PDF《upon Mutation》

Evaluation of the Stability of Folding Nucleus upon Mutation Mathieu Lonquety1,2 , Zo? e Lacroix1,3 , and Jacques Chomilier2

1 Scienti?c Data Management Laboratory, Arizona State University Tempe AZ 85281-5706, USA

2 IMPMC, Universit? e Pierre et Marie Curie, UMR

7590 CNRS

140 rue de Lourmel,

75015 Paris, France

3 Pharmaceutical Genomics Division, Translational Genomics Research Institute

13400 E Shea Blvd, Scottsdale AZ 85259, USA Abstract.

The development of a method that accurately predicts pro- tein folding nucleus is critical at least on two points. On one hand, they can participate to misfolded proteins and therefore they are related to several amyloid diseases. On the other hand, as they constitute struc- tural anchors, their prediction from the sequence can be valuable to improve database screening algorithms. The concept of Most Interact- ing Residues (MIR) aims at predicting the amino acids more likely to initiate protein folding. An alternative approach describes a protein 3D structure as a series of Tightened End Fragments (TEF). Their spatially close ends have been shown to be mainly located in the folding nucleus. While the current sequence-driven approach seems to capture all MIR, the structure-driven method partially fails to predict known folding. We present a stability-based analysis of protein folding to increase the recall and precision of these two methods. Results: Prediction of the folding nucleus by MIR algorithm is in agree- ment with mutation stability prediction. Availability: The database is available at: http://bioinformatics.eas.asu.edu/Stability/index.php. The MIR calcu- lation program is available at: http://bioserv.rpbs.univ-paris-diderot.fr/cgi-bin/MIR and the TEF pro- gram at: http://bioserv.rpbs.univ-paris-diderot.fr/TEF. Contact: [email protected] Keywords: Protein folding, folding nucleus, structure stability, point mutations.

1 Introduction Structural bioinformatics has been particularly productive for the past decade partially thanks to the contribution of physics disciplines. One of its main fo- cuses is the study of protein folding and, in particular, the prediction of folding nuclei. Modeling and predicting protein folding mechanisms is critical because M. Chetty, A. Ngom, and S. Ahmad (Eds.): PRIB 2008, LNBI 5265, pp. 54C65, 2008. c Springer-Verlag Berlin Heidelberg

2008 Evaluation of the Stability of Folding Nucleus upon Mutation

55 a misfolded protein may result in the formation of aggregates that may play a role in most misfolding diseases such as amyloid ones [1,2,3,4]. The folding nucleus model [5,6,7,8,9] is based on the assumption that protein folding begins with just a few amino acids that strongly interact with each other. These strong interactions initiate the folding that is completed by a successive folding of the remaining parts of the structure to constitute a compact globule. Within this model, a precise and accurate prediction of the main amino acids responsible for initiating the folding provides enough constraints to simulate the whole folding mechanism. For that purpose, Papandreou et al. developed an algorithm devoted to the search of the Most Interacting Residues (MIR) [10]. The current algorithm has a good recall however its precision needs improvement as several studies consider that the minimal number of amino acids needed to initiate a folding process is signi?cantly less than the

15 % found in average with the MIR prediction algorithm [8,11,12,13]. Proteins can be described as a succession of Tightened End Fragments or TEF [14,15,16,17] which spatially close ends (lower than