Alma de Nogal : Los Chalchaleros


Page 1

Chemical Research in Toxicology is published by the American Chemical Society.1155 Sixteenth Street N.W., Washington, DC 20036ArticleGlyphosate Formulations Induce Apoptosis and Necrosisin Human Umbilical, Embryonic, and Placental CellsNora Benachour, and Gilles-Eric Se#raliniChem. Res. Toxicol., 2009, 22 (1), 97-105 • DOI: 10.1021/tx800218n • Publication Date (Web): 23 December 2008Downloaded from on February 9, 2009More About This ArticleAdditional resources and features associated with this article are available within the HTML version:•Supporting Information•Access to high resolution figures•Links to articles and content related to this article•Copyright permission to reproduce figures and/or text from this article


Page 2

Chemical Research in Toxicology is published by the American Chemical Society.1155 Sixteenth Street N.W., Washington, DC 20036


Page 3

Glyphosate Formulations Induce Apoptosis and Necrosis in HumanUmbilical, Embryonic, and Placental CellsNora Benachour and Gilles-Eric Séralini*UniVersity of Caen, Laboratory Estrogens and Reproduction, UPRES EA 2608, Institute of Biology,Caen 14032, FranceReceiVed June 16, 2008We have evaluated the toxicity of four glyphosate (G)-based herbicides in Roundup (R) formulations,from 105times dilutions, on three different human cell types. This dilution level is far below agriculturalrecommendations and corresponds to low levels of residues in food or feed. The formulations have beencompared to G alone and with its main metabolite AMPA or with one known adjuvant of R formulations,POEA. HUVEC primary neonate umbilical cord vein cells have been tested with 293 embryonic kidneyand JEG3 placental cell lines. All R formulations cause total cell death within 24 h, through an inhibitionof the mitochondrial succinate dehydrogenase activity, and necrosis, by release of cytosolic adenylatekinase measuring membrane damage. They also induce apoptosis via activation of enzymatic caspases3/7 activity. This is confirmed by characteristic DNA fragmentation, nuclear shrinkage (pyknosis), andnuclear fragmentation (karyorrhexis), which is demonstrated by DAPI in apoptotic round cells. G provokesonly apoptosis, and HUVEC are 100 times more sensitive overall at this level. The deleterious effectsare not proportional to G concentrations but rather depend on the nature of the adjuvants. AMPA andPOEA separately and synergistically damage cell membranes like R but at different concentrations. Theirmixtures are generally even more harmful with G. In conclusion, the R adjuvants like POEA changehuman cell permeability and amplify toxicity induced already by G, through apoptosis and necrosis. Thereal threshold of G toxicity must take into account the presence of adjuvants but also G metabolism andtime-amplified effects or bioaccumulation. This should be discussed when analyzing the in vivo toxicactions of R. This work clearly confirms that the adjuvants in Roundup formulations are not inert.Moreover, the proprietary mixtures available on the market could cause cell damage and even deatharound residual levels to be expected, especially in food and feed derived from R formulation-treatedcrops.IntroductionHumans are exposed daily to a great number of xenobioticsand their metabolites, present as pollutants (1). They act asmixtures having compensatory, multiplicative, or synergisticeffects, as we have shown (2) with others (3, 4). The mainglyphosate (G) formulations, commercialized as Roundup (R)from the Monsanto Company, are themselves already mixturesof G and various adjuvants at different concentrations. We havestudied these products, which are the major nonselectiveherbicides worldwide (5); moreover, their use and presence inthe food chain (6) are increasing since more than 75% ofgenetically modified edible plants have been designed to toleratehigh levels of these compounds (7). G and its major metaboliteaminomethylphosphonic acid (AMPA) were classified amongthe first contaminants in rivers (8). The adjuvants, less measuredin the environment, are usually considered as inert and areprotected as a “trade secret” in manufacturing (9). However,among them, the predominant one appears to be the polyethoxy-lated tallowamine or POEA (10, 11), which has itself sometoxicity (12), such as causing ocular burns, redness, swellingsand blisters, short-term nausea, and diarrhea. In combinationwith G, the mixture becomes more active (13). These products,like detergents, could allow facilitated G penetration throughplasmatic membranes, potentialization of its action, increasedstability, and bioaccumulation (14, 15).The dose- and time-dependent cytotoxicity of R Bioforce (360g/L of G, R360) on human placental and embryonic cells (15)could explain at least in part some reproductive problems (16).Among the two lines, 293 embryonic cells have proven to bevery suitable for estimating the hormonal activity for xenobiotics(17), and JEG3 cells are also considered a useful model forexamining placental toxicity (18). These lines may be equallyor even less sensitive to xenobiotics than primary cultures (19).In the present study, we also tested the mechanism by which Rmixtures affect human primary cells of the umbilical vein cordendothelial cells (HUVEC) for comparative purposes.The endothelial lining of blood vessels constitutes a permeablebarrier between the blood and the underlying tissues. Theendothelium also plays an important role in various physiologi-cal processes, such as metabolism of vasoactive substances andmaintenance of antithrombotic factors on the vessel wall (20).The endothelial cells are exposed directly to chemicals circulat-ing in the blood of the umbilical cord and pass through theplacenta (21). It is known that HUVEC cells may be a targetfor adverse effects of xenobiotics activated into reactivemetabolites (22, 23). Other somatic cell types have been usedto study pesticide toxicity and apoptosis such as HeLa (24) andJurkat (25), but none was before treated by glyphosate.In human cells, we have demonstrated that G mixed withadjuvants in R360 was cytotoxic through alteration of succinatedehydrogenase SD (14, 15). With isolated rat liver mitochondria,it is demonstrated that R depresses the mitochondrial complexes* To whom correspondence should be addressed. Tel: 33(0)2-31-56-56-84. Fax: 33(0)2-31-56-53-20. E-mail: Res. Toxicol. 2009, 22, 97–1059710.1021/tx800218n CCC: $40.75 2009 American Chemical SocietyPublished on Web 12/23/2008


Page 4

II (SD) and III (26). In sea urchin eggs, R deteriorated cell cyclecheckpoints, and G with its adjuvants inhibited hatching enzymetranscription synergistically (27, 28). Recently, it was shownin this model to activate the DNA damage checkpoint CDK1/cycline B of the first cell cycle of development (29, 30) forcommitment to cell death by apoptosis in the case of failure ofDNA repair.This work focuses on the cell death mechanism in humancells induced by four different G formulations with a largenumber of agricultural applications. We have chosen RoundupExpress (R7.2), Roundup Bioforce or Extra 360 (R360),Roundup Grand Travaux (R400), and Roundup Grand TravauxPlus (R450) at subagricultural dilutions. We tested them on threeimportant enzymatic biomarkers. First, at the membrane level,we measured adenylate kinase (AK) activity after its release inthe medium (31), revealing cytoplasmic membrane rupture,corresponding to a necrosis and/or a secondary necrosis at theend of apoptosis (32). Second, at the mitochondrial respirationlevel, we measured succinate dehydrogenase (SD) activity (33).Third, we tested the cytosolic level with caspase 3 and 7activities to determine the apoptosis pathway (34-36) and insitu DNA fragmentation (DAPI). Necrosis is evinced bycytoplasmic swelling, rupture of the plasma membrane, swellingof cytoplasmic organelles (particularly mitochondria), and somecondensation of nuclear chromatin, whereas apoptosis is mani-fested by cytoplasmic and nuclear condensation (pyknosis),nuclear fragmentation (karyorrhexis), normal morphologicalappearance of cytoplasmic organelles, and an intact plasmamembrane; following nuclear fragmentation, the cell disaggre-gates into a number of membrane-bound apoptotic bodies(37, 32). By contrast, cell death is now known to be perpetratedthrough a variety of mechanisms. It can be classified into fourdifferent types, based upon morphological characteristics: apo-ptosis (type 1), autophagy (type 2), necrosis (oncosis, type 3),and mitotic catastrophe (37).The three human cell types allowed us to establish not onlythe differential sensitivity of these models but also the generalhuman cell pathways of G-based pesticides actions from 1 ppm(0.0001%); these were produced by G itself, its major metaboliteAMPA, and the main adjuvant POEA, singly or in combination.Materials and MethodsChemicals. N-Phosphonomethyl glycine (glyphosate, G, PM169.07) and its major metabolite AMPA (PM 111.04) werepurchased from Sigma-Aldrich (Saint Quentin Fallavier, France).Herbicide Roundup formulations (Monsanto, Anvers, Belgium)were available on the market: Roundup Express 7.2 g/L of G,homologation 2010321 (R7.2); Bioforce or Extra 360 at 360 g/Lof G, homologation 9800036 (R360); Grands Travaux 400 g/Lof G, homologation 8800425 (R400); and Grands Travaux plus450 g/L of G, homologation 2020448 (R450). A 2% solution ofRoundup (1 or 2% is recommended by the company foragricultural use) and an equivalent solution of glyphosate toRoundup Bioforce were prepared in serum-free medium andFigure 1. Cytotoxic effects of four Roundup formulations (R) on three human cell types. The R (from 10 to 2 × 104ppm) contain differentglyphosate (G) concentrations (7.2, 360, 400, or 450 g/L) and adjuvants. G alone was used as control at equivalent quantities to R360 and at similarpH 5.8. The cells were either primary from neonate umbilical cord (HUVEC) or lines from embryo (293) or placenta (JEG3). The actions on themitochondrial succinate dehydrogenase (SD) activity (cellular viability in %, A) and on the release of cytoplasmic adenylate kinase (AK) activity[cell death in relative luminescence units (RLU), B] were compared in serum-free medium after 24 h of exposure. The 50% lethal dose (LD50) isindicated by a dashed line. SEs are shown in all instances (n)12).98 Chem. Res. Toxicol., Vol. 22, No. 1, 2009Benachour and Séralini


Page 5

adjusted to pH 5.8 of the 2% Roundup Bioforce solution. Themajor adjuvant of Roundup, polyethoxylated tallowamine (POEAat 785 g/L), was a gift from Pr. Robert Bellé (UMR 7150 CNRS/UPMC, Station Biologique de Roscoff, France). Successivedilutions were then obtained with serum-free medium. 4′,6′-Diamidino-2-phenylindole, dihydrochloride (DAPI) nucleic acidstain powder was obtained from Lonza (Saint Beauzire, France).3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide(MTT) and all other compounds, otherwise precised, wereobtained from Sigma-Aldrich. MTT was prepared as a 5 mg/mL stock solution in phosphate-buffered saline, filtered througha 0.22 µm filter before use, and diluted to 1 mg/mL in a serum-free medium.Cell Cultures. Human Primary Cells. The human primary cellsused in this work were HUVEC (C2519A) provided by Lonza. Cells(passage 5 or 6) were grown according to the supplier, in specificendothelial growth medium EGM-2 SingleQuots (CC-4176) con-taining hEGF, hydrocortisone, GA-1000 (Gentamicin, Amphoteri-Figure 2. Nonlinear dose effects of R formulations. The LD50(%) measured by SD are compared for the 4 R (see the Figure 1A legend) and Gfor the three cell types in similar conditions.Figure 3. Cytotoxicity of R adjuvant (POEA) and glyphosate (G) metabolite (AMPA) on three human cell types. G and R360 were used as controlsin similar conditions as in Figure 1 (see legend), in comparison to R adjuvant POEA and G metabolite AMPA (1-105ppm). The 50% lethal dose(LD50) is indicated by a dashed line. SEs are shown in all instances (n)12).Glyphosate Formulations Toxicity in Human CellsChem. Res. Toxicol., Vol. 22, No. 1, 2009 99


Page 6

cin-B), FBS (fetal bovine serum), VEGF, hFGF-B, R3-IGF-1,ascorbic acid, and heparin. Fifty thousand cells per well were grownat 37 °C (5% CO2, 95% air) over a 24 h period to 80% confluencein 48 well plates and were washed with serum-free EGM-2.Human Cell Lines. The human embryonic kidney 293 cell line(ECACC 85120602) and the human choriocarcinoma-derivedplacental JEG3 cell line (ECACC 92120308) were provided byCERDIC (Sophia-Antipolis, France). Cells were grown in phenolred-free Eagle’s modified minimum essential medium (EMEM;Abcys, Paris, France) containing 2 mM glutamine, 1% nonessentialamino acid, 100 U/mL antibiotics (a mix of penicillin, streptomycin,and fungizone; Lonza), 10 mg/mL of liquid kanamycin (DominiqueDutscher, Brumath, France), and 10% FBS (PAA, les Mureaux,France). The JEG3 cell line was supplemented with 1 mM sodiumpyruvate. Fifty thousand cells per well were grown at 37 °C (5%CO2, 95% air) over a 48 h period to 80% confluence in 48 wellplates and were washed with serum-free EMEM.Cell Treatments. Cells were exposed for 24 h in serum-freemedium to various dilutions of the different treatments includingthe four Roundup formulations (R7.2, R360, R400, and R450), G,AMPA, or POEA (14 concentrations from 10 ppm to 2%) and,particularly for POEA, were tested at the very low concentrationsof 1 and 5 ppm; for AMPA, we tested in addition 4, 6, 8, and10%. In another case, cells were incubated with G, AMPA, andPOEA mixtures by pairs at the final nontoxic dilution on SD of0.5% on the human cell lines (293 or JEG3) and 0.05% on thehuman primary cells (HUVEC) in comparison to R360.For the details, in each cell type, three combinations were studied.For the two cell lines, the first mixture was the combination of G(0.4999%) with POEA (0.0001%); the second was the combinationof G (0.4%) with AMPA (0.1%), and the third was AMPA(0.4999%) plus POEA (0.0001%). For the primary HUVEC cells,the first mixture was G (0.04999%) with POEA (0.0001%); thesecond was G (0.04%) with AMPA (0.01%), and the third wasAMPA (0.04999%) plus POEA (0.0001%).Cell Death Measurements. Mitochondrial Activity Mea-surement. This measure was based on the cleavage of MTT intoa blue-colored product (formazan) by the mitochondrial enzymesuccinate dehydrogenase (38, 39, 33); it was used to evaluate humancell viability. After cell treatments, the supernatants were recoveredfor the ToxiLight bioassay, and adherent cells were washed withserum-free medium and incubated with 200 µL MTT per well aftereach treatment. The 48 well plates were incubated for 3 h at 37°C, and 200 µL of 0.04 N hydrochloric acid-containing isopropanolsolution was added to each well. The plates were then vigorouslyshaken to solubilize the blue formazan crystals formed. The opticaldensity was measured at 570 nm using a luminometer Mithras LB940 (Berthold, Thoiry, France).Cell Membrane Damage Assay. The bioluminescent ToxiLightbioassay (Lonza) was a nondestructive cytotoxicity highly sensitiveassay designed to measure toxicity in mammalian cells and celllines in culture. It quantitatively measured the release of cytosolicAK from the membranes of damaged cells (40, 31). AK is a robustprotein present in all eukaryotic cells, which is released into theculture medium when cells die, described as an important necrosismarker. The enzyme actively phosphorylated ADP, and the resultantATP was then measured using the bioluminescent firefly luciferasereaction with the ToxiLight reagent. After 24 h of differenttreatments, 50 µL of cell supernatants was deposited in 96 wellblack plates. Then, 50 µL of the AK detection reagent (AKDR)was added by well. Plates were then placed under agitation for 15min safe from the light, and then, luminescence was measured usingthe luminometer Mithras LB 940 (Berthold) at 565 nm. The serum-free medium was the negative control, and a positive control wasthe active reagent AKDR mixed with cells treated in the serum-free medium to determine the basal activity.Figure 4. Combined effects of G, AMPA, and POEA on three human cell types. The cells were incubated in serum-free medium for 24 h, and theproducts were tested by pairs to a final concentration, where they are nontoxic alone on succinate dehydrogenase, of 0.05 (HUVEC) and 0.5% (293,JEG3). Results of cellular death are evaluated through AK activity in relative units in comparison to nontreated cells (control)1), and values areblank-subtracted (blank)no AK); see the Materials and Methods. R360 and G are used as controls. SEs are shown in all instances (n)16;**p<0.01).100 Chem. Res. Toxicol., Vol. 22, No. 1, 2009Benachour and Séralini


Page 7

Apoptotic Cell Death Measurements. The Caspase-Glo 3/7assay (Promega, Paris, France) was a luminescent kit designed forautomated high-throughput screening of caspases activity or apo-ptosis. It can measure caspase 3 and 7 activities in purified enzymepreparations or cultures of adherent or suspension cells (41, 42, 36).The assay provided a pro-luminescent caspase 3/7 substrate, whichcontains the tetrapeptide sequence DEVD active group. Thissubstrate was cleaved to release amino-luciferin, a substrate ofluciferase used in the production of light. The Caspase-Glo 3/7reagent was optimized for caspase activity, luciferase activity, andcell lysis. The addition of the single Caspase-Glo 3/7 reagent, inan “add-mix-measure” format, resulted in cell lysis followed bycaspase cleavage of the substrate and generation of a “glow type”luminescent signal. The Caspase-Glo 3/7 bioassay was carried outin 96 well white plates.After cell cultures and their treatments by 50 µL of variousdilutions, an equal volume of the reagent was added to each well.Plates were then agitated for 15 min safe from the light, to stabilizethe light signal before measuring luminescence. Again, the negativecontrol was the serum-free medium, and the positive control wasthe active reagent mixed with cells treated in the serum-free mediumto determine the basal activity of the caspases 3/7. Luminescencewas measured using the luminometer Mithras LB 940 (Berthold)at 565 nm.Cell Microscopy. At the end of the 24 h cell treatment, theserum-free medium was removed, and cells were fixed in absoluteethanol-chloroform-acetic acid (6:3:1, v/v/v) for 1 day at-20°C. Each well was washed with PBS (pH 7.4) and incubated with1 µg/mL DAPI solution (43). Staining of DNA with DAPI wasexamined with a microscope using a fluorescent mode (model LeïcaLMD 6000, Rueil Malmaison, France). Labeled DNA of viablecells was scattered throughout the nucleus, and bright condensationof chromatin revealed apoptotic cells (magnification, 400×). Atthe end of the cell treatment, the microphotographs (magnification,100×; blue filter) of cells without coloration were also obtainedwith the Leïca Microscopy Systems (model Leïca DC 100,Germany).Statistical Analysis. The experiments were repeated at least threetimes during different weeks on three independent cultures eachtime. All data were presented as the means(standard errors (SEs).Statistical differences were determined by a Student’s t test usingsignificant levels of 0.01 (**).ResultsWe have studied for the first time the mechanism of cellularaction of different R on human cells, from placenta, embryonickidney, and neonate. The first surprising results show that thefour R herbicides and G cause cellular death for all types ofhuman cells, with comparable toxicity for each one but atdifferent concentrations. For instance, 20 ppm for R400 at 24 h,Figure 5. Time-dependent apoptosis through caspases 3/7 induction by R and G in three human cell types. R360 and G, at similar concentrationsand pH (as in Figure 1), were incubated for 6, 12, 18, or 24 h. The apoptotic pathway was tested by the Caspase-Glo 3/7 assay, and results arepresented in relative units to nontreated cells (control)1). SEs are shown in all instances (n)8).Glyphosate Formulations Toxicity in Human CellsChem. Res. Toxicol., Vol. 22, No. 1, 2009 101


Page 8

the most toxic, corresponds approximately to 47 µM G (8 ppm)with adjuvants (Figure 1). However, 4-10 ppm G alone isnontoxic; its toxicity begins around 1%. The mechanism isconstant for all R: There is a release of AK, indicative of cellmembrane damage, and an inhibition of the mitochondrial SD(Figure 1). For all R, the membrane damage (AK) is 1.5-2times more sensitive than mitochondrial activity (SD) for 293and JEG3 or equally sensitive for HUVEC. By contrast, Ginduces mitochondrial toxicity without cell membrane damage.Unexpectedly, R400 is more toxic than another formulationcontaining more G, such as R450; the latter is in turn moreharmful than R360, R7.2, and G in last, but all of them aredetrimental nonproportionally to the G concentration that theycontain. This is illustrated in Figure 2.The mitochondrial SD inhibition measures cell asphyxia. Itis obvious from Figure 2 that 7.2 or 360 g/L G with adjuvantsin R formulations has closely comparable actions on cell death,while 400 or 450 g/L gives inversely proportional effects inanother range. This is not an artifact since the embryonic andplacental cell lines behave remarkably similarly in that regardand the primary umbilical cord cells have sensitivity for all Rand G just analogous to these cell lines (Figure 2). The mortalityin all cases is not linearly linked to G. The hypothesis that othersubstances are implicated has thus to be investigated in theformulation of the product.Consequently, the major G metabolite, AMPA, and thesurfactant POEA, the main claimed adjuvant by the manufac-turer (the exact composition is a secret of formulation), havebeen tested separately in a first approach, in comparison to Gand R360 as controls, and in similar conditions as in Figure 1,from very low subagricultural dilutions (10-6if used pure likeclaimed by some farmers and 10-4if diluted as recommendedat 1%).G is claimed by the manufacturer to be the active ingredient,and it is claimed to be not toxic for human cells but toxic forvegetable ones when mixed with inert components. Our studydemonstrates for the first time that all products including AMPAand POEA provoke SD and AK effects in human cells, andthus mortality (Figure 3), but at different concentrations.Astonishingly, the supposed inert product POEA is the mostpotent one. From 1 ppm, it begins to alter SD in HUVEC andAK in 293 and JEG3. The mixture R is then more poisonousthan G or AMPA. The metabolite AMPA itself destroys thecell membrane (AK release), whatever the cell type. This isnot observed with G, which is, however, 3-8 times moreinhibitory on SD than AMPA, with some differences betweencells. However, because the cell membrane damage is generallymore sensitive, the metabolite AMPA is finally more toxic thanG on human cells. POEA is the most toxic; if it was the onlyadjuvant of R360, its maximal concentration would be around1-24 ‰, according to the cells. Thus, POEA could beconsidered as the active ingredient on human cell death andmore damaging than G. As R is more viscous than 1‰ POEAplus G, it is obvious that other compounds are in the mixture.Thus, it was necessary to study the combined effects on cellmembrane integrity (by AK release). We have tested thecompounds by pairs at maximal levels where alone they do notinfluence SD (Figure 4). This was to assess the respective roleof each one, knowing that R contains all tested compounds whenmetabolized. In contrast to previous results, the cells reacteddifferently. The mixtures were more disrupting on embryonicand umbilical cells, respectively, while placental carcinoma cellsappeared to be more membrane-resistant but to mixtures only.It is very clear that if G, POEA, or AMPA has a small toxiceffect on embryonic cells alone at low levels, the combinationof two of them at the same final concentration is significantlydeleterious (Figure 4).We have thus elucidated that R- and G-induced cell deathcan be due, at least in part, to apoptosis via caspases 3/7induction (Figure 5). The caspases are activated from 6 h witha maximum at 12 h in all cases, but umbilical primary cells are60-160 times more sensitive than lines (293 and JEG3,respectively) at this level. Moreover, G and R360 enhanceexactly at the same concentration caspases, from 50 ppm(HUVEC). The adjuvants do not appear to be necessary torender G as a death inducer at this level. Even G alone is 30%more potent on this pathway than R. Surprisingly, G acted veryrapidly at concentrations 500-1000 times lower than agriculturaluse on human cell apoptosis. This apoptotic pathway was alsoactivated at levels 200 times lower for G on caspases than itsaction on SD for umbilical cells, and for R at levels 60 timeslower, in a four times shorter period (6-24 h). After 24 h oftreatment, the caspases returned to basal level when SD andAK react significantly. These data are consistent with a gradualloss of caspases 3/7 activity in apoptotic cells that undergosecondary necrosis in vitro (44).Our results are confirmed by the morphology of the cells aftertreatment by R (for instance R400, Figure 6B,D,E) in compari-son with the normal cell types (A, C, F). Indeed, the very weakR concentration of 0.005% causes a very important cell death,lack of adhesion, shrinking, and fragmentation in apoptoticbodies. This is confirmed in Figure 7 with the DNA fluorescentlabeling with DAPI, for example, with R360 at 0.5% over 24 h.The characteristic fluorescence of apoptotic cells evidencingFigure 6. Microphotographs of R-treated human cells. The cell typeswere without coloration (magnification, 100×; blue filter), HUVEC(A, B), 293 (C, D), and JEG3 (E, F), and were incubated with 0.005%of R400 or not (controls) in serum-free medium for 24 h. Micropho-tographs were obtained with the Leïca Microscopy Systems (modelLeïca DC 100).102 Chem. Res. Toxicol., Vol. 22, No. 1, 2009Benachour and Séralini


Page 9

DNA condensation is more visible with the herbicide than incontrols (A, D, G) and more after R treatment (C, F, I) thanwith G alone (B, E, H), for cell lines. The primary cells aresimilarly sensitive to G than to R, as for caspases activation inFigure 5.DiscussionWe had previously demonstrated (14) that G-based formula-tions were able to affect human placental cell viability atsubagricultural doses (0.1% in 18 h) and sexual steroidbiosynthesis at lower nontoxic doses (0.01%) and that this wasdue at least in part to G, but its action was highly amplified byadjuvants, the so-called inert ingredients of R formulations, keptconfidential by the companies (9). However, the question of aspecific cell line action or a time reversible effect remained open.Benachour et al. (15) demonstrated that in embryonic cells aswell as in normal human placenta and equine testis, there wasa similar G-dependent endocrine disruption, through aromataseinhibition, at nontoxic levels. The embryonic cells were evenmore sensitive: It was discovered that the cell mitochondrialactivity was also reached in time- and dose-dependent mannersby the G formulation R360. The cytotoxicity was amplifiedaround 14 times between 24 and 72 h (15), suggesting either abioaccumulation or a time-delayed effect and suggesting acumulative impact, after endocrine disruption, of very low dosesaround G acceptable daily intake (ADI: 0.3 mg/kg/j), accordingto the nature of the adjuvants.To understand in vivo effects through the interpretation ofthe in cell impacts described above, it is necessary to haveknowledge of the dilution and of the processes leading to anelimination of the product in the body. This must be taken intoaccount in regard to its bioaccumultation potential and time-delayed effects. This is why we have measured the caspasesactivities at different times and G or R concentrations, afterhaving previously demonstrated their effects amplified with timewithin 3 days, on SD in embryonic and placental cells (15).Moreover, the metabolism of the herbicide has to be considered,and the tests in this study of all the above-cited productsapproach this question.All cell types, including primary cultures, react similarly atthe membrane and mitochondrial level, justifying the hypothesisthat the cell lines used provide excellent models to study humancell toxicity, for instance in placental cells (18). We show forthe first time that embryonic and umbilical cells also havecomparable sensitivity. The most reactive level reached appearsto be the cell membrane level for the different formulations,but not for G. The supposed “inert ingredients” play obviouslyand differently the role of cell membrane disruptors, indepen-dently to G, as we have previously proposed (14), and this wassuggested in fish, amphibians, and microorganisms (27, 45) orin plants (46). We now demonstrate that in human cells.The second level is the mitochondrial membrane and theenzymatic reaction in it, SD, localized in the internal membranein complex II of the respiratory chain (47). It is altered in acomparable way, not proportional to G but relatively to thenature and the quantity of the adjuvants that we have previouslylisted (15). This means that the toxicity of G clearly varies withformulations that must imperatively now be used in in vivo teststo study any toxicity (45); this also means that the ADI of Gmust take into account its formulation, since 7.2 or 360 g/L ofG may have comparable effects, considerably different to 400g/L. It would even be more correct to use precisely an ADI ofR instead of G. It may also be time-dependent. These ideas arenot taken into account yet for regulatory legislation.The necessity to study combined effects also appears fromour results. In fact, the body is always exposed to mixtures andnot to single compounds. We have previously demonstrated thatmixtures could amplify toxicity for other widely spread pollut-ants (2). For embryonic or neonatal cells, POEA, the majoradjuvant, has the highest toxicity, either by itself or amplified2-5 times in combination with G or AMPA. It has already beenshown that POEA is highly toxic for sea urchin embryos,impinging on transcription (28). It is also known that in anFigure 7. Increase of DNA condensation (DAPI test) in R360- or G-treated human cells. The cell types HUVEC (A-C), 293 (D-F), and JEG3(G-I) were incubated for 24 h with or without 0.5% R360 or G at equivalent concentrations. Staining of DNA with DAPI was examined with amicroscope model Leïca LMD 6000, using a fluorescent mode. Labeled DNA of viable cells was scattered throughout the nucleus, and brightcondensation of chromatin revealed apoptotic cells (magnification, 400×).Glyphosate Formulations Toxicity in Human CellsChem. Res. Toxicol., Vol. 22, No. 1, 2009 103


Page 10

aquatic environment, POEA has higher effects than R and Gon bacteria, microalgae, protozoa, and crustaceans (12). Inaddition, the known metabolism of G in the soil or plants issupposed to detoxify it in AMPA (11); however, here, wedemonstrate that AMPA is more toxic than G in human cells,especially on cell membrane. AMPA is also more stable in soil(48), in plants, and in food or feed residues (49), and morepresent in wastewater (2-35 ppm) than G [0.1-3 ppm; (50)].It is not toxic alone at these concentrations in our experiments,but it amplifies G or POEA toxicity in combination. Thesynergic toxicity of all of these compounds is now more obvious.The induced resistance of placental cancer cells (51) couldexplain a specific difference for JEG3 cells at these levels. Theplacental cells could form an efficient barrier to mixtures beforetheir death, since the membranes are more resistant, and thiscould be due to the fact that there are carcinoma-derived cellsthat have acquired a capacity to excrete xenobiotics.The caspases 3/7 inducing apoptosis were in fact activatedfirst within 6 h, and then, they decreased with cell mortality.This corroborates the timing observed for another compound(36). The caspases induction by G alone is observed at dosesthat do not provoke cell or mitochondrial membrane damages,indicating a clear G-apoptotic pathway always at subagriculturaldoses. Mixed with adjuvants, G in R formulations reached theother end points. This suggests that the adjuvants could alsoplay a role in total cell death, through necrosis characterizedby organelle alterations with mitochondrial and cell membranesswelling and ruptures (52). The most sensitive are umbilicalHUVEC cells, for which apoptosis has been described (53-55),but very rarely induced by a pesticide, for example, in the caseof diallyl trisulfide (56). Surprisingly, this phenomenon wasobserved for G and R at similar and low concentrations, as if acell membrane death receptor was activated (57, 32), with noG penetration necessary. The modification of a dependencyreceptor is another pathway that could be studied (58). Theapoptotic cell appearance was microscopically confirmed. Then,our next step could be to study the necrotic/apoptotic ratio withinshort times.In conclusion, mixtures called “formulations” change cellpermeability, toxicity, and pathways of xenobiotics: In all cases,cell death is induced more by R than by AMPA or G, and thelatter provokes apoptosis (from 50 ppm in HUVEC cells)without membrane damage. By contrast, G mixed with adjuvantsin R formulations disrupts cell and mitochondrial membranesand promotes necrosis. It becomes obvious that the “threshold”level of action of the herbicide should take into account theperiod and length of exposure, the presence of adjuvants, inparticular POEA, metabolism, and bioaccumulation or time-delayed effects. All of the above effects are demonstrated belowthe recommended herbicide agricultural dilutions (from 104ppm). This clearly confirms that the adjuvants in Roundupformulations are not inert. Moreover, the proprietary mixturesavailable on the market could cause cell damage and even deatharound residual levels to be expected, especially in food andfeed derived from R formulation-treated crops.Acknowledgment. We thank CRIIGEN, Regional Councilof Basse-Normandie, and Human Earth Foundation. This workwas also supported by a grant from Fondation Denis Guichardunder the aegis of the Fondation de France. We declare that wehave no competing financial interest. We thank Dr. CarineTravert for scientific revision of the manuscript.References(1) Feron, V. J., Cassee, F. R., Groten, J. P., van Vliet, P. W., and vanZorge, J. A. (2002) International issues on human health effects ofexposure to chemical mixtures. EnViron. Health Perspect. 110, 893–899.(2) Benachour, N., Moslemi, S., Sipahutar, H., and Séralini, G. E. (2007)Cytotoxic effects and aromatase inhibition by xenobiotic endocrinedistrupters alone and in combination. Toxicol. Appl. Pharmacol. 222,129–140.(3) Tichy, M., Borek-Dohalsky, V., Rucki, M., Reitmajer, J., and Feltl,L. (2002) Risk assessment of mixtures: Possibility of prediction ofinteraction between chemicals. Int. Arch. Occup. EnViron. Health 75,S133-S136.(4) Monosson, E. (2005) Chemical mixtures: Considering the evolutionof toxicology and chemical assessment. EnViron. Health Perspect. 113,383–390.(5) Acquavella, J. F., Bruce, H., Alexander, B. H., Mandel, J. S., Gustin,C., Baker, B., Champan, P., and Bleeke, M. (2004) Glyphosatebiomonitoring for farmers and their families: Results from the farmfamily exposure study. EnViron. Health Perspect. 112, 321–326.(6) Takahashi, M., Horie, M., and Aoba, N. (2001) Analysis of glyphosateand its metabolite, aminomethylphosphonic acid, in agriculturalproducts by HPLC. Shokuhin Eiseigaku Zasshi 42, 304–308.(7) Clive, J. (2007) The global status of the commercialized biotechno-logical/genetically modified crops: 2006. Tsitol. Genet. 41, 10–12.(8) IFEN (2006) Report on pesticides in waters: Data 2003-2004. InstitutFrançais de l’Environnement, Orleans, France. Dossier 5, 15–20.(9) Cox, C. (2004) Glyphosate. J. Pest. Reform. 24, 10–15.(10) Acquavella, J. F., Weber, J. A., Cullen, M. R., Cruz, O. A., Martens,M. A., Holden, L. R., Riordan, S., Thompson, M., and Farmer, D.(1999) Human ocular effects from self-reported exposures to Roundupherbicides. Hum. Exp. Toxicol. 18, 479–86.(11) Williams, G. M., Kroes, R., and Munro, I. C. (2000) Safety evaluationand risk assessment of the herbicide Roundup and its active ingredient,glyphosate, for human. Regul. Toxicol. Pharmacol. 31, 117–65.(12) Tsui, M. T., and Chu, L. M. (2003) Aquatic toxicity of glyphosate-based formulations: comparison between different organisms and theeffects of environmental factors. Chemosphere 52, 1189–1197.(13) Cox, C. (1998) Glyphosate (Roundup). J. Pest. Reform. 18, 3–17.(14) Richard, S., Moslemi, S., Sipahutar, H., Benachour, N., and Séralini,G. E. (2005) Differential effects of glyphosate and Roundup on humanplacental cells and aromatase. EnViron. Health Perspect. 113, 716–720.(15) Benachour, N., Sipahutar, H., Moslemi, S., Gasnier, C., Travert, C.,and Séralini, G. E. (2007) Time and dose-dependent effects of Roundupon human embryonic and placental cells and aromatase inhibition.Arch. EnViron. Contam. Toxicol. 53, 126–133.(16) Savitz, D., Arbuckle, T., Kaczor, D. M., and Curtis, K. (1997) Malepesticide exposure and pregnancy outcome. Am. J. Epidemiol. 146,1025–1036.(17) Kuiper, G. G., Lemmen, J. G., Carlsson, B., Corton, J. C., Safe, S. H.,Van der Saag, P. T., Van der Burg, B., and Gustafsson, J. A. (1998)Interaction of estrogenic chemicals and phytoestrogens with estrogenreceptor β. Endocrinology 139, 4252–4263.(18) Letcher, R. J., Van Holstein, I., Drenth, H. J., Norstrom, R. J., Bergman,A., Safe, S., Pieters, R., and van den Berg, M. (1999) Cytotoxicityand aromatase (CYP19) activity modulation by organochlorines inhuman placental JEG-3 and JAR choriocarcinoma cells. Toxicol. Appl.Pharmacol. 160, 10–20.(19) L’Azou, B., Fernandez, P., Bareille, R., Beneteau, M., Bourget, C.,Cambar, J., and Bordenave, L. (2005) In vitro endothelial cellsusceptibility to xenobiotics: Comparison of three cell types. Cell Biol.Toxicol. 21, 127–137.(20) Thorin, E., and Shreeve, S. M. (1998) Heterogeneity of vascularendothelial cells in normal and disease states. Pharmacol. Ther. 78,155–166.(21) Peters, R. J. B. (2005) Man-Made Chemicals in Maternal and CordBlood. TNO report B&O-A R 2005/129.(22) Annas, A., Granberg, A. L., and Brittebo, E. B. (2000) Differentialresponse of cultured human umbilical vein and artery endothelial cellsto Ah receptor agonist treatment: CYP-dependent activation of foodand environmental mutagens. Toxicol. Appl. Pharmacol. 169, 94–101.(23) Choi, W., Eum, S. Y., Lee, Y. W., Hennig, B., Robertson, L. W., andToborek, M. (2003) PCB 104-induced proinflammatory reactions inhuman vascular endothelial cells: Relationship to cancer metastasisand atherogenesis. Toxicol. Sci. 75, 47–56.(24) Rathinasamy, K., and Panda, D. (2006) Suppression of microtubuledynamics by benomyl decreases tension across kinetochore pairs andinduces apoptosis in cancer cells. FEBS J. 273, 4114–4128.(25) Kannan, K., Holcombe, R. F., Jain, S. K., Alvarez-Hernandez, X.,Chervenak, R., Wolf, R. E., and Glass, J. (2000) Evidence for the104 Chem. Res. Toxicol., Vol. 22, No. 1, 2009Benachour and Séralini


Page 11

induction of apoptosis by endosulfan in a human T-cell leukemic line.Mol. Cell. Biochem. 205, 53–66.(26) Peixoto, F. (2005) Comparative effects of the Roundup and glyphosateon mitochondrial oxidative phosphorylation. Chemosphere 61, 1115–1122.(27) Marc, J., Mulner-Lorillon, O., Boulben, S., Hureau, D., Durand, G.,and Bellé, R. (2002) Pesticide Roundup provokes cell divisiondysfunction at the level of CDK1/cyclin B activation. Chem. Res.Toxicol. 15, 326–331.(28) Marc, J., Le Breton, M., Cormier, P., Morales, J., Bellé, R., andMulner-Lorillon, O. (2005) A glyphosate-based pesticide impinges ontranscription. Toxicol. Appl. Pharmacol. 203, 1–8.(29) Bellé, R., Le Bouffant, R., Morales, J., Cosson, B., Cormier, P., andMulner-Lorillon, O. (2007) Sea urchin embryo, DNA-damaged cellcycle checkpoint and the mechanisms initiating cancer development.J. Soc. Biol. 201, 317–27.(30) Le Bouffant, R., Cormier, P., Cueff, A., Bellé, R., and Mulner-Lorillon,O. (2007) Sea urchin embryo as a model for analysis of the signalingpathways linking DNA damage checkpoint, DNA repair and apoptosis.Cell. Mol. Life Sci. 64, 1723–1734.(31) Squirrell, D., and Murphy, J. (1997) Rapid detection of very lownumbers of micro-organisms using adenylate kinase as a cell marker.A Practical Guide to Industrial Uses of ATP Luminescence in RapidMicrobiology, pp 107-113 Cara Technology Ltd., Lingfield, Surrey,United Kingdom.(32) Taatjes, D. J., Sobel, B. E., and Budd, R. C. (2008) Morphologicaland cytochemical determination of cell death by apoptosis. Histochem.Cell. Biol. 129, 33–43.(33) Scatena, R., Messana, I., Martorana, G. E., Gozzo, M. L., Lippa, S.,Maccaglia, A., Bottoni, P., Vincenzoni, F., Nocca, G., Castagnola,M., and Giardina, B. (2004) Mitochondrial damage and metaboliccompensatory mechanisms induced by hyperoxia in the U-937 cellline. J. Biochem. Mol. Biol. 37, 454–459.(34) Los, M., Wesselborg, S., and Schulze-Osthoff, K. (1999) The role ofcaspases in development, immunity, and apoptotic signal transduction:Lessons from knockout mice. Immunity 10, 629–639.(35) Omezzine, A., Chater, S., Mauduit, C., Florin, A., Tabone, E., Chuzel,F., Bars, R., and Benahmed, M. (2003) Long-term apoptotic cell deathprocess with increased expression and activation of caspase-3 and -6in adult rat germ cells exposed in utero to flutamide. Endocrinology144, 648–661.(36) Liu, J. J., Wang, W., Dicker, D. T., and El-Deiry, W. S. (2005)Bioluminescent imaging of TRAIL-induced apoptosis through detec-tion of caspase activation following cleavage of DEVD-aminoluciferin.Cancer Biol. Ther. 4, 885–892.(37) Galluzzi, L., Maiuri, M. C., Vitale, I., Zischka, H., Castedo, M.,Zitvogel, L., and Kroemer, G. (2007) Cell death modalities: Clas-sifications and pathophysiological implications. Cell Death Differ. 14,1237–43.(38) Mosmann, T. (1983) Rapid colorimetric assay for cellular growth andsurvival: Application to proliferation and cytotoxicity assays. J. Im-munol. Methods 65, 55–63.(39) Denizot, F., and Lang, R. (1986) Rapid colorimetric assay for cellgrowth and survival modifications to the tetrazolium dye proceduregiving improved sensitivity and reliability. J. Immunol. Methods 89,271–277.(40) Crouch, S. P. M., Kozlowski, R., Slater, K. J., and Fletcher, J. (1993)The use of ATP bioluminescence as a measure of cell proliferationand cytotoxicity. J. Immunol. Methods 160, 81–88.(41) O’Brien, M., Moravec, R., Riss, T., and Promega Corporation (2003)Caspase-Glo 3/7 assay: Use fewer cells and spend less time with thishomogeneous assay. Cell Notes 6, 13-15.(42) Ren, Y. G., Wagner, K. W., Knee, D. A., Aza-Blanc, P., Nasoff, M.,and Deveraux, Q. L. (2004) Differential regulation of the TRAIL deathreceptors DR4 and DR5 by the signal recognition particle. Mol. Biol.Cell 15, 5064–5074.(43) Travert, C., Carreau, S., and Le Goff, D. (2006) Induction of apoptosisby 25-hydroxycholesterol in adult rat Leydig cells: Protective effectof 17beta-estradiol. Reprod. Toxicol. 22, 564–570.(44) Riss, T. L., and Moravec, R. A. (2004) Use of multiple assay endpointsto investigate the effects of incubation time, dose of toxin, and platingdensity in cell-based cytotoxicity assays. Assay Drug DeV. Technol.2, 51–62.(45) Cox, C., and Surgan, M. (2006) Unidentified inert ingredients inpesticides: Implications for human and environmental health. EnViron.Health Perspect. 114, 1803–806.(46) Haefs, R., Schmitz-Eiberger, M., Mainx, H. G., Mittelstaedt, W., andNoga, G. (2002) Studies on a new group of biodegradable surfactantsfor glyphosate. Pest. Manage. Sci. 58, 825–833.(47) Thomas, P. K., Cooper, J. M., King, R. H., Workman, J. M., Schapira,A. H., Goss-Sampson, M. A., and Muller, D. P. (1993) Myopathy invitamin E deficient rats: Muscle fibre necrosis associated withdisturbances of mitochondrial function. J. Anat. 183, 451–461.(48) Gard, J. K., Feng, P. C., and Hutton, W. C. (1997) Nuclear magneticresonance timecourse studies of glyphosate metabolism by microbialsoil isolates. Xenobiotica 27, 633–644.(49) Scalla, R. (1997) Sécurité alimentaire biotransformation des herbicidespar les plantes transgéniques résistantes. Oléagineux, Corps Gras,Lipides. Dossier: Génie Génét. Oléagineux 4, 113–119.(50) Ghanem, A., Bados, P., Kerhoas, L., Dubroca, J., and Einhorn, J.(2007) Glyphosate and AMPA analysis in sewage sludge by LC-ESI-MS/MS after FMOC derivatization on strong anion-exchange resinas solid support. Anal. Chem. 79, 3794–801.(51) Xu, J., Peng, H., and Zhang, J. T. (2007) Human multidrug transporterABCG2, a target for sensitizing drug resistance in cancer chemo-therapy. Curr. Med. Chem. 14, 689–701.(52) Okada, H., and Mak, T. W. (2004) Pathways of apoptotic and non-apoptotic death in tumour cells. Nat. ReV. Cancer 4, 592–603.(53) Solovyan, V. T., and Keski-Oja, J. (2005) Apoptosis of humanendothelial cells is accompanied by proteolytic processing of latentTGF-beta binding proteins and activation of TGF-beta. Cell DeathDiffer. 12, 815–826.(54) Song, Y., Shen, K., and He, C. X. (2007) Construction of autocatalyticcaspase-3 driven by amplified human telomerase reverse transcriptasepromoter and its enhanced efficacy of inducing apoptosis in humanovarian carcinoma. Zhonghua Fu Chan Ke Za Zhi 42, 617–622.(55) Qiu, J., Gao, H. Q., Li, B. Y., and Shen, L. (2008) Proteomicsinvestigation of protein expression changes in ouabain inducedapoptosis in human umbilical vein endothelial cells. J. Cell. Biochem.104, 1054–1064.(56) Xiao, D., Li, M., Herman-Antosiewicz, A., Antosiewicz, J., Xiao, H.,Lew, K. L., Zeng, Y., Marynowski, S. W., and Singh, S. V. (2006)Diallyl trisulfide inhibits angiogenic features of human umbilical veinendothelial cells by causing Akt inactivation and down-regulation ofVEGF and VEGF-R2. Nutr. Cancer 55, 94–107.(57) Ryu, H. Y., Emberley, J. K., Schlezinger, J. J., Allan, L. L., Na, S.,and Sherr, D. H. (2005) Environmental chemical-induced bone marrowB cell apoptosis: Death receptor-independent activation of a caspase-3to caspase-8 pathway. Mol. Pharmacol. 68, 1087–1096.(58) Kroemer, G., Galluzzi, L., and Brenner, C. (2007) Mitochondrialmembrane permeabilization in cell death. Physiol. ReV. 87, 99–163.TX800218NGlyphosate Formulations Toxicity in Human CellsChem. Res. Toxicol., Vol. 22, No. 1, 2009 105