Limits...
Towards the systematic discovery of signal transduction networks using phosphorylation dynamics data.

Imamura H, Yachie N, Saito R, Ishihama Y, Tomita M - BMC Bioinformatics (2010)

Bottom Line: The number of identified phosphoproteins and their phosphosites is rapidly increasing as a result of recent mass spectrometry-based approaches.We found that peptides extracted from the same intracellular fraction (nucleus vs. cytoplasm) tended to be close together within this phosphorylation dynamics-based network.The network was then analyzed using graph theory and compared with five known signal-transduction pathways.

View Article: PubMed Central - HTML - PubMed

Affiliation: Institute for Advanced Biosciences, Keio University, Tsuruoka, Yamagata, Japan.

ABSTRACT

Background: Phosphorylation is a ubiquitous and fundamental regulatory mechanism that controls signal transduction in living cells. The number of identified phosphoproteins and their phosphosites is rapidly increasing as a result of recent mass spectrometry-based approaches.

Results: We analyzed time-course phosphoproteome data obtained previously by liquid chromatography mass spectrometry with the stable isotope labeling using amino acids in cell culture (SILAC) method. This provides the relative phosphorylation activities of digested peptides at each of five time points after stimulating HeLa cells with epidermal growth factor (EGF). We initially calculated the correlations between the phosphorylation dynamics patterns of every pair of peptides and connected the strongly correlated pairs to construct a network. We found that peptides extracted from the same intracellular fraction (nucleus vs. cytoplasm) tended to be close together within this phosphorylation dynamics-based network. The network was then analyzed using graph theory and compared with five known signal-transduction pathways. The dynamics-based network was correlated with known signaling pathways in the NetPath and Phospho.ELM databases, and especially with the EGF receptor (EGFR) signaling pathway. Although the phosphorylation patterns of many proteins were drastically changed by the EGF stimulation, our results suggest that only EGFR signaling transduction was both strongly activated and precisely controlled.

Conclusions: The construction of a phosphorylation dynamics-based network provides a useful overview of condition-specific intracellular signal transduction using quantitative time-course phosphoproteome data under specific experimental conditions. Detailed prediction of signal transduction based on phosphoproteome dynamics remains challenging. However, since the phosphorylation profiles of kinase-substrate pairs on the specific pathway were localized in the dynamics-based network, our method will be a complementary strategy to explore new components of protein signaling pathways in combination with previous methods (including software) of predicting direct kinase-substrate relationships.

Show MeSH

Related in: MedlinePlus

Characteristics of the phosphorylation dynamics-based network. (A) We generated the dynamics-based network by connecting pairs of peptides with similar (R > 0.99) time courses of phosphorylation activities. The network was visualized using Cytoscape (version 2.6.1) [45] and eXpanda (version 1.0.6) [46]. (B) Network density of the whole dynamics-based network and of the cytoplasmic and nucleic subnetworks. (C) Cumulative proportion, P ≥ (k), for the node degrees (k) based on analysis of the whole dynamics-based network simultaneously (i.e., not separately as in Figure D) (R > 0.99). For each group of cytoplasmic and nucleic nodes in the network, circles represent the proportions of proteins having more than k interacting partners. (D) Cumulative proportion for the node degrees with the cytoplasmic and nucleic subnetworks analyzed separately. (E, I, K) Patterns of the cellular fractions (cytoplasmic and nucleic): (E) binary, (I) triangular, and (K) square motifs that appeared in the dynamics-based network. The names of each motif pattern appear under the corresponding diagram: T, triangular; B, binary; S, square. (F-H, J, L) Appearance of each motif (proportion of total) in the dynamics-based network. (F-H) Triangular motifs appeared in the dynamics-based network of (F) R > 0.99, (G) R > 0.98, and (H) R > 0.97. (J) Binary and (L) square motifs appeared in the dynamics-based network with R > 0.99. Black bars represent percentages of the corresponding motif patterns in the real dynamics-based network; white bars represent the mean values of the percentages estimated using negative controls generated by random edge rewiring (RER, n = 1000). Error bars represent standard deviations. Significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2875242&req=5

Figure 1: Characteristics of the phosphorylation dynamics-based network. (A) We generated the dynamics-based network by connecting pairs of peptides with similar (R > 0.99) time courses of phosphorylation activities. The network was visualized using Cytoscape (version 2.6.1) [45] and eXpanda (version 1.0.6) [46]. (B) Network density of the whole dynamics-based network and of the cytoplasmic and nucleic subnetworks. (C) Cumulative proportion, P ≥ (k), for the node degrees (k) based on analysis of the whole dynamics-based network simultaneously (i.e., not separately as in Figure D) (R > 0.99). For each group of cytoplasmic and nucleic nodes in the network, circles represent the proportions of proteins having more than k interacting partners. (D) Cumulative proportion for the node degrees with the cytoplasmic and nucleic subnetworks analyzed separately. (E, I, K) Patterns of the cellular fractions (cytoplasmic and nucleic): (E) binary, (I) triangular, and (K) square motifs that appeared in the dynamics-based network. The names of each motif pattern appear under the corresponding diagram: T, triangular; B, binary; S, square. (F-H, J, L) Appearance of each motif (proportion of total) in the dynamics-based network. (F-H) Triangular motifs appeared in the dynamics-based network of (F) R > 0.99, (G) R > 0.98, and (H) R > 0.97. (J) Binary and (L) square motifs appeared in the dynamics-based network with R > 0.99. Black bars represent percentages of the corresponding motif patterns in the real dynamics-based network; white bars represent the mean values of the percentages estimated using negative controls generated by random edge rewiring (RER, n = 1000). Error bars represent standard deviations. Significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Mentions: We constructed three dynamics-based networks, using threshold values of R > 0.97, 0.98, and 0.99. The dynamics-based network reconstructed with a threshold value of R > 0.99 contained 4,907 edges and 851 nodes (of which 377 and 474 nodes were cytoplasmic and nucleic, respectively; Figure 1A and Additional File 1). The network with R > 0.98 consisted of 10,626 edges and 959 nodes (of which 423 and 536 were from the cytoplasm and nucleus, respectively; Additional File 2), and the network with R > 0.97 consisted of 16,481 edges and 1,015 nodes (of which 442 and 573 were from the cytoplasm and nucleus, respectively; Additional File 3).


Towards the systematic discovery of signal transduction networks using phosphorylation dynamics data.

Imamura H, Yachie N, Saito R, Ishihama Y, Tomita M - BMC Bioinformatics (2010)

Characteristics of the phosphorylation dynamics-based network. (A) We generated the dynamics-based network by connecting pairs of peptides with similar (R > 0.99) time courses of phosphorylation activities. The network was visualized using Cytoscape (version 2.6.1) [45] and eXpanda (version 1.0.6) [46]. (B) Network density of the whole dynamics-based network and of the cytoplasmic and nucleic subnetworks. (C) Cumulative proportion, P ≥ (k), for the node degrees (k) based on analysis of the whole dynamics-based network simultaneously (i.e., not separately as in Figure D) (R > 0.99). For each group of cytoplasmic and nucleic nodes in the network, circles represent the proportions of proteins having more than k interacting partners. (D) Cumulative proportion for the node degrees with the cytoplasmic and nucleic subnetworks analyzed separately. (E, I, K) Patterns of the cellular fractions (cytoplasmic and nucleic): (E) binary, (I) triangular, and (K) square motifs that appeared in the dynamics-based network. The names of each motif pattern appear under the corresponding diagram: T, triangular; B, binary; S, square. (F-H, J, L) Appearance of each motif (proportion of total) in the dynamics-based network. (F-H) Triangular motifs appeared in the dynamics-based network of (F) R > 0.99, (G) R > 0.98, and (H) R > 0.97. (J) Binary and (L) square motifs appeared in the dynamics-based network with R > 0.99. Black bars represent percentages of the corresponding motif patterns in the real dynamics-based network; white bars represent the mean values of the percentages estimated using negative controls generated by random edge rewiring (RER, n = 1000). Error bars represent standard deviations. Significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2875242&req=5

Figure 1: Characteristics of the phosphorylation dynamics-based network. (A) We generated the dynamics-based network by connecting pairs of peptides with similar (R > 0.99) time courses of phosphorylation activities. The network was visualized using Cytoscape (version 2.6.1) [45] and eXpanda (version 1.0.6) [46]. (B) Network density of the whole dynamics-based network and of the cytoplasmic and nucleic subnetworks. (C) Cumulative proportion, P ≥ (k), for the node degrees (k) based on analysis of the whole dynamics-based network simultaneously (i.e., not separately as in Figure D) (R > 0.99). For each group of cytoplasmic and nucleic nodes in the network, circles represent the proportions of proteins having more than k interacting partners. (D) Cumulative proportion for the node degrees with the cytoplasmic and nucleic subnetworks analyzed separately. (E, I, K) Patterns of the cellular fractions (cytoplasmic and nucleic): (E) binary, (I) triangular, and (K) square motifs that appeared in the dynamics-based network. The names of each motif pattern appear under the corresponding diagram: T, triangular; B, binary; S, square. (F-H, J, L) Appearance of each motif (proportion of total) in the dynamics-based network. (F-H) Triangular motifs appeared in the dynamics-based network of (F) R > 0.99, (G) R > 0.98, and (H) R > 0.97. (J) Binary and (L) square motifs appeared in the dynamics-based network with R > 0.99. Black bars represent percentages of the corresponding motif patterns in the real dynamics-based network; white bars represent the mean values of the percentages estimated using negative controls generated by random edge rewiring (RER, n = 1000). Error bars represent standard deviations. Significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Mentions: We constructed three dynamics-based networks, using threshold values of R > 0.97, 0.98, and 0.99. The dynamics-based network reconstructed with a threshold value of R > 0.99 contained 4,907 edges and 851 nodes (of which 377 and 474 nodes were cytoplasmic and nucleic, respectively; Figure 1A and Additional File 1). The network with R > 0.98 consisted of 10,626 edges and 959 nodes (of which 423 and 536 were from the cytoplasm and nucleus, respectively; Additional File 2), and the network with R > 0.97 consisted of 16,481 edges and 1,015 nodes (of which 442 and 573 were from the cytoplasm and nucleus, respectively; Additional File 3).

Bottom Line: The number of identified phosphoproteins and their phosphosites is rapidly increasing as a result of recent mass spectrometry-based approaches.We found that peptides extracted from the same intracellular fraction (nucleus vs. cytoplasm) tended to be close together within this phosphorylation dynamics-based network.The network was then analyzed using graph theory and compared with five known signal-transduction pathways.

View Article: PubMed Central - HTML - PubMed

Affiliation: Institute for Advanced Biosciences, Keio University, Tsuruoka, Yamagata, Japan.

ABSTRACT

Background: Phosphorylation is a ubiquitous and fundamental regulatory mechanism that controls signal transduction in living cells. The number of identified phosphoproteins and their phosphosites is rapidly increasing as a result of recent mass spectrometry-based approaches.

Results: We analyzed time-course phosphoproteome data obtained previously by liquid chromatography mass spectrometry with the stable isotope labeling using amino acids in cell culture (SILAC) method. This provides the relative phosphorylation activities of digested peptides at each of five time points after stimulating HeLa cells with epidermal growth factor (EGF). We initially calculated the correlations between the phosphorylation dynamics patterns of every pair of peptides and connected the strongly correlated pairs to construct a network. We found that peptides extracted from the same intracellular fraction (nucleus vs. cytoplasm) tended to be close together within this phosphorylation dynamics-based network. The network was then analyzed using graph theory and compared with five known signal-transduction pathways. The dynamics-based network was correlated with known signaling pathways in the NetPath and Phospho.ELM databases, and especially with the EGF receptor (EGFR) signaling pathway. Although the phosphorylation patterns of many proteins were drastically changed by the EGF stimulation, our results suggest that only EGFR signaling transduction was both strongly activated and precisely controlled.

Conclusions: The construction of a phosphorylation dynamics-based network provides a useful overview of condition-specific intracellular signal transduction using quantitative time-course phosphoproteome data under specific experimental conditions. Detailed prediction of signal transduction based on phosphoproteome dynamics remains challenging. However, since the phosphorylation profiles of kinase-substrate pairs on the specific pathway were localized in the dynamics-based network, our method will be a complementary strategy to explore new components of protein signaling pathways in combination with previous methods (including software) of predicting direct kinase-substrate relationships.

Show MeSH
Related in: MedlinePlus