Exposure to an anti-androgenic herbicide negatively impacts reproductive physiology and fertility in Xenopus tropicalis

Amphibians are threatened on a global scale and pollutants may be contributing to population declines, but how chemicals impact on their reproduction is poorly understood. We conducted a life cycle analysis to investigate the impacts of early life exposure to two anti-androgens (exposure until completion of metamorphosis;stage 66): flutamide, (50 µg/L)/linuron (9 and 45 µg/L)) on sexual development and breeding competence in Xenopus tropicalis. Our analyses included: mRNA levels of dmrt1, cyp17, amh, cyp19, foxl2 and ar (tadpoles/metamorphs), gonadal histomorphology (metamorphs/adults), mRNA levels of ar/gr (adult male brain/gonad/forelimb), testosterone/corticosterone levels (adult males), secondary sexual characteristics (forelimb width/nuptial pad: adult males) and breeding competence (amplexus/fertility: adult males). Compared to controls, feminised sex ratios and increased number of spermatogonia (adults) were observed after exposure to flutamide and the lower linuron concentration. Exposure to the lower linuron concentration also resulted in demasculinisation of secondary sexual characteristics and reduced male fertility. Flutamide exposure resulted in masculinisation of the nuptial pad and elevated mRNA levels of dmrt1, cyp17, amh and foxl2 in brains (metamorphs). Testosterone levels were higher in all treatment groups, however, overall few effects were observed in response to the higher linuron concentration. Our findings advance understanding of reproductive biology of X. tropicalis and illustrate negative effects of linuron on reproductive processes at a concentration measured in freshwater environments.

Supplementary methods S3 -Adult male histomorphology For testis histomorphology, digital photos of a section cut through the centre of the right testis were captured with a photo-micrographic camera (Leica DFC 550, Leica AB, Kista). For each section analysed, a grid was overlaid and all seminiferous tubules that contained crossing gridlines (spacing: 0.7 mm) were selected for analysis (3-23 tubules, depending on the size of the testis). For each seminiferous tubule analysed, the number of germ cell nests (cyst-like structures within the luminal margins of the seminiferous tubule) and each nest was classified according to the most mature cell type observed: spermatocytes, spermatids or spermatozoa, using established criteria 5 . In addition, the number of spermatogonia per tubule and the number of spermatocytes in the largest spermatocyte nest in each tubule (2-20 nests per tubule) were counted. The amount of spermatozoa in the lumen of the tubule was assessed and assigned a score number (spermatozoa: 1 = tubules with little/no spermatozoa, 2 = tubules with spermatozoa in half of the lumen, 3 = tubules with lumen filled with spermatozoa) 6 . Testis and tubule diameter length and width (average calculated for data analysis) and the total number of seminiferous tubules were recorded. Testicular morphology was compared between treatments using mean values for each measured endpoint across tubules within individuals. Analysis was done directly or using ImageJ software (National Institute of Health, Bethesda, MD, USA) where appropriate. All slides were analysed without knowledge of exposure group.

Supplementary methods S5 -Breeding trials
Breeding of male X. tropicalis took place 6 months post-metamorphosis. During the week prior to breeding, males were weighed and individually marked within toe webbings (Visible Implant Elastomer Tags, Northwest Marine Technology, USA). A competitive breeding system was used, whereby 2 experimental males were placed with 1 unexposed female. Unexposed female frogs were of a similar age (~1.5 years) and size (~20 grams) and were obtained from Xenopus 1 (Dexter, USA) 2 weeks prior to the beginning of the breeding trials. Two breeding trials were undertaken with each experimental male, ensuring a similar recovery time between breeding trials (minimum of 10 days 10 ). In total 158 trials were undertaken (92 trials for the first breeding, and 66 for the second breeding).
One day prior to the breeding trial, the selected males (4 or 5 sets of two males from each treatment) and females (4 or 5 individuals) were placed in holding tanks (each set of two males in 6L tanks; all female frogs in a 15L tank) in the breeding room in water at a lower temperature (24 ± 1 °C) to stimulate breeding 1 . To induce breeding, male and female frogs were given two injections of hCG: priming (20 IU) and boosting (23 hours later, 100 IU).
Immediately prior to the priming injection, male frogs were weighed, and immediately prior to the boosting injection, photographs were taken of the forelimb and nuptial pad (Nikon D70, objective AF micro Nikkon 60 mm 1:2:8D). Photographs were analysed with ImageJ (National Institute of Health, Bethesda, MD -USA) to determine forelimb width and length and Adobe Photoshop CS6 for nuptial pad size and colour (see . Photographs were randomised and all analyses were undertaken with no knowledge of treatment. Following the hCG boosting injection, each set of two males was placed in a breeding tank (20L, darkened with plastic covering, containing 3 glass petri dishes), and a pre-weighed female was introduced. To minimise confounding effects, the boosting injection of the first or second male from each pair was alternated. Male frogs were given 20-25 minutes recovery time prior to addition of the female, and the order in which females were added to breeding tanks was randomised. The breeding trial for each tank was initiated when the female was placed in the breeding tank with the male pair. After 60 minutes, and thereafter every 45 minutes, tanks were assessed to identify the individual male in amplexus and whether spawning had occurred. Six hours after the first amplexus was observed -or if the frogs were not observed in amplexus for two of the consecutive measured time points -all frogs were removed from the breeding tanks (this always occurred within 10 hours of initiation of breeding trial).
Photographs were taken of the tanks containing spawn immediately following removal of the frogs to determine the number of eggs spawned. A sub sample of the eggs spawned was then removed from each fertilisation tank and placed into glass petri dishes (2L, 26.50C in water channel). Fertilisation rate was assessed by comparing photographs taken immediately after their collection from the breeding tanks and those taken 26 hours after initiation of amplexus, when the hatched fertilised eggs could be distinguished by early embryo development (elongation of the spherical egg). See Supplemental methods S6 below for details on method optimisation.
Immediately following the second breeding, male frogs were anaesthetised and sacrificed by pithing. Their snoutvent length was measured, and weights of the body and left testis weight were measured for calculation of gonadosomatic index (GSI). The right testis was fixed in NBF (4%) for histological analysis. 1. Practicality: Development over 20 hours after amplexus had ceased (~26 hours after initiation of amplexus) allowed time to remove the fertilized eggs and get the tanks ready for the next batch of eggs for the following breeding trials (depending on amplexus behaviour, there could be as little as 30 minutes between the two). 2. Optimal timing: As embryo development starts directly at fertilization, 20 hours after amplexus ceased fertilized oocytesd should have reached at least Nieuwkoop and Faber 25 to 33 (depending on temperature) and fertilized eggs should be obvious/easy to identify.
Fertilization success was measured at 18, 19, 20 and 21 hours after amplexus ceased. In aquarium 1 ( Figure 1) and aquarium 2 ( Figure 2) 20 hours was shown to be sufficient to capture the vast majority of the fertilization that had occurred.   Figure S1. Measurement of forelimb width and length.
Before the second boosting injection, each male frog was photographed. A piece of graph paper was held beneath the forearm. The camera (Nicon D70, objective AF micro Nikkor 60 mm 1:2:8D) was held by a camera support, approximately 30 cm over a bench, facing it, and a torch was used to illuminate the arm. The forelimb length and width of the male frogs were measured in the photos using an image analysis program, ImageJ. The graph paper beneath the forelimb on the picture was used as a scale. The length was measured on the inside of the forearm, from the bend at the elbow, down to the wrist. The arm width was measured by rotating this drawn line 90°, then moving it to where the "length line" (first measured line described above) ended at the elbow end, and the width was measured at this point. Figure S2. Measurement of nuptial pad size and colour.
The same photograph was used for forelimb width and nuptial pad measurements. The size and colour intensity of the nuptial pad were analysed using Adobe Photoshop CS6. The nuptial pad was selected using the Quick Selection Tool, and the area (in number of pixels) and colour intensities were recorded. Colour intensity ranged from 0 (black) to 255 (white). Reflections on the arm from water drops have an intense white colour which interferes with the colour intensity measurements. To eliminate this artefact the reflections were removed using the Spot Healing Brush Tool. This tool allows the removal of the reflections by replacing these areas with a composite colour sampled from the skin surrounding the reflection. For each photo the area was calculated by comparing the number of pixels in the selected part with the number of pixels in a 1x2 mm selected area in the graph paper. Each photograph was analysed twice without knowledge of treatment and the mean of these measurements was used for data analysis.  Figure S5. Body weights and gonado-somatic index (GSI) in adult males. No differences between groups were observed (mean ± S.E.).   The size of the largest spermatocyte-nest in each seminiferous tubule was determined by using score numbers one to three. b The number of spermatozoa in the seminiferous tubular lumen was estimated using score numbers one to three. * Significantly different from control (*p < 0.05, ** p < 0.01), one-way ANOVA and Holm-Sidak .   Figure S10. Gene expression in tadpole brains in control organisms. Males (filled symbols) and females (open symbols) did not differ in expression of genes. ar expression (bold) increased during ontogeny in both males (black) and in females (grey). Ar and foxl2 expression was related in males (Pearsons' , p < 0.001). Expression of cyp17, ar or foxl2 (NB: dmrt1, amh and cyp19 were not tested) were not sexually dimorphic. Expression of foxl2 was correlated with ar in males (p < 0.001; R2 = 0.70). Expression of ar in the brain increased during ontogeny in males and females, but no changes were seen for cyp17 or foxl2.    and ar (cyp17 expression was very low and did not differ from expression observed in controls (Fig. S9)).  Table S7. Effects of treatment on the relationship between expressed target genes in males and females.

Male
In both males and females, the interrelationships between the expressed genes differed in response to treatments, compared with their respective controls (shown in Figure 3). In response to linuron low/high treatments in males foxl2 was correlated with dmrt1/amh (p < 0.01; R2 > 0.43) that was not observed in the controls (Figure 3). Significant positive correlations see in control females (dmrt1 v. cyp17/amh/cyp19; ar v. foxl2; amh v. cyp19; Figure  3) were not observed in the linuron low or flutamide treatment groups (p < 0.05; R2 < 0.67). -0.59* -0.08 Asterices in dicate sig nificant corr elations between analysed gene s (normal t ext, no diff erence compared to control) or significant differences in the linear regression compared to the control (bold text [ANCOVA, with stage as a covariate]) with the arrows indicating the direction of the change. *p < 0.05, **p < 0.01, ***p < 0.001. Table S8. Effects of treatment on brain/testis ar and arm/testis gr.