Healthy adult CPBs were collected from Suifenhe, Heilongjiang Province, and cultured with fresh potato leaves in the Safety Laboratory of Suifenhe Customs Technology Center, Heilongjiang Province. The temperature of the greenhouse was 26 ± 2°C, the relative humidity was 40–60%, and the photoperiod was 16 L:8D.

Insect feeding behavior is initiated by the volatiles emanating from the leaves of a host plant, and it is primarily governed by olfaction. To explore the factors influencing feeding selection behavior in CPBs, fresh potato leaves were immersed in a mixture of five antibiotics (tetracycline, penicillin, ofloxacin, ciprofloxacin, and ampicillin) at specific concentrations (50 μg/mL, 100 μg/mL, and 1 mg/mL) for 1 h. The five antibiotics in the mixture were mixed at the same ratio, and 1% Tween-20 was added to improve the adsorption capacity of the antibiotics on potato leaves. The experimental potato beetles were fed leaves soaked in each concentration of antibiotic mixture for 3 days, and the soaked potato leaves were replaced once a day. The control group was fed sterile water and potato leaves soaked in 1% Tween-20 for 3 days, and the development and feeding conditions were the same to eliminate the influence of other factors on their feeding selection behavior. The feeding of the experimental and control group was stopped 6 h before the test. The insect olfactory behavior instrument was divided into treatment and control areas. Fresh potato leaves (30 g) were placed in the treatment area, and the control area was blank. Each test time was 30 min, and the number of selected insects on each wall was recorded after the test time. Each experiment was repeated three times. We maintained a constant light intensity in the experimental area. The attraction index was calculated as the treatment group minus the control group divided by the total.

For 16S sequencing of the gut microbes of antibiotic-treated and non-treatment control CPBs, genomic DNA was extracted using the Fast DNA Stool Mini Kit (Qiagen, Germany). The quality and quantity of the extracted DNA were measured on a 1% agarose gel and using a Nanodrop, respectively. This DNA was used as a template for amplifying 16S V3–V4 fragments by RT-PCR using primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACH VGGGTWTCAAT-3′) (Frank et al., 2013). The libraries were constructed using amplified PCR products, which were then sequenced using an Ion Torrent S5TM XL platform.

The PureLink™ RNA Mini Kit was used to extract RNA from whole heads of CPBs in the antibiotic treatment and control group, and the Illumina Hiseq™ sequencing platform was used for RNA transcriptome sequencing. Expression levels were assessed in terms of TPM values (transcripts per kilobase per million reads), which were calculated based on the number of mapped transcript fragments corrected for transcript length and sequencing depth through featureCounts (http://subread.sourceforge.net) software. Differential expression analysis of two samples was performed using the DESeq2 R package (Love et al., 2014). The software applied the negative binomial generalized linear regression model to fit gene expression, evaluate the variation of dispersion and difference multiple, and Wald test was used for statistical test to obtain p value. Differentially expressed gene screening requirements were|log2fold change| > 1, and a p-value <0.05 (Trapnell et al., 2010).

The qRT-PCR was performed in a 20-μL reaction mixture (including 1 μL of cDNA, 10 μL of 2× TransStart®Green qPCR SuperMix, 1 μL of 10 μM forward and reverse primers, and 7 μL of nuclease-free water) on an ABI 7900 system (Applied Biosystems, CA, United States) to compare the differences in the expression of OR genes between the antibiotic treatment, control, and Rescue group in the heads of CPBs. It was performed under the following conditions: 95°C for 2.5 min and 36 cycles at 94°C for 15 s, 60°C for 15 s, and extension at 72°C for 15 s. The primers (5′-3′) used were as follows: OR9 forward primer, CTTTTCTAGGAGCCCAAC; OR9 reverse primer, ATTACTCTTGAAAGGCCAT; OR17 forward primer, ATAGTATTATGACTAATAGCGAAG; and OR17 reverse primer, GAAAATTATGGGCGCTTT. The melting curve was analyzed to assure the specificity of the primers after each reaction, and the RP18 gene was used as the reference gene (Shi et al., 2013). The 2^−ΔΔCT method (Livak and Schmittgen, 2001) was used to calculate the expression level of each OR gene. Each sample type was repeated three times. The differences between the relative expression levels of OR genes were analyzed using a t-test. The experimental data were statistically analyzed using GraphPad Prism 5 (GraphPad Software, CA, United States).

We used traditional microbial separation and purification techniques to isolate the intestinal tracts of adult CPBs from Heilongjiang province. We obtained different monoclonal strains and cloned and sequenced 16S rDNA for each. We cultured three Pseudomonas strains (P. rhodesiae 39B5, P. rhodesiae VTT E-031889, and P. plecoglossicida sp. RL104) separately and diluted the concentration of the bacterial solution to 10⁵ cells/mL. To prepare the recovery group for experiments, we mixed 50 mL of each of the above three bacterial solutions, soaked fresh potato leaves in 150 mL of the bacterial mix solution for 1 h, and then fed to CPB adults after 3 days of antibiotic treatment (1 mg/mL). The feeding process was repeated for 3 consecutive days. We compared the feeding choice behavior among the three groups, which were the Rescue group and two control groups (one treated with antibiotics and one treated with sterile water using the method in section “Effects of gut microbiota treatment on the feeding selection behavior of CPBs”). The methods of behavior testing and attraction index calculation were the same as those described in section “Effects of gut microbiota treatment on the feeding selection behavior of CPBs”.

To investigate the relationship between intestinal microbes and olfactory-mediated feeding behavior in CPBs, we fed adult CPBs with a compound antibiotic solution consisting of tetracycline, ciprofloxacin, penicillin, ofloxacin, and ampicillin to alter the abundance of microbes. The effects of antibiotics on the olfactory behavior of CPBs were investigated via insect olfactory analysis. The different effects of the mixed antibiotic solution on the leaf selection behavior of CPBs were analyzed. The feeding behavior of CPBs was significantly changed by feeding on the leaves soaked in the antibiotic mixture, and the selection of host plant leaves was significantly reduced. After feeding with antibiotics, CPBs lost their ability to distinguish between leaves and blank controls (the attraction index was close to zero). We selected different concentrations of antibiotics for feeding treatment, including 1 mg/mL, 100 μg/mL, and 50 μg/mL. No significant difference was observed in the selection of host plant leaves by CPBs with different concentrations of antibiotics (Figure 1). These results indicate that a dose of 50 μg/mL of antibiotics may be sufficient to reduce the olfactory discrimination ability of CPBs to host plants, which can be developed and used as a new green control method in the future.

Based on the changes in the olfactory behavior of adult CPBs after feeding with a mixture of antibiotics, we performed 16S sequencing of intestinal microorganisms in adult CPBs treated with 1 mg/mL antibiotics and in non-treatment control CPBs to explore the differences in the composition of intestinal microorganisms between the two groups. A total of 46 phyla, 65 classes, 154 orders, 286 families, 779 genera, and 951 species were obtained from the intestinal microorganisms of the CPBs in the two groups. The number of annotations obtained for the two groups at different taxonomic levels is listed in Table 1. The results showed that the gut microbiota diversity of the control group was slightly higher than that of the antibiotic group, and the relative abundance of bacteria between the two treatments differed significantly at each classification level (Figure 2). The relative abundance of Pseudomonas was significantly higher in the control group than that in the antibiotic group (p = 0.030).

To investigate the changes in OR gene expression induced by gut microbiota, we screened the expression differences of all genes through transcriptome analysis with a reference genome. According to the results of all samples and splicing, the transcript per million (TPM) value was used to screen differentially expressed genes. Compared with the non-treatment group, 1728 significantly differentially expressed genes in the three antibiotic treatment groups were identified, of which 822 were upregulated and 906 were downregulated. According to the mean TPM value ≥5, |log₂fold change| > 1, and a p-value <0.05, only one upregulated OR gene, LdecOR9, and one downregulated OR gene, LdecO17, were identified (Figures 3A,B). The detailed data can be found in Supplementary Data 1. The OR genes LdecOR9 and LdecOR17, with significant differences, were validated via quantitative real-time fluorescence PCR, with the RP18 gene as the internal reference gene. Comparing the results of qPCR with the TPM results of transcriptome sequencing, the expression trends of LdecOR9 and LdecOR17 were consistent (Figures 3C–F), which further indicates that the transcriptome sequencing results were reliable.

The use of antibiotics has a great impact on gut microbiota; thus, we adopted a feeding method to restore intestinal gut microbiota in CPBs. We then compared the changes in CPB feeding behavior in three strains of Pseudomonas (P. rhodesiae 39B5, P. rhodesiae VTT E-031889, and P. plecoglossicida sp. RL104; bacterial concentration of 10⁵). These strains were isolated from the intestinal tract of adult CPBs from Heilongjiang province and were chosen based on previous 16S rDNA sequencing analysis results and the literature review. After feeding antibiotic-treated adult CPBs, which had marked changes in feeding selection behavior, we used the Pseudomonas strains to verify the behavior experiments and investigate the effects of gut microbiota on the olfactory behavior and olfactory-related genes in adult CPBs.

The fresh potato leaves were soaked in the Pseudomonas bacterial mixture at a concentration of 10⁵ to feed the antibiotic-treated adult CPBs. We then observed whether their feeding choice behavior was restored. As can be seen from the results, the index of attraction to host plant leaves was approximately 28% in the non-treatment group and approximately 0% in CPBs fed with leaves soaked in a 1 mg/mL antibiotic mixture, which shows changes in feeding selection behavior. The feeding selection behavior of adult CPBs was significantly recovered, reaching approximately 15%. The results showed that antibiotic feeding had a significant effect on the olfactory behavior of adult CPBs. The effects of antibiotics could be partially recovered, but incomplete recovery may be mainly due to the complex gut microbial population. Feeding only the above target bacteria could not completely recover the bacterial population dynamics in CPBs. We further verified the two olfactory genes LdecOR9 and LdecOR17 using quantitative real-time PCR and found that gene expression was partially restored (Figure 4), which was correlated with the changes in intestinal microorganisms, further proving that gut microbiota influence feeding behavior through changes in OR gene expression in CPBs.