In:
Science, American Association for the Advancement of Science (AAAS), Vol. 376, No. 6590 ( 2022-04-15)
Abstract:
Learning is mediated by experience-dependent plasticity of neuronal function. Although we have a detailed knowledge of synaptic and dendritic plasticity in vitro, learning-induced functional changes in vivo are mostly assessed by use of soma-centric methods such as unit recordings or imaging of somatic Ca 2+ activity. However, these methods do not reveal the complexity of dendritic signaling and its regulation by local neuronal circuitry. Moreover, because dendrites are functionally independent subcellular compartments that dynamically integrate incoming information and thereby affect a neuron’s input-output function, the questions arise whether compartmentalized plasticity occurs during learning and which mechanisms ultimately control somatic output in behaving animals. RATIONALE To investigate dendritic function and plasticity in vivo, we focused on the lateral amygdala (LA), a subcortical brain structure that is central to classical auditory fear conditioning, a fast and robust form of associative learning. During auditory fear conditioning, an auditory conditioned stimulus (CS; typically a pure tone or white noise) is paired with an aversive unconditioned stimulus (US; typically a foot shock), which results in the induction of Hebbian activity-dependent synaptic potentiation at auditory synaptic inputs onto LA principal neurons (PNs). This view has recently been extended by studies reporting that similar proportions of neurons up- and down-regulate their CS response upon fear conditioning, suggesting that fear learning involves more diverse forms of plasticity. However, these studies relied on measurements of somatic activity, whereas dendritic activity and plasticity during associative learning was not explored. RESULTS To image the activity of dendrites and somas of amygdala PNs deep in the brain of awake mice undergoing classical fear conditioning, we used gradient-index lens–based high-resolution two-photon microscopy across multiple days. We show that sensory stimulation induces compartmentalized dendritic responses controlled by dendrite-targeting somatostatin-expressing (SST+) interneurons. Spontaneous inputs to PN dendrites are suppressed by SST+ interneurons, whereas salient sensory stimuli transiently alleviate SST+ interneuron–mediated inhibition of PN dendrites likely through VIP+ interneurons. In most cases, this evokes highly correlated somatic and dendritic sensory responses. However, sensory input can also lead to isolated dendritic responses without concomitant somatic output, indicating that dendrites of LA PNs can integrate auditory inputs locally. The relief of SST+ interneuron–mediated dendritic inhibition is necessary to amplify dendritic CS responses during conditioning, which is consistent with the notion that disinhibition through SST+ interneurons opens a temporal window during which CS inputs are eligible for the induction of associative dendritic plasticity upon concomitant exposure to an aversive US. Fear conditioning induces bidirectional plasticity of somatic CS responses that correlates with learning at the behavioral level. On average, dendritic CS responses increase in neurons with both up-regulated (CSup) or down-regulated (CSdown) somatic CS responses. However, fear conditioning also increases the variance of CS responses across the dendritic tree of a given neuron, indicating that not all dendrites are undergoing similar levels of plasticity during learning. Moreover, dendritic spines that show up-regulated CS responses are more likely to be located on dendrites that exhibit increased CS responses after learning. Last, even though dendritic CS responses of CSdown neurons are potentiated, their somatic responses are reduced by enhanced Parvalbumin-expressing (PV+) interneuron–mediated perisomatic inhibition, counteracting the learning-induced increased synaptic drive. CONCLUSION Our findings demonstrate that LA PNs locally integrate dendritic sensory inputs in a compartmentalized manner and that fear conditioning–induced plasticity of dendritic and somatic sensory responses can be uncoupled. Both compartmentalized dendritic integration and uncoupling of dendritic and somatic plasticity are regulated by local inhibitory circuits that specifically target dendrites or the perisomatic region, two distinct subcellular compartments of postsynaptic PNs. The regulation of compartmentalized dendritic and somatic plasticity increases the computational capacity of amygdala circuits, possibly enhancing an animal’s behavioral flexibility in the face of danger. Fear conditioning induces compartmentalized plasticity. ( Left ) During auditory stimuli, dendritic disinhibition through SST+ interneurons amplifies dendritic Ca 2+ responses. ( Right ) Fear conditioning induces a correlated increase of somatic and dendritic CS responses in a population of LA PNs (CSup neurons), whereas CSdown neurons show increased dendritic but decreased somatic CS responses. This compartmentalized plasticity is mediated by soma-targeting PV+ interneurons.
Type of Medium:
Online Resource
ISSN:
0036-8075
,
1095-9203
DOI:
10.1126/science.abf7052
Language:
English
Publisher:
American Association for the Advancement of Science (AAAS)
Publication Date:
2022
detail.hit.zdb_id:
128410-1
detail.hit.zdb_id:
2066996-3
detail.hit.zdb_id:
2060783-0
SSG:
11
Permalink