Loading...

Summer Research Fellowship Programme of India's Science Academies

FindSim: Development and optimization of computational multi-scale model for neural network of biochemical signaling pathways

Vinodkumar Ugale

Department of Pharmaceutical Chemistry, R. C. Patel Institute of Pharmaceutical Education and Research, Shirpur, Dhule, Maharashtra 425405

Guided by:

Professor Upinder Singh Bhalla

Department of Neurobiology, National Centre for Biological Sciences, Bengaluru, Karnataka, 560065

Abstract

Many distinct signaling pathways allow the cell to receive, process, and respond to information. In this project, we aim at development and optimization of fully-functional multi-scale simulation model for neural pathways of inflammation. Computational simulations have been employed for comprehensive step by step understanding of biochemical signaling involved. We have developed a network model in stages. Herein, we modeled individual pathways first, and then we examined experimentally defined combinations of two or three such individual pathways and tested these combined models against published data. Pathways are linked by two kinds of interactions: (i) Second messengers such as arachidonic acid (AA) and diacylglycerol (DAG) produced by one pathway will be used as inputs to other pathways. (ii) Enzymes whose activation was regulated by one pathway will be coupled to substrates belonging to other pathways. The stage of analysis used two interconnected pathways. AA is preferentially incorporated into sn-2 position of membrane phospholipids. Phospholipase A2 (PLA2) is an enzyme that specifically deacylates sn-2 position and plays a vital role in hormonal regulation of arachidonic acid release and eicosanoid production. Mitogen-associated protein kinases (MAPK) can phosphorylate and activate cytosolic phospholipase A-2 (cPLA2), and AA produced by cPLA2 acts synergistically with DAG to activate protein kinase C (PKC). Here, a positive feedback system involving PKC and MAPK was analyzed. In order to ease the challenging task of parameter optimization and validation of model, we used an interface called FindSim: Framework for Integrating Neuronal Data and Signaling Models using MOOSE (Multi-Scale Object Oriented Simulation Environment) platform. This interface integrates a curated database of experimental conditions and readouts with detailed models to drive simulations that map to experiments. Each simulated experiment receives a score based on how closely the model outcome matches the experimental data. The score is then utilized in a series of parameter sweeps and automated parameter optimizations to improve the fidelity of model. These studies indicated that simple biochemical reactions with appropriate coupling may be used to store information. The simulation results will enable us to identify pathways contributing to disease and provide solution to modify those pathways as a potential cure. Moreover, the model can be helpful for neurologists as well-documented database of the neural reactions. A refined model will be used to make pharmacological predictions, drug interventions, side effects, combinatorial treatments, and also tackle genetic compensatory effects in control and diseased state.

Keywords: Archidonic acid, cPLA2, MAPK, PKC

Abbreviations

Abbreviations
SRFSummer Research Fellowship
IAS Indian Academy of Sciences 
FindSim   Framework for Integrating Neuronal Data and Signaling Models  
MOOSE   Multi-Scale Object Oriented Simulation Environment  
AA   Arachidonic acid
APC  Arachodonylphosphatidylcholine 
DAG   Diacylglycerol 
LTDlong term depression
PLA2   Phospholipase A2  
MAPK  Mitogen-associated protein kinases 
cPLA2   cytosolic phospholipase A-2 
  PKC Protein kinase C  
PIP2   Phosphatidylinositol-4,5-bisphosphate 
  PA Phosphatidic acid 
GPC   Glycerophosphocholine 
PS  Phosphatidylserine 
  PI Phosphatidylinositol 

INTRODUCTION

Background/Rationale

Computer modeling plays a vital role in bridging the gap between precise theories of simplified systems and complexity of experimental biology through in-silico experiments [1]. Modeling harnesses computer power to scale theoretical calculations up from equations describing individual reactions, to more biological systems comprising thousands of interacting molecules. The start and end-point of this process is experiment: from data to predictions to further data. The ideal of modeling process is to gain insights into system function along the way [2]. Intracellular signaling pathways communicate extracellular information to modulate cellular functions in response to external stimuli. Signaling pathways allow the cell to receive, process, and respond to information through neural network [3]. The processing occurs both via summation of inputs and through temporal characteristics of pathways [4]. FindSim (Framework for Integrating Neuronal Data and Signaling Models) is a pipeline for systematic, data-driven construction of large biologically detailed models of neuronal signaling. The key advantgaes of using this interface are:

(1) This simulation interface combines a database of structured experimental data with each model, to systematically generate scores of how well the model fits the entire dataset.

(2) The FindSim maps models of neural and cellular signaling to experimental protocols and readouts. It runs experiment on model, and provides a score that reports how closely two match [5].

FindSim functions by combining with python-driven MOOSE (Multi-Scale Object Oriented Simulation Environment) simulation engine for multiscale models (Fig. 1) [6-7]. We examined how to systematically use experimental data to parameterize multiscale neuronal signaling models reproducibly, scalably, openly, and in a generally applicable manner. It is clearly desirable to have a standard for facile mapping between experiments and models, especially in rapidly expanding domain of neural physiology and signaling. Herein, we have also explored the role of FindSim interface in experiment-driven simulation specification in a production environment.

Types of Neuronal models.png
    Types of neuronal models for signaling network [7].

    Statement of the Problems

    Initially, signaling pathways were studied in a linear fashion, and it was shown that many important biological effects are obtained through linear information transfer [8]. Signaling pathways linked with one another and the final biological response is outcome of interaction between pathways [9]. These interactions result in networks that are quite complex and may have properties that are nonintuitive. A systematic analysis of interactions between signaling pathways could be useful in understanding the properties of these networks [10]. A computational study for signaling pathways proposed that the duration of long term depression (LTD) is prolonged by a positive feedback loop [11]. This positive feedback loop is thought to include several enzymes, such as protein kinase C (PKC), mitogen-activated protein kinase (MAPK), and phospholipase A2 (PLA2), and these enzymes mutually activate each other. Here, signaling network connected by two interactions: (i) Second messengers such as arachidonic acid (AA) and diacylglycerol (DAG) produced by one pathway will be used as inputs to other pathways [12]. (ii) Enzymes whose activation was regulated by one pathway will be coupled to substrates belonging to other pathways. Hence the stage of analysis use two interconnected pathways (Fig. 2). MAPK can phosphorylate and activate cPLA2, and AA produced by cPLA2 acts synergistically with DAG to activate PKC. Here, a positive feedback system involving PKC and MAPK was analyzed [12-14]. Inhibition of PLA2, which should disrupt the postulated positive feedback loop, produces the predicted changes in the dependence of LTD, suggesting that this loop acts downstream of Ca2+ to initiate LTD [15]. Networking results in several emergent properties that the individual pathways do not have. We developed network model in stages and selected subset PLA2 of implicated pathways for this study based on reported roles, and on the degree to which the pathways have been characterized. Herein, We examined individual pathways first, and then we examined experimentally defined combinations of two or three such individual pathways and screened these combined pathways against published data. We have analyzed these interactions by modeling multiple inputs and the process of self-modifying feedback.

    Pathway.png
      Diagrammatic representation of signaling pathway involving arachidonic acid [2].

      Objectives of the Research

      a. To identify pathways contributing to disease and provide solution to modify those pathways as a potential cure.

      b. To develop and optimize pathways that can be helpful for neuroscientists as well-documented database of neural reactions.

      c. To study pathways for making pharmacological predictions, drug interventions, side effects, combinatorial treatments, and also tackle genetic compensatory effects in control and diseased state.

      LITERATURE REVIEW

      Information

      Signalling pathways are abstractions that help neuroscientists to structure the coordination of neuronal activity. Interaction between pathways appears due to distinct reasons, for example to integrate signals, to produce a variety of responses to a signal, to reuse proteins between pathways and also accounts for complex signalling behaviours. Modeling has tended to be the preserve of a computer scientist, but its greatest value is in hands of experimentalist. There are currently several systems which strive toward this goal in the field of signaling pathways and genetic networks. These include Vcell, MCell, Dbsolve, Jarnac, GENESIS/Kinetikits, and others [10].

      There are several major drawbacks of this almost universal modeling process.

      a) First, it is idiosyncratic;

      b) Second, most models are highly specific for individual questions posed by the developers;

      c) Third, by necessity, all such models are tiny subsets of known signaling [16]; and

      d) Fourth, models rarely venture across scales, that is cross electrical and biochemical, or structural and genetic.

      Summary

      Neuronal signaling is a complex, multiscale phenomenon comprising genetic, biochemical, transport, structural, protein synthesis, electrical and network components. There is an abundance of models of specific parts of this landscape, with a special focus on electrophysiological properties of neurons and biochemical signaling in plasticity [8, 17]. Recent development has been the emergence of simulators designed for multiscale signaling as well as the incorporation of multiscale features in existing simulators [18-19]. With these developments the most common scale crossover, between spatially detailed electrical and chemical models, is greatly facilitated. To develop models of signaling pathways, it is necessary to consider the mechanisms by which signal transfer occurs. MAPK cascade communicates signals from growth factors that bind to receptor tyrosine kinases to transcriptional machinery and other cellular effectors. PKC in turn stimulates MAPK 1,2 through activation of the protein kinase c-Raf. MAPKs activate cPLA2. The arachidonic acid produced by cPLA2 also stimulates PKC. These pairwise connections create a potential positive-feedback loop in the MAPK network. This positive feedback loop involved PKC, MAPK, and PLA2 [14], and it is postulated that these enzymes mutually activate each other as shown in Fig. 3. The MAPK, PKC and PLA2 system is network that regulates many cellular machines, including cell cycle machinery and autocrine/paracrine factor synthesizing machinery.

      The current study examines how to systematically use experimental data to parameterize multiscale neuronal signaling models reproducibly, scalably, openly, and in rapidly expanding domain of neural physiology and signaling. We report two core developments: how to unambiguously and scalably match experimental observations to models, and how to manage development of models having thousands of components needing thousands of experimental constraints. Both are combined in FindSim, the Framework for Integration of Neuronal Data and Signaling Models. We explain FindSim and illustrate a model development pipeline capable of handling such models and their associated experiments.

      Literature review.png
        Positive feedback loop involving PKC, MAPK, and PLA2 in signaling network [18-19].

        METHODOLOGY

        Concepts

        We illustrate our approach using small subset of model for biochemical signaling which is designed to be embedded in a single-compartment model.

        Model development and parameterization pipeline

        In this study, we have developed a model and subject it to battery of experimental tests, defined in an open, extendable, and structured form (Fig. 4). A subset of model is simulated according to instructions derived from the experimental dataset, and the outcome for each such simulation is scored according to how well a model fits the data. To run through model, we implemented a python based script that reads the model, experiment definitions and launches MOOSE simulator to execute an experiment. This wrapper script then examines the outputs from MOOSE and compares these with those expected from experiment. The comparison is scored according to a user-defined scoring function specified as part of experiment definition. We have implemented a subset of model with structure of experiment specification, and a FindSim pipeline to systematically test the model against each experiment (Fig. 5) [4]. Each of these is in an open format and is accessible by other models and simulation tools.

        My model.png
          Specification of model subset for signaling neural network.
          Methdology.png
            Block diagram of the FindSim pipeline to test a model against experiment [4].

            Methods

            Current experiments touch only small but overlapping parts of very complex subcellular signaling networks in neurons.We illustrate our approach using a large core model of biochemical signaling which is designed to be embedded in a single-compartment electrical model. Based on experience with development of neuronal signaling models, both within NCBS group and from the published literature, we chose three categories of experiments for our initial set. These were time-series, dose-response and multi-stimulus response. Our reference model is a composite of several modeling studies linked together based on known interactions. To run through the models, we have implemented a Python based script that reads the model and experiment definitions and launches the MOOSE simulator to execute the experiment.

            Basic parameters for pathway involving model subset for PLA2.
            Name Value
            Name PLA2
            Notes Source of data: Leslie and Channon BBA 1045 (1990) pp 261-270 [20]. Fig 6 is Ca curve. Fig 4a is PIP2 curve. Fig 4b is DAG curve. Many inputs activate PLA2. The Km and Vmax of these active complexes is scaled according to the relative activation reported in the papers.
            Molecules Enzyme sites Reactions
            8 5 7
            List of molecules involved in subset of PLA2.
            Sr. No. Name Initial Conc (uM) Buffered Notes
            1 PLA2-cytosolic 0.4 0 cPLA2 IV form has mol wt of 85 Kd. Leslie and Channon use about 400 nM. Decent match, Use 400 nM.
            2 PLA2* 0 0 Phosphorylated PLA2. The site differs from the site phosphorylated by PKC [21].
            3 PLA2*-Ca 0 0 Phosphorylated and active form of PLA2.
            4 DAG-Ca-PLA2* 0 0 DAG and Ca complexed with PLA2. This form is moderately active.
            5 PLA2-Ca* 0 0 The generic Ca-activated form of PLA2.
            6 PIP2-PLA2* 0 0 PIP2 bound to PLA2. This form is moderately active.
            7 PIP2-Ca-PLA2* 0 0 PIP2 complexed with Ca and PLA2. This form is moderately active.
            8 APC 30 01 Arachodonylphosphatidylcholine is the favoured substrate [22-23]. Their assay used 30 uM substrate, which is what the kinetics in this model are based on. For now it is treated as a buffered metabolite.
            Reactions List for pathway PLA2.
            Sr. No Name Kf Kb Substrate Product Notes
            1 PLA2*-Ca-act 6 0.1 PLA2*Ca PLA2*-Ca Nemenoff et al report a 2X to 4x activation of PLA2 by MAPK, which seems dependent on Ca as well.
            2 DAG-Ca-PLA2-act 0.003 4 DAG PLA2-Ca* DAG-Ca-PLA2* Synergistic activation of PLA2 by Ca and DAG. The Kd is rather large and may reflect the complications in measuring DAG. For this model, DAG is held fixed.
            3 PLA2-Ca-act 1 0.1 PLA2-cytosolic Ca PLA2-Ca* Direct activation of PLA2 by Ca (Fig 7).
            4 PIP2-Ca-PLA2-act 0.012 0.1 temp-PIP2PLA2-Ca* PIP2-Ca-PLA2* Synergistic activation of PLA2 by Ca and PIP2.
            5 PIP2-PLA2-act 0.0012 0.5 temp-PIP2 PLA2- PIP2-PLA2* Activation of PLA2 by PIP2. In this model we don’t really expect any PIP2 stimulus.
            6 Dephosphorylate-PLA2* 0.17 0 PLA2* PLA2-cytosolic Dephosphorylation reaction to balance MAPK phosphorylation of PLA2. Rates determined to keep the balance of phosphorylated and non-phosphorylated PLA2 reasonable.
            7 Degrade-AA 0.4 0 AA cytosolic APC Degradation pathway for AA. For the purposes of the full model we use a rate of degradation of 0.4/sec to give a dynamic range of AA comparable to what is seen experimentally.
            List of Enzymes involved in pathway.
            Sr. No. Molecule Km Vmax Ratio (k2/k3) Substrates Products Notes
            1 PLA2*-Ca 20 120 4 APC AA This form should be 3 to 6 times as fast as the Ca-only form [24].
            2 DAG-Ca- PLA2* 20 60 4 APC AA Based on Leslie and Channon.
            3 PLA2-Ca* 20 5.4 4 APC AA Based on Leslie and Channon in relation to the other PLA2 inputs. Ca alone is rather a weak input.
            4 PIP2-Ca- PLA2* 20 36 4 APC AA Based on AA generation by different stimuli.
            5 PIP2-PLA2* 20 11.04 4 APC AA Based on Leslie and Channon.

            RESULTS AND DISCUSSION

            Experiment No. 1 Anionic phospholipids stimulate an arachidonoyl-hydrolyzing phospholipase A2 from macrophages and reduce the calcium requirement for activity

            Anionic phospholipids stimulate an arachidonoyl-hydrolyzing phospholipase A2 from macrophages and reduce the calcium requirement for activity. We computed the PLA2 pathway with its substrate, product and enzyme involved. Here we outlined the moelcules involved and studied from this paper.

            Molecules involved in neural pathway of PLA2
            Sr. No. Name Initial Conc Buffered Notes
            1 PLA2-cytosolic 0.4 cPLA2 IV form has mol wt of 85 Kd. Leslie and Channon use about 400 nM.
            2 PIP2-PLA2* 0 PIP2 bound to PLA2. This form is moderately active.
            3 PLA2-Ca* 0 The generic Ca-activated form of PLA2.
            4 DAG-Ca-PLA2* 0 DAG and Ca complexed with PLA2. This form is moderately active.

            This study looks at the PLA2 activity using arachidonyl-GPC as the substrate. PLA2 was extracted from RAW264.7 murine macrophage cell line. 2 ug of the enzyme was used with 30 uM of substrate and 1 uM CaCl2 in the absence of phosphatidylethanolamine (PE) for 1 min at 37 °C in a final volume of 50 ul.

            In this study, the effect of anionic phospholipids, DAG and PE on the activity of a partially purified, intracellular, arachidonoyl-hydrolyzing PLA2 has been performed. For these experiments PLA2 activity was assayed in presence of 1 μM calcium by measuring the hydrolysis of AA from sonicated dispersions of ether-linked substrate, l-O-hexadecyl-2[3H]arachidonoylglycerophosphocholine. All the anionic phospholipids tested, including phosphatidylserine (PS), phosphatidic acid (PA), phosphatidylinositol (PI) and phosphatidylinositol-4,5-bisphosphate (PIP2), stimulated PLA2 activity. PLA2 activity was assayed by measuring the hydrolysis of AA from sonicated dispersions of l-O-hexadecyl-2[3H]arachidonoyl-GPC (30 µM) that contained various amounts of the anionic phospholipids.

            FindSim Experiment Results

            The FindSim results for this experiment are depicted in Fig. 6 and 7 Score = 0.458 for LeslieC1990_Fig4A_MsOAoeA.tsv Elapsed Time = 0.2 s time to convert 0.21042323112487793

            PIP2_1.png
              Dose-response curve looking at PLA2 activity using various doses of PIP2.
              pip2_2.png
                Dose-response curve looking at PLA2 activity using various doses of PIP.

                This experiment suggests that a maximum activity is seen by 0.3uM pf PIP2, while no saturation occurs in 1 min for each does if I ignore the AA degradation to APC.

                Experiment No. 2. Direct activation of PLA2 by Ca-Lesli1990.

                The effects of calcium dose on PLA2 activity. PLA2 activity against 1-0-hexadecyl-2-[3H]arachidonoyl-GPC was measured as a function of calcium concentration in presence of 24 uM PE. This will use the pathway PLA2-Ca*. To ensure that the shape of the calcium dose-response in the presence of PIP, and PE was not as a result of this higher degree of substrate hydrolysis, the experiment was repeated using decreasing levels of enzyme.

                FindSim Experiment Results

                The FindSim results for this experiment is depicted in Fig. 8; Score = 0.050 for Direct_activation_of_PLA2_by_Ca-Lesli1990_EjN4hy6.tsv Elapsed Time = 0.3 s time to convert 0.21422529220581055 (Fig. 8).

                Ca2+.png
                  Direct activation of PLA2 by Ca-Lesli1990.

                  Experiment No. 3.

                  PLA2 enzyme activity is also assesesed by using various concentrations of DAG. This will use the pathway DAG-Ca-PLA2-act.

                  DAG-Ca-PLA2-act
                  1. Name
                  1. Kf
                  1. Kb
                  1. Substrate
                  1. Product
                  1. Notes
                  DAG-Ca-PLA2-act 0.003 4 DAG PLA2-Ca* DAG-Ca-PLA2* Synergistic activation of PLA2 by Ca and DAG. The Kd is rather large and may reflect the complications in measuring DAG. For this model it is not critical since DAG is held fixed.

                  FindSim Experiment Results

                  The FindSim results for this experiment are displayed as Fig. 9 and Fig.10 Score = 0.542 for _Fig_4a_DAG_curve.tsv Elapsed Time = 0.5 s time to convert 0.16009521484375

                  DAG Figure 4B.png
                    Figure for DAG response curve.
                    DAG Fig by FindSIM.png
                      Dose-response curve looking at PLA2 activity using various doses of DAG.

                      As the results are not satisfactory, I am trying to optimize this graph.

                      Experiment No 4.

                      Experiment Aims: Time course of hydrolysis of arachidonoyl phospholipids.

                      Experiment Detail: An 100‐kDa arachidonate‐mobilizing phospholipase A2 in mouse spleen and the macrophage cell line J774.

                      Molecule

                      Pathway of PLA2
                      Sr. No. Name Initial Conc (uM) Buffered Notes
                      1 PLA2-cytosolic 0.4 0 cPLA2 IV form has mol wt of 85 Kd. Decent match, Use 400 nM [20, 25].

                      Reaction list for pathway

                      Reaction list for pathway
                      Sr. No Name Kf Kb Substrate Product Notes
                      1 Degrade-AA 0.4 0 AA cytosolic APC Degradation pathway for AA. APC is a convenient buffered pool to dump it back into, though the actual metabolism is probably far more complex. For the purposes of the full model we use a rate of degradation of 0.4/sec to give a dynamic range of AA.

                      They have identified a PLA2 in murine macrophages with a much higher apparent molecular mass (70 kDa by gel chromatography), whose activity was increased after prior activation of cellular protein kinase C with phorbol myristate acetate. This and other properties of macrophage PLA2 strongly suggested that the enzyme is responsible for the PKC dependent mobilization of arachidonic acid seen in intact macrophages. Some of its enzymatic properties and the possibility that it acts as a substrate for PKC have also been investigated.

                      Experimental Conditions:

                      The standard assay mixture contained 0.2-0.5 nmol l-stearoyl-2-['H]AchPtdCho (20000 dpm) as sonicated vesicles, CaCI2, (625 nmol; equal to 400 pM free Ca2+), fatty-acid-depleted bovine serum albumin (100ug) and 1-10u1 enzyme in a total volume of 525 ul buffer A (without dithioerythritol). The mixture was incubated for 30 min at 37°C and stopped by the addition of 2.5 ml chloroform/methanol/l0 M HCI, (2:1:0.01 by vol.), unlabeled PtdCho and arachidonic acid as carrier. After centrifugation, the lipid phase was applied to a silicic acid column. Fatty acids were eluted with 1 ml chloroform and phospholipids with 2.5 ml methanol [22-23].

                      FindSim Experiment Results

                      The FindSim results for this experiment are appeared as Fig. 11 and Fig. 12 Score = 0.661 for FINDSim_WIJKANDER_and_SUNDLER_1991_vsE7jnR.tsv Elapsed Time = 0.1 s time to convert 0.17006564140319824

                      Time course of hydrolysis of arachidonoyl phospholipids.png
                        Relative hydrolysis of arachidonoyl phospholipids
                        Fig Wijkander.png
                          Time course of hydrolysis of arachidonoyl phospholipids.

                          Experiment No. 5.

                          Translocation of the 85-kDa phospholipase A2 from cytosol to the nuclear envelope in rat basophilic leukemia cells stimulated with calcium ionophore or IgE/Antigen [26].

                          Experiment Aims: To understand the cPLA2---> APC --> Arachidonic acid pathway.

                          Experiment Design: The rat mast cell line RBL-2H3.1 contains an 85-kDa cytosolic phospholipase A2 (cPLA2) that is very likely involved in liberating arachidonate from membrane phospholipid for the synthesis of eicosanoids following stimulation with either calcium ionophore or IgE/antigen.

                          Experiment Detail:

                          ModelSubset: PLA2/PLA2_p,PLA2/Arachidonic_Acid

                          ModelLookup: PLA2:PLA2/PLA2_p,Arachidonicacid:PLA2/Arachidonic_Acid

                          Parameter Change ConInit: PLA2/Arachidonic_Acid

                          Stimuli: PLA2 Readouts:Arachidonic_Acid

                          Observations:

                          Release of radiolabeled arachidonic acid from RBL-2H3.1 cells stimulated with Ca2+ ionophore or IgE/antigen. Release is expressed as the percent of total cellular radiolabeled arachidonic acid that was released into the culture medium. Activation with calcium ionophore produces a 10-fold larger release of arachidonate than does stimulation with IgE/antigen. Thus, the results suggest that the extent of membrane binding of cPLA2 correlates with the release of arachidonate and that the site of arachidonate liberation is the nuclear envelope where many of the enzymes that oxygenate this fatty acid are located.

                          FindSim Experiment Results

                          The FindSim results for this experiment are appeared as Fig. 13; Score = 0.6700 for Experiment_No_4__Sarah_Glover_et_al_1995_oDUOMYC.tsv Elapsed Time = 0.0 s time to convert 0.24033355712890625

                          Fig 2 Glover.png
                            cPLA2---> APC --> Arachidonic acid pathway.

                            CONCLUSION AND RECOMMENDATIONS

                            Together with the underlying Python-driven MOOSE simulation engine for multiscale models, FindSim is an open, standards driven, and scalable approach to develop reliable, large-scale models. These experimental methods allowed me to observe the predicted systems behavior only in a qualitative manner. To determine quantitative accuracy of our models, techniques that quantitatively and selectively measure biochemical reactions within the cell must be developed. Nevertheless, this first glimpse into the systems capabilities of a well-understood and widely used cell signaling system allowed me to start unraveling the underlying concepts of neural network. I am able to understand and analyze cellular signaling through basic concepts such as definition of reactions, rates, and concentrations. Further optimzation of the model by using FindSim may result in more beficial results. In conclusion, these studies indicate that simple biochemical reactions can, with appropriate coupling, be used to store information. Thus, reactions within signaling pathways may constitute one locus for the biochemical basis for learning and memory.

                            REFERENCES

                            1. Bhalla, U. S., Ram, P. T., & Iyengar, R. (2002). MAP kinase phosphatase as a locus of flexibility in a mitogen-activated protein kinase signaling network. Science, 297(5583), 1018-1023.

                            2. Bhalla, U. S., & Iyengar, R. (1999). Emergent properties of networks of biological signaling pathways. Science, 283(5400), 381-387.

                            3. Bhalla, U. S. (2002). Use of Kinetikit and GENESIS for modeling signaling pathways. Methods in enzymology, 345, 3-23.

                            4. Viswan, N. A., HarshaRani, G. V., Stefan, M. I., & Bhalla, U. S. (2018). FindSim: A Framework for Integrating Neuronal Data and Signaling Models. Frontiers in neuroinformatics, 12.

                            5. Bhalla, U. S. (2002). Biochemical signaling networks decode temporal patterns of synaptic input. Journal of computational neuroscience, 13(1), 49-62.

                            6. Ray, S., & Bhalla, U. S. (2008). PyMOOSE: interoperable scripting in Python for MOOSE. Frontiers in neuroinformatics, 2, 6.

                            7. Gleeson, P., Crook, S., Cannon, R. C., Hines, M. L., Billings, G. O., Farinella, Barnes, S. R., & Bhalla, U. S. (2010). NeuroML: a language for describing data driven models of neurons and networks with a high degree of biological detail. PLoS computational biology, 6(6), e1000815.

                            8. Bhalla, U. S., & Bower, J. M. (1993). Exploring parameter space in detailed single neuron models: simulations of the mitral and granule cells of the olfactory bulb. Journal of Neurophysiology, 69(6), 1948-1965.

                            9. Bhalla, U. S. (2002a). The chemical organization of signaling interactions. Bioinformatics 18, 855–863.

                            10. Bhalla, U. S. (2002b). Use of kinetikit and GENESIS for modeling signaling pathways. Methods Enzymol. 345, 3–23.

                            11. Bhalla, U. S., and Iyengar, R. (2001). Functional modules in biological signalling networks. Novartis Found. Symp. 239, 4–13; discussion 13–15, 45–51.

                            12. Kuroda, S., Schweighofer, N., and Kawato, M. (2001). Exploration of signal transduction pathways in cerebellar long-term depression by kinetic simulation. J. Neurosci. 21, 5693–5702.

                            13. Tanaka, K., Khiroug, L., Santamaria, F., Doi, T., Ogasawara, H., Ellis-Davies, G.C., Kawato, M., and Augustine, G.J. (2007). Ca2+ requirements for cerebellar long-term synaptic depression: role for a postsynaptic leaky integrator. Neuron, 54, 787–800.

                            14. Leitges, M., Kovac, J., Plomann, M., and Linden, D.J. (2004). A unique PDZ ligand in PKCa confers induction of cerebellar long-term synaptic depression. Neuron, 44, 585–594.

                            15. Dura-Bernal, S., Suter, B. A., Gleeson, P., Cantarelli, M., Quintana, A., & McDougal, R. A. (2019). NetPyNE, a tool for data-driven multiscale modeling of brain circuits. Elife, 8, e44494.

                            16. Heil, K. F., Wysocka, E., Sorokina, O., Kotaleski, J. H., Simpson, T. I., Armstrong, J. D., & Sterratt, D. C. (2018). Analysis of proteins in computational models of synaptic plasticity. BioRxiv, 254094.

                            17. Wils, S., & De Schutter, E. (2009). STEPS: modeling and simulating complex reaction-diffusion systems with Python. Frontiers in neuroinformatics, 3, 15.

                            18. Narayanan, R., & Johnston, D. (2010). The h current is a candidate mechanism for regulating the sliding modification threshold in a BCM-like synaptic learning rule. Journal of neurophysiology, 104(2), 1020-1033.

                            19. Tanaka, K., & Augustine, G. J. (2008). A positive feedback signal transduction loop determines timing of cerebellar long-term depression. Neuron, 59(4), 608-620. 20. Leslie, C. C., & Channon, J. Y. (1990). Anionic phospholipids stimulate an arachidonoyl-hydrolyzing phospholipase A2 from macrophages and reduce the calcium requirement for activity. Biochimica et Biophysica Acta (BBA)-Lipids and Lipid Metabolism, 1045(3), 261-270.

                            21. Nemenoff, R. A., Winitz, S., Qian, N. X., Van Putten, V., Johnson, G. L., & Heasley, L. E. (1993). Phosphorylation and activation of a high molecular weight form of phospholipase A2 by p42 microtubule-associated protein 2 kinase and protein kinase C. Journal of Biological Chemistry, 268(3), 1960-1964.

                            22. Wijkander, J., & Sundler, R. (1991). An 100‐kDa arachidonate‐mobilizing phospholipase A2 in mouse spleen and the macrophage cell line J774: Purification, substrate interaction and phosphorylation by protein kinase C. European journal of biochemistry, 202(3), 873-880.

                            23. Wijkander, J., & Sundler, R. (1992). Macrophage arachidonate-mobilizing phospholipase A2: role of Ca2+ for membrane binding but not for catalytic activity. Biochemical and biophysical research communications, 184(1), 118-124.

                            24. Lin, L. L., Wartmann, M., Lin, A. Y., Knopf, J. L., Seth, A., & Davis, R. J. (1993). cPLA2 is phosphorylated and activated by MAP kinase. Cell, 72(2), 269-278.

                            25. Wijkander, J. & Wykle, R. L. (1995). 5-Lipoxygenase products modulate the activity of the 85-kDa phospholipase A2 in human neutrophils. Journal of Biological Chemistry, 270(44), 26543-26549.

                            26. Diez, E., & Mong, S. (1990). Purification of a phospholipase A2 from human monocytic leukemic U937 cells. Calcium-dependent activation and membrane association. Journal of Biological Chemistry, 265(24), 14654-14661.

                            ACKNOWLEDGEMENTS

                            I would like to express my special thanks to my adviser, Professor Upinder Bhalla, Department of Neurobiology, National Centre for Biological Sciences, Bengaluru, for his guidance, motivation, enthusiasm, and immense knowledge. His guidance helped me in all the time of this project. Our interactions were always quite informal and friendly. I consider myself quite fortunate to have had such an understanding and caring adviser, throughout the summer research fellowship. I also extend my sincere thanks to Nisha, Surbhit, and all the lab members for being there always, to discuss the subject with me. These discussions were very insightful and helpful.

                            I am deeply indebted and express deep sense of gratitude to Professor Dr. Sanjay J. Surana, Principal, R. C. Patel Institute of Pharmaceutical Education and Research, Shirpur and Management of Shirpur Education Society who gave me the opportunity to work at National Centre for Biological Sciences, Bengaluru by allowing me to pursue a summer research fellowship.

                            I am thankful to Indian Academy of Sciences for providing me the opportunity to be a part of Upi’s Lab during Summer, 2019.

                            More
                            Written, reviewed, revised, proofed and published with