G Protein-Coupled Receptor Signaling Networks from a Systems Perspective.

The signal-transduction network of a mammalian cell integrates internal and external cues to initiate adaptive responses. Among the cell-surface receptors are the G protein-coupled receptors (GPCRs), which have remarkable signal-integrating capabilities. Binding of extracellular signals stabilizes intracellular-domain conformations that selectively activate intracellular proteins. Hereby, multiple signaling routes are activated simultaneously to degrees that are signal-combination dependent. Systems-biology studies indicate that signaling networks have emergent processing capabilities that go far beyond those of single proteins. Such networks are spatiotemporally organized and capable of gradual, oscillatory, all-or-none, and subpopulation-generating responses. Protein-protein interactions, generating feedback and feedforward circuitry, are generally required for these spatiotemporal phenomena. Understanding of information processing by signaling networks therefore requires network theories in addition to biochemical and biophysical concepts. Here we review some of the key signaling systems behaviors that have been discovered recurrently across signaling networks. We emphasize the role of GPCRs, so far underappreciated receptors in systems-biology research.

Systems medicine: helping us understand the complexity of disease.

Advances in genomics and other -omic fields in the last decade have resulted in unprecedented volumes of complex data now being available. These data can enable physicians to provide their patients with care that is more personalized, predictive, preventive and participatory. The expertise required to manage and understand this data is to be found in fields outside of medical science, thus multidisciplinary collaboration coupled to a systems approach is key to unlocking its potential, with concomitant new ways of working. Systems medicine can build on the successes in the field of systems biology, recognizing the human body as the multidimensional network of networks that it is. While systems medicine can provide a conceptual and theoretical framework, its practical goal is to provide physicians the tools necessary for harnessing the rapid advances in basic biomedical science into their routine clinical arsenal.

Domain-oriented reduction of rule-based network models.

The coupling of membrane-bound receptors to transcriptional regulators and other effector functions is mediated by multi-domain proteins that form complex assemblies. The modularity of protein interactions lends itself to a rule-based description, in which species and reactions are generated by rules that encode the necessary context for an interaction to occur, but also can produce a combinatorial explosion in the number of chemical species that make up the signalling network. The authors have shown previously that exact network reduction can be achieved using hierarchical control relationships between sites/domains on proteins to dissect multi-domain proteins into sets of non-interacting sites, allowing the replacement of each 'full' (progenitor) protein with a set of derived auxiliary (offspring) proteins. The description of a network in terms of auxiliary proteins that have fewer sites than progenitor proteins often greatly reduces network size. The authors describe here a method for automating domain-oriented model reduction and its implementation as a module in the BioNetGen modelling package. It takes as input a standard BioNetGen model and automatically performs the following steps: 1) detecting the hierarchical control relationships between sites; 2) building up the auxiliary proteins; 3) generating a raw reduced model and 4) cleaning up the raw model to provide the correct mass balance for each chemical species in the reduced network. The authors tested the performance of this module on models representing portions of growth factor receptor and immunoreceptor-mediated signalling networks and confirmed its ability to reduce the model size and simulation cost by at least one or two orders of magnitude. Limitations of the current algorithm include the inability to reduce models based on implicit site dependencies or heterodimerisation and loss of accuracy when dynamics are computed stochastically.

Quantifying gene network connectivity in silico: scalability and accuracy of a modular approach.

Large, complex data sets that are generated from microarray experiments, create a need for systematic analysis techniques to unravel the underlying connectivity of gene regulatory networks. A modular approach, previously proposed by Kholodenko and co-workers, helps to scale down the network complexity into more computationally manageable entities called modules. A functional module includes a gene's mRNA, promoter and resulting products, thus encompassing a large set of interacting states. The essential elements of this approach are described in detail for a three-gene model network and later extended to a ten-gene model network, demonstrating scalability. The network architecture is identified by analysing in silico steady-state changes in the activities of only the module outputs, communicating intermediates, that result from specific perturbations applied to the network modules one at a time. These steady-state changes form the system response matrix, which is used to compute the network connectivity or network interaction map. By employing a known biochemical network, the accuracy of the modular approach and its sensitivity to key assumptions are evaluated.

Signal processing at the Ras circuit: what shapes Ras activation patterns?

A systems biology approach is applied to gain a quantitative understanding of the integration of signalling by the small GTPase Ras. The Ras protein acts as a critical switch in response to signals that determine the cell's fate. In unstimulated cells, Ras switching between an inactive GDP-binding and active GTP-binding state is controlled by the intrinsic catalytic activities of Ras. The calculated high sensitivity of the basal Ras-GTP fraction to changes in the rate constant of GTP-hydrolysis by Ras can account for the carcinogenic potential of Ras mutants with decreased GTPase activities. Extracelluar stimuli initiate Ras interactions with GDP/GTP exchange factors such as SOS, and GTP-hydrolysis activating proteins such as RasGAP. Our data on freshly isolated hepatocytes stimulated with epidermal growth factor (EGF) show transient SOS activation and sustained Ras-GTP patterns. We demonstrate that these dose-response data can only be explained by transient RasGAP activitation, and not by merely switching off the SOS signal, e.g. by inhibitory phosphorylation of SOS. A transient RasGAP activity can be brought about by a number of mechanisms. A comprehensive kinetic model of the EGF receptor (EGFR) network was developed to explore feasible molecular scenarios, including the receptor-mediated recruitment of SOS and RasGAP to the plasma membrane, phosphorylation of RasGAP and p190 RhoGAP by soluble tyrosine kinases, and RasGAP interactions with phosphoinositides and p190 RhoGAP. We show that a transient RasGAP association with EGFR followed by the capture of RasGAP through the formation of complexes with p190 RhoGAP can account for data on hepatocytes. In summary, our results demonstrate that a combination of experimental monitoring and integrated dynamic analysis is capable of dissecting regulatory mechanisms that govern cellular signal transduction.

Occurrence of paradoxical or sustained control by an enzyme when overexpressed: necessary conditions and experimental evidence with regard to hepatic glucokinase.

It is widely assumed that the control coefficient of an enzyme on pathway flux decreases as the concentration of enzyme increases. However, it has been shown [Kholodenko and Brown (1996) Biochem. J. 314, 753-760] that enzymes with sigmoidal kinetics can maintain or even gain control with an increase in enzyme activity or concentration. This has been described as 'paradoxical control'. Here we formulate the general requirements for allosteric enzyme kinetics to display this behaviour. We show that a necessary condition is that the Hill coefficient of the enzyme should increase with an increase in substrate concentration or decrease with an increase in product concentration. We also describe the necessary and sufficient requirements for the occurrence of paradoxical control in terms of the flux control coefficients and the derivatives of the elasticities. The derived expression shows that the higher the control coefficient of an allosteric enzyme, the more likely it is that the pathway will display this behaviour. Control of pathway flux is generally shared between a large number of enzymes and therefore the likelihood of observing sustained or increased control is low, even if the kinetic parameters are in the most favourable range to generate the phenomenon. We show that hepatic glucokinase, which has a very high flux control coefficient and displays sigmoidal behaviour within the hepatocyte in situ as a result of interaction with a regulatory protein, displays sustained or increased control over an extended range of enzyme concentrations when the regulatory protein is overexpressed.

[Kinetic modeling of energy metabolism and generation of active forms of oxygen in hepatocyte mitochondria]. / Kineticheskoe modelirovaniie énergeticheskogo metabolizma i generatsii aktivnyhkh form kisloroda v mitokhondriiakh gepatotsita.

Direct nonenzymatic oxidation of semiquinone by oxygen is one of the main sources of superoxide radicals (O2.-) in mitochondria. By using all the known data on hepatocyte mitochondria, we have revealed the correlation between the rate of superoxide generation by the bc1 complex and the transmembrane potential (delta psi). If the main electrogenic stage of the Q cycle is suggested to be the electron transfer between the cytochrome b hemes, then the rate of superoxide generation sharply increases when delta psi grows from 150 mV to 180 mV. However, this interrelation is ambiguous. Indeed, the increase of the generation rate with the growth of the potential can occur faster when succinate dehydrogenase is inhibited by malonate than when external ADP is exhausted. When the potential is changed by adding phosphate or potassium (K+), the rate of O2.- production remains constant, although the comparison of the rate values at the same delta psi reveals the effect of phosphate or potassium. It turned out that the rate of O2.- generation is a function of delta mu H rather than any of its components. Phosphate and K+ have practically no influence on delta mu H, since the change in delta psi is compensated by delta pH. The rate of superoxide generation by the bc1 complex is a multiple function of the electron-transfer activity of enzymes, the processes determining the membrane potential (e.g., loading), and of the oxygen concentration. The kinetic model proposed in this work may serve a tool to understand how the superoxide production is regulated.

Diffusion control of protein phosphorylation in signal transduction pathways.

Multiple signalling proteins are phosphorylated and dephosphorylated at separate cellular locations, which potentially causes spatial gradients of phospho-proteins within the cell. We have derived relationships that enable us to estimate the extent to which a protein kinase, a phosphatase and the diffusion of signalling proteins control the protein phosphorylation flux and the phospho-protein gradient. Two different cellular geometries were analysed: (1) the kinase is located on one planar membrane and the phosphatase on a second parallel planar membrane, and (2) the kinase is located on the plasma membrane of a spherical cell and the phosphatase is distributed homogeneously in the cytoplasm. We demonstrate that the control contribution of protein diffusion is potentially significant, given the measured rates for protein kinases, phosphatases and diffusion. If the distance between the membranes is 1 microm or greater, the control by diffusion can reach 33% or more, with the rest of the control (67%) shared by the kinase and the phosphatase. At distances of less than 0.1 microm, diffusion does not limit protein phosphorylation. For a spherical cell of radius 10 microm, a protein diffusion coefficient of 10(-8) cm(2). s(-1) and rate constants for the kinase and the phosphatase of approx. 1 s(-1), control over the phosphorylation flux resides mainly with the phosphatase and protein diffusion, with approximately equal contributions of each of these. The ratio of phospho-protein concentrations at the cell membrane and the cell centre (the dynamic compartmentation of the phospho-protein) is shown to be controlled by the rates of the protein phosphatase and of diffusion. The kinase can contribute significantly to the control of the absolute value of the phospho-protein gradient.

Engineering a living cell to desired metabolite concentrations and fluxes: pathways with multifunctional enzymes.

With molecular genetics enabling modulation of the concentrations of cellular enzymes, metabolic engineering becomes limited by the question of which modulations of the enzyme concentrations are required to bring about a desired pattern of cellular metabolism. In an earlier paper (Kholodenko et al. (1998). Biotechnol. Bioeng. 59, 239-247) we derived a method to determine the required modulations. This method, however, cannot be immediately applied to cellular pathways with enzymes catalyzing more than one step in metabolism (multifunctional enzymes). In the present paper we show to which extent the presence of multifunctional enzymes limits biotechological ambitions, which one might otherwise pursue in vain. In particular, it is impossible to change the concentration of a single intermediate and leave the rest of metabolism unperturbed if that intermediate interacts directly with a multifunctional enzyme. The analytical machinery of Metabolic Control Analysis is used to relate the desired and ensuing changes in the metabolic pattern. An explicit solution to this problem of engineering metabolism is then given in the form of a single matrix equation.

Kinetics and control of oxidative phosphorylation in rat liver mitochondria after chronic ethanol feeding.

Changes in the kinetics and regulation of oxidative phosphorylation were characterized in isolated rat liver mitochondria after 2 months of ethanol consumption. Mitochondrial energy metabolism was conceptually divided into three groups of reactions, either producing protonmotive force (Deltap) (the respiratory subsystem) or consuming it (the phosphorylation subsystem and the proton leak). Manifestation of ethanol-induced mitochondrial malfunctioning of the respiratory subsystem was observed with various substrates; the respiration rate in State 3 was inhibited by 27+/-4% with succinate plus amytal, by 20+/-4% with glutamate plus malate, and by 17+/-2% with N,N,N',N'-tetramethyl-p-phenylenediamine/ascorbate. The inhibition of the respiratory activity correlated with the lower activities of cytochrome c oxidase, the bc(1) complex, and the ATP synthase in mitochondria of ethanol-fed rats. The block of reactions consuming the Deltap to produce ATP (the phosphorylating subsystem) was suppressed after 2 months of ethanol feeding, whereas the mitochondrial proton leak was not affected. The contributions of Deltap supply (the respiratory subsystem) and Deltap demand (the phosphorylation and the proton leak) to the control of the respiratory flux were quantified as the control coefficients of these subsystems. In State 3, the distribution of control exerted by different reaction blocks over respiratory flux was not significantly affected by ethanol diet, despite the marked changes in the kinetics of individual functional units of mitochondrial oxidative phosphorylation. This suggests the operation of compensatory mechanisms, when control redistributes among the different components within the same subsystem.

Why cytoplasmic signalling proteins should be recruited to cell membranes.

It has been suggested that localization of signal-transduction proteins close to the cell membrane causes an increase in their rate of encounter after activation. We maintain that such an increase in the first-encounter rate is too small to be responsible for truly enhanced signal transduction. Instead, the function of membrane localization is to increase the number (or average lifetime) of complexes between cognate signal transduction proteins and hence increase the extent of activation of downstream processes. This is achieved by concentrating the proteins in the small volume of the area just below the plasma membrane. The signal-transduction chain is viewed simply as operating at low default intensity because one of its components is present at a low concentration. The steady signalling level of the chain is enhanced 1000-fold by increasing the concentration of that component. This occurs upon 'piggyback' binding to a membrane protein, such as the activated receptor, initiating the signal-transduction chain. For the effect to occur, the protein translocated to the membrane cannot be free but has to remain organized by being piggyback bound to a receptor, membrane lipid(s) or scaffold. We discuss an important structural constraint imposed by this mechanism on signal transduction proteins that might also account for the presence of adaptor proteins.

Cellular information transfer regarded from a stoichiometry and control analysis perspective.

Metabolic control analysis (MCA) allows one to formalize important aspects of information processing in living cells. For example, information processing via multi-level enzyme cascades can be quantified in terms of the response coefficient of a cellular target to a signal. In many situations, control and response coefficients cannot be determined exactly for all enzymes involved, owing to difficulties in 'observing' all enzymes experimentally. Here, we review a number of qualitative approaches that were developed to cope with such situations. The usefulness of the concept of null-space of the stoichiometry matrix for analysing the structure of intracellular signaling networks is discussed. It is shown that signal transduction operates very efficiently when the network structure is such that the null-space matrix can be block-diagonalized (which may or may not imply that the network consists of several disconnected parts) and some enzymes have low elasticities to their substrates.

Negative feedback and ultrasensitivity can bring about oscillations in the mitogen-activated protein kinase cascades.

Functional organization of signal transduction into protein phosphorylation cascades, such as the mitogen-activated protein kinase (MAPK) cascades, greatly enhances the sensitivity of cellular targets to external stimuli. The sensitivity increases multiplicatively with the number of cascade levels, so that a tiny change in a stimulus results in a large change in the response, the phenomenon referred to as ultrasensitivity. In a variety of cell types, the MAPK cascades are imbedded in long feedback loops, positive or negative, depending on whether the terminal kinase stimulates or inhibits the activation of the initial level. Here we demonstrate that a negative feedback loop combined with intrinsic ultrasensitivity of the MAPK cascade can bring about sustained oscillations in MAPK phosphorylation. Based on recent kinetic data on the MAPK cascades, we predict that the period of oscillations can range from minutes to hours. The phosphorylation level can vary between the base level and almost 100% of the total protein. The oscillations of the phosphorylation cascades and slow protein diffusion in the cytoplasm can lead to intracellular waves of phospho-proteins.

Quantification of short term signaling by the epidermal growth factor receptor.

During the past decade, our knowledge of molecular mechanisms involved in growth factor signaling has proliferated almost explosively. However, the kinetics and control of information transfer through signaling networks remain poorly understood. This paper combines experimental kinetic analysis and computational modeling of the short term pattern of cellular responses to epidermal growth factor (EGF) in isolated hepatocytes. The experimental data show transient tyrosine phosphorylation of the EGF receptor (EGFR) and transient or sustained response patterns in multiple signaling proteins targeted by EGFR. Transient responses exhibit pronounced maxima, reached within 15-30 s of EGF stimulation and followed by a decline to relatively low (quasi-steady-state) levels. In contrast to earlier suggestions, we demonstrate that the experimentally observed transients can be accounted for without requiring receptor-mediated activation of specific tyrosine phosphatases, following EGF stimulation. The kinetic model predicts how the cellular response is controlled by the relative levels and activity states of signaling proteins and under what conditions activation patterns are transient or sustained. EGFR signaling patterns appear to be robust with respect to variations in many elemental rate constants within the range of experimentally measured values. On the other hand, we specify which changes in the kinetic scheme, rate constants, and total amounts of molecular factors involved are incompatible with the experimentally observed kinetics of signal transfer. Quantitation of signaling network responses to growth factors allows us to assess how cells process information controlling their growth and differentiation.

Spatial gradients of cellular phospho-proteins.

If a protein is rapidly phosphorylated and dephosphorylated at separate cellular locations and protein diffusion is slow, then a spatial gradient of the phosphorylated form of the protein may develop within the cell. We have estimated the potential size of such gradients using measured values of protein diffusion coefficients and protein kinase and phosphatase activities. We analysed two different cellular geometries: (1) where the kinases is located on the plasma membrane of a spherical cell and the phospatase is distributed homogenously in the cytoplasm and (2) where the kinase is located on one planar membrane and the phosphatase on a second parallel planar membrane. The estimated gradients of phospho-proteins were potentially very large, which has important implications for cellular signalling.

A model of O2.-generation in the complex III of the electron transport chain.

Oxidation of semiquinone by O2 in the Q cycle is known to be one of the sources of superoxide anion (O2.-) in aerobic cells. In this paper, such a phenomenon was analyzed using the chemical kinetics model of electron transfer from succinate to cytochrome c, including coenzyme Q, the complex III non-heme iron protein FeSIII and cytochromes bl, bh and cl. Electron transfers from QH2 to FeSIII and cytochrome bl were assumed to occur according to direct transfer mechanism (dynamic channelling) involving the formation of FeS(red)III-Q.- and Q.--cytochrome bl complexes. For oxidation/reduction reactions involving cytochromes bh and bl, the dependence of the equilibrium and elementary rate constants on the membrane potential (deltapsi) was taken into consideration. The rate of O2.- generation was found to increase dramatically with increase in deltapsi above the values found in State 3. On the other hand, the rate of cytochrome c reduction decreased sharply at the same values of the membrane potential. This explains experimental data that the O2.- generation at State 4 appears to be very much faster than at State 3. A mild uncoupling in State 4 can markedly decrease the superoxide generation due to a decrease in deltapsi below the above mentioned critical level. DeltapH appears to be equally effective as deltapsi in stimulation of superoxide production which depends, in fact, upon the deltamuH+ level.

Subtleties in control by metabolic channelling and enzyme organization.

Because of its importance to cell function, the free-energy metabolism of the living cell is subtly and homeostatically controlled. Metabolic control analysis enables a quantitative determination of what controls the relevant fluxes. However, the original metabolic control analysis was developed for idealized metabolic systems, which were assumed to lack enzyme-enzyme association and direct metabolite transfer between enzymes (channelling). We here review the recently developed molecular control analysis, which makes it possible to study non-ideal (channelled, organized) systems quantitatively in terms of what controls the fluxes, concentrations, and transit times. We show that in real, non-ideal pathways, the central control laws, such as the summation theorem for flux control, are richer than in ideal systems: the sum of the control of the enzymes participating in a non-ideal pathway may well exceed one (the number expected in the ideal pathways), but may also drop to values below one. Precise expressions indicate how total control is determined by non-ideal phenomena such as ternary complex formation (two enzymes, one metabolite), and enzyme sequestration. The bacterial phosphotransferase system (PTS), which catalyses the uptake and concomitant phosphorylation of glucose (and also regulates catabolite repression) is analyzed as an experimental example of a non-ideal pathway. Here, the phosphoryl group is channelled between enzymes, which could increase the sum of the enzyme control coefficients to two, whereas the formation of ternary complexes could decrease the sum of the enzyme control coefficients to below one. Experimental studies have recently confirmed this identification, as well as theoretically predicted values for the total control. Macromolecular crowding was shown to be a major candidate for the factor that modulates the non-ideal behaviour of the PTS pathway and the sum of the enzyme control coefficients.

Implications of macromolecular crowding for signal transduction and metabolite channeling.

The effect of different total enzyme concentrations on the flux through the bacterial phosphoenolpyruvate:carbohydrate phosphotransferase system (PTS) in vitro was determined by measuring PTS-mediated carbohydrate phosphorylation at different dilutions of cell-free extract of Escherichia coli. The dependence of the flux on the protein concentration was more than linear but less than quadratic. The combined flux-response coefficient of the four enzymes constituting the glucose PTS decreased slightly from values of approximately 1.8 with increasing protein concentrations in the assay. Addition of the macromolecular crowding agents polyethylene glycol (PEG) 6000 and PEG 35000 led to a sharper decrease in the combined flux-response coefficient, in one case to values of approximately 1. PEG 6000 stimulated the PTS flux at lower protein concentrations and inhibited the flux at higher protein concentrations, with the transition depending on the PEG 6000 concentration. This suggests that macromolecular crowding decreases the dissociation rate constants of enzyme complexes. High concentrations of the microsolute glycerol did not affect the combined flux-response coefficient. The data could be explained with a kinetic model of macromolecular crowding in a two-enzyme group-transfer pathway. Our results suggest that, because of the crowded environment in the cell, the different PTS enzymes form complexes that live long on the time-scale of their turnover. The implications for the metabolic behavior and control properties of the PTS, and for the effect of macromolecular crowding on nonequilibrium processes, are discussed.