Elsevier

Neural Networks

Volume 22, Issue 8, October 2009, Pages 1174-1188
Neural Networks

2009 Special Issue
Dopamine-modulated dynamic cell assemblies generated by the GABAergic striatal microcircuit

https://doi.org/10.1016/j.neunet.2009.07.018Get rights and content

Abstract

The striatum, the principal input structure of the basal ganglia, is crucial to both motor control and learning. It receives convergent input from all over the neocortex, hippocampal formation, amygdala and thalamus, and is the primary recipient of dopamine in the brain. Within the striatum is a GABAergic microcircuit that acts upon these inputs, formed by the dominant medium-spiny projection neurons (MSNs) and fast-spiking interneurons (FSIs). There has been little progress in understanding the computations it performs, hampered by the non-laminar structure that prevents identification of a repeating canonical microcircuit. We here begin the identification of potential dynamically-defined computational elements within the striatum. We construct a new three-dimensional model of the striatal microcircuit’s connectivity, and instantiate this with our dopamine-modulated neuron models of the MSNs and FSIs. A new model of gap junctions between the FSIs is introduced and tuned to experimental data. We introduce a novel multiple spike-train analysis method, and apply this to the outputs of the model to find groups of synchronised neurons at multiple time-scales. We find that, with realistic in vivo background input, small assemblies of synchronised MSNs spontaneously appear, consistent with experimental observations, and that the number of assemblies and the time-scale of synchronisation is strongly dependent on the simulated concentration of dopamine. We also show that feed-forward inhibition from the FSIs counter-intuitively increases the firing rate of the MSNs. Such small cell assemblies forming spontaneously only in the absence of dopamine may contribute to motor control problems seen in humans and animals following a loss of dopamine cells.

Introduction

The striatum is a large subcortical nucleus that forms the principal input structure of the basal ganglia. Diseases that directly affect the striatum or its primary afferents–such as Huntington’s or Parkinson’s disease–lead to profound deficits in motor control. In particular, loss of dopamine cells in Parkinson’s disease and its animal models leads to motor symptoms of rigidity, akinesia, and tremor (Ferro et al., 2005, Kirik et al., 1998, Schwarting and Huston, 1996), and the striatum is the main locus of dopamine’s action, containing the highest density of dopamine receptors in the vertebrate brain (Dawson et al., 1986, Hurd et al., 2001, Richtand et al., 1995). Moreover, an intact dopamine system also seems to be critical for many forms of learning (Ferro et al., 2005, Whishaw and Dunnett, 1985), consistent with reported correlations between dopamine cell firing and the prediction error required by reinforcement learning theories (Redgrave and Gurney, 2006, Schultz, 2007). An intact striatum is similarly required for successful acquisition of many instrumental conditioning tasks (Yin & Knowlton, 2006). An understanding of the striatum’s computational operation would thus shed light on a fundamental contributor to both motor control and learning.

Within the striatum lies a complex, predominantly GABAergic, microcircuit (Bolam et al., 2006). Medium spiny projection neurons (MSNs) are the only output neurons and comprise up to 95% of the cell population in rats, with GABAergic and cholinergic interneurons forming most of the remaining cell population. Despite their comparatively small number, the GABAergic fast-spiking interneurons (FSIs), in particular, exert a very strong influence on the MSNs (Koos & Tepper, 1999), receive input from similar sources, and are interconnected by both chemical synapses and gap junctions. Dopamine has multiple effects on these neuron types, via multiple receptor types: indeed, the exact effects of dopamine receptor activation on the MSN have been much debated (Surmeier, Ding, Day, Wang, & Shen, 2007). Seemingly ideal for underpinning its multiple functional roles, the striatum receives massive convergent input from the neocortex, thalamus, hippocampal formation, and amygdala (Glynn and Ahmad, 2002, Groenewegen et al., 1999, McGeorge and Faull, 1989, Smith et al., 2004), and dopamine modulates the striatal neurons’ responses to them.

Despite, or perhaps due to, this complexity of structure and input, there are few well-quantified theories of the striatum’s computational role. Many theories of striatal-specific or global basal ganglia function draw explicit attention to the role of the inhibitory local MSN collaterals as a substrate for competitive dynamics (e.g. Beiser and Houk, 1998, Frank, 2005, Pennartz et al., 1994, Wickens et al., 1991), whether that competition be labelled ‘decision making’, ‘motor program selection’ or ‘pattern classification’. Wickens and colleagues’ domain hypothesis is the most developed, and proposes that the basic computational element of the striatum is the set–or “domain”–of all MSNs that are mutual inhibitory (see e.g. Alexander and Wickens, 1993, Wickens et al., 1991, Wickens et al., 1995). In simulation, they have shown that winner-takes-all like competition occurs within a single domain, while winners-share-all dynamics (multiple active neurons) occur in networks composed of multiple overlapping domains (Alexander and Wickens, 1993, Wickens et al., 1991). Similar results have been obtained in analytical studies of general mutually inhibitory neural networks (Fukai & Tanaka, 1997).

All such theories of competitive dynamics are faced by the problems that the inhibition provided by the local MSN collaterals is weak (Czubayko and Plenz, 2002, Jaeger et al., 1994, Koos et al., 2004, Taverna et al., 2004, Tunstall et al., 2002), so that a single MSN is only contacted by between 12%–18% of MSNs in its dendritic field (Tepper, Koos, & Wilson, 2004), and that mutual inhibition is the exception rather than rule (Tepper et al., 2004, Tunstall et al., 2002).

Some theories do predict such weak connections. Bar-Gad, Morris, and Bergman (2003) have proposed that the striatum compresses information relayed to it from the cortex, transmitting back the compressed version via the basal ganglia output nuclei. They noted that the two layer network formed by the striatum and the output nuclei can be mapped to standard neural network implementations of principal components analysis, and that these require weak correlation in a layer corresponding to the striatum. While an interesting hypothesis, this mapping does not account for the microcircuit of the striatum, or the effects of the numerous neuromodulators within it. Other models of the whole basal ganglia circuit do not rely on the local collaterals within the striatum for their computations, rather proposing that the striatum is both an integrator of diverse cortical information and a filter on weak cortical inputs, as the first stage of an input selection mechanism implemented by the whole basal ganglia (as opposed to just the striatum), (Gurney et al., 2001, Humphries et al., 2006) — but these models also do not account for the striatal microcircuit.

Our aim is to find out what computations can be supported by the intrinsic circuitry of the striatum, what–if any–“basic computational elements” exist, and develop computational theories of function on this basis. In particular, we wish to understand the role of the dominant GABAergic circuits of the striatum: the rare, but powerful, FSIs, and the weak, asymmetrical, but comparatively plentiful MSN local collaterals. Understanding the contribution of all the striatum’s elements ideally requires large-scale models (Djurfeldt, Ekeberg, & Lansner, 2008) that replicate the neuron types, numbers, and connectivity at a one-to-one scale. Such models can give deep insight into the role of each neuron class in local circuit dynamics.

The purpose of this paper is twofold. First, we draw together, for the first time, a series of techniques we have developed for leveraging anatomical and physiological constraint data, some of which promise general applicability (beyond the striatum) in microcircuit construction: (1) a powerful computational neuroanatomy method for extracting the best connectivity statistics from impoverished data; (2) the development of reduced models for dopamine modulation of striatal neurons, which replicate the output of detailed compartmental models; and (3) a rigorous method for spike generation which allows good approximation to cortical input. We add to these here by introducing: (1) a gap junction model tunable to known membrane properties; (2) a principled method for parameterising the spike generation tool based on anatomical and physiological data; and (3) a novel method for detecting patterns in multi-unit activity at multiple time-scales, with general applicability to simulation or experimental data.

Second, we begin the identification of computational elements within the striatum, and examine how these might support hypotheses for competitive dynamics underpinned by the GABAergic neurons of the striatum. Specifically, we construct a three-dimensional model of the striatal microcircuit’s connectivity, and instantiate this with our dopamine-modulated neuron models of the MSNs and FSIs. We apply our multiple spike-train analysis to the outputs of this model to find groups of synchronised neurons at multiple time-scales. We then show that, with realistic in vivo background input, small assemblies of synchronised MSNs spontaneously appear, consistent with experimental observations (Carrillo-Reid et al., 2008), and that the number of assemblies and the time-scale of synchronisation is strongly dependent on the simulated concentration of dopamine.

Section snippets

Creating the striatal microcircuit

Building large-scale models at up to 1:1 scale, neuron for neuron, is an ambitious aim. In particular, as recognised by the Blue Brain Project (Markram, 2006), these models are severely limited by the need for accurate connectivity. There is a wealth of studies showing how the structure of a network is a strong determinant of its dynamics (see e.g. Galan, 2008, Kwok et al., 2007, Nishikawa et al., 2003), and that the typical fall-back of completely regular or random networks give false

Model neurons

The model striatal network forms the basis for our study of its dynamics. If we are to build at such scales, we require individual neuron models that are simple enough to be computationally tractable, but sufficiently complex to capture key membrane properties that contribute to the characteristic behaviour of a neuron species. Our neuron model of choice is the recent canonical spiking model of Izhikevich (2007), which has been employed in some notably large-scale models (Izhikevich, Gally, &

Detecting groups of synchronised cells in multi-unit data

We sought to identify potential candidates for the basic computational elements of the striatum from the dynamics of our large-scale models under background input. For our present purposes, we wanted to find groups of co-active or mutually antagonistic MSNs that could form the basis for competitive dynamics within the striatum. In addition, we studied this input regime to see if the reported striatal cell clusters, spontaneously formed in vitro (Carrillo-Reid et al., 2008), could be identified

Results

We now have the necessary tools–models of anatomy, neurons, and input, and suitable analysis methods–to begin addressing the problem of identifying the computational elements of the striatum. We use in this paper a small striatal region of 250 μm3, which gives us 1400 neurons, 1359 MSNs and 41 FSIs. This made a thorough analysis of both the network itself and all its outputs computationally tractable, and we keep this size throughout for consistency. We randomly split the MSNs into two equal

Discussion

To study the striatal GABAergic microcircuit, we have brought together for the first time a detailed model of striatal anatomy, models of its main neurons, their modulation by dopamine, and connection by gap junctions, and models of cortical input. Further, we proposed a new algorithm for finding structure in the multiple spike train data-sets resulting from the striatum model. We used this method to gain a unique insight into the computations of the microcircuit, and identify potential “basic

Acknowledgements

This work was funded by the EU Framework 6 Project IST-027819-IP, EPSRC Research Grant EP/C516303/1, and the EPSRC “CARMEN” e-Science Pilot Project. We thank Nathan Lepora for discussions and prior contributions to the single neuron models.

References (92)

  • D. Kirik et al.

    Characterization of behavioral and neurodegenerative changes following partial lesions of the nigrostriatal dopamine system induced by intrastriatal 6-hydroxydopamine in the rat

    Experimental Neurology

    (1998)
  • H. Kita et al.

    Parvalbumin-immunoreactive neurons in the rat neostriatum: A light and electron microscopic study

    Brain Research

    (1990)
  • A.J. McGeorge et al.

    The organization of the projection from the cerebral cortex to the striatum in the rat

    Neuroscience

    (1989)
  • E.K. Miller et al.

    All my circuits: using multiple electrodes to understand functioning neural networks

    Neuron

    (2008)
  • C.M. Pennartz et al.

    The nucleus accumbens as a complex of functionally distinct neuronal ensembles: An integration of behavioural, electrophysiological and anatomical data

    Progress in Neurobiology

    (1994)
  • D. Plenz

    When inhibition goes incognito: feedback interaction between spiny projection neurons in striatal function

    Trends in Neurosciences

    (2003)
  • N.M. Richtand et al.

    Regional quantification of D1, D2, and D3 dopamine receptor mRNA in rat brain using a ribonuclease protection assay

    Molecular Brain Research

    (1995)
  • W. Schultz

    Behavioral dopamine signals

    Trends in Neurosciences

    (2007)
  • R.K. Schwarting et al.

    The unilateral 6-hydroxydopamine lesion model in behavioral brain research. analysis of functional deficits, recovery and treatments

    Progress in Neurobiology

    (1996)
  • Y. Smith et al.

    The thalamostriatal system: a highly specific network of the basal ganglia circuitry

    Trends in Neurosciences

    (2004)
  • D.J. Surmeier et al.

    D1 and D2 dopamine-receptor modulation of striatal glutamatergic signaling in striatal medium spiny neurons

    Trends in Neurosciences

    (2007)
  • J.M. Tepper et al.

    GABAergic microcircuits in the neostriatum

    Trends in Neurosciences

    (2004)
  • I.Q. Whishaw et al.

    Dopamine depletion, stimulation or blockade in the rat disrupts spatial navigation and locomotion dependent upon beacon or distal cues

    Behavioural Brain Research

    (1985)
  • D.G. Beiser et al.

    Model of cortical-basal ganglionic processing: Encoding the serial order of sensory events

    Journal of Neurophysiology

    (1998)
  • J.D. Berke

    Uncoordinated firing rate changes of striatal fast-spiking interneurons during behavioral task performance

    Journal of Neuroscience

    (2008)
  • M.D. Bevan et al.

    Selective innervation of neostriatal interneurons by a subclass of neuron in the globus pallidus of the rat

    Journal of Neuroscience

    (1998)
  • K.T. Blackwell et al.

    Quantitative estimate of synaptic inputs to striatal neurons during up and down states in vitro

    Journal of Neuroscience

    (2003)
  • J.P. Bolam et al.

    Microcircuits in the striatum

  • E. Bracci et al.

    Dopamine excites fast-spiking interneurons in the striatum

    Journal of Neurophysiology

    (2002)
  • E. Bracci et al.

    Excitatory GABAergic effects in striatal projection neurons

    Journal of Neurophysiology

    (2006)
  • E.N. Brown et al.

    Multiple neural spike train data analysis: State-of-the-art and future challenges

    Nature Neuroscience

    (2004)
  • R.E. Burke et al.

    A parsimonious description of motoneuron dendritic morphology using computer simulation

    Journal of Neuroscience

    (1992)
  • G. Buzsaki

    Large-scale recording of neuronal ensembles

    Nature Neuroscience

    (2004)
  • L. Carrillo-Reid et al.

    Encoding network states by striatal cell assemblies

    Journal of Neurophysiology

    (2008)
  • D. Centonze et al.

    Receptor subtypes involved in the presynaptic and postsynaptic actions of dopamine on striatal interneurons

    Journal of Neuroscience

    (2003)
  • J. Cho et al.

    Dopamine depletion causes fragmented clustering of neurons in the sensorimotor striatum: evidence of lasting reorganization of corticostriatal input

    Journal of Comparative Neurology

    (2002)
  • U. Czubayko et al.

    Fast synaptic transmission between striatal spiny projection neurons

    Proceedings of the National Academy of Sciences of the United States of America

    (2002)
  • T.M. Dawson et al.

    D-1 dopamine receptors in the rat brain: A quantitative autoradiographic analysis

    Journal of Neuroscience

    (1986)
  • M. Djurfeldt et al.

    Large-scale modeling a tool for conquering the complexity of the brain

    Frontiers in Neuroinformatics

    (2008)
  • M.J. Frank

    Dynamic dopamine modulation in the basal ganglia: a neurocomputational account of cognitive deficits in medicated and nonmedicated Parkinsonism

    Journal of Cognitive Neuroscience

    (2005)
  • T. Fukai et al.

    A simple neural network exhibiting selective activation of neuronal ensembles: from winner-take-all to winners-share-all

    Neural Computation

    (1997)
  • R.F. Galan

    On how network architecture determines the dominant patterns of spontaneous neural activity

    PloS One

    (2008)
  • M. Galarreta et al.

    A network of fast-spiking cells in the neocortex connected by electrical synapses

    Nature

    (1999)
  • G. Glynn et al.

    Three-dimensional electrophysiological topography of the rat corticostriatal system

    Journal of Comparative Physiology A

    (2002)
  • F. Gonon

    Prolonged and extrasynaptic excitatory action of dopamine mediated by d1 receptors in the rat striatum in vivo

    Journal of Neuroscience

    (1997)
  • N. Gorelova et al.

    Mechanisms of dopamine activation of fast-spiking interneurons that exert inhibition in rat prefrontal cortex

    Journal of Neurophysiology

    (2002)
  • Cited by (111)

    • Distributed cell assemblies spanning prefrontal cortex and striatum

      2022, Current Biology
      Citation Excerpt :

      This would pose a theoretical conundrum because the vast majority of striatal neurons are inhibitory and have sparse local connectivity,57 and it would thus be unlikely that cell assemblies could emerge only from local striatal circuit interactions. Instead, consistent with the known anatomical projections,54 conduction delays,40 and computational simulations,58,59 timed prefrontal inputs could leverage synchronous striatal activity and/or oscillatory coherence with striatum for distributed assembly activations. Our data are consistent with PFC monosynaptically driving the STR during assembly activations because of the results of the Granger analysis of oscillations, as well as the asymmetry of the spike pair cross-correlograms, although precisely delayed driving inputs from a third structure cannot be excluded.

    View all citing articles on Scopus
    View full text