Grammatical rules spell out new drugs
Learning the language of antibiotics could help fight superbugs.
By shuffling protein segments like words in a sentence, researchers have created what they hope is a language for finding potent new antibiotics.
Such drugs are sorely needed because of the rise of 'superbugs' that are resistant to most existing drugs. The bacteria become resistant by mutating the protein attacked by an antibiotic.
Researchers are hopeful that new antibiotics might be found in a group of small bug-killing proteins, which occurs naturally in secretions ranging from human sweat to scorpion venom. Bacteria cannot evade such peptides with a simple mutation because they interact with multiple molecules in the bacterial wall. But researchers know of only a few hundred such naturally occurring peptides, and none of them have proven potent and safe enough to make new human antibiotics yet.
So Gregory Stephanopoulos at Massachusetts' Institute of Technology, Cambridge, and his team wanted to see if they could artificially create a suite of new peptides with a similar structure to the natural ones, but with safer and more potent features.
The researchers tackled the problem as if the peptides were a language, and each of the amino acid building blocks a word. They used a pattern-recognition algorithm that is partly based on language analyses to compare all the known peptides and find what features they have in common.
The analysis revealed that the diverse peptides do contain specific arrangements that the team calls grammars. In the same way that a written sentence might have a noun, pronoun and verb, the peptides too contain specific types of amino acids that tend to come in certain places. Just as one noun can be swapped for another while retaining sense ('He kicked the ball' to 'He kicked the tyre'), some amino acids can be swapped for others while retaining a working peptide.
Armed with this peptide language, the team next used a computer to generate a variety of new amino acid sequences that fit the grammatical rules, but are also quite different from all known peptides.
The team synthesised 42 new peptides this way and found that nearly half killed bacteria. One killed the hospital superbug Staphylococcus aureus and the bioterror agent Bacillus anthracis at low concentrations. By comparison, peptides containing exactly the same amino acids, but arranged in a non-grammatical way, had little effect on the bugs. The work is published in Nature1.
Stephanopoulos says it would be easy to generate another 50,000 artificial peptides, and he is talking to drug companies about this possibility. Until now, drug makers have struggled to design more effective antimicrobial peptides, and no such drugs have reached the market.
There are, however, problems with using peptides as antibiotics. Unlike most conventional antibiotics, they are large molecules that would be difficult to deliver in a pill. And there have been reports that crafty bacteria can acquire resistance even to these drugs.
The same approach might be used to identify other useful patterns in the peptides, such as those that indicate which drugs are toxic in human tissues, or those that specifically tackle E. coli. "That's the big advantage," Stephanopoulos says.
The method does seem novel, says Michael Zasloff, an expert in the peptides at Georgetown University, Washington DC. "These are not the tools we would normally use in protein analysis," he says. "It's funny to think I could actually assemble a group for development of new antimicrobial peptides and justify the inclusion of a linguist."
But Robert Hancock, who is carrying out similar studies at the University of British Columbia, Vancouver, questions how different the linguistics technique is from other computational methods used to find similarities between protein sequences. "What's new is the catchy title," he says.
Visit our newsblog to read and post comments about this story.
- Loose C., et al. Nature, 443 , doi 10.1038/nature05233 (2006).