Limits...
VENNTURE--a novel Venn diagram investigational tool for multiple pharmacological dataset analysis.

Martin B, Chadwick W, Yi T, Park SS, Lu D, Ni B, Gadkaree S, Farhang K, Becker KG, Maudsley S - PLoS ONE (2012)

Bottom Line: An improved appreciation of the connectivity between multiple, highly-complex datasets is crucial for the next generation of data analysis of genomic and proteomic data streams.Applied to complex pharmacological datasets, VENNTURE's improved features and ease of analysis are much improved over currently available Venn diagram programs.This study highlights the potential for such a program in fields such as pharmacology, genomics, and bioinformatics.

View Article: PubMed Central - PubMed

Affiliation: Metabolism Unit, Laboratory of Clinical Investigation, National Institute on Aging, National Institutes of Health, Baltimore, Maryland, United States of America.

ABSTRACT
As pharmacological data sets become increasingly large and complex, new visual analysis and filtering programs are needed to aid their appreciation. One of the most commonly used methods for visualizing biological data is the Venn diagram. Currently used Venn analysis software often presents multiple problems to biological scientists, in that only a limited number of simultaneous data sets can be analyzed. An improved appreciation of the connectivity between multiple, highly-complex datasets is crucial for the next generation of data analysis of genomic and proteomic data streams. We describe the development of VENNTURE, a program that facilitates visualization of up to six datasets in a user-friendly manner. This program includes versatile output features, where grouped data points can be easily exported into a spreadsheet. To demonstrate its unique experimental utility we applied VENNTURE to a highly complex parallel paradigm, i.e. comparison of multiple G protein-coupled receptor drug dose phosphoproteomic data, in multiple cellular physiological contexts. VENNTURE was able to reliably and simply dissect six complex data sets into easily identifiable groups for straightforward analysis and data output. Applied to complex pharmacological datasets, VENNTURE's improved features and ease of analysis are much improved over currently available Venn diagram programs. VENNTURE enabled the delineation of highly complex patterns of dose-dependent G protein-coupled receptor activity and its dependence on physiological cellular contexts. This study highlights the potential for such a program in fields such as pharmacology, genomics, and bioinformatics.

Show MeSH

Related in: MedlinePlus

Cellular context modification of downstream receptor signaling activity in SH-SY5Y cells.(A) Comparison of phosphoprotein/GO term group/canonical signaling pathway activation in the non-stimulated (blue circle) and multiple MeCh-stimulated conditions (red circles) in either control-state (solid line) or CMP-state (dashed lines) SH-SY5Y cells. (B) Summary of potential non-stimulated and MeCh-induced signaling functionality (aggregate of highest-scoring GO-term group and canonical signaling pathway enrichments) in a dose-dependent manner compared between control-state or CMP-state cells.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3351456&req=5

pone-0036911-g007: Cellular context modification of downstream receptor signaling activity in SH-SY5Y cells.(A) Comparison of phosphoprotein/GO term group/canonical signaling pathway activation in the non-stimulated (blue circle) and multiple MeCh-stimulated conditions (red circles) in either control-state (solid line) or CMP-state (dashed lines) SH-SY5Y cells. (B) Summary of potential non-stimulated and MeCh-induced signaling functionality (aggregate of highest-scoring GO-term group and canonical signaling pathway enrichments) in a dose-dependent manner compared between control-state or CMP-state cells.

Mentions: Using VENNTURE, we were able to discern that many of the functional outputs of MeCh-mediated muscarinic receptor stimulation were strongly dose-dependent and also largely distinct at each different dose. We further investigated how this signaling diversity was specifically affected by dose between our two experimental cellular contexts, i.e. control state versus CMP state. We found that for phosphoproteins that were specific to MeCh-stimulated VENNTURE sets (Figure 5A–B, Venn sets 2–62) a similar percentage, in both cellular contexts, of these proteins were found in ‘dose-unique’ sets only (36% control-state, 38% CMP-sate). When analyzing the VENNTURE distribution of the GO terms and canonical signaling pathways, we found a largely similar distribution of factors to these ‘dose-unique’ Venn sets again for both cellular contexts, i.e. approximately 36–39% of the MeCh-stimulated GO term/signaling pathways (excluding phosphoproteins common to non-stimulated samples) were ‘dose-unique’ and in the CMP groups 35–38% were ‘dose-unique’ (data not shown). We then sub-divided these ‘dose-unique’ sectors into the respective dose-specific proportions of phosphoproteins/GO terms/signaling pathways (Figure 6). At the phosphoprotein level the percentage bias of each dose within the total number of ‘dose-unique’ phosphoproteins was generally similar in the control state cells (10 nM-15%, 100 nM-15%, 1 µM-21%, 10 µM-22%, 100 µM-27%). In contrast, for the CMP-state cells, a reduction in the percentage of phosphoproteins unique to the lower doses was observed (10 nM-7%, 100 nM-11%: Figure 6-Protein) compared to control-state cells. As proteins can possess multiple functions, and therefore could be grouped into several GO term groups or signaling pathways, it would be expected that a divergence in relative dose-specific proportions in these functional groups would occur. A stark contrast was observed with respect to the ‘dose-unique’ profile of GO term groups between the control and CMP-state cells. The relative proportions of ‘dose-unique’ GO term groups were roughly similar in control-state cells (Figure 6-GO term). However in CMP cells more than half (54%) of the total number of ‘dose-unique’ significantly-populated GO term groups were found at the 100 µM dose. Concomitantly, a large reduction of the low-MeCh dose-specific GO term groups was observed (10 nM-4% in CMP cells versus 24% in control cells: Figure 6-GO term). At the signaling pathway level, the dose-related functional output differences were even more diverse between control and CMP-state cells. In each cell context, one specific MeCh dose appeared to generate the most extensive signaling repertoire, i.e. ability to stimulate many forms of potential signaling cascades. Such activity is indicated by the largest numbers of significantly-populated signaling pathways populated by protein phosphorylation events induced by a specific ligand dose. In the control state cells this dose was 100 nM MeCh, while in the CMP-state cells this dose was 10 µM MeCh (Figure 6-Signaling pathway). No significant signaling pathways were populated by the lowest MeCh dose (10 nM) in the CMP-state cells, while 11% of total ‘dose-unique’ signaling pathways were stimulated specifically by 10 nM MeCh in the control-state cells. Such ligand dose- and cellular context-dependent alteration of receptor signaling profiles may be suggestive of altered populations of muscarinic receptors in stable isomeric conformations that are ‘hard-wired’ into specific signaling paradigms [8], [9], [13], [29], [30]. Reinforcing the considerable differences observed between ‘dose-unique’ proportions of phosphoprotein/GO term/signaling pathways in the control- or CMP-state cells, we found that in a dose-by-dose comparison there was minimal cross-over between proteins/GO terms/signaling pathways between these two cell contexts (Figure 7A). Combining and summarizing the functional annotation information (both GO term and canonical signaling pathway enrichment) for each dose of MeCh, it was apparent that distinct potential functional outcomes were generated by different MeCh doses within a single cellular context (Figure 7B). Moreover, in a dose-to-dose comparison, between cellular context models, profoundly distinct signaling outcomes were generated between these two cellular states (control versus CMP: Figure 7B). This may suggest that the functional outcomes of receptor stimulation may be considerably disrupted by the multiple proteomic/genomic alterations induced by CMP treatment which, in-part, mimics cell aging [14]. While the CMP treatment may affect dynamic responses to receptor stimulation, it clearly also exerts multiple and differential effects upon steady state (non-stimulated experiments) protein expression as well (Figure 7B:Tables S17, S23, S33, S34). Hence in non-stimulated control-state cells cyclin-dependent kinases signaling was evident, while fatty acid synthesis and apoptotic mechanisms were evident in non-stimulated CMP-state cells (Figure 7B). With respect to the dynamic signaling responses with 10 nM MeCh stimulation of control cells, an output related to spindle kinetics, molecular motor function and energy regulatory pathways was predicted, while similar MeCh-stimulation of CMP-treated SH-SY5Y cells suggested only effects upon cell motility. With 100 nM MeCh stimulation of control state cells, stimulation of microtubule-associated protein kinases, DNA regulatory behavior and neuroprotective kinase (phosphoinositide-3-kinase) signaling was predicted, while a distinct output phenotype was present in CMP cells, i.e. stimulation of synaptic signaling and endocytic activity was predicted. With 1 µM stimulation, again a strong divergence of signaling activity for MeCh was seen between control and CMP cellular states. In control cells, at a 1 µM dose, stimulation of histone modifying activity and mRNA processing was observed, while in CMP-cells signaling activity appeared to center more upon neuronal filament/stress fiber modification, amine metabolism and neurotrophic signaling. With 10 µM stimulation, MeCh-stimulation of control-state cells regulated nucleotide-dependent kinase activity as well as signaling involved in pro-excitatory neurotransmission. The 10 µM stimulation of CMP-state cells however was less phenotypically robust and was instead linked to cation binding, genomic stabilization and lipid-mediated stress signaling. At the highest dose of MeCh employed (100 µM), stimulation of stress fiber, proteasomal and gene regulatory activity was observed in control-state cells, while in CMP-state cells we noted a strong neuronal differentiation, metabolic and biosynthetic bias of signaling (Figure 7B:Tables S31, S32-GO terms; S35, S36-canonical signaling pathways).


VENNTURE--a novel Venn diagram investigational tool for multiple pharmacological dataset analysis.

Martin B, Chadwick W, Yi T, Park SS, Lu D, Ni B, Gadkaree S, Farhang K, Becker KG, Maudsley S - PLoS ONE (2012)

Cellular context modification of downstream receptor signaling activity in SH-SY5Y cells.(A) Comparison of phosphoprotein/GO term group/canonical signaling pathway activation in the non-stimulated (blue circle) and multiple MeCh-stimulated conditions (red circles) in either control-state (solid line) or CMP-state (dashed lines) SH-SY5Y cells. (B) Summary of potential non-stimulated and MeCh-induced signaling functionality (aggregate of highest-scoring GO-term group and canonical signaling pathway enrichments) in a dose-dependent manner compared between control-state or CMP-state cells.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3351456&req=5

pone-0036911-g007: Cellular context modification of downstream receptor signaling activity in SH-SY5Y cells.(A) Comparison of phosphoprotein/GO term group/canonical signaling pathway activation in the non-stimulated (blue circle) and multiple MeCh-stimulated conditions (red circles) in either control-state (solid line) or CMP-state (dashed lines) SH-SY5Y cells. (B) Summary of potential non-stimulated and MeCh-induced signaling functionality (aggregate of highest-scoring GO-term group and canonical signaling pathway enrichments) in a dose-dependent manner compared between control-state or CMP-state cells.
Mentions: Using VENNTURE, we were able to discern that many of the functional outputs of MeCh-mediated muscarinic receptor stimulation were strongly dose-dependent and also largely distinct at each different dose. We further investigated how this signaling diversity was specifically affected by dose between our two experimental cellular contexts, i.e. control state versus CMP state. We found that for phosphoproteins that were specific to MeCh-stimulated VENNTURE sets (Figure 5A–B, Venn sets 2–62) a similar percentage, in both cellular contexts, of these proteins were found in ‘dose-unique’ sets only (36% control-state, 38% CMP-sate). When analyzing the VENNTURE distribution of the GO terms and canonical signaling pathways, we found a largely similar distribution of factors to these ‘dose-unique’ Venn sets again for both cellular contexts, i.e. approximately 36–39% of the MeCh-stimulated GO term/signaling pathways (excluding phosphoproteins common to non-stimulated samples) were ‘dose-unique’ and in the CMP groups 35–38% were ‘dose-unique’ (data not shown). We then sub-divided these ‘dose-unique’ sectors into the respective dose-specific proportions of phosphoproteins/GO terms/signaling pathways (Figure 6). At the phosphoprotein level the percentage bias of each dose within the total number of ‘dose-unique’ phosphoproteins was generally similar in the control state cells (10 nM-15%, 100 nM-15%, 1 µM-21%, 10 µM-22%, 100 µM-27%). In contrast, for the CMP-state cells, a reduction in the percentage of phosphoproteins unique to the lower doses was observed (10 nM-7%, 100 nM-11%: Figure 6-Protein) compared to control-state cells. As proteins can possess multiple functions, and therefore could be grouped into several GO term groups or signaling pathways, it would be expected that a divergence in relative dose-specific proportions in these functional groups would occur. A stark contrast was observed with respect to the ‘dose-unique’ profile of GO term groups between the control and CMP-state cells. The relative proportions of ‘dose-unique’ GO term groups were roughly similar in control-state cells (Figure 6-GO term). However in CMP cells more than half (54%) of the total number of ‘dose-unique’ significantly-populated GO term groups were found at the 100 µM dose. Concomitantly, a large reduction of the low-MeCh dose-specific GO term groups was observed (10 nM-4% in CMP cells versus 24% in control cells: Figure 6-GO term). At the signaling pathway level, the dose-related functional output differences were even more diverse between control and CMP-state cells. In each cell context, one specific MeCh dose appeared to generate the most extensive signaling repertoire, i.e. ability to stimulate many forms of potential signaling cascades. Such activity is indicated by the largest numbers of significantly-populated signaling pathways populated by protein phosphorylation events induced by a specific ligand dose. In the control state cells this dose was 100 nM MeCh, while in the CMP-state cells this dose was 10 µM MeCh (Figure 6-Signaling pathway). No significant signaling pathways were populated by the lowest MeCh dose (10 nM) in the CMP-state cells, while 11% of total ‘dose-unique’ signaling pathways were stimulated specifically by 10 nM MeCh in the control-state cells. Such ligand dose- and cellular context-dependent alteration of receptor signaling profiles may be suggestive of altered populations of muscarinic receptors in stable isomeric conformations that are ‘hard-wired’ into specific signaling paradigms [8], [9], [13], [29], [30]. Reinforcing the considerable differences observed between ‘dose-unique’ proportions of phosphoprotein/GO term/signaling pathways in the control- or CMP-state cells, we found that in a dose-by-dose comparison there was minimal cross-over between proteins/GO terms/signaling pathways between these two cell contexts (Figure 7A). Combining and summarizing the functional annotation information (both GO term and canonical signaling pathway enrichment) for each dose of MeCh, it was apparent that distinct potential functional outcomes were generated by different MeCh doses within a single cellular context (Figure 7B). Moreover, in a dose-to-dose comparison, between cellular context models, profoundly distinct signaling outcomes were generated between these two cellular states (control versus CMP: Figure 7B). This may suggest that the functional outcomes of receptor stimulation may be considerably disrupted by the multiple proteomic/genomic alterations induced by CMP treatment which, in-part, mimics cell aging [14]. While the CMP treatment may affect dynamic responses to receptor stimulation, it clearly also exerts multiple and differential effects upon steady state (non-stimulated experiments) protein expression as well (Figure 7B:Tables S17, S23, S33, S34). Hence in non-stimulated control-state cells cyclin-dependent kinases signaling was evident, while fatty acid synthesis and apoptotic mechanisms were evident in non-stimulated CMP-state cells (Figure 7B). With respect to the dynamic signaling responses with 10 nM MeCh stimulation of control cells, an output related to spindle kinetics, molecular motor function and energy regulatory pathways was predicted, while similar MeCh-stimulation of CMP-treated SH-SY5Y cells suggested only effects upon cell motility. With 100 nM MeCh stimulation of control state cells, stimulation of microtubule-associated protein kinases, DNA regulatory behavior and neuroprotective kinase (phosphoinositide-3-kinase) signaling was predicted, while a distinct output phenotype was present in CMP cells, i.e. stimulation of synaptic signaling and endocytic activity was predicted. With 1 µM stimulation, again a strong divergence of signaling activity for MeCh was seen between control and CMP cellular states. In control cells, at a 1 µM dose, stimulation of histone modifying activity and mRNA processing was observed, while in CMP-cells signaling activity appeared to center more upon neuronal filament/stress fiber modification, amine metabolism and neurotrophic signaling. With 10 µM stimulation, MeCh-stimulation of control-state cells regulated nucleotide-dependent kinase activity as well as signaling involved in pro-excitatory neurotransmission. The 10 µM stimulation of CMP-state cells however was less phenotypically robust and was instead linked to cation binding, genomic stabilization and lipid-mediated stress signaling. At the highest dose of MeCh employed (100 µM), stimulation of stress fiber, proteasomal and gene regulatory activity was observed in control-state cells, while in CMP-state cells we noted a strong neuronal differentiation, metabolic and biosynthetic bias of signaling (Figure 7B:Tables S31, S32-GO terms; S35, S36-canonical signaling pathways).

Bottom Line: An improved appreciation of the connectivity between multiple, highly-complex datasets is crucial for the next generation of data analysis of genomic and proteomic data streams.Applied to complex pharmacological datasets, VENNTURE's improved features and ease of analysis are much improved over currently available Venn diagram programs.This study highlights the potential for such a program in fields such as pharmacology, genomics, and bioinformatics.

View Article: PubMed Central - PubMed

Affiliation: Metabolism Unit, Laboratory of Clinical Investigation, National Institute on Aging, National Institutes of Health, Baltimore, Maryland, United States of America.

ABSTRACT
As pharmacological data sets become increasingly large and complex, new visual analysis and filtering programs are needed to aid their appreciation. One of the most commonly used methods for visualizing biological data is the Venn diagram. Currently used Venn analysis software often presents multiple problems to biological scientists, in that only a limited number of simultaneous data sets can be analyzed. An improved appreciation of the connectivity between multiple, highly-complex datasets is crucial for the next generation of data analysis of genomic and proteomic data streams. We describe the development of VENNTURE, a program that facilitates visualization of up to six datasets in a user-friendly manner. This program includes versatile output features, where grouped data points can be easily exported into a spreadsheet. To demonstrate its unique experimental utility we applied VENNTURE to a highly complex parallel paradigm, i.e. comparison of multiple G protein-coupled receptor drug dose phosphoproteomic data, in multiple cellular physiological contexts. VENNTURE was able to reliably and simply dissect six complex data sets into easily identifiable groups for straightforward analysis and data output. Applied to complex pharmacological datasets, VENNTURE's improved features and ease of analysis are much improved over currently available Venn diagram programs. VENNTURE enabled the delineation of highly complex patterns of dose-dependent G protein-coupled receptor activity and its dependence on physiological cellular contexts. This study highlights the potential for such a program in fields such as pharmacology, genomics, and bioinformatics.

Show MeSH
Related in: MedlinePlus