The structurally constrained protein evolution model accounts for sequence patterns of the LβH superfamily

GUSTAVO PARISI; JULIAN ECHAVE

BMC EVOLUTIONARY BIOLOGY

BIOMED CENTRAL LTD

Lugar: Londres; Año: 2004 p. 41 - 47

1471-2148

Background: Structure conservation constrains evolutionary sequence divergence, resulting in observable sequence patterns. Most current models of protein evolution do not take structure into account explicitly, being unsuitable for investigating the effects of structure conservation on sequence divergence. To this end, we recently developed the Structurally Constrained Protein Evolution (SCPE) model. The model starts with the coding sequence of a protein with known three-dimensional structure. At each evolutionary time-step of an SCPE simulation, a trial sequence is generated by introducing a random point mutation in the current coding DNA sequence. Then, a "score" for the trial sequence is calculated and the mutation is accepted only if its score is under a given cutoff, λ. The SCPE score measures the distance between the trial sequence and a given reference sequence, given the structure. In our first brief report we used a "global score", in which the same reference sequence, the ancestral one, was used at each evolutionary step. Here, we introduce a new scoring function, the "local score", in which the sequence accepted at the previous evolutionary time-step is used as the reference. We assess the model on the UDP-Nacetylglucosamine acyltransferase (LPXA) family, as in our previous report, and we extend this study to all other members of the left-handed parallel beta helix fold (LβH) superfamily whose structure has been determined. Results: We studied site-dependent entropies, amino acid probability distributions, and substitution matrices predicted by SCPE and compared with experimental data for several members of the LβH superfamily. We also evaluated structure conservation during simulations. Overall, SCPE outperforms JTT in the description of sequence patterns observed in structurally constrained sites. Maximum Likelihood calculations show that the local-score and global-score SCPE substitution matrices obtained for LPXA outperform the JTT model for the LPXA family and for the structurally constrained sites of class i of other members within the LβH superfamily. Conclusion: We extended the SCPE model by introducing a new scoring function, the local score. We performed a thorough assessment of the SCPE model on the