28 chemicals


The aim of this work is to introduce a method that is considered reliable for assessing the toxicity of chemical substances. This goal of safety for human health cannot be achieved through the use of standard toxicology methods, which are still largely based on animal experiments. No animal species is a biological model for another species, including man. This was clearly demonstrated in the second half of the 20th century’s discoveries in the field of genetics. We can now define a species in terms of its reproductive isolation which results from the fact that genes and chromosomes are unique to each species and hence preclude reproduction between different species. Genes determine protein synthesis and biological functions. As any given species has a unique set of genes, it follows that those biological functions will be species-specific.

Penicillin kills guinea pigs whereas it has saved millions of humans ; aspirin can cause birth defects in rats or dogs whereas this has not been observed in humans. Many more examples illustrate the fact that no animal species is a reliable model for another. Claude Bernard, the physiologist who, in the 19th century, promulgated animal experiments as the main source of knowledge for human medicine, stressed the similarities between animals and humans, rather than the differences. He also justified his use of animals on ethical grounds, as wanting to avoid experiments that could harm humans.

We are now in the 21st century and possess many non-invasive methodologies for the study of human medicine. Toxicological studies can be performed on cultured human cells. In addition, for the past 15 years, we also possess powerful new tools with which to monitor cells exposed to a given chemical – DNA chips. These chips can reveal which genes are active in a cell at a given time, even as soon as 24 hours after the cells have been exposed to the test chemical. Knowing the function of these genes, we can observe the response of the cell to the chemical and whether this response drives the cell towards a pathological state. Toxicogenomics is the study of the genes expressed following exposure to a chemical.

Antidote Europe has developed a novel approach to toxicogenomics by using minituarized DNA chips, in combination with sequential exposure of two different cell types, approximating what would occur in a whole body. We have termed this approach “Scientific Toxicology Program” (STP) and what follows are the results of 28 substances tested according to the STP.

Selection of test substances

France is the second largest consumer of pesticides in the world. One hundred and fifty thousand people a year die in France from cancer. A further 600,000 persons suffer from Alzheimer’s disease and, in common with other European countries, 15% of couples are infertile. Several renowned researchers and medical scientists have pointed out that the widespread use of chemicals is very likely the main cause of these shocking health statistics. Greenpeace and the WWF have shown that dozens of chemicals are present in the air we breathe in our homes. Worse still, chemicals in the blood of pregnant mothers are ‘off-loaded’ on to the unborn foetus in the womb. Yet, chemical manufacturers maintain that the link between chemical use and disease has not been established.

In 2001, the European Commission launched the REACH (Registration, Evaluation and Authorization of CHemicals) project, claiming that little or no safety data existed with respect to almost 100,000 chemical substances produced by industry. Following mounting pressure from manufacturers, this number was subsequently reduced to 30,000 chemicals.

While acknowledging the need for such a project, Antidote Europe cautions against the use of unreliable test methods for assessing toxic risk. Classical toxicological methods still largely rely on animal experiments, which, as we have explained, are not reliable for humans. Instead, we urge the European Commission to incorporate STP as an integral part of its testing strategy, based on human exposure data. Many observations have been made following occupational or accidental exposure to chemicals. This should be employed as a reference with respect to both the toxicity of these chemicals and as a standard for the validation of relevant toxicological methods. Since our aim was not to present the regulatory authorities with data of unknown chemicals but simply to demonstrate the validity of STP, we selected 28 chemicals already known or suspected of being toxic. We therefore selected 15 pesticides (abamectin, aldicarb, aldrin, carbaryl, chlorpyriphos, dicofol, fenazaquin, fipronil, heptachlor, lindane, methoxychlor, paraquat, permethrin, phosmet and rotenone), 2 food additives (benzoic acid E210 and quinoline E104), 5 cosmetic ingredients (3-aminophenol, 4-aminobiphenyl, 2-butoxyethanol, benzophenone-3 and propyl paraben E214), 1 prescription drug (acetaminophen paracetamol) and 5 additional substances commonly used in industry (1,4-dioxane, acetonitril, acrylamide, bisphenol A and ethylene glycol). Some of these substances are multi-purpose. Although some of the pesticides have been banned in many European countries, they may still be present in our environment and in our water supply, thereby contaminating our bodies.

What is the STP?

The first report on STP was published in Biogenic Amines, 2003, vol. 18, pp 41-54. STPis based on human cell cultures, DNA chips and knowledge of genetics.

Culturing human cells has been routine in labs since about 1930. Some centers, especially in Germany, the UK and the US, keep many different cell lines obtained from human tissues and organs. These cells lines are commercially available. In spite of being cancerous cells, their characteristics are well known and they are used as biological models for research worldwide. Our study used HepG2 liver cells and SH-SY5Y neuronal cells. Once the culture box was covered by one layer of cells (a monolayer of cells), the chemical to be tested was introduced and left in the culture medium for 24 or 48 hours, at two different concentrations.

We chose liver cells because the liver, with its detoxifying function, is crucially in contact with all chemicals circulating in the blood, and will metabolize these chemicals in order to facilitate their elimination, although this metabolism can result in substances becoming even more dangerous than the original, depending on the enzymatic capability of the liver. It is important to note that the enzymatic capability can differ significantly between different animal species, which is one of the reasons for the failure of animal-based toxicology with respect to humans. We chose neuronal cells because many insecticides target the nervous system of their victims and wanted to observe this effect on human cells.

A novel idea of our scientific team was to expose liver and neuronal cells sequentially rather than in parallel, thus approximating the whole body physiological response, in which the nervous system and other tissues and organs are exposed to the metabolites circulating in the blood once the liver has modified the original substances. And indeed we observed that some of the tested chemicals did not elicit much response from the liver while they (or rather their metabolites) significantly affected the neuronal cells.

Another novel idea was to construct minituarized DNA chips containing 6 families of genes known to be implicated in the 6 metabolic pathways we had selected for our study. These pathways were already known for their association to particular diseases. Extensive literature already exists regarding the role of each of these genes in normal cell metabolism, cycle and function, as well as in pathological states. In contrast with the huge number (many thousands of genes), expensive and time-consuming processes of DNA chips developed in the US, for example, our cost-effective and very easy to process minituarized DNA chips provide a key solution to the problem of dealing with large automated platforms capable of testing tens of thousands of substances (and many of their combinations) in reasonable time and cost frames.

Our DNA chips contained 51 genes comprising 6 families, and designed for the study of :

1. Cell stress: 5 genes for monitoring the response to oxidative stress (GSS, GPX1, SOD1, GSTM3 and EPHX1), 2 genes implicated in the survival of a cell in a stress situation (TRPM2 and HSPA9B) and 2 genes implicated in the inflammatory response (PTGS2 Vioxx’s target and NOS2A). If these genes are elicited following the introduction of the substance to the culture medium, it means the cell is under duress, and making attempts to repair cell damage. The cell may subsequently ‘commit suicide’ if it cannot cope with the damaging effects of the substance. According to the kind of damage and if systemic mechanisms are not available for repair, exposed individuals could develop different diseases: cancer, due to free radical activity, inflammation or auto-immune disease, etc.

2. DNA damage: 3 genes implicated in DNA repair (RAD50, RAD51 and NFKB1), 3 genes preventing cell cycle (cell life and reproduction) following damage (CDC25C, CDK4 and CDKN1) and 3 genes triggering apoptosis (cell death) if the damage is not repaired (APAF1, ATM and BAX). If these genes are themselves impaired, cells with damaged DNA will eventually grow and multiply, resulting in cancer or developmental abnormalities, for example, if systemic repair mechanisms are not available.

3. Cell cycle control: DNA replicates before cell division gives birth to two cells, each receiving one copy of the original DNA. Many proteins have crucial roles in the accurate and quality control of DNA replication as well as in authorizing and controlling the progression of the cell cycle from step to step. Our chips harboured 2 genes controlling cell proliferation (FOS and JUN) and 7 genes implicated in stopping cell division and apoptotic signal (BCL2, GADD45A, MDM2, TP53, EGF, PPARA and TUBA1). Abnormal expression of these genes has been observed in cancer.

4. Neurotoxicity: 8 genes implicated (but not exclusively) in nerve transmission and nervous system development (ACHE, CTSB, DRD2, TH, BZRP, THBS1, HOXD1 and ROBO1). Abnormal expression of these genes, in the absence of compensating mechanisms, could lead to neurological or even psychiatric disorders, or malformations of nervous, muscular or cardiac systems if gene expression is affected during foetal or early childhood development.

5. Hormonal response: ranging from glycaemia to reproduction, hormones regulate many physiological functions. We have selected 10 genes under hormonal control (TFF1, CTSD, PGR, RAN, AR, CREB1, ESR1, CALR, CYP19A1 and ALB). Some of these genes code for steroid hormone receptors and their deregulation could explain some of the increase in cases of genital malformation, reduced male fertility and even breast or prostate cancer.

6. Protein conformation: the tridimensional structure of proteins is crucial for their ability to perform their function. Correct amino acid sequence but wrong folding is thought to lead to accumulation of non functional proteins contributing to conformational diseases such as Alzheimer’s, type I diabetes, etc. Our chips harboured 6 genes implicated in quality control of newly synthesized and cell loading of proteins (HSPA5, XBP1, ATF6, ERN1, C12orf8 and A2M).


Full results are available on our website or upon request from Antidote Europe. Here, we present a brief summary of the reactions elicited by the tested chemicals.

1. 1,4-dioxane: 3 out of the 6 marker families were impaired in liver cells, all 6 families were impaired in neuronal cells, especially the cell stress markers. This substance can lead to all 6 of the pathological pathways explored.

2. 2-butoxyethanol: 41 of the 51 genes present on our DNA chips were significantly repressed (for some of them, gene expression was reduced a hundred-fold). All studied cellular functions were severely compromised in both cell lines.

3. 3-aminophenol: 36 of the 51 genes were repressed, ten-fold for some of them. All studied cellular functions were compromised in both cell lines.

4. 4-aminobiphenyl: 4 genes were overexpressed in liver cells (markers for cell stress,DNA damage, cell cycle control and hormonal response) ; 25 genes were repressed in neuronal cells, in all 6 pathological pathways explored.

5. Abamectin: 44 of the 51 genes were strongly repressed, a hundred-fold in some cases. 27 genes were repressed in both cell lines. This insecticide (targeting GABAsynapses) has an unequivocal pathological potential.

6. Acetaminophen (paracetamol): 30 genes were repressed quite strongly, 2 genes were overexpressed, affecting all cellular functions under study in both cell lines.

7. Acetonitril: 2 genes were repressed and 3 overexpressed in liver cells (markers for cell stress, DNA damage, cell cycle control, neurotoxicity and hormonal response). 15 genes were repressed in neuronal cells, affecting all studied cellular functions.

8. Benzoic acid (E210): 38 of the 51 genes were repressed, significantly in some cases. All studied cellular functions were disturbed in both cell lines.

9. Acrylamide: 3 genes were overexpressed, 30 were repressed, significantly in some cases. In neuronal cells, the response was mainly observed after exposure to the highest dose. In contrast, liver cells were also affected at low doses. All studied functions were affected in both cell lines.

10. Aldicarb: expression of 20 genes was affected in both cell lines and a further 24 genes were affected in one or the other cell line, totalling 44 markers elicited. All studied cellular functions were affected in both cell lines.

11. Aldrin: 14 genes were affected in liver cells, mainly overexpressed. 11 genes were affected in neuronal cells, of which, 2 were dramatically overexpressed. All studied cellular functions were affected in both cell lines, with very different reactions for each gene.

12. Benzophenone-3: 2 genes were repressed (cell stress and neurotoxicity) and 1 overexpressed (DNA damage) in liver cells. 16 genes were repressed in neuronal cells, affecting all studied functions in this cell line.

13. Bisphenol A: 1 gene was repressed in liver cells after exposure to high dose. 7 genes were overexpressed in liver cells after exposure to low dose. 20 genes were repressed in neuronal cells, affecting all studied functions.

14. Carbaryl: 4 genes were overexpressed and 7 were repressed in liver cells, affecting all studied functions in this cell line. The effect was dramatic on neuronal cells, in which 48 genes were significantly repressed (a hundred-fold or more).

15. Chlorpyriphos: expression of 42 genes was affected in both cell lines and a further 7 genes were affected in one or the other cell line, totalling 49 markers elicited out of the 51 harboured by our DNA chips. Genes were repressed except for one neurotoxicity marker, which was overexpressed in neuronal cells.

16. Dicofol: only one gene (a neurotoxicity marker) was slightly repressed in liver cells. But 41 genes were repressed in neuronal cells, affecting all studied functions in this cell line.

17. Ethylene glycol: 29 genes were repressed in liver cells, 48 genes were repressed in neuronal cells, affecting all studied functions in both cell lines.

18. Fenazaquin: 22 genes were strongly repressed in liver cells, 11 genes were significantly repressed in neuronal cells, affecting all studied functions in both cell lines.

19. Fipronil: 2 cell stress markers and 2 DNA damage markers were affected in both cell lines ; 1 neurotoxicity and 1 hormonal response marker were repressed in liver cells ; 1 cell cycle control, 2 hormonal response and 1 protein conformation marker were repressed in neuronal cells.

20. Heptachlor: 1 DNA damage marker was overexpressed in liver cells. 17 genes were repressed in neuronal cells, affecting all studied functions mostly after low dose but long exposure time.

21. Lindane: 11 genes were repressed in both cell lines and another 32 genes were repressed in one or the other cell line, totalling 43 markers repressed, thus affecting all studied functions in both cell lines.

22. Methoxychlor: 20 genes were affected in both cell lines and a further 23 genes were repressed in one or the other cell line, totalling 43 markers elicited, thus affecting all studied functions in both cell lines. Except for 1 overexpressed neurotoxicity marker in liver cells, fairly significant repression was observed for all the other markers.

23. Paraquat: 46 genes were strongly repressed in both cell lines, affecting all studied functions. Most dramatic effects were seen for DNA damage and cell cycle control markers. For the long time exposure, most of the elicited markers in neuronal cells were repressed a hundred-fold.

24. Permethrin: 7 genes were affected in liver cells, with respect to cell stress, DNAdamage and hormonal response functions. Neuronal cells were more significantly affected, with 33 genes repressed, especially after long exposure. All studied functions were affected in neuronal cells.

25. Phosmet: 2 genes wre overexpressed and 11 repressed in liver cells. 41 genes were significantly repressed in neuronal cells. All studied functions were affected in both cell lines.

26. Propyl paraben (E214): 6 out of 6 studied functions were affected in both cell lines. Neurotoxicity markers were only slightly elicited in liver cells.

27. Quinolin (E104): 5 out 6 studied functions were affected in liver cells, with most of elicited markers repressed two-fold. All studied functions were affected in neuronal cells.

28. Rotenone: 29 genes repressed in liver cells, 27 in neuronal cells. The most dramatic effects (hundred-fold repression) were observed on neuronal cells after long exposure.

These being the first studies done with the STP, we do not have all the keys for translating laboratory observations into specific prediction of disease risk. However, we clearly see that each substance elicits a specific response. In order to accurately hypothesize possible effects on the individual, many parameters would have to be taken into account. For example, water-soluble substances will easily circulate in body fluids and be quickly eliminated whereas lipid-soluble ones will be stored in adipocytes (fat cells) and can also remain in the lipidic sheaths of the neuronal cells, thus posing a risk to potential target cells and tissues over long periods of time. Whereas a substance capable of damaging the central nervous system must cross the blood-brain barrier, a xenoestrogen will easily reach its target simply by being transported in the blood, as is the case with steroid hormones.

Effects already observed in humans

Clues about the validity of STP can be found by considering what is known about the tested substances following human accidental or occupational exposure. According to the International Agency for Research in Cancer :

  • workers exposed to 4-aminobiphenyl have developed bladder cancer
  • urinary tract cancer has been observed in acetaminophen users in Australia ; others have reported hepatotoxicity of this drug
  • nervous system problems have been observed in workers exposed to acrylamide
  • aldicarb has already been shown to induce DNA damage and mutations in human cultured cells
  • chromosomal aberrations have been induced by aldrin in cultured human lymphocytes (white blood cells) ; aldrin inhibited intercellular communication in human cell systems
  • slight excess of lung cancer cases was observed in workers exposed to heptachlor
  • 4 cases of leukaemia were reported in men exposed to lindane ; cases of aplastic anaemia have also been associated with exposure to lindane.
  • P. Darbre et al have found parabens in tissue samples from breast tumors and state that these chemicals are known xenoestrogens (J Appl Toxicol, 2004, 24, 1-4).
  • Professor Charles Sultan states, after a study on exposure to pesticides, that babies born to farmers have twice the normal risk of presenting genital malformations.


All of this data is a small part of a much bigger picture since cancer, neurological disorders and most of other diseases under consideration may have very long latent periods of onset, and therefore the link between the triggering substance and the disease will not be established as exposure will have ceased long before the appearance of the first symptoms. However initial these observations, they are nevertheless consistent with what could be hypothesized as the chemical’s effects on the basis of our STP results.

We can improve STP

The results presented here are the very first obtained with STP based on limited resources. Many improvements are possible.

1. Cell types. For substances likely to be absorbed through inhalation or feeding, lung and gut cells should be tested as these organs will be exposed first. Blood and kidney cells should also be tested routinely. For very important chemicals (prescription drugs, for example), STP could be a first step before microdosing tests in healthy volunteers and could give more valuable information than animal tests, after which only 1 drug out of 12 makes it to the market. STP is potentially a more powerful tool than QSAR and other simulations since STP is not a simulation, but a direct observation of what is taking place in human cells.

2. Markers. The human genome contains about 25,000 genes. About 1100 are known to respond to chemical exposure. Although the 51 set selected for our DNA chips have key functions in the processes we intended to study, there is still room for a greater number of markers on chips designed to monitor unexpected effects of substances.

3. Time and dose. Our results represent a snapshot of the state of gene expressions 24 and 48 hours after the introduction of the test substance into the culture medium. It would be interesting to observe more snapshots at shorter, intermediate and longer periods of exposure. The same is true for concentrations of test compounds, especially for suspected xenoestrogens as the endocrine system is sensitive to infinitesimal doses.


As imperfect as it is, STP has already proven much more reliable that animal-based toxicology and should replace it at once. Carcinogenicity tests, which take several months and yield unreliable results in rodents, can be performed in a few days with STP, in conjunction with other tests (neurotoxicity, immunotoxicity, acute toxicity, etc.), thereby sparing the lives of countless animals. STP has the advantage of dealing with human genes, thus allowing the identification of sensitive individuals or groups of individuals to a particular substance, based on our knowledge of human polymorphism, unique to humans and not predictable through animal-based toxicology.

With STP, automatization and optimisation of tests is possible at the very early stages of a new chemical’s development, thus allowing early screening and disqualification of dangerous or ineffective substances, with consequent gain of time and money. The short test times employed by STP would be able to cope with the original 100,000 chemicals initially included in the REACH project, and even many combinations of molecules (not feasible using on animals), in a reasonable period of time (much less than the 12 years considered by the European Commission for only 30,000 substances with the current means).

With toxicogenomics programs receiving substantial funding in the US and Japan, Europe should seriously consider allocating resources to improve STP whose novel approach could put the EU at the cutting edge of this promising new technology. These techniques are likely to become the standard in the coming years, and manufacturers should help to implement them rather than spending yet money on useless animal-based tests – and before public opinion loses confidence in an industry that avoids using the best available technology to assess the safety of its products.