Plutella xylostella used in this study were previously the focus of a genomics analysis (You et al., 2013). The colony was established in 2004 and individuals have been continuously reared on radish seedlings using the method described by Xia et al. (2015).

Enterobacter sp. (JQ396388), Serratia sp. (JQ396393), and Enterococcus sp. (KC150018), which were previously isolated from P. xylostella guts were cultured in LB medium (see Xia et al., 2015). The bacteria were incubated on a rotary shaker at 150 rpm at 37°C overnight in 100 mL of LB medium. The cultures were subsequently centrifuged and the LB medium was removed. The bacteria were washed with double distilled (dd) H₂O to remove residues of the culture medium, and then diluted with ddH2O to a concentration of OD₆₀₀ = 1.0. In order to study the effects of heat treatment on P. xylostella gut bacteria, the three bacterial strains, at densities of OD₆₀₀ = 1.0, were heat killed at 70°C for 15 min and then 10 μL of the heat killed bacteria solution was spotted onto LB medium and cultured at 37°C for 24 h. Each experiment was replicated three times.

Content of chlorpyrifos in solution can be determined by ultraviolet (UV) spectrophotometry (Xie et al., 2005), so we constructed a standard curve by dissolving 25 mL of chlorpyrifos (100 mg L⁻¹) in petroleum ether. The chlorpyrifos solution was diluted to concentrations of 0.0, 3.125, 6.25, 12.5, 25, 50, and 100 mg L⁻¹ using petroleum ether, for measurement using a UV spectrophotometer at 293 nm absorbance. The standard curve, obtained using linear regression, was used to determine the concentration of degraded chlorpyrifos by the three bacterial strains.

To determine the degradation efficiency of the isolated strains of Enterobacter sp., Serratia sp., and Enterococcus sp., we first enriched the strains in LB liquid medium. When OD₆₀₀ nm was about 0.5, the cells were harvested by centrifugation at 5,000 r min⁻¹ for 5 min, before being washed twice in minimal salt (MS) solution (1.5 g K₂HPO₄, 0.5 g KH₂PO₄, 0.5 g (NH₄)₂SO₄, 0.5 g NaCl, 0.2 g MgSO₄, 0.05 g CaCl₂, 0.02 g FeSO₄, and 20 g agar in 1 L of ddH₂O; pH 7.0) and re-suspended in MS solution to OD₆₀₀ = 1.0. The bacteria (2% MS solution) were inoculated into 100 mL of MS medium that contained 50 mg L⁻¹ of chlorpyrifos; the control containing an equal quantity of chlorpyrifos was not inoculated with bacteria. The control and bacteria treated insecticide MS media were incubated on the rotary shaker at 200 rpm at 37°C. Following incubation for 24 h, 4 mL of the culture medium was removed and an equal volume of petroleum ether was added before vortexing the mixture for 1 min. The organic and aqueous phases separated completely after 1 h, at which point the chlorpyrifos were extracted for analysis, where the organic phase solution of the upper layer was used to determine concentration at 293 nm absorbance. All experiments were replicated three times.

The degradation efficiency of chlorpyrifos was calculated as:

where X represented percent degradation efficiency; C_b was the concentration of chlorpyrifos in the medium degraded by a specific strain of bacteria (_b) after culturing for 24 h; and, C_CK was the concentration of chlorpyrifos in the control group (no bacteria) that after culturing for 24 h under the same conditions. The degradation efficiency of pesticide was analyzed using one-way ANOVA, followed by an LSD post-hoc test.

In order to detect colonization by the three strains of bacteria in the P. xylostella gut, cabbage leaves were dipped in a bacterial suspension of Enterobacter sp., Serratia sp. or Enterococcus sp. at OD₆₀₀ = 1.0 for 10 min, and then air dried in Petri dishes. Cabbage leaves dipped in ddH₂O and air dried as before were used as a control. There were three replicates of each treatment. Fifty first instar larvae were placed in a Petri dish containing cabbage leaves, until they reached the third instar. Control and treated cabbage leaves were renewed daily.

Quantitative PCR (qPCR) detection of gut bacteria was analyzed from randomly sampled 50 × third instar larvae. The surface of the larvae was sterilized by immersing the larvae in 75% ethanol for 90 s and then rinsing with sterilized ddH₂O for three times. The larvae were dissected to remove the gut contents, which were homogenized with 1 mL sterilized ddH₂O in a vortex mixer for 10 min to lyse the bacteria, before the DNA was extracted using a QIAamp®DNA Stool Mini Kit (QIAGEN, Gene Company Limited, China) following the manufacturer's protocol.

DNA of each of the bacteria-reared lines was analyzed using qPCR, by targeting the 16S-rRNA genes of Enterococcus, Serratia, and Enterobacteriaceae. Since the Enterobacter primers were difficult to design, we used the abundance of Enterobacteriaceae as a proxy to assess changes in abundance for Enterobacter. Primers are listed in Table S1. The general bacteria (Eub) primer set with total gut microbial DNA (Denman and Mcsweeney, 2006) was used to calibrate the relative abundance of the strains. Detailed methods for quantifying the bacteria are described in Xia et al. (2013). A Student's t-test was used to compare mean level of gut colonization by the three strains of bacteria.

Inhibition effects of antibiotics on the isolated gut bacteria were analyzed in vitro in MS medium (1 g yeast extract, 1.5 g K₂HPO₄, 0.5 g KH₂PO₄, 0.5 g (NH₄)₂SO₄, 0.5 g NaCl, 0.2 g MgSO₄, 0.05 g CaCl₂, 0.02 g FeSO₄, and 20 g agar in 1 L of ddH₂O; pH 7.0) containing 1 mg mL⁻¹ ciprofloxacin, 1 mg mL⁻¹ levofloxacin, and 2 mg mL⁻¹ metronidazole. The medium was spotted with 10 μL solution of each bacterium (Enterobacter sp. Serratia sp. and Enterococcus sp.; OD₆₀₀ = 1.0) and the bacteria were cultured at 37°C for 24 h. In order to detect the efficiency of antibiotics against gut bacteria in vivo, cabbage leaves were dipped in a solution of 1 mg mL⁻¹ ciprofloxacin, 1 mg mL⁻¹ levofloxacin, and 2 mg mL⁻¹ metronidazole in ddH₂O for 10 min, air dried, and then put into Petri dishes. Control cabbage leaves were dipped in ddH₂O. Next, 50 × first instar larvae were placed in a Petri dish and reared on the cabbage leaves until they reached the third instar. The cabbage leaves were renewed daily and there were three replicates of the treatments. The surface of the third instar larvae was sterilized by immersing in 75% ethanol for 90 s, before rinsing with sterilized ddH₂O. The larvae were dissected to remove the gut contents and these were homogenized in 1 mL sterilized ddH₂O. Next, 10 μL of the gut content suspension was used to culture bacteria in MS medium at 37°C for 24 h to assess the effect of the antibiotics on P. xylostella gut bacteria.

Vitamin C is an antioxidant that affects immune response (Hardie et al., 1991; Molina-Cruz et al., 2008; El-Gendy et al., 2010) and acetylsalicylic acid is believed to trigger eicosanoids (Claria and Serhan, 1995) that are an important component of immunity (Stanley and Miller, 2006). Previous studies have revealed that vitamin C repairs damage in Drosophila melanogaster caused by acephate (Rajak et al., 2017) and acetylsalicylic acid increases larval survival in L. dispar treated with Bt toxins (Broderick et al., 2010), so these two compounds that innately modulate immunity were selected to analyze their effect on the development of insecticide resistance in P. xylostella. Cabbage leaves were dipped in a suspension of vitamin C (5 mg ml⁻¹), acetylsalicylic acid (10 mg ml⁻¹) or ddH₂O (as a control) for 10 min and air dried, before being added to Petri dishes, along with10 × first instar larvae. The leaves were renewed every day until the larvae had reached the third instar. There were three replicates of each treatment.

Cabbage leaves were dipped in a solution of chlorpyrifos (50 g L⁻¹) or in ddH₂O (control) for 10 min, air dried and then put into Petri dishes, to which 10 × third instar larvae, which had been reared under the different bacteria, antibiotics, vitamin C and acetylsalicylic acid conditions, were added, after having been starved for 6 h. Survival was assessed at 24 h and 36 h, and there were three replicates of each treatment.

In order to study the effect of heat-killed bacteria on insecticide resistance, bacteria of the three strains (OD₆₀₀ = 1.0) were heat-killed at 70°C for 15 min, and cabbage leaves were dipped in the suspensions for 10 min. Then, 10 × first instar larvae were added to Petri dishes, and reared on the cabbage leaves until the third instar. The third instar larvae were collected as for bioassay, as detailed above, and survival was assessed at 24 and 36 h; there were three replicates of each treatment. One-way ANOVA followed by an LSD post-hoc test was used to compare treatment means, using SPSS v. 23.

Batches of 10 × third instar larvae were randomly selected and starved for 6 h, before being transferred to Petri dishes and reared, for 24 h, on cabbage leaves treated with the different bacteria, antibiotic, vitamin C and acetylsalicylic acid conditions, as detailed above. Whole body total RNA of the larvae was extracted using qPCR following the methods described by Xia et al. (2015).

The antibiotic treatment containing ciprofloxacin, levofloxacin, and metronidazole reduced bacterial growth in vitro (Figure S1) and similarly, gut bacteria collected from P. xylostella were inhibited by the antibiotics (Figure S2). We found that gut contents collected from P. xylostella that had been reared on a diet containing antibiotics contained no bacteria when cultured on antibiotic-free medium (Figure S3). When we examined the introduction of bacteria into the gut of P. xylostella, we found that all three strains of bacteria increased in abundance compared with the control (Enterococcus sp.: t = 3.43, P = 0.003; Enterobacter sp.: t = 4.00, P = 0.003; and, Serratia sp.: t = 8.24, P < 0.001; Figure 1).

The insecticide resistance bioassay indicated that Enterococcus sp. significantly enhanced insecticide resistance in P. xylostella after 24 h [survival rate: 86.7 ± 15.3% compared to the 63.3 ± 15.3% in the control, F_{(4, 10)} = 8.02, P < 0.05] and 36 h [survival rate: 46.7 ± 5.8% compared to the 23.3 ± 5.8% in the control, F_{(4, 10)} = 8.71, P < 0.05], but Serratia sp. decreased insecticide resistance at 24 h [survival rate: 40.0% ± 10.0% compared to the 63.3 ± 15.3% in the control, P < 0.05], while Enterobacter sp. had no effect [survival rate: 80.0 ± 10.0% compared to the 63.3 ± 15.3% in the control, P = 0.07; Figure 2A]. Moreover, antibiotics enhanced the insecticide resistance significantly at 36 h [survival rate: 36.7 ± 5.8% compared to the 23.3 ± 5.8% in the control, P < 0.05; Figure 2A]. Heat-killed bacteria (Figure S4) had no effect on P. xylostella insecticide resistance, irrespective of strain, at 24 h [F_{(3, 8)} = 1.58, P = 0.268] or 36 h [F_{(3, 8)} = 0.90, P = 0.487] (Figure 2B).

The survival rate of P. xylostella was higher for larvae reared on vitamin C treated cabbage leaves (80.0 ± 10.0%) at 24 h than the control (60.0 ± 10.0%) [F_{(2, 6)} = 4.43, P < 0.05], and the survival rate of P. xylostella reared on cabbage leaves treated with acetylsalicylic acid (70.0 ± 10.0%) at 36 h was higher than the control (33.3 ± 5.8%) [F_{(2, 6)} = 22.20, P < 0.01; Figure 2C].

All three strains of bacteria degraded chlorpyrifos in vitro, with mean degradation efficiencies of 34.1 ± 4.6%, 36.7 ± 4.6%, and 33.0 ± 6.0% for Enterococcus sp., Enterobacter sp., and Serratia sp. respectively, but this difference was not significant overall [F_{(2, 6)} = 0.36, P = 0.711; Figure 3].

We found that Enterococcus sp. induced the expression of gloverin [F_{(4, 10)} = 225.24, P < 0.01] and lysozyme [F_{(4, 10)} = 113.50, P < 0.01], but down-regulated expression of cecropin [F_{(4, 10)} = 560.12, P < 0.05] and had no effect on moricin expression [F_{(4, 10)} = 41.88, P = 0.377]. Enterobacter sp. induced cecropin [P < 0.01], moricin [P < 0.01], and lysozyme [P < 0.01] expression, but not that of gloverin [P = 0.184]. Serratia sp. induced moricin expression [P < 0.01] (Figure 4).

We studied the effect of heat-killed bacteria on expression of the P. xylostella immune genes, and found a different expression profile compared to the live cells. The heat-killed Enterococcus sp. and Serratia sp. induced the antimicrobial peptides (AMPs) of cecropin [F_{(3, 8)} = 680.05, P < 0.01], moricin [F_{(3, 8)} = 49.88, P < 0.01], gloverin [F_{(3, 8)} = 256.42, P < 0.01], and lysozyme [F_{(3, 8)} = 1005.25, P < 0.01]. Heat-killed Enterobacter sp. only induced lysozyme [P < 0.01] (Figure 5).

Antibiotics were found to induce expression of moricin [P < 0.01] and lysozyme [P < 0.05], but down-regulate cecropin expression [P < 0.01] with no effect on gloverin expression [P = 0.573] (Figure 4). Vitamin C up-regulated the expression of gloverin [F_{(2, 6)} = 43.22, P < 0.01] and lysozyme [F_{(2, 6)} = 165.97, P < 0.01], but down-regulated cecropin [F_{(2, 6)} = 55.69, P < 0.01] and moricin [F_{(2, 6)} = 129.41, P < 0.01]. Finally, we found that acetylsalicylic acid up-regulated moricin [P < 0.01] and gloverin [P < 0.01], but down-regulated cecropin [P < 0.01] (Figure 6).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.