Wang et al., (2009). PI3K-dependent cross-talk interactions converge with Ras as quantifiable inputs integrated by Erk.
November 2013, model of the month by Lu Li
Original model: BIOMD0000000288
Cross-talks of signalling pathways are of vital importance, because they deliver the extracellular information among a wide variety of intracellular targets, and increase the spectrum of cellular responses. However, to evaluate and assess links between pathways is difficult because of their complex nature (Hunter, 2000). This is especially true for platelet-derived growth factor (PDGF) activated phosphoinositide 3-kinase (PI3K) pathway and Ras/extracellular signal-regulated kinase (ERK), which interact in various ways and in a context-dependent manner.
Here, Wang et al. (2009)  have demonstrated how cross links of signalling cascades can be analysed systematically via combing experimental and modelling approaches.
First of all, the authors assessed the connection between Ras and PI3K pathways by applying large-scale quantitative immunoblotting on protein extract from NIH 3T3 fibroblasts after different stimulations and treatments. More specifically, inhibiting Ras activation, reduces ERK responses upon all PDGF stimulating doses and durations; whereas incubating cells with PI3K inhibitor, impairs ERK activation at low doses of PDGF, not at saturating concentrations (Figure 1AB). At these high doses, inhibition PI3K does not affect peak ERK activation but its sustainability. Blocking both activities completely abolishes ERK's response to PDGF. Furthermore, restraining the function of MEK, significantly reduces the response of ERK to PDGF (Figure 1D). These data indicate that Ras and PI3K signalling cascades are responsible for PDGF induced ERK phosphorylation in these fibroblast cells. Although they activate ERK through different pathways and mechanisms, their efforts converge on the phosphorylation of MEK.
Interestingly, the authors revealed that expression of constitutively active Ras reduces the peak level of ERK activity induced by PDGF (Figure 1C). And these observations are complemented by increased expression of MAPK phosphatase (MKP-1), which affects both threonine and tyrosine phosphorylation sites on ERK. It seems there is a negative feedback on ERK activation arising from ERK itself because of its influence on the level of MKP-1.
Figure 2 Quantitative Ras-GTP loading measurements: characterization of PI3K-dependent cross-talk and ERK-dependent feedback regulation. NIH 3T3 fibroblasts were stimulated with PDGF as indicated; pretreatments were control (0.2% DMSO), LY294002 (PI3K inhibitor, 100 μM), or PD098059 (MEK inhibitor, 50 μM). Ras-GTP levels were measured using a quantitative enzymatic assay and normalized as described in the paper; values are reported as mean±s.e.m. (n=3). Figures are taken from .
Figure 3 Conceptual model of the PDGF receptor signalling network. Figures are taken from .
Based on these observed conceptual relationships between PI3K and Ras cascades and the constrains set by these dose-response experimental readouts, the authors developed a mathematical model as shown in Figure 3. This model incorporates 34 unspecified parameters, estimated based on Metropolis algorithm (Metropolis, Rosenbluth, Rosenbluth, Teller, & Teller, 1953). Finally, 10000 sets of parameters are generated, which fit equally well with experimental data. Based on these parameter sets and simulation results produced, the authors compared the catalytic efficiencies between PI3K and Ras pathway on MEK phosphorylation. As shown in Figure 4A, There is a consistent ratio between PI3K and Ras dependent contribution to the first phosphorylation site of MEK. However, Ras shows a more potent effect on dually phosphorylated MEK. Besides, the ensemble of parameters indicates the coexistence of PI3K dependent and independent Ras activation mechanism, and shows that they play equally important roles (Figure 4B).
Finally, the authors predicted that at low PDGF concentration, blocking either the upstream or downstream of PI3K and Ras cross-link would place the level of ERK phosphorylation in between control and PI3K inhibited condition. However, at high PDGF concentration, inhibiting the cross-talk that is at the upstream of Ras will not influence ERK activation at all, whereas impairing the mechanism at downstream of Ras will generate similar ERK phosphorylation as if PI3K is inhibited (Figure 5).
Therefore by applying data-driven analysis of the kinetic model, the authors revealed the unique kinetic signature of the cross-talk between PI3K and Ras signalling pathway: At low PDGF concentrations, the effect of PI3K and Ras on ERK activation follows AND logic; whereas they follows OR when PDGF concentration is high. These findings can be verified in future experiments.
Figure 1 Systematic analysis of PDGF-stimulated ERK phosphorylation kinetics. (A-D) Quantification of ERK phosphorylation, normalized as described in the paper, comparing either Ras inhibition (A; n=6), PI3K inhibition (B; n=5), constitutively active Ras (C; n=6), or MEK inhibition (D; n=5) with their respective controls. Values are reported as mean±s.e.m., and comparisons to control in (A, B) are by Student's t-test: *P<0.05; **P<0.01. Figure taken from .
Using a coupled enzymatic assay, the authors measured how PI3K influences the kinetics of Ras activation. At low concentration (but not high doses) of PDGF stimulation, impairing PI3K function attenuates transient Ras activation, demonstrating that PI3K pathway influences upstream of Ras. Blocking MEK dramatically increases Ras activity level, indicating the negative feedback from ERK to Ras (Figure 2).
Figure 4 Quantitative analysis of PI3K-dependent cross-talk to Ras/Erk. (A) Quantification of Ras- and PI3K-dependent MEK phosphorylation pathways. The quantity Cxij is defined as the maximum catalytic efficiency of pathway i (i=1, Ras dependent; i=2, PI3K dependent) toward site j on MEK divided by the catalytic efficiency of the corresponding phosphatase reaction. On the dashed line, the two pathways are equally potent by this measure. (B) Quantification of the PI3K-independent and -dependent modes of Ras-GEF recruitment, characterized by the model parameters KGR and KGP respectively, plotted here for each parameter set in the ensemble. The factor of 123 is a scaling factor; when KGP=KGR/123 (dashed line), the two models contribute equally to Ras-GEF recruitment in the limit of low PDGF concentration. Figure taken from .
Figure 5 Model predictions based on ensemble averaging. Predictions are based on hypothetical interventions by which the PI3K-dependent cross-talk upstream or downstream of Ras are selectively blocked. Solid curves represent ensemble means, and dashed curves are mean±s.d. (n=10000). PDGF concentrations are: red, 1nM; black, 30pM. Figure taken from .
- Hunter, T. Signaling--200 and beyond. Cell. (2000), Jan 7;100(1):113-27.
- Metropolis et al. Equation of state calculations by fast computing machines. J Chem Phys. (1953), 21, 1087.
- Wang et al. PI3K-dependent cross-talk interactions converge with Ras as quantifiable inputs integrated by Erk. Mol Syst Biol. (2009), 5:246.