Prefrontal Area 32 Links Cognitive and Affective Areas
Mary Kate P. Joyce, Miguel Ángel García-Cabezas, Yohan J. John, and Helen Barbas
(see pages 8306–8328)
Responding appropriately to threats is essential for survival, but hypersensitivity to potential threats can be maladaptive. Therefore, activity in area 25 (A25) of the ventromedial prefrontal cortex, which is thought to drive physiological and behavioral responses to threats and stress, may be dampened by dorsolateral prefrontal cortex (DLPFC) to inhibit disadvantageous threat responses. Consistent with this hypothesis, DLPFC is hypoactive and A25 is hyperactive in people with major depression, and successful treatment increases DLPFC activity while reducing A25 activity. Because DLPFC has only sparse connections with A25, however, any effects of DLPFC on A25 must be indirect. Joyce et al. posited that the pregenual anterior cingulate area A32, which has strong connections with both areas, serves as an intermediary. They therefore used confocal and electron microscopy to explore its connections in more detail.
Tracers injected into monkey A32 revealed that this area received input primarily from superficial layers in DLPFC and it densely innervated all layers of A25. Notably, presynaptic boutons formed by A32 axons in A25 were larger, more likely to have mitochondria, and more likely to contact spines with perforated synapses and a spine apparatus than surrounding excitatory boutons.
Approximately 75% of neurons targeted by A32 axons in A25 were likely pyramidal neurons, based on their shape and/or lack of expression of the calcium-binding proteins calbindin, calretinin, and parvalbumin, which identify different types of GABAergic neurons. Among the inhibitory neurons targeted by A32, calretinin-expressing neurons were the most frequently contacted in the superficial layers, whereas parvalbumin-expressing neurons were more frequently targeted in deep layers. Calbindin-expressing neurons received an intermediate proportion of inputs in all layers.
These results confirm that A32 receives input from DLPFC and provides dense, strong innervation of A25. Given that A32 primarily targets pyramidal neurons and calretinin interneurons (which typically inhibit other interneurons) in the superficial layers of A25, it is likely to have a net excitatory effect in these layers. In contrast, by targeting parvalbumin-expressing neurons in deep layers, A32 may exert strong inhibitory control over A25 output. Future work with more targeted labeling should map this circuit in more detail to determine whether DLPFC targets A32 neurons that innervate deep or superficial layers of A25.
Thalamic Nucleus Reuniens Influences Sequencing in mPFC
David Angulo-Garcia, Maëva Ferraris, Antoine Ghestem, Lauriane Nallet-Khosrofian, Christophe Bernard, et al.
(see pages 8343–8354)
The hippocampus works closely with the medial prefrontal cortex (mPFC) to store and retrieve memories that guide behavior. Within each area, memories are thought to be encoded by ensembles of neurons that fire in a precise temporal sequence. Repeated activation of these ensembles strengthens synaptic connections within them, and thus consolidates memories. This occurs frequently during slow-wave sleep. Specifically, ensembles in mPFC are repeatedly activated during the depolarized UP states of slow oscillations, while hippocampal ensembles are activated during sharp-wave ripple events (SWRs). The thalamic nucleus reuniens is thought to coordinate ensemble activity in the two areas; this is required for memory consolidation.
How nucleus reuniens coordinates mPFC and hippocampal activity has been unclear. To find out, Angulo-Garcia, Ferraris, et al. recorded from all three areas in anesthetized rats. In the nucleus reuniens and mPFC, most recorded neurons fired in reliable temporal sequence at the beginning of UP states. Unexpectedly, neurons in the nucleus reuniens also tended to fire in a spatial sequence from dorsal to ventral, suggesting a previously unrecognized topographical organization in this area. Hippocampal neuron activity was not as tightly tied to UP states as that in nucleus reuniens and mPFC, but hippocampal neurons showed sequential activation during SWRs, as expected.
The temporal sequence of neuronal firing across UP states was most consistent in nucleus reuniens, somewhat lower in mPFC, and minimally consistent in hippocampus. Importantly, sequences in nucleus reuniens always preceded those in mPFC. Moreover, when the nucleus reuniens was inhibited via muscimol injection, the consistency and number of neurons participating in temporal sequences was reduced in both mPFC (during UP states) and hippocampus (during SWRs).
These results suggest that neurons within ensembles in nucleus reuniens are activated in reliable spatiotemporal sequences at the beginning of UP states and that this activity plays an important role in driving temporal firing patterns in neuronal ensembles in hippocampus and mPFC. Future work should confirm that these patterns occur in natural sleep as well as anesthesia, assess the importance of temporal sequences in nucleus reuniens for memory consolidation, and investigate how nucleus reuniens affects activity during hippocampal SWRs, which occur at variable times relative to the onset of UP states.
Footnotes
This Week in The Journal was written by Teresa Esch, Ph.D.