Oxytocin receptor gene loss influences expression of the oxytocin gene in C57BL/6j mice in a sex- and age-dependent manner

Abstract

Parental care and sensory stimulation are critical environmental factors that influence oxytocin (OXT) and its receptor (OXTR). Because developmental Oxt mRNA expression is enhanced by sensory-rich early life experience, and reduced by sensory deprivation, we predicted that compared to wild-type (WT) littermates, mice with congenital loss of OXTR (OXTR KO), as a genetically-induced deprivation, would show impaired Oxt mRNA expression in the offspring hypothalamus during development. Oxt mRNA levels of male and female OXTR KO mice were not different from WT littermates from P0-P6, but that by P8, OXTR KO showed significantly decreased Oxt mRNA expression in the hypothalamus compared to WT littermates. At P14, male and female OXTR KO mice had significantly decreased Oxt mRNA expression specifically in the paraventricular nucleus (PVN), but not the supraoptic nucleus (SON), compared to WT littermates. We assessed if this effect persisted in adulthood (P90) and found a significant genotype by sex interaction where male OXTR KO mice displayed a reduction in Oxt expression specific to the PVN compared to male WT littermates. In contrast, male and female OXTR KO adults had increased Oxt mRNA levels in the SON. These findings suggest that OXTR plays a role in developmental Oxt mRNA expression with sex by genotype interactions apparent at adulthood. We then measured OXT and neural activation in the PVN and SON at P14. We observed more OXT immunoreactive cells in the PVN of OXTR KO mice, but significantly fewer c-Fos immunoreactive cells. There were no genotype differences in immunoreactivity for OXT and no c-Fos activity in the SON at P14. Combined, these data suggest that OXTR WT P14 mice have more PVN activity and are more likely to release OXT than OXTR KO mice. Future experiments are warranted to understand which OXTR-expressing neural circuits modulate the development of the PVN oxytocin system.

Introduction

The oxytocin (OXT) system has been hypothesized to act as both a mediator (effector) and target of early life experiences [1, 2]. As an effector, OXT may facilitate the acquisition of socially salient information in the neonatal environment [3, 4], specifically, sensory experiences in a social context, as demonstrated in adults [5–7]. As a target, the activity and/or development of the OXT system can be altered by early life experiences. While the neural mechanisms of the effector and target roles of OXT in social development are unclear, these roles may be linked in a feedforward virtuous cycle: some enriched early experiences may lead to enhancement of infant OXT production, which could be used to further encode the social environment during developmental sensitive periods for attachment and socialization. In contrast, sensory and/or social deprivation may lead to reduced OXT production, and an impairment in OXT-dependent development.

The OXT system is a mediator of developmental experience. Exposure to OXT during postnatal development has been shown to have long-lasting effects on adult physiology and social behaviors such as parental behavior and partner preference behavior [8, 9]. Physiological effects of neonatal subcutaneous OXT injections include: increased weight gain and increased nociceptive threshold in male and female rats at adulthood [10], rescue of feeding behavior in neonatal Magel2 Knock-out mice (a model for Prader-Willi syndrome) [11], and decreased blood pressure in adult male and female rats [12]. Neonatal intraperitoneal OXT injections have sexually-dimorphic effects on adult social behavior [13] and OXT immunoreactivity in the PVN at postnatal day 21 (P21) [14]. Neonatal central OXT manipulation also affects early infant social behavior. Intra-cisternal OXT injections at P10, but not at P7 causes more compact huddling within litters of rat pups [15]. The effects of OXT in the newborn include significant changes in neural activity that are important for the transition into postnatal life [16, 17]. Conversely, the absence of OXT or OXTR during postnatal development has significant effects on pup behaviors essential to elicit maternal care such as ultrasonic vocalizations (USVs). OXT Knock-out (KO) mouse pups display significantly decreased ultrasonic vocalizations (USVs) compared to the wild-type (WT) pups [18] and male OXTR KO mouse pups display significantly fewer USVs compared to their WT age-matched controls at P7 [19]. Adult OXTR KO male mice show deficits in social discrimination and increased aggressive behavior compared to wild-type controls [19]. The absence of OXT/OXTR signaling during development may contribute to the abnormal maturation of circuits important for typical adult social behaviors.

The OXT system is a target of developmental experience [20, 21]. In response to social isolation, prairie vole pups show enhanced c-Fos activity and OXT immunoreactivity in the PVN at P9. This increase can be interpreted as either an increase in OXT production in response to a stressor, or abnormally low levels of OXT release during social isolation at this age [22]. Sensory deprivation by whisker removal or dark-rearing of C57BL/6J mice since birth, reduces Oxt mRNA and OXT peptide in the PVN, and OXT levels in the CSF [23]. Conversely, mice reared in an enriched environment from birth to P14, show increased Oxt mRNA levels in the hypothalamus as well as increased OXT peptide signaling levels in the somatosensory cortex and visual cortex at P14 [23].

As both a mediator and a target of developmental experience, Oxt mRNA expression may rely in part on intact experience-dependent OXT signaling through the OXTR. Congenital loss of OXTR may impair Oxt mRNA expression. Given that OXTR in some brain areas has a transient dynamic developmental expression [24, 25], the effects of OXTR loss on Oxt mRNA expression might also be dynamically regulated during specific developmental stages. We hypothesize that typical Oxt mRNA expression levels through development requires the Oxtr gene. To test our hypothesis, we used reverse transcriptase quantitative PCR (RT-qPCR) to measure gene expression changes in Oxt mRNA in WT and OXTR KO mice on a C57BL/6J background at different developmental ages. If activity at infant OXTR is required for the full complement of Oxt mRNA expression, then Oxtr null pups should show a reduced level of Oxt gene expression in the hypothalamus compared to their wild-type littermates. We also tested the effect of Oxtr gene loss on OXT peptide content and neural activation indicated by c-Fos in P14 mice. We tested this hypothesis using both male and female mice.

Materials and Methods

Mice

Oxtr mice (Oxtr^tm1.1knis) [26] were bred within our laboratory after genetically-confirmed backcrossing to C57BL/6J [24]. All procedures were performed after approval by The Institutional Animal Care and Use Committee, Florida State University (protocol #1722 and #1746) in accordance with state and federal guidelines (Guide for the Care and Use of Laboratory Animals of the National Institutes of Health). Animals were housed on a 12L:12D forward cycle in wire top caging with wood chip bedding, a nestlet, and ad libitum access to food (LabDiet 5001) and water. Pregnancies were not timed; heterozygous breeding pairs were checked daily, with the first appearance of a litter established as P0. Litters for RT-qPCR were collected on P0, P2, P4, P6, P8 and P14. At P21, litters were weaned into same sex cages of littermates until P90. On the day of sample collection of pre-weanling litters, parents were removed from the homecage, the entire litter was transferred onto a heating pad that maintained the bedding temperature at 28 ± 2°C through the entire duration of sample collection. Each pup was separated from the litter, weighed and euthanized by decapitation. Tail samples were collected for genotyping and whole brains were dissected out of the skull. Whole brains were snap frozen in isopentane on dry ice and stored at −80°C. All samples were collected between 12pm and 6pm. A total of 54 litters were generated from 20 breeder pairs to generate the sample sizes for the RT-qPCR experiments. A separate cohort of samples were collected at P14 for immunohistochemistry: once every ten minutes, a pup was selected at random from the home cage and housed individually in a 3 oz (88mL) plastic cup for 1 hour on a heating pad that maintained the temperature at 28 ± 2°C. A total of 5 litters from 5 separate breeder pairs were used for OXT immunohistochemistry, which did not use a within-litter design. Only one pup of the same genotype and sex from each litter contributed to the sample size.

Sex and Oxtr genotyping

Genomic DNA was collected from tail samples for genotyping for sex and Oxtr by PCR [24]. Briefly, to determine genetic sex of neonates, the forward primer (5’-ccgctgccaaattctttgg-3’) and the reverse primer (5’-tgaagcttttggctttgag-3’) generated a 290 bp product from the Smcy gene on the Y chromosome, and a 330 bp product from the Smcx homolog on the X chromosome under the following thermal cycling conditions: 95°C for 7 min; 35 cycles of 93°C for 30 s, 58°C for 30 s, 72°C for 30 s; 72°C for 10 min. Oxtr genotypes were determined by PCR with the forward primer (5’-ctggggctgagtcttggaag-3’) and the reverse primers for WT (5’-ctcgatactccagttggctgc-3’) or KO (5’-gttgggaacagcggtgatta-3’). These primers generated a 665 bp product for the WT allele, a 450 bp product for the KO allele under the following thermal cycling conditions: 94°C for 5 min; 37 cycles of 94°C for 30 s, 57°C for 45 s, 72°C for 90 s; 72°C for 7 min.

RNA extraction and cDNA synthesis

Frozen whole brains stored at −80°C were micro-dissected in a cryostat (Leica CM1850) at −20°C. The diencephalon was dissected out from whole brains for P0-P8. At P14 and P90, PVN-containing tissue and SON-containing tissue were micro-dissected from a 2–3 mm thick coronal section of the brain containing the diencephalon. Depending on the developmental age, diencephalon tissue or PVN-containing and SON-containing tissue was homogenized in 500 μl TRIzol (Invitrogen) using a pestle, and further homogenized using a Sonic Dismembrator (Fisher Scientific). A standard phenol-chloroform method was used to extract total RNA. The RNA concentration was measured using a Nanodrop™ Spectrophotometer (ND-1000). Total RNA for each sample was diluted to 600 ng/μl. Total RNA (1 μl) was run on a 1% agarose gel with a 1kb ladder (Invitrogen). Presence of 28S and 18S ribosomal RNA bands confirmed the integrity of extracted RNA. Equal concentrations of input RNA (4.2μg) were treated with DNase. Total RNA was digested at 37°C for 30 minutes with RQ1 RNase-Free DNase (1 μl/μg of RNA) (Promega) and incubated at 65°C for 5 minutes with stop solution (Promega). Half of the total reaction volume (10 μl) was used for cDNA synthesis and the rest was stored at −80°C and later used to check for genomic DNA contamination through qPCR. DNase digested RNA (10 μl) was reverse-transcribed to cDNA using the Applied Bio Hi Capacity cDNA kit. The following cycling conditions were used: 25° C for 10 min, 37° C for 120 min, 85° C for 5 min. cDNA was diluted 1:2 and run on a 1% agarose gel with a 1kb ladder. The cDNA concentration for qPCR was calculated to be ~50ng/μl, under the assumption that 1–5% of total RNA is mRNA.

Quantitative PCR

Samples of cDNA (~ 50 ng) were quantified by quantitative PCR and SYBR-Green detection methods in a Biorad CFX thermal cycler (BioRad, Hercules, CA). Prior to the experimental run, a standard curve was generated with serial dilutions of cDNA, testing GAPDH and OXT primer pairs. The coefficients of efficiencies were calculated using Biorad CFX manager software for the primer sets: GAPDH (E= 99.2%, R²= 0.994) and OXT (E= 111.1%, R²= 0.996). The primer sequences used were: Gapdh forward (5`-aatggtgaaggtcggtgt-3`), Gapdh reverse (5`-gtggagtcatactggaacatgtag-3`), Oxt forward (5`- ccgaagcagcgtccttt-3`), and Oxt reverse (5`-cttggcttactggctctgac-3`). A gradient PCR was run to optimize the annealing temperature for the primer sets, and the presence of a single specific amplicon for each gene was confirmed by agarose gel electrophoresis. The cycling conditions were as follows: 94° C for 5 min followed by 40 cycles of denaturation at 94° C for 30 seconds and annealing at 60° C for 30 seconds. Melt curve data were analyzed to confirm the presence of a single specific PCR amplicon for all primers. Experimental plates were set-up to include sibling pairs on the same run. Every plate included samples in triplicates, and a no template control (NTC) and an inter-plate normalizer. The inter-plate normalizer showed consistent performance across all plates. Males and females were run on different plates for the comparison between WT and OXTR KO sibling pairs, and on the same plates for within litter sex comparisons in WT only (see supplemental materials). Additionally, the DNase-treated RNA for every sample was run with Gapdh primers to confirm the absence of genomic DNA.

Relative expression calculations

Raw Cq values from 3 technical replicates were used to calculate mean Cq for the reference gene (Gapdh), and mean Cq for the gene of interest (Oxt), for each sample. ΔCq were calculated as mean Cq(Oxt) - mean Cq(Gapdh) for each sample. Relative expression was calculated as 2^(-ΔCq).

Immunohistochemistry (IHC)

To observe OXT and c-Fos staining with bright field microscopy, perfused brains from P14 mice were used in IHC. Because of the robust effect size beyond littermate comparisons at P14 for Oxt mRNA, in this immunohistochemical study we did not restrict our analysis to littermate pairs. Mice were overdosed with sodium pentobarbital and transcardially perfused with 0.1M phosphate buffer pH 7.4 followed by 4% PFA, pH 7.4. Brains were post-fixed in 4% PFA overnight and immersed in 10%, 20% and 30% sucrose in PBS for 24 hours each. Brains were then sectioned in 3 series in the coronal plane at 40 μm on a microtome (Microm HM450, Thermoscientific). Sections were stored in freezing media (25% glycerol, 30% ethylene glycol, in PBS) at −20°C until immunostaining. Sections were washed 4 × 5 min in PBS, incubated for 10 minutes in 0.5%H₂O₂ in PBS, then washed 4 × 5 min in PBS. Sections were washed 3 × 10 minutes in 0.1M Tris-glycine (pH 7.4), sections were pre-blocked by 4 × 5-min washes in 4% Blotto (non-fat dry milk in PBS) and 0.2% Triton-X 100, followed by incubation with a 1:100 mouse monoclonal anti-OXT antibody (MAB AI-28, a generous gift from Dr. A.J. Silverman, Columbia U) in 4% Blotto for 72 hours. After incubation, slides were washed in 4% Blotto 5 × 5 minutes, followed by incubation in 1:1000 biotinylated donkey anti-mouse IgG in 4% Blotto for 60 min. Sections were washed in 4% Blotto 5 × 5 min, followed by wash in PBS 2 × 5 minutes and incubated with Vectastain ABC elite (VectorLabs) for 1 hour. Sections were washed in PBS 5 × 5 minutes and then reacted in 0.05% DAB and 0.01% H₂O₂ in PBS for 3 minutes. Sections were washed 5 × 5 minutes in PBS, mounted on gelatin coated slides, air dried, dehydrated through a graded alcohol series, defatted in Citrisolv and coverslipped with DPX. Images were captured on a bright field microscope (Keyence BZ-X710, Keyence Corp., Osaka, Japan).

Immunoreactivity for c-Fos was performed on an adjacent series for each specimen, following the same protocol as above with a rabbit anti-c-Fos antibody (1:20,000; Oncogene Ab-5 cat# PC-38) and a biotinylated donkey anti-rabbit secondary (1:1,000; cat# 711–065-152, Jackson Immunoresearch). The biotin signal was amplified with Vectastain ABC standard (VectorLabs). The DAB reaction was enhanced with nickel cobalt (0.05% DAB, 1% nickel sulfate, 1% cobalt chloride, 0.01% hydrogen peroxide).

Quantification of OXT and c-Fos immunostaining

Images of the PVN and SON were captured at 2X magnification and arranged along the rostro-caudal axis. A total of 5 ordered sections per sample were imaged at 10X. OXT immunoreactive neurons and c-Fos immunoreactive nuclei were counted manually using the multi-point tool in Image J [27].

Statistical Analysis

All RT-qPCR samples had Cq values between 15–35 with Gapdh Cq values ranging from 15– 23 (Supplemental FigS1). Gapdh Cq values were not different based on genotype or sex, although, as expected Gapdh Cq values did vary with age. Statistical analyses were conducted on 2^(-ΔCq) values. RT-qPCR data were analyzed using mixed-effects ANOVA to control for variation across litters tested [28]. For determining the impact of Oxtr gene loss on Oxt expression by RT-qPCR, litters were considered the experimental subjects, with sex as a between-subjects factor and genotype as a within-subjects factor for each developmental age. Because Gapdh Cq values did not differ for ages P0-P6, data for this age range were analyzed by three-way mixed-effects ANOVA with developmental age, sex and genotype as factors. Statistically significant interactions were followed up by Post-hoc Sidak’s multiple comparisons test. Statistical significance was set at p <0.05. At P14 and P90, the PVN and SON were dissected apart and analyzed separately. For analysis on the total number of OXT-immunoreactive neurons in the PVN and SON, data were tested for normality. Normally distributed data were analyzed by a two-tailed t-test. We were underpowered to detect any sex-differences in the number of OXT-positive neurons and therefore combined sexes to compare genotypes. All raw data are available upon request.

Results

Oxytocin mRNA expression is not different in OXTR KO compared to WT littermates aged P0-P6

To assess the developmental regulation of Oxt gene expression from birth, we sampled same-sex, WT/ OXTR KO sibling pairs from 4 different litters at each time point (16 litters per sex): P0, P2, P4 and P6. With developmental age, sex, and genotype as factors, no significant main effects or interaction effects were revealed in Oxt mRNA expression [p > 0.05, Three- way Mixed-Effects ANOVA], (Figure 1; Table 1).

Figure 1. Oxt mRNA expression is not different in OXTR KO compared to WT littermate neonates younger than P8.

Same-sex WT and OXTR KO littermates (n = 4 litters per sex) were assessed from P0 to P6. Data are presented as mean 2^(-ΔCq) values ±SEM. No significant effects of age, sex, or genotype were revealed in hypothalamic Oxt mRNA expression from P0 to P6, p>0.05, Three-way Mixed-effects ANOVA.

Table 1:

Mixed-effects Analysis of Variance of P0-P6 Oxt mRNA expression in the hypothalamus

Variable	F-Value (df)	P Value
Genotype	0.136 (1,48)	0.714
Sex	0.077 (1,48)	0.783
Age	2.513 (3,48)	0.070
Age × Sex	0.613, F (3, 48)	0.610
Age × Genotype	2.090, F (3, 48)	0.114
Sex × Genotype	0.133, F (1, 48)	0.717
Age × Genotype × Sex	0.480, F (3, 48)	0.697

Oxytocin mRNA expression is reduced in OXTR KO compared to WT littermates at P8

We tested a sample of within-sex, WT/ OXTR KO sibling pairs from 4 litters for males and 4 litters for females at P8. Oxt mRNA expression revealed a significant main effect of genotype [F (1,6) = 8.04, p = 0.030, Mixed-effects Two-Way ANOVA] in Oxt mRNA expression. OXTR KO had significantly lower levels than their WT littermates (Figure 2, Table 2). These results are consistent with data from a separate pilot study we performed in males with an alternative set of primers (supplemental figure 2).

Figure 2. Oxt mRNA expression is reduced in OXTR KO compared to WT littermates at P8.

Each data point value represents a sample. Same-sex WT and OXTR KO littermates (n = 4 litters per sex) are connected by a dashed line. Bars represent mean 2^(-ΔCq) values ±SEM. Hypothalamic Oxt mRNA expression is higher in WT pups compared to KO pups. Significant main effect of genotype, *p<0.05, Two-way Mixed-effects ANOVA.

Table 2:

Mixed-effects Analysis of Variance of P8 Oxt mRNA expression

Variable	F-Value (df)	P Value
Genotype	8.049 (1,6)	0.030
Sex	0.025 (1,6)	0.881
Genotype × Sex	4.708 (1,6)	0.073

As an extension to a previously published study of sex differences in perinatal Oxt expression in WT mice [29], we assessed Oxt mRNA expression in WT pups across development and evaluated littermate and non-littermate comparisons. With non-littermate comparisons, we observed a significant main effect of age [p = 0.021, F (4, 30) = 3.383] on Oxt mRNA expression, and no significant main effects of sex [p = 0.508, F (1, 30) = 0.450] or interaction effects [p = 0.110, F (4, 30) = 2.065] in a Two-way ANOVA (Supplementary FigS3). Littermate comparisons revealed a non-significant main effect of sex [p = 0.050, F (1,15) = 4.529, Two-way Mixed-effects ANOVA] and a significant age by sex interaction [p = 0.008, F (4, 15) = 3.229, Two-way Mixed-effects ANOVA] with P8 WT females expressing lower Oxt mRNA compared to WT males, and no significant main effect of age [p = 0.484, F (4, 15) = 0.908] (Supplemental FigS3). Our littermate comparisons between WT males and WT females were obtained from OXTR^+/+ by OXTR^+/+ and OXTR^+/+ by OXTR^+/− in addition to the OXTR^+/− by OXTR^+/− breeder pairs needed for the within litter genotype comparisons.

Oxytocin mRNA expression in the PVN, but not SON, is reduced in OXTR KO compared to WT littermates at P14

To assess if the effect observed at P8 persists at P14, an age previously established to show significant experience-dependent effects on Oxt expression [23], we sampled same-sex WT/ OXTR KO littermates at P14 from 4 different litters per sex to measure Oxt mRNA expression in the PVN and SON. Oxt mRNA expression in the PVN revealed a statistically significant main effect of genotype [F (1,6) = 8.916, p = 0.024] with OXTR KO expressing lower Oxt mRNA levels compared to WT littermates, and a significant main effect of sex [F (1,6) = 17.84, p = 0.005] (Figure 3A, Table 3), with females expressing higher levels of Oxt mRNA compared to males.

Figure 3. Oxt mRNA expression in the PVN, but not SON, is reduced in OXTR KO compared to WT littermates at P14.

Individual samples from same-sex WT and OXTR KO littermates (n = 4 litters per sex) are connected by a dashed line. Bars represent mean 2^(-ΔCq) values ±SEM. A. PVN Oxt mRNA expression is higher in WT compared to KO littermates at P14. PVN Oxt mRNA expression is higher in females compared to males (non-littermate comparison) at P14. Significant main effect of genotype, *p < 0.05; Significant main effect of sex, **p < 0.01; Two-way Mixed-effects ANOVA B. SON Oxt mRNA expression is not different in WT compared to KO littermates at P14. SON Oxt mRNA expression is higher in females compared to males (non-littermate comparison) at P14. Significant main effect of sex, ***p < 0.001, Two-way Mixed-effects ANOVA.

Table 3:

Mixed-effects Analysis of Variance of P14 Oxt mRNA expression in PVN

Variable	F-Value (df)	P Value
Genotype	8.916 (1,6)	0.024
Sex	17.841 (1,6)	0.006
Genotype × Sex	0.433 (1,6)	0.535

Oxt mRNA expression in the SON revealed a significant main effect of sex [F (1,6) = 52.33, p < 0.001] (Figure 3B, Table 4) with females expressing higher levels of Oxt mRNA compared to males. There was no main effect of genotype in the SON at P14, nor a genotype by sex interaction (Table 4).

Table 4:

Mixed-effects Analysis of Variance of P14 Oxt mRNA expression in SON

Variable	F-Value (df)	P Value
Genotype	0.070 (1,6)	0.801
Sex	52.330 (1,6)	<0.001
Genotype × Sex	2.260 (1,6)	0.183

Sex specific reduction in OXT mRNA in the PVN, in male OXTR KO compared to WT littermates at P90

Next, to assess if congenital loss of OXTR impacts Oxt expression in adult mice (P90), we sampled same-sex WT/ OXTR KO littermates co-housed since weaning from 4 different litters. Oxt mRNA expression in the PVN revealed a significant genotype by sex interaction [F (1,7) = 7.231, p = 0.030, Two-way Mixed-effects ANOVA] (Figure 4A, Table 5). Sidak’s multiple comparison post-hoc indicated that OXTR KO males had lower levels of Oxt mRNA compared to WT male littermates.

Figure 4. Sex specific reduction in Oxt mRNA in the PVN, but not in the SON, in male OXTR KO compared to WT littermates at P90.

Individual samples from same-sex WT and OXTR KO littermates (n = 4 litters per sex) are connected by a dashed line. Bars represent mean 2^(-ΔCq) values ±SEM. A. PVN Oxt mRNA expression is higher in adult male WT compared to KO littermates. PVN Oxt mRNA expression is not different in adult female WT compared to KO littermates, nor were females different compared to males (non-littermate comparison). Significant interaction of genotype × sex, p < 0.05 Two-way Mixed-effects ANOVA, Sidak’s multiple post-hoc comparison ^#corrected p = 0.030 B. SON Oxt mRNA expression is higher in adult KO compared to adult WT mice. Significant main effect of genotype, *p < 0.05, Two-way Mixed-effects ANOVA.

Table 5:

Mixed-effects Analysis of Variance of P90 Oxt mRNA expression in PVN

Variable	F-Value (df)	P Value
Genotype	2.496 (1,7)	0.158
Sex	1.357 (1,7)	0.282
Genotype × Sex	7.231 (1,7)	0.031

Oxt mRNA expression in the SON is increased in OXTR KO compared to WT littermates at P90

We also measured Oxt mRNA expression in the SON at P90. Oxt mRNA expression in the SON at P90 revealed a significant main effect of genotype [F (1, 6) = 8.814, p = 0.025, Two-way Mixed-effects ANOVA] (Figure 4B, Table 6). OXTR KO mice expressed higher levels of Oxt mRNA compared to WT littermates in the SON at P90.

Table 6:

Mixed-effects Analysis of Variance of P90 Oxt mRNA expression in SON

Variable	F-Value (df)	P Value
Genotype	8.814 (1,6)	0.025
Sex	3.404 (1,6)	0.115
Genotype × Sex	0.681 (1,6)	0.441

Oxytocin immunoreactivity in the PVN, but not SON, is significantly different between WT and OXTR KO at P14

We followed up our gene expression results to assess OXT immunoreactivity in the PVN and SON neurons at P14. Due to the observed effect sizes at P14 beyond littermate comparisons, we sampled non-littermate WT (3 males, 4 females) and OXTR KO (4 males, 5 females) weanlings and measured OXT-immunoreactive neurons in the PVN and SON. Because we counted cells in every 40 μm section sampled in 3 series, samples with damaged or missing sections were not included. After sample loss, the final sample size for OXT IHC was 4 WT (1 male, 3 females) and 7 OXTR KO (4 males, 3 females). Thus, for this analysis, we only compared WT and KO, without considering sex. A student’s t-test indicated that OXTR KO mice at P14 had greater numbers of OXT immunoreactive cells in the PVN compared to WT mice (Figure 5A; t_(d.f.9) = 4.419; p = 0.001). There were no genotype differences in the number of OXT immunoreactive cells in the SON at P14 (Figure 5B; (t_(d.f.9) = 1.375; p = 0.202)).

Figure 5. At P14, in the PVN, OXTR WT has reduced OXT immunoreactivity and increased c-Fos immunoreactivity compared to OXTR KO, with no differences in the SON.

A. Representative images of OXT-immunoreactive neurons in the PVN. Scale bar, 100 μm. Number of OXT-positive neurons is reduced in WT compared to OXTR KO mice in the PVN at P14. B. Representative images of OXT- immunoreactive neurons in the SON. Scale bar, 100 μm. Number of OXT-positive neurons is not different between WT and OXTR KO mice in the SON at P14. C. Representative images of c-Fos- immunoreactive neurons in the PVN. Scale bar, 100 μm. Number of OXT-positive neurons is not different between WT and OXTR KO mice in the PVN at P14. D. Representative images of c-Fos-immunoreactive neurons in the SON. Scale bar, 100 μm. Number of OXT-positive neurons is not different between WT and OXTR KO mice in the SON at P14.

WT mice at P14 had higher levels of c-Fos immunoreactivity in the PVN

We hypothesized that this contradictory result (decreased OXT immunoreactivity in WT) could be explained by increased neural activity in the PVN of the WT hypothalamus, as this would release the available pool of OXT in the PVN for detection. In adjacent sections to those stained for OXT, we stained and counted the number of c-Fos immunoreactive nuclei in the PVN and SON. The final sample size after accounting for section loss in the c-Fos analysis was 6 WT (3 male, 3 females) and 7 OXTR KO (3 males, 4 females). In the PVN, OXTR WT mice had significantly more c-Fos immunoreactive nuclei in the hypothalamus compared to OXTR WT (Figure 5C; (t_(d.f.11) = 12.304; p = 0.042). In the SON, c-Fos was not detected, so it was not counted (Figure 5D).

Discussion

In this study we show through littermate comparisons that congenital absence of OXTR activity by Oxtr gene loss alters Oxt gene expression in the hypothalamus at various developmental ages, summarized in Figure 6. This is consistent with our hypothesis that OXTR may be used to drive Oxt expression during development, suggesting a role for OXTR in a positive feedforward system for socially acquired exogenous OXT and/or a positive feedback system for endogenous OXT. Sensory experience requires a mechanism to shape OXT production in development. OXTR signaling may be a part of that mechanism. Through pairwise within-litter comparisons between WT and OXTR KO male and female pups, we did not find any statistically significant main effects or interaction effects in Oxt gene expression in the hypothalamus from P0 through P6, with samples collected on every even day. However, by P8, male and female OXTR KO mice had lower levels of Oxt mRNA in the whole hypothalamus compared to WT littermates. At P14, we observed that male and female OXTR KO have decreased Oxt mRNA levels in the PVN but not in the SON compared to WT littermates. While OXTR gene loss was associated with reduced Oxt mRNA levels, OXT immunoreactive cell number was increased in the PVN at P14. These contradictory observations may be explained by reduced neural activation and subsequently reduced release of OXT in the OXTR KO mice when measured at P14.

Figure 6.

Summary of the effects of OXTR KO on Oxt mRNA expression and OXT and c-Fos immunoreactivity in mice.

At P90, male OXTR KO have decreased Oxt mRNA levels in the PVN compared to WT littermates, while female OXTR KO and their WT littermates did not show any significant differences in Oxt mRNA in the PVN. These data suggest that some compensatory mechanism exists in the females which restores the Oxt mRNA levels to normal between P14 and P90. Alternatively, the levels of Oxt mRNA in the WT female could be at a floor for detecting potential detrimental genotype effects if females require a more enriched environment than what is provided in standard laboratory caging, or perhaps Oxtr genotype differences in Oxt expression would re-appear in adult females with pregnancy and parturition. The adult mice used in our study were all sexually naïve.

Interestingly we observed an increase in Oxt mRNA levels in the SON in OXTR KO mice compared to WT at P90 in both sexes. Magnocellular neurons in the SON project to the pituitary to release OXT into the bloodstream, but they can also release OXT into the cerebrospinal fluid (reviewed in [30]). The release of OXT from magnocellular OXT neurons in the SON seems to be regulated by parvocellular OXT neurons in the PVN [31]. If there is a decrease in OXT production from P14 to P90 in the PVN this could impair the modulatory control of the PVN on the SON.

In the only other study to measure the effect of Oxtr genotype on Oxt mRNA expression, no differences in Oxt mRNA levels in the hypothalamus of OXTR KO mice were observed [19]. It is important to note that those data were not collected from within-litter WT and KO sibling pairs during development. Furthermore, grouping the PVN and SON can result in dilution of significant effects in one of the hypothalamic nuclei. While littermate controls are a unique strength of the current study, there are some limitations to the current study design: while we measured Oxt mRNA at various ages, we only measured OXT immunoreactivity at one age, and, these were not littermate comparisons. Additionally, the study lacks circuit level and mechanistic detail because the OXTR KO mice lack OXTR from all OXTR-expressing cells. Thus, we cannot determine which population of OXTR-expressing cells are most important and when they are most important. Finally, we did not manipulate the social or sensory environment. These caveats are elaborated in further detail below.

First, the study design focuses on changes in Oxt mRNA and does not thoroughly evaluate quantitative changes in OXT peptide production and release in OXTR WT compared to OXTR KO. We chose to focus on Oxt mRNA expression across time rather than OXT immunoreactivity as a more stable and trait-like dependent variable to evaluate the organizational effects of Oxtr gene loss. To determine if Oxtr gene loss was also associated with OXT immunoreactivity, we counted the number of OXT immunoreactive neurons in the PVN and SON at P14 and found more OXT immunoreactive cells in OXTR KO mice. The most parsimonious explanation for the discrepancy between Oxt mRNA and OXT protein is that activity just prior to tissue collection can influence the available quantity of OXT peptide for measurement (an activational effect), but not mRNA (Oxt is not an immediate early gene). Experimental handling of animals, such as social isolation prior to harvest may result in OXT release from neurons making it difficult to interpret the effects of gene loss on protein levels measured with immunohistochemistry due to the poor temporal resolution of that anatomical technique. Immunohistochemistry for immediate early gene products, like c-Fos in this study, provide direct support for an effect of Oxtr gene loss on PVN activity. For our hypothesis, Oxt mRNA represents a valuable measure of organizational trait-like effects while OXT immunohistochemistry is likely to be influenced by a combination of organizational and activational effects which are difficult to dissect apart with a single baseline measure. Sometimes measurements of mRNA and peptide may yield consistent results and sometimes not, making this a challenging research area. Changes at multiple regulatory steps between transcription and translation might be taking place resulting in a lack of correlation between mRNA and protein measurements. Perhaps, decreases in mRNA levels are compensated by higher translational efficiency resulting in higher levels of protein in the cell as a buffering mechanism. Importantly, while we counted the numbers of OXT immunoreactive neurons, we did not quantify the amount of OXT produced within these neurons. Future studies should include a time course of OXT release profiles, with and without stimulation. This is beyond the scope of the current study, which focuses on the developmental time course of Oxt mRNA production with and without the Oxtr gene.

Second, given that we used a congenital mouse model, we cannot determine which brain, or even peripheral, populations of OXTR-expressing cells are most responsible for the current findings, nor when they play the largest role. We have observed OXTR in both the neonatal central nervous system [24] and in the periphery [32]. In the current study, OXTR is eliminated in both compartments, so we cannot differentiate between central and peripheral effects of OXTR on Oxt expression. Thus, we also are unable to differentiate between a direct role for OXTR within the hypothalamus on Oxt expression or a role for OXTR in some other node in a circuit connected with the hypothalamus and regulating the expression of Oxt. To address this issue, in the future, we will use a Cre-LoxP approach to create a targeted deletion of OXTR expression in neonatal mice and quantify gene expression in the hypothalamus. We can also modulate the temporal expression of OXTR to determine if there is a developmental sensitive window for the expression of OXTR on adult social behavior outcomes.

Finally, we have only used a genetic manipulation rather than a gene by environment manipulation. If the OXT/OXTR system can modify the sensory detection and/or perception of the environment, then OXTR KO mice should be impaired in their ability to be shaped by environmental enrichment during development. Sensory deprivation (whisker removal or dark-rearing) since birth reduces both Oxt mRNA expression and the number of OXT-immunoreactive neurons in the PVN, but not the SON, when measured at P14 in males and females [23]. While we did not manipulate the external sensory experience of the mice, our findings with RT-qPCR are consistent with an interpretation that OXTR activity may modulate or enhance the impact of the sensory environment resulting in the observed changes in Oxt mRNA levels.

During development, OXTR KO mice still displayed production of Oxt mRNA and OXT peptide and showed some c-Fos activity of the PVN, suggesting that OXT signaling is still possible in these mice. However, without the OXTR, any effects of OXT would be limited to signaling through the vasopressin 1a receptor (AVPR1A) and the vasopressin 1b receptor (AVPR1B) [33], and perhaps the V2 receptor (AVPR2), although this receptor is not widely considered to have expression in the nervous system. This poses an interesting question for future research regarding the balance of receptor signaling during development and which brain circuits are shaped by experience-dependent stimulation. For example, AVPR1A has a different expression pattern and developmental time course compared to OXTR [34]. We suspect that the OXT peptide system in the OXTR KO is less well tuned to be activated by the same stimuli that activate those neurons in WT mice. That is an exciting empirical question for future study.

To guide future research, we hypothesize that OXTR [32] are involved in the sensory circuits engaged during received parental care and sensory stimulation to enhance postnatal Oxt expression. This study provides a foundation for future mechanistic studies to understand how OXTR activity and environmental factors control Oxt expression and translation efficiency to produce and release OXT. For example, we hypothesize that socially acquired OXT acting through OXTR on peripheral sensory neurons may play a role in the sensory-dependent development of the infant PVN (e.g. [35]). While the current study is underpowered to evaluate modest sex differences, exploratory analysis of the data hint at the possibility of a stronger developmental effect of Oxtr gene loss in male mice. This would be an important question for future studies that are adequately powered to resolve this question. The dynamic spatial and temporal expression pattern of OXTRs during development (reviewed in [25]), coupled with sensory-dependent OXT activity, may aid in the maturation of specific brain circuits at developmental milestones. The largest effect sizes we observed of Oxtr gene loss on PVN Oxt expression were in P8 and P14 mice. This period of time coincides with peak OXTR ligand binding in the developing brain in areas that show robust expression during development and minimal expression in adulthood, such as layers II/III of the neocortex [24, 25]. Importantly, this peak in OXTR expression and sensitivity of Oxt expression to genetic manipulation or environmental enrichment also coincides with an important sensitive period for infant attachment by classical conditioning [36] and coincides with observed effects of Oxt and Oxtr gene manipulation on infant ultrasonic vocalizations [19]. Perhaps the OXT/OXTR system is poised for peak experience-dependent plasticity during this phase of social neurodevelopment and may be part of the molecular regulation of the sensitive period for infant attachment.