Chemical analysis
Pirimicarb (99.9% purity) and diazepam (98.0% purity) were purchased from Fujifilm Wako Pure Chemical Corporation (Osaka, Japan). To determine the concentrations of the chemicals in the water samples, each sample was diluted with acetonitrile: pure water (60:40, v/v). The chemicals were quantified by liquid chromatography-mass spectrometry (LC-MS/MS) at a flow rate of 0.3 mL/min using a column at 40 °C (LC: 1290 HPLC; MS/MS: 6460 Triple Quad LC-MS/MS, Agilent Technologies, CA, USA; column: Acquity UPLC HSS T3, 1.8 μm × 2.1 mm × 100 mm, Waters, MA, USA). The operating conditions and mobile phase were in accordance with the procedure outlined by Ishimota et al. (2020a). The monitored precursor/product ions for pirimicarb and diazepam were at 239.1/72.1 and 285.0/193.1 m/z, respectively. The peak area response for each chemical was determined using a data processor and the calibration curve was constructed by plotting the amount against the peak area of the calibration standard using the least-squares method.
In our previous study, the recovery rates for pirimicarb at 0.5 µg/L and 160 mg/L were 95% to 104% and the relative standard deviations (RSDs) were 1% to 2% in the same test medium and the chemical was very stable under these test conditions for 48 h (Ishimota et al., 2020a). Thus, the analytical accuracy, variation, and stability under the test conditions were checked only for diazepam in the present study. The limit of quantification was 0.5 μg/L for each chemical. The recovery rates and RSDs at 0.5 µg/L and 20 mg/L for diazepam were calculated from three spiked water samples. The media containing diazepam (20 mg/L) were prepared in triplicate. After a 48-h period under test conditions, the diazepam concentration was determined. The diazepam was found to be extremely stable in the water samples during the 48-h exposure period. Thus, the concentrations for each test group were only measured at the beginning of exposure in the present study. The RSDs of the diazepam concentrations in all samples were calculated as per Eqn. (1):

where, SD and mean denote the standard deviation and mean of the diazepam concentrations (n = 3), respectively.
Test organisms and culture methods
C. yoshimatsui (NIES clone) was obtained from the National Institute for Environmental Studies (Ibaraki, Japan). The test species had been continuously maintained in ISO medium with a slight modification (CaCl2・2H2O: 36.9 mg/L, MgSO4・7H2O: 6.1 mg/L, NaHCO3: 64.8 mg/L, KCl: 8.60 mg/L, thiamine hydrochloride: 75.0 µg/L, cyano-cobalamin: 1.00 µg/L, and biotin: 0.75 µg/L) (Ishimota et al., 2020a) and were fed TetraFin (Tetra, Melle, Germany) at least five times per week at a rate of 0.1 g per container. The test species were cultured in 5 L stainless-steel containers (length: 40 cm, width: 27 cm, and height: 7 cm) with a net containing 1.5 L of the medium and 1 kg of glass beads. The culture medium was changed at least once per week. The test species were maintained under standardized conditions, i.e., a photoperiod of 16:8 h light: dark and a light intensity of approximately 800 lux at 23 ± 1 °C. The egg masses that were produced were collected daily and were used to conduct all the experiments and establish the next generation in the multi-generational study.
Multi-generational experiment
In our previous multi-generational study of tolerance to chemicals in chironomids, we used < 24-h-old larvae; however, we used the > 24-h and < 48-h-old (hereafter, < 48-h-old) larvae for all experiments in the present study because the size of this aged larvae was large enough to detect each of the mRNA levels, whereas that of the < 24-h-old larvae was not adequate (Ishimota et al., 2020a). Chironomid larvae (< 48-h-old) were exposed to low effective concentrations of chemicals and were cultured until an appropriate number of egg masses was obtained for each generation (five successive generations: F0–F4) (Fig. 1). Changes in the larval sensitivity were investigated based on the 48-h 50% effective concentration (EC50) value across the five generations. Additionally, we measured CAT, CYP450 and Hb mRNA levels in the larvae exposed to the target chemicals in each generation.
Nominal concentrations (0, 5, 10, 20, 40, and 80 mg/L for pirimicarb and 0, 1.3, 2.5, 5, 10, and 20 mg/L for diazepam) were prepared based on the 48-h EC50 value for each chemical in the preliminary tests. To prevent the chironomids from becoming trapped on the surface of the solution, each test solution contained 2 µL/L Tween 80. After the larvae were fed, five larvae (< 48-h-old) were exposed to 50 mL test solutions in 100 mL glass beakers for 48 h without any food under standardized conditions. We prepared four sets of glass beakers and 20 individuals from each treatment were exposed. After 48-h exposure, the number of larvae that could not change their position during 15 s of gentle agitation of the beaker (immobility) was counted for each concentration and the EC50 values were calculated from the immobility results for the F0 samples. The 48-h EC50 values for pirimicarb and diazepam were determined according to the acute toxicity test of the Organization for Economic Co-operation and Development (OECD) guideline “No. 235 Chironomus sp., acute immobilization test” (OECD, 2011). To check the 48-h EC50 value in each generation (F1, F2, F3, and F4 samples), the larvae were exposed to the same method as described for the F0 generation. This experiment for each generation was performed in triplicate using the larvae collected from three egg masses.
To establish the next generation group (control and treatment groups), chironomids were pre-exposed to each chemical as follows. A total of 400 larvae (< 48-h-old) were exposed to the 500 mL solutions (control group without any exposure and treatment groups containing each chemical, i.e., 0 and 15 mg/L of pirimicarb and 0 and 5 mg/L of diazepam in 500 mL glass beakers without food under standardized conditions). After 48-h exposure, 300 chironomids that had survived were randomly selected from each group and cultured separately without chemicals until enough egg masses for each experiment were obtained. In a previous study, the chironomids developed tolerance to pirimicarb, which caused a decrease in mobility but no mortality (Ishimota et al., 2020a). Thus, the nominal concentrations for each chemical were set at similar levels as those in the previous study for pirimicarb. Controls (0 µg/L) for each generation in culture medium containing 2 µL/L Tween 80 for 48 h were extracted using the same method described previously. The procedure for culturing the chironomids for each generation was the same as that described in section 2.2.
CAT, CYP450, Hemoglobin and mRNA levels
We tested how CAT, CYP450 and Hb mRNA levels changed in chironomids exposed to three different concentrations of pirimicarb and diazepam. After the seventy larvae (<48-h-old) were fed, they were exposed to 500 mL of the test solutions in 500 mL glass beakers for 48 h without any food under standardized conditions. We prepared three sets of glass beakers, and 70 individuals from each treatment (i.e., pirimicarb: 0, 15, and 40 mg/L; diazepam: 0, 2.5, and 5.0 mg/L, with each solution containing 2 µL/L Tween 80) were exposed to each test solution without food, under standardized conditions for 48 h. We measured the CAT, CYP450 and Hb mRNA levels in the exposed chironomids in each concentration group. Additionally, the CAT, CYP450 and Hb mRNA levels in the chironomids from the control and treatment groups were measured in the multi-generational study. After the chironomid larvae obtained from the control and treatment groups in each generation were fed, they were exposed to 500 mL of each chemical solution (15 mg/L for pirimicarb and 5 mg/L for diazepam containing 2 µL/L Tween 80) without food under standardized conditions for 48 h. The concentrations were the same as those used for the pulsed exposure method in the multi-generational study. These experiments were performed in triplicate.
Total RNA was extracted from the 70 chironomids using a NucleoSpin® XS RNA kit (Takara BIO Inc., Shiga, Japan) according to the manufacturer’s instructions. RNA samples were reverse-transcribed into complementary DNA using a PrimeScript™ RT reagent kit with gDNA Eraser (Takara BIO Inc.) following the manufacturer’s instructions. A TB Green® Premix EX Taq™II kit (Takara BIO Inc.) was used to detect the amplified products. The CAT, CYP450 and Hb genes were amplified using quantitative polymerase chain reaction (Real Time PCR System Quant Studio TM 12k Flex, Applied Biosystems, Thermo Fisher Scientific Inc. MA, USA) with the following primers. For CAT (mRNA; GenBank accession No. JL641904.1), the forward primer 5′-CGTGATCTTCGTGGTTTTGCTG-3′ and reverse primer 5′- GGATTGGATCGCGGATGAAG -3′ were used (Nair et al., 2011). For CYP450 (mRNA; GenBank accession No. FJ541450.1), the forward primer 5′-GACATTGATGAGAATGATGTTGGT-3′ and reverse primer 5′-TAAGTGGAACTGGTGGGTACA-3′ were used, whereas for the Hb gene (Hb IIB gene, GenBank accession No. AJ003807.1), the forward primer 5′-ATTCGCTGGAAAGGATGTTG-3′ and reverse primer 5′-TATGAGACGAGTGAGGCACG-3′ were used (Ha and Choi, 2008; Lencioni et al., 2016; Park et al., 2009). Actin (mRNA; GenBank accession No. AB070370.1) was used as the reference gene, with the forward primer 5′-GATGAAGATCCTCACCGAAC-3′ and the reverse primer 5′-CCTTACGGATATCAACGTCG-3′ (Lencioni et al., 2016). Amplification efficiencies of CAT, CYP450, Hb, and actin in chironomids were 100%, 100%, 103%, and 102%, respectively. The expression data from the triplicate experiments were expressed relative to those of actin to normalize any difference in the reverse transcriptase efficiency and they were analyzed using the standard curve method. In the control and treatment groups for the multi-generational study, the ratio of mRNA levels in the treatment (15 mg/L for pirimicarb and 5 mg/L for diazepam) against each control (0 mg/L) in each generation were used as an index.
To confirm whether the CAT, CYP450 and Hb mRNA levels were related to the sensitivity, we investigated the relationship between sensitivity and mRNA levels using the generalized linear model (GLM) approach.
Statistical analysis
The data were analyzed using R version 3.6.1 (R Development Core Team, 2019). The EC50 values and standard errors were estimated by fitting acute toxicity data to a two-parameter log-logistic model, where the lower limit was fixed at 0 and the upper limit was fixed at 1 (Eqn. 2):

where, f(x) is the probability of immobilization, x is the nominal concentration of pirimicarb, e is the inflection point of the fitted line (which is equivalent to the concentration required to cause a 50% response (EC50)), and b is the relative slope around e. The triplicate testing of each test group was considered a random effect in the medrc package (Gerhard and Ritz, 2017). Significant differences between the EC50 values for the control and treatment groups from F0 to F4 were determined using the EDcomp function in the drc package (Ishimota et al., 2020b; Ritz et al., 2006). In this function, the ratios of the EC50 values were compared to 1. Then, the significance level in each pair was adjusted using Holm’s method (Holm, 1979).
For the CAT, CYP450 and Hb mRNA levels in the chironomids exposed to the four concentrations of each chemical, the variance of the groups was confirmed to be heterogeneous between the control and treatment groups (Levene’s test) following Dunnett’s test using the multcomp package (Hothorn et al., 2008). Significant differences between the EC50 values for the control and treatment groups from F0 to F4 were determined using the EDcomp function in the drc package (Ishimota et al., 2020b; Ritz et al., 2006). In this function, the ratios of the EC50 values were compared to 1. Then, the significance level in each pair was adjusted using Holm’s method (Holm, 1979). CAT, CYP450 and Hb mRNA levels in the samples from the multi-generational study were compared with a pairwise t-test using the R software program. For the multi-generational study, GLM analysis was used to estimate the effects of CAT, CYP450 and Hb mRNA levels and generational alternation (F0–F4) on the EC50 values. The EC50 values were assumed to be gamma (log-link) distributed. The selection of the best performance model (model with the lowest Akaike information criterion [AIC]) was based on AIC (Akaike, 1974). Full models included all variables (CAT, CYP450, and Hb mRNA levels, generational alteration (GA), the interaction between the GA and each mRNA level. We tested the significance of the variables using the χ2 test (P < 0.05) based on the Wald statistic (Hosmer and Lemeshow, 2000; Marques et al., 2008).