Biochemical Basis for the Functional Switch That Regulates Hepatocyte Growth Factor Receptor Tyrosine Kinase Activation^†

Detailed Mathematical Model

The model describes activation of c-MET by 28 biochemical events that include 15 intermediate species that are central to the activation process of the RTK. Coupled differential equations were used to express the time-dependent concentrations of reaction intermediates and products shown in Figure 1b. These sets of equations provide a mathematical description of the feed-forward loop shown in Figure 1a. Given initial conditions, rate constants for the forward and reverse reactions (k_on and k_off, respectively), catalytic efficiencies for the enzymatic reactions, and the time interval, Mathematica 4.1 was used to solve our set of simultaneous differential equations to give concentrations of any species at any given time point.

This model describes a minimal number of reactions required to generate a downstream signal (Figure 1b). In its simplest form, the c-MET receptor is at the cell surface as a nonsignaling monomer that can bind ATP and a tyrosine-containing effector molecule (Tyr) (reactions 1−4). The abbreviation Tyr, which is used throughout the paper, represents a generic effector substrate that contains a Tyr residue that undergoes phosphorylation. The monomeric, nonphosphorylated receptor either can become autophosphorylated (reaction 17, Figure 1b) or may phosphorylate the Tyr effector (reaction 19, Figure 1b). Similarly, the phosphorylated monomer can bind ATP and Tyr to form a ternary complex (reactions 5−8, Figure 1b). The phosphorylated and nonphosphorylated c-MET may dimerize in response to extracellular ligand binding (reactions 23, −23, 24, and −24, Figure 1b). Also, the phosphorylated receptor can be dephosphorylated (reactions 25 and 26, Figure 1b) by cellular protein tyrosine phosphatases (PTPs). Dephosphorylation of the Tyr effector is included in the model as this down-regulatory reaction is present in cells. Not included in the current model is the formation of multiple phosphorylated c-MET species, although the addition of multiple reactions and catalytic efficiencies can be easily done when experimental data become available for these reactions. The kinetic and thermodynamic parameters are for fully phosphorylated c-MET, as those are the data currently available either from our laboratory or from the literature as described below.

A complete set of the kinetic equations used to describe the activation of the c-MET receptor can be found in Table 1. For example, the concentration of the binary complex ([MET–ATP]) at any given time is equal to the rate of production of the binary complex less its consumption. The production of the binary complex can be described by the rate of formation from free MET and free ATP, the rate of dissociation of the ternary complex ([MET–ATP–Tyr]) into the MET–ATP complex and free Tyr, the rate of monomerization of the MET₂–ATP complex into the MET–ATP complex, and the dephosphorylation of the pMET–ATP complex into the MET–ATP complex. The rate of consumption of the binary MET–ATP complex is equal to the reverse of the reactions mentioned above with the loss due to the kinetic autophosphorylation reaction instead of the dephosphorylation reaction. Mathematically, this can be expressed as

where, ν_x is the rate of reaction x from Figure 1b. The exact values of the rate equations are enumerated in Table 2. In the overall activation process, the receptor concentration is assumed to be conserved such that [MET]_total = [MET]_free+[MET]+[MET–ATP]+[MET–Tyr]+[MET–ATP-Tyr]+[pMET]+[pMET–ATP]+[pMET–Tyr]+[pMET–ATP–Tyr] + [MET₂] + [MET₂–ATP] + [MET₂–Tyr] + [MET₂–ATP–Tyr] + [pMET₂] + [pMET₂–ATP] + [pMET₂–Tyr] + [pMET₂–ATP–Tyr] = constant. The initial response in the computer model easily accounts for this assumption. In cells, c-MET is internalized and degraded through endosomal trafficking, each of which occurs in the time frame of minutes (31, 32). Since we focus on initial activation of the receptor which occurs within ∼200 s, the contribution of receptor internalization and degradation is assumed to be negligible.

Parameter Values

Our detailed activation model containing the 28 biochemical events described earlier utilizes 48 different parameters (42 rate constants and six Michaelis–Menten constants). As described below, 36 of these 48 parameters are either directly measured or derived from the measured parameters for the c-MET system. This represents a considerable number of accurate c-MET experimental parameters for inclusion in a computational model of this size.

In our model, we use k_on and k_off values that had been previously determined in our laboratory for reactions 5−8 and 13−16 (22). Similarly, the kinetic catalytic efficiencies for reactions 20 and 22 were described by our laboratory (21). Although we have not directly measured k_on and k_off values for the dephosphorylated forms of c-MET, other groups showed that equilibrium dissociation constants between phosphorylated and dephosphorylated RTK catalytic domains differ by a factor of 2 (11). Thus, it was reasonable to estimate that the on and off rates for the ATP and effector to the dephosphorylated c-MET were on the same order of magnitude as those of the c-MET-phosphorylated form. This assumption was tested with our model, and order of magnitude changes to on and off rates for substrate binding to dephosphorylated c-MET did not affect the shape of the response curves for the total phosphorylated receptor and pTyr.

Naldini et al. (33) assessed the phosphorylation of exogenous substrates by phosphorylated and dephosphorylated c-MET immunoprecipitated from cell lysates and observed the phosphorylated receptor had ∼10-fold higher kinase activity than the dephosphorylated receptor. Our kinetic constants for phosphorylated active dimeric TPR–MET were similar to the values obtained by Naldini et al. (33) for the phosphorylated immunoprecipitated c-MET. Therefore, we assumed a 10-fold difference in catalytic efficiencies for the dephosphorylated c-MET to phosphorylate exogenous tyrosine substrates compared to phosphorylated c-MET (reactions 19 and 21 vs reactions 20 and 22). Similarly, the ability of unstimulated c-MET to undergo autophosphorylation was assumed to mimic the monomer's ability to phosphorylate exogenous effectors. Therefore, reactions 17 and 19 were assumed to have similar catalytic rates.

Although the mechanism of c-MET autophosphorylation is unknown, evidence of both intermolecular and intramolecular phosphorylation exists for different RTKs (10, 34–41). As a starting point for our model, monomer autophosphorylation was assumed to be an intramolecular reaction. Preliminary computation analyses using the autophosphorylation k_cat assigned to monomeric c-MET and a trans mode of autophosphorylation predicted similar levels of c-MET phosphorylation (data not shown). The catalytic efficiency of dephosphorylated dimeric c-MET is unknown. Since extracellular ligand stimulation induces an ∼10-fold increase in the autophosphorylation catalytic activity of the IR (42), the activity of reaction 18 was similarly increased relative to that of reaction 22.

Dephosphorylation rate constants for Shc and EGFR were used to approximate the dephosphorylation rates of exogenous effectors (Tyr) and c-MET (1). Shc is phosphorylated by c-MET when the receptor is activated by HGF (43, 44). It is a well-characterized substrate for c-MET and has been exhaustively studied in the context of c-MET signaling (44). Shimizu et al. (24) describe an ∼10-fold increase in the sensitivity of monomeric versus dimeric PDGFR toward dephosphorylation by recombinant phosphatases. Our studies measured a 10-fold decreased substrate specificity of recombinant PTPβ for phosphorylated dimeric TPR–MET as compared to phosphorylated monomeric cytoMET (20). We include a 10-fold difference in V_max between the two c-MET dephosphorylation reactions while keeping the K_m equal.

The number of c-MET molecules on the cell surface is unknown. However, in the EGFR system, there are multiple studies that measure ∼1−3 × 10⁵ molecules/cell, which translates into effective cellular EGFR concentrations within the range of 20−100 nM (1). We assumed c-MET and EGFR concentrations to be similar and estimated the c-MET concentration to be 100 nM for our model. The effect of c-MET concentration on the signaling output was simulated using this model. Increases in c-MET concentration have a linear impact on the signaling output, implying c-MET concentration does not play an important role in the synergy highlighted in the paper. The concentrations of tyrosine-containing effector molecules (Tyr) that were phosphorylated by c-MET were taken from previously published results to be 150 nM (1). The ATP concentration was assumed to be 1 mM, which was seen to be saturating for this system (45). Concentrations of c-MET and Tyr were assumed to be constant throughout the early time points investigated in our model. The ternary complexes of c-MET, ATP, and tyrosine-containing substrate or effector (Tyr) were assumed to be the necessary intermediate complex leading to effector phosphorylation. The mechanistic details of this phosphorylation reaction do not change the time-dependent buildup of phosphorylated c-MET and pTyr. All enzymatic reactions were assumed be irreversible, and PTPs were assumed to follow Michealis–Menten kinetics. The activation model used first-order mass-action kinetics and sets of differential equations described by specific kinetic and thermodynamic parameters.

Internalin B (InlB) was used as the extracellular ligand to activate c-MET in computer simulations and corresponding cell culture experiments. InlB was recently identified as a bacterial agonist for c-MET RTK and was shown to mimic c-MET activation and endocytosis following stimulation with a cognate HGF ligand (32, 46, 47). Furthermore, InlB initiated c-MET signaling as measured by phosphorylation of Gab1, Cbl, and Shc, along with activation of the PI3K pathway (48). K_d values for InlB binding to the extracellular domain of c-MET are in the nanomolar range (49). On and off rates for InlB–Met binding, which were more crucial for developing time-dependent models, are unknown. Thus, we use InlB on and off rates consistent with the equilibrium binding constant and the time constraints of the signaling process (50). With a k_on of 10 μM⁻¹ s⁻¹ and a k_off of 0.01 s⁻¹, the receptor reached a steady-state value within 2 min. Increasing or decreasing the on and off rates by a factor of 10 did not significantly affect the response curves that described buildup of phosphorylated c-MET or pTyr.

Experimental Validation of the c-MET Activation Model

The quantitative and qualitative predictions of the activation model were tested experimentally. The Vero cells (ATCC) used for these studies endogenously expressed c-MET. Prior to any stimulation with extracellular ligand, the cells were grown to ∼60% confluency and serum-starved for 12−24 h. For ligand stimulation experiments, cells were stimulated with 1, 5, and 20 nM InlB. At 0, 15, 60, 120, and 300 s post-InlB stimulation, the cells were washed twice with ice-cold PBS for 3 min and subsequently lysed in Triton X-100 lysis buffer [50 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X-100, 3 mM Na₃VO₄, 2 mM EDTA, 2 mM NaF, and protease inhibitor cocktail]. Protein concentrations in the lysates were determined using a Bio-Rad DC Assay, and 500−1000 μg of total protein was used for c-MET immunoprecipitation with the primary antibody to the c-MET extracellular region (antibody DO-24). Quantitative analyses to determine the percentage of total receptor that was phosphorylated used the strategy described by Kholodenko et al. (1). Briefly, cell lysates at each time point were divided into equal parts for parallel immunoprecipitation with the anti-phosphotyrosine (PY)–agarose conjugate and anti-MET (Upstate). The anti-PY immunoprecipitates and the c-MET immunoprecipitates at indicated time points post-ligand stimulation were run on the same gel for immunoblotting with anti-MET antibody (Figure 3a, left panel). InlB-induced c-MET phosphorylation was assessed at early time points when internalization is negligible (32). Immunoprecipitation was performed according to standard protocols. Briefly, lysates were incubated with the primary antibody overnight at 4 °C and then incubated with 100 μL of protein A agarose in PBS for 3−4 h at 4 °C to capture the immune complex. The tubes were spun, and beads were washed three or four times in lysis buffer and then resuspended in SDS–PAGE buffer. Samples were separated with 7.5% polyacrylamide gels and SDS–PAGE for subsequent immunoblotting. After an overnight incubation in blocking buffer (1% BSA in TBST) at 4 °C, the membranes were incubated with anti-MET antibody (typically at a 1:1000 dilution) for 1 h at room temperature. After three successive washes in TBST, the membranes were incubated with secondary antibody (1: 10000 dilution) for 45 min. The proteins were visualized with an ECL Plus chemiluminescent kit (Amersham) and exposure to film (Kodak X-OMAT AR) (Figure 3a, left panel). The fraction of total c-MET phosphorylated (Figure 3a) in Vero cells was determined from the quantified blots as the ratio of signal from the anti-MET antibody in precipitates from the anti-PY antibody to that from the anti-MET antibody for each time point (1). X-ray films containing the protein bands were imaged in a MultiImage Light Cabinet (Alpha Innotech Corp.), and densitometry was performed using ChemiImager provided by the manufacturer. Experiments were performed within the dynamic range of the antibody signal and the exposure time.

The pervanadate stimulation experiments have been previously described (20), and quantitative data for these experiments are presented (Figure 3b). Briefly, cells were stimulated with the addition of pervanadate-containing media. At specific time points poststimulation, the cells were lysed and subjected to immunoprecipitation with the anti-MET antibody. The precipitated proteins were separated on denaturing polyacrylamide gels and transferred to a PVDF membrane using protocols outlined above. The membranes were probed with the anti-PY antibody and reprobed with the anti-MET antibody (Figure 3b). Quantitative analyses were used to obtain the fraction of c-MET phosphorylated. At PY signal saturation, it was assumed that all active c-MET receptors were fully phosphorylated. In that case, the fraction of phosphorylated receptor would be the ratio of signal at any time point to the signal measured at saturation.

Development of the Kinetic Model

To quantitatively study c-MET activation, a detailed mathematical model was developed that incorporated 28 major biochemical reaction events associated with c-MET dimerization and signaling (Figure 1). As detailed in Experimental Procedures, ∼75% of the parameter values in this model are from published kinetic and thermodynamic parameters for monomeric cytoMET and the dimeric TPR–MET complex (20–22). The few rate constants in our model that were not available from the c-MET system were derived from experimental data in related RTKs (1, 5, 16, 24, 33). Protein–protein interactions were described by mass action equations. Rate equations used for generating our c-MET activation model are listed in Table 2. Starting concentrations of c-MET and Tyr-containing effectors were taken from published results (1). The ATP concentration was set at 1 mM (45), although the model predictions do not change with higher ATP concentrations (data not shown). Activation of c-MET occurs in response to binding the physiological ligand HGF or the Internalin B (InlB) protein of Listeria monocytogenes. InlB has been shown to mimic many of the features of HGF-induced c-MET activation, endocytosis (32, 46, 47), and signaling that are relevant for this study (48). Accordingly, InlB was used as the c-MET agonist for the simulations and cell culture studies.

c-MET Phosphorylation in the Absence of Agonist

We determined the population distribution of different c-MET species in the absence of an extracellular ligand. With no external ligand stimulus, our model predicts that c-MET would exist predominantly as an ATP-bound unphosphorylated monomeric receptor (denoted MET–ATP in Figure 1b). Since the c-MET monomer has kinase activity (20–23), we tested whether significant levels of phosphorylated c-MET monomers can accumulate in the absence of an extracellular ligand. Our model predicts that in the absence of an extracellular ligand the steady-state concentration of phosphorylated c-MET monomers (defined as [pMET] + [pMET–ATP] + [pMET–Tyr] + [pMET–ATP–Tyr]) would be <1% of the total c-MET concentration in cells (data not shown). We could not detect c-MET phosphorylation in unstimulated serum-starved Vero cells using immunoblotting (data not shown). This is likely due to the predicted levels of c-Met being below the level of detection, since small amounts of EGFR basal phosphorylation in unstimulated cells have been reported using a more sensitive FRET-based detection system (51). Significantly, this level of monomeric phosphorylated receptor is approximately 2 orders of magnitude smaller than the steady-state levels of phosphorylated dimeric receptor generated in response to added extracellular ligand (Figure 2a).

The basal level of c-MET phosphorylation in the absence of extracellular ligand stimulation directly correlates with changes to the modeled c-MET monomer autophosphorylation (reaction 17, Figure 1b) and dephosphorylation (reaction 26, Figure 1b) rates. Therefore, we examined the impact of setting the c-MET monomer dephosphorylation rate (reaction 26, Figure 1b) to zero. In this extreme case, our model predicts a relatively slow yet substantial buildup of phosphorylated c-MET (Figure 2c). This prediction has been reported in previous cell culture studies, where the buildup of phosphorylated PDGFR (18), IR (19), and c-MET (20) was observed in the presence of phosphatase inhibitors. Our experimental data on c-MET phosphorylation kinetics in the presence of the PTP inhibitor pervanadate (Figure 3b) correlate qualitatively and quantitatively with our model's prediction (Figure 2c). Moreover, our model predicts that the rate of phosphorylated monomeric c-MET accumulation upon inhibition of receptor dephosphorylation is directly related to the c-MET autophosphorylation k_cat (reaction 17, Figure 1b). Thus, our model clearly shows c-MET phosphorylation and dephosphorylation processes are likely balanced, resulting in a dynamic steady state.

c-MET Phosphorylation in the Presence of an Extracellular Ligand

Using our c-MET activation model (Figure 1), we calculated the rate and magnitude of c-MET autophosphorylation and exogenous effector phosphorylation in response to an extracellular ligand (Figure 2a,b). Our model predicted that the concentration of total phosphorylated c-MET (defined as [pMET] + [pMET–Tyr] + [pMET–ATP] + [pMET–ATP–Tyr] + [pMET₂] + [pMET₂–Tyr] + [pMET₂–ATP] + [pMET₂–ATP–Tyr]) was not linearly related to the concentration of added extracellular ligand (Figure 2a). At low InlB concentrations (<5 nM), our model predicted that the phosphorylation of c-MET would reach a maximum steady-state level within seconds (Figure 2a). In contrast, our model predicted that c-MET phosphorylation kinetics at higher InlB concentrations (>5 nM) would produce a rapid transient spike response followed by a steady-state phosphorylation level that was lower than the maximum observed during the transient spike (Figure 2a). The magnitude of the phosphorylated c-MET transient spike and steady-state levels were dependent on the concentration of added extracellular ligand.

We validated our simulations using Vero cells as a model cell culture system (Figure 3a). Vero cells have been extensively used to characterize c-MET signaling following activation with HGF and InlB (46, 52). Predictions of c-MET phosphorylation were compared to c-MET phosphorylation levels measured in cultured Vero cells in response to different amounts of InlB (1, 5, and 20 nM). We observe a ligand-dependent nonlinear increase in the level of c-MET phosphorylation in serum-starved Vero cells. In addition, a rapid transient spike in the level of phosphorylated c-MET is observed at 20 nM InlB. Our measured c-MET phosphorylation responses (Figure 3a) correlate closely to our model predictions (Figure 2a), implying our c-MET activation model is sufficient to accurately reproduce the salient features of c-MET activation. Unlike earlier modeling studies (1), our activation model does not require the effector-bound state of the receptor to be “inaccessible” to cellular phosphatases to simulate the transient spike in receptor phosphorylation. Instead, the observed transient spike in the level of the phosphorylated receptor following stimulation with high ligand concentrations results from rapid c-MET autophosphorylation prior to cellular phosphatases reaching full activity. In addition, our model interprets the observed steady-state receptor phosphorylation levels as a result of the balance between kinase and phosphatase kinetic reactions once a quasi-equilibrium state is reached. As discussed below, varying parameters associated with dimeric c-MET autophosphorylation (reaction 18, Figure 1b) and dephosphorylation (reaction 25, Figure 1b) reactions significantly modulate the shape and height of the phosphorylated c-MET transient spike in response to ligand stimulation. Our simulations accurately reproduce all tested cell culture observations, validating the applicability of our activation model in improving our understanding of the dynamic processes that regulate c-MET activation.

Computational Analysis of Effector Binding and Phosphorylation

Signaling from activated c-MET requires effectors to bind to c-MET phosphotyrosines, resulting in effector protein phosphorylation and/or activation (26). Binding, phosphorylation, and dephosphorylation reactions for c-MET kinase substrates were included in our model. Although multiple effectors may interact with phosphorylated c-MET simultaneously, as a first approximation only one effector is included in our c-MET activation model. Additional complexities associated with multiple-effector binding will be subsequently included in our model once the relevant experimental data are available. The time course for the buildup of the phosphorylated effector (pTyr) is simulated for a range of added extracellular ligand concentrations (Figure 2b). At each extracellular ligand concentration, our c-MET activation model shows that pTyr levels will increase monotonically in response to ligand-induced dimerization; no transient spike is observed in pTyr levels. This pattern of effector phosphorylation was observed for Shc in the related EGFR RTK system (1). In addition, our model predicted pTyr accumulated only after significant c-MET phosphorylation, resulting in a lag time between extracellular ligand stimulation and effector phosphorylation (Figure 2). The rate of effector phosphorylation is biphasic (most clearly observed at high extracellular ligand concentrations), with a higher initial rate of effector phosphorylation followed by a slower rate of phosphorylation as the dynamic system reaches steady state (Figure 2b). Finally, the amount of Tyr bound to the c-MET (defined as [MET–Tyr] + [MET–ATP–Tyr] + [pMET–Tyr] + [pMET–ATP–Tyr] + [MET₂–Tyr] + [MET₂–ATP–Tyr] + [pMET₂–Tyr] + [pMET₂–ATP–Tyr]) correlates closely with the accumulation of phosphorylated c-MET (data not shown). These observations imply that c-MET phosphorylation is necessary for effective effector recruitment and/or phosphorylation, consistent with previous reports that mutants deficient in kinase activity did not result in effector activation in response to InlB or HGF.

Sensitivity of c-MET Signaling to Biochemical Parameters

c-MET activation, described by the magnitude and specificity of c-MET and effector phosphorylation in response to extracellular ligand-mediated dimerization, is controlled by a variety of thermodynamic and kinetic parameters. The sensitivity of dimeric c-MET and effector phosphorylation to these different biochemical parameters was studied in an effort to understand their relative importance in c-MET activation. Sensitivity analysis of c-MET phosphorylation to kinase ATP and effector binding on and off rates reveals that these parameters have little impact on the behavior of the c-MET activation model (data not shown). In contrast, changing the kinetic parameters associated with dimeric c-MET phosphorylation and dephosphorylation reactions has a significant impact on c-MET activation as measured by the level of accumulation of phosphorylated dimeric c-MET and pTyr.

We systematically varied the kinetic parameters associated with dimeric c-MET autophosphorylation (k₁₈), dephosphorylation (V₂₅), and effector phosphorylation (k₂₁ and k₂₂) reactions to deepen our understanding of the relative importance of these processes to c-MET activation. Panels a and b of Figure 4 show the time-dependent accumulation of phosphorylated c-MET and pTyr as a function of dimer autophosphorylation k_cat (k₁₈) at a saturating (i.e., 20 nM) extracellular ligand concentration. Clearly, the predicted rates of c-MET phosphorylation and steady-state levels of phosphorylated c-MET are dependent on c-MET dimer autophosphorylation values. Furthermore, by affecting the level of phosphorylated c-MET, the autophosphorylation k_cat significantly changes the rate of pTyr production (Figure 4b). Thus, the c-MET activation response is sensitive to the dimer autophosphorylation rate. In contrast, changes to the effector phosphorylation k_cat (k₂₁ and k₂₂) have little effect on the rate of buildup of phosphorylated c-MET (Figure 4c). However, the rate of pTyr buildup is directly dependent upon the magnitude of the effector phosphorylation k_cat (Figure 4d). Phosphorylated c-MET steady-state levels and the rate of pTyr formation are significantly changed when the c-MET dephosphorylation V_max (V₂₅) is varied, implying a critical role for cellular phosphatases in modulating c-MET signaling potential (Figure 4e,f). Changes to the steady-state level of phosphorylated c-MET modify the sharpness of the phosphorylated c-MET transient spike observed at high extracellular ligand concentrations (Figure 4e). Thus, the phosphatase-catalyzed dephosphorylation rates dictate the amplitude as well as the duration of c-MET activation.

The simulations described above evaluated the relative importance of kinetic and thermodynamic parameters associated with dimeric c-MET, the predominant activated species. A similar parameter sensitivity analysis was applied to monomeric (i.e., unstimulated) c-MET. In this case, no significant differences are observed in the buildup of phosphorylated c-MET or pTyr when single thermodynamic or kinetic parameters are varied (k₁₇, k₁₉, k₂₀, and k₂₆) (data not shown). In contrast, as discussed in greater detail below, significant changes in the buildup of phosphorylated c-MET and pTyr are observed when several kinetic parameters for monomeric c-MET are simultaneously changed. On the basis of these results, it can be inferred that dimerization is a reversible mechanism for simultaneously adjusting several c-MET kinetic parameters to form the active signaling state. Nonstimulated monomeric c-MET exists in a tightly regulated inactive signaling state that is relatively insensitive to individual changes to its kinetic and thermodynamic parameters.

Synergistic Changes to c-MET Biochemical Properties

The k_on and k_off rates for the thermodynamic processes, the kinetic catalytic constants for autophosphorylation and effector phosphorylation, and rates of dephosphorylation of c-MET by phosphatases are the main biochemical parameters that differ between monomeric and dimeric c-MET. To better understand how oligomerization-induced changes to the thermodynamic, kinetic, and dephosphorylation properties of c-MET impact its activation, these properties were simultaneously varied for monomeric c-MET and the buildup of phosphorylated c-MET and pTyr was monitored (Table 3). These modeling studies were normalized to baseline simulations using monomeric and dimeric c-MET biochemical reaction parameters that were based on experimentally determined values (as described above). c-MET activation in the absence of extracellular ligand stimulation (and, hence, c-MET dimerization) was monitored by the level of accumulation of phosphorylated monomeric c-MET and pTyr. For comparison, normalized values of phosphorylated c-MET and pTyr in the presence of InlB are tabulated. As summarized in Table 3, c-MET and effector phosphorylation levels do not significantly change relative to the baseline simulation unless a biochemical parameter changed >100-fold relative to its default value. However, small changes in two or more kinetic properties have significant synergistic effects on c-MET's signaling competence. For example, decreasing the dephosphorylation rate (V₂₆) 10-fold and increasing autophosphorylation k_cat (k₁₇) and effector phosphorylation k_cat (k₁₉ and k₂₀) 10-fold relative to the monomeric c-MET default parameters result in a buildup of phosphorylated c-MET and pTyr comparable to that observed at 20 nM InlB (Table 3). An order of magnitude change in the effector binding rate constants (k₂, k₃, k₆, and k₇) increases the basal levels of phosphorylated c-MET to ∼0.5% of the total level of c-MET; however, these levels are several-fold lower than that generated in the presence of ligand. Thus, the generation of an activated c-MET state appears to be the result of synergistic effects between dynamic processes regulating receptor kinase activity (autophosphorylation and effector phosphorylation) and dephosphorylation.

If oligomerization synergistically modifies multiple biochemical parameters to regulate the phosphorylation state and hence the signaling activity of c-MET, will our model account for constitutive receptor activation? The TPR–MET complex is an excellent well-characterized example of a mutant c-MET protein that causes constitutive activation of c-MET in a manner independent of ligand stimulation and via receptor dimerization (23). The TPR–MET complex lacks mutations in its kinase domain but is able to dimerize and is tumorigenic due to the enhanced kinetic and thermodynamic properties associated with the dimeric species. Alternatively, it is clear that small synergistic changes in c-MET kinetic properties can result in a several-fold increased level of accumulation of phosphorylated c-MET and pTyr (Table 3). Interestingly, c-MET gain-of-function mutations (for example, D1228V, V1220I, Y1230H, Y1230C, D1228N, and V1220I) have been shown to lead to deregulated and prolonged kinase activity, which is critical to their transforming potential (54). The possible structural basis of ligand-independent activation via these mutations is discussed by Wang et al. (55), using the c-MET X-ray crystal structure. These structural changes may alter the biochemical properties of c-MET. The exact kinetic parameters of these mutants have not been measured directly, although it has been observed that some of these mutant forms of c-MET have high basal kinase activity in the absence of ligand (in their monomeric form) (54). The increased kinase activity of c-MET could directly impact the autophosphorylation (reaction 17) and effector phosphorylation (reactions 19 and 20) rates. If a c-MET point mutation increases the rates of autophosphorylation (reaction 17) and effector phosphorylation (reactions 19 and 20) 10-fold relative to that of the wild-type monomeric receptor, our model predicts that the unstimulated c-MET mutant would lead to an ∼14-fold increase in the level of phosphorylated c-MET and an ∼130-fold increase in the level of pTyr relative to that of the unstimulated wild-type receptor. The pTyr levels predicted from the point-mutated receptor are quantitatively similar to the amount of pTyr generated by the wild-type receptor following stimulation with 3 nM InlB (Table 3). c-MET point mutations that decrease the susceptibility of monomeric c-MET to dephosphorylation, and simultaneously increase the kinase autophosphorylation and effector phosphorylation catalytic efficiencies relative to that of wild-type c-MET, will result in a constitutively activate monomeric c-MET and phosphorylated c-MET and pTyr levels comparable to those observed for wild-type c-MET in response to saturating levels of ligand.