Time- and Dose-Dependent Effects of Roundup on Human Embryonic and Placental Cells
Roundup® is the major herbicide used worldwide, in particular on genetically modified plants that have been designed to tolerate it. We have tested the toxicity and endocrine disruption potential of Roundup (Bioforce®) on human embryonic 293 and placental-derived JEG3 cells, but also on normal human placenta and equine testis. The cell lines have proven to be suitable to estimate hormonal activity and toxicity of pollutants. The median lethal dose (LD50) of Roundup with embryonic cells is 0.3% within 1 h in serum-free medium, and it decreases to reach 0.06% (containing among other compounds 1.27 mM glyphosate) after 72 h in the presence of serum. In these conditions, the embryonic cells appear to be 2–4 times more sensitive than the placental ones. In all instances, Roundup (generally used in agriculture at 1–2%, i.e., with 21–42 mM glyphosate) is more efficient than its active ingredient, glyphosate, suggesting a synergistic effect provoked by the adjuvants present in Roundup. We demonstrated that serum-free cultures, even on a short-term basis (1 h), reveal the xenobiotic impacts that are visible 1–2 days later in serum. We also document at lower non-overtly toxic doses, from 0.01% (with 210 μM glyphosate) in 24 h, that Roundup is an aromatase disruptor. The direct inhibition is temperature-dependent and is confirmed in different tissues and species (cell lines from placenta or embryonic kidney, equine testicular, or human fresh placental extracts). Furthermore, glyphosate acts directly as a partial inactivator on microsomal aromatase, independently of its acidity, and in a dose-dependent manner. The cytotoxic, and potentially endocrine-disrupting effects of Roundup are thus amplified with time. Taken together, these data suggest that Roundup exposure may affect human reproduction and fetal development in case of contamination. Chemical mixtures in formulations appear to be underestimated regarding their toxic or hormonal impact.
Mammals and humans may be exposed to Roundup herbicide residues by agricultural practices (Acquavella et al. 2004) or when the residues enter the food chain (Takahashi et al. 2001); glyphosate is also found as a contaminant in rivers (Cox 1998).
In our previous work, we have demonstrated that the major herbicide used worldwide, Roundup, was toxic for a human placental cell line at concentrations below that recommended for agricultural use (1–2 %, i.e., with 21–42 mM glyphosate) and had endocrine-disrupting potential on estrogen synthesis at lower nontoxic doses. These cell culture experiments were performed with or without serum only on one cell model and up to 18 or 48 h, respectively (Richard et al. 2005). Roundup is believed to be rather specific and less toxic to the ecosystem than other pesticides; transgenic plants tolerant to this compound have even been developed following this argument (Vollenhofer et al. 1999, Williams et al. 2000). Roundup is in fact a mixture of an isopropylamine salt of glyphosate, quantitatively a minor compound called the active ingredient, and various adjuvants (Cox 1998, Cox 2004) usually considered as surfactants forming an inert part of the composition and a secret of manufacturing. All these adjuvants can be differently used depending on the formulations. Among them are ammonium sulfate, benzisothiazolone, 5-chloro-2-methyl 3(2H)-isothiazolone, FD&C Blue No. 1, glycerine, 3-iodo-2-propynyl butylcarbamate, isobutane, isopropylamine, light aromatic petroleum distillate, methyl p-hydroxybenzoate, methyl pyrrolidinone, pelargonic acid, polyethoxylated tallowamine or alkylamine (POEA), potassium hydroxide, propylene glycol, sodium sulfite, sodium benzoate, sodium salt of o-phenylphenol, and sorbic acid. These products allow for glyphosate penetration through plasmatic membranes, potentialization of its action, increased stability, and potential bioaccumulation. Glyphosate does not appear to have an herbicide action by itself.
A differential effect was noticed in our previous study in favour of Roundup, in contrast to pure glyphosate. The purpose of the present work was to study in more detail the dose-and time-dependent cytotoxicity of both compounds, up to 72 h, comparing the effects on two cell lines from human embryonic kidney and placenta. Moreover, we wanted to examine the combined effects of this chemical mixture Roundup (Bioforce® herein) on a new cellular model. We also tested the hypothesis that Roundup and glyphosate would inhibit aromatase activity at doses lower that those producing overtly toxic effects. We determined the aromatase disruption potential in 293 cells transfected with aromatase cDNA, and examined the temperature-dependent and direct mechanism of inhibition of aromatase by glyphosate on preparations of fresh human placenta and equine testis, a tissue known to be aromatase-rich (Lemazurier et al. 2001).
This was of particular interest since Roundup and/or glyphosate were suggested to disturb human (Savitz et al. 1997) and rat pregnancies (Daruich et al. 2001, Beuret et al. 2005), mouse kidney (Peluso et al. 1998), rabbit spermatogenesis (Yousef et al. 1995), and other human tissues (Monroy et al. 2005).
The cytotoxic and/or genotoxic effects of glyphosate have been reported at several checkpoints of the ecosystem, for instance on fish (Jiraungkoorskul et al. 2003), tadpoles and other aquatic species (Pettersson and Ekelund 2006), but also on urchin eggs (Marc et al. 2002, 2004, 2005) and human cells (Richard et al. 2005, Monroy et al. 2005). The endocrine disruption provoked by this compound is less documented. However, it has a very clear target at two crucial steps of steroidogenesis in mammals: at the first rate-limiting level of mitochondrial cholesterol transport (Walsh et al. 2000), and at the last irreversible conversion of sexual steroids androgens into estrogens, via a direct action on the aromatase enzyme (Richard et al. 2005).
Aromatase is an evolutionarily well conserved cytochrome P450 enzyme. Its superfamily includes numerous proteins able to metabolize xenobiotics (Nelson 1998). Its catalytic action is ensured by the product of the CYP19 gene (Bulun et al. 2003) associated with another moiety, the ubiquitous NADPH-dependent reductase as electron donor. It is considered a limiting factor involved in estrogen synthesis and, thus, in physiologic functions, including female and male gametogenesis (Carreau 2001), reproduction, sex differentiation, and even bone growth. It is also pharmacologically controlled in the treatment of estrogen-dependent cancers (Séralini and Moslemi 2001).
The cytotoxic effect of Roundup on cells, and the direct action of glyphosate on aromatase, could explain some reproduction problems at least in part. Among the two lines, the 293 cells have proven to be very suitable to estimate hormonal activity for xenobiotics after transfection (Kuiper et al. 1998). In contrast, JEG3 cells present natural aromatase activity and are also considered a useful model to examine placental toxicity (Letcher et al. 1999). These cell lines may be even less sensitive to xenobiotics than primary cultures (L’Azou et al. 2005); in this case, the effects measured could well be an indication of human placental toxicity in vivo, if sufficient contamination occurs, because the phenomena appear to be amplified with time in cells.
Materials and Methods
N-(Phosphonomethyl) glycine (glyphosate) was purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). The herbicide Roundup used in this work is the formulation available on the market called Roundup Bioforce®, which contains 360 g/L acid glyphosate, equivalent to 480 g/L of isoproplyamine salt of glyphosate, homologation 9800036, Monsanto, Anvers, Belgium. A 2% solution of Roundup (1 or 2% is recommended by the company for agricultural use, i.e., 21–42 mM glyphosate) and an equivalent solution of glyphosate were prepared in Eagle’s modified minimum essential medium (EMEM; Abcys, Paris, France). When their effects were compared, the pH of glyphosate solution was adjusted to the pH of the 2% Roundup solution (∼ pH 5.8). Successive dilutions were then obtained with serum-free or serum-containing EMEM. 3-(4,5-Dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide (MTT) and all other compounds, unless specified otherwise were obtained from Sigma-Aldrich. MTT was prepared as a 5-mg/mL stock solution in phosphate-buffered saline, filtered through a 0.22-μm filter before use, and diluted to 1 mg/mL in serum-free EMEM.
The human embryonic kidney 293 cell line (ECACC 85120602) and the human choriocarcinoma-derived placental JEG3 cell line (ECACC 92120308) were provided by CERDIC (Sophia-Antipolis, France). Cells were grown in phenol red-free EMEM containing 2 mM glutamine, 1% non-essential amino acid, 100 U/mL of antibiotics (mix of penicillin, streptomycin, and fungizone), and 10% fetal calf serum (Biowhittaker, Gagny, France). The JEG3 cell line was suplemented with 1 mM sodium pyruvate. Fifty thousand cells per well were grown at 37°C (5% CO2, 95% air) during 48 h to 80% confluence in 24-well plates, washed with serum-free EMEM and then exposed to various concentrations of Roundup (0.01, 0.05, 0.1, 0.5, 0.8, 1, 2%), or the equivalent concentrations of glyphosate, in EMEM serum free or not, for various times:1, 24, 48, and 72 h.
This enzymatic test, based on the cleavage of MTT into a blue-coloured product (formazan) by the mitochondrial enzyme succinate-dehydrogenase (Mossmann 1983), was used to evaluate human cell viability. Cells were washed with serum-free EMEM and incubated with 250 μL MTT per well after each treatment. The plates were incubated for 3 h at 37°C and 250 μL of 0.04 N-hydrochloric acid-containing isopropanol solution were added to each well. The plates were then vigorously shaken in order to solubilize the blue formazan crystals formed. The optical density was measured using a spectrophotometer (Stratagene, Strasbourg, France) at 560 nm for test and 720 nm for reference. The differential effects between glyphosate and Roundup are measured by the surfaces between the curves by the calculation of integrals.
Measurement of Aromatase Activity in Cells
Aromatase activity was evaluated according to the tritiated water release assay (Thompson and Siiteri 1974) with a slight modification as previously described (Dintinger et al. 1989). This method is based on the stereo-specific release of 1β-hydrogen from the androstenedione substrate, which forms tritiated water during aromatization. The 293 cells were transfected with the human aromatase cDNA (Auvray et al. 1998), exposed to nontoxic concentrations of glyphosate alone or Roundup, and were washed with serum-free EMEM and incubated for 45 min with 200 nM [1β-3H] androstenedione at 37°C (5% CO2, 95% air).
The reaction was stopped by placing the plates on ice for 5 min and then centrifuging at 2700g, at 4°C for 10 min. After adding 0.5 mL of charcoal/dextran T-70 suspension (7%:L5%), the mixture was left at 4°C for 5 min, and then centrifuged similarly. Supernatant fractions were assessed for radioactivity by scintillation counting.
Preparation of Microsomes
Microsomal fractions (containing endoplasmic reticulum) were obtained from full-term placentas of young healthy and nonsmoking women (Centre Hospitalier Régional de Caen, France) and equine testis by differential centrifugations (Moslemi et al. 1997). Briefly, tissues were washed with 0.5 M KC1, homogenized in 50 mM phosphate buffer (pH 7.4) containing 0.25 M sucrose and 1 mM Dithiothreiol DTT, and centrifuged at 20,000g. The supernatant was then ultracentrifuged at 100,000g, and the final pellet was washed twice, dissolved in the same buffer containing 20% glycerol, and stored at −70°C until use. All steps of the preparation were carried out at 4°C.
Measurement of Microsomal Aromatase Activity
Microsomal aromatase activity was evaluated by tritiated water release from radiolabelled substrate [1β-3H] androstenedione as described above. Human placental microsomes (50 μg proteins) were incubated with radiolabelled androstenedione (100 pmol/tube) at 37°C for 15 min, in the presence or absence of various concentrations of Roundup or glyphosate in 1 mL total volume of 50 mM Tris-maleate buffer, pH 7.4. The reaction was started by adding 100 μL of 60 μM H+-NADPH and stopped with 1.5 mL chloroform and then centrifuged at 2700g at 4°C for 5 min. After adding 0.5 mL of charcoal/dextran T-70 suspension (7%:1.5%) into the preparation, the centrifugation was repeated for 10 min. Aromatase activity was determined by measuring the radioactivity of 0.5 mL aqueous phase.
Purification of Aromatase Moieties and Measurement of Reductase Activity
Reductase was prepared by chromatographic separation using (ωD-aminohexyl-Sepharose 4B followed by adenosine 2′-5′-diphosphate-agarose, hydrophobic interaction, and affinity columns (Vibet et al. 1990), Protein concentration was determined as previously described (Bradford 1976). Reductase activity was determined by the measurement of the increasing absorbance of the preparation, corresponding to the reduction of the cytochrome C in the presence of H+-NADPH (Vibet et al. 1990) at 550 nm for 2 min at 20°C using a Kontron-Uvikon 860 spectrophotometer. The pH of the preparation was adjusted to 7.4 by adding an appropriate volume of 10 N NaOH. After equilibration, the reaction was started by adding cytochrome C.
The inactivation was carried out as previously described (Moslemi and Séralini 1997) by pre-incubation of equine testicular microsomes (200 μg proteins) for different times (0 to 30 min) at 20°C in a 0.5 mL final volume of 50 mM Tris-maleate buffer, pH 7.4, in the presence of saturating concentration of Roundup (11.6%) or in its absence (control). Androstenedione (400 nM) or H+-NADPH (60 μM) were included or not in the preincubation medium. After preincubation, the free Roundup and androstenedione were removed by adding 100 μL of charcoal/dextran T-70 suspension (2%:1%) into the medium. The mixture was then gently mixed and left at 4°C for 15 min; this was followed by a centrifugation at 350g at 4°C for 10 min. Residual aromatase activity was then evaluated by incubating 70 μL of the aqueous phase with 200 nM tritiated androstenedione for 15 min at 25°C, in 0.5 mL of 50 mM Tris-maleate buffer, pH 7.4, containing 60 μM H+-NADPH. The efficiency of Roundup adsorption by charcoal/dextran was previously tested without preincubation.
The experiments were repeated at least 3 times in different weeks on 3 independent cultures each time (n = 9). All data were presented as the mean ± standard error (S.E.M.). Statistical differences were determined by a Student t-test using significant levels with p < 0.01 (**) and p < 0.05 (*).
We tested the toxicity potential of Roundup on 293 cells derived from a human embryo, at doses (from 0.01 to 2%, i.e., containing 210 μM to 42 mM glyphosate among adjuvants) below that recommended for agricultural use. We tested its effect on cell viability up to 72 h in comparison to glyphosate. We also compared the results of similar exposures on human placental JEG3 cells. The Roundup dilutions and equivalent quantities of glyphosate were adjusted to the same pH, to avoid measuring a specific action of glyphosate acidity.
Aromatase Activity Inhibition
In this work, we demonstrated a cytotoxic effect of Roundup for the first time on human embryonic cells, as well as endocrine disruption in this new model at lower nontoxic levels. This major herbicide is used worldwide and composed of glyphosate and a mixture of various adjuvants. The 293 cells were shown to be suitable for the estimation of hormonal activity of xenobiotics after aromatase transfection (Kuiper et al. 1998), in particular since they are themselves deprived of steroidogenesis. Our results also confirmed and extended our previous study on the human placental JEG3 cell line (Richard et al. 2005). This cell line is considered to be a useful model for examining placental toxicity (Letcher et al. 1999). Our studies also revealed that the embryonic cells are more sensitive than the placental ones.
The use of transformed or cancer-derived cell lines allows longer experiments than in primary cultures; moreover, the established cell lines may be less sensitive to xenobiotics than their normal counterparts (L’Azou et al. 2005), but still we measure here important impacts of Roundup. In this case, the timing and effects measured may be more important in vivo if living tissues are exposed to comparable contamination. Of course, the metabolism in the body will moderate these actions. However, we demonstrate irreversible inhibition and the exposures are also often longer in vivo. Thus, our models offer at least a good indication of the potential toxicity of Roundup during agricultural use. We have also worked here with fresh human placenta to determine whether the endocrine disruption by Roundup observed in the cell lines could also be evoked in the microsomal fraction obtained from fresh, normal tissue.
When used in agricultural practice, the formulated concentrated commercial Roundup is diluted on the farm. The farmers are then often exposed to concentrated solutions (100%, i.e., 2.13 M glyphosate), and then during spraying to more diluted solutions, up to 1–2%, the latter corresponding to the maximal concentrations used on the cells in this work. Pregnant women with embryonic and placental cells could be exposed during repeated herbicide preparations and generally only a few precautions are applied, since Roundup is believed to be one of the most environmentally friendly pesticides (US EPA 1998, Williams et al. 2000).
Our data demonstrated that as little as 0.01% Roundup, within only 24 h, provoked a significant reduction of 19% of estrogen production in transfected 293 cells. Estrogens are known to be necessary for normal fetal development. This Roundup dose became toxic after 72 h of exposure. Serum-binding proteins, including albumin, can buffer the xenobiotic bioavailability (Seibert et al. 2002), as we have observed, and our serum-free cultures allowed a shortening of the experiments to mimic longer-term effects, since within 1 h we obtained results comparable to those after 1–2 days in serum.
The endocrine effect was linked to glyphosate, which was directly able to inhibit aromatase in cells, and in the microsomes formed not only by the endoplasmic reticulum out of placental fresh cells but also from equine testis. Glyphosate also inhibited aromatase activity independently of its acidity, and on both enzyme moieties (reductase and cytochrome P450 aromatase). However, the acidity presented very little partial impact in contrast to the formulation Roundup. This interaction was not only demonstrated to be direct with the aromatase active site (Richard et al. 2005), it was also found to be temperature-dependent in our work on enzymatic catalytic activity, and all these impacts were promoted by the adjuvants in all instances. It is also suggested that the adjuvants allow a better solubilization of glyphosate and the latter are more active with the increase of the temperature. An indirect pathway on the aromatase gene expression was also observed in JEG3 cells (Richard et al. 2005). The action of Roundup disturbing the transcriptional activity of another crucial enzyme has been demonstrated (Marc et al. 2005) for the hatching of the sea urchin eggs. In addition, when the cytotoxic effect was noticed in this work, it was due to disruption of the mitochondrial enzyme succinate-dehydrogenase, implicated in a cellular viability process.
Our models are then pertinent to the study of Roundup toxicity. If the agents that it contains bioaccumulate, in case of contamination of a pregnant woman, it is likely that the placenta and embryo will be reached by significant levels of those. Pesticide adjuvants and surfactants, which are present in Roundup, are used in herbicide formulations to favor stability and penetration of the active ingredient into cell membranes (Cox 1998). These adjuvants amplify the cellular effects of herbicides not only in plants but also in animal models (Marc et al. 2002, Walsh et al. 2000, Nosanchuk et al. 2001). Some of them may eventually stick to DNA and bioaccumulate in new and not usually detected forms (Peluso et al. 1998). Thus, partial Roundup elimination does not exclude the action of some metabolites at cellular levels, since at least some of the Roundup residues have been demonstrated to be strongly and more permanently bound to mammalian tissues.
This potential bioaccumulation could also induce or explain amplified effects with time. Thus, our results are in favour of the recruitment of other synergistic signalling pathways of action. This is why we have also analyzed the cell viability from 1 to 72 h, which has confirmed a drastic amplification of the cytotoxic effect of Roundup with time. Unfortunately, farmers are exposed often at least for many weeks to this regularly used product, which is also a common contaminant of rivers (Cox 1998). Considering all the data taken together, we cannot conclude like Williams et al. (2000) that the effects of the surfactants are antagonistic rather than synergistic.
Finally, we characterized in this work the differential sensitivity for Roundup and glyphosate of human embryonic cells, placental-derived cell lines, and fresh tissue extracts from human placenta and mammalian testis. Moreover, we confirmed the potential endocrine disruption of Roundup in all models on estrogen synthesis. As Roundup was more active than its claimed active ingredient in all instances, the formulation adjuvants probably allow a better cell penetration and stabilization of the product. Chemical mixtures in formulations may thus be underestimated regarding their toxic or hormonal impact (Tichy et al. 2002, Lydy et al. 2004, Monosson 2005). Most of the tests undertaken in a regulatory context are in fact performed with the active ingredient alone in vivo for one or two years (Williams et al. 2000). For instance, toxicity was not measured for Roundup treatments during more than 22 days with rats and rabbits. The potency for endocrine modulation was not assessed with the Roundup mixture at all, but only with glyphosate or POEA alone (Williams et al. 2000). Consequently, our experiments with Roundup should be also conducted on entire organisms in vivo. As emphasized by Brian et al. (2005), we can conclude that the failure to account for the combined effects, in particular with adjuvants, will undoubtedly lead to the underestimation of potential hazards, especially at the endocrine disruption level, and hence to erroneous conclusions at a regulatory level regarding the risk that they provoke. Thus, the toxic or hormonal impact of chemical mixtures in formulations appears to be underestimated.
This work was accomplished in the Biochemistry Laboratory EA2608 and the other authors’ affiliations represent their present postal address. For financial support and fellowships, we thank La Fondation Denis Guichard and The Human Earth Foundation, CRIIGEN (Committee for Independent Research and Information on Genetic Engineering), and le Conseil Regional de Basse-Normandie. The authors declare they have no competing financial interest.
- Cox C (1998) Glyphosate (Roundup). J Pest Reform 18:3–17Google Scholar
- Cox C (2004) Glyphosate. J Pest Reform 24:10–15Google Scholar
- Lydy M, Belden J, Wheelock C, Hammock B, Denton D (2004) Challenges in regulating pesticide mixtures. Ecol Society 9:1–15Google Scholar
- Monroy CM, Cortes AC, Sicard DM, de Restrepo HG (2005) Cytotoxicity and genotoxicity of human cells exposed in vitro to glyphosate. Biomedica 25:335–345Google Scholar
- Moslemi S, Séralini GE (1997) Inhibition and inactivation of equine aromatase by steroidal and non-steroidal compounds. A comparison with human aromatase inhibition. J Enzyme Inhib 12:241–254Google Scholar
- Nelson DR (1998) Cytochrome P450 nomenclature. Methods Mol Biol 107:15–24Google Scholar
- Savitz DA, Arbuckle T, Kaczor D, Curtis KM (1997) Male pesticide exposure and pregnancy outcome. Am J Epidemiol 146:1025–1036Google Scholar
- Thompson EA, Siiteri PK (1974) Utilization of oxygen and reduces nicotinamide adenine dinucleotide phosphate by human placental microsomes during aromatization of androstenedione. J Biol Chem 249:5364–5372Google Scholar
- Tichy M, Borek-Dohalsky V, Rucki M, Reitmajer J, Feltl L (2002) Risk assessment of mixtures: possibility of prediction of interaction between chemicals. Int Arch Occup Environ Health 75(Suppl):S133–S136Google Scholar
- U.S. Environmental Protection Agency (1998) Endocrine Disruptor Screening and Testing Advisory Committee (ECSTAC) Final Report, August 1998. U.S. Environmental Protection Agency, Washington, DCGoogle Scholar
- Yousef MI, Salem MH, Ibrahim HZ, Helmi S, Seehy MA, Bertheussen K (1995) Toxic effects of carbofuran and glyphosate on semen characteristics in rabbits. J Environ Sci Health B30:513–534Google Scholar