Cortical and thalamic inputs to the dorsal striatum
The corticostriatal input is overall topographically organized. Corticostriatal pyramidal cells have multiple projections sites including ipsi- but also contralateral striatum (Shepherd, 2013). Recent studies have focused on systematically mapping the precise excitatory projection patterns to the entire striatum (Hunnicutt et al., 2016), revealing striatal subregions and their boundaries by systematic analysis of these input patterns from individual cortical and thalamic subregions (Oh et al., 2014; Hunnicutt et al., 2016). They found clear boundaries separating the three traditional striatal domains and uncovered a fourth subdivision in the posterior striatum. Interestingly, the posterior striatum has also been reported to be innervated by a specific dopamine neurons population (Menegas et al., 2015). The dorsomedial striatum seem to show the highest degree of cortical input heterogeneity, suggesting it serves as an information hub. In addition, thalamic subregions receiving basal ganglia outputs are preferentially interconnected with motor-related cortical subregions.
Thalamic subregions are defined by cytoarchitectural boundaries to delineate ~40 nuclei. The organization of thalamostriatal inputs, which provide ~1/3 of all glutamatergic synapses in the striatum (Huerta-Ocampo et al., 2014), particularly originating from the intralaminar thalamic nuclei (Smith et al., 2004; Doig et al., 2010), has been less studied than corticostriatal inputs.
Cortical and thalamic neurons regulate the activity of SPNs in the striatum through long-range glutamatergic excitatory projections (Landry et al., 1984; Wilson, 1987; Graybiel et al., 1994; Lovinger and Tyler, 1996; Reiner et al., 2003; Kress et al., 2013; Hunnicutt et al., 2016), while inhibition is mediated by feed-forward and feed-back circuits (Tepper et al., 2008). The main feed-forward circuit is characterized by GABAergic striatal interneurons receiving excitatory inputs from the cortex and monosynaptically inhibiting SPNs. The feed-back circuit comprises SPNs interconnections via local axon collaterals (Calabresi et al., 1991; Kawaguchi, 1993; Kita, 1993, 1996; Kawaguchi et al., 1995; Tunstall et al., 2002; Planert et al., 2010). SPN to SPN synapses are typically unidirectional, predominantly localized onto distal dendrites of other SPNs (Tunstall et al., 2002; Plenz, 2003; Koos et al., 2004; Taverna et al., 2004, 2005). dSPNs preferentially innervate other dSPNs, whereas iSPNs innervate both subtypes equally (Taverna et al., 2008). More recent work using optogenetic and electrophysiology approaches to activate SPNs has inferred an even higher degree of connectivity (Chuhma et al., 2011; Wei et al., 2017).
Each SPN receives several thousands of corticostriatal synapses on its dendrites, but each individual corticostriatal axon makes only a few contacts with each SPN, typically making one or two en passant synapses (Kincaid et al., 1998; Parent and Parent, 2006). Corticostriatal and thalamostriatal glutamatergic synapses are similar morphologically and intermingled along the dendrites of SPNs (Smith et al., 2004; Raju et al., 2006). The synapses they form on SPNs are nearly equal in number, corticostriatal synapses being slightly more numerous (Smith et al., 2004). The vast majority of synapses target spines rather than dendritic shafts and the proportion seems more elevated in corticostriatal than thalamostriatal synapses (Fujiyama et al., 2006; Raju et al., 2006, 2008; Moss and Bolam, 2008; Doig et al., 2010; Lei et al., 2013; Zhang et al., 2013; Huerta-Ocampo et al., 2014) (Figure Intro 5).
[Figure Intro 5: Pie charts illustrate the proportions of targets for corticostriatal (A) and thalamostriatal (B) synapses in mice. Axospinous, synapses onto spines; Axodendritic, contacts onto dendrites; Axo?, contacts onto undetermined targets. From Zhang et al 2013].
A study from Wall et al 2013, using a monosynaptic rabies virus system, also showed that innervation of D1-SPN and D2-SPN seems anatomically biased towards specific brain regions (Wall et al., 2013). They found that thalamostriatal input and dopaminergic input seem similar onto both pathways. Motor cortex preferentially targets the indirect pathway while sensory cortical and limbic structures preferentially innervate the direct pathway. Others find that pyramidal tract axons innervate both type of SPN (Kress et al., 2013).
Striatal dopamine inputs
There are ten DA-producing nuclei in the mammalian brain (Björklund and Dunnett, 2007; Tritsch and Sabatini, 2012) designated A8 to A17 (Figure intro 6). Projections from a given subset of DA neurons target one region of the brain (Yetnikoff et al., 2014), but projections to a given subset of DA neurons arise from many different regions as shown by studies using combinations of techniques using CLARITY and Rabies virus injections (Watabe-Uchida et al., 2012; Menegas et al., 2015). The substantia nigra pars compacta (SNc; field A9) and VTA (field A10), which project to the dorsal and ventral striatum, and form the nigrostriatal and mesocorticolimbic pathways respectively; each contain in the rodent ≈20,000 – 30,000 neurons bilaterally (German and Manaye, 1993; Zaborszky and Vadasz, 2001), and a total of neurons between 160 000–320 000 in monkeys and 400 000–600 000 in humans, with >70% of the neurons located in the SN (Björklund and Dunnett, 2007). These neurons form widely spread highly dense axonal arborizations in the striatum (Prensa and Parent, 2001; Matsuda et al., 2009). Individual SNc neurons extend highly branched axons of half a meter in total length that densely ramify throughout up to 1 mm3 of tissue (Matsuda et al., 2009). Dopaminergic synaptic boutons represent ≈10% of all synapses in the striatum (Groves et al., 1994). The closest distance in between each dopaminergic bouton is only ∼1.18 μm (Arbuthnott and Wickens, 2007). Some of these terminals are found at spine necks abutting cortical synapses (Smith et al., 1994; Moss and Bolam, 2008), but many dopaminergic terminals have been found against dendritic shafts with no detectable electron-dense postsynaptic structure (Groves et al., 1994; Hanley and Bolam, 1997).
[Figure intro 6: Distribution of DA neuron cell groups in the adult rodent brain as illustrated schematically on a sagittal view. The numbering of cell groups, from A8 to A16, was introduced in the classic study of Dahlström and Fuxe in 1964. A17 Retinal dopaminergic neurons are not shown on this drawing. The principal projections of the DA cell groups are illustrated by arrows. Adapted from Bjorklund and Dunnett 2007] Midbrain DA neurons are autonomous pacemakers spontaneously active at low frequencies (1–8 Hz) in vivo (Guzman et al., 2009; Kreitzer, 2009), suggesting that each neuron provides a basal DA tone to many target neurons, adjusted by either phasic bursts or transient pauses of activity, critical for normal striatal function (Schultz, 2007a). This basal DA tone most likely activates high-affinity DA receptors of the D2-type (D2–D4) (Richfield et al., 1989). Bursts of action potentials that briefly elevate striatal extracellular DA are fired by DA neurons in response to behaviorally relevant stimuli (Schultz, 2007b). These phasic spikes of DA are activating both the high-affinity D2 type receptors and the lower-affinity DA receptors of the D1 type (D1, D5) (Richfield et al., 1989). The phasic and tonic firings of DA neurons allow encoding transient responses, for instance to reward prediction error, or longer timescale responses like uncertainty (Fiorillo et al., 2003; Schultz, 2007b). Dopamine thus plays a crucial role in the control of motor, cognitive and emotional behaviors and their adaptation by learning in response to reward.
Neurotransmitter receptors and signaling pathways in the dorsal striatum
DA receptors belong to the large superfamily of guanine nucleotide binding protein-coupled receptors (GPCRs). These metabotropic receptors share interaction with G-proteins, structure with seven alpha-helices transmembrane domains that are interconnected by alternating intracellular and extracellular loops. The heterotrimeric G-proteins are formed by a combination of an α-subunit and a βγ dimer that can each lead to activation of signaling effectors. These receptors do not signal exclusively through heterotrimeric G proteins and may also engage in G protein-independent signaling events (Luttrell, 2014; Peterson and Luttrell, 2017; Luttrell et al., 2018; Pack et al., 2018).
DA binds and activates two families of GPCRs (Kebabian and Calne, 1979; Andersen et al., 1990; Sibley and Monsma, 1992; Greengard et al., 1999; Beaulieu and Gainetdinov, 2011; Tritsch and Sabatini, 2012): the D1 family (D1 and D5 subtypes) (Tiberi et al., 1991) and the D2 family (D2, D3 and D4 subtypes) with different affinities for DA ranging from nanomolar to micromolar range, but less different between individual subtypes within a family. The affinity of D2-like receptors for DA is generally reported to be 10- to 100-fold greater than that of D1-like receptors, with D3 and D4 receptors displaying the highest sensitivity for DA and D1 receptors the lowest (Beaulieu and Gainetdinov, 2011) (Table 2). But D1 and D2 receptors can exist in both high and low affinity states.
The organization of the genes of D1- and D2-class DA receptors is also different. The D1 and D5 DA receptor genes do not contain introns in their coding regions, so do not generate splice variants. The genes encoding the D2-class receptors have several introns, with six introns in the gene that encodes the D2R receptor, five in the gene for the D3R, and three in the gene for the D4R (Gingrich and Caron, 1993). The alternative splicing of an 87-base-pair exon between introns 4 and 5 lead to different isoforms. The two main isoforms are the short and long variants of D2 receptors (D2S and D2L, respectively mediating pre- and post- synaptic signaling (De Mei et al., 2009). D2S and D2L isoforms differ in the presence of an additional 29 amino acids in the third intracellular loop. Variants of D3 and D4 receptors have also been described (Callier et al., 2003). The D3 splice variants encode proteins essentially nonfunctional (Giros et al., 1991). The D4 polymorphic variants have a 48 bp repeat sequence in the third cytoplasmic loop, with up to 11 repeats reported (Van Tol et al., 1992).
The D1 and D5 DA receptors are 80% identical in their transmembrane domains, whereas the D3 and D4 DA receptors are 75% and 53% identical, respectively, with the D2R. Whereas the NH2-terminal domain has a similar number of amino acids in all of the DA receptors, the COOH-terminal for the D1-class receptors is seven times longer than for the D2-class receptors (Gingrich and Caron, 1993).
DA receptors are broadly expressed in the CNS, their distribution matching the density of innervation by DA fibers. D1- and D2-like receptors are expressed in both SPNs and interneurons in the striatum, and in subpopulations of pyramidal neurons, interneurons, and glial cells in cortex (Tritsch and Sabatini, 2012) (Table 2). DA receptors are located both synaptically and extrasynaptically (Gerfen and Surmeier, 2011b).
As mentioned above, D1 receptors (D1R) are expressed by striatonigral neurons of the direct pathway (D1-dSPNs), whereas D2 receptors (D2R) are expressed by striatopallidal neurons of the indirect pathway (D2-iSPNs) (Figure intro 3). The segregation in two distinct populations is high but not complete (Le Moine and Bloch, 1996; Bertran-Gonzalez et al., 2008, 2010; Matamales et al., 2009) and changes in technical approaches have allowed to further detail these repartitions (Valjent et al., 2009; Ade et al., 2011; Durieux et al., 2011). In the dorsal striatum, only 5% of the striato-pallidal iSPN express D1R and 5% of the striato-nigral dSPN express D2R (Le Moine and Bloch, 1995; Ince et al., 1997; Valjent et al., 2009) (Table 3). In the nucleus accumbens (Le Moine and Bloch, 1995), a higher proportion of iSPN, mainly originating from the shell of the nucleus accumbens rather than from the core, express D1-like receptors (Robertson and Jian, 1995; Bertran-Gonzalez et al., 2010; Gangarossa et al., 2013a). The D3R are also mostly present in the ventral regions of the striatum. Along the rostro-caudal axis of the mouse dorsal striatum, D1R- and D2R-expressing SPNs are randomly distributed in most of the dorsal striatum, except a specific region in the caudal striatum, adjacent to the GPe. This region exclusively comprises D1R-expressing dSPNs and lacks neurons expressing markers for iSPN, and especially D2R (Gangarossa et al., 2013b).
[Table 3: Cellular distribution of DA receptors in the cortex and striatum of rodents. This table reports semiquantitative expression levels of various DA receptor subtypes (+++, highest expression; +, low expression; -, mRNA not detected) and their relative cellular distribution (in parentheses) within defined cortical and striatal neuronal populations.
As mentioned above, D2R are also expressed at the presynaptic level, on dopaminergic neurons where they act as autoreceptors. D2R have also been reported on non-dopaminergic afferent fibers to the striatum among which the glutamatergic cortical and thalamic afferents and that innervate SPNs and interneurons (Sesack et al., 1994; Wang and Pickel, 2002). D1R have also been observed in a small number of presynaptic glutamatergic terminals in the striatum (Huang et al., 1992; Dumartin et al., 2007). Some interneurons have also been reported to bare DA receptors: D2 and D5 receptors on cholinergic interneurons, D5 receptors on somatostatin/neuropeptide Y interneurons, on parvalbumin and on calretinin GABAergic interneurons (Yan and Surmeier, 1997; Rivera et al., 2002) (Table 3).
Glutamate is the major excitatory neurotransmitter in the vertebrate central nervous system responsible for the fast excitatory transmission. Glutamate acts on two types of glutamate receptors: the ionotropic (iGluR), which are ligand-gated ion channels, and the metabotropic glutamate receptors (mGluR).
Striatal ionotropic glutamate receptors (iGluRs)
The iGluRs comprise two major groups termed after their selective synthetic ligands: the N-methyl-d-aspartate (NMDA) receptors (NMDAR) and the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate receptors (AMPAR) (Dingledine et al., 1999). NMDAR and AMPAR are heterotetramers and their subunit composition controls their properties (Seeburg and Hartner, 2003; Traynelis et al., 2010; Paoletti et al., 2013).
NMDA- and AMPA/kainate iGluR are expressed presynaptically in dopaminergic and glutamatergic terminals, but also postsynaptically in SPNs where they produce postsynaptic excitatory currents (Bernard et al., 1997; Bernard and Bolam, 1998; Tarazi et al., 1998; Tarazi and Baldessarini, 1999; Gardoni and Bellone, 2015). A study has shown that AMPAR, but not NMDAR, are located on glutamatergic corticostriatal and thalamostriatal terminals (Fujiyama et al., 2004). AMPA, NMDA and kainate receptors are also expressed on cholinergic and GABAergic striatal interneurons (Tallaksen-Greene and Albin, 1994; Deng et al., 2007, 2010).
AMPARs, heteromers composed of GluA1-4 subunits, are the main type of glutamate receptor mediating the fast excitatory response of neurons to glutamate (Traynelis et al., 2010). They are permeable to cations and, at normal resting membrane potential, depolarize the cell by letting Na+ flow in and K+ flow out. AMPAR lacking the GluA2 subunit can also let Ca2+ enter the cell. GluA2 subunits impair Ca2+ permeability of AMPAR. RNA editing and alternative splicing generate sequence variants, and those variants, as in GluA2-4 AMPA receptor subunits, generally show different properties (Seeburg and Hartner, 2003; Penn et al., 2012; Wright and Vissel, 2012; La Via et al., 2013; Wen et al., 2017).
NMDAR are heterotetramers composed of GluN1 subunits and GluN2 (A-D) and/or GluN3 (A-B) subunits encoded by 7 different genes (Stroebel et al., 2018). Subunit composition determines receptor properties. The GluN2 regulatory subunits are responsible for glutamate binding (Laube et al., 1997) and exist as four different isoforms, mostly GluN2A and GluN2B in the striatum (formerly known as NR2A and NR2B, respectively). To open NMDAR requires binding not only of glutamate but also a coagonist, namely glycine (Johnson and Ascher, 1987; Kleckner and Dingledine, 1988) or D-serine (Mothet et al., 2000) at GluN1 or GluN3, the obligatory subunit to form the ion channel. Importantly, at negative membrane potential (i.e., resting potential or partial depolarization), NMDA channels are blocked by extracellular Mg2+ ions (Nowak et al., 1984). This obstruction is lifted at positive membrane potential (Mayer et al., 1984) and the channels then become permeable to Na+, K+ and Ca2+ ions; providing a major entry point for Ca2+-dependent intracellular pathways. NMDA channels are coincidence detectors that are activated when the presynaptic terminal releases glutamate and when the postsynaptic element is depolarized.
Regarding synaptic plasticity, long term potentiation (LTP) usually corresponds to an increased function of AMPAR,with first an increased permeability of existing channels followed by an increased number of channels at the postsynaptic side, with possible changes in subunit composition. LTP is usually triggered by a massive postsynaptic Ca2+ influx resulting from the opening of NMDAR. Long-term depression (LTD), in contrast, often results from a decreased permeability and mostly a decreased number of AMPAR in the synaptic region and is triggered by a small increase in postsynaptic Ca2+.
Striatal metabotropopic glutamate receptor (mGluRs)
The mGluR are GPCR. Eight mGluRs are classified into three groups, namely the group I mGlu receptors (that includes the mGlu1 and mGlu5 receptors), the group II mGlu receptors (that includes the mGluR2 and mGluR3 receptors) and the group III mGlu receptors (that includes the mGluR4, mGluR6, mGluR7, and mGluR8 receptors)(David et al., 2005; Ferraguti and Shigemoto, 2006).
Histological studies have revealed that mGlu receptors are densely distributed within the striatum (Ferraguti and Shigemoto, 2006). Although the vast majority (90%) of mGlu receptors is thought to be preferentially located postsynaptically and less in presynaptic level, it has been discussed wether mGluR, notably group I, were located at the presynaptic level which seems to be the case (Pittaluga, 2016). Postsynaptic mGluR group I receptors appear to be concentrated in perisynaptic and extrasynaptic area (Nicoletti et al., 2011). Main localizations of mGluR in the basal ganglia circuits are summarized in Figure intro 7 (Conn et al., 2005).
mGluR1 and mGluR5 are Gq protein-coupled and stimulate phosphoinositide hydrolysis. Their activation induces mobilization of intracellular Ca2+ stores and activation of phospholipase C (PLC)(Conn et al., 2005). Activation of PLC catalyzes the cleavage of phosphatidylinositol-4,5-bisphosphate to inositol trisphosphate (IP3) and diacylglycerol (DAG). The latter activates protein kinase C (PKC) in the presence of Ca2+. mGluR5 regulates IP3-induced and indirectly Ca2+-induced Ca2+-release (CICR) from the endoplasmic reticulum (ER) via IP3 and ensuing stimulation of IP3 receptors. mGluR5 also modulates L-type VGCC in a PLC/PKC dependent way (Fieblinger et al., 2014b). The activation of the presynaptic Group II mGluRs inhibits cyclic AMP (cAMP) and cAMP-dependent protein kinase A (PKA) signaling since they are coupled to Gi/o proteins. Group III, similarly to Group II mGluRs, are negatively coupled to adenylate cyclase (AC) activity and found presynaptically in the glutamatergic terminals of the striatum. mGluR exist as either homo or heterodimers (Nicoletti et al., 2011). When they are presynaptic, they regulate the release of glutamate and of various transmitters, including GABA, dopamine, noradrenaline, and acetylcholine (Musante et al., 2008; Vergassola et al., 2018).
cAMP production and actions
In the intact striatum, DA regulates the cAMP pathway in opposite directions in dSPN and iSPN. In D1-dSPNs, DA activates D1R, which increase cAMP production by activating the AC. In D2-iSPNs, DA activates D2R, which decrease cAMP production by inhibiting the AC (Greengard et al., 1999). D1R in dSPNs and A2AR in iSPNs are coupled to Golf proteins, whereas D2R to Go and Gi proteins (Neve et al., 2004; Beaulieu and Gainetdinov, 2011). Gs and Golf proteins stimulate AC. Golf is a heterotrimeric G protein involved in olfaction, very closely related to Gs (88% amino acid homology) (Jones and Reed, 1989). In the neostriatum, expression of Gαs is very low, whereas Gαolf is abundantly expressed (Hervé et al., 1993). The heterotrimeric olfactory type G-protein Golf comprising the Gαolf, Gβ2 and Gγ7 subunits is required for DA-activated AC in the striatum, expressed in the two types of SPNs. Its alpha subunit, Gαolf, is necessary to couple D1 receptors in D1R expressing-SPNs but also A2A receptors in D2R expressing-SPNs, to the AC (Zhuang et al., 2000; Corvol et al., 2001). The G-protein γ subunit, Gγ7, associates with Gαolf to couple the receptors to the AC (Wang et al., 2001). In mutant KO mice for Gαolf or Gγ7, the activation of AC by DA or adenosine is severely impaired (Zhuang et al., 2000; Corvol et al., 2001; Hervé et al., 2001; Schwindinger et al., 2003). The main AC isoform in SPNs is AC5 (Visel et al., 2006; Hervé, 2011), whose deletion has severe functional consequences (Kheirbek et al., 2009). AC5 is inhibited by Ca2+ at concentration < 1 µM (Hanoune and Defer, 2001).
The major target of cAMP in neurons is the regulatory subunit (R) of PKA, a heterotetramer kinase containing two regulatory and two catalytic (C) subunits (Girault, 2012). The binding releases the C subunits, which become fully active, phosphorylate membrane-bound and cytosolic substrates, and can penetrate the nucleus to phosphorylate nuclear targets. R subunits interact with a family of partners called A-kinase-associated proteins (AKAPs), which enrich PKA at specific subcellular locations such as postsynaptic sites or perinuclear region (Logue and Scott, 2010). cAMP can also activate directly guanine nucleotide exchange factors (cAMP-GEF or EPACs) highly enriched in the striatum (Gloerich and Bos, 2010). The importance of these targets is increasing but not fully understood yet.
In the striatum, several protein substrates of PKA were identified as highly enriched in DA-innervated regions including 32-kDa DA and cAMP-regulated phosphoprotein (DARPP-32) (Walaas et al., 1983), and a number of other PKA substrates (Girault et al., 1990; Girault, 2012). Recent approaches using phosphoproteomics have extended the early work of Greengard and colleagues and identified numerous proteins phosphorylated in response to stimulation of D1R (Nagai et al., 2016b).
Table of contents :
1- The dorsal striatum and the basal ganglia-thalamo-cortical motor circuit
1.1 General organization of the striatum
1.2 Striatal projection neurons and main efferent pathways of the dorsal striatum
1.3 Cortical and thalamic inputs to the dorsal striatum
1.4 Striatal dopamine inputs
2- Neurotransmitter receptors and signaling pathways in the dorsal striatum
2.1 DA receptors
2.2 Glutamate receptors
2.3 cAMP production and actions
2.6 The ERK Cascade
2.7 Integration of multiple signaling pathways activated by DA, roles of D1R and D2R
2.8 Signaling crosstalk between glutamate and DA
2.9 Adenosine receptors and adenosine signaling in SPNs
2.10 Calcium signaling
3- PD and striatal alterations in the absence of dopamine
3.1 Animal models of PD
3.2 Striatal alterations in PD and dopamine deficiency
3.3 Physiological and signaling alterations in SPNs in the absence of dopamine
4- L-DOPA-induced dyskinesia
4.1 Clinical features
4.2 Model of basal ganglia circuit alterations in the dyskinetic state
4.3 Postsynaptic and presynaptic mechanisms of LIDs
4.4 Changes in structural and synaptic plasticity in LIDs
4.5 Molecular bases of LID
4.6 Role of glutamate transmission in LID
4.7 Role of PDEs
4.8 Role of adenosine in LIDs
5- Introduction to biosensor live imaging
5.1 FRET principles
5.2 An historical perspective on fluorescent probes
5.3 cAMP sensors
5.4 ERK sensors
5.5 Genetically encoded Ca2+ indicators (GECIs)