Submitted by: O'Connor, Simon
Authors: Simon O'Connor, Cardiff School of Biosciences,
Cardiff University,
Tim J.C. Jacob, Cardiff School of Biosciences, Cardiff
University,
Title: A GENESIS model of dendrodendritic feedback inhibition of the mitral cell in the olfactory bulb.
Abstract: In an early breakthrough in neuron modelling and understanding of mitral cell inhibition Rall et al.1 , modelling olfactory bulb field potentials, deduced that the mechanism must be dendrodendritic in nature. In further electrophysiological studies (e.g. Mori et al.2 ) this feedback mechanism was thoroughly characterised. Realistic models have been produced of individual mitral and granule cells (Bhalla and Bower3 ). In this GENESIS modelling study the Bhalla and Bower models are being adapted using the neuropharmacological data obtained from cell culture studies (Trombley and Westbrook4 ). We attempt to model the mitral cell feedback mechanism and replicate the electrophysiological data of Mori et al.
The model currently consists of a single mitral cell and single granule cell, which interact across dendritic synapses. These interactions are provoked either by antidromic current injection into the mitral cell short axon compartment or by orthodromic current injection into the mitral cell primary dendrite.
References
1 Rall, W., Shepherd, G.M., Reese, T.S., and Brightman, M.W. (1966) Dendrodendritic synaptic pathway for inhibition in the olfactory bulb. Exp. Neurol. 14, 44-56.
2 Mori, K., Kogure, S., and Takagi, S.F. (1977) Alternating responses of olfactory bulb neurons to repetive lateral olfactory tract stimulation. Brain Res. 133, 150-155.
3 Bhalla,
4 Trombley, P.Q. Westbrook, G.L. (1990) Excitatory Synaptic Transmission in Primary Cultures of Rat Olfactory Bulb, J. Neurophysiol. 64, 598-606.
Submitted by: Khurshudyan, Ara
Authors: Ara Khurshudyan, Neuronal Systems Mathematical Modeling Laboratory,
Orbeli Institute of Physiology, Yerevan,
Armen Sargsyan,
Neuronal Systems Mathematical Modeling Laboratory, Orbeli
Institute of Physiology, Yerevan,
Gilles van Luijtelaar, NICI-Biological
Psychology, PSY. LAB., Radboud Nijmegen University, Nijmegen,
Hovhannes Mkrtchian,
Neuronal Systems Mathematical Modeling Laboratory, Orbeli
Institute of Physiology, Yerevan,
Title: A model of cortical neuronal circuit with dynamic synapses
Abstract: We present a model of cortical neuronal circuitry that allows to investigate the possible mechanisms responsible for different types of oscillations that may emerge in this circuit, namely, the sleep spindles and spike and wave discharges (SWD), and how they are related to each other. It is known that history-dependent synaptic plasticity strongly defines the oscillatory properties of the network and plays a significant role in self-organizing the network to produce highly synchronous discharges. Synaptic plasticity may play an important role in transition from sleep spindles to SWDs: pathological changes in dynamic properties, or frequency characteristics, of either excitatory or inhibitory (or both) cortical synapses may lead to an imbalance between excitation and inhibition, resulting in increased excitability of cortical pyramidal cells. The model consists of cortical pyramidal cells (PC) and cortical interneurons involved in feed-forward (FFI) and recurrent inhibition (RI), lateral inhibition and lateral excitation. The neurons are represented by compartmental models of different complexity, depending on the type of neuron. An arbitrary number of PCs and interneurons may be included in this model. Each included PC may receive its own thalamic input, which is independent from the others and may be of arbitrary pattern. Each modelled PC may have different position along vertical (cortical depth) direction. The substantial advantage of this model is introducing plastic synaptic connections, that is, the strength of a connection is not constant but changes depending on temporal pattern of presynaptic activity, both for excitatory and inhibitory synapses. For this, our newly developed model of synaptic transmission that combines the abilities to undergo both short- and long-term history-dependent changes in synaptic efficacy (diachronic model) was used. The model is implemented by means of the GENESIS neurosimulator. The diachronic synapse and the random spike train generator (for uniform, Poisson and Gaussian distributions) used to drive the thalamic inputs were realized as new GENESIS objects. We carried out simulations with several modifications of the network model, differing in types of synapses used for PC-PC, PC-RI excitatory, and RI-PC, RI-RI inhibitory connections. Our results prove, that even a small network of pyramidal cells interconnected with excitatory connections and inhibitory ones via interneurons, is capable to demonstrate synchronized periodic firing of pyramidal cells even in the case of random external inputs. The highest synchrony was achieved when both excitatory and inhibitory interconnections between pyramidal cells were strong. It is shown that the degree of synchronization substantially depends on plasticity properties of each synapse and their various combinations (when the rest of model parameters is kept unchanged).
Submitted by: Sargsyan, Armen
Authors: Armen Sargsyan, Neuronal Systems Mathematical Modeling Laboratory,
Orbeli Institute of Physiology, Yerevan,
Gilles van Luijtelaar, NICI-Biological
Psychology, PSY. LAB., Radboud Nijmegen University, Nijmegen,
Albert Melkonyan, Neuronal Systems
Mathematical Modeling Laboratory, Orbeli Institute
of Physiology, Yerevan,
Title: Simulation of sleep spindles and spike and wave discharges using a novel method for field potential.
Abstract: We suggest a method for calculation of extracellular field potentials generated by a large population of pyramidal cells (PCs), using a single PC compartmental model. Similar methods described earlier use the assumption that the intracellular potential or current distributions of the cells within the population are much alike as a result of simultaneous activation at about the same longitudinal location (i.e., all the PCs in the population are located on the same level and are ideally synchronized). However, the degree of synchronization of natural firing even during synchronized rhythmic discharges in the cortex is not as high. We introduce the possibility to vary the degree of synchronization of the PCs’ activity in the population, thus taking into account disperse timing of cortical pyramidal cells’ firing. The temporal variability in cell firing is described by Gaussian distribution, the width of which defines the degree of synchronization/desynchronization. In addition, the suggested method allows for certain spatial spread of PCs in the population along longitudinal axis of the PCs. The method was applied to test the assumption that the transition from sleep spindles to rhythmic spike and wave discharges (SWDs) observed in absence epilepsy may occur due to an increase in pyramidal cells' firing synchronization. The simulations were performed using the GENESIS neurosimulator. The field potential calculation algorithm was implemented as a new GENESIS object, allowing for calculation and visualization of field potentials at several different spatial locations simultaneously, directly during the simulation run. We show that in case of weak synchronization of PC firing in the population, the shape of field potential during rhythmic thalamic input is similar to the oscillations during a sleep spindle, while at stronger synchronization of PCs, it looks much more as a SWD, with clear expressed spikes and waves. This suggests that in large population of pyramidal cells the changes in the degree of synchronization of cell firing may sufficiently explain the changes in the shape of field potential from spindle oscillations to SWDs and vice versa, and no additional mechanisms or factors are necessary.
Submitted by: Lytton, William
Authors: William W. Lytton, Depts of Physiology, Pharmacology, Neurology, SUNY Downstate
Title: Neural Query System -- data-mining from the simulation environment
Abstract: We have developed a database management system call Neural Query System (NQS) inside the
NEURON simulator. This software package allows the user to set up and manage a
relational database and provides query facilities modeled on those available with SQL.
A modeling tool such as NEURON or GENESIS is typically viewed as an
adjunct to data-mining. However, a database system has three major roles to play in
simulation: 1. management of the simulations themselves; 2. organization and
analysis of simulation output; 3. comparison of simulations and experimental data.
The enormous number of parameters involved in compartment simulations with active
dendrites makes it hard to summarize parameters or readily reorganize them to
produce comparable dynamics in a different dendritic tree. The inclusion of these
models within large networks adds still greater complexity. We have been using
NQS for single cell model and network definition, as well as for retrospective
graphical analysis to confirm that a desired simulation design was implemented
correctly.
As a technique, neural modeling requires large amounts of experimental data and
creates large amounts of simulation data. We have begun to use NQS to organized,
access and compare these data. Centralizing data organization within the
simulator should also permit better understanding of the effects of parameters at
different levels. For example, we would use NQS to evaluate the relative
contributions of ion channel dynamics and connectivity parameters on the dynamics
of a network.
Submitted by: Publio, Rodrigo
Authors: Rodrigo Publio,Dept. de Física e Matemática, FFCLRP,USP, Ribeirão Preto, SP
Antonio C. Roque,Dept. de Física e Matemática, FFCLRP,USP, Ribeirão Preto, SP
Title: Simulation of a rod photoreceptor for use in a realistic retina model
Abstract: We are developing a biologically accurate model of the mammalian retina in NEURON and we report here the first results relative to simulations of the rod photoreceptor and of a layer of such photoreceptors interconnected by gap junctions. The rod model was adapted from a model by Kamiyama et al. (Kamiyama, Y., Ogura, T. and Usui, S., Ionic current model of the vertebrate rod photoreceptor, Vision Res., 36:4059-4068, 1996). It contains the following ionic currents placed at its inner segment: hyperpolarization activated current (Ih), delayed rectifying potassium current (IKv), potassium current (IKx) and calcium current (ICa). The effect of light transduction was simulated by a photocurrent applied directly into the inner segment. The rod model was able to replicate experimentally found behavior of rod photoreceptors when stimulated with light (Baylor, D. A, G. Matthews, and B.J. Nunn. Location and function of voltage-sensitive conductances in retinal rods of the salamander, Ambystoma tigrinum.J. Physiol. 354:203-223, 1984).
Submitted by: Robertson, Richard
Authors: Richard Robertson, Dept. of Mathematics, Calif. State Polytechnic U,Pomona
Kerstin Menne, Inst. für Signalverarbeitung, Universität zu Lübeck,
Title: Toward the creation of a subiculum-like model, to study the role of the subiculum in human TLE
Abstract: Starting with Kerstin Menne’s freely-available GENESIS model of CA3, we have taken some preliminary steps toward making our model more closely resemble the subiculum. Because of its strategic position in controlling input and output activity from the hippocampal formation, various experimentalists and epileptologists have postulated that the subicular complex might play a key role in temporal lobe epilepsy. Cohen et al. ("On the Origin of Interictal Activity in Human Temporal Lobe Epilepsy in Vitro", Science, 15 Nov 02, Vol. 298) discovered that a spontaneous, rhythmic activity, reminiscent of interictal epileptic discharging, was frequently initiated in the subicula of slices resected from human patients with TLE. This sort of spontaneous activity was almost never observed by them in any other region of the parahippocampal formation. However, until recently, very little was known about this potentially pivotal structure.
Liset Menendez de la Prida, an expert on (rat) subicular complex electrophysiology and morphology, has definitively classified its basic cell types: IB+, “strong bursting”, IB-, “weak bursting”, and RS, “regular spiking” pyramidal cells, with FS, “fast spiking” interneurons. Furthermore, she has estimated their relative numbers and has begun to uncover their intrinsic cellular mechanisms and network connectivities. It seems that the bursting cells, especially the IB+ type, are the major contributors to a subicular output derived from local activity. Others, like her colleague, Richard Miles, have made similar observations using human subicular slices, mostly from TLE patients.
To date, we have taken the best, most recent data on subicular cells and their connectivities to re-model Kerstin’s CA3, to make it as “biologically realistic” a ‘subiculum-like’ structure as such a small (< 100 neurons now) network could be. We present preliminary data here to demonstrate several ways in which it displays behaviors suggestive of a real subiculum, as well as others where its performance, so far, is less promising.
Our perspective is that such a model, built on the flexible, modular, and biologically-sound GENESIS platform, can continue to be refined and expanded to behave more and more like a real subiculum. Our goal is to eventually produce a computer model that can assist theoreticians, experimentalists, and perhaps even clinicians in studying the possible role of the subiculum in the pathophysiology of mental disorders such as Alzheimer’s disease and the generation and/or propagation of TLE seizures.
Submitted by: Robbins, Kay
Authors: Egle Pilipaviciute, Dept of Computer Science, UTSA, San Antonio, TX
Kay A. Robbins, Dept of Computer Science, UTSA, San Antonio, TX
David M. Senseman, Dept. of Biology, UTSA, San Antonio, TX
Title: Using Davis to Compare Neural Models
Abstract: Davis (Data Viewing System) is a general-purpose data viewer designed for the simultaneous display and comparison of dynamic data sets. Inspired by the need to explore data from imaging experiments and large computational models of the cerebral cortex, Davis allows scientists to study the detailed behavior of individual elements and the interaction of these elements to achieve cortical function. Davis provides direct views that animate activity over time. Davis also supports remapped space-time activity diagrams, parallel coordinate latency visualizations and low dimensional-projections of correlated behavior that facilitate comparison of structure across data sets. We demonstrate how Davis can be used to understand and compare the dynamic behavior across models, as well as to reveal relationships between underlying variables.
Submitted by: Roth, Arnd
Authors: Arnd Roth, Wolfson Institute for Biomedical Research and Department of
Physiology, University College London, London,
Title: Dendrites: Bug or Feature?
Abstract: How does dendritic morphology shape the functional architecture
of different types of neurons? Using compartmental models of
reconstructed neurons endowed with the same distribution of active
conductances we isolate morphology as the only variable. We show that
the spread of subthreshold synaptic potentials, the forward- and
backpropagation of action potentials in dendrites as well as the
interaction of somatic and dendritic action potential initiation sites
is tuned by subtle details of the dendritic branching pattern.
Submitted by: Cannon, Robert
Authors: Robert C. Cannon, Informatics, University of
Edinburgh,
Title: Design Patterns for Engineering with Spikes, Synapses and Dendrites
Abstract: The process of building systems that use (for example) only spikes,
synapses and dendrites (SSD), instead of normal programming
constructs, is a form of constrained engineering that can be used to
generate candidate mechanisms for comparison with natural neural systems.
It is analogous to archaeologists deciding to live for six
months with only bronze age tools and technology in order to gain a
deeper understanding of the people they are studying.
We explore a range of design patterns that prove useful in building
SSD systems for tasks involving learning, decision making and
spontaneous activity. One goal of this work is to replace the
very difficult task of invention that faces the traditional top-down
or bottom-up modeler in looking at a natural implementation of an
unknown mechanism, with the much easier task of recognition.
Engineers who build complex SSD systems may find that some of the
structures ("design patterns") they invent relate closely to natural
structures. This can then provide comparison points and a conceptual
framework for further study.
Another goal is to establish the family of constructs that can be
reliably and robustly implemented with SSD. These can then form the
basis of a higher level specification language (declarative, not
procedural as programming languages are) for tackling more
complex tasks at lower computational cost. Such a language should be
far more accessible to a broad range of researchers than are
conventional programming languages.
Submitted by: Kondra, Shripad
Authors: Shripad A. Kondra, Institute of Signal Processing, University of Luebeck, Luebeck,
Kerstin M. L. Menne, Institute
of Signal Processing, University of Luebeck, Luebeck,
Steffen Klatt, Institute of
Signal Processing, University of Luebeck, Luebeck,
Ulrich G. Hofmann, Institute of Signal Processing, University
of Luebeck, Luebeck,
Title: A small network with huge artificially grown pyramidal cells
Abstract: In this work, we present our efforts to improve a small CA3 network consisting of 72 pyramidal cells and 18 interspersed interneurons to simulate multisite neuronal recordings with more natural spike output. Up till now, the existing 100 cell network was able to provide challenging signals, very well suited to validate spike detection algorithms with its ground truth. Unfortunately, no clustering algorithm could be tested on these data sets, since all cells fired the same and unchanging spike shape. We attributed this behaviour to the “clone”-like similarity of the 60 compartment, Traub-related pyramidal cells utilised in this network.
To cope with this problem, we will upgrade the model by replacing the “clone”-like identical pyramidal cells with synthetically, in computer grown, way more complex pyramidal cells. These cells were designed by Lindenmayer-growth grammar of natural pyramidal cells, each consisting of up to 6000 morphologically realistic compartments.
We are using a Sunfire 64-node SMP machine with mpGENESIS (MPI communication protocol) and are currently setting up a Linux-Cluster for pGENESIS (PVM protocol) based on the cheap Microsoft Xbox-platform.
At the time of this writing, up to 12 nodes on the SMP machine are used for simulating the old model, but we will expand the model to include more classes of cells and utilise more nodes. We have studied the individual response of the artificially grown cells. We are convinced that the possibility to simulate the model with our artificially grown cells will generate recordings with more classes of spikes, which will assist to program robust and complete spike-sorting algorithms.
Keywords: Biologically realistic site recordings, artificial cells, parallel programming, GENESIS, GROGRA, Lindenmayer-systems
Submitted by: Sautois, Bart
Authors: - Alan Roberts, School of Biological Sciences,
University of Bristol, Bristol, BS8 1UG,
- Wen-Chang Li, School of Biological
Sciences, University of Bristol, Bristol, BS8 1UG,
- Steve R Soffe, School of Biological
Sciences, University of Bristol, Bristol, BS8 1UG,
- Ervin S Wolf, Department of Anatomy, University of Debrecen,
Debrecen,
- Bart Sautois, Dept. of Applied
Mathematics and Computer Science, Ghent University, Ghent,
Title: Modeling the frog tadpole nervous system and behavior
Abstract: Very young frog tadpoles can exhibit 2 types of movement: they swim (alternating, low amplitude, head to tail waves) when touched and struggle (alternating, high amplitude, tail to head waves) when held. The tadpole nervous system is very simple with less than 2000 neurons in total controlling these responses; neurons which belong to possibly less than 10 classes of spinal neurons. Moreover, these neurons are so small that they can be accurately modeled as single-compartment neurons. So this system is definitely eligible for modeling.
Previous modeling has always used a single type of model for all neurons. Our recent whole-cell patch results show that different types of neurons have different properties. We want to assess the significance of this, by first building individual neuron models for each type, and then linking them in a network that matches the biological findings. Being a numerical computer scientist and working a lot with dynamical systems, the modeling so far is done by working directly on the differential equations and their coefficients, not by using special software as GENESIS or NEURON. The work is still in an early stage of development.
Submitted by: Raman, Karthik
Authors: Karthik Raman, Supercomputer
Education and Research Centre/Bioinformatics Centre, Indian Institute of Science,
Bangalore,
Preethi Rajagopalan,
Bioinformatics Centre, Indian Institute of Science, Bangalore,
Nagasuma Chandra, Supercomputer
Education and Research Centre/Bioinformatics Centre, Indian Institute of Science,
Bangalore,
Title: Flux-based analysis of the mycolic acid pathway in Mycobacterium tuberculosis
Abstract: Tuberculosis is one of the leading health concerns in the world, caused by the pathogen Mycobacterium tuberculosis. Creation of an in silico model of the microbe can help understand the molecular basis of the disease and provide a platform for the design of better drugs.
The unique production of mycolic acid is a critical step for the survival of M. tuberculosis. Mycolic acid provides a protective coat to the pathogen, acting as a barrier to hydrophilic solutes and is responsible for the resistance of M. tuberculosis to common anti-bacterial drugs such as Ampicillin, Cephalosporins, Tetracyclins etc. The mycolic acid pathway has been the target for several anti-tuberculosis drugs, such as isoniazid and ethionamide.
A quantitative simulation of this pathway can give us an insight into the pharmacology of the drug administered. However, kinetic parameters for only few of the concerned reactions are available in literature --- the information available is still insufficient to set up a simulation. In the absence of such quantitative information, a (steady state) metabolic model for these pathways can be established. Flux analysis on this model can give us information on the behaviour of the system. Such a model can also be used to analyse the effect of drugs such as isoniazid on the system.
The flux-based modelling and simulation of the mycolic acid pathway and the implications of simulation will be discussed. Such metabolic models can be of general use in understanding drug action.
Submitted by: McCormick, Bruce
Authors: David M. Mayerich, Dept. of Computer Science, Texas A&M University, College Station, TX
B.L. Busse, Dept. of Computer Science, Texas A&M University, College Station, TX
Louise C. Abbott, Dept. of Veterinary Integrated Biosciences, Texas A&M University, College Station, TX
Bruce H. McCormick, Dept. of Computer Science, Texas A&M University, College Station, TX
Title: Spatial Distribution and Morphology of Mouse Neurons
Abstract: We have produced the first detailed map of 3D brain microstructure across a whole mammalian brain, that of the C57BL/6J mouse. Two brains were used: one in Nissl stain, showing the spatial distribution and morphology of the stained cell bodies; and one in Golgi-Cox stain, to elucidate morphology of selective cells; approximately 1% of neurons stain in the Golgi stain. Three-dimensional reconstruction of only selected regions of Golgi-stained microstructure has been attempted to date, in view of the computational burden entailed.
A unique instrument of our invention, the Knife-Edge Scanning Microscope (KESM), scanned each entire mouse brain at 300nm sampling resolution and created an aligned volume data set of ~7 terabytes (uncompressed) representing the tissue microstructure. New tissue preparation techniques have been developed for en bloc staining and embedding with Nissl and Golgi-Cox stained tissue. In addition, new 3D parallel image-processing algorithms, including our polymerization algorithm, have been developed to geometrically reconstruct and visualize the cell bodies and dendritic and axonal arborization of the neurons. Lastly, an I/O-intensive cluster computer supports the endeavor.
Support for the Brain Networks Laboratory contributed by: NSF-MRI Grant 0079874, Texas Higher Education Coordinating Board Grant ATP-00512-0146-2001 and the Office of the Vice President, Texas A&M University.
Submitted by: McCormick, Bruce
Authors: B.L. Busse, Dept. of Computer Science, Texas A&M University, College Station, TX
David M. Mayerich, Dept. of Computer Science, Texas A&M University, College Station, TX
Louise C. Abbott, Dept. of Veterinary Integrated Biosciences, Texas A&M University, College Station, TX
Bruce H. McCormick, Dept. of Computer Science, Texas A&M University, College Station, TX
Title: Biologically Accurate Modeling of Mouse Brain Requires Biologically Accurate Networks
Abstract: Biologically accurate networks will soon be made possible by a combination of new methods in electron and light microscopy. These instruments, both using physical sectioning, are respectively, the Serial Block Face Scanning Electron Microscope (SBF-SEM), invented by Winfried Denk, Max Planck Institute, Heidelberg, and the Knife-Edge Scanning Microscope (KESM), a light microscope invented by our Texas A&M group. The SBF-SEM necessarily relies on heavy element staining of tissue to provide image contrast. The KESM light microscope is unique in its ability to acquire brain microstructure from tissue stained in common with identical heavy element stains.
Shown here are exploratory tracings and reconstructions of mouse brain tissue stained with osmium tetroxide and scanned with the KESM. Segmenting the osmium datasets poses a unique challenge because, unlike Nissl or Golgi stains, osmium staining gives a rich cellular detail throughout the tissue. The osmium images, at lower resolution, look not unlike those seen at higher resolution in the SBF-SEM. Our exploratory image analysis of brain tissue stained with heavy elements, both with light microscopy and with electron microscopy, has forced us to adopt a new image analysis paradigm: Stain all neurons; reconstruct selectively. This reverses the conventional wisdom held previously: Stain sparsely, reconstruct exhaustively. In the new strategy we can return to the original data set for additional data, if required; in the old strategy, there is no additional data to return to, at least not in the same mouse.
Support for the Brain Networks Laboratory contributed by: NSF-MRI Grant 0079874, Texas Higher Education Coordinating Board Grant ATP-00512-0146-2001 and the Office of the Vice President, Texas A&M University.
Submitted by: Choe, Yoonsuck
Authors: Yoonsuck Choe, Department of Computer Science, Texas A&M University
Yingwei Yu, Department of Computer Science, Texas A&M University
Title: Role of the thalamic reticular nucleus in selective propagation of the results of cortical computati
Abstract: The intricate circuitry involving the thalamus, the thalamic reticular
nucleus (TRN), and the cortex are known to serve a diverse set of functions.
These functions include attentional modulation, binding through
synchronization,
mediation of cortico-cortical communication, corollary dispatchment of sensory
signals to motor areas, and more broadly, consciousness and awareness.
A common theme penetrating all of the above is that of selective gating, or
filtering. However, it has not been clear what kind of general computational
principle can be underlying such a process. A careful analysis of the temporal
characteristics of this circuit (e.g., membrane time constant and axonal delay)
reveals that the TRN may be implementing a selective filter for corticothalamic
feedback that has a cortical origin, rather than a peripheral origin (e.g.,
deriving from sensory afferents). In other words, the TRN seems to be
selectively enhancing the results of cortical computation, and suppressing the
input-driven cortical activity that triggered those results. For example, given
a sensory input as a question, the thalamus-TRN-cortex circuit can generate a
cascade of answers, where each time the last answer gets promoted. In this
poster, we will present results from GENESIS simulations of Hodgkin-Huxley
neurons supporting this idea. The results indicate that well-known temporal
properties of the circuit, such as slow activation of TRN neurons and slow
corticothalamic feedback, work together to implement such a selective
propagation mechanism which favors the results of cortical computation. The
parameters and results from the model are detailed enough to be verified
experimentally.
Submitted by: Scheler, Gabriele
Authors: Gabriele Scheler ISLE Stanford, Ca. 94305
Title: Effects of cocaine sensitization on storage and learning capacity in Nucleus Accumbens
Abstract: Data by White etal(2004,2002,1998) have shown that 5 day exposure to
cocaine and 2 day withdrawal leads to significant changes of ion channel
conductances in accumbens medium spiny neurons (MSNs).
Synaptic plasticity is affected in a similar manner (ThomasBonci2001).
The sensitized accumbens exhibits an altered activation
pattern, with abnormally low responses to rewarding stimuli
and a reduction of the capacity to acquire new rewards.
We have constructed a model that employs parameter updates
in ion channel conductances to develop individually adapted neurons
and we showed how pattern recognition can be performed by these neurons.
With the data on cocaine sensitization, we now show that
the sensitization changes in activation functions and synaptic weights
correspond to information loss for previous stimuli and
a reduction of new learning capacity. We attempt to measure this
capacity quantitatively.
Submitted by: Fontanini, Alfredo
Authors: Alfredo Fontanini - Volen Center for Complex Systems, Brandeis University, Waltham, MA, 02454
James M. Bower - Research Imaging Center, University of Texas Health Science Center at San Antonio and Cajal Neuroscience Center, University of Texas San Antonio, San Antonio, TX, 78284-6240
Title: Coupling between olfactory system activity and respiration in ketamine/xylazine anesthetized rsts
Abstract: The study of the relationship between cerebral cortical oscillations and behavioral states has been a major focus of neuroscience since the first EEG recordings were obtained several decades ago. Perhaps the most enduring relationships found have been the association of increased cortical activity (e.g. waking and REM sleep) with high frequency, low amplitude oscillations, and presumed periods of cortical inactivity (e.g. slow wave sleep - SWS -, or deep states of anesthesia) with low frequency, high amplitude oscillations modulating periodic bursts of high frequency activity .
In this study we have characterized slow and fast oscillation at several stages of olfactory processing under light and deep ketamine/xylazine anesthesia in the albino rat. While monitoring the animal’s respiration, we also obtained field potentials from the olfactory bulb and cortex and simultaneously recorded membrane potentials in piriform (olfactory) cortex pyramidal cells. Our results demonstrate that oscillations are generally found at higher frequencies under lighter and lower frequencies under deeper anesthesia. In previous studies of cerebral cortex, similar results in ketamine/xylazine anesthetized animals have been interpreted to correspond with the higher frequencies found during waking, and lower frequencies found in the sleep state.
Correlation and coherence analysis between data obtained in the bulb and cortex reveals a clear difference in coupling depending on the anesthetic state of the animal. Specifically, activity recorded in the whole system is highly correlated with respiration during deep anesthesia, whereas only the olfactory bulb, and not the cortex, is correlated with respiration during light anesthesia.
This data suggests that global activity in the piriform cortex is actually more directly tied to peripheral slow respiratory input during slow wave than fast wave states and that the coupling between olfactory structures can be dynamically modulated by the level of anesthesia and therefore presumably by different brain states as well.
Submitted by: Gomez-Molina, Juan
Authors: Juan F. Gomez-Molina, Dept. of Biology, UTSA, San Antonio, TX
Kay A. Robbins, Dept. of Computer Science, UTSA, San Antonio, TX
David M. Senseman, Dept. of Biology, UTSA, San Antonio, TX
Title: Differences in the Spatiotemporal Dynamics of Cortical Waves Evoked by Stationary and Moving Stimuli
Abstract: A large-scale GENESIS model of the turtle cerebral cortex was used to study differences in the spatiotemporal dynamics of traveling waves evoked by a stationary stimulus or stimuli moving either in a nasal (N) or temporal (T) direction. Using Davis (Data Visualization System), we found that each stimulus condition produced a propagating wave of excitation in the cortical model that traveled in a rostrocaudal direction with distinctly different spatiotemporal characteristics. These differences were mostly easily studied by calculating low-dimensional subspaces using Karhunen-Loeve (KL) decompostion and then quantitatively analyzing the projections of subspace pairs that captured movement along the rostrocaudal axis. One obvious difference in the spatiotemporal characteristics was the presence of a partial wave reflection at the caudal boundary only for a stimulus moving in a T-to-N direction. The generality of this and other differences are current under investigation.
Submitted by: Voicu, Horatiu
Authors: Horatiu Voicu, Dept. of Neurobiology and Anatomy, UTH, Houston, TX
Michael D. Mauk, Dept. of Neurobiology and Anatomy, UTH, Houston, TX
Title: Homeostatic plasticity - too much of a good thing
Abstract: It has been hypothesized that homeostatic plasticity is the mechanism
used by the brain for maintaining neurons within an optimal dynamic
firing frequency range. Moreover, it has been suggested that in order
to compensate for the positive feedback produced by local
correlational learning rules, homeostatic plasticity should be
ubiquitous in the brain.
In this paper we simulate a detailed model of the cerebellum to show
that too many homeostatic plasticity mechanisms can have a detrimental
effect on the performance of the network. This decline in performance
is due to the reduced number of degrees of freedom that control the
resulting neural network.
Submitted by: Li, Guoshi
Authors: G. Li, Ph.D. student, and S. S. Nair, Professor, Mechanical & Aerospace Engineering
S. Lee, Research Associate, and F. Booth, Professor, Veterinary Biomedical Sciences
University of Missouri, Columbia MO 65211
Title: REGULATION OF THE G2/M TRANSITION BY P53 IN MAMMALIAN CELL CYCLE
Abstract: A computational model of the regulation of G2/M transition by p53 in a mammalian cell cycle has been developed to investigate control mechanisms associated with the G2 checkpoint. This model has 19 states and includes CDK/Cyclin B complexes, p53, p21, Chk1, Cdc25, Mdm2 and 14-3-3. A detailed network model of G2/M regulation is developed from which a “core” subsystem is then extracted. Based on the existing model of Mitosis control, a reduced order model is developed which consists of the two most important pathways regulating G2 cycle arrest in response to DNA damage. The DNA damage signal is modeled as a first order system with constant DNA repair rate. The model shows that the p53 dependent pathway is not required for initial G2 arrest since the Chk1/Cdc25 pathway can arrest the cell in G2 right after DNA damage. However, p53 and p21 expression is important for a more sustained G2 arrest by inhibiting the Thr161 phosphorylation by CAK. By eliminating the phosphorylation effect of Chk1 on p53, two completely independent pathways are obtained and it is shown that it does not affect the G2 arrest time much. Hence, the p53/p21 pathway makes an important and independent contribution to G2 arrest in response to DNA damage, and any defect in this pathway may lead to genomic instability and predisposition to cancer. We conjecture that such double-check mechanisms are widely present in biological systems as a self-protection and self-survival means to various environmental changes. Also, the controversial issue on the mechanism of inactivation of Cdc2 by p21 is addressed and simulation results indicate that G2 arrest would not be affected much by considering the direct binding of p21 to Cdc2/Cyclin B given that the inhibition of CAK by p21 is already present.
Submitted by: Howard, Harry
Authors: Harry Howard, Spanish & Portuguese, Tulane University, New Orleans, LA
Title: Liquid State Machines for sensory, executive, and motor neocortex
Abstract: The Liquid State Machine framework for modeling neocortical microcircuits (Maass et al. 2002 and posterior work) aims at a high degree of neural realism, yet paradoxically does not rely on a high degree of neural mimesis. This paradox is finessed by the claim that the goal of the intricate (and not entirely mapped, much less understood) connectivity of neurons in a microcircuit (say, a minicolumn) is to provide a sufficiently high-dimensional and richly dynamic fading analog memory from which simple (i.e. linear) read-out neurons can extract a pattern instantly (‘anytime computing’) or within a short time window (‘real-time computing’). Thus the degree of neurological detail that the modeler needs to implement is only so much as to ensure a sufficient variety of synapses, neurons, and connections. As Maass et al. (2003) put it: “It … is not necessary to construct circuits to achieve substantial computational power. Instead sufficiently large and complex “found” circuits … tend to have already large computational power, provided that the reservoir from which their units are chosen is sufficiently rich and diverse.”
To summarize in a nutshell, a “liquid state” x(t) consists of the vector of contributions of all the neurons in the microcircuit to the membrane potential at time t of a generic read-out neuron and is all the information about the state of the microcircuit to which the read-out neuron has access. It is assumed to vary continuously over time and, as we said, to be sufficiently sensitive and high-dimensional to contain all the information needed for a given task.
In contrast to general purpose simulators like GENESIS and NEURON which are free-standing packages, the liquid-state framework is implemented in MATLAB, though its computational units (integrate-and-fire neurons) are written in C++ and compiled in MATLAB as MEX files. Similar to GENESIS and NEURON, the liquid-state simulator, the neural Circuit SIMulator CSIM, is distributed under the GNU General Public License from a Web site, <http://www.lsm.tugraz.at/csim/index.html>.
In this paper, we outline liquid-state computing, demonstrate how it distinguishes sensory from executive from motor neocortex, and time permitting, compare it to GENESIS and NEURON.
LSM References
Bertschinger, N., & Natschläger, T. (2004). Real-Time Computation at the Edge of Chaos in Recurrent Neural Networks. Neural Computation, 16(7), 1413-36.
Häusler, S., Markram, H., & Maass, W. (2003). Perspectives of the high dimensional dynamics of neural microcircuits from the point of view of low dimensional readouts. Complexity,
Maass, W., Natschläger, T., & Markram, H. (2002). Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural Computation, 14(11), 2531-60.
Maass, W., Natschläger, T., & Markram, H. (2003). A Model for Real-Time Computation in Generic Neural Microcircuits. In S. Becker, S. Thrun, K. Obermayer (Eds.), Proc. of NIPS 2002. Advances in Neural Information Processing Systems, volume 15. (pp. 229-36MIT Press.
Natschläger, T., & Maass, W. (2004). Dynamics of information and emergent computation in generic neural microcircuit models.
Natschläger, T., & Maass, W. (2004). Information dynamics and emergent computation in recurrent circuits of spiking neurons. In S. Thrun, L. Saul, B. Schölkopf (Eds.), Proc. of NIPS 2003, Advances in Neural Information Processing Systems, volume 16. (pp. 1255-62). Cambridge: MIT Press.
Submitted by: Weaver, Christina
Authors: Christina M. Weaver, Center for Computational Biology, Computational Neurobiology and Imaging Center, Dept. of Neuroscience, Mount Sinai School of Medicine, New York, NY
Susan L. Wearne, Center for Computational Biology, Computational Neurobiology and Imaging Center, Dept. of Neuroscience, Mount Sinai School of Medicine, New York, NY
Title: Role of the objective function and boundary management in constrained optimization of neuronal compartment models.
Abstract:
Compartmental neuron models typically have large numbers of free parameters which interact nonlinearly, only loosely constrained within physiologically plausible ranges. These parameters must be estimated when fitting simulations to electrophysiological data, but the complexity of this problem renders manual parameterization very difficult. Accordingly, the use of automated parameter search methods has become increasingly common. An important step in parameter fitting is to select an objective function that accurately represents the key differences between the model and the experimental data. Action potential (AP) shape has been shown to be a critical determinant of neuronal firing dynamics. We describe construction of an objective function that incorporates both AP shape error and errors in the statistics of firing rate and firing regularity. Furthermore, applying physiological boundary constraints requires an automated parameter search method which operates intelligently within them. We implement a variant of simplex-based simulated annealing, a stochastic optimization method which is robust for a range of modeling applications. The search algorithm replaces any point outside the boundaries it encounters in parameter space by a random point within a neighborhood of the current minimum, improving convergence to the best minimum within the boundary constraints. NEURON's Multiple Run Fitter is used to implement and to evaluate the objective function and optimization method. We show how our choice of objective function allows the model to capture essential features of neuronal firing patterns, and why our boundary management technique is superior to previous approaches.
Submitted by: Maye, Alexander
Authors: Alexander Maye, Dept.
of Scientific Visualization, Zuse Institute Berlin
(ZIB),
Jan-Felix Evers, Institute of Neurobiology, Free University
of Berlin,
Carsten Duch,
Institute of Neurobiology, Free University of Berlin,
Ulrich Nowak, Dept. of Numerical Analysis and Modelling, Zuse Institute Berlin
(ZIB),
Peter Deuflhard, Dept. of Numerical
Analysis and Modelling, Zuse
Institute Berlin (ZIB),
e.g. - John Q. Smith, Dept. of Physiology, UTHSCSA, San Antonio, TX
Title: Morphological dependence of temporal signal integration properties in an insect motoneuron during
Abstract: Our aim is to elucidate the role of changing dendritic morphology and synapse distribution to support changing behavioral demands on computation during reshaping of dendritic projection patterns. We have chosen the motoneuron 5 of the hawkmoth Manduca sexta as model system. This neuron is changing from a slow motoneuron involved in crawling behavior in larvae to a fast motoneuron, controlling a major part of the main downstroke flight muscles in the flying adult moth. This changing behavior is accompanied by severe changes in dendritic morphology, making this neuron a valuable biological model for studying structure-function relationships.
Using a novel method for semi-automatic tracing of neurites, we fitted a cylindrical model precisely to confocal images of four developmental stages (larva, wander stage, pupal stage 5 and adult). Second channel image data of immune-histochemically labeled presynaptic proteins (Synapsin-I, Synaptotagmin) were interpreted as spatial probability distributions for afferent synapses. The membrane potential dynamics resulting from synaptic activity were described by a passive conductance model. For numerical integration of the PDEs we employed the
fast and efficient solver LIMEX.
For each morphology we were interested in the efficiency of generating an axonal depolarization response depending on the number of activated synapses and the degree of synchrony of the activation. To this end we sampled synapse localization from the synapse density distribution and generated synapse activation events
according to different exponential distributions.
Our findings indicate significant influence of changing neuronal morphology on temporal integration of synaptic activity. Currently, we investigate integration
dependency on spatial distribution of synapse density.
Submitted by: Howard, Harry
Authors: Harry Howard; Spanish & Portuguese, Linguistics, Neuroscience; Tulane University, New Orleans, LA
Title: Liquid State Machines for sensory, executive, and motor neocortex (delete other)
Abstract: Liquid-state machines for sensory, executive, and motor neocortex
The Liquid State Machine framework for modeling neocortical microcircuits (Maass et al. 2002 and posterior work) aims at a high degree of neural realism, yet paradoxically does not rely on a high degree of neural mimesis. This paradox is finessed by the claim that the goal of the intricate (and not entirely mapped, much less understood) connectivity of neurons in a microcircuit (say, a minicolumn) is to provide a sufficiently high-dimensional and richly dynamic fading analog memory from which simple (i.e. linear) read-out neurons can extract a pattern instantly (‘anytime computing’) or within a short time window (‘real-time computing’). Thus the degree of neurological detail that the modeler needs to implement is only so much as to ensure a sufficient variety of synapses, neurons, and connections. As Maass et al. (2003) put it: “It … is not necessary to construct circuits to achieve substantial computational power. Instead sufficiently large and complex “found” circuits … tend to have already large computational power, provided that the reservoir from which their units are chosen is sufficiently rich and diverse.”
Having established such a circuit, a read-out neuron can be trained to recognize some aspect of its activity in the following manner: states x(t) of the microcircuit are recorded at various time points in response to numerous (training) inputs, and a supervised learning algorithm is applied to a set of training examples of the form [x(t),y(t)] to train a read-out function f for which the actual outputs f(x(t)) are as close as possible to the target outputs y(t).
There are several advantages to this approach. One is that all temporal processing is done implicitly in the circuit, so is that it is not necessary to take any temporal aspects into account for the learning task. Another advantage is that no a-priori decision must be made regarding the neural code by which information about preceding inputs is encoded in the current liquid state of the circuit. A third is that the rich dynamics of the microcircuit support many different read-out functions working in parallel, through the training of a separate readout neuron for each target output. A final advantage is that there is no a-priori limitation on the power of this model, though a larger circuit is required to implement computations that require greater memory capacity or greater noise-robust pattern discrimination.
In contrast to free-standing simulators like GENESIS and NEURON, the liquid-state framework is implemented in MATLAB, though its computational units (integrate-and-fire neurons) are written in C++ and compiled in MATLAB as MEX files. Similar to GENESIS and NEURON, the liquid-state simulator, the neural Circuit SIMulator CSIM, is distributed under the GNU General Public License from a Web site, <http://www.lsm.tugraz.at/csim/index.html>.
In this paper, we outline liquid-state computing, demonstrate how to vary the architecture of a liquid-state network so as to mimic sensory, executive, and motor neocortex, and, time permitting, compare CSIM to GENESIS and NEURON.
LSM References
Bertschinger, N., & Natschläger, T. (2004). Real-Time Computation at the Edge of Chaos in Recurrent Neural Networks. Neural Computation, 16(7), 1413-36.
Häusler, S., Markram, H., & Maass, W. (2003). Perspectives of the high dimensional dynamics of neural microcircuits from the point of view of low dimensional readouts. Complexity,
Maass, W., Natschläger, T., & Markram, H. (2002). Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural Computation, 14(11), 2531-60.
Maass, W., Natschläger, T., & Markram, H. (2003). A Model for Real-Time Computation in Generic Neural Microcircuits. In S. Becker, S. Thrun, K. Obermayer (Eds.), Proc. of NIPS 2002. Advances in Neural Information Processing Systems, volume 15. (pp. 229-36). MIT Press.
Natschläger, T., & Maass, W. (2004). Dynamics of information and emergent computation in generic neural microcircuit models.
Natschläger, T., & Maass, W. (2004). Information dynamics and emergent computation in recurrent circuits of spiking neurons. In S. Thrun, L. Saul, B. Schölkopf (Eds.), Proc. of NIPS 2003, Advances in Neural Information Processing Systems, volume 16. (pp. 1255-62). Cambridge: MIT Press.
Submitted by: Gu, Yuqiao
Authors: Yuqiao Gu, Dept. of Biometry and Engineering,
SLU, Uppsala,
Hans Liljenström, Dept. of Biometry
and Engineering, SLU, Uppsala,
Title: Modeling the Neurodynamics of Insect Olfaction - With Applications to Biological Control
Abstract: Based on experimental facts about anatomical circuits, physiological data and hypotheses of the olfactory system of insects, we have developed a dynamical neural network model to simulate recognition and discrimination of host plant odors and mating pheromones. In particular, we model how the spatio-temporal representation patterns of sex pheromone components and host plant odors emerging in the glomeruli of the antennal lobe (AL) rely on the interplay and synchronous oscillation of the local and global AL circuits. We also investigate the dose-response characteristics of glomerular activity in relation to the network structure and complex dynamics. We further study how weak signals can be amplified (sensitivity to individual odour molecules), and how a specific mixture can be discriminated, based on the neuronal and network properties and stochastic resonance dynamics.
Using Hodgkin-Huxley type models we have simulated the dynamics of isolated single projection neurons (PN) and local interneurons (LN) in the AL, as well as small networks of glomeruli, composed of pairs of PNs and LNs. Simulation results demonstrate synchronous oscillation characteristics, as experimentally observed in the AL of many insects. When separating the input stimulus into two parts, background and odor signal, computer simulations show that the PNs and LNs oscillate in lower frequency under constant background input. While the odor stimulus pulse is presented to the network, the oscillatory frequency increased. In addition, a smaller difference between the input odor pulse strength to the two PNs induces bigger difference in output dynamics. This indicates that the glomerular circuit is very sensitive to the difference in inputs. It is found that this output difference decrease with decreasing connection strength from PN to LN. If we connect the two PNs with cholinergic synapses, this difference varies with the connection strength.
Submitted by: Holmes, William
Authors: William R. Holmes, Dept. Biological Sciences, Ohio University, Athens, OH
Lawrence M. Grover, Dept. Physiology, Marshall University School of Medicine, Huntington, WV
Title: Fitting experimental data to models that use morphological data from public databases
Abstract: Ideally, detailed models should make use of morphological and electrophysiological data from the same cell. However, this rarely happens. The assumption of the modeler is that the morphological data used is representative of the particular cell type and that experimental data comes from an equally representative cell. Here we seek to determine how different the parameter value fits to the same experimental data would be with the use of different morphological reconstructions. We model morphological data for 4 CA1 pyramidal cells obtained from 3 different databases. Experimental data was obtained from 19 CA1 pyramidal cells, 6 of which were exposed to zd-7288 to block H channels. The multiple run fitter in NEURON was used to fit parameter values in each of the 4 morphological models to match experimental data for each of the 19 cells. Excellent fits were obtained in almost all cases, but average fitted parameter values for Ra, Rm, Cm and (for non zd-7288 treated cells) H conductance were very different among the 4 reconstructions. Average Ra and Cm values were similar for cells exposed or not exposed to zd-7288 for a given cell morphology. The widely different parameter values obtained for the different morphological models can be explained by highly different measurements of diameter, total length, membrane area, and volume among the reconstructed cells.
Submitted by: McCormick, Allyson
Authors: Allyson V. McCormick, Dept. of Genome Sciences, UW, Seattle, WA
James H. Thomas, Dept. of Genome Sciences, UW, Seattle, WA
Title: Analysis of UNC-43 control of neuronal synchrony in Caenorhabditis elegans
Abstract: Mutations in a few C. elegans genes can result in a convulsive phenotype marked by repeated end-to-end full body contractions caused by the simultaneous excitation of dorsal and ventral body wall muscles. unc-43 encodes the only CaM Kinase II homologue in C. elegans. unc-43(lf) worms have rare spontaneous convulsions that are intensified in frequency and severity by exposure to neurostimulants (e.g., pentylenetetrazole [PTZ] and pilocarpine).
To investigate the underlying molecular mechanisms of these convulsions, we are looking in detail at the neuronal requirement for UNC-43. Using neural subtype-specific promoters, we have delineated the requirement of UNC-43 to the motor neurons. In addition, we have demonstrated that this neural control of convulsions requires only that UNC-43 be present in the adult nervous system by using the heat-shock promoter hsp16-2 which implies that the convulsion phenotype of unc-43(lf) is separable from its defects in early neuron development (1). To explore the neural output of the motor neurons involved in the convulsing worm, we have expressed the in vivo calcium-sensor, cameleon (2), specifically in these neurons. Our data confirms that unc-43(lf) motor neurons display aberrant synchronized activity whereas wild type does not.
These experiments characterize the role of UNC-43 in controlling simultaneous excitation of neuronal networks in C. elegans. Similar types of neuronal synchrony may cause human epilepsies, since a knockout of the a-CaM Kinase II (3) gene in mouse is susceptible to spontaneous and drug-induced seizures. We believe that our use of C. elegans, as a well-studied and tractable model organism combining genetics and a mapped nervous system (4) may be a powerful way to characterize key molecular components controlling seizures. Furthermore, our project is primed for more sophisticated computational analysis to better understand both normal and seizuring neural activity.
1 Rongo, C. and Kaplan, J. M. (1999). Nature 402:195-199
2 Rex, K. et al. (2000). Neuron 26: 583-594
3 Butler, L.S., et al. (1995). PNAS 92:6852-6855
4 White, J.G., et al. (1986). Phil. Trans. R. Soc. Lond. 314:1-340
Submitted by: Gall, John
Authors: J. Gall, Ph.D. student University of Missouri-Columbia
W. Smith, Ph.D. student University of Missouri-Columbia
S. Nair, Ph.D. University of Missouri-Columbia
J. Nair, M.D. University of Missouri-Columbia
D. Xu, Ph.D. University of Missouri-Columbia
P. Kalivas, Ph.D., Medical University of South Carolina
Title: Transition to bistable behavior for the Nacc medium spiny neuron due to cocaine
Abstract: The transition to bistable behavior for an Nacc medium spiny neuron is investigated using reduced order modeling techniques. Evidence of a stable hyperpolarized (Down) and a depolarized (Up) state is a well known response to input stimulus. A biologically realistic model of an Nacc neuron developed by our group is taken as the starting point for the development. Reduction of order is considered via analysis of bifurcations. Examination of the full system at a stable steady state via destabilization helps determine the important mechanisms that contribute to this bistability. Uncoupling feedback interactions to determine the role that individual elements play in maintenance of local stability retains the mechanisms that contribute to the nonlinear behaviors, while allowing for reduction of order by assumption of constant values for the stable states. The effects of glutamate and dopamine modulation on ionic currents are not fully understood at the present time. Model reduction may be achieved by considering the reward dependent dopamine release in terms of a modulatory factor that affects key ionic currents (Gruber, 2003). This facilitates examination of current interaction and the impact these dynamics have on membrane potential bifurcations. Bistable behaviors often emerge in response to enhancement of inward currents as well as NMDA. Glutamate release, symptomatic of cocaine use introduces bistable behavior via synaptic channels. The nature of firing as it is modulated by glutamate input is carefully examined in the reduced order model. Transition to multiple steady states from a single stable state is generated by model bifurcations, and reduction of order modeling facilitates analysis of these via phase plane analysis of two variable interactions. Further analysis of bistability in response to multivariable feedback interaction is considered. Multistability in the higher order system may be examined by studying the feedback interactions of the subsystems for monotonicity and monostability. Analysis of subcellular emergent properties allows for modeling of signaling pathways, and feedback interactions are characterized. A system exhibiting monostable steady state response to constant input as well as monotonicity in positive feedback in subsystems may be examined for bistable properties without order reduction. Model predictions will be compared with experimental results.
Submitted by: Iyoho, Anthony
Authors: A. Iyoho, University of Missouri-Columbia
S. Nair, Ph.D, University of Missouri-Columbia
J. Nair, M.D., University of Missouri-Columbia
B. Beitman, M.D., University of Missouri-Columbia
Title: A computational model of the basolateral amygdalar neuron
Abstract: The rate of return to craving and drug seeking behavior in cocaine addicts is high, and is further exacerbated by exposure to certain environmental stimuli. Three types of environmental stimuli that can precipitate relapse include drug exposure, stress, and drug-associated cues. Studies have shown that the basolateral amygdala (BLA) and central amygdala nucleus (CeA) play a key role in cue-primed and stress-primed reinstatement, respectively. The amygdala can be grouped into three distinct subdivisions that include 1) the superficial or cortical-like group 2) the deep or basolateral group; and 2) the centromedial group. Approximately, 70% of the neuronal cells in the BLA can be described as pyramidal or class I cells (different from pyramidal cells in the cortex) and the rest of the cell population are stellate-like or class II cells . The BLA receives glutamatergic inputs from the thalamus and cortex among other places, and projects glutamatergically primarily to the centromedial amygdala nuclei, but also to the prefrontal cortex, nucleus accumbens, and thalamus. The majority of cells in the lateral and medial division of the CeA have been labeled “medium spiny neurons” similar in structure to striatal neurons. Projections from the CeA are primarily GABAergic. The brain circuitry involving cue-primed and stress-primed reinstatement is distinct, but converges to similar anatomical structures. For instance, cue-primed reinstatement follows from dopamine innervation from the ventral tegmental area (VTA) to the BLA, then from the BLA to the anterior cingulate, and finally from the anterior cingulate to the core of the nucleus accumbens (NAcore). Whereas, stress-primed reinstatement involves adrenergic input from the lateral tegmental nucleus to the extended amygdala (CeA, bed nucleus of the stria terminalis (BNST), and shell of the nucleus accumbens), and it is hypothesized that a connection linking the extended amygdala region to the VTA and/or dorsal prefrontal cortex concludes this circuit. In particular, it has been suggested that dopamine (DA) participates in modulating the BLA in cocaine-seeking behavior during reinstatement. Specifically, the D1 receptors seem to be the sight of modulation as selective DA D1 antagonists attenuated conditioned reinstatement behavior. Thus, modeling the BLA would benefit the research community concerned with cocaine addiction and relapse. The model will consist of the pyramidal neuron found in the BLA including the major synaptic inputs with an emphasis on studying dopaminergic modeling of the BLA in regards to cue-primed reinstatement.
Submitted by: Burns, Gully
Authors: Gully APC Burns, Dept. of Neurobiology, University of Southern California, Los Angeles, CA
Title: Extracting and managing model parameters from the literature.
Abstract: Models must be constrained by parameters. These are usually derived from experimental observations published in the literature compiled into a literature review by the researcher building the model. Biologically accurate models tend to require a great number of parameters and collating the relevant information from the literature can become a challenging knowledge management task for researchers. Several factors contribute to the difficulty of this: some parameters may not be documented (requiring assumptions to be made & documented); some published information may not be reliable; the literature is continually evolving; and the nomenclature used may be non-standardized and variable. Computational support for this process may make it easier to build models and may improve the quality of the definition of the model. This presentation concerns NeuroScholar, an open-source knowledge management system for the neuroscientific literature. Essentially, NeuroScholar builds knowledge representations from a library of fragments derived from full-text articles (or scanned pages when an online full-text article is not available). Additional functionality includes the capability of managing the following data types: schematic diagrams (to illustrate the organization of a model system), and delineations of regions of brain tissue drawn on neuroanatomical atlases. The system’s schema follows object-oriented design principles and incorporates a dynamic graph-based schema-viewer in the user-interface. Since connectivity data is commonly used within modeling studies, we demonstrate how the system can be used to build a representation of tract-tracing data with a detailed example describing circuits concerned with defensive behavior system in the rat.
Submitted by: Lim, Heejin
Authors: Heejin Lim, Dept. of Computer Science, Texas A&M Univ. College Station, TX
Yoonsuck Choe, Dept. of Computer Science, Texas A&M Univ. College Station, TX
Title: Facilitatory neural dynamics for extrapolation and delay
Abstract: In flash-lag effect, the position of a moving object is perceived to be
ahead of a briefly flashed stimulus when they are actually physically
co-localized at the time of the flash. A potential explanation for this
phenomena is the motion extrapolation model. Given that nervous systems
have conduction delay, the delayed object-location data have to be
extrapolated so that the perceived location of the object at a given
moment is the same as that in the environment, at that same instant.
According to the motion extrapolation hypothesis, flash-lag effect is
caused by such a delay compensation mechanism embedded in our nervous
system. However, the neural mechanism of such an extrapolatory process has
not been fully investigated. In this poster, we propose that facilitating
synapses can serve as an effective neural mechanism for implementing
extrapolation. Facilitating synapses have been thought to be contributing
to memory-related processes, however, it turns out that they can actually
help compensate for neural conduction delay as well. We tested our idea in
the visual luminance flash-lag effect domain. The results indicate that
the extrapolated firing rate in our facilitating synapse model accurately
reflects that of experimental data. In sum, facilitatory dynamics found in
neurons may be contributing to delay compensation, which is expressed in
the flash-lag effect. We expect our perspective to shed new light on the
role of facilitatory neural dynamics.
Submitted by: Cline, C. Harvey O.
Authors: C. H. Cline, K. Pietarila, T. Jang, Ph.D. students
S. Nair, Ph.D., J. Nair, M.D., and B. Beitman, Ph.D.
University of Missouri-Columbia
Title: COMPUTATIONAL MODELING OF AN OFC PYRAMIDAL CELL
Abstract: Two neurotransmitters related to obsessive compulsive disorder (OCD) and attention deficit hyperactivity disorder (ADHD) are serotonin and dopamine. The underlying mechanisms for these disorders would be of interest since ADHD involves short attention span while OCD involves excessive focusing. The Orbital Frontal Cortex (OFC) is an important brain region for these disorders. Biologically realistic models of a pyramidal cell in the OFC is developed with focus on the intracellular pathways of serotonin and dopamine systems, as part of studies to model the disorders. The software package GENESIS is used for intracellular and membrane potential level modeling.
OCD. In a healthy serotonergic system, serotonin is released into the synaptic cleft to act on both postsynaptic receptors, for signal transmission, and presynaptic receptors, for regulation of serotonin production and release. Inactivation occurs via membrane carriers which transport serotonin from the synaptic cleft back into the presynaptic neuron, reuptake [3]. Seven classes of serotonin receptors have been identified thus far, with all of the receptor classes, except for 5-HT3 , being coupled to G-proteins. The literature indicates possible participation from the 5-HT1D receptor in OCD. Data suggests that SSRIs may improve obsessional symptoms through the down-regulation of the 5HT1D receptor in the OFC, thus facilitating serotonergic transmission here The principal cell of the orbital frontal cortex is the pyramidal cell. The computational model being developed includes the important receptors and all the relevant intracellular signaling pathways.
ADHD. The influences of the dopamine receptor D2, D3 and
D4 genes are thought to be critical for inhibition and attention. D4 is particularly
associated with ADHD and is mainly found in the STR as well as the PFC. Dopamine
2-type receptors, such as D4, reduce cyclic adenosine mono phosphate (cAMP)
levels thus inactivating protein kinease A (PKA). PKA is known to phosphorylate
many receptors and ion channels. The dopamine transporter (DAT1) protein may
play a role in ADHD. DAT1 exists in the surface of dopamine secreting cells,
regulates the action of dopamine by reuptake of the neurotransmitter into
the presynaptic neuron (acts by active transport of the dopamine molecule
from the synapse across the cell membrane against a concentration gradient
into the presynaptic neuron). ADHD individuals have on average an abnormally
high density of DAT within the striatum. Current research indicates complex
interactions between other neurotransmitters including 5-HT and norepinephrine
both of which contribute to ADHD both individually and in interactions with
DA. All the relevant mechanisms will be modeled for the pyramidal cell in
the orbital frontal cortex.
Submitted by: Smith, Wesley
Authors: Wesley Smith, University of Missouri
John Gall, University of Missouri
Satish Nair Ph.D., University of Missouri
Jyotsna Nair Ph.D., University of Missouri
Bernard Bietman M.D., University of Missouri
Peter Kalivas Ph.D., Medical University of South Carolina
Title: Neuroplasticity in the Glutamatergic Pathway for a Nucleus Accumbens Medium Spiny Neuron
Abstract: A biologically realistic computational model of the medium spiny neuron in the nucleus accumbens has been developed with focus on the glutamatergic PFC-Nacc pathway. The model considers neuroplasticity effects due to cocaine, including dendritic spine growth. A multi-level framework is proposed to incorporate known gene to membrane potential level dynamics, with the capability to integrate new data as they become available.
The framework includes trans- extra- and intra-cellular level details. Extracellular interactions are modeled by synaptic injections of dopamine and glutamate. These systems are modeled with reuptake mechanisms, synaptic concentrations, and bound concentrations. The trans-neuronal subsystem is characterized by neurotransmitter interactions including, NMDA and AMPA binding of Glutamate, D1 and D2 binding of dopamine. Membrane potential is regulated via ionic channels. Standard ionic channels for outward rectifying potassium are included. Dopamine modulation of slowly inactivating potassium currents, PKA modulated L- and T- type calcium, as well as PKA phosphorylated fast inward sodium are also modeled. The intraneuronal subsystem contains cAMP and Ca2+ diffusion dynamics with the dendrites. The DARPP-32 PP-1 cascade is modeled within the compartments. Somatic, dendritic, and spine compartments are modeled, with glutamate and dopamine synapses at spine locations. Features of membrane potential firing are considered, including bistable behavior. Up and Down state events occurring in the medium spiny neuron are a result of the modeled interactions. Excitatory glutamatergic input from the cortical afferents excites NMDA and AMPA receptors, responsible for instantiation of plateau potentials, while GABAA influences the duration of the Up state. Dopamine transmission in PFC during cocaine reinstatement affects the glutamate release to the accumbens via AGS3, and this affect modulates the synaptic input. The membrane potential dynamics predicted from the model are compared with experimental data.
Submitted by: Summers, Jonathan
Authors: J. P. Summers, undergrad student, University of Missouri, Columbia, MO
G. Li, Ph.D. student, University of Missouri, Columbia, MO
C. H. Cline, Ph.D. student, University of Missouri, Columbia, MO
S. Nair, Ph.D., University of Missouri, Columbia, MO
J. Nair, M.D., University of Missouri, Columbia, MO
B. Beitman, M.D., University of Missouri, Columbia, MO
Title: Preliminary network model for a neuropsychiatric disorder
Abstract: A network-level model of the relevant brain structures has been developed for modeling the interactions of the different brain regions associated with obsessive compulsive disorder (OCD). OCD is characterized by hyperactivity of the orbital frontal cortex (OFC) and caudate nucleus resulting in the hallmark symptoms of obsessions and compulsions. It is believed that an overactive OFC projects preferentially to the ventromedial region of the head of the caudate nucleus resulting in an overstimulation of the so-called direct pathway through the basal ganglia (BG). The direct pathway through the BG, through the output structures, provides inhibition to the thalamus which projects back to the cortex creating a neural “loop”. With these BG output structures inhibited by the overactivated striatum, they fail to provide the thalamus with adequate inhibition resulting in excessive projection back to the OFC. It is the resulting amplification of this loop that is believed to cause the symptoms of OCD.
The computational model is developed using the GENESIS software. Individual
region response as well as combined region interaction are of interest when
attempting to characterize the system. Simplified model neurons are used to
create the individual regions and utilizing a two-compartment, Hodgkin-Huxley
structure. Each nuclei (caudate, putamen, globus pallidus - interna and externa,
substantia nigra - pars reticulata and pars compacta, subthalamic nucleus,
and the thalamus) is divided into sub-regions according to organization reported
in the literature. The spherical sub-regions are internally organized and
connected to one another using a spatial organization scheme. The cortical
input into the model is represented by random field inputs. To study the model
responses and outputs, EEG elements as well as graphical representations of
neural activity are used.
Submitted by: Blackwell, Kim
Authors: Jeanette Hellgren Kotaleski:Dept Numerical Analysis
and Computer Science, Royal Institute of Technology, S-100 44 Stockholm,
Maria Lindskog: Dept Numerical
Analysis and Computer Science, Royal Institute of Technology, S-100 44 Stockholm,
Kim T Blackwell: George Mason University, School of Computational
Science and the Krasnow Institute for Advanced study,
MS 2A1, Fairfax, VA 22030,
Title: Modulation of long term synaptic plasticity by dopamine activated second messenger pathways
Abstract: Recent evidence points to time critical interactions between cortical inputs and back-propagating action potentials in the control of calcium, which plays a role in synaptic plasticity. In other brain regions these time critical interactions control not only calcium but also synaptic plasticity. In the striatum, where dopamine is required for plasticity, an open question is whether the temporal interval between dopamine and cortical inputs is critical in producing plasticity of cortico-striatal synapses. Long term potentiation of this synapse results from increased phosphorylation of AMPA channels, produced by activation of protein kinase A and inactivation of protein phosphatase 1. These two enzymes interact through the intracellular phosphoprotein DARPP-32, expressed at high levels in spiny projection neurons in the striatum. The biochemical pathways leading from dopamine and glutamatergic inputs to DARPP-32 has been delineated, but the interaction of these biochemical reaction pathways is not well understood.
To investigate the time course of interactions, and the effect of temporal pattern of stimulation on DARPP-32 modulation and consequent synaptic plasticity, we developed a computer model of the biochemical reaction pathways involved in the phosphorylation of DARPP-32 on threonine 34 and threonine 75. The ordinary differential equations describing the biochemical reactions were implemented in a single compartment model using the software XPP. Reaction rate constants were obtained from the biochemical literature. Simulation results show that the model exhibits sensitivity to the order of dopamine and glutamate stimulation: PKA activation is greater when glutamate (calcium) occurs prior to dopamine. In addition, PP1 activity is smaller when glutamate occurs prior to dopamine. This is due to interactions with DARPP-32 which inhibits PP1 after PKA phosphorylates threonine 34. Further simulations examine mechanisms whereby synaptic plasticity depends on the interactions between dopamine and glutamate stimulation.
Submitted by: Tripp, Patrick
Authors: Patrick G. Tripp, Research Imaging Center, The University of Texas Health Science Center at San Antonio
Fidel Santamaria, Department of Neurobiology, Duke University Medical Center, Durham, NC 27710
James M. Bower, Research Imaging Center, The University of Texas Health Science Center at San Antonio
Title: Parallel fiber modulation of cerebellar Purkinje cell responses to paired afferent inputs
Abstract: We have previously proposed that cerebellar granule cells have two distinctly different physiological effects on Purkinje cells: a modulatory effect mediated through the parallel fibers (and associated molecular layer interneurons), and a direct excitatory effect mediated through synapses made by the ascending granule cell axonal segments (for review see Bower, 2002). In this combined modeling and physiological study, we have evaluated the direct effects of paired pulse inputs on Purkinje cell spiking responses under different levels of background synaptic input. In the model, when low levels of background parallel fiber and molecular layer inhibitory input were present, the Purkinje cell spiking response to the second of two pulses was either suppressed or enhanced depending on the specific inter-pulse interval (ISI). These nonlinear effects were less apparent when background synaptic levels were elevated. This specific model prediction was tested physiologically by recording from individual Purkinje cells in the tactile regions of cerebellar folia Crus IIA, while applying paired upper lip tactile stimuli. Paired pulse responses were then evaluated in the presence and absence of a continually varying background air-puff stimulation intended to activate most or all tactile regions of the folium. The results support a new interpretation of the physiological and computational organization of cerebellar cortex.
This will be presented in poster form.
Submitted by: Migliore, Michele
Authors: M. Migliore, Dept of Neurobiology, Yale University School of Medicine, ML Hines, Dept of Neurobiology, Yale University School of Medicine, GM Shepherd, Dept of Neurobiology, Yale University School of Medicine
Title: The role of distal dendritic gap junctions in synchronization of mitral
cell axonal output
Abstract: One of the first and most important stages of odor processing occurs in the glomerular units of the olfactory bulb and most likely involves mitral cell synchronization. Using a detailed model constrained by a number of experimental findings, we show how the intercellular coupling mediated by intraglomerular gap junctions (GJs) in the tuft dendrites could play a major role in sychronization of mitral cell action potential output in spite of their distal dendritic location. The model suggests that the high input resistance and active properties of the fine tuft dendrites are instrumental in generating local spike synchronization and an efficient forward and backpropagation of action potentials between the tuft and the soma. The model also gives insight into the physiological significance of long primary dendrites in mitral cells, and provides evidence against the use of reduced single compartmental models to investigate network properties of cortical pyramidal neurons.
Submitted by: Huffman, Todd
Title: Automated Microtubule Segmentation and 3-Dimensional Reconstruction
from Sequential Transmission Electron Micrographs
Huffman TM, Ramos H, Shah K, Uchida M, Roberson RW.
Arizona State University, Tempe AZ
Abstract:
Reconstruction of cellular features in sequential images currently requires extensive manual labeling and expertise, slowing efforts to understand the role(s) of microtubules in secretory events critical for polarized cell growth, such as in tube-like fungal cells called hyphae. The purpose of the present work is to automate the process of image segmentation and 3-D reconstruction of microtubules in sequential transmission electron micrographs. Microtubules are segmented in the micrographs using data generated templates, and a series of filters are applied to reduce the numbers of false positives. Further false positive rejection and 3-D classification is accomplished through Hough transforms. The techniques used for microtubule reconstruction can be extended to other cellular features, and integrated with existing cellular reconstruction software.
Submitted by: Esquivel, Angelica
Authors:
Angelica V. Esquivel, Research Imaging Center, UTHSCSA, San Antonio, TX
Huo Lu, Research Imaging Center, UTHSCSA, San Antonio, TX
James M. Bower, Research Imaging Center, UTHSCSA, Cajal Neuroscience Research
Institute, UTSA, San Antonio, TX
Title: Differential distribution of inhibitory synapses on the Purkinje
cell dendrite
Abstract: Previous studies using serial electron microscopy have shown that synapses associated with parallel fibers, and those associated with the ascending segment synapses, terminate on different and non overlapping parts of the Purkinje cell dendrite (Gundappa-Sulur et al., 1999). Both our network and single cell modeling efforts suggest that molecular layer inhibition specifically interacts with the parallel fiber system to regulate the response properties of the Purkinje cell dendrite (Jaeger et al., 1997; Santamaria et al., 2002; Santamaria and Bower, 2005). In this study, we have returned to the use of serial electron micrographic 3-D reconstruction techniques to determine whether synapses associated with molecular layer inhibition are distributed uniformly across the Purkinje cell dendrite, or are restricted to the location of the parallel fibers. To date, 7 serial reconstructions of Purkinje cell dendrites (n=7) containing parallel fiber synapses have all been associated with inhibitory synaptic connections while none of the reconstructions made of ascending segment axons have included inhibitory synapses. If molecular layer inhibition is, in fact, restricted to parallel fiber synaptic locations, as our preliminary data suggests, it would both reinforce our model-based reinterpretation of the functional structure of cerebellar cortex (Bower, 2002) and also provide important new structural information for the next generation of realistic cerebellar network models.
Submitted by: Smith, Spencer LaVere
Authors: Spencer LaVere Smith, Dept. of Neurobiology, UCLA, Los Angeles, CA.
Title: The computational advantages of spontaneous firing in the cerebellar cortex.
Abstract: The spontaneous firing rates of Purkinje neurons and molecular layer interneurons are not simply background noise or a static tone of the cerebellar system. In fact, these firing rates are not fixed and can be changed based on previous synaptic activity. Moreover, spontaneous activity bestows distinct advantages for analog computations. A possible role for firing rate plasticity will be discussed with data from a model. Also, unpublished data will be presented that demonstrates how spontaneous firing interacts with synaptic plasticity.
Submitted by: Simoes de Souza, Fabio
Authors: Fabio M. Simoes de Souza, Psychobiology Sector,Universidade de Sao
Paulo at Ribeirao Preto, Brazil.
James M. Bower, Research Imaging Center, The University of Texas Health Science
Center at San Antonio,TX.
Title: What the Olfactory Stimuli Tells to the Piriform Cortex
Abstract: This research has been utilizing a biologically accurate biophysical model of the major structure of the olfactory cortex -the piriform cortex-to study the computation of the afferent olfactory input in this structure. The goal of this work is to explore the relationship among the synaptic organization, the electrical properties of the piriform cortex and the olfactory function. A special attention is focused in the origin and roles of the electric field potential oscillations and on the dynamic of the electrotonic properties of the neurons during the simulation of one SNIFF cycle -a theta burst. Also, it is explored how anterior and posterior areas of the model piriform cortex has been responding to inputs coming from the olfactory bulb. It has explored how the dynamical electrotonic properties of cells -electrotonic length and time constant- are related with the spatiotemporal integration of the olfactory stimuli and the processing of the olfactory stimuli.
Submitted by: Fransén, Erik
Authors: Erik Fransén, Department of Numerical analysis and computer
science, Royal Institute of Technology, SE-10044 Stockholm, Sweden
Alexander Kozlov, Department of Numerical analysis and computer science, Royal
Institute of Technology, SE-10044 Stockholm, Sweden
Yuecong Xie, Department of Numerical analysis and computer science, Royal
Institute of Technology, SE-10044 Stockholm, Sweden
Carl Christensen, Department of Numerical analysis and computer science, Royal
Institute of Technology, SE-10044 Stockholm, Sweden
Mikael Djurfeldt, Department of Numerical analysis and computer science, Royal
Institute of Technology, SE-10044 Stockholm, Sweden
Örjan Ekeberg, Department of Numerical analysis and computer science,
Royal Institute of Technology, SE-10044 Stockholm, Sweden
Anders Lansner, Department of Numerical analysis and computer science, Royal
Institute of Technology, SE-10044 Stockholm, Sweden
Title: Evaluation of model scalability in parallel neural simulators
Abstract:
A long standing belief in neuroscience has been that the brain and specifically the neocortex obtains its computational power by massive parallelism. Albeit conceptually appealing, this notion that effective processing requires large networks has not been possible to test in detailed simulations. In one project, we intend to study the generation of theta activity in the entorhinal-hippocampal system. Several simulation studies indicate that frequency and synchronization of the oscillation generated may depend on density of connectivity and/or geometry of connections. In a second project, we are studying how a model of early visual processing scales towards realistic sizes. To effectively evaluate the model, it must be scaled up to sizes where processing demands from the input given are sufficiently high, and where network size is made sufficiently large to process this information.
We have in preliminary studies tested two parallel simulators. One is a version of pGENESIS supporting MPI from University of Sunderland, UK. The other is Split, a software produced in our own laboratory. Both have been tested on an Itanium2 cluster. Tests include variable number of processors and scaling number of neurons/compartments or number of synapses. In these simulations, average spike frequency in the network is also varied. The aim is to identify main bottle-necks. For instance, we foresee the need to parallelize the construction/layout of synapses.
Submitted by: Parsons, Lawrence
Title: What can realistic modeling tell us about cognitive function?
Abstract:
Cognitive neuroimaging data has significantly affected
ideas about the
functional organization of human brain. However, very often the interpretation
of neuroimaging activity in a specific region is made without any consideration
of the nature of the microcircuitry in specific brain areas. I will discuss
examples of this tendency, and provide counterexamples in which researchers
have cast their interpretations accommodating known facts of about local
microcircuitry.
Submitted by: Aworinde, Buky
Authors:
Buky Aworinde, Research Imaging Center, The University of Texas Health Science Center at San Antonio
Huo Lu, Research Imaging Center, The University of Texas Health Science Center at San Antonio
James M. Bower, Research Imaging Center, The University of Texas Health Science Center at San Antonio
Title: A Structural Comparison Between Rat, Turtle, and Guinea Pig Purkinje
cells
Abstract: Arguably one of the most complex neurons in the brain, the last 10 years have seen a considerable and growing interest in constructing realistic compartmental models of the cerebellar Pukinje cell. These models have been used to make a number of unusual but important predictions regarding the active electrical properties of this cell's enormous dendrite and the effect of those intrinsic properties on its processing of a correspondingly large number of synaptic inputs. Most of this work to date, however, has been based on the same Guinea Pig Purkinje cell, originally reconstructed by Rapp, Yarom, and Segev in the 1980s. In this study, we have used light microscopic reconstruction procedures to compare the detailed morphology of this Guinea Pig neuron to both rat and turtle Purkinje cells. We have quantitatively examined the size and branching structure of each set of dendrites, looking for general principles of organization. Because the general structure of the cerebellar cortical circuitry in which the Purkinje cell is embedded is highly conserved throughout phylogeny, one would expect that there should also be conserved features in the detailed branching structure of Purkinje cells. These anatomical results will also provide the basis for future comparative single cell modeling efforts.
Submitted by: Megan Sullivan
Authors: Megan R. Sullivan(1), Axel Nimmerjahn(2), Dmitry V. Sarkisov(1), Fritjof Helmchen(2) and Samuel S.-H. Wang(1); (1)Princeton University, NJ (2)Max Planck Institute for Biomedical Research, Heidelberg, Germany
Title: In vivo calcium imaging of circuit activity in cerebellar cortex
Abstract: In vivo two-photon calcium imaging provides the opportunity to monitor activity in multiple components of neural circuitry at once. Here we report the use of bulk-loading of fluorescent calcium indicators to record from axons, dendrites, and neuronal cell bodies in cerebellar cortex in vivo. In cerebellar folium crus IIa of anesthetized rats we imaged the labeled molecular layer and identified all major cellular structures: Purkinje cells, interneurons, parallel fibers, and Bergmann glia. Using extracellular stimuli we evoked calcium transients corresponding to parallel fiber beam activity. This beam activity triggered prolonged calcium transients in interneurons, consistent with in vitro evidence for synaptic activation of NMDA receptors via glutamate spillover. We also observed spontaneous calcium transients in Purkinje cell dendrites that were identified as climbing fiber-evoked calcium spikes by their size, time course, and sensitivity to AMPA receptor antagonist. Two-photon calcium imaging of bulk-loaded cerebellar cortex is thus well suited to optically monitor synaptic processing in the intact cerebellum.
Submitted by: Michael E. Hasselmo, Department of Psychology Center for Memory and Brain, Boston University, 2 Cummington St., Boston, MA, 02215
Modeling the role of entorhinal cortex, prefrontal cortex and hippocampus in memory guided behavior
Simulation of experimental data on multiple levels of analysis helps to constrain the structure of computational models. This talk will describe coordinated modeling at different levels of detail which addresses the role of cortical structures in memory-guided behavior. Experimental data shows that lesions of the modulatory input from septum to hippocampus impairs memory guided behavior in tasks such as spatial alternation. Other data demonstrates that blockade of muscarinic cholinergic receptors or lesions of cortical cholinergic innervation impair performance in tasks including delayed matching and long term recognition. Detailed models on multiple levels link this behavioral data to physiological data from entorhinal cortex, hippocampus and prefrontal cortex. Modeling includes compartmental biophysical simulations of entorhinal cortical neurons (with Erik Fransen and Angel Alonso), which link data from slice preparations to unit recording and the effect of cholinergic lesions on delayed matching. These entorhinal mechanisms are included in larger scale integrate-and-fire simulations of the hippocampal formation (with Randal Koene and Robert Cannon), which guide the behavior of a virtual rat and link cellular changes during theta rhythm to unit recording data from behaving rats. This work is guided by models using simplified neurons which replicate theta phase precession and context-dependent responses during spatial alternation tasks. The action selection processes in behavioral tasks are simulated with network models of prefrontal cortex linking unit recording data to goal-directed decision making.
Fransen, E., Alonso, A.A. and Hasselmo, M.E. (2002) Simulations of the role of the muscarinic-activated calcium-sensitive non-specific cation current I(NCM) in entorhinal neuronal activity during delayed matching tasks. J. Neurosci. 22(3):1081-1097 . Hasselmo, M.E., Bodelon, C., Wyble, B.P. (2002) A proposed function of hippocampal theta rhythm: Separate phases of encoding and retrieval enhance reversal of prior learning. Neural Comp 14: 793-817.
Hasselmo, M.E. (2005) A model of prefrontal cortical mechanisms for goal directed behavior. Journal of Cognitive Neuroscience, in press.
Koene, R.A., Gorchetchnikov, A., Cannon, R.C. and Hasselmo M.E. (2003) Modeling goal-directed spatial navigation in the rat based on physiological data from the hippocampal formation. Neural Networks 16(5-6):577-84
Koene, R.A. and Hasselmo, M.E. (2005) An integrate and fire model of prefrontal cortex neuronal activity during performance of goal-directed decision making. Cerebral Cortex, in press.