Prefrontal cortical (PFC) circuits mediating higher cognitive
operations. A. The PFC expands tremendously in brain evolution,
comprising a very small proportion of the brain in rodents such as the rat
and increasing dramatically in primates, with special prominence in the
human brain. The PFC is highlighted in blue. An inset of the human dlPFC is
shown at the right; this Nissl-stained section shows the six layers of
dlPFC. B. The dlPFC microcircuits subserving working memory,
discovered by Goldman-Rakic (1995) [1]. Pyramidal cells in deep layer III receive
visuospatial information from the parietal association cortex. Pyramidal
cells with similar spatial inputs excite each other through connections on
dendritic spines to maintain persistent firing throughout the delay period.
The spatial tuning of the neuron’s response is sharpened by lateral
inhibition from parvalbumin-containing GABAergic interneurons, such as the
Basket cell (B) shown in this figure. Note that chandelier cells also serve
this function (not shown). The red rectangle highlights an axo-spinous
synapse enlarged in C and D. These dlPFC microcircuits are the ones most
afflicted in schizophrenia, where there is loss of neuropil (including loss
of dendritic spines) in deep layer III [100], and reduced parvalbumin GABAergic
function [101].
C. A working model of the cAMP-potassium channel signaling
mechanisms in spines that dynamically weaken synaptic efficacy and gate out
network inputs to the neuron. cAMP directly opens HCN channels, while cAMP
activation of PKA signaling increases the open state of KCNQ channels. cAMP
generated by calcium build up, e.g., feedback fatigue via NMDA or mGluR1/5,
or actively generated by stress exposure, e.g., via D1 or β1 receptor
stimulation. D. NE or guanfacine stimulation of α2A receptors
on spines inhibits cAMP production and closes HCN and KCNQ channels,
strengthening network connectivity, increasing neuronal PFC firing, and thus
improving PFC regulation of behavior, thought and emotion. See Figure 5 for data supporting the model. C
and D artistically adapted from Arnsten et al., 2010 [71].