The first 2 authors contributed equally in this article.
Pathway sensitivity analysis for detecting pro-proliferation activities of oncogenes and tumor suppressors of epidermal growth factor receptor-extracellular signal-regulated protein kinase pathway at altered protein levels
Version of Record online: 23 JUN 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 18, pages 4246–4263, 15 September 2009
How to Cite
Li, H., Ung, C. Y., Ma, X. H., Liu, X. H., Li, B. W., Low, B. C. and Chen, Y. Z. (2009), Pathway sensitivity analysis for detecting pro-proliferation activities of oncogenes and tumor suppressors of epidermal growth factor receptor-extracellular signal-regulated protein kinase pathway at altered protein levels. Cancer, 115: 4246–4263. doi: 10.1002/cncr.24485
- Issue online: 4 SEP 2009
- Version of Record online: 23 JUN 2009
- Manuscript Accepted: 9 FEB 2009
- Manuscript Revised: 12 JAN 2009
- Manuscript Received: 3 OCT 2008
- Academic Research Fund. Grant Number: R-148-000-081-112 of 101
- National University of Singapore
- epidermal growth factor receptor-extracellular signal-regulated protein kinase pathway;
- pathway simulation;
- biologic networks;
- system biology
Mathematic models and sensitivity analyses of biologic pathways have been used for exploring the dynamics and for detecting the key components of signaling pathways.
The authors previously developed a mathematic model of the epidermal growth factor receptor-extracellular signal-regulated protein kinase (EGFR-ERK) pathway using ordinary differential equations from existing EGFR-ERK pathway models. By using prolonged ERK activation as an indicator that may lead to cell proliferation under certain circumstances, in the current study, a pathway sensitivity analysis was performed to test its capability of detecting pro-proliferative activities through altered protein levels to examine the effects on ERK activation.
The analysis revealed that 12 of 20 oncoproteins and 4 of 5 tumor suppressors were detected, consistent with reported experimental works. Because pathway dynamics depend on many factors, some of which were not included in the current models, failure to detect all known oncogenes and tumor suppressors can be because of the failure to include relevant crosstalk to other pathway components.
Overall, the current results indicated that pathway sensitivity analysis is a useful approach for detecting and distinguishing pro-proliferation activities of oncoproteins and suppressed proliferative activities of tumor suppressors at altered protein levels at least in the EGFR-ERK model. Cancer 2009. © 2009 American Cancer Society.
Mathematic models of biologic pathways have been explored extensively for simulating the dynamics of signaling networks1-8 and for studying network responses to drugs and other regulators.9, 10 In addition to the studies of signaling dynamics and responses, sensitivity analyses of pathway models have been conducted and have demonstrated promising potential for identifying regulators that affect cell fate,11 key network components,12 robust and fragile points,13-16 and signal-altering single point mutations in specific oncogenes and tumor suppressor genes.17, 18 We believed that it would be interesting to evaluate further the capability of pathway sensitivity analysis for detecting pro-proliferation activities of oncoproteins and tumor suppressors that may lead to deregulated proliferative activities at altered protein levels.
Oncogenes and tumor suppressors are key genes that promote tumorigenesis through mutations and altered expression.19 Because there are strong correlations between their protein levels with proliferative signals, and because many of these proteins have been included in available mathematic models of pathways, it is now of particular interest to test the capability of pathway sensitivity analysis in detecting tumorigenesis activities of these proteins. In the current work, by using a well studied mathematic model of the epidermal growth factor receptor-extracellular signal-regulated protein kinase (EGFR-ERKP pathway that we developed previously,20 we tested the capability of pathway sensitivity analysis for detecting pro-proliferation activities of oncogenes and tumor suppressors in the EGFR-ERK pathway at altered protein levels.
The EGFR-ERK pathway is involved in the regulation of normal cell proliferation, survival, and differentiation.21, 22 Aberrant regulation of this pathway contributes to cancer23 and other diseases, such as developmental disorders,24 cardiovascular disease,25, 26 and urinary bladder dysfunction.27 Deregulation of this pathway frequently arises from somatic mutations and altered expression levels of some components of the EGFR-ERK pathway, such as EGFR, Ras, and Raf, leading to cancer,28, 29 developmental disorders,24 Noonan syndrome,30 and Costello syndrome.31 Some of these components have been used as primary targets for drug discovery in the treatment of relevant diseases.32
To a large extent, the outcomes of EGFR-ERK pathway regulation are determined by the duration, magnitude, and subcellular compartmentalization of ERK activation.21, 22, 33-35 The degree and duration of ERK activation are important both in disease processes, such as cancer, and in normal cellular processes in response to external stimuli.22 Although transient ERK activation leads to proliferation in specific cell types, such as PC12 nerve cells, in many other cell types, such as fibroblasts, Rat-1, hepatocytes, T lymphocytes, and smooth muscle cells, sustained ERK activation is necessary for proliferation.35, 36 Therefore, for many cell types, activities that lead to sustained ERK activation may be considered pro-proliferation activities. The strength of ERK activity also is a determinant of cell fate, but the actual outcome is regulated by complicated interactions involving multiple pathways.22 Many of these interactions are not included in the currently available models of the EGFR-ERK pathway. Hence, these currently available models may not be used in a straightforward manner for studying the influence of the strength of ERK activity on the outcome of cell fate without adding relevant interactions. In the current work, we used prolonged ERK activation as an indicator of the pro-proliferation signal that possibly may lead to proliferation in certain cell types under specific conditions.
The model of the EGFR-ERK pathway that we used in this study is illustrated in Figure 1, which includes all of the components of the relevant pathway models published in the literature.1-6, 20 Although not all of the cell fate-regulatory elements are included, this model is 1 of the most extensively studied, and it captures a variety of observed behaviors.1-6, 20 Therefore, it represents the best available model for testing the capability of pathway sensitivity analysis for detecting pro-proliferation activities in the EGFR-ERK pathway at altered protein levels. The results of pathway sensitivity analysis were compared with relevant experimental studies for their activities against ERK and their roles in tumorigenesis.
MATERIALS AND METHODS
Model Development and Collection of Kinetics Parameters
One of the most commonly used approaches to model the dynamics of biologic pathway systems is that of ordinary differential equations (ODEs). In general, a differential equation can be used to describe the chemical reaction rate that depends on the change of participating species over time. The temporal dynamic behavior of molecular species in the biologic signaling pathway network can be captured by a set of coupled ODEs.37 The overall pathway internal links and architecture of our mathematic model is illustrated in Figure 1. The information for the topology of the crosstalk was collected from various published works.1-6 All equations for molecular interactions in this study were derived based on laws of Mass Action. The kinetic parameters for forming protein-protein interaction complexes in forward (Kf) and complex dissociation in backward reactions (Kb) were used. For enzymatic and irreversible reactions, the reaction constant (K) and turn over rates (Kcats) were used instead of the Michaelis-Menten constants (Km), which are applicable primarily to steady-state models.
The kinetic parameters that were used in this study were obtained mostly from published experimental data. For those kinetic parameters that were unavailable from experimental data, similarly based strategies were used to derive putative kinetic constants. Ranges of kinetic parameters were constrained based on the literature data and in vivo measurements of signaling kinetics.2-6, 38 A set of coupled ODEs was used to describe the reaction network. Our model contains 205 equations and interactions with 194 distinct molecular species characterized by 313 kinetic parameters and 38 initial molecular concentrations. These ODEs were then solved using the Ode45 solver of MatLab (MathWorks, Natick, Mass).39
Model Optimization and Validation
Mathematic models that are developed at the systems level generally are not expected to reproduce exact quantitative values in all systems but are capable of producing known behavior or trends of specific systems that are close to the system based on which model is developed. For instance, mathematic models that are developed for a biologic pathway from parameters obtained experimentally from 1 cell type can behave slightly different in another cell type. The difference in the model's behavior in these cell types can be because of the presence or absence of crosstalk (ie, the topology and, hence, the boundary of the mathematic model) and variation in values of the kinetic parameters that are used. In the current study, we tentatively used a generic model of the EGFR-ERK signaling pathway to investigate the capability of pathway sensitivity analysis.
The simulated results are represented in curves of concentrations of a molecular species over time that are validated against available experimental data. If the trend or dynamics of the curves of some particular reactants or products behave as the experimental data suggest, then the model is considered to be optimized and can be used to analyze and predict unknown biologic phenomena within the boundary of the model. If the simulation results were not in fair agreement with known experimental facts, then the definition as well as the boundary of the model has to be revisited to examine possible errors, such as incorrect interaction kinetics or values of kinetics parameters.
Optimized parameters obtained from previous mathematic models were not necessarily optimized in the current study, because the boundaries of these models were different. Because a biologic network is robust and binding affinity of protein-protein interactions for proteins of similar families that mediate similar types of biochemical reactions (such as Ras and Ras homolog gene family, member A [RhoA]) differ within a 10-fold range, the values of kinetic parameters obtained from previous models are optimized within these ranges. Likewise, for parameters that were unavailable from previous models, parameters of proteins in similar families were used tentatively and were optimized further. For instance, the parameters for the RhoA activation cycle were obtained from those for the Ras activation cycle and were optimized further in 10-fold ranges. The cycle of optimization and validation was repeated until the simulated results were in fair agreement with known experimental trends.
The dynamics of biologic pathways depend on multiple factors, and all components collectively affect and regulate each other to produce a dynamic behavior that cannot be predicted from just a few simple correlations among some of the components in the pathway. Thus, currently, the most widely used approach for investigating the regulation of biologic pathways is to develop a mathematic model of biologic pathway and simulate its dynamics behavior. However, the developed model is limited by the definition of the pathway boundary, which determines the extent of the components of a pathway that can be simulated. The development of a pathway model, thus, is closely related to the biologic question that is addressed. In this work, we built a mathematic model of the EGFR-ERK pathway that includes scaffold-mediated ERK activation with crosstalk to small GTPase RhoA. By using this model, we intended to test the capability of pathway sensitivity analysis in detecting cancer-related genes in the EGFR-ERK pathway, which is the most characterized to date.
Our mathematic model has been validated, as described in our previous work20 in which the simulated profiles of time-dependent protein concentrations were in reasonable agreement with the available experimental data and with reported computational studies. Table 1 provides a summary of the results of pathway sensitivity analyses. The simulated activity profile of ERK at different levels of EGF, Ras GTPase-activating protein (RasGAP), Raf, ERK, RhoA, Rho guanine nucleotide exchange factor (RhoGEF), Src homologous and collagen (Shc), and mitogen-activated protein kinase kinase 1 (MEKK1), are displayed in Figure 2. Based on the observation that elevations and reductions of oncoproteins are normally within a 10-fold range relative to normal levels, as described above (see Materials and Methods), in pathway sensitivity analysis, we altered the level of a particular protein by 10-fold for both up-regulation and down-regulation while keeping the levels of other proteins unchanged.
|Signaling Molecule||Predicted Effect at Elevated Expression Levels||Predicted Effect at Reduced Expression Levels||Literature-Reported Functional Role and Expression Profile of the Gene Product in Cancer (References)||Consistency Between Predicted Proliferative Implication and Literature Reports (References)|
|Expression Level at 10 Fold Elevation, μM||Effect on ERK Activation and Possible Implication for Proliferation||Expression Level at 10 Fold Reduction, μM||Effect on ERK Activation and Possible Implication for Proliferation|
|EGF||8.1967e-2||Significantly prolonged duration of transient ERK activation, slightly favoring proliferation||8.1967e-4||Reduced duration of transient ERK activation, significantly reduced active ERK concentration||Oncogene of EGFR-Ras-ERK pathway; significant relations between EGF expression and tumor size (Suo 2002,40 Zhang 200541)||Consistent|
|EGFR||3||Unaffected transient ERK activation||0.03||Unaffected transient ERK activation||Oncogene of EGFR-Ras-ERK pathway, successful anticancer target; overexpressed in breast, head, and neck cancers (Do 200442); mutations and amplifications in breast, head and neck, lung, colon, brain tumors, and glioblastoma (Zhang 200541)||Partly consistent: Enhanced EGFR levels through reducing Cbl-mediated endocytosis prolonged ERK activation, and vice versa (Ung 200820)|
|SHP||1||Unaffected transient ERK activation||0.01||Unaffected transient ERK activation||High-frequency silencing of hematopoietic cell-specific protein-tyrosine phosphatase SHP1 gene by promoter methylation has been detected in various kinds of leukemias and lymphomas and in many hematopoietic cell lines, suggesting that functional loss of SHP1 is associated with the pathogenesis of leukemias/lymphomas (Oka 200243)||Inconsistent: It has been demonstrated that tyrosine phosphatase SHP1 positively regulates the induction of interferon-beta in toll-like receptor signaling, which was not included in this study (O'Neill 200844)|
|Shc||10||Sustained ERK activation, favoring proliferation||0.1||Reduced duration of transient ERK activation, reduced active ERK concentration||A common substrate of EGFR and other RTKs (Nolan 199745); expression correlated with proliferation in prostate (Veeramani 200546), gastric (Yukimasa 200547), breast (Jackson 200048), and chronic myelogenous leukemias (Bonati 200049)||Consistent|
|Grb2||10||Reduced duration of transient ERK activation, slightly reduced active ERK concentration||0.1||Significantly reduced duration of transient ERK activation, slightly reduced active ERK concentration||Grb2 is involved in various cancers and specifically in metastasis-related processes (Giubellino 2005,50 Dharmawardana 2006,51 Watanabe 200052)||Consistent|
|SOS||3||Reduced duration of transient ERK activation, substantially reduced active ERK concentration||0.03||Unaffected transient ERK activation||Expression of SOS proteins was substantially increased in all human bladder cancer cell lines (Watanabe 200052)||Inconsistent: Sos is a bifunctional GEF that activates Ras and displays RacGEF activity; the complex Sos-1-E3b1-Eps8, endowed with Rac-specific GEF activities (Innocenti 200253) and PI3K, activates Rac by entering into a complex with Eps8, Abi1, and Sos (Innocenti 200354), which were not considered in this study|
|Ras||1.5||Significantly prolonged duration of transient ERK activation, slightly reduced active ERK concentration, favoring proliferation||0.015||Slightly reduced duration of transient ERK activation||Oncogene of EGFR-Ras-ERK pathway; involved in leukemias and in lung, ovarian, colon, and pancreatic cancers; overexpression is associated with a poor prognosis (Zhang 200555)||Consistent|
|Raf||5||Sustained ERK activation, favoring proliferation||0.05||Slightly reduced duration of transient ERK activation||Oncogene of EGFR-Ras-ERK pathway; amplified in several tumors, including bladder, hormone-resistant prostate, nasopharyngeal carcinoma. and anaplastic large cell leukemia (Zhang 2005,55 Simon 2001,56 Mao 2003,57 Tupputi 1992,58 Hui 200259)||Consistent|
|MEK||6.8||Sustained ERK activation, favoring proliferation||0.068||Slightly reduced duration and substantially reduced amplitude of transient ERK activation||Oncogene of EGFR-Ras-ERK pathway and anticancer target; aberrant expression frequently observed in cancers (Zhang 2005,55 Smith 200660)||Consistent|
|ERK||4||Significantly prolonged duration of transient ERK activation, slightly favoring proliferation||0.04||Significantly prolonged duration of transient ERK activation, reduced active ERK concentration||Oncogene of EGFR-Ras-ERK pathway; overexpressed in various tumors of epithelial origin (Kiyokawa 1994,61 Zhang 200555)||Consistent|
|PP2A||0.2||Reduced duration of transient ERK activation, reduced active ERK concentration||0.002||Sustained ERK activation, favoring proliferation||Tumor suppressor against certain Ras actions; dephosphorylated kinases in Raf-MEK-ERK pathway (Fukukawa 2005,62 Letournex 200663) and Ras effectors c-Myc and RalA. (Mumby 200764), expression levels distinctly lower in liver cancer (Kitamura 199265), brain tumors (Colella 200166)||Consistent|
|MKP3||0.03||Reduced duration of transient ERK activation, reduced active ERK concentration||0.0003||Sustained ERK activation, favoring proliferation||Tumor suppressor against ERK actions; directly dephosphorylated ERK (Bermudez 200867), down-regulated in invasive pancreatic cancer (Furukawa 200368)||Consistent|
|RasGAP||1||Slightly reduced duration of transient ERK activation||0.01||Prolonged duration of transient ERK activation||A major down-regulator of ras activity, under expressed in human trophoblastic tumors, and presumably acts as a tumor suppressor gene product in these neoplasms (Davidson 199869); RasGAP expression in PC12 cells resulted in substantial inhibition of sustained MAP kinase activity after NGF treatment (Yao & Cooper 199570)||Consistent|
|PI3K||0.1||Slightly reduced duration of transient ERK activation||0.001||Unaffected transient ERK activation||PI3K is overexpressed in various cancer types, including carcinoma of the breast, prostate, colon, and endometrium (Zhao & Vogt 200871)||Inconsistent: Components of PI3K-AKT pathway, such as mTOR, TSC1/TSC2, and Rheb (Franke 200872), were not included in this model|
|Akt||1||Unaffected transient ERK activation||0.01||Unaffected transient ERK activation||The Akt gene frequently is hyperactivated in tumors, and its amplification has been demonstrated in a several types of human cancers (Han 200873)||Inconsistent: Components of PI3K-AKT pathway. such as mTOR, TSC1/TSC2, and Rheb (Franke 200872), were not included in this model|
|PTEN||1||Unaffected transient ERK activation||0.01||Unaffected transient ERK activation||Tumor-suppressor of PI3K-AKT pathway (Salmena 200874); PTEN activity lost at high frequency in many primary and metastatic human cancers (Blanco-Aparicio 200775)||Inconsistent: Components of PI3K-AKT pathway, such as mTOR, TSC1/TSC2, and Rheb (Franke 200872), were not included in this model|
|RacGEF||1||Unaffected transient ERK activation||0.01||Unaffected transient ERK activation||Tiam1, a GEF of Rac, is overexpressed in a subset of human colorectal tumors and regulates cell adhesion, migration, and apoptosis in colon tumor cells (Minard 200676)||Inconsistent: Components of Rac-Tiam1 signaling, such as p38 MAPK, PAK, and actin cytoskeleton, were not considered in this model (Mertens 200377)|
|Rac||2||Unaffected transient ERK activation||0.02||Unaffected transient ERK activation||Activation of Rac in various cancers, such as lung cancer (Kissil 200778)||Inconsistent: Components of Rac-Tiam1 signaling, such as p38 MAPK, PAK, and actin cytoskeleton, were not considered in this model (Mertens 200377)|
|RhoA||1.5||Significantly prolonged duration of transient ERK activation, slightly reduced active ERK concentration, slightly favoring proliferation||0.015||Slightly reduced duration of transient ERK activation||Cross-talked with EGFR-Ras-ERK pathway to favor proliferation (Sahai 200179); overexpression associated with progression in breast cancers (Bellizzi 200880) and colorectal cancer (Takami 200881) and with a poor prognosis in liver cancer (Xiaorong 200882)||Consistent|
|SHP2||1||Slightly reduced duration of transient ERK activation||0.01||Slightly prolonged duration of transient ERK activation||Inhibition of SHP2 led to a mesenchymal-to-epithelial transition in breast cancer cells (Zhou & Agazie 200883)||Consistent|
|RhoGEF||1||Significantly prolonged duration of transient ERK activation, slightly reduced active ERK concentration, slightly favoring proliferation||0.01||Slightly reduced duration of transient ERK activation, slightly increased active ERK concentration||RhoGEF activity activates RhoA and is required for transformation (Sahay 200884), leukemias, and neck squamous cell carcinoma (Bourguignon 200685)||Consistent|
|ROCK||6.8||Significantly prolonged duration of transient ERK activation, slightly reduced active ERK concentration||0.068||Slightly reduced duration of transient ERK activation||ROCK is involved in cell movement and metastasis of cancers (Joshi 2008,86 Rosel 200887)||Consistent|
|Src||5.18||Reduced duration of transient ERK activation, reduced active ERK concentration||0.0518||Sustained ERK activation, favoring proliferation||Expression and kinase activity frequently increased in a wide array of cancers, including tumors from breast, colon, pancreas, lung, ovary, and CNS (Watanabe & Yamada 199988)||Inconsistent: Signaling components of Src, such as Csk, paxillin, and Gab1 (Ren 200489). were not considered in this work|
|Complex of Cbl and CIN85||8||Significantly reduced duration of transient ERK activation||0.08||Significantly prolonged duration of transient ERK activation.||CIN85 and Cbl are important suppressors of growth factor receptor signaling for the invasive activities of breast cancer cells (Nam 200790)||Consistent|
|MEKK1||5||Significantly prolonged duration of transient ERK activation, slightly favoring proliferation||0.05||Reduced duration of transient ERK activation, reduced active ERK concentration||Research anticancer target; expression of lethal doses of MEKK1 strongly activated ERK, JNK, and p38 (Boldt 200391)||Consistent|
Four known oncogenes (EGF, Ras, Raf, and ERK) in the pathway had significantly prolonged duration of ERK activation from a transient signal into a more sustained-like signal with a 10-fold elevation of protein level, whereas a 10-fold reduction in these protein levels primarily produced a transient-like profile (Fig. 2a-d). Thus, the results suggested that these 4 oncogenes exhibited pro-proliferation activities at elevated levels, consistent with their reported oncogenic roles, as summarized in Table 1. At elevated levels, RhoA, RhoGEF, Shc, and MEKK1 significantly prolonged ERK activation and, thus, exhibited pro-proliferation activities (Fig. 2e-h). RhoA is an oncoprotein of the RhoA/signal transducer and activator of transcription (STAT) pathway and the RhoA/RhoA-associated kinase (ROCK)/cytoskeleton pathway92, 93 that reportedly crosstalks with the EGFR-Ras-ERK pathway to favor proliferation.39 RhoGEF is an oncoprotein of the RhoA-STAT pathway and the RhoA-ROCK-cytoskeleton pathway that exerts its functions in part through the regulation of RhoA.92, 93 Shc is a common substrate of EGFR and other receptor tyrosine kinases,45 and its expression reportedly is correlated with proliferation in prostate cancer,46 gastric cancer,47 breast cancer,91 and chronic myelogenous leukemias.69 In addition, it has been reported that the expression of lethal doses of MEKK1 strongly activate ERK, c-Jun N-terminal kinase (JNK), and p38.92 Hence, pathway sensitivity analysis is able to detect the pro-proliferative activities of these proteins through ERK activation and is consistent with reported experimental findings.
Figure 3a,b illustrates the 10-fold reduction of 2 tumor suppressors (protein phosphatase 2A [PP2A] and mitogen-activated protein kinase phosphatase 3 [MKP3]) in the pathway, which prolonged ERK activation; whereas a 10-fold elevation produced a transient-like profile, indicating the suppressive role of PP2A and MKP3 in cell proliferation, consistent with their reported roles as tumor suppressors, as summarized in Table 1. At reduced levels, the Ras GTPase-activating protein (RasGAP) prolonged ERK activation and, thus, pro-proliferation activities were observed (Fig. 3c). RasGAP is a negative regulator of Ras activity that reportedly is under expressed in human trophoblastic tumors and that presumably acts as a tumor suppressor gene product in these neoplasms.69 It also was reported that there was no expression in neurofibromatosis type 1 sarcomas94 or in colon cancer specimens.95 RasGAP expression in PC12 cells resulted in substantial inhibition of sustained mitogen activated protein kinase (MAPK) activity after nerve growth factor treatment.70 Hence, the predicted activity of RasGAP on ERK activation seems to be consistent with these experimental findings. Nonetheless, RasGAP expression has been observed in some percentage of colonic, gastric, and lung cancer cell lines.95 Hence, combinations of multiple factors most likely define the signaling outcome that leads to proliferation, and the duration of ERK activation is just 1 of the important indicators for assessing the likelihood that a protein contributes to the proliferative signaling process.
Because the performance of pathway sensitivity analysis is highly dependent on the boundary of a pathway model, the lack of some components in the current pathway model (including those that mediate crosstalk with other pathways) may lead to superficially insensitive, dynamic behavior upon ERK activation that deviates from true behavior. EGFR is known as an oncoprotein of the EGFR-Ras-ERK pathway that is overexpressed in breast cancer and in head and neck cancers79 and is an anticancer target. However, pathway sensitivity analysis in this work indicated that ERK activation is relatively insensitive to altered levels of EGFR in the 10-fold range. This is because the commonly used initial concentration of EGFR, which was also used in this work, is at a near-saturated level. However, the inclusion of additional components, such as receptor endocytosis, revealed that the oncogenic properties of EGFR may be produced. Our earlier work demonstrated that mutations that impaired the c-Cbl-EGFR association delayed EGFR endocytosis and produced higher mitogenic signals.20 Figure 4a illustrates that increased levels of the 85-kDa Cbl-interacting protein (Cbl-CIN85) enhanced internalized endophilin A1-associated EGFR through receptor endocytosis, which reduced the duration of ERK activation (Fig. 4b). Hence, mutational analysis using this model revealed that EGFR possesses pro-proliferative activity.
Src is an oncogenic kinase that frequently has increased expression in a wide array of cancers, including tumors of the breast, colon, pancreas, lung, ovary, and central nervous system.96 Our pathway sensitivity analysis indicated that increased Src levels reduced the duration of ERK activation, whereas decreased Src levels caused sustained ERK activation, inconsistent with previous experimental work that demonstrated the oncogenesis properties of Src. Currently, to the best of our knowledge, there is no report indicated that reduced Src levels prolong ERK activation. Although no studies have demonstrated that Src down-regulation leads to prolonged ERK activation through the conventional Ras-Raf-mitogen-activated or extracellular signal-regulated protein kinase (MEK)-ERK cascade, there is experimental evidence suggesting that Src inhibition prolongs ERK activation through the delta opioid receptor.97 Audet and coworkers showed that Src inhibitor PP2 also prolonged ERK stimulation by D-Pen-2,5-enkephalin. It did so by maintaining sustained activation of the kinase at approximately 20% of the maximum after an initial rapid reduction in the response.97 However, whether this process is mediated by the Ras-Raf-MEK-ERK cascade remain to be ruled out. In our model, Src is activated through phosphorylation by active EGFR. The activated Src subsequently phosphorylates both RhoGEF98 and RhoGAP.99 RhoGEF activates RhoA through an enhanced exchange of GDP for GTP that subsequently promotes ERK activation. Conversely, RhoGAP promotes GTP hydrolysis of active RhoA, leading to its inhibitory function. Hence, in our current model, Src has dual functions that, at the same time, promote and decrease ERK activation through RhoA. The dual functions of Src are expected to coordinate with other pathway components to produce its known oncogenic properties. Signaling components of Src, such as C-terminal Src kinase, paxillin, growth factor receptor-bound protein 2-associated binding protein 1,89 and their relevant components, are not considered in this work, which most likely is part of the reason for the simulated insensitivity of ERK activation to the alteration of Src levels.
It has been demonstrated that phosphatidylinositol 3-kinase (PI3K) is overexpressed in various types of cancer, including cancers of the breast, prostate, colon, and endometrium.71 In addition, the Akt gene frequently is hyperactivated in tumors and reportedly is amplified in several human cancers.73 Moreover, the lost of phosphatase and tensin homolog (PTEN) phosphatase activity occurs at high frequency in many primary and metastatic human cancers.75 In our model, ROCK (which is a mediator of small GTPase RhoA) phosphorylates PTEN, which leads to its activation.100, 101 The activated PTEN catalyzes the conversion of plasma membrane intrinsic protein 3 (PIP3) to PIP2, which subsequently inhibits the activity of protein kinase B (AKT). Hence, PTEN is a phosphatase that acts as a tumor suppressor of the PI3K-AKT pathway by through inhibiting the activity of AKT.74 However, our current pathway sensitivity analysis indicated that ERK activation is relatively insensitive to altered levels of PI3K, AKT, and PTEN. The inconsistency of our analysis with reported experimental works most likely is caused in part to the lack of other components of the PI3K pathway, such as mammalian target of Rapamycin, tuberous sclerosis gene 1 (TSC1)/TSC2, and Ras homolog enriched in brain,72 and their relevant components in this model. Other oncogenic and tumor suppressor proteins that do not sensitively affect ERK activation at altered levels in our sensitivity analysis are son of sevenless (SOS), Rac, RacGEF, and protein-tyrosine phosphatase (SHP). The lack of relevant crosstalk in these proteins is most likely a key reason for the failure to detect these proteins, as summarized in Table 1.
Overall, pathway sensitivity analysis had reasonably good capacity for detecting the pro-proliferative activities of oncogenes and the suppressed proliferative activities of tumor suppressors. Of 20 oncogenic proteins and 5 tumor suppressors, 12 oncoproteins and 4 tumor suppressors were detected by pathway sensitivity analysis, consistent with reported experimental works, as summarized in Table 1. Such capacity is expected to be enhanced further with expanded knowledge of the networks, interactions, and crosstalk that collectively regulate ERK activation, proliferation, and other processes.22 Apart from detecting the pro-proliferation activities of oncogenes, tumor suppressors, and cancer-related genes, the identified sensitive components of the EGFR-ERK pathway (defined as the proteins that, at altered levels, sensitively affect ERK activation) also are consistent with those of the reported sensitivity studies. For instance, Mayawala and coworkers reported that MEK and ERK are sensitive components of the MAPK cascade.13 In another analysis, Pant and Ghosh identified MEK, ERK, Raf, Ras, PP2A, MKP1, SOS, and RasGAP and as sensitive components of the canonical EGFR pathway.17 Table 1 indicates that all of these proteins were predicted as sensitive components by our analysis.
Our study and earlier studies consistently have suggested that sensitivity analysis of biologic pathways provides useful hints regarding the distinctive activities of oncogenes, tumor suppressors, and other disease-related proteins11, 17, 18 in addition to the ability to analyze network robustness and identify fragile and regulatory points of biologic networks.12-16 Thus, simulation studies at the systems level can complement other high-throughput and screening technologies, such as microarray102, 103 and proteomics analyses,104, 105 for disease analysis and for gene and target discovery. Active signaling proteins such, as phosphorylated forms, usually are present at low amounts that are difficult to detect using mass spectrometry in proteomic analysis.104 Some malfunctions and disorders are caused by mutations that have little effect on expression levels. These activities cannot be captured easily by microarray or proteomic analyses. These problems can be alleviated in part by simulation and sensitivity analysis of biologic pathways. Such studies also facilitate the identification of network regulatory points for multicomponent therapies.106-112
Conflict of Interest Disclosures
Supported by Academic Research Fund (R-148-000-081-112 of 101), National University of Singapore.
- 39MathWorks. Matlab. Natick, Mass: The MathWorks, Inc.; 2007.
- 58Anesthesia and pregnancy in oral medicine [in Italian]. Minerva Anestesiol. 2003; 58: 1051-1056., , , .
- 85Hyaluronan-CD44 interaction with leukemia-associated RhoGEF and epidermal growth factor promotes Rho/Ras co-activation, phospholipase C epsilon-Ca2+ signaling, and cytoskeleton modification in head and neck squamous cell carcinoma cells. J Biol Chem. 2006; 281: 14026-14040., , , , .