Neurotensin Modulates Emotional Valence Assignment in the Basolateral Amygdala Through Neuromodulator Gain

Based on previous experience, learning to avoid or seek certain specific stimuli again in the future is crucial for survival. Our brains are wired to assign a particular valence—either positive or negative—as a result of sensory stimuli, and it is this valence that serves as the foundational motivation for our subsequent actions. Simply put, all motivational actions fall into two categories: pleasure-seeking behavior guided by positive emotional valence, and pain-avoiding behavior driven by negative emotional valence [1]. The ability to shift from one emotional valence to another is an important characteristic of affective states, while the instability of emotional states underlies many psychiatric disorders, highlighting the clinical importance of managing and understanding these fluctuations. This ability to adapt emotional responses can be attained by modulating the gain across distinct neural pathways, thus enabling the nuanced and smooth assignment of valence through the strengthening or weakening of circuit activity [1, 2].

The amygdala has long been a focus of research on emotions [3–5]. It is involved in processing and integrating various sensory information [6]. Within the amygdala, the basolateral nucleus (BLA) can also process input information and output it to different nuclear complexes to regulate both positive and negative emotions [3]. Previous studies have reported that the projection from the BLA to the nucleus accumbens (NAc) is associated with reward learning and exciting this circuit facilitates reward-seeking behavior [7, 8]. On the other hand, the projection from the BLA to the central amygdala (CeA), especially its medial division (CeM), is associated with punishment learning [9]. Yet the CeA plays a key role in processing both sensory and negative emotional-affective components of neuropathic pain, and long-term depression (LTD) at the amygdaloid BLA-CeA synapse underlies the comorbid aversive and depressive symptoms in neuropathic pain [10]. The functional differences between these two types of BLA neurons may be attributed to changes in synaptic electrophysiological properties and gene expression. To confirm this, Namburi et al. labeled these two types of neurons in the BLA by injecting retrograde viruses into the NAc and CeM separately and conducted associative learning tasks in mice. The mice were trained to associate a sound with a reward (positive valence) or an electric shock (negative valence) [11]. The results showed that the relative synaptic strength of the BLA-CeM projection increased after punishment learning and decreased after reward learning, while the BLA-NAc projection exhibited an inverse pattern. Strikingly, upon conducting RNA sequencing on discrete neuronal populations, it emerged that a scant number of genes exhibited variant expression levels. In addition, RNA sequencing of these neuron populations revealed that only a small set of genes, including the neurotensin receptor 1 gene (Ntsr1), showed differing expression levels, suggesting a genetic basis for the development of these distinct valence learning processes. Moreover, the perpetual mutability of our natural environs needs a versatile modulation between emotional states spanning the valence spectrum, reflective of the complex adaptations that occur as a result of experiential learning.

The ceaselessly dynamic expanse of our world is fraught with countless potential rewards and perils. These diverse positive and negative experiences prompt changes in the electrophysiological properties of distinct neuronal clusters within the BLA. Such transformations guide the information flow through the BLA [2], consequently influencing the processing of emotional valence. For animals to survive in this challenging world, adopting flexible behavioral strategies is essential. The heterogeneity of neurons in the BLA and the opposing behaviors driven by valence-specific neurons projecting to different downstream targets may form the physiological basis for the flexibility of behavioral selection [11, 12].

From the perspective of spatial distribution, BLA neurons projecting to different downstream targets are intermingled throughout the nucleus. BLA-CeA neurons encoding negative valence exhibit higher density in the dorsal region, while BLA-NAc neurons encoding positive valence are more concentrated in the ventral part. Furthermore, these projection terminals form local microcircuits that interact with each other. Projection-defined BLA populations have different impacts on the local networks recruited. The BLA-CeA projection neurons are more likely to suppress the activity of surrounding neurons, while the BLA-NAc projection neurons are more likely to promote activity [13]. This flexibility in recruiting other neurons may depend on the influence of inputs from the paraventricular thalamus (PVT) to the internal state of the BLA (Fig. 1A), which contains both neurotensin (NTs) and glutamate [14, 15]. However, the synaptic-level interactions between neural circuits and their changes under different behavioral states are still unknown. It has been speculated that neuropeptides or neuromodulation signals targeting different receptor sub-types or even receptor expression levels can increase or decrease the relative gain of signal propagation along different pathways, thereby selectively conveying plasticity to neurons encoding positive or negative valence. There are gene expression differences between the populations of BLA neurons projecting to the NAc and CeA, and Ntsr1 is more abundantly expressed in the BLA neurons projecting to the CeA than in the BLA-NAc neurons [16]. Furthermore, the NTs signal in the BLA has been shown to influence long-term potentiation and fear learning, suggesting that NTs released by PVT neurons can encode valence information by modifying synaptic plasticity and guide the allocation of valence information in the BLA.

Fig. 1.

Role of the basolateral amygdala (BLA) in valence assignment. A Simplified diagram of how the BLA receives input from neurotensin (NTs) projection neurons from the paraventricular thalamus (PVT) and influences the divergent transmission of the valence information to the nucleus accumbens (NAc) and the central amygdala (CeA). Neurons projecting to the CeA (BLA-CeA) are associated with negative valence (electrical shock), while neurons projecting to the NAc (BLA-NAc) are associated with positive valence (sucrose). B More specific details of the synaptic strength dynamics in different BLA projection neurons. Synaptic strength in the BLA-CeA pathway decreases with punishment learning and that in the BLA-NAc pathway increases with reward learning. Ntsr1, Neurotensin receptor 1; AMPAR, Alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; NMDAR, N-methyl-D-aspartic acid receptor.

In order to further investigate the specific effects of the PVT projection to BLA NTs neurons on the allocation of emotional valence in mice, Li et al. ingeniously combined retrograde tracing virus technology with the CRISPR-Cas9 method to selectively knock out NTs in this neuronal pathway (injecting retrograde virus carrying NTs guide gene downstream of the circuit and injecting anterograde virus carrying spCas9 upstream) [16]. This manipulation led to diminished responsiveness in mice to both reward and punishment stimuli, suggesting an alteration in their assessment of the stimuli’s valence. In addition, to selectively monitor the dynamic changes in NTs concentration in BLA neurons during learning processes, they collaborated with Yulong Li to develop a genetically encoded fluorescent NTs sensor, GRAB_NTs1.0. The results revealed an increase in NTs concentration after reward learning and a decrease after punishment learning. Li et al. speculated that when the valence of environmental stimuli is uncertain, NTs concentration decreases, but it increases or further decreases when positive or negative valence is detected. Activation of this pathway through optogenetic techniques was found to enhance reward learning, while inhibition promoted punishment learning, indicating that NTs in the BLA neuronal cluster guide the allocation of emotional valence in the BLA through concentration-dependent modulation of glutamatergic signaling (Fig. 1B).

In conclusion, the NTs neurons in the PVT regulate the activity of glutamatergic neurons in the BLA by modulating the concentration of neurotensin, thus influencing the dominance of the BLA over the downstream nuclei NAc and CeA, and consequently the allocation of emotional valence in associative learning. The implications of this research extend beyond the elucidation of physiological mechanisms, offering a novel locus for therapeutic intervention in the amelioration of emotional dysregulation syndromes including anxiety, addiction, and the like. Furthermore, this study raises other issues worthy of further exploration. Firstly, the BLA contains functionally opposing types of neurons and they branch out to each other, a process known as collateralization [13, 17]. These projection branches may facilitate cross-regional information transmission and regulation. This could enable the transmission of emotional valence signals to other related neural networks, potentially affecting emotion formation and regulation. Secondly, targeted manipulation of neural assemblies during the processes of associative learning and memory retention could precipitate a recalibration of inherent emotional biases, reversing naturally occurring valences. One conceivable application of this might be the attenuation of the predisposition towards negative memory in individuals suffering from depression, ultimately diminishing their experience of adverse emotions [18]. Thirdly, the inherent plasticity of neural circuits responsible for the distribution of emotional weight could underpin the conceptualization and creation of sophisticated artificial neural networks in forthcoming technological advances [19]. To illustrate, Lin et al. established a biomimetic neural network structure based on the innate fear information processing circuit induced by visual stimuli in the mouse brain [20]. Overall, these findings advance the understanding of affective neuroscience and open the door to potential therapeutic interventions for emotional dysregulation disorders, enhancing our grasp of the neural circuitry underpinning emotional valence processing and adaptive behaviors.

Conflict of interest

The authors declare no competing interests.