Introduction

Differences between direct an indirect MSNs are not obvious from morphology or electrophysiology. However, the two cell types and pathways can be separated and selectively manipulated using genetic tools, such as lines of reporter mice that express GFP in either D1 or D2 expressing MSNs, so many recent studies have tried to sort apart the pathways. This paper is a review of more classic studies on DA’s differential effects through D1 versus D2 signalling on glutamatergic signaling, intrinsic activity, and plasticity.

Classic View of DA Modulation

Albin hypothesized that DA acts through D1 receptors to excite MSNs and through D2 receptors to inhibit MSNs. This has largely borne out. Though DA acts through GPCRs, it can ‘excite’ or ‘inhibit’ MSNs by “modulating the gating and trafficking” of voltage-dependent and channels and ionotropic receptors.

D1 receptor activation increases glutamatergic signaling

DA acts through G_olf to activate AC, increasing cAMP and PKA signaling, which phosphorylates intracellular targets such as DARPP-32. D1 receptor activation of PKA can increase surface expression of both AMPA and NMDA receptors.

It’s controversial whether D1R stimulation rapidly affects glutamate receptor gating. While PKA phosphorylation of NR1 subunits can enhance NMDA currents, it is unclear whether this occurs in MSNs and it’s hard to directly study because the D1Rs are out on dendrites, which are not amenable to patch clamping.

Results conflict from different preparations. When intracellular cesium is used to block voltage-gated channels (and allow better voltage clamping of dendrites and transmission of small conductances from the dendrites back to the cell body), D1 agonists seem to have no effect on AMPA or NMDA signaling in the dorsal striatum. However, an earlier study that didn’t use cesium showed D1 receptor stimulation rapidly enhances NMDA receptor signaling. One possible explanation for this conflict is that D1Rs enhance NMDA currents through modulation of voltage-dependent dendritic conductances, which also jives with a study showing voltage-gated L-type Ca2 channels (which are phosphorylated by PKA, enhancing their opening) are necessary for D1-receptor mediated enhancement of NMDA current and newer theories (based on work in cortex) that dendrites active role in computation.

D1 Receptor Activation Modulates Intrinsic Excitability through Channels

D1 signalling can decrease voltage-dependent Na+ conductance, by making them enter a non-conducting slowly inactivated state.

D1R agonists’ effects on MSNs depend on the ‘state’ of the MSNs. If MSNs are voltage-clamped near the ‘upstate’ ~ -60mV agonists have different effects, then if MSNs are clamped near the down state of ~ -80mV. (These states seem to be regulated in part by the activity of channels; in the up state Kir2, Kv1, and Kv4 are all closed, increasing the membrane resistance and depolarizing the cell, as well as DA-mediated enhancement of L-type Ca2+ channels and NMDARs.)

In the up state D1 receptor stimulation increases MSN response to current injection. One mechanism by which D1 agonists increase excitability, is through PKA-mediated phosphorylation of L-type Ca2+ channels enhancing their opening. Because L-type channels open near -60mV, enhancement of these channels will mainly effect MSNs in the up state.

D2 activation decreases glutamatergic signaling

Act through G_i/o which inhibits AC. D2 receptor activation decreases AMPA signaling and tissue and culture possibly through dephosphorylation of a GluR1 site leading AMPA to be trafficked away from the synaptic membrane. D2 signalling also decreases the presynaptic release of glutamate.

D2 signalling decreases MSN excitability

Causes negative modulation of Cav1.3, slow inactivation of Na+ channels, and opening of K+ channels.

DA’s effect on striatal cholinergic interneurons

All cholinergic interneurons seem to express DA receptors, but the most well characterized effects are on the giant cholinergic Tonically Active Neurons (TANS) which express D2 receptors. D2 signalling onto TANs decreases Ach release, both by inhibiting spiking and by decreasing calcium in the presynapse and therefore ntx release. The cited Graybiel paper shows that after classical conditioning, cholinergic neurons all throughout the striatum decrease in firing during presentation of the US (consistent with DA modulation), and that this effect is abolished during 6-OH-DA lesion. Graybiel proposes that cholinergic neurons may provide synchronization for striatum and help solve the ‘binding problem’ by helping to allow distal regions of the striatum (and other brain regions) to combine information in a meaningful way. This somewhat clashes with excessive synchronization theories of Parkinson.

ACh in turn has many intrastriatal targets including DA terminals, glutamtergic terminals, and MSNs. ACh increases the excitability and responsiveness of MSNs to glutamate. So because DA decreases ACh signaling, this will cause an indirect decrease in MSN activation.

LTD of corticostriatal synapses requires D2 actviation in direct and indirect MSNs

In slice preparations when postsynaptic depolarization is paired with HFS of glutamatergic fibers, a non-NMDA LTD (HFS-LTD) is seen in almost all MSNs. This LTD depends on activation of L-type Ca2+ channels, mGluR1 and mGluR5 receptors and the generation of endocannabinoids (ECs). ECs act on CB1 receptors presynaptically.

The mechanism of why D2 activation is required for HFS-LTD is unclear, but it is intriguing that D2 signalling is required for HFS-LTD in the non-D2 expressing Direct MSNs.

Two theories are proposed to explain why Direct Pathway MSNs need D2 signaling for HFS-LTD. One theory is that DA allows HFS-LTD by acting on the D2 receptors of TANs. This inhibits ACh release, causing decreased M1 Receptor activation, and therefore decreasing M1 receptor-mediated inhibition of L-type Ca2+ channels. The other theory posits that HFS-LTD seen in Direct Pathway MSNs is due to ECs produced by Indirect Pathway MSNs that diffuse to neighboring terminals where they induce LTD.

It’s unclear to me how much to read into these studies of plasticity using superphysiological stimuli in slice.

LTP of corticostriatal glutamatergic transmission

HFS-LTP in the striatum isn’t nearly as well studied as its depressive counterpart. HFS-LTP requires co-activation of D1 and NMDA receptors. Induction of LTP in MSNs has been shown in vivo (reviewed in Mahon 2004, which looks interesting).

STDP at corticostriatal synapses

Has been reported by Fino et al. 2005, but it is unknown if DA modulates this signal.

D1 and D2 receptors are found on different types of corticostriatal synapses

D2 indirect MSNs receive 4x as many projections from cortical neurons that project to the pyramidal tract. D1 direct MSNs receive 2x as many projections from neurons projecting exclusively to the striatum.

If you lesion striatum, do VTA neurons still show RPE? – Is striatal plasticity involved in reward prediction learning? Or just in increasing behaviors that lead to rewards?

Other random facts:

Thalamic inputs to striatum

Up to 40% of glutamatergic inputs to MSNs come from the thalamus, but no one knows much about plasticity at these synapses. In primates, the ‘associative’ intralaminar nuclei of the thalamus target primarily the direct pathway. In rodents, ‘motor nuclei’ VA/VL of the thalamus primarily target the indirect pathway.

Random Action Selction Hypothesis Thoughts

Even if BG cells are too slow to direcly activate or inhibit the current action, perhaps they have these effects on subsequent actions. Once you start one action, you need to suppress subsequent ‘co-activation of incompatible action programs,’ similarly you may need to encourage cortical activation of the next movement. This may also help explain why the basal-ganglia is so involved in sequenced motor learning.