Background
Autism spectrum disorder (ASD) is a neurodevelopmental disorder affecting upwards of 1 in 100 children [
1‐
3]. ASD is characterized by impairments in social skills, sensory processing, and communication as well as stereotypic behaviors [
4]. Symptoms typically present in early childhood and effective therapies are lacking [
5,
6]. Causal mechanisms underlying changes in early brain development leading to ASD are not well defined. The genetic causes of ASD are exceptionally diverse, and hundreds of candidate loci have been identified [
7‐
11]. Many ASD-associated genes are expressed in both developing excitatory and inhibitory neurons and play known roles in synaptic transmission [
11]. Given the complex etiology of ASD, a deeper understanding of the role of these ASD-associated genes in early development is needed.
NMDA receptors (NMDARs) are glutamate-gated, calcium-permeable ion channels that mediate a slow component of glutamatergic transmission [
12]. They play key roles in early brain development [
13] and the plasticity underlying higher-order processes [
14,
15]. NMDARs are obligate heterotetramers, comprised of two GluN1 subunits and typically some combination of GluN2(A-D) subunits [
12]. The various GluN2 subunits impart distinct functional and cell biological properties to the receptor and play myriad roles in brain development. Accordingly, mutations in NMDAR subunits are associated with numerous neurodevelopmental disorders including ASD, epilepsy, intellectual disability, and schizophrenia [
16‐
18].
GRIN2B, which encodes the GluN2B subunit, is a high-confidence ASD gene [
8,
11,
19]. GluN2B is highly expressed in humans during early brain development and has critical functions in neurogenesis and circuit formation [
20‐
22]. The complex activities of GluN2B in early development remain incompletely defined, in part because murine knockouts of GluN2B do not survive past perinatal stages [
23]. As a highlight of this complexity, mutations causing both loss and gain of GluN2B function are associated with ASD as well as other neurodevelopmental disorders [
24,
25].
Here, we investigate the validity and robustness of the zebrafish model to study GluN2B in development. This system affords detailed analysis of the role of GluN2B in early brain development and the emergence of complex behaviors, offering numerous readouts of the developing nervous system. Additionally, zebrafish are highly social creatures [
26], facilitating the study of the role of GluN2B in social preference development. In zebrafish, the GluN2B subunit is encoded by two paralogous genes:
grin2Ba and
grin2Bb [
27]. We find that fish with frameshift mutations in both
grin2B paralogues, and thus lacking all GluN2B, survive into adulthood and are fertile. This offers a unique opportunity to study the developmental roles of GluN2B and characterize its requirements in brain development and the formation of various zebrafish behaviors. We find that zebrafish lacking GluN2B show social deficits, despite having no significant changes in whole-brain size, glutamatergic neuron density, or overall excitatory/inhibitory (
E/
I) balance. Nevertheless, we find significant reduction in a marker of inhibitory neurons in a small subset of regions, most notably the subpallium, an area containing subcortical structures whose dysfunction is associated with ASD phenotypes [
28,
29].
Discussion
GRIN2B is a high-confidence ASD-associated gene in ASD [
8,
11,
19]. How it contributes to clinical phenotypes, however, is poorly understood. In addition to well-studied functions for NMDARs in synaptic transmission at mature excitatory synapses, NMDARs also play key roles in development [
13]. NMDARs regulate neurogenesis [
20], cell migration [
59], and neuronal differentiation [
22]. Here, we establish zebrafish as an in vivo model to study GluN2B in development and dysfunction. Unlike mouse knockouts for
GRIN2B, which fail to suckle and die shortly after birth [
23], zebrafish lacking GluN2B (
grin2B−/− fish) survive into adulthood and are fertile (Fig.
2). Notably, many aspects of these
grin2B−/− fish appear normal including movement (Fig.
5A–C) and learning (Fig.
5E, F) which may be driven by the compensational expression of other NMDAR genes (Fig.
6). Nevertheless, one behavior—a juvenile social preference—is deficient in
grin2B−/− fish (Fig.
3). Social deficits are common in ASD patients, and the specificity of this behavioral phenotype in this model is a powerful tool to study
grin2B in brain dysfunction. Altered excitatory/inhibitory (
E/
I) balance, often caused by reductions in inhibitory neuron populations, is central to ASD pathogenesis [
58,
60]. We did not see dramatic changes in
E/
I balance when globally observing the brain in
grin2B−/− fish (Fig.
7). However, we observed a significant reduction in an inhibitory neuron marker in the subpallium, a region that will later mature into putative homologues of amygdaloid, septal, and striatal nuclei, all structures previously implicated in ASD [
28,
29,
61,
62] (Additional file
1: Table S5).
Due to an ancient genome duplication, zebrafish have two
grin2B paralogues (
grin2Ba and
grin2Bb) [
63]. Proteins generated from both paralogues share considerable sequence homology with each other and human GluN2B (hGluN2B), only diverging in sequence in the amino-terminal (ATD) and carboxy-terminal (CTD) domains (Fig.
1). The CTD is the region of greatest variance in sequence and length among NMDAR subunits of the same species and is involved in intracellular signaling [
15]. Despite this, zebrafish GluN2B proteins interact with various conserved intracellular components, including Ca
2+/calmodulin-dependent protein kinase (CAMKII) [
64] and ERK [
65]. As suggested by the conservation in sequence in ligand-binding, transmembrane, and pore-forming domains, zGluN2Bb formed functional heterotetramers with hGluN1 in a heterologous expression system (Fig.
1C). These cross-species NMDARs showed glutamate-gated inward current and had prolonged deactivation, a hallmark of GluN2B-containing NMDARs (Fig.
1E). These results demonstrate a high degree of functional similarity between human and zebrafish GluN2B proteins.
There are multiple lines of evidence supporting the claim that there is no residual GluN2B function in our
grin2B−/− zebrafish. Firstly, we generated our CRISPR-Cas9 mutations directly before or in the critical M3 pore-lining helix (Fig.
2A–C). Due to the importance of this highly conserved region, any alternative-splicing event that restored the reading frame of the protein would still render a non-functional channel. Additionally, no wild-type mRNA was detected for either
grin2Ba or
grin2Bb in our
grin2B−/− zebrafish (Fig.
2D). The observation of NMD and potential transcriptional adaptation also supports that the mutant mRNA is targeted for degradation (Fig.
6).
Zebrafish are social creatures. Non-specific block of all NMDARs with MK-801 decreases social preference in juveniles [
47,
66] and adults [
66,
67]. Our findings refine this pharmacological work by implicating GluN2B, rather than general NMDAR dysfunction, in the development of social preference (Fig.
3D). Additionally, acute pharmacological block precludes the understanding of any developmental roles of NMDARs. It is possible that MK-801-driven disruptions in movement control may also contribute to observed social deficits in these assays [
34,
47,
66]. Our findings suggest that developmental alterations in specific brain regions, rather than alterations in movement control (Fig.
5), are likely generating these phenotypes.
Disease-associated missense mutations in
GRIN2B are more likely to result in neurodevelopmental disorders, such as ASD, as compared to other subunits, highlighting the significant developmental requirements for GluN2B [
68]. Interestingly, although not as numerous as in GluN2B, rare variants associated with ASD are also found in GluN1, GluN2A, GluN2C, and GluN2D [
69]. Zebrafish lacking other NMDAR subunit paralogues (
grin1a, grin1b, grin2Aa/b, grin2Da, or grin2Db) did not result in a social deficit at 3 wpf. Only through testing
grin2C and
grin2D double mutants (as done here with
grin2A) can we definitively rule out the involvement of these subunits. However, given the lack of social behavioral phenotype in fish lacking the widely expressed GluN2A subunit, mutations in other subunits may confer some effects through association with GluN2B, possibly in triheteromeric receptors.
Behavior and cellular phenotypes of
fmr1 mutant fish have been previously studied [
70‐
72]. Interestingly, our findings on
fmr1 social deficits (Fig.
4A, B) differ from previous studies which report
fmr1 mutant zebrafish developing social preference earlier and more robustly than wild type [
73]. These studies use a different social behavior paradigm that may be more influenced by well-documented hyperactivity in
fmr1 mutant zebrafish [
73,
74].
Zebrafish lacking GluN2B displayed wild-type spontaneous swimming, visual motor response, photic startle responses and could capture paramecia, albeit at a slightly diminished capacity (Fig.
5). Further, these fish were capable of learning within the prey capture paradigm (Fig.
5F). One possibility is that synaptic plasticity mediated by GluN2B-containing NMDARs is not essential to learning the prey capture task. Alternatively, the action of other NMDARs, such as those containing GluN2A or GluN2D, can compensate for the lack of GluN2B. Indeed, in rodents, increase in levels of GluN1 and GluN2A was observed during memory consolidation, while GluN2B levels remained unaltered [
75].
NMD of mutant RNA can result in transcriptional adaptation [
55]. Compensatory increases in other NMDAR subunits were observed in our assays (Fig.
6) and may contribute to both the viability of the
grin2B−/− fish into adulthood, as well as their wild-type spontaneous locomotion, photic-evoked responses, and capacity to learn in prey capture assays. These findings, however, further highlight the specificity of GluN2B in the development of social behavior, as upregulation of other NMDAR subunits is not sufficient to rescue this phenotype.
Whole-brain scans of zebrafish larvae lacking GluN2B and containing transgenic lines labeling excitatory and inhibitory populations outlined no structural or whole-brain changes in neuronal populations. Similar whole-brain larval confocal scans have been done on valproic acid (VPA)-treated fish [
40]. VPA exposure in zebrafish also causes social deficits [
76,
77] and is associated with increased risk of ASD in humans [
78,
79]. Unsurprisingly, due to the broad disruption caused by global VPA treatment, gross structural and compositional changes were reported in many regions [
40]. Given the shared social phenotype of VPA-treated and
grin2B−/− fish, our model offers a more refined view of the potential regions involved in establishing this phenotype. In particular, we observed a significant reduction in the intensity of the marker labeling inhibitory neurons in the subpallium of 6 dpf larvae. Changes were apparent in a cell-rich anterior domain and in neuropil-rich regions surrounding the preoptic area.
Nuclei of the precommissural subpallium are outlined on a schematic coronal section of the telencephalon (Additional file
1: Fig. S8). At this stage, major regions of the forebrain are not yet mature; however, neurons derived from the subpallium later contribute to putative zebrafish homologues of the amygdala, septum, and striatum [
80‐
82] (Additional file
1: Table S5). While precise homologies of specific zebrafish nuclei remain unresolved, there are indisputably domains of the zebrafish ventral telencephalon (subpallium) that molecularly and functionally represent these structures. Our data suggest that changes in these regions may underlie the abnormal social behavior observed in 3–4 weeks old
grin2B−/− fish. Abnormalities in amygdalar, septal, and striatal development and function have been implicated in ASD and are thought to contribute to the social impairment that characterizes this disorder [
28,
29,
61,
62]. Given the autonomic dysfunction and sensory processing atypia associated with ASD, it is unsurprising that we also uncovered substantial differences in inhibitory neuron signal in additional anatomical areas related to these functions [
83‐
85]. These findings and their associations warrant further investigation.
We establish zebrafish as a clinically relevant model for investigating developmental roles of GluN2B. The viability, homology, and ability to assay social preference together facilitate the study of GluN2B in development and ASD etiology. Given the myriad roles of NMDAR in development, further study into the specific mechanisms by which GluN2B perturbation gives rise to the ASD phenotype are necessary. These should include effects on neurogenesis, connectivity, and synaptic alterations with a specific focus on the nuclei of the subpallium. In addition to knockout models, the advent of precise gene-editing in zebrafish [
86], coupled with the high homology of GluN2B in zebrafish, facilitates the future study of ASD-associated GluN2B missense mutations in zebrafish.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.