Background
The maturation of neural circuitry across postnatal development is a dynamic and multifaceted process, which is vital for healthy adult brain function and cognition. A well-characterized and important aspect of this developmental program is the change in the subunit composition of the
n-methyl-
d-aspartate (NMDA) receptor. The heteromeric NMDA receptor is composed of two GluN1 subunits and a combination of two other subunits, which in the cortex and hippocampus are predominantly either two GluN2A subunits, two GluN2B subunits, or one of each [
1,
2]. The subunit composition of NMDA receptors influences their channel properties, such as calcium permeability [
3‐
5] and open probability [
6], and hence impacts their role in synaptic plasticity (for review, see [
7]). Across postnatal development in rodents and humans, the relative expression of the GluN2B subunit of the NMDA receptor decreases, while the expression of GluN2A increases [
2,
8‐
14]. As a result, the NMDA receptor becomes less sensitive to blockade by GluN2B antagonists ifenprodil, CP101,606, and Ro25-6981 [
9,
15‐
18]. These changes are accompanied by concurrent changes in electrophysiological properties, such as increasing excitatory postsynaptic current (EPSC) amplitude and decreasing decay time [
15,
17], and reflect the healthy, activity-dependent maturation of the brain [
19‐
21].
NMDA receptor signaling dysregulation has been implicated in schizophrenia, autism spectrum disorder (ASD), epilepsy, and intellectual disability, which are considered disorders of brain development [
22‐
26]. NMDA receptor antagonists ketamine and PCP can induce psychosis symptoms in healthy individuals and worsen such symptoms in individuals with schizophrenia [
27‐
29]. Altered NMDA receptor signaling in the prefrontal cortex of individuals with schizophrenia has been identified using a paradigm in which intracellular signaling downstream to receptor activation was monitored [
30,
31]. In addition, altered expression of NMDA receptor subunits was shown in the whole brain homogenates [
32] and in the postsynaptic density (PSD) [
33], a cellular micro-domain which is a hub for postsynaptic signaling events. Genetic variants in the GluN2B gene, GRIN2B, have been implicated in sporadic ASD [
34‐
38], while mutations of GRIN2B and GRIN2A (the GluN2A gene) have been associated with epilepsy and intellectual disability [
39‐
42]. NMDA receptor knockout mice exhibit a range of neurological and behavioral deficits relevant to both schizophrenia and ASD [
43‐
47]. Given that NMDA receptor abnormalities are implicated in schizophrenia and ASD, it is plausible that developmental disruption of the GluN2B-GluN2A switch may play a role in the pathogenesis of these disorders.
If the developmental GluN2B-GluN2A switch does play a role in the emergence of schizophrenia and ASD, then this process may also be impacted by risk factors for these illnesses. Two key factors which influence risk for schizophrenia and ASD are sex [
48‐
51] and genetic vulnerability [
52‐
54]. Evidence that sex modifies risk for schizophrenia and ASD comes from epidemiological studies which reveal that both disorders are more common in males, particularly ASD [
55‐
57]. On average, schizophrenia is diagnosed earlier in males than in females [
48‐
51], around adolescence and young adulthood when sexual dimorphism in the brain increases [
58]. Males also experience greater severity of some schizophrenia symptoms [
59,
60] even prior to conversion to psychosis [
61]. In ASD, some evidence supports the theory that the brain is excessively masculinized [
62,
63]. Genetic vulnerability also impacts risk for highly heritable neurodevelopmental disorders such as schizophrenia [
52‐
54] and ASD [
64,
65], likely via the combined influence of multiple risk variants. Among the genes which may increase the susceptibility to schizophrenia and ASD is DTNBP1 (dysbindin). Dysbindin is developmentally regulated [
66] and contains single nucleotide polymorphisms (SNPs) identified as possible schizophrenia risk variants in genetic association studies prior to the GWAS era [
67‐
70]. Some of these DTNBP1 SNPs have been associated with more severe psychotic symptoms [
71]. They may also contribute to differences in structural brain development and cognitive ability in the general population [
72‐
74]. DTNBP1 is also contained within a region on chromosome 6 which has been linked to ASD [
75,
76] and is regulated by MeCP2 [
77], the gene whose mutation causes Rett syndrome [
78]. DTNBP1 mRNA and protein expression are decreased in the hippocampus and dorsolateral prefrontal cortex of individuals with schizophrenia [
79,
80], while the DTNBP1 promoter is hypermethylated in the saliva and brain in individuals with schizophrenia [
81,
82]. DTNBP1 null mutant mice, which do not express dysbindin protein, display cellular and functional abnormalities in the brain of relevance to schizophrenia, such as decreased GRIN1 mRNA expression [
83], increased surface GluN2A protein expression [
84], altered prepulse inhibition [
85,
86], decreased hippocampal long-term potentiation [
84,
87], and impaired working memory [
83,
85]. It is not known whether sex and DTNBP1 mutation impact the developmental GluN2B-GluN2A switch in the brain, thereby increasing risk for schizophrenia and/or ASD.
Therefore, in this study, we investigated the effects of sex and dysbindin null mutation on the GluN2B-GluN2A switch in the frontal cortex and hippocampus. Using female and male wild-type (WT), heterozygous DTNBP1 null mutant [DTNBP1(+/−)] and homozygous null mutant [DTNBP1(−/−)] mice aged 7, 14, 28, and 56 days, we focused on events at the synapse using a biochemical approach to enrich for PSD proteins. This PSD enrichment enabled targeted quantification of proteins at the synapse, where NMDA receptor signaling plays a key role in excitatory neurotransmission and cognition [
88]. We hypothesized that sex and/or DTNBP1 null mutation would disrupt the normal patterns of synaptic expression of GluN2B and GluN2A subunits and associated proteins across postnatal development.
Discussion
In this study, we provide molecular evidence that the developmental shift in the balance of GluN2B and GluN2A in the cortex differs in females and males and in the hippocampus, can be disrupted by DTNBP1 null mutation. In males compared to females, the balance of GluN2 subunits was shifted toward a greater abundance of GluN2B-containing NMDA receptors at the cortical synapse, accompanied by underlying increases in Y1472-GluN2B phosphorylation, Fyn tyrosine kinase abundance, and PLCγ abundance. In the hippocampus, the developmental trajectory of GluN2B maturation was disrupted in DTNBP1(−/−) mice compared to WT mice, alongside increases in the relative abundance of GluN2A. These changes occurred in the context of genotype effects on Y527-Src phosphorylation, Fyn abundance in females only, and PLCγ abundance. Overall, these findings suggest that sex and DTNBP1 genotype can influence GluN2 subunit balance, in a process possibly involving Fyn and PLCγ. They also support the hypothesis that differences in GluN2 subunit balance, arising developmentally during the GluN2B-GluN2A switch, may underlie some sex- or genotype-related individual differences in risk for neurodevelopmental disorders.
It is interesting to note that the molecules that showed sex differences at the cortical synapse may be related functionally and mechanistically. Males had more abundant representation of Fyn and PLCγ, and higher pY1472-GluN2B:total GluN2B, GluN2B:GluN2A, and GluN2B:GluN1 ratios, than females in PSD enrichments from the frontal cortex. In the pre-adolescent period, the increased synaptic abundance in males of the tyrosine kinase Fyn, which phosphorylates GluN2B at Y1472 [
100], may lead to male-specific increases in Y1472-GluN2B phosphorylation. This in turn may lead to increased stability of GluN2B-containing NMDA receptors at the synapse in males [
94,
97,
98], reflected in increased GluN2B:GluN2A and GluN2B:GluN1 ratios in the PSD. Consistent with this scenario, we observed that levels of Fyn in the PSD were positively correlated with pY1472-GluN2B:total GluN2B ratio within the P28 age group in the frontal cortex. However, at P28 when qualitative sex differences in Fyn were greatest, the sex differences in pY1472-GluN2B phosphorylation were most subtle. Furthermore, Fyn was negatively correlated with pY1472-GluN2B phosphorylation across the lifespan as a whole, decreasing with postnatal age while pY1472-GluN2B phosphorylation increased. This suggests that, while Fyn may fine-tune levels of GluN2B phosphorylation in pre-adolescence, other mechanisms may also contribute to sex differences and drive the developmental increase pY1472-GluN2B phosphorylation across postnatal life as a whole. Alternatively, increased pY1472-GluN2B phosphorylation in males may also arise from increased abundance or activity of PLCγ, which can facilitate GluN2B phosphorylation [
101], was increased in males at the ages when pY1472-GluN2B was also different in males, and was positively correlated with pY1472-GluN2B levels at P28. Sex differences in PLCγ may also indicate a compensatory change, aimed at driving mGluR5-mediated increases in GluN2A to “normalize” the increased relative abundance of GluN2B. The emergence of sex differences in the GluN2B:GluN2A ratio, pY1472-GluN2B:total GluN2B ratio, and PLCγ abundance in the PSD appeared to coincide with the juvenile (P14-P28) period, prior to the pubertal surge in sex hormones during adolescence. This suggests that the divergence of males and females may be due to organizational effects of sex hormones in males during the early life surge in testosterone or due to sex differences in autosomal gene/protein expression [
10,
104]. The emergence of sex differences during discrete developmental epochs reinforces the value of using developmental trajectories, rather than isolated developmental endpoints, to assess molecular parameters of relevance to brain function and behavior.
Further studies are required to understand the consequences of sex differences in GluN2B-GluN2A balance in the PSD of the frontal cortex. Typically, GluN2B-containing NMDA receptors have been viewed as most critical to healthy brain function early in postnatal development (coincident with their high relative abundance), during which time they are involved in synapse formation, stabilization, and plasticity [
105,
106]. Later in life, GluN2B-containing NMDA receptors have been considered key contributors to extra-synaptic NMDA receptor signaling, responsible for the damaging excitotoxicity which occurs in response to brain injury [
107] (for review, see [
108,
109]). In this study, however, relative GluN2B abundance and phosphorylation were increased in males during adolescence and/or adulthood in PSD enrichments (i.e., predominantly within the synapse), suggesting that any effects of these sex differences would be mediated via synaptic actions of GluN2B-containing NMDA receptor complexes later in life. It is possible that GluN2B-related sex differences may impact a substantial proportion of NMDA receptors in the adult cortex if tri-heteromeric GluN1/GluN2A/GluN2B-containing NMDA receptors are more common than previously thought, as has been indicated in the hippocampus [
110]. Indeed, GluN2B-containing NMDA receptors have recently been shown to play a critical role in layer 5 pyramidal neurons of the prefrontal cortex in adulthood, a role which is acquired during adolescence [
111]. Throughout the lifespan, pY1472 phosphorylation may also be important, as it is modulated by NMDA receptor activity [
112], mediates some of the detrimental effects of neonatal hypoxic brain injury [
113], and may play a role in pathological processes such as alcohol dependency [
114,
115]. Future work exploring whether observations in this study are driven by sex differences in specific cell types, such as pyramidal neurons or interneurons, may also shed light on the possible functional consequences of such differences and their relevance to ASD, schizophrenia, or other neurodevelopmental disorders.
Future work may also determine whether sex differences in GluN2B-GluN2A balance, as observed here in rodents, also exist in the human cortex, and if such sex differences relate to other sex differences in the brain structure, neural activity, cognitive function, and risk for psychiatric illness. The brains of males and females have been found to have differences in molecular [
104,
116,
117], structural [
118‐
128], and functional [
129‐
133] components. Females and males are also differentially susceptible to neurodevelopmental disorders such as ASD [
55,
56] and schizophrenia [
48,
49,
59], as well as other psychiatric illnesses such as major depressive disorder [
134,
135]. In experimental settings, male and female rodents are differentially sensitive to disturbance of brain circuits and behavior following postnatal ketamine, cocaine, morphine, and nicotine exposure [
136‐
139]. Interestingly, males have more severe behavioral deficits than females in a rodent model of NMDA receptor deficiency [
140] and after early postnatal NMDA receptor blockade [
141]. If the sex differences observed here extend to humans, it is plausible that pathophysiologic mechanisms in neurodevelopmental disorders which converge on GluN2A- or GluN2B-mediated NMDA receptor signaling may differentially impact males and females.
In this study, we identified effects of DTNBP1 mutation on the GluN2B-GluN2A shift, particularly on the developmental trajectory of GluN2B and the overall synaptic abundance of GluN2A in the hippocampus. Dysbindin 1 (protein product of the DTNBP1 gene) is a component of the BLOC-1 complex and has diverse functions in the brain including supporting dendritic spine formation and facilitating glutamate release (for review, see [
142]). In the whole cortex, dysbindin 1A expression has been reported to decrease across postnatal development, particularly up to P28 [
66]. In our study, the greatest disruption to GluN2B at the PSD occurred up to P28, after which time, the GluN2B:GluN1 ratio in DTNBP1(−/−) mice approached WT levels. We also observed increased overall GluN2A in DTNBP1(−/−) mice across postnatal life in the hippocampal PSD. However, we did not observe a main effect of genotype effects in the GluN2B:GluN2A ratio, potentially as a result of concurrent differences at some developmental ages in both GluN2B and GluN2A. Our findings with GluN2A mirror previous work in DTNBP1(−/−) mice [
84], in which GluN2A surface expression was increased in cultured hippocampal neurons from DTNBP1(−/−) mice (harvested at E18, cultured 16 days), hippocampal EPSCs were increased with faster decay time in pre-adolescent DTNBP1(−/−) mice (P26–31), and hippocampal long-term potentiation (LTP) was increased in adult DTNBP1(−/−) mice (P56) [
84]. However, a contradictory decrease in LTP in adult DTNBP1(−/−) mice at P35–50 and P90–120 has also been reported [
87]. Decreased amplitude of NMDA-evoked currents in cortical neurons at P45–60, accompanied by decreased GRIN1 gene expression and impaired working memory, has also been described [
83]. Although some of this previous work identified frontal cortex-related deficits, our work suggests that the hippocampus may be more sensitive than the frontal cortex to the effects of DTNBP1 mutation on the GluN2B-GluN2A switch, as assessed using our techniques. Our findings are consistent with work investigating another schizophrenia risk factor, neuregulin 1. Neuregulin 1 (Nrg1) is a candidate schizophrenia risk gene [
143] which has been found to increase GluN2B phosphorylation via PLCγ-dependent mechanism in cortical neurons [
101]. This process involves ErbB4-TrkB interaction, which is decreased in the brain in individuals with schizophrenia [
101]. In Nrg1 transmembrane heterozygous knockout mice, pY1472-GluN2B is reduced, in parallel with increased baseline and decreased evoked power of gamma frequency neural oscillations, reproducible features of schizophrenia [
144]. Future work may determine whether GluN2B phosphorylation, and/or other developmental events related to the GluN2B-GluN2A switch, represent points of convergence for putative risk genes for neurodevelopmental disorders. Overall, our findings highlight the potential for genetic variants in DTNBP1 and other developmentally regulated genes to influence maturational programs, such as the GluN2B-GluN2A switch, which are relevant to neurodevelopmental disorders.
Using biochemical fractionation and Western blotting, we describe a decrease in the ratio of GluN2B:GluN2A across the lifespan, from a peak of approximately 1.4:1 in neonates (P7) in the hippocampus. This decrease is consistent with, but more subtle than, calculations of subunit ratios based on EPSC decay times in CA1 of the hippocampus [
6]. This difference may arise because of differences in sensitivity between the two methods. It is also possible that GluN2B subunits may be present in a greater percentage of postnatal NMDA receptors than suggested by electrophysiological studies if tri-heterormeric GluN1/GluN2A/GluN2B receptors represent a substantial proportion of receptors in later postnatal life and have faster decay times than assumed during analysis of electrophysiological data [
6]. Future studies incorporating electrophysiological measurement of EPSC decay times and ifenprodil sensitivity in discrete brain subregions would be valuable to confirm and extend the findings of this study.
There were a number of limitations of the current study. Firstly, proteins from the PSD were enriched in our samples using a fractionation method based on two rounds of Triton X-100 resuspension/precipitation. This method did not involve use of a sucrose gradient, which is a well-characterized method to separate pre- and parasynaptic proteins from PSD proteins [
91,
92]. As such, although presynaptic proteins synaptophysin and Rab3 were scarce in our PSD enrichments (consistent with samples generated using sucrose gradient fractionation [
145]), it is likely that non-PSD proteins (particularly presynaptic proteins) were also present in our samples. Secondly, the Western blotting approach used in this study is versatile but is not a gold standard method for protein quantification. We confirmed using loading standard curves that proteins could be quantified effectively. Such standard curves demonstrated that the linear range of quantification for all proteins measured in this study extended sufficiently to allow accurate quantification of at least twofold increases and decreases in samples at the extremes of protein abundance across the lifespan. However, in future work, we will confirm the current findings using another proteomic method, such as mass spectrometry. Thirdly, because traditional loading controls (such as β-actin) are often developmentally regulated, normalization of data to control for loading amount is a challenge. Therefore, we investigated biologically relevant ratios of proteins of interest (such as GluN2B:GluN2A and pY1472-GluN2B:total GluN2B) which are inherently normalized and are also of functional relevance at the synapse. Fourthly, we did not include DTNBP1 heterozygous mice in analysis of NMDA receptor signaling in the hippocampus. Since DTNBP1 heterozygous mice may most closely mimic dysbindin deficits in human psychiatric illness, inclusion of this genotype group in future studies of the hippocampus may expand their relevance to human disease. We also did not detect dysbindin 1A or 1C in our hippocampus PSD enrichments using the Abcam ab133652 antibody. Both proteins have been reported in whole brain mouse PSD enrichments, albeit less abundantly than in synaptosomal membrane fractions [
146]. Other antibodies or protein quantification methods may be required to link the developmental trajectory of dysbindin to the emergence of genotype effects in DTNBP1 null mutant mice. Finally, endogenous levels of sex hormones were not measured or controlled for and thus, contributions of sex hormones to sex differences in post-pubertal animals were not examined in this study. Future work employing gonadectomy and hormone replacement may be valuable for determining the extent to which sex differences in NMDA receptor signaling are influenced by sex hormones.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
DS, CGH, and KBW designed the study; GC provided DTNBP1 mutant mice; DS, JC, and MM performed the experiments; DS and JC analyzed the data; and all authors contributed to writing the manuscript. All authors read and approved the final manuscript.