Survey of US wastewater for carbapenem-resistant Enterobacteriaceae

Abstract

A survey for antibiotic-resistant (AR) Escherichia coli in wastewater was undertaken by collecting samples from primary clarifiers and secondary effluents from seven geographically dispersed US wastewater treatment plants (WWTPs). Samples were collected at each WWTP in cool and summer months and cultured using selective media. The resulting isolates were characterized for resistance to imipenem, ciprofloxacin, cefotaxime, and ceftazidime, presence of carbapenemase and extendedspectrum beta-lactamase (ESBL) genes, and phylogroups and sequence types (STs). In total, 322 AR E. coli isolates were identified, of which 65 were imipenem-resistant. Of the 65 carbapenem-resistant E. coli (CREC) isolates, 62% were positive for more than one and 31% were positive for two or more of carbapenemase and ESBL genes targeted. The most commonly detected carbapenemase gene was bla_VIM (n=36), followed by bla_KPC (n=2). A widespread dispersal of carbapenem-resistant STs and other clinically significant AR STs observed in the present study suggested the plausible release of these strains into the environment. The occurrence of CREC in wastewater is a potential concern because this matrix may serve as a reservoir for gene exchange and thereby increase the risk of AR bacteria (including CR) being disseminated into the environment and thence back to humans.

INTRODUCTION

Escherichia coli, a member of the family Enterobacteriaceae, is the predominant enterobacterial gut resident in humans and other mammals and is commonly used as an indicator of fecal pollution in source, drinking, and recreational waters (Elmund et al., 1999; Edberg et al., 2000; Odonkor and Ampofo, 2013). Antibiotic resistance, including to extended-spectrum cephalosporins (ESCs) and carbapenems, is increasingly prevalent among Enterobacteriaceae. According to the Centers for Disease Control and Prevention (CDC), prior to 2002 less than 1% of E. coli clinical isolates were ESC-resistant or produced extended-spectrum β-lactamases (ESBLs), whereas in a recent report approximately 22% of clinical E. coli isolates were ESC-resistant (Weiner et al. 2016). Prior to 2010, reports of carbapenem-resistant Enterobacteriaceae (CRE) infections were uncommon in the USA; however, according to the CDC’s National Healthcare Surveillance Network, the percentage of CRE isolates identified from hospital acquired infections ranged from 3.3 to 10.9%, depending on the site of infection (Weiner et al. 2016).

Historically, carbapenem antibiotics have been effective against multi-drug-resistant Gram-negative bacilli and a mainstay of therapy for infections due to such organisms. Carbapenem-resistant E. coli (CREC) is increasingly prevalent, mainly due to the emergence of novel carbapenemases. Currently, the three carbapenemases categorized as being most important are Klebsiella pneumoniae carbapenemase (KPC; Ambler class A), certain metalloenzymes (VIM, NDM; Ambler class B), and OXA-type enzymes (e.g., OXA-48; Ambler class D) (Nordmann et al., 2012). The associated carbapenemase genes are frequently found on mobile genetic elements and have the potential to spread to other Gram-negative bacteria (Nordmann et al., 2012).

Although CREC have been recognized primarily in health care settings, investigators at the US Environmental Protection Agency (EPA) proposed that antimicrobial resistance surveillance involving sewage isolates could be informative as to the occurrence and dissemination of specific clonal groups or sequence types within a given community or population (Boczek et al. 2007). Wastewater treatment plants (WWTPs) typically use a multi-stage treatment approach, consisting of preliminary, primary, and secondary treatment with effluent disinfection. Primary treatment occurs when wastewater received by the treatment plant is collected into settling basins, where solids and grease are removed from the liquid portion using screens and gravity. Such processes are successful at removing approximately 60% of suspended solids from the wastewater. This primary effluent is then subjected to some type of secondary treatment, typically involving biological treatment to further breakdown the organic material in the wastewater. Following secondary treatment, the effluent is disinfected either with a chemical oxidizing agent, such as chlorine, or a physical treatment such as exposure to UV radiation (Manual of Practice 8; Design of Wastewater Treatment Plants 1998; Boczek et al. 2010).

This study is follow-on research to preliminarily survey wastewater, specifically primary and secondary effluents, from seven locations in the USA for antibiotic-resistant (AR) E. coli to ESCs and carbapenem, to characterize the resulting isolates for the presence of carbapenemase and ESBL genes and identify sequence types that have been associated with extraintestinal infections in humans.

MATERIALS AND METHODS

Study sites, sample collection and titer determination

Seven geographically dispersed WWTPs in the continental US were selected opportunistically for sampling. Permission was granted by WWTP operators on the condition of plant anonymity. The WWTPs were located in New Jersey, Maryland, Ohio, Texas, Colorado, northern California, and southern California, and served four urban and three rural/suburban areas. All participating plants used conventional activated sludge for secondary treatment of primary clarified effluents. Samples were collected at each WWTP once during cooler months (November 2012 or April 2013) and once during the summer (July 2013 or August 2013), for a total of 28 samples, four per WWTP. Each sample consisted of either 1 liter from the primary clarifier effluent or an effluent from secondary treatment after disinfection. The samples were collected in sterile polypropylene bottles, shipped overnight on ice to the US EPA, Cincinnati, OH, and analyzed within 24 hours of collection.

E. coli titers were determined for all samples using membrane filtration method 9222 I (Standard Methods for the Examination of Water and Wastewater 2017). E. coli identification was confirmed using BBL Crystal Kits and Crystal Mind Software with the Autoreader (BD Bioscience, Sparks, MD). Confirmed E. coli isolates were then assigned a unique identification number and stored in 10% glycerol at _80 _C until further analysis.

Isolation of antibiotic-resistant E. coli and antimicrobial susceptibility testing

AR E. coli were isolated using a membrane filtration procedure, using mFC agar (Becton Dickinson, Franklin Lakes, NJ) supplemented with 1 mg/L imipenem, 4 mg/L ciprofloxacin, 4 mg/L cefotaxime, or 16 mg/L ceftazidime (Sigma-Aldrich, St. Louis, MO), reflecting the respective resistance breakpoints specified by the Clinical and Laboratory Standards Institute (CLSI 2012). Briefly, serial dilutions ranging from 10 mL to 0.01 mL of sample were filtered through 0.45 μm polycarbonate filters and transferred to mFC plates supplemented with antibiotics and incubated at 44.°C. Filters that contained countable colonies (<100) were transferred to mFC supplemented with 4-methylumbelliferyl-β-D-glucuronide (MUG) to obtain AR E. coli titers and isolates. All MUG positive isolates were chosen for testing.

The four antibiotics were selected because of the 2015 Centers for Disease Control and Prevention’s definition for CRE, i.e., full resistance to two third generation cephalosporins and intermediate to full resistance to a carbapenem. Ciprofloxacin was chosen to ensure coverage for the H30R subclone within STc131. This subclone is known for its characteristic resistance to ciprofloxacin and for causing most multi-drug resistant E. coli infections in the USA (Johnson et al. 2010).

Minimum inhibitory concentrations (MICs) for ciprofloxacin, imipenem, ceftazidime, and cefotaxime were determined using E-Test™ Strips (bioMerieux, Marcy’Etoile, France). Isolates were classified as resistant, intermediate, or susceptible according to specified breakpoints (CLSI, 2012). For data analysis, intermediate isolates were regarded as resistant.

Real-Time PCR for carbapenemase and ESBL genes

CREC isolates were screened by real-time polymerase chain reaction (PCR) for four types of ESBL-encoding genes and five types of carbapanemase-encoding genes (Table 1). PCR was performed in a QuantStudio 6 Flex instrument (Applied Biosystems, Foster City, CA) as previously described (Ryu et al., 2012). Each PCR plate included controls to check for sample cross-contamination and positive controls. Dissociation curves were examined for evidence of potential primer-dimers and other non-specific reaction products.

Table 1-.

Sequence and reference information for qPCR analysis, with target genes, primer sequences, amplicon sizes and annealing temperatures

Target Gene	Primer Sequence (5’−3’)	Amplicon size (bp)	Annealing Temperature	References	Control Strains
blaTEM	TEM_F: AAGATGCTGAAGATCA TEM_R: TTGGTATGGCTTCATTC	425	44 °C	Speldooren et al., 1998	E. coli T5
blaSHV	SHV_F: GCGAAAGCCAGCTGTCGGGC SHV_R: GATTGGCGGCGCTGTTATCGC	304	62 °C	Henriques et al., 2006a	E. coli 52
blaCTX-M	CTX_F:58 SCVATGTGCAGYACCAGTAA CTX_R: G55CTGCCGGTYTTATCVCC	652	55 °C	Lu et al., 2010	E. coli 85.01
blaIMP	IMP_F: GAATAGAGTGGCTTAATTGTC IMP_R: GGTTTAAYAAAACAACCACC	232	55 °C	Henriques et al., 2006b	NTU 92/99 (IMP-1) and P. aeruginosa
blaVIM	VIM_F: GATGGTGTTTGGTCGCATATCG VIM_R: GCCACGTTCCCCGCAGACG	475	58 °C	Henriques et al., 2006a	P. putida
blaKPC	KPC_F: CATTCAAGGGCTTTCTTGCTGC KPC_R: ACGACGGCATAGTCATTT	538	55 °C	Dallenne et al, 2010	E. coli USVAST-0600
blaGES	GES_F: AGTCGGCTAGACCGGAAAG GES_R: TTTGTCCGTGCTCAGGAT	399	57 °C	Dallenne et al., 2010	E. coli GES
blaNDM	NDM_F: GGGCAGTCGCTTCCAACGGT NDM_R: GTAGTGCTCAGTGTCGGCAT	405	60 °C	Manchanda et. al, 2011	E. coli MH01

Phylotyping and sequence typing

Phylogenetic groups of the AR E. coli isolates were defined by the quadraplex PCR method of Clermont, et al. (2012). Clonal lineages were identified using 2-locus sequence analysis of fumC and fimH (CH typing) (Weissman et al., 2012). Each allele combination (i.e., CH type) was assigned to a putative sequence type (ST) based on the known association of CH types with STs (Weissman et al., 2012). In ambiguous situations, additional housekeeping gene loci were sequenced to clarify the most likely associated ST, according to Enterobase (http://enterobase.warwick.ac.uk).

Results and Discussion

Titers of total E. coli, AR E. coli, and CREC were calculated for the WWTP samples to assess relative percentages of resistant organisms. Table 2 presents a summary of total E. coli titers along with the number of AR E. coli and CREC isolates obtained from each WWTP. Only results from the primary clarifiers are presented, as no AR E. coli isolates were obtained from the secondary effluents. Overall, 322 isolates were obtained that were resistant to ≥1 of the four antibiotics tested. Among the 322 AR E. coli isolates, the prevalence of resistance to individual agents declined sequentially as follows: cefotaxime (74%), ciprofloxacin (72%), ceftazidime (68%), and imipenem (20%). Additionally, 235 (73%) isolates qualified as multidrug resistant, defined here as having resistance to ≥2 antibiotics tested. Notably, based on biochemical reactions, antibiotic resistance patterns, and sequence typing (results for the latter two are discussed below), and because no enrichment media were used, the AR isolates seemed unlikely to include replicates of the same clone from a given sample in the isolate collection.

Table 2.

Summary of WWTP locations, Escherichia coli loads in primary clarifier effluents and antibiotic-resistant E. coli isolates obtained from each site

WWTP Site	Average Facility Capacity	Dates of Collection	Total E. coli (CFU/100 mL)	AR E. coli isolates	CR E. coli isolates	Raw wastewater composition
OHIO	13 MGD	July 2013	9.00E+06	51	8	Domestic, urban
		Sept. 2013	2.70E+06
NEW JERSEY	45 MGD	July 2013	3.50E+06	84	22	Domestic, urban
		Nov. 2012	4.50E+06
N. CALIFORNIA	33 MGD	July 2013	1.00E+07	8	1	Agriculture, domestic, suburban
		April 2013	TNTC
TEXAS	7.0 MGD	July 2013	1.50E+06	14	1	Agriculture, domestic, rural
		April 2013	2.00E+07
MARYLAND	5.0 MGD	July 2013	3.00E+07	32	8	Domestic, industrial, suburban
		April 2013	5.50E+07
COLORADO	135 MGD	July 2013	7.50E+06	29	10	Agriculture, domestic, industrial, urban
		April 2013	3.00E+07
S. CALIFORNIA	275 MGD	July 2013	1.70E+07	104	15	Industrial, agriculture, domestic, urban
		Sept. 2013	1.00E+07

To estimate the overall prevalence of AR E. coli within the WWTP-associated E. coli population, colony counts for all 14 primary effluent samples were summed and an average was calculated for total E. coli and for E. coli resistant to imipenem, ciprofloxacin, cefotaxime, and ceftazidime. The average AR fraction within the total E. coli population, by agent, was 0.0022% for imipenem, 0.38% for cefotaxime, 0.031% for ceftazidime, and 3.1% for ciprofloxacin (Table S1, available with the online version of this paper). Comparable percentages were reported previously for raw wastewater from Poland, although no carbapenem-resistant organisms were found in that study (Łuczkiewwicz et al. 2010). It is difficult to compare the percentages of AR E. coli seen in wastewater to clinical studies, but percentages of CREC ranged from 0.7 to 1.9% in a survey of clinical infections, depending on the site of infection (Weiner et al. 2016). These higher carbapenem resistance percentages, as compared to the present findings in wastewater, would be expected because clinical settings will naturally concentrate AR isolates.

Antibiotic resistance in E. coli is more concerning when it occurs in strains capable of causing clinical infections, due to the potential for dissemination and the resulting morbidity and mortality. Phylogenetic analyses have shown consistently that most clinical E. coli isolates from human extraintestinal infections are derived from phylogroups B2 and D (Clermont et al. 2012). However, transfer of antibiotic resistance genetic elements from commensal to pathogenic strains is suspected to occur in reservoirs where these groups co-mingle, including WWTPs (Marshall et al. 2009; Bailey et al. 2010). Table 3 presents the phylotypes and associated resistance profiles of the present 322 AR E. coli isolates. Phylotypes B2 and D accounted collectively for nearly half (n=181, 45%) of the sewage-source AR isolates (Table 3). Adding potentially pathogenic phylogroup F (Clermont et al. 2012) increased this to 56% of the total.

Table 3.

Phylotyping of AR Escherichia coli with associated resistance profiles.

Phylotype	Number of Isolates	Imipenem	Ciprofloxacin	Cefotaxime	Ceftazidime
A	64	6	51	45	41
B1	44	9	30	38	36
B2	62	10	47	37	33
C	30	12	27	25	25
D	84	25	42	67	60
E	3	0	3	2	2
F	35	3	32	25	23
Totals	322	65 (20)	232 (72)	239 (74)	220 (68)

Percentage of total number of isolates resistant to specific antibiotic in parentheses

Overall, 65/322 (20%) isolates were imipenem-resistant. CREC (n=65) were concentrated in virulence-associated phylogroups B2, D and F (38/181 (21%)), as compared with the other phylogroups combined (27/141 (19%)). By contrast, all phylogroups had a high prevalence of multi-resistant isolates, including resistance specifically to ciprofloxacin and ESCs. On a percentage basis, phylogroup C had the highest percentages of total AR E. coli.

The proportions of CRE per phylogroup are as follows (# CRE/# in phylogroup, %): A (6/64, 9%), B1 (9/44, 20%), B2 (10/62, 16%), C (12/30, 40%), D (25/84, 30%), and F (3/35, 9%). Thus, most CREC isolates (38/651/4 58%) were from phylogroups B2, D, or F and could represent extraintestinal pathogenic E. coli (ExPEC) which have an enhanced potential for causing disease in humans.

To further characterize the AR E. coli isolates, STs were determined to identify lineages associated with extraintestinal infections (Table 4). CH typing identified several ExPEC-associated clonal complexes (Johnson et al. 2010, 2017; Pitout 2012; Xia et al. 2017), namely (number of isolates, % of 322), STc131 (28, 8.7%), STc648 (17, 5.3%), ST1193 (25, 7.8%), and STc405 (12, 3.7%). Regarding the temporal and geographical distribution of these AR isolates from ExPEC-associated STs, carbapenem-resistant STc131 was identified in three separate samples from two locations (NJ, both summer and cooler temperatures; southern CA, summer only). Likewise, STc648 was identified in five locations, including two carbapenem-resistant isolates from two separate samples from one location (NJ, cooler and summer months); ST1193 was identified in five locations, including three CREC isolates from one sample in one location (NJ, summer); and STc405 was identified in five locations, including two CREC isolates from one sample in one location (MD, summer). These data indicate a widespread dispersal of these clinically significant AR STs with respect to locale and seasonality.

Table 4.

Number of sequence types complexes found among AR E. coli known to be associated with clinically significant infections and their associated antibiotic resistance.

Antibiotic Resistance	B2- ST131-O25b and B2-ST131-O16	D-STc405	STc 648	ST 1193
	(n=28)	(n=12)	(n=18)	(n=27)
Imipenem	4 (14)	2 (17)	2 (11)	3 (11)
Ciprofloxacin	21 (75)	11 (92)	17 (94)	22 (81)
Ceftazidime	15 (54)	9 (75)	9 (50)	10 (37)
Cefotaxime	18 (64)	11 (92)	12 (67)	12 (44)

Cell number represents number of isolates resistant to antibiotic, with percent in parentheses

Determining the mechanisms that impart carbapenem resistance is imperative for epidemiological reasons. The major reason for carbapenem resistance is the presence of a carbapenemase-encoding gene. Many such genes are located on mobile genetic elements (Tait 1993), and tracking their dissemination can provide important information to health care professionals. Screening of our 65 CREC isolates by PCR for five carbapenemase genes and four ESBL genes (Table 5) showed that 45 (62%) of the isolates contained ≥1, and 20 (31%) contained ≥2, of the nine studied carbapenemase and ESBL genes, mainly VIM (36, 55%) and TEM (24, 37%), but also GES, SHV, IMP, OXA, KPC, CTX, and NDM. Few of these isolates harbored bla_KPC (n=2), and none carried bla_NDM or bla_OXA, which are the focus of multiple clinical case studies. The 20 (31%) remaining CREC isolates with no detected target gene conceivably could harbor antibiotic resistance genes not targeted in the study, or an alternate carbapenem resistance mechanism, such as loss of outer membrane proteins or efflux pumps/porin mutations, which have been demonstrated previously (Kong et al. 2018).

Table 5.

Gene panel profiles of carbapenem and ESBL resistant E. coli isolates from wastewater

No. of genes positive	blaIMP	blaKPC	blaTEM	blaCTX-M	blaVIM	No. of isolates
0	−	−	−	−	−	20
1	−	−	−	+	−	2
	−	−	−	−	+	16
	−	−	+	−	−	7
2	−	−	+	−	+	14
	−	+	−	−	+	1
3	−	+	−	+	+	1
	−	−	+	+	+	3
	+	−	+	−	+	1
	1	2	25	6	36	65

Samples were all negative for bla_NDM, bla_SHV, bla_GES and bla_OXA genes

The 36 bla_VIM-containing CREC isolates were distributed by phylogroup as follows (% of 36): group D (19 isolates, 53%), B2 (6, 17%), C (5, 14%), B1 (3, 8%), and A (3, 8%). Thus, 70% of the bla_VIM-positive isolates belonged to extraintestinal virulence-associated phylogroups B2 and D. As for distribution by ST, bla_VIM occurred in three STc131 isolates (one of which also had bla_KPC), one STc405 isolate, and one ST1193 isolate. The high prevalence of bla_VIM in our study is notable considering the typical occurrence of this gene within integrons on broadhost-range plasmids. We are unaware of previously published data demonstrating such extensive prevalence of bla_VIM among clinical E. coli isolates. Metallo-β-lactamases such as VIM were reported previously to be common worldwide, but rare in Enterobacteriaceae in the USA (Gupta et al.2011; Nordmann et al. 2011).

CONCLUSIONS

This one-year study of water samples from seven geographically dispersed WWTPs in the USA identified 65 CREC isolates, of which 38 were from ExPEC-associated phlyogroups, including 11 from ExPEC-associated STs. Carbapenem-resistant STc131 was identified in three separate samples from two locations, and other clinically important STs were identified in multiple samples from other locations, demonstrating the plausible release of these organisms into receiving bodies of water.

Antibiotic resistance, including resistance to carbapenems, was not limited to pathogenic E. coli phylotypes or STs, as evidenced by the commensal phylotypes (A and C) also harboring multi-drug resistance patterns. Conceivably, this could be indicative of ectotherms harboring AR E. coli in their intestinal tracts as part of their normal microbiota.

The occurrence of CREC and other AR E. coli with potentially mobile resistance genes in wastewater is concerning because wastewater may serve as a reservoir for gene exchange. This implies a risk of AR E. coli and CREC being disseminated into the environment and thence back to humans, with routes of exposure potentially including contaminated recreational water and land application of biosolids. Our ability to understand the extent of this risk will depend, in part, on more detailed information concerning the fate of AR E. coli, including CREC, introduced in indigenous microbial communities, the rate of horizontal gene transfer in these communities, and how best to define the relative risk associated with both introduced and indigenous AR bacteria (Pepper et al. 2018).

Future studies should address the impact of multiple factors (e.g., geographic location, season, types of wastewater, and treatment plant configuration) on the prevalence of AR in wastewater. Additionally, rapid and reliable methods are needed to allow enumeration and characterization of AR organisms with respect to phylogenetic and clonal background, pathogenic potential, and resistance mechanisms. Identification of genetic factors that impart antibiotic resistance is essential when utilizing wastewater as an epidemiological screening tool for detection of emerging resistance patterns in comparison with clinical isolates from a given geographical locale.