Analysis of adult damselfly fecal material aids in the estimation of antibiotic-resistant Enterobacterales contamination of the local environment

Antibiotics

All antimicrobial agents, including cefotaxime (CTX), ampicillin (AMP), kanamycin (KAN), tetracycline (TET), chloramphenicol (CHL), ciprofloxacin (CIP), and sulfapyridine (SPY), were purchased from Sigma-Aldrich. In addition, we selected representative antibiotics that are used as the major veterinary and/or human pharmaceuticals in Japan (Yamasaki et al., 2018).

Sample collection

A total of 315 adult damselflies were captured from June–August 2016 and in June 2017 using an insect net. The damselflies were collected from seven locations: Higashitonden (north latitude: 43.147353; east longitude: 141.337327), Shinoroshinkawa (north latitude: 43.120752; east longitude: 141.415881), Ohno (north latitude: 43.074279; east longitude: 141.341688), Shinoro (north latitude: 43.159162; east longitude: 141.364069), Tonneusu (north latitude: 43.169276; east longitude: 141.408628), Gotenzan (north latitude: 43.048401; east longitude: 141.258540), and Yasuharu (north latitude: 43.144301; east longitude: 141.315791). Each sampling site was located near a local river or pond with distinct natural environments in Sapporo City, Japan (Fig. 1). In specific, the most urban location was “Pond: Ohno”, located within the grounds of Hokkaido University, near Sapporo Station, in the center of the city (Fig. 1. See “Green density”). In contrast, the “River: Gotenzan” sampling site was located in a forest and was therefore the most “natural” of the sampling locations (Fig. 1. See “Green density”). Sixty-eight larvae were also collected from three of the locations (Shinoroshinkawa, Tonneusu, and Yasuharu) in June 2017 (Fig. 1). Environmental temperatures at each of the sampling locations were obtained from the Japan Weather Association (http://www.tenki.jp/).

Species determination for the captured adult damselflies

The species of each of the captured adult damselflies was determined morphologically based on distinguishing colors and patterns (lines and spots) on the body and wing (See Fig. 2A). Because of the difficulties associated with morphological species determination of larvae, the species of damselfly larvae samples could not be identified.

Collection of fecal materials

Fecal samples were collected as per the protocol outlined in Fig. 3. Briefly, adult damselflies wrapped in clean paper, along with larvae that had been immersed in sterile phosphate-buffered saline (sPBS) after washing with sPBS to prevent contamination from the exoskeleton, were stored at room temperature for at least a week. Dead individuals were omitted from the subsequent experiments. Fecal materials discharged onto the paper or into the sPBS were collected into fresh sPBS solution and homogenized. Aliquots of the homogenized solutions were inoculated onto soybean casein digest (SCD) agar plates (BD Biosciences) minus antibiotics (SCD-w/o) as an indicator of total damselfly gut bacteria, SCD agar plates supplemented with CTX (50 mg/liter) or TET (50 mg/liter) (SCD-A) as an indicator of environmental bacteria that frequently resist to these antibiotics (Huang et al., 2015), and MacConkey agar plates supplemented with AMP (10 mg/liter), TET (10 mg/liter), or KAN (10 mg/liter) (MacConkey-A) as an indicator of Gram-negative bacteria, including human-derived pathogenic Enterobacterales species. All plates were incubated at 25 °C before colony enumeration to prevent over growth on agar plates covered by environmental bacteria characteristic of rapid growth above 30 °C temperature. Results were then expressed as colony-forming units (CFU) per individual damselfly or larva. Any SCD-A or MacConkey-A plate with more than one colony was considered positive for antibiotic-resistant bacteria, followed by comparing these prevalence among sampling sites; the prevalence is defined to be the number of sample showing agar plates (SCD-A or MacConkey-A plate) with more than one colony. Morphologically representative colonies on MacConkey-A plates were also isolated and used for antimicrobial susceptibility testing according to the method described below.

Bacterial species determination

Bacterial species were identified using an API 20E kit according to the manufacturer’s protocol (bioMérieux, Marcy-l’Étoile, France). The MALDI Biotyper system (Bruker Daltonics, Billerica, MA, USA) was used as per the manufacturer’s instructions to identify those species that could not be determined using the API kit. Meanwhile, if the API value was low (below 80%), bacterial 16S rDNA typing was performed to define bacterial species according to the protocol described previously (Horn et al., 1999).

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was performed for each of the identified bacterial isolates using the agar-dilution method to determine the minimum inhibitory concentration (MIC) values for six antimicrobial agents (AMP, KAN, TET, CHL, CIP, and SPY) on Mueller-Hinton II agar (BD Biosciences, Franklin Lakes, NJ, USA). Assays were carried out according to the criteria determined by the CLSI (Wiegand, Hilpert & Hancock, 2008). Staphylococcus aureus ATCC29213 and Escherichia coli ATCC25922 were used as quality control strains.

Clustering and phylogenic analysis

Cluster analysis was performed using Cluster 3.0 for Mac OS X (Clustering Library 1.52) as described previously (Okubo et al., 2017). Phylogenetic trees were generated from the aligned population structures by Cluster 3.0 (see below) and visualized in Java TreeViewX (version 0.5.0). Specifically, the obtained data were transformed to a log scale, and then processed with the setting (>80% filtering and Pearson correlation (centered)). While the data associated with specific variables (temperature), total bacterial numbers on SCD-w/o, the prevalence of bacteria on SCD-A, the prevalence of bacteria on MacConkey-A were converted to an equivalent range (0–1), the MIC data were used without any alterations. The resulting data were then processed using the setting (>80% filtering and Spearman rank correlation) and converted to “cdt” files using Cluster 3.0 then visualized using TreeViewX.

Ethics

The study reported in this manuscript did not involve any human participants, human data, human tissue, data pertaining to specific individuals, or animal experiments.

Statistical analysis

Comparison of total CFU numbers between sampling locations was conducted using a Bonferroni/Dunn test. A p-value of < 0.01 was considered significant. Correlations among factors (temperature, total bacterial numbers on SCD-w/o, the prevalence of bacteria on SCD-A, the prevalence of bacteria on MacConkey-A) were identified by Pearson’s correlation coefficient test. A correlation coefficient value of >0.5 or <−0.05, with a P-value of < 0.05, was considered significant. Comparisons between the prevalence of bacteria on SCD-A and MacConkey-A agar plates were conducted using a Chi-square for independence test. Comparisons between the prevalence of antibiotic-resistant isolates obtained from MacConkey-A plates at each of the sampling locations were also conducted using a Chi-square for independence test. A combination with a P-value of < 0.05 was considered statistically significant. All calculations were conducted using Excel for Mac (2001) with Statcel3C.

Damselflies are ubiquitously but focally present in local environments

Because our study relied on the domestic behaviors of damselflies at each location, we first assessed whether focal inhabitation by adult damselflies could be observed at each of the sampling sites. A total of 315 adult damselflies were captured between June and August 2016, and in June 2017, at seven locations (Fig. 1). Morphological species identification results for the captured adult damselflies are summarized in Fig. 2A. Although all sampling sites were located within a 10-km² area, the prevalence of each of the 11 identified damselfly species varied between the sites (Fig. 2B). Coenagrion ecornutum was the most dominant species at the “River: Higashitonden” location, while Enallagma circulatum was most commonly identified at the “River: Shinoroshinkawa”, “River: Shinoro”, and “Marsh: Tonneusu” sites. The dominancy of Paracercion hieroglyphicum was higher at the “River: Yasuharu” and “Marsh: Tonneusu” sites compared with the other locations, while Lestes sponsa was more frequently identified at the “River: Gotenzan” site compared with all other sampling sites. In addition, 68 damselfly larvae were collected from three of the sampling sites (Shinoroshinkawa, Tonneusu, and Yasuharu) in June 2017. However, morphological species identification was not possible for the larvae. Thus, as expected, the population density of adult damselflies changed depending on the sampling location, indicating that damselflies ubiquitously but focally inhabit each of the local environments around Sapporo City.

Relationship between temperature and total number of bacteria in the damselfly gut and the prevalence of antibiotic-resistant bacteria

Fecal materials from both adult and larval damselflies were collected, and the bacterial numbers were estimated according to the protocol outlined in Fig. 3. When estimated using SCD-w/o medium, the total number of bacteria per adult/larval damselfly was fairly consistent across all sampling locations (∼10⁸ CFU per adult/larva), except for the “Pond: Ohno” site (Fig. 4). We also assessed the relationship between the total bacterial load (SCD-w/o) or the prevalence of antibiotic-resistant bacteria (SCD-A and MacConkey-A plates) and environmental temperature using a Pearson’s correlation coefficient test and cluster analysis with a Spearman’s correlation coefficient by rank test. The results showed that temperature changes were significantly but negatively correlated with the total bacterial load on SCD-w/o plates (r coefficient −0.677 vs. “Lowest temperature”, P < 0.0001; r coefficient −0.590 vs. “Highest temperature”, P < 0.0001) (Fig. 5A). Meanwhile, no significant correlation between temperature changes and the prevalence of colonies on SCD-A and MacConkey-A agar plates was observed. Furthermore, cluster analysis based on three different indicators (environmental temperature, total bacterial numbers in fecal samples, the prevalence of antibiotic resistant bacteria) revealed several clades grouped by sampling location (Fig. 5B, blue bars). Interestingly, samples obtained from the “River: Shinoroshinkawa” site were assigned to two different groups, which, together, had the highest prevalence of antibiotic resistant bacteria (Fig. 5B, red bars with asterisks). Thus, while the fecal bacterial load on SCD-w/o medium was fairly constant and was negatively correlated with environmental temperature, the prevalence of antibiotic-resistant bacteria changed depending on the sampling location and was not related to environmental temperature.

Prevalence of antibiotic-resistant bacteria and identification of representative antibiotic-resistant Enterobacterales

To determine which antibiotic-resistant bacterial strains were most prevalent in the sampled environments, we estimated the prevalence of antibiotic-resistant bacteria in the damselfly fecal samples. SCD-A medium was used as an indicator of environmental bacteria, while MacConkey-A medium was used to select for pathogenic bacteria, including Enterobacterales, derived from humans. Results showed that the prevalence of environmental and human-derived pathogenic bacteria changed depending on the sampling site, irrespective of whether the samples came from adult or larval damselflies (Table S1). However, at almost all sampling locations, environmental antibiotic-resistant bacteria were significantly more prevalent than human-derived pathogenic bacteria (adult damselfly: “River: Shinoro”, P = 0.0487; “Marsh: Tonneusu”, P < 0.0001; larvae: “River: Yasuharu”, P = 0.0329; “Marsh: Tonneusu”, P < 0.0001). Interestingly, the samples obtained from the “River: Shinoroshinkawa” site uniquely showed a bias towards antibiotic-resistant human-derived pathogenic bacteria (P = 0.0149). Thus, the results revealed that the prevalence of antibiotic-resistant bacteria in the fecal materials of both adult and larval damselflies significantly differed between sampling sites, reflecting differences in bacterial contamination of each of the domestic environments.

Amongst the sampling locations, the “River: Shinoroshinkawa” site appeared to be statistically unique, with an increased prevalence of antibiotic-resistant Enterobacterales on MacConkey-A plates. To confirm this, 59 representative antibiotic-resistant Enterobacterales isolates from adult damselfly fecal material were selected from MacConkey-A plates and identified using an API kit and the MALDI Biotyper system with MIC assessments (Table S2). The predominant bacterial species identified were Serratia fonticola, Serratia liquefaciens, Enterobacter aerogenes, Enterobacter cloacae, Klebsiella oxytoca, and Stenotrophomonas maltophilia. As expected, the site with the highest prevalence of antibiotic-resistant bacteria was “River: Shinoroshinkawa” (χ² test: p < 0.001) (Fig. 6). The MICs (MIC₅₀ and MIC₉₀) of AMP, KAN, TET, CHL, CIP, and SPY for each of the identified bacteria were also determined and used as a basis for clustering analysis (Table S1). Although there were no obvious differences in the MIC ranges between the sampling sites except for Shinoroshinkawa with high values of MIC90 for KAN and CIP (Fig. 7A), cluster analysis revealed that the identified bacteria derived from the “River: Shinoroshinkawa” site formed a cluster and showed higher MIC values and a greater prevalence of multidrug resistance compared with other sampling locations (Fig. 7B). Thus, the “River: Shinoroshinkawa” site was unique amongst the sampling locations, with the highest frequency of antibiotic-resistant Enterobacterales.