All chemicals used were purchased from Sigma-Aldrich (Castle Hill, NSW, Australia), unless otherwise specified. Culture media were purchased from Vitrolife (Göteborg, Sweden).

All mice were kept in the School of BioSciences animal housing facility at the University of Melbourne, maintained under a 12 h light: 12 h dark lighting regimen and fed ad libitum a soy-free diet (Specialty Feeds, Perth, WA, Australia) to avoid the effects of phytoestrogens. Six-week old female C57BL/6J (Melbourne Bioresources Platform) mice were mated overnight with 12-week-old C57BL/6J males. Mating was determined by presence of a copulatory plug at day 0.5 of pregnancy. Pregnant females were then individually housed and randomly allocated to one of two treatment groups: Control (CON, n = 5) and Atrazine (ATZ, n = 5). Atrazine was initially dissolved in DMSO, prior to being added in distilled drinking water (< 0.5% DMSO v/v). Pregnant females (n = 5 per group) were administered from day 9.5 of gestation either a vehicle treatment (< 0.5% DMSO, CON) or atrazine treatment (5 mg/kg bw/day, atrazine), based on our previous studies (33) and standard water consumption rates. The selected dose is 10-fold higher than the NOEL used to calculate the acceptable daily intake (ADI) for drinking water in Australia (37). Oral administration of this dose is known to expose mouse foetuses to a similar in utero concentration to that measured in maternal plasma (39). Pups were weaned at four weeks of age, as per standard animal husbandry practices, with only male pups utilised for this study. These males were housed individually with continued exposure to the treatment water (CON or ATZ). Within each treatment littermates were split up and maintained until either i) 12 weeks of age (n = 15) or ii) 26 weeks of age (n = 20). A soy-free diet and water were provided ad libitum and their consumption was monitored for six males from each treatment group for five weeks. Body weights were recorded weekly. All experiments were approved by the University of Melbourne Animal and Ethics committee (AEC 1513481.5) in accordance with the Australian National Health and Medical Research Council guidelines (40).

At 11 and 25 weeks of age males (n > 5, at least one from each litter) from each treatment group, and age cohort, were randomly selected for mating. Three-week old C57BL/6J female mice (n > 20 per treatment per age cohort) were superovulated using a standard protocol (41), with an intraperitoneal injection of 0.25 IU/g pregnant mare serum gonadotrophin (PMSG; Folligon, Intervet, Bendigo, Australia) followed 48 h later by 0.25 IU/g human chorionic gonadotrophin (hCG; Chorulon, Intervet). Individual males and females were housed overnight for mating and 2 h post-hCG injection, pronucleate oocytes (2PN) were recovered from female tracts and placed in G-MOPS handling medium supplemented with 5 mg/ml human serum albumin (GMOPS+; Vitrolife, Göteborg, Sweden). Pronucleate oocytes were denuded of cumulus cells via incubation in GMOPS containing 300 IU/ml hyaluronidase for 20 s (bovine testes, type IV; Sigma-Aldrich) followed by washing in GMOPS+. Denuded pronucleate oocytes were immediately washed in GMOPS+ and transferred in groups of 10 to 20 µl drops of G1 media (Vitrolife) under 3.5 ml paraffin oil (Ovoil) for 72 h at 6% CO₂, 5% O₂ and 89% N₂ at 37°C as previously detailed (42). Embryos were then assessed for development and transferred to pre-equilibrated G2 media (Vitrolife) for a further 48 h. After a total 96 h culture, pre-implantation embryos were assessed for development stage and subjected to differential staining to identify the allocation of cells to the inner cell mass (ICM) and the trophectoderm (TE) of blastocysts, as described previously (41). Briefly, blastocysts were placed in 0.5% v/v pronase (Sigma-Aldrich) until the zona pellucida disbanded, followed by washing in GMOPS+ for 5 min. Embryos were then incubated in 10 mM 2,4,6-trinitrobenzene sulfonic acid (Sigma-Aldrich) for 10 min then washed in GMOPS+ for 5 min, before a 10 min incubation in 0.1 mg/ml anti-dinitrophenol (Sigma-Aldrich). Blastocysts were subsequently washed for 5 min in GMOPS+, then incubated in 10% v/v guinea pig serum with 25 mg/ml propidium iodide (IMVS, Adelaide, Australia) for 5 min. Blastocysts were transferred to 0.1 mg/ml bisbenzimide (Hoechst, 33342) in 10% v/v ethanol for 15 min, washed in GMOPS+ and finally mounted in glycerol on glass slides under coverslips (Thermo Fisher, Scoresby, VIC, Australia). Cells were visualised and photographed using a fluorescent microscope (Nikon Eclipse TS100) equipped with a Nikon Digital Sight DS-L2 camera (Nikon, Tokyo, Japan). Cell numbers were counted using ImageJ Version 1.47 (43), by a technician blinded to treatment groups. Five biological embryo culture replicates per age cohort were undertaken.

Mice were killed via cervical dislocation at i) 12 weeks of age (n = 15 males from five litters) or ii) 26 weeks of age (n = 20 males from five litters). At post-mortem body weight was recorded, the epididymides were immediately dissected out and transferred to a tissue culture dish (Thermo Fisher) containing G-MOPS+ at 37°C. An epidydimal sperm count and live:dead sperm stain was performed, as described below. The liver, testes, seminal vesicles, perigonadal fat, and retroperitoneal fat were removed and weighed. Samples of liver and testis were snap frozen in liquid nitrogen and stored at -80°C for gene expression analysis. Additional samples of liver and testis were fixed in 4% (v/v in 0.9% saline) paraformaldehyde (PFA) for histological analysis.

Sections of liver and testis fixed in 4% paraformaldehyde (v/v) were washed in 0.9% saline and stored in 70% ethanol at 4°C (44). These tissues were then embedded in paraffin wax, sectioned at 7 µm thickness and stained with Harris Haemotoxylin & Eosin Y (H & E) following standard protocols and as previously described (45). Sections were visualised using an Olympus BX51 Microscope (Olympus, Tokyo, Japan) and images were captured using an Olympus DP70 Camera (Olympus). Liver and testis sections were assessed at x 20 magnification for the presence of gross abnormalities by a technician blinded to treatment groups. For each liver sample five random sections were imaged at x 40 magnification, and scored for macrovesicular and microvesicular steatosis, where macrovesicular steatosis is defined by the vacuoles displacing the nucleus to the side, while microvesicular steatosis do not, as previously described (46). The median steatosis scores from the five sections of each animal were used to determine the median steatosis score, which was used to calculate the median for each litter, group, and age cohort.

Epididymides were transferred to an organ-well culture dish containing 500 μl of GMOPS+ culture medium at 37°C. The epididymides were punctured ten times with a 23-gauge needle (Becton Dickinson, Scoresby, VIC Australia), placed on a heated stage at 37°C and covered from light for 10 min to allow sperm to swim out (47). For each sample, a 40 µl aliquot of the sperm solution was diluted 1:10 in water (v/v) and 10 μl of this solution was pipetted onto each chamber of a haemocytometer (Neubauer, Wertheim, Germany). Sperm were counted on a Nikon Eclipse TS100 (Nikon, Tokyo, Japan) at 200 x magnification, according to previously defined methods (48). Duplicate counts were undertaken and averaged per sample prior to calculating sperm concentration by a technician blinded to treatment groups.

Live/dead sperm staining was performed based on previously described methods (49). Briefly, 10 µl of propidium iodide (0.5 mg/ml) was diluted 1:10 (v/v) with GMOPS+ culture medium and 5 µl of this solution, combined with one drop of H33342 stain (NucBlue; Thermo Fisher, Scoresby, VIC, Australia) was added to a 40 µl aliquot of sperm solution and incubated for 5 min at 37°C. Subsequently, the reaction was stopped using 5 µl of 0.1% (v/v) neutral buffered formalin (NBF) and 2 µl of bovine serum albumin (BSA; 100 mg/ml, MP Biomedical, NSW, Australia) was added to prevent sperm agglutination. The solution was centrifuged at 500 g for 5 min at room temperature and the supernatant discarded, the solution was then resuspended in 20 µl of GMOPS+ culture medium. The stained sperm solution (10 µl) was mounted on a superfrost microscope slide (Platinum Pro, Thermo Fisher) and visualised under a fluorescent microscope (Nikon Eclipse TS100) with the appropriate filters. A minimum of 10 images per sample were captured randomly using a Nikon digital sight camera. Captured images were later processed, and counted using Image J (43). Sperm nuclei stained with propidium iodide were counted as “dead”. A minimum of 200 spermatozoa were counted per slide (48). The percentage of live sperm was then calculated from these data.

Daily sperm production (DSP) was determined using a partial portion of testis, as previously described (50). Briefly, a portion of snap-frozen testis from each male was weighed and sonicated for 30 s at 22.5 kHz using a VirSonic ultra cell disruptor 100 (VirTis, Gardiner, NY, USA) in 500 µl of DSP saline buffer that contained 0.9% NaCl and 0.05% Trition-x-100 (v/v). The solution was stained with 10 µl of 0.4% trypan blue (v/v saline buffer) and vortexed for 20 s at room temperature to mix. A 20 µl aliquot of the sperm solution was diluted 1:1 with H₂O to facilitate counting accuracy. A 10 μl volume of this solution was pipetted onto each side of a haemocytometer (Neubauer). Spermatids were counted by a technician blinded to treatment groups using light microscopy on a Nikon Eclipse TS100 (x 200) according to previously defined methods (50). Duplicate counts were undertaken and averaged for each sample. To obtain the daily sperm production, the total number of spermatids was divided by 4.84, which corresponds to the time that developing spermatids spend in steps 14 to 16 during spermatogenesis in the mouse (51) and corrected based on the weight of the original tissue portion used.

A portion of testis for each male (n = 4 from different litters per group per age cohort) were homogenised and total RNA extracted using GenElute™ Mammalian total RNA miniprep kit (Sigma-Aldrich), following the manufacturer’s instructions. A section of liver from each male (n = 4 from different litters per group per age cohort) was homogenised and extracted using TRIzol reagent (Thermo Fisher) according to the manufacturer’s instruction. Extracted RNA samples were then DNase treated using Ambion TURBO DNA-free (Invitrogen, Scoresby, VIC, Australia) following the manufacturer’s specifications. Total RNA quantity and purity was assessed using the NanoDrop One/One^C UV-Vis Spectrophotometer (Thermo Fisher). A minimum inclusion criterion of 1.8 for the 260:280 ratio was used as a measure of quality and for samples to be used. The cDNA was synthesised from total RNA (400 ng/ml liver or 2 ng/ml testis) using the Superscript^®III First-strand synthesis system for RT-PCR (Invitrogen) per the manufacturer’s instruction and using random hexamer primers.

Quantitative real time RT-PCR was performed in triplicate 10 µl reaction volumes on 96 well reaction plates using SensiFAST™ SYBR Lo-ROX (Bioline, Eveleigh, Australia), as previously described (41). Both template free and minus RT triplicates were included on the plate to act as controls for genomic contamination. Reactions were run using the Viia™7 thermocycler (Applied BioSystems, Mulgrave, Australia). The genes investigated were for the liver samples: Ldlr, Acacα, Slc27a5, Pparα and Cpt1a, for the testis samples Nr5a1, Hsd17β11, Cyp19a1, and Srd5α1, with Tbp and β-actin as reference genes (see Table 1 for a full list of gene names and primer sequences). Genes of interest were selected primarily from previously identified targets for atrazine action in both tissues, including the expression of fatty acid and cholesterol pathway members in the liver (26, 28), and steroidogenic pathway enzymes and transcription factors in the testis (20). Primers were selected from previous publications or designed using Primer Express (Thermo Fisher) and NCBI primer-BLAST (52) and were synthesised by IDT (Singapore). Primer specificity and efficiency were calculated prior to use. Cycling conditions were: 50°C for 5 min, 95°C for 10 min, then amplified for 40 cycles, where each cycle included denaturation for 15 s at 94°C, annealing for 30 s at 60°C and extension for 30 s at 72°C, followed by a final extension at 72°C for 5 min. Gene expression was quantified using Pfaffl’s (53), with Tbp as a reference gene and expressed as fold changes. Tbp was selected as the reference gene, due to consistent expression levels in treatment, individuals and cohorts.

All data were tested for normality and homogeneity using boxplots, residual plots, and the Shapiro-Wilks test. Data subjected to repeated measures analysis were tested for sphericity using the Mauchly’s sphericity test. Embryo compaction and blastocyst rates were first arc-sin transformed prior to undertaking analyses, as common undertaken for embryo development studies (54, 55). For each age cohort, body weight, cumulative weight gain, as well as food and water intake data were analysed separately and combined for each age cohort using a repeated measures ANOVA, with treatment as a fixed factor and litter as a covariate, using SAS/STAT^® version 9.2 (SAS Institute). All other statistical analyses were undertaken using Rstudio version 1.0.143. For each age cohort, birth weight, relative tissue weight, sperm concentration, blastocyst cell counts (ICM, TE, total, ICM:TE ratio and %ICM) and gene expression were analysed using a one-way ANOVA, with data analysed by litter where appropriate. Non-normally distributed data (survival to weaning, % live sperm, DSP and liver histology assessment) were analysed by litter using a non-parametric Mann-Whitney test. Sex ratio was compared with an expected 50:50 ratio as well as between groups by a corrected χ² procedure and was double-checked by binominal analysis. Results were considered significant when P ≤ 0.05 and all data are expressed as the mean ± SEM, unless otherwise stated.

There was no effect of atrazine treatment on the sex ratio of pups when compared with controls (proportion male; CON 0.64 (n = 18 male, n = 10 female), ATZ 0.50 (n = 18 male, n = 18 female; P > 0.1). Also, no effect of atrazine treatment compared with controls on average pup birth weight (CON 3.8 ± 0.4 g, n = 28; ATZ 4.3 ± 0.2 g, n =36; P > 0.1) or survival to weaning (CON n = 27/28 (96.4%); ATZ n = 34/36 (94.4%; P > 0.1).

From weaning until the end of the study, no mortality post-weaning was observed among the mice. Food and water intake were monitored for five weeks and no differences were observed between the control and treatment groups (P > 0.1). Average water intake for control and atrazine treatment was 25.9 ± 0.6 ml/week and 25.7 ± 0.8 ml/week, respectively, while average food intake was 27.6 ± 0.9 g/week and 27.9 ± 0.9 g/week, respectively.

Atrazine (5 mg/kg bw/day) exposure from prenatal E9.5 did not cause a significant increase in cumulative body weight in the 12-week ( Figure 1A , n = 15 males from five litters, P > 0.1) or 26-week age cohorts ( Figure 1B , n = 20 males from five litters, P > 0.1). This result was evident, irrespective of whether actual of cumulative body weight were analysed by age cohort or with 12-week and 26-week cohort data combined. In addition, there were no differences observed in the absolute or relative weights of the seminal vesicles, testes, retroperitoneal fat pads or perigonadal fat pads between the atrazine treated and the control groups in either age cohort ( Figure 2 , P > 0.1). However, atrazine exposure caused a significant reduction in mean absolute liver weight (CON 1.2 ± 0.06 g, ATZ 0.92 ± 0.05 g) and relative liver weight ( Figure 2A , P = 0.01) compared to the control in the 12-week cohort, although this difference was not observed in the 26-week cohort ( Figure 2B , P > 0.1). Visual appraisal of liver and testis sections stained with H & E revealed no changes in gross morphology in the 12- or 26-week cohorts between treatments ( Figure 3 ). A more detailed study of the liver identified there was also no difference between treatments in the median score of macrovesicular or microvesicular steatosis in the livers of 12-week or 26-week cohorts ( Table 2 ).

We observed a significant decrease in epididymal sperm concentration of the atrazine exposed males in the 12-week cohort ( Table 3 , P = 0.02). However, no differences in epididymal sperm concentration were identified in the 26-week cohort ( Table 3 , P > 0.1). Similarly, exposure to atrazine had no effect on daily sperm production (DSP) in the testis or the percentage of live sperm in the epididymis at both 12 and 26 weeks of age ( Table 3 , P > 0.1).

The expression of genes related to lipid homeostasis was analysed in the liver. Atrazine exposure from E9.5 to 12 weeks of age caused an increase in the expression of several genes involved in lipid uptake into the liver. Specifically, Slc27a5, which encodes fatty acid transporter protein 5, was increased in the 12-week atrazine group compared with the control ( Figure 4A , P = 0.02). In addition, the expression of Ldlr, which encodes a protein responsible for the transport of low-density lipoproteins into the liver, was significantly increased ( Figure 4A , P = 0.02). The expression of the transcriptional factor Pparα which is a regulator of fatty acid metabolism, in particular Slc27a5, was also significantly increased (P = 0.05) in the atrazine treated group compared to the control group ( Figure 4A ). No change in the expression of Acacα or Cpt1a was determined ( Figure 4A , P > 0.1). In the 26-week age cohort, a significant decrease was identified in the expression of Pparα ( Figure 4B , P = 0.03), while no differences were observed in the expression of any of the other genes studied ( Figure 4B , P > 0.1).

Exposure to atrazine from E9.5 to 12 weeks of age caused a change in the expression of steroidogenic genes in the testis. There was a significant increase in the expression of Cyp19a1 ( Figure 5A , P = 0.05), the gene that encodes for the aromatase, which is responsible for the conversion of androgens to oestrogens. Conversely, there was a decrease in the expression of Srd5α1 ( Figure 5A , P = 0.008), the gene that encodes for 5-α reductase, which is responsible for the conversion of testosterone to the more potent androgen DHT. The expression levels of Nr5a1 and Hsd17β11 were not found to be affected by the exposure to atrazine ( Figure 5A , P > 0.1). When the exposure period to atrazine was increased to 26 weeks of age, no significant differences were found between any of these genes in the testis ( Figure 5B , P < 0.1).

Preimplantation embryos derived from sperm of males exposed to atrazine had no effect on key embryo development parameters, with compaction and blastocysts rates unaffected in both age cohorts ( Table 4 , P > 0.1). Despite this, paternal atrazine exposure affected the number of cells within the preimplantation embryo. Sperm from the 12-week cohort of atrazine exposed males generated embryos with fewer ICM (P = 0.02), TE (P = 0.05) and total cells (P = 0.01) than the embryos generated using sperm from unexposed males ( Figure 6A ). However, in the 26-week cohort, there was no difference in cell counts of embryos generated from atrazine exposed fathers compared to the unexposed fathers ( Figure 6B ). In addition, no differences were observed between treatments in the ICM:TE ratio, or %ICM in either age cohort (P > 0.1, Table 4 ).

This study showed that atrazine exposure, starting prenatally, caused negative metabolic and reproductive effects in adult male mice, which differed in severity depending on the duration of atrazine exposure. These effects were relatively subtle, as would be expected, based on conflicting evidence reported from human studies (15) and animals in the natural environment (3, 17). This does not detract from the importance of identifying them, with additional endpoint measures, such as determining the circulating or local steroid concentrations, would help identify possible mechanisms. The findings of this and other studies also highlight the importance of assessing the effects of atrazine at numerous life stages, as well as the effects over multiple generations. However, these measures and their transgenerational implications are generally not assessed in traditional toxicological assessments. Due to the widespread and continued use of atrazine, it is essential that the implications and health risks associated with its use are understood and that a broader range of endpoints are investigated.