Associative memory neurons are recruited in PFC-centered circuits to encode schizophrenia-like behavior by dopaminergic receptor-II

Abstract

The severe stresses induce fear memory and mental disorders including anxiety, depression and schizophrenia. Their molecular and cellular mechanisms are expectedly revealed to develop therapeutic strategies. We aim to identify the stress-induced cellular units and neural circuits that are essential for fear memory and schizophrenia in cerebral cortices by behavior tasks, molecular biology, neural tracing and electrophysiology. The social stress by the resident/intruder paradigm leads to the fear memory specific to a resident CD1 mouse and schizophrenia-like behaviors as well as the synapse interconnections among medial prefrontal, auditory and S1Tr cortical neurons in intruder mice. This stress-induced synapse interconnection enables these cortical neurons be recruited as associative memory neurons that are featured by receiving the convergent synapse innervations from the interconnected areas and encoding the stressful signals including the battle sound and the pain signal from trunk-injury area generated in the social stress. The knockdown of dopaminergic receptor-II in the medial prefrontal cortex precludes the recruitment of associative memory neurons and the formation of fear memory and schizophrenia-like behaviors. Eticlopride as a dopaminergic receptor-II antagonist in the medial prefrontal cortex weakens the activities of associative memory neurons and relieves schizophrenia-like behavior. These associative memory neurons recruited by the social stress in the medial prefrontal, auditory and S1Tr cortices through dopaminergic receptors-II are essential for fear memory and schizophrenia.

Introduction

Schizophrenia as the most severe psychological disorder is featured by the mania including hallucination, delusion, misbelief and weird thought as well as the negative mood including social withdrawal, anhedonia and anxiety [1–6]. Schizophrenia’s etiology is thought of as the interactions between genetic predisposition and environment factors [7–10]. Genetic-correlated development abnormality in the brain elevates individual’s vulnerability to psychological traumas [11–14], such as social violence, abuse, neglect, stressful family relations and deviant communications [12, 15–20]. These psychological traumas in the genetic vulnerable individuals may induce the neuronal deterioration, cognitive impairment and emotion instability during the postnatal development, e.g., schizophrenia [21–28]. In terms of molecular pathogenesis, the dopaminergic synapse transmission in the mesolimbic projection and the mesocortical projection is thought to be imbalanced in schizophrenia patients [29–36]. In clinical practices, the antagonists of dopamine receptors-II have been applied to treat mania-dominant signs for decades [4, 10, 37–42]. However, the role of dopaminergic receptor-II in the formation of schizophrenia-correlated neural circuits is largely unclear. The comprehensive view of cellular infrastructures for schizophrenic pathogenesis remains elusive [43–47].

Physical and psychological stress in social activities may induce fear memories and psychotic deficits in the schizophrenia variety [15, 18, 20, 48–65]. Stress-induced social defeats are thought to be one of critical reasons for schizophrenia [61, 66]. The stress-induced schizophrenia is often associated with the weird memories [67–69] and the waning cognitions including the disorganizations of associative thinking, logical reasoning and working memory that are presumably encoded by prefrontal cortical neurons [70–73]. These data indicate the ongoing link from stress-induced fear memory and anxiety to bipolar disorder and schizophrenia [74, 75]. How those cortical neurons are recruited to encode fear memory and schizophrenia remains unknown. In terms of the locations for the fear memory and schizophrenia, the prefrontal cortex has been found to be correlated [76–78]. Schizophrenia subjects are associated with the abnormality of the prefrontal cortex [72, 79–89]. The prefrontal cortex is considered to be the target of antipsychotics for schizophrenia patients [90]. Based on these data, those memory neurons to encode the stress-induced fear memory and schizophrenia are hypothetically recruited in the prefrontal cortex. The downregulation of prefrontal cortical memory neurons may improve the symptoms and signs of stress-induced schizophrenia.

In addition to the amygdala and the limbic system, schizophrenia-correlated neural circuits include the interactions between the prefrontal cortex and the thalamus which relays exogenous signals to the sensory cortices [36, 91, 92]. Associative memory neurons have been found in the sensory cortices that receive the sensory signals from olfactory, somatic tactile and gustatory systems [93–99] as well as in the prefrontal cortices whose synapse inputs derive from the sensory cortices [100, 101]. The associative memory neurons that encode those signals from the stressful social activity have been recruited in auditory and somatosensory cortices [99, 102, 103]. How the prefrontal cortex and sensory cortices interact each other in social stresses to constitute the interconnected neural circuits and to recruit associative memory neurons for encoding schizophrenia is a primal goal in the present study, especially the role of dopamine receptor-II in the recruitment of schizophrenia -correlated neural circuits and associative memory cells in these cortices. Our study is expected to provide crucial data for drawing a comprehensive diagram of fear memory and schizophrenia.

Our strategies to study these questions are listed below. Intruder C57 mice experienced the social stress by attacks from a resident CD1 mouse in a resident/intruder paradigm [53, 104–109]. The fear memory specific for this resident mouse in the intruder mice was identified by the social interaction test (SIT). Their anxious state was examined by the elevated-plus maze (EPM). Their depressive mood including anhedonia and loss of interest was examined by the sucrose preference test (SPT) and the Y-maze test (YMT). Their schizophrenia-like behaviors were identified by the modified pre-pulse inhibition test (mPPI) and the persecutory delusion test (PDT). In cellular level, the mutual axon projection and synapse formation were examined by neural tracing, and the neuronal responses to the stressful signals including the battle sound and the pain signal in somatic injury regions were recorded by in vivo electrophysiology in the prefrontal cortex and sensory cortices. The synapse interconnections of associative memory neurons among cross-modal cortices were studied by microinjecting adeno- associated viruses (AAV) that carried genes of encoding fluorescent proteins in the source areas of cortices and by detecting the expression of gene-coded fluorescent proteins in their targeted cortices, or the other way around. The recruitment of associative memory neurons was ensured when the cortical neurons morphologically received new synapse contacts between fluorescent- labelled presynaptic axon boutons and postsynaptic spines along with innate synapse contacts on neuronal dendrites in the convergent manner [93, 95, 96] and when the cortical neurons expressed the strengthened spike-encoding in response to stress signals [93, 98]. The roles of dopaminergic receptor-II in the formation of fear memory and schizophrenia signs as well as in the recruitment of associative memory neurons were examined by short-hairpin RNA (shRNA) that was specific to silence those dopaminergic receptor-II mRNAs.

Materials and Methods

The use of animals

Experiments were accorded with the guidelines and regulations by the Administration Office of Laboratory Animal in Beijing, China. All of the experiment protocols were approved by Institutional Animal Care and Use Committee in Administration Office of Laboratory Animal at Beijing, China (B10831).

C57BL/6J-Thy1-YFP mice (Jackson Laboratory, USA) were used in our studies. Glutamatergic neurons in their cerebral brain were genetically labeled by yellow fluorescent protein (YFP) [110, 111]. These mice were accommodated in the sterile barrier facility under the circadian of twelve hours for daytime and night, respectively, with the sufficient food and water. The ambient temperature at 22 ± 2°C and the relative humidity at 55 ± 5% were set for a live condition of the specific pathogen free (SPF). The C57 male mice with well-developed body in their postnatal weeks three were chosen for our experiments in the groups of control and social stress by the resident/intruder paradigm. The reason to use male mice was due to the fact that CD1 resident mice appeared less to attack female intruder C57 mice. The qualified control and intruder mice were also based upon their higher activity and lower anxious state. These mice were taken into the laboratory for them to be familiar with the experimental operators and the training apparatus for one week. During the adaptation period, these C57 mice were allowed to be familiar with the cages for the social interactions without the resident CD1 mouse, such that the appearance of CD1 resident mouse, the attacks from a resident CD1 mouse and the battle sound were new stressful signals for C57 mice in this resident/intruder paradigm. These intruder and control C57 mice had the healthy capability of social interactions, which was measured by placing them in a social interaction cage to collect their self-control data about the stay time in the interaction zone [53, 62]. The anxious state of the C57 mice was examined by using the elevated-plus maze. Those C57 mice, which had the ratio of the stay time in the interaction zone above 0.5, the stay time in the open arms of the elevated-plus maze above 10% and these values consistently within mean±2 SD, were chosen to be qualified mice for our experiments. The criteria are based on the rule of the consistency in the physical measures and psychological state of animals used among experiment groups. These C57 mice were also examined by the Y-maze test, the sucrose preference test, the pre-pulse inhibition test and the persecutory delusion test in the adaptation period to have their self-control data. As illustrated in Figure SM1-2 (figure one and two for supplementary methods), C57 mice in the control group and the intruder subgroups appear well-homogeneity in these tests.

After an adaptation period, these qualified C57 mice were randomly divided into the control group and the social stress group. Either of these groups experienced experiment manipulations in the control for three weeks or in the social stress (a resident/intruder paradigm) once a day for three weeks. Subsequently, the mice were examined in the formation of their fear memories and schizophrenia-like behaviours as well as the recruitment of associative memory cells by multiple disciplinary approaches. The timeline for our experiments in Fig. 1A was the adaptation period for the self-control data, the resident/intruder paradigm and the studies including behavior tasks, neural tracing, electrophysiology in vivo as well as molecular and pharmacological manipulations.

Fig. 1. Social stress elicits fear memory and schizophrenia-like behaviors.

A Schematic overview of the experimental design. B Heatmaps illustrate the activity trace of mice in the open field in response to CD1+ and CD-. No target, the absence of CD1; Target, the presence of CD1. C A barplot of the stay time of mice in the interaction zone (t(34) = 12.91, P < 0.001. n = 18 for Ctrl and Intruder groups mice). D Comparison of the time spent in the open arms (t(38) = 12.57, P < 0.001. Each group n = 20 mice). E Time spent in the interaction-arm with a companion in the EPM (t(34) = 8.084, P < 0.001. Each group n = 18 mice). F Percentage of sucrose preference (t(36) = 7.997, P < 0.001. Each group n = 19 mice). G Modified pre-pulse inhibition test of different decibels. H Plots of the pre-pulse response intensity ratio at 120 dB (t(28) = 2.795, P = 0.0093. Each group n = 15 mice). I Jumping times under 70 dB and 80 dB conditions, with each decibel level involving 15 mice. Each mouse was stimulated 5 times, resulting in a total of 75 responses per group. (Chi-square test, 80 dB, χ² = 3.972, P = 0.046. 70 dB, χ² = 1.349, P = 0.246). J Changes in response strengths (Two-way ANOVA with Fisher’s LSD for multiple comparisons. dB × Group F(1, 35) = 0.3934, P = 0.5346. 90 dB, Ctrl, n = 10. Intruder, n = 13. 80 dB, Ctrl, n = 6. Intruder, n = 10). K, L Representative traces and statistics of behavioral tests for tremor frequency (t(21) = 5.281, P < 0.0001. n = 12 and 11 for Ctrl and Intruder groups mice). M, N Representative traces and statistics of fluctuations of included angles in body arch (t(21) = 4.983, P < 0.0001. n = 12 and 11 mice). O, P Representative traces and statistics of motion distance (t(21) = 4.119, P = 0.0005. n = 12 and 11 mice). Q Diagram of social stress induces fear memory and schizophrenia-like behaviors. The darkness of color represents the severity. The mouse images were produced based on the platform from BioRender.com. Data are represented as mean ± s.e.m.

The CD1 male mice selected to be aggressive residents (aggressors) in the resident/intruder paradigm were based on the criterion that the latency of their attacks to the unfamiliar C57 mice was within two minutes when they were placed together. In order to have the resident CD1 male mice more aggressive in this resident/intruder paradigm, we placed a pair of male and female CD1 mice to live in a normal cage (29 × 17.5 × 12.5 cm) more than four days, or one sexual cycle, for their “marriage” relationships [53, 62, 107–109].

The social stress was induced by the resident/intruder paradigm

In the resident/intruder paradigm [104–108], the attacks of a resident CD1 male mouse to intruder C57 mice were more realistic to mimic the social stress in lifespan than the electrical shocks to mouse feet as the stress used in other studies [107–109]. After the adaptation period, the qualified C57 mice at postnatal weeks three were divided into four groups, i.e., control, intruder, intruder plus scramble control and intruder plus dopaminergic receptor-II knockdown (Drd2-KD). A knockdown of dopaminergic receptor-II mRNA in the medial prefrontal cortex (mPFC) was done by the injection of AAV2-CMV-U6-mDR2-GFP in the mPFC (Figure SM3), which produced short-hairpin RNAs specifically to silence dopaminergic receptor-II mRNA. shRNA-scramble control was done by the injection of AAV2-CMV-U6-GFP into the mPFC.

During the attacks by the resident CD1 male mouse, intruder C57 mice received the stressful signals, such as the battle sound from the auditory system, the pain signal of somatic injury areas from the somatosensory system as well as the image of CD1 mouse plus the battle field from the visual system. These signals were inputted to those cross-modal sensory cortices of intruder C57 mice that encoded relevant sensory signals, such as the battle sound to the auditory cortex, the images including resident CD1 mouse and battle scenes to the visual cortex and the painful signal from body injury areas to the S1Tr cortex, leading to the associative learning. These physical and psychological stress signals were thought to associatively evoke the fear memory of intruder C57 mice to a resident CD1 mouse [53, 62, 112, 113]. It is noteworthy that the battle sounds from the attacks of the resident CD1 mouse to intruder C57 mouse were collected by an audio recorder with high fidelity for the future uses of auditory stimulations in behavioral tasks and electrophysiology in vivo.

The C57 mice in the intruder subgroups, intruder, intruder plus shRNA scramble control and intruder plus dopaminergic receptors-II knockdown, experienced the resident/intruder paradigm for their social stress. In the first two weeks, each of these intruder C57 mice was placed into the living cage of resident CD1 mice in every afternoon, in which the aggressive CD1 male mouse was present and the female CD1 mouse was taken out. The duration for each intruder mouse to stay in the CD1-living cage was based on the attacks when the resident mouse had bitten this intruder mouse five times on the back of its body. In weeks three, each of these intruder mice was placed into this CD1-living cage once two days. Through this procedure, the intruder mice were thought of as experiencing the social stress from the attack of the resident mouse. The stressful signals in resident CD1 attacks were dissected to be the sound signal during their battle, the images of this aggressive CD1 resident and the pain stimulus from the body injury regions bitten by the resident mice. These intruder C57 mice have associatively learnt the stress signals inputted from auditory, somatosensory and visual systems. The stimulations based on these stress signals were used to detect the behavior responses of intruder mice to these associated stress signals in order to test the onset of associative fear memory and schizophrenia as well as used to analyze the responses of neurons in auditory, S1Tr and medial prefrontal cortices to these associated signals in order to confirm the recruitment of associative memory neurons.

The test of fear memory formation

This test within the social interaction cage was used to examine whether the intruder C57 mice were able to memorize the resident CD1 mouse that had attacked them in the resident/intruder paradigm. The avoidance to the resident CD1 mouse with less interaction indicated the formation of fear memory in intruder C57 mice to this resident CD1 mouse. After the period for C57 mice in control group and the resident/intruder paradigm period for C57 mice in the intruder subgroups, the formation of fear memory to the resident CD1 mouse was tested. The object used to test their fear memory was a resident CD1 male mouse that had attacked the intruder mice. The emergence of fear memory was examined in an interaction cage that included one small box of holding this resident CD1 mouse and the interaction zone around this small box in an open field cage (Fig. 1A). The identification of fear memory formation in the intruder mice was based upon the fact that the intruder C57 mice avoided the box of holding this resident CD1 mouse as well as less accessed toward the interaction zone, but not avoided to the box of holding a familiar C57 mouse. The stay time for intruder mice in the interaction zone with the presence of a resident mouse and the stay time for intruder mice in the interaction zone with the absence of this resident were measured for the comparisons between the self-control before the treatment and the data after the treatment, between the control and the intruder, as well as between intruders plus dopaminergic receptor-II knockdown and intruders plus scramble control. The significant reductions of the stay time in the interaction zone with the presence of a resident mouse before and after the social stress as well as the significant reductions of the stay time in the interaction zone with the presence of the resident mouse among these groups indicate the formation of fear memory in intruder mice specifically to the resident mouse.

It is noteworthy that intruder and control mice were separately housed in their own cages in the adaptation period, the intervals of the resident/intruder paradigm and the intervals of the tests of behavior tasks. That is, there were no chances for those mice among inter-groups to the direct interactions for establishing their social communications, empathy and mind infection. In addition, intruder C57 mice had no loss of the social interaction capability as they did not avoid the empty box and other C57 mice [53, 112], except for the resident CD1 mouse. Furthermore, C57 intruder mice had no loss of auditory ability since they responded to the battle sound as well as the sound pulses (Figure SM4).

The test of schizophrenia-like behaviors

Schizophrenia as one severe psychiatric disorder is featured by psychological mania and negative moods. The symptoms and signs of psychological mania mainly include hallucination, delusion and misbelief. The symptoms and signs of negative mood include anhedonia, social withdrawal and anxiety [1–6, 10]. The hallucination refers to the status in that persons believe to sense some signals and messages, especially auditory signals, from their environments, but not realistically present [114, 115]. The hallucination state in the animals has been examined by the pre-pulse inhibition test to measure their hypersensitivity in response to their environmental clues [116, 117], which is also one feature in schizophrenic patients [118]. This hypersensitivity mainly results from the decrease of the sensory threshold to detect the signals from their environments. The decrease of the sensory threshold in the auditory sensation may cause the weak sounds in the environments to be amplified to an alert sound or the scare signal, and even create new types of sensation, e.g., the pain from the internal ears. The decrease of the sensory threshold in the tactile sensation may cause the normal touch to be felt as the pain stimulus. In this regard, the stimulus threshold in the pre-pulse inhibition test may also be changed. Thus, the measurements of the pre-pulse inhibition and the response threshold to sound pulses can be used to test the hypersensitivity of the hallucination status.

The delusion refers to the state that manic persons believe the presence of certain situations around themselves, which are not realistically present, or called as misbelief [119–122]. The typical delusion in schizophrenic patients is characterized as the persecutory delusion. The persecutory delusion is presumably developed when their behaviors in response to the scared environments emerge under the condition of the absence of the scared signals [123, 124]. Based on this definition of the persecutory delusion for schizophrenic patients, we have designed and developed a persecutory delusion test for the rodents including mice, rats and so on. The persecutory delusion is expressed when the scare behaviors including the body arches, limbs’ tremor and interrupted steps in response to the scare signals (Figure SM5-6) emerge under the normal condition or without the presence of the stress signals (Fig. 1K-P). In other words, the result in that intruder mice show the scare behaviors under the normal condition or without receiving any of scared signals is judged to be the emergence of the persecutory delusion (please compare Figure S6 and Fig. 1K-P). Through examining the emergence of the body arches, limb tremor and interrupted steps in intruder mice in the absence of the stress signals, or the absence of a CD1 resident mouse, we are able to conclude whether the persecutory delusion emerges or not.

When the intruder C57 mice showed the hallucination-like behavior, delusion-like behavior, depression-like behavior and anxious state, they were thought of as emerging schizophrenia-like behaviors induced by the social stress.

Hallucination-like behaviors are identified by the tests of pre-pulse inhibition and sensory threshold

The hypersensitivity of schizophrenic mania was examined by the pre-pulse inhibition test and the responsive threshold test to sound pulses. The standard approach of the pre-pulse inhibition test [116, 117, 125–127] was utilized in our experiments to examine the emergence of the hypersensitivity in schizophrenia-like behaviors from those intruder mice in comparison with the control mice. Each of those mice from controls and various intruder subgroups in the pre-pulse inhibition test was conducted in a small open-field cage that was attached on a pressure sensor to sensitively detect the pulse weight due to mouse jumps. A timeline for the pre-pulse inhibition test included an adaptation period, pulse one, irregular pulses’ intervals and pulse two. The mice in this adaption period received 65 dB background sound about 2 min in this open-field cage. Pulse one was set at 120 dB for the sound strength and 40 milliseconds (ms) by 5 times in total 30 seconds for the sound duration. Irregular sound pulses in the intervals consisted of 65-120 dB and 40 milliseconds with 40-100 millisecond intervals about 20 min. The features of pulse two were identical to pulse one. That is, the protocol of the pre-pulse inhibition test consisted of two sound pulses with 120 dB in the strength and 30 seconds in the duration plus the intervals of 20 min for irregular sound pulses (please see one Table in Figure SM5A). The digital traces of mouse responses to sound pulses are presented in Figure SM4B-C.

In the judgement of the hypersensitivity of experimental mice, the following theories have been taken into account. Normally, the mice appeared frequent jumps on an open-field cage in response to the strong sound signals, and their response strengths were larger in pulse one than pulse two. The difference of response strengths between pulse two and pulse one was negative under the normal condition or in the control mice. The ratio of response strength in pulse two to that in pulse one was less than one under the normal condition or in control mice. The decreased response to sound pulses after the first sound pulse and the irregular sound pulses was thought of as the role of inhibitory neural circuits in the sensation and behaviors [116, 117, 128]. As GABAergic neuronal circuit dysfunction has been detected in the schizophrenia patients [129–131], those intruder mice with schizophrenia-like behavior might show the decreased ratio of response two to response one in the pre-pulse inhibition test. The measurements in our studies included the difference of the response strength in pulse two and the response strength in pulse one (R2-R1) or the ratio of this difference to the response strength in pulse one (R2-R1)/R1.

In addition to a decreased response in the pre-pulse inhibition test, the hypersensitivity due to the decreased sensory threshold might cause weak environment stimulations to be amplified to the alert sound and scared signal. The stimulus threshold of the pre-pulse inhibition test might be reduced, such that we have modified the pre-pulse inhibition test. In this modified test, the strengths of sound pulses after the pre-pulse inhibition test were reduced sequentially from 120 dB to 70 dB to merit the threshold of mouse responses (the times and strengths of mouse jumps on the open-field cage) to these sound pulses. The measurements of the response thresholds to sound pulses included the minimal stimulus for mice to jump in response to these sound pulses (stimulus threshold) as well as the mouse response strength that was the ratio of the dynamic weight due to jumps to the static weight. The significant reductions of the stimulation threshold after the pre-pulse inhibition test as well as the increase of their response strength indicated the possibility that these sound stimulations had been converted into the scared signal or the painful signal in the auditory system in those mice with schizophrenia-like behaviors. This shifting of the sensation to sound pulses toward the sensation to the pain in the ears by these sound pulses or toward some scared signals in the emotion might cause the situation similar to the hallucination based on the hypersensitivity in the sensations to various stimulations and in the emotion to the scared signals, which were not present realistically.

Delusion-like behaviors are identified by persecutory delusion test

Based on the definition of the delusion, i.e., the false belief about the presence of unrealistic situations, the persecutory delusion was presumably expressed in the mice when their behaviors in response to the scared environments emerged under the non-scared condition. In terms of the mechanisms underlying the persecutory delusion, the decrease of the sensory threshold in the schizophrenic patients or mice might be associated with the expansion of these super sensitivity and activity in the sensory cortices to the prefrontal cortices and other brain areas, so that they expressed weird memories and disorganized thoughts. In other words, the decreased sensory thresholds in the pre-pulse inhibition test might be changed further to cause false beliefs and disorganized thoughts, or the delusion in mind. What mouse behaviors in response to the scared environments emerged under non-scared conditions was thought of as the expression of the persecutory delusion.

By using the self-programmed artificial intelligence (AI) to recognize behaviors in C57 mice under the condition of facing to a resident CD1 mouse, we measured the following parameters of behaviors from C57 mice and defined their fear responses to this resident CD1 mouse, including the angles of their back arch, the frequencies of their body shaking and the traces of their motion in the cage. The measurements and statistical analyses of these parameters are shown in Figure SM6. The behaviors of C57 mice in their back arch to be the smaller included-angle, the frequent body shaking and the less motion due to the higher tension of limbs’ muscles were thought of as their fear response to the resident CD1 mouse. In the measurement of the back arch of C57 mice based on images from a video recorder that were annotated by using DeepLabCut (DLC) to label their bodies, three points were set for AI’s recognitions in those C57 mice, including the nose tip, the middle top in spinal back and the tail root. These data about the labels of the mouse bodies were used for training the AI’s neural network to learn and to estimate these points accurately, and to read out these points precisely in subsequent analyses. With the connection of such three points, the included angle based on the middle top of the spinal back were measured. As those C57 mice fearing to a resident CD1 mouse appeared to be frequent bend in their back and body shaking, the included angles based on the middle top of their spinal back might become smaller and quickly fluctuated. In the measurement of the body shaking of C57 mice, the frequencies in their quick motions in these three points of their bodies were recognized and measured by this AI’s neural network. The C57 mice fearing to a resident CD1 mouse appeared the high frequency of body shaking. In addition, this self-programmed software utilized for AI’s recognition was also used to monitor the total traces of mouse motions in the cages.

The measurements of fear responses in C57 mice have been conducted in the cages with the presence of the resident CD1 mouse (the scared environment) and without the presence of this resident (none scare environment). It is noteworthy that the interval of the persecutory delusion test between with the scared environment and the none scared environment was above three hours to prevent the interactive influence from these two conditions, and that the cages for this test were not for the living houses for resident mice and intruder mice. What the fear behaviors in the scared environment emerged in the absence of this resident mouse were thought of the persecutory delusion. The comparison of those C57 mice among the groups of control, intruder, intruder plus dopaminergic receptor-II knockdown and intruder plus scramble control in face to the resident CD1 mouse and with no resident CD1 mouse would indicate whether those C57 mice expressed persecutory delusion-like behavior. If the intruder C57 mice demonstrated significantly fear behaviors without the presence of the resident mouse, including the smaller included angles, the frequent body shaking and shorter motion traces, they were presumably in the status of the persecutory delusion-like behaviors.

Depression-like behaviors are identified by the sucrose preference test and the Y-maze test

Anhedonia, interest loss and social withdrawal as depression-like behaviors were assessed after C57 mice experienced resident/intruder paradigm or were treated as the control for three weeks. Anhedonia was evaluated by the sucrose preference test (SPT). Loss of interest to their partners and social withdrawal were assessed by using the Y-maze test (YMT) [57, 132–136]. The SPT was performed by measuring mouse ingestions of 1% sucrose water versus pure water in two hours. The SPT values were the ratio of ingested sucrose water to total water including the sucrose water plus pure water. The YMT was operated by monitoring the mouse stay time in a special arm and other arms. The end of this special arm included a female mouse (i.e., M-arm). In five min of measurements in the YMT, the ratios of the stay time in the M-arm to the stay time in three arms were calculated. The SPT and YMT were given before and after the resident/ intruder paradigm. All of these measurements and carefulness in the detailed protocols were given in our previous publications [135–137]. With the sufficiency of these two tests above for assessing depressive mood, we did not used the tail suspension test and the forced swimming test in that the stressful condition may influence the judgement of mood state.

Depression-like behaviors were accounted when intruder mice showed the decreases in the sucrose preference and M-arm stay time, in comparison with the values during their self-control period (week one for an adaption) and in control mice. The significant changes in these tests for each mouse were accepted if the SPT and YMT values attenuated over 20% of their self-controls. These criteria were based on the averaged values in our previous studies [58, 59, 135, 136, 138, 139]. The mice with significant changes in these two tests were thought of as depression-like mice induced by a resident/intruder paradigm [58, 59, 138]. The control mice and intruder mice with depression-like behaviors and other symptoms were further studied in neuronal functions, synapse innervations and mRNA/protein analyses in the medial prefrontal cortex, the auditory cortex and the S1Tr cortex.

Anxiety-like behaviors were identified by an elevated-plus maze

The anxious state in C57 mice induced by the resident/intruder paradigm was evaluated by an elevated-plus maze (EPM), which was thought to be a validated and classic method to assess the level of anxiety in rodents [140, 141]. In the typical EPM, two open with 30 cm in the length, 5 cm in the width and 0 cm in the wall height were opposite to two closed arms with 30 cm in the length, 5 cm in the width and 15.25 cm in the wall height. Such crossed arms were extended from a central platform (5 cm × 5 cm). The height of the EPM’s arms and central plate was 40 cm above the floor. All of these experiments were performed between 8:00 to 14:00. In general, those mice avoided the open field, however, they were going to explore new environments for the food and social partners. The avoidance of the mice to the open field was measured by the duration when these mice stayed in the closed arms, or the duration in the closed arms versus total experiment time. The exploration of the mice to the new environment was measured by the entry times into open arms. Thus, the exploration times and the stay duration in the closed arms were utilized to evaluate the level of anxious state, which were recorded by an automatic video-tracking system for five min. The C57 mice were placed at the central platform of the EPM in face to one of the closed arms at the beginning of experiments. The behavior that the mice spent more time in the closed arms and had low exploration times to the open arms was presumably the higher level of anxious state [142, 143].

The identification of associative memory cell

Associative memory neurons were defined as the neurons that received convergent synapse innervations including newly formed synapses and previously formed synapses as well as encoded multiple signals brought by these synapse inputs [65, 93, 95, 96, 144]. The associative memory neurons were recruited based on the principle of coactivity together and interconnections together among the neurons by a chain reaction including the intensive action potentials, epigenetic events as well as gene and protein expressions in relevance to new axonal projection and synapse formation. The associative memory neurons among cross-modal cortices and within intramodal cortex were featured by their synapse interconnections, such that each of them received new synapse innervations from active neurons alongside innate synapse inputs as well as encoded the signals inputted by these axons and synapses [65, 93, 95, 96, 99, 144]. The stressful signals in the resident/intruder paradigm were dissected into distinct modal signals that could be sensed by their correspondent sensory systems of intruder mice. The battle sound generated in the attack of the resident CD1 mouse to intruder C57 mice was detected and transmitted to the auditory cortex by the auditory system. The painful signal from body-injury regions bitten by the resident CD1 mouse was detected and transmitted into the S1Tr cortex by the somatosensory system. The images of resident CD1 mouse and their fighting were detected and transmitted into the visual cortex by the visual system. This resident/intruder paradigm drove intruder C57 mice to associatively learn these auditory, somatosensory and visual signals. In other words, the stress signals in the joint storage and the reciprocal retrieval of this associative fear memory included the auditory signal (the battle sound generated during their fighting), the somatosensory signal (the pain signal from body-injury regions) and the visual signal from the resident CD1 image and fight scene. The retrievals of fear memory to the resident CD1 mouse in intruder C57 mice might be induced by seeing this resident CD1 mouse, hearing the battle sound or receiving the painful stimulus in body-injury areas. New synapse interconnections might be detected by neural tracing morphologically among S1Tr, auditory and visual cortices. Convergent synapse innervation might be detected on prefrontal cortical neurons by neural tracing. Associative memory neurons that encoded somatic stimuli, battle sound and CD1 images might be recorded by electrophysiological approach in these cortical regions [99].

Neural tracing to localize associative memory neurons

The morphological identification of interconnections among medial prefrontal cortex, S1Tr cortex and auditory cortex was traced by gene-coded fluorescent proteins carried by adeno-associated viruses (AAV) [93–96, 98, 99]. A few subtypes of AAV2s with CMV-promoter were used in our experiments, e.g., AAV2-CMV-GFP, AAV2-CMV-BFP, and AAV2-CMV-tdTomato (OBiO Inc., Shanghai China). In the study of synapse interconnections among these cortices, AAV2-CMV-GFP was injected in the S1Tr cortex (-1.5 mm posterior to the bregma, 1.5 mm lateral to the middle line and 0.5 mm depth away from the bregma; from the brain map [145]), AAV2-CMV-tdTomato was injected in the auditory cortex (-2.0 mm posterior to the bregma, 4.0 mm lateral to the middle line and 1.6 mm depth away from the bregma) as well as AAV2-CMV-EBFP was injected in the medial prefrontal cortex (1.8 mm anterior to the bregma, 0.4 mm lateral to the middle line and 1.7 mm depth away from the bregma) three days before the resident/intruder paradigm. These microinjections were done by using glass pipettes controlled from the microsyringe held with three-dimensional stereotaxic apparatus (RWD Life Science, Shenzhen, China). The AAV microinjections were about 0.2 μl in the volume and 30 min for the duration. AAV-CMV-GFP was uptaken and expressed in S1Tr cortical neurons, where the green fluorescent protein (GFP) was produced. The GFP was transported to entire axons at the target areas in an anterograde manner, so that axon boutons and terminals were labelled by the GFP. AAV-CMV-tdTomato was uptaken and expressed in the auditory cortical neurons, where the red fluorescent protein (RFP) was produced. The RFP was transported to entire axons at the target areas in an anterograde manner, so that axon boutons and terminals were labelled by the RFP. AAV-CMV-EBFP was uptaken and then expressed in the prefrontal cortical neurons, where the blue fluorescent protein (BFP) was produced. The BFP was transported to entire axons at the target areas in an anterograde manner, so that axon boutons and terminals were labelled by the BFP [93–96, 98, 146].

In the study of the interconnections among medial prefrontal, auditory and S1Tr cortices, our strategies and approaches were the detection of fluorescent-labelled axonal boutons in the target areas, as fluorescent proteins had been injected in the source areas [98, 99]. As the RPF was injected and expressed in the auditory cortex, the detection of RFP -labelled axons in the S1Tr cortex implied the projection of axons of auditory cortical neurons to the S1Tr cortex. As the GPF was injected and expressed in the S1Tr cortex, the detection of GFP -labelled axons in the auditory cortex implied the projection of axons of S1Tr cortical neurons to the auditory cortex. These two results together endorsed the formation of the interconnections between the S1Tr cortex and the auditory cortex in the intruder mice after the resident/intruder paradigm. Similar strategies were used to examine synapse interconnections between the medial prefrontal cortex and the auditory cortex as well as between the medial prefrontal cortex and S1Tr cortex, based on the approach that the BFP expressed in the medial prefrontal cortex, the RFP expressed in the auditory cortex and the GFP expressed in the S1Tr cortex.

In the study of convergent synapse innervations on medial prefrontal cortical neurons from the auditory cortex and the S1Tr cortex, or associative memory neurons in the medial prefrontal cortex, AAV2-CMV-tdTomato was microinjected in the auditory cortex and AAV2-CMV-GFP was injected into the S1Tr cortex. These AAVs were uptaken and expressed at those neurons in the injected areas, where fluorescent proteins were produced. The RFP in the auditory cortex and the GFP in the S1Tr cortex were transported toward the entire axons at the target areas, where these RFP- and GFP-labelled axon boutons and terminals could be detected. When the RFP- and GFP-labelled axon boutons were detected in the medial prefrontal cortex, especially on those spines from individual dendrites of medial prefrontal cortical neurons, these prefrontal cortical neurons were recruited as the associative memory neurons by receiving the convergent synapse innervations, or the morphological identification of the associative memory neurons [93–96, 98]. This strategy was also used to identify the recruitment of associative memory neurons in the auditory cortex and the S1Tr cortex. AAV injections three days before a resident/intruder paradigm allowed the mice to be recovered from this injection operation for experiencing the subsequent experimental manipulations [99].

After those AAV-carried genes were microinjected in intruder and control mice about three weeks when the resident/intruder paradigm was administered, these mice were anesthetized by 2% pentobarbital sodium through the intraperitoneal injections, as well as perfused by 50 ml of 0.9% saline and then 50 ml of 4% paraformaldehyde through the left ventricle until their bodies were rigid. The brains were rapidly isolated and post-fixed in 4% paraformaldehyde for additional one day. The cerebral brains were sliced by the vibratome in a series of coronal sections with the thickness of 100 μm. In order to clearly show three-dimensional images about new synapses in cortices, brain slices were placed in a Sca/eA2 solution for 10 min to make them transparent [95, 147]. These brain slices were rinsed by the phosphate buffer solution for three times, air-dried and cover-slipped. The images of cortical neurons, dendrites, dendritic spines, axon boutons and synapse contacts were taken and collected under a confocal microscope with a 60X lens for high magnification (Nikon A1R plus). The anatomic images of the cerebral brains were taken by this confocal microscope with a 4X lens for low magnification. In C57BL/6 J Thy1-YFP mice, postsynaptic neuron dendrites and spines were genetically labelled by the YFP. Presynaptic axon boutons were labelled by the GFP, BFP and/or RFP produced from the AAV-CMV-FPs being microinjected, respectively. Those contacts between yellow dendritic spines and green, blue or red presynaptic axon boutons with less than 0.1 μm space cleft were chemical synapses presumably [95, 98, 99]. The wavelength of excitation laser beam 488 nm was used to activate the GFP and the YFP. The wavelength of the excitation laser beam 561 nm was utilized to activate the RFP. The wavelength of the excitation laser beam 405 nm was used to activate the BFP. The wavelengths of the emission spectra of the BFP, GFP, YFP and RFP were 412-482 nm, 492-512 nm, 522-552 nm and 572-652 nm, respectively. The images of dendritic spines, axon boutons and synapse contacts were analysed by ImageJ and Imaris quantitatively [95, 99]. Associative memory neurons were accepted when axonal buttons from two sources convergently innervated to dendritic spines on one of YFP-labelled cortical neurons [65, 93, 99].

Electrophysiological neuron recordings to identify associative memory neurons

Before the electrophysiological recording of the neurons in the prefrontal cortex, the auditory cortex and the S1Tr cortex, those mice in the intruder group with fear memory and schizophrenia like-behaviors or control group were anesthetized by intraperitoneal injections of urethane (1.5 g/kg) for surgical operations. The body temperature was kept at 37°C by computer-controlled heating blanket. The craniotomy (2 mm in diameter) was made on the mouse skull above the left side of the prefrontal, auditory or S1Tr cortices [98]. The in vivo electrophysiological recordings at these cortical neurons was conducted in the mice under light anesthetic condition with the withdrawal reflex by pinching fingers, the eyelid blinking reflex by air-puff and the muscle relax. The unitary discharges of these cortical neurons in the category of local field potential (LFP) were recorded in layers II-III of these cortical areas by using glass pipettes filled with a standard solution (150 mM NaCl, 3.5 mM KCl and 5 mM HEPES). The resistance of those recording pipettes was 30 MΩ. The electrical signals of these cortical neurons in their spontaneous spikes and evoked-spikes by the battle sound replayed by an audio recorder or the somatic stimulus to injury areas were recorded and acquired by the AxoClamp-2B amplifier and the Digidata 1322 A and as well as analyzed by a pClamp 10 system (Axon Instrument Inc. CA, USA). These spiking signals were digitized at 20 kHz and filtered by low-pass at 5 kHz. The 100-3000 Hz band-pass filter and the second-order Savitzky -Golay filter were used to isolate spike signals. The normalized spike frequencies in response to the battle sound and the somatic stimulus were the ratios that the spike frequencies in response to the stimulations were divided by spontaneous spike frequencies 30 seconds before the stimuli. If the ratio of the evoked-spike frequencies to spontaneous spike frequencies was 1.5 and above, cortical neurons were deemed as the response to the stimulations [93–98, 148]. Associative memory neurons were accepted by the situations that those neurons in the medial prefrontal cortex, auditory cortex and S1Tr cortex responded to two sources of the stress signals [65, 93, 99].

In terms of associative memory neurons in primary versus secondary in nature, we assumed that the associative memory neurons in the prefrontal cortex were secondary and the associative memory neurons in auditory and S1Tr cortices were primary. This assumption was based on the fact that the prefrontal cortex was allocated at the downstream of sensory cortices, such as the auditory cortex and the S1Tr cortex anatomically, in the signal flow from sensory systems to the cognition system and the motion system [65]. In this regard, we accepted those associative memory neurons in the prefrontal cortex as the secondary identity if their responses to sensory inputs disappeared, when the activities of the neurons in its upstream areas including the auditory cortex and the S1Tr cortex were blocked. Experiments were conducted by recording the activities of prefrontal cortical neurons in response to the battle sound and somatic stimulus as well as by blocking the activities of auditory and S1Tr cortical neurons with 100 μM CNQX and 50 μM D-AP5 that were the antagonists of inotropic glutamatergic receptor-channels. When the responses of associative memory neurons in the medial prefrontal cortex to the battle sound and the somatic stimulus disappeared by using the antagonists of inotropic glutamate receptors, the associative memory neurons in the medial prefrontal cortex were secondary in nature, i.e., the second-order of associative memory neurons.

The study of molecular mechanisms for the recruitment of associative memory neurons

In the study of roles of dopaminergic receptor-II in the formation of new synapse innervation and in the recruitment of associative memory cells in the medial prefrontal cortex and its connecters, dopaminergic receptor-II mRNA was downregulated by its specific shRNA that was carried with AAVs and microinjected in the medial prefrontal cortex in intruder plus dopaminergic receptor-II knockdown mice [149–153]. In the meantime, the control was done by the shRNA scramble microinjection in the medial prefrontal cortex in the intruder plus scramble control mice. By the microinjections of AAV2-CMV-U6-DR2-GFP (pAAV[shRNA]-GFP-U6-DR2) in the medial prefrontal cortex (1.8 mm anterior the bregma, 0.4 mm lateral to the middle line and 1.7 mm depth away from the bregma) three days before the resident/intruder paradigm for associative learning, this approach was expected to reduce dopaminergic receptor-II expression in the prefrontal cortices (Figure SM3) as well as to prevent the formation of new synapse innervations from auditory and S1Tr cortices and the recruitment of associative memory cells in the medial prefrontal cortex. Experiments in dopaminergic receptor-II knockdown were jointly conducted with AAV-mediated neural tracing and electrophysiological recordings to examine the effectiveness of this molecular manipulation on the morphology and the function of associative memory neurons’ recruitment. After an associative learning by the resident/intruder paradigm, those mice from the subgroups in the intruder group including intruder plus dopaminergic receptor-II knockdown and intruder plus scramble controls were examined in their behavioral tasks in relevance to fear memory and schizophrenia-like signs (Figures SM7-9), the morphological convergent synapse innervations on medial prefrontal cortical neurons and the electrophysiological recording of prefrontal cortical neurons in response to the battle sound and the somatic stimulus. The quantities of the medial prefrontal cortical neurons in response to the two stress signals were analyzed in two subgroups. The effectiveness of shRNA specific for dopaminergic receptors-II on new synapse formations and associative memory neuron recruitment would be confirmed if the numbers of the new synapse contacts and associative memory neurons in the group of intruder plus dopaminergic receptor-II knockdown mice were significantly lowered compared with the scramble control group.

In terms of the production of DR2 shRNA, we commissioned Obio Technology Corp., Ltd. in Shanghai, China to design shRNA that targeted DR2 based on the transcripts of mouse DR2 gene, to pack the precursor of shRNA into multiple cloning site of AAV2 as well as to synthesize primers for quantitative RT-PCR. Three sequences were selected based on the homology to mouse DR2 mRNA (NM_010077.3). To prevent nonspecific binding, we assessed these sequences by using a NCBI Basic Local Alignment Search Tool. These sequences included CCGTTATCATGAAGTCTAATG (DR2RNAi01), CCCAGGATTGCCAAGTTCTTT (DR2RNAi02) and CATTGTTCTTGGTGTGTTCAT (DR2RNAi03). The sequence for the scramble control of DR2 shRNA (CCTAAGGTTAAGTCGCCCTCG) did not correspond to any one of known sequences for mouse species. Furthermore, we verified the efficacy of these sequences to knockdown DR2 mRNA by using quantitative PCR (qPCR) and DR2 protein by using western-blot analysis. For the experiments in this study, we have selected sequence 3, as it showed more effectiveness in the pilot experiment (experimental data in Figure SM3). That is, the shRNA designed to knock down dopaminergic receptor-II has been proved to be their effectiveness on the expression levels of dopaminergic receptor-II.

The role of dopaminergic receptor-II in supporting the function of associative memory cells in the medial prefrontal cortex was tested by injecting the antagonist of dopaminergic receptor-II, eticlopride [154–156], into the medial prefrontal cortex. The concentration of eticlopride being injected in the medial prefrontal cortex was 1.325 nM in the optimal effectiveness with a lower dosage as possible, based on our study in the dose-response (Figure SM10A-F). The role of dopaminergic receptor-II in fear memories and schizophrenia-like behaviors was examined by the intraperitoneal injections of eticlopride. Its dosage for intraperitoneal injection was 0.075 mg/kg nM in an optimal effectiveness with a lower dosage as possible, based on our study in the dose-response (Figure SM10G-L). The behaviors in response to the presence of the resident CD1 mouse are represented in Figure SM11 as the index for examining the behaviors of those mice in the absence of resident CD1 mouse and for finding out the changes of their schizophrenia-like behaviors.

Statistical analyses

The ANOVA was used for the comparisons of experiment data including behavioral tasks, axon boutons, synapse contacts and neuronal responses to the battle sound and the pain stimulus in those mice from control and resident/intruder paradigm groups. The ANOVA was also used for the statistical comparison of the changes in neuronal activities and morphology from two groups of intruder plus shRNA scramble control mice and intruder plus dopaminergic receptor-II knockdown (DR2-KD) mice. The X²-test was used for the statistical comparisons of the alternations in the percentages of associative memory neurons identified by electrophysiological study in vivo. The variation of the behavioral tasks was calculated as covariances (please refer to supplement tables one and two, Table S1 and Table S2). The sample size was set by no less than nine samples for behavioral tasks and no less than twenty neurons from five mice in cellular morphological and functional studies.

Results

In this section, we present experimental evidences about the stress-induced recruitment of associative memory neurons in mPFC-centered neural circuits that encode the fear memory and schizophrenia by dopaminergic receptors-II. After intruder mice experienced the social stress in a resident/intruder paradigm, a social interaction test was used to examine the formation of their fear memory to resident mouse. The pre-pulse inhibition test, the persecutory delusion test and other mental tests were used to examine the formation of schizophrenia-like behavior. The neural tracing by AAV-carried genes of fluorescent proteins and electrophysiological recordings in vivo were used to examine the recruitment of associative memory neurons among medial prefrontal, auditory and S1Tr cortices. The knockdown of mRNA that encoded dopaminergic receptor-II by its specific short-hairpin RNA was used to test the roles of dopaminergic receptor-II in the formation of new synapses and the recruitment of associative memory cells for schizophrenia-like behaviors induced by the social stress.

The social stress by resident/intruder paradigm induces fear memory and schizophrenia

The formation of fear memory specific to the resident CD1 mouse in intruder C57 mice was examined by the social interaction test [53, 62, 112, 113, 157]. The avoidance of intruder mice to this resident mouse as the index of fear memory onset was merited by seeing the longer differences of the stay time in the interaction zone between the presence of the resident mouse and the absence of this resident. Heat-maps in Fig. 1B show that one of intruder mice appears to stay away from an interaction zone in the presence of the resident mouse (target), compared to stay near this interaction zone in the absence of the resident (no target), similar to control mice that stay nearby the interaction zone. The differences of the stay time in the interaction zone between the presence of a resident mouse and the absence of this resident mouse are 33.3 ± 8.4 seconds in control mice (n = 18) and -91.5 ± 4.8 seconds in intruder mice (n = 18, p < 0.0001, ANOVA, Fig. 1C). The avoidance of those intruder mice to the resident mouse indicates the emergence of the fear memory specific to this resident mouse in these intruder mice.

In schizophrenia patients, the psychological mania is featured by hallucination and delusion, and the negative mood includes the anhedonia, social withdrawal and anxiety [1–6, 10]. We assessed schizophrenia-like behaviors by the modified pre-pulse inhibition test for hypersensitivity and hallucination, the persecutory delusion test for delusion, the elevated-plus maze test for anxiety, the sucrose preference test for anhedonia and the Y-maze test for social withdraw. When intruder mice showed hallucination-like, delusion-like, anxiety-like and depression-like behaviors, they were thought of as schizophrenic mice.

Anxiety-like behaviors were examined by an elevated-plus maze, as presented in heat-maps of figure for supplementary result (Figure SR1A). Less access to open arms in mice indicated their avoidance to open arms. The reduced percentage of stay time in open arms over total time on an elevated-plus maze after resident/intruder paradigms in intruder mice indicated their avoidance to open arms, or anxiety-like state. In this Figure SR1A, one of intruder mice stays in closed arms and away from open arms, and one of control mice seems to access open arms. The percentages of the stay time in open arms over the total time are 23.4 ± 1.1% in control mice (n = 20) and 4.2 ± 0.8% in intruder mice (n = 20; p < 0.001, ANOVA; Fig. 1D). Thus, the social stress by the resident/ intruder paradigm induces anxiety in intruder mice with fear memory, which endorses the data in the social interaction test (Fig. 1B, C).

Depression-like behaviors were examined by the sucrose preference test for anhedonia and the Y-maze test for interest loss and social withdrawal [135–137, 158]. In heat-maps of Figure SR1B, one intruder mouse prefers to stay away from the interaction arm including its confidante, in comparison with one of control mice. The percentages of the stay time in the interaction arm over the total time within the Y-maze are 22.0 ± 3.3 seconds in intruder mice (n = 18) and 55.4 ± 2.5 seconds in control mice (n = 18, p < 0.0001, ANOVA; Fig. 1E). This result implies that intruder mice are loss of interest to their confidantes in the social activity. In the sucrose preference test, the percentages of sucrose water ingestion in total water ingestion are 48.0 ± 2.4% in intruder mice (n = 19) and 72.1 ± 1.9% in control mice (n = 19, p < 0.0001, ANOVA; Fig. 1F). Intruder mice express anhedonia to sugar. The social stress by the resident/intruder paradigm induces depression-like behaviors in intruder mice with fear memory.

The hypersensitivity of schizophrenic mania in intruder mice was examined by the pre-pulse inhibition test and the response capability to sound pulses. In the pre-pulse inhibition test (Fig. 1G), the response strength was calculated by the ratio of response differences between pulse two and pulse one to response one, (R2-R1)/R1. The values of this ratio are -0.18 ± 0.03 in control mice (n = 15) and−0.08 ± 0.02 in intruder mice with fear memory (n = 15; p = 0.0093, ANOVA; Fig. 1H). Moreover, the response capability to sound pulses was measured by the response threshold and the response strength. The response threshold to sound pulses was the response to the minimal sound pulse for mice to jump, which was calculated by jump times over 75 sound pulses. The response strengths were the levels of responses at the given sound pulses, which were calculated by the ratio of the dynamic weight due to the jumps to the static weight. The times of mouse jumps in 75 sound pulses at 80 dB are 21 in intruder mice with fear memory (n = 15) and 11 in controls (n = 15, p < 0.05, X²-test; the left panel in Fig. 1I). The times of mouse jumps in 75 sound pulses at 70 dB are 2 in intruder mice with fear memory and 4 in control mice (the right panel of Fig. 1I), which is less than mean±2 SD in total pulses. That is, the mouse jumps at 70 dB did not reach to the response threshold. In addition, the response strengths to sound pulses at 80 dB are 1.14 ± 0.02 in intruder mice (n = 10) and 1.06 ± 0.01 in control mice (n = 6, p < 0.05, ANOVA; the left panel of Fig. 1J). The response strengths to sound pulses at 90 dB are 1.15 ± 0.02 in intruder mice (n = 13) and 1.1 ± 0.01 in control mice (n = 10, p < 0.05, ANOVA; the right panel of Fig. 1J). The decreased response threshold and the increased response strength in the intruder mice with fear memory indicate the stress-induced hypersensitivity to sound signals in these intruder mice, i.e., schizophrenia-like mania.

The delusion of schizophrenic mania in the intruder mice was examined by the persecutory delusion test. The delusion of persecution refers to the situation that the scare-induced behaviors express in the absence of the scared signals, or spontaneously. In face to the resident mouse, the intruder mice appeared their back arches to small included-angles, the frequent tremors of their bodies and the less motion in open fields, or scare-induced fear responses to the resident mouse. If intruder mice showed such behaviors in the absence of the resident mouse, they were thought to be the state of persecutory delusion. In Fig. 1K, intruder mice appear frequent body tremors (red trace), compared to control mice (blue trace). The shaking frequencies are 0.27 ± 0.07 Hz in control mice (n = 12) and 2.39 ± 0.41 Hz in intruders (n = 11; p < 0.001, ANOVA; Fig. 1L). Moreover, the body arch in intruder mice appears smaller included-angle (red trace in Fig. 1M), compared with control mice (blue trace). The included angles of mouse back arch (degrees) are 104.0 ± 3.2 in control mice (n = 12) and 76.2 ± 4.7 in intruders (n = 11; p < 0.001, ANOVA; Fig. 1N). Additionally, an intruder mouse with fear memory appears less motion in an open field (red trace in Fig. 1O), compared with a control mouse (blue trace). The motion distances (cm/10 min) in this open field are 1.30 ± 0.2 ×10⁵ in control mice (n = 12) and 0.5 ± 0.1 ×10⁵ in intruders (red bar, n = 11; p < 0.001, ANOVA; Fig. 1P). The frequent body tremors, the dominant body arches and the less motion in the absence of the resident mouse imply that the intruder mice suffer from persecutory delusion.

Intruder mice show the fear memory specific to resident mouse as well as the schizophrenic mania with the hyperactivities for hallucination and delusion plus the negative moods including depression-like and anxiety-like behaviors (Fig. 1Q). We further examined the stress-induced formation of neural circuits in relevance to fear memory and schizophrenia in prefrontal, auditory and somatosensory cortices. Our focus on these cortices to reveal cellular mechanisms, especially memory cells, was based on the following thoughts. Associative memory neurons as basic units of memory trace have been found in the sensory cortices and the prefrontal cortex in associative learning and memory under physiological conditions [65, 93–101]. Fear signals in a resident/intruder paradigm to those intruder mice included the pain signal from body-injury areas to somatosensory cortices and the battle sound to the auditory cortex in associative learning under pathological conditions [53, 62, 112, 113, 157]. The recruitment of associative memory neurons to encode such fear signals for fear memory and schizophrenia in intruder mice was examined by morphological and functional approaches.

The social stress induces interconnections among auditory, S1Tr and medial prefrontal cortices

Morphological interconnections among mouse cortices were examined by the neural tracing. AAV-carried genes of green or red fluorescent proteins (GFP or RFP) were injected in their source areas. GFP-labelled or RFP-labelled axon boutons were searched in their target areas [93, 95, 96, 99, 144]. Because cortical neurons in C57BL/6JThy1-YFP mice were genetically labelled by yellow fluorescent proteins (YFP), the contacts between GFP- or RFP-labelled axon boutons and YFP-labelled dendritic spines were presumably synapses [95, 98, 99, 144]. The rationale to study the interconnections and interactions among the medial prefrontal cortex, the auditory cortex and the S1Tr cortex is based upon the common view that the stressful signals including the battle sound and the painful signal from body-injury areas during resident/intruder paradigms are inputted to the auditory cortex and the S1-Tr cortex, respectively, and secondarily transmitted to the prefrontal cortex [65].

In the study of target areas of S1Tr cortical neurons and auditory cortical neurons, AAV-CMV -Oregon Green and AAV-CMV-tdTomato were injected in the S1Tr cortex and the auditory cortex, respectively (Fig. 2A). In comparison with control mice, the S1Tr cortical areas in intruder mice receives RFP-labeled axons from the auditory cortex (left panel in Fig. 2B), the auditory cortex areas in intruder mice receive GFP-labelled axons from the S1Tr cortex (middle panel in Fig. 2B), and the medial prefrontal cortices in intruder mice receive convergent synapse innervations from RFP-labelled axons of auditory cortical neurons and GFP-labelled axons of S1Tr cortical neurons (right panel in Fig. 2B). The densities of RFP-labelled axon boutons per mm³ in S1Tr cortices are 0.33 ± 0.01 ×10⁵ in control group (n = 29 cubes from 9 mice) and 0.57 ± 0.05 ×10⁵ in intruders (n = 30 cubes from 9 mice, p = 0.0003, ANOVA; Fig. 2C). The densities of GFP-labelled axon boutons per mm³ in auditory cortices are 0.04 ± 0.01 ×10⁵ in controls (n = 28 cubes from 9 mice) and 0.57 ± 0.06 ×10⁵ in intruder group (n = 30 cubes from 9 mice, p < 0.0001, ANOVA; Fig. 2C). The social stress induces axon interconnections between S1Tr and auditory cortices. Furthermore, the densities of GFP-labelled axon boutons from S1Tr cortical neurons per mm³ in the medial prefrontal cortices are 0.20 ± 0.06×10⁵ in controls (n = 38 cubes from 9 mice) and 1.21 ± 0.21×10⁵ in intruders (n = 41 cubes from 9 mice, p < 0.0001, ANOVA; Fig. 2D). The densities of RFP-labelled axon boutons from auditory cortical neurons per mm³ in the medial prefrontal cortices are 0.5 ± 0.08 ×10⁵ in controls (n = 39 cubes from 9 mice) and 1.09 ± 0.20 ×10⁵ in intruders (n = 40 cubes from 9 mice, p = 0.012, ANOVA; Fig. 2D). These data indicate stress-induced axon projections to the medial prefrontal cortex from S1Tr and auditory cortices.

Fig. 2. The social stress induces interconnections among auditory, S1Tr and mPFC.

A Virus schematics (top) and representative a coronal section confocal image of the injection site (bottom). Scale bar, 1 mm. B Confocal microscopy images illustrate YFP-labeled postsynaptic dendrites and axonal innervations labeled with GFP and RFP in the S1Tr, AC, and mPFC. Red arrows denote axon boutons originating from the AC, green arrows indicate from the S1Tr. Scale bar, 20 μm. C Statistical analyses of the densities of fluorescently labeled axon boutons in the S1Tr and AC. (Area ×Group F(1, 113) = 9.446, P = 0.0027). D Same as C, but for the mPFC area. (Area ×Group F(1, 154) = 1.658, P = 0.1998). E Illustrating synapse contacts in S1Tr, AC, and mPFC. White boxes represent the newly formed synapse contacts. Axonal boutons from the S1Tr are labeled with GFP, axonal boutons from the AC are labeled with RFP. Scale bar = 10 μm; inset scale bar = 2 μm. F Statistical analyses synapse contacts in S1Tr and AC per 100 μm dendrite. (Area ×Group F(1, 129) = 0.0884, P = 0.7668) G Same as F, but for the mPFC area. (Area ×Group F (1, 250) = 52.05, P < 0.0001). Two-way ANOVA with Fisher’s LSD for multiple comparisons. H Top, virus schematics. Bottom,coronal section confocal images of the injection site. Scale bar, 1 mm. I Axonal innervations (left panel) and newly formed synapse contacts (right panel) in S1Tr. Red arrows indicate boutons, white boxes represent the newly contacts. Scale bar = 20 μm, scale bar = 10 μm, inset scale bar = 2 μm. J, K A barplot of axonal innervations and synapse contacts in the S1Tr, respectively. (J t(63) = 12.03. K t(70) = 5.702). L Same as I, but for the AC area. M, N Same as J, K but for the AC area. (M t(63) = 5.231. N t(62) = 5.739). Independent-samples t-test. S1Tr: primary somatosensory cortex, AC auditory cortex, mPFC medial prefrontal cortex. Data are represented as mean ± s.e.m.

The axon innervations on target neurons to make new synapses were examined by screening those attachments between the GFP- or RFP-labeled axon boutons from presynaptic neurons and the YFP-labeled dendritic spines on postsynaptic neurons, or synapse contacts [95, 98, 99, 144]. In comparison with controls, the left panel in Fig. 2E shows the reception of new synapse contacts on S1Tr cortical neurons from auditory cortical area in intruder mice, the middle panel in Fig. 2E shows the reception of new synapse contacts on auditory cortical neurons from S1Tr cortical area, as well as the right panel in Fig. 2E shows the reception of new synapse contacts on prefrontal cortical neurons from auditory and S1Tr cortical areas. Synapse contacts per 100 μm dendrite in S1Tr cortices are 1.6 ± 0.3 in control group (n = 37 dendrites from 9 mice) and 3.7 ± 0.3 in intruder group (n = 35 dendrites from 9 mice, p = 0.0001, ANOVA; Fig. 2F). Synapse contacts per 100 μm dendrite in auditory cortices are 3.1 ± 0.5 in controls (n = 26 dendrites from 7 mice) and 5.4 ± 0.5 in intruders (n = 35 dendrites from 7 mice, p = 0.001, ANOVA; Fig. 2F). These data indicate stress-induced synapse interconnections between S1Tr and auditory cortices. Moreover, the synapse contacts from S1Tr cortical neurons per 100 μm dendrite in prefrontal cortices are 0.7 ± 0.2 in controls (n = 71 dendrites from 12 mice) and 4.6 ± 0.4 in intruder group (n = 56 dendrites from 10 mice, p < 0.001, ANOVA; Fig. 2G). The synapse contacts from auditory cortical neurons per 100 μm dendrite in prefrontal cortices are 0.7 ± 0.1 in controls (n = 71 dendrites from 12 mice) and 1.40 ± 0.2 in intruders (n = 56 dendrites from 10 mice, p < 0.01, ANOVA; Fig. 2G). These data indicate the stress-induced convergent synapse innervations onto medial prefrontal cortical neurons from the S1Tr and auditory cortices.

In the study of axon targets of medial prefrontal cortical neurons, AAV2-CMV-tdTomato was injected in medial prefrontal cortices (Fig. 2H) and RFP-labelled axon boutons were screened in S1Tr and auditory cortices. In comparison with control mice, S1Tr cortical neurons in the intruder mice appear to receive more axon projections from the prefrontal cortex (left panel in Fig. 2I) and more synapse contacts from these axon boutons (right panel). Axon boutons per mm³ in S1Tr cortices are 0.19 ± 0.02 ×10⁵ in control group (n = 35 cubes from 9 mice) and 0.67 ± 0.04 ×10⁵ in intruder group (n = 30 cubes from 9 mice, p < 0.0001, ANOVA; Fig. 2J). Synapse contacts per 100 μm dendrite on S1Tr cortical neurons are 0.0 ± 0.0 in controls (n = 34 dendrites from 6 mice) and 1.0 ± 0.20 in intruders (n = 38 dendrites from 8 mice, p < 0.0001, ANOVA; Fig. 2K). In addition, auditory cortical neurons in intruder mice appear to receive more axon projections from the prefrontal cortex (left panel in Fig. 2L) and more synapse contacts from these axon boutons (right panel), in comparison to control mice. Axon boutons per mm³ in auditory cortices are 0.08 ± 0.01 ×10⁵ in control group (n = 30 cubes from 9 mice) and 0.29 ± 0.03 ×10⁵ in intruder group (n = 35 cubes from 9 mice, p < 0.001, ANOVA; Fig. 2M). Synapse contacts per 100 μm dendrite on auditory cortical neurons are 0.0 ± 0.0 in controls (n = 29 dendrites from 8 mice) and 1.4 ± 0.20 in intruders (n = 26 dendrites from 8 mice, p < 0.001, ANOVA; Fig. 2N). Thus, the S1Tr and auditory cortical neurons in intruder mice with fear memory and schizophrenia-like behaviors receive new synapse innervations from the medial prefrontal cortex.

From these morphological data, the social stress by the resident/intruder paradigm induces the synapse interconnections among auditory, somatosensory and medial prefrontal cortices, so that the cortical neurons in any one of these regions receive the synapse innervations from other two areas. Their synapse interconnections endorse the cortical neurons to encode and memorize associative fear signals in the social stress including the battle sound and the somatic signal for the formation of fear memories and schizophrenia-like behaviors. The recruitment of associative memory cells [65] in these cortices was examined by in vivo electrophysiological recordings.

The social stress recruits associative memory neurons in prefrontal, S1Tr and auditory cortices

Stress-induced synapse interconnections among the neurons in these cortices recruit them as associative memory neurons to encode the stress signals for fear memories and schizophrenia, similarly to the recruitment of associative memory neurons in associative learning [65]. For instance, auditory cortical neurons by receiving synapse innervations innately from the internal geniculate body and newly from the S1Tr cortex are recruited to encode battle sound and pain signal in the resident/intruder paradigm, or the other way around.

Electrophysiological recordings in vivo were conducted in S1Tr cortices in the meantime to give battle sound and somatic signal. In comparison with one sample neuron in response to the somatic stimulation to back regions in control mice (blue trace in Fig. 3A), one of S1Tr cortical neurons in intruder mice appears to encode battle sound and somatic stimulus, i.e., associative memory neuron (red trace in Fig. 3A). The percentages of S1Tr cortical neurons in response to both sound and somatic signals are 3.91% in control group (n = 128 neurons in totally recorded from 9 mice) and 10.75% in intruder group (n = 120 neurons from 9 mice; p < 0.05, X²-test in Fig. 3B). The proportions of associative memory neurons and non-associative memory neurons are presented in Figure SR2A for control and intruder mice. Moreover, the activity strengths of S1Tr cortical neurons in response to battle sound and somatic stimulus were analyzed. Normalized spike frequencies at S1Tr cortical neurons in response to the battle sound are 1.2 ± 0.1 in control mice (n = 128 neurons) and 1.80 ± 0.3 in intruders (n = 120 neurons; p < 0.05, ANOVA; the left panel of Fig. 3C). Normalized spike frequencies at S1Tr cortical neurons in response to the somatic stimulus are 1.3 ± 0.1 in controls (n = 128 neurons) and 1.70 ± 0.2 in intruders (red bar, n = 120 neurons; p < 0.05, ANOVA; the right panel of Fig. 3C). More S1Tr cortical neurons to encode new battle sound and innate somatic signal as well as their high response strengths in intruder mice indicate the recruitment of these neurons to be the associative memory neurons and their functional upregulation induced by the social stress.

Fig. 3. Social stress recruits associative memory neurons.

A Representative traces of spontaneous and evoked spikes in S1Tr neurons. Gray boxes represent the application of two stimuli, AS, battle sound. SS, somatic stimulus, which last for 30 s. B Percentage diagram illustrates the response of S1Tr neurons to both auditory and somatosensory signals (Ctrl, n = 128 neurons. Intruder, n = 120 neurons. Chi-square test, χ² = 4.265, P = 0.0389). C Statistical comparison of normalized spike frequencies in S1Tr (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 466) = 0.1715, P = 0.6790. AS: Ctrl, n = 128 neurons. Intruder, n = 120 neurons. SS: Ctrl, n = 128 neurons. Intruder, n = 120 neurons). D Examples of spike recordings in the AC neurons. E Compared to other neuronal ensembles, the neurons ensemble in the Intruder group contained a higher proportion of AMC (Ctrl, n = 120 neurons. Intruder, n = 128 neurons. Chi-square test, χ² = 4.864, P = 0.0286). F Normalized spike frequencies in AC (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 492) = 0.1316, P = 0.9711). G Same as A, but for the mPFC area. H Same as B, but for the mPFC area. (Ctrl, n = 122 neurons. Intruder, n = 115 neurons. Chi-square test, χ² = 6.574, P = 0.0103). I Same as C, but for the mPFC area (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 524) = 0.2040, P = 0.6517). J Strategy for the application of antagonists to ionotropic glutamatergic receptors, coupled with the recording of neuronal activity via electrodes in the mPFC. K Example traces depict the recording of both spontaneous and evoked spikes before and after the application of antagonists. Blue triangles point to the application of antagonists, gray boxes indicate the reapplication of two stimuli (AS and SS). L, M Discharge frequency statistics standardized for antagonists and saline group (AS, t(8) = 5.880, P = 0.0004. SS, t(8) = 5.353, P = 0.0007. n = 9 mice for CNQX/D-AP5 and saline groups). CNQX 6-Cyano-7-nitroquinoxaline-2,3-dione, D-AP5 D-2-Amino-5-phosphonovaleric acid. Data are represented as mean ± s.e.m.

Electrophysiological recordings in vivo were also done in auditory cortices in the meantime to give battle sound and somatic signal. In comparison with one sample neuron in control mice in response to battle sound (blue trace in Fig. 3D), one of auditory cortical neurons in intruder mice appears to encode the battle sound and somatic stimulus (red trace in Fig. 3D). The percentages of auditory cortical neurons in response to somatic and sound signals are 5.83% in controls (n = 120 neurons from 9 mice) and 16.41% in intruders (n = 128 neurons from 9 mice; p < 0.05, X²-test in Fig. 3E). The proportions of associative memory neurons and non-associative memory neurons are shown in Figure SR2B for control and intruder mice. Moreover, normalized spike frequencies at auditory cortical neurons in response to battle sound are 1.2 ± 0.1 in controls (n = 120 neurons) and 1.7 ± 0.2 in intruder (n = 128 neurons; p < 0.01, ANOVA; the left panel of Fig. 3F,). Normalized spike frequencies at these auditory cortical neurons in response to the somatic stimulus are 1.3 ± 0.1 in controls (n = 120 neurons) and 1.90 ± 0.2 in intruders (n = 128 neurons; p < 0.05, ANOVA; the right panel in Fig. 3F). More auditory cortical neurons in intruder mice to encode innate battle sound and new somatic signal as well as their higher responsive strength indicate the recruitment of these neurons to associative memory neurons and their functional upregulation induced by the social stress.

Electrophysiological recordings in vivo have also been done in medial prefrontal cortices in the meantime to give battle sound and somatic stimulus. In comparison with a sample neuron from control mouse in response to the somatic stimulus (blue trace in Fig. 3G), one of medial prefrontal cortical neurons in intruder mice appears to encode both somatic stimulus and battle sound (red trace in Fig. 3G). The percentages of medial prefrontal cortical neurons in response to somatic and sound signals are 4.92% in controls (n = 122 neurons in total recorded from 9 mice) and 14.78% in intruders (n = 115 neurons from 9 mice; p < 0.05, X²-test in Fig. 3H). The portions of associative memory neurons and non-associative memory neurons are shown in Figure SR2C for control mice and intruder mice. Moreover, normalized spike frequencies at medial prefrontal cortical neurons in response to the battle sound are 1.30 ± 0.1 in controls (n = 122 neurons) and 1.90 ± 0.1 in intruders (n = 115 neurons; p < 0.01, ANOVA; the left panel of Fig. 3I). Normalized spike frequencies at these medial prefrontal cortical neurons in response to the somatic stimulus are 1.3 ± 0.1 in control mice (n = 122 neurons) and 2.0 ± 0.2 in intruders (n = 115 neurons; p < 0.001, ANOVA; the right panel in Fig. 3I). More medial prefrontal cortical neurons to encode battle sound and somatic signal as well as their high responsive strengths in intruder mice indicate the recruitment of these neurons to associative memory neurons and their functional upregulation induced by the social stress.

Therefore, the social stress recruits medial prefrontal, auditory and S1Tr cortical neurons to be associative memory neurons. As the signal flows in the cerebral cortex originate from sensory cortices to their downstream cortices in the frontal lobe, these associative memory neurons may fall into different grades, such as the first order in sensory cortices and the second order in the prefrontal cortex [65]. We have tested this hypothesis by the pharmacological blocking of sensory cortices to see whether the activity of associative memory neurons in medial prefrontal cortices disappeared.

Electrophysiological recordings in vivo were conducted in the medial prefrontal cortices of the intruder mice to examine their responses to the battle sound and the somatic stimulus, while the antagonists of ionotropic glutamatergic receptors were used in S1Tr and auditory cortices to block their neuronal activity (Fig. 3J). A timeline to identify the secondary associative memory neurons in the medial prefrontal cortex was to record their responsiveness to battle sound and somatic stimulus, to perfuse CNQX and D-AP5 to S1Tr and auditory cortices in an interval period, and to record the activity of these associative memory neurons again (Fig. 3K). In comparison with the injection of saline in S1Tr and auditory cortices, the injections of CNQX/D-AP5 to these areas significantly suppress the spike-encoding of associative memory neurons to battle sound and somatic stimulus (Fig. 3L, M). These data support the hypothesis that associative memory neurons in the medial prefrontal cortex are the second-order in nature.

In the investigation of the role of secondary associative memory neurons within the medial prefrontal cortex in fear memories and schizophrenia-like behaviors of intruder mice, dopamine receptor-II was knocked down to examine whether this manipulation blocked the stress-induced recruitment of associative memory neurons and the stress-induced emergence of fear memories and schizophrenia-like behaviors.

Dopaminergic receptors-II is needed for stress-induced recruitment of associative memory cells

The knockdown of dopaminergic receptors-II mRNA (D2R-KD) in the medial prefrontal cortex was conducted by microinjecting AAV2-CMV-U6-DR2-EGFP (pAAV[shRNA]-EGFP-U6-DR2) in this area before the C57 mice experienced the resident/intruder paradigm, or intruder plus D2R-KD. In the meantime, shRNA-scramble control was microinjected into the medial prefrontal cortex in intruder plus scramble control mice. In the morphological identification of associative memory neurons, AAV2-CMV-BFP, AAV2-CMV-tdTomato and AAV2-CMV-GFP were respectively injected in S1Tr, auditory and medial prefrontal cortices, while the pAAV[shRNA]-GFP-U6-DR2 was injected into the medial prefrontal cortex. After the mice experienced the resident/intruder paradigm, the electrophysiological recordings in vivo were conducted in medial prefrontal cortical neurons, and neural tracings were conducted in the S1Tr, auditory and medial prefrontal cortices in those mice from the groups of intruder plus DR2-KD and intruder plus scramble.

In the study of the influence of dopaminergic receptor-II knockdown on the interconnections between S1Tr cortical neurons and auditory cortical neurons, AAV-CMV-tdTomato was injected in the auditory cortex and AAV-CMV-BFP was injected in the S1Tr cortex (Fig. 4A). In comparison with intruder plus scramble control, the knockdown of dopaminergic receptor-II in intruder mice (intruder plus DR2-KD) appears to dilute the RFP-labeled axons from auditory cortical neurons to the S1Tr cortex (left panel in Fig. 4B), the BFP-labelled axons from S1Tr cortical neurons to the auditory cortex (middle panel) and their convergent synapse innervations onto medial prefrontal cortical neurons (right panel). RFP-labelled axon boutons per mm³ in S1Tr cortices are 0.60 ± 0.06 ×10⁵ in intruder plus scrambles (n = 27 cubes from 9 mice) and 0.3 ± 0.04 ×10⁵ in intruder plus DR2- KD (n = 34 cubes from 9 mice, p < 0.001, ANOVA, left bars in Fig. 4C). BFP-labelled axon boutons per mm³ in auditory cortex are 0.20 ± 0.05×10⁵ in intruder plus scramble (n = 29 cubes from 9 mice) and 0.06 ± 0.01 ×10⁵ in intruder plus DR2-KD (n = 37 cubes from 9 mice, p = 0.008, ANOVA; right bars in Fig. 4C). These data indicate that the dopaminergic receptor-II knockdown prevents the stress-induced interconnections between the S1Tr cortex and the auditory cortex. In addition, BFP-labelled axon boutons per mm³ from S1Tr cortical neurons to medial prefrontal cortices are 1.08 ± 0.22 ×10⁵ in intruder plus scramble (n = 29 cubes from 8 mice) and 0.13 ± 0.04 ×10⁵ in intruder plus DR2-KD (n = 30 cubes from 8 mice, p < 0.001, ANOVA; left bars in Fig. 4D). RFP-labelled axon boutons per mm³ from auditory cortical neurons to medial prefrontal cortices are 0.99 ± 0.18 ×10⁵ in intruder plus scrambles (n = 39 cubes from 9 mice) and 0.5 ± 0.08 ×10⁵ in intruder plus DR2-KD (n = 40 cubes from 9 mice, p < 0.001, ANOVA; right bars in Fig. 4D). These data indicate that the dopamine receptor-II knockdown prevents stress-induced axon innervations to medial prefrontal cortical neurons from auditory and S1Tr cortices.

Fig. 4. Drd2 knockdown reduces interconnections among AC, S1Tr and mPFC.

A Virus schematics (top) and representative a coronal section confocal image of the injection site (bottom). Scale bar, 1 mm. B Representative confocal images depict axonal innervations from fluorescently labeled axons in the scramble and Drd2-KD groups. Red triangles point to boutons inputted from the AC, green triangles point to boutons inputted from the S1Tr. Scale bar, 20 μm. C Bar graph depicts the densities of axon boutons labeled with fluorescent markers in the S1Tr and AC (Area ×Group F(1, 123) = 2.461, P = 0.1193). D Same as C, but for the mPFC area (Area ×Group F(1, 134) = 2.362, P = 0.1267). E Illustrating synapse contacts in S1Tr, AC and mPFC. White boxes represent the newly formed synapse contacts. EBFP-labeled axonal bouton from S1Tr, RFP-labeled axonal boutons from the AC. Scale bar = 10 μm; inset scale bar = 2 μm. F Bar graph depicts Drd2-KD reduced the synapse contacts in the S1Tr and AC regions (Area ×Group F(1, 155) = 23.87, P < 0.0001). G Same as F, but for the mPFC area (Area ×Group F(1, 222) = 1.64, P = 0.2016). Two-way ANOVA with Fisher’s LSD for multiple comparisons. H Intersectional viral strategy for Drd2 knockdown of mPFC. Scale bar, 1 mm. I Axonal innervations (left panel) and newly formed synapse contacts (right panel) in S1Tr. Green arrows indicate boutons, white boxes represent the newly contacts. Scale bar, 20 μm, scale bar = 10 μm, scale bar = 2 μm. J, K A barplot of axonal innervations and synapse contacts in the S1Tr, respectively. J t(99) = 11.18. Each group, n = 9. K t(100) = 7.481. Each group, n = 9. L Same as I, but for the AC area. M, N Same as J, K, but for the AC area. M t(57) = 8.937. Scramble, n = 8. Drd2-KD, n = 9. N t(63) = 5.838. Each group, n = 9. Independent-samples t-test. Data are represented as mean ± s.e.m.

The influence of dopaminergic receptor-II knockdown on the synapse innervations to target neurons was analyzed by detecting the synapse contacts between FP-labeled axon boutons from presynaptic neuron and YFP-labeled dendritic spines on postsynaptic neuron. In comparison with intruder plus scramble control (bottom panels in Fig. 4E), dopaminergic receptor-II knockdown in intruder mice appears to attenuate new synapse contacts made by auditory cortical neuronal axons to S1Tr cortical neurons (left panel), new synapse contacts made by S1Tr cortical neuronal axons to auditory cortical neurons (middle panel) as well as convergent synapse contacts made by the axons from auditory and S1Tr cortices onto medial prefrontal cortical neurons (right panel). The synapse contacts per 100 μm dendrite on S1Tr cortical neurons are 1.60 ± 0.20 in intruder plus scrambles (n = 39 dendrites from 9 mice) and 0.60 ± 0.20 in intruder plus D2R-KD (n = 42 dendrites from 9 mice, p = 0.005, ANOVA; left bars in Fig. 4F). The synapse contacts per 100μm dendrite on auditory cortical neurons are 3.60 ± 0.40 in intruder plus scrambles (n = 37 dendrites from 10 mice) and 0.20 ± 0.10 in intruder plus DR2-KD (n = 41 dendrites from 10 mice, p < 0.001, ANOVA; right bars in Fig. 4F). These data indicate that dopaminergic receptors-II knockdown prevents the stress- induced synapse interconnections between S1Tr and auditory cortices. In addition, the synapse contacts per 100 μm dendrite onto medial prefrontal cortical neurons from S1Tr cortical neurons are 1.2 ± 0.2 in intruder plus scramble (n = 48 dendrites from 9 mice) and 0.40 ± 0.10 in intruder plus DR2-KD (n = 65 dendrites from 12 mice, p < 0.01, ANOVA; left bars in Fig. 4G). Synapse contacts per 100 μm dendrite onto medial prefrontal cortical neurons from auditory cortical neurons are 0.70 ± 0.1 in intruder plus scrambles (n = 48 dendrites from 9 mice) and 0.20 ± 0.09 in intruder plus DR2-KD (n = 65 dendrites from 12 mice, p < 0.01, ANOVA; right bars in Fig. 4G). These data indicate that the dopaminergic receptors-II knockdown prevents stress- induced convergent synapse innervations onto medial prefrontal cortical neurons from S1Tr and auditory cortices.

In the study of the influence of dopaminergic receptor-II knockdown on synapse innervation at target areas of prefrontal cortical neurons, AAV-CMV-GFP was injected in the medial prefrontal cortex (Fig. 4H) and GFP-labelled axon boutons of prefrontal cortical neurons were screened in S1Tr and auditory cortices. In comparison with intruder plus scramble control, the dopaminergic receptor-II knockdown in intruder mice appears to lower axon projections from medial prefrontal cortical neurons to the S1Tr cortex (left panel in Fig. 4I) and the synapse contacts of the axons onto S1Tr cortical neurons (right panel). Axon boutons per mm³ in the S1Tr cortices are 1.7 ± 0.01 ×10⁴ in intruder plus scramble (n = 54 cubes from 9 mice) and 0.1 ± 0.01 ×10⁴ in intruder plus DR2- KD (n = 47 cubes from 9 mice, p < 0.0001, ANOVA; Fig. 4J). Synapse contacts per 100 μm dendrite on S1Tr cortical neurons are 2.6 ± 0.32 in intruder plus scrambles (n = 39 dendrites from 9 mice) and 0.52 ± 0.09 in intruder plus DR2-KD (n = 63 dendrites from 9 mice, p < 0.0001, ANOVA; Fig. 4K). In addition, the knockdown of dopaminergic receptor-II in intruder mice appears to reduce the axon projection from medial prefrontal cortical neurons to the auditory cortex (left panel in Fig. 4L) and the synapse contacts of these axonal boutons on auditory cortical neurons (right panel), in comparison with intruder plus scramble control. Axonal boutons per mm³ in auditory cortices are 1.4 ± 0.01 ×10⁴ in intruder plus scramble (n = 24 cubes from 8 mice) and 0.0 ± 0 in intruder plus DR2-KD (n = 35 cubes from 9 mice, p < 0.001, ANOVA; Fig. 4M). Synapse contacts per 100 μm dendrite on auditory cortical neurons are 1.1 ± 0.2 in intruder plus scramble (n = 29 dendrites from 9 mice) and 0 ± 0 in intruder plus DR2-KD (n = 36 dendrites from 9 mice, p < 0.001, ANOVA; Fig. 4N). The knockdown of dopaminergic receptors-II in the medial prefrontal cortex prevents the stress-induced reception of axon projections and the innervations of new synapses on S1Tr and auditory cortical neurons from the prefrontal cortex.

The influence of the dopaminergic receptor-II knockdown in the medial prefrontal cortex on the recruitment of associative memory neurons was also examined by in vivo electrophysiological recordings in S1Tr, auditory and medial prefrontal cortical neurons, in the meantime to apply the battle sound and the somatic stimulus (Fig. 5). In comparison with one of medial prefrontal cortical neurons from a scramble control mouse that encodes battle sound and somatic stimulus (blue trace in Fig. 5A), one of medial prefrontal cortical neurons from an intruder plus D2R-KD mouse appears not to encode battle sounds and somatic stimuli (red trace). The percentages of medial prefrontal cortical neurons in response to both sound and somatic signals are 18.26% in intruder plus scrambles (n = 115 neurons from 9 mice) and 4.43% in intruder plus DR2-KD (n = 113 neurons from 9 mice; p < 0.01, X²-test in Fig. 5B). Normalized spike frequencies at medial prefrontal cortical neurons in response to battle sounds are 1.7 ± 0.2 in intruder plus scrambles (n = 115 neurons) and 1.30 ± 0.1 in intruder plus DR2-KD (n = 113 neurons; p < 0.05, ANOVA; left bars in Fig. 5C). Normalized spike frequencies at medial prefrontal cortical neurons in response to the somatic stimulus are 1.80 ± 0.20 in intruder plus scramble (n = 115 neurons) and 1.20 ± 0.1 in intruder plus DR2-KD (n = 113 neurons; p < 0.05, ANOVA; right bars in Fig. 5C). Less recruitment of associative memory neurons in the medial prefrontal cortex by the dopaminergic receptor-II knockdown indicates that dopaminergic receptor-II is required for the social stress- induced recruitment of associative memory neurons.

Fig. 5. Drd2 knockdown weakens the recruitment of associative memory neurons.

A Representative spontaneous and evoked spikes traces. B Proportion of mPFC neurons responsive to AS and SS compared to those showing no response to these signals (Scramble, n = 115 neurons. Drd2-KD, n = 113 neurons. Chi-square test, χ² = 10.80, P = 0.0010). C Normalized spike frequencies in mPFC cortical neurons (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 438) = 0.0006, P = 0.9806. Scramble, n = 115 neurons. Drd2-KD, n = 113 neurons). D Example of bandpass continuous data for spike recordings in the S1Tr. E Lower proportion of response to both sound and somatic signals cells in the Drd2-KD group than the Scramble group (Scramble, n = 118 neurons. Drd2-KD, n = 119 neurons. Chi-square test, χ² = 5.943, P = 0.0147). F Statistical graph shows normalized spike frequencies in S1Tr (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 467) = 0.0046, P = 0.9458. Scramble, n = 118 neurons. Drd2-KD, n = 119 neurons). G Same as A, but for the AC area. H Same as B, but for the AC area. (Scramble, n = 126 neurons. Drd2-KD, n = 114 neurons. Chi-square test, χ² = 12.41, P = 0.0162). I Same as C, but for the AC area (Two-way ANOVA with Tukey’s Multiple Comparisons. Stimulus × Group F(1, 468) = 0.4116, P = 0.5215. Scramble, n = 126 neurons. Drd2-KD, n = 114 neurons). S1Tr primary somatosensory cortex, AC auditory cortex, mPFC medial prefrontal cortex. Data are represented as mean ± s.e.m. **P < 0.01, *P < 0.05.

The influence of dopaminergic receptor-II knockdown in the media prefrontal cortex on the recruitment of associative memory neurons was also examined in S1Tr and auditory cortices. In comparison with one of S1Tr cortical neurons from a scramble control mouse that encodes both battle sound and somatic stimulus (blue trace in Fig. 5D), one of S1Tr cortical neurons from an intruder plus DR2-KD mouse appears not to encode such two signals (red trace in Fig. 5D). The percentages of S1Tr cortical neurons in response to battle sound and somatic signals are 20.80% in intruder plus scrambles (n = 118 neurons from 9 mice) and 5.26% in intruder plus DR2-KD group (n = 119 neurons from 9 mice; p < 0.05, X²-test in Fig. 5E). Normalized spike frequencies at S1Tr cortical neurons in response to battle sound are 1.9 ± 0.2 in intruder plus scramble (n = 118 neurons) and 1.3 ± 0.1 in intruder plus DR2-KD (n = 119 neurons, p < 0.05, ANOVA; left bars in Fig. 5F). Normalized spike frequencies at S1Tr cortical neurons in response to the somatic stimulus are 2.0 ± 0.20 in intruder plus scrambles (n = 118 neurons) and 1.5 ± 0.1 in intruder plus D2R-KD (n = 119 neurons; p < 0.05, ANOVA; right bars in Fig. 5F). Thus, the reductions in the recruitment of S1Tr cortical neurons to encode two signals and in their activities by the dopamine receptor-II knockdown in the medial prefrontal cortex strongly indicate that the dopaminergic receptor-II is required for the stress-induced recruitment of associative memory neurons in the S1Tr cortex and interactions between these two cortices.

Moreover, compared with one of auditory cortical neurons from a scramble control mouse that encodes somatic stimulus and battle sound (blue trace in Fig. 5G), one of auditory cortical neurons from an intruder plus D2R-KD mouse appears not to encode these two signals (red trace in Fig. 5G). The percentages of auditory cortical neurons in response to sound and somatic signals are 16.59% in intruder plus scramble control (n = 126 neurons from 9 mice) and 6.72% in intruder plus D2R-KD (n = 114 neurons from 9 mice; p < 0.05, X²-test in Fig. 5H). Normalized spike frequencies at auditory cortical neurons in response to battle sound are 1.90 ± 0.2 in intruder plus scrambles (n = 126 neurons) and 1.2 ± 0.1 in intruder plus DR2-KD (n = 114 neurons, p < 0.01, ANOVA; left bars in Fig. 5I). Normalized spike frequencies at auditory cortical neurons in response to somatic stimulus are 1.7 ± 0.1 in intruder plus scrambles (n = 126 neurons) and 1.2 ± 0.1 in intruder plus DR2-KD (n = 114 neurons, p < 0.05, ANOVA; right bars in Fig. 5I). Thus, the reductions in the recruitment of auditory cortical neurons to encode these two signals and in their activities by dopaminergic receptor-II knockdown in the medial prefrontal cortex indicate that dopamine receptor-II is required for the stress-induced recruitment of associative memory neurons in the auditory cortex and interactions between these two cortices.

It is noteworthy that the roles of dopaminergic receptor-II in the recruitment of associative memory neurons by the electrophysiological study is also granted by the effects of dopaminergic receptor-II knockdown on the proportions of associative memory neurons versus non-associative memory neurons in medial prefrontal, S1Tr and auditory cortices (Figure SR3). As the type two of dopaminergic receptors as metabotropic receptor act onto G-protein coupled intracellular signal pathways [159], the formation of synapse interconnections among auditory, S1Tr and mPFC as well as the recruitment of associative memory neurons in these cortical areas induced by the social stress may be initiated by G-protein coupled intracellular signal pathways.

Morphological and functional studies above suggest that the dopaminergic receptor-II in the medial prefrontal cortex plays the important role in the stress-induced recruitment of associative memory cells and strengthening of their activities in the prefrontal, somatic and auditory cortices based on their interconnections and interactions. If such associative memory neurons are critical for stress-induced fear memory and schizophrenia-like behaviors, the dopaminergic receptors-II knockdown in the prefrontal cortex is expected to prevent fear memories and schizophrenia-like behaviors.

Dopaminergic receptors-II knockdown prevents fear memory and schizophrenia

After the mice in groups of intruder plus DR2-KD and intruder plus scramble experienced the resident/intruder paradigm, the emergence of their fear memory specific to resident CD1 mouse was examined by the social interaction test. Heat-maps in Fig. 6A show that one C57 mouse in intruder plus DR2-KD group appears not to stay away from the interaction zone in the presence of the resident mouse, in comparison with its performance in the absence of this resident mouse as well as the stay in the interaction zone from one mouse in intruder plus scramble group. The differences of the stay time in the interaction zone between the presence of the resident mouse and the absence of this resident mouse are -59.5 ± 9 seconds in intruder plus scramble mice (n = 17) and -15.8 ± 3.9 seconds in intruder plus DR2-KD mice (n = 17, p < 0.001, ANOVA; Fig. 6B). Less avoidance to the resident mouse in intruder plus DR2-KD mice indicates that the dopaminergic receptor-II knockdown in the medial prefrontal cortex prevents the stress-induced fear memory specific to the resident mouse.

Fig. 6. Drd2 knockdown alleviates fear memory and schizophrenia-like behaviors.

A Heatmaps demonstrate the impact of Drd2 knockdown in the open field in response to CD1+ and CD-. B A barplot of the staying time in the interaction zone comparing Scramble and Drd2-KD mice (t(32) = 4.411, P = 0.001. n = 17 for Scramble and Drd2-KD groups mice). C Percentage statistics of the stay time in open arms (t(24) = 6.331, P < 0.0001. Each group, n = 13 mice). D Depicts alterations during the Y-maze test in the arm containing a conspecific companion (t(22) = 5.445, P < 0.001. n = 10 and 14 for Scramble and Drd2-KD groups mice). E Sucrose preference percentage of Scramble and Drd2-KD groups (t(28) = 3.677, P = 0.0001. Each group, n = 15 mice). F Modified pre-pulse inhibition test of different decibels following Drd2 knockdown. G Lower pre-pulse response intensity ratio at 120 dB in the Drd2-KD group than the Scramble group (t(28) = 2.086, P = 0.0462. Each group, n = 15 mice). H Depicts the impact of Drd2 knockdown on mouse jumping times at 70 dB and 80 dB (Chi-square test, 80 dB, χ² = 4.167, P = 0.041. 70 dB: χ² = 0.667, P = 0.414). I Changes in response strengths under 90 dB and 80 dB conditions (Two-way ANOVA with Tukey’s Multiple Comparisons. dB × Group F(1, 30) = 0.076, P = 0.7842. 90 dB, Scramble, n = 11. Drd2-KD, n = 5. 80 dB, Scramble, n = 11. Drd2-KD, n = 7). J, K The effect of Drd2 knockdown on changes tremor levels. Drd2-KD mice appear less frequent body tremor (t(20) = 4.072, P = 0.006. n = 10 and 12 for Scramble and Drd2-KD groups mice). L, M The impact of Drd2 knockdown on alterations in the angular levels. Drd2-KD mice displayed lower fluctuations of included angles in body arch (t(20) = 3.219, P = 0.0043. n = 10 and 12 mice). N, O Representative traces and statistics of motion distance (t(21) = 3.219, P < 0.0001. n = 11 and 12 mice). P A diagram elucidates the role of Drd2 knockdown in the formation of fear memory and the manifestation of schizophrenia-like behaviors. The darkness of color represents the severity. Data are represented as mean ± s.e.m. Details of the statistical information are provided in Supplementary Data 1.

As described previously, the schizophrenia-like behaviors were measured by the pre-pulse inhibition test for their hypersensitivity and hallucination, the persecutory delusion test for their persecutory delusion, the elevated-plus maze test for their anxious state as well as the sucrose preference and Y-maze tests for their depressive mood.

The anxiety-like behaviors were examined by an elevated-plus maze. One mouse in intruder plus DR2-KD group appears to stay in open arms, while a mouse in intruder plus scramble group appears not to access these open arms (Figure SR4A). The percentages of the stay time in open arms to the total time on the elevated-plus maze are 4.80 ± 0.7% in intrude plus scramble control mice (n = 13) and 12.9 ± 1.1% in intruder plus DR2-KD mice (n = 13; p < 0.001, ANOVA; Fig. 6C). The dopaminergic receptor-II knockdown in the medial prefrontal cortex prevents the stress-induced anxiety in those intruder mice with fear memory.

The depression-like behaviors were examined by the sucrose preference test for anhedonia and the Y-maze test for loss of interest and social withdrawal. Heat-maps from the Y-maze test in Figure SR4B show that one mouse in intruder plus D2R-KD group appears not to stay away from the interaction arm including its confidante, compared with one mouse in intruder plus scramble group. The values of the stay time in the interaction arm are 42.5 ± 2.1 seconds in intruder plus DR2-KD mice (n = 10) and 23.7 ± 2.8 seconds in intruder plus scramble mice (n = 14, p < 0.001, ANOVA; Fig. 6D). These data indicate that the mice in an intruder plus D2R-KD group remain interested in their confidantes during the social activity. In the sucrose preference test, the percentages of sucrose water ingestion in total water ingestion are 57.1 ± 2.2% in intruder plus DR2-KD mice (n = 15) and 46.8 ± 1.8% in intruder plus scramble mice (n = 15, p < 0.001, ANOVA; Fig. 6E). These data indicate that the mice in intruder plus DR2-KD express a sugar preference. The dopaminergic receptor-II knockdown in the medial prefrontal cortex prevents the stress-induced depression-like behaviors in intruder mice with fear memory.

The hypersensitivity of schizophrenic mania was examined by the pre-pulse inhibition test and the response threshold to stimulus pulses (Fig. 6F). The relative response strength of the mice in the pre-pulse inhibition test was calculated by the ratio of the differences of responses between pulse two and pulse one over responses to pulse one. The values about this ratio are -0.07 ± 0.02 in intruder plus scramble mice (blue symbols in Fig. 6G; n = 15) and -0.14 ± 0.02 in intruder plus DR2-KD (red symbols; n = 15; p < 0.05, ANOVA). In addition, the response threshold to sound pulses was minimal sound pulses for mice to jump, and the response strength was a ratio of the dynamic weight due to their jumps to the static weight. In Fig. 6H, response thresholds are 80 dB for intruder plus DR2-KD mice (red bar, n = 10/75) and 70 dB for intruder plus scramble mice (blue bar, n = 6/75). Response strengths to sound pulses at 80 dB are 1.11 ± 0.02 in intruder plus DR2-KD mice (n = 7) and 1.19 ± 0.03 in intruder plus scramble (n = 11, p < 0.05, ANOVA; right bars in Fig. 6I). Response strengths to sound pulses at 90 dB are 1.11 ± 0.02 in intruder plus DR2-KD mice (n = 5) and 1.21 ± 0.03 in intruder plus scramble mice (n = 11, p < 0.05, ANOVA; left bars in Fig. 6I). These data indicate that the knockdown of dopaminergic receptor-II in the medial prefrontal cortex prevents stress-induced hypersensitivity to sound signals, i.e., schizophrenia-like mania.

The delusion of schizophrenic mania in mice was examined by the persecutory delusion test. Fear responses to the resident mouse were featured by the mouse back arch to small included- angles, the frequent shaking of their bodies and the less movements in the open field. If intruder mice expressed these behaviors without the presence of the resident mouse, they were thought to be the emergence of the persecutory delusion. As shown in Fig. 6J, one mouse in intruder plus DR2-KD group appears less frequent tremor in the body (red trace) than a mouse in intruder plus scramble control group (blue trace). The shaking frequencies are 2.6 ± 0.6 Hz in intrude plus scramble mice (n = 10) and 0.4 ± 0.1 Hz in intruder plus DR2-KD mice (n = 12; p < 0.001, ANOVA; Fig. 6K). Moreover, the included angles of back arch appear larger in a mouse in intruder plus DR2-KD group (red trace in Fig. 6L) than a mouse in intruder plus scramble control group (blue trace). The included angles of back arch (degrees) are 76.7 ± 6.5 in intruder plus scramble mice (n = 10) and 98.9 ± 3.3 in intruder plus DR2- KD mice (n = 12; p < 0.01, ANOVA; Fig. 6M). In addition, the motions in an open field appear more active in a mouse in intruder plus DR2-KD group (red trace in Fig. 6N) than a mouse in intruder plus scramble control (blue trace). Motion distances in this open field (cm/10 min) are 0.17 ± 0.01 ×10⁵ in intruder plus scramble mice (n = 11) and 1.02 ± 0.01 ×10⁵ in intruder plus DR2-KD mice (n = 12; p < 0.001, ANOVA; Fig. 6O). Therefore, the dopamine receptor-II knockdown may prevent the stress-induced persecutory delusion including frequent body tremors, dominant body arch and less movement in the absence of scared signals. Figure 6P illustrates the diagram about the role of dopaminergic receptor-II in schizophrenia-like behaviors.

Dopaminergic receptors-II antagonist suppresses fear memory and schizophrenia

In addition to the prevention of the recruitment of associative memory neurons for the fear memory and schizophrenia-like behaviors, we studied whether the blockade of the dopaminergic receptor-II suppressed the function of the stress-recruited associative memory neurons and their encoded schizophrenia-like behaviors. In intruder mice with fear memory and schizophrenia, we injected dopaminergic receptor-II antagonist eticlopride [154–156] in bilateral medial prefrontal cortices (Fig. 7A). Compared with the saline injection (bottom trace in Fig. 7B), the eticlopride (1.325 nM) injection appears to block the responses of associative memory neurons in the medial prefrontal cortex to the battle sound and the somatic stimulus (top trace). The normalized spike frequencies of associative memory neurons in response to the battle sound are 2.80 ± 0.5 in self-controls and 0.90 ± 0.2 after an eticlopride uses (left symbols in Fig. 7C; n = 10 neurons from 10 mice, p < 0.01, ANOVA). Their normalized spike frequencies in response to somatic stimulus are 4.2 ± 0.9 in self-controls and 1.7 ± 0.4 after an eticlopride uses (right symbols in Fig. 7C, n = 10 neurons from 10 mice, p < 0.05, ANOVA). Furthermore, the normalized spike frequencies of these associative memory neurons in response to the battle sound are 3.0 ± 0.5 in self-controls and 3.0 ± 0.7 after the saline uses (left symbols in Fig. 7D; n = 9 neurons from 9 mice, p = 0.943, ANOVA). Their spikes in response to the somatic stimulus are 4.1 ± 0.6 in self-controls and 3.0 ± 0.3 after the saline uses (right symbols in Fig. 7D, n = 9 neurons from 9 mice, p = 0.123, ANOVA). The inhibition of dopaminergic receptor-II suppresses the function of associative memory cells in the medial prefrontal cortex.

Fig. 7. DR2 antagonist suppresses fear memory and schizophrenia-like behaviors.

A Schematic of DR2 antagonist (Eticlopride, Eti) injection into mPFC. B Example traces of spontaneous and evoked spikes recording before and after Eti application. Blue triangles point to the application of antagonist, blue dashed-line boxes indicate the reapplication of two stimuli (AS and SS). C Discharge frequency statistics standardized for Eti (AS, t(9) = 3.392, P = 0.0080. SS, t(9) = 2.731, P = 0.0232. n = 10 mice). D Same as C, but for the Saline group (AS, t(8) = 0.0742, P = 0.9428. SS, t(8) = 1.723, P = 0.1232. n = 9 mice). E Effect of Eti on the duration of staying in the interaction zone (t(16) = 2.933, P = 0.0098. Each group, n = 9 mice). F The Eti group displayed a higher percentage of staying in the open arm during the EPM (t(16) = 5.026, P = 0.0001. Each group, n = 9 mice). G Interaction time of mice in the Y-maze (t(16) = 3.280, P = 0.0047. Each group, n = 9 mice). H Sucrose preference percentage of Eti and saline groups (t(16) = 2.414, P = 0.0281. Each group, n = 9 mice). I Lower pre-pulse response intensity ratio at 120 dB in the Eti group than the saline group (t(18) = 2.713, P = 0.0143. Each group, n = 10 mice). J Impact of Eti on mouse jumping times at 70 dB and 80 dB (Chi-square test, 80 dB, χ² = 10.169, P = 0.001. 70 dB: χ² = 0.207, P = 0.649). K The Eti group displayed a lower response intensity index after resident/intruder paradigm (Two-way ANOVA with Tukey’s Multiple Comparisons. dB × Group F(1, 24) = 0.0008, P = 0.9778. 90 dB, Eti, n = 4. Saline, n = 8. 80 dB, Eti, n = 6. Saline, n = 10). L, M The effect of Eti on changes in mice tremor levels. Eti mice appear less frequent body shaking (t(18) = 5.221, P < 0.0001. Each group, n = 10 mice). N, O Included angles curve and statistical graph (t(18) = 2.540, P = 0.0205. Each group, n = 10 mice). P, Q Representative traces and statistics of motion distance (t(18) = 2.192, P = 0.0418. Each group, n = 10 mice). AS auditory stimulus, SS somatic stimulus, Data are represented as mean ± s.e.m.

We also examined the effectiveness of eticlopride-suppressed associative memory neurons in the medial prefrontal cortex on fear memory and schizophrenia-like behaviors in intruder mice. In the social interaction test, the differences of the stay time in the interaction zone between the presence of the resident mouse and the absence of this resident mouse are -83.0 ± 11.5 seconds in the saline use (n = 9 mice) and -30.4 ± 13.8 seconds in the eticlopride use (n = 9 mice, p < 0.01, ANOVA; Fig. 7E). Less avoidance to the resident mouse by eticlopride uses indicates that the dopaminergic receptor-II blockade in the medial prefrontal cortex relieves the stress-induced fear memory. In the anxiety test, the percentages of the stay time in open arms to the total time on the elevated-plus maze are 1.0 ± 0.6% in the saline use (n = 9) and 9.20 ± 1.5% in the eticlopride use (n = 9 mice; p < 0.001, ANOVA; Fig. 7F). Thus, the dopaminergic receptor-II blockade relieves stress-induced anxiety-like behaviors. In the Y-arm test, the values of the stay time in interaction arm are 22.80 ± 5.20 seconds in the saline use (n = 9 mice) and 45.6 ± 4.6 seconds in the eticlopride use (n = 9 mice, p < 0.01, ANOVA; Fig. 7G). In the sucrose preference test, the percentages of sucrose water ingestion in total water ingestion are 43.2 ± 4.7% in the saline use (n = 9 mice) and 55.4 ± 1.9% in the eticlopride use (n = 9 mice, p < 0.05, ANOVA; Fig. 7H). Thus, the dopaminergic receptor-II blockade in the medial prefrontal cortex relieves the stress-induced depression-like behaviors.

In the pre-pulse inhibition test for the hypersensitivity of schizophrenic mania, the ratios of the differences between responses to pulse two and pulse one over responses to pulse one are 0.04 ± 0.03 in a saline use (n = 10 mice) and -0.08 ± 0.03 in an eticlopride use (n = 10; p < 0.05, ANOVA; Fig. 7I). In the measurement of the response threshold to those sound pulses, the thresholds in response to minimal sound pulses for the mice to jump are 80 dB for eticlopride use (n = 8/75) and for the saline use (n = 24/75; left bars in Fig. 7J). Eticlopride raises the response threshold. The strengths in response to sound pulses at 80 dB are 1.12 ± 0.02 in a saline use (n = 10 mice) and 1.08 ± 0.01 in an eticlopride use (n = 6 mice, p < 0.05, one-way ANOVA; right bars in Fig. 7K). The strengths in response to sound pulses at 90 dB are 1.13 ± 0.01 in the saline use (n = 8 mice) and 1.09 ± 0.01 in the eticlopride use (n = 4 mice, p < 0.05, ANOVA; left bars in Fig. 7K,). Therefore, dopaminergic receptor-II blockade in the medial prefrontal cortex relieves the stress-induced hypersensitivity in schizophrenic mania.

The delusion of schizophrenic mania in the intruder mice was examined by the persecutory delusion test. In Fig. 7L, one DR2-blockade mouse appears less frequent shaking in the body (red trace) than a saline-use mouse (blue trace). Their body-shaking frequencies are 1.6 ± 0.1 Hz in a saline use (n = 10 mice) and 1.0 ± 0.1 Hz in an eticlopride use (n = 10 mice; p < 0.001, ANOVA; Fig. 7M). Moreover, the included angles of mouse back arch appear larger in a DR2- blockade mouse (red trace in Fig. 7N) than one saline-use mouse (blue trace). The included angles of mouse back arch (degrees) are 75.20 ± 3.7 in a saline use (n = 10 mice) and 85.20 ± 1.3 in an eticlopride use (n = 10 mice; p < 0.05, ANOVA; Fig. 7O). In addition, a mouse with DR2-blocake in the prefrontal cortex appears more motions in the open field (red trace in Fig. 7P), in comparison with one saline use mouse (blue trace). The motion distances (cm/10 min) in this open field are 0.5 ± 0.01 ×10⁵ in a saline use (n = 10 mice) and 1.0 ± 0.1×10⁵ in an eticlopride use (n = 10 mice; p < 0.05, ANOVA; Fig. 7Q). Therefore, the blockade of dopaminergic receptor-II in the medial prefrontal cortex relieves the stress-induced persecutory delusion.

Discussion

The social stress by a resident/intruder paradigm induces mice to express fear memory and schizophrenia-like behaviors, in which the stress signals include the battle sound and the painful signal from somatic injury areas (Fig. 1). The social stress evokes new synapse interconnections among the neurons in medial prefrontal, auditory and S1Tr cortices in the intruder mice with fear memories and schizophrenia-like behaviors (Fig. 2). Certain neurons in these cerebral cortices from intruder mice received the convergent synapse innervations and became able to encode the battle sound and the somatic stimulus, i.e., associative memory neurons (Fig. 3), in which such associative memory neurons in the medial prefrontal cortex are the second order, compared with primary associative memory neurons in auditory and S1Tr cortices. The recruitment of associative memory neurons as well as the emergence of fear memory and schizophrenia-like behaviors are downregulated by dopamine receptor-II knockdown in the medial prefrontal cortex (Figs. 4–6). The function of associative memory neurons and the expression of schizophrenia-like behaviors are precluded by the antagonist of dopaminergic receptors-II (Fig. 7). These data indicate that dopaminergic receptor-II plays the crucial roles in the recruitment and the function of associative memory neurons in mPFC-centered neural circuits correlated to fear memory and schizophrenia.

The social stress leads to the fear memory for posttraumatic stress disorders and the weird memory for schizophrenia [15, 18, 20, 48–57, 60, 61, 63–69]. These types of stress-correlated memories and waning cognition are presumably encoded by the neurons in the prefrontal cortex [70–73, 76–78]. Schizophrenic patients are associated with by the abnormality of the prefrontal cortex [72, 79–81, 83–89]. The prefrontal cortex has been thought of as the main target for the antipsychotics in schizophrenia [90]. To questions how the prefrontal cortex becomes the center of neural circuits and how these cortical neurons are recruited to encode the stress signals for fear memory and schizophrenia, our data indicate that the synapse interconnections among medial prefrontal, auditory and S1Tr cortices emerge in stress-induced fear memory and schizophrenia- like behaviors. Certain neurons in any one of these three cortices become to receive convergent synapse innervations from other two cortices and to encode the fear signals including the battle sound and the somatic stimulus, i.e., the recruitment of associative memory neurons, in the mice with fear memory and schizophrenia-like behavior (Figs. 2, 3). Importantly, the downregulation of the formation and the function of associative memory neurons in the medial prefrontal cortex precludes fear memory and schizophrenia-like behaviors (Figs. 4–7). Thus, associative memory neurons are recruited in mPFC-centered neural circuit essential for stress-induced fear memory and schizophrenia. A fact that the downregulation in the activity of the medial prefrontal cortex blocks the synapse innervations from sensory cortices to the mPFC by reducing its attraction to axons and from the mPFC to sensory cortices by its active projection strengthens a hypothetical principle about coactivity together and interconnection together among the neurons [65]. In addition, our current study also indicates that mPFC-centered neural circuits also include the synapse interconnections of the mPFC with the visual cortex, entorhinal cortex, ventral tegmental area, amygdala, ventral hippocampus and substantial nigra for fear memories and schizophrenia-like behaviors.

In terms of the order of associative memory neurons in medial prefrontal, auditory and S1Tr cortices, associative memory neurons in sensory cortices are presumably primary, and those in the medial prefrontal cortex are likely secondary [65]. This viewpoint is based on the following facts. Neural signals in the brain flow from sensory cortices to their downstream regions, such as the prefrontal cortex [160, 161]. In physiology, associative memory neurons have been found in sensory cortices that bring various sensory signals from the olfactory, gustatory and tactile systems [93–99] and in the prefrontal cortex whose synapse inputs come from sensory cortices [100, 101, 162]. In pathology, the schizophrenia-correlated neural circuits include the interactions between the prefrontal cortex and the thalamus-sensory cortices [36, 91]. The associative memory neurons that encode stressful signals have been detected in auditory and somatosensory cortices [99, 102, 103]. Particularly, the function of associative memory neurons in the medial prefrontal cortex is attenuated by blocking neuronal activities in auditory and S1Tr cortices (Fig. 3). Therefore, the associative memory neurons in the medial prefrontal cortex are secondary and the associative memory neurons in sensory cortices are primary. Taken all of these data with stress-induced synapse interconnections among medial prefrontal, auditory and S1Tr cortices (Fig. 2), we suggest that the neural circuit including the associative memory neurons in the medial prefrontal, auditory and S1Tr cortices is formed during the social stress for the formation of fear memory and the occurrence of schizophrenia (Fig. 8). It is noteworthy that the downregulation of mPFC activities by Drd2 knockdown attenuates the recruitment and function of associative memory cells in S1Tr and auditory cortices (Figs. 4, 5), strengthening a viewpoint about the interaction of associative memory cells between sensory cortices and mPFC [65]

Fig. 8. The diagram of the changes of neural circuits and psychological behaviors.

The left half of the figure shows that social stress induced by the resident/intruder paradigm leads to fear memory specific to a CD1 resident mouse and schizophrenia-like behaviors. It induces synapse interconnections among medial prefrontal, auditory, and S1Tr cortical neurons. These cortical neurons are recruited as associative memory neurons, which are characterized by receiving new synaptic innervations and encoding stressful signals, including battle sounds and somatic stimuli. The right half of the figure shows that dopaminergic receptor-II knockdown in the medial prefrontal cortex prevents the recruitment of associative memory neurons and the onset of fear memory and schizophrenia-like behaviors.

Dopaminergic synapse transmissions in mesolimbic and mesocortical pathways are thought to be imbalance in schizophrenia patients [6, 29, 30, 32–36]. The antagonists of dopaminergic receptors-II have been applied to treat the mania-dominant signs of schizophrenia patients for decades [4, 10, 37, 38, 40–42]. Cellular targets for the actions of dopaminergic receptors-II antagonists, particularly in the formation and function of schizophrenia-correlated neural circuits, are largely unknown [43, 45–47]. Our data demonstrate that an application of eticlopride (dopaminergic receptor-II antagonist) in the medial prefrontal cortex relieves fear memory and schizophrenia-like behavior (Fig. 7), indicating the consistency of schizophrenia relief in our animal model with clinical data. This application of eticlopride is able to block the response of associative memory neurons in the medial prefrontal cortex to fear signals (Fig. 7A), indicating the cellular targets of dopamine receptor-II antagonists. Furthermore, the knockdown of dopaminergic receptors-II in the medial prefrontal cortex prevents the new synapse interconnections among medial prefrontal, auditory and S1Tr cortices (Fig. 2) as well as the recruitment of associative memory neurons in these cortical areas, which are correlated to fear memory and schizophrenia (Figs. 2, 3). Our studies reveal that the downregulation of dopaminergic receptors-II can be used to treat schizophrenia and to prevent the formation of schizophrenia-correlated neural circuits as well, which provides new insights for designing the therapeutic strategies of stress-induced schizophrenia, especially individuals with high vulnerability and minor symptoms/signs. It should keep in mind about the effective period in the uses of DR2 knockdown and antagonists. The DR2 knockdown in the mPFC can prevent the emergence of stress-induced schizophrenia-like behaviors and the recruitment of the synapse interconnections and associative memory cells permanently. The DR2 blockade by its antagonist in the mPFC attenuates the expression of stress-induced schizophrenia-like behaviors and cellular alternations that have been formed, in which the effective period depends upon its removal by the local circulations of the blood and brain-spinal fluid.

The symptoms and signs of schizophrenia are featured by the mania including hallucination and delusion as well as the negative mood including social withdrawal, anhedonia and anxiety [1–6, 10]. The identification of schizophrenia onsets in animal models has been based on a pre-pulse inhibition test [116, 117]. In addition to identifying stress-induced schizophrenic mania, we have developed approaches to merit response threshold and response strength to sound pulses in the pre-pulse inhibition, a modified pre-pulse inhibition, for the assessment of mania-relevant hypersensitivity. We have also developed approaches to merit body tremors, back arches and motion state in the absence of the scared signals, which normally express in the presence of the scared signals (e.g., resident CD1 mice), for the assessment of persecutory delusion. As those data from the modified pre-pulse inhibition test and the persecutory delusion test are linearly correlated (Figure SR5), the two tests for the indices of the hallucination and the persecutory delusion can be jointly used to assess schizophrenic mania and misbelief. Our studies provide new approaches for assessing schizophrenia-correlated behaviors. Taken these measurements with the assessment of negative moods by the elevated-plus maze test for anxiety, the sucrose preference test for anhedonia and the Y-maze test for loss of interest, our studies to estimate the schizophrenia-like behaviors are more convincing, compared with previous studies by using one of the merits.

The cortical neurons after their coactivity are synaptically interconnected and become able to encode the associative signals inputted from multiple sources of the synapses. These recruited associative memory neurons work for the joint storage and reciprocal retrieval of the associated signals under physiological conditions [93–101, 162]. In the social stress, learning the associated stressful signals induces the onset of fear memories and the recruitment of associative memory cells that encode these stress signals in the animals that express stress-induced schizophrenia-like behaviors. This pathological memory and associative memory neurons are associated with anxiety-, depression- and schizophrenia-like behavior (Figs. 1–3). Our data endorse the findings that the acute stress induces fear memory for anxiety, the chronic stress induces memories to negative outcome and defeat for depression, and the severe stress induces weird memory in relevance to schizophrenia [15, 18, 20, 48–61, 63–69]. Our data also indicate a progressive chain from stress-induced fear memory and posttraumatic stress disorder toward bipolar disorder and schizophrenia [74, 75]. It seems to be possible that the associative memory neurons can be recruited specifically to associate physiological signals or pathological signals and to be involved in the relevant processes [65]. Our data strengthen a concept of associative memory cells as basic units in memory traces or engrams for associative learning and memory [65]. This strengthened conclusion is encouraging future investigations in memory-relevant events to examine whether the associative memory neurons and their assemblies by the synapse interconnections are recruited, in addition to the molecular markers, such as immediate early genes that are nonspecific for showing the activity strength of various cellular events.

In addition to revealing the recruitments of associative memory neurons in mPFC-centered neural circuits essential for stress-induced fear memories and psychological disorders, we have extended our study to the prevention of these pathological changes and the transmission of fear memories among the involved rodents. The psychological stress by observing a resident/intruder paradigm induces the recruitment of associative memory neurons in the medial prefrontal cortex and sensory cortices for encoding the fear memories by the dynamical axon transportation. This stress-induced fear memories can be prevented and weakened by the social interactions among those mice that experience the psychological stress, i.e., their social interactions raise the chance for them to be resilience of suffering from psychosis. The social interactions among the mice with the psychological stress attenuate the recruitment of associative memory neurons for encoding fear memory in sensory cortices by intracellular signaling cascades. Moreover, this stress-induced fear memory can be transmitted from the mice with stress-induced fear memories to their living partners that have never experienced this psychological stress. The associative memory neurons of encoding stress signals are recruited in the auditory cortex of those living partners. These data grant the hypotheses that the social interaction can reduce the likelihood of suffering from the psychosis in stress-experienced subjects, and endorses the confidantes without experiencing the stress to be the alertness about the harmfulness of the stresses, all of which are based upon the recruitment of associative memory cells of encoding these stressful signals [65].

In terms of molecular cascades from dopaminergic receptors-II to the formation of synapse interconnection and the recruitment of associative memory neurons correlated to fear memory and schizophrenia, our thoughts are given below. The dopaminergic receptors-II are coupled with G-proteins that link to some intracellular signal cascades, such as PIP2-turnover, protein kinases, genes’ expression and DNA methylation [163–167]. These molecules in turn trigger the activity of brain cells including the morphological extension of neuronal processes, the formation of synapses and the excitability of neurons. Based on our previous analysis in molecular profiles by high throughput sequencing of miRNA and mRNA, dopaminergic receptors-II are upregulated, and numerous signals, such as cAMP pathway, Wnt pathway, cell adhesion molecules and other kinds of synapse pathways, are involved in this stress-induced fear memory by resident/intruder paradigm. KEGG analysis indicates the links of dopaminergic receptor-II with many of these signal molecules [112], indicating the interactions among dopaminergic receptor-II and many of such signal pathways. Although our current study is focused on the role of dopaminergic receptor-II in the recruitment of associative memory neurons correlated to fear memory and schizophrenia, the involvement of these molecules is worthy to be examined in the future studies, especially to find out the primary molecules.

In summary, the formation of new synapse interconnection among cross-modal cortices and the recruitment of associative memory neurons to encode the stress signals for fear memory and schizophrenia-like behaviors are functionally and morphologically identified in a mouse model of the social stress by a resident/intruder paradigm. The stress-induced psychological behaviors and associative memory neurons in mPFC-centered neural circuit are based on dopamine receptors-II. The diagram in Fig. 8 illustrates the alternations of neural circuits and psychological behaviors. In addition to revealing the cellular target (associative memory cells) for dopaminergic receptor-II antagonists to treat schizophrenic mania, our data present the viewpoint that the formation of schizophrenia-correlated neural circuits is based on dopaminergic receptors-II, i.e., dopaminergic receptor-II antagonists may be used to prevent a progress of minor schizophrenia. Moreover, the recruitment of the associative memory neurons that encode the stress-induced fear memory and schizophrenia strengthens the concept of associative memory neurons recruited in other types of associative learning [65]. These indications from our data have not been stated by other studies in the field of memorioscience [102, 103].

Author contributions

Lei Wang, Lichuang Geng, Bingchen Chen, Jiayi Li, Yang Xu contributed to experiments and data analyses. Jiajia Zhang contributed to write the computer software for AI pattern recognition and data analysis. Lei Wang contributed to write figure legends. Jin-Hui Wang contributed to the concepts, project design, paper writing and funds’ acquisition.

Funding

This study is funded by Natural Science Foundation of China (81971027, U2241209 and 81930033) to Jin-Hui Wang.

Data availability

The datasets supporting the conclusions of this article are included within the article and its supplementary file. All of the raw date will be available based on the request of those readers who do not use them as the commercial purpose.

Ethics approval and consent to participate

The experiments were accorded with the guidelines and regulations by the Administration Office of Laboratory Animal in Beijing, China. All experimental protocols were approved by the Institutional Animal Care and Use Committee in Administration Office of Laboratory Animal at Beijing, China (B10831).

Competing interests

All authors declare no competing interest. All authors have approved the final version of the manuscript.

Consent for publication

This study is not relevant to individual’s details, images, videos and voices.

Supplementary information

The online version contains supplementary material available at 10.1038/s41380-025-03388-0.