Limits...
Geomfinder: a multi-feature identifier of similar three-dimensional protein patterns: a ligand-independent approach.

Núñez-Vivanco G, Valdés-Jiménez A, Besoaín F, Reyes-Parada M - J Cheminform (2016)

Bottom Line: Nevertheless, many of the protein structures stored in public databases do not provide information about the location and characteristics of ligand binding sites and/or other important 3D patterns such as allosteric sites, enzyme-cofactor interaction motifs, etc.Thus: (a) Geomfinder detected identical pairs of 3D patterns in a series of monoamine oxidase-B structures, which corresponded to the effectively similar ligand binding sites at these proteins; (b) we identified structural similarities among pairs of protein structures which are targets of compounds such as acarbose, benzamidine, adenosine triphosphate and pyridoxal phosphate; these similar 3D patterns are not detected using sequence-based methods; (c) the detailed evaluation of three specific cases showed the versatility of Geomfinder, which was able to discriminate between similar and different 3D patterns related to binding sites of common substrates in a range of diverse proteins.Although the software is not intended for simultaneous multiple comparisons in a large number of proteins, it can be particularly useful in cases such as the structure-based design of multitarget drugs, where a detailed analysis of 3D patterns similarities between a few selected protein targets is essential.

View Article: PubMed Central - PubMed

Affiliation: Escuela de Ingeniería Civil en Bioinformática, Universidad de Talca, Avenida Lircay s/n, Talca, Chile ; Centro de Bioinformática y Simulación Molecular, Universidad de Talca, 2 Norte 685, Talca, Chile.

ABSTRACT

Background: Since the structure of proteins is more conserved than the sequence, the identification of conserved three-dimensional (3D) patterns among a set of proteins, can be important for protein function prediction, protein clustering, drug discovery and the establishment of evolutionary relationships. Thus, several computational applications to identify, describe and compare 3D patterns (or motifs) have been developed. Often, these tools consider a 3D pattern as that described by the residues surrounding co-crystallized/docked ligands available from X-ray crystal structures or homology models. Nevertheless, many of the protein structures stored in public databases do not provide information about the location and characteristics of ligand binding sites and/or other important 3D patterns such as allosteric sites, enzyme-cofactor interaction motifs, etc. This makes necessary the development of new ligand-independent methods to search and compare 3D patterns in all available protein structures.

Results: Here we introduce Geomfinder, an intuitive, flexible, alignment-free and ligand-independent web server for detailed estimation of similarities between all pairs of 3D patterns detected in any two given protein structures. We used around 1100 protein structures to form pairs of proteins which were assessed with Geomfinder. In these analyses each protein was considered in only one pair (e.g. in a subset of 100 different proteins, 50 pairs of proteins can be defined). Thus: (a) Geomfinder detected identical pairs of 3D patterns in a series of monoamine oxidase-B structures, which corresponded to the effectively similar ligand binding sites at these proteins; (b) we identified structural similarities among pairs of protein structures which are targets of compounds such as acarbose, benzamidine, adenosine triphosphate and pyridoxal phosphate; these similar 3D patterns are not detected using sequence-based methods; (c) the detailed evaluation of three specific cases showed the versatility of Geomfinder, which was able to discriminate between similar and different 3D patterns related to binding sites of common substrates in a range of diverse proteins.

Conclusions: Geomfinder allows detecting similar 3D patterns between any two pair of protein structures, regardless of the divergency among their amino acids sequences. Although the software is not intended for simultaneous multiple comparisons in a large number of proteins, it can be particularly useful in cases such as the structure-based design of multitarget drugs, where a detailed analysis of 3D patterns similarities between a few selected protein targets is essential.

No MeSH data available.


Score values for similar pairs of 3D patterns detected in 1E8W and 2O3P. In this figure, eight pairs of similar 3D patterns are shown. The 3D patterns corresponding to the quercetin binding sites are shown in green. The 3D patterns whose residues are located near of the quercetin and imidazole binding sites, are shown in pink. The 2 non-obvious similar 3D patterns found, are shown in yellow. In the red box, are shown the final similarity scores of Geomfinder (GScore)
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4834829&req=5

Fig9: Score values for similar pairs of 3D patterns detected in 1E8W and 2O3P. In this figure, eight pairs of similar 3D patterns are shown. The 3D patterns corresponding to the quercetin binding sites are shown in green. The 3D patterns whose residues are located near of the quercetin and imidazole binding sites, are shown in pink. The 2 non-obvious similar 3D patterns found, are shown in yellow. In the red box, are shown the final similarity scores of Geomfinder (GScore)

Mentions: We compared the crystal structures of the protein phosphatidyl-inositol 4,5-bisphosphate 3-kinase [PDBid: 1E8W] (961 residues) with the oncogene serine/threonine protein kinase [PDBid: 2O3P] (293 residues). These proteins are very dissimilar and have only 8 % of identity in their primary sequences. The sequence alignment yielded a similarity of 34.3 % with almost 350 residues aligned (including gaps). In our analysis, the initial radius of the distance used to build the virtual grid of coordinates was set in 10 Å. Herewith, 4535 and 14,702 virtual coordinates of reference were created into the protein structures of 2O3P and 1E8W respectively (see an example in the Table 2). This generated more than 66 million of pairs of 3D patterns, which were compared in 5 minutes and 20 seconds. Our results revealed the existence of several pairs of 3D patterns with a GScore higher than 50 %, one of which corresponds to the flavonoid quercetin binding site. This is in agreement with a previous report [55], which showed similar chemical interactions between quercetin and the residues of the binding site in both proteins. Among the highest GScore found, five pairs of these similar patterns were shown to contain residues located near to the co-crystallized ligands quercetin and imidazole, another common ligand. Interestingly, two other non-obvious similar 3D patterns were detected (Fig. 9). In fact, one of these pairs exhibited the highest GScore value (86.3 %) determined after comparison of both proteins. The patterns in this pair were detected from the virtual reference coordinate Atom459 (in 1E8W) and Atom151 (in 2O3P), and showed values of 100 % of similarity for the non-bonded energies (SNbE) and the sequence component (SSc) parameters. It should be noted that in order to align the sequences of these 3D patterns (ASP637, PRO671, SER636, GLU638 and ARG641 to the Atom459, and ASP114, PRO113, SER115, GLU70 and ARG112 to Atom151), a sequence-based method must incorporate at least 40 gaps, with the corresponding decrease of the alignment significance.


Geomfinder: a multi-feature identifier of similar three-dimensional protein patterns: a ligand-independent approach.

Núñez-Vivanco G, Valdés-Jiménez A, Besoaín F, Reyes-Parada M - J Cheminform (2016)

Score values for similar pairs of 3D patterns detected in 1E8W and 2O3P. In this figure, eight pairs of similar 3D patterns are shown. The 3D patterns corresponding to the quercetin binding sites are shown in green. The 3D patterns whose residues are located near of the quercetin and imidazole binding sites, are shown in pink. The 2 non-obvious similar 3D patterns found, are shown in yellow. In the red box, are shown the final similarity scores of Geomfinder (GScore)
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4834829&req=5

Fig9: Score values for similar pairs of 3D patterns detected in 1E8W and 2O3P. In this figure, eight pairs of similar 3D patterns are shown. The 3D patterns corresponding to the quercetin binding sites are shown in green. The 3D patterns whose residues are located near of the quercetin and imidazole binding sites, are shown in pink. The 2 non-obvious similar 3D patterns found, are shown in yellow. In the red box, are shown the final similarity scores of Geomfinder (GScore)
Mentions: We compared the crystal structures of the protein phosphatidyl-inositol 4,5-bisphosphate 3-kinase [PDBid: 1E8W] (961 residues) with the oncogene serine/threonine protein kinase [PDBid: 2O3P] (293 residues). These proteins are very dissimilar and have only 8 % of identity in their primary sequences. The sequence alignment yielded a similarity of 34.3 % with almost 350 residues aligned (including gaps). In our analysis, the initial radius of the distance used to build the virtual grid of coordinates was set in 10 Å. Herewith, 4535 and 14,702 virtual coordinates of reference were created into the protein structures of 2O3P and 1E8W respectively (see an example in the Table 2). This generated more than 66 million of pairs of 3D patterns, which were compared in 5 minutes and 20 seconds. Our results revealed the existence of several pairs of 3D patterns with a GScore higher than 50 %, one of which corresponds to the flavonoid quercetin binding site. This is in agreement with a previous report [55], which showed similar chemical interactions between quercetin and the residues of the binding site in both proteins. Among the highest GScore found, five pairs of these similar patterns were shown to contain residues located near to the co-crystallized ligands quercetin and imidazole, another common ligand. Interestingly, two other non-obvious similar 3D patterns were detected (Fig. 9). In fact, one of these pairs exhibited the highest GScore value (86.3 %) determined after comparison of both proteins. The patterns in this pair were detected from the virtual reference coordinate Atom459 (in 1E8W) and Atom151 (in 2O3P), and showed values of 100 % of similarity for the non-bonded energies (SNbE) and the sequence component (SSc) parameters. It should be noted that in order to align the sequences of these 3D patterns (ASP637, PRO671, SER636, GLU638 and ARG641 to the Atom459, and ASP114, PRO113, SER115, GLU70 and ARG112 to Atom151), a sequence-based method must incorporate at least 40 gaps, with the corresponding decrease of the alignment significance.

Bottom Line: Nevertheless, many of the protein structures stored in public databases do not provide information about the location and characteristics of ligand binding sites and/or other important 3D patterns such as allosteric sites, enzyme-cofactor interaction motifs, etc.Thus: (a) Geomfinder detected identical pairs of 3D patterns in a series of monoamine oxidase-B structures, which corresponded to the effectively similar ligand binding sites at these proteins; (b) we identified structural similarities among pairs of protein structures which are targets of compounds such as acarbose, benzamidine, adenosine triphosphate and pyridoxal phosphate; these similar 3D patterns are not detected using sequence-based methods; (c) the detailed evaluation of three specific cases showed the versatility of Geomfinder, which was able to discriminate between similar and different 3D patterns related to binding sites of common substrates in a range of diverse proteins.Although the software is not intended for simultaneous multiple comparisons in a large number of proteins, it can be particularly useful in cases such as the structure-based design of multitarget drugs, where a detailed analysis of 3D patterns similarities between a few selected protein targets is essential.

View Article: PubMed Central - PubMed

Affiliation: Escuela de Ingeniería Civil en Bioinformática, Universidad de Talca, Avenida Lircay s/n, Talca, Chile ; Centro de Bioinformática y Simulación Molecular, Universidad de Talca, 2 Norte 685, Talca, Chile.

ABSTRACT

Background: Since the structure of proteins is more conserved than the sequence, the identification of conserved three-dimensional (3D) patterns among a set of proteins, can be important for protein function prediction, protein clustering, drug discovery and the establishment of evolutionary relationships. Thus, several computational applications to identify, describe and compare 3D patterns (or motifs) have been developed. Often, these tools consider a 3D pattern as that described by the residues surrounding co-crystallized/docked ligands available from X-ray crystal structures or homology models. Nevertheless, many of the protein structures stored in public databases do not provide information about the location and characteristics of ligand binding sites and/or other important 3D patterns such as allosteric sites, enzyme-cofactor interaction motifs, etc. This makes necessary the development of new ligand-independent methods to search and compare 3D patterns in all available protein structures.

Results: Here we introduce Geomfinder, an intuitive, flexible, alignment-free and ligand-independent web server for detailed estimation of similarities between all pairs of 3D patterns detected in any two given protein structures. We used around 1100 protein structures to form pairs of proteins which were assessed with Geomfinder. In these analyses each protein was considered in only one pair (e.g. in a subset of 100 different proteins, 50 pairs of proteins can be defined). Thus: (a) Geomfinder detected identical pairs of 3D patterns in a series of monoamine oxidase-B structures, which corresponded to the effectively similar ligand binding sites at these proteins; (b) we identified structural similarities among pairs of protein structures which are targets of compounds such as acarbose, benzamidine, adenosine triphosphate and pyridoxal phosphate; these similar 3D patterns are not detected using sequence-based methods; (c) the detailed evaluation of three specific cases showed the versatility of Geomfinder, which was able to discriminate between similar and different 3D patterns related to binding sites of common substrates in a range of diverse proteins.

Conclusions: Geomfinder allows detecting similar 3D patterns between any two pair of protein structures, regardless of the divergency among their amino acids sequences. Although the software is not intended for simultaneous multiple comparisons in a large number of proteins, it can be particularly useful in cases such as the structure-based design of multitarget drugs, where a detailed analysis of 3D patterns similarities between a few selected protein targets is essential.

No MeSH data available.