Introduction

Irritability is a mood disturbance characterized by heightened sensitivity to emotional stimuli, along with an increased propensity for anger, frustration and aggression [1,2,3]. This negative emotional state has not attracted enough attention in biological research, while it is commonly observed in a wide range of psychiatric disorders from childhood to adulthood, including autism spectrum disorder (ASD), attention deficit hyperactivity disorder, disruptive mood dysregulation disorder, anxiety, depression and bipolar disorder [4,5,6]. Irritability could be a symptom of withdrawal from addictive substances, nicotine or alcohol [7,8,9]. Severe and persistent irritability may cause considerable distress, impairment in daily functioning or suicidality [5, 10]. Despite its high prevalence, less is known about the neural mechanisms underlying irritability. In children and adolescents, irritability has been linked to dysfunction in treat and reward processing [1, 6]. Indeed, human functional neuroimaging studies show that decreased functional connectivity between the amygdala and prefrontal cortex underlies increased irritability to face emotion recognition [11], and that people with irritability present exaggerated responses in the anterior cingulate, rostromedial prefrontal and posterior cingulate cortices [12].

Individuals with ASD often show comorbid increase of irritability [13, 14]. A recent meta-analysis found that around 43% of children with ASD have elevated levels of irritability [15]. However, the mechanism linking ASD to yielding irritability remains largely unclear. Genetic factors are the main contributors to ASD pathogenesis and more than 100 risk genes have been identified so far [16, 17]. Multiple ASD-relevant genetic mouse models have been developed for preclinical investigations of pathophysiological mechanisms and therapeutic avenues [18, 19]. We reasoned whether genetic mouse models of ASD could be invaluable tools for investigating mechanisms underlying irritability. Here, we focused on the Coiled-coil and C2 domain containing 1a (Cc2d1a) gene, which encodes a scaffold in routing the signals to multiple intracellular signaling pathways critical for neuronal function [20]. Previous studies have demonstrated that mice with conditional deletion of Cc2d1a in forebrain excitatory neurons recapitulate autistic-like core symptoms of social communication deficits and excessive repetitive behaviors [21,22,23,24,25]. Nonetheless, it remains unclear whether Cc2d1a deficiency could lead to heightened irritability-like behavior.

Aberrant oxytocin (OXT) system is implicated in ASD-related behaviors and social abnormalities, and OXT treatments show beneficial effects on social abnormalities in children with ASD [26, 27]. There is also evidence that knockout mice lacking the ASD-risk genes, contactin-associated protein-like 2 (Cntnap2) and SH3 and multiple ankyrin repeat domains protein 3b (Shank3b), result in reduced central OXT levels, and OXT treatment restores behavioral deficits in these models [28, 29]. In this study, we investigated whether Cc2d1a deletion from forebrain excitatory neurons might lead to increased irritability, and whether exogenous and evoked OXT could rescue aberrant irritability-like behavior in this ASD mouse model. We additionally identified brain areas and interconnected circuits underlying irritability.

Materials and methods

Detailed information about the materials and methods is described in the Supplementary Information.

Animals

We used the following mouse lines: Emx1-Cre mice and Cc2d1a conditional knockout (cKO) mice generated by crossing Cc2d1a-floxed (Cc2d1af/f) mice with Emx1-Cre mice. All experimental procedures were performed in accordance with the regulations of the National Cheng Kung University IACUC (Authorization Approval No. 110077).

Stereotaxic viral injections and chemogenetic manipulation

Stereotaxic viral injections and chemogenetic manipulations were performed as previously described [28, 30].

Irritability-like behavior

The bottle-brush test (BBT) was performed as previously described [31, 32]. A within-subject design was used. Testing consisted of 10 trials with 10-s intertrial intervals per mouse in a clean plastic cage (37.5 cm × 17 cm × 18 cm) with fresh bedding. During each trial, the mouse started at the back of cage. Each trial comprised of five steps in sequence (Fig. 1a). Step 1: the brush was rotated rapidly toward the mouse for ~2 s. Step 2: the brush was rotated around the mouse’s whiskers for ~2 s. Step 3: the brush was then retreated to the front of the cage for ~2 s. Step 4: the brush was rotated at the start site for ~2 s. Step 5: the brush was hung vertically for ~10 s before being removed from the cage. Both aggressive (exploring, following and tail rattling) and defensive responses (escaping, digging, jumping, rearing and grooming) across all trials were determined.

Fig. 1: Male Cc2d1a cKO mice exhibit increased irritability-like behavior in the BBT.
figure 1

a Schematic showing the experimental schedule for the BBT to trigger and assess irritability-like behavior (aggressive and defensive responses). b Comparison of time spent by male and female WT and Cc2d1a cKO mice in aggressive responses to the BBT (mouse number: Nwt/male = 16, NcKO/male = 13, Nwt/female = 10, NcKO/female = 9; two-way ANOVA, genotype: F(1,44) = 5.13, p = 0.029; sex: F(1,44) = 19.31, p < 0.0001; interaction: F(1,44) = 7.78, p = 0.0078). c Comparison of time spent by male and female WT and Cc2d1a cKO mice in defensive responses to the BBT (mouse number: Nwt/male = 16, NcKO/male = 13, Nwt/female = 10, NcKO/female = 9; two-way ANOVA, genotype: F(1,44) = 14.05, p = 0.0005; sex: F(1,44) = 1.27, p = 0.2659; interaction: F(1,44) = 0.013, p = 0.9117). d Comparison of time spent by male WT and Cc2d1a cKO mice in individual aggressive responses to the BBT (mouse number: Nwt = 16, NcKO = 13; two-tailed unpaired Student’s t-test). e Comparison of time spent by male WT and Cc2d1a cKO mice in individual defensive responses to the BBT (mouse number: Nwt = 16, NcKO = 13; two-tailed unpaired Student’s t-test). f Comparison of time spent by female WT and Cc2d1a cKO mice in individual aggressive responses to the BBT (mouse number: Nwt = 10, NcKO = 9; two-tailed unpaired Student’s t-test). g Comparison of time spent by female WT and Cc2d1a cKO mice in individual defensive responses to the BBT (mouse number: Nwt = 10, NcKO = 9; two-tailed unpaired Student’s t-test). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 as compared with as compared with each other group.

Full size image

Immunohistochemistry and quantification

Immunohistochemistry follows protocols as previously described [30]. For quantification of c-Fos immunopositivity, c-Fos+ neurons were determined only when cells were co-localized with DAPI staining.

Statistical analysis

All results are presented as means ± SEM and analyzed by the GraphPad Prism 6 software. Two-tailed paired Student’s t-test was used for within-group comparison. The difference between multiple groups was calculated by two-way ANOVA followed by Sidak’s multiple comparisons post hoc test. Differences were considered as significant at p < 0.05. The statistical results are described in figure legends.

Results

Cc2d1a cKO mice show increased irritability-like behavior

We first examined whether Cc2d1a cKO mice might exhibit heightened irritability-like behavior in response to the BBT (Fig. 1a), in which aggressive and defensive responses to repeated attacks by a mechanical stimulus provide a measure of irritability-like and avoidance-like behavior, respectively [31, 32]. Subject mice were individual housed for 3 days before behavioral testing. To assess whether there was a sex difference related to genotype in irritability-like behavior, we conducted a two-way ANOVA and found a significant effect of genotype, sex and genotype × sex interaction (Fig. 1b). Post hoc analysis showed that male but not female Cc2d1a cKO mice spent more time in aggressive responses than WT mice. We also examined defensive responses in the BBT, and unexpectedly found no sex differences related to genotype (Fig. 1c). When analyzing individual aggressive responses, male Cc2d1a cKO mice spent more time in exploring but not following or tail rattling compared to WT mice (Fig. 1d). When analyzing individual defensive responses, male Cc2d1a cKO mice spent less time in digging but not escaping, jumping, rearing or grooming than WT mice (Fig. 1e). There were no significant differences between female Cc2d1a cKO and WT mice in exploring, following or tail rattling (Fig. 1f). Female Cc2d1a cKO mice showed a significance decrease of time in digging but not escaping, jumping, rearing or grooming compared to WT mice (Fig. 1g).

Cc2d1a cKO mice exhibit a trend toward increased aggression in the resident-intruder test

To explore a more general role of Cc2d1a deletion in aggression, we utilized the resident-intruder test (RIT). Since we observed only male Cc2d1a cKO mice exhibiting increased irritability-like behavior, we hereafter only focused on male mice. To increase aggression to intruder mice, male subject mouse was co-housed with an ovariectomized female mouse for 3 days before the RIT. Since not all subject mice showed aggressive behaviors toward intruder mice, the subject mice that attacked intruder mouse during the testing period are referred as aggressors and subject mice that did not attack are referred as nonaggressors [33]. There was no significant difference in the proportion of aggressors in Cc2d1a cKO compared to WT mice (Fig. S1a). The aggressors exhibited distinct aggressive-like behaviors, including anogenital sniffing, chasing, tail rattling, biting or attack. While there was a trend toward an increase in time spent in anogenital sniffing and chasing, the differences did not reach statistical significance (Fig. S1b, c). We observed no differences between groups in tail rattling, biting or attack, latency to attack or number of attack (Fig. S1d–g). Although the proportion of aggressors was comparable between Cc2d1a cKO and WT mice in 10 min, more aggressors was detected in Cc2d1a cKO than WT mice during the first phase (0–5 min) of the RIT (Fig. S1h).

Cc2d1a cKO mice show selective loss of OXT-expressing neurons in the PVN

Aberrant OXT system has been implicated in ASD-related behaviors and social abnormalities [26, 27]. We analyzed the number of OXT-expressing neurons in the paraventricular nucleus (PVN) (Fig. 2a) and supraoptic nucleus (SON) of the hypothalamus (Fig. 2b). We observed less OXT-immunoreactive neurons in the PVN (Fig. 2c) but not the SON (Fig. 2e) in male Cc2d1a cKO mice, whereas the total cell numbers in the PVN (Fig. 2d) and SON (Fig. 2f) did not differ between groups. However, no significant difference was found in the number of OXT-immunoreactive neurons in the PVN between female Cc2d1a cKO and WT mice (Fig. S2a–c). Because OXT and arginine vasopressin (AVP) neurons constitute most of PVN neurons, we also compared the percentage of AVP-expressing neurons in male Cc2d1a cKO and WT mice. P2X4R immunoreactivity was used to reflect the total number of OXT- and AVP-expressing neurons in the PVN [34]. Consistently, we found a significant decrease in the percentage of OXT-immunoreactive neurons (Fig. 2i) and P2X4R-immunoreactive neurons in the PVN (Fig. 2m) in male Cc2d1a cKO compared to WT mice, whereas no significant difference was found between groups in the percentage of OXT+P2X4R+ neurons to total P2X4R+ neurons (Fig. 2j). Male Cc2d1a cKO mice showed a significance decrease in the percentage of P2X4R-immunoreactive neurons (Fig. 2o) but not AVP-immunoreactive neurons (Fig. 2k) compared to WT mice. Although there was a trend toward increase in the percentage of AVP+P2X4R+ neurons to total P2X4R+ neurons (Fig. 2l), the difference did not reach statistical significance. The total cell numbers in the PVN were comparable between male Cc2d1a cKO and WT mice (Fig. 2n, p). Since Emx1-Cre transgenic mouse is commonly to direct genetic recombination in forebrain excitatory neurons [35], we tested whether endogenous CC2D1A signaling functions in a cell autonomous or non-cell autonomous manner to regulate OXT expression in the PVN. The number of CC2D1A+ cells in hippocampal CA1 region was significantly reduced in Cc2d1a cKO mice compared to WT mice (Fig. S3a–c). We observed no significant differences between male Cc2d1a cKO and WT mice in the percentages of CC2D1A+ cells (Fig. S3d, e) and OXT+CC2D1A+ neurons to total OXT+ neurons in the PVN.

Fig. 2: Male Cc2d1a cKO mice show reduced number of OXT-expressing neurons in the PVN.
figure 2

Representative images of OXT immunoreactivity in different anteroposterior levels of the PVN (a) and SON (b) in male WT and Cc2d1a cKO mice. Scale bar, 200 μm. Quantification of percentage of OXT+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 3.35, p = 0.0052; c) and total number of DAPI+ cells (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 0.52, p = 0.61; d) in the PVN of male WT and Cc2d1a cKO mice. Quantification of the percentage of OXT+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 0.93, p = 0.48; e) and total number of DAPI+ cells (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 0.49, p = 0.63; f) in the SON of male WT and Cc2d1a cKO mice. g Representative images of colocalization of OXT immunoreactivity and P2X4R immunoreactivity in the PVN of male WT and Cc2d1a cKO mice. Scale bar, 200 μm. h Representative images of colocalization of AVP immunoreactivity and P2X4R immunoreactivity in the PVN of male WT and Cc2d1a cKO mice. Scale bar, 200 μm. Quantification of percentage of OXT+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 2.74, p = 0.017; i), OXT+P2X4R+/P2X4R+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 1.15, p = 0.27; j), P2X4R+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(11) = 2.65, p = 0.0224; m) and total number of DAPI+ cells (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(11) = 1.15, p = 0.2736; n) in the PVN of male WT and Cc2d1a cKO mice. Quantification of the percentage of AVP+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 0.085, p = 0.93; k), AVP+P2X4R+/P2X4R+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 2.73, p = 0.02; l), P2X4R+/DAPI+ neurons (mouse number: Nwt = 8, NcKO = 7; two-tailed paired Student’s t-test, t(13) = 2.29, p = 0.0396; o) and total number of DAPI+ cells (mouse number: Nwt = 8, NcKO = 7; two-tailed unpaired Student’s t-test, t(13) = 0.27, p = 0.7907; p) in the PVN of male WT and Cc2d1a cKO mice. Data are presented as mean ± SEM. *p < 0.05 and **p < 0.01 as compared with WT group.

Full size image

OXT treatment rescues irritability-like behavior in Cc2d1a cKO mice

We performed additional experiments to explore a causal link between reduced OXT expression in the PVN and higher irritability-like behavior in male WT mice. Using an adeno-associated viral vector expressing a short hairpin RNA targeting mouse Oxt mRNA (shmOxt) (Fig. 3a), we reduced the number of OXT-expressing neurons in the PVN, whereas the total cell numbers in the PVN did not differ between shmOxt- and scramble-treated groups (Fig. 3b–d). As expected, shmOxt-treated male mice showed increased aggressive responses (Fig. 3e) and reduced defensive responses (Fig. 3f) to the BBT compared to scramble-treated male mice.

Fig. 3: Intranasal administration of OXT improves irritability-like behavior in male Cc2d1a cKO mice.
figure 3

a Schematic representation of the experimental design. Two weeks after stereotaxic injection of AAV5-pU6-scramble-EGFP or AAV5-pU6-shmOxt-EGFP into the PVN, mice were single housed for 3 days before the BBT. b Representative image showing the expression of oxytocin and GFP in PVN of shmOxt- and scramble-treated male WT mice. Scale bar: 100 μm. Magnified images of rectangle indicate OXT+GFP+ cells. Scale bar, 20 µm. c Quantification of the percentage of GFP+/DAPI+ cells (mouse number: Nscramble = 10, NshmOxt = 10; two-tailed unpaired Student’s t-test, t(18) = 0.197, p = 0.8454). d Quantification of the percentage of OXT+GFP+/GFP+ cells (mouse number: Nscramble = 10, NshmOxt = 10; two-tailed unpaired Student’s t-test, t(18) = 20.66, p < 0.0001). e Summary bar graphs depicting the effects of shmOxt and scramble treatment on time spent by male WT in aggressive responses to the BBT (mouse number: Nscramble = 10, NshmOxt = 10; two-tailed unpaired Student’s t-test, t(18) = 3.572, p = 0.0022). f Summary bar graphs depicting the effects of shmOXT and scramble treatment on time spent by male WT in defensive responses to the BBT (mouse number: Nscramble = 10, NshmOxt = 10; two-tailed unpaired Student’s t-test, t(18) = 3.114, p = 0.006). g Schematic representation of the experimental design. Male WT or Cc2d1a cKO mice were single housed for 3 days before the BBT. Mice were intranasal administration of OXT (0.2 mg/kg in a volume of 5 μl) or an equivalent volume of vehicle (saline) 30 min before the BBT, respectively. h Summary bar graphs depicting the effect of vehicle and OXT treatment on time spent by male WT or Cc2d1a cKO mice in aggressive responses to the BBT (mouse number: Nwt/Vehicle = 15, NcKO/Vehicle = 9, Nwt/OXT = 15, NcKO/OXT = 9; two-way ANOVA, genotype: F(1,44) = 18.47, p < 0.0001; treatment: F(1,44) = 23.69, p < 0.0001; interaction: F(1,44) = 15.09, p = 0.0003). i Summary bar graphs depicting the effect of vehicle and OXT treatment on time spent by male WT or Cc2d1a cKO mice in defensive responses to the BBT (mouse number: Nwt/Vehicle = 15, NcKO/Vehicle = 9, Nwt/OXT = 15, NcKO/OXT = 9; two-way ANOVA, genotype: F(1,44) = 11.30, p = 0.0016; treatment: F(1,44) = 6.11, p = 0.0174; interaction: F(1,44) = 1.88, p = 0.1778). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 as compared with each other group.

Full size image

We next used three complementary approaches to more thoroughly investigate whether OXT treatment could rescue increased irritability-like behavior in male Cc2d1a cKO mice. First, we treated the mice with exogenous OXT. WT or Cc2d1a cKO mice were individually housed for 3 days and examined their aggressive and defensive responses to the BBT 30 min after intranasal administration of vehicle or OXT (Fig. 3e). The OXT dose was chosen based on a previous study showing that it effectively restores social behavior deficit in Cntnap2 KO mice [28]. A two-way ANOVA on aggressive responses revealed a significant effect of genotype, treatment and genotype × treatment interaction (Fig. 3h). Post hoc analysis showed that OXT treatment alleviated aggressive responses in Cc2d1a cKO mice. OXT treatment did not significantly alter aggressive responses in WT mice. A two-way ANOVA on defensive responses revealed a significant effect of genotype and treatment but no effect of genotype × treatment interaction (Fig. 3i).

Considering that the melanocortin 4 receptor (MCR4) has been shown to be expressed in PVN OXT neurons [36] and activation of MCR4 can increase central OXT release [28, 37], we investigated whether the selective MC4R agonist, RO27-3225, could ameliorate irritability-like behavior in Cc2d1a cKO mice. We administered vehicle or RO27-3225 intraperitoneally and aggressive and defensive responses to the BBT were examined 30 min after treatment (Fig. S4a). A two-way ANOVA analysis on aggressive responses revealed a significant effect of genotype, treatment and genotype × treatment interaction (Fig. S4b). Post hoc analysis showed that RO27-3225 treatment ameliorated aggressive responses in Cc2d1a cKO mice. RO27-3225 treatment did not alter aggressive responses in WT mice. A two-way ANOVA analysis on defensive responses revealed a significant effect of treatment and genotype × treatment interaction but no effects of genotype (Fig. S4c). Post hoc analysis showed that RO27-3225 treatment promoted defensive responses in Cc2d1a cKO mice. To investigate the potential role of OXT signaling in mediating the effect of RO27-3225, we pretreated WT and Cc2d1a cKO mice with OXT receptor antagonist, L-368,899, 15 min before RO27-3225 injection (Fig. S4d). The rescue effects of RO27-3225 on aggressive and defensive responses were eliminated by L-368,899 pretreatment (Fig. S4e, f).

If increased irritability-like behavior in Cc2d1a cKO mice results from reduced number of PVN OXT-expressing neurons, it would be possible to rescue irritability-like behavior by driving OXT neurons. We applied a chemogenetic approach to specifically activate PVN OXT neurons during the BBT. We bilaterally injected the PVN with an AAVDJ vector expressing hM3D(Gq)-mCherry under the control of the endogenous mouse OXT (mOXT) promoter [AAVDJ-mOXT-hM3D(Gq)-mCherry] (Fig. 4a). We verified the neuronal specificity of viral expression by imaging mCherry, whose expression was confined to OXT-immunoreactive neurons in the PVN (Fig. 4b, c). A two-way ANOVA analysis on aggressive responses revealed a significant effect of genotype, treatment and genotype × treatment interaction (Fig. 4d). Post hoc analysis showed that CNO treatment reduced aggressive responses in hM3D(Gq)-mCherry-expressed Cc2d1a cKO mice. CNO treatment did not alter aggressive responses in hM3D(Gq)-mCherry-expressed WT mice. A two-way ANOVA analysis on defensive responses revealed significant effects of genotype, treatment and genotype × treatment interaction (Fig. 4e). Post hoc analysis showed that CNO treatment reduced defensive response in hM3D(Gq)-mCherry-expressed Cc2d1a cKO mice. To further confirm the role of OXT signaling in mediating CNO’s effect, we pretreated WT and Cc2d1a cKO mice with L-368,899 15 min before CNO application (Fig. 4f). L-368,899 treatment blocked the effects of CNO on aggressive and defensive responses in hM3D(Gq)-mCherry-expressed Cc2d1a cKO mice (Fig. 4g–J).

Fig. 4: Chemogenetic activation of OXT-expressing neurons in the PVN rescues irritability-like behavior in male Cc2d1a cKO mice.
figure 4

a Schematic representation of the experimental design. Three weeks after stereotaxic injection of AAVDJ-mOXT-hM3D(Gq)-mCherry into the PVN, mice were single housed for 3 days before the BBT. Mice were injected intraperitoneally with vehicle (0.5% DMSO in saline) or CNO (3 mg/kg) 30 min before the BBT, respectively. Representative images and bar graphs showing percentage of colocalization of OXT+mCherry+/OXT+ and OXT+mCherry+/mCherry+ neurons in the PVN of male WT (b) and Cc2d1a cKO mice (c). Scale bar, 100 μm. d Summary bar graphs depicting the effects of systemic vehicle and CNO injections on time spent by male WT or Cc2d1a cKO mice that received bilateral intra-PVN AAVDJ-mOXT-hM3D(Gq)-mCherry injections in aggressive responses to the BBT (mouse number: Nwt/Vehicle = 18, NcKO/Vehicle = 13, Nwt/CNO = 18, NcKO/CNO = 13; two-way ANOVA, genotype: F(1,58) = 30.78, p < 0.0001; treatment: F(1,58) = 40.38, p < 0.0001; interaction: F(1,58) = 35.37, p < 0.0001). e Summary bar graphs depicting the effects of systemic vehicle and CNO injections on time spent by male WT or Cc2d1a cKO mice that received bilateral intra-PVN AAVDJ-mOXT-hM3D(Gq)-mCherry injections in defensive responses to the BBT (mouse number: Nwt/Vehicle = 18, NcKO/Vehicle = 13, Nwt/CNO = 18, NcKO/CNO = 13; two-way ANOVA, genotype: F(1,58) = 56.16, p < 0.0001; treatment: F(1,58) = 4.30, p = 0.0426; interaction: F(1,58) = 4.05, p = 0.0488). f Schematic representation of the experimental design. Three weeks after stereotaxic injection of AAVDJ-mOXT-hM3D(Gq)-mCherry into the PVN, mice were single housed for 3 days before the BBT. L-368,899 (10 mg/kg) was injected intraperitoneally 10 min before CNO (3 mg/kg) application and CNO was injected intraperitoneally 30 min before the BBT. Representative images and bar graphs showing percentage of colocalization of OXT+mCherry+/OXT+ and OXT+mCherry+/mCherry+ neurons in the PVN of male WT (g) and Cc2d1a cKO mice (h). Scale bar, 100 μm. i Summary bar graphs depicting the effects of systemic L-368,899 and L-368,899 + CNO injections on time spent by male WT or Cc2d1a cKO mice that received bilateral intra-PVN AAVDJ-mOXT-hM3D(Gq)-mCherry injections in aggressive responses to the BBT (mouse number: Nwt/L-368,899 = 4, NcKO/L-368,899 = 7, Nwt/L-368,899+CNO = 4, NcKO/L-368,899+CNO = 7; two-way ANOVA, genotype: F(1,18) = 7.17, p = 0.0154; treatment: F(1,18) = 10.66, p = 0.0043; interaction: F(1,18) = 0.06, p = 0.81). j Summary bar graphs depicting the effects of systemic L-368,899 and L-368,899 + CNO injections on time spent by male WT or Cc2d1a cKO mice that received bilateral intra-PVN AAVDJ-mOXT-hM3D(Gq)-mCherry injections in defensive responses to the BBT (mouse number: Nwt/L-368,899 = 4, NcKO/L-368,899 = 7, Nwt/L-368,899+CNO = 4, NcKO/L-368,899+CNO = 7; two-way ANOVA, genotype: F(1,18) = 0.27, p = 0.61; treatment: F(1,18) = 0.49, p = 0.4915; interaction: F(1,18) = 0.56, p = 0.4639). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 as compared with each other group.

Full size image

Inhibition of MeApv–VmHvl projection alleviates irritability-like behavior in Cc2d1a cKO mice

To decipher brain regions and interconnected circuits underlying increased irritability-like behavior in Cc2d1a cKO mice, we used c-Fos immunohistochemistry to assess the neuronal activity in several brain regions 90 min after the BBT. Among these brain areas, more c-Fos+ cells in Cc2d1a cKO than WT mice were found in the MeApv, VmHvl and dorsolateral periaqueductal gray (DLPAG) (Fig. S5a–d). No significant differences between groups were found in the MeA posterodorsal subnucleus (MeApd), bed nucleus of the stria terminalis (BNST) and basolateral amygdala (BLA) (Fig. S5d). To further characterize the neuronal populations of the MeApv activated during the BBT, we conducted co-immunolabelling with antibodies against CaMKIIα (excitatory neuronal marker) or GAD67 (GABAergic neuronal marker). Significantly higher c-Fos+CaMKIIα+ cells were observed in the MeApv, VmHvl and DLPAG (Fig. S5a–c, e) of Cc2d1a cKO than WT mice. No significant differences in the numbers of c-Fos+CaMKIIα+ cells between groups were observed in the MeApd, BNST and BLA (Fig. S5e). We found no significant differences in the numbers of c-Fos+GAD67+ cells between groups in the MeApv, VmHvl, DLPAG, MeApd, BNST and BLA (Fig. S5f).

The MeA and related circuits have been implicated in the escalation of aggressive behaviors induced by traumatic stress [33]. Based on our c-Fos expression results, we hypothesized that increased activation of MeApv excitatory neurons is responsible for the expression of heightened irritability-like behavior in Cc2d1a cKO mice. To test this, we bilaterally injected AAVDJ-hSyn-DIO-mCherry or AAVDJ-hSyn-DIO-hM4D(Gi)-mCherry into the MeApv of Cc2d1a cKO mice for chemogenetic inhibition during the BBT (Fig. 5a). We found that CNO treatment reduced aggressive responses (Fig. 5b) and restored defensive responses (Fig. 5c) in hM4D(Gi)-mCherry-expressed Cc2d1a cKO mice. Functional validation of AAV-infected neurons using ex vivo whole-cell patch clamp recordings confirmed that cell treated with 50 μM CNO reliably diminished spiking induced by current injection into the soma of hM4D(Gi)+ neurons (Fig. 5d, e). As a complementary approach, we tested whether activation of the MeApv excitatory neurons is sufficient to enhance irritability-like behavior to the BBT. We chemogenetic activation of MeApv excitatory neurons by bilaterally delivering AAVDJ-hSyn-DIO-mCherry or AAVDJ-hSyn-DIO-hM3D(Gq)-mCherry into the MeApv of Emx1-Cre mice (Fig. 5f). We found that CNO treatment increased aggressive responses (Fig. 5g) and reduced defensive responses (Fig. 5h) in hM3D(Gq)-mCherry-expressed Emx1-Cre mice. Function of the hM3D(Gq)+ infected neurons were validated by the ex vivo whole-cell patch recording and confirmed that cell treated with 50 μM CNO reliably enhanced spiking induced by current injection into the soma of hM3D(Gq)+ neurons (Fig. 5i, j). To explore MeApv outputs necessary for governing irritability-like behavior, we focused on the VmHvl that has been identified as a downstream target of the MeApv and potential neuroanatomical substrate implicated in aggressive behavior [38, 39]. To manipulate the MeApv–VmHvl pathway selectively, we bilaterally injected AAVrg-hSyn-DIO-mCherry or AAVrg-hSyn-DIO-hM4D(Gi)-mCherry into the VmHvl of Cc2d1a cKO mice (Fig. 5k). We found that intra-MeApv infusions of CNO reduced aggressive responses (Fig. 5l) and restored defensive responses (Fig. 5m) in hM4D(Gi)-mCherry-expressed Cc2d1a cKO mice.

Fig. 5: The MeApv–VmHvl pathway mediates irritability-like behavior.
figure 5

a Schematic representation of the experimental design. AAV5-hSyn-DIO-hM4D(Gi)-mCherry or AAV5-hSyn-DIO-mCherry was bilaterally injected into the MeApv of male Cc2d1a cKO mice. Three weeks after viral infection, mice were individually housed for 3 days and then subjected to the BBT 30 min after vehicle (0.5% DMSO in saline) or CNO injection (3 mg/kg). b Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Cc2d1a cKO mice that received bilateral intra-MeApv injections of either AAV5-hSyn-DIO-hM4D(Gi)-mCherry or AAV5-hSyn-DIO-mCherry in aggressive responses to the BBT (mouse number: NmCherry/Vehicle = 4, NhM4D(Gi)/Vehicle = 10, NmCherry/CNO = 4, NhM4D(Gi)/CNO = 10; two-way ANOVA, vector: F(1,24) = 3.44, p = 0.0762; treatment: F(1,24) = 20.05, p = 0.0002; interaction: F(1,24) = 11.78, p = 0.0022). c Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Cc2d1a cKO mice that received bilateral intra-MeApv injections of either AAV5-hSyn-DIO-hM4D(Gi)-mCherry or AAV5-hSyn-DIO-mCherry in defensive responses to the BBT (mouse number: NmCherry/Vehicle = 4, NhM4D(Gi)/Vehicle = 10, NmCherry/CNO = 4, NhM4D(Gi)/CNO = 10; two-way ANOVA, vector: F(1,24) = 2.67, p = 0.1142; treatment: F(1,24) = 1.48, p = 0.2352; interaction: F(1,24) = 3.76, p = 0.0642). d Representative traces showing responses of infected (hM4D(Gi)+) neurons to 500 ms depolarizing current pulse under whole-cell current clamp before and after bath application of CNO (50 μm) in the ex vivo MeApv slices. e Representative image showing the co-expression of hM4D(Gi)-mCherry and biocytin in MeApv of cKO mice. Scale bar: 25 μm. f Schematic representation of the experimental design. AAV5-hSyn-DIO-hM3D(Gq)-mCherry or AAV5-hSyn-DIO-mCherry was bilaterally injected into the MeApv of male Emx1-Cre mice. Three weeks after viral infection, mice were individually housed for 3 days and then subjected to the BBT 30 min after vehicle or CNO injection. g Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Emx1-Cre that received bilateral intra-MeApv injections of either AAV5-hSyn-DIO-hM3D(Gq)-mCherry or AAV5-hSyn-DIO-mCherry in aggressive responses to the BBT (mouse number: NmCherry/Vehicle = 6, NhM3D(Gq)/Vehicle = 11, NmCherry/CNO = 6, NhM3D(Gq)/CNO = 11; two-way ANOVA, vector: F(1,30) = 9.41, p = 0.0045; treatment: F(1,30) = 7.05, p = 0.0126; interaction: F(1,30) = 11.93, p = 0.0017). h Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Emx1-Cre that received bilateral intra-MeApv injections of either AAV5-hSyn-DIO-hM3D(Gq)-mCherry or AAV5-hSyn-DIO-mCherry in defensive responses to the BBT(mouse number: NmCherry/Vehicle = 6, NhM3D(Gq)/Vehicle = 11, NmCherry/CNO = 6, NhM3D(Gq)/CNO = 11; two-way ANOVA, vector: F(1,30) = 4.30, p = 0.0046; treatment: F(1,30) = 6.85, p = 0.0138; interaction: F(1,30) = 2.27, p = 0.1424). i Representative traces showing responses of infected (hM3D(Gq)+) neurons to 500 ms depolarizing current pulse under whole-cell current clamp before and after bath application of CNO (50 μm) in the ex vivo MeApv slices. j Representative image showing the co-expression of hM3D(Gq)-mCherry and biocytin in MeApv of Emx1-Cre mice. Scale bar: 25 μm. k Schematic representation of the experimental design. AAVrg-hSyn-DIO-hM4D(Gi)-mCherry or AAVrg-hSyn-DIO-mCherry was bilaterally injected into the VmHvl of male Cc2d1a cKO mice and CNO (4 mM, 0.5 μl) was micro-infused into the MeApv. l Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Cc2d1a cKO mice that received bilateral intra-VmHvl injections of either AAVrg-hSyn-DIO-hM4D(Gi)-mCherry or AAVrg-hSyn-DIO-mCherry in aggressive responses to the BBT (mouse number: NmCherry/Vehicle = 4, NhM3D(Gq)/Vehicle = 7, NmCherry/CNO = 4, NhM3D(Gq)/CNO = 7; two-way ANOVA, Vector: F(1,18) = 7.27, p = 0.0147; Treatment: F(1,18) = 9.15, p = 0.0073; Interaction: F(1,18) = 4.10, p = 0.058). m Summary bar graphs depicting the effects of vehicle and CNO treatment on time spent by male Cc2d1a cKO mice that received bilateral intra-VmHvl injections of either AAVrg-hSyn-DIO-hM4D(Gi)-mCherry or AAVrg-hSyn-DIO-mCherry in defensive responses to the BBT (mouse number: NmCherry/Vehicle = 4, NhM3D(Gq)/Vehicle = 7, NmCherry/CNO = 4, NhM3D(Gq)/CNO = 7; two-way ANOVA, vector: F(1,18) = 1.88, p = 0.1872; treatment: F(1,18) = 4.70, p = 0.0438; interaction: F(1,18) = 11.16, p = 0.0036). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 as compared with each other group.

Full size image

Discussion

High levels of irritability are common in individuals with ASD [13, 14], but the neural mechanisms underlying irritability in ASD remain elusive. Here, we have coupled chemogenetic manipulations, pharmacological treatments and behavioral assays in a Cc2d1a cKO mouse model of ASD to illustrate that a dysfunctional OXT system with less OXT-expressing neurons in the PVN is associated with increased irritability and uncover a pivotal role of the MeApv–VmHvl pathway for governing irritability-like behavior. More importantly, our data suggest that exogenous and evoked OXT can effectively rescue irritability-like behavior in male Cc2d1a cKO mice.

Cc2d1a cKO mice exhibit several key autistic-like phenotypes, including deficits in social communication, restricted and repetitive behaviors [21,22,23,24,25]. Here, we extend these findings and demonstrate that Cc2d1a deletion leads to increased irritability-like behavior in male but not female mice. The absence of increased irritability-like behavior in female mice is likely due to normal OXT expression in the PVN of female Cc2d1a cKO mice. Similar sex-specific behavioral deficits in Cc2d1a cKO mice were also observed in the Morris water maze test to assess spatial learning and reference memory [40]. The observed sex-specific differences in susceptibility to Cc2d1a deletion might be due to compensatory mechanisms or differential allocation of molecular and cellular functions between males and females [41]. Furthermore, our results clearly demonstrate an inverse correlation between a decrease in the number of OXT-expressing neurons and an increase in irritability-like behavior in male Cc2d1a cKO mice. Consistently, a loss of OXT neurons has also been observed in other knockout mice lacking the ASD-risk genes, Cntnap2 and Shank3b [28, 29]. These findings highlight a prominent role for OXT deficiency in core autistic features and raise the possibility that OXT-based treatments may be helpful in treating these behavioral abnormalities [42,43,44]. In line with this notion, our results show that intranasal administration of OXT or evoked endogenous OXT release consistently rescues irritability-like behavior in Cc2d1a cKO mice. We further show that indirect activation of PVN OXT neurons by application of MC4R agonist effectively alleviates irritability-like behavior in Cc2d1a cKO mice. Thus, our results strongly support the idea that the central OXT system is crucial in regulating irritability-like behavior. However, it remains unclear how Cc2d1a deletion causes reduced OXT expression in the PVN. Since our Emx1-Cre-mediated recombination is restricted to forebrain excitatory neurons and does not directly affect Cc2d1a expression within PVN OXT neurons (Fig. S3), it is possible that Cc2d1a deletion in excitatory neurons may have a broad impact on PVN OXT neuron development, resulting in reduced OXT expression in the PVN. Further studies are needed to characterize this possibility.

What are the neural substrates underlying irritability? Previous functional neuroimaging studies largely focused on brain regions in the classical treat and reward systems, such as the amygdala, prefrontal cortex and striatum [1, 6]. The MeA has been proposed to play a central role in aggression [33, 45], thus promoting us to consider it as a relay center, which integrates incoming information about external emotional stimuli with the internal state and then transmits this information to downstream executive regions to mediate irritability. Concordantly, activity mapping using c-Fos expression strongly implicated the MeApv and MeApd in behavioral responses to the BBT. The MeApv and MeApd are thought to be differentially involved in defensive behavior [46,47,48]. Given we selectively ablated Cc2d1a expression from excitatory neurons [24], along with the observation that excitatory neurons are more abundant in the MeApv than MeApd [49], we chose to focus on the MeApv. Our findings confirm the importance of MeApv excitatory projection neurons in mediating irritability-like behavior. Although the MeApv is known to send densely glutamatergic projections to multiple brain regions [50], our circuit-level analysis identified the MeApv-VmH pathway as the main contributor in mediating irritability-like behavior. We found that chemogenetic inhibition of the MeApv–VmHvl pathway alleviates irritability-like behavior in Cc2d1a cKO mice and, conversely, chemogenetic activation of the MeApv–VmHvl pathway enhances irritability-like behavior in Emx1-Cre mice. Our findings are consistent with earlier literature showing that projection-specific pathways from the MeApv differentially regulated aggressive-like behavior [33, 51], and further expand on these by implicating a pivotal role for the MeApv–VmHvl pathway in mediating irritability. Despite the evidence supporting the involvement of the VmHvl in irritability-like behavior, it remains unclear the downstream regions of the VmHvl that contribute to execute irritability. One potential candidate is the DLPAG, which receives strong inputs from the VmHvl and represents an important premotor region involved in executing attack [38, 39, 48, 52]. Correspondingly, our results revealed a consistent increase in c-Fos induction in the DLPAG in response to the BBT in Cc2d1a cKO mice.

Irritability phenotype is assessed using measures based on rodent’s aggressive and defensive responses. The BBT is commonly employed to trigger and evaluate irritability-like behavior in rodents, especially those induced by alcohol, cannabis and nicotine withdrawal [32, 53, 54]. Relative to other behavioral paradigms that trigger aggressive and defensive behaviors, such as the social dominance-subordination paradigm and RIT, the BBT was found to be more controllable to provoking these behaviors [54, 55]. While previous studies commonly employed the sum of aggressive and defensive responses to reflect the severity of irritability [32, 53, 54], our study indicates that the effects of Cc2d1a deletion on aggressive and defensive responses are not in the same direction. Cc2d1a cKO mice showed increased aggressive responses, especially the exploring behavior, and decreased defensive responses, especially the digging behavior. The decreased defensive response to the BBT might be explained by the increased aggressive response. Although our data do not reveal that a relationship exists directly between aggressive and defensive responses, this result is consistent with a previous study showing that increased aggression can be concomitant with decreased defensive responses [52]. We were surprised to find that Cc2d1a cKO mice show only mild signs of aggression in the RIT, although not statistically significant. One potential explanation for this observation is that Cc2d1a cKO mice exhibit significantly decreased social approach [24], which may result in reduced time spent in interaction with the novel intruder conspecific. While the BBT and RIT are similar in that they are paradigms driven by the sensory cues, it is important to make the distinction between the two as the RIT is strongly influenced by social factors in eliciting aggressive behavior. Accordingly, abnormal social approach may affect aggressive behavior toward intruder conspecific in the RIT.

There are two weaknesses in this study. First, although the BBT is a standard behavioral test to assess irritability-like behavior [54, 55], which may not fully encapsulate the complexity of irritability in ASD, we need to use other behavioral paradigms that elicit frustration to measure irritability. Second, we do not know whether chronic early postnatal treatment with OXT may restore OXT deficiency, shape the MeApv-VmH pathway, and exert more lasting beneficial effects on irritability-like behavior in Cc2d1a cKO mice. Further studies are required for improving on these weaknesses.

In conclusion, this study uncovers a causal link between OXT deficiency and irritability-like behavior in Cc2d1a cKO mice. We further identify a pivotal role of the MeApv–VmHvl pathway for governing irritability-like behavior. Although it remains to be seen whether these findings have translational validity in humans, our current studies provide empirical evidence supporting the therapeutic value of OXT to treat comorbid irritability in individuals with ASD.