Nearly ten years ago an article published in Science [Lockless SW, Ranganathan R (1999) Science 286:295–299] got a lot of attention. It described a method of demonstrating signal transfer in proteins by comparing their amino acid sequence.

The authors recorded a statistical method of showing how certain parts of proteins change together through evolution, i.e. if a change had taken place in one part a change simultaneously took place in another part of the protein. They found a network of parts that seemed to belong together and, within this network, signal transfer was deemed to take place.

In a new PNAS article, Uppsala researchers present results of experiments they say contradicts that theory.

Proteins govern nearly all chemical processes in the body's cells. A fundamental property of proteins is their ability to transfer signals – both within and between proteins. It is known, for example, that such signal transfer is vital to haemoglobin, which transports oxygen in the body. In that instance the mechanism has largely been clarified.

"But in other instances very little is known about the mechanisms or whether such signal transfer even occurs," says Per Jemth, who together with his research group at Uppsala University is studying whether signal transfer also occurs in small proteins.

The Uppsala researchers said they saw several things that were not right about the results in the much discussed article, and by means of experiments they can now show that no more signals occur in this network than with other parts of the protein. They instead found, completely logically, that nearby parts of the protein interact more with each other than parts that are a long way apart.

"Our results thus question whether statistical methods can demonstrate signal transfer within proteins, and emphasise the importance of precise experiments to substantiate computer-based methods in protein chemistry," says Per Jemth.

The ability to predict proteins' function down to the smallest detail on the basis of their amino acid sequence is a goal that has preoccupied many researchers ever since human DNA became known. This study emphasises that experiments are needed to improve and refine the computerised methods currently in use.

"When theory, computer simulation and experiments provide the same answers the long-term goal has been attained, but there's still a long way to go."

Article: Celestine N. Chi, Lisa Elfström, Yao Shi, Tord Snäll, Åke Engström, and Per Jemth, 'Reassessing a sparse energetic network within a single protein domain', PNAS published online March 13, 200810.1073/pnas.0711732105