Predictive toxicogenomics uses the genetic hand you're dealt to forecast your reaction to environmental chemicals. Lisa Melton reports

Predictive toxicogenomics uses the genetic hand you’re dealt to forecast your reaction to environmental chemicals. Lisa Melton reports

Every poker player knows that it takes more than a lucky hand to win. You need to have the measure of your opponent and, crucially, to understand the odds of beating them. 


Toxicogenomic researchers are using a similar approach to tackle disease. They know that nearly all diseases involve a combination of genetic factors and environmental hazards such as stress, drug side-effects and even food molecules.

The big question is how much our health depends on the genetic hand we are dealt at conception, and how much on the toxic assault we experience throughout our lives. It’s a complex mix which has, so far, been hard to quantify. 

But toxicogenomics is now beginning to measure the influence of these environmental and genetic factors. Researchers have revealed that environmental factors can alter a person’s genetic makeup, particularly during very early, developmental stages. And the genetic code, in turn, influences how people will respond to a particular drug treatment, or determine the impact of exposure to a toxin.

William Suk, a toxicogenomic researcher from the US National Institute of Environmental Health Sciences (NIEHS) in Research Triangle Park, North Carolina, US, describes it as a ripple effect. ’It’s about understanding environmental exposure and how it cascades through the [human] system over time to cause disease,’ he says. ’If you can understand where the ripples are, you can understand how to smooth the bumps and prevent disease,’ he adds.

The consequences of toxicogenomics are immediate and practical. If a person is genetically susceptible to Parkinson’s disease, for example, they would do well to avoid certain pesticides that ramp up their risk of developing the condition. Understanding the interplay of genetics and toxicology in this way could help millions to stay healthier for longer.

Early encounters   

The completion of the human genome sequence, coupled with powerful gene chip technologies helped kick-start a new era in toxicogenomic research. Until then, investigators had been interrogating one gene at a time. Microarrays allowed them to ramp up the numbers, tracking thousands of genes at once. 

In November 2000, with genomic information still hot off the sequencers, the NIEHS launched a toxicogenomics initiative to track the genomic responses to certain chemicals. Five US academic centres joined the consortium to harness microarray gene expression profiling, proteomic technologies and bioinformatics to answer perplexing gene-environment questions.  


Microarrays allow researchers to track thousands of genes at a time

Rebecca Fry’s work on arsenic exposure is a dramatic example of the insights microarray profiling can provide. ’Could we, with a handful of genes, figure out whether a baby has been exposed to arsenic in the womb?’ says Fry, a research scientist at the Massachusetts Institute of Technology, in Boston, US and part of the NIEHS-funded consortium.  

Fry uncovered a ’smoking gun’ for pre-natal arsenic exposure. Working with Mathuros Ruchirawat of the Chulabhorn Research Institute in Thailand, and Leona Samson, director of the MIT’s Center for Environmental Health Sciences, she studied mothers and their offspring living in a province of Thailand where tin mining leads to heavy arsenic contamination in groundwater. Arsenic exposure increases cancer risk later in life.  

It is possible to single out newborns exposed to arsenic in the womb, says Fry. She found a subset of 11 genes - mostly related to inflammation and stress response - that are dramatically altered if exposure has taken place. The results published in PLoS Genetics  have generated much attention, especially in countries where arsenic water poisoning is common, and population screening highly desirable.1 

Most scientists agree the time will come when individual toxicogenomic profiles will become a major tool in medicine. Cancer, heart disease, Parkinson’s disease and Alzheimer’s disease among many, are known to result from genes and the environment acting in synergy.  

According to Helmut Zarbl, a toxicogenomics researcher at the University of Medicine and Dentistry of New Jersey, US, researchers already have the knowledge to scour the genome for these toxic fingerprints. ’The key is to connect those biological response indicators to mechanisms of disease. We [can] make informed guesses as to which of the affected biological networks lead to disease,’ he explains.  

Zarbl has already pinned down several molecular pathways that may boost the risk of breast cancer. Among them is the hormone prolactin. Rat strains in which prolactin signalling is strongly activated are prone to develop mammary carcinomas, Zarbl has shown.2 And drugs known to increase prolactin levels in serum also boost the risk of breast cancer, and he is going one step futher - using toxicogenomic approaches to search for compounds that dampen those gene expression profiles. 

Spotting toxicant signatures and susceptibilities will, eventually, provide people with individualised advice on how to prevent disease. ’If a person carries a combination of genes that puts them at higher risk of disease with certain type of environmental exposure, then we could take steps either to reduce their exposure or use dietary interventions or drugs to counter the effects of that exposure,’ Zarbl explains.  

Small dangers 

But environmental hits are minute and take place over long periods. They could be fiendishly difficult to track. ’Dose is always the problem in toxicology,’ says Andrew Smith, a toxicologist from the MRC Toxicology Unit, University of Leicester, UK. ’When you study environmental toxins at such low level, are they of any significance? Putting all the data together to create a feasible picture is mind-bogglingly difficult.’  



Toxicogenomics can identify those at risk of developing cancer from arsenic exposure via drinking water

Working with Smith at the University of Leicester is molecular biologist Tim Gant. He points out that natural and innocuous fluctuations in the environment, for instance, temperature changes, also prompt gene expression changes. ’Just because you can measure loads of genes going up or down doesn’t mean anything. The organism may be responding in a perfectly natural way,’ he says.  

But Gant is impressed by the dramatic speed with which pharmaceutical compounds trigger changes in gene expression. ’You can see gene expression patterns that predict pathology [changing] very quickly, within minutes, and it is possible to quantify these changes.’  

He is focusing his research efforts on anti-cancer agents. Very unpleasant side-effects of doxorubicin, a compound used to treat many types of tumours, are well-known in the clinic - the drug causes vomiting and hair-loss. But the cardiac toxicity associated with this compound is not understood at the molecular level. Using microarray data, Gant and his colleagues have tracked gene expression changes in heart cells. ’These gene expression patterns tell us about the potential mechanisms that underpin those harmful changes in cardiac cells,’ he explains. 

Even widely-prescribed drugs, such as paracetamol, are astonishingly toxic to some individuals. ’People are given paracetamol all the time and now and again somebody comes into the clinic with an overdose and toxicity,’ notes William Kaufmann, a professor of pathology and laboratory medicine. ’It is the most common cause of acute liver failure, but it’s not clear why some people react with a toxic response and others don’t.’ 

His team at the University of North Carolina at Chapel Hill, US has identified a clear gene expression pattern or signature for liver toxicity in mice treated with sub-lethal doses of paracetamol. If the same patterns crop up in humans, it will be possible to warn those people who may be at risk and should avoid taking the drug; doctors could soon be running lab-on-a-chip tests to screen for this type of toxicity.  

But in the clinic, tissue access for testing could be an obstacle. Most microarray analyses in humans have, thus far, used biopsies. Kaufman is pursuing the idea that if toxicity patterns show up in blood cells, these would provide an easy-to-access, surrogate marker for what is going on in other tissues.  

Doomed drugs 

Early alarm bells to warn of toxicity could also avoid the blind alleys that plague drug discovery (see Chemistry World  , June 2007, p58). At the early phases, the rate of attrition is astonishingly high. According to Adriano Henney, AstraZeneca’s director of global discovery, enabling capabilities and sciences, most compounds directed at novel targets will never make it to pharmacists’ shelves. ’60 per cent of compounds fail at the preclinical stage. Toxicity is the main reason,’ he notes.  

Researchers at the Abbot Laboratories in Abbot Park, Illinois, US are wielding toxicogenomic tools to weed out the bad from the maybes. ’If we do a toxicology study and some toxicity occurs, we can use gene expression analysis to understand the mechanism,’ explains Jeff Waring, group leader of Abbot’s toxicogenomics division. ’Is the toxicity an on-target effect? If so, that may not be a good target to pursue,’ he points out. ’If it is an off-target effect, it could be possible to design compounds that steer away from that toxicity.’  


Stevens-Johnson syndrome is a serious and unpredictable effect of drug toxicity

Gene chip testing can give an indication of whether a compound is associated with toxic liabilities in three to five days, while traditional toxicology tests take between two weeks and six months to complete. ’Using small quantities of compound, it allows us to quickly select compounds with an acceptable safety margin,’ Waring adds. Such rapid screening translates into fewer false starts and improves the quality of drug candidates moving along the pipeline. 

There is no denying that all medicines - even approved ones - can have serious adverse effects in some people, and these are, to some degree, influenced by genetics. In September 2007, the International Serious Adverse Events Consortium (SAEC) was formed to try to identify genetic DNA-variants that could be useful in predicting that risk. Leading pharmaceutical companies joined forces with academic institutions and the US Food and Drug Administration to collect and analyse the data. 

The programme will collect DNA from individuals who have experienced drug-related liver toxicity or a serious, drug-related skin condition called Stevens-Johnson syndrome. ’A few people get serious side-effects to drugs, but most don’t. We’d like to know what the bases of those differences are,’ says Waring, whose company, Abbott, is participating in the consortium along with GlaxoSmithKline, Johnson & Johnson, Pfizer, Roche, Sanofi-aventis and Wyeth. Other partners include the UK’s Newcastle University and Eudragene 
(a European academic consortium). 

Drenched in data 

At first glance, microarrays whizzing through more than 40,000 human genes in a single analysis may seem an unprecedented opportunity to understand how the system responds to a particular perturbation. ’We were told, "take a voyage and let us know what you discover",’ enthuses Kaufmann of the University of North Carolina, who also participates in the NIEHS consortium.  

But the veritable flood of information from these assays is proving more than a challenge. ’The size of the data set is daunting,’ he adds. ’We spend a lot of time trying to figure out how we are going to extract anything meaningful out of this. There are genes going in one direction and in another. Some patterns we can attribute to known biochemical signalling within cells, for others we have no biochemical explanation - the versatility in the biological response is extraordinary.’ 

Pattern recognition algorithms to analyse the data are, nonetheless, spurring on the research. What is now needed is a full integration of ’omics’ data with biological network information to understand how these perturbed genes and pathways fit into the context of the whole organism. ’The new buzzword is "systems" as opposed to the old days when we talked about individual genes,’ says Gant at the University of Leicester. 

The food and cosmetics industries are also vigorously exploring the ’omics’ route to identify which substances could, say, irritate skin. ’Our intention is to find biomarkers to predict in the lab what is going on in vivo,’ explains Dan Scott at Unilever’s plant in Colworth Park, UK. ’For the moment, we are still using toxicogenomics as a research tool rather than a front-line assay.’ 

Testing for genotoxic chemicals is top of the agenda in Europe. A project funded by the European Commission aims to bring together academia and industry to develop high-throughput assays to identify compounds with carcinogenic risk, for example. Such toxicogenomic techniques could enable safety decisions without using animal tests.  

Suk, currently acting director of the NIEHS consortium, believes the field is poised for take off: ’In the last five years we have been standardising the tools, building our capacity. Now we can begin to apply what we have learnt and ask how environmental exposure perturbs the system.’ If all goes to plan, predicting a person’s risk of disease will have more in common with poker than with snap. 

Lisa Melton is a science journalist based in London, UK 

Further Reading

1 R C Fry et al  PLoS Genet.   2007, 3, e207 (DOI:10.1371/journal.pgen.0030207) 
2 X Ren et al Carcinogenesis, 2008, 29, 177