Physiological plasticity in auditory cortex: Rapid induction by learning

The amygdala, cholinergic basal forebrain, and higher-order auditory cortex (HO-AC) regulate brain-wide plasticity underlying auditory threat learning. Here, we perform multi-regional extracellular recordings and optical measurements of acetylcholine (ACh) release to characterize the development of discriminative plasticity within and between these brain regions as mice acquire and recall auditory threat memories. Spiking responses are potentiated for sounds paired with shock (CS+) in the lateral amygdala (LA) and optogenetically identified corticoamygdalar projection neurons, although not in neighboring HO-AC units. Spike- or optogenetically triggered local field potentials reveal enhanced corticofugal—but not corticopetal—functional coupling between HO-AC and LA during threat memory recall that is correlated with pupil-indexed memory strength. We also note robust sound-evoked ACh release that rapidly potentiates for the CS+ in LA but habituates across sessions in HO-AC. These findings highlight a distributed and cooperative plasticity in LA inputs as mice learn to reappraise neutral stimuli as possible threats.

Long-term potentiation (LTP) and long-term depression (LTD) are key cellular processes that support memory formation. Whereas increases of synaptic strength by means of LTP may support the creation of a spatial memory ‘engram’, LTD appears to play an important role in refining and optimising experience-dependent encoding. A differentiation in the role of hippocampal subfields is apparent. For example, LTD in the dentate gyrus (DG) is enabled by novel learning about large visuospatial features, whereas in area CA1, it is enabled by learning about discrete aspects of spatial content, whereby, both discrete visuospatial and olfactospatial cues trigger LTD in CA1. Here, we explored to what extent local audiospatial cues facilitate information encoding in the form of LTD in these subfields. Coupling of low frequency afferent stimulation (LFS) with discretely localised, novel auditory tones in the sonic hearing, or ultrasonic range, facilitated short-term depression (STD) into LTD (>24 h) in CA1, but not DG. Re-exposure to the now familiar audiospatial configuration ca. 1 week later failed to enhance STD. Reconfiguration of the same audiospatial cues resulted anew in LTD when ultrasound, but not non-ultrasound cues were used. LTD facilitation that was triggered by novel exposure to spatially arranged tones, or to spatial reconfiguration of the same tones were both prevented by an antagonism of the metabotropic glutamate receptor, mGlu5.

These data indicate that, if behaviourally salient enough, the hippocampus can use audiospatial cues to facilitate LTD that contributes to the encoding and updating of spatial representations. Effects are subfield-specific, and require mGlu5 activation, as is the case for visuospatial information processing. These data reinforce the likelihood that LTD supports the encoding of spatial features, and that this occurs in a qualitative and subfield-specific manner. They also support that mGlu5 is essential for synaptic encoding of spatial experience.

This article is part of the Special Issue entitled ‘Metabotropic Glutamate Receptors, 5 years on’.

To thrive in a changing environment, organisms evolved strategies for rapidly modifying their behavioral responses to sensory stimuli. In this review, we investigate the role of sensory cortical circuits in these flexible behaviors. First, we provide a framework for classifying tasks in which flexibility is required. We then present studies in animal models which demonstrate that responses of sensory cortical neurons depend on the expected outcome associated with a stimulus. Last, we discuss inactivation studies which indicate that sensory cortex facilitates behavioral flexibility, but is not always required for adapting to changes in environmental conditions. This analysis provides insights into the contributions of cortical and subcortical sensory circuits to flexibility in behavior.

Primary (“early”) sensory cortices have been viewed as stimulus analyzers devoid of function in learning, memory, and cognition. However, studies combining sensory neurophysiology and learning protocols have revealed that associative learning systematically modifies the encoding of stimulus dimensions in the primary auditory cortex (A1) to accentuate behaviorally important sounds. This “representational plasticity” (RP) is manifest at different levels. The sensitivity and selectivity of signal tones increase near threshold, tuning above threshold shifts toward the frequency of acoustic signals, and their area of representation can increase within the tonotopic map of A1. The magnitude of area gain encodes the level of behavioral stimulus importance and serves as a substrate of memory strength. RP has the same characteristics as behavioral memory: it is associative, specific, develops rapidly, consolidates, and can last indefinitely. Pairing tone with stimulation of the cholinergic nucleus basalis induces RP and implants specific behavioral memory, while directly increasing the representational area of a tone in A1 produces matching behavioral memory. Thus, RP satisfies key criteria for serving as a substrate of auditory memory. The findings suggest a basis for posttraumatic stress disorder in abnormally augmented cortical representations and emphasize the need for a new model of the cerebral cortex.

Extinction of auditory fear conditioning induces a temporary inhibition of conditioned fear responses that can spontaneously reappear with the passage of time. Several lines of evidence indicate that extinction learning relies on the recruitment of specific neuronal populations within the basolateral amygdala. In contrast, post-extinction spontaneous fear recovery is thought to result from deficits in the consolidation of extinction memory within prefrontal neuronal circuits. Interestingly, recent data indicates that the strength of gamma oscillations in the basolateral amygdala during auditory fear conditioning correlates with retrieval of conditioned fear responses. In the present manuscript we evaluated the hypothesis that post-extinction spontaneous fear recovery might depend on the maintenance of gamma oscillations within the basolateral amygdala by using single unit and local field potential recordings in behaving mice. Our results indicate that gamma oscillations in the basolateral amygdala were enhanced following fear conditioning, whereas during extinction learning gamma profiles were more heterogeneous despite similar extinction learning rates. Remarkably, variations in the strength of gamma power within the basolateral amygdala between early and late stages of extinction linearly predicted the level of post-extinction spontaneous fear recovery. These data suggest that maintenance of gamma oscillations in the basolateral amygdala during extinction learning is a strong predictive factor of long term spontaneous fear recovery.