Appl Microbiol Biotechnol DOI 10.1007/s00253-013-5498-1
ENVIRONMENTAL BIOTECHNOLOGY
Seasonal variations in bacterial communities and antibiotic-resistant strains associated with green bottle flies (Diptera: Calliphoridae) Ting Wei & Ryuichi Ishida & Kazuhiko Miyanaga & Yasunori Tanji
Received: 13 November 2013 / Revised: 14 December 2013 / Accepted: 23 December 2013 # Springer-Verlag Berlin Heidelberg 2014
Abstract Green bottle flies occur frequently around human environments in Japan. Many species of green bottle flies have been studied with regard to their importance in forensic examinations or clinical therapies, but the bacterial communities associated with this group of flies have not been comprehensively investigated. In this research, 454 pyrosequencing was used to reveal the bacterial communities in green bottle flies collected in different seasons. Meanwhile, the bacteria were screened with selective media and tested for antibiotic susceptibility. Samples collected in three different seasons harbored distinctive bacterial communities. The predominant genera associated with green bottles flies were Staphylococcus in spring, Ignatzschineria in summer, and Vagococcus, Dysgonomonas, and an unclassified Acetobacteraceae in autumn. An upward trend in bacterial community diversity was observed from spring to autumn. Changes in climatic conditions could be the cause of these seasonal variations in fly-associated bacterial communities. The species of isolated antibiotic-resistant bacteria also differed across seasons, but it was difficult to correlate seasonal changes in antibiotic-resistant bacteria with changes in whole communities. A number of multiple-antibiotic-resistant bacteria were isolated, and some of these strains were closely affiliated with pathogens such as Enterococcus faecalis and Enterococcus faecium, which could cause serious threats to public health. Overall, this research provided us with information about the composition and seasonality of bacterial communities in green bottle flies, and highlighted the risks of fly-mediated dissemination of antibiotic-resistant pathogens. T. Wei : R. Ishida : K. Miyanaga : Y. Tanji (*) Department of Bioengineering, Tokyo Institute of Technology, 4259 J2-15 Nagatsuta-cho, Midori-ku, Yokohama 226-8501, Japan e-mail: [email protected]
Keywords Green bottle fly . Bacterial community . 454 pyrosequencing . Multidrug resistance
Introduction Studies of bacterial communities associated with insects have long been of great interest to many researchers. Insects provide an important source of bacterial diversity and a potentially ideal model for studying bacteria–host interactions. Bacteria associated with insects play important roles in the hosts’ physiology, such as nutrition and digestion (Dillon and Dillon 2004; Douglas 2009), as well as in the functions of the innate immune system (Cirimotich et al. 2011; Kaltenpoth 2009). Among animal symbionts, pathogens are considered to be a minority in most cases; however, the potential threat to human health posed by insect-mediated transmission of pathogens should never be overlooked, especially regarding insects that are found in close proximity to human environments (Graczyk et al. 2001). In order to more clearly understand the impact that diverse bacteria have on their insect hosts, and on other animals (including humans) when those hosts serve as vectors, it is necessary to obtain better knowledge of these bacterial communities. Earlier studies mostly resorted to the conventional culture method to recover bacterial strains, and usually focused solely on pathogenic strains (Förster et al. 2007; Graczyk et al. 2001). However, the data acquired through these methods were limited, because ∼99 % of all bacteria in the environment cannot be cultured (Rappé and Giovannoni 2003). In recent years, extensive in-depth surveys using a culture-independent method, pyrosequencing, have been carried out to reveal bacterial communities in fruit flies (Wong et al. 2011), black soldier flies (Zheng et al. 2013), and other insects (Kautz et al. 2013; Osei-Poku et al. 2012).
Appl Microbiol Biotechnol
Green bottle flies include many species of blowflies of the genera Lucilia and Phaenicia, and this group of flies can be found worldwide. Many species are important in forensic examinations because they are the first colonizer of cadavers, and therefore serve as biological clock in measurements of the time of death (Amendt et al. 2004). In addition, the common green bottle fly (Lucilia sericata) is well-known because its larvae were widely used for maggot therapy before the use of antibiotics and other medicines. Green bottle flies are synanthropic flies that occur as frequently as houseflies in human environments in Japan. Given that green bottle flies commonly breed in carcasses and other decaying organic matter, which are easily colonized by various microorganisms, we can safely assume that these flies harbor diverse bacterial communities including pathogens. To date, however, minimal information has been obtained regarding the bacterial communities associated with green bottle flies. In addition to revealing whole bacterial communities, it is also important to investigate the antibiotic-resistant bacteria associated with synanthropic flies. Due to overuse of antibiotics, antibiotic resistance has become a serious problem that undermines clinical treatment efficacy and poses significant risks to public health. Flies with large populations in close contact with humans could serve as ideal vectors for the transmission of antibiotic resistance. Antibiotic-resistant bacteria, especially multipleantibiotic-resistant bacteria have been identified in flies in hospital environments, near animal production sites, and other community environments (Ahmad et al. 2011; Graham et al. 2009; Hayes and English 2004; Rahuma et al. 2005; Unno et al. 2010). The potential risk of fly-mediated dissemination of antibioticresistant bacteria should be taken seriously in the context of efforts to improve conditions related to public health. In a previous study, we investigated bacterial communities and antibiotic-resistant strains associated with houseflies at different developmental stages (Wei et al. 2013). The flies used as experimental material were collected in residential areas of Tokyo and subsequently cultured in the lab, and may therefore have carried only subsets of the bacterial communities associated with the originally captured flies (Chandler et al. 2011). In this study, we collected samples directly from several densely populated metropolitan areas in which the flies could easily transmit to humans their microbial symbionts, including antibiotic-resistant bacteria, thereby affecting public health. We also sought to increase the sample size by collecting more individual flies and to recover the bacterial communities and antibiotic-resistant bacteria as thoroughly as possible by both pyrosequencing and culture methods. Moreover, we compared samples collected in different seasons in order to improve our understandings of the factors that shaped the bacterial communities. The results of this research fill a gap in studies of bacterial symbionts associated with green bottle flies, which are among the most common insects in human environments.
Materials and methods Fly collection Adult green bottle flies were captured in five metropolitan districts of Tokyo (Shibuya, Shinjuku, Ikebukuro, Osaki, and Ueno) and one suburban district of Yokohama (Nagatsuta) in spring, summer, and autumn of 2011. The exact dates for sample collection were as follows: 14 May (spring), 24 August (summer), and 25 October (autumn). The location of the sampling sites is shown in Fig. 1. In spring, flies were collected only from Shibuya and Nagatsuta due to the low emergence rate. In summer and autumn, flies were collected from the five districts of Tokyo. Five to six individuals were collected from each district in one season, so the sample sizes were 10 individuals in spring and 27 individuals each in summer and autumn. Collected flies were individually washed in 0.1 % sodium dodecyl sulfate solution to remove microorganisms attached to the surface of the fly body. The entire gut was dissected, weighed, and homogenized in 1 ml of sterile phosphate buffered saline (PBS) solution. Each homogenized mixture was divided into two equal samples: one was used for DNA extraction, and the other was stored as a glycerol stock at −80 °C. DNA extraction Genomic DNA was extracted using the ISOFECAL kit (Nippon Gene, Japan) according to the manufacturer’s protocol. 454 Pyrosequencing The 16S rRNA gene universal primers 27 F (5′-AGAGTTTG ATCMTGGCTCAG-3′) and 519R (5′-GWATTACCGCGG CKGCTG-3′) were used to amplify approximately 500 bp of the variable regions V1 to V3. Different tags were attached to the forward primer in order to distinguish different samples. The individuals collected from the same district in each season were combined for this PCR; in total, 12 combined samples (Nagatsuta, spring; Shibuya, spring; Osaki, summer; Ikebukuro, summer; Shibuya, summer; Shinjuku, summer; Ueno, summer; Osaki, autumn; Ikebukuro, autumn; Shibuya, autumn; Shinjuku, autumn; and Ueno, autumn) were used as templates. The conditions for PCR were as follows: an initial denaturation step at 95 °C for 5 min, followed by 30 cycles of amplification (denaturation at 95 °C for 30 s, annealing at 58 °C for 30 s, and elongation at 72 °C for 60 s), and a final extension step for 5 min at 72 °C. PCR products were purified with phenol/chloroform extraction, and all samples were multiplexed by combining 10 ng of purified DNA from each. Roche 454 GS-FLX Titanium pyrosequencing was performed by Hokkaido System Science Co., Ltd.
Appl Microbiol Biotechnol Fig. 1 Map of sampling sites. a Map of Japan. Shaded area indicates Greater Tokyo; b map of Greater Tokyo; c magnified image of the framed zone in b. Triangles indicate sampling sites in the five metropolitan districts of Tokyo and a suburban district of Yokohama
a
b Greater Tokyo Tokyo City
Yokohama
c Shinjuku Shibuya
Tokyo Bay
Ikebukuro Ueno
Osaki Tokyo Bay Nagatsuta
Analysis of pyrosequencing results All 16S rRNA gene pyrosequencing reads were analyzed using QIIME, version 1.5.0 (Caporaso et al. 2010). Quality criteria were as follows: minimum sequence length of 350 bp, maximum sequence length of 560 bp, and minimum average quality score of 50. Using stringent quality-control parameters, we allowed no ambiguous bases or mismatches in the primer sequence and no barcode errors, and a maximum homopolymer length of 6 bp. Sequences were clustered into operational taxonomic units (OTUs) at the threshold of 97 % similarity, and a representative sequence from each OTU was assigned taxonomic data using Ribosomal Database Project (RDP) classifier (Cole et al. 2005). Measures of alpha diversity including the number of observed OTUs, Chao1 nonparametric richness estimator (Chao 1984), and Shannon diversity index (Hill et al. 2003) were computed as part of QIIME’s pipeline. Beta diversity representing the explicit comparison between all 12 combined samples based on their composition was calculated using the default beta diversity metrics of weighted and unweighted UniFrac (Lozupone and Knight 2005; Lozupone et al. 2007) as implemented in QIIME. Similarity of samples for the weighted and unweighted UniFrac analysis was visualized using principal coordinate analysis (PCoA) plots. Jackknife-supported unweighted pair-group method with arithmetic mean (UPGMA) was used for hierarchical clustering of the communities, and the tree representing the relationships among the samples was displayed with FigTree, version 1.4.0. (http://tree.bio.ed.ac.uk/ software/figtree/). Isolation of antibiotic-resistant bacterial strains Homogenized gut samples were serially (×10) diluted with sterile PBS and drop-plated onto Luria–Bertani (LB) agar in Petri plates supplemented with ampicillin (Amp, 50 μg/ml),
8 km
N
cefpodoxime (Cef, 10 μg/ml), tetracycline (Tet, 50 μg/ml), kanamycin (Kan, 50 μg/ml), nalidixic acid (Nal, 30 μg/ml), or ciprofloxacin (Cip, 5 μg/ml), or without antibiotics as a control. Colonies on each plate were counted, and the resultant counts were used to calculate cell density in the original samples. Ten colonies from each antibiotic plate were picked and purified twice by streaking. The purified colonies were used as templates for 16S rRNA gene PCR with universal primers 27 F and 1492R (5′-GGCTACCTTGTTACGACTT3′). The PCR products were classified by restriction fragment length polymorphism (RFLP). The DNA fragments with representative restriction patterns were purified by phenol/ chloroform extraction and subjected to conventional Sanger sequencing using Big Dye Terminator ver. 3.1 (Applied Biosystems) on an Applied Biosystems 3730xl DNA Analyzer. The sequencing results were analyzed with the Basic Local Alignment Search Tool (BLAST) (Altschul et al. 1997). In total, 129 bacterial strains (including 27 in spring, 64 in summer, and 38 in autumn) were sequenced and compared with sequences in the GenBank database. Antibiotic susceptibility test The isolated bacterial strains were tested for susceptibility to the five antibiotics listed in the previous section using a standard agar dilution assay as previously described (Andrews 2001). Mueller-Hinton agar plates were supplemented with twofold serially diluted antibiotic solutions to give final concentrations ranging from 1 to 512 μg/ml. This range of concentration was expected to cover the susceptibility of the isolated strains as predicted from the antibiotic concentrations used in the screening step. Inocula prepared from overnight cultures of isolated strains were delivered onto the agar surface using a multichannel pipette. The agar plates were incubated for 18–24 h at 37 °C. The minimum inhibitory
Appl Microbiol Biotechnol
concentration (MIC) was determined as the lowest concentration of antibiotics at which no visible growth of the organism was observed. Nucleotide sequence accession number The sequences of isolated bacterial species are deposited in GenBank under accession numbers KF767872 to KF767999. 454 pyrosequencing data are available in GenBank’s Sequence Read Archive under accession number SRP032324. Data analysis Kruskal–Wallis test followed by Dunn’s multiple-comparison test was used to assess differences in cell counts among flies collected in spring, summer, and autumn. Differences were considered significant when P was less than 0.05. The statistical software Prism version 5.0c (GraphPad Software, San Diego, CA, USA) was used for analysis and graphing.
Results
samples are combined and annotated as “other”. The results indicated that the bacterial communities in the two samples collected in spring were highly similar: in both, the predominant genus was Staphylococcus (approximately 86 %) and the second most abundant genus was Pseudomonas (approximately 10 %). Ignatzschineria dominated (21–96.5 %) all five samples of summer, while Dysgonomonas, Vagococcus, and Lactobacillus were the second most abundant genera in samples from Ikebukuro, Shibuya, and Shinjuku, respectively. Pseudomonas was common among the five summer samples, with varying relative abundance. The compositions of major bacterial communities in the five autumn samples were quite diverse. The predominant genera were Dysgonomonas in the sample from Shinjuku (63.3 %), Vagococcus in samples from Osaki (26.6 %), Ikebukuro (43.2 %), Ueno (19.9 %), and an unclassified Acetobacteraceae in the sample from Shibuya (26.7 %). The number of observed OTUs, Chao1 estimator, and Shannon index of all 12 samples after rarefaction to 634 reads are listed in Table 1. The highest values of these three parameters were observed in the samples from Ueno in autumn. Overall, samples collected in autumn had significantly higher alpha diversity than samples collected in spring and summer.
Diversity and composition of bacterial communities After quality filtering, a total of 13,574 reads were recovered from all 12 samples. The average read length was 498 bp. After quality control, the number of reads per sample ranged from 639 to 1,851. Sequencing reads were clustered into 658 OTUs. Of all qualified reads, 99.6 % could be assigned to a known phylum using the QIIME analysis pipeline, and 12 phyla were recovered from the bacterial communities associated with green bottle flies captured in three seasons. Two ubiquitous phyla, Proteobacteria and Firmicutes, occupied 47 and 35.8 % of all reads, respectively. Bacteroidetes was recovered with a relative abundance of 12.6 %, and was present in all samples except for the one from Osaki in summer. Aside from these three phyla, Acidobacteria, Actinobacteria, Cyanobacteria, Deferribacteres, Elusimicrobia, Fusobacteria, Planctomycetes, TM7, and Tenericutes were also detected. Each of these phyla constituted no more than 2.5 % of all reads, and was present in a few of the 12 samples. Nonetheless, the phylum Tenericutes was noted because it accounted for 24.5 % of reads in the sample from Shibuya in autumn. Classification of the results down to order level revealed that the order Pseudomonadales was ubiquitous, constituting 0.6–12.7 % of reads in each sample. Other frequently detected orders across samples included Bacteroidales, Lactobacillales, Xanthomonadales, Clostridiales, Actinomycetales, and Enterobacteriales. Further classification at the genus level is summarized in Fig. 2. Genera constituting the major bacterial communities are represented by different patterns, whereas genera that individually constituted less than 10 % of reads in all the 12
Comparison of bacterial communities across seasons In addition to revealing the seasonal variations of the flyassociated bacterial communities, we also assessed the potential influence of different sampling sites on these communities. In the jackknife-supported PCoA plots based on unweighted UniFrac distances (Fig. 3), the two spring samples clustered closely, indicating that the bacterial communities in these samples were highly similar even though the samples were collected from sites more than 25 km apart. Furthermore, this cluster was clearly distinct from the other samples in all three plots. The five summer samples could also be grouped together, although not as closely. By contrast, the autumn samples scattered in different quadrants, with the sample from Shibuya especially distant from the other four samples. The bootstrapped tree resulting from UPGMA hierarchical clustering provided a clearer view of the relationship among the bacterial communities in the 12 samples (Fig. 4). Spring and summer samples formed two separate clusters with more than 50 % jackknife support, indicating that for these two seasons, samples from different sites were closely related. Three autumn samples (from Shinjuku, Ikebukuro, and Osaki) clustered together. The sample from Ueno could be grouped with all summer samples and the three aforementioned autumn samples in a larger cluster, but with lower confidence (45 and 55 % support for internal nodes). As in the PCoA analysis, the sample from Shibuya in autumn was distant from all other samples. It was also clear that bacterial communities
Appl Microbiol Biotechnol Fig. 2 Taxonomic classification of bacterial reads retrieved from pooled samples collected at different sites in spring, summer, and autumn. Sequences were classified at genus level using RDP classifier. Genera constituting less than Z of reads in any of the 12 samples were combined and annotated as “other”
varied significantly among samples collected from the same site in different seasons. Bacterial culture To assess the prevalence of culturable bacteria, including antibiotic-resistant strains, each individual fly sample was processed separately, and then the samples were analyzed in groups based on season (Fig. 5). The abundance of total culturable bacteria was calculated from the number of colonies formed on LB agar plates without any antibiotics. This concentration varied over two orders of magnitude, ranging from 107 to 109 colonyforming units (CFU)/g fly gut. Summer samples harbored significantly higher numbers of total culturable bacteria than samples collected in the other two seasons. Across all samples, the concentration of culturable bacteria that were resistant to at least one antibiotic ranged from 103 to 108 CFU/g fly gut. Statistical analysis revealed that the concentration of ampicillin-resistant bacteria in summer samples Table 1 The number of observed operational taxonomic units (OTUs), Chao1 estimator, and Shannon index after rarefaction to 634 reads Sample
OTUs
Chao1
Shannon
Nagatsuta, spring Shibuya, spring Osaki, summer Ikebukuro, summer Shibuya, summer Shinjuku, summer Ueno, summer Osaki, autumn Ikebukuro, autumn Shibuya, autumn Shinjuku, autumn Ueno, autumn
24 25 15 24 25 73 17 79 57 135 53 139
61.4 38.0 18.6 29.5 33.1 89.7 39.6 108.6 96.9 271.3 95.0 281.6
1.494 2.094 0.857 2.644 2.675 4.771 0.351 4.619 3.487 4.942 2.695 5.626
was significantly higher than in spring or autumn, and that the concentration of cefpodoxime-resistant bacteria was significantly lower in autumn samples than in the other two seasonal groups. No significant difference among seasons was detected in the concentrations of bacteria resistant to tetracycline, kanamycin, nalidixic acid, or ciprofloxacin. Diversity and antibiotic resistance of isolated strains A total of 129 bacterial strains were isolated from antibioticsupplemented LB agar plates (27 in spring, 64 in summer, and 38 in autumn). The closest matches are listed in Table 2. By alignments with records in GenBank, the bacterial isolates were affiliated with 10 species for spring samples, 21 species for summer samples, and 14 species for autumn samples. In spring samples, the most frequently isolated strains were closely related to Pseudomonas putida (eight strains), Bacillus subtilis, and another Pseudomonas sp. (four strains each). Staphylococcus, the predominant genus revealed in the pyrosequencing analysis, was not recovered among the resistant strains. In summer samples, species closely related to Enterococcus hirae (16 strains) were most abundant, followed by Enterococcus faecalis (eight strains) and Carnobacterium maltaromaticum (seven strains). Staphylococcus sp. (six strains) and Vagococcus teuberi (five strains) were also isolated at high frequency. In autumn samples, E. faecalis was the most frequently isolated species (13 strains), followed by Lactobacillus plantarum (seven strains) and E. hirae (six strains). Isolated antibiotic-resistant strains were tested for their susceptibility to five of the antibiotics used in the screening step. For logistical reasons, cefpodoxime was not used in this step of analysis. Thirteen strains isolated from summer samples and 16 from autumn samples were multidrug-resistant strains that exhibited resistance to at least three antibiotics. The MICs of the five antibiotics tested for these strains are listed in Table 3. There was no clear pattern of correlation between antibiotic-resistance profiles and seasons. Although
Appl Microbiol Biotechnol
Fig. 3 Jackknifed replicate PCoA plots of bacterial communities from combined samples subjected to pyrosequencing. Variation among replicates is displayed as confidence ellipsoids around the sample dots
summer and autumn samples shared some common species, such as L. plantarum, E. faecalis, and E. hirae, very different antibiotic-resistance profiles were observed among strains classified as the same species.
Discussion Bacterial communities associated with animals often provide essential functions to their hosts. However, due to the
complexity of the microbial communities, knowledge regarding the identities and significance of these microbial symbionts is still very limited. To date, most studies on the diversity and functions of bacterial communities have been performed using a limited range of model organisms, including some insects. In this study, we sought to reveal the diversity of bacterial communities and antibiotic-resistant strains associated with flies that are found near humans and could potentially affect our health.
Appl Microbiol Biotechnol
Fig. 4 Bootstrapped tree of UPGMA hierarchical cluster analysis, aimed at interpreting phylogenetic relationships among bacterial communities associated with green bottle flies collected at different sites in different seasons. The numbers at the nodes indicate jackknife support values
Pyrosequencing results showed that the two phyla, Proteobacteria and Firmicutes, were present in all samples and predominant in most of them. Bacteroidetes was detected in most samples, less frequently than the two predominant phyla, and was missing from only one sample. Based on the results of several previous studies (Chandler et al. 2011; Corby-Harris et al. 2007; Gupta et al. 2012; Wong et al. 2011), these three phyla could be considered as typical members of the major bacterial communities of fruit flies and houseflies. We found one order common to all samples, Pseudomonadales, but its prevalence was relatively low, and it was predominant in none of the samples. When the sequences were classified down to lower phylogenetic levels, no strict core bacterial microbiome was detected across all seasons. However, samples collected from different sites in
Spring
Summer
Autumn
10 10 10 9
CFU/g fly gut
10 8 10 7 10 6 10 5 10 4 10 3 10 2 10 1 LB
Amp
Cef
Tet
Kan
Nal
Cip
Fig. 5 Cell counts of bacteria isolated from green bottle flies in spring, summer, and autumn on LB agar and LB supplemented with antibiotics: Amp ampicillin, Cef cefpodoxime, Tet tetracycline, Kan kanamycin, Nal nalidixic acid, Cip ciprofloxacin. Significant differences are represented by asterisks (*P
ENVIRONMENTAL BIOTECHNOLOGY
Seasonal variations in bacterial communities and antibiotic-resistant strains associated with green bottle flies (Diptera: Calliphoridae) Ting Wei & Ryuichi Ishida & Kazuhiko Miyanaga & Yasunori Tanji
Received: 13 November 2013 / Revised: 14 December 2013 / Accepted: 23 December 2013 # Springer-Verlag Berlin Heidelberg 2014
Abstract Green bottle flies occur frequently around human environments in Japan. Many species of green bottle flies have been studied with regard to their importance in forensic examinations or clinical therapies, but the bacterial communities associated with this group of flies have not been comprehensively investigated. In this research, 454 pyrosequencing was used to reveal the bacterial communities in green bottle flies collected in different seasons. Meanwhile, the bacteria were screened with selective media and tested for antibiotic susceptibility. Samples collected in three different seasons harbored distinctive bacterial communities. The predominant genera associated with green bottles flies were Staphylococcus in spring, Ignatzschineria in summer, and Vagococcus, Dysgonomonas, and an unclassified Acetobacteraceae in autumn. An upward trend in bacterial community diversity was observed from spring to autumn. Changes in climatic conditions could be the cause of these seasonal variations in fly-associated bacterial communities. The species of isolated antibiotic-resistant bacteria also differed across seasons, but it was difficult to correlate seasonal changes in antibiotic-resistant bacteria with changes in whole communities. A number of multiple-antibiotic-resistant bacteria were isolated, and some of these strains were closely affiliated with pathogens such as Enterococcus faecalis and Enterococcus faecium, which could cause serious threats to public health. Overall, this research provided us with information about the composition and seasonality of bacterial communities in green bottle flies, and highlighted the risks of fly-mediated dissemination of antibiotic-resistant pathogens. T. Wei : R. Ishida : K. Miyanaga : Y. Tanji (*) Department of Bioengineering, Tokyo Institute of Technology, 4259 J2-15 Nagatsuta-cho, Midori-ku, Yokohama 226-8501, Japan e-mail: [email protected]
Keywords Green bottle fly . Bacterial community . 454 pyrosequencing . Multidrug resistance
Introduction Studies of bacterial communities associated with insects have long been of great interest to many researchers. Insects provide an important source of bacterial diversity and a potentially ideal model for studying bacteria–host interactions. Bacteria associated with insects play important roles in the hosts’ physiology, such as nutrition and digestion (Dillon and Dillon 2004; Douglas 2009), as well as in the functions of the innate immune system (Cirimotich et al. 2011; Kaltenpoth 2009). Among animal symbionts, pathogens are considered to be a minority in most cases; however, the potential threat to human health posed by insect-mediated transmission of pathogens should never be overlooked, especially regarding insects that are found in close proximity to human environments (Graczyk et al. 2001). In order to more clearly understand the impact that diverse bacteria have on their insect hosts, and on other animals (including humans) when those hosts serve as vectors, it is necessary to obtain better knowledge of these bacterial communities. Earlier studies mostly resorted to the conventional culture method to recover bacterial strains, and usually focused solely on pathogenic strains (Förster et al. 2007; Graczyk et al. 2001). However, the data acquired through these methods were limited, because ∼99 % of all bacteria in the environment cannot be cultured (Rappé and Giovannoni 2003). In recent years, extensive in-depth surveys using a culture-independent method, pyrosequencing, have been carried out to reveal bacterial communities in fruit flies (Wong et al. 2011), black soldier flies (Zheng et al. 2013), and other insects (Kautz et al. 2013; Osei-Poku et al. 2012).
Appl Microbiol Biotechnol
Green bottle flies include many species of blowflies of the genera Lucilia and Phaenicia, and this group of flies can be found worldwide. Many species are important in forensic examinations because they are the first colonizer of cadavers, and therefore serve as biological clock in measurements of the time of death (Amendt et al. 2004). In addition, the common green bottle fly (Lucilia sericata) is well-known because its larvae were widely used for maggot therapy before the use of antibiotics and other medicines. Green bottle flies are synanthropic flies that occur as frequently as houseflies in human environments in Japan. Given that green bottle flies commonly breed in carcasses and other decaying organic matter, which are easily colonized by various microorganisms, we can safely assume that these flies harbor diverse bacterial communities including pathogens. To date, however, minimal information has been obtained regarding the bacterial communities associated with green bottle flies. In addition to revealing whole bacterial communities, it is also important to investigate the antibiotic-resistant bacteria associated with synanthropic flies. Due to overuse of antibiotics, antibiotic resistance has become a serious problem that undermines clinical treatment efficacy and poses significant risks to public health. Flies with large populations in close contact with humans could serve as ideal vectors for the transmission of antibiotic resistance. Antibiotic-resistant bacteria, especially multipleantibiotic-resistant bacteria have been identified in flies in hospital environments, near animal production sites, and other community environments (Ahmad et al. 2011; Graham et al. 2009; Hayes and English 2004; Rahuma et al. 2005; Unno et al. 2010). The potential risk of fly-mediated dissemination of antibioticresistant bacteria should be taken seriously in the context of efforts to improve conditions related to public health. In a previous study, we investigated bacterial communities and antibiotic-resistant strains associated with houseflies at different developmental stages (Wei et al. 2013). The flies used as experimental material were collected in residential areas of Tokyo and subsequently cultured in the lab, and may therefore have carried only subsets of the bacterial communities associated with the originally captured flies (Chandler et al. 2011). In this study, we collected samples directly from several densely populated metropolitan areas in which the flies could easily transmit to humans their microbial symbionts, including antibiotic-resistant bacteria, thereby affecting public health. We also sought to increase the sample size by collecting more individual flies and to recover the bacterial communities and antibiotic-resistant bacteria as thoroughly as possible by both pyrosequencing and culture methods. Moreover, we compared samples collected in different seasons in order to improve our understandings of the factors that shaped the bacterial communities. The results of this research fill a gap in studies of bacterial symbionts associated with green bottle flies, which are among the most common insects in human environments.
Materials and methods Fly collection Adult green bottle flies were captured in five metropolitan districts of Tokyo (Shibuya, Shinjuku, Ikebukuro, Osaki, and Ueno) and one suburban district of Yokohama (Nagatsuta) in spring, summer, and autumn of 2011. The exact dates for sample collection were as follows: 14 May (spring), 24 August (summer), and 25 October (autumn). The location of the sampling sites is shown in Fig. 1. In spring, flies were collected only from Shibuya and Nagatsuta due to the low emergence rate. In summer and autumn, flies were collected from the five districts of Tokyo. Five to six individuals were collected from each district in one season, so the sample sizes were 10 individuals in spring and 27 individuals each in summer and autumn. Collected flies were individually washed in 0.1 % sodium dodecyl sulfate solution to remove microorganisms attached to the surface of the fly body. The entire gut was dissected, weighed, and homogenized in 1 ml of sterile phosphate buffered saline (PBS) solution. Each homogenized mixture was divided into two equal samples: one was used for DNA extraction, and the other was stored as a glycerol stock at −80 °C. DNA extraction Genomic DNA was extracted using the ISOFECAL kit (Nippon Gene, Japan) according to the manufacturer’s protocol. 454 Pyrosequencing The 16S rRNA gene universal primers 27 F (5′-AGAGTTTG ATCMTGGCTCAG-3′) and 519R (5′-GWATTACCGCGG CKGCTG-3′) were used to amplify approximately 500 bp of the variable regions V1 to V3. Different tags were attached to the forward primer in order to distinguish different samples. The individuals collected from the same district in each season were combined for this PCR; in total, 12 combined samples (Nagatsuta, spring; Shibuya, spring; Osaki, summer; Ikebukuro, summer; Shibuya, summer; Shinjuku, summer; Ueno, summer; Osaki, autumn; Ikebukuro, autumn; Shibuya, autumn; Shinjuku, autumn; and Ueno, autumn) were used as templates. The conditions for PCR were as follows: an initial denaturation step at 95 °C for 5 min, followed by 30 cycles of amplification (denaturation at 95 °C for 30 s, annealing at 58 °C for 30 s, and elongation at 72 °C for 60 s), and a final extension step for 5 min at 72 °C. PCR products were purified with phenol/chloroform extraction, and all samples were multiplexed by combining 10 ng of purified DNA from each. Roche 454 GS-FLX Titanium pyrosequencing was performed by Hokkaido System Science Co., Ltd.
Appl Microbiol Biotechnol Fig. 1 Map of sampling sites. a Map of Japan. Shaded area indicates Greater Tokyo; b map of Greater Tokyo; c magnified image of the framed zone in b. Triangles indicate sampling sites in the five metropolitan districts of Tokyo and a suburban district of Yokohama
a
b Greater Tokyo Tokyo City
Yokohama
c Shinjuku Shibuya
Tokyo Bay
Ikebukuro Ueno
Osaki Tokyo Bay Nagatsuta
Analysis of pyrosequencing results All 16S rRNA gene pyrosequencing reads were analyzed using QIIME, version 1.5.0 (Caporaso et al. 2010). Quality criteria were as follows: minimum sequence length of 350 bp, maximum sequence length of 560 bp, and minimum average quality score of 50. Using stringent quality-control parameters, we allowed no ambiguous bases or mismatches in the primer sequence and no barcode errors, and a maximum homopolymer length of 6 bp. Sequences were clustered into operational taxonomic units (OTUs) at the threshold of 97 % similarity, and a representative sequence from each OTU was assigned taxonomic data using Ribosomal Database Project (RDP) classifier (Cole et al. 2005). Measures of alpha diversity including the number of observed OTUs, Chao1 nonparametric richness estimator (Chao 1984), and Shannon diversity index (Hill et al. 2003) were computed as part of QIIME’s pipeline. Beta diversity representing the explicit comparison between all 12 combined samples based on their composition was calculated using the default beta diversity metrics of weighted and unweighted UniFrac (Lozupone and Knight 2005; Lozupone et al. 2007) as implemented in QIIME. Similarity of samples for the weighted and unweighted UniFrac analysis was visualized using principal coordinate analysis (PCoA) plots. Jackknife-supported unweighted pair-group method with arithmetic mean (UPGMA) was used for hierarchical clustering of the communities, and the tree representing the relationships among the samples was displayed with FigTree, version 1.4.0. (http://tree.bio.ed.ac.uk/ software/figtree/). Isolation of antibiotic-resistant bacterial strains Homogenized gut samples were serially (×10) diluted with sterile PBS and drop-plated onto Luria–Bertani (LB) agar in Petri plates supplemented with ampicillin (Amp, 50 μg/ml),
8 km
N
cefpodoxime (Cef, 10 μg/ml), tetracycline (Tet, 50 μg/ml), kanamycin (Kan, 50 μg/ml), nalidixic acid (Nal, 30 μg/ml), or ciprofloxacin (Cip, 5 μg/ml), or without antibiotics as a control. Colonies on each plate were counted, and the resultant counts were used to calculate cell density in the original samples. Ten colonies from each antibiotic plate were picked and purified twice by streaking. The purified colonies were used as templates for 16S rRNA gene PCR with universal primers 27 F and 1492R (5′-GGCTACCTTGTTACGACTT3′). The PCR products were classified by restriction fragment length polymorphism (RFLP). The DNA fragments with representative restriction patterns were purified by phenol/ chloroform extraction and subjected to conventional Sanger sequencing using Big Dye Terminator ver. 3.1 (Applied Biosystems) on an Applied Biosystems 3730xl DNA Analyzer. The sequencing results were analyzed with the Basic Local Alignment Search Tool (BLAST) (Altschul et al. 1997). In total, 129 bacterial strains (including 27 in spring, 64 in summer, and 38 in autumn) were sequenced and compared with sequences in the GenBank database. Antibiotic susceptibility test The isolated bacterial strains were tested for susceptibility to the five antibiotics listed in the previous section using a standard agar dilution assay as previously described (Andrews 2001). Mueller-Hinton agar plates were supplemented with twofold serially diluted antibiotic solutions to give final concentrations ranging from 1 to 512 μg/ml. This range of concentration was expected to cover the susceptibility of the isolated strains as predicted from the antibiotic concentrations used in the screening step. Inocula prepared from overnight cultures of isolated strains were delivered onto the agar surface using a multichannel pipette. The agar plates were incubated for 18–24 h at 37 °C. The minimum inhibitory
Appl Microbiol Biotechnol
concentration (MIC) was determined as the lowest concentration of antibiotics at which no visible growth of the organism was observed. Nucleotide sequence accession number The sequences of isolated bacterial species are deposited in GenBank under accession numbers KF767872 to KF767999. 454 pyrosequencing data are available in GenBank’s Sequence Read Archive under accession number SRP032324. Data analysis Kruskal–Wallis test followed by Dunn’s multiple-comparison test was used to assess differences in cell counts among flies collected in spring, summer, and autumn. Differences were considered significant when P was less than 0.05. The statistical software Prism version 5.0c (GraphPad Software, San Diego, CA, USA) was used for analysis and graphing.
Results
samples are combined and annotated as “other”. The results indicated that the bacterial communities in the two samples collected in spring were highly similar: in both, the predominant genus was Staphylococcus (approximately 86 %) and the second most abundant genus was Pseudomonas (approximately 10 %). Ignatzschineria dominated (21–96.5 %) all five samples of summer, while Dysgonomonas, Vagococcus, and Lactobacillus were the second most abundant genera in samples from Ikebukuro, Shibuya, and Shinjuku, respectively. Pseudomonas was common among the five summer samples, with varying relative abundance. The compositions of major bacterial communities in the five autumn samples were quite diverse. The predominant genera were Dysgonomonas in the sample from Shinjuku (63.3 %), Vagococcus in samples from Osaki (26.6 %), Ikebukuro (43.2 %), Ueno (19.9 %), and an unclassified Acetobacteraceae in the sample from Shibuya (26.7 %). The number of observed OTUs, Chao1 estimator, and Shannon index of all 12 samples after rarefaction to 634 reads are listed in Table 1. The highest values of these three parameters were observed in the samples from Ueno in autumn. Overall, samples collected in autumn had significantly higher alpha diversity than samples collected in spring and summer.
Diversity and composition of bacterial communities After quality filtering, a total of 13,574 reads were recovered from all 12 samples. The average read length was 498 bp. After quality control, the number of reads per sample ranged from 639 to 1,851. Sequencing reads were clustered into 658 OTUs. Of all qualified reads, 99.6 % could be assigned to a known phylum using the QIIME analysis pipeline, and 12 phyla were recovered from the bacterial communities associated with green bottle flies captured in three seasons. Two ubiquitous phyla, Proteobacteria and Firmicutes, occupied 47 and 35.8 % of all reads, respectively. Bacteroidetes was recovered with a relative abundance of 12.6 %, and was present in all samples except for the one from Osaki in summer. Aside from these three phyla, Acidobacteria, Actinobacteria, Cyanobacteria, Deferribacteres, Elusimicrobia, Fusobacteria, Planctomycetes, TM7, and Tenericutes were also detected. Each of these phyla constituted no more than 2.5 % of all reads, and was present in a few of the 12 samples. Nonetheless, the phylum Tenericutes was noted because it accounted for 24.5 % of reads in the sample from Shibuya in autumn. Classification of the results down to order level revealed that the order Pseudomonadales was ubiquitous, constituting 0.6–12.7 % of reads in each sample. Other frequently detected orders across samples included Bacteroidales, Lactobacillales, Xanthomonadales, Clostridiales, Actinomycetales, and Enterobacteriales. Further classification at the genus level is summarized in Fig. 2. Genera constituting the major bacterial communities are represented by different patterns, whereas genera that individually constituted less than 10 % of reads in all the 12
Comparison of bacterial communities across seasons In addition to revealing the seasonal variations of the flyassociated bacterial communities, we also assessed the potential influence of different sampling sites on these communities. In the jackknife-supported PCoA plots based on unweighted UniFrac distances (Fig. 3), the two spring samples clustered closely, indicating that the bacterial communities in these samples were highly similar even though the samples were collected from sites more than 25 km apart. Furthermore, this cluster was clearly distinct from the other samples in all three plots. The five summer samples could also be grouped together, although not as closely. By contrast, the autumn samples scattered in different quadrants, with the sample from Shibuya especially distant from the other four samples. The bootstrapped tree resulting from UPGMA hierarchical clustering provided a clearer view of the relationship among the bacterial communities in the 12 samples (Fig. 4). Spring and summer samples formed two separate clusters with more than 50 % jackknife support, indicating that for these two seasons, samples from different sites were closely related. Three autumn samples (from Shinjuku, Ikebukuro, and Osaki) clustered together. The sample from Ueno could be grouped with all summer samples and the three aforementioned autumn samples in a larger cluster, but with lower confidence (45 and 55 % support for internal nodes). As in the PCoA analysis, the sample from Shibuya in autumn was distant from all other samples. It was also clear that bacterial communities
Appl Microbiol Biotechnol Fig. 2 Taxonomic classification of bacterial reads retrieved from pooled samples collected at different sites in spring, summer, and autumn. Sequences were classified at genus level using RDP classifier. Genera constituting less than Z of reads in any of the 12 samples were combined and annotated as “other”
varied significantly among samples collected from the same site in different seasons. Bacterial culture To assess the prevalence of culturable bacteria, including antibiotic-resistant strains, each individual fly sample was processed separately, and then the samples were analyzed in groups based on season (Fig. 5). The abundance of total culturable bacteria was calculated from the number of colonies formed on LB agar plates without any antibiotics. This concentration varied over two orders of magnitude, ranging from 107 to 109 colonyforming units (CFU)/g fly gut. Summer samples harbored significantly higher numbers of total culturable bacteria than samples collected in the other two seasons. Across all samples, the concentration of culturable bacteria that were resistant to at least one antibiotic ranged from 103 to 108 CFU/g fly gut. Statistical analysis revealed that the concentration of ampicillin-resistant bacteria in summer samples Table 1 The number of observed operational taxonomic units (OTUs), Chao1 estimator, and Shannon index after rarefaction to 634 reads Sample
OTUs
Chao1
Shannon
Nagatsuta, spring Shibuya, spring Osaki, summer Ikebukuro, summer Shibuya, summer Shinjuku, summer Ueno, summer Osaki, autumn Ikebukuro, autumn Shibuya, autumn Shinjuku, autumn Ueno, autumn
24 25 15 24 25 73 17 79 57 135 53 139
61.4 38.0 18.6 29.5 33.1 89.7 39.6 108.6 96.9 271.3 95.0 281.6
1.494 2.094 0.857 2.644 2.675 4.771 0.351 4.619 3.487 4.942 2.695 5.626
was significantly higher than in spring or autumn, and that the concentration of cefpodoxime-resistant bacteria was significantly lower in autumn samples than in the other two seasonal groups. No significant difference among seasons was detected in the concentrations of bacteria resistant to tetracycline, kanamycin, nalidixic acid, or ciprofloxacin. Diversity and antibiotic resistance of isolated strains A total of 129 bacterial strains were isolated from antibioticsupplemented LB agar plates (27 in spring, 64 in summer, and 38 in autumn). The closest matches are listed in Table 2. By alignments with records in GenBank, the bacterial isolates were affiliated with 10 species for spring samples, 21 species for summer samples, and 14 species for autumn samples. In spring samples, the most frequently isolated strains were closely related to Pseudomonas putida (eight strains), Bacillus subtilis, and another Pseudomonas sp. (four strains each). Staphylococcus, the predominant genus revealed in the pyrosequencing analysis, was not recovered among the resistant strains. In summer samples, species closely related to Enterococcus hirae (16 strains) were most abundant, followed by Enterococcus faecalis (eight strains) and Carnobacterium maltaromaticum (seven strains). Staphylococcus sp. (six strains) and Vagococcus teuberi (five strains) were also isolated at high frequency. In autumn samples, E. faecalis was the most frequently isolated species (13 strains), followed by Lactobacillus plantarum (seven strains) and E. hirae (six strains). Isolated antibiotic-resistant strains were tested for their susceptibility to five of the antibiotics used in the screening step. For logistical reasons, cefpodoxime was not used in this step of analysis. Thirteen strains isolated from summer samples and 16 from autumn samples were multidrug-resistant strains that exhibited resistance to at least three antibiotics. The MICs of the five antibiotics tested for these strains are listed in Table 3. There was no clear pattern of correlation between antibiotic-resistance profiles and seasons. Although
Appl Microbiol Biotechnol
Fig. 3 Jackknifed replicate PCoA plots of bacterial communities from combined samples subjected to pyrosequencing. Variation among replicates is displayed as confidence ellipsoids around the sample dots
summer and autumn samples shared some common species, such as L. plantarum, E. faecalis, and E. hirae, very different antibiotic-resistance profiles were observed among strains classified as the same species.
Discussion Bacterial communities associated with animals often provide essential functions to their hosts. However, due to the
complexity of the microbial communities, knowledge regarding the identities and significance of these microbial symbionts is still very limited. To date, most studies on the diversity and functions of bacterial communities have been performed using a limited range of model organisms, including some insects. In this study, we sought to reveal the diversity of bacterial communities and antibiotic-resistant strains associated with flies that are found near humans and could potentially affect our health.
Appl Microbiol Biotechnol
Fig. 4 Bootstrapped tree of UPGMA hierarchical cluster analysis, aimed at interpreting phylogenetic relationships among bacterial communities associated with green bottle flies collected at different sites in different seasons. The numbers at the nodes indicate jackknife support values
Pyrosequencing results showed that the two phyla, Proteobacteria and Firmicutes, were present in all samples and predominant in most of them. Bacteroidetes was detected in most samples, less frequently than the two predominant phyla, and was missing from only one sample. Based on the results of several previous studies (Chandler et al. 2011; Corby-Harris et al. 2007; Gupta et al. 2012; Wong et al. 2011), these three phyla could be considered as typical members of the major bacterial communities of fruit flies and houseflies. We found one order common to all samples, Pseudomonadales, but its prevalence was relatively low, and it was predominant in none of the samples. When the sequences were classified down to lower phylogenetic levels, no strict core bacterial microbiome was detected across all seasons. However, samples collected from different sites in
Spring
Summer
Autumn
10 10 10 9
CFU/g fly gut
10 8 10 7 10 6 10 5 10 4 10 3 10 2 10 1 LB
Amp
Cef
Tet
Kan
Nal
Cip
Fig. 5 Cell counts of bacteria isolated from green bottle flies in spring, summer, and autumn on LB agar and LB supplemented with antibiotics: Amp ampicillin, Cef cefpodoxime, Tet tetracycline, Kan kanamycin, Nal nalidixic acid, Cip ciprofloxacin. Significant differences are represented by asterisks (*P
