SOCIABILITY DEVELOPMENT IN MICE WITH CELL-SPECIFIC DELETION OF THE NMDA RECEPTOR NR1 SUBUNIT GENE

Abstract

Social affiliative behavior is an important component of everyday life in many species and is likely to be disrupted in disabling ways in various neurodevelopmental and neuropsychiatric disorders. Therefore, determining the mechanisms involved in these processes is crucial. A link between N-methyl-D-aspartate (NMDA) receptor function and social behaviors has been clearly established. The cell types in which NMDA receptors are critical for social affiliative behavior, however, remain unclear. Here, we use mice carrying a conditional allele of the NMDA R1 subunit to address this question. Mice bearing a floxed NMDAR1 (NR1) allele were crossed with transgenic calcium/calmodulin-dependent kinase IIα (CaMKIIα)-Cre mice or parvalbumin (PV)-Cre mice targeting postnatal excitatory forebrain or PV-expressing interneurons, respectively, and assessed using the three-chambered Social Approach Test. We found that deletion of NR1 in PV-positive interneurons had no effect on social sniffing, but deletion of NR1 in glutamatergic pyramidal cells resulted in a significant increase in social approach behavior, regardless of age or sex. Therefore, forebrain excitatory neurons expressing NR1 play an important role in regulating social affiliative behavior.

Introduction

Engaging in and interpreting social interactions are both beneficial in everyday life and crucial for survival in most species. Reproduction, avoidance of aggression, and reciprocal collaborations are all dependent upon social ability. Deficits in social affiliative behaviors are disabling core symptoms of neurodevelopmental and neuropsychiatric disorders, including autism spectrum disorder, anxiety disorder, and schizophrenia ^1–6. Therefore, understanding the circuits and cell populations involved in social affiliation is important.

Many brain areas have been implicated in social affiliative behavior, including the amygdala, hippocampus, prefrontal cortex, and anterior cingulate cortex^7–18. Likewise, a number of neurotransmitters, neuropeptides, and receptor types have been shown to be important in the regulation of social behavior, including dopamine, serotonin, oxytocin, and glutamate receptors^19–28. Specifically, N-methyl-D-aspartate receptors (NMDARs), ionotropic glutamate receptors important in learning and memory and synaptic plasticity, have been associated with social affiliative behavior, and may play an important role in disorders such as autism spectrum disorder and schizophrenia^29–44.

Both knockdown and antagonism of NMDAR result in disruption of social behaviors. Patients with anti-NMDAR encephalitis exhibit deficits in social behavior, including social cognitive impairments and aggressive or asocial behavior^45,46. Treatments with NMDAR antagonists such as MK-801 and ketamine have been shown to cause decreased social behaviors in mice, rats, and zebrafish^23,47–54. In particular, the obligatory NR1 subunit of the heterotetrameric NMDAR complex is implicated in social functioning and neuropsychiatric conditions. The NMDAR subunit NR1 is encoded by a single gene, GRIN1. Downregulation or deletion of NR1 in various cell types and brain areas of rodents results in social deficits^36,55–66. Specifically, deletion of NR1 in the mPFC increased social novelty preference, whereas NR1 deletion in the CA3 of the hippocampus, in the thalamus, or in the nucleus accumbens decreased social approach behavior^60,61,64. Depletion of NR1 in interneurons of the hippocampus and cortex result in decreased social behavior, and some studies have reported decreases in social behavior following ablation of NR1 in excitatory forebrain neurons in older mice or parvalbumin-positive interneurons^57,65,67. Therefore, the NR1 NMDAR subunit is an important component in the study of social behaviors. However, what is less clear are the specific mechanisms through which NR1 signaling affects social affiliative behaviors across juvenile and young adult development, periods of particular relevance for neurodevelopmental disorders like autism and schizophrenia that exhibit social deficits.

In order to investigate the cell-specific effects of NR1 on social approach behaviors, we used the Cre/Lox system to delete GRIN1 from subsets of neurons and analyzed social approach behavior in male and female, juvenile and young adult mice. Specifically, mice with a floxed NR1 allele were crossed with mice with Cre recombinase expression in postnatal pyramidal cells of the forebrain, driven by the calcium/calmodulin-dependent kinase IIα (CaMKIIα) promoter, or with mice with Cre recombinase expression driven by the parvalbumin (PV) promoter, which constitute 40-50% of GABAergic interneurons throughout the brain^68–73. We observed the effects of ablation of NR1 in either excitatory forebrain cells or in a subset of inhibitory cells on behavior in the three-chambered Social Approach Test (SAT), a well-established assay with good reliability and validity in mice, which are highly social animals^74–80.

Materials and Methods

Animal Housing:

All mice were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, and all animal procedures were approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC). Breeding cages contained two dams and a sire. Same-sex littermates were weaned, ear tagged, and tail samples were collected for genotyping at ~21 days of age. Weaned littermates were then group-housed 2 - 5 per cage in a temperature- and humidity-controlled environment with a 12-hour light-dark cycle (lights on at 7 am). All mice had access to food (LabDiet 5010, Purina Mills) and water ad libitum.

Mouse lines:

The mice in this study were bred from the following three progenitor mouse lines that had been purchased from The Jackson Laboratory (Bar Harbor, ME): 1) mice with the NMDA receptor NR1 subunit gene (NR1 or Grin1; ~12 kbp including the transmembrane domain and C-terminal region)) flanked by loxP sites on a C57BL/6J (B6) genetic background (B6.129S4-Grin1^tm2Stl/J mice (JAX stock #005246)), referred to as NR1^flox/flox mice; 2) mice with a Cre transgene driven by a parvalbumin (PV) promoter on a B6-129P2 genetic background (B6;129P2-Pvalb^tm1(cre)Arbr/J mice (JAX stock # 017320)), referred to as PV-Cre⁺ mice, which express Cre, starting at E14, in over 90% of PV+ cells and were shown electrophysiologically to be specific to inhibitory neurons^81,82; and 3) mice with a Cre transgene driven by a CaMKIIα promoter on a B6 genetic background (B6.Cg-Tg(Camk2a-cre)T29-1Stl/J mice (JAX stock #005359)), referred to as CaMKII-Cre⁺ mice, in which Cre is driven by a promoter for CaMKII, which begins to weakly express shortly after birth and increases around postnatal day 5. These mice have been shown to lack Cre in GABAergic interneurons^65,73,83. These 3 lines were bred to produce the following two lines:

PV-Cre⁺; NR1^flox/flox mouse line:

PV-Cre⁺ mice were crossed to NR1^flox/flox mice to produce PV-Cre^+/-; NR1^flox/wt offspring, which were backcrossed with PV-Cre⁺ mice to produce PV-Cre⁺; NR1^flox/wt mice. For this study, PV-Cre⁺; NR1^flox/wt mice were intercrossed to obtain mice of the following genotypes: PV-Cre⁺; NR1^flox/flox, PV-Cre⁺; NR1^flox/wt, and PV-Cre⁺; NR1^wt/wt. The behavioral experiments compared PV-Cre⁺; NR1^flox/flox (n=14 males, tested at 30d and 70d; n=16 females, tested at 30d and 70d) to PV-Cre⁺; NR1^wt/wt littermates (n=13 males; n=16 females, tested at 30d and 70d). In vitro whole-cell hippocampal recordings from PV-Cre⁺; NR1^flox/flox mice confirm loss of NMDAR currents in PV+ cells⁸⁴.

CaMKII-Cre⁺; NR1^flox/flox mouse line:

CaMKIIα-Cre⁺ mice were crossed with NR1^flox/flox mice to produce CaMKIIα-Cre^+/−; NR1^flox/wt offspring, which were backcrossed with CaMKIIα-Cre⁺ mice to produce CaMKIIα-Cre⁺; NR1^flox/wt mice. These CaMKIIα-Cre⁺; NR1^flox/wt mice were intercrossed to obtain mice of the following genotypes: CaMKII-Cre⁺; NR1^flox/flox, CaMKII-Cre⁺; NR1^flox/wt, and CaMKII-Cre⁺; NR1^wt/wt. Behavioral experiments compared CaMKII-Cre⁺; NR1^flox/flox (n=14 males, tested at 30d and 70d; n=15 females, tested at 30d and 70d) to CaMKII-Cre⁺; NR1^wt/wt littermates (n=11 males; n=22 females, tested at 30d and 70d). CaMKII-Cre⁺; NR1^flox/flox mice exhibit significant NR1 mRNA loss in the forebrain as well as NMDAR current in hippocampal pyramidal neurons⁶⁵.

Stimulus Mice for the Social Approach Test (SAT):

A/J mice were gonadectomized by the age of 4 weeks at the Jackson Laboratory, and then were shipped to University of Pennsylvania at 5-6 weeks of age. Gonadectomized A/J mice were used as stimulus mice in the SAT in order to minimize aggressive and sexual behaviors directed toward them by test mice. At least a week after arrival at University of Pennsylvania, A/J mice were habituated to the SAT apparatus by being placed in a cylinder in the apparatus for 5 minutes on 3 consecutive days, prior to being used in the SAT. A/J mice were almost always used as a social stimulus once per day with a minimum of 2 days between subsequent testing. However, on 3 separate occasions (out of 242 tests run), an A/J mouse was used as a social stimulus twice in one day, with 1-2 hours rest between testing.

Behavior:

Cohorts of PV-Cre⁺; NR1^flox/flox and PV-Cre⁺; NR1^wt/wt littermates (1-4 mice per genotype and sex from 11-18 litters), as well as cohorts of CaMKII-Cre⁺; NR1^flox/flox and CaMKII-Cre⁺; NR1^wt/wt littermates (1-4 mice per genotype and sex each from 9-16 litters), underwent Social Approach Testing (see below) during their light phase at ~30 days of age (27-34 days of age; juveniles) and again at ~70 days of age (69-75 days of age; adults). Male and female test mice were assayed on different, alternating days.

The Social Approach Test:

The Social Approach Test (SAT) procedure was conducted as previously described^74,85,86 with a three-chambered top- and bottom-less black Plexiglas arena (10 X 20.5 X 9 in) placed on clean bench paper, replaced for each mouse, which was placed on a clear Plexiglass stand lit from underneath by infrared light panels. Each end chamber contained a transparent Plexiglas cylinder with holes for air circulation and olfactory exploration. As in our previous studies^29,74,80,87, experiments were conducted under dim lighting (<2 lux), in order to encourage exploration and decrease anxiety-related behavior, and were videotaped (Sony-DCR-SR85 Handycam camcorder, Sony Corporation) in infrared-sensing ‘nightshot’ mode. Mice were given ~90 min to acclimate to the testing room before the start of the experiment.

For Phase 1, a test mouse was placed in the apparatus and was allowed to explore for 10 minutes. In Phase 2 a gonadectomized, same-sex A/J mouse (social stimulus) was placed in one cylinder (the social cylinder) and a novel object (purple, plastic half-sphere paper weight, 2.8 X 2.4 X 1.4 in, 2.4 oz) into the other cylinder (the nonsocial cylinder) for 10 minutes, during which the test mouse was allowed to move freely throughout the apparatus and sniff the cylinders. Novel object and novel mouse placement side (left or right chamber) were alternated for each trial.

At the end of each test, the soiled bench paper was discarded, and all equipment (arena, cylinders, object, etc.) was rinsed profusely with water at a sink outside of the test room and then dried with fresh paper towels. Clean bench paper was placed prior to the next test. At the end of all of the tests for the day, 70% ethanol was used to wipe down all surfaces following water rinsing.

All behavioral tracking was performed by TopScan software (Version 2.00, Clever Sys. Inc.). We have previously demonstrated that this software is highly reliable and consistent with scoring from a manual observer⁷⁴. Experimenters were blinded to genotype during testing and analysis. The main dependent variable of interest was time that the test mouse spent sniffing the cylinders in each phase. We have previously shown that cylinder sniffing time, rather than time spent in the chamber, is a more reliable and ecologically valid measure of social approach behavior. Here we use time spent sniffing the social cylinder and time spent sniffing the nonsocial cylinder as the main dependent variables of interest. In the Supplementary Materials, we also present time spent in chambers; time spent sniffing the social cylinder minus the time spent sniffing the nonsocial cylinder (which we will refer to as the “delta sniffing” variable) to control for variability in sniffing times^88–90,74; and distance traveled, a measure of locomotor activity.

Statistical Analysis:

Mixed-effects model linear regression was used for all data. For all analyses, main effects of Genotype, Age, Sex, and Phase were tested, along with all two-way and three-way interactions.

Statistics were performed using R (https://www.R-project.org/)⁹¹ and STATA (StataCorp LLC, College Station, TX). Data are presented as mean ± SEM. The threshold for significance was p < 0.05. All statistical results are presented in tables.

Results

We tested male and female mice as juveniles and again as adults in the Social Approach Test (SAT). We measured the amount of time spent sniffing each of two cylinders in two separate phases. In Phase 1, both cylinders were empty, and in Phase 2, one cylinder contained a novel object (nonsocial cylinder) and the other a novel mouse (social cylinder). To focus on social affiliative behavior and reduce potential sexual and aggressive motivations of the test mouse, we used a same-sex, gonadectomized mouse of a docile strain, A/J, as the social stimulus. We used a mixed effects linear regression model to test main effects of Genotype (Cre⁺; NR1^wt/wt vs Cre⁺; NR1^flox/flox), Age (30 d vs 70 d, repeated measures), Sex, and Phase (1, empty cylinders, or 2, cylinders containing a mouse or object, repeated measures), as well as two- and three-way interactions.

When measuring sniffing time of the nonsocial cylinder in PV-Cre⁺; NR1^flox/flox mice compared to their PV-Cre⁺; NR1^wt/wt littermates, there were no main effects or interactions (Fig. 1 and Table 1). Social cylinder sniffing time analysis in these mice uncovered a main effect of Phase (p<0.001), indicating that mice spent more time sniffing the social cylinder in Phase 2 when it contained a mouse compared to Phase 1 when it was empty, as expected. There was also a significant Age x Phase interaction (p<0.001), demonstrating that 30d animals of both genotypes and sexes spent more time than 70d animals in sniffing the social cylinder in Phase 2 (Fig. 2 and Table 2). There were no additional significant main effects or interactions. We also analyzed delta sniffing time, the amount of time the mouse spent sniffing the social cylinder minus the time spent sniffing the nonsocial cylinder, to indicate sniffing preference. There was no main effect of Genotype, Age, or Sex, but there was a main effect of Phase (p<0.001). The only significant interaction was Age x Phase (p<0.001), indicating that delta sniffing time was higher for juvenile mice than adult mice in Phase 2 of the Social Approach Test (Fig. S1 and Table S1). Therefore, ablation of NR1 in PV+ cells does not affect social affiliative behavior, as reflected by the lack of genotype effects or interactions.

Figure 1.

Nonsocial cylinder sniffing time in mice with reduced NR1 in PV+ interneurons. There were no main effects of Genotype, Age, Sex, or Phase, and no significant interactions; PV-Cre⁺; NR1^flox/flox mice exhibited similar nonsocial cylinder sniffing times as their PV-Cre⁺; NR1^wt/wt littermates.

Table 1 –

Nonsocial cylinder sniffing time in PV-Cre+; NR1 mice (Fig. 1).

PV-Cre+; NR1	Nonsocial Cylinder Sniffing
Effect	p-value
Genotype	0.567
Age	0.395
Sex	0.656
Phase	0.918
Genotype x Age x Sex	0.356
Genotype x Age x Phase	0.077
Genotype x Sex x Phase	0.294
Sex x Age x Phase	0.054
Genotype x Age	0.063
Genotype x Sex	0.744
Genotype x Phase	0.072
Age x Sex	0.991
Age x Phase	0.974
Sex x Phase	0.722

Figure 2.

Social cylinder sniffing time in mice with reduced NR1 in PV+ interneurons. Mixed effects regression model revealed a main effect of Phase (p<0.001) and a significant Age x Phase interaction (p<0.001). *** = p<0.001.

Table 2 –

Social cylinder sniffing time in PV-Cre+; NR1 (Fig. 2).

PV-Cre+; NR1	Social Cylinder Sniffing
Effect	p-value
Genotype	0.711
Age	0.997
Sex	0.501
Phase	<0.001
Genotype x Age x Sex	0.200
Genotype x Age x Phase	0.784
Genotype x Sex x Phase	0.160
Sex x Age x Phase	0.155
Genotype x Age	0.454
Genotype x Sex	0.342
Genotype x Phase	0.270
Age x Sex	0.776
Age x Phase	<0.001
Sex x Phase	0.134

A mixed-effects model revealed a main effect of Phase (p=0.030) in the time PV-Cre⁺; NR1 mice spent in the nonsocial chamber; regardless of genotype, age, or sex, animals spent more time in the nonsocial chamber in Phase 1 than in Phase 2 (Fig. S2 and Table S2). There were no significant main effects or interactions in time spent in the social chamber in these mice (Fig. S3 and Table S3).

For distance traveled, there were main effects of Genotype (p=0.005), Sex (p=0.013), and Phase (p<0.001), indicating that PV-Cre⁺; NR1^flox/flox mice traveled less than their PV-Cre⁺; NR1^wt/wt littermates, females as a whole traveled more than males, and generally, mice traveled more in Phase 1 than Phase 2. There was a significant Genotype x Age x Sex interaction as well (p=0.005; Fig. S4 and Table S4).

For CaMKII-Cre⁺; NR1 mice in nonsocial cylinder sniffing time, there were no significant main effects, but a significant Genotype x Age x Sex interaction (p=0.043) (Fig. 3 and Table 3). Fig. 4 shows time spent sniffing the social cylinder and a main effect of Phase (p<0.001), demonstrating that mice spent significantly more time sniffing the social cylinder in Phase 2 when it contained a novel mouse compared to Phase 1 when it was empty. In addition, there were significant Genotype x Age x Phase (p=0.008) and Genotype x Phase (p<0.001; Table 4) interactions, establishing that CaMKII-Cre⁺; NR1^flox/flox mice sniffed the social cylinder significantly more than their CaMKII-Cre⁺; NR1^wt/wt littermates in Phase 2. For delta sniffing time (Fig. S5), there was a main effect of Phase (p<0.001) and a significant Genotype x Phase interaction (p=0.002; Table S5). In Phase 1, when both cylinders were empty, all mice exhibited delta sniffing times near 0, indicating no preference for one cylinder over another. In Phase 2, all mice showed high levels of delta sniffing, indicating a preference for the social cylinder containing the novel mouse over the nonsocial cylinder containing the novel object. However, CaMKII-Cre⁺; NR1^flox/flox mice, regardless of age or sex, exhibited higher delta sniffing times than their CaMKII-Cre⁺; NR1^wt/wt littermates. Therefore, both male and female juvenile and young adult mice lacking NR1 in excitatory cells showed significantly increased sociability compared to their wild-type littermates.

Figure 3.

Nonsocial cylinder sniffing time in mice with reduced NR1 in glutamatergic pyramidal cells. There were no significant main effects of Genotype, Age, Sex, or Phase, but a significant Genotype x Age x Sex interaction (p=0.043).

Table 3 –

Nonsocial cylinder sniffing time in CaMKII-Cre+; NR1 mice (Fig. 3).

CaMKII-Cre+; NR1	Nonsocial Cylinder Sniffing
Effect	p-value
Genotype	0.054
Age	0.453
Sex	0.831
Phase	0.518
Genotype x Age x Sex	0.043
Genotype x Age x Phase	0.093
Genotype x Sex x Phase	0.826
Sex x Age x Phase	0.568
Genotype x Age	0.885
Genotype x Sex	0.951
Genotype x Phase	0.327
Age x Sex	0.545
Age x Phase	0.389
Sex x Phase	0.539

Figure 4.

Social cylinder sniffing time in mice with reduced NR1 in glutamatergic pyramidal cells. There was a significant main effect of Phase (p<0.001) and significant Genotype X Age X Phase (p=0.008) and Genotype X Phase (p<0.001; Table 4) interactions. CaMKII-Cre⁺; NR1^flox/flox mice sniffed the social cylinder significantly more than their CaMKII-Cre⁺; NR1^wt/wt littermates in Phase 2. *** = p<0.001.

Table 4 –

Social cylinder sniffing time in CaMKII-Cre+; NR1 mice (Fig. 4).

CaMKII-Cre+; NR1	Social Cylinder Sniffing
Effect	p-value
Genotype	0.407
Age	0.629
Sex	0.768
Phase	<0.001
Genotype x Age x Sex	0.153
Genotype x Age x Phase	0.008
Genotype x Sex x Phase	0.354
Sex x Age x Phase	0.835
Genotype x Age	0.781
Genotype x Sex	0.665
Genotype x Phase	<0.001
Age x Sex	0.756
Age x Phase	0.996
Sex x Phase	0.364

A mixed-effects model of time spent in the nonsocial chamber revealed a main effect of Phase (p=0.001; Fig. S6 and Table S6), but no main effects of Genotype, Age, or Sex, and no significant interactions. Mice spent less time in the nonsocial chamber in Phase 2 than in Phase 1. Likewise, a main effect of Phase (p=0.008; Fig. S7 and Table S7) in duration in the social chamber shows that mice spent more time there in Phase 2 than in Phase 1.

Locomotor activity showed main effects of Age (p=0.004) and Phase (p=0.014), indicating that in general, 70d mice traveled more than 30d mice and animals had higher distance traveled in Phase 1 than Phase 2, likely because they were sniffing more than walking in Phase 2. There were also significant Genotype x Age x Sex (p=0.014) and Age x Phase (p=0.009) interactions which show that in Phase 2, 70d animals traveled more than 30d animals (Fig. S8 and Table S8).

Discussion

Elucidating the cell types involved in social affiliative behavior, here measured by the three-chamber Social Approach Test, is critical to understand a behavior that is extremely important in the everyday lives of most species, and one that is susceptible to disruption in neurodevelopmental, neuropsychiatric, and neurodegenerative disorders. NMDAR have been a focus of interest for their role in social affiliative behaviors as well as a potential therapeutic target in disorders such as autism and schizophrenia that cause social withdrawal.

In the current study, we used the Cre/loxP system to ablate the NMDAR subunit NR1 in excitatory or inhibitory cells throughout the brain. Mice containing loxP sites flanking the NR1 gene (Grin1) were mated to mice with Cre recombinase expression in postnatal excitatory forebrain neurons driven by the calcium/calmodulin-dependent kinase II α (CaMKII α ) promoter^92,93, and the resulting CaMKII-Cre⁺; NR1^flox/flox mice exhibited an increase in sociability. These mice were previously shown to lack Cre in GABAergic interneurons and exhibit a decrease of NR1 mRNA of 57% in the cortex and 66% in the hippocampus. Loss of NMDAR current in hippocampal pyramidal neurons was confirmed as well⁶⁵. In general, CaMKII-Cre⁺; NR1^flox/flox mice may spend slightly more time sniffing overall, including empty cylinders and the cylinder containing an object than their CaMKII-Cre⁺; NR1^wt/wt littermates. However, when considering the delta sniffing values, they have a clear, robust increase in social behavior. Therefore, the deletion of NR1 in neurons of the hippocampus, cortex, striatum, and amygdala⁹⁴ resulting in increased delta sniffing time (increased social approach), compared to that of littermates positive for Cre but wild-type for the NR1 floxed allele, indicates that postnatal excitatory forebrain neurons play an important role in modulating social affiliative behavior in juvenile and young adult mice.

Interestingly, in our hands, the deletion of NR1 in parvalbumin cells, using the same NR1-floxed mouse line mated to a Cre line with a PV promoter, had no effect on social behavior. Approximately 40-50% of interneurons express the calcium-binding protein, parvalbumin, and those cells are found in the somatosensory cortex, motor cortex, entorhinal cortex, striatum, hippocampus, amygdala, thalamic reticular nucleus, globus pallidus, and cerebellum^69–71,95. This PV-Cre line has been shown to exhibit recombination limited to PV-positive interneurons and expresses Cre in over 90% of PV+ cells^67,82. Therefore, this population of interneurons may not be crucial to social affiliative behavior in juvenile and young adult mice.

Our data did not reveal any sex differences in social affiliative behavior with manipulation of NR1 in forebrain excitatory cells or PV-positive interneurons in juveniles or young adults. There was a main effect of sex in locomotor activity in PV-Cre+; NR1 mice, with females traveling more than males. It was important to include both males and females because, while many brain regions express NR1 at similar levels between males and females, some differences have been reported in specific brain regions during development^96–99. In addition, much of the previous literature includes only male subjects. We were also interested in whether manipulation of NR1 expression in these cell types would have different effects in the juvenile versus young adult phase. Levels of NR1 appear more dynamic during prenatal brain development, with less changes over time postnatally, although some brain areas exhibit decreased NR1 levels over postnatal development, including cortex, hippocampus, hypothalamus, and thalamus^100–103. Surprisingly, we uncovered a main effect of age on social approach behavior in mice carrying the PV-Cre allele, regardless of whether they also expressed the floxed NR1. In both cases, 30-day-old juvenile males and females exhibited higher levels of social behavior than adults. However, the same result was not demonstrated by the CaMKII-Cre-positive mice. Regardless of age or sex, presence of both CaMKII-Cre and NR1 floxed alleles were necessary to increase social approach. In these mice, there was an effect of age only on distance traveled, with 30d animals traveling less than 70d animals in Phase 2, in both CaMKII-Cre⁺; NR1^flox/flox and CaMKII-Cre⁺; NR1^wt/wt males and females.

This difference between the lines in the effect of age is likely the result of the presence of Cre in different cell types in the two lines and should be considered as a limitation to these experiments. In addition, onset of Cre expression, and therefore the start of cell-type specific ablation of NR1, occurs at different points in development in these two lines; CaMKII-Cre expression begins shortly after birth and increases dramatically around postnatal day 5, whereas PV-Cre expression begins much earlier, at E14^73,82,83. However, despite the lack of genetic deletion of NR1 in excitatory cells very early in development, disruption postnatally is sufficient to modulate social approach behavior.

Importantly, in our studies, genetic background was controlled for in our breeding strategies, as this can have significant impact on behavioral outcomes¹⁰⁴. We also used same-sex, gonadectomized non-aggressive A/J mice as social stimuli to avoid aggressive and sexual social motivations. As controls, we used littermates positive for Cre expression but lacking floxed NR1 alleles, but we did not include mice positive for floxed NR1 but lacking Cre. It is possible there is a non-specific effect of the presence of Cre, and this can be considered a caveat to the study. However, fear conditioning experiments in our lab comparing CaMKII-Cre mice to WT littermates showed no genotype differences in males or females¹⁰⁵. Because this Cre line had no effect on an amygdala-dependent fear conditioning behavior, it seems unlikely that this Cre would affect social approach behavior, which also involves amygdala activity, according to our previous studies⁷. Relatedly, our breeding scheme included both male and female Cre and floxed mice to obtain experimental mice. In some cases, Cre undergoes germline recombination, which can result in full knockout of the gene, rather than deletion in the desired specific pattern. This may be an issue particularly in males carrying the Cre allele. However, the PV-Cre line used here was reported to undergo less germline recombination than other available lines due to very weak expression in Leydig cells and no expression in spermatids¹⁰⁶. Furthermore, our PCR analysis genotyping protocol uncovered no unexpected recombination via faint or missing bands or bands of incorrect number or size¹⁰⁷. Therefore, we do not expect this issue to have affected our results or interpretations.

An additional caveat is the baseline difference in sniffing in CaMKII-Cre and PV-Cre control mice lacking floxed NR1. While this may indicate a nonspecific effect of Cre that results in a sociability deficit which is rescued by deletion of NR1 in forebrain excitatory cells, a possibility that should be ruled out in future experiments, it is more likely due to variability in social behavior across experiments, which is observable in a number of social approach studies^{88,89,108,109}. Finally, the mice used in these experiments were tested both at 30 days and again at 70 days, but the time spent sniffing at 70 days did not appear to be affected by the single social exposure more than one month earlier, compared to previously collected data with naïve mice of the same age. Other groups have also reported no effects of earlier non-stressful behavioral testing^109,110.

Previous literature has revealed an important but complex role for NR1 in social behavior. Several groups have reported that global reduction of NR1 causes social impairment. NR1^neo−/− mice with constitutively reduced NR1 expression (~5-10% of wild-type) exhibited reduced social approach in a resident-intruder test and a three-chamber social approach test^{35,36,66,111,112}. In addition, disrupting the phosphorylation site of the NR1 subunit resulted in deficits in homecage social interaction⁵⁸. Other studies have shown mixed results regarding the effect of removal of NR1 from PV cells on sociability. One group reported a social deficit following early postnatal elimination of NR1 in roughly half of cortical and hippocampal interneurons, most of which were PV-positive, using a Ppp1r2-Cre line⁵⁷. Comparatively, our PV-Cre experiments resulted in NR1 loss limited to PV-expressing neurons and excluded other types of interneurons, but included loss in more brain areas, rather than just the hippocampus and cortex⁵⁷. In addition, onset of NR1 loss in our mice was at approximately E14 versus the postnatal elimination in the previous study^57,82.

There are also several published studies using the same genetic mouse strains as those reported here (PV-Cre and NR1-flox). Some used test apparatus and social stimulus mice similar to those employed in our current study and describe social approach deficits resulting from NR1 ablation in PV-positive interneurons^67,113. However, mice in those studies were significantly older than those used in our current study. In addition, previous results were from singly-housed mice fitted with implanted recording electrodes, and mice had been subjected to several behaviors prior to social testing¹¹³. Another group using the same PV-Cre⁺; NR1^flox/flox mice found no social behavioral deficits resulting from deletion of NR1 in PV cells, which aligns with our findings of lack of social effect of ablation of NR1 in PV-positive cells¹¹⁴. Therefore, specific cell types in which NR1 is deleted and the developmental timepoint at which the deletion occurs have important effects on social behavior. Likewise, age at testing, housing conditions, experimental procedures, and prior experience likely have important effects on social behavior and modulate the behavioral effects of NR1 ablation on social behavior.

In our current results, deletion of NR1 from CaMKII-positive excitatory neurons gave rise to a significant increase in social sniffing time. A previous publication reported a social approach deficit in the same genetic mouse line with similar experimental parameters. However, the subjects of that study were significantly older, singly-housed males with previous behavioral and surgical exposure. Testing was also shorter and did not include habituation to the arena⁶⁵. Therefore, the age and previous experiences of the mice seem to be extremely important for social behavior, particularly if some of the previous experiences are stressful. Specifically, single housing has been shown to have negative effects on sociability^{76,110,115–117}. Other groups have also found excitatory neurons to be important in social behavior. When Tye and colleagues optogenetically stimulate CaMKII-positive cells in the basolateral amygdala that project either to the medial prefrontal cortex or the ventral hippocampus, mice exhibit fewer social behaviors than a control group. Likewise, optogenetic inhibition of those CamKII-expressing projection neurons induces social behaviors^11,12. It would be interesting to determine whether those cells being manipulated expressed NR1. Taken in the context of previous reports, our results indicate that NR1 plays an important role in social approach behaviors, but that the effects of NR1 on social approach behaviors are quite complex, and may vary strikingly depending on the age at which NR1 expression is altered, the specific cell types and brain areas affected, the prior experience and housing conditions of the mice, and age at which social approach testing is conducted.

Future experiments should be aimed at determining the contribution of NR1 to social behavior in specific brain regions. One relevant brain area may include the basolateral amygdala, as it has been shown to be important for social approach behavior^7,11,12. The medial prefrontal cortex and the CA3 region of the hippocampus may also play a role; deletion of NR1 in the mPFC increased social novelty preference while deletion in CA3 decreased social approach⁶⁰. In another study, deletion of NR1 in nucleus accumbens decreased social interaction⁶¹.

Determining the mechanisms by which NR1 modulates social behavior is critical as well. As the obligatory subunit of NMDAR, its absence in some cells may affect NMDAR expression in other cells, or expression of other molecules or receptors. Manipulation of NR1 has been shown to cause changes in AMPAR-mediated synaptic transmission and AMPAR expression^58,118. Reduction of NR1 also causes decreases in spine density¹¹⁹. Therefore, altering NR1 expression patterns likely initiates a number of significant downstream effects that may contribute to alterations in social behaviors. Finally, other subunits may also be important for social behaviors (see Zhou and Sheng, 2013¹²⁰ for review). Mice that overexpress NR2B exhibit increased social recognition, but those that overexpress NR2A display deficits in olfaction and social memory^121–123. Similarly, reduction of the signaling factor C-terminal Src kinase (Csk) led to increased levels of NR2B and enhanced social memory¹²⁴. A transgenic mouse model in which NR2B subunits were replaced with NR2A subunits exhibited decreased social exploration¹²⁵. Studies of children with NR2B mutations also have atypical social behavior including impulsive, overly friendly and lacking boundaries, or disruptive^126,127. NR2D may even play a role in some types of social behaviors, despite its limited expression in the mature brain; a knockout mouse model demonstrated normal social interaction but impaired social memory¹²⁸.

In conclusion, NR1 are important for social affiliative behavior, particularly in excitatory neurons of the forebrain. Postnatal deletion of these cells results in an increase in social approach and sniffing in juveniles and young adult mice, while deletion in PV-positive GABAergic cells has no effect at this age. Age at which NR1 expression is altered, cell type, and brain area, and age of behavioral testing are all critical factors in the precise effects of NR1 expression on sociability, as well as type of social test and social stimulus. Future studies are needed to continue to elucidate NR1 mechanisms of social affiliative behaviors, in order to ultimately develop improved understanding of social withdrawal and improved treatments for social withdrawal in various neuropsychiatric conditions.