Limits...
Predicting Binding Free Energy Change Caused by Point Mutations with Knowledge-Modified MM/PBSA Method.

Petukh M, Li M, Alexov E - PLoS Comput. Biol. (2015)

Bottom Line: While the approach is rigid body approach and does not explicitly consider plausible conformational changes caused by the binding, the effect of conformational changes, including changes away from binding interface, on electrostatics are mimicked with amino acid specific dielectric constants.This provides significant improvement of SAAMBE predictions as indicated by better match against experimentally determined binding free energy changes over 1300 mutations in 43 proteins.The final benchmarking resulted in a very good agreement with experimental data (correlation coefficient 0.624) while the algorithm being fast enough to allow for large-scale calculations (the average time is less than a minute per mutation).

View Article: PubMed Central - PubMed

Affiliation: Computational Biophysics and Bioinformatics, Department of Physics, Clemson University, Clemson, South Carolina, United States of America.

ABSTRACT
A new methodology termed Single Amino Acid Mutation based change in Binding free Energy (SAAMBE) was developed to predict the changes of the binding free energy caused by mutations. The method utilizes 3D structures of the corresponding protein-protein complexes and takes advantage of both approaches: sequence- and structure-based methods. The method has two components: a MM/PBSA-based component, and an additional set of statistical terms delivered from statistical investigation of physico-chemical properties of protein complexes. While the approach is rigid body approach and does not explicitly consider plausible conformational changes caused by the binding, the effect of conformational changes, including changes away from binding interface, on electrostatics are mimicked with amino acid specific dielectric constants. This provides significant improvement of SAAMBE predictions as indicated by better match against experimentally determined binding free energy changes over 1300 mutations in 43 proteins. The final benchmarking resulted in a very good agreement with experimental data (correlation coefficient 0.624) while the algorithm being fast enough to allow for large-scale calculations (the average time is less than a minute per mutation).

No MeSH data available.


Related in: MedlinePlus

The effect of dielectric constant variation for charged, polar and other residues in calculations of EE and SP on the correlation coefficient between experimental and calculated values of the change in binding free energy for the tDB.Only EE, VE and SP components were taken into account for the multiple linear regression analysis.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4492929&req=5

pcbi.1004276.g001: The effect of dielectric constant variation for charged, polar and other residues in calculations of EE and SP on the correlation coefficient between experimental and calculated values of the change in binding free energy for the tDB.Only EE, VE and SP components were taken into account for the multiple linear regression analysis.

Mentions: Since structural refinement with NAMD was done in implicit solvent model with dielectric constant 1, it is expected that the same value should be used to calculate the electrostatic components of the energy. However, initial testing showed that the obtained correlation of SAAMBE predicted energy changes and experiments is not impressive. This combined with our previous work on predicting folding free energy changes [36], we decided to test the possibility that better correlation can be obtained if amino acids with different physico-chemical properties are modeled with different dielectric constants. Previous investigations indicated that charged and polar amino acid should be assigned relatively large dielectric constant as compared with hydrophobic groups [36]. However, the work was done for predicting folding free energy changes and the results may not be directly transferrable to model the changes of the binding free energy. In SAAMBE protocol, we assume that there are three groups of residues with specific dielectric constants ε1, ε2 and ε3 for charged, polar and other groups, respectively (see Method section). We varied systematically the dielectric constant for charged groups from 5 to 15, for polar from 3 to 13 and other residues from 3 to 13 with a step of 2. Then multiple linear regression analysis was performed for SAAMBE formula containing EE, VE and SP components only. This was done for computational efficiency only. Fig 1 shows contour maps of the correlation coefficients for fixed ε1 of charged residues and varied ε2 of polar residues (on the x-axis) and ε3 for other types of residues (on the y-axis). The grey color represents the area with the maximum correlation coefficient, while the black one—its minimum for given combination of dielectric constants. From Fig 1 one can see that the area with maximum correlation coefficient increases with the increase of dielectric constant of the charged residues, reach its maximum at ε1 = 9 and then decreases. The correlation coefficient has the highest value when the ε2 for the polar residues is 8 and ε3 for other types of residues is 7. Thus SAAMBE protocol uses dielectric constants of 9, 8, and 7 for charged, polar and other amino acids, respectively, to calculate the SP energy component. The EE component is calculated with the lowest dielectric constant, ε = 7, for the entire protein and protein complex.


Predicting Binding Free Energy Change Caused by Point Mutations with Knowledge-Modified MM/PBSA Method.

Petukh M, Li M, Alexov E - PLoS Comput. Biol. (2015)

The effect of dielectric constant variation for charged, polar and other residues in calculations of EE and SP on the correlation coefficient between experimental and calculated values of the change in binding free energy for the tDB.Only EE, VE and SP components were taken into account for the multiple linear regression analysis.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4492929&req=5

pcbi.1004276.g001: The effect of dielectric constant variation for charged, polar and other residues in calculations of EE and SP on the correlation coefficient between experimental and calculated values of the change in binding free energy for the tDB.Only EE, VE and SP components were taken into account for the multiple linear regression analysis.
Mentions: Since structural refinement with NAMD was done in implicit solvent model with dielectric constant 1, it is expected that the same value should be used to calculate the electrostatic components of the energy. However, initial testing showed that the obtained correlation of SAAMBE predicted energy changes and experiments is not impressive. This combined with our previous work on predicting folding free energy changes [36], we decided to test the possibility that better correlation can be obtained if amino acids with different physico-chemical properties are modeled with different dielectric constants. Previous investigations indicated that charged and polar amino acid should be assigned relatively large dielectric constant as compared with hydrophobic groups [36]. However, the work was done for predicting folding free energy changes and the results may not be directly transferrable to model the changes of the binding free energy. In SAAMBE protocol, we assume that there are three groups of residues with specific dielectric constants ε1, ε2 and ε3 for charged, polar and other groups, respectively (see Method section). We varied systematically the dielectric constant for charged groups from 5 to 15, for polar from 3 to 13 and other residues from 3 to 13 with a step of 2. Then multiple linear regression analysis was performed for SAAMBE formula containing EE, VE and SP components only. This was done for computational efficiency only. Fig 1 shows contour maps of the correlation coefficients for fixed ε1 of charged residues and varied ε2 of polar residues (on the x-axis) and ε3 for other types of residues (on the y-axis). The grey color represents the area with the maximum correlation coefficient, while the black one—its minimum for given combination of dielectric constants. From Fig 1 one can see that the area with maximum correlation coefficient increases with the increase of dielectric constant of the charged residues, reach its maximum at ε1 = 9 and then decreases. The correlation coefficient has the highest value when the ε2 for the polar residues is 8 and ε3 for other types of residues is 7. Thus SAAMBE protocol uses dielectric constants of 9, 8, and 7 for charged, polar and other amino acids, respectively, to calculate the SP energy component. The EE component is calculated with the lowest dielectric constant, ε = 7, for the entire protein and protein complex.

Bottom Line: While the approach is rigid body approach and does not explicitly consider plausible conformational changes caused by the binding, the effect of conformational changes, including changes away from binding interface, on electrostatics are mimicked with amino acid specific dielectric constants.This provides significant improvement of SAAMBE predictions as indicated by better match against experimentally determined binding free energy changes over 1300 mutations in 43 proteins.The final benchmarking resulted in a very good agreement with experimental data (correlation coefficient 0.624) while the algorithm being fast enough to allow for large-scale calculations (the average time is less than a minute per mutation).

View Article: PubMed Central - PubMed

Affiliation: Computational Biophysics and Bioinformatics, Department of Physics, Clemson University, Clemson, South Carolina, United States of America.

ABSTRACT
A new methodology termed Single Amino Acid Mutation based change in Binding free Energy (SAAMBE) was developed to predict the changes of the binding free energy caused by mutations. The method utilizes 3D structures of the corresponding protein-protein complexes and takes advantage of both approaches: sequence- and structure-based methods. The method has two components: a MM/PBSA-based component, and an additional set of statistical terms delivered from statistical investigation of physico-chemical properties of protein complexes. While the approach is rigid body approach and does not explicitly consider plausible conformational changes caused by the binding, the effect of conformational changes, including changes away from binding interface, on electrostatics are mimicked with amino acid specific dielectric constants. This provides significant improvement of SAAMBE predictions as indicated by better match against experimentally determined binding free energy changes over 1300 mutations in 43 proteins. The final benchmarking resulted in a very good agreement with experimental data (correlation coefficient 0.624) while the algorithm being fast enough to allow for large-scale calculations (the average time is less than a minute per mutation).

No MeSH data available.


Related in: MedlinePlus