Introduction
Insect–microbe associations are very common across many orders of insects. As a group of phytophagous insects, which mainly feed on nitrogen-poor fruits, Tephritidae often harbor many commensal bacteria in their gut (Capuzzo et al., Reference Capuzzo, Firrao, Mazzon, Squartini and Girolami2005; Mazzon et al., Reference Mazzon, Piscedda, Simonato, Martinez-Sañudo, Squartini and Girolami2008, Reference Mazzon, Martinez-Sanudo, Simonato, Squartini, Savio and Girolami2010; Kounatidis et al., Reference Kounatidis, Crotti, Sapountzis, Sacchi, Rizzi, Chouaia, Bandi, Alma, Daffonchio, Mavragani-Tsipidou and Bourtzis2009; Prabhakar et al., Reference Prabhakar, Sood, Kapoor, Kanwar, Mehta and Sharma2009, Reference Prabhakar, Sood, Kanwar, Sharma, Kumar and Mehta2013; Naaz et al., Reference Naaz, Choudhary, Prabhakar and Maurya2016). Gut microbiota are involved in the carbon and nitrogen cycles and help the host fruit flies synthesize essential amino acids and minerals (Miyazaki et al., Reference Miyazaki, Boush and Baerwald1968; Lauzon et al., Reference Lauzon, Sjogren and Prokopy2000; Lauzon et al., Reference Lauzon, Bussert, Sjogren and Prokopy2013; Behar et al., Reference Behar, Yuval and Jurkevitch2005). They can produce volatiles such as 2, 5-dimethyl pyrazine to attract flies (Robacker et al., Reference Robacker, Lauzon and He2004, Reference Robacker, Lauzon, Patt, Margara and Sacchetti2009). Meanwhile, gut bacteria can help flies degrade toxic purine compounds from host plants (Lauzon et al., Reference Lauzon, Sjogren and Prokopy2000), improve fruit fly fitness via prevention of the colonization of pathogenic bacteria and via an increase in pesticide resistance (Behar et al., Reference Behar, Yuval and Jurkevitch2008a; Cheng et al., Reference Cheng, Guo, Riegler, Xi, Liang and Xu2017), and they also help with mate selection (Damodaram et al., Reference Damodaram, Ayyasamy and Kempraj2016). It is clear that gut microbiota play an important role in the development, fitness, and reproduction of fruit flies; therefore, understanding the function of the gut microbiota of fruit flies will certainly help find new pest management strategies aimed at commensal bacteria (Lauzon et al., Reference Lauzon, Bussert, Sjogren and Prokopy2013).
Commensal bacteria play an important role in pest management of fruit flies. For example, some commensal bacteria and their chemical constitutes can be used to attract and monitor fruit flies (Hadapad et al., Reference Hadapad, Prabhakar, Chandekar, Tripathi and Hire2016). Removing commensal bacteria using antibiotics can be effectively used to control fruit flies. Treatment of the larvae of the melon fruit fly B. cucurbitae with the antibiotics oxytetracycline and sulfanilamide destroyed the symbiotic microorganisms in the mycetocytes of the midgut region and eventually led to reduced larval survival rates (Chinnarajan et al., Reference Chinnarajan, Jayaraj and Narayanan1972). Another fruit fly management tactic can be adding related bacteria in fly diet to prolong the survival of sterile male insects, and in order to improve the success of the sterile insect technique (SIT), this was used for Mediterranean fruit flies, Ceratitis capitata (Ben et al., Reference Ben, Yuval and Jurkevitch2010). SIT is a widely effective method for controlling fruit flies. Irradiation-induced sterile flies obtained from mass rearing are released in the field, they cross with wild population subsequently suppressing the production of normal offspring, and they ultimately successfully control the wild population growth (Ben et al., Reference Ben, Yuval and Jurkevitch2010). Increasing Klebsiella oxytoca to a level of stable colonization and decreasing the abundance of Pseudomonas spp. accelerates mating competitiveness of irradiated males compared with wild males in C. capitata (Behar et al., Reference Behar, Jurkevitch and Yuval2008b). Enterococcus phoeniculicola, a gut bacterium, also is known to promote larval and pupal developmental, whereas Lactobacillus lactis has negative effects on B. dorsalis development (Khaeso et al., Reference Khaeso, Andongma, Akami, Souliyanonh, Zhu, Krutmuang and Niu2018). Another pest management benefit of commensal bacteria is that they can create a suitable environment for parasitoid rearing. Commensal bacteria can improve the quality of parasitoids by increasing their size, fecundity, and longevity (Eben et al., Reference Eben, Benrey, Sivinski and Aluja2000; Cicero et al., Reference Cicero, Sivinski and Aluja2012).
The oriental fruit fly Bactrocera dorsalis (Hendel) is an invasive fruit fly pest in some parts of Asia and Africa and causes considerable economic losses (Clarke et al., Reference Clarke, Armstrong, Carmichael, Milne, Raghu, Roderick and Yeates2005; Schutze et al., Reference Schutze, Mahmood, Pavasovic, Bo, Newman, Clarke, Krosch and Cameron2015). The damage caused by this fruit fly can be attributed to its high fecundity and wide host range (Li et al., Reference Li, Wu, Chen, Wu and Li2012). In recent years, B. dorsalis has received a lot of attention because of its expanding geographic range and increase in the severity of damage it causes (Hu et al., Reference Hu, Chen and Li2014). A combination of molecular and morphological methods have been used to identify many commensal bacteria from the whole body, gut, and reproductive system of B. dorsalis, and Enterobacteriaceae species belonging to genera such as Klebsiella, Citrobacter, and Enterobacter were found to be the major species (Jang & Nishijima, Reference Jang and Nishijima1990; Sun et al., Reference Sun, Cui and Li2007; Wang et al., Reference Wang, Jin and Zhang2011, Reference Wang, Jin, Peng, Zhang, Chen and Hua2014; Shi et al., Reference Shi, Wang and Zhang2012; Andongma et al., Reference Andongma, Wan, Dong, Li, Desneux, White and Niu2015; Liu et al., Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016). Bacterial diversity among laboratory-reared, laboratory-sterile sugar-reared, and field-collected populations, also among different development stages (egg, larva, pupa, adult), and lastly among different invasion regions have been compared; the results showed the bacterial diversity was associated with foods and host plants (Wang et al., Reference Wang, Jin and Zhang2011; Andongma et al., Reference Andongma, Wan, Dong, Li, Desneux, White and Niu2015; Liu et al., Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016).
There have been three reports on the function of commensal bacteria in B. dorsalis. Bacteria strains isolated from the intestines and metabolites of one strain (Bacillus cereus) showed significant attraction to B. dorsalis adults (Wang et al., Reference Wang, Jin, Peng, Zhang, Chen and Hua2014). Phosphatase hydrolase genes, which can enhance resistance of B. dorsalis to the organophosphate insecticide trichlorfon have been identified in the gut microbiota Citrobacter sp. (Cheng et al., Reference Cheng, Guo, Riegler, Xi, Liang and Xu2017). Compared with the antibiotic-treated female adults, male flies were attracted to and ejaculated more sperm into females harboring the microbiota (Damodaram et al., Reference Damodaram, Ayyasamy and Kempraj2016). With the exception of pesticide resistance, attractiveness, and mate selection, the effects of gut microbiota on host fly reproduction and plant choice are still unknown and need more investigation.
Infecting or removing the target bacteria is often used to investigate the effects of commensal bacteria on host insects (Lee et al., Reference Lee, Park, Lee, Jang, Eo, Jang, Kim, Ohbayashi, Matsuura, Kikuchi, Futahashi, Fukatsu and Lee2017). Because many commensal bacteria cannot be cultured on media, removing target bacteria through antibiotic or high-temperature treatments is a much better choice than infecting host with bacteria. In fruit fly (Drosophila melanogaster), the model organism, antibiotic was used to investigate whether the shifts of microbiota can alter the hosts’ resistance to parasitism (Chaplinska et al., Reference Chaplinska, Gerritsma, Dini-Andreote, Salles and Wertheim2016). In Chinese citrus fly, Bactrocera minax, supplementing adult fly diets with bacterial isolates or antibiotics helped reveal the role of bacteria in the fitness of host flies. Antibiotic-treated B. minax showed significant decrease in fecundity, but increase in male and female longevity (Rashid et al., Reference Rashid, Andongma, Dong, Ren and Niu2018). Not only in fruit flies but also in many other insects, antibiotics were used to change bacteria diversity and to study the function of gut microbiota on that basis. Although antibiotics have been used in many studies to alter the gut bacteria diversity in order to study their function in host insects (Peterson et al., Reference Peterson, Stewart and Scharf2015; Raina et al., Reference Raina, Rawal, Singh, Daimei, Shakarad and Rajagopal2015; Taylor et al., Reference Taylor, Johnson and Dively2016; Yao et al., Reference Yao, Wang, Li, Cai, Lemaitre and Zhang2016), few reports have elucidated how bacterial diversity changes after antibiotic treatment (Lin et al., Reference Lin, Kang, Pan and Liu2015; Thakur et al., Reference Thakur, Dhammi, Saini and Kaur2016). In this study, the effects of three types of antibiotics on gut bacterial diversity of laboratory-reared B. dorsalis were investigated and results found may aid in improving understanding of the function of specific gut bacteria in insects.
Materials and methods
Insect rearing
In this study, we used laboratory-reared population of B. dorsalis to test gut bacterial diversity and their dynamic changes. Bactrocera dorsalis were originally collected from the Yunnan province and reared in an incubator (RXZ-500B, Ningbo) for at least 20 generations at 25 ± 0.5°C, 75 ± 5% relative humidity, and 14:10 h (light: dark) photoperiod. Adult fruit flies were initially maintained in steel-framed cages (35 × 35 × 35 cm) covered with gauze and supplied with deionized water and artificial diets. Artificial diet included powdered diet and agar-like diet. The powdered diet was composed of sucrose and soy peptone in a 1:3 ratio. The agar-like diet was as follows: sucrose 120 g, soybean peptone 10 g, dry yeast 40 g, Agar 10 g, methyl-p-hydroxybenzoate 1.2 g, sorbic acid 1 g, L-ascorbic acid 6.6 g, and enough deionized water to bring the volume to 1000 ml. The diet composition was optimized from Krainacker & Vargas (Reference Krainacker and Vargas1989). The diet was replaced every 2 days. Water was supplied in glass bottles with cotton balls. After adults developed to sexual maturity, they were provided papaya for oviposition and eggs were carefully collected and transferred into glass bottles with tampons (6 cm in diameter, 10 cm in height) containing agar-like diet. The third-instar larvae were picked out, and then were put into sterilized moist sand, waiting for pupation and eclosion.
Fruit fly treatment with antibiotics
The antibiotics, tetracycline, streptomycin, and ampicillin were tested (Yao et al., Reference Yao, Wang, Li, Cai, Lemaitre and Zhang2016). Antibiotics were mixed into artificial diet at four different concentrations. The larvae were transferred from normal diet into antibiotic diet at the first-instar stage and were then fed the antibiotic diet until pupation. The adults emerging from the treated larvae were continuously fed with antibiotic diet and antibiotic water at the same concentrations as larvae. The composition and concentration of the antibiotic in each treatment are listed in table 1. Each treatment group was divided into four sub-groups (T1, T2, T3, and T4) according to antibiotic concentration. The flies supplied with normal diet were the control group (CK). There were three replications in each group and 30 individuals in each replication.
Table 1. Composition and concentration of antibiotic for control and treatment groups.
CK, represents the control group supplied with normal diet; T, represents the treatment group supplied with diet with antibiotic at different concentrations.
Isolation and identification of cultivable gut bacteria
The sterilization and dissection of the third-stage larvae and 10-day-old adults followed the procedure of Liu et al. (Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016). For each control and treatment group, three biological replications were performed. Each replication included four individuals each of larvae, males, and females. The digestive system, including the gut and esophageal bulb, were dissected aseptically in a petri dish with sterile water under a stereoscope, and then was put in a 1.5 ml centrifuge tube containing 200 µl of sterile water. The tissue suspension of the dissected digestive systems was blended and then diluted into three concentrations (10−1, 10−2, and 10−3). Each suspension was spread on plate count agar (PCA) medium on a clean worktable and was then cultured at 35 ± 1°C in a constant temperature incubator. Number of bacterial colonies for each concentration was counted and compared with the control after 48–72 h. Bacterial colonies were isolated and purified by spreading methods (this step was just used in larvae, because there were only a few or even no cultivable bacteria in adults after feeding on antibiotic-laced diets). Individual colonies from these plates were picked and re-inoculated onto fresh PCA medium. This procedure was repeated thrice. Each purified cultivable bacteria colony with a different size, color, and morphological characteristics was identified via 16s rDNA sequencing. The universal primer of bacteria and the method for DNA extraction and sequence amplification were consistent with those reported by Liu et al. (Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016).
Identification of gut bacteria
Genomic DNA was extracted from digestive systems using E.Z.N.A™ Mag-Bind Soil DNA Kits (OMEGA, D5625-01) following the manufacturer's instructions. Gene libraries were constructed by analyzing the V3-V4 region of 16s rDNA using a universal primer comprising the A-linker, the B-linker, and the sample-specific tag (16s 341F, 16s 805R). The primer sequences are listed in table 2.
Table 2. Primer sequences used for V3−V4 region amplification.
Each polymerase chain reaction (PCR) was conducted in a reaction volume of 30 µl containing 1 µl of Bar-PCR primer F (10 µM), 1 µl of Primer R (10 µM), 20 ng of genomic DNA, 15 µl of 2 × Taq Master Mix, and enough ddH2O to bring the volume to 30 µl. The amplification program was as follows: 94°C for 3 min, then five cycles (94°C for 30 s, 45°C for 20 s, 65°C for 30 s), 20 cycles (94°C for 20 s, 55°C for 20 s, 72°C for 30 s), and finally 72°C for 5 min and 10°C indefinitely. A second round of amplification following the first involved the introduction of Illumina bridge PCR-compatible primers, with an amplification program as follows: 95°C for 3 min, then five cycles (94°C for 20 s, 55°C for 20 s, 72°C for 30 s), 72°C for 5 min, and 10°C preserved.
PCR productions were sent to the Sangon Biotech Company (Shanghai, China) for 16S rDNA library construction. Gut bacteria were identified via high-throughput metagenomic microbial sequencing.
Illumina sequencing and data analysis
The data analysis of Illumina sequencing included data processing and statistics, operational taxonomic unit (OTU) cluster, α diversity, and bacteria classification of gut bacteria.
Data processing and statistics
Quality check of reads was analyzed using Prinseq 0.20.4 (Schmieder & Edwards, Reference Schmieder and Edwards2011), cutadapt 1.2.1 (Chen et al., Reference Chen, Khaleel, Huang and Wu2014), and PEAR 0.9.6 (Zhang et al., Reference Zhang, Kobert, Flouri and Stamatakis2014) software. Sequences with low quality were removed through USEARCH 8.1 package (Edgar, Reference Edgar2010). Chimeras were identified and removed by using UCHIME 4.2.40 (Edgar et al., Reference Edgar, Haas, Clemente, Quince and Knight2011).
OTU cluster
USEARCH 8.1 package was used for OTU analysis at the 97% similarity. One representative sequence with the highest similarity was chosen from each OTU for Venn and phylogenetic analysis, and was then compared with the RDP database to obtain the taxonomic information of each OTU. Phylogenetic distance among samples was calculated using Bray–Curtis with the vegan package 2.0-10 of R software 3.2 (Noyce et al., Reference Noyce, Fulthorpe, Gorgolewski, Hazlett, Honghi and Basiliko2016). Venn diagram, which showed the similarity and difference among control and treatment groups, was obtained from the Venn diagram package 1.6.16 of R software. Bray tree and PCA analysis displaying the correlation among samples were obtained using vegan package 2.0-10 of R (Noyce et al., Reference Noyce, Fulthorpe, Gorgolewski, Hazlett, Honghi and Basiliko2016).
The α diversity
In order to estimate the adequacy of sampling, Chao rarefaction curves were generated. The Chao1 richness estimator (Chao, Reference Chao1984), abundance-based coverage estimators (ACE), Coverage, and Shannon indices were calculated using Mothur ver. 1.30.1 (Schloss et al., Reference Schloss, Westcott, Ryabin, Hall, Hartmann, Hollister, Lesniewski, Oakley, Parks, Robinson, Sahl, Stres, Thallinger, Van Horn and Weber2009).
Bacterial classification
Representative sequence from each OTU was Blast using the NCBI 16S rDNA and 18S rDNA database: http://ncbi.nlm.nih.gov/. Blast results with the similarity and coverage higher than 90% were used for analysis, otherwise all the sequences were regarded as unclassified sequences. The phylogenetic trees of identified bacterial species at the genus level were obtained using ete3 package of python 2.6.6 (Huerta-Cepas et al., Reference Huerta-Cepas, Dopazo and Gabaldón2010).
Bacterial diversity comparisons
Four types of bacterial diversity comparisons were performed. The first one compared cultivable bacterial diversity among the control and four treatments of third-stage larvae. Another three comparisons, including cultivable and gut bacterial diversity, focused on the control and four treatments of larvae (CK-L, T1-L, T2-L, T3-L, T4-L), males (CK-M, T1-M, T2-M, T3-M, T4-M) and females (CK-F, T1-F, T2-F, T3-F, T4-F). Clone percentage across a number of PCR libraries was used to describe the relative abundance of each genus.
Statistical analysis
SPSS software was used to analyze the number of cultivable gut bacteria in the digestive system by one-way analysis of variance (ANOVA) with the square root transformation to stabilize variances but untransformed data are reported. Significant ANOVAs were followed by multiple comparisons of mean values (P ≤ 0.05) (Bertacco et al., Reference Bertacco, Dehner, Caturegli, D'Amico, Morotti, Rodriguez, Mulligan, Kriegel and Geibel2017).
Results
Isolation and identification of cultivable gut bacteria in digestive systems
The number of cultivable gut bacteria was gained in five groups of larvae, male adults, and female adults after 48–72 h of culturing on PCA medium. The bacteria suspension diluted 100 times was chosen to count colony numbers, where the average colony numbers of five groups (CK, T1, T2, T3, and T4) were 1232, 125, 76, 62, and 34, respectively, for each larval group; 906, 1, 1.5, 0, and 0, respectively, for each adult female group; and 700, 1, 0.5, 0, and 0, respectively, for each adult male group. The results showed that, even for treatment 1 group of larvae, female, and male adults, the cultivable colony number was significantly smaller (P < 0.05). However, there were no significant differences among the four treatment groups (T1, T2, T3, and T4) of larvae, female adults, and male adults (fig. 1). Based on NCBI Blast results, four genera, including Arthrobacter, Bacillus, Lactobacillus, and Morganella, were found in the control group of larvae. However, Morganella was killed in all the treatment groups, Lactobacillus was killed in the T2 and T4 groups, and Bacillus was killed in only the T4 group. Actinokineospora also appeared only in the T4-L group (table 3).
Fig. 1. Number of cultivable bacteria per larva before and after feeding on antibiotic. Black: larvae; light gray: female adult; dark gray: male adult. CK: control group; T1−T4: four treatment groups feeding on antibiotics in four combinations according to table 1. *P < 0.05.
Table 3. Isolated bacteria from different groups of larvae.
T, means treatment group at different antibiotic concentrations; +, represents bacteria found in control or treatment groups.
Identification of all gut bacteria in digestive systems
Bacterial 16S rDNA libraries
The 16S rDNA libraries were constructed for 15 samples. The quality control results for control and treatment groups are shown in Supplementary table 1. The sequence and OTU number and α indices are shown in Supplementary table 2. The higher the Simpson index was, the lower the diversity of the community. ACE index refers to the index of community richness. The rarefaction curves for most of the clone libraries tended to be saturated at 3% difference between sequences (Supplementary fig. 1). The coverage was higher than 95% suggesting that the number of clones sampled was enough to provide an adequate estimation of B. dorsalis bacterial diversity. All the 16S rDNA sequences were submitted to NCBI (SRA accession number: SRP145819).
Bacterial diversity
The first ten dominant bacterial genera in the control and treatment groups of larvae, female adults, and male adults are shown in table 4. For larvae, comparing with the control group, the content of two genera (Lactobacillus and Brevundimonas) decreased and that of five genera (Bacillus, Lactococcus, Geobacillus, Acinetobacter, and Streptococcus) increased in the treatment groups. For antibiotic-treated females, the content of three genera, Pseudomonas, Enterobacter, and Leclercia decreased, but Leclercia was killed. There was also an increase for five genera, Bacillus, Lactobacillus, Acinetobacter, Stenotrophomonas, and Streptococcus. For antibiotic-treated males, Enterobacter, Acinetobacter, Leclercia, and Achromobacter decreased, and Stenotrophomonas, Bacillus, Streptococcus, and Lactococcus increased. Achromobacter and Leclercia were also killed, while Pseudomonas did not show any changes.
Table 4. The top 10 genera of gut bacteria in control and treatment group of B. dorsalis
CK, the control group supplied with normal diet; T, the treatment group supplied with diet with antibiotic at different concentrations; L, larvae; F, female adults; M, male adults.
Dominant populations of bacteria
The dominant populations of CK-L were Bacillus, Lactococcus, Lactobacillus, and Brevundimonas (81.56%), while those of T1-L, T2-L, T3-L, and T4-L were Bacillus and Lactococcus (64.74, 67. 01, 64, and 62.29%, respectively). After being fed with antibiotics, two genera of commensal bacteria of larvae were reduced, and these were Lactobacillus and Brevundimonas (fig. 2a). Pseudomonas was the dominant population of CK-F, with a percentage of 75.65, Compared with CK-F, the dominant population of bacteria changed from Pseudomonas to Bacillus and Lactococcus (fig. 2b). The percentages of T1-L, T2-L, T3-L, and T4-L were 75.65, 50.3, 24.96, 27.39, and 34.12, respectively. Compared with CK-M, the dominant populations of bacteria changed from Enterobacter and Acinetobacter to Bacillus, Lactococcus, Enterococcus, and Pseudomonas (fig. 2c). The percentages of four treatment groups were 88.89, 47.62, 62.39, 54.29, and 44.34, respectively.
Fig. 2. Frequency of the detected gut bacteria at the genus level. a: larvae; b: female adult; c: male adult. CK: control group; T1−T4: four treatment groups feeding on antibiotics in four combinations according to table 1. Different colors represent different genera. The bar means the frequency of each genus in the reads.
Population differences
According to the Venn diagram, there were 1416 OTUs in the CK and four treatments of larvae in all, but only 125 OTUs were shared by all of them (fig. 3a); there were 1886 OTUs in all in the five groups of females, but only 18 OTUs were shared (fig. 3b); and there were 1602 bacteria in all in the five groups of males, but only 15 OTUs were shared by all of them (fig. 3c). The shared OTU number decreased from larva to adult.
Fig. 3. Venn diagram of genus number among the control and four treatment groups in Bactrocera dorsalis. a: larvae; b: female adult; c: male adult. CK: control group; T1−T4: four treatment groups feeding on antibiotics in four combinations according to table 1. Blue: control group; orange: T1 treatment group; pink: T2 treatment group; green: T3 treatment group; purple: T4 treatment group.
A multi-sample comparison tree was used to compare the community similarity of each sample at the OTU composition and phylogenetic level. As Supplementary fig. 2a shows, bacteria communities of T1-L, T2-L, T3-L, and T4-L were more similar, and their phylogenetic distance was closer compared with that of CK-L. The same results were seen for female and male adults (Supplementary fig. 2b and c). For female adults, the T3-F and T4-F treatments, representing the highest concentrations of antibiotics, are the farthest distance from CK-F (Supplementary fig. 2b), while for male adults, the T4-M treatment was in the same position with T3-F and T4-F (Supplementary fig. 2c). Principal component analysis (PCA) was also used to compare the community structure similarity of each group. Each of the scatter points represents each group in the graph. The more similar the composition of the sample, the closer the distance is in the PCA map. Bacterial communities of the four treatment groups are similar, as they gather together at the top left (larvae) or right (female and male adults) of the figure (fig. 4). The distance between treatment and control groups in larvae is closer than that of male and female adults. The Bray tree and the PCA figure show the same results on community similarity.
Fig. 4. Principal component analysis of gut bacteria diversity among the control and treatment groups. a: larvae, b: female adult; c: male adult. Black circle: control group; yellow square: T1 treatment group; blue rhombus: T2 treatment group; blue regular triangle: T3 treatment group; purple inverted triangle: T4 treatment group.
The relative abundance and diversity of bacteria after the host flies being fed with antibiotics are shown in Supplementary fig. 3. The first 30 OTUs from the larvae group were distributed in nine classes and five phyla. Lactobacillus and Brevundimonas were only in the control larvae and had almost been removed by antibiotic treatment. These two bacterial genera belong to the Bacilli (Fimicutes) and α-proteobacteria (Proteobacteria) classes, respectively (Supplementary fig. 3a). For female adult flies, the first 30 OTUs were distributed in six classes and three phyla. Pseudomonas, Citrobacter, Enterobacter, and Leclercia genera, which belonged to the same class of γ-proteobacteria, showed a high proportion of these four genera in the control group. The decreased proportion in the treatment groups meant that these four genera were removed effectively by antibiotic treatment. Most of the other genera (except Lactococcus) were in treated female adult flies, which meant that many new bacteria appeared after antibiotic treatment (Supplementary fig. 3b). In the male groups, the first 30 OTUs were distributed in seven classes and four phyla, among which Enterobacter and Achromobacter were almost completely removed after the antibiotic was ingested. The remaining three phyla, including Firmicutes, Actinobacteria, and Bacteroidetes, mostly appeared in treatment groups. Brochothix genus was only found in T4 treatment groups, while most α-proteobacteria showed the same pattern of Brochothix genus (Supplementary fig. 3c).
Discussion
Bacteria diversity comparison
This is the first study to compare the before and after gut bacterial diversity using different development stages of B. dorsalis fed on diets laced with three different antibiotics. Additionally, this is the first study that has investigated the dynamic changes in gut bacteria after antibiotic treatment of B. dorsalis. In this study, three types of antibiotics were used together to change gut microbe populations. After being fed with antibiotics, B. dorsalis larval bacterial diversity did not change much because the antibiotics were still in the process of action, but colonies of Lactobacillus and Brevundimonas were decreased. In adults, the bacterial diversity was significantly increased because the antibiotic had worked fully by that life stage. The inhibitor of dominant bacteria allowed the increase of some other gut bacteria. Increase in gut bacterial diversity and the degree of bacterial removal were not influenced by increased antibiotic concentration.
Many reports have revealed that gut bacterial diversity in insects varies according to the host plants, rearing environment and geographical populations, because of the long time that has been spent on feeding the diet (Andongma et al., Reference Andongma, Wan, Dong, Li, Desneux, White and Niu2015; Morrow et al., Reference Morrow, Frommer, Shearman and Riegler2015; Chaplinska et al., Reference Chaplinska, Gerritsma, Dini-Andreote, Salles and Wertheim2016). The results of the bacteria community makeup before antibiotic treatment in this study are consistent with those of Andongma et al. (Reference Andongma, Wan, Dong, Li, Desneux, White and Niu2015), who reported that female and male adults had similar bacteria communities, but they were different from those in larvae, eggs, and pupae. Even after antibiotic treatment, the male and female adult groups shared similar bacteria diversity. Andongma et al. showed that Proteobacteria dominated in immature stages, while Firmicutes dominated in adult stages of B. dorsalis. In this study, though the genera were different among larvae and adults, most of the dominant genera were proteobacteria. Liu et al. (Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016) reported the bacteria composition of male adults from 14 geographic populations, and retrieved 13 genera. Our current study obtained similar communities in the control samples of male adults, and reterived ten main genera, of which Enterobacter, Pseudomonas, Stenotrophomonas, and Lactoccus were the same as that reported by Liu et al. (Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016). When we analyzed our results at the family level, all the ten main genera can be attributed to Enterobacteriaceae, Streptococcaceae, Xanthomonadaceae, Pseudomonadaceae, and Bacillaceae, and the first four had also been found in B. dorsalis by Liu et al. Enterobacteriaceae, Enterococcaceae, and Bacillaceae were also confirmed to be the main bacterial community in the intestinal tract of B. dorsalis by Wang et al. (Reference Wang, Jin, Peng, Zhang, Chen and Hua2014). Though there are some differences at the genus level, all the reports showed samiliar gut bacterial diversity in B. dorsalis at the family level. The small difference can be attributed to the different artifical diets used for rearing B. dorsalis. The three antibiotics (tetracycline, ampicillin, and streptomycin) used in the current study have been used by Damodaram et al. (Reference Damodaram, Ayyasamy and Kempraj2016) to change the bacteria in B. dorsalis; however, they focused only on the change in mating behavior and did not compare the bacteria diversity before and after antibiotic treatment.
Despite the fact that there are many studies indicating gut bacterial diversity of insects can be changed by antibiotic treatment, there appears to be only one study that elucidates the dynamic changes in bacterial communities in Spodoptera litura (Lepidoptera: Noctuidae) after antibiotic treatment (Thakur et al., Reference Thakur, Dhammi, Saini and Kaur2016). Unlike our current study, Thakur et al. (Reference Thakur, Dhammi, Saini and Kaur2016) used only one antibiotic (streptomycin) and three concentrations (0.03, 0.07, and 0.15%), and only cultivable bacteria of third-instar larvae of S. litura were detected. Planococcus citreus, Staphylococcus sciuri, and Enterobacter cloacae were totally removed, while only Microbacterium paraoxydans, Bacillus methylotrophicus, and B. amyloliquefaciens with lower abundance were isolated from the gut of larvae feeding on streptomycin-laced diet (Thakur et al., Reference Thakur, Dhammi, Saini and Kaur2016). Just like S. litura, Bacillus genus cannot be removed totally by antibiotics in larvae or adults of B. dorsalis. Enterobacter was similar to S. litura in that it was only retrieved in adults of B. dorsalis. Though Enterobacter can be decreased to a very low level, it cannot be totally removed by antibiotics. Our results showed high resistance to antibiotics in gut bacteria of B. dorsalis and this may explain why B. dorsalis can be an invasive pest in China. Though our study involved more gut bacteria than in Thakur et al. (Reference Thakur, Dhammi, Saini and Kaur2016), because of sequencing method, we obtained similar results, which showed that the different concentrations of streptomycin sulfate did not result in changes of gut microbial diversity of larvae.
Advantages and disadvantages of the use of antibiotics
Though antibiotics have been useful tools in investigating the function of gut bacteria, there are some additional considerations. This method is advantageous in that it makes the in vivo uncovering of gut bacteria possible, especially for some uncultivable bacteria (Li et al., Reference Li, Evans, Li, Zhao, DeGrandi-Hoffman, Huang, Li, Hamilton and Chen2017; Myint Khaing et al., Reference Myint Khaing, Yang, Zhao, Zhang, Wang, Wei and Liang2017). However, the non-lethal and toxic levels of antibiotics can influence the result (Xu et al., Reference Xu, Buss and Boucias2016). Teasing out the influence of antibiotics themselves on host insects becomes a key question. To answer this question, wild-strain control groups without bacteria should be fed antibiotics in order to determine the influence of the antibiotic itself (Yao et al., Reference Yao, Wang, Li, Cai, Lemaitre and Zhang2016; Lee et al., Reference Lee, Park, Lee, Jang, Eo, Jang, Kim, Ohbayashi, Matsuura, Kikuchi, Futahashi, Fukatsu and Lee2017). However, for insects that have no target commensal bacteria-free strains, bacteria ingestion may be a better choice to study bacteria function (Lee et al., Reference Lee, Park, Lee, Jang, Eo, Jang, Kim, Ohbayashi, Matsuura, Kikuchi, Futahashi, Fukatsu and Lee2017). In future, bacteriophages may also be used in place of antibiotics for bacteria removal (Louradour et al., Reference Louradour, Monteiro, Inbar, Ghosh, Merkhofer, Lawyer, Paun, Smelkinson, Secundino, Lewis, Erram, Zurek and Sacks2017). Tetracycline and ampicillin can be used to remove gut bacteria selectively. Homona magnanima was coinfected with Wolbachia and Spiroplasma. Tsugeno et al. (Reference Tsugeno, Koyama, Takamatsu, Nakai, Kunimi and Inoue2017) used tetracycline to selectively remove Wolbachia from female adults, while Spiroplasma remained positive. The use of tetracycline facilitated the determination of the function of Spiroplasma, which is responsible for male killing in the Oriental Tea Tortrix (Tsugeno et al., Reference Tsugeno, Koyama, Takamatsu, Nakai, Kunimi and Inoue2017). However, this study used three kinds of broad-spectrum antibiotics, which affected most gut bacteria. The related phenotype after using broad-spectrum antibiotic was caused not by one special bacteria, but is the combined action of many bacteria, which posed a barrier to the study of the function of one single gut bacteria. Research and development of new selective antibiotics will offer a better chance for studying the function of individual species of gut bacteria. Meantime, antibiotic sensitivity of bacteria in B. tau was studied to find useful symbiotic bacteria, and the results showed that all the bacteria are sensitive to co-trimoxazole, gentamycene, and nofloxacin (Sood & Prabhakar, Reference Sood and Prabhakar2009). As in the Sood & Prabhakar (Reference Sood and Prabhakar2009) study involving B. tau, our study is also restricted to the effects of antibiotics on a laboratory-reared population. Studying different bacterial composition between reared and field populations, between different geographic populations, extending the research to the natural population of fruit flies would help increase our understanding of the eco-friendly use of bacteria in pest management (Chaplinska et al., Reference Chaplinska, Gerritsma, Dini-Andreote, Salles and Wertheim2016).
Function research on gut bacteria
Previous research reported possible functions of gut bacteria in fruit flies, among which providing the necessary nutrition and helping degrade toxins in host plants are the most important functions (Liu et al., Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016). In the case of B. dorsalis, functions include insecticide resistance, attractiveness, and mating behavior regulation (Damodaram et al., Reference Damodaram, Ayyasamy and Kempraj2016; Liu et al., Reference Liu, Martinez-Sanudo, Mazzon, Prabhakar, Girolami, Deng, Dai and Li2016; Cheng et al., Reference Cheng, Guo, Riegler, Xi, Liang and Xu2017). Besides, antibiotic can decrease the lipoid level and fecundity of female feeding on a full diet, via changing the microbiota composition (Ben-Yosef et al., Reference Ben-Yosef, Jurkevitch and Yuval2008). The potential use of microbiota for improving pupal and adult productivity, and shortening rearing duration particularly for males, also facilitates the use of SIT in pest control (Augustinos et al., Reference Augustinos, Kyritsis, Papadopoulos, Abd-Alla, Cáceres and Bourtzis2015). However, the most important requirement for determining the function of gut bacteria is uncovering the bacterial composition. Thus, our study provides a foundation for research on the function of gut bacteria of B. dorsalis, and will facilitate future investigation of hypothesized functions of specific gut bacteria. In the future, research involving the ingestion of just single bacterial species may help identify the main gut bacteria that help B. dorsalis to have a broad host plant range, a reproduction ability, and the ability to invade new areas.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485318000834.
Acknowledgement
We thank our colleague George Opit, professor in Oklahoma State University, and Renfu Shao, research fellow in University of the Sunshine Coast who helped us review earlier versions of this manuscript. Financial support was provided by the National Key Research and Development Project of China (No.2016YFC1200605) and the Natural Science Foundation of Beijing (No. 6174043).
