Computational modelling identifies key determinants of subregion-specific dopamine dynamics in the striatum
- Department of Neuroscience, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark
- Department of Bioinformatics, H Lundbeck A/S, Denmark
- Behavioural and Clinical Neuroscience Institute, University of Cambridge, United Kingdom
- Department of Psychology, University of Cambridge, United Kingdom
- Department of Psychiatry, University of Cambridge, United Kingdom
Figures
Figure 1 with 4 supplements
Large-scale 3D model of the dorsal striatum.
(A) Self-enveloped simulation space of 100 µm3 with approximately 40,000 release sites from 150 neurons. Colours of individual release sites are not matched to neurons. (B) Simulation of a single release event after 5 ms and 10 ms. Colour-coded by DA concentration. (C) Comparison of analytical solution and simulation of diffusion after a single release event at three different time points. (D) Representative snapshot of steady state DA dynamics at 4 Hz tonic firing with parameters mirroring the dorsal striatum. (E) Cross-section of temporal dynamics for a midway section through the simulation space shown in (d). (F) Histogram of DA concentrations ([DA]) across the entire space in (d). (G) DA release during three burst activity scenarios for all release sites in a 10 x 10 × 10 µm cube (black boxes) and spill-over into the surrounding space. Burst simulated as an increase in firing rate on top of continued tonic firing of the surrounding space. Traces on top are average DA concentrations for the marked cubes, with bursts schematised by coloured lines below. The first image row is at the end of the burst, and the second row is 100 ms after. Scale bars for traces are 200 ms and 500 nM. Scale bar for the images is 20 µm. (H) Top: representative [DA] trace for a voxel with a release site during pacemaker and burst activity. Bottom: Occupancy of D1Rs and D2Rs for the same site. (I) Zoom on a DA burst as in (h), with [DA] in blue and D1R occupancy in teal with line style indicating different affinities. The shaded area indicates the period of bursting with 6 APs at 20 Hz. (J) Effect of complete pause in firing for 1 s on both average [DA] and D1R and D2R occupation.
-
Figure 1—source code 1
Source code used to generate data in A, D, E, and F.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code1-v1.zip
-
Figure 1—source code 2
Source code used to generate data in B and C.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code2-v1.zip
-
Figure 1—source code 3
Source code used to generate data in G.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code3-v1.zip
-
Figure 1—source code 4
Source code used to generate data in J.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code4-v1.zip
-
Figure 1—source code 5
Source code used to generate data in H.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code5-v1.zip
-
Figure 1—source code 6
Source code used to generate data in I.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-code6-v1.zip
Figure 1—figure supplement 1
Average concentration and receptor kinetics.
(A) Average DA concentration across a 100 x 100 × 100 µm volume of simulated DS at pacemaker activity. (B) Mean concentration change in response to a single event with 60% release probability (see supplementary notes on electrical stimulation) in all neurons at time zero. Blue trace is output directly from simulation, grey trace is the predicted FSCV measurement. (C) Modelling of predicted FSCV measurement and representative snapshot of simulation space just after a release event. (D) Peak DA concentration reached at different distances from the area with phasic activity for the three firing scenarios. (E) Volume of space exposed to greater than 100 nM DA after firing relative to volume of space where terminals actively burst. (F) Least-squares fit linear regression of the dLight and GRABDA sensors based on reported kinetics (Labouesse and Patriarchi, 2021) and newest experimental characterisation of D2R (Ågren et al., 2021). Shaded areas indicate 95% C.I. (G) Top: representative [DA] trace for a voxel with a release site during pacemaker and a triple burst scenario (300 ms long bursts of 3 APs at 10 Hz, three times in a row with 300 ms in between). Bottom: Occupancy of D1Rs and D2Rs for the same site. (H) Effect of complete pause in firing for 1 s on both average [DA] and D2R occupation at different affinities.
-
Figure 1—figure supplement 1—source code 1
Source code used to generate data in Figure 1—figure supplement 1.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-figsupp1-code1-v1.zip
Figure 1—figure supplement 2
Simulation size and granularity.
(A) Schematic of different simulation sizes. (B) Concentration percentiles at different simulation diameters. Results are not robust until a diameter close to 50 µm is reached (line runs behind 100 µm line). (C) Schematic of simulation granularity. (D) Effect of simulation voxel diameter on concentration percentiles. Virtually no difference in concentration profiles below the 99.9th percentile of [DA], with the 1.0 µm voxel size still following 0.1 and 0.5 µm well above that level. The inset shows 99.75th to 100th percentile with y-axis matching main y-axis. (E) Absolute difference in [DA] between simulations at 0.1 µm voxel diameter and 0.5, 1.0, and 2.0 µm. The inset shows 99th to 100th percentile with y-axis matching main y-axis. (F) Percentage difference in [DA] between simulations at 0.1 µm voxel diameter and 0.5, 1.0, and 2.0 µm.
-
Figure 1—figure supplement 2—source code 1
Source code used to generate data in Figure 1—figure supplement 2.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig1-figsupp2-code1-v1.zip
Figure 1—video 1
Representative video of steady-state dynamics at 4 Hz tonic firing in a 100 x 100 × 100 µm volume with parameters mirroring those experimentally observed in dorsal (left) and ventral (right) striatum.
Slowed down 10 x for illustrative purposes.
Figure 1—video 2
Representative video of a cross-section of a burst firing of 6 action potentials (AP) at 20 Hz the centre of the plane (circle) for the dorsal (upper row) and ventral (bottom row) striatum during steady state dynamics at 4 Hz tonic firing.
The left column shows DA concentration, middle D1 occupancy and right D2 occupancy. The scale bar in the upper left-hand field shows 10 µm and is kept identical for all views. Slowed down 15 x for illustrative purposes.
Figure 2 with 1 supplement
Regional differences in uptake greatly impact DA dynamics.
(A) Representative snapshots of steady state dynamics at 4 Hz tonic firing with parameters mirroring the dorsal (left) and ventral striatum (right). (B) Cross-section of temporal dynamics for data shown in a. The bottom row shows concentrations of the dashed lines in the top panels. (C) Normalised density of DA concentration of simulations in (a). Thick lines are for the entire space, and thin lines are across time for five randomly sampled locations. Dashed red line is for simulation of the ventral striatum with lowest reported innervation density in the literature. (D) Same data as in (c), but for concentration percentiles. Note that even the lowest percentiles of VS were above 10 nM in [DA]. (E) Convolved model response (Figure S1c) to mimic FSCV measurements mirroring the experimentally tested stimulation paradigm in May and Wightman, 1989 for the dorsal (left) and ventral striatum (right) (F) DA release during three burst activity scenarios for all release sites in a 10 x 10 × 10 µm cube (black boxes) and spill-over into the surrounding space. Burst simulated as an increase in firing rate on top of continued tonic firing of the surrounding space. Traces on top are average DA concentrations for the marked cubes, with bursts schematised by coloured lines below. The first image row is at the end of the burst, and the second row is another 100 ms after. Scale bars for traces are 200 ms and 500 nM. Scale bar for the images is 20 µm. (G) Top: representative [DA] trace 1 µm away from a release site during pacemaker and burst activity. Bottom: Occupancy of D1Rs and D2Rs for the same site. Occupancy data from the corresponding DS simulation on Figure 1k shown as a dotted line. (H) Peak occupancy at different distances from the area bursting, normalised to maximal and minimum occupancy.
-
Figure 2—source code 1
Source code used to generate data in A-F.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig2-code1-v1.zip
-
Figure 2—source code 2
Source code used to generate data in G and H.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig2-code2-v1.zip
Figure 2—figure supplement 1
Histochemical gradient of DAT and VMAT2 fluorescence.
(A) Representative image of the mouse striatal slices analysed in B. Dashed white line indicates the quantified dorsoventral gradient (length 2 mm). (B) Relative intensity of the DAT and VMAT2 immunosignal in the dorsoventral axis of striatal mouse brain slices from Sørensen et al., 2021. All slices show a drop at the anterior commissure (AC). Shaded areas around lines denote S.E.M. (C) Mean relative intensity of the DAT and VMAT2 signal before and after AC. Two-sided t-test, VS-DAT:VMAT2, p=0.012(*), n=4 mice; one-sided t-tests, DAT-DS:VS, p=0.0021(**), VMAT2-DS:VS, p=0.0086(**), n=4 mice. (D) Peak DA concentration reached at different distances from area with phasic activity for the three firing scenarios in VS. (E) Volume of space in VS exposed to greater than 100 nM after firing relative to volume of space where terminals actively burst. (F) Effect of complete pause in firing in VS for 1 s on both average [DA] and D1R and D2R occupation.
Figure 3 with 2 supplements
Sensitivity of the model to parameter changes.
(A) Schematic of the fraction of active release sites. Black dots are inactive sites, and green dots indicate actively releasing sites. (B) Effect of changing fraction of active release sites on DA concentrations. Blue line, DS peak DA concentration (99.5th percentile); Red line, VS peak DA concentration (99.5th percentile); Dotted blue line, DS tonic DA concentration (50th percentile); Dotted red line, VS tonic DA concentration (50th percentile). (C) Ratio between peak (99.5th percentile) and tonic (50th percentile) concentrations across fractions of active release sites in the DS (blue line) and VS (red line) as a measure of DA signal focality. (D) Schematic of changing quantal size (Q). (E) Effect of changing quantal size on tonic and peak DA concentrations in DS (blue lines) and VS (red lines). (F) Ratio between peak and tonic concentrations across various quantal sizes in in DS (blue line) and VS (red line). (G) Relative difference between the DS and VS for peak (black line) and tonic DA (dotted line) at different quantal sizes. (H) Schematic of changing DAT Km. (i) Effect of changing DAT Km on DA concentrations in DS (blue lines) and VS (red lines). (J) Schematic of changing DAT Vmax. (K) Effect of changing DAT Vmax on DA concentrations. Shaded areas are median Vmax of the two regions (DS and VS) as found in the literature shown in Appendix 2—table 2 ± 50%. (L) Effect of changing DAT Vmax, with tonic (50th percentile) and peak (99.5th percentile) DA concentrations normalised to their value at 2 µm s–1 (median value for VS). The shaded area indicates median Vmax for VS found in the literature shown in Appendix 2—table 2 ± 50%.
-
Figure 3—source code 1
Source code used to generate data in B, C, E-G.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig3-code1-v1.zip
-
Figure 3—source code 2
Source code used to generate data in I.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig3-code2-v1.zip
-
Figure 3—source code 3
Source code used to generate data in K and L.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig3-code3-v1.zip
Figure 3—figure supplement 1
Model parameter testing.
(A) Schematic of our definitions of tonic (50th percentile/median, dashed lines) and peak (99.5th percentile, solid line) DA for both the dorsal and ventral striatum. (B) Effect of changing release probability (R%) on DA concentrations. (C) Relative difference between the ventral and dorsal striatum at different percentiles for different release probabilities. (D) Ratio between 99.5th and 50th percentiles as a measure of focality for both regions. As R% increases, the concentrations become more homogeneous. (E) Effect of changing firing rate on DA concentrations. (F) Relative difference between the ventral and dorsal striatum at different percentiles for different firing rates. (G) Ratio between 99.5th and 50th percentiles for both regions. As the firing rate increases, the concentrations become more homogeneous.
-
Figure 3—figure supplement 1—source code 1
Source code used to generate data in Figure 3—figure supplement 1.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig3-figsupp1-code1-v1.zip
Figure 3—figure supplement 2
Fold change during inhibition, Vmax-sensitivity at different release parameters and release-uptake balance.
(A) Fold change over baseline (Km of 210 nM) for mean DA concentration in the dorsal (DS) and ventral striatum (VS) with changing DAT Km. (B) Relative difference between the dorsal and ventral striatum for both phasic and tonic DA at different Km values. (C) Effect of changing DAT Vmax on DA concentrations for three different quantal sizes (Q). [DA] normalised to highest values within each Q. Shaded area indicates median Vmax for DS and VS as found in the literature shown in Appendix 2—table 2 with ±50%. (D) Effect of changing DAT Vmax on DA concentrations for three different release probabilities (R%). [DA] normalised to highest values within each R%. Shaded area indicates median Vmax for DS and VS as found in the literature shown in Appendix 2—table 2 with ±50%. (E) Least-square fit linear regression between release rate and autocorrelation decay rate (τ) (Ejdrup et al., 2023). Shaded area highlights 95% C.I. (F) Partial regression plot error of the regression in (f) and error between DA response to amphetamine as measured by microdialysis and the release rate from Ejdrup et al., 2023 to show that the less release and uptake correlate, the less release rate can explain the microdialysis response, suggesting release and uptake are partially independent of each other. Shaded area highlights 95% C.I.
-
Figure 3—figure supplement 2—source code 1
Source code used to generate data in Figure 3—figure supplement 2A-D.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig3-figsupp2-code1-v1.zip
Figure 4
DAT nanoclustering reduces uptake and shows regional variation.
(A) Schematic of dense DA cluster. White dots represent individual DAT molecules, and colour gradient the surrounding DA concentration. (B) Effective transport rate dependent on local concentration. (C) Top view of unfolded DA varicosity. Black shapes denote clusters of DAT. A dashed white line indicates placement of cross-section shown in (d). (D) Cross-section showing DA concentration in space surrounding varicosity unfolded in c. The grey line at the bottom is the surface of the varicosity. Colour-coded for DA concentration. (E) Top view from (c), but colour coded for DA concentration immediately above membrane surface at different DAT cluster sizes. (F) Changes in [DA] from 15 nM unclustered (Un.) steady state with constant release after changing to four different cluster size scenarios (ø=diameter). (G) Clearance of 100 nM [DA] for different DAT cluster sizes. (H) Difference between DA concentration at the centre of clusters (or general surface of varicosity for unclustered) and mean concentration of the full simulation space (I) Concentrations across a cross section of a surface with 80 nm diameter clusters. Shaded areas highlight cluster locations. (J) Location of images of the dorsal (DS) and ventral striatum (VS) in striatal slices from mice as imaged in Sørensen et al., 2021 with direct stochastic optical reconstruction microscopy (dSTORM). (K) Two representative DA varicosities from DS and VS with VMAT2 in white and DAT in magenta. Images are 1.5x2 µm (scale bar 0.5 µm). (L) Individual DAT localisations (locs.) from images in (k) coloured by clustering. Black indicates localisation identified as clustered based on DBSCAN with parameters 80 nm diameter and 40 localisations. Grey indicates unclustered localisations. (M) Quantification of clustering across all images in (j) with parameters in (l). Welch’s two-sample t-test, p=0.012(*), n=12 (DS) and 13 (VS). (N) Absolute difference in percentage of clustering as assessed with DBSCAN across a range of parameters. VS has a higher propensity to cluster across cluster sizes typically reported for DAT clusters.
-
Figure 4—source code 1
Source code for simulation.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig4-code1-v1.zip
-
Figure 4—source code 2
Source code used to generate data in A-E and G-I.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig4-code2-v1.zip
-
Figure 4—source code 3
Source code used to generate data in F.
- https://cdn.elifesciences.org/articles/105214/elife-105214-fig4-code3-v1.zip
Tables
Table 1
List of variables used in the simulation of the dorsal striatum.
| Variable | Abbreviation | Value | Reference |
|---|---|---|---|
| Firing rate | 4 Hz | Paladini et al., 2003 | |
| Release probability | 6% | Dreyer et al., 2010 | |
| DA molecules per vesicle | 3000 | Klaus et al., 2019 | |
| Diffusion coefficient | 763 µm2 s-1 | Nicholson, 1995 | |
| Tortuosity | 1.54 | Rice and Nicholson, 1991 | |
| Vmax | 6.0 µm s-1 | See Appendix 2—table 2 | |
| Km | 210 nM | Hovde et al., 2019 | |
| Active terminal density | - | 0.04 µm-3 | Liu et al., 2021 |
| Extracellular volume fraction | 0.21 | Rice and Nicholson, 1991 | |
| Number of neurons in simulation space | - | 150 | Matsuda et al., 2009 |
Key resources table
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
|---|---|---|---|---|
| Software, algorithm | Python | https://www.python.org | v3.9.7 | |
| Software, algorithm | Spyder IDE | https://www.spyder-ide.org | v5.1.5 | |
| Software, algorithm | Model code | https://github.com/GetherLab/striatal-dopamine-modelling | Developed for this work | |
| Strain, strain background (Mus musculus) | C57Bl/6 J | Details provided in Sørensen et al., 2021 | Data used are from Sørensen et al., 2021 | |
| Antibody | anti-DAT Nt (rat monoclonal) | Sigma-Aldrich | MAB369 RRID:AB_2190413 | IF (1:200) Sørensen et al., 2021 |
| Antibody | Anti-VMAT2 (rabbit polyclonal) | Kind gift from Dr Gary W. Miller, Columbia University Sørensen et al., 2021 | IF (1:4000) Sørensen et al., 2021 |
Appendix 2—table 1
Overview of reports on dopaminergic density and release in the striatum.
Only studies that assessed both regions in rodents are included. Studies that stimulate directly in the striatum are omitted due to the large activation of nicotinic receptors on DA terminals (1, 2). A.U.=arbitrary units, DS = dorsal striatum, VS=ventral striatum, TH = tyrosine hydroxylase.
| Measure | Ratio | DS | VS | Units | Species | Source |
|---|---|---|---|---|---|---|
| TH expression density | 100 % | ~90 | ~90 | A.U. | Mouse | Alberquilla et al., 2020 |
| TH immunoreactivity | 95 % | ~68 | ~64 | A.U. | Mouse | Kuroda et al., 2010 |
| TH protein content | 150 % | 0.07 | 0.11 | ng TH/µg prot. | Mouse | Salvatore et al., 2016 |
| DA content | 90 % | ~155 | ~140 | ng DA/mg prot. | Mouse | Salvatore et al., 2016 |
| TH immunoreactivity | 75 % | 2.8 | 2.1 | A.U. | Rat | Huang et al., 2019 |
| TH protein content | 66 % | 0.36 | 0.24 | ng TH/µg prot. | Mouse | Salvatore et al., 2005 |
| FSCV - [DA]p | 91 % | 57 | 52 | nM | Rat | May and Wightman, 1989 |
| FSCV - [DA]p | 76 % | 89.3 | 67.5 | nM | Rat | Garris and Wightman, 1994 |
| Median | 90 % | - | - | - | - | - |
Appendix 2—table 2
Overview of reported Vmax values for DA uptake in the striatum.
Only studies that assessed both regions in rodents are included. DS = dorsal striatum, VS=ventral striatum, FSCV = fast scan cyclic voltammetry.
| Method | VS/DS Ratio | DS (uM/s) | VS (uM/s) | Species | Source |
|---|---|---|---|---|---|
| FSCV | 29 % | 7.0 | 2.0 | Mouse | Calipari et al., 2012 |
| FSCV | 28 % | 6.0 | 1.7 | Rat | Calipari et al., 2012 |
| FSCV | 47 % | 3.0 | 1.4 | Rat | May and Wightman, 1989 |
| FSCV | 31 % | 6.5 | 2.0 | Mouse | Siciliano et al., 2014 |
| FSCV | 44 % | 5.0 | 2.2 | Rat | Ferris et al., 2014 |
| Median | 31 % | 6.0 | 2.0 | - | - |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Computational modelling identifies key determinants of subregion-specific dopamine dynamics in the striatum
eLife 14:RP105214.
https://doi.org/10.7554/eLife.105214.3
