DBP3: Multiscale Modeling of DAT Function in Dopamine (DA) Neurons

DBP3: Multiscale Modeling of DAT Function in Dopamine (DA) Neurons

Back to table

A. Collaborating Investigators: Alexander Sorkin,1 Simon Watkins,1 Ethan R. Block,2 GonzaloTorres,3 Ivet Bahar,1 James Faeder,1 Robert Murphy,4 Terry Sejnowski,5 Tom Bartol4

B. Institutions: 1Pitt, 2Chatham U, 3U of Florida, 4Carnegie Mellon U, 5Salk

C. Funding Status: R01DA14204(Sorkin) 5/1/01-4/7/17; R01DA038598 (Torres); 9/15/14 - 6/30/19

 

D. Driving relationship between DBP3 and TR&Ds: Two important advances in the past funding term now open the way to multiscale structure-based, spatiotemporally realistic simulations of DA reuptake by DA neurons. The first is the structural resolution of Drosophila melanogaster DA transporter, dDAT,46 and its interactions with antidepressants47 and psychostimulants,48 which permitted us (TR&D1) to elucidate the structural dynamics of human DAT (hDAT),49 as well as the molecular mechanism of its interactions with cocaine and AMPH.50 The second breakthrough is the quantitative characterization of the spatiotemporal distribution of DAT in the striatum by the Sorkin lab. This was enabled by novel immunofluorescence labeling methods with transgenic knock-in mice that express hemagglutinin (HA) epitope-tagged DATs.51 The data was essential to building a first MCell model (Fig VII.3) which indicated that the geometry near the active zone and the spatial distribution of DATs significantly affect the effectiveness of DA reuptake. Data obtained by others52,53 also suggest that in the striatum firing patterns alter the efficiency of DA receptor activation. Further data will be supplied by Pitt and U of Florida collaborators for the effect of psychostimulants/drugs on DA transport. Our goal is to construct the first spatiotemporally realistic model of DA reuptake and examine its dynamics in response to different types of firing patterns and in the presence of different drugs or psychostimulants.

Picture3

Fig VII.3. Modeling and simulation of dopamine (DA) uptake by DA neurons.

A. Distribution of HA-DAT in mouse brain striatum. Slices were labeled with DAT-Nt antibodies, and conjugated with Cy3 (HA, red). The fluorescence image consists here of 18 slices with 0.4 μm. B. 3D reconstruction of A, visualized with Blender. Different colors shows different axonal segments. C. Isometric view of the simulation box. The remaining portions are filled with cells that do not express DAT. D. Snapshots from MCell simulations. Color code: red, DA; white, outwards facing DAT; green, inward-facing DAT. E. Distribution of average extracellular (EC) DA level from runs with different settings: blue, well-mixed model; green, realistic geometry with uniform distribution of DAT; and black, realistic geometry with heterogeneous distribution of DAT derived from fluorescence images. Results are shown for 2 different firing patterns, tonic and phasic, both with same average frequency. F. Time evolution of EC DA concentration for tonic (top) and phasic (bottom) firing, for different numbers of active axonal termini (right) and their histograms (left).

E. Innovation: This will be the first development of a spatially realistic model for DA axonal arbors and DAT localization near active zones, which will provide a framework for simulating DA reuptake using as input time-resolved super-resolution image data with resolution of 120 nm (lateral) x 240 nm (axial). This, combined with advances in BTRR technology will permit a first molecular-to-cellular structure-based study of DA signaling and its modulation by psychostimulants at the time scale of milliseconds to minutes.

F. Methods and Procedure: Aim 1. In silico reconstruction of DA neurons morphology and molecular distributions. The Sorkin lab will provide stacks of images obtained with the technology described recently,54 which includes immuno-histochemical analysis of brain slices, confocal fluorescence microscopy, EM and immunogold-silver labeling of DAT with transgenic HA knock-in mouse brain. These will drive the development and implementation of (i) high-quality image registration and segmentation methods and (ii) a geometry preparation pipeline for MCell/CellBlender, by TR&D2 and TR&D4. Thus far, we have used a semi-automated 3D reconstruction method55 to build a simulation box of 772 m3 containing 16 axon terminals (Fig VII.3 A-C). The spacing between DA neurons (void fraction: 0.298) and the spatial distribution of DATs (220,000 of them) were deduced from EM images. Using newly developed technologies, TR&D2 & 4 will reconstruct with high- fidelity the full data (382 micrographs for 9000–17,000 m2 striatum tissue) with/without AMPH.56

Aim 2. Development of a kinetic model for DA/psychostimulant binding/unbinding and DAT alternating access mechanics. This aim will drive the development of the new hybrid methodology that combines coMD28,57 and WE58,59 for generating molecular data on the populations of substates and their rates of transitions (TR&D1 aims 1 and 3) . These will serve as input for MCell. They will be benchmarked and iteratively refined using electrophysiology and kinetics data from both DBP1 and DBP3 experiments and the literature, in addition to earlier applicable results from expensive full-atomic studies.50,60 We already studied the structural dynamics of hDAT61 and its structural homologs,62-67 and the effects of psychostimulants on DAT dynamics.50 These will be used in the hybrid coMD-WE methodology (TR&D1 aim 3) that will be iteratively refined using DAT transport cycle data as benchmark. We will perform free energy perturbation (FEP) and adaptive biasing force (ABF) calculations, and assess ligand selectivity (see e.g.68), and current-voltage relations.

Aim 3. MCell simulations of DA reuptake events under different conditions, and trajectory analysis. This will require enhancement of the simulation and interfacing capabilities of MCell (TR&D2 aims 1 and 3) in collaboration with TR&D3 (e.g. spatially structured, multi-state, multi-component molecules). Vesicular DA discharge will be modeled as quantal release from active zones. A distance threshold of 20 Å will be adopted for DA-DAT recognition, consistent that observed in our simulations for DAT or LeuT.50,69-72 Our preliminary runs show that (i) DA molecules diffuse away from the active zone within ~5 ms (Fig VII.3D), (ii) axonal geometry and heterogeneous distribution of DATs significantly affect the distribution of DA levels in the striatal EC region (Fig VII.3E), underscoring the importance of doing MCell simulations; and (iii) the propensity to reach the EC levels (of 10 nM) required for D2R activation is much higher in the presence of phasic release of DA, as opposed to tonic (Fig VII.3F). The variance is greatly reduced when a fraction of striatal axons are rendered inactive (e.g. PD model). We will establish the conditions under which normal DA transmission is ensured, and study how drugs alter the dynamics. Image and electrophysiology data from our collaborators73-75 will serve as test bed for validating our results. Cocaine blocks the DA binding site,50,76-78 resulting in increased EC [DA]; we will examine under which conditions these levels approach that (~1mM)79 required for activating low affinity DA receptors. Data from fast scan cyclic voltammetry experiments in the presence of AMPH and L-DOPA80 will further help test the predictions. Data from DBP1 will further help elucidate AMPH-stimulated events.

Pitfalls and alternative strategies. The time scale of occupancy of binding site on DAT by cocaine (hours)81 is longer compared to the timescales (seconds to minutes) of DA signaling. We may need long simulations in order to obtain statistically significant data. Alternatively, simulations with mixed (explicit and implicit) interactions will be considered. Similar strategies may be required for PD models where the DA neuronal activity is slowed down,82 or for long-time exposures (>15 min) to AMPH or L-DOPA.

 

Copyright © 2020 National Center for Multiscale Modeling of Biological Systems. All Rights Reserved.