Antimicrobial resistance, conjugative plasmids and pathogenicity in wastewater and freshwater Escherichia spp. in Stockholm, Sweden

Abstract

To investigate antibiotic resistance genes (ARGs), conjugative plasmids, and virulence genes in Swedish waterborne Escherichia spp., water samples were collected from urban freshwater and Baltic Sea beaches and a primary wastewater treatment plant (WWTP) in Stockholm, Sweden. During the summer of 2022, 68 isolates were recovered using ESBL-selective and non-selective agar, including 40 from wastewater and 28 from fresh or brackish water. Isolates were characterised by phenotypic antibiotic susceptibility testing, conjugation assays, and whole-genome sequencing. Antibiotic residues were quantified, with higher concentrations detected at WWTP inlets and outlets than in natural water sources. Overall, 28 isolates (41.17%) were phenotypically multidrug-resistant (MDR), and 18 (26.47%) carried ≥3 ARGs. WWTP-derived isolates showed a significantly higher prevalence of extended-spectrum β-lactamase (ESBL) genes than freshwater isolates (p < 0.0001). Isolates represented diverse multilocus sequence types (MLST), and most harboured ≥1 plasmid. Sixteen strains transferred conjugative plasmids encoding resistance to cefotaxime, tetracycline, streptomycin, and trimethoprim. ESBL genes, including bla_CTX-M-15, were located on IncF, IncN, IncB/O/K/Z, and IncI plasmids, with IncF plasmids showing lower transfer frequencies than IncN and IncI1 conjugative plasmids. These findings identify WWTPs as a major source of MDR and pathogenic E. coli, highlighting their role in environmental dissemination.

Introduction

Antimicrobial resistance (AMR) in bacteria is recognised as a critical and escalating global health threat^1–3. The incidence of antibiotic-resistant bacteria (ARB) and multidrug-resistant (MDR) bacteria has increased due to overuse and misuse of antibiotics in healthcare, animal husbandry, industrial waste, and inadequate treatment of diseases^4,5. Unsafe drinking water sources can cause illness in humans and animals upon consumption, and contaminated or inadequately purified environmental waters and wastewaters may contribute to spread and exposure to microbial pathogens^6–9. The spread of AMR in pathogenic species has hence emerged as a significant global public health issue due to lack of treatment options.

The increased levels of antibiotic residues found in wastewater, and particularly untreated hospital wastewater, may provide the necessary selective pressure for the retention and spread of antibiotic resistance genes (ARGs)^10,11. Additionally, the mixture of environmental and pathogenic strains in wastewater treatment plants (WWTP) may allow for the exchange, transfer and/or acquisition of ARGs. The exchange of ARGs between environmental strains and pathogenic strains of clinical importance poses a threat for human health.

Escherichia coli (E. coli) is typically used as an indicator of faecal pollution in water sources since it is an almost universal commensal resident in the intestines of warm-blooded animals, including humans. Although commensal E. coli live in symbiosis with their hosts, pathogenic subspecies of E. coli harbouring virulence genes are common causes of intestinal diarrheagenic (DEC) and extraintestinal infections (ExPEC) in humans and animals¹². DEC are often water- or foodborne pathogens with increased prevalence in natural waters during heavy precipitation and flooding in endemic areas^6,13. ExPEC are often recovered from sewage, WWTPs or downstream of WWTPs in rivers^14–17. E. coli is also often studied for the presence of ARGs since it is easily isolated and can transfer ARGs through horizontal gene transfer as well as through distribution in anthropogenically influenced water systems, including agricultural areas, urban sewage, and WWTPs.

The growing concern of the emergence of AMR in pathogenic water-borne E. coli has resulted in several studies determining the prevalence of MDR bacteria in aquatic environments^15,16,18. Several studies have established higher levels of ARGs, including extended-spectrum β-lactamases (ESBL), in E. coli and other species collected from wastewater samples compared to environmental water isolates^5,16,19. Although hospital sewage and WWTPs are considered hotspots for the presence of ARB and ARGs¹⁸, the environment is a significant reservoir and vehicle in zoonotic infections and emerging human pathogens as well as ARB²⁰. Antibiotics from urban and agricultural sources persist in soil and water as well as in wastewater and may drive horizontal gene transfer and development of new ARGs and ARB.

The emergence of ESBL-producing strains has been recognised since the 1990s, when mostly TEM/SHV beta-lactamases were described together with emerging bla_CTX-M-2²¹. Since 2000, a dramatic shift in the prevalence of particularly bla_CTX-M-15 has been reported in E. coli, where the global expanse is probably mediated by the uropathogenic clone ST131 that predominantly carries bla_CTX-M-15²². Wastewater sample studies frequently isolate MDR and ESBL carrying isolates belonging to the ExPEC pathotype. The isolates frequently detected include clonal groups associated with urinary tract infections (UTI) and bloodstream infections (BSI) such as ST131, ST648, ST69, ST95, ST410, ST127, and ST38^17,23–25.

The present study aimed to investigate the presence of pathogenic and/or MDR E. coli isolates and antibiotic residues present in freshwater, brackish water and wastewater sites in Stockholm, Sweden as well as characterise their ability to transfer mobile genetic elements carrying ARGs.

The results show that MDR bacteria with conjugative plasmids carrying ARGs, as well as pathogenic bacteria, were present at multiple sites, which highlights the importance of surveillance and studies of the transmission of ARB and ARGs from the environment back to humans.

Results

Genomic analysis of isolates revealed presence of diverse Escherichia species across WWTP inlets/outlets and natural water sources in Stockholm, Sweden

A total of 68 Escherichia spp. isolates were collected from multiple locations (fresh, brackish and wastewater), whereof 40 were wastewater treatment plant (WWTP) isolates (20 isolated from ESBL selective agar, 20 from non-selective agar) and 28 isolates from fresh/brackish water (all isolated from non-selective agar, no colonies were found on ESBL selective agar) (Fig. 1, Table 1 and Supplementary Fig. 1). All isolates were analysed for phenotypic antimicrobial resistance (AMR) by disc diffusion analyses and whole genome sequenced (WGS) followed by analyses for multi-locus sequence type (MLST), antibiotic resistance genes (ARGs), presence of plasmids and virulence genes (Fig. 1, Table 1 and Supplementary Table. 1).

Fig. 1. Tree of kmer distance of all Escherichia spp. isolates from Stockholm indicating isolation site (WWTP vs fresh/brackish water), plasmid replicons, phenotypic antibiotic-resistance and presence of antibiotic resistance genes (ARGs).

The first column represents the isolate number. The second column indicates multi-locus sequence type (MLST). The third column indicates isolation source, where the filled squares indicate isolation from WWTP compared to fresh/brackish water as empty squares. The fourth column indicates presence/absence of plasmid replicons belonging to the plasmid incompatibility groups. The fifth column indicates presence/absence of phenotypic resistance. The sixth and final column indicates presence/absence of antibiotic resistance genes (ARGs). The MLST was determined through the Achtman Scheme; Plasmid Replicon Groups were determined using CGE pipeline PlasmidFinder; Phenotypic Resistance was determined through the disc diffusion method; and the ARGs were determined using CGE pipeline ResFinder. The phylogenetic tree was produced using Mashtree and annotations were added in iTOL.

Table 1.

MLST, total antibiotic resistance gene (ARG) and plasmid content of all 69 Escherichia spp. isolates described in this study

Isolate	MLST	Plasmids	Total n° ARGs	ARGs								Accession Number
				Aminoglycosides	β-lactams	Tetra-cyclines	Sulfon-amides	Quinolones	Macrolides	Trime-thoprims	Phenolics
1. SE-N-W-Ec10	10980	IncFIB(pHCM2), IncFIB(AP001918), IncI1, Col(MG828)										SAMN48681364
2. SE-HDO-W-EcE4	127	IncFIB(AP001918), IncFIA, IncFII(pRSB107)	8	aph(3”)-lb, aph(6”)-ld	blaCTX-M-15, blaTEM-1B	tet(B)	sul2		mph(A)	dfrA14		SAMN48681365
4. SE-N-S-Ec1	3247	IncFIC(FII), IncFIB(AP001918)										SAMN48681366
5. SE-SI-W-Ec1	404	IncFIB(AP001918), IncFII(29)_1_pUTI89, Col156	1							dfrA1		SAMN48681367
6. SE-R-W-Ec1	6496	IncFII(pCoo), IncFIB(AP001918), pEC4115										SAMN48681368
7. SE-HDI-W-Ec5	607	IncFIB, IncY, Col440I										SAMN48681369
8. SE-HDI-W-Ec2	710	IncFII, IncFIB(K)_1_1_Kpn3, ColRNAI, Col4401, IncR										SAMN48681370
9. SE-HDO-W-EcE5 (p)	38	ColRNAI	2		blaCTX-M-15			qnrS1				SAMN48681371
10. SE-HDI-W-Ec8 (p)	607	IncY, IncFIA(HI1), Col440I, IncR										SAMN48681372
11. SE-H-S-Ec1-1 (p)	133											SAMN48681373
12. SE-SI-W-Ec8 (p)	357	IncFIC(FII)										SAMN48681374
13. SE-HD-W-EcE6 (p)	131	IncFIB(AP001918), IncFII(pRSB107), IncFIA, Col156	9	aph(3”)-lb, aph(6”)-ld, aadA5	blaCTX-M-27	tet(A)	sul1, sul2		mph(A)	dfrA17		SAMN48681375
14. SE-SI-W-EcE2 (p)	3873	IncFII, IncBOKZ, IncI2, IncFIB(AP001918), ColRNAI, ColpVC	7	aadA5	blaCTX-M-15	tet(B)	sul1	qnrS1	mph(A)	dfrA18		SAMN48681376
15. SE-SI-W-EcE5 (p)	38	IncI2, IncFII(pCoo), ColRNAI, Col(MG828), ColpVC	3		blaCTX-M-15	tet(B)		qnrS1				SAMN48681377
16. SE-R-S-Ec3 (p)	2026	IncFIB(AP001918)										SAMN48681378
17. SE-HD-W-Ec7 (p)	196											SAMN48681379
18. SE-HDI-W-EcE1 (p)	683	IncFIB(AP001918), IncN, ColRNAI	6		blaCTX-M-15, blaTEM-1B	tet(A)		qnrS1		dfrA14	floR_2	SAMN48681380
19. SE-HD-W-Ec1 (p)	472	IncHI1A, IncHI1B(R27), Col(Ye4449), ColRNAI, Col4401										SAMN48681381
20. SE-SI-W-EcE8 (p)	38	IncFIB(AP001918), IncFIB(K)_1_Kpn3, IncI1, IncX3, ColRNAI, Col8282, Col(MG828), Col440II	3		blaCTX-M-15, blaSHV12			qnrS1				SAMN48681382
22. SE-N-W-Ec1 (p)	2521	IncFIA, IncFIB(AP001918), p0111										SAMN48681383
23. SE-HDO-W-EcE1 (p)	7326	IncFII(pCoo), IncFIB(pB171), IncFII	6	aadA5	blaCTX-M-15		sul1	qnrS1	mph(A)	dfrA17		SAMN48681384
35. SE-N-W-Ec1 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681385
36. SE-N-W-E2 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681386
37. SE-N-W-Ec3 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681387
38. SE-N-W-Ec4 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681388
39. SE-N-W-Ec5 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681389
40. SE-N-W-Ec6 COL-R	13281	IncFIC(FII), IncFIB(AP001918)										SAMN48681390
41. SE-HDI-W-EcE2	6872	IncI, IncFIB(AP001918), IncFII	6	aph(3”)-lb, aph(6”)-ld	blaCTX-M-15		sul2	qnrS1		dfrA8		SAMN48681391
42. SE-HDI-W-EcE3 (G4)	10	IncI1, IncI2, IncFIB(AP001918), IncFII, IncFII(pCoo)	1		blaCTX-M-1							SAMN48681392
43. SE-HDI-W-EcE4	38		3	ant(3”)-la	blaCTX-M-14					dfrA1		SAMN48681393
44. SE-HDI-W-EcE5	38	ColRNAI	2		blaCTX-M-15			qnrS1				SAMN48681394
45. SE-HDI-W-EcE6	131	IncFIA, IncFIB(AP001918), IncFII(pRSB107), Col156	9	aph(3”)-lb, aph(6”)-ld, aadA5	blaCTX-M-27	tet(A)	sul1, sul2		mph(A)	dfrA17		SAMN48681395
46. SE-HDI-W-EcE7	58	IncN, IncFII, IncFIB(AP001918)	3		blaCTX-M-15			qnrS1		dfrA14		SAMN48681396
47. SE-SI-W-EcE3	1442	IncFIB(AP001918), IncFIC(FII), IncI1, Col8282, ColRNAI	1		blaCTX-M-3							SAMN48681397
48. SE-SI-W-EcE4	636	IncFIA, IncFIB(AP001918), IncFII(pRSB107), Col440II, Col(MG828), Col8282, ColRNAI, Col(BS512)	6	aph(3”)-lb, aph(6”)-ld, ant(3”)-la	blaCTX-M-15		sul2			dfrA1		SAMN48681398
49. SE-SI-W-EcE1	2522	IncN	3		blaCTX-M-15			qnrS1		dfrA14		SAMN48681399
50. SE-SI-W-EcE6	38	IncFII, IncP1, Col156, Col8282	2		blaCTX-M-15			qnrS1				SAMN48681400
51. SE-SI-W-EcE7	18204	IncFIB(K)_1_Kpn3, IncFIB(AP001918)	8	aph(3”)-lb, aph(6”)-ld	blaCTX-M-15, blaTEM-1B	tet(A)	sul2	qnrS1		dfrA14		SAMN48681401
52. SE-HDO-W-EcE8	131	IncFIB(AP001918), IncFII(29)_1_pUTI89, Col156	10	aph(3”)-lb, aph(6”)-ld, aadA5	blaCTX-M-27, blaTEM-1B	tet(A)	sul1, sul2		mph(A)	dfrA17		SAMN48681402
53. SE-H-S-Ec1	133											SAMN48681403
54. SE-H-S-Tc-CTX2	2026	IncFIB(AP001918)										SAMN48681404
55. SE-H-S-Tc-CTX3	133											SAMN48681405
56. 1SE-N-W-Ec2	10980	IncFIB(pHCM2), IncFIB(AP001918), IncI1, Col(MG828)										SAMN48681406
57. 1SE-N-W-Ec11	10980	IncFIB(pHCM2), IncFIB(AP001918), IncI1, Col(MG828)										SAMN48681407
63. SE-HDI-W-Ec1	607	IncFIA(HI1), IncY, IncX1, IncR, IncHI1A, IncHI1B(R27), pENTAS02										SAMN48681409
64. SE-HDI-W-Ec3	164											SAMN48681410
65. SE-HDI-W-Ec4	18202	IncFII, Col440I, Col440II, ColRNAI, Col8282, Col(MG828)	5	aph(3”)-lb, aph(6”)-ld	blaCTX-M-15	tet(B)		qnrS1				SAMN48681411
66. SE-HDI-W-Ec6	607	IncFIB(pQil), IncY, Col440I										SAMN48681412
67. SE-HDI-W-Ec7	10	IncFIA, IncFIB(AP001918), IncFII(pRSB107), Col156, ColRNAI	6	aph(3”)-lb, aadA5			sul1, sul2		mph(A)	dfrA17		SAMN48681413
68. SE-R1-W-Ec2	6496	IncFIB(AP001918), IncFII(pCoo), pEC4115										SAMN48681414
71. SE-R1-S-Ec4	362	IncFII(pCoo), IncI1, ColRNAI										SAMN48681415
72. SE-H-S-Ec1-1	133											SAMN48681416
73. SE-HDO-Ec7	196											SAMN48681417
74. SE-HDO-Ec8	607	IncY, Col440I										SAMN48681418
75. 1SE-T-W-Ec1	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681419
76. 1SE-T-W-Ec2	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681420
77. 1SE-T-W-Ec3	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681421
78. 1SE-N-S-Ec1	3247	IncFIB(AP001918), IncFIC(FII)										SAMN48681422
79. 1SE-T-W-Ec4	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681423
80. 1SE-T-W-Ec5	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681424
81. 1SE-T-W-Ec6	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681425
82. 1SE-T-W-Ec7	351	IncFIA, IncFIB(AP001918), IncFIC(FII), IncX1, IncX3, IncY	1			tet(A)						SAMN48681426
83. SE-HDO-W-Ec3	635	IncFIB(K)_1_Kpn3, Col440I										SAMN48681427
84. SE-HDO-W-Ec4	34	IncFIB(pB171), IncFII(pHN7A8), Col8282, Col440I	5	aph(3”)-lb, aph(6”)-ld	blaTEM-1B	tet(B)	sul2					SAMN48681428
86. SE-HDO-W-Ec6	95	IncFIB(AP001918), IncFII(29)_1_pUTI89, Col156										SAMN48681429
88. SE-SI-W-Ec3	1567	IncFIA, IncFIB(AP001918), Col440I										SAMN48681430
89. SE-SI-W-Ec4	108											SAMN48681431
90. SE-SI-W-Ec5	607	IncY, IncX1, IncFIA(HI1), pENTAS02										SAMN48681432

The first column describes the isolate (the number which the isolate is referred to as throughout the paper, and the code which describes its origin where SE= Sweden; the following letters correspond to isolation source (Supplementary Fig. 1): Nynäshamn (N), Rålambshovsparken (R), Brunnsviken in Hagaparken (H), Sickla Inlet (SI), Henriksdal Inlet (HDI/HD), Henriksdal Outlet (HDO), Tyresö Fiskarholmen (T); followed by sample type: Water (W) or Sediment (S); and lastly, the isolate number on the plate where Ec isolates originate from non-selective plates and EcE isolates originate from ESBL selective plates); the second column shows the Multi-locus sequence type (MLST); the third column shows all plasmids present in the isolate; the fourth column is a summary of the total number of ARGs found in the isolate; and columns 5–12 indicate the specific ARGs found in the isolate which cause resistance to the different indicated antibiotic classes.

All isolates were typed by MLST using the Achtman system²⁶ and a phylogenetic tree was generated (Fig. 1). The MLST typing revealed different ST groups present in the different environments, with no overlap of MLSTs between WWTP and fresh/brackish water isolates.

For fresh/brackish water isolates, the 28 isolates belonged to 11 different MLSTs: ST10980, ST3247, ST6496, ST133, ST2026, ST2521, ST13281, ST6496, ST362, ST351 and ST3247 (Fig. 1, Table 1) where the most frequent MLSTs were ST351 (25%, 7/28) and ST13281 (21.4%, 6/28). Six isolates (ST133 and ST6496) branched out and were identified as Escherichia marmotae, previously designated as Escherichia coli (E. coli) clade V (two from Rålambshovsparken and four from Brunnsviken) (Fig. 1 and Supplementary Fig. 1). Three of these isolates were phenotypically MDR, however the corresponding resistance genes were not identified. The 40 WWTP isolates belonged to 24 different MLSTs (Fig. 1 and Table 1), where the most frequent MLSTs were ST607 (15%, 6/40) and ST38 (15%, 6/40). In addition, three WWTP isolates belonged to ST131, whereof two were isolated from the inlet and one from the outlet.

To determine the plasmid content, the WGS data was analysed by PlasmidFinder. The isolates carried plasmid replicons belonging to Incompatibility (Inc) groups: IncFII, IncFIA, IncFIA(HI1), IncFIB(AP001918), IncFIB(pQil), IncFIBp(pHCM2), IncFIB(K)_1_Kpn3, IncFIB(pB171), IncFIC(FII), IncFII, IncFII(pHN7A8), IncFII(pRSB107), IncFII(pCoo), IncFII(29)_1_pUTI89, IncBOKZ, IncI1, IncI2, IncN, IncR, IncY, IncX1, IncX3, IncH1A, IncHI1B(R27), Col440I, Col440II, ColpVC, Col(Ye4449), p0111, pENTAS02, Col(Mg828), Col8282, Col156, ColRNAI, pEC4115 and Col(BS512) (Fig. 1, Table 1).

WWTP inlet and outlet samples harbour a higher prevalence of ESBL E. coli isolates compared to natural water sources in Stockholm, Sweden

AMR profiles determined by disc diffusion showed that 41.17% (28/68) isolates were phenotypically resistant to ≥3 antibiotics classes (10 classes tested). Of the 68 isolates, 26.47% (18/68) carried ≥3 ARGs. All isolates contained the intrinsic chromosomal mdf(A) ARG, however, 52.9% (36/68) of the isolates solely carried mdf(A), the majority of these (52.77%, 19/36) came from fresh/brackish water sites collected close to beaches and recreational areas (Fig. 1, Table 1 and Supplementary Fig. 1). Mdf(A) is a membrane transporter which confers tolerance to alkaline conditions and resistance to several antibiotics²⁷.

Of the E. coli isolated from WWTP samples (inlet and outlet), 52.5% (21/40) were phenotypically resistant to ≥3 antibiotics, and 45% (18/40) carried ≥3 ARGs (Fig. 1 and Table 1). Of the WWTP isolates, 55% (22/40) were extended-spectrum β-lactamases (ESBL)-carriers, where the most predominantly isolated ARG was bla_CTX-M-15 (68.2%, 15/22), which was either chromosomal or carried on IncN, IncI and IncF plasmids (Table 1). In 22.7% (5/22) of the isolates carrying bla_CTX-M ARGs, there was co-carriage of bla_CTX-M and either bla_SHV or bla_TEM. Overall, the most predominantly found ARGs from WWTP isolates were bla_CTX-M-15 (37.5%, 15/40), qnrS1 (35%, 14/40), aph(3”)-lb (25%, 10/40) and aph(6”)—ld (25%, 10/40).

Moreover, the three ST131 wastewater isolates (13, 45, 52) harboured IncF and col156 plasmids. ST131 isolates are a global MDR uropathogenic clonal group which frequently carry bla_CTX-M genes and the Stockholm ST131 isolates carried the highest number of ARGs ( ≥ 9) among the isolates in this study. All ST131 isolates, apart from mdf(A), which is intrinsic in E. coli, also carried aadA5, aph(3”)-lb, aph(6)-ld, bla_CTX-M-27, dfrA17, mph(A), sul1, sul2 and tet(A) (Fig. 1 and Table 1). Isolate 52 also carried bla_TEM-1B and was the isolate with most ARGs (10) identified in this study. Isolate 2 (ST127), carried the third highest number of ARGs (8) including mdf(A), bla_CTX-M-15, sul2, aph(3”)-Ib, aph(6)-Id, bla_TEM-1B, tet(B), mph(A) and dfrA14 (Fig. 1 and Table 1).

In contrast, no ESBL-carrying E. coli were identified in the fresh/brackish water samples. However, 25% (7/28) of fresh/brackish water isolates were phenotypically MDR, where the predominant ARGs found were mdf(A) (100%) and tet(A) (25%, 7/28) (Fig. 1, Table 1 and Supplementary Fig. 2). The analysed wastewater isolates were significantly more often ESBL positive (p > 0.0001, Fisher’s exact test) compared to fresh/brackish water isolates, however, there was no significant difference in phenotypic MDR between the groups and it is important to note that total number of colonies on ESBL and non- selective plates, respectively, were not recorded in this study.

The bla_CTX-M-15 gene was the most common ESBL and the most predominantly found ARGs in wastewater were bla_CTX-M-15 and qnrS1 (Supplementary Fig. 2). Furthermore, the co-occurrence of bla_CTX-M-15 and qnrS1 was the most common (Supplementary Fig. 2). bla_CTX-M-15 has been showed to be associated with qnr genes located close together on conjugative plasmids²⁸ which supports the findings of this study.

No carbapenem ARGs were found in the sequenced isolates, although, this does not entail that carbapenemases are not present at the sampling sites as further sampling would be required to accurately measure carbapenem ARG levels. However, one isolate (57) was phenotypically resistant to imipenem (IMP) (Fig. 1), although no bla_OXA or bla_CMY genes were found in the annotation. The isolate contained OmpC but lacked OmpF, which may be associated with imipenem resistance²⁹.

Levels of antibiotic residues at WWTP inlets and outlets are higher than in natural water sources in Stockholm, Sweden

Filtered water samples were analysed using UHPLC-MS/MS at ILVO (Belgium) to determine the levels of antibiotic residues³⁰. Twelve different residues were detected, including β-lactam antibiotics, quinolones, amphenicols, macrolides, sulphonamides, tetracyclines and rifaximin (Table 2). A maximum of ten different residues were identified in one sample. Low concentrations of antibiotics were detected in natural water sources, whereas the inlets and outlets of the Henriksdal WWTP contained higher levels up to 0.25 µg/L (Table 2). The highest concentrations were observed at the inlet of the Sickla WWTP and the outlet of the Henriksdal WWTP. A reduction in both the concentration and the number of antibiotics is noticeable between the Sickla inlet and the Henriksdal outlet, however, in Henriksdal outlet 1, the concentration of ciprofloxacin (CIP) is similar to that of Sickla inlet and exceeds the predicted no effect concentration³¹ (PNECs) for CIP (0.064 µg/L) (Table 2).

Table 2.

Antibiotic residue concentrations that were at least detected once in water samples collected from freshwater, brackish water and wastewater in Stockholm, Sweden

		Antibiotics residues detected (µg/L)
Location	Type of water	CIP	TYLV	CHL	CLA	SMX	SPY	TMP	TET	OTC	OFL	CFR	RIFX
Brunnsviken i Hagaparken	Freshwater (urban)			<0.01					0.058
Rålambshovsparken	Freshwater (urban)			<0.01					0.058
Karlbergskanalen	Freshwater (urban)	<0.1	0.047
Tyresö (Fiskarholmen)	Brackish water			<0.01
Nynäshamn	Brackish water			<0.01
Henriksdal outlet 1	Wastewater	1.1			0.038	0.064	0.245	0.087
Henriksdal outlet 2	Wastewater	0.015			0.04	0.081	0.179	0.108
Sickla inlet	Wastewater	1.335			0.038	0.345	0.566	0.12	0.298	<0.038	0.044	<0.019	<0.011

Ciprofloxacin (CIP), Tylvalosine (TYLV), Chloramphenicol (CHL), Clarithromycin (CLA), Sulfamethoxazole (SMX), Sulfapyridine (SPY), Trimethoprim (TMP), Tetracycline (TET), Oxytetracycline (OTC), Ofloxacin (OFL), Cefodroxil (CFR), Rifaximine (RIFX).

The results containing ”<” suggests the residue concentration detected is below the level of detection (LOD), the value shown is the LOD³⁰. Blank cells represent no detection of antibiotic residues.

Conjugative plasmids carrying ARGs are transferred at frequencies ranging from 10⁻² to 10⁻⁷

To determine if the ARG-carrying plasmids were conjugative, all isolates with phenotypic resistance to one or more of ciprofloxacin (CIP), tetracycline (TET), cefotaxime (CTX), streptomycin (STR) and trimethoprim (TMP) were tested (n = 39). The transfer of conjugative plasmids carrying ARGs from freshwater, brackish water and WWTP-derived E. coli donors to the recipient strain E. coli CV601 occurred at a frequency of 41.03%, where 16 of the 39 E. coli conjugated with CV601 (Table 3). Conjugation assays using CTX or TET as selective agents were successful, whereas those using CIP, TMP and STR did not yield transconjugants. However, when CTX or TET were used, the ARG for TMP (dfrA) was transferred along with the ARGs for CTX (bla_CTX-M) in isolates 18, 46 and 49; the ARGs for STR (aph(3”)-lb, aph(6”)-ld), bla_TEM and the ARG for TET (tet(B)) in isolate 84.

Table 3.

Description of plasmids, antibiotic resistance genes (ARGs) and transfer frequency (TF) of the Transconjugants produced in this study

Trans-conjugant	Plasmids in donor	ARGs in donor	TF	Plasmids transferred	ARGs transferred						Accession numbers
					Amino-glycosides	β-lactams	Tetra-cycline	Sulfon-amides	Quino-lones	Trime-thoprim
18-CV601	IncFIB(AP001918), IncN, ColRNAI	blaCTX-M-15, blaTEM-1B, dfrA14, floR, mdf(A) qnrS1, tet)(A).	1.33 × 10−2 ± 0.00587	IncN		blaCTX-M-15			qnrS1	dfrA14	SAMN54758342
41-CV601	IncFIB(AP001918), IncFII, IncI1	aph(3”)-lb, aph(6)-ld, blaCTX-M-15, dfrA8, mdf(A), qnrS1, sul2	3.31 × 10−6 ± 0.00000124	IncI		blaCTX-M-15			qnrS1		SAMN54758343
42-CV601	IncFIB(AP001918), IncFII(pCoo), IncFII, IncI1, IncI2	blaCTX-M-1, mdf(A)	4.97 × 10−5 ± 0.0000497	IncI1		blaCTX-M-1					SAMN54758344
46-CV601	IncFIB(AP001918), IncFII, IncN	blaCTX-M-15, dfrA14, mdf(A), qnrS1	1.24 × 10−3 ± 0.0000713	IncN		blaCTX-M-15			qnrS1	dfrA14	SAMN54758345
47-CV601	IncFIB(AP001918), IncFIC(FII), IncI1, ColRNAI, Col8282	blaCTX-M-3, mdf(A)	1.12 × 10−3 ± 0.000589	IncI1, Col8282, ColRNAI		blaCTX-M-3					SAMN54758346
49-CV601	IncN	blaCTX-M-15, dfrA14, mdf(A), qnrS1	1.59 × 10−3 ± 0.000910	IncN		blaCTX-M-15			qnrS1	dfrA14	SAMN54758347
50-CV601	IncFII, IncP1, Col8282, Col156	blaCTX-M-15, mdf(A), qnrS1	1.77 × 10−5 ± 0.0000112	IncFII, IncP1, Col156		blaCTX-M-15			qnrS1		SAMN54758348
65-CV601	IncFII, Col440I, Col440II, ColRNAI, Col8282, Col(MG828)	aph(3”)-lb,aph(6)-ld, blaCTX-M-15, mdf(A), qnrS1, tet(B)	1.16 × 10−5 ± 0.00000581	IncFII		blaCTX-M-15			qnrS1		SAMN54758349
75-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	1.44 × 10−7 ± 0.0000000563	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758350
76-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	1.44 × 10−7 ± 0.0000000563	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758351
77-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	1.44 × 10−7 ± 0.0000000997	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758352
79-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	1.76 × 10−7 ± 0.000000142	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758353
80-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	2.47 × 10−7 ± 0.000000228	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758354
81-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	2.12 × 10−7 ± 0.000000118	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758355
82-CV601	IncFIA, IncFIB(AP001918), IncFIC(FII), IncY, IncX1, IncX3	mdf(A), tet(A)	1.76 × 10−7 ± 0.0000000815	IncFIA, IncFIB(AP001918), IncFIC(FII)			tet(A)				SAMN54758356
84-CV601	IncFIB(pB171), IncFII(pHN7A8), Col8282, Col440I	aph(3”)-lb, aph(6)-ld, blaTEM-1B, mdf(A), sul2, tet(B)	1.99 × 10−5 ± 0.0000154	IncFII(pHN7A8)	aph(3”)-lb, aph(6”)-ld	blaTEM-1B	tet(B)	sul2			SAMN54758357

The columns show the transconjugant name (original donor code – CV601); plasmids and antibiotic resistance genes (ARGs) found in the original donors; transfer frequency (TF); plasmids and ARGs transferred to CV601 from WWTP (18, 41, 42, 46, 47, 49, 50, 65, 84) and Fresh/Brackish water (75, 76, 77, 79, 80, 81, 82) isolates. The TF displayed is the mean with standard deviation.

Of the 68 Escherichia spp. isolates, plasmid replicons were not identified in nine isolates (five from fresh/brackish water and four from WWTP), possibly suggesting lack of plasmids in these isolates. Interestingly, one of these isolates (isolate 43, ST38), with the presumed lack of plasmids, carried ant(3”)-la and bla_CTX-M-14 on the chromosome. Several other isolates carrying only smaller Col plasmids, such as isolates 9 and 44 (both ST38), also contained bla_CTX-M-15 and qnrS1, where presence of these ARGs in the largest contig (contig size: 153 kb and 156 kb, respectively) suggest they are chromosomal. Additionally, in both isolates, the bla_CTX-M-15 and qnrS1 genes were flanked by transposon genes and genes not commonly found on plasmids (e.g. araE, ygeI), highlighting the possibility of chromosome insertion. In line with this, ARG transfer to CV601 did not occur when conjugation was tested with these isolates which further supported that the ESBL resistance was chromosomal. ST38 has previously been reported to frequently harbour ESBL genes in the chromosome^32,33.

Among the donors that transferred ARG-carrying plasmids to CV601, 56.3% (9/16) were wastewater isolates and belonged to ST683, ST6872, ST10, ST58, ST1442, ST2522, ST38, ST34, and ST18202. The transferred wastewater plasmids were IncN (33.3%, 3/9), IncF (22.2%, 2/9) IncF + IncP (11.1% 1/9), and IncI (33.3%, 3/9) (Table 3, and Fig. 2A, B). All carried ESBL ARGs (bla_CTX-M (88.9%, 8/9) or bla_TEM (11.1%, 1/9). A subset also co-carried qnrS1 (60%, 6/9) and dfrA14 (30%, 3/9), and only one plasmid co-carried sul, aph and tet(B) ARGs (Table 3). Additionally, isolates 47 and 50 also transferred smaller col plasmids (colRNAI, col8282, col156). The wastewater plasmids had a transfer frequency (TF) range of 10⁻²–10⁻⁷ (Fig. 2). The IncN plasmids had a higher TF range of 10⁻²–10⁻⁴ compared to IncF plasmids (10⁻⁵–10⁻⁷) (Table 3 and Fig. 2a).

Fig. 2. Transfer frequencies (TF), phylogeny and transferred plasmids of all transconjugants.

a Transfer frequencies (TFs) of all obtained transconjugants displayed on a segmented linear axis, including error bars showing the mean. The blue bars indicate the TFs of IncN plasmids, the orange bars indicate IncI TFs, the chequered bar indicates IncF/IncP TF; and the green bars indicate IncF TFs. The graph indicates that the IncN plasmids transferred at the highest TF, where transconjugant 18-CV601 had the highest TF of all transconjugants. The IncN plasmids had higher TFs compared to the IncF plasmids, where the TFs were varied. b Tree of kmer distance of all transconjugants and recipient CV601 showing clustering of transconjugants carrying plasmids of the same Inc group, the IncF plasmids from wastewater isolates cluster differently from the brackish water isolate IncF plasmids, note that transconjugant 50-CV601 branches out due to transfer of two plasmids (IncF and IncP). c Conjugative IncN, IncI and IncF plasmids using alignment of full contigs. The plasmid figures contain multiple different tracks, the outermost is the antibiotic resistance gene (ARG) track, followed by the Bakta annotation track, genes involved in plasmid transfer track and then the remaining tracks vary between each plasmid, where the backbones are the plasmids discussed in this study, and the remaining tracks are plasmids where sequence comparison using BLAST from sequences derived from GenBank Accession numbers were used. The three IncN plasmids were identical, hence displayed as one plasmid; the three IncI plasmids were not identical, therefore they are displayed as three separate plasmids; one IncP plasmid was isolated but carried no antibiotic resistance genes (ARGs) nor virulence genes and therefore no figure was made; lastly, two IncF plasmids (from 50-CV601 and 65-CV601) were identical, all IncF plasmids isolated from 75-CV601 - 82-CV601 were identical and displayed as one plasmids, and the plasmid from 84-CV601 showed differences compared to all other IncF plasmids. Note that 50-CV601 contained both IncP and IncF plasmids but IncF contained ARGs and represent the reported TF in (a). The bar chart was generated using PRISM, the phylogenetic tree was generated using Mashtree and ITOL, the plasmids were generated using Proksee.

The plasmid with the highest TF (1.33 × 10⁻²) was the IncN plasmid from isolate 18 (18-CV601 transconjugant), which is significantly higher compared to other TFs (p < 0.0001, Tukey’s multiple comparisons test). The IncN plasmids transferred from isolates 46 (46-CV601) and 49 (49-CV601) had similar TFs (1.24 × 10⁻³ and 1.59 × 10⁻³, respectively) (Table 3 and Fig. 2). Comparing the content of these three plasmids, there is no genetic variation which could account for the difference in TF. Additionally, upon aligning all three IncN plasmids, they appear almost identical (Fig. 2c). IncI plasmids had the second highest TFs followed by IncF plasmids (Table 3 and Fig. 2a). Furthermore, the three ST131 wastewater isolates (13, 45, 52) did not transfer conjugative plasmids, despite harbouring IncF plasmids.

Forty-one percent (7/16) of the conjugative plasmids originated from fresh/brackish water isolates, in these isolates the only plasmids transferred were IncF plasmids containing tet(A) and all donors belonged to ST351 (100%). The transfer frequencies of these plasmids were lower than those originating from wastewater isolates (Table 3 and Fig. 2a). Interestingly, TET ARGs were not transferred by any of the WWTP donors.

IncF plasmids were the only plasmids transferred from both wastewater (IncFII and IncFII(pHN7A8)) and fresh/brackish water (IncFIA, IncFIB(AP001918), IncFIC(FII)) isolates to CV601 (Table 3). The TFs varied where the IncF plasmids from wastewater donors had higher TFs compared to the IncF plasmids from fresh/brackish water donors (Fig. 2a). Additionally, in the isolates which carried both IncN and IncF plasmids (18 and 46), the IncN plasmid seemed to be favoured and transferred through conjugation (Table 3). The same pattern was observed for IncI and IncF, where IncI was prioritised (41 and 47). The donors which transferred IncF plasmids only had IncF plasmids, no IncN or IncI plasmids. The results highlight that most Escherichia spp. isolates carry plasmids, not all contain ARGs, and of those that do, a subset conjugate.

IncN, IncI, IncF and IncP plasmids were not the only plasmids transferred to CV601 during conjugation (Table 3). Isolate 47 transferred IncI1, Col8282 and ColRNAI, whereas isolate 50 transferred IncFII, IncP1 and Col156. Isolate 50 also carried a Col8282 plasmid which was not transferred to CV601. Interestingly, three other isolates carried similar Inc plasmids but did not transfer smaller Col plasmids during conjugation where larger IncN or IncF plasmids were transferred: isolates 18 (ColRNAI), 65 (ColRNAI, Col440I, Col440II, Col8282, Col(MG828)) and 84 (Col8282, Col440I).

Detection of virulent E. coli in WWTP inlets/outlets and in natural water sources

Since most of the isolates (60/68) carried plasmids but a subset lacked ARGs, the prevalence of virulence genes associated with human disease was studied (virulence factor database (VFDB) and VirFinder (CGE)). Sequenced transconjugants were analysed and no virulence genes were detected on the transferred plasmids.

The annotations of the sequenced isolate genomes revealed that most of the E. coli isolates contained commonly found virulence genes, such as adhesins fimH and fdeC. However, a subset of isolates (Supplementary Tables 1 and 2) contained genes related to specific E. coli pathotype species: EAEC (enteroaggregative E. coli), EIEC (enteroinvasive E. coli), ExPEC (extra-intestinal E. coli), EPEC (enteropathogenic E. coli), STEC (shigatoxin producing E. coli), ETEC (enterotoxigenic E. coli), and UPEC (uropathogenic E. coli).

A total of twenty-five different genes coding for adhesins were identified and associated with different pathotypes (Supplementary Table 2). In addition, ten different toxin genes were identified: sepA, pic, sat, sta2 senB, cdt, cnf1, hlyA, vat and tcpC (Supplementary Tables 1 and 2). Multiple adhesins were associated with the EAEC pathotype, including aaiC, aap, aar, aatA, agg4A, aggA, aggR, air/eaeX, and eilA. Wastewater isolates 23 (ST7326) and 84 (ST34) carried multiple of these genes, including the transcription factor aggR and toxins pic and sat, indicating that they are EAEC, however, these isolates did not carry the heat-stable enterotoxin astA often associated with the EAEC pathotype (Supplementary Table 1).

Whereas aufA, papA, papB, papC, papG-II, sfaA, and yfcV are associated with the ExPEC pathotype, all these genes plus afaA, except yfcV are also associated with the UPEC pathotype (Supplementary Tables 1 and 2). This may explain why many isolates expressing these adhesins are frequently isolated from UTI, BSI and uro-sepsis cases, confirming association with both the ExPEC and UPEC pathotypes. The wastewater ST types (isolate 2 (ST127), 5 (ST404), 13 (ST131), 43 (ST38), 45 (ST131), 52 (ST131) and 86 (ST95)) have been associated with UTIs, BSIs and invasive infections³⁴, while the fresh/brackish water isolates 36, 37, 38, 39 and 40 (all ST13281) are not yet described in relation to disease to the author’s knowledge. These isolates carried combinations of toxins sat, senB, vat cnf1, tcpC and hlyA commonly found in ExPEC and UPEC.

Additionally, two genes were found to be associated with the STEC/EPEC pathotypes: bfpB (required for formation of type IV bundle-forming pili) and lpfA (long polar fimbriae). The lpfA gene was found in 48.52% (33/68) of isolates, comparatively, only 5.88% (4/68) of isolates carried bfpB. Isolate 14 co-carried both genes, although none of these isolates carried the hallmark virulence gene for shiga-like toxin stx (Supplementary Tables 1 and 2).

Three adhesins were associated with the ETEC pathotype: nfaE (diffuse adherence fibrillar adhesin gene), cfaE (tip-localised adhesive subunit of colonisation factor antigen I fimbriae CFA/I) and tia (cell invasion determinant, also associated with STEC). Isolate 41 carried both cfaE and tia as well as the sta2 gene encoding the heat-stable enterotoxin confirming this isolate to be enterotoxigenic E. coli (ETEC) (Supplementary Table 1). This isolate is the first ETEC documented in Swedish wastewater, to the authors’ knowledge³⁵.

Four genes coding for colicins or microcins were identified: mchB (Microcin H47), mchC (MchC protein), mchF (ABC transporter protein MchF) and cma (Colicin M). Only isolate 23 carried all mch genes, associated with STEC/EPEC pathotypes, whereas three isolates carried mchF only. A total of thirteen isolates (18.84%) carried the cma gene and there was no co-carriage of the mch and cma genes (Supplementary Table 2).

The iss gene, coding for Increased serum survival protein, was found in 52.94% (36/68) of all isolates, from both wastewater and fresh/brackish water. Carriage of this gene is associated with the ExPEC pathotype. Another gene implicated in serum resistance, cvaC, coding for a serum resistance protein was found in only 5.88% (4/68) of all isolates, of which all 5.88% co-carried iss. Carriage of cvaC is associated with the UPEC pathotype (Supplementary Table 2).

Discussion

This study characterised conjugative plasmids, identified ARGs as well as investigated the virulence genes carried by Escherichia spp. isolated from fresh/brackish water and wastewater sites in Stockholm, Sweden by whole genome sequencing (WGS) of both antibiotic-resistant and non-resistant isolates to compare the E. coli diversity found at fresh/brackish water and wastewater sites. Different MLST types and pathotypes were identified in fresh/brackish water and in wastewater, with no overlap of MLST types between the two environments. This might suggest that strain specific adaptation to the different locations exists, which has previously been seen, especially when comparing wastewater and freshwater sites^25,36. Naturalised wastewater E. coli, strains which have evolved to live and survive in WWTPs, are commonly resistant to wastewater treatment and are commonly MDR, carrying multiple ARGs^37,38.

Antibiotic residues, ARB and ARGs are widespread in wastewater, often considered a hotspot for the dissemination of AMR^39–41. Release of resistant bacteria via effluent discharge is cause for concern in the surrounding environment, potentially promoting the selection of AMR in natural environments^42–44.

In line with this, all WWTP water samples in this study contained detectable levels of antibiotics. The physico-chemical properties of the residues and environmental conditions play a role in affecting degradation and adsorption processes⁴⁵ and varying treatment methods in WWTPs impact the breakdown of antibiotics⁴⁶. Henriksdal outlet showed equally high concentrations of antibiotics as Sickla inlet, a trend is therefore observed, where MDR E. coli are isolated from WWTPs with detectable levels of antibiotic residues. This implies that the antibiotic selection pressure together with the MDR observed in E. coli ST types frequently detected in WWTPs selects for certain pathotypes in WWTPs. The results from this study show that WWTPs can introduce antibiotic residues into the environment. The concentrations of ciprofloxacin (CIP) at Sickla inlet (1.34 µg/L) and Henriksdal outlet (1.1 µg/L) both exceeded the predicted no effect concentration (PNEC) at 0.064 µg/L established by Bengtsson-Palme and Larsson (2016)³¹ implying selection for AMR. Considering this, our finding of high frequencies of co-carriage of bla_CTX-M-15 and qnrS suggests that selection pressure by CIP in WWTPs may drive carriage of both bla_CTX-M-15 and qnrS in WWTP isolates. Although it has been widely established that β-lactam antibiotics are easily hydrolysed in water⁴⁷, so it is not possible to correlate these residues with ARGs, although they can still be active and drive antibiotic resistance. Regardless, the finding of frequent co-occurrence of bla_CTX-M-15 and qnrS is interesting and bla_CTX-M-15 has been shown to be associated with qnr genes located close together on conjugative plasmids²⁸, which supports the findings of this study.

Previous studies have found low MLST diversity in ESBL-selected and wastewater isolates, and that the most predominant ST type was ST131^25,36,48. Correlating with the results from this study, previous studies have shown that ExPEC populations, particularly UPEC, are the predominantly isolated E. coli pathotype from wastewater and appear to be more capable of surviving wastewater treatment^9,17,49,50. Zhi et al.²³ found clinically relevant ST types (ST131, ST95, ST127 and ST640) in wastewater as well as identifying UPEC strains which were found to be almost identical to clinical UPEC strains, whilst also carrying multiple ARGs. The predominantly isolated ESBL-producing MLST in this study was ST38 (n = 6) followed by ST131 (n = 3). Isolates belonging to ST38 have been reported to be well-known causes of extra-intestinal infections in humans⁵¹ and are frequently associated with ARGs³² and detected in WWTPs as well as in environmental waters^15,52. Three ST131 isolates were isolated in this study, whereof two from Henriksdal outlet (isolates 13 and 52) and one from the inlet (isolate 45). It has recently been reported that ST131 isolates are increasing in Europe and Sweden and becoming increasingly resistant to last-line antibiotics⁵³. This correlates with the findings from this study, where all three ST131 isolates carried nine or more ARGs. In this study, a predominance of MDR and pathogenic ExPEC/UPEC-associated MLST types were found in wastewater. One ETEC was isolated in wastewater inlet samples while no EIEC or shiga-like toxin (stx) genes were found indicating absence of STEC/EHEC, although other virulence genes associated to STEC were identified.

The environment has been shown to play a role in the AMR crisis, where the environment acts as a reservoir of ARGs and ARB where acquired resistance has been driven mostly by anthropogenic use of antibiotics^8,54,55. In this study, levels of AMR, ARGs, antibiotic residues and virulence genes were expected to be low in the samples and strains isolated from fresh/brackish water sites. The antibiotic residue concentrations at the fresh/brackish water sites were around the limit of detection (LOD) but CIP, chloramphenicol (CHL) and/or tetracycline (TET) were detected at multiple sites. The environment close to WWTP outlets could therefore be affected with an increase in MDR, possibly conjugative and/or pathogenic, E. coli. Other studies have shown virulence markers are common in E. coli isolated from WWTP effluents¹⁴. More concerningly, if pathogenic strains possess stress responses which allow for better survival through wastewater treatment, it may be the case that they are better adapted to surviving in the environment surrounding the outlets.

For E. coli isolated from fresh/brackish water sources, 75% (21/28) carried several virulence genes associated with ExPEC and UPEC pathotype species, but fewer genes associated with other pathotype species, such as ETEC, EAEC, STEC and fewer virulence genes overall compared to isolates from wastewater. However, since these E. coli were isolated from environmental recreational sites, the risk of exposure to humans is higher compared to wastewater. Studies have suggested that water may be an important vehicle of transmission for ExPEC⁵⁶, where UTIs caused by UPEC strains have been linked to exposure to contaminated recreational water⁵⁷. In line with this, a study in the UK found that the prevalence of ExPEC/UPEC ST38, ST3103, ST131 and ST167, and ST1421 carrying bla_CTX-M-15 was increased in river sediment collected downstream of a WWTP¹⁵. A study of AMR Enterobacteriaceae in rivers and canals in Switzerland identified carbapenemase genes in ExPEC and UPEC- associated STs ST38, ST73, ST167, ST410 and ST648⁵². ST38 is considered a high-risk clone responsible for the spread of bla_OXA-48⁵².

Six E. marmotae isolates were identified in freshwater samples located within Stockholm city. E. marmotae has been identified as the most prevalent Escherichia clade present in the environment and has been found in bird faeces, water and soil⁵⁸. E. marmotae has also been reported to infect humans and the isolates carried virulence genes (astA, hlyF and kpsMII_K5). These findings indicate that several species of environmental Escherichia might be able to cause disease upon exposure to recreational water.

Plasmids have been shown to be the main source of ARGs for E. coli isolated from wastewater⁵⁹. Presence of E. coli carrying ARG-containing conjugative plasmids in WWTPs is reason for concern as bacterial species are in proximity, risking both inter- and intra-species conjugation, therefore the risk of ARG spread is high.

The present study found high transfer frequencies (TF) of IncN, IncI and IncP plasmids while IncF plasmids had lower TF. Conjugative plasmids’ TF has been shown to differ between plasmid Inc types. IncI plasmids have been reported to spread at rates of 1 × 10⁻²–1 × 10⁻⁴ during both in vitro and in vivo experiments⁶⁰. IncF plasmids, including plasmids from ST131, have been shown to display low or absent TFs, while IncF transferred from a ST69 donor had a TF of 1 × 10^{−4 60}. These results correlate with the results of this study where variations were observed in TF of IncF plasmids, with no transfer of IncF plasmids from ST131 isolates while a TF of 1 × 10⁻⁵ was recorded in isolates 65 (ST18202) and 84 (ST34). Isolates from sediment and river water downstream of a WWTP found TFs to an E. coli recipient between 10⁻⁷ and 10⁻⁴ including donors ST131 and ST38¹⁵. The TF to CV601 for an IncP plasmid has previously been reported to be 1 × 10^{−2 61}, while the same author reported TFs of 7.8 × 10⁻³ for an IncFIIY plasmid⁶². Conjugation assays where Bolivian river water microbiota was used as donor found only IncN plasmids⁶³. These results are in accordance with studies of transfer from hospital sewage where IncN plasmid transfer to CV601 was found to be common⁶⁴. The results from the conjugation assays in this study suggest that IncN plasmids in general have high TFs which supports multiple findings of IncN plasmids when water microbiota has been used as donor.

High TFs of IncN plasmids has been reported previously⁶⁵. The present study shows an agreement with several studies and indicates that transfer frequencies of IncN, IncP and IncI1 are higher than the commonly found IncF plasmids in E coli.

In this study, we also observed that IncN and IncI plasmids were preferentially transferred over IncF plasmids in isolates containing a combination of these plasmids. IncF plasmid transfer inhibition by IncI plasmids has previously been described^66–70, but there are no reports of IncF transfer inhibition by IncN plasmids. However, transfer inhibition of IncP plasmids by IncN, IncF and IncI plasmids is described in several studies^71–73. Nevertheless, isolate 50 in this study transferred both IncP and IncF plasmids.

Pallares-Vega et al.⁷⁴ showed that low temperatures and nutrient limitations (environmental conditions) greatly affected TFs of a conjugative IncP plasmid, which highlights that TFs observed in optimal conditions (e.g. 37 °C and high nutrients) may lead to an over estimation of TFs. It was noted that lower nutrients impacted TF more than lower temperatures. However, albeit at a very low level, conjugation still occurred at low temperatures and in nutrient-lacking conditions, which could suggest that conjugation may occur in the environment at low rates. In addition, when considering WWTPs, the temperature is low, but nutrients are often present, which further highlights WWTPs as possible hotspots for conjugation.

E. coli isolates carrying ARGs may carry a fitness cost in terms of decreased growth rate, competitive ability and/or virulence^75,76. In an antibiotic free environment, the cost of carrying ARGs has been shown to be highly variable, ranging from costs of resistance as high as >50%, to little, if any cost^77–80. Vogwill and MacLean found that AMR by plasmid acquisition confers a smaller cost compared to chromosomal mutations causing AMR. The relatively low cost of carrying an ARG-containing plasmid could explain why plasmids play a principal role in the evolution of resistance⁸¹. Due to this low cost on fitness, the isolates have no reason to lose the plasmid, highlighting that MDR isolates carrying conjugative ARG plasmids released in effluent from wastewater, are likely to keep the plasmid once they enter the environment, in absence of selective pressures faced in the WWTP.

However, in another more recent study, plasmids carrying multiple ARGs were shown to reduce bacterial fitness by 2.9% per generation⁸². Individually, the ARGs were mostly cost-free, apart from bla_CTX-M-15 and tetAR, which were responsible for the entire cost of the plasmid. Therefore, finding isolates in the environment which carry such plasmids might suggest that there are other selective forces acting on environmental MDR E. coli.

The risks associated with the spread of clinically relevant ARGs in the environment are not yet fully understood. This study contributes to current knowledge of E. coli ecology and AMR by emphasising the importance of investigating both antibiotic-resistant and non-antibiotic-resistant isolates to generate comprehensive and unbiased surveillance data. Furthermore, the findings underscore the importance of integrated surveillance of WWTP effluents and adjacent environmental sites to identify potential increases in pathogenic and/or AMR bacteria. While this study focused on E. coli due to its widespread use as an indicator of anthropogenic contamination, it should be noted that ARGs may also be transferred by other bacterial species. Metagenomic studies conducted in the Baltic sea have detected ARGs, including those conferring resistance to beta-lactams, at all sampling locations; although high abundances of Gammaproteobacteria were present in fresh and brackish waters, the prevalence of Enterobacterales was low and Escherichia spp. were not detected⁸³.

Sampling was conducted at a single time point during the summer season in Sweden, precluding assessment of seasonal variation. As higher prevalences of pathogenic bacteria are globally associated with warm and wet seasons, the present data represent a temporal snapshot and may overestimate pathogen occurrence in the environment surrounding Stockholm, Sweden. Regardless, the results from this study underscore the importance of environmental AMR surveillance, where natural and naturalised E. coli may act as ARG reservoirs which, when acquired by virulent and pathogenic isolates, could complicate treatment outcomes.

Methods

Isolation of Escherichia coli from water

Water samples were collected from five environmental sites in and around Stockholm city in Sweden in 2022. Freshwater samples were collected at Brunnsviken in Hagaparken, Karlbergskanalen, and Rålambshovsparken, all located in Stockholm city, and Brackish water from Tyresö, and Nynäshamn located outside the city (Supplementary Fig. 1). Samples were also collected from four wastewater sites (Henriksdal inlet, outlet, Sickla inlet, and Henriksdal biofilter). At each sampling site, 1.5 l water (W) was collected, and at all sites except the wastewater sites and Karlbergskanalen, 25 ml sediment (w/v) (S) was collected in sterile bottles or tubes and directly transported to the laboratory.

For sediment samples, bacteria were extracted by centrifugation (500 rpm, 4 °C), followed by washing with 25 ml Phosphate Buffered Saline (PBS) and shaking for 1–2 h. Liquid from sediment samples were directly plated on RAPID’E.coli 2 Agar (Biorad) and Thermo Scientific™ Brilliance™ ESBL Agar (37 °C, 24 h) for Escherichia coli (E. coli) isolation.

For all water samples (fresh/brackish and wastewater), 250 ml was vacuum-filtered using 0.45 µm filters (MF-Millipore^TM Membrane Filter HAWP02500). The filtered water was used to dilute the original fresh/brackish and wastewater samples to 10⁻¹, 10⁻². The diluted samples (total volume of 250 ml) were filtered as above, and these filters were placed on the surface of RAPID’E.coli 2 and Brilliance^TM ESBL Agar (37 °C, 24 h).

Presumptive E. coli colonies were re-streaked on RAPID’E.coli 2 and Brilliance^TM ESBL Agar and confirmed using MALDI-TOF (Bruker). The confirmed isolates were stored at −80 °C in LB broth (Invitrogen^TM Luria Broth Base powder 12795084) with 20% glycerol.

Antibiotic susceptibility testing (AST)

The antibiotic resistance profiles of all isolates were determined using the disc diffusion method. The isolates were grown overnight in Luria Broth (LB) with supplemented antibiotics for the transconjugants at 37 °C, 220 rpm. The cultures were adjusted to OD₆₀₀ = 0.1 before inoculation of 100 µl on Mueller Hinton agar (MH) (Thermo Scientific^TM Mueller Hinton Agar dehydrated CM0337B). Antimicrobial susceptibility discs (Oxoid^TM) were placed on the surface of the MH agar within 10 min of inoculation. The plates were incubated overnight at 37 °C and inhibition halos measured. Breakpoints were measured according to EUCAST clinical breakpoints, following recommended disc concentrations (The European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters. Version 13.0, 2023. http://www.eucast.org).

The antibiotics tested included ampicillin (AMP), cefotaxime (CTX), streptomycin (STREP), ciprofloxacin (CIP), ceftazidime (CAZ), compound sulphonamides (SUL), trimethoprim (TRIM), azithromycin (AZM), norfloxacin (NOR), nalidixic acid (NA), tetracycline (TET), colistin (CT), amoxicillin/clavulanic acid (AMC), rifampicin (RIF), meropenem (MEM), Imipenem (IPM) and nitrofurantoin (F).

Antibiotic residue analysis

The filtered water samples were frozen upon filtration. Five hundred millilitres of the undiluted filtered water was sent to ILVO, Belgium for antibiotic residue analysis. One hundred millilitres of a water sample was screened for the presence of 78 antibiotic residues from 10 different classes using a UHPLC-MS/MS method validated in compliance with EU Commission Implementing Regulation 2021/808. The analytical procedure involved solid phase extraction using OASIS HLB columns (600 cc, 500 mg), followed by separation in an ACQUITY UPLC H-class (Waters) over a reversed-phase ACQUITY UHPLC BEH C18-column (2.1 × 150 mm;1.7 μm, 100 Å) (Waters), which was coupled to a Xevo TQ-XS spectrometer (Waters) in electrospray ionisation modus. This multiresidue method³⁰ enabled the quantification of residues of 20 β-lactams, 22 sulphonamides, 13 (fluoro)quinolones, 9 macrolides, 4 tetracyclines, 4 amphenicols, 2 lincosamides, 2 pleuromutilins, trimethoprim and rifaximin.

Conjugation assays

Conjugation assays were performed as described previously^63,84 with minor modifications. E. coli donors n = 39 (isolated from the fresh/brackish water and wastewater and resistant to selected antibiotics: CTX, CIP, TMP, STP and/or TET) were grown overnight in LB at 37 °C, 220 rpm. The recipient (E. coli CV601 (Kan^R, Rif^R, Lac⁻, GFP)) was grown overnight in LB with 50 mg/L KAN at 30 °C, 220 rpm⁸⁵.

Aliquots (1 mL) of the donors and recipient were washed in PBS. The aliquots were adjusted to OD₆₀₀ = 1. The donor and recipient strains were mixed in equal volumes (100 µL each) in sterile 1.5 ml centrifuge tubes or 96-well plates to obtain the conjugation mixture. Donors and recipient were also included as controls separately (100 µL of each). All tubes/96-well plates were incubated at 30 °C for 2 h to allow conjugation to occur, which was disrupted by pipetting. The donor, recipient, and conjugation mixture were spotted on LB agar with and without single selection (KAN or antibiotic of choice) and double selection (KAN+ antibiotic of choice) and incubated for up to 48 h at 30 °C. In this study single selection refers to: LB + KAN (200 mg/L) and LB + CTX (5 mg/L) or LB + TET (30 mg/L), and double selection: LB + CTX + KAN or LB + TET + KAN.

Transconjugants were re-streaked on LB agar with double selection as described above, on MacConkey agar (Thermo Scientific^TM), and disc diffusion was performed as described above to assess transferred resistances. The transconjugants were stored at −80 °C in LB broth with 20% glycerol.

Calculating conjugation transfer frequencies

The isolates which conjugated in the above assay were selected for further assays to determine transfer frequency. Transfer frequencies were calculated as colony-forming units (CFU) of the transconjugants divided by the CFU of the recipient. Once conjugation ability was confirmed using the method described above, the same method was employed but deviated after disrupting conjugation. The recipient (CV601) was serially diluted to 10⁻⁷ and the conjugation mixture was serially diluted to 10⁻⁴. For the recipient (CV601), 100 µL of each dilution was spread on LB + KAN (200 mg/L). For the conjugation mixture, 100 µL of each dilution was spread on the double selective plates (LB + KAN + antibiotic of choice). The donor, recipient, and conjugation mixture were spotted on LB, LB + KAN and LB + KAN+antibiotic of choice as controls. All plates were incubated for up to 24 h at 30 °C. The CFU of the recipient and presumptive transconjugants were counted and the transfer frequency was calculated. This was repeated in biological triplicates with two technical replicates.

Whole genome sequencing (WGS) and data analysis

Whole genome sequencing (WGS) was performed for all isolates, both donors and transconjugants. Isolates were grown overnight in LB at 37 °C, 220 rpm. Transconjugants were grown overnight in LB supplemented with 200 mg/L KAN and appropriate antibiotic (5 mg/L CTX or 30 mg/L TET) at 37 °C, 220 rpm. Total DNA isolation was performed using DNeasy Blood and Tissue Kits (Qiagen) following the manufacturer's protocol without modification. Isolated DNA was measured using either Qubit™ dsDNA Quantification Assay Kits or Nanodrop (ThermoFisher) and stored at −20 °C. For WGS, all DNA samples were shipped to Eurofins Genomics Europe (Germany) for Illumina paired-end, NovaSeq sequencing (read length: 2 × 150 bp).

Sequencing data was processed using the BACTpipe assembly and annotation pipeline⁸⁶(v. 2.6.1, https://github.com/ctmrbio/BACTpipe)⁸² and re-annotated using Bakta. The presence of ARGs, virulence genes, Multi-locus sequence types (MLST) profiles, and plasmid typing (Inc) was determined using the Centre for Genomic Epidemiology (CGE) pipeline, including ResFinder^87,88, VirFinder^89,90, and PlasmidFinder^88,91. MLSTs were determined using the Achtman scheme²⁶. Neighbour-joining trees of k-mer distances were produced using Mashtree⁹² and visualised in ITOL v6⁹³. Additionally, MOB-suite⁹⁴, and MOBFinder⁹⁵ were also used to verify the presence of plasmids/plasmid replicons, in addition, contigs were manually curated and plasmid replicons were identified in the same contigs as, for example, ARGs to identify ARG-carrying plasmids. The surrounding genetic environment was also studied to ensure correlation with plasmid or chromosome content. Figures of confirmed plasmids were visualised in Proksee⁹⁶.

Statistical analyses

The transfer frequencies were analysed in GraphPad Prism using one-way ANOVA and Tukey’s multiple comparisons test. The significance of ESBL presence in wastewater isolates compared to environmental isolates was calculated using Fisher’s exact test.

Author contributions

A.J.N. analysed the resistance profiles and conjugation frequencies, analysed and visualised the data, and drafted the manuscript together with Å.S. K.T. performed bioinformatical analyses on the sequenced isolates together with A.J.N. L.T. and G.R. analysed antibiotic residues. E.J., A.J.M.R. and Å.S. conceived the experiments and collected the isolates. All authors read and approved the final manuscript.

Funding

Open access funding provided by University of Gothenburg.

Data availability

The raw data and assembled genome data presented and discussed in this study can be found on GenBank online repository under bioproject PRJNA1266621. SRA accession numbers: SRR35509390-SRR35509458 and biosample accession numbers: SAMN48681410-SAMN48681432; SAMN48681364–SAMN48681385 and SAMN48681386 – SAMN48681409. The assembled genome data for the transconjugants can be found on GenBank online repository under bioproject PRJNA1347068.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s44259-026-00208-5.