Demo-3
A Model of Signal Transduction from the Epidermal Growth Factor Receptor (EGFR) to the Cyclin D1 Gene
A Guided Tour

Demo-3 is designed to provide a glimpse of how Pathway Logic can be used by a researcher to assemble experimentally determined signaling processes into a model, and then explore the model -- perhaps to design a new experiment.  Because the current viewer is just a prototype, the full power of Pathway Logic for modeling and analyzing signaling networks (or other biological systems) will not be obvious.  However, as Pathway Logic is further developed for online use it will become clear that it does not merely "connect the dots", but offers a versatile set of tools for logically reasoning about very complex biology.

Like all Pathway Logic models, the heuristic model illustrated in Demo-3 is based on conclusions curated from published literature concerning biochemical reactions and processes that are well established experimentally.  Demo-3 is focused on intracellular signaling pathways initiated by binding of the epidermal growth factor (EGF) to its receptor EGFR (ErbB1).  As in Demo-1 and -2, the biochemical reactions comprising these pathways were converted into Maude rules, a Petri net was generated from these rules, and the Petri net was loaded as a file capable of being displayed by the Pathway Logic viewer.

Here we describe an analysis of the classic or canonical EGF-to-ERK signaling pathway.  EGF binds to the EGFR and activates its protein tyrosine kinase function, the adaptor protein Grb2 binds to the activated (phosphorylated) EGFR and recruits the guanine nucleotide exchange factor Sos, the Sos complex activates a Ras family GTPase, activated Ras activates a member of the Raf serine/threonine protein kinase family, which activates the dual-specificity protein kinase MEK1/2, which activates ERK1/2.  In this classic pathway, activated ERK1/2 phosphorylates transcription factor targets that then induce new gene expression.  However, you will see that the network displayed in Demo-3 contains much more information than just this linear pathway.  To isolate and display a particular pathway within this network, you must instruct the viewer accordingly.  Notice that the instructions described below assume some knowledge of viewer commands (see 'Using Pathway Logic' for a summary of these commands).

The rules and references specific for each rule can be found in rules3.maude.

(1) Find the Canonical Pathway From EGF to ERK2

First Operation: Show all the pathways leading from EGF-EGFR to ERK2 activation. Clearly, there are several pathways from EGF binding to the EGFR leading to Erk2 activation.  In fact, the command 'Show Relevant Subnet' works by starting at the goal and then proceeding 'backwards' to EGF.  Any step in the original network that is not involved in the transduction of a signal from EGF to Erk2 is removed, but there are still many steps that are relevant to reaching the goal.

Second Operation: Find the shortest pathway leading from EGF-EGFR to ERK2 activation.

The result of this operation is to display a version of the classic pathway, with a few additional details.  To simplify the display it is possible to remove the input nodes (grey) by the following operation:

(2) The Key Under The Lamppost (Or How Do You Know What's Real?)

Why did the 'Find A Pathway' command use Rafb instead of Raf1 to activate Mek1?   The reason is that, in this model, the pathway between EGF and Rafb is shorter than that between EGF and Raf1.  However, this result does not reflect nature, but is an artifact of the amount of information originally entered into the model.  In the literature there is considerably more information describing the activation of Raf1 than there is for the activation of Rafb.  Until recently, review articles relegated Rafb into the background with statements such as "B-Raf is activated by both Ras and Rap" [11294822].   Therefore, in the absence of more detailed information about the actiation of Rafb, our rule 775 states simply that 'activated Ras activates Rafb'.  In contrast, there is enough literature information concerning Raf1 activation [12127063] to write rule 280, which states that activation of Raf1 requires Ras as well as Pak1, Src, PP2A, PKCz and 14-3-3.

What happens if Rafb is removed from the model by the following operation?

Surprisingly, now the display shows that Erk2 is activated downstream of Rac1 -- a pathway commonly associated with Jnk activation.

(3) Is It True If It's In The Literature? (A Small Caveat)

The example described above illustrates an important point not only about Pathway Logic models of signaling networks but also about models of biological networks in general.  That is, the models are only as predictive as the information they contain.  In the case of Pathway Logic, all biological and chemical information is curated from the literature and then translated into rules that the system can read and use to assemble networks of hypothetical pathways, which evolve from a given starting state.  A strength of Pathway Logic is that it provides a researcher with both the opportunity to display the logical consequences of the interaction of documented biochemical processes or reactions, and the opportunity to test or question them.

If the evidence supporting a particular rule seems weak, try removing it to observe what happens to the signaling network.  For example, consider rule [148.Mek.is.act-2].  This rule can be paraphrased as 'activated Mekk1 activates Mek1 and/or Mek2'.  Although it was originally reported that Mekk1 can activate Mek1 [9305638], later it was found that EGF could still activate Erk1/2 in mouse embryo fibroblasts deficient for Mekk1 expression [12048245].  See what happens when Mekk1 is removed from the model by the following operation.

Now the classic EGF-to-Erk pathway is displayed with all the extra requirements for Raf1 activation as inputs.

(4) Why Are Fibronectin and Integrin-5-Beta-1 In the Model?

Interestingly, the displayed EGF-to-Erk pathway also includes inputs from Fibronectin (FN) binding to its receptor integrin-5-beta-1.  Here FN binding to this integrin receptor causes activation of p21-activated kinase 1 (Pak1) [12167697] which is required for activation of Raf1.  Whether there are other protein kinases that phosphorylate Raf1 at the same site is unknown, and therefore only Pak1 activation is included in the present model.  Biologically, attachment to a suitable substrate is necessary for most if not all adherent normal cells to survive and proliferate, particularly through Erk activation [8106557] [10777598].  See what happens when FN is removed from the model by the following operation. The result, 'Goal node is not reachable', is consistent with the requirement for FN engagement (substrate adherence) for the activation of Erk2 in response to EGF stimulation.

(4) Turning On the Cyclin D1 Gene

In various cycling cells (e.g., in the early G1 phase of the cell cycle), activation of the EGF-to-Erk pathway may result in the induction of proliferation-associated genes such as cyclin D1 [9710644].  The model in Demo-3 includes some of the inputs that have been reported to stimulate increased transcription of this gene.  Continuing the analysis described above, explore the role of Erk2 activation in this process with the following operation. The result is a collection of input pathways, including the classical EGF-to-Erk pathway, leading from EGF binding to the EGFR to the activation (turning on) of transcription of the cyclin D1 gene.  Further, because new protein synthesis also requires activation of the cellular translation apparatus, the following operation adds to this emerging model of cyclin D1 regulation by EGF.The result is an interesting Pathway Logic illustration of the overlap between or among signaling pathways that regulate the expression of genes such as cyclin D1 at both the transcriptional and translational levels.

The Maude model is available in Demo3Model

To try Demo 3, click on the appropriate link:  [light]  [heavy]

Back to the main Pathway Logic Demo page.