Urban and agricultural soils in Southern California are a reservoir of carbapenem‐resistant bacteria

Abstract Carbapenems are last‐resort β‐lactam antibiotics used in healthcare facilities to treat multidrug‐resistant infections. Thus, most studies on identifying and characterizing carbapenem‐resistant bacteria (CRB) have focused on clinical settings. Relatively, little is still known about the distribution and characteristics of CRBs in the environment, and the role of soil as a potential reservoir of CRB in the United States remains unknown. Here, we have surveyed 11 soil samples from 9 different urban or agricultural locations in the Los Angeles–Southern California area to determine the prevalence and characteristics of CRB in these soils. All samples tested contained CRB with a frequency of <10 to 1.3 × 104 cfu per gram of soil, with most agricultural soil samples having a much higher relative frequency of CRB than urban soil samples. Identification and characterization of 40 CRB from these soil samples revealed that most of them were members of the genera Cupriavidus, Pseudomonas, and Stenotrophomonas. Other less prevalent genera identified among our isolated CRB, especially from agricultural soils, included the genera Enterococcus, Bradyrhizobium, Achromobacter, and Planomicrobium. Interestingly, all of these carbapenem‐resistant isolates were also intermediate or resistant to at least 1 noncarbapenem antibiotic. Further characterization of our isolated CRB revealed that 11 Stenotrophomonas, 3 Pseudomonas, 1 Enterococcus, and 1 Bradyrhizobium isolates were carbapenemase producers. Our findings show for the first time that both urban and agricultural soils in Southern California are an underappreciated reservoir of bacteria resistant to carbapenems and other antibiotics, including carbapenemase‐producing CRB.

seem to support this hypothesis. However, further studies are needed to fully understand the role of the environment as a reservoir of CRB and carbapenem resistance genes.
Knowledge about the environmental distribution and characteristics of CRB is especially lacking in the United States. For example, there have only been three studies about CRB in freshwater environments in the United States (Ash et al., 2002;Aubron et al., 2005;Harmon et al., 2019) and no specific studies about the prevalence or characteristics of CRB in U.S. soils. However, recent studies in soil and related environmental samples from Africa and Europe suggest that soil may be an underappreciated reservoir of CRB. For example, CRB including CP-CRB have been isolated from agricultural and nonagricultural soil samples from Algeria, Spain, England, Germany, Denmark, and Norway (Gudeta et al., 2016) and Croatia (Hrenovic et al., 2019), as well as from swine and poultry farms from Germany (Borowiak et al., 2017;Fischer et al., 2013), and natural soil samples from Algeria (Djenadi, Zhang, Murray, & Gaze, 2018), among other locations.
Although there are no specific studies about the prevalence or characteristics of CRB in U.S. soils, a few studies suggest that CRB may also be prevalent in U.S. soils. For example, a study on soil samples from the Midwestern United States that used penicillins as selective agents identified three isolates that were carbapenem-resistant (Crofts et al., 2018). CRB and CP-CRB have also been isolated from fecal samples from dairy farms in New Mexico and Texas (Webb et al., 2016), as well as from fecal and environmental samples recovered from a swine nursery in Ohio (Mollenkopf et al., 2017).
These findings are very significant because farm animal feces are routinely used as manure, which may lead to the spread of CRB and carbapenemase genes to the soil, water, and other environments.
To contribute to addressing the information gap about the role of U.S. soils as potential sinks and sources of CRB, we report here the first study specifically aimed at determining the prevalence and characteristics of CRB in soil from the West Coast of the United States. Our findings indicate that both urban and agricultural soils from the highly populated Los Angeles-Southern California area are a significant reservoir of CRB and CP-CRB, which we found to be also resistant to other classes of antibiotics as well.

| Collection of soil samples and isolation of carbapenem-resistant bacteria
We collected 11 different soil samples from 9 different locations in the Los Angeles (California) area between June 2016 and January 2019. The location ( Figure 1) and characteristics of sampling sites are summarized in Table 1. For each sample, we collected surface soil in 50-ml sterile conical tubes and immediately transported the sample to the laboratory. We then weighed 4 g of the soil sample into a sterile 15-ml conical tube, added 10 ml of sterile saline (0.85% NaCl), and vortexed the mixture continuously for 5 min to homogenize the sample and extract the bacteria present in the soil. Soil debris was then removed by centrifugation for 10 min at 1,000 × g, and the supernatant containing the extracted soil bacteria collected for subsequent analyses.
The total count of bacteria was determined using MacConkey medium (Fisher Scientific) as a primary selection for enteric bacteria and gram-negatives, which were the main target in our study. The bacterial count was determined by direct plating of 100 µl of soil supernatant as well as by spot plating of 10 µl of a 10 0 to 10 -4 dilution bank of soil supernatants in sterile saline on MacConkey agar plates, followed by incubation for 24 hr at 37°C. The count of carbapenem-resistant bacteria (CRB) was determined by the same procedure except for using MacConkey agar plates containing 4 µg/ml of meropenem  , 2018). Growth in at least 4 μg/ml of meropenem was confirmed for nearly all patched colonies. In total, we selected 40 CRB isolates-up to 8 distinct CRB isolates per sample, prioritizing those that grew in 16 μg/ml of meropenem-for culturing, longterm storage at −80°C, and preparation of cell suspension templates for PCR, as previously described .

| Identification of CRB by PCR and sequencing of the 16S rRNA gene, and oxidase test
The 40 selected soil CRB isolates were identified following the procedures described in Harmon et al. (2019). Briefly, we used PCR amplification of the 16S rRNA gene of each selected isolate, followed by Sanger sequencing, BLAST analysis (Altschul et al., 1997) of the obtained sequences, and oxidase test analysis. The oxidase test was used to further distinguish between closely related S. maltophilia, which is oxidase negative, and Pseudomonas species, most of which are oxidase-positive (Bergey & Holt, 1994).
Besides, we constructed a phylogenetic tree for each genus isolated in our study (Achromobacter, Bradyrhizobium, Cupriavidus, Enterococcus, Planomicrobium, Pseudomonas, and Stenotrophomonas) to further characterize the taxonomic relationship between our soil isolates across different locations, as well as between our isolates and isolates from previous studies. We used MEGA X 10.1 software (Hall, 2013) to align the 16S rRNA genes and construct phylogenetic

| Identification of carbapenemaseproducing isolates by the CarbaNP and mCIM assays, and detection of the L1 carbapenemase gene in Stenotrophomonas isolates
We identified carbapenemase-producing CRB isolates using the CarbaNP assay (Dortet, Poirel, & Nordmann, 2012a, 2012bNordmann, Poirel, & Dortet, 2012). The assay was performed as PCR amplification to confirm the presence of the L1 carbapenemase gene (bla L1 ) in carbapenemase-producing Stenotrophomonas isolates was performed using the primers and program described by Henriques et al. (2012) to amplify bla L1 as previously described .

| Distribution, frequency, and identification of carbapenem-resistant bacteria in soil samples from the Los Angeles-Southern California area
We analyzed 11 different soil samples from 9 different urban and agricultural locations in the Los Angeles-Southern California area (United States; Figure 1; Table 1). Using meropenem as a selective agent, we found that all soil samples analyzed contained CRB.
The frequency of CRB in these samples was between <10 and 1.3 × 10 4 cfu per gram of soil (Table 1). Interestingly, S4 and S5, the two samples with the most abundance of CRB, were obtained from the soil of a private urban chicken coop, which suggests that animal TA B L E 2 Summary of the number and characteristics of soil carbapenem-resistant bacteria isolated from samples described in Table 1 Genus

Number of CP a isolates Antibiotic resistant/intermediate (number of isolates) b
Achromobacter S10 1 0 MP (1), CF (1) Bradyrhizobium S11 1 1 MP (1) Table 3. feces might be an important contributor to soil CRB. Overall, samples could be classified into those with a low relative frequency of CRB (<1%) compared to the total bacterial counts obtained (S1-S3 and S11-S12; mostly urban soils) and those with a high relative frequency of CRB (18%-80%, urban chicken coop, and most agricultural soil samples) compared to the total bacterial count obtained (S4-S10; Table 2).
We selected a total of 40 CRB isolates for further identification and characterization. We identified them using their 16S rRNA gene sequence as well as phylogenetic analyses (Figure 2 and Figures A1-A7). We also used the oxidase test to distinguish between members of the Stenotrophomonas genus and closely related members of the genus Pseudomonas. We preliminarily identified our isolates as 1 Stenotrophomonas maltophilia isolates ( Figure 2; Tables 2 and 3).
Interestingly, the majority of the urban soil isolates belonged to the genera Pseudomonas and Stenotrophomonas, whereas the most represented agricultural soil isolates belonged to the genus Cupriavidus ( Figure 2). Overall, we identified carbapenem-resistant (CR) Pseudomonas in 5 (all urban soils) out the 11 samples analyzed; CR Stenotrophomonas maltophilia in 3 samples (2 urban and 1 agricultural soil); CR Cupriavidus in 1 urban and 3 agricultural soil samples; and CR Enterococcus in 3 samples (2 agricultural and 1 urban soil),

| Characterization of the antibiotic susceptibility profile of CRB isolates
We next characterized the antibiotic susceptibility profile of the 40 identified CRB isolates using disk diffusion experiments with the two most clinically used carbapenems (meropenem and imipenem) and 4 noncarbapenem antibiotics (cefotaxime, ciprofloxacin, gentamicin, and tetracycline; Tables 2 and 3; and Figure 3). All 40 isolates were resistant to meropenem, confirming them as CRB. Moreover, most of the isolates were also resistant or intermediate to imipenem (55% of the isolates) and cefotaxime (83% of isolates), which although not a carbapenem, it is also a β-lactam (third-generation cephalosporin; Figure 3; Table 3). In contrast, the number of isolates that were resistant to the three different classes of non-β-lactam antibiotics tested was much lower. Overall, 43% and 28% of the CRB isolates characterized were resistant or intermediate to aminoglycoside gentamicin and tetracycline, respectively ( Figure 3; Table 3). Furthermore, only one CRB isolate, identified as Bradyrhizobium elkanii, was resistant to the fluoroquinolone ciprofloxacin ( Figure 3; Table 3). These findings highlight the importance of Southern California soils as reservoirs of CRB, including CRB that are also resistant to other antibiotics.

| Identification of CRB isolates that produce carbapenemases
Given the importance of carbapenemase genes in spreading resistance to carbapenems, we next used the CarbaNP test to determine which CRB isolates produce carbapenemases. Interestingly, 16 out of the 40 CRB isolates tested (40%) were positive for carbapenemase production when tested by the CarbaNP using both meropenem and imipenem, and as confirmed by the mCIM test (Tables 2 and   3). These carbapenemase-positive isolates were 1 Bradyrhizobium elkanii, 1 E. gallinarum, 1 P. putida, 2 P. vranovensis, and all 11 S. maltophilia (Table 3). To our knowledge, this is the first report of carbapenemase production for E. gallinarum and P. vranovensis as well as in the genus Bradyrhizobium. both meropenem and imipenem, and all were confirmed as positives using the mCIM test. Carbapenemase production was inducible on all carbapenemase-producing isolates except for S. maltophilia isolates S1-2 and S1-3-2.

| D ISCUSS I ON
2009; Ssekatawa et al., 2018). This gap in knowledge is especially significant in the United States, where only three specific studies about the prevalence of CRB in the environment, all three in freshwater, have been performed (Ash et al., 2002;Aubron et al., 2005;Harmon et al., 2019). CRB in the United States have also been found in fecal samples from dairy farms in New Mexico and Texas and a swine nursery in Ohio (Mollenkopf et al., 2017;Webb et al., 2016).
Thus, not only clinical facilities but also farms may contribute to spread CRB to the environment. Recent findings in other parts of the world, especially in Europe, have found CRB in agricultural and nonagricultural soil samples (Borowiak et al., 2017;Djenadi et al., 2018;Fischer et al., 2013;Gudeta et al., 2016;Hrenovic et al., 2019) and suggest that soil may be an underrecognized reservoir of CRB.
In the United States, 3 CRB isolates were identified among a collection of penicillin-resistant isolates obtained from soil samples from the Midwestern United States (Crofts et al., 2018). However, studies that specifically address the distribution and characteristics of CRB (a hiking trail sample), the relative frequency of CRB compared to the total bacterial counts obtained was less than 1%, which is similar to the relative frequencies of CRB we had previously observed in freshwater environments from the Los Angeles-Southern California area . In contrast, most agricultural soil samples (and the urban chicken coop soil samples) had a much higher relative frequency of CRB to the total bacterial count (from 18% up to 80% in soil S7, which was obtained adjacent to a produce farm).
Although further studies comparing soil samples from locations at different proximities from farms are necessary, our results support the hypothesis that the use of antibiotics (or the use of manure from antibiotic-treated animals) in farms might contribute to the spread of CRB to the environment (Mollenkopf et al., 2017;Webb et al., 2016), including CP-CRB and CRB also resistant to other antibiotics.
In a previous study, Hrenovic et al. (2019) used a similar approach than the one we used in our study, but a different growth medium (CHROMagar™ Acinetobacter medium with CR102 supplement in their study, compared to MacConkey agar medium supplement with meropenem in our study) and temperature (37°C and 42°C in their study, compared to 37°C in our study) to determine the presence of CRB in different soils samples from Croatia. Hrenovic et al. (2019) found that at 37°C, most soil isolates were S. maltophilia, except for two soil samples in which they were absent. As is further discussed below, S. maltophilia are widespread in soil and other environments, and are intrinsically resistant to carbapenems (Brooke, 2012;Harmon et al., 2019;Tacão et al., 2015;Youenou et al., 2015).
They also found that isolating CRB at 42°C, which suppresses the growth of S. maltophilia, increased the diversity of CRB recovered from their samples, including CRB of potential anthropogenic origin (Hrenovic et al., 2019). In the future, as we expand our studies to additional soil samples and locations, it will be interesting to analyze our samples at both 37°C and 42°C to compare the abundance and diversity of CRB obtained at both temperatures. However, of the 40 CRB isolates identified and characterized in the present study, only 11 of them (from 3 different soil samples) were S. maltophilia (Tables 2 and 3    and have been found before both in clinical settings and in soil, freshwater, animal feces, and other environments (Aubron et al., 2005;Brooke, 2012;Centers for Disease Control & Prevention, 2013b;Djenadi et al., 2018;Gudeta et al., 2016;Hrenovic et al., 2019;Tacão et al., 2015;Webb et al., 2016). However, this is to our knowledge the first report of carbapenem-resistant P. alkylphenolica and P. vranovensis isolates. Resistance to carbapenems in Pseudomonas can occur by different mechanisms such as the production of different carbapenemases, overexpression of efflux pumps, and decreased outer membrane permeability (Papp-Wallace et al., 2011;Rizek et al., 2014;Rodríguez-Martínez et al., 2009 (Brooke, 2012;Harmon et al., 2019;Tacão et al., 2015;Youenou et al., 2015). Using PCR, we could confirm that this carbapenemase gene was also present in all our S. maltophilia isolates (data not shown).
The third most abundant CR soil isolates obtained belonged to the genus Cupriavidus, which we identified in four different samples.

| CON CLUS IONS
In conclusion, our findings show for the first time that soils from the Los Angeles-Southern California area are a previously underappreciated reservoir of different species of CRB that are also resistant to other antibiotics, including carbapenemase-producing CRB. Our study also shows a much higher relative frequency of CRB on most soils from locations adjacent to farms, compared to most soils from urban locations, which suggest a potential role of farms in spreading bacteria resistant to carbapenems and other antibiotics. Ruiz.

CO N FLI C T S O F I NTE R E S T
None declared.

E TH I C S S TATEM ENT
None required.