Contact : 25 rue Jacques Callot - 66000 Perpignan - France
Tel : 33 (0)4 68 80 53 32
Lundi et jeudi de 10h30 à 18h / Monday and Thursday from 10h30 to 18h
Scientific Toxicity Assessment of pesticides, drugs
and other chemicals
in : Biogenic Amines, Vol. 18, No. 1, pp.41-54 (2003) Claude
REISS (1), André MENACHE (2), Simone PARVEZ (3) and Hasan PARVEZ (3)
Abstract - People living in developed countries are exposed to over 100 000 chemically pure, man-made substances (and an immense number of their combinations). 98% of these chemicals have never been tested for their effects on our health or environment. Traditional toxicity testing in animal models has proven unreliable. With the help of concepts, methods and tools developed in modern biology, scientific toxicity assessment is at present possible. Various fields of toxicology are reviewed and the scientific methods allowing their reliable assessments are summarized. The implementation of scientific toxicity testing is expected to save yearly millions of lives in developed countries.
European Union authorities are right in deciding that the 100,000 man-made chemicals to which we are exposed must be assessed for their adverse effects on our health and our environment. However, the prospect that these assessments will be carried out at the manufacturers' expense, that they will involve a very large number of animals and that they will not be completed for several decades at least, is cause for concern. Those most concerned are consumer organisations and private charities whose aim it is to promote human health. Some of these groups have challenged the scientific validity of such observations and results, since they would be obtained from animal-based research, and contend that the adverse effects of chemicals can only be reliably assessed for the species tested and are unreliable for any other species, humans in particular. They claim also that health statistics data fully support their case.
It is in fact
striking to observe the huge progress made over the past century by
the chemical, petrochemical, agrochemical and pharmaceutical industries
in the development of products, whilst the methods for assessing the
risks of chemical products have remained almost unchanged and still
depend exclusively on animal models. It is therefore timely, before
proceeding with the project of risk assessments of the 100 000 chemicals,
to carefully examine whether recent scientific progress could provide
benefits, especially by improving consumer safety. To this end, we
will examine the following six topics:
MORBIDITY AND MORTALITY STATISTICS IN THE EU AND OTHER DEVELOPED COUNTRIES: ALARMING TRENDS
Despite the fact that life expectancy in EU countries is high, this benefit is offset by high morbidity rates. Several million EU citizens suffer debilitating neurodegenerative conditions (Alzheimer's and Parkinson's disease, multiple sclerosis (MS), autism, etc.). Although for most of these diseases the increase in the number of cases correlates roughly with the increase in life expectancy, a rapid increase in neurodegenerative conditions has been observed among people between 20 and 40 years of age (MS in particular) and even in children (autism). The steepest rise in morbidity and mortality has, however, been seen in cancer. In France, for instance, the primary cause of death since 1990 for people aged 35 to 65 years has been cancer. The proportion of deaths due to all cancers, except lung cancer, among people aged 40 to 45 years has increased six-fold between 1950 and 1980, and 300 000 new cancers are diagnosed annually, with a significant increase in the number of cancer cases likely to be linked to hormones. One woman out of thirteen was diagnosed as having breast cancer in 1970. Today, it is one in seven.
It is generally agreed that 5-10% of all cancers are linked to genetic defects, and this figure has remained fairly constant. Hence, exogenous factors, especially lifestyle (smoking, alcohol, dietary excess, etc) and carcinogenic products present in our food and/or in our environment are responsible for nine out of ten cancers. Lifestyle tends to become more reasonable. According to the WHO, smoking among adult males dropped from 50% in 1950 to 30% at present. Balanced diets have been intensively promoted over the past decades, although obesity (a factor indirectly favouring cancer, because many cancerogenic products tend to accumulate in adipous tissues, and exposing local cells persistently to high carcinogene doses) tends to increase. It is therefore likely that environmental carcinogens are now the main culprits responsible for causing the 1.7 million new cancers diagnosed in EU countries annually. This is clear evidence that these products have either not been tested for their carcinogenic potential or have been tested by methods which have failed to detect this danger. To assess the efficacy of these methods, let us examine how they perform in an area where they are applied most stringently: the assessment of prescription drug toxicity. Despite the fact that many years of research are invested in any particular drug development and testing, adverse drug reactions (side effects) rank as the fourth leading cause of death in the EU, claiming 20 000 lives annually in France (and some 120 000 lives in the EU as a whole). Incautious or indiscriminate prescriptions of drugs may contribute to these figures, but GMPs can hardly be blamed for that, as they usually follow strictly drug instructions provided by the manufacturer.
It is obvious, then, that current testing methods are failing to protect public health efficiently enough. What are these methods? As required by law, toxicity testing in general, and for prescription drugs in particular, must be performed on animals, i.e. 'models' which are believed to display biological reactions similar to those of humans. It is therefore worthwhile analysing the relevance of the 'animal model' concept in relation to human health.
IS RESORTING TO ANIMAL MODELS FOR HUMAN-HEALTH ISSUES BASED ON RATIONAL PRINCIPLES?
There is remarkably simple yet clear proof that no animal species can substitute as a reliable biological model for another species. A species is defined in terms of its reproductive isolation, meaning that members from different species cannot interbreed. This is because a given species has its own unique genetic make-up (from number, organisation and structure of chromosomes, through to regulation and control of gene expression). Modern biology has clearly demonstrated that the genetic make-up of an individual determines the precise biological activities of its cells, tissues, and organs. Hence, individuals from different species have different genetic make-ups and therefore display different biological activities, even if some appear similar in the short term. The statement that members of a given species can substitute as reliable biological models for other species is therefore invalid.
In particular, the assumption that results obtained in some mammalian species are valid for humans is unfounded and seriously compromises human health. Consider, for instance, the chimpanzee, our closest relative (in evolutionary terms). If exposed to the human immunodeficiency virus (HIV), the chimpanzee does not respond - in humans it causes AIDS; if injected with the hepatitis B virus, one out of ten or so chimpanzees might develop a mild form of hepatitis and will recover quickly - in humans, the virus causes chronic hepatitis and sometimes liver cancer; and the chimpanzee, when injected with the Ebola virus, dies of haemorrhagic fever, as do humans. In other words, the best animal model ever behaves in an opposite, different or identical way to humans, depending on the challenging factor. Nobody could have forecasted these results, which can only be arrived at after observing the test in both species. Testing animal models is therefore useless at best and is at worst dangerous, sometimes fatal, to humans: the French blood scandal occurred because 'experts', noting that the chimpanzee showed no response, approved of HIV-contaminated blood samples going onto the market.
A conservative estimate of the number of deaths in France, resulting from this flawed methodology of testing both prescription drugs (responsible for 20 000 deaths) and carcinogenic products alone (which claim an estimate of 50% to 70% of the 150 000 deaths due to cancer), ranges from 100 000 to 120 000 a year. Assuming that similar rates per capita are valid in other EU nations, some 600 000 to 750 000 citizens die prematurely within the EU year after year, just because of side effects of drugs and carcinogenic products present in our environement.
WHAT IS THE BASIS OF VALID TOXIC RISK ASSESSMENT FOR HUMANS?
Resorting to animals for assessing toxic risks in humans goes back to medieval times, as it was then the only way to get a vague indication of the risk. At present however, the scientific revolution provides us with far more reliable means for toxic risk assessment. This science-based toxicology is built on two basic concepts.
First, toxicology is the science of life in an environment of toxic products (xenobiotics). Over the past half-century, biology has made unprecedented leaps, moving away from empiricism and towards science, and even almost exact science when it comes to cellular and molecular biology. Toxicology can benefit from the concepts, methods and tools developed in such modern biology and thereby achieve also the status of an almost exact science.
Second, the cell is where life starts. It is therefore not surprising (and fully supported by modern biology) that the answers to practically all biological problems must first be sought at the level of the cell. Human diseases almost invariably have a cellular origin, whether the cause is endogenous (in the organism) or exogenous (outside the organism). This holds true for cancer, neuropathologies and cardiovascular diseases, to cite the most frequent and life-threatening diseases in EU countries. It follows that harm done to the cell by a toxic substance might be the first step to diseases linked to the body's environment (including food, polluting substances etc).
The stage is thus set for a science-based toxicology (SBT), as opposed to the traditional toxicity assessment by means of animal models. SBT has its roots in modern molecular and cellular biology, from which it selects and adapts the methods and tools best suited for its goals.
The study of human cells in the environment of toxic products will therefore be the first step for reliable toxicity assessment in humans. Modern biology has also made impressive progress in the study of integrated systems at the tissue, organ and systemic levels. Non-invasive methods (various tomographies, functional testing of biochemical activities in organs, etc.) are available, allowing us to complete at the systemic level molecular and cellular human-risk assessments, in particular for substances to which consumers are extensively exposed (prescription drugs, food additives, pesticides, etc.).
SCIENCE-BASED TOXICOLOGY (SBT)
Our purpose here is not to advance a detailed scientific programme of SBT but to sketch the main outlines of molecular and cellular SBT, based on fifteen years of experience in this field. Indeed, not only have some of the scientists who contributed to this report been active in carrying out research and developing methods and tools for SBT, but they have also organised international workshops aimed at bringing together world-class specialists in their own particular fields of molecular and cellular toxicology - namely, First European Workshop in Molecular Toxicology, Sophia-Antipolis, France (1996), Second European Workshop in Molecular Toxicology, Paris (1999) - and published the proceedings of these workshops: Molecular Toxicology, (1997) (VSP publishing, Reiss, Parvez and Labbe Editors), Molecular Responses to Xenobiotics, (2001) (Elsevier publishing, Parvez, Reiss and Labbe Editors) and Special Issue of Toxicology 153 (2000, numbers 1-3, Guest Editors Parvez and Reiss).
Toxicity responses can be acute and systemic; they can also be delayed, either because of an accumulation of minor damages, which manage finally to overcome cellular defence and repair mechanisms (a major cause of liver and kidney diseases) or because the development of the disease takes a long time following its induction. A typical example of this delay is cancer, as it takes on average five to ten years between the onset of the proliferation of a cell and the diagnosis of the resulting tumour. Neurodegenerative conditions follow a similar time-course. Hence the necessity to assess both the short- and long-term toxic responses.
Cellular studies are ideally suited to this end. They are best performed on primary cultures, but established cell lines allow for easy preliminary investigations. An interesting complement to the methods described below is the in silico evaluation of the toxic effects of a molecule, derived from its chemical structure (structure-activity relationship), which is increasingly reliable in forecasting the adverse biological activities of the molecule even before it has been synthesised (Cronin et al., 2003).
and cell toxicology
Intracellular toxicity assessment. Once the xenobiotic or its metabolite has entered the cell, the effect on the latter and its fate must be monitored. In response to even mild aggression, the cell will mobilise a series of genes, either to protect itself or to have the damage repaired. Many members of the families of genes involved (stress genes and various repair enzyme genes) are known and can be recruited as 'reporters', which provide information on the target of the xenobiotic, the extent of the damage and the ability of the cell to overcome the damage (Mumtaz et al., 2002). Reporters also monitor the fate of the cell, exposed to various doses of the product, telling us about its ability to survive and how it will cope with the product in the long run. At present, reporter gene-loaded cells are commercially available, allowing the fast and inexpensive tracing of xenobiotics responsible for stress (including oxidative stress), various kinds of DNA damage, membrane damage, etc.
The disadvantage with reporters is the necessity of guessing the gene(s) targeted by the xenobiotic. This problem is overcome with commercially available DNA chips, carrying hundreds or thousands of gene elements known to be involved in toxic response (Toxicogenomics). By standard biochemical manipulations, the expression of each of these genes can be individually visualised on the chip. DNA chips allow observation of the simultaneous transcription behaviour in the cell's nucleus of all genes present on the chip, whether the genes are stimulated, repressed or unaffected by the xenobiotic (Kramer and Kolaja, 2002).
DNA chips are presently the ultimate tools for monitoring the first part - the transcription - of gene expression. In order to have a full view of the effect of xenobiotic activity on gene expression, the second part of expression - translation - during which the gene product is actually synthesised, must also be monitored. This can be done with the tools of Toxicoproteomics (2D gel or capillary electrophoresis, protein chips, mass spectroscopy, and many new methods under rapid development). Toxicoproteomics account for the xenobiotic-induced protein modifications (chemical or structural), modifications of proteolytic processes, aggregation, etc., which have been identified recently as representing important stages in many severe diseases (neurodegenerative disorders, dementia, diabetes type 2, etc.) (Tomer and Merrick, 2003).
Genotoxic activities of xenobiotics need to be identified and monitored with particular attention, since the failure to assess their effects accurately is the main factor responsible for the steep rise in cancer incidence observed over the past 50 years. These effects and activities can lead to DNA mutations, which can be monitored by a wealth of techniques (directly on DNA, precisely (sequencing) or roughly ('Comet' test); or indirectly, by monitoring the expression of DNA repair genes). Tumorigenesis is promoted by mutation or inactivation of genes involved in the regulation of cell growth and division. This leads to deregulation of pathways in charge of controlling programmed cell death, response to growth factors, cell migration, etc. (Amundson et al., 2001). DNA chips are available that allow characterisation of the transcription status of several hundreds of the genes involved in tumorigenesis when the cell is exposed to some xenobiotic.
Carcinogenic events can also be induced by non-genotoxic mechanisms, occurring at various steps (check points) of the cell cycle (division) or at the level of the higher organisation of the genetic material (chromatin) (Salleh et al., 2003). Of particular interest are the so-called tumour suppressors, as, for instance, the much-celebrated protein p53, one of the 'guardians of the genome'. This protein has control over the integrity of the genetic material of the cell. If some mutagenic event has occurred, p53 will immediately interrupt the cell cycle until the damage has been repaired. If the repair has not been achieved within a few hours, p53 will force the cell into apoptosis (suicide), thereby preventing transmission of the damage to the cell's progeny. Xenobiotics targeting p53 (or other tumour suppressors), either by mutating its gene or by modifying its structure, will abolish the 'guardian' activity of the protein, so that mutations can carry over to the cell's progeny and eventually induce uncontrolled cell proliferation - over half of solid tumours carry inactive forms of p53. Many commercial kits are available for monitoring the state of tumour suppressors and checkpoint gene expression, at the transcription or the protein level, especially for cancers thought to depend on hormones (breast cancer and oestrogen receptor signalling, prostate cancer and androgen signalling). DNA chips are also available for monitoring the expression, in the presence of a xenobiotic, of hundreds of human genes involved in the major steps of tumorigenesis, from deregulation of pathways in cell growth and division, DNA damage response, genome stability and repair, to cell adhesion, invasion, metastasis, angiogenesis etc.
Cytotoxic xenobiotics target the cell's organisation, its equipment, its metabolism, etc. This toxicity is often signalled by intense expression of stress genes (in particular oxidative stress - see 'reporter' gene section above). Cells initially respond by producing 'chaperones' and other enzymes to protect themselves from the xenobiotic's action. The severity and length of the exposure can elicit more global cellular responses, such as growth arrest, senescence, necrosis, cell death by apoptosis and even cell proliferation and carcinogenesis. The targets of cytotoxicity are the various cellular compartments and components, in particular mitochondria, the energy-factory of the cell which can trigger apoptosis (Moggs and Orphanides, 2003). For instance, xenobiotics can trigger the apoptosis pathway involving the Bcl2 gene family. The corresponding proteins oligomerise and insert themselves into the mitochondrial membrane, inducing the release of its content (including the CARD family), some of which trigger cell death. Cytotoxicity can be established by monitoring the expression of a long list of housekeeping genes and genes involved in necrosis, apoptosis, growth arrest, senescence, etc.
The methods mentioned so far can be applied to any type of cell, but those from organs most exposed to xenobiotics (liver, kidney, skin) should be assessed first.
Specific methods exist, of which the following are examples, for the molecular assessment of a variety of xenobiotics which target particular biological functions.
Reproductive toxicity can be monitored at the cell level, by, for instance, studying the activation by a xenobiotic of hormone receptors present at the surface of specific cells (Wong and Gill, 2002). As an example, 'endocrine proliferators' can mimic natural hormones and unduly induce the hormone-specific signal, or they can saturate specific hormone receptors and thereby preclude access to the normal hormone and delivery of its signal (Heinlein and Chang, 2003). Xenobiotics, including a large class of pesticides, which can induce abnormalities in male reproductive organs or promote tumours in tissues and organs that are under hormonal control (breast, ovary, prostate), can be identified in cultures of cells taken from these organs and tissues (Su et al., 2003).
Developmental toxicity (including teratogenesis) linked to xenobiotics targeting the cell cycle, growth factors or cell components involved in development signalisation (classes of RNAs, chromatin components, etc.) can be assessed by monitoring the interaction of the xenobiotic with these molecular components of the cell (Wang et al., 2002). DNA chips are available for monitoring the action of xenobiotics on many human genes involved in the regulation of the cycle phases (cyclins, cyclin-dependent kinases and their regulators: inhibitors, phosphatases and kinases).
can be the consequence of the action of a xenobiotic on molecules
involved in neuronal communication - a majority of insecticides target
these molecules! Several standard methods make it possible to assess
this class of neurotoxicity on neuronal cells in culture (Dam et al.,
2003). Electrochemical signals sent down the length of a neuron, or
from one neuron to another, are mediated by three classes of ion-specific
channels (passive channels, which maintain resting membrane potential;
chemically gated channels, which recognise neurotransmitters and initiate
an action potential, which is then propagated by voltage-gated channel);
and by two broad classes of neurotransmitters (those acting directly
on chemically gated ion channels and causing them to open, and those
acting indirectly and more slowly, involving G-protein coupled receptors
and the production of secondary messengers). DNA chips carrying genes
of ion channels and neurotransmitter transporters (including neurotrophins,
which are thought to play an important role in neuronal development)
are available for monitoring the effect of a xenobiotic on these essential
actors in neurobiology.
Immunotoxicity and inflammatory response. Inflammation is both the normal response of the body to pathogenes and a key intermediate of disease states such as allergies, asthma and arthritis that can be induced by xenobiotics. The signalling cascade of inflammatory response is propagated by the secretion of small glycoproteins (cytokines) and their binding on receptors of target cells. DNA chips bearing tens of human cytokine genes involved in the inflammatory response, and of the genes of their receptors, are available, allowing simultaneous determination of their expression profiles in cells exposed to some immunitoxic xenobiotic (Burchiel et al., 2001).
Molecular methods exist to assess membrane toxicity (modification of polarity, size and structure of lipid rafts, etc.) and epigenetic toxicity (xenobiotic-producing DNA methylation, acylation or phosphorylation of chromatin, which can seriously affect the programme of gene expression), etc. (Gore et al., 2000). Because of the polymorphism of the human population, assessment of cellular toxicity for classes of the population sharing common polymorphic patterns could be made, using class-specific DNA chips, which would, for instance, allow one to list xenobiotics that are especially harmful, or conversely safe, for members of the class (polymorphism-specific and idiosyncratic toxicology). For example, an extensive human polymorphism is found in the P450 family of metabolising enzymes.
Signal transduction and intercellular toxicity. This targets cell-to-cell and cell-extracellular matrix (gap junction) interactions, endocrine or exocrine processes, etc. Many cell types receive external signals (from hormones) which they transduce (translate) by some mechanism (signal-transduction pathway) into changing their behaviour or characteristics. G-protein coupled receptors, for instance, form a large family of cell-surface receptors involved in signal transduction. They are activated by a large variety of ligands, including chemicals and most drugs. Chemical stress can activate signal transduction via the NFkB transcription factor, released by the phosphorylation of the inhibitory IkB family of proteins or the Rel subunits. Once NFkB translocates in the nucleus, it induces transcription and expression of many genes, such as those encoding cytokines, adhesion molecules, inhibitors of apoptosis, etc. MAP kinase signalling operates through a cascade of kinases that also activate transcription factors of a variety of genes. Signalling by the TGFb super-family causes growth inhibition. Again, DNA chips carrying hundreds of human genes involved in signal transduction can monitor the action of xenobiotics on these genes (Waring et al., 2002).
The fate of the extracellular matrix (the substratum to which cells attach via adhesion molecules on the cell surface to help define tissue shape, structure and function), when exposed to a xenobiotic, can be monitored via the expression profile of molecules involved in cell-to-cell and cell-tissue interaction, such as cell-adhesion molecules (integrins, cadherins, catenins, selectins), extracellular matrix proteins (lamins, fibronectin, fibrinogen), protease (matrix metalloproteinases, serine and cystein protinases, cathepsin) and protease inhibitors (maspin). DNA arrays are available for monitoring the expression of hundreds of the genes involved, providing valuable information on the effect of a xenobiotic on primary steps in tissue and organ development.
In summary, using techniques based on molecular toxicity, we can obtain a clear view of the mechanism through which the substance or product is harmful, at what doses the cell can resist and, most importantly, the long-term effect on the cell. The experiment takes a few days on average, can be performed in large parallel screening set-ups (various cell types or doses, for instance), is relatively inexpensive, easy to standardise and requires tiny amounts of the xenobiotic (important in drug testing). The results are quantitative (large range of linear dose-response), reproducible and, most significantly, are valid for the species which provided the cells. These points represent definite scientific and economic advantages, even though advanced technical skills are required for most of these methods.
assessment of organ, tissue and systemic toxicity
MERITS OF SBT (SCIENCE-BASED TOXICOLOGY) OVER TRADITIONAL (ANIMAL-BASED RESARCH) TOXICITY ASSESSMENT
As shown above, the biological reactions of individuals of a given species are unique. Individuals from different species may occasionally display similar gross responses when exposed to the same toxic product, but one should never be misled by these chance phenomena. First and foremost, the mechanism through which a product induces some pathological reaction can be quite different in different species. Of the drugs in use, 60% are metabolised in humans by the same member of the P450 family (which can lead to synergistic drug activation), but several different members of the P450 family are involved in the case of apes, dogs and rodents. Secondly, long-term effects in humans are impossible to assess in species with a shorter life expectancy. In mice, spontaneous cancer development commences at the age of ten months, whereas in humans it usually starts after the age of forty years, and the mechanisms of cancer promotion are known to be rather different in different species. The strain-to-strain susceptibility for cancer in mice can vary a hundredfold. Certain strains tolerate, with no apparent ill-effects, oestrogen doses that are many times higher than doses harmful for other strains. Even if the gross response in two different species looks alike in the short term, the underlying mechanism which determines the long-term outcome is very likely different and can therefore lead to vastly different outcomes over the years. Incidentally, it would be useless to perform SBT for a given species on cells or cell cultures belonging to a different species.
CONCLUSION: BENEFITS OF SBT FOR INTERESTED PARTIES
The most obvious benefit would be consumer safety. SBT allows one to understand the mechanism by which a substance produces its adverse effects, which then allows one to forecast its long-term effects. By identifying cancer-promoting products, cancer prevention would be significantly boosted. As a result of science-based assessment of carcinogenic substances, we can estimate that cancer morbidity figures could be halved within the next three to five years. The reliable assessment of drug toxicity could save tens of thousands of lives a year. Neurotoxic substances (80% of insecticides are neurotoxic to insects: what about humans?) could be identified and removed from the market, preventing damage to the neuronal development of children - according to the FDA, this could be the case with the insecticide rotenone. Detection and removal of endocrine proliferators would prevent both abnormal development of sex organs and of most hormone-dependent cancers (breast, ovary, prostate).
Improved consumer safety would result in the immediate alleviation of socio-economic costs resulting from diseases, rates for which are presently soaring in EU countries.
The benefits of SBT assessments for industry are many. SBT experiments (as mentioned above) take a few days on average, can be performed in large parallel screening set-ups (with various cell types or doses, for instance), are relatively inexpensive, easy to standardise and require tiny amounts of the xenobiotic (important in drug testing). The results are quantitative (large range of linear dose-response), reproducible and, most importantly, are reliable and valid for the species under investigation. Furthermore, understanding the mechanism of the adverse effect may allow in-house chemists to modify the xenobiotic and remove, or alleviate, its toxicity (with the help of Structure-Activity-Relationship models, for example) and improve the quality of the product - and the image of the manufacturer! Since SBT tests are easily standardised and are independent of subjective parameters, the tests remain valid across political borders, which allows for unrestricted circulation of the SBT-tested goods among EU countries and abroad. These are definite scientific and economic advantages.
Whilst it is true that advanced technical skill is required for most of the SBT methods, and the requisite laboratory equipment is expensive, this would be more than counterbalanced by the relatively speedy scientific, economic and greatly improved public-health advantages.
SBT procedures also apply to other species and can therefore be used to assess environmental toxicities in any animal or plant species.
Since SBT procedures necessarily avoid 'model' species, they would also satisfy animal-welfare organizations.
Finally, implementing SBT methods would improve the image of health authorities, both inside the country (consumers would be grateful for better protection of their health) and outside, as the authorities could lead the way to effective improvement of environmental health issues worldwide.
steps could provide a rapid and practical implementation of SBT:
Bittner M., Meltzer P., Trent J. and Fornace A.J. Jr. (2001). Physiological
function as regulation of large transcriptional programs: the cellular
response to genotoxic stress, Comp. Biochem. Physiol. B, Biochem.
Mol. Biol. 129, 703-710.