BioEd Online

When chemists design new detergents, shampoos, paints, and lubricants, they don't immediately consider whether their molecules will have toxic side-effects. That task has traditionally been left to toxicologists further down the production line.

But synthetic chemists can and should take earlier responsibility for the safety of their molecules, urges a group led by Julie Zimmerman at Yale University in New Haven, Connecticut.

In a paper published in Green Chemistry¹, the researchers show how obeying two key rules of thumb greatly reduces the chances of a molecule being acutely toxic to fish and other aquatic organisms. They plan to follow this up with similar design guidelines to avoid other types of damage, such as toxicity to birds.

That such design guidelines are possible would hardly surprise toxicologists. But the research demonstrates to synthetic chemists how our growing understanding of toxicity puts the onus on them to deliberately avoid making molecules that fall in the danger zone, argues co-author Paul Anastas, who is currently on leave from Yale as science adviser to the US Environmental Protection Agency (EPA).

Decades of safety tests have generated enough data for researchers to learn how chemicals produce toxic effects, and computer models are picking out the molecular properties that underlie this activity². "Rather than measuring how bad something is after it's made, and then going back to the drawing board, you can start to design molecules in what is likely to be a safer chemical space," says Anastas.

Avoid the danger zone

Zimmerman's team analysed data on hundreds of chemicals that have already been tested by the EPA and the Japanese environment ministry for their acute toxicity to three aquatic species — the fathead minnow (Pimephales promelas), the tiny Japanese medaka fish (Oryzias latipes) and the water flea Daphnia magna.

Following established toxicology research, the team were not surprised to find that the most toxic chemicals tended to be quite soluble in fat relative to water, because this makes it easier for a molecule to pass through a cell's membrane — the first hurdle for a chemical to overcome in order to have a biological effect. They also found, as expected, that the toxic chemicals are adept at ripping electrons from other molecules, a property that makes it more likely they will cause damage once inside a cell.

Combining the data on all species, the team quantified a range of values for variables relating to these two properties, within which fell 77% of the "desirable" chemicals; those with low or no toxicity. When they applied these rules to chemicals that had been tested on the green alga Pseudokirchneriella subcapitata, they found that 23 out of 29 desirable compounds fell inside the range.

Staying within the safer range increases the chances of designing a compound with very low acute aquatic toxicity by two- to fivefold, the authors found.

Toxicology for beginners

"I like this idea, but scientifically of course, this is nothing new," says Thomas Hartung, a toxicology expert at the Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland. As he points out, computational models that correlate biological activity and toxicity with chemical structure — known as quantitative structure–activity relationships (QSARs) — already indicate the key variables leading to acute aquatic toxicity. "This is just an illustrative example, in an area where toxicity prediction is working and we already know the descriptors," Hartung says.

But Hartung agrees with the Yale team's contention that this message needs to be absorbed by synthetic chemists, who tend not to be educated in toxicology. To chemists, QSARs seem like after-the-fact mathematical algorithms that don't give any indication of how to change a structure to avoid toxicity, says Adelina Voutchkova, another member of the Yale team. "We need QSARs, but we also need chemists to have a better chance of designing a molecule that passes the QSAR stage and in vivo tests," she says.

The design of inherently safer molecules is a goal that Anastas has been working towards for years. He is well known as one of the co-founders of the 'green chemistry' movement, which aims to cut down on toxic chemicals and processes. He admits that there are many aspects of toxicity (such as reproductive toxicity) that are currently not sufficiently well understood for design 'rules' to be created that would steer past them. This point is also emphasized by Mark Thompson, director of chemical company DuPont's Haskell Global Centers for Health and Environmental Sciences in Newark, Delaware. He says that, in general, DuPont agrees with the Yale team's approach, but adds that "there's still a long way to go in our capability to predict toxicity".

But that is just a matter of more research, Anastas thinks. "The key is to reduce the complexity to a small number of molecular properties which can be manipulated. That is a level very accessible to the molecular designers, the synthetic chemists," he says.

Hartung notes that the approach meshes well with the EPA's embrace of a 2007 report from the US National Research Council, which sets out a vision for twenty-first century toxicology that uses alternative methods, including probabilistic computational models. The question now, he says, is whether industrial chemists will start to adopt these methods. "Most chemical companies — with the exception of the really big ones that are forward-thinking — have no toxicology departments," he asserts.