SPECIFIC AIM 1:
Aim #1a - Model kernel library for elements of networks involving CREB: Models will be developed for elements of an intracellular system that is important for a variety of cellular responses that are dependent on gene expression, and exemplify many of the features of complex genetic networks such as autoregulation (both negative and positive), convergence and crosstalk. This system is the network of genes regulated by CREB and related transcription factors.
One important and complex biochemical pathway in which a vital role is played by the CREB gene network is that leading to the formation of long-term memory (LTM). Strengthening of specific synapses is thought to be essential for the consolidation of LTM, and such synaptic enhancement requires gene transcription regulated by the CREB family of transcription factors. The figure below diagrams some of the elements of the CREB gene network and the dynamical properties that can emerge form the network (Smolen et al., 2001b; see also DeCesare and Sassone-Corsi, 2000). This network contains two regulatory motifs. First, there is a positive-feedback loop. Activation of Protein Kinase A (PKA) by the ligand-binding process (i.e., 5-HT stimulation of adenylyl cyclase) leads to phosphorylation of a transcription factor (i.e., CREB) and increased transcription/translation of feedback proteins such as ubiquitin hydrolase, which in turn, leads to increased activity of PKA (i.e., positive-feedback). Second, there is a "gate" that regulates activity in the positive-feedback loop. The ligand-binding process activates an additional kinase (e.g., MAP Kinase or MAPK), which in turn, phosphorylates the transcriptional repressor CREB2. Phosphorylation of CREB2 reduces its ability to repress transcription, and thereby allows the positive-feedback loop to function. Thus, the MAPK/CREB2 pathway "gates" the positive-feedback loop. This network manifests important dynamical properties. First, it can integrate signals over relatively long periods of time. Second, it has a threshold for inducing long-term changes. Third, it translates brief stimuli (with specific temporal characteristics) into long-lasting cellular responses (e.g., LTM).
Studies that help to understand the dynamics of CREB activation and gene expression in neurons undergoing synaptic enhancement will contribute to understanding how the CREB genetic regulatory system functions in other cells or tissues, because the intracellular signaling pathways activated as a result of stimuli that induce LTM are the same or similar to those activated in other cell types following hormonal or pharmacological stimuli that influence transcription of genes regulated by CREB. In other cells and tissues, the CREB gene network is important for mediating aspects of immunity, carcinogenesis, and pathologies such as diabetes.
The figure below summarizes key elements of the Drosophila circadian rhythm generator. The transcriptional activator dCLOCK activates per transcription (Bae et al., 2000; Lee et al., 1999). Subsequent to per transcription, dCLOCK appears to be bound by PER protein (in combination with another protein called TIM) so as to mask its DNA binding activity (Bae et al., 2000; Lee et al., 1999). This leads to repression (removal of activation) of per transcription, forming a negative-feedback loop in which PER represses its own formation. As indicated by the dashed box, PER undergoes multiple phosphorylation. However, all forms of PER appear capable of repressing per transcription. Subsequent to these phosphorylations, PER is degraded, allowing for another round of per transcription and another circadian oscillation in the level of PER and dCLOCK. These proteins also regulate other circadian "output" genes responsible for the behavioral aspects of circadian rhythms. Recent results indicate that positive feedback also characterizes this system. In Drosophila, dCLOCK protein represses dclock transcription. PER activates dclock transcription (Bae et al., 1998). A positive feedback loop can be envisioned as follows. If dclock expression is slightly activated, then the resulting activation of per transcription by dCLOCK results in binding of dCLOCK by PER. Thus, repression of dclock is relieved, and the level of dCLOCK increases further. The role of positive feedback is not understood, either in Drosophila or in Neurospora where similar interactions occur (Lee et al., 2000). One function of our model kernel will be to help assess the feasibility of proposed roles for positive feedback. As shown in the top portion of the figure, experimental data suggests substantial time delays between the regulation of per and dclock transcription and subsequent changes in the rate of appearance of PER or dCLOCK protein (Glossop et al., 1999; So and Rosbash, 1997).
Aim #2 - Experimental validation for the model of the
biochemical network involving CREB: This model will be validated
with experiments that utilize neurons of Aplysia in which
synaptic strengths can be regulated by the CREB network. These
experiments are essential in two ways. First, they will provide
insights into the ways in which specific elements of the gene network,
such as positive-feedback loops, govern the operation of the network.
Second, they will help determine key parameters, such as time
constants, that govern the response of the network to stimuli. Reporter
genes, regulated by CREB, will be introduced into these neurons and the
dynamics of their expression following neurotransmitter application
studied.
Aim #3a - Carry out simulations with various tools of
the Bio-SPICE project: To examine the current state of
the BioSPICE system, we performed a usability study of the BioSPICE
Dashboard and four simulators: BioSketchPad/Charon, BioSpreadSheet,
Jarnac/JDesigner and JigCell. Our goal was to examine the usability of
each simulator, both independantly and within the Dashboard, and the
interoperability of the various tools, using SBML as a language of
exchange between the different applications. Each tool provides a model
builder component and a simulator component. We examined the four
simulators with four models that produce oscillatory behavior or time
course of a long duration. Four models where used in our present study,
a circadian rhythm model (M1), a cell division cycle model (M3), an
allosteric model for glycolytic oscillations (M4) and a memory
induction model (M5). M1. The circadian rhythm model correspond to the
oscillations in the levels of core gene expression. Our model
uses a transcription factor (TF) which undergoes multiple
phosphorylation steps. Over the space of a day, TF protein becomes
fully phosphorylated and then degrades. This relieves TF repression so
that another "burst" of TF transcription can occur. M3. The cell
division cycle model describes the interaction of the proteins cyclin
and cdc2 and the activation of maturation promoting factor, a
heterodimer of cyclin and cdc2. The model for certain parameter regime
showed limit cycle solutions. M4. The allosteric model for glycolytic
oscillations is a
reduced model, of 2 ODEs, using a quasi-steady-state hypothesis and
dimensionless variables; and showed oscillations of substrate and
product concentrations. M5. The memory model contains 16 ODEs (partial
list below), several of which are highly nonlinear equations. The model
represents short-, intermediate-, and long-term phases of protein
kinase A (PKA) activation, as well as represents phosphorylation of the
transcription factor CREB1 by PKA and consequent induction of the
immediate-early gene Aplysia ubiquitin hydrolase (Ap-uch), which is
essential for long-term synaptic facilitation (LTF). The above models
were used to study the Dashboard and four applications of the BioSPICE
project: BioSpreadSheet, BioSketchPad with Charon, JigCell, and SBW
Jarnac and JDesigner. The goal was to simulate the models with each
tool, graph the resulting time course, and convert the model to SBML,
to compare usability and interoperability.
Aim #3b - Develop a Getting Started manual for first-time users
of Bio-SPICE: Develop
a Tutorial Manual that will guide
first-time users of BioSPICE through the processes of downloading and
installing BioSPICE, creating models, running simulations, and
analyzing
data. The tutorial will be based on three
realistic use-cases, which implement
previously published models of molecular processes that underlie
circadian
rhythms, cell division and glycolytic oscillations.