Antimicrobial Resistance Linked to Septic System Contamination in the Indiana Lake Michigan Watershed

Extended-spectrum β-lactamases confer resistance to a variety of β-lactam antimicrobials, and the genes for these enzymes are often found on plasmids that include additional antimicrobial resistance genes (ARG). We surveyed aquatic environments in the Indiana Lake Michigan watershed in proximity to areas with high densities of residential septic systems to determine if human fecal contamination from septic effluent correlated with the presence of antimicrobial resistance genes and phenotypically resistant bacteria. Of the 269 E. coli isolated from environmental samples and one septic source, 97 isolates were resistant to cefotaxime, a third-generation cephalosporin. A subset of those isolates showed phenotypic resistance to other β-lactams, fluoroquinolones, sulfonamides, and tetracyclines. Quantitative PCR was used to quantify human-associated Bacteroides dorei gene copies (Human Bacteroides) from water samples and to identify the presence of ARG harbored on plasmids from E. coli isolates or in environmental DNA. We found a strong correlation between the presence of ARG and human fecal concentrations, which supports our hypothesis that septic effluent is a source of ARG and resistant organisms. The observed plasmid-based resistance adds an additional level of risk, as human-associated bacteria from septic systems may expand the environmental resistome by acting as a reservoir of transmissible resistance genes.


Introduction
The β-Lactam antibiotic family is the most prescribed class of antibiotics worldwide, making up 65% of the antibiotic sales in the USA alone [1]. β-Lactams encompass a wide range of drug families, including penicillins, cephalosporins, monobactams, and carbapenems, each consisting of various classes and generations. These drugs are all characterized by their active component, the β-lactam ring, which binds to penicillinbinding proteins (PBP) and inhibits peptidoglycan synthesis, resulting in the loss of bacterial viability [2]. Recently, increased detection of β-lactam-resistant Enterobacteriaceae is threatening the efficacy of these drugs in combating bacterial diseases.
First detected in 1979, extended-spectrum β-lactamases (ESBL) are enzymes that bind to and hydrolyze the β-lactam ring, rendering a wide variety of cephalosporin classes ineffective-including cefotaxime [3]. The β-lactamase family itself consists of four classes, A-D, with class A having the most predominant ESBL genes detected globally currently, including bla TEM , bla SHV , and bla CTX-M . The bla CTX-M family is divided into five main phylogenetic groups: CTX-M-1, CTX-M-2, CTX-M-8, CTX-M-9, and CTX-M-25, of which the first three are the most prevalent [4]. The CTX-M-1-15 variant is the most dominant across Asia, Africa, Europe, and North America, followed by CTX-M-9-14 [5]. In addition to the predominant class A genes, class C AmpC β-lactamases, encoded by the CMY-2 gene, are known to confer broad spectrum resistance to β-lactams and may be found on either plasmids or chromosomes [6].
ESBL and CMY-2 encoding genes are primarily located on transferable plasmids and can spread among bacteria via horizontal gene transfer (HGT) mechanisms [7]. The ability to spread antibiotic resistance genes via conjugation allows clinical and non-clinical environments to become reservoirs for the dissemination and proliferation of ARG in the environment, including pathogenic E. coli strains identified globally in surface waters [8][9][10][11]. The widespread use of antibiotics in pharmaceutical and agricultural sectors, and their resulting release into the environment, has continuously driven selection towards resistant organisms in settings such as groundwater runoff, wastewater treatment plants, and local creeks and streams [12]. Natural resistance has been previously observed in aquatic environments; however, the augmentation of the environmental resistome via human fecal pollution may serve to accelerate this process.
Waterbodies in proximity to residential areas with a high density of septic systems, such those within the Indiana Lake Michigan watershed and many other coastal or rural areas, may be impacted by septic effluent in a variety of ways [13,14]. Contamination by commensal bacteria, such as E. coli, has already been linked to overflows from failing septic tanks in coastal waters, well-waters, and even the Great Lakes [15][16][17]. Furthermore, effluent from onsite sanitation systems has been shown to spread antibiotic-resistant bacteria (ARB) into the environment via human fecal pollution [18]. The presence of ARB in aquatic systems presents a serious public health issue, particularly for waters with designated uses such as recreation [19], drinking water intake, and crop irrigation [20]. Therefore, more tracking is required to assess the threat from potential sources of ARB, particularly those that may disseminate resistance genes in environmental reservoirs. In this study, we examined the connection between septic system contamination and the prevalence of antibiotic resistance genes and resistant organisms in nearby aquatic environments.

E. coli Concentrations in Water Samples
E. coli concentrations varied widely within waterbodies at different locations and on different sampling days (Table 1). Over two-thirds of the samples (69%) exceeded the state water quality criterion (235 MPN E. coli/100 mL) for bacterial ambient water quality and full-contact recreational use, and two samples exceeded the capacity of the assay, with a most probable number of E. coli >2419.6 per 100 mL. The Human Bacteroides (HB) assay detected the Bacteroides dorei 16S rRNA gene (DNQ or higher values) at 23 sites and was present in high concentrations (>1000 CN/100 mL) in six different samples from three waterbodies: Deer Creek, Salt Creek, and Trail Creek. Deer Creek had three positive samples, including one sample with values > 6 × 10 5 CN/100 mL, which is nearly 10% of the value of pure septic waste (~7 × 10 6 CN/100 mL).
E. coli were isolated from all sites, with 1-12 isolates obtained per sample. The percent of isolates that could grow in the presence of cefotaxime (CTX) ranged broadly (8.3-100%) and was often variable among sites within the waterbody or samples collected at different times. The number of resistant isolates tended to be higher from sites with higher densities of E. coli and HB copy numbers, but some sites had a low yield of resistant isolates despite significant human fecal inputs. For example, sample E00336 from Salt Creek had only 1/12 E. coli isolates that showed resistance to CTX, despite environmental concentrations of >900 MPN/100 mL and HB > 8000 CN/100 mL.

Genetic Determinants of Antimicrobial Resistance in Isolates and the Environment
All CTX-resistant isolates had one or more plasmids of varying sizes, as visualized by gel electrophoresis of plasmid prep extractions. Genes associated with resistance to βlactams and fluroquinolones were amplified from plasmids and environmental DNA from sample sites ( Figure 2). The genes most frequently detected in CTX-resistant E. coli were CMY-2 (100%) and blaCTX-M-1-15 (96.8%). Other genes were amplified in the majority of isolates, including blaCTX-M-9-14 (77.4%), blaSHV-2 (61.3%), blaKPC (58.1%), and qnrS (70.9%). The mcr1 (colistin resistance) or blaNDM genes were not detected in any of the isolates.
Resistance genes were less frequently detected in community DNA from water samples, which is not surprising given the relatively low abundance of E. coli in natural bacterial communities. CMY-2, which was found in 100% of plasmid extracts and the septic sample, was not found in any environmental DNA ( Figure 2). Similarly, very few samples showed positive amplification for blaCTX-1-15, which was the second most common gene in isolates. The β-lactamase-associated genes were intermittently detected in the same waterbodies at different times or sites within the waterbody. The blaSHV-2 gene was amplified most frequently (in 55.3% of environmental DNA samples), followed by blaCTX-9-14, blaKPC (31%), and qnrS (27.6%).

Genetic Determinants of Antimicrobial Resistance in Isolates and the Environment
All CTX-resistant isolates had one or more plasmids of varying sizes, as visualized by gel electrophoresis of plasmid prep extractions. Genes associated with resistance to β-lactams and fluroquinolones were amplified from plasmids and environmental DNA from sample sites ( Figure 2). The genes most frequently detected in CTX-resistant E. coli were CMY-2 (100%) and bla CTX-M-1-15 (96.8%). Other genes were amplified in the majority of isolates, including bla CTX-M-9-14 (77.4%), bla SHV-2 (61.3%), bla KPC (58.1%), and qnrS (70.9%). The mcr1 (colistin resistance) or bla NDM genes were not detected in any of the isolates.
Resistance genes were less frequently detected in community DNA from water samples, which is not surprising given the relatively low abundance of E. coli in natural bacterial communities. CMY-2, which was found in 100% of plasmid extracts and the septic sample, was not found in any environmental DNA ( Figure 2). Similarly, very few samples showed positive amplification for bla CTX-1-15 , which was the second most common gene in isolates. The β-lactamase-associated genes were intermittently detected in the same waterbodies at different times or sites within the waterbody. The bla SHV-2 gene was amplified most frequently (in 55.3% of environmental DNA samples), followed by bla CTX-9-14 , bla KPC (31%), and qnrS (27.6%).

Environmental Factors Contributing to Antimicrobial Resistance
E. coli concentrations across all samples did not correlate strongly with HB copy numbers (Kendall's τ = 0.103, p = 0.44), indicating that the fecal contamination at some of the sites was not predominately due to human fecal inputs. The percentage of CTX-resistant isolates from the first round of screening negatively correlated with either E. coli (τ =−0.189, p = 0.85) or HB (τ =−0.153, p = 0.26) concentrations, although neither relationship was statistically significant. The only significant correlation between environmental resistance and site characteristics was the number of resistance genes detected by PCR in the environmental DNA and the HB concentration of the sample (τ = 0.51, p < 0.001). The Kendall's τ value of 0.51 indicates a very strong correlation, supporting our hypothesis that septic effluent is a source of resistance elements to the environment.
Principal component analysis (PCA) of the environmental sites was performed using the percent of CTX isolates from the original sample, the presence of specific resistance genes, and E. coli concentrations as the variables (Figure 3). PC1 explained 32.1% of the variation among the sites, and PC2 explained 20.5% of the variation. PC1 indicated that differences in the environmental sites were mainly explained by detection of the resistance genes blaCTX-1-14, blaKPC, blaSHV-2, and qnrS. Variation was explained, to a lesser extent, by PC2, the E. coli concentrations, and the percent of initial CTX-resistant isolates. While HB was not included as variable in the PCA, both the site ordination and the overlay of HB on the input variables shows that it would explain site differences similarly to the four genes listed above. This analysis suggests that the number and type of ARG in the environment drives the differences among the sites more than general fecal pollution or number of single-resistance isolates. The number and diversity of genes detected also correlates well with the degree of human fecal contamination indicated by the HB assay.  Table 3 shows the isolates, their sites, and samples of origin, and the genes that were amplified from each. Genes given in bold are genetic determinants of β-lactam resistance. Genes that amplified in plasmid DNA from isolates were frequently not detected in the environment from which they came, with the exception of the septic sample. Occasionally, a gene was amplified in environmental DNA but not in any of the isolates from that sample. For example, sample E00205 was positive for bla CTX-1-15 and blaKPC. The only concordance among the environmental and plasmid DNA was the highly human-impacted sample E00215 from Deer Creek, with five of the same genes amplified in both and one additional gene from the isolate.

Environmental Factors Contributing to Antimicrobial Resistance
E. coli concentrations across all samples did not correlate strongly with HB copy numbers (Kendall's τ = 0.103, p = 0.44), indicating that the fecal contamination at some of the sites was not predominately due to human fecal inputs. The percentage of CTX-resistant isolates from the first round of screening negatively correlated with either E. coli (τ = −0.189, p = 0.85) or HB (τ = −0.153, p = 0.26) concentrations, although neither relationship was statistically significant. The only significant correlation between environmental resistance and site characteristics was the number of resistance genes detected by PCR in the environmental DNA and the HB concentration of the sample (τ = 0.51, p < 0.001). The Kendall's τ value of 0.51 indicates a very strong correlation, supporting our hypothesis that septic effluent is a source of resistance elements to the environment.  I00729  I00730  I00731  I00732  I00733  I00734  I00735  I00736  I00737  I00738  I00739  I00740  E00307  Burns Ditch  I00654  I00655  I00656  I00657  E00308  Burns Ditch  I00659  P00041  Septic  I00644  I00645  I00646  I00647  I00648  I00649  I00650  I00651 Principal component analysis (PCA) of the environmental sites was performed using the percent of CTX isolates from the original sample, the presence of specific resistance genes, and E. coli concentrations as the variables (Figure 3). PC1 explained 32.1% of the variation among the sites, and PC2 explained 20.5% of the variation. PC1 indicated that differences in the environmental sites were mainly explained by detection of the resistance genes bla CTX-1-14 , bla KPC , bla SHV-2 , and qnrS. Variation was explained, to a lesser extent, by PC2, the E. coli concentrations, and the percent of initial CTX-resistant isolates. While HB was not included as variable in the PCA, both the site ordination and the overlay of HB on the input variables shows that it would explain site differences similarly to the four genes listed above. This analysis suggests that the number and type of ARG in the environment drives the differences among the sites more than general fecal pollution or number of single-resistance isolates. The number and diversity of genes detected also correlates well with the degree of human fecal contamination indicated by the HB assay.

Septic Systems Threaten Environmental Water Quality in the Lake Michigan Watershed
The Lake Michigan watershed in Northwest Indiana is a mix of highly urbanized and rural environments. Unincorporated areas that lack municipal sewerage, as well as many densely populated coastal areas, rely on septic systems for the management of residential waste because of the unique landscape of the coastal dunes. In addition, homeowners may neglect obligations for regular maintenance pump-out or replacement of septic tanks that are past their intended functional age [21,22]. The local soil characteristics in the Indiana coastal zone exacerbate these issues, as over 75% of soils in regions of septic use are not suitable for proper retention of septage to ensure removal of nutrients and pathogens [23]. This creates a potential risk for human health, as septage can move quickly through the highly permeable sandy soils into groundwater and surface water, introducing active pathogens along with pharmaceuticals and personal care products (PPCPs) [13,24].
All waterbodies in our study were adjacent to residential areas with septic systems as the sole or predominant waste management system. Sample sites were located within 500 feet of known septic systems or downstream from septic areas. Therefore, we expected that sites impacted by human fecal contamination would also exhibit bacterial water quality impairment [25]. Furthermore, we hypothesized that at least some of the fecal indicator organisms isolated from environmental sources would share characteristics with those from septic systems [26]. While most samples exceeded the state water quality criterion of 235 MPN E. coli/100 mL (Indiana Administrative Code 327 IAC 2-1.5-8), not all sites with elevated E. coli concentrations also showed evidence of human fecal inputs. Conversely, not all sites with evidence of human fecal contamination exceeded the bacterial water quality criterion (e.g., Dunes Creek E00210, Trail Creek E00209; Table 1). While many nonhuman sources can contribute to fecal contamination [27,28], low E. coli concentrations due to diluted septic inputs may underestimate the health risks associated with these waters.

Septic Systems Threaten Environmental Water Quality in the Lake Michigan Watershed
The Lake Michigan watershed in Northwest Indiana is a mix of highly urbanized and rural environments. Unincorporated areas that lack municipal sewerage, as well as many densely populated coastal areas, rely on septic systems for the management of residential waste because of the unique landscape of the coastal dunes. In addition, homeowners may neglect obligations for regular maintenance pump-out or replacement of septic tanks that are past their intended functional age [21,22]. The local soil characteristics in the Indiana coastal zone exacerbate these issues, as over 75% of soils in regions of septic use are not suitable for proper retention of septage to ensure removal of nutrients and pathogens [23]. This creates a potential risk for human health, as septage can move quickly through the highly permeable sandy soils into groundwater and surface water, introducing active pathogens along with pharmaceuticals and personal care products (PPCPs) [13,24].
All waterbodies in our study were adjacent to residential areas with septic systems as the sole or predominant waste management system. Sample sites were located within 500 feet of known septic systems or downstream from septic areas. Therefore, we expected that sites impacted by human fecal contamination would also exhibit bacterial water quality impairment [25]. Furthermore, we hypothesized that at least some of the fecal indicator organisms isolated from environmental sources would share characteristics with those from septic systems [26]. While most samples exceeded the state water quality criterion of 235 MPN E. coli/100 mL (Indiana Administrative Code 327 IAC 2-1.5-8), not all sites with elevated E. coli concentrations also showed evidence of human fecal inputs. Conversely, not all sites with evidence of human fecal contamination exceeded the bacterial water quality criterion (e.g., Dunes Creek E00210, Trail Creek E00209; Table 1). While many non-human sources can contribute to fecal contamination [27,28], low E. coli concentrations due to diluted septic inputs may underestimate the health risks associated with these waters.

Antibiotic Resistance Genes (ARG) and Antibiotic Resistant E. coli in Septic Effluent
The presence, abundance, and proliferation of antibiotic resistant bacteria and genetic elements that confer resistance in municipal wastewater has been well established [29]. Similarly, surface waters receiving wastewater effluent typically contain more ARB and resistance genes compared to unimpacted waters [30,31]. However, less is known about the contribution of nonpoint sources, such as septic systems [32], and few studies have directly assessed the presence and diversity of resistant organisms and genetic elements in septic tanks or effluent plumes [32]. Therefore, our analysis of a septic tank sample as source material provides key information about resistant E. coli, their genes, and community resistance markers. CTX-resistant isolates from the septic sample were generally more phenotypically resistant and harbored more ARG (Table 3) compared to isolates from most environmental samples. A notable exception was the isolates from the Deer Creek sample, which was heavily impacted by human fecal inputs (HB > 6 × 10 5 CN/100 mL; Table 1). The community DNA sample from the septic system was positive for five of the six antibiotic resistance genes assayed, more than any environmental samples.
Several of the isolates from the septic sample and sites with the greatest human fecal contamination showed the phenotypic and genotypic resistance to multiple classes of antibiotics. However, none of the isolates were resistant to colistin or imipenem based on disk diffusion assays, nor did any show amplification of mcr1 and bla NDM genes that would confer these resistance phenotypes. Neither gene was amplified from any of the community DNA samples, including the septic sample. This is likely due to the relative rarity of these types of resistance compared to cephalosphorins.
Co-occurrence of human and agricultural fecal sources in rural areas further complicates the attribution of ARB and ARG to septic systems [33,34], as farm animals are also a significant source of resistance elements and organisms to the environment [35]. Associations between the presence or abundance of human fecal indicators and ARG and/or resistant isolates are therefore commonly used to establish these links [32,34,36]. Burch et al. found that detection of the ARG sul1, tetA, and tetX and the integron intl1 in well water was significantly correlated to the number of septic systems within the drainage area [34]. A recent report from a multi-year monitoring project of a mixed-used watershed showed that all ARG tested were detected in surface waters only when human fecal markers were also present [32]. Furthermore, ARG in all samples correlated strongly with human fecal contamination, like our findings that water samples with higher HB copy numbers tested positive to more ARG (Figure 3). Mapping of ARG hotspots revealed that septic systems were the most likely source of the inputs [32].

Mobile Resistance Genes in Environmental Isolates Increase Potential Health Risks
Our study took a One Health approach to monitoring a common and increasingly frequent form of AMR by assessing the crossover of β-lactam-resistant bacteria from an unregulated source of human fecal bacteria (septic systems) into the aquatic environment. We were specifically interested in E. coli bearing plasmids encoding ESBLs, as the CDC categorizes ESBL-producing Enterobacteriaceae a "serious threat" [37]. ESBL genes are rarely found on E. coli chromosomes [38] since their initial jump from the chromosome of a Kluyvera spp. to a conjugative plasmid [39]. Therefore, it is not surprising that CTX-resistant E. coli isolated from our environmental and the septic sample harbored plasmids with genes conferring this resistance (Table 3).
While most E. coli are non-pathogenic commensals, pathogenic ESBL-producing MDR phenotypes are common and are responsible for many nosocomial infections [40]. Such organisms have been isolated from drinking water [40,41], recreational waters [19], and surface waters [11]. One of the major concerns associated with the introduction of septic system effluent into environmental waters is that they will expand the reservoir of resistance genes [41]. ESBL genes are often associated with conjugative plasmids [7,40,41], introducing the additional risk of transfer to susceptible organisms. Studies have shown that resistance plasmids can be transmitted to human commensals from a variety of sources, including companion animals [42], exogenous sources such as contaminated food [43], or drinking water. These risk scenarios are highly relevant to the Lake Michigan watershed, as waterbodies throughout the region are used for recreational activities such as swimming and fishing. Furthermore, unincorporated areas that lack municipal sewer lines may rely on well water for their drinking water source. The plasmid-based resistance observed in our environmental isolates may expand the environmental resistome by acting as a reservoir of transmissible resistance genes.

Study Sites, Sample Collection
The sites sampled in this study were part of a region-wide survey of the impacts of septic systems on adjacent waterbodies, specifically how they may cause beneficial use impairment due to bacterial pollution. E. coli concentrations were determined for each site to assess the presence and severity of fecal pollution. The Human Bacteroides (HB) assay was used to determine if human feces were present and therefore contributed to the E. coli load. We hypothesized that sites with higher levels of human fecal pollution would tend to have a greater number and diversity of resistant E. coli.
Samples were collected in the late spring through fall of 2019 from 12 waterbodies, including beach sand and coastal waters of Lake Michigan, and rivers and creeks that are tributaries to the lake ( Figure 4, Table 1). The sampling sites included tributary waters in residential areas with a high density of septic systems. Water samples were collected into sterile, 1L Nalgene ® bottles (Thermo Fisher Scientific, Waltham, MA, USA). Each bottle was rinsed three times with the water to be sampled before the final composite sample was obtained. Samples were stored on ice during transport to the lab and were processed within 6 h of collection.
Antibiotics 2023, 12, x FOR PEER REVIEW 10 of 16 that resistance plasmids can be transmitted to human commensals from a variety of sources, including companion animals [42], exogenous sources such as contaminated food [43], or drinking water. These risk scenarios are highly relevant to the Lake Michigan watershed, as waterbodies throughout the region are used for recreational activities such as swimming and fishing. Furthermore, unincorporated areas that lack municipal sewer lines may rely on well water for their drinking water source. The plasmid-based resistance observed in our environmental isolates may expand the environmental resistome by acting as a reservoir of transmissible resistance genes.

Study Sites, Sample Collection
The sites sampled in this study were part of a region-wide survey of the impacts of septic systems on adjacent waterbodies, specifically how they may cause beneficial use impairment due to bacterial pollution. E. coli concentrations were determined for each site to assess the presence and severity of fecal pollution. The Human Bacteroides (HB) assay was used to determine if human feces were present and therefore contributed to the E. coli load. We hypothesized that sites with higher levels of human fecal pollution would tend to have a greater number and diversity of resistant E. coli.
Samples were collected in the late spring through fall of 2019 from 12 waterbodies, including beach sand and coastal waters of Lake Michigan, and rivers and creeks that are tributaries to the lake ( Figure 4, Table 1). The sampling sites included tributary waters in residential areas with a high density of septic systems. Water samples were collected into sterile, 1L Nalgene ® bottles (Thermo Fisher Scientific, Waltham, MA, USA). Each bottle was rinsed three times with the water to be sampled before the final composite sample was obtained. Samples were stored on ice during transport to the lab and were processed within 6 h of collection.

Water Quality Assessment
Colilert Quanti-tray assays (IDEXX™, Westbrook, ME, USA) were performed following the manufacturer's instructions to estimate the most probable number of E. coli concentrations for each water sample. Water samples (100 mL) were also processed via vacuum filtration onto 0.45 µm pore-size filters (Thermo Fisher Scientific, Waltham, MA, USA) and then placed onto modified membrane Thermotolerant E. coli (mTEC) agar (Fisher Scientific, Waltham, MA, USA). The plates were incubated for 24 h at 37 • C to select for cefotaxime-resistant E. coli. Colonies of E. coli were differentiated from other enteric or environmental bacteria by their purple color.
Water samples from each site were processed in duplicate by vacuum filtration through 0.22 µm pore-sized filters (EMD Millipore Membrane Filters™, Thermo Fisher Scientific, Waltham, MA, USA). Filters were frozen and stored at −80 • C for future DNA extraction.

Sample Processing and Purification of Cefotaxime-Resistant Escherichia coli Isolates
Up to 12 E. coli colonies per sample were randomly selected from the modified mTEC plates and streaked onto Eosin Methylene Blue (EMB) agar (Thermo Fisher Scientific, Waltham, MA, USA) to obtain pure isolates. E. coli growth on EMB was confirmed by the presence of metallic green or dark purple colonies. Isolates were plated on modified mTEC supplemented with 4 µg/mL cefotaxime (CTX-4). Isolates that were able to grow on CTX-4 were quadrant-streaked as a final purification step to produce single colonies of pure, resistant isolates. Isolates were then stored in glycerol at −80 • C.

Minimum Inhibitory Concentration and Multi-Drug Resistance Tests
E-test strips were used to determine the minimum inhibitory concentration (MIC) of cefotaxime and cefepime for a subset of 36 isolates. Isolate densities were standardized in liquid medium using a 0.5 McFarland Standard and spread onto Mueller Hinton II agar (Fisher Scientific, Waltham, MA, USA) using the tri-swab method to ensure proper coverage. Cefotaxime and cefepime E-test strips (Liofilchem™ MTS™, Thermo Fisher Scientific, Waltham, MA, USA), with concentrations from 0.016 to 256 µg/mL, were placed in the middle of the streaked plates using sterile forceps. The plates were then incubated overnight at 37 • C. The concentration at which growth was inhibited was compared to the Clinical and Laboratory Standards Institute (CLSI) guidelines to classify each isolate as either resistant or susceptible.

Environmental DNA and Plasmid Extractions
Frozen filters from environmental waters were crushed with a sterile spatula, and DNA was extracted from the filter pieces with a Powerlyzer Powersoil extraction kit (Qiagen ® , MD, USA). The manufacturer's protocol was modified slightly by adding the class beads and lysis reagent to the filter tube. Extracted DNA was quantified on a NanoDrop2000™ (Thermo Fisher, MA, USA) and stored at −20 • C until analysis.
Plasmid extractions were performed to identify antibiotic resistance genes (ARG) that could be mobilized in the environment. LB Broth (Fisher Scientific, Waltham, MA, USA) supplemented with both cefotaxime and yeast extract (19 g/L) was used to grow 100 mL isolate cultures for 16 h at 37 • C in a shaking incubator set to 200 rpm. The ZymoPureII™ Plasmid Midiprep kit (Zymo Research) was used with the vacuum filtration protocol for the extraction of plasmids. Plasmid DNA (30 µL) was eluted, and the purity and concentration of DNA samples were quantified using the NanoDrop2000. Extracted plasmids were stored at −20 • C. Gel electrophoresis was performed to estimate the number and approximate size of plasmids from each isolate.

qPCR Assays for Antibiotic Resistance Genes and Human Fecal Marker Genes
Gene amplification was performed via qPCR to detect the presence of genetic resistance determinants. Plasmids extracted from CTX-resistant isolates and DNA from the environmental water samples were tested for bla CTX-M-1-15 , bla CTXM-9-14 , bla SHV2 , bla KPC , bla NDM , CMY-2, qnrS, and mcr1 using primer sets and cycling conditions previously described (Table 4; [45][46][47][48][49]). Final melt curve analysis and gel electrophoresis were performed to confirm that amplicon characteristics that matched those of the positive controls. Human fecal contamination in environmental samples was quantified using the EPA Human Bacteroides HF183/BacR287 TaqMan ® (Thermo Fisher Scientific, Waltham, MA, USA) probe qPCR assay (HB Assay; Table 4) originally described by Green et al. [50]. All DNA samples were analyzed in duplicate, and all qPCR runs included a standard curve, technical replicates, sample spike qPCR, no template controls, and an internal standard.

Data Analysis
All data analyses were conducted in R [51]. Non-normality of data distribution was determined using the Shapiro-Wilk normality test in the stats package. Non-parametric correlations were performed using the Kendall method in cor.test. Principal Component Analysis (PCA) was conducted with FactoMineR [52].

Conclusions
This study investigated the impacts of septic systems on waterbodies within the southern Lake Michigan watershed with respect to bacterial contamination and, more specifically, the introduction of antibiotic-resistant bacteria into the natural environment. Through comparison of the genotypic and phenotypic traits of E. coli isolates from the environment and a septic source, quantification of human fecal contamination, and detection of ARG in the environment, we saw a strong correlation between the diversity of resistance potential and the impact of effluent from nearby septic systems. Our results are somewhat limited by the simple detection versus the quantification of ARG from environmental DNA. Future studies could further explore the relative abundance of ARG in each environment compared to the total bacterial abundance to better estimate the proportion of resistant organisms and provide a better correlation with the HB human fecal markers. Despite these data limitations, our study does provide further support for the observation [32] that septic systems can contribute to the expansion of the aquatic resistome.