Vinclozolin (VZ) is an endocrine-disrupting chemical that is able to interact with the endocrine system of mammalian models directly via receptor-binding and indirectly via the epigenetic regulation of DNA methylation. Gastropods often respond sensitively to chemicals classified as endocrine disrupting chemicals, but information on epigenetics is almost absent for this group. In order to investigate potential phenotypic and epigenetic effects of VZ in gastropods, we performed a two-generation study with the freshwater snail Physella acuta. The parental generation (F0) of P. acuta was exposed to 0.01, 0.1, 10 and 100 µg l−1 VZ. VZ-exposed F0 and nonexposed offspring of the control, 0.1 and 100 µg l−1 groups (F1 and F2) were investigated for phenotypic (reproduction, growth and embryo development) and epigenetic (5-methyl cytosine content) effects. VZ exposure of F0 did not induce an adverse reproductive phenotype in P. acuta. However, the size of F1 adults and the mortality of F2 embryos significantly increased when F0 adults were previously exposed to 100 µg l−1 VZ. At the level of global DNA methylation, we did not observe a VZ effect, but global DNA methylation of P. acuta decreased with increasing age. VZ affected the offspring of 100 µg l−1 VZ-exposed individuals, most probably due to an indirect VZ exposure of F1 (embryonic stage) and F2 (germline). The postulated relationship between VZ phenotype and epitype was not shown by the present study. However, our results demonstrate for the first time the presence and age-dependence of DNA methylation in P. acuta.
Molluscs have been shown to be sensitive to a wide range of environmental contaminants, including substances that are classified as endocrine disrupting chemicals (EDCs) due to their ability to interfere with the hormonal system of vertebrates (OECD, 2010). One of the best documented examples of the high sensitivity of molluscs to such substances is the masculinizing effect of tributyltin in female gastropods, which causes the development of imposex or intersex (Oehlmann et al., 2007; Oehlmann, Pottinger & Sumpter, 2012). Vinclozolin (VZ)—a fungicide with antiandrogenic activity in vertebrates—induced imposex and a reduction of accessory sex-organ expression in the caenogastropods Marisa cornuarietis, Nassarius reticulatus and Nucella lapillus (Tillmann et al., 2000). In adults of the freshwater pulmonate snail Lymnaea stagnalis, VZ exposure resulted in infertility and impaired egg production (Ducrot et al., 2010), while the nature and amplitude of the effects of VZ in L. stagnalis highly depend on the conditions of exposure (Lagadic, Coutellec & Caquet, 2007). Giusti et al. (2014) reported the development of polyembryonic eggs and an increased number of eggs per clutch in response to VZ exposure in L. stagnalis.
Furthermore, VZ is a prominent example of an epigenetically active EDC in mammals (e.g. Anway et al., 2005; Anway, Rekow & Skinner 2008; Guerrero-Bosagna et al., 2010). There is further evidence that VZ affects pathways that drive epigenetic inheritance in unexposed offspring generations (e.g. Anway et al., 2008; Guerrero-Bosagna et al., 2010). VZ exposure of the parental (F0) generation during gonadal sex determination (100 mg kg−1 VZ d−1 from embryonic day 8 to 14/15) caused increasing sperm cell apoptosis and several diseases and abnormalities in VZ-unexposed F1 to F3/F4 individuals (Anway et al., 2005, 2008). In these studies, phenotypic changes were associated with locus-specific hypo- and hypermethylation and differential expression of epigenetic key enzymes (e.g. DNA methyltransferases, histone deacetylases, histone methyltransferases). For invertebrates, however, information on epigenetic effects of chemicals is scarce. In the waterflea Daphnia magna, epigenetic alterations in response to VZ exposure (overall DNA hypomethylation) persisted to the nonexposed F2 offspring (Vandegehuchte et al., 2010). VZ exposure of the mosquito Aedes albopictus led to a reversible alteration of global DNA methylation levels for two unexposed offspring generations (Oppold et al., 2015).
In molluscs, the existence of DNA methylation as one of the key features of the epigenetic system has been demonstrated for the freshwater Biomphalaria glabrata, the terrestrial Helix pomatia and the marine Pecten islandicus, Patella sp. and Zeacumantus subcarinatus (Regev, Lamb & Jablonka, 1998; Fneich et al., 2013; Joe, 2013). Furthermore, cellular components for DNA methylation are present in a number of other gastropods (Aplysia californica, Lymnaea, Tritonia, Pleurobranchaea and Clione, unpublished observations by Moroz and Kohn reported by Walters & Moroz, 2009; Z. subcarinatus, Joe, 2013; Pomacea canaliculata, Ottaviani et al., 2013). However, information on whether environmental contaminants such as VZ also induce epigenetic modifications in gastropods is completely missing.
The present study aims to investigate whether (1) VZ induces epigenetic modifications in Physella acuta (Draparnaud, 1805), (2) epigenetic modifications are related to alterations in its reproductive phenotype and (3) epigenetic and phenotypic modifications are detectable if only the germline is exposed. Therefore, we performed a two-generation study in which the F0 was directly, and F1 and F2 germline indirectly, exposed to VZ (Fig. 1).
We selected the freshwater snail P. acuta as the gastropod model by reason of its short generation time (35 d until first reproduction at 25 °C; Tsitrone, Jarne & David, 2003), its successful use in multigeneration studies and well-known endpoints (Jarne et al., 2000; Tsitrone et al., 2003; Seeland et al., 2013) and its general responsiveness to chemicals acting as EDCs in mammals (increased egg production; De Castro-Català et al., 2013) and to VZ in particular (reduced hatching rate; Sánchez-Argüello, Aparicio & Fernández, 2012).
MATERIAL AND METHODS
Physella acuta individuals used for the experiments originated from our in-house culture (Department of Aquatic Ecotoxicology, Goethe University, Frankfurt am Main) that was derived from material from fish aquaria in 2010. Physella acuta was reared in 15–60 l aquaria filled with ISO medium (OECD, 2004) and placed in a climatic chamber (22.6 °C ± 1.22 °C) with a light-dark cycle of 16:8 h. ISO medium in the aquaria was renewed weekly. Snails were fed ad libitum with a 1:1 mixture of TetraPhyll and TetraMin twice a week.
Snails of a minimum shell height of c. 7 mm and a minimum age of 70 d posthatching were used for the reproduction tests. At this life stage, the snails are sexually mature and their growth rate is stable (oocytes >80 µm, spermatozoa class 3; Brackenburry & Appleton, 1991). To test the effects of VZ on P. acuta in the two-generation study (see below), VZ (CAS 50471-44-8, 99.9% purity, Sigma-Aldrich) was dissolved in ISO medium. The VZ exposure was not investigated directly by chemical analysis, because of the low VZ concentrations (10 and 100 ng l−1, 10 and 100 µg l−1) and the high limit of quantification of the available method (13.3 µg l−1, Oppold et al., 2015). Furthermore, VZ has been reported to dissipate quickly from water systems (Szeto et al., 1989; Wong et al., 1995; Euling et al., 2002; Kavlock & Cummings, 2005) and to have a very low bioconcentration/bioaccumulation potential (PPDB, 2015).
However, in order to follow the fate of VZ, a degradation experiment using a high VZ concentration was performed. The dissipation time of VZ was investigated during 24 h exposure of three snails each in triplicated 200-ml test vessels filled with 2.5 mg l−1 VZ in ISO medium (1% acetone) under the experimental conditions described above. Snails were fed once ad libitum with a 1:1 mixture of TetraPhyll and TetraMin. Analytical samples were collected after 0, 15, 30, 45 and 60 min and 2, 6 and 24 h and then immediately measured by reverse-phase high-performance liquid chromatography (Dionex Corporation, Sunnyvale, CA, USA) using a C18 column (particle size 5 µm and pore size 120 Å, 4.3 × 150 mm) at a flow rate of 1 ml min−1, with a UV detector set to 200 nm (Oppold et al., 2015). To determine the 50% dissipation time (DT50), a one-phase decay model (based on peak areas) was used.
The experimental design of the two-generation study included two reproduction tests (Fig. 1). In the first reproduction test (R1), 140–170 d old adult snails (7–10 mm) were exposed to four VZ concentrations (10 and 100 ng l−1, 10 and 100 µg l−1 in ISO medium) and (for the control group) to ISO medium. For all control and VZ treatment groups, the experimental design has four replicates for measuring phenotypic parameters and three replicates for the quantification of global DNA methylation and physicochemical water parameters.
Three F0 snails per replicate were placed in 250-ml glass beakers filled with 200 ml of ISO medium without (F0 controls) or with VZ (F0 VZ treatments). The VZ exposure period was 28 d. Test vessels were aerated using glass Pasteur pipettes. The experimental conditions (temperature, light-dark cycle) were the same as described above. Twice a week, all test media, including controls, were renewed and snails were fed with 5 mg of a 1:1 mixture of TetraPhyll and TetraMin snail−1 d−1.
Before each medium renewal, ammonium (quick test, Aquamerck, Merck), nitrite (Quantofix, Macherey-Nagel) and oxygen concentrations (CellOx 325) as well as pH (SenTix 41) and conductivity (TetraCon® 325, WTW sensors) were measured in the test vessels. Ammonia concentrations were calculated based on ammonium concentration, temperature and pH (Hobiger, 1996). Furthermore, all egg masses were removed, counted, transferred separately to 12-well cell culture plates and investigated for the number of eggs with the use of an inverse microscope (Axiovert 40c, Zeiss) and a stereo microscope (S6E, Leica). Dead embryos or embryos unable to reach the veliger stage were classified in the category ‘mortality’. Embryos showing strong deformations or a developmental retardation from veliger stage onwards (compared to the control group) were counted as ‘deformations’ (for more details see Seeland et al., 2013).
The offspring of F0 controls (F1 controls) and of two selected F0 VZ treatments (100 ng l−1, 100 µg l−1 VZ; F1 VZ treatments) were transferred to 5-l glass aquaria filled with 4 l ISO medium and cultured for 6–8 weeks until they reached sexual maturity. In order to detect potential delayed VZ effects, the reproductive capacity of F1 adults was assessed in a second reproduction test (R2, Fig. 1). In this case, 70–98 d old fertile adult F1 snails (7–8.5 mm) were cultured in 500-ml test vessels with ISO medium for 28 d, while all other aspects of the experimental design were similar to R1. At the start and the end of both reproduction tests, body weight (wet weight), size and mortality of adult snails were recorded. The weight of snails was measured with a precision of 0.1 mg after blotting of snails with tissue paper. Size of snails was represented by their maximum shell heights, measured with a precision of 0.1 mm using a binocular (Olympus, Stemi SZX7, CellsSensEntry v. 1.6.9464.0).
DNA extraction and DNA methylation assay
Specimens of P. acuta were randomly sampled from the replicates of both reproduction tests at day 0 (start of tests), day 15 (only in R1) and day 28 (end of tests) with nt0,R1 = 9, nt15 and t28,R1 = 3 and nt0 and t28,R2 = 5. In order to document possible changes in DNA methylation levels during early juvenile development, snails from five different egg masses were sampled after 4, 10, 20, 30, 41 and 50 d posthatching. All snails were dried on cellulose tissue and stored at −80 °C until DNA extraction.
Frozen material was treated with lysis buffer [cetyl trimethylammonium bromide (CTAB), ethylenediaminetetraacetic acid, sodium chloride] and proteinase K and ground for 30–60 s (tissue lyser, MM 400, Retsch, Haan). The CTAB protocol (modified from Doyle & Doyle, 1987) was used to facilitate complex formation with polysaccharides from snail mucus. DNA was extracted from homogenates with chloroform/isoamyl alcohol (24:1). RNA still present in samples was digested using RNAse A and removed by an additional chloroform/isoamyl alcohol extraction step. Finally, DNA was precipitated with ice-cold isopropanol, redissolved in TE buffer and checked for quality (agarose gel) and quantity (260 nm, Hellma Tray cell, type 105.800, Hellma Analytics, BioSpectrometer Kinetic photometer, Eppendorf).
The 5-methyl cytosine (5 mC) level of P. acuta DNA was determined using the MethylFlash Methylated DNA Quantification Kit (Epigentek) and the assay performed according to the manufacturer's instructions. Briefly, a specific capture antibody combined with a detection antibody allowed the photometric detection of 5 mC at 450 nm (M200 PRO, Tecan). The 5 mC levels in the samples were quantified in duplicate using a standard curve (linear regression) prepared by mixing unmethylated and methylated control DNA (0.25–2.5% methylated DNA, in triplicate).
Physicochemical data measured in R1 and R2 were tested for differences with F-tests. All percentage data were arcsine-transformed and in case of heterogeneous variances additionally log-transformed (embryo mortality in R2, 5 mC levels in R1). The homogeneity of variances was determined by means of Bartlett's tests. Means of egg production, increase of adult weight and size, embryo mortality and deformations, and 5 mC levels of P. acuta were tested for VZ-related differences with one-way ANOVAs and subsequent Tukey post-hoc tests.
The cumulative egg production of P. acuta in R1 and R2 was tested for deviation from linearity with the Runs test. Reproduction parameters in the controls and the overall 5 mC levels were tested for differences regardless of VZ treatments using an unpaired t-test with previous F-test. Linear and nonlinear (one-phase decay model) regressions were computed to detect a potential relationship between the 5 mC level and the age of snails. Furthermore, ammonium vs oxygen concentrations, ammonium/oxygen concentrations vs 5 mC levels and 5 mC levels vs the age of snails were tested for potential relationships in one-tailed Spearman correlation analyses. Nonparametric Spearman analysis was used because of the non-Gaussian distribution (D'Agostino and Pearson omnibus normality test). All statistical analyses were performed using the software Graph Pad Prism v. 5.0 and at the probability level of 0.05.
VZ and physico-chemical parameters
Directly after spiking test vessels with 2.5 mg l−1 VZ (nominal concentration), 127% of VZ was recovered. The DT50 in ISO medium in the presence of snails was 110.1 min (64.7–380.3 min, 95% confidence interval). After 24 h, 0.015 ± 0.009 mg l−1 VZ (0.38 ± 0.20%) was detected.
Conductivity and pH in the test media were similar in both reproduction tests (Table 1). In contrast, oxygen, ammonium and ammonia concentrations differed significantly between R1 and R2. Average oxygen concentration was lower in R1 than in R2. Ammonium was higher and more variable in R1 than in R2. Oxygen and ammonium concentrations fluctuated concomitantly: ammonium concentrations increased with decreasing oxygen concentrations (Spearman r = –0.63, P < 0.001). Ammonia concentration was always below 0.06 mg l−1. Toxic nitrite was not detected during the study.
|Conductivity (µS cm−1)||747.9 ± 56.5||730.6 ± 57.3||1.029||0.92|
|pH||7.46 ± 0.18||7.71 ± 0.13||1.917||0.08|
|Oxygen (mg l−1)||4.92 ± 0.54||6.13 ± 0.31||3.034||0.003|
|Ammonium (mg l−1)||4.12 ± 2.30||1.49 ± 0.46||25.00||<0.001|
|Ammonia (mg l−1)||0.06 ± 0.03||0.03 ± 0.01||11.26||<0.001|
|Nitrite (mg l−1)||0.00 ± 0.00||0.00 ± 0.00||–||–|
|Conductivity (µS cm−1)||747.9 ± 56.5||730.6 ± 57.3||1.029||0.92|
|pH||7.46 ± 0.18||7.71 ± 0.13||1.917||0.08|
|Oxygen (mg l−1)||4.92 ± 0.54||6.13 ± 0.31||3.034||0.003|
|Ammonium (mg l−1)||4.12 ± 2.30||1.49 ± 0.46||25.00||<0.001|
|Ammonia (mg l−1)||0.06 ± 0.03||0.03 ± 0.01||11.26||<0.001|
|Nitrite (mg l−1)||0.00 ± 0.00||0.00 ± 0.00||–||–|
Significant difference is indicated by bold numbers.
Mortality, weight and size of adults
Mortality of Physella acuta did not significantly differ among controls and the VZ treatments in R1 and R2. The F0 adult mortality was ≤10% in all groups except for the 10 ng l−1 VZ treatment (19%) and the F1 adult mortality was ≤5% for all groups. In experiment R1, the F0 control individuals had an average initial weight of 77 mg and a final weight of 102 mg (increase of 25 mg = 32.5%; Fig. 2A), and an average initial size of 8.2 mm and a final size of 9.4 mm (increase of 1.3 mm = 14.6%; Fig. 2B). The F1 individuals whose parents were not exposed to VZ during R1 (F1 control) weighed 66 mg at the start and 97 mg at the end of R2 (increase of 32 mg = 47.0%; Fig. 2A), and were 7.2 mm at the start and 8 mm at the end of R2 (increase of 0.8 mm = 11.1%; Fig. 2B). The increase of adult weight and size did not significantly differ between controls and VZ treatments in F0 and F1, except for the size of the 100 µg l−1 F1 VZ treatment group, which increased significantly compared with the F1 control (increase of 1.8 mm, P < 0.001).
Reproduction, embryo mortality and embryo deformations
The total egg number, the number of egg masses per snail and the egg number per egg mass were continuous throughout the 28 d in R1 and R2 (linear increase of cumulative data: R²(R1) = 0.64–0.96; R²(R2) = 0.88–0.96). Reproductive parameters did not differ significantly among controls and the VZ treatments in R1 and R2 (Fig. 3A). Reproduction of the F0 control group was 395 ± 91.8 eggs per snail and 12.3 ± 2.5 egg masses per snail. Thus, R1 control animals produced significantly less offspring than observed in R2 (F1 control group with 649 ± 81.9 eggs per snail and 23.0 ± 3.1 egg masses per snail, P < 0.001), although the egg masses of F0 controls (31.9 ± 2.7 eggs per egg mass−) were slightly larger (28.3 ± 3.1 eggs per egg mass, P < 0.05).
The average mortality and deformation rate of P. acuta embryos produced by the F0 control in R1 (F1 control embryos) was 9.3 and 4.6%, respectively (Fig. 3B). Neither parameter differed significantly between controls and VZ treatments with exception of F2 embryos in the 100 µg l−1 VZ group. When compared with F2 control embryos, the embryo mortality and deformation rate in the 100 µg l−1 VZ F2 group were increased by 12.4% (P = 0.04) and 6.8% (P = 0.35), respectively.
Global DNA methylation after exposure to VZ
The 5 mC levels of P. acuta changed during R1 and R2, although a clear response pattern to the VZ treatment was not found (Fig. 4). There was a general trend to decreased 5 mC levels over the time of R1 in all treatments, including the control. Levels decreased by 30–60% from an initial value of 0.40 ± 0.04% (significant for the control and for the 10 ng l−1 and 10 µg l−1 VZ treatments). This pattern was not recovered during R2 with the unexposed offspring generation and the two VZ treatments show contrasting 5 mC level changes. In the low-concentration VZ treatment the 5 mC level significantly increased over time, in line with the trend of change of 5 mC level in the respective controls. In the high-concentration VZ treatment, however, a significant decrease of the 5 mC level was observed over time. Notably, 5 mC levels measured at the end of both reproduction tests were significantly correlated with the average ammonium (Spearman r = −0.69, P = 0.035) and oxygen concentrations (Spearman r = 0.83, P = 0.008).
DNA methylation in relation to age
The F1 individuals had a significantly higher 5 mC level (0.629 ± 0.030%) than the F0 individuals (0.33 ± 0.01%, P < 0.001). F1 individuals were 42–70 d younger than F0 individuals at the start of R2 and R1, respectively. The cross-generational comparison of the 5 mC levels of P. acuta controls with their respective age showed a linear decrease in 5 mC levels with increasing age (Fig. 5, inset), which was, however, not significant (Spearman r = −0.80, P = 0.07). Nevertheless, the additional analysis of younger P. acuta individuals (4–50 d posthatch) corroborated the trend for an age-dependent decrease of 5 mC levels. Juvenile snails had undergone an exponential demethylation with ageing (Fig. 5). After 3–9 d posthatching, 50% of the DNA was demethylated (half-life 5 mC 0–50 d = 4.75 d (3.25–9.03 d, 95% CI). When combining both datasets (0–200 d), 5 mC levels and age were negatively correlated (Spearman r = 0.94, P < 0.001).
At the phenotypic level, 100 µg l−1 VZ treatment affected the size of F1 adults and the mortality and deformation of F2 embryos of Physella acuta, probably due to an indirect VZ exposure of F1 (embryonic stage) and F2 (germline). Despite the fact that VZ did not induce epigenetic modifications, our results demonstrate for the first time the presence and age-dependence of DNA methylation in P. acuta.
In aqueous solutions, VZ has a short half-life (20–42 h; Szeto et al., 1989; Oppold et al., 2015). The half-life (DT50) was considerably shorter in our study (c. 2 h), probably due to the medium used. Typically, VZ is rapidly hydrolysed to the stable metabolites M1 (2-[[(3,5-dichlorophenyl)-carbamoyl]oxy]-2-methyl-3-butenoic acid) and M2 (3,5-dichloro-2-hydroxy-2-methylbut-3-enanilide), which are EDCs with an antiandrogenic activity not present in the parent molecule (Wong et al., 1995; Euling et al., 2002; Kavlock & Cummings, 2005). Therefore, P. acuta was exposed to these VZ metabolites too. The toxic VZ metabolite M3 (3,5-dichloroanilin) was most likely not relevant for the short-term exposure used in the present study, given that M3 only forms at low concentrations after several weeks (Szeto et al., 1989). The proportion of VZ and its metabolites in water does not necessarily reflect the VZ concentration in animals. For example, after 96 h of exposure, the internal VZ concentration in Lymnaea stagnalis represented 67.52 ± 8.81% of the total internal residue concentration, and M1 and M2 2.02 ± 0.92 and 6.79 ± 2.20%, respectively (Ducrot et al., 2010). Overall, the active compound(s) in P. acuta remain unknown.
Lack of direct but presence of indirect VZ effects on adults
An increase in growth was observed in P. acuta F1 adults whose parents were directly exposed to 100 µg l−1 VZ (Fig. 2B). While this is in contrast to studies that have observed direct growth inhibition after adult exposure to VZ (Kiparissis et al., 2003; Ducrot et al., 2010), the early VZ exposure of the F1 germline (together with F0 adults) and embryos (egg masses deposited in VZ-containing medium for maximum of 4 d) in the present study may have produced a different phenotype.
Growth was likewise not directly affected in VZ-exposed L. stagnalis (Giusti et al., 2014), which is in contrast to the observation of VZ-induced growth inhibition by Ducrot et al. (2010). Such a discrepancy may be attributed to a difference in snail sexual maturity (older adults used by Giusti et al., 2014,vs younger adults by Ducrot et al., 2010). These results circumstantially support the results of the present study, i.e. the absence of a VZ effect in older adults (R1) and its presence in younger adults (R2). Thus, the responsiveness to VZ may be age-dependent. A life-stage-speciﬁc toxicity has already been demonstrated for P. acuta exposed to the fungicide pyrimethanil (Seeland et al., 2013).
VZ did not affect the egg production of P. acuta at the low to medium concentrations of 10 ng l−1 to 100 µg l−1 (28 d exposure; Fig. 3A), whereas 5 mg l−1 VZ reduces the hatching of embryos and ≥5 mg l−1 VZ is genotoxic for the species (Sánchez-Argüello et al., 2012). For comparison, the egg production of L. stagnalis is reduced at 250 µg l−1 and 2.5 mg l−1 VZ, but stimulated at 25 µg l−1 VZ after 21 d of exposure (Ducrot et al., 2010). For the same species, Giusti et al. (2014) reported a reduction of clutch size and an increased number of polyembryonic eggs at 10–110 ng l−1.
Indirect VZ effects on embryo development
Deposited F1 egg masses were exposed to the VZ-polluted medium for 3–4 d, until the water was renewed. One might assume that the egg mass envelope and the egg integument protect the eggs against exposure to VZ and its metabolites. Indeed, the capsular membrane around eggs and the egg mass envelope of Biomphalaria sudanica and L. stagnalis protect early embryos from microorganisms and are more permeable to water than to inorganic ions (Beadle, 1969). However, which compounds are able to enter the egg is determined by their molecular weight: molecules of 342 g mol−1, but not 504 g mol−1, can permeate the capsular membrane of B. sudanica. In L. stagnalis, molecules with a weight of 504 g mol−1, but not 3000 g mol−1, are able to permeate the capsular membrane (Beadle, 1969). Such information is not available for P. acuta. Nevertheless, the F1 egg capsules may very likely have been permeated by VZ (molecular weights: VZ = 286 g mol−1, M1 = 304 g mol−1, M2 = 260 g mol−1; cf. Wong et al., 1995) to expose the F1 embryos to it.
With regard to embryo development, VZ induced significant deformations in F2 embryos, whose ‘grandparents’ (together with F1 germline and probably F1 embryos) were exposed to 100 µg l−1 VZ (Fig. 3B). A VZ exposure during R1 can likewise be assumed for the F2 germ cells, because the primordial germ cells in molluscs can be traced back to early derivatives of D-cells (Fioroni, 1992). The D-cell can already be distinguished in the four-cell stage of many molluscs (e.g. Dreissena polymorpha, Nassarius reticulatus) because of its larger size (‘D-cell inequality’; Fioroni, 1992). The adverse VZ phenotype in the 100 µg l−1 VZ F2 embryos could therefore have been caused by an indirect VZ exposure of the F2 D-cells already existing in the F1 embryos in R1.
Phenotype vs epitype
We observed specific VZ-induced F1 and F2 phenotypes, but indefinite epitypes, in P. acuta. This is in contrast to other studies, in which it appears to be more difficult to detect phenotypic than epigenetic VZ effects. For example, Vandegehuchte et al. (2010) exposed F0 daphnids to VZ and detected an overall DNA hypomethylation in the VZ unexposed F2, but no phenotypic effects in the unexposed offspring F1 to F3. After in utero VZ exposure of gestating rats (F0), Anway et al. (2008) detected apoptotic germ cells in VZ-unexposed offspring only until F2, but there were still striking changes at the epigenetic and transcriptomic level in VZ-unexposed F3. Thus, the adverse VZ effect on the F1 embryos and F2 germline detected in P. acuta in the present study may have occurred during R1. This is a more reasonable assumption than the indirect transgenerational transmission of VZ effects via epigenetic mechanisms.
Indeed, we observed a significant demethylation in 100 µg l−1 VZ F1 individuals in the course of R2 together with an adverse F1 and F2 phenotype (Figs 2B, 3B), but no alterations of the global 5 mC if comparing control and different VZ treated F0 individuals in R1. A balance between locus-specific hypo- and hypermethylation in response to VZ could have masked effects on the global 5 mC level in the present study. However, hypomethylation seemed to be generally related to adverse P. acuta phenotypes regardless of the VZ treatment. The strongest decrease of 5 mC levels co-occured with the highest adult mortality (10 ng l−1 in R1), the lowest weight:size ratio, and the highest embryo deformation and mortality (100 µg l−1 in R2). Moreover, the general lower 5 mC content detected in R1 was associated with an overall lower egg production, higher ammonium/ammonia concentrations and often a higher age of adult snails if compared with R2 (Figs 3–5, Table 1).
Hypomethylation and detoxication
The significant demethylation during R1 and the decreased reproductive success of P. acuta in R1 suggest the presence of some additional factors compared to R2 (Figs 3A, 4). One unintended stressor in R1 could have been the elevated ammonium/ammonia level and lowered oxygen concentration (Table 1). Differences in ammonia and oxygen concentrations between R1 and R2 are likely to be linked to the difference in the age and size of the snails. Older and larger snails are more active and produce more faeces as compared with younger, smaller snails. Physella acuta tolerates high ammonium concentrations, as indicated by the 50% lethal concentration of 580 mg l−1 of the salt tetrabutyl ammonium in juveniles (shell length <9 mm; Bernot, Kennedy & Lamberti, 2005). Besides, the distribution of gastropod species (including P. acuta) monitored at 981 sites in German rivers is significantly related to a number of environmental variables, but not to ammonium (Früh, Stoll & Haase, 2012). Such observations provide evidence for a reasonable ammonium tolerance of P. acuta, which may be due to its effective detoxication mechanisms (Ma et al., 2014).
Lower 5 mC levels may reflect physicochemical stress (with regard to ammonia and oxygen concentrations). Ammonium is conjugated by the phase II detoxication enzyme glutathione S-transferase to glutathione (GSH) for elimination (e.g. Kosenko et al., 1998; Ching et al., 2009). A chronic consumption of GSH increases the need for GSH synthesis, which is based on homocysteine. Homocysteine is in turn an important member of the methionine pathway as it is the key molecule for S-adenosylmethionine synthesis, being the key methyl donor for DNA methylation. Thus hypomethylation can result from a depletion of methyl groups caused by a shift of homocysteine consumption towards GSH synthesis (and detoxication) rather than for DNA methylation (Szyf, 2007; Lee, Jacobs & Porta, 2009). Such an imbalance in response to ammonium would explain the reduced 5 mC levels of the controls at the end of R1 (Fig. 4).
The same mechanism can explain the hypomethylation observed in the 10 ng l−1 to 100 µg l−1 VZ groups in R1 and the 100 µg l−1 VZ group in R2 (Fig. 4). Edlich & Lyr (1992) considered unspecific oxidative damages, including lipid peroxidation, as the main mode of action of VZ. Indeed, oxidative stress has been induced by VZ in mammalian cells, where GSH decreased with increasing oxidative stress at increasing VZ concentrations (Radice et al., 1998). Our assumption is further supported by findings in the zebrafish Danio rerio that VZ affects the expression of genes belonging to both methionine and GSH metabolism (Martinović-Weigelt et al., 2011). Experimental evidence for chemically-induced parallel effects on the methionine pathway and global methylation was further provided by Wang et al. (2009), who linked a tributyltin-induced global hypomethylation in the rockfish Sebastiscus marmoratus with a reduced availability of S-adenosylmethionine for the methionine pathway. Nevertheless, a clear relationship between phase II detoxication and DNA hypomethylation has not yet been demonstrated for gastropods or for VZ. There is a clear need for future research in this area.
Comparison of methylation levels
Our study provides the first evidence for the existence and the plasticity of global DNA methylation in P. acuta. The 5 mC levels of P. acuta (0.4–2.4%) are in the same range of DNA methylation as in ctenophores (0.1–1.1%, Dabe et al., 2015) and other invertebrates such as Drosophila melanogaster (0.1–0.4%, Lyko, Ramsahoye & Jaenisch, 2000), the mosquito Aedes albopictus (0.1–0.5%, Oppold et al., 2015) and Daphnia magna (0.17–2.0%, Vandegehuchte, Lemière & Janssen, 2009; Menzel et al., 2011; Asselmann et al., 2015). The average 5 mC level of adult P. acuta (0.4%) is very low compared with 2.1–4.9% in adults of Biomphalaria glabrata, Pecten islandicus, Helix pomatia and Patella sp. (Regev et al., 1998; Fneich et al., 2013). One reason might be methodological differences (Lisanti et al., 2013).
The decrease of 5 mC levels from juvenile to adult life stages of P. acuta (Fig. 5) is in general accordance with observations on Crassostrea gigas, D. melanogaster and the marine annelid Chaetopterus variopedatus (del Gaudio, di Giaimo & Geraci, 1997; Lyko et al., 2000; Riviere et al., 2013). No or very low 5 mC levels were detected in adult stages (and fly ovaries), whereas their embryonic DNA (and annelid sperm DNA) was considerably methylated. 5 mC levels in D. melanogaster decreased from 0.4 to 0.1% during the first 14–15 h of embryo development, indicating a functional role of DNA methylation in development and ageing (Lyko et al., 2000). Likewise, DNA methylation in C. gigas shifted after only 9 h of embryogenesis (Riviere et al., 2013). The heavily and increasingly methylated DNA in oocytes up to the morula stage (3 h postfertilization, hpf) lost most of its methyl groups when reaching the gastrula stage (9 hpf, Riviere et al., 2013). The congruent observation of hypomethylation in P. acuta juveniles with increasing age (Fig. 5) also implies the involvement of DNA methylation in regulating functions for mollusc ageing—a role that has been widely accepted for mammalian epigenetics.
In conclusion, VZ exposure had a phenotypic impact on offspring generations of Physella acuta (100 µg l−1 F1 and F2 treatments), but global DNA methylation in F0 and F1 adults was not affected. Our results imply that a direct exposure of F0 adults and a simultaneous, indirect exposure of F1 embryos and F1 and F2 germ cells (rather than epigenetic mechanisms) resulted in the observed VZ effects. However, knowledge of the molluscan epigenome, its susceptibility to chemical manipulation and transgenerational inheritance of epigenetic changes, is still very limited. In that connection, we report age-dependency of global DNA methylation in gastropods and suggest that DNA methylation is involved in ontogenetic processes in P. acuta.
We gratefully acknowledge the technical support of our late colleague Olaf Dittberner and of Maximilian Behr (VZ measurements). We also thank three anonymous reviewers for their helpful comments. R.M., S.C., A.O., J.O. and M.W. initiated and conceptualized the present study. Experimental work and data management were done by S.C., C.S. and A.O. R.M., S.C. and A.O. performed statistics. All authors were involved in writing and editing the manuscript.