A survey of antimicrobial resistance in Enterobacteriaceae isolated from the Chesapeake Bay and adjacent upper tributaries

Abstract In recent years, the rise in antimicrobial resistance (AR) in the healthcare setting as well as the environment has been recognized as a growing public health problem. The Chesapeake Bay (CB) and its upper tributaries (UT) is a large and biologically diverse estuary. This pilot study evaluated the presence of AR of gram‐negative bacteria isolated from water samples collected at various sites of the Chesapeake Bay. Bacterial organisms were identified and antimicrobial susceptibility testing was performed by phenotypic and genotypic methods. Ninety‐two distinctly different gram‐negative bacteria were identified; Klebsiella pneumoniae, Enterobacter cloacae, Enterobacter aerogenes, Serratia marcescens, and Escherichia coli were most often isolated. Serratia marcescens was more frequently isolated in samples from the UT compared to the CB. Antimicrobial resistance was more frequently detected in organisms from the CB by phenotypic and genotypic methods. Antimicrobial resistance to ampicillin, imipenem, tetracycline, and chloramphenicol were the most frequently observed resistance patterns. ACT‐1, CMY, and SHV genes were the most frequently detected resistance genes, with predominance in organism isolated from the CB. The results from this study emphasize the importance for further developing comprehensive surveillance programs of AR in bacterial isolates in the various environments, such as recreational and other water systems.


| INTRODUC TI ON
During the recent two decades, antimicrobial resistance (AR) in bacteria has been recognized as a critical public health problem (Hawken & Snitkin, 2019). Besides the fundamental utility of antibiotics in improving human health, antibiotics are widely used for treatment and prevention of infections in animals and plants, as well as for promoting growth in animal farming (Cabello, 2006;McManus, Stockwell, Sundin, & Jones, 2002;Singer et al., 2003;Smith, Harris, Johnson, Silbergeld, & Morris, 2002). However, during the past two decades, development and spread of AR in many bacteria has been recognized with increasing frequency, and now presents a global health crisis (Hawken & Snitkin, 2019;Wattkins & Bonomo, 2016). Each year in the United States alone, approximately 2 million infections due to AR bacteria occur, resulting in at least 23,000 deaths (Centers for Disease Control & Prevention, 2013). The economic impact of AR is tremendous and healthcare costs due to infections with AR bacteria continue to increase and pose a significant burden on societies (Wattkins & Bonomo, 2016). While AR has been described in almost all bacterial pathogens, AR and specifically emerging multidrug resistance (MDR) among gram-negative bacteria represents a unique and immediate threat (Hawken & Snitkin, 2019;Lautenbach & Perencevich, 2014). Furthermore, gram-negative bacteria and their associated AR have certainly the highest implication for human health, and would be expected to be most likely acquired from external sources. The development of AR is a complex process that is driven by multiple factors, including overuse of antimicrobial agents in healthcare, inadequate adherence to infection control practices, global travel and tourism, antimicrobial overuse in agriculture, and poor sanitation and contaminated water systems (Wattkins & Bonomo, 2016). In recent years, these complex factors contributing to AR have been well recognized and become important components of monitoring AR evolution in the greater context of global health, as exemplified in One Health Initiative (2018).
Previous studies and reports suggested that as much as 80% of the total production of antimicrobial agents in the United States is used in agriculture, animal farming, and veterinary medicine (Aarestrup, 1999;Bates, Jordens, & Griffiths, 1994;Ferber, 2003;FDA, 2014;NRC, 1999;Witte, 1998). Furthermore, a significant portion of those antimicrobial agents used in agriculture and animal husbandry are also important antimicrobial agents used for the treatment of common infections in humans (FDA, 2014;vanBoeckel et al., 2015). Therefore, it comes as no surprise, that the appearance of AR is directly linked to the use and overuse of antimicrobial agents; this fact was previously described for clinical healthcare settings as well as veterinary medicine and farming (Aarestrup, 1999;Bates et al., 1994;Chantziares, Boyen, Callens, & Dewulf, 2014;Ferber, 2003;NRC, 1999;Singer et al., 2003;Smith et al., 2013;vanBoeckel et al., 2015;Witte, 1998).
In addition, the World Health Organization (WHO) previously recognized that the use of antimicrobial agents in animals clearly affects the occurrence of AR in bacteria responsible for infections in humans (Martinez, 2009;WHO, 2000). While it is important to acknowledge that many of the antimicrobial agents currently used in human healthcare settings were initially discovered as compounds produced by various environmental microorganisms, it is equally important to understand that many of the AR genes in human pathogenic microorganisms, commonly acquired by horizontal gene transfer, also originated in environmental organisms (Aminov, 2009;Aminov & Mackie, 2007;Campagnolo et al., 2002;Martinez, 2009). Therefore, it seems intuitive that environmental pollution by antimicrobial agents and their residues serve as a contributing factor in the evolution of AR in natural microbial ecosystems (Martinez, 2009). In addition, the run-off from agriculture, hospitals and other healthcare settings, as well as wastewater treatment plants, among other sources, may also contribute to increased presence of antimicrobial agents, their residues, as well as pathogenic bacteria (Aminov, 2009;Aminov & Mackie, 2007;Börjesson, Matussek, Melin, Löfgren, & Lindgren, 2010;Campagnolo et al., 2002;Hooda, Edwards, Anderson, & Miller, 2000;Kulkarni et al., 2017;Martinez, 2009;Rosenberg Goldstein et al., 2012). However, despite the growing number of studies and associated evidence, little is known about the overall effects of antimicrobial agents on the dynamics of the various larger ecosystems or the microspheres within such systems.
We previously described unusual resistance patterns in bacteria implicated in wound infections in patients seen within our hospital network (Parrish, Luethke, Dionne, Carroll, & Riedel, 2011). The infections were related to recreational activities on the Chesapeake Bay waters. These findings and others alike reported in the literature (Ceccarelli et al., 2015;McNicol et al., 1980;Morgan, Guerry, & Colwell., 1976;Shaw et al., 2014;Shaw, Sapkota, Jacobs, He, & Crump, 2015), were the basis to conduct a pilot survey of the Chesapeake Bay and upper tributaries to assess bacterial diversity and AR. Specifically, this study evaluated antimicrobial susceptibility patterns for enteric gram-negative bacteria isolated from water samples obtained from various locations of the Chesapeake Bay and its upper tributaries in Maryland. Our findings provide a more comprehensive analysis of such a kind for bacterial AR in gram-negative bacteria in the Chesapeake Bay area.

| MATERIAL S AND ME THODS
We conducted a pilot survey of the Chesapeake Bay and its upper tributaries in July 2012 to assess the presence of AR in enteric gram-negative bacteria. Considering the rapid emergence of AR in gram-negative bacteria and the associated increasing risk to human health, as described above, we focused this pilot survey initially on isolation of gram-negative bacteria. Water samples were collected at 10 locations as outlined in Figure 1. These 10 sites were selected based on those included in a previous survey conducted and published in 1976 (Morgan et al., 1976), therefore serving as a comparative reference; in addition, sites were selected because of their close proximity to industrial agriculture, human habitation, and wastewater treatment plants. Specifically The harbor serves as a major tourist attraction, including historic ships at anchor, piers for cruise ships, water taxis, and other tourist water activities. All of the collection sites in the upper tributaries are located in the State of Maryland, with the exception of the site at the Shenandoah River, which is in West Virginia. All sites in the upper tributaries were in close proximity to agricultural and farming operations. For each of these ten locations, global positioning system (GPS) coordinates were recorded for each sampling site as shown in Figure 1. Water sampling, sample processing, organism identification, and antimicrobial resistance testing were performed as described in the following detailed procedures. One sampling site located at the Point of Rocks, on the Potomac River, was located immediately downstream from the Point of Rocks wastewater treatment facility, whereas the other Potomac River site, was located immediately upstream from the Shepherdstown wastewater treatment facility.

| Water sampling procedure
For the collection sites in the Chesapeake Bay, surface water samples were collected just below the surface using 500-ml sterile, glass, screw-capped bottles that were opened immediately after being submerged in the water. Sampling was performed in two steps; one sample was collected within a few meters from the shore (proximal sample), and a second sample was collected further away from the shore, at a distance of approximately 150 m (distal sample). Sampling at the upper tributary sites (i.e., rivers and creeks) was performed in a similar fashion; however, only one sample was collected several meters into the river, where the water was deeper and the current was more rapid than in closest proximity to the riverbank. All personnel collecting water samples wore clean gloves to avoid contamination of bottles and/or samples with human skin flora. Upon completion of the collection process at each site, all samples were placed on ice and transported back to our laboratory for further processing.

| Physical and chemical water quality measurements
Water column depth, water temperature, and water pH were measured on every sampling date and at every sampling collection.

| Water and sample processing
Water samples received at the laboratory were poured through glass filtration units, using vacuum filtration through presterilized 0.2-µm bacterial recovery filters (Millipore, Billerica, MA).
Subsequently, each membrane filter was cut into sections using sterile scissors and forceps; the sections were then placed into Trypticase soy broth (Becton, Dickinson & Co., Sparks, MD) and incubated at 35°C for 24 hr. After the initial incubation period, aliquots from broth tubes exhibiting turbidity were subcultured onto sheep blood agar and MacConkey agar. Agar plates were incubated at 35°C in 5% CO 2 for 24 hr. Broth tubes that did not demonstrate turbidity at 24 hr were re-incubated for an additional 24-hr period.
If these broth tubes did not demonstrate turbidity after the additional incubation period, subcultures onto sheep blood agar were performed before the sample was deemed negative for bacterial growth.

| Organism identification and antimicrobial resistance testing
All distinct colonies recovered from positive broth cultures and on the initial agar media were further processed for organism identification. Organism identification was performed using routine microbiological methods, including use of additional differential and selective agar media (e.g., Hektoen enteric agar, Columbia agar with colistin and nalidixic acid, and MacConkey agar), Gram stain, various bench tests (e.g., catalase, oxidase), and commercial bench identification methods, including the API 20E and API 20NE tests (bioMérieux). All Antimicrobial susceptibility testing (AST) was performed for all gram-negative enteric bacteria as well as gram-negative nonfermentative bacteria, using disk-diffusion and E-test methods, following CLSI guidelines and standard laboratory procedures (CLSI, 2012(CLSI, ,2016, determining either zone diameters (mm) for growth inhibition or minimum inhibitory concentrations, MICs, (µg/ml). These results were then interpreted as susceptible AST results, it is common practice to select antimicrobial agents that tested "S" for treatment; however, antimicrobial agents that resulted in the interpretations "R" are not considered for treatment. It is furthermore common practice to consider antimicrobial agents that tested "I" as being less suitable for treatment,

| Statistical analysis
Phenotypic and genotypic AR were summarized using frequencies and percentages. The percent of organisms with AR to various antimicrobial agents in the Chesapeake Bay were compared to those in the Upper Tributary using the Fisher's exact test. Statistical analysis was performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC).
All tests were two-sided and p < 0.05 was considered statistically significant.   Tables 4 and 5. Overall, resistance genes were detected in 71% (35/49) of organisms isolated from CB sample sites, while only 27% (11/41) of all organisms isolated from UT sample sites (Table 5) had resistance genes detected. This difference was statistically significant (p < 0.001). The CMY-2 and CMY-70 genes, both of which belong to large families of plasmid-mediated Amp-C β-lactamases
chromosomally encoded AmpC β-lactamase genes (ACT-1/MIR-1, CMY-2, CMY-47, CMY-70) were detected in various organisms (Escherichia coli, C. freundii, and Enterobacter species); in addition, the SHV-G156 and SHV-G238 genes were detected in 92% (12/13) and 54% (7/13) of K. pneumoniae isolates, respectively. In addition, the SHV-G156 gene was also detected in one isolate of Enterobacter aerogenes and Raoultella terrigena. The predominance of detecting the SHV-G156 gene in isolates from the CB sites compared to the UT sites was statistically significant (p = 0.001). In addition, the difference in detection of the ACT-1 gene isolates from the CB compared to the UT sampling sites was also statistically significant (p = 0.007; see Table 5).

| D ISCUSS I ON
This study investigated the occurrence of AR in gram-negative bacteria isolated from various sample sites at the Maryland CB and its adjacent UT. We identified resistance against several antimicrobial agents in a variety of gram-negative Enterobacteriaceae isolated from various surface water samples from a variety of sampling sites surrounding the CB and UT. Generally, a larger number of bacterial organisms were recovered from the sampling sites in the CB (including the Baltimore Inner Harbor) when compared to the UT sampling sites. Interestingly, phenotypic AST revealed generally no statistically significant difference in distribution of antimicrobial resistant organisms between the CB and UT sampling sites, albeit, a statistically significant difference was observed for imipenem resistance.
Imipenem resistance was more often observed in organisms isolated from the CB samples. However, a statistically significant difference was more frequently observed in the detection of genotypic resistance from organisms isolated from the CB samples compared to the UT samples; this difference was most pronounced for the detection of plasmid mediated AmpC β-lactamases as well as the plasmid mediated SHV-type extended-spectrum β-lactamases. The results from our study demonstrate that Enterobacteriaceae occurring in surface waters are important reservoirs for AR for a number of antibiotic classes, including β-lactam antimicrobial agents, carbapenems, tetracyclines, and chloramphenicol. Several of the isolates showed high levels of resistance to select antimicrobial agents, specifically some of the β-lactam antimicrobial agents. The antimicrobial agents that were tested in this study represent those that may be typically used to treat a variety of clinical infections. Some of these infections (e.g., wound infections) could be acquired through recreational activities when wounds become contaminated with water, soil, or other environmental sources, which may in turn contain bacteria that are resistant to the various antimicrobial agents that one would use for treatment. Furthermore, bacteria that are harboring AR and that are present in surface waters could be the source for transferring such resistance genes to yet other, still susceptible bacteria that are present in the water. Lastly, surface waters may not be just be important because of human recreational activities, but surface waters and wastewaters are also affecting agriculture and animal husbandry; in these situations, antimicrobial-resistant bacteria may be further spread among animals for food production. The threat to human and global health is significant: humans may become infected by AR bacteria from livestock, or by consumption of contaminated food and water; humans may also become colonized with such AR bacteria; and finally, AR may be spread through the above referenced means on a more global scale (Wattkins & Bonomo, 2016).
TA B L E 5 Genotypic detection of antimicrobial resistance of bacteria, overall and by location Mycobacterium marinum isolated from a patient with a soft tissue infection following a fish-hook injury after fishing activities on the Chesapeake Bay (Parrish et al., 2011). With regard to mycobacteria,  (McNicol et al., 1980). Furthermore, the authors commented on the fact that water samples from various sites of the CB had a high level of pollution with enteric gram-negative bacteria. Interestingly, these findings are consistent with the findings in one of the earlier studies investigating the water quality and AR in the CB (Morgan et al., 1976). In recent years, investigations of water quality and specifically the detection of AR in enteric bacteria as well as other bacteria commonly isolated from aquatic environments have been recognized as important components of monitoring AR evolution in the greater context of the One Health Initiative (Allen et al., 2011;Edge & Hill, 2005;Hamelin et al., 2006;Stange, Sidhu, Tiehm, & Toze, 2016;Wright, 2010;One Health Initiative, 2018). One Health has been defined as "the collaborative effort of multiple disciplines working locally, nationally, and globally to attain optimal health for people, animals, and the environment" (One Health Initiative, 2018).
Here, we specifically referenced studies investigating the detection of bacterial organism burden and AR in various aquatic environments as a comparison to our study design. In the studies referenced here, the investigators detected predominantly resistance to the following classes of antimicrobial agents: ampicillin/amoxicillin, tetracyclines, trimethoprim/sulfamethoxazole; it is of note that resistance to aminoglycosides and chloramphenicol was also detected in some bacterial organisms (e.g., Vibrio) but not in others (e.g., Aeromonas). In comparison to all these referenced studies, the findings in our study of the CB and UT isolates have similarities to the findings in some but not all of the other studies, specifically with respect to identifying resistance to ampicillin/amoxicillin, tetracycline, and chloramphenicol. It is also of interest to note that the majority of the isolates in our current study demonstrated resistance to chloramphenicol, whereas bacterial isolates from earlier studies (McNicol et al., 1980;Morgan et al., 1976) of CB water samples did not detect such resistance with the exception of Shaw et al. whose study described chloramphenicol resistance in Vibrio species (Shaw et al.., 2014). While such differences in results from these various studies are readily apparent, it is important to consider that most studies, including our own study, were only conducted during a limited time period and/or season. None of these studies present data for longitudinal, long-term, ongoing surveillance over the course of an entire year or even years. Furthermore, most studies focused on specific groups of bacteria, for example, Vibrio species (Shaw et al., 2014), Aeromonas species (McNicol et al., 1980), Enterobacteriaceae (Morgan et al., 1976). It is likely that the diversity of bacterial organisms present in various aquatic environments will undergo seasonal changes and is further influenced by a variety of other factors (e.g. human activities, environmental factors, weather events, etc.).
Additional studies that also include longitudinal surveillance of AR will be necessary to establish a better and more in-depth understanding of the AR in bacterial organisms in various aquatic environments and their subsequent potential impact on human health.
Considering the One Health concepts with respect to emerging AR in various clinical and nonclinical settings and its importance to human health, we believe that our data underscore the importance of efforts to monitor aquatic and other environments with respect to the presence of enteric and other bacteria with AR, as such bacteria are likely to contribute to the growing burden of AR globally.
Humans continue to be exposed to aquatic and other environments through various activities, including recreational activities of all kinds, and one should recognize the potential of accidental injuries with subsequent exposure and possible infection with pathogenic bacteria with higher levels of AR. The CB is North America's largest and most biologically diverse estuary; however, since the 1970s its water quality has declined significantly (Ruhl & Rybicki, 2010;Savage & Ribaudo, 2013;Wainger, 2012). This decline of water quality has been largely attributed to the human population increase as well as aggressive agricultural production and animal husbandry practices (Bernhardt & Pelton, 2017;Land, 2012;Randall, 2003).
Since the 1980s, the CB and its adjacent watershed has been the focus of numerous State and Federal initiatives, mainly focused on the reduction of nutrient pollution from agriculture and other sources of human activities. Success of such measures has been reported to a limited extent (Ruhl & Rybicki, 2010;Savage & Ribaudo, 2013;Wainger, 2012). The results from our study demonstrate that in various locations around the CB, a significant amount of enteric bacteria are present in the surface water. Such bacteria are likely to originate from human and animal biowaste; in addition, we identified the presence of AR against various antimicrobial agents in these bacteria. Some of these antimicrobial agents are commonly used in animal husbandry as growth promoters and to prevent and/or treat infections of farm animals.
We recognize that our study has several limitations; samples were collected only at a single point in time, and no consecutive sampling was performed. Furthermore, our study focused on gramnegative bacteria, specifically Enterobacteriaceae, which were characterized for genetic determinants of resistance which indicated differences between most strains suggesting they were not clonal isolates. A more detailed genomic characterization of each isolate was beyond the scope of this study and was not performed, thus the extent of clonality was not determined. However, these limitations may suggest potential avenues for future research to augment the understanding of emerging AR in bacterial isolates from various environmental sources in relation to human activities.
In summary, the results from this study contribute to a better understanding of AR and the mobility of resistance genes in organisms isolated from aquatic environments, specifically those in close proximity to areas of human recreational and other activities, as well as animal husbandry activities. While measures for management and control of water quality have been partly implemented, our data suggest that such initiatives could be augmented and broader surveillance of water quality should also include the assessment of bacterial enteric burden together with surveillance of antimicrobial resistance in such bacterial isolates. Longitudinal, sustained surveillance studies may be necessary to further enhance our understanding of the complex issues related to the emergence of antimicrobial resistance at the interface been the environment and human activities.

ACK N OWLED G M ENTS
This work was funded in part from a grant from the Sherrilyn and Translational Sciences of the National Institutes of Health. In addition, the authors thank the following volunteers for their contributions during the water sampling process: Philip Pasco and Erin Weddle. Finally, we thank OpGen, Inc. for providing us with the Acuitas Resistome test for the analysis of all bacterial isolates in this study. The mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation and/or endorsement of these products by the authors, singly or collectively.

CO N FLI C T O F I NTE R E S T S
Dr. Stefan Riedel serves as a member of the Clinical Advisory Board and as a paid consultant to OpGen, Inc., which provided the Acuitas Resistome Test for the analysis of the resistance genes identified in the bacterial organisms in this study. The mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation and/or endorsement of these products by the authors, singly or collectively.

E TH I C S S TATEM ENT
The study protocol was reviewed and approved by the Johns Hopkins University, School of Medicine, Institutional Review Board (IRB) and considered non-human subject research. Therefore no further IRB review was required.

DATA ACCE SS I B I LIT Y
The data that support the findings of this study are not publicly available. However, data are available from the authors upon reasonable request and with permission of Drs. Nicole Parrish and Stefan Riedel, who are the co-principal investigators of this study.