C&SP19: Mapping the Supramolecular Organization of Dendritic Spines to Model the Regulation of Synaptic Signal Transfer

A. Collaborating Investigators: Mark H. Ellisman,¹ Andreas Herz,²Terry Sejnowski,³ Tom Bartol³

B. Institutions: ¹University of California, San Diego, ²Ludwig-Maximilians U, Munich, and ³Salk

C. Funding Status of Project: NIDA R01-DA038896-02 ‘Deciphering the dynamical multi-scale structure-function of dendritic spines (Ellisman) 7/1/14 – 6/30/19

D. Biomedical Research Problem

Synapses are the sites of communication between neurons and most excitatory synapses occur onto specialized structures called dendritic spines.³⁴ Generally, spines consist of a spine head and a thin spine neck. The role of spines in influencing electrical and biochemical flow at the synapse is still debated; however, spines can assume a variety of shapes and sizes, and it has recently been shown that spine size is highly correlated with synaptic activation history and network function.¹⁰ Spines are filled with a high concentration of filamentous actin, but the density of actin can vary within and across brain regions.³⁵ The role of the actin network in spine function has not been well studied. The highly branched filamentous actin carries electrostatic charges, which causes ions to condense in its vicinity. Within the condensation profile, known as the electrical double-layer, cations, the main charge carriers for depolarizing phenomena, should propagate with great ease due to the high local concentration. Thus, a significant fraction of the current within the volume of the spines is constrained to flow in nano-slits along charged structures. The multiscale modeling of this process is to be explored in this collaboration.

The Ellisman lab is generating high-resolution (~5 nm) serial EM tomographic volumes of entire spine heads from cerebellum, hippocampus, and striatum. The tissue is prepared using high pressure freezing / freeze substitution to optimally preserve actin network structure and membrane profiles.³⁶ Multi-tilt acquisition and direct electron detection further improve image quality for subsequent feature extraction. Using these datasets, the Herz lab is building spatially accurate 3D models of ionic current propagation in topologically consistent segmentations of the filamentous actin network (Fig VIII.2). These models will be used to perform simulations of reaction-diffusion dynamics using MCell, and multiscale simulations of electro-diffusion dynamics, while enforcing consistent biophysical boundary conditions and accurate ion-channel kinetics.

E. Methods and Procedures

Aim 1: Create a multiscale model. Models will be generated using CellBlender that will include all relevant actors (spine apparatus, postsynaptic density) necessary to computationally reconstitute synapses and explore dynamics of Ca²⁺ and other molecules within spines and spine necks with MCell.

Aim 2: Determine role of actin network in spines of various sizes/strengths. Since spine size is strongly correlated with synaptic strength, we will study differences in actin morphology and its influence on diffusion in spines of various sizes. We will choose spines in the CA1 stratum radiatum that are representative of large (>0.1um3), average (0.02-0.04 um3), and small (<0.01um3) spines. MCell will be used to model diffusion in the CellBlender models created in Aim 1.

Aim 3: Compare role of actin in spines from different brain regions. Dendritic spines of the cerebellum have been found to contain considerably higher concentrations of actin than spines in CA1 stratum radiatum. Spines in the striatum appear to contain intermediate amounts of actin. We will model actin in spines from these three regions and use MCell to determine how the cytoskeleton affects diffusion in spines representing realistic variations in actin concentration.