Researchers at IBM’s Thomas J. Watson Research Center and Columbia University have used a small (single-rack) Blue Gene system from IBM to discover a new phenomenon having to do with the folding of a melittin protein in water.
Specifically, the Blue Gene system’s processing power allowed the researchers to perform simulations that were impractical to run on previously available high-performance computing systems.
In their simulations, the researchers studied what is called the dewetting process associated with the folding of a melittin protein tetramer in water. The focus of the simulations was to look at variations in the folding process under different conditions.
Understanding the mechanisms of melittin folding in water has implications in several areas. For instance, melittin inhibits lipid-protein interactions in some membranes (See: “Effects of melittin on lipid-protein interactions in sarcoplasmic reticulum membranes,” Biophysics Journal, vol. 63, page 1513, 1992.)
Melittin is a small protein that forms a needle-like shape in water. This shape allows it to puncture cell membranes. (Melittin is a component of bee venom, and this cell-puncturing characteristic helps deliver other venomous ingredients -- particularly the disruptive enzyme phospholipase A -- into a cell.)
The science behind the IBM and Columbia University simulations has to do with the way melittin folds. The protein is self-folding and the researchers were looking at the role a thin layer of water (within the folding protein) plays as the folding process takes place.
Essentially, there is a very small channel (i.e., nanoscale-sized; about the width of a few molecules) of water that influences the folding process. The dewetting process investigated has to do with the water layer drying out, leading to a collapse of the protein unto itself.
The simulations varied the channel size and other conditions. “We ran many different configurations,” says Ruhong Zhou, of the Computational Biology Center at IBM’s Thomas J. Watson Research Center. “We used ten different melittin protein sequences with ten different starting configurations each -- so 100 simulations all together.”
He notes that the key to doing each simulation (and the large number of simulations) was Blue Gene’s processing power. “We ran the simulations on a single-rack system with about a thousand [processing] nodes,” says Zhou. “In the past, simulations on similar systems took about 20 weeks of CPU time to run. With the single rack Blue Gene, we were able to reduce this to one week.”
The simulations give researchers the ability to better understand the basic science involved in protein folding. In fact, Zhou and his colleagues found a mechanism that switched the channel from wet to dry. “This was not expected for real protein systems,” says Zhou. “It’s a new phenomenon we’re seeing.”
The findings were published in the recent issue of Nature. (“Observation of dewetting transition in the collapse of the melittin tetramer,” Nature, September 1, 2005, Volume 437, Number 7055, p.159.)
Besides Zhou, the others involved in this research project are Pu Liu (the paper’s lead author) and Xuhui Huang (both of the Department of Chemistry at Columbia University) and B.J. Berne of the Department of Chemistry at Columbia University and the IBM Thomas J. Watson Research Center.
New Blue Gene Project to Model the Brain
First Blue Gene Paper Published
IBM, SGI Earn Top Bragging Rights at SC2004