Comparative core-genome MLST of vancomycin-resistant Enterococcus faecium supports the utility of wastewater-based surveillance: a pilot study

TL;DR

This study shows that wastewater can be used to track antibiotic-resistant bacteria, like vancomycin-resistant Enterococcus faecium, and aligns with human surveillance data.

Contribution

The study demonstrates the utility of wastewater-based surveillance for tracking AMR in VREfm using core-genome MLST.

Findings

VREfm was isolated from 22% of wastewater samples, with no significant variation across treatment plants.

Most wastewater isolates clustered closely with human isolates, showing genetic similarity.

Wastewater-based surveillance aligns with national human surveillance data for VREfm.

Abstract

Wastewater-based surveillance (WBS) could complement clinical data and be used as an early warning tool for population-level monitoring of priority pathogens such as vancomycin-resistant Enterococcus faecium (VREfm). In this pilot study, we isolated VREfm using CHROMAgar VRE from 77 composite wastewater influent samples from ten wastewater treatment plants (WWTPs), assessed their phenotypic antibiotic susceptibility profiles using the broth microdilution assay, and used core genome MLST (cgMLST) to examine the genetic relatedness of human and wastewater isolates in Finland. VREfm was isolated from 17 samples (22%), with no significant difference in the isolation rate across the ten WWTPs (p = 0.407). The phenotypic antimicrobial susceptibility testing (AST) revealed that all isolates were resistant to ampicillin, ciprofloxacin, erythromycin, teicoplanin, and vancomycin and were susceptible to chloramphenicol, daptomycin, and linezolid. All 17 VREfm isolates belonged to CC17, and 12 of them were sequence type ST80. Altogether, 76% (n = 13) of the wastewater isolates harboured the vanA gene, 18% (n = 3) had the vanB gene, and 6% (n = 1) had both vanA and vanB genes. cgMLST revealed that most VREfm strains from wastewater clustered closely (< 10 allelic differences) with those from human surveillance isolates across the country. The ST80 isolates belonged to the clusters identified during 2022–2023 as part of national VRE surveillance of screening samples. Therefore, WBS closely aligns with what is observed through human surveillance. We believe that this pilot supports the utility of WBS as a sensitive tool for AMR surveillance. This study also contributes to the One Health approach, needed to fully understand the transmission dynamics of pathogenic bacteria and effectively manage the antimicrobial resistance (AMR) challenge.

The online version contains supplementary material available at 10.1186/s12866-025-04610-3.

Figures

Full text

Introduction

The 2024 World Health Organization (WHO) bacterial priority pathogen list classifies vancomycin-resistant Enterococcus faecium(VREfm) as a high-priority pathogen [1]. As the first of the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) VREfm is a major cause of healthcare-associated outbreaks and multidrug-resistant infections with increasing global prevalence [2]. In Europe, the proportion of VREfm rose from 8.1% (95% CI: 6.7–9.7) in 2012 to 19.0% (95% CI: 16.8–21.5) in 2018 [3], with global estimates ranging from 1 to 55% [4]. Clinically, E. faeciumhas emerged as a key nosocomial pathogen, causing endocarditis, urinary tract infections, and septicemia [5], and has been implicated in exacerbating inflammatory bowel disease [6]. The rising burden of VREfm also translates into substantial healthcare costs and prolonged hospital stays [7, 8].

In Finland, VRE prevalence remains comparatively low [3]. The incidence rates ranged between 0.4 and 5 cases per 100,000 inhabitants annually between 2016 to 2023, with E. faecium accounting for 97% of isolates and both vanA and vanB genotypes detected [9]. Whole-genome sequencing has identified seven VREfm outbreak clusters since 2016 [9–12]. While clinical and laboratory-based molecular surveillance remain central to outbreak detection, these methods provide a limited view of pathogen dissemination beyond healthcare settings.

Wastewater-based surveillance (WBS) offers a complementary, population-level approach aligned with the One Health framework, which integrates human, environmental, and microbial health. By monitoring pathogens circulating within communities, WBS can reveal spatial and temporal trends that extend beyond clinical case detection and support early warning systems in resource-limited settings [13]. Finland has already applied WBS to monitor multiple pathogens, including SARS-CoV-2, norovirus, and antimicrobial resistance genes [13–17] . Previous studies have also indicated epidemiological links between wastewater and clinical VREfm isolates [18–20], yet few have systematically examined the genetic diversity of VREfm across human and wastewater sources.

Here, we pilot a culture-based WBS approach to evaluate its potential as a complementary tool for assessing the carriage and diversity of VREfm in Finland. This integrative strategy aims to strengthen national antimicrobial resistance surveillance and contribute to One Health-informed monitoring of emerging bacterial threats. We utilized short-read genomic sequencing to assess the genetic relatedness of wastewater and sequenced human surveillance and clinical VREfm isolates.

Materials and methods

Study sites and phenotypic characterization of vancomycin-resistant E. faecium isolates

The methodology is similar to that utilized in the other studies under the WASTPAN project [13]. Briefly, the project analyzed 77 composite wastewater (WW) samples (1 L each) from across ten across Finland. The cities serviced by these wastewater treatment plants (WWTPs) were Espoo, Helsinki, Kuopio, Lappeenranta, Oulu, Pietarsaari, Rovaniemi, Seinäjoki, Tampere, and Turku. Sampling took place between February 2021 and January 2022. The ten sampling sites were representative of 44 municipalities and served over 2.24 million people (Table S1). These municipalities collectively represent 40% of Finland's population. The samplings were done eight times during the year to obtain two samples per season in Finland. Other parameters for wastewater (WW) samples are shown in Table S2. Previously, we have shown that monthly trends and incidence of ESBL-producing Escherichia coliwithin the catchment area of the 10 WWTPs sampled in this study closely mirrored those in some 28 WWTPs and the nationwide trends of the same pathogen [13].

Upon the immediate transportation of the samples to our bacteriology laboratory, we cultivated VRE by direct plating 10 µL of pre-enriched medium (Mueller–Hinton Broth supplemented with 2 mg/L vancomycin) for each sample onto CHROMagar VRE agar plates (Paris, France). These plates were incubated for 18–24 h at 37 ˚C. Presumptive E. faecium isolates were sub-cultured on fresh CHROMagar VRE plates and finally purified on nutrient agar (Oxoid, Basingstoke, UK). One to three colonies were picked from each plate and were identified using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-ToF) (Brüker, Bremen, Germany) following the Biotyper protocol using the best match score value of ≥ 2.30. The broth microdilution assay was used to determine the minimum inhibitory concentration against antibiotics infused into the Thermofisher EUVENC Sensititre plates (chloramphenicol, vancomycin, teicoplanin, gentamycin, ampicillin, erythromycin, ciprofloxacin, daptomycin, tetracycline, linezolid, and synercid (quinupristin/dalfopristin). The ATCC 8459 strain of E. faeciumwas used as a control strain for the phenotypic assays. The antibiotic susceptibility profiles were based on the epidemiological cut-off values (ECOFF), as previously described [13]. The human isolates consist of both screening and clinical isolates and were obtained using a similar methodology as described above. All VREfm isolates from all wellbeing centers in Finland are sent to the central laboratory, where all confirmed isolates are collated and sequenced. No routine phenotypic antibiotic susceptibility testing is conducted on the isolates. Hence, we could not compare phenotypic resistance profiles. The human VREfm sequences were obtained from samples obtained between 2016–2023.

Whole genome sequencing of vancomycin-resistant E. faecium isolated from wastewater influents

The total DNA of the isolates was extracted using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) in a QIAcube Connect instrument (Qiagen, Hilden, Germany) following the manufacturer’s instructions. The purified DNA was quantified with a Qubit 4 Fluorometer (Invitrogen, Singapore). DNA sequencing was outsourced to a commercial sequencing company (Novogene GmbH, Munich, Germany). Pooled DNA samples from each isolate were sequenced on the Illumina MiSeq platform using the 2 × 150 paired-end read approach (Illumina Inc., San Diego, CA, USA). The raw sequencing read output had its adaptor trimmed using Trimmomatic *v.*0.36, and the quality of the reads was assessed using the FastQC tool *v.*0.11.9. The raw reads were de novo assembled into contigs using Velvet v1.1.04.

The raw sequences have been deposited at the European Nucleotide Archive (BioProject PRJEB78611) with the individual accession numbers listed in Supplementary Table S3. The assembled contigs were fed into the Bruker MBioSEQ Ridom Typer v.11.0.5 (Ridom GmbH, Münster, Germany). This bacterial analytical pipeline utilized the NCBI AMR Finder plus tool *v.*4.0.15 at the default threshold of 90% identity and 60% minimum length to identify resistance genes. The CGE analytical pipeline was used for the identification of the plasmid replicons (PlasmidFinder tool v. 2.1) using the recommended settings. The quality metrics for the 17 WW sequences are shown in Supplementary Table S4.

Comparison of wastewater and human vancomycin-resistant E. faecium isolates

Raw reads from the wastewater VREfm isolates were analysed with Bruker MBioSEQ Ridom Typer *v.*11.0.5 to perform a core genome MLST (cgMLST) analysis in which the WW isolates (n = 17) were compared with all human VREfm isolates from national VRE surveillance (2016–2023, n = 754). Raw reads from the WW and human isolates were de novo assembled using the Velvet algorithm v. 1.1.04, and the public cgMLST scheme for E. faecium was used with default parameters.

Sequence data from human VREFm isolates were obtained from the national infectious diseases register. Finnish clinical microbiology laboratories are required to notify all VRE findings from clinical and screening specimens to the National Infectious Diseases Register (NIDR) and send corresponding isolates to the Finnish Institute for Health and Welfare (THL) for further characterization. Whole-genome sequencing has been performed on all VRE isolates sent to THL since 2016.

Data analysis

Data were presented as proportions. The differences in the VREfm isolation rate between the ten WWTPs were tested using logistic regression with Bonferroni correction and adjusted for population size using SPSS v.29®.

Results

Isolation and phenotypic susceptibility assay

The isolation rate for VREfm was 22% (n = 17/77) (Table S5). Other Enterococcus strains detected in the untreated sewage include E. avium (n = 2), E. faecalis (n = 2), and E. gallinarum (n = 3). There were no significant differences in the isolation rate of VREfm across the ten WWTPs (p = 0.407). In terms of counts, more isolates were retrieved from Turku (n = 6/8). No VREfm was isolated from the sewage collected from Espoo or Rovaniemi. The phenotypic antibiotic susceptibility testing (AST) revealed that all isolates were resistant to ampicillin, teicoplanin, vancomycin, erythromycin, ciprofloxacin, and had significant resistance to gentamycin (Table 1). All isolates were, however, susceptible to chloramphenicol, daptomycin, and linezolid (Table S6).

Table 1. Phenotypic antibiotic susceptibility profiles of vancomycin-resistant Enterococcus faecium isolated from wastewater (n = 17)^*^The upper range for gentamycin in the EUVENC Sensititre plate is 1024 mg/L; ECOFF—Epidemiological cut-off; Synercid—Quinupristin and dalfopristin. Bold vertical lines indicate the ECOFF values for resistance for E. faecium as of 28.5.2024. The ECOFF is established by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and can be accessed here: https://mic.eucast.org/search/. Grey fields denote the range of dilutions tested for each substance. Values above the range denote MIC values greater than the highest concentration in the range. MICs equal to or lower than the lowest concentration tested are given as the lowest concentration

Genotypic profile of wastewater vancomycin-resistant E. faecium isolates

All of the 17 VREfm isolates belonged to the CC17 clonal complex, an ancestral clone of the hospital-associated clade A1 [18]. Of these, 12 isolates belonged to the epidemic-prone ST80 variant. Other STs were 787 (n = 2), 78 (n = 1), 117 (n = 1), and 612 (n = 1). Most of the isolates harboured the vanA gene (76.5%), while three isolates harboured the vanB gene (17.6%), and sample VR69 harboured both the vanA and vanB genes. There was a wide diversity of antibiotic resistance genes (ARGs), with the aminoglycoside resistance gene (aac(6')-I), macrolide resistance gene (msr(C)), eat(A), and van gene clusters being conserved in all the WW isolates (Table 2). B-lactam and fluoroquinolone resistance was mediated by chromosomal point mutations in the pbp5 and gyrA p.S82Y and parC p.S80I genes, respectively (Supplementary Table S3).Table 2. Diversity of antibiotic resistance genes and plasmids in wastewater VREfm isolates (n = 17)ARGN (%)Plasmid repliconsN (%)aac(6')-I17 (100)repUS1517 (100)vanA/vanH-A/vanR17 (100)rep211 (64.7)msr(C)17 (100)rep18b11 (64.7)eat(A)17 (100)repUS1210 (58.8)aph(3')-IIIa11 (64.7)repUS4310 (58.8)erm(B)10 (58.8)rep178 (47)dfrG9 (52.9)rep18a7 (41.2)tet(M)6 (35.3)rep116 (35.3)sat44 (23.5)rep146 (35.3)aph(2'')-Ia3 (17.6)repUS14 (23.5)erm(T)2 (11.8)rep292 (11.8)

Core genome MLST of human and wastewater vancomycin-resistant E. faecium

cgMLST revealed that WW contained some of the most prevalent STs of VREfm in circulation in Finland. Our study demonstrated the clonal spread of VREfm in Finland, with three clusters detected among the WW isolates (Fig. 1a). The distance matrix for WW isolates is shown in Supplementary Table S7. When compared with human isolates, WW isolates formed eight clusters. Sample VR40 (ST612, CT1026 from Turku) clustered with 16 human isolates, with one to ten allelic differences between the isolates. The closest isolate was from the wellbeing services county of Southwest Finland, where the city of Turku is located (Fig. 1b). The two ST787 isolates retrieved from Southwest Finland and Pirkanmaa were identical and clustered with 9 human isolates that were mostly obtained from the Pirkanmaa region (Fig. 1c). VR62-ST117/CT5630 and VR72-ST80/CT3302 (two of the three samples from the Helsinki region) and VR38-ST78/CT8498 did not cluster with any human isolates, suggesting possible introduction through international (trans-border) travel. The only WW ST80:CT1470 strain was genomically indistinguishable from several human samples obtained from the same region (Kuopio or North Savo) and had only a few (0 to 6) allelic differences when compared with 44 other human isolates (Fig. 1d). In the same vein, the four WW ST80:CT2046 strains isolated from Turku and Helsinki clustered with 100 human isolates from Southwest Finland (Fig. 1e) with a maximum of 3 allelic differences. The three sequences from Oulu (ST80:CT2099/6196) clustered with human isolates obtained in 2020 from the same region (wellbeing services county of North Ostrobothnia) (Fig. 1f). The only isolate (VR69-ST80/CT2262) that harboured both the vanA and vanB genes was related and clustered with human isolates that harboured either vanA or vanB but not both from Kymenlaakso, North Savo, South Karelia, and Päijät-Häme wellbeing counties (Fig. 1g). The WW sample VR30 (obtained from Turku in May 2021) was identical to one human sample that was isolated in 2019 in the wellbeing services county of Southwest Finland. Another isolate, VR46 (obtained from Pietarsaari in August 2021), was identical to one human sample (no allelic difference) that was isolated in 2021 in the wellbeing services county of Ostrobothnia.Fig. 1a Minimum spanning tree showing the relatedness of wastewater vancomycin-resistant E. faecium (n = 17). The Bruker MBioSEQ Ridom Typer MST was based on 1260 genes using a cluster distance threshold of 10. The numbers indicate the allelic differences between the genomes (indicating genetic relatedness. b-g Minimum spanning tree showing the relatedness of wastewater vancomycin-resistant E. faecium to human isolates across Finland. The Bruker MBioSEQ Ridom Typer MST was based on 1260 genes using a cluster distance threshold of 10. CT- Cluster type 7; ST – Sequence type; WW- wastewater. The red digits indicate the allelic differences between the genomes (indicating genetic relatedness)

Discussion

Here, we report the detection and genomic analysis of VREfm strains in WW influent in Finland. To be useful as a preparedness tool and be a complementary source of data for public health decision-making, we used cgMLST to compare human and WW VREfm sequences. The core genomes were closely related, highlighting the importance of WBS as a sensitive tool for population-level surveillance.

Vancomycin-resistant Enterococcus faecium (VREfm) remains a World Health Organization critical priority pathogen, predominantly of human origin and strongly associated with hospital environments [21–24]. Our findings reveal that VREfm in municipal wastewater (WW) exhibits multidrug resistance profiles consistent with those reported across Europe [3]. Studies have similarly documented high resistance to ampicillin [22, 25, 26], gentamicins [27], erythromycin [22, 28], teicoplanin [22], ciprofloxacin [29], synercid [22] and tigycycline [22] in both clinical and environmental isolates. The intrinsic resistance of Enterococcus spp. to aminoglycosides likely arises from their robust cell wall structure [21]. Consistent with our results, European Antimicrobial Resistance Surveillance Network (EARS-NET) data encompassing over 100,000 E. faecalis and E. faecium isolates show low prevalence of resistance to linezolid and daptomycin [3, 30, 31]. Comparable susceptibility patterns were also observed in WW VREfm isolates from Brno, Czech Republic [22].

The high phenotypic resistance to macrolides and synercid (quinupristin/dalfopristin) observed in this study likely results from carriage of the msrC and eatA genes, both members of the ABC-F family of ribosomal protection proteins [32–35]. Previous work has linked van gene-mediated vancomycin resistance with collateral sensitivity to pleuromutilin antibiotics through epistatic interactions between van clusters and msrC [33, 35]. However, evidence remains limited regarding whether similar mechanisms contribute to reduced susceptibility to linezolid or daptomycin in environmental isolates.

The repertoire of resistance genes was similar to those previously reported [20, 23, 24, 36],, and one of these studies detected 28 antibiotic resistance genes in the hospital-adapted clade, of which 23 were represented in bloodstream, hospital sewage, and municipal WW isolates [20]. Data from the metagenomic analysis of the Global Sewage Surveillance Project revealed that Enterococcuswas assigned the largest number of ARGs, and mainly shared these with other known Gram-positive species [37]. This largely confirms observations from human isolates and suggests that the current choice of Escherichia coli and Enterococcusas indicator species for surveillance of AMR is appropriate [37, 38].

Vancomycin resistance in E. faecium is primarily mediated by the vanA or vanB gene clusters located on the Tn1546 transposon. Our findings, consistent with national surveillance data, show a predominance of vanA-positive isolates circulating during 2021–2022 [10]. These clusters typically contain nine genes: orf1 and orf2 for transposition; vanR and vanS for signal transduction; vanH, vanA, vanX, and vanY for vancomycin resistance; and vanZ for teicoplanin resistance [39]. Expression of the vanH/vanA/vanX cassette, under control of the vanR/vanS regulatory system, is essential for the resistant phenotype [40, 41]. Globally, vanA predominates except in Australia, where vanB is more common, a trend mirrored in our data [41].

The importance of genomics in identifying or tracing outbreaks of nosocomial pathogens such as VREfm cannot be overemphasized. Recently, more studies have demonstrated that clustering according to the core genome multilocus sequence type (cgMLST) for VREfm is more informative of the population structure than traditional MLST [42]. cgMLST revealed that WW VREfm isolates clustered closely with human VREfm from most of the wellbeing service counties in Finland, with some isolates displaying no allelic differences. This is in accordance with a study in the UK in which VREfm was isolated from all 20 sampled WW treatment plants, with highly related isolates shared between a major teaching hospital in the east of England and nine WWTPs [20]. The study concluded that the widespread distribution of hospital-adapted VREfm beyond acute healthcare settings was consistent with the extensive release of VREfm into the environment in the east of England [20]. Our cgMLST revealed that WBE can be useful in detecting the population-level spread of VREfm. It could also help detect the introduction of a new strain or changing dynamics among already established strains.

Our study also revealed that WW surveillance could detect the same proportions of vanA/B, ST, and circulating strains as the human samples. The culture-based WBS approach identified three of the most common VREfm outbreak-associated STs in Finland (ST80, ST612, and ST787). The epidemic-prone ST80 strains have been reported to be the most prevalent ST in Sweden [43], Denmark [44], Germany [45, 46], Ireland [47, 48], and across the globe. ST612 is not globally distributed, as only a few studies have reported this ST [49–52], and ST787 has also been reported in Ireland [47]. Of the three WW isolates that did not cluster with any human isolates, two were from Helsinki and were more likely to be newly introduced into the country or could be an emerging strain. Hence, they should be monitored over the next few years.

Although Enterococcus spp. are common gut commensals, selective culture methods can capture clinical and resistant strains in WW. While treatment processes reduce bacterial load, ARGs frequently persist post-treatment [20]. The detection of human-adapted CC17 lineages in WW is of public health concern, as ARGs and resistant bacteria may disseminate into surface waters, agricultural systems, and food chains, facilitating reintroduction into human and animal populations [21]. These findings highlight the importance of adopting a One Health framework to elucidate AMR transmission dynamics and guide effective intervention strategies [1].

The primary limitation of this study was the small number of VREfm isolates analyzed. Also, the intra-sample diversity may have been underestimated due to single-colony selection, potentially missing low-abundance or viable-but-non-culturable (VBNC) strains. Future studies could mitigate this through targeted enrichment and metagenomic sequencing. Additionally, sporadic bacterial release, environmental acquisition of ARGs, and non-human sources may influence VREfm prevalence in WW, limiting generalizability. Despite these constraints, our results underscore the utility of WW surveillance in AMR monitoring and provide a valuable framework for future longitudinal studies.

Conclusion

This pilot study reports the detection and relatedness of human and WW VREfm isolates from the influents of ten WWTPs in Finland. The isolates expressed phenotypic multidrug-resistant profiles, and there was high concordance between phenotypic and genotypic profiles. cgMLST revealed that outbreak-associated VREfm strains from WW clustered closely with those from human isolates across the country. We believe this study supports the utility of WBS as a tool for AMR surveillance. To limit the spread of VREfm, it is essential to enforce strict infection prevention and control (IPC) as well as biosecurity measures and optimize antibiotic usage in humans and animals through AMR stewardship programs. In the future, we need to develop tools to be able to simultaneously capture and characterize multiple pathogens and isolates from WW samples.

Supplementary Information

Supplementary Material 1.