A Population-Based Surveillance Study of Shared Genotypes of Escherichia coli Isolates from Retail Meat and Suspected Cases of Urinary Tract Infections

Community-acquired urinary tract infection caused by Escherichia coli is one of the most common infectious diseases in the United States, affecting approximately seven million women and costing approximately 11.6 billion dollars annually. In addition, antibiotic resistance among E. coli bacteria causing urinary tract infection continues to increase, which greatly complicates treatment. Identifying sources of uropathogenic E. coli and implementing prevention measures are essential. However, the reservoirs of uropathogenic E. coli have not been well defined. This study demonstrated that poultry sold in retail stores may serve as one possible source of uropathogenic E. coli. This finding adds to a growing body of evidence that suggests that urinary tract infection may be a food-borne disease. More research in this area can lead to the development of preventive strategies to control this common and costly infectious disease.

C ommunity-acquired urinary tract infection (CA-UTI) can greatly impact the quality of life of affected individuals and cause considerable economic burden (1)(2)(3)(4). From the U.S. registry data, Taur and Smith estimated that seven million patients seek health care for uncomplicated UTIs in the United States each year (1). Uropathogenic Escherichia coli (UPEC) bacteria represent 80% of the pathogens that cause CA-UTI (4).
Escherichia coli strains that cause infections outside the intestinal tract such as UTIs, bloodstream infections (BSI), meningitis, and wound infections are referred to as extraintestinal pathogenic E. coli (ExPEC) (5). Intestinal pathogenic E. coli (IPEC), such as E. coli O157:H7, and other Shiga toxin-producing E. coli (STEC), are well-recognized major food-borne pathogens, but the sources of ExPEC have remained undefined. The widespread dissemination of a single lineage of uropathogenic E. coli in the United States in 1999 to 2000 led to a suggestion that CA-UTI may be a food-borne disease (6).
Subsequently, a growing number of studies have suggested that nonhuman reservoirs, especially food products, might be an important source of UPEC (7)(8)(9)(10)(11)(12)(13). Ramchandani et al. demonstrated that one E. coli isolate from beef shared 95% similarity by pulsed-field gel electrophoresis (PFGE) genotyping test to that of a human UPEC strain belonging to ST69 defined by multilocus sequence typing (MLST) (8). A study conducted in the Netherlands in 2011 identified TEM-producing ST10 UPEC from human urine samples and chicken (10). In Canada, E. coli ST117 and 131 isolates with related PFGE profiles have been identified in human infections and poultry, and ST95 was separately identified in human UTI cases and in a honeydew melon (7). Another study in 2011 also identified extended-spectrum ␤-lactamase (ESBL)-producing E. coli ST10 from retail meats, rectal swabs from healthy humans, and blood cultures (14). Despite these reports, the impact and magnitude of food as a source of UPEC or ExPEC are not well established.
The objective of this study was to determine the frequency of recovery of ExPEC genotypes from retail meat samples collected from the same geographic region where human cases of CA-UTIs were studied. Here, we assessed the impact of retail meat as a potential source of UPEC causing CA-UTIs in one university community by concurrently comparing the genotypes of E. coli strains isolated from urine and meat samples obtained from retail stores in Northern California counties surrounding the university community in 2016 to 2017.

RESULTS
E. coli isolates from human urine samples. Between September 2016 and May 2017, we collected 1,087 nonduplicate urine samples. E. coli was isolated from 230 (21%) of these samples, yielding 233 E. coli isolates. Of these isolates, 225 were assigned to 61 unique STs; 8 could not be classified into a known ST (see Fig. 2A  Distribution of MLST genotypes of urine and meat E. coli isolates. The most common STs among E. coli isolates from urine samples were ST95, ST127, ST73, ST69, ST131, and ST10. These STs were found in Ͼ60% of all urine isolates collected in our study. The most common genotypes among E. coli isolates from meat were ST117, ST10, ST69, ST58, ST399, and ST101. These six genotypes were found in approximately  20% of E. coli isolates from meat isolates (Fig. 2). The values for Simpson's diversity index were 0.91 for the urine isolate genotypes and 0.99 for the meat isolate genotypes.
Phylogenetic relationship of STs of urine and meat E. coli isolates. We compared the E. coli multilocus sequence types for putative phylogenetic relationship of the STs a Numbers in parentheses represent the percentage of isolates from each type of meat.
based on a minimal spanning tree using globally optimized eBURST (goeBURST) (15, 16) (Fig. 4A). A distinct separation of clusters of STs by nonpoultry meat (green) and human (blue) E. coli STs is observed, whereas the STs of the isolates from poultry (red) are distributed across the entire spectrum of the goeBURST diagram. The human isolates cluster more frequently with the poultry isolates (Fig. 4A). Figure 4B shows the distribution of STs shared by poultry and human isolates. Putative founder STs included both urine and poultry isolates belonging to clonal complex 10 (CC10), CC38, CC88, CC117, and CC569, while founder STs that contained only human isolates belonged to CC73 and CC95. Single-locus variants were considered to belong to the same CC and are connected with thick black lines in the figures. Another poultry CC strain (1167) was connected to human isolates by a solid thick black line. Thus, at the CC level, one more potentially related lineage could be observed.

Food-Borne Uropathogenic Escherichia coli
Antimicrobial susceptibility of urine and meat E. coli isolates belonging to the 12 shared genotypes. We tested isolates belonging to the 12 shared genotypes for antimicrobial susceptibility. Overall, among the 12 shared MLST genotype isolates, 27 (60%) of 43 meat E. coli isolates and 15 (29.4%) of 51 human isolates were pansusceptible (P Ͻ 0.005). Resistance phenotypes found in meat isolates were found in human isolates from four genotypes (ST10, ST69, ST38, and ST131) ( Table 2). For example, E. coli isolates from both meat and urine samples with ST10 were resistant to ampicillin and trimethoprim-sulfamethoxazole (TMP-SMZ). However, with ST10, resistance to gentamicin was found only among human isolates. Among the meat and urine E. coli isolates belonging to ST101 and ST569, resistance to ampicillin was found only in human isolates, and all the meat isolates were pan-susceptible. All meat and urine E. coli isolates belonging to ST1844, ST80, ST906, and ST2562 were pan-susceptible.
␤-Lactamase gene types among ampicillin-resistant strains of the 12 shared genotypes. Ampicillin-resistant ST10 and ST69 E. coli strains containing bla TEM were found among both meat and human isolates (Table 3). Ampicillin-resistant meat and human ST38 isolates did not carry any bla TEM -type, bla CTX-M -type, bla OXA -type, bla SHV -type, or bla AmpC -type genes. Among ampicillin-resistant ST88 isolates, bla TEM -type genes were found only in a human isolate. Ampicillin-resistant ST101, ST131, and ST569 E. coli strains harboring any ␤-lactamase gene were found only among human isolates.  (1) 1.00

DISCUSSION
We prospectively analyzed the genotypes and antimicrobial susceptibility of E. coli isolates from human urine samples and retail meat products that were concurrently collected in the same geographic region of Northern California. We found 12 genotypes shared among human urine and meat samples. E. coli isolates from meat samples showed a greater genotypic diversity than UTI isolates. More than 80% of meat isolates that shared genotypes with human isolates were isolated from poultry sources. Of the 12 shared genotypes, ST10, ST69, and ST131 were among the most common lineages in human urine E. coli isolates. They accounted for 17% of E. coli isolates from urine samples and 10% of those from poultry samples. These STs are recognized worldwide as pandemic lineages of ExPEC (17)(18)(19)(20).
Our findings are consistent with observations reported in other studies assessing E. coli isolates collected in a similar overlapping temporal sampling scheme from food and human sources (7,10,11,14,(21)(22)(23)(24). Studies that concurrently compared genotypes of UTI and meat E. coli isolates from Montreal, Canada, found three STs shared by human and meat sources; chicken was the most common source of these UPEC strains (7,11). They found ST117 and ST131 in chicken, ST69 in pork only, and ST95 in a honeydew melon only. In our study, we found 12 STs shared between UTI and meat samples. The difference in the number of shared STs in this study compared to the Canadian study may be due to differences in source farms, meat processing, distribution, or study design. Nevertheless, it is striking that the three STs found in meat in the Canadian study were also found in meat in our study. In our study, eight STs were shared between human and chicken isolates, and 4 other human STs were shared by other meat types (Table 1). In addition, we found 13 other meat-associated E. coli genotypes documented in the MLST Enterobase database among our human UPEC isolates (https://enterobase.warwick.ac.uk/species/index/ecoli). Thus, we found 25 distinct UPEC STs shown either by this study or reported elsewhere to be isolated from meat sources.
Interestingly, human isolates were significantly more likely to be antibiotic resistant than meat isolates, although 20 (39%) of 51 human isolates and 29 (67%) of 43 retail  (1) ST69 bla TEM type (14) bla TEM type (6) bla TEM type ϩ bla CTX-M group 1 (1) Other (2) ST101 bla TEM type ϩ bla CTX-M group 1 (1) NA ST38 Other (1) Other (1) ST131 bla CTX-M group 1 ϩ bla OXA type (4) NA bla TEM type ϩ bla CTX-M group 9 (1) bla OXA type (1) bla TEM type (3) ST88 bla TEM type (1) Other (1) Other (1) ST569 bla TEM type (1)  meat isolates had identical drug susceptibility patterns. The overall resistance frequency of the meat isolates was lower than that observed among ExPEC isolates from retail meats in Georgia, Maryland, Oregon, and Tennessee surveyed by Xia et al. in 2006 (25). Changes in the management of antibiotics in animal husbandry over time may have contributed to the differences in resistance frequency of E. coli isolates from retail meats. All the ␤-lactamase genes observed among meat isolates were identical and seen among human isolates, but not the other way around, suggesting the direction of transmission of UPEC is mainly from meat to people. A variety of food and environmental sources contain saprophytic organisms that can harbor mobile drug resistance genes (26,27), and thus, it is conceivable that the human intestine is routinely colonized with bacteria that carry such genes. Certainly, there has been considerable interest in the gut microbiota as a potential source of antimicrobial resistance ("the gut resistome") (28)(29)(30)(31). For example, Sommer et al. found that nearly half of the resistance genes in cultured aerobic gut isolates were identical to resistance genes harbored by major pathogens (31). If so, it may be speculated that drug-susceptible meat strains may acquire drug resistance genes in the human intestine.
Phylogenetic analyses by goeBURST showed clustering of human isolates with poultry isolates (Fig. 4). Many of the STs assigned as founder STs contained both human and poultry isolates, which precluded determination of the direction of transmission. Single-locus variants linked to the founder STs included strains from human and poultry sources also, and thus, they did not provide information needed to suggest transmission direction. A larger sample size or comparison by a higher-resolution genotyping method such as whole-genome sequencing may be needed to demonstrate with more confidence if poultry is a source of human UPEC. Nevertheless, given the large number of STs and the well-validated application of MLST to depict phylogenetic relationships of E. coli (32), the observation in this population-based study of an overlap of STs of E. coli isolates from poultry and human urine samples suggests that poultry is a source of UPEC. Of a total of 121 different STs identified from both urine and poultry samples in this study, 10 (8%) overlapped.
ST95 (a pandemic ExPEC lineage) was the most common ST among UPEC isolates in this study, accounting for 39 (17%) of all the suspected cases of UTIs. We did not find any ST95 isolates in retail meat. Nevertheless, Vincent et al. reported the isolation of an ST95 strain from a honeydew melon in Canada (7), and studies in Europe documented ST95 strains from poultry (24,33). ST95, sometimes described as avian-pathogenic E. coli (APEC), has been isolated from both domestic and wild birds (34,35). We also did not find ST73 and ST127 (other pandemic lineages) among the meat E. coli isolates. This finding suggests that UPEC ST73, ST95, and ST127 may have other sources in the United States. Other potential sources include vegetables, fruit, seafood, or environmental sources we did not examine.
An important limitation of our study was that we did not obtain clinical, travel or food histories to link potential sources of the isolates recovered from the patients. However, we note that this study was not designed to show direct connection of the meat isolates as the causative agent of human UTIs. It was mainly designed to determine how much of the meat E. coli isolates belonged to recognized ExPEC lineages isolated from CA-UTI cases in the same geographic region. The UTI isolate collection overlapped 6 months of a 10-month meat collection period. Unlike gastrointestinal illnesses in which the timing of exposure (contaminated food ingestion) can be estimated from the incubation period, such exposure history cannot be easily documented with CA-UTI. The duration of human intestinal colonization by an E. coli genotype can vary greatly (36), and hence, the time of ingestion of a suspected food item cannot be approximated.
We also note that the demonstration of identical E. coli STs in two sources does not necessarily show disease causality. Here, disease causality is assumed for some of the STs because of their membership in recognized ExPEC STs, especially those that are pandemic. As previously mentioned, other studies have suggested food as a potential source of ExPEC. What this study attempted to do further was to assess the magnitude of food as a potential reservoir of ExPEC. This was a population-based study where E. coli bacteria from food and UTI samples were prospectively analyzed contemporaneously in the same geographic setting. We took advantage of an extensive retail meat survey being conducted by the U.S. Food and Drug Administration (FDA) in the very region where the UTI study was simultaneously conducted. To date, no other studies have identified so many ExPEC lineages in meat samples.
In conclusion, this investigation identified 12 MLST E. coli genotypes shared between urine and retail meat E. coli isolates. We found that poultry may be a common reservoir of UPEC and a potential source of its transmission. Additional research is needed to examine the importance and magnitude of food products as a source of human CA-UTIs. With ever-increasing prevalence of CA-UTIs caused by multiple-drug-resistant (MDR) UPEC strains, a better understanding of sources of UPEC could help devise novel public health control interventions to combat this common infectious disease problem.

MATERIALS AND METHODS
Study design. We prospectively cultured E. coli from urine samples from patients suspected to have UTIs at a Northern California university health center and from meat products collected from retail stores in three Northern California counties (Fig. 1).
Isolation of E. coli from urine samples. Between 19 September 2016 and 4 May 2017, we tested urine samples that were collected consecutively from patients suspected to have UTIs seen at the university health service (Fig. 1). All urine samples were first tested at the health service by dipstick, and those specimens found to test positive for leukocytes, nitrates, protein, blood, or glucose were collected for our study and subjected to further microscopic examination. We defined a case of UTI as a symptomatic patient with clean-catch urine specimen that contained more than 10 2 CFU of E. coli per ml. We cultured a 10-l aliquot of urine on a MacConkey agar plate to isolate Gram-negative bacteria and presumptively identified lactose-positive and indole-positive colonies as E. coli for further analysis.
Isolation of E. coli from retail meat products. Between 29 November 2016 and 29 September 2017, we obtained retail raw meat samples: skin-on/bone-in chicken; chicken breast, leg, wing, and thigh; ground turkey; pork chops; and ground beef. Retail meat products were collected as part of the National Antimicrobial Resistance Monitoring System (NARMS), a national public health food surveillance system involving collaboration of state and local public health departments, Centers for Disease Control and Prevention (CDC), the U.S. Food and Drug Administration (FDA), U.S. Department of Agriculture (USDA), and Food Safety Inspection Service (FSIS) (37). Retail meat products (chicken, ground turkey, ground beef, and pork chops) were purchased at local grocery stores in California. The Contra Costa County Public Health Laboratory selected sampling zip codes that are at least within a 50-mi radius of the laboratory and purchased the retail meat samples from local grocery stores within these zip codes by random selection from a list obtained from the Northern California municipality. The university health service is located within this radius. Twice a month, the laboratory purchased a total of 40 meat samples, including 20 samples each of chicken, 10 samples of ground turkey, 5 samples of ground beef, and 5 samples of pork chops. When bone-in/skin-on chicken breasts were not available for sampling, chicken wings were used as the preferred substitution, followed by legs and thighs. The laboratory recorded information, including the store name, store location, brand name, sell-by date, purchase date, and laboratory processing date for purchased meat on monthly log records.
Retail meat products were kept on ice during transport from the local grocery stores to Contra Costa County Public Health Laboratory. Samples were refrigerated and processed within 24 h of purchase. Each retail meat sample was transported aseptically into a sterile sample bag containing 225-ml buffered peptone water. Several chicken pieces were used to achieve a 25-g sample when substituting chicken breast with another chicken part; wing, leg, or thigh. One pork chop from each retail package was aseptically placed into a sterile bag containing 225-ml buffered peptone water. Samples (25 g) of ground beef and ground turkey were aseptically transferred from each package into a sterile bag containing 225-ml buffered peptone water (38). The sealed bags were shaken by a mechanical shaker at 200 rpm for 15 min. The samples were tested for Salmonella and Campylobacter at the Contra Costa County Public Health Laboratory. After the samples were processed, we transported the retail meat samples in the sealed bags to the project laboratory at the University of California, Berkeley, for E. coli isolation.
A 15-ml aliquot of the peptone water from each bag containing meat samples was preincubated in double strength MacConkey broth in a 1:1 dilution at 35°C for 24 h. A separate 10-l aliquot of preincubated peptone water from each bag containing meat was cultured on a MacConkey agar plate to isolate Gram-negative bacteria. Lactose-positive and indole-positive colonies were presumed to be E. coli and selected for further analysis.
Strain typing. Five colonies recovered on plate from each sample were randomly picked and subtyped by the ERIC2-PCR assay, as described previously (39,40). Single colonies from tryptic soy agar plates were selected and inoculated into 2 ml tryptic soy broth and incubated in a shaking incubator for 15 h at 37°C. The 1-ml aliquots of grown cultures were centrifuged, and the pellets were resuspended in a test tube with 350 l of distilled water, boiled for 10 min in a water bath, and then cooled on ice for 2 min. The samples were centrifuged for 2 min at 13,000 rpm, and the supernatants were stored at Food-Borne Uropathogenic Escherichia coli