Differential Striatal Axonal Arborizations of the Intratelencephalic and Pyramidal-Tract Neurons: Analysis of the Data in the MouseLight Database

Abstract

There exist two major types of striatum-targeting neocortical neurons, specifically, intratelencephalic (IT) neurons and pyramidal-tract (PT) neurons. Regarding their striatal projections, it was once suggested that IT axons are extended whereas PT axons are primarily focal. However, subsequent study with an increased number of well-stained extended axons concluded that such an apparent distinction was spurious due to limited sample size. Recent work using genetically labeled neurons reintroduced the differential spatial extent of the striatal projections of IT and PT neurons through population-level analyses, complemented by observations of single axons. However, quantitative IT vs. PT comparison of a large number of axons remained to be conducted. We analyzed the data of axonal end-points of 161 IT neurons and 33 PT neurons in the MouseLight database (http://ml-neuronbrowser.janelia.org/). The number of axonal end-points in the ipsilateral striatum exhibits roughly monotonically decreasing distributions in both neuron types. Excluding neurons with no ipsilateral end-point, the distributions of the logarithm of the number of ipsilateral end-points are considerably overlapped between IT and PT neurons, although the proportion of neurons having more than 50 ipsilateral end-points is somewhat larger in IT neurons than in PT neurons. Looking at more details, among IT subpopulations in the secondary motor area (MOs), layer 5 neurons and bilateral striatum-targeting layer 2/3 neurons, but not contralateral striatum-non-targeting layer 2/3 neurons, have a larger number of ipsilateral end-points than MOs PT neurons. We also found that IT ipsilateral striatal axonal end-points are on average more widely distributed than PT end-points, especially in the medial-lateral direction. These results indicate that IT and PT striatal axons differ in the frequencies and spatial extent of end-points while there are wide varieties within each neuron type.

Antipsychotic drugs selectively decorrelate long-range interactions in deep cortical layers

“…Consistent with this, an increase in striatal dopamine activates auditory cortex (Maia and Frank, 2017) and mediates hallucination-like perceptions in mice (Schmack et al, 2021). It is intriguing, however, to speculate that the effects we observed in L5 IT neurons, which provide a dominant input to striatum (Morita et al, 2019), may be an alternate site of action. Feedback from striatum to cortex is exclusively mediated through thalamus, and neither L4, the primary thalamo-recipient layer in cortex, nor our brain wide recordings exhibited any of the effects we saw in L5 IT neurons.…”

Opponent learning with different representations in the cortico-basal ganglia circuits

“…Since SR-based TD learning also enables sensitive adaptation to changes in outcome values [20, 21], these results would also be consistent with the possibility that IT neurons, but not PT neurons, are engaged in SR-like representation, while the results also suggest that SR-based learning should occur from negative feedbacks (i.e., devaluation) at least to some extent. Local connections between IT neurons are less reciprocal (i.e., more unidirectional) than those between PT neurons [44], and axonal projections of individual IT neurons to the striatum tend to have more widespread endpoints than those of PT neurons [45]. These properties might also be in line with the possibility that IT neurons have SR-like state representation, because SR is based on the directional transition relationships between states and also calculation of state value using SR requires access to not only that state but also all the states which are reached from that state through transitions.…”

“…Since SR-based TD learning also enables sensitive adaptation to changes in outcome values (Momennejad et al, 2017;Russek et al, 2017), these results would also be consistent with the possibility that IT neurons, but not PT neurons, are engaged in SR-like representation, while the results also suggest that SR-based learning should occur from negative feedbacks (i.e., devaluation) at least to some extent. Local connections between IT neurons are less reciprocal than those between PT neurons (Morishima et al, 2011), and axonal projections of individual IT neurons to the striatum tend to have more widespread endpoints than those of PT neurons (Morita et al, 2019). These properties might also be in line with the possibility that IT neurons have SR-like, rather than punctate IR-like, state representation.…”

Benchmarking of tools for axon length measurement in individually-labeled projection neurons

“…For the foreseeable future, these high-end platforms will remain limited to industrial-scale research facilities. At present they remain focused in producing open-access datasets of adult "standard" model species brains intended as reference for other brain circuit studies [6,[19][20][21].…”