Genomic characterization of extended-spectrum β-Lactamases (ESBLs)-producing Escherichia coli recovered from bloodstream infections in China Authors: Danni Bao, Xiaohong Xu, Yizhang Wang, Yanhong Wu, Tingfei Chen, Jiayu Shao, Fengjiao Zhu, Huiying Ye, Hongzhang Li Published: 2026 ABSTRACT AbstractBackgroundBloodstream infections (BSIs) caused by Escherichia coli represent a major clinical challenge, particularly among elderly patients and those with urinary catheters. The rising prevalence of extended-spectrum β-lactamase (ESBL)-producing strains further complicates therapeutic management and infection control. This study aimed to characterize the clinical features, antimicrobial resistance profiles, and genomic features of ESBL-producing E. coli isolates associated with BSIs in a county-level hospital in Zhejiang Province, China.MethodsA total of 236 non-duplicate E. coli isolates recovered from BSIs were collected between January 2021 and December 2023. Antimicrobial susceptibility testing (AST) was performed for all isolates. Of these, 144 isolates were successfully preserved for subsequent molecular analysis, among which 65 ESBL-producing isolates underwent whole-genome sequencing (WGS) for the detection of ESBL genes, phylogenetic grouping, multilocus sequence typing, serotyping, fimH typing, and core genome SNP analysis.ResultsMost patients were elderly women, and urinary tract infection was identified as the primary source of bacteremia. Resistance rates were low to piperacillin-tazobactam (2.14%) and cefoperazone-sulbactam (1.28%) but high to levofloxacin (35.74%), ceftriaxone (47.01%), cefuroxime (50.21%), and trimethoprim-sulfamethoxazole (56.17%); all isolates remained fully susceptible to imipenem, ertapenem, amikacin, and tigecycline. Among 144 isolates, 65 (45.1%) were ESBL producers, predominantly carrying blaCTX−M−27, followed by blaCTX−M−14, blaCTX−M−55, and blaCTX−M−15. Phylogenetic analysis showed that the majority of ESBL-producing isolates were assigned to phylogroups B2 and D, with ST131 as the dominant sequence type, strongly associated with O25:H4 and fimH30. Core genome SNP analysis revealed substantial genetic diversity, with closely related isolates mainly observed within the same patient over time.ConclusionsThis study highlights a substantial burden of ESBL-producing E. coli BSIs, dominated by the ST131-fimH30-O25:H4 clone carrying blaCTX−M−27, underscoring the need for enhanced surveillance, targeted antimicrobial therapy, and effective infection control strategies.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12866-026-05039-y.Keywords: Bloodstream infections (BSIs), Escherichia coli, Whole genome sequencing, Antimicrobial resistance, blaCTX−M−27, Sequence types, FimH subtyping Introduction Bloodstream infections (BSIs), largely attributed to extraintestinal pathogenic Escherichia coli, are critical complications leading to extended hospitalization and increased mortality [1]. A number of patient-related factors have been linked to the occurrence of BSIs, including older age, longer hospital stays, and the presence of underlying diseases [2]. Urinary tract infections (UTIs) remain the leading source of infection, followed by biliary tract and other intra-abdominal infections [3]. In high-income settings, this burden is especially evident among elderly individuals, where outcomes are often less favorable [1]. The growing emergence of multidrug-resistant (MDR) E. coli has further complicated clinical management, increasing the risk of bloodstream invasion and adverse outcomes [3]. China, where antibiotic use remains widespread in both healthcare and agricultural sectors, continues to face rising antimicrobial resistance [4]. Data from the China Antimicrobial Surveillance Network (CHINET) in 2023 indicate that resistance rates of E. coli to third-generation cephalosporins have surpassed 50% [4], largely driven by the emergence of ESBLs [5]. ESBL-producing E. coli are resistant to multiple β-lactams, limiting empirical therapy options and contributing to poor clinical outcomes. Risk factors independently associated with ESBL-producing strains include nosocomial infections, biliary tract infections, the use of stomach tube catheters, and previous administration of cephalosporins [2]. In addition, ESBL production, hospital-acquired infection, and malignancy have been identified as predictors of mortality [2]. Since their initial identification in the 1980s, ESBLs have gradually become widespread among Enterobacterales, initially dominated by TEM and SHV, but now largely driven by CTX-M enzymes [5]. These resistance genes are often located on plasmids and often associated with mobile genetic elements such as transposons and insertion sequences, facilitating horizontal gene transfer. Notably, the global population structure of ESBL-producing E. coli is largely driven by the high-risk clone ST131, which is characterized by enhanced virulence and transmission capacity. Understanding the molecular epidemiology, resistance mechanisms, and transmission dynamics of ESBL-producing E. coli in BSIs is essential to develop targeted interventions and improve patient outcomes. This study aimed to characterize the antimicrobial resistance profiles, genomic characteristics, and clonal dissemination of ESBL-producing E. coli isolates from BSIs in a county-level hospital. Methods Study design and isolate collection Between January 2021 and December 2023, a total of 236 non-duplicate E. coli isolates were obtained from patients with BSIs at Sanmen People’s Hospital. All isolates underwent AST as part of routine laboratory analysis. Among these, 144 isolates were successfully preserved for subsequent molecular analyses. Within this group, 65 isolates were identified as ESBL-producing E. coli based on phenotypic AST, and confirmed by WGS. This study was reviewed and approved by the Ethics Committee of Sanmen People’s Hospital, which waived the requirement for informed consent due to the retrospective and observational design. The detailed workflow is presented in Fig. 1. Clinical information was extracted from electronic medical records. Workflow for the genomic and epidemiological analysis of E. coli bloodstream infection isolates Blood culture and bacterial identification For adult patients, 8–10 mL of whole blood was collected per bottle and inoculated into aerobic and anaerobic blood culture bottles (BACT/ALERT FA Plus and FN Plus, bioMérieux, France). For pediatric patients, 3–5 mL of blood was inoculated into pediatric bottles (BACT/ALERT PF, bioMérieux, France) according to the manufacturer’s recommendations. The culture was maintained for 5 days or until a positive result was obtained in an automatic blood culture system (BACT/ALERT 3D). Matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry systems (bioMérieux, Lyon, France) were used for species identification. Antimicrobial susceptibility testing AST for the isolates was performed using the VITEK 2 Compact system (bioMérieux, France). The susceptibility of the isolates to amikacin, amoxicillin-clavulanic acid, cefoperazone-sulbactam, piperacillin-tazobactam, ceftazidime, ceftriaxone, cefepime, cefoxitin, cefuroxime, levofloxacin, imipenem, ertapenem, tigecycline and trimethoprim-sulfamethoxazole was determined according to the manufacturer’s guidelines. AST was performed, and minimum inhibitory concentrations (MICs) were interpreted according to the Clinical and Laboratory Standards Institute M100 guidelines (2023). E. coli ATCC 25,922 was used as the quality control strain. As CLSI breakpoints for tigecycline in Enterobacterales are not available, interpretation followed the criteria established by the European Committee on Antimicrobial Susceptibility Testing (EUCAST). For cefoperazone-sulbactam, susceptibility results was interpreted using the CLSI breakpoints for cefoperazone. Whole-genome sequencing and assembly Whole-genome sequencing of the 65 ESBL-producing isolates was performed using the Illumina HiSeq X-Ten platform (Illumina, San Diego, CA, USA) with a 150 bp paired-end strategy. Raw sequencing reads were quality-filtered and trimmed using fastp v0.23.2 [6]. De novo genome assembly was conducted using Shovill (https://github.com/tseemann/shovill), and contigs with low coverage (< 10×) or short length (< 250 bp) were excluded from downstream analyses. Genomic characterization and phylogenetic analysis Phylogenetic grouping of E. coli isolates was assigned using the Clermont scheme. BacWGSTdb 2.0 online server was used to identify sequence types [7]. Serotype, fimH type, virulence factors, and antimicrobial resistance genes were detected based on SerotypeFinder v2.0, CHTyper v1.0, VirulenceFinder v2.0, and ResFinder v3.2, respectively, available from the Center for Genomic Epidemiology (CGE) (https://www.genomicepidemiology.org/services/). Core genome single nucleotide polymorphisms (cgSNPs) were identified by mapping sequencing reads against the genome of isolate 2021-0146 from this study, using Snippy v4.6.0 (https://github.com/tseemann/snippy). Recombination sites were removed using Gubbins v3.3 with default parameters (Croucher et al., 2015). The maximum-likelihood phylogeny based on the concatenated dataset was inferred using IQ-TREE 2 under the HKY model [8]. The resulting phylogenetic tree was visualized using the Interactive Tree of Life (iTOL) web server [9]. The WGS data of all E. coli strains involved in this study were deposited in the NCBI database under the BioProject accession number PRJNA1180495. Statistical analysis Statistical analyses were performed using SPSS version 28 (IBM Corp., Armonk, NY, USA). Categorical variables were compared using the Chi-square test or Fisher’s exact test where appropriate. Continuous variables were expressed as median and interquartile range (IQR) and compared using the Mann-Whitney U-test. Associations between virulence genes and specific sequence types (e.g., ST131) were evaluated using Fisher’s exact test. To control for multiple comparisons in virulence gene analysis, the false discovery rate (FDR) correction using the Benjamini-Hochberg method was applied. A two-tailed P value < 0.05 was considered statistically significant. Results The median age of the patients was 72 years (IQR 60–79, range 17–97) and 156 (66.1%) of 236 patients were women. Among them, 148 individuals exhibited abnormalities in routine urinalysis, and 62 had undergone urinary catheterization. Similar to other studies, the urinary tract was the most frequent source of bacteremia. In this cohort, four patients died upon discharge, resulting in a mortality rate of 1.7%. AST of 236 E. coli isolates showed varying resistance rates. Piperacillin-tazobactam (2.1%) and cefoperazone-sulbactam (1.3%) showed the lowest resistance rates, followed by amoxicillin-clavulanic acid (6.9%), cefoxitin (7.3%), ceftazidime (16.2%), and cefepime (26.4%). In contrast, levofloxacin (35.7%), ceftriaxone (47.0%), cefuroxime (50.2%), and trimethoprim-sulfamethoxazole (56.2%) exhibited higher resistance, but all susceptible to imipenem, ertapenem, amikacin, and tigecycline (Table 1). Table 1Comparison of antibiotic resistance rates between all E. coli isolates (n = 236), ESBL-producing isolates (n = 65) and no ESBL-producing isolates (n = 79)All isolates (n = 236)ESBL-producing isolates (n = 65)No ESBL-producing isolates (n = 79)S (%)I/SDD (%)R (%)S (%)I/SDD (%)R (%)S (%)I/SDD (%)R (%)TZP96.21.72.1100.00010000AMC79.813.36.978.518.53.086.17.66.3CAZ65.118.716.229.243.127.791.15.13.8FEP66.68.026.418.529.252.389.96.33.88CXM47.22.650.200100.0815.113.9SCF96.62.11.392.34.63.110000CRO53.0047.000100.094.905.1FOX87.15.67.386.212.31.592.407.6IPM99.60.40100.00010000ETP100.000100.00010000AMK99.60.40100.00010000LVX24.739.635.712.335.452.339.23822.8TGC99.60.40100.00098.71.30SXT43.8056.232.3067.754.4045.6R resistant, I intermediate, SDD susceptible dose-dependent, S susceptible, TZP piperacillin/tazobactam, AMC amoxicillin/clavulanic acid, CAZ ceftazidime, FEP cefepime, CXM cefuroxime, SCF cefoperazone/sulbactam, CRO ceftriaxone, FOX cefoxitin, IPM imipenem, ETP ertapenem, AMK amikacin, LVX levofloxacin, TGC tigecycline, SXT trimethoprim/sulfamethoxazole Comparison of antibiotic resistance rates between all E. coli isolates (n = 236), ESBL-producing isolates (n = 65) and no ESBL-producing isolates (n = 79) R resistant, I intermediate, SDD susceptible dose-dependent, S susceptible, TZP piperacillin/tazobactam, AMC amoxicillin/clavulanic acid, CAZ ceftazidime, FEP cefepime, CXM cefuroxime, SCF cefoperazone/sulbactam, CRO ceftriaxone, FOX cefoxitin, IPM imipenem, ETP ertapenem, AMK amikacin, LVX levofloxacin, TGC tigecycline, SXT trimethoprim/sulfamethoxazole Among the 236 isolates, 144 were successfully preserved for subsequent data analyses, of which 65 (45.1%) were identified as ESBL-producers and 79 non-ESBL isolates. ESBL-producing isolates exhibited markedly higher resistance rates to third- and fourth-generation cephalosporins, with resistance to ceftriaxone and cefuroxime reaching 100%. Resistance to levofloxacin (52.3%) and trimethoprim-sulfamethoxazole (67.7%) was also more frequent among ESBL-producing isolates. In contrast, non-ESBL isolates showed substantially lower resistance rates to these antibiotics. Both groups remained highly susceptible to carbapenems, amikacin, and tigecycline (Table 1). Among the ESBL-producing isolates, blaCTX−M−27 (31/65, 47.7%) was the most frequent ESBL-encoding gene, followed by blaCTX−M−14 (13/65, 20.0%), blaCTX−M−55 (13/65, 20.0%), blaCTX−M−15 (7/65, 10.8%), blaCTX−M−3 (3/65, 4.6%), and blaCTX−M−65 (1/65, 1.53%). Notably, strains 2023 − 0230 and 2023 − 0956 carried both blaCTX−M−27 and blaCTX−M−15, with strain 2023 − 0956 harboring blaCTX−M−55 and blaCTX−M−65 genes. Additionally, strain 2023-103014 carries two copies of the blaCTX−M−55 genes (Supplementary Table 1). Isolates carrying blaCTX−M-type genes exhibited high resistance rates to third- and fourth-generation cephalosporins, including cefotaxime, ceftazidime, and cefepime, which is consistent with the known resistance spectrum of CTX-M-type ESBL enzymes. In addition, the ESBL-producing isolates harbor multiple resistance genes. The most commonly detected were sulfonamide and tetracycline resistance genes, sul2 (70.8%) and tet(A) (70.8%), respectively. Aminoglycoside resistance was also prevalent, with aph(6)-Id (66.2%) and aph(3’’)-Ib (64.6%) being the predominant genes. Among β-lactam resistance genes, blaEC−5 was found in 58.5% of cases, while macrolide resistance was mainly mediated by mph(A) (61.5%). Trimethoprim resistance was attributed to dfrA17 (55.4%), and fosfomycin resistance was conferred by fosA3 (9.2%) (Table 2, Supplementary Table 2). Table 2Major antibiotic resistance genes detected in 65 ESBL-producing E. coli isolatesResistance geneCountPercentage (%)Aminoglycosides aadA5 3553.8 aph(3’’)-Ib 4264.6 aph(6)-Id 4366.2 aac(3)-IId 2030.8Beta-lactams (ESBLs & others) bla CTX-M-14 1320.0 bla CTX-M-15 710.8 bla CTX-M-27 3147.7 bla CTX-M-55 1320.0 bla TEM-1 2843.1 bla EC-5 3858.5 bla EC-8 1116.9 bla EC-18 812.3Sulfonamides sul1 3350.8 sul2 4670.8 dfrA17 3655.4Tetracyclinestet(A)4670.8PhenicolsfloR1218.5Macrolidesmph(A)4061.5Only resistance genes with frequency ≥10% are shown in the main table Major antibiotic resistance genes detected in 65 ESBL-producing E. coli isolates Only resistance genes with frequency ≥10% are shown in the main table The two most common phylogroups were B2 (39/65, 60.0%) and D (10/65, 15.4%), followed by B1 (8/65, 12.3%), F (4/65, 6.2%), A (2/65, 3.1%), and G (1/65, 1.5%) (Table 3). Based on the phylogenetic classification applied in this study, 44 isolates (67.7%) were assigned to group 1 (B2, F, and G), 11 (16.9%) to group 2 (A, B1, and C), and 10 (15.3%) to group 3 (D). MLST analysis classified into 24 STs, and three isolates belonged to newly identified STs. ST131 was the most prevalent (27/65, 41.5%), followed by ST1193 (6/65, 9.2%), ST69 (3/65, 4.5%) and ST38 (3/65, 4.5%) (Table 3; Fig. 2). Table 3Number of isolates by sequence type (ST) within each phylogroup in our collection of 65 ESBL-producing E. coli isolatesPhylogroupSequence type (ST)Phylogroup A (n = 2)ST10 (n = 1), ST93 (n = 1)Phylogroup B1 (n = 8)ST155 (n = 2), ST162 (n = 2), ST1196 (n = 1), ST602 (n = 1), ST101 (n = 1), ST10512 (n = 1),Phylogroup B2 (n = 39)ST12 (n = 2), ST131 (n = 27), ST1193 (n = 6), ST998 (n = 1), ST4456 (n = 2), STnew (n = 1)Phylogroup C (n = 1)ST410 (n = 1)Phylogroup D (n = 10)ST38 (n = 3), ST69 (n = 3), ST393 (n = 2), ST3052 (n = 1), ST10184 (n = 1)Phylogroup F (n = 4)ST648 (n = 2), novel ST (n = 2),Phylogroup G (n = 1)ST117 (n = 1) Number of isolates by sequence type (ST) within each phylogroup in our collection of 65 ESBL-producing E. coli isolates Sankey diagram based on the metadata of 65 ESBL-producing E. coli strains. The lines are drawn connecting strain number, MLST, phylogroup, serotype and fimH of 65 ESBL-producing E. coli strains. The width of the lines reflects the number of isolates for each category, going from left to right. Lines are colored based on different classification criteria The analysis of the serotypes among the bacterial strains revealed a diverse distribution, with clear associations to specific phylogroups and sequence types. The most prevalent serotype identified was O25:H4 (24.6%), followed by O16:H5 (15.3%), both of which were predominantly associated with phylogroup B2 and ST131. Similarly, O75:H5 was strongly linked to ST1193 in phylogroup B2. In phylogroup D, a higher diversity of serotypes was observed, including O15:H18 (ST10184 and ST393), O17:H18 (ST69), and O45:H6 (ST38) (Fig. 2, Supplementary Table 1). Subtyping based on fimH revealed 17 subtypes, and one isolate was not typable. fimH30 was most frequently typed among ST131 isolates (15/27, 55.5%), followed by fimH41 (10/27, 37.0%) and fimH54 (2/27, 7.4%) (Fig. 2, Supplementary Table 1). Because the isolates analyzed in this study were recovered from bloodstream infections, they are most consistent with extraintestinal pathogenic E. coli (ExPEC). ExPEC strains typically harbor virulence determinants involved in adhesion, iron acquisition, and toxin production, which facilitate colonization, invasion, and systemic dissemination within the host. The predominance of phylogroups B2 and D observed in this study is consistent with previous reports indicating that these phylogenetic groups are strongly associated with ExPEC lineages responsible for invasive infections. Similar observations have been reported in recent genome-based studies of pathogenic E. coli, which highlight the genetic diversity and virulence potential of ExPEC populations across different hosts and environments [10]. A diverse array of virulence factors was identified among the E. coli isolates, with 145 virulence-associated genes detected in total (Supplementary Table 3). Overall, virulence genes associated with adhesion, iron acquisition, and secretion systems were commonly detected. Several conserved genes, including ompA, as well as genes encoding curli fimbriae and type 1 fimbriae, were present in the majority of isolates. In addition, iron acquisition systems such as enterobactin were universally detected, while the yersiniabactin cluster was also highly prevalent. Notably, several virulence genes were significantly associated with specific sequence types. In particular, chuA, kpsD, kpsM, pap operon genes, and sat were significantly associated with ST131 lineage (P < 0.05) (Supplementary Table 4). A core genome-based phylogenetic tree was constructed for the 65 E. coli isolates (Fig. 3). The isolates clustered into multiple lineages and phylogroups, with clear separation observed among major lineages. Most isolates belonging to the same lineage tended to cluster within the same clade, supporting clonal expansion. In contrast, identical ESBL genes were detected across multiple unrelated lineages, suggesting frequent horizontal transfer via mobile genetic elements. Analysis of pairwise SNP differences showed a high level of genomic diversity among the 65 ESBL-producing isolates, with distances ranging widely from 2 to 74,270 SNPs. Notably, five pairs of strains displayed SNP differences of fewer than 20. These were as follows: strain pair 2023-1865 and 2022-0821 differed by 13 SNPs; strain pair 2023-1133 and 2023-0874 showed a difference of only 2 SNPs; strain pair 2022-1289 and 2022-1315 had a 4 SNP difference; strain pair 2023-0901 and 2023-1127 differed by 5 SNPs; strain pair 2023-0915 and 2023-0757 exhibited a 3 SNP difference. Of these pairs, only 2023-1865 and 2022-0821 were isolated from different patients. The other four pairs were recovered from the same individuals but at different time points during hospitalization, suggesting possible within-host persistence or microevolution. Phylogenetic tree of 65 ESBL-producing E. coli isolates analyzed in this study. The tree was constructed based on core genome SNP analysis. The colored annotation tracks indicate the sequence type (ST), phylogenetic group, and ESBL genes for each isolate. Different colors represent distinct categories as shown in the legend Discussion E. coli remains the most common cause of bloodstream infections globally and is linked to considerable morbidity and mortality, particularly in vulnerable populations. The growing prevalence of antimicrobial resistance, especially among ESBL-producing strains, has made clinical management increasingly challenging and, in many cases, reduced available treatment options. Understanding local epidemiological and molecular characteristics is therefore essential for guiding empirical treatment and infection control strategies. In the present study, we examined 236 E. coli isolates from BSIs collected over a three-year period, with particular attention given to 65 ESBL-producing strains. Our findings demonstrate a high prevalence of ESBL-producing isolates and reveal the predominance of the high-risk ST131 lineage carrying blaCTX−M−27, providing important epidemiological insights for the management of E. coli bloodstream infections. The study cohort had a median age of 72 years, and 66.1% were female, which generally reflects the higher incidence of E. coli BSIs among older individuals and women [3]. As the global elderly population continues to grow at an unprecedented rate, it is expected that the incidence of E. coli bacteremia will similarly increase unless preventive measures are implemented [3]. According to the Blood Bacterial Resistant Investigation Collaborative System (BRICS) report, E. coli resistance notably increased from 29.1% to 38.5% during from 2014 to 2019 [11]. Most cases are likely to occur in individuals aged 65–74 years, a group that currently represents a large proportion of the elderly population, and this trend is expected to persist [3]. These findings highlight the increasing burden of E. coli bacteremia among elderly populations. The findings also emphasize the urinary tract as the most common source of bacteremia, with 148 patients exhibiting abnormalities in routine urinalysis and 62 having undergone urinary catheterization, confirming urinary tract infection as a major risk factor [3, 12, 13]. The overall mortality rate in this study was relatively low (1.7%), which is lower than the mortality rates of approximately 10–13% reported for E. coli bloodstream infections in previous studies [14, 15]. This discrepancy may be explained by several factors. First, mortality in this study was defined as in-hospital death recorded during hospitalization, and post-discharge outcomes were not systematically tracked. Also, some critically ill patients may have been discharged at the terminal stage, which could lead to an underestimation of mortality in retrospective analyses. AST revealed varying resistance profiles among the 236 E. coli isolates. The lowest resistance rates were observed for carbapenems (imipenem, ertapenem), amikacin, and tigecycline (0%). This pattern is generally consistent with global surveillance data, where carbapenems are still widely considered reliable options for treating severe infections caused by resistant E. coli [1]. Similarly, low resistance rates were also observed for piperacillin-tazobactam (2.1%) and cefoperazone-sulbactam (1.3%), suggesting that these antimicrobial agents remain effective for treating E. coli BSI, particularly when resistance to other antibiotics is present. By contrast, relatively high resistance was observed for levofloxacin (35.7%), third-generation cephalosporins such as ceftriaxone (47.0%) and cefuroxime (50.2%), and trimethoprim-sulfamethoxazole (56.2%), consistent with findings from previous studies [11]. This increasing resistance may be attributed to the widespread use and overuse of these antibiotics in both outpatient and hospital settings. These findings highlighting the urgent need for improved antimicrobial stewardship programs. A considerable proportion of the isolates (45.1%, 65/144) were identified as ESBL producers. This proportion is broadly comparable to national surveillance data reported in China between 2014 and 2019, where prevalence rates ranged from 48.8% to 58.2% [11]. Compared with the overall cohort of 236 isolates, ESBL-producing E. coli demonstrated markedly higher resistance rates. This difference was particularly evident for third-generation cephalosporins, including ceftazidime (27.7% vs. 16.2%), cefuroxime (100% vs. 50.2%) and ceftriaxone (100% vs. 47.0%), as well as the fourth-generation cephalosporin cefepime (52.3% vs. 26.4%). In addition, co-resistance to non-β-lactam antibiotics was frequently observed, with ESBL producers exhibiting notably higher resistance to levofloxacin (52.3% vs. 35.7%) and trimethoprim-sulfamethoxazole (67.7% vs. 56.2%). This co-resistance pattern may be related to the co-existence of multiple resistance determinants on plasmids, which facilitates both co-selection and horizontal transfer of resistance traits. When compared with non-ESBL isolates, the differences in resistance were even more pronounced. Non-ESBL isolates exhibited substantially lower resistance rates to cephalosporins, including ceftriaxone (5.1%), ceftazidime (3.8%), and cefepime (3.8%). Similarly, resistance to levofloxacin and trimethoprim-sulfamethoxazole was also lower among non-ESBL isolates, highlighting the strong association between ESBL production and multidrug resistance. Among the ESBL-producing isolates, blaCTX−M−27 was the most prevalent gene (47.7%). Previous studies from different regions have also reported the increasing predominance of blaCTX−M−27 among ESBL-producing E. coli [16–18]. In particular, the rapid global emergence of the ST131 C1-M27 lineage carrying blaCTX−M−27 has been documented in several surveillance studies from Europe and North Africa, highlighting its expanding dissemination worldwide [19, 20]. Consistent with these observations, blaCTX−M−27 was the most prevalent ESBL gene identified in our study, suggesting that the ST131 C1-M27 lineage may also play an important role in the epidemiology of ESBL-producing E. coli. The spread of blaCTX−M−27 is likely facilitated by mobile genetic elements carried on plasmids, which enable efficient horizontal gene transfer among Enterobacteriaceae and contribute to its rapid dissemination [21, 22]. Together, these findings highlight the growing importance of blaCTX−M−27-producing strains and underscore the need for continued molecular surveillance. Molecular typing revealed a wide clonal distribution among the E. coli isolates from BSIs [23]. In our study, the most common lineage was ST131, which comprised 41.5% of all ESBL-producing strains. ST131 is known for its association with multidrug resistance and enhanced virulence [24], which has been implicated in a number of global E. coli outbreaks [25]. It frequently carries multiple antimicrobial resistance genes, including blaCTX−M−27, blaCTX−M−15 and blaCTX−M−9 [26, 27]. In addition, exhibits enhanced adhesion and invasion capabilities due to the presence of virulence factors such as fimH30, iroN, ompA, and ibeA [26]. Isolates belonging to ST131 and ST1193 also showed relatively limited sequence variation, which may reflect the clonal expansion of these successful lineages. Other lineages, such as ST1193 [28] and ST69 [29] were also prevalent, while the remaining isolates were distributed across multiple sequence types. Most of these lineages were represented by only one or a few isolates. The presence of ESBL genes across multiple lineages suggests that plasmid-mediated horizontal gene transfer may play an important role in their dissemination. Phylogenetic analysis indicated that the most prevalent phylogroups among the ESBL-producing strains were B2 (60%) and D (15.4%), both of which are commonly associated with ExPEC. The combined proportion of these two phylogroups (75.4%) is consistent with previous reports on bacteraemic isolates, where similar distributions of extraintestinal phylogroups (approximately 67–80%) have been described [1, 30, 31]. Phylogroup B2, in particular, is frequently linked to urovirulent strains and was predominantly represented by the ST131 clonal group. These findings suggest that certain phylogroups, particularly B2, are overrepresented in bloodstream infections, which may be related to their increased virulence and ability to invade and disseminate within the host. Serotype analysis revealed a heterogeneous distribution overall, although certain clear associations between serotypes, phylogroups, and sequence types were observed. The most prevalent serotype identified was O25:H4 (24.6%), followed by O16:H5 (15.3%). Both of these serotypes were predominantly associated with phylogroup B2 and the highly virulent ST131 strain [32]. The strong association of O25:H4 and O16:H5 with ST131 suggests that these serotypes may play a key role in clinical infections of this clone. The O25:H4 serotype constitutes the predominant ST131 clade and is mainly characterized by fimH30, whereas the O16:H5 serotype represents a more divergent lineage, typically associated with fimH41 [32, 33]. Subtyping based on fimH revealed significant variability in adhesin types across the isolates. The fimH30 subtype was most prevalent in ST131 strains, consistent with previous studies showing the association of fimH30 with enhanced virulence in E. coli. When introduced into an isogenic ST131 background, fimH30 exhibited increased adhesion to and invasion of human cells, along with enhanced biofilm-forming capacity compared with other variants [34]. ExPEC, particularly uropathogenic E. coli (UPEC), harbor multiple virulence determinants that facilitate colonization of the urinary tract and subsequent systemic dissemination. These determinants mainly include adhesins, toxins, and iron-acquisition systems, which collectively promote bacterial adhesion, invasion, and survival within the host [35]. The ompA gene, which encodes outer membrane protein A, was detected in 100% of the isolates in this study. OmpA plays a pivotal role in pathogenesis by promoting bacterial adhesion, invasion, and intracellular persistence, while also contributing to immune evasion and the induction of pro-inflammatory cytokines [36]. Among these, the adhesion-related gene fimH (98.5%), fyuA (84.6%) and cnf1 (27.7%) were particularly notable for its strong association with urinary tract infections, especially in catheterized patients [37–39]. Additionally, iutA, the most prevalent gene encoding aerobactin receptor in uropathogenic UPEC [40], was detected in 75.4% of the strains in our study. Although the ibeA (4.6%) gene, encoding the invasion of brain endothelium A protein, is found at a very low prevalence, it has been shown to be important for invasion of the blood-brain barrier [41]. The focA, focC, focD, focF, focG and focH genes (1.5–3.1%) encode F1C fimbriae, which are commonly linked to uropathogenic strains and mediate binding to glycosphingolipids on epithelial cells of the bladder, urethra, and kidneys, thereby promoting bacterial adhesion and colonization [42]. In our cohort, several virulence genes were significantly associated with the ST131 lineage, including chuA, chuS, chuT, kpsD, kpsM, pap operon, and sat. The Chu system facilitates heme uptake and has been shown to be more abundant in uropathogenic UPEC than in commensal strains [43]. The presence of kpsD and kpsM indicates an increased capacity for capsule biosynthesis. Previous studies have consistently reported that kpsM and related genes are closely associated with the virulence and pathogenicity of ST131 [44]. Furthermore, the pap operon is closely linked to urinary tract colonization [45]. The sat gene encodes a secreted autotransporter toxin (SAT), which has been implicated in the pathogenic potential of UPEC. SAT exerts cytopathic activity on host epithelial and endothelial cells, leading to direct tissue injury and disruption of cellular integrity [46]. Moreover, this toxin may facilitate the entry of pyelonephritogenic strains into the bloodstream by specifically damaging the glomeruli and proximal tubules [47]. The observed SNP differences among the 65 E. coli strains suggest considerable genetic diversity within the collection. However, the identification of closely related strains with fewer than 20 SNP differences highlights potential clonal relationships [48], particularly among strains isolated from the same patient over time. Four of these pairs were obtained from the same patient at different time points, suggesting within-patient persistence or recurrence of the same clonal lineage, possibly due to ongoing colonization or reinfection. One pair (2023-1865 and 2022-0821) was obtained from different patients and showed 13 SNP differences. However, due to the lack of epidemiological data (e.g., temporal overlap or shared hospital wards), possible transmission between patients cannot be confirmed. Therefore, these findings should be interpreted cautiously. Although this study covered a three-year period (2021-2023), no clear year-to-year change in the prevalence of ESBL-producing isolates was observed. However, the consistent detection of CTX-M-27-producing ST131 strains throughout the study period suggests that this high-risk lineage has become established in our hospital. Continuous surveillance will therefore be important to monitor potential future expansion. Although the predominance of ST131 carrying CTX-M-type ESBLs has been widely reported in tertiary hospitals worldwide, data from county-level hospitals remain relatively limited. County hospitals represent an important component of the healthcare system in China and often serve large rural populations. Our findings suggest that the molecular epidemiology of ESBL-producing E. coli in this county-level hospital largely mirrors patterns reported in larger tertiary care centers, highlighting the widespread dissemination of high-risk clones such as ST131 across different healthcare settings. This observation underscores the need for genomic surveillance not only in large academic hospitals but also in regional and county-level healthcare institutions. This study has several limitations that should be acknowledged. First, it was performed at a single center, which may restrict the generalizability of the results. In addition, only 144 of the 236 collected isolates (61.0%) were successfully preserved and available for downstream analysis. Due to routine laboratory workflow and limitations in strain preservation during the early phase of the study period, not all isolates were stored. This incomplete strain preservation may introduce potential selection bias, as the ESBL prevalence reported in this study was calculated based on the preserved subset rather than the entire collection. Although AST was performed for all isolates, genomic characterization was limited to the available strains. Future studies with systematic strain preservation would help provide a more comprehensive evaluation of the molecular epidemiology of E. coli. Second, whole-genome sequencing was performed only on ESBL-producing isolates. Therefore, the phylogenetic structure and virulence gene distribution described in this study represent only the ESBL-producing subset rather than the entire population of E. coli causing bloodstream infections. As a result, direct comparisons between ESBL and non-ESBL isolates could not be performed. Future studies including genomic analysis of both ESBL and non-ESBL isolates would provide a more comprehensive understanding of the molecular epidemiology of E. coli bloodstream infections. Third, the available clinical data were relatively limited because of the retrospective design of the study. Important clinical variables such as comorbidities, prior antibiotic exposure, and infection source were not systematically analyzed. Finally, mortality was defined as in-hospital death, and post-discharge outcomes were not available. In some cases, critically ill patients may have been discharged at the terminal stage, which may lead to an underestimation of the true mortality associated with E. coli bloodstream infections. Conclusion This study provides a genomic epidemiological analysis of E. coli bloodstream infections, highlighting the substantial burden of antimicrobial resistance, particularly among ESBL-producing strains. Importantly, this work was conducted in a county-level (primary care) hospital setting, which offers valuable insight into the local epidemiology of bloodstream infections in resource-limited healthcare environments. The findings emphasize the importance of routine AST and molecular diagnostics, alongside the need for strengthened infection control practices, especially in high-risk populations such as elderly patients and those with urinary catheterization. From the perspective of a primary healthcare facility, these results also reflect the practical challenges of managing multidrug-resistant infections in routine clinical practice. Continued surveillance of E. coli bloodstream infections, together with a better understanding of their genetic and phenotypic diversity, will likely be important for informing future treatment strategies and guiding public health interventions at both regional and broader levels. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Acknowledgements We thank the staff of the microbiology laboratory of Sanmen People’s Hospital for their assistance with sample processing and bacterial identification. We also acknowledge the support provided by colleagues involved in sequencing and bioinformatics analysis. AI-assisted language editing statement The authors declare that AI-based tools were used only to assist with language refinement and improving clarity. These tools had no role in study design, data analysis, interpretation, or the generation of scientific content. The authors assume full responsibility for the accuracy and integrity of the manuscript. Abbreviations Bloodstream Infections Whole-genome Sequencing extended-spectrum β-lactamase Urinary Tract Infections Multidrug-Resistant China Antimicrobial Surveillance Network Core genome single nucleotide polymorphisms Antimicrobial Susceptibility Testing Minimum Inhibitory Concentrations Clinical and Laboratory Standards Institute European Committee on Antimicrobial Susceptibility Testing Extraintestinal pathogenic E. coli Blood Bacterial Resistant Investigation Collaborative System Uropathogenic E. coli Secreted Autotransporter Toxin Authors’ contributions DNB, XHX, and YZW performed bacterial isolation, antimicrobial susceptibility testing, and genomic data analysis. YHW and TFC assisted with whole-genome sequencing, genome assembly, and annotation. JYS provided bioinformatics support and contributed to phylogenetic analysis and data visualization. FJZ contributed to data interpretation and literature review. HYY and HZL conceived and designed the study, supervised the research, and critically revised the manuscript. All authors contributed to manuscript writing and approved the final version. Funding This study is supported by Zhejiang Provincial Natural Science Foundation of China (LQ24H200003), Zhejiang Provincial Medical and Health Science and Technology Plan (2023KY414, 2023KY288 and 2024KY544), and Taizhou Science and Technology Plan (23ywb169). Data availability The datasets generated and/or analyzed during the current study are publicly available in the NCBI BioProject database under the accession number PRJNA1180495. Declarations Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of the Sanmen Peopleʼs Hospital (Approval No. 20230012). The requirement for informed consent was waived by the Ethics Committee of the Sanmen People’s Hospital due to the retrospective nature of the study and the use of anonymized data, with no identifiable patient information involved. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Huiying Ye, Email: yehuiying@zjsmyy.com. Hongzhang Li, Email: lihongzhang@zjsmyy.com. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Data Availability Statement The datasets generated and/or analyzed during the current study are publicly available in the NCBI BioProject database under the accession number PRJNA1180495.