episodes at a global level made the WHO to consider sepsis as a Global Health Threat in 2017 (https://www.global-sepsis-alliance.org/news/2017/5/26/wha-adopts-resolution-on-sepsis). The problem affects more than 30 million of people and increases 9-13% in Western countries where constitutes the third cause of mortality (3–6). The current knowledge suggests differences between countries but such information is highly fractionated in multicentric studies(3), mostly focused on antibiotic resistant BSI isolates (7), and performed at variable periods of time using different sampling criteria (3–8). Only studies from the UK, where surveillance of BSI is compulsory since 2011, provide a long-term comprehensive analysis of BSI and the population structure and the dynamics of E. coli causing BSI (6, 9).
The gut microbiota is often the origin of extraintestinal infections caused by E. coli (10) and vary between humans of different age and lifestyle (10, 11). Among the 7 major phylogenomic groups of the species, the B2 phylogenetic group is currently predominantly detected in clinical and fecal isolates from adults and children of Western areas (10–12). Moreover, the seven E. coli phylogenetic groups (A, B1, B2, C, D, E, F) and also the 10 B2-E. coli subgroups, are consistently overrepresented by certain sequence type complexes (STcs) [e.g.B2-I (STc131), B2-II (STc73), B2-IX (STc95), D (STc69) or F (STc648)], in isolates of different origins although their distribution vary between human populations (13–16).
Despite the apparent persistent structure of E. coli in the gut of Western individuals, clonal expansions of emerging STcs periodically occurred (15, 17). The STc131 represents the most emblematic example of a global emerging clone nowadays (17). A multicentric study in the UK describes how an emerging and pandemic clone STc131 contributes but only transiently alter the persistent population structure of E. coli (9). Nonetheless, the factors behind such E. coli clonal expansions are not well understood and different evolutionary pathways including the acquisition of antibiotic resistance, exist for different STcs (18).
The number of works related to the E. coli population structure contrasts with the scarcity of longitudinal studies necessary for the understanding of long-term population dynamics of the lineages, and the local evolutionarily stable strategies for each lineage that led to local lineage conservation and coexistence (19). Local settings are important sources of information because can reflect the dimensions of human populations structure (age, sex, interconectedness), and offer stability in terms of the intensity of selection (e.g. common policies to control antimicrobial resistance such as antibiotic use, infection control strategies), models of healthcare delivery, and measuring approaches (diagnostic tools) (3, 8).
We retrospectively studied a randomized sample of 649 isolates drawn from a collection of 7165 E. coli isolates, which represented all BSI episodes registered at our institution between 1996 and 2016. This period coincided with the global emergence and amplification of various blaESBLs genes and B2 E. coli clones and with the overall increase in the frequency of E. coli BSIs. The aim of this study was to infer the local diversity and dynamics of E. coli causing BSIs, focusing on the B2 phylogroup and the most abundant contemporary B2 subtype, the pandemic B2-I STc131.
METHODS
Study design
Ramón y Cajal University Hospital is a tertiary-level public health center with 1155 beds that provides attention to 600,000 habitants in the Northern area of Madrid (Spain), which reflects a pyramid-age of “declining type”, has full accessibility to primary attention care and have a predominant medium-high socioeconomic level. Of the total 21,695 positive blood cultures detected between January 1996 and December 2016, we identified 7165 E. coli that represented 1 isolate per patient and per BSI episode. A sample of nearly 10% of this E. coli collection, stratified by sex, age, and antimicrobial resistance pattern was sorted by statistical randomization and was further analyzed (649 E. coli isolates from 339 females and 310 males; < 1-98 years of age). The study was approved by the Ethics Committee of our hospital.
BSIs are classified as hospital acquired (HA), community acquired (CA), and community-onset healthcare-associated (HCA), according to the date of the sample collection after patient admission and the patient exposure to hospitals before the BSI episode (8). Due to the inaccessibility to all the medical records of patients enrolled in this study, and their advanced age, we classified the episodes into HA (if the blood culture was obtained at the ICU, surgical, or medical areas after 48 h of admission) and CA+HCA (if the blood culture was obtained at the hospital emergency wards or at the day care centers) categories. UTIs were considered as the origin of BSI if E. coli was recovered from both the urine and blood samples, with a difference of ±24 h.
Characterization of the bacterial isolates
Blood culture isolates of E. coli are routinely frozen and stocked in skimmed milk at −70°C, and were subcultured onto brain-heart infusion agar prior to analysis. Bacterial susceptibility against 13 antibiotics (ampicillin, amoxicillin-clavulanic acid, cefotaxime, ceftazidime, meropenem, nalidixic acid, ciprofloxacin, streptomycin, kanamycin, gentamicin, tetracycline, chloramphenicol, and trimethoprim/sulfamethoxazole) was performed by the disk-diffusion method (20).
Multiplex PCR assays allowed classifying the E. coli isolates into major phylogenetic groups A, B1, B2, C, D, E, and F (21); B2 subgroups (B2 I-X) (22); and B2-I-STc131 E. coli serotypes (O16/O25b), clades (H41, H22, H30), and the H30 subclades C0 [H30, fluoroquinolone susceptible (FQS)], C1 [H30-R, fluoroquinolone resistant (FQR)], C2 (H30Rx, FQR+blaCTX-M-15), and C1-M27 (H30-Rx, FQR+ blaCTX-M-27) (22, 23). STc131 isolates were further analyzed by pulsed field gel electrophoresis (PFGE). The presence of 29 virulence-associated genes (VAGs) was determined for all B2 isolates by PCR (2, 24).
Statistical analysis
To calculate statistical significance, the chi-squared test, a 2-sample t-test for normally distributed variables, and Kendall’s correlation were used, considering p-values <0.05 to be statistically significant.
RESULTS
Epidemiology of the 7165 E. coli isolates causing BSIs at Hospital Ramón y Cajal
The incidence of BSIs caused by E. coli doubled from 1996 to 2016 in our institution (5.5 to 10.8 BSI episodes/1000 hospitalizations). Although the overall number of BSI episodes in the hospital and community settings was similar at the beginning of the study in the mid 1990s, we observed a steady increase in both HA-BSIs and CA+HCA-BSIs from 1995 to 2002 (CA/HA ratio >1–2) followed by waves of alternative predominance of either CA-BSIs or HA-BSIs. The increases of BSI episodes in the community predated those in the hospital setting and would explain the wave dynamics between the hospital and the community-based populations and the overall shift in the ratio of BSI acquired in the community and the hospital (Figure 1). The analysis of antimicrobial resistance records in our department for the blood E. coli isolates revealed a coincidental increase in the rise of BSI infections and the increasing trends of E. coli resistant to fluoroquinolones (FQR) from 1994, and resistant to third-generation cephalosporins (3GC) from 2003. Rates of resistance to other antibiotics remained stable during the period of study (data not shown).
UTIs were identified as the origin of the BSI in one-third of the cases. This finding was more frequent in women than in men (42.33% vs. 26.77%, p≤.005) but not significantly different between patients of different age groups (p >.05). The proportion of polymicrobial BSIs was 9% (n = 57/649), and was similar for men and women (11.8% vs. 7.3% of the BSI cases, respectively).
Clonal diversity of E. coli causing BSI
Figure 2 shows the occurrence of various E. coli populations, namely phylogenetic groups, B2 subgroups, and B2-I (STc131) subclades. The predominant phylogenetic group was B2 (348/649; 53.06%), followed at a much lower frequency by D (11,4%), B1 (7,24%), A (6.3%), C (5.,1%), F (6.8%), E (0.9%) and Clades I and II (0.9%). Half of the B2 isolates corresponded to subgroups B2-I (25.6%) and B2-II (25.1%), which were followed by B2-IX (14.2%), B2-VI (9.5%) and others. Almost 10% of the isolates were catalogued as The STc131 isolates, clearly predominant within the B2-I subgroup, represents 12.3% of the total BSI E. coli isolates and 21.8% of the B2 phylogroup. The STc131 isolates (serogroups O25b and O16 representing 92% and 8%, respectively) split in clade A (7/82, 9%), clade B (16/82, 20%), and clade C (59/82, 71%). Clade C comprised isolates of subclades C1 (29/82, 35.4%), C2 (22/82, 26.8%), C0 (6/82, 7.3%), and C1-M27 (2/82, 2.4%).
Distribution by Age
All phylogenetic groups were present in patients of all age groups, although differences were observed for subgroups B2-I (STc131), B2-II, B2-VI, and B2-IX. This distribution implies the clear predominance of B2-I in the elderly (>80 years) and B2-II in individuals younger than 45 years. Others showed a bimodal distribution as B2-IX and B2-VI (Figure S1).
Acquisition of the BSIs
Most B2 isolates were community based although intertwined waves of CA+HCA-BSI and HA-BSI episodes of strains of both B2 and non-B2 phylogroups, which occurred synchronously until 2008, when B2 clearly overtook non-B2 BSI episodes and STc131 become transiently predominant (Figure S2). Almost half of patients with STc131 (46%) were admitted to the medical emergency ward with an established BSI, suggesting the frequent presence of this clone in the community. As an example, we detected among these patients several coincidences of strains with the same PFGE pattern corresponding to clades H22 (2000), C1 (2004, 2005), and C2 (2008-2011) (data not shown).
Temporal variation
Except for phylogroups B2-I (ST131) and phylogroup D, the trends of the phylogroups did not significantly change during the period analyzed. However, the occurrence of major subgroups (B2-II, B2-IX, B2-VI, B-IV, B2-VII) greatly varied through the years suggesting episodes of transmission with transient amplifications as occurred for B2-I (STc131) (Figure 3).
The STc131 E. coli was the only group increasingly recovered coinciding with its global amplification (17). The strains of ST131 clade B were initially detected in 1996, and remained steadily identified since then. In the current study, the ST131 of clade A was first detected in 2004, but we retrospectively identified STc131 clade A in clinical isolates of TEM-4 and TEM-24 producers from 1991 and 2000, respectively (25, 26) For the predominant ST131 clade C, the subclade C0 (H30, FQS) emerged in 2000, followed by C1 (H30-R, FQR) in 2004, C2 (H30-Rx, FQR blaCTX-M-15) in 2006, and C1-M27 (H30-Rx, FQR blaCTX-M-27) in 2016 (Figure 4). The number of E.coli strains of phylogroup D decreased from the beginning of the study, but it varied significantly through the years.
Antimicrobial susceptibility
Figure 5 shows the antimicrobial susceptibility patterns in the sample of 649 E. coli isolates, revealing that FQR appears in between 35% and 50% of the isolates of each phylogroup, with B2 being less prevalent (17%), although it is of note that most FQR B2 isolates correspond to STc131 (70% vs. 3.5% in non-ST131 B2). Only 6.1% of the total number of B2 isolates showed a 3GCR phenotype compatible with the production of an extended-spectrum beta-lactamase (ESBL), further identified as CTX-M-15, CTX-M-14, and CTX-M-27, and was only detected among STc131 isolates. Resistance to ampicillin (70.4%), streptomycin (39.6%), nalidixic acid (32.8%), and cotrimoxazole (45%–50%) was frequent among the isolates of the various phylogroups. Remarkable differences were observed for amoxicillin-clavulanic acid, kanamycin, gentamicin, tetracycline, and chloramphenicol, mostly due to the phylogroup C isolates. Susceptibility to the 13 antimicrobials tested was observed in 17% of the total number of isolates tested, with the frequency of susceptible isolates higher among those of phylogroups A, B1, B2, and F (15% each) than those of phylogroups D (10%), E (1%), and C (0%). Within B2, a pan-susceptibility pattern was more frequent for non-STc131 than for STc131 isolates (26.8% vs. 3.7%). The B2-STc131 strains showed a multiresistant profile (resistance against ≥4 antimicrobial agents) more frequently than other B2 members (56.6% vs. 11.6%; p <.001).
Virulence-associated gene profiling in the B2 phylogroup
According to the content of the VAG genes, the B2 E. coli strains were clustered into 2 large groups and 10 subgroups, each comprising numerous gene combinations and showing variable redundancies (Figure 6). Despite some VAG variability within each B2 subgroup, the results suggest a conserved genetic structure related to virulence and colonization in the B2 strains analyzed in this study. More than half (56%) of the B2-I (STc131) strains clustered in VAG 1.5. Most (54%–75%) of the E. coli phylogenetic subgroup isolates B2-II, B2-III, B2-V, B2-VII, B2-IX, and B2-X clustered in the VAG 2.6, 2.2, 2.9, 2.6, 2.7, and 2.4 groups, respectively (ordered by frequency). A detailed analysis of virulence content is provided in Supplementary Text and Figures S3, S4, S5, S6.
DISCUSSION
This work, one of the very few studies -and the first in Spain- to analyse the dynamics of E. coli strains for a long period of time at a single centre, reveals intertwined waves of CA+HCA-BSI and HA-BSI episodes of strains of both B2 and non-B2 phylogroups, which occurred synchronously until B2 overtook non-B2 BSI episodes and the B2-I-STc131 became transiently predominant and predate new waves of other major B2 subgroups. This dynamics is similar to that observed in the UK (18) but further suggest that the burden of CA-BSI and HA-BSI infections are closely related. Besides the risk of cross-infections/colonizations in the hospital that led to local increases in BSIs in particular periods, species or clones which are part of the microbiota of healthy individuals in the community might occasionally have an epidemic structure, in our case providing a source of E. coli strains influencing the incidence of hospital BSIs.
E. coli bacteremia, which was not considered common at the beginning of the 20th century, has steadily increased from the past to the present, according to long surveys performed at Boston City Hospital between 1935 and 1972 (27), at St. Thomas hospital in London between 1969 and 1986 (28), and more recently, at our and other hospitals in Western countries (most cross-sectional or longitudinal multicentric studies) (3, 6, 8, 9, 29). The local survey performed during the 1980s (28) reflected, for the very first time, a relatively constant proportion of E. coli among all cases of BSI with where the incidence peak was coincidental with a community-based clonal outbreak which led to the introduction of an E. coli O15:H12 clone in the hospital setting and a subsequent increase in nosocomial BSI cases (28). Shortly afterwards, community-based and hospital clonal outbreaks by E. coli of phylogroup D were extensively reported during the 1990s, namely ST69 in the US and ST393 O15:H12 in the UK and Spain (15). During the 2000s, the increasing of CA outbreaks of B2 E. coli lineages as STc73, STc95 and more recently, STc131, became common (9, 14). Massot et al demonstrated that the shift in the population structure of commensal E. coli from A and D groups in 1970s to B2 and F in 2000s could be influenced by human interventions as antibiotic use (12). However, clonal expansions at local level have been poorly analyzed and always under the perspective of antimicrobial resistance. Significant changes in healthcare delivery and in the human population structure (age, interconnectedness, diet), have been recently suggested to explain both the transmission (amplification) of microorganisms and the acquisition of resistance even in the absence of personal history of antibiotic consumption and other known risk factors for antimicrobial resistance (30–32).
The most comprehensive recent analysis, using European Antimicrobial Resistance Surveillance System (EARSS) data from 2002 to 2008, highlighted a remarkable average annual increase of 29.9% in the number of reported bacteremias caused by E. coli isolates resistant to 3GCs. However, such increases were less noticeable in countries from Southern Europe (3), probably because of the overrepresentation of major E. coli in hospitals predates the acquisition of genes coding for resistance to first-line beta-lactam antibiotics (9, 13). What we should be aware of the steady increase in the occurrence of FQR in all non-B2 groups since the early 1990s that would reflect the selection, amplification and stabilization of certain E. coli lineages as some of the phylogroups D, and more recently, of STc131 of phylogroup B2. These lineages would have also been circulating in the community for decades before the emergence of STc131 clade C, a FQR lineage that, in contrast with others, is predominant in individuals >80y. FQR has recently been considered an advantage for the fitness of high-risk clones of E. coli and other species (33, 34). Some FQR clones are predominant in the elderly, with a sustained pattern in hospital (or day-care-center) admissions and personal clinical history that facilitates the acquisition of other antibiotic resistances (35). An increasing cumulative “elderly microbiota reservoir” of antimicrobial-resistant subpopulations of various bacterial species would be congruent with the overrepresentation of community-based individuals with a BSI by STc131 at admission, with the age of the patients in our sample, and with the identification of identical PFGE types in groups of unrelated patients. However, this does not fully explain the overrepresentation of certain B2 lineages which are amplified time to time as it reflects the waves of different B2 subgroups in Figure 3.
B2-I (STc131) was the only lineage that significantly increased during the period analyzed, and this transient increase was due to the expansion of both C1 and C2 subclades, and probably, to CTX-M-27 in the last years not included in the study according to the emergence of this subclade in the community (36). Although the epidemiology of ST131 has extensively been documented (17), this work and that by Kallonen (9) are the only ones that reveals the evolution and coexistence of all STc131 subclades which, after a transient amplification, are incorporated into the E. coli population structure (17, 36). The other predominant B2 subgroups varied greatly through the period studied, suggesting some peaks of prevalent B2-II, B2-IX, B2-VI clones and also others as B2-VII and B2-X and non-I/II which are poorly documented in the literature. The correlation between VAG profiles and B2 subgroups reflects an apparent structure of B2 populations in agreement with the niche specialization hypothesis (10, 37–39). Although the concordance of specific VAGs with isolates of B2-II and B2-IX subgroups is congruent with their categorization as uropathogenic E. coli, such a way explaining their bimodal distribution in different age groups, most B2-I strains (STc131 clade C) exhibit a low number of classical EC-VAGs while retaining some traits that would facilitate colonization and persistence in the gut (40, 41). To explain the expansion of the STc131 and other emerging lineages, such as STc648 (phylogroup F), studies that have applied mathematical models to a high number of genomes suggest that negative frequency-dependent selection of previously rare populations might have favored their increase (16, 41), and will foster the expansion of some others in the future. However, these studies are fully focused on the microorganisms and do not allow us to associate the observed changes with the type of hosts located either in hospital or community compartments.
We are aware that the size of the sample analyzed (despite of the high number of isolates analyzed it represents the 10% of the total number of BSI in our institution), and the lack of whole genome comparative analysis might limit our conclusions about the clonal dynamics. However, the long period analyzed at a single institution helps identifying relevant epidemiological conditions that explain the expansion or particular E. coli populations observed in our and other areas during recent years. Currently, the B2 E. coli strains are relevant units of BSI pathogenicity, which should correlate with their success in particular microecological landscapes, in part determined by interventions exerted on particular human populations (particularly elderly patients). This view is in agreement with the concept of incorporating ecological features in the identification of the fundamental units of bacterial diversity (38) and pathogenicity. In fact, particular clones and lineages are differentially represented in a “human microbiota reservoir” (elderly being a reservoir of high-risk B2 lineages) flowing from the community to the hospital and vice versa, where they can either be selected or coexist as predicted by an evolutionarily stable strategy (38). The evolutionary trajectories of recently amplified major lineages indicate the relevance of undetected selective events, which could be further amplified by the acquisition of antimicrobial resistance in settings under (or not) antibiotic pressure (31, 32, 42). The early detection of the abundance and diversity of B2 subgroups of clinical significance for BSI in metagenomic samples and the analysis of common causes that enhance their selection, either in the hospital, nursing homes or the community, are priority research challenges that warrant attention in an era that favor the pandemics of microorganisms and antimicrobial resistance (43).
FUNDING
This work was supported by the European Commission (grants ST131TS AC00043/17 and EVOTAR-FP7-HEALTH-282004) and the Instituto de Salud Carlos III of Spain, co-financed by the European Development Regional Fund (A Way to Achieve Europe program; Spanish Network for Research in Infectious Diseases grant RD12/0015/0004 and RD16/0016/0011, an CIBER CB06/02/0053). During the performance of this study, I.R. was a recipient of a postdoctoral contract “Sara Borrell” (Ref. CD12/00492) from the Instituto de Salud Carlos III (Spain), MS and ASF were recipients of an ErasmusPlus fellowship, and PM was funded by the Programa Operativo de Empleo Juvenil (PEJ-2017-AI/BMD-7200).
CONFLICTS OF INTEREST
None.
AUTHORS’ CONTRIBUTIONS
I.R. performed the wetlab work of the isolates collected from 1996–2012, assisted by C.R., and participated in the study design, in the data analysis, and the writing of the initial draft of the manuscript. A.S.F and M.S., assisted by PM, performed the wetlab work of the isolates collected from 2012–2016, the categorization of all ST131s included in the study, the data analysis, and the revision of the manuscript. V.F.L performed the bioinformatic and statistical analysis of the results, most of the figures, and the revision of the manuscript. J.Z, E.L., and R.C. participated in the study design and the revision of the manuscript. C.J.B. provided expertise and participated in the revision of the manuscript. F.B. and T.M.C. participated in the study design, data analysis, and wrote the manuscript. All the authors have read and approved the final version of this manuscript.
Supplementary Text
Virulence-associated gene profiling in the B2 phylogroup of Escherichia coli
The presence of 29 virulence-associated genes (VAGs) was determined for all B2 isolates by PCR [1,2]. The B2 E. coli strains were clustered into 2 large groups and 10 subgroups on the basis of the content of the VAG genes. Each group comprises numerous gene combinations and showed variable redundancies. Most E. coli isolates analyzed here harbored fyuA, an outer membrane protein acting as a receptor for ferric-yersiniabactin (fyu) and a marker of PAIJ86IV (a “high pathogenicity island”) [3,4]; malX, a phosphotransferase system enzyme and marker of PAICFT073 [5]; ompT (an extracellular protease active against host cationic peptides located on E. coli plasmids or chromosomes [6,7]; and traT, only located on F plasmids, which act as a surface exclusion protein, conferring serum resistance. [8]. However, the isolates differ in the presence of type P fimbriae, all of which contain various combinations of siderophores, toxins, and types S and I fimbriae, which are usually located on emblematic E. coli pathogenicity islands (PAIs). Despite some VAG variability within each B2 subgroup, more than half (56%) of the B2-I (STc131) strains clustered in VAG 1.5. Most (54%–75%) of the E. coli phylogenetic subgroup isolates B2-II, B2-III, B2-V, B2-VII, B2-IX, and B2-X clustered in the VAG 2.6, 2.2, 2.9, 2.6, 2.7, and 2.4 groups, respectively (ordered by frequency) (Figures 6, S3, S4, S5, S6).
The VAG 1.5 group only comprised isolates of the B2-I (STc131) and B2-VII subgroups (representing 54% and 33% of the isolates of these B2 subgroups, respectively). They contained the genes iha, sat, iutA, kpsMT, and sporadically hlyA, all frequently located on PAICFT073-pheV, and to a lesser extent, afa/draBC (25%) or ibeA. Some B2-I (STc131) isolates were also associated with VAG 1.8 (lacking traits of PAICFT073-pheV but containing ibeA and often iroN), VAG 2.4 (containing iroN, ibeA, and papABCED), and VAG2.9 (papABCED). The analysis of STc131 isolates in Figure S6 reflects the diversity of VAGs within the ST131 groups H30 (or subclade C), H22 (or subclade B), and H41 (or subclade A); many of the isolates exhibiting VAGs could not be associated with any of the virotypes proposed by Dahbi et al. [9]
The VAG 2.7 was predominant in isolates of the B2-IX group (63%, p <.005) and was only associated with this B2 subgroup. This VAG profile was characterized by papACEF genes, ireA (encoding a protein involved in iron acquisition, adhesion to epithelial cells, resistance to stress, and a high pH, a trait frequently associated with isolates causing UTIs and BSIs), and a variable presence of genes in PAIU189 (iss, cvaC, and K1-capsule).
The VAG 2.6 profile comprised more than half of B2-II and B2-VII isolates and was linked to the presence of type P fimbriae and with various traits associated with PAIJ96 (hlyA, cnf1, sfa, ΔpapGIII), PAICFT073-serX (sfa/focDE, iroN), and PAICFT073-pheV (iutA, iha, iroN, sat, kpsMTK5).
The VAG 2.9 group was mostly associated with B2-V (50% of the isolates), and differs from VAG 2.6 in the absence of iroN and K5. VAG 2.2 comprises most B2-III (73 of the isolates) and VAG 2.4, most B2-X isolates. With the exception of B2-I and B2-IV, genes of most strains of B2-II, B2-IX, and B2-IV have traditionally been associated with UTIs or BSIs.