A Nucleus Accumbens Tac1 Neural Circuit Regulates Avoidance Responses to Aversive Stimuli

Neural circuits that control aversion are essential for motivational regulation and survival in animals. The nucleus accumbens (NAc) plays an important role in predicting aversive events and translating motivations into actions. However, the NAc circuits that mediate aversive behaviors remain elusive. Here, we report that tachykinin precursor 1 (Tac1) neurons in the NAc medial shell regulate avoidance responses to aversive stimuli. We show that NAcTac1 neurons project to the lateral hypothalamic area (LH) and that the NAcTac1→LH pathway contributes to avoidance responses. Moreover, the medial prefrontal cortex (mPFC) sends excitatory inputs to the NAc, and this circuit is involved in the regulation of avoidance responses to aversive stimuli. Overall, our study reveals a discrete NAc Tac1 circuit that senses aversive stimuli and drives avoidance behaviors.


Introduction
Reward and aversion are critical for motivated behaviors and are associated with many mood disorders. Unexpected stimuli and threats drive aversive behaviors, an innate response crucial to the survival of animals [1]. Aversive stimuli engage negative emotions and contribute to prominent psychiatric disorders. Enormous advances have been made in understanding the neural circuits underlying reward [2][3][4][5][6][7]. However, the neural circuits underlying aversion remain elusive.
It is widely thought that the nucleus accumbens (NAc) is a critical brain region in the reward and aversion circuits that integrate different inputs, leading to motivated behaviors [8][9][10][11][12][13][14]. Anatomically, the NAc can be divided into the core, lateral shell, and medial shell [15]. It has been found that distinct NAc neural circuits are involved in different brain functions [16][17][18][19][20][21]. Dopamine transmissions from the ventral tegmental area have been linked to reward and aversion processing [18]. Glutamatergic inputs from the thalamic paraventricular nucleus to the NAc regulate aversion [19,22]. How the NAc regulates opposite behaviors at the same time remains elusive. Thus, it is worth investigating whether distinct NAc subregions are included in discrete neural circuits involved in aversion.
The NAc has received attention as a crucial convergence point of reward and aversion circuits, as it receives multiple projections from the ventral tegmental area (VTA), medial prefrontal cortex (mPFC), basolateral amygdala (BLA), and hippocampus [33,34]. The mPFC is strongly related to neural circuits encoding aversion and decision making [35][36][37]. The prelimbic and infralimbic regions of the mPFC have been implicated in aversion [38][39][40]. However, studies have yielded conflicting findings. How the mPFC regulates aversion through specific neural circuits remains underexplored.
Here, we show that tachykinin precursor 1 (Tac1) neurons in the NAc medial shell mediate avoidance responses to aversive stimuli. Neural tracing and electrophysiological data show that NAc Tac1 neurons project inhibitory signals to the lateral hypothalamic area (LH) and modulate avoidance behavior in the presence of aversive stimuli. Additionally, neurons in the NAc medial shell receive inputs from mPFC glutamatergic (mPFC Glut ) neurons, and optogenetic manipulation of the mPFC Glut → NAc circuit regulates aversive behaviors. These results indicate the essential role of Tac1 neurons in encoding aversive stimuli and regulating behavioral responses.

NAc Tac1 Neurons Regulate Avoidance Behavior in Response to Aversive Stimuli
To investigate the expression of Tac1 neurons in the NAc, we crossed the Tac1-internal ribosome entry site 2 (IRES2)-Cre mouse line [41] with a Cre-dependent tdTomato reporter line, Ai9 [42] ( Figure 1A). We observed that Tac1-tdTomato cellular expression closely matched endogenous substance P and Dopamine Receptor 1 ( Figure 1B-G). To mimic aversion in mice, Tac1-Cre male mice were given an injection of formalin in the plantar surface of a hindpaw, as previously described [43,44]. Patch-clamp recordings were performed on Tac1 neurons in the NAc ( Figure 1H). We observed decreased excitability of Tac1 neurons in the medial shell, but not in the lateral shell (Figures 1I,J and S1A,B). These data indicate that Tac1 neurons in the NAc medial shell are involved in the circuit regulating aversion.
To determine whether NAc Tac1 neurons in the NAc medial shell regulate aversive behaviors, we performed chemogenetics using designer receptors exclusively activated by designer drugs (DREADDS). To selectively manipulate the activity of Tac1 neurons, we bilaterally injected AAV-DIO-hM3D(Gq)-mCherry, AAV-DIO-hM4D(Gi)-mCherry, and AAV-DIO-mCherry into the NAc medial shell of Tac1-Cre male mice ( Figure 1K). Formaldehyde has been shown to act as an unfamiliar aversive stimulus for rodents without altering their motor activity [45]. We thus measured the approach-avoidance behaviors of male mice to an aversive stimulus (formaldehyde) while inhibiting or activating the activity of NAc Tac1 neurons. A piece of cotton dipped in 5% formaldehyde was placed on one side of a three-chamber arena. Mice tend to explore a novel object, but animals display strong avoidance behaviors when exposed to formaldehyde. Mice were introduced into the chamber containing formaldehyde. Interactions with formaldehyde were recorded for 5 min. The opposite chambers of the arena were designated the 'safe' area and 'center' area. We observed that hM3D(Gq)-injected mice spent significantly more time exploring the aversive stimulus than hM4D(Gi)-and mCherry (control)-injected mice ( Figure 1L-O). To investigate whether the activity of Tac1 neurons regulates interactions with a neutral stimulus in mice, we performed the approach experiment and replaced the piece of 5% formaldehyde cotton with a piece of regular cotton. We found that the activity of NAc Tac1 neurons did not affect time spent interacting with neural stimuli ( Figure S2A-C). Moreover, olfaction and locomotion were not affected by hM4D(Gi) or hM3D(Gq) injection ( Figure S3A,B). These results suggest that NAc Tac1 neurons in the medial shell are crucial to avoidance behaviors in response to aversive stimuli.   Tac1-IRES2-Cre and Ai9-tdTomato reporter mice to label Tac1-positive neurons in the NAc, compared to the Allen Institute for Brain Science. Scale bar: 500 µm and 100 µm, respectively. D: dorsal; V: ventral; L: lateral; M: medial. (B) Representative image of acute striatal slice from Tac1-Cre; Ai9 mice stained with antibodies against substance P. Scale bar: 20 µm. (C) Percentage of Tac1positive neurons that co-localized with substance P; co-localization ratio (TdT/DAPI: 43.55 ± 4.096%, SP/DAPI: 43.55 ± 4.096%, TdT/SP: 100%; one-way ANOVA test, F (2,12) = 1.669, p < 0.0001). N = 5. (D) Representative image of acute striatal slice from Tac1-Cre; Ai9 mice stained with antibodies against NeuN. Scale bar: 20 µm. (E) Percentage of Tac1-positive neurons that co-localized with NeuN; co-localization ratio (TdT/DAPI: 43.90 ± 4.801%, NeuN/DAPI: 77.43 ± 3.455%, TdT/NeuN: 54.72 ± 5.014%; one-way ANOVA test, F (2,9)  NAc msh: nucleus accumbens medial shell. All data are means ± s.e.m. * p < 0.05; ** p < 0.01; *** p < 0.001; n.s.: not significant.

NAc Tac1 Neurons Project to the LH
To identify possible downstream targets of NAc Tac1 neurons that may encode aversive stimuli, we injected AAV-DIO-mCherry into the NAc medial shell of Tac1-Cre mice. Four weeks later, the animals were euthanized, and the distribution of neurons that NAc Tac1 neurons target in the brain was examined (Figure 2A-C). The whole-brain mapping results indicated that dense mCherry-labeled terminals were found in the lateral hypothalamic area (LH) ( Figure 2D-F, Figure S4).

NAc Tac1 -to-LH Projection Mediates Avoidance Behaviors in Response to Aversive Stimuli
To assess whether the NAc Tac1 →LH circuit regulates avoidance responses to aversive stimuli. Male Tac1-Cre mice were unilaterally injected with AAV-DIO-mCherry and AAV-DIO-ChR2-mCherry ( Figure 3A,B). Six weeks later, we carried out an approachavoidance assay for evaluation ( Figure 3C). A piece of cotton dipped in 5% formaldehyde

NAc Tac1 -to-LH Projection Mediates Avoidance Behaviors in Response to Aversive Stimuli
To assess whether the NAc Tac1 →LH circuit regulates avoidance responses to aversive stimuli. Male Tac1-Cre mice were unilaterally injected with AAV-DIO-mCherry and AAV-DIO-ChR2-mCherry ( Figure 3A,B). Six weeks later, we carried out an approach-avoidance assay for evaluation ( Figure 3C). A piece of cotton dipped in 5% formaldehyde was placed in one corner of a square chamber. Mice were introduced into the chamber. Their interactions with formaldehyde were recorded. Compared with the control stimulation, the selective delivery of blue light (5 ms pulse, 20 Hz for 5 min) to the LH of ChR2-expressing terminals elicited a significant increase in interaction time with formaldehyde ( Figure 3D,E). We also calculated the total distance traveled by the mice in the arena and found that locomotion was not affected ( Figure 3F). was placed in one corner of a square chamber. Mice were introduced into the chamber. Their interactions with formaldehyde were recorded. Compared with the control stimulation, the selective delivery of blue light (5 ms pulse, 20 Hz for 5 min) to the LH of ChR2expressing terminals elicited a significant increase in interaction time with formaldehyde ( Figure 3D,E). We also calculated the total distance traveled by the mice in the arena and found that locomotion was not affected ( Figure 3F).  We then selectively inhibited the NAc Tac1 terminals in the LH by delivering continuous yellow light to the LH of male mice bilaterally infected with AAV-DIO-NpHR-eYFP in the NAc medial shell ( Figure 3G,H). In the approach-avoidance assay, the photoinhibition of NAc Tac1 →LH projection significantly decreased interaction time with formaldehyde without affecting locomotion ( Figure 3I-L). We also carried out a real-time place aversion (RTPA) assay and found that the photoinhibition of NAc Tac1 →LH projection elicited avoidance of the photoinhibition-paired chamber ( Figure S6A-D). Taken together, these data indicate that the NAc Tac1 →LH circuit is crucial to avoidance behaviors in response to aversive stimuli.

mPFC Glut Inputs Activate NAc Neurons
Next, we sought to identify upstream brain regions of NAc Tac1 neurons that might mediate aversive behaviors. We employed a monosynaptic viral tracing strategy in Tac1-Cre mice. The NAc medial shell of Tac1-Cre mice was injected with AAV-DIO-RVG and AAV-DIO-TVA-GFP. Four weeks later, RV-EnVA-dsRed was injected into the LH ( Figure 4A-C). We found that the medial prefrontal cortex (mPFC) was projected to NAc Tac1 neurons ( Figure 4D). Based on emerging studies [37,38,[46][47][48] showing that the mPFC is critical for neural circuits of aversion, we focused on neurons in the mPFC projecting to NAc Tac1 neurons. To determine the kinds of mPFC neuron that are involved in the NAc Tac1 circuit, we performed immunofluorescence experiments and found that mCherry-labeled neurons co-expressed the glutamatergic marker VGLUT2 (Figures 4E and S7).
We next evaluated the synaptic function of mPFC Glut neurons projecting to NAc neurons. We first expressed ChR2 in mPFC Glut neurons by injecting AAV-CaMKIIα-ChR2-eYFP into the mPFC, and then, selectively activated NAc neurons that were receiving projections from mPFC Glut neurons via optogenetic stimulation (5 ms pulse, 20 Hz) ( Figure 4F,G). In the whole-cell patch-clamp configuration, excitatory postsynaptic currents (EPSCs) were recorded in 32 out of 60 NAc neurons ( Figure 4H). However, no inhibitory postsynaptic currents were recorded. EPSCs were eliminated via pretreatment with the AMPA receptor antagonist CNQX (Figures 4I,J and S8). To identify whether Tac1 neurons in the NAc medial shell received inputs from the mPFC Glut , we expressed ChR2 in mPFC Glut neurons in Tac1-Cre; Ai9 mice. In the patch-clamp recording, EPSCs were recorded in Tac1 neurons in the NAc medial shell ( Figure S9A-D). These results suggest that mPFC Glut neurons project excitatory signals to Tac1 neurons in the NAc medial shell. We next evaluated the synaptic function of mPFC Glut neurons projecting to NAc neurons. We first expressed ChR2 in mPFC Glut neurons by injecting AAV-CaMKIIα-ChR2-eYFP into the mPFC, and then, selectively activated NAc neurons that were receiving projections from mPFC Glut neurons via optogenetic stimulation (5 ms pulse, 20 Hz) ( Figure  4F,G). In the whole-cell patch-clamp configuration, excitatory postsynaptic currents (EP-SCs) were recorded in 32 out of 60 NAc neurons ( Figure 4H). However, no inhibitory postsynaptic currents were recorded. EPSCs were eliminated via pretreatment with the

The mPFC Glut -to-NAc Circuit Modulates Avoidance Behaviors in Response to Aversive Stimuli
To investigate whether activation of the mPFC Glut →NAc circuit decreases avoidance behaviors in response to aversive stimuli. Male mice were unilaterally injected with AAV-CaMKIIα-eYFP and AAV-CaMKIIα-ChR2-eYFP ( Figure 5A,B). Six weeks later, we carried out an approach-avoidance assay for evaluation ( Figure 5C). Compared with the control stimulation, the selective delivery of blue light (5 ms pulse, 20 Hz for 5 min) to ChR2-expressing terminals in the NAc medial shell elicited a significant increase in interaction time with formaldehyde ( Figure 5D,E). We also calculated the total distance traveled by mice in the arena and found that locomotion was not affected ( Figure 5K). mPFC Glut neurons project excitatory signals to Tac1 neurons in the NAc medial shell.

The mPFC Glut -to-NAc Circuit Modulates Avoidance Behaviors in Response to Aversive Stimuli
To investigate whether activation of the mPFC Glut →NAc circuit decreases avoidance behaviors in response to aversive stimuli. Male mice were unilaterally injected with AAV-CaMKIIα-eYFP and AAV-CaMKIIα-ChR2-eYFP ( Figure 5A,B). Six weeks later, we carried out an approach-avoidance assay for evaluation ( Figure 5C). Compared with the control stimulation, the selective delivery of blue light (5 ms pulse, 20 Hz for 5 min) to ChR2expressing terminals in the NAc medial shell elicited a significant increase in interaction time with formaldehyde ( Figure 5D,E). We also calculated the total distance traveled by mice in the arena and found that locomotion was not affected ( Figure 5K). Next, we selectively inhibited mPFC Glut terminals in the NAc medial shell by delivering continuous yellow light to the LH of male mice bilaterally infected with AAV-CaMKIIα-eNpHR3-mCherry in the mPFC ( Figure 5G,H). In the approach-avoidance assay, inhibition of the mPFC Glut →NAc pathway significantly reduced interaction time with formaldehyde without affecting locomotion ( Figure 5I-L). We also carried out a real-time place aversion (RTPA) assay and found that inhibition of the mPFC Glut →NAc pathway elicited avoidance of the photoinhibition-paired chamber ( Figure S10A-D).
Furthermore, we determined whether the activation of mPFC GLUT neurons could attenuate aversive behaviors following the inhibition of NAc Tac1 neurons. AAV-CaMKIIα-hM3D(Gq)-mCherry or AAV-CaMKIIα-mCherry was injected in the mPFC, while AAV-DIO-hM4D(Gi)-mCherry was injected in the NAc ( Figure S10A). We found that the activation of the mPFC neurons was able to attenuate avoidance behaviors in response to aversive stimuli ( Figure S11B,C). Taken together, these data indicate that the mPFC Glut →NAc circuit is crucial to avoidance behaviors in response to aversive stimuli.

Discussion
Using neural circuit tracing, chemogenetics, electrophysiology, and optogenetics approaches, we found that NAc Tac1 neurons in the medial shell mediate avoidance responses to aversive stimuli in 10-14-week-old male mice. NAc Tac1 neurons send inhibitory inputs to the LH, and the Nac Tac1 →LH circuit is required for aversive behaviors in mice. Moreover, Nac neurons receive glutamatergic inputs from mPFC Glut neurons, and mPFC Glut →Nac projection regulates behavioral responses in the presence of aversive stimuli.
Our mapping study of output circuits demonstrated that Nac Tac1 neurons in the medial shell project to LH neurons. The LH is a brain region that contains heterogeneous cell populations [49] and is involved in the regulation of multiple behaviors, such as feeding, aversion, and reward-seeking [50,51]. Previous studies have reported that the activation of LH neurons causes avoidance and aversive behaviors [52][53][54]. This is consistent with our data, which show that the optogenetic inhibition of NAc Tac1 terminals in the LH induced aversive behaviors in mice. However, the LH receives multiple excitatory and inhibitory inputs from both cortical and subcortical structures, further research is needed to fully resolve the neuron populations receiving inhibitory inputs from NAc Tac1 neurons and the mechanisms underlying aversion in the LH.
Previous studies have reported that mPFC neurons project to the NAc and that these neurons are able to elicit avoidance [55]. It has also been reported that projections from the mPFC to the NAc have no effect on aversion [37]. These conflicting results may partially be caused by the heterogeneity of the different regions of the NAc: the core, medial shell, and lateral shell. These regions contain similar classes of medial spiny projection neurons (MSNs). NAc medial shell MSNs have been described as "medium-small spiny neurons" with low density [56]. In addition, the NAc medial shell shows a perplexing phenotype that opposes the classical direct and indirect pathway model [16,57,58]. In addition, previous work indicates that D1 neurons in the NAc also represent a portion of the classical indirect pathway and are activated by aversive stimuli [59,60]. Taken together, the classical striatal direct and indirect pathway models are not applied to the NAc. In this study, we found that mPFC Glut neurons send excitatory signals to neurons in the NAc medial shell. Moreover, this mPFC Glut →NAc circuit is involved in the regulation of aversion. However, aversion is a multidimensional construct, and we cannot rule out the possibility that other neurons in the NAc receive inputs from the mPFC and contribute to aversive behaviors.
In summary, we delineated distinct Tac1 neurons as encoding aversive stimuli. Furthermore, we dissected the dedicated function of this circuit and identified it as a critical component of the aversion circuit. These results may improve our understanding of the aversion circuit. By understanding the structure and mechanisms underlying aversion and negative prediction, it will be possible to design intervention strategies for pathological depressive conditions.

Animals
All experimental procedures were approved by the Animal Advisory Committee of Northeast Normal University, China. The laboratory was kept under specific pathogen-free (SPF) conditions. All mice were maintained on a 12-12 h light-dark cycle (lights on from 6:00 to 18:00 every day), with food and water provided ad libitum. All behavioral tests were performed during the light period. C57BL/6J mice were obtained from Huafukang Animal Center, Beijing, China. Tac1-IRES2-Cre mice (Jax No. 021877) were obtained from Jackson Laboratory (USA). Ai9 mice (Jax No. 007905) were kindly provided by Prof. Chunjie Zhao from Southeast University.

Viral Tracing
For output mapping, the NAc medial shell of Tac1-Cre mice was injected with with AAV-EF1α-DIO-mCherry (200 nL). For input mapping, AAV-EF1α-DIO-His-EGFP-2a-TVA and AAV-EF1α-DIO-RG (1:1, total 150 nL) were injected into the NAc medial shell of Tac1-Cre mice. Four weeks later, 300 nL of RV-ENVA-∆G-dsRed was injected into the LH. Thus, we only infected NAc Tac1 neurons in the medial shell that projected to the LH, and traced their inputs. The mice were sacrificed one week after RV injection.

Stereotaxic Injection
Mice were anesthetized with 1.0% sodium pentobarbital (0.1 g/kg body weight, i.p.). Viruses were delivered at a rate of 100 nL/min using a stereotaxic instrument (RWD Co, Shenzhen China) and a 5 µL syringe (Hamilton, Sigma, USA). After each injection, the syringe was left in place for 15 min, and then, slowly withdrawn. Experiments were performed at least 4-6 weeks after virus injection.
Stereotaxic coordinates were derived from the Paxinos and Franklin Mouse Brain Atlas and empirically adjusted. The coordinates for injection into the NAc medial shell (total volume of 400 nL) were +1.9 mm AP, ±0.6 mm ML, and −4.4 mm DV. The coordinates for injection into the LH (total volume of 150 nL) were −1.5 mm AP, ±0.9 mm ML, and −5.1 mm DV. The coordinates for injection into the mPFC (total volume of 400 nL) were +2.2 mm AP, ±0.3 mm ML, and −1.35 mm DV. For monosynaptic circuit tracing and the ChR2 experiment, viruses were delivered unilaterally. For other functional analysis, viruses were delivered bilaterally.
Whole-cell patch-clamp recordings were performed using an EPC-10/2 amplifier (HEKA, Germany). The recording pipettes were pulled from borosilicate glass tubes (Sutter Instruments, USA) and had a resistance of 3-6 MΩ; only whole-cell patches with a series resistance < 15 MΩ were used for recordings. EPSC and IPSC were recorded by holding the membrane potential at −70 mV.

Behavioral Assays
All mice used for the behavioral assays were male mice and their littermates. An experimenter blinded to the genotypes performed all the tests.

Approach-Avoidance Test
The avoidance test was conducted to measure avoidance of an unfamiliar aversive stimulus. For the chemogenetics experiments, control-, hM3D(Gq)-, and hM4D(Gi)-injected mice were i.p. injected with clozapine N-oxide (CNO; 5 mg/kg or JHU37160; 0.5 mg/kg), and introduced into the chamber half an hour later. The chamber (70 cm × 70 cm) contained three sides (safe, center, and form). A piece of cotton dipped in 5% formaldehyde was placed on the form side. The opposite sides of the chamber were designated the 'safe' area and 'center' area. As previously described [18], mice tend to explore a novel object, but animals display strong avoidance behaviors when exposed to formaldehyde. Interactions with formaldehyde were recorded and analyzed by the EthoVision XT system (Noldus, Wageningen, The Netherlands).
For the optogenetics (ChR2 and NpHR) experiments, after recovery from the surgery for virus injection and fiber implantation, mice were introduced into an arena (40 cm × 40 cm). A piece of cotton dipped in 5% formaldehyde was place in a corner of the arena. Interactions with formaldehyde were recorded in 5 min segments and analyzed using the EthoVision XT system (Noldus, Wageningen, The Netherlands).

RTPA Assay
On the day of habituation, the mice were introduced into a Plexiglas box with two chambers (30 cm × 30 cm × 50 cm each) and allowed to explore the chamber freely for 15 min. One chamber was randomly designated the stimulation side, and the other was designated the non-stimulation side. The time spent in each of the chambers was recorded. Mice that spent more than 60% of the total time in either compartment were excluded from the experiments. On the day of the experiment, the mice were randomly introduced into either chamber and received continuous 589 nm yellow light (or 20 Hz 465 nm blue light) every time they entered the stimulation chamber until they moved into the non-stimulation chamber. The time spent in each chamber was recorded and analyzed using the EthoVision XT system.

Olfaction Test
Before the test, all pellets were removed from the home cage, but the water bottle was kept in place. On the day of the experiments, a mouse was introduced into a clean cage containing clean bedding with a depth of 3 cm. The animal was allowed to explore the arena freely for 5 min. The animal was then transferred to an empty clean cage. In the cage containing the bedding, food was buried approximately 1 cm beneath the surface in a random corner. The surface of the bedding was smoothed out, and the animal was reintroduced into the cage. The latency to find the buried food was recorded. The food was considered uncovered when the mouse started to eat it.

Quantification and Statistical Analyses
All experimental procedures and data analyses were conducted in a blinded manner. The number of replicates (N or n) indicated in the figure legends refers to the number of experimental subjects independently treated in each experiment. All statistical analyses were performed in Graph Pad Prism (GraphPad Software) unless otherwise stated. For normally distributed data, a Student's test and one-way ANOVA, followed by Scheffe's post hoc test, were used to analyze the significance between groups. For non-normally distributed data, Mann-Whitney U-tests were used to calculate the significance between groups. All statistical data can be found in the figure legends. Statistical significance was set at * p < 0.05, ** p < 0.01, and *** p < 0.001. The data are presented as the means ± s.e.m.

Data Availability Statement:
The authors confirm that the data supporting the findings of this study are available within the article and its Supplementary materials. Raw data that support the findings of this study are available from the corresponding author, upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.