Antibiotic Resistance and Virulence of Extraintestinal Pathogenic Escherichia coli (ExPEC) Vary According to Molecular Types

Extraintestinal pathogenic Escherichia coli (ExPEC) can cause many human extraintestinal infections. Resistance and virulence of ExPEC are inextricably linked to its phylogenetic background. However, studies on type-specific distribution of resistance and virulence and the connection between resistance/virulence and molecular typing are lacking. Here, 411 ExPEC strains were collected and characterized using antimicrobial susceptibility testing and molecular typing. Among these, 74 representative strains were selected for whole genome sequencing and the Galleria mellonella killing assay. CH40-30-ST131, CH37-27-ST405, CH40-41-ST131, and CH13-5-ST12 isolates had high resistance rates to all antimicrobials tested. BlaCTX–M played a significant role in the β-lactam resistance of ExPEC isolates. CH14-64-ST1193, CH40-30-ST131, and CH35-27-ST69 isolates were highly virulent in the G. mellonella model. Virulence factors (VFs) involved in adherence (papB, papI, papX, and fimA), autotransporter (sat), invasion (aslA, kpsD), iron uptake (except for entD), or toxin (senB) might be responsible for pathogenicity in vivo. Specific antibiotic resistance genes (ARGs) or VFs were prevalent in specific types of strains, including papB, papI, fimA, sat, kpsD, senB, and aerobactin genes in CH14-64-ST1193 isolates; blaCTX–M–15, aac(6′)-Ib-cr, papB, papI, sat, iucA, iucB, iucC, chuT, chuX, and shuU in CH40-30-ST131 isolates; tetB in CH35-27-ST69 and CH13-5-ST12 isolates. Type distribution also differed by VF score. CH37-27-ST405 and CH26-5-ST38 isolates carried more ARGs and VFs indicating that they had a high resistance and virulence potential. This study demonstrates the type-specific distribution of resistance and virulence thus providing a basis for further research, prevention and treatment of ExPEC infections.


INTRODUCTION
Extraintestinal pathogenic Escherichia coli (ExPEC) is one of the most common bacterial pathogens isolated from clinical specimens. It causes urinary tract infection, bacteremia, neonatal meningitis, and other infections at non-intestinal sites (Biran and Ron, 2018). Extensive use of antibiotics has led to a significant increase in antimicrobial resistance (especially against thirdgeneration cephalosporins and fluoroquinolones) in ExPEC isolates severely affecting its effective treatment (WHO, 2019). ExPEC strains have a great impact on public health and economic burden due to the high percentage of ExPEC infections and its antimicrobial resistance.
Extraintestinal pathogenic Escherichia coli as an opportunistic pathogen, often causes infection in immunocompromised individuals, which relies primarily on antimicrobials as a first line of defense against infection. Virulence and antimicrobial resistance in ExPEC play an important role in the infection process. Resistance and virulence are inextricably linked to the phylogenetic background of the strain. Most ExPEC isolates belong to phylogenetic group B2 and to a lesser extent to group D, whereas most commensal isolates belong to groups A and B1 (Pitout, 2012). Isolates from group B2 demonstrated a high frequency of virulence factors (VFs) while group A strains displayed a low prevalence of VFs (Salipante et al., 2015). Up to 30% of all ExPEC, 60-90% of fluoroquinoloneresistant ExPEC, and 40-80% of ExPEC producing extendedspectrum β-lactamase (ESBL) belong to ST131 (Pitout and DeVinney, 2017). Within the ST131 isolates bla CTX−M−15 and aac(6 )-Ib-cr were more prevalent in the H30-Rx sublineage (Peirano and Pitout, 2014). These 12 VFs (iha, fimH, sat, astA, fyuA, iutA, kpsM II-K2, kpsM II-K5, usp, traT, ompT, and malX) were significantly more prevalent in ST131 than in non-ST131 isolates (Johnson et al., 2010). With the exception of ST131 isolates, there are less studies on the characterization of antibiotic resistance and virulence of ExPEC belonging to other molecular types. Multilocus sequence typing (MLST) is the most common method of characterizing ExPEC-associated clonal lineages based on the sequencing of seven housekeeping genes (Manges et al., 2019). High-resolution CH (fumC/fimH) typing was also developed for identifying the population structure of ExPEC based on the sequencing of fumC and fimH (Weissman et al., 2012). And specific CH types correspond to specific STs with 95% accuracy (Roer et al., 2018). Although molecular typing, antibiotic resistance, and virulence of ExPEC have been detected and analyzed in some studies (Matsumura et al., 2017;Nüesch-Inderbinen et al., 2017;Bozcal et al., 2018;La Combe et al., 2019), the complete analysis of antibiotic resistance genes (ARGs) and VFs using genome sequencing is limited, and the comprehensive analysis and connection between antibiotic resistance/virulence and molecular typing are also limited or lacking. In this study, we conducted a systematic study of ExPEC to investigate the specific distribution of antibiotic resistance and virulence with regards to molecular types. We performed antimicrobial susceptibility testing, CH or MLST, screening for acquired ARGs and VFs using genome sequencing, and Galleria mellonella killing assay on ExPEC isolates. The study shows that antibiotic resistance and virulence of ExPEC strains differ according to molecular types. It provides a basis for further research targeting prevention and treatment.

Bacterial Isolates and Antimicrobial Susceptibility Testing
A total of 411 non-duplicate ExPEC isolates were obtained from clinical samples (Supplementary Table S5) of patients in the Second Hospital of Tianjin Medical University (Tianjin, China) in two successive periods (2011-2012 and 2018). Antimicrobial susceptibility testing and ESBL detection were evaluated using the Vitek 2 automated system (Sysmex-bioMérieux, Marcy I'Etoile, France). The tested antimicrobials, which are commonly used in clinical practice, included amikacin, gentamicin, amoxillin/clavulanic acid, ampicillin, ampicillin/sulbactam, piperacillin, piperacillin/tazobactam, aztreonam, cefatriaxone, cefazolin, cefepime, cefoperazone/sulbactam, cefotaxime, cefoxitin, ceftazidime, cefuroxime, ertapenem, imipenem, meropenem, trimethoprim/sulfamethoxazole, nitrofurantoin, ciprofloxacin, levofloxacin, minocycline, tetracycline, and tigecycline. E. coli ATCC 25922 was used as a control strain. Interpretation of the results was performed using the guidelines from the Clinical & Laboratory Standards Institute (CLSI, 2019). Intermediate and resistant results were analyzed as resistant. Multidrug-resistant (MDR) isolates were those resistant to ≥3 of the following drug classes: aminoglycosides, beta-lactams, trimethoprim/sulfamethoxazole, nitrofurantoin, fluoroquinolones, and tetracyclines. The resistance score was defined using the number of resistant antimicrobial agents for each isolate. A score of 1 indicates resistant; 0.5, intermediary; and 0, sensitive.

Detection of Clonal Groups
Clonal groups were defined using CH (fumC/fimH) typing as described by Weissman et al. (2012). For fimH-null isolates (isolates without fimH), MLST was performed according to the protocol described by Wirth et al. (2006). Briefly, overnight cultured ExPEC isolates were heated to 100 • C for 10 min for crude DNA extraction. The two (fumC and fimH) or seven (adk, fumC, gyrB, icd, mdh, purA, and recA) housekeeping genes were amplified by PCR. The amplified PCR products were then sequenced, and the results were analyzed using a Web tool for CH typing 1 (CHTyper 1.0) or sequence type (ST) (MLST 2.0) 2 .

Whole-Genome Sequencing and Sequence Analysis
Apart from 7, 2, and 19 isolates respectively in CH11-54, CH26-65, and "Others" group, about one-quarter of the isolates in each group were also sampled equidistantly between the lowest and highest resistance rates in each group (Supplementary Table S1). A total of 74 representative isolates were selected for whole-genome sequencing. They were subjected to DNA extraction and purification using a Bacteria genomic DNA kit (CWbiotech, Beijing, China) following the manufacturer's instructions. A total amount of 1 µg DNA was used for the generation of sequencing libraries using NEBNext Ultra DNA Library Prep Kit for Illumina (NEB, United States) following manufacturer's recommendations. The library was then sent to Novogene (Tianjin, China) for genome sequencing on an Illumina HiSeq Xten platform using a 150 bp paired-end strategy. Acquired ARGs carried by the isolates were analyzed using KmerResistance v 2.0 (Clausen et al., 2016) with raw reads. KMA v1.2.0 (Clausen et al., 2018) was used to detect VFs by aligning raw reads directly against the Virulence Factors of Pathogenic Bacteria Database (VFDB; "core dataset" downloaded on 21 June 2020) (Liu et al., 2019). Results were reported for ≥90% nucleotide identity, ≥90% coverage of the query, and a sequence depth of ≥10×. ARG and VF scores were defined as the number of unique ARG and VF detected for each isolate, respectively. STs of the strains were analyzed from raw reads using MLST v 2.0 (Larsen et al., 2012), with results reported for thresholds of 100% nucleotide identity, 100% coverage of the query, and a sequence depth of >50×. Raw reads were assembled into scaffolds using SPAdes v 3.13.1 (Nurk et al., 2013). The major ExPEC phylogenetic group (A, B1, B2, C, D, E, and F) was determined by analyzing the assembled scaffolds using the online service ClermonTyping 3 . The sequence data were deposited in NCBI Sequence Read Archive under the accession number SAMN14395432-SAMN14395505.

Infection of G. mellonella Larvae
The 74 representative strains (Supplementary Table S1) were also used for the G. mellonella killing assay. Overnight cultures of ExPEC isolates were washed using phosphate-buffered saline (PBS, pH 7.0) and further adjusted with PBS to concentrations of 1 × 10 7 colony-forming units (CFU)/mL. For each strain, 10 larvae were injected respectively with 10 µL of bacterial suspension (Alghoribi et al., 2014). In addition, there were three different control groups (10 larvae per group). The first control group was injected with E. coli DH5α to evaluate effects of nonpathogenic bacterial infection, the second group was injected with PBS to measure effects of physical injury, and the third group was not manipulated. Larvae were then incubated at 37 • C in the dark and monitored for an additional 96 h. We recorded the number of larvae killed. All the experiments were replicated three times.

Statistical Analysis
Comparisons between groups were analyzed using Kruskal-Wallis test or Mann-Whitney U-test for continuous variables and chi-square or Fisher's exact test for categorical variables. A two-sided P value <0.05 was considered statistically significant. 3 http://clermontyping.iame-research.center/ All statistical analyses were performed using IBM SPSS statistics version 25.0 (IBM Corp., Armonk, NY, United States).
Most of the 411 isolates were from urine (63.5%), sputum (15.6%), and blood (10%) samples (Supplementary Table S5). Compared with non-CH14-64 isolates, more CH14-64 isolates were from urine, which suggested that this type strain was a major contributor to urinary tract infection. However, fewer CH40-30 isolates were from urine compared with non-CH40-30 isolates. Interestingly, 9 out of 10 CH13-5 isolates were from sputum samples of infants with bronchopneumonia (2011.12-2012.04), which might be strongly implicated in the prevalence of pediatric bronchopneumonia during the winter and spring. Type distribution differed by sources of ExPEC.
e Resistance score was defined as the sum of inactive in vitro antimicrobial agents for each isolate. A score of 1 was attributed for a resistant, 0.5 for an intermediary, and 0 for a sensitive isolate; a higher score thus indicated a more resistant isolate.
The difference between each of all of the detected VFs (Supplementary Table S3) and corresponding numbers of larvae killed was analyzed using Mann-Whitney U-test to further explore key VFs in the infection model ( Table 2). ExPEC isolates carrying VFs involved in adherence (papB, papI, papX, and fimA), autotransporter (sat), invasion (aslA and kpsD), iron uptake (as in Table 2, except for entD), or toxin (senB) killed more larvae than corresponding VF negative strains (P < 0.05). The abovementioned results indicated that these VFs played an important role in the model. Unusually, ExPEC isolates carrying VFs involved in adherence (afaA, afaB-I, afaC-I, draP, and daaF), iron FIGURE 1 | The number of larvae killed in a G. mellonella infection model according to CH or ST types among the 74 ExPEC isolates. Lethality in larvae (10 larvae/E. coli strain) was observed for 4 days after infection with test strains. Each dot corresponds to one test strain, its corresponding abscissa represents the average number of three independent experiments for the strain, and its corresponding ordinate represents the CH or ST types of the strain. Non-infected larvae and PBS challenged larvae showed no mortality. Low larvae mortality (≤20%) was recorded following challenge with E. coli DH5α. The Mann-Whitney U test showed the number of larvae killed in CH14-64, CH40-30, and CH35-27 groups were not statistically different, but higher than that in any of the remaining groups (P < 0.05). uptake (only entD), or secretion system (as in Table 2) killed less larvae than corresponding VF negative strains (P < 0.05). The results of VFs involved in adherence (afaA, afaB-I, afaC-I, draP, and daaF) might not be reliable because they only had 2-3 (2.7-4.1%) positive strains ( Table 2). Each of the 74 ExPEC isolates was found to carry 39-121 VFs (Supplementary Table S3) and infection was not linked to a single VF but rather to multiple traits (Johnson et al., 2006;Ciesielczuk et al., 2015;Johnson et al., 2019a). VFs involved in iron uptake (only entD) or secretion system (as in Table 2) might play a very weak role in the model.
Among these types, CH14-64-ST1193, CH40-30-ST131, and CH35-27-ST69 isolates showed high virulence in the infection model. ST69 isolates were reported to show higher virulence than ST95 and ST131 isolates (Alghoribi et al., 2014) while our research specifically showed that ST131-H30 (H30 subclone of sequence type 131) and ST69 isolates (both having equal virulence) had higher virulence than ST131-H41 and ST95 isolates. Compared with PCR surveillance of 29-55 VFs in previous similar studies (Johnson et al., 2006;Alghoribi et al., 2014;Merino et al., 2020), we detected the known VFs in VFDB to investigate VFs responsible for pathogenicity in vivo. Consistent with the literature (Johnson et al., 2006), pap FIGURE 2 | ARG (A) and VF (B) scores according to CH or ST types among the 74 ExPEC isolates. ARG and VF scores were defined as the number of unique ARG and VF detected for each isolate, respectively. ARG scores in CH37-27 isolates were significantly higher than in CH14-64 or CH13-5 isolates while ARG scores in CH26-5 isolates were significantly higher than in CH40-30 or CH13-5 isolates (P < 0.05). VF scores in CH13-5 isolates were significantly higher than in any of these 12 groups (except CH26-65) while VF scores in CH11-54 isolates were significantly lower than in any of these 12 groups (except ST648 and Others) (P < 0.05).
(P fimbriae elements) and fyuA were closely associated with the virulence of ExPEC. Other key VFs ( Table 2) associated with the virulence of ExPEC in this study were different from reports in previous studies (Johnson et al., 2006;Ciesielczuk et al., 2015;Johnson et al., 2019a). The difference is more likely to come from the differences of infection models and detected VFs between this study and these previous investigations. Apart from sat and senB which were frequently detected in ExPEC, a large number of VFs involved in iron uptake (as in Table 2, except for entD) were identified as playing a role in the experimental virulence of ExPEC. These VFs might play a more important role in pathogenicity in vivo. Besides, a large number of VFs involved in secretion system were detected and supposed to be negatively associated with the virulence of ExPEC. Some (espL, espR, espX, and espY) of these VFs encoded type III secretion system (T3SS) effectors in enterohemorrhagic E. coli (EHEC) O157:H7 (Tobe et al., 2006) and others (gspC, gspD, gspE, gspF, gspG, gspH, gspI, gspJ, gspK, gspL, and gspM) probably encoded secretory pathway proteins of type II secretion system (T2SS) in Shigella dysenteriae Sd197 (Yang et al., 2005). The products of these VFs were effectors and secretory pathway proteins, these components could not form a complete T3SS or T2SS, and they could not perform the function of T3SS or T2SS (Green and Mecsas, 2016). This might be a main reason that these VFs involved in T3SS or T2SS were negatively correlated with the virulence of ExPEC. Moreover, these VFs involved in T3SS or T2SS did not perfectly matched (100% identity and 100% coverage) with corresponding reference genes in VFDB, their nucleotide identities ranged from 90.1 to 99% (data not shown), and the small variations might disrupt the function of these VFs. This might be another reason for these VFs negatively correlated with the virulence of ExPEC. Further studies are needed to determine the functions of these above-mentioned VFs.
We also analyzed ARG and VF scores, differential ARGs and VFs according to molecular types among the 74 ExPEC isolates to explain their phenotype (Supplementary Table S4). Compared with previous study on ST131 strains producing bla CTX−M−15 and aac(6 )-Ib-cr (Alsharapy et al., 2018), this study clearly indicated that a high percentage of bla CTX−M−15 and aac(6 )-Ib-cr was detected in CH40-30-ST131 isolates but not in isolates belonging to other molecular types. The prevalence of tetB in CH35-27-ST69 and CH13-5-ST12 isolates was also reported in this study. Currently, available studies with which to compare differential VFs according to molecular types are scarce. In this study, VFs differed according to molecular types, such as CH14-64-ST1193 isolates with a high prevalence of papB, papI, papX, sat, senB, aerobactin and yersiniabactin genes; CH40-30-ST131 isolates with a high prevalence of sat, aerobactin and yersiniabactin genes; CH35-27-ST69 isolates with a high percentage of yersiniabactin genes; CH11-54 isolates with a low percentage of VFs involved in K1 capsule (kps), and iron uptake (chu and yersiniabactin genes). These VFs might be responsible for the experimental virulence of different types of strains. This suggests that the key VFs are probably different in the same pathogenic process induced by different types of strains. ARG and VF scores according to molecular types were analyzed in this study. Perhaps more significantly, CH37-27-ST405 and CH26-5-ST38 isolates carried more ARGs and VFs among these molecular types, which meant that they had a high resistance and virulence potential and might become more widespread in the future.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

AUTHOR CONTRIBUTIONS
YL, DM, YD, and HR conceived, coordinated the study, and provided the financial support. SL, YC, YD, SZ, HG, and LZ collected the strains and did antimicrobial susceptibility testing. HG, LZ, and YD carried out CH (or ST) typing, whole-genome sequencing, and the G. mellonella killing assay. YD, SL, HG, and ZW analyzed the data. YD drafted the manuscript. YL, DM, and HR revised the manuscript. All authors read and approved the final manuscript.