Modification of the synaptic cleft under excitatory conditions.

The synaptic cleft is the extracellular part of the synapse, bridging the pre- and postsynaptic membranes. The geometry and molecular organization of the cleft is gaining increased attention as an important determinant of synaptic efficacy. The present study by electron microscopy focuses on short-term morphological changes at the synaptic cleft under excitatory conditions. Depolarization of cultured hippocampal neurons with high K+ results in an increased frequency of synaptic profiles with clefts widened at the periphery (open clefts), typically exhibiting patches of membranes lined by postsynaptic density, but lacking associated presynaptic membranes (18.0% open clefts in high K+ compared to 1.8% in controls). Similarly, higher frequencies of open clefts were observed in adult brain upon a delay of perfusion fixation to promote excitatory/ischemic conditions. Inhibition of basal activity in cultured neurons through the application of TTX results in the disappearance of open clefts whereas application of NMDA increases their frequency (19.0% in NMDA vs. 5.3% in control and 2.6% in APV). Depletion of extracellular Ca2+ with EGTA also promotes an increase in the frequency of open clefts (16.6% in EGTA vs. 4.0% in controls), comparable to that by depolarization or NMDA, implicating dissociation of Ca2+-dependent trans-synaptic bridges. Dissociation of transsynaptic bridges under excitatory conditions may allow perisynaptic mobile elements, such as AMPA receptors to enter the cleft. In addition, peripheral opening of the cleft would facilitate neurotransmitter clearance and thus may have a homeostatic and/or protective function.

Optimization of protocols for pre-embedding immunogold electron microscopy of neurons in cell cultures and brains.

Immunogold labeling allows localization of proteins at the electron microscopy (EM) level of resolution, and quantification of signals. The present paper summarizes methodological issues and experiences gained from studies on the distribution of synaptic and other neuron-specific proteins in cell cultures and brain tissues via a pre-embedding method. An optimal protocol includes careful determination of a fixation condition for any particular antibody, a well-planned tissue processing procedure, and a strict evaluation of the credibility of the labeling. Here, tips and caveats on different steps of the sample preparation protocol are illustrated with examples. A good starting condition for EM-compatible fixation and permeabilization is 4% paraformaldehyde in PBS for 30 min at room temperature, followed by 30 min incubation with 0.1% saponin. An optimal condition can then be readjusted for each particular antibody. Each lot of the secondary antibody (conjugated with a 1.4 nm small gold particle) needs to be evaluated against known standards for labeling efficiency. Silver enhancement is required to make the small gold visible, and quality of the silver-enhanced signals can be affected by subsequent steps of osmium tetroxide treatment, uranyl acetate en bloc staining, and by detergent or ethanol used to clean the diamond knife for cutting thin sections. Most importantly, verification of signals requires understanding of the protein of interest in order to validate for correct localization of antibodies at expected epitopes on particular organelles, and quantification of signals needs to take into consideration the penetration gradient of reagents and clumping of secondary antibodies.

Stimulation-induced differential redistributions of clathrin and clathrin-coated vesicles in axons compared to soma/dendrites.

Clathrin-mediated endocytosis plays an important role in the recycling of synaptic vesicle in presynaptic terminals, and in the recycling of transmitter receptors in neuronal soma/dendrites. The present study uses electron microscopy (EM) and immunogold EM to document the different categories of clathrin-coated vesicles (CCV) and pits (CCP) in axons compared to soma/dendrites, and the depolarization-induced redistribution of clathrin in these two polarized compartments of the neuron. The size of CCVs in presynaptic terminals (~ 40 nm; similar to the size of synaptic vesicles) is considerably smaller than the size of CCVs in soma/dendrites (~ 90 nm). Furthermore, neuronal stimulation induces an increase in the number of CCV/CCP in presynaptic terminals, but a decrease in soma/dendrites. Immunogold labeling of clathrin revealed that in presynaptic terminals under resting conditions, the majority of clathrin molecules are unassembled and concentrated outside of synaptic vesicle clusters. Upon depolarization with high K+, label for clathrin became scattered among de-clustered synaptic vesicles and moved closer to the presynaptic active zone. In contrast to axons, clathrin-labeled CCVs and CCPs were prominent in soma/dendrites under resting conditions, and became inconspicuous upon depolarization with high K+. Thus, EM examination suggests that the regulation and mechanism of clathrin-mediated endocytosis differ between axon and dendrite, and that clathrin redistributes differently in these two neuronal compartments upon depolarization.

Activity-dependent redistribution of CaMKII in the postsynaptic compartment of hippocampal neurons.

Calcium/calmodulin-dependent protein kinase II (CaMKII), an abundant protein in neurons, is involved in synaptic plasticity and learning. CaMKII associates with multiple proteins located at or near the postsynaptic density (PSD), and CaMKII is known to translocate from cytoplasm to PSD under excitatory conditions. The present study examined the laminar distribution of CaMKII at the PSD by immunogold labeling in dissociated hippocampal cultures under low calcium (EGTA or APV), control, and stimulated (depolarization with high K+ or NMDA) conditions. The patterns of CaMKII distribution are classified with particular reference to the two layers of the PSD: (1) the PSD core, a layer within ~ 30-40 nm to the postsynaptic membrane, and (2) the PSD pallium, a deeper layer beyond the PSD core, ~ 100-120 nm from the postsynaptic membrane. Under low calcium conditions, a subpopulation (40%) of synapses stood out with no CaMKII labeling at the PSD, indicating that localization of CaMKII at the PSD is sensitive to calcium levels. Under control conditions, the majority (~ 60-70%) of synapses had label for CaMKII dispersed evenly in the spine, including the PSD and the nearby cytoplasm. Upon stimulation, the majority (60-75%) of synapses had label for CaMKII concentrated at the PSD, delineating the PSD pallium from the cytoplasm. Median distance of label for CaMKII to postsynaptic membrane was higher in low calcium samples (68-77 nm), than in control (59-63 nm) and stimulated samples (49-53 nm). Thus, upon stimulation, not only more CaMKII translocated to the PSD, but they also were closer to the postsynaptic membrane. Additionally, there were two relatively infrequent labeling patterns that may represent intermediate stages of CaMKII distribution between basal and stimulated conditions: (1) one type showed label preferentially localized near the PSD core where CaMKII may be binding to NR2B, an NMDA receptor concentrated at the PSD core, and (2) the second type showed label preferentially in the PSD pallium, where CaMKII may be binding to Shank, a PSD scaffold protein located in the PSD pallium. Both of these distribution patterns may portray the initial stages of CaMKII translocation upon synaptic activation. In addition to binding to PSD proteins, the concentrated CaMKII labeling at the PSD under heightened excitatory conditions could also be formed by self-clustering of CaMKII molecules recruited to the PSD. Most importantly, these accumulated CaMKII molecules do not extend beyond the border of the PSD pallium, and are likely held in the pallium by binding to Shank under these conditions.

Immunogold labeling of synaptic vesicle proteins in developing hippocampal neurons.

Synaptic vesicles (SV) contain high concentrations of specific proteins. How these proteins are transported from soma to synapses, and how they become concentrated at SV clusters at presynaptic terminals were examined by immunogold electron microscopy in dissociated rat hippocampal neurons at 3-6 days in culture, a developmental stage when axonal transport of SV proteins is robust. In neuronal somas, labels for the SV integral membrane proteins (synaptophysin, SV2, VAMP/synaptobrevin, and synaptotagmin) were localized at Golgi complexes and other membranous structures that were dispersed in the cytoplasm as individual vesicle/vacuoles. These vesicles/vacuoles became aggregated in axons, with the size of aggregates ranging from 0.2 to 2 µm in length. Pleomorphic vesicle/vacuoles within the aggregate were typically larger (50-300 nm) than SVs, which were uniform in size at ~ 40 nm. These pleomorphic vesicles/vacuoles are probably transport cargos carrying SV integral membrane proteins from the soma, and then are preferentially sorted into axons at early developmental stages. Serial thin sections of young axons indicated that many labeled aggregates were not synaptic, and in fact, some of these axons were without dendritic contacts. In contrast, labels for two SV-associated proteins, synapsin I and α-synuclein, were not localized at the Golgi complexes or associated with membranous structures in the soma, but were dispersed in the cytoplasm. However, these SV-associated proteins became highly concentrated on clusters of SV-like vesicles in axons, and such clusters were already distinctive in axons as early as 3 days in culture. These clusters consisted of ~ 4-30 vesicles in single thin sections, and the vesicles were of a uniform size (~ 40 nm). Serial sectioning analysis showed that these clusters could be part of nascent synapses or exist in axons without any dendritic contact. Importantly, the vesicles were intensely labeled for SV integral membrane proteins as well as SV-associated proteins. Thus, these EM observations reveal that the two groups of proteins, SV integral membrane and SV-associated, proceed through different routes of biosynthesis and axon transport, and are only sorted into the same final compartment, SV clusters, when they are in the axons.

Postsynaptic densities fragment into subcomplexes upon sonication.

Postsynaptic density (PSD) fractions were isolated from rat forebrain and sonicated. Pellets from sonicated samples examined by electron microscopy revealed particles with an electron density similar to PSDs that appeared to be fragments of PSDs. Immuno-gold labeling confirmed that some of these contained PSD-95 and/or SynGAP. Biochemical analysis of supernatant and pellet fractions from sonicated samples showed almost complete recovery of several major PSD components (SynGAP, PSD-95, Shank3, Homer and Glutamate receptors) in the pellet, while the supernatant contained known contaminants of PSD fractions, such as glial acidic fibrillary protein and neurofilament protein, as well as actin and α-actinin, indicating susceptibility of these cytoskeletal elements to mechanical disruption. Size distributions of particulate material in control and sonicated samples were clearly different, with particles in the 40-90 nm range observed only in sonicated samples. Fragmentation of the PSD into subcomplexes containing major constituents suggests a patchwork structure consisting of weakly bound modules, that can be readily dissociated from each other through mechanical disruption. Modular organization and weak association between modules would endow the PSD with lateral structural flexibility.

Stimulation induces gradual increases in the thickness and curvature of postsynaptic density of hippocampal CA1 neurons in slice cultures.

Activity can induce structural changes in glutamatergic excitatory synapses, including increase in thickness and curvature of the postsynaptic density (PSD); these structural changes can only be documented by electron microscopy. Here in organotypic hippocampal slice cultures where experimental conditions can be easily manipulated, increases in thickness and curvature of PSDs were noticeable within 30 s of stimulation and progressed with time up to 3 min. These structural changes were reversible upon returning the samples to control medium for 5-10 min. Thus, the postsynaptic density is a very dynamic structure that undergoes rapid reorganization of its components upon stimulation, and recovery upon cessation of stimulation. The gradual increase in thickness of PSD could result from a gradual translocation of some PSD proteins to the PSD, and the increase in curvature of the PSD is likely led by postsynaptic elements.

FAM81A protein, a novel component of the postsynaptic density in adult brain.

Analysis of affinity-purified PSD-95 complexes had previously identified a 'hypothetical protein', product of the gene FAM81A [1]. The present study examined the tissue and subcellular distribution of FAM81A protein and its expression levels during development. Comparison of different organs indicates selective expression of FAM81A protein in brain. FAM81A is expressed late in development, with a post-natal gradual increase in brain levels that parallels the expression of PSD-95. Comparison of subcellular fractions from adult brain shows that the distribution of FAM81A protein is similar to that of PSD-95, with a drastic enrichment in the postsynaptic density fraction. Immuno-electron microscopy of adult brain tissue reveals specific immunogold labeling for FAM81A protein at postsynaptic densities in the forebrain. The label for FAM81A protein is concentrated at the cytoplasmic edge of the electron-dense core of the postsynaptic density, with a mean distance of ∼33 nm from the postsynaptic membrane. These observations firmly establish FAM81A protein as a component of the postsynaptic density in the adult brain, suggesting a role in synaptic function.

Neural precursor cells form integrated brain-like tissue when implanted into rat cerebrospinal fluid.

There is tremendous interest in transplanting neural precursor cells for brain tissue regeneration. However, it remains unclear whether a vascularized and integrated complex neural tissue can be generated within the brain through transplantation of cells. Here, we report that early stage neural precursor cells recapitulate their seminal properties and develop into large brain-like tissue when implanted into the rat brain ventricle. Whereas the implanted cells predominantly differentiated into glutamatergic neurons and astrocytes, the host brain supplied the intact vasculature, oligodendrocytes, GABAergic interneurons, and microglia that seamlessly integrated into the new tissue. Furthermore, local and long-range axonal connections formed mature synapses between the host brain and the graft. Implantation of precursor cells into the CSF-filled cavity also led to a formation of brain-like tissue that integrated into the host cortex. These results may constitute the basis of future brain tissue replacement strategies.

Distribution of densin in neurons.

Densin is a scaffold protein known to associate with key elements of neuronal signaling. The present study examines the distribution of densin at the ultrastructural level in order to reveal potential sites that can support specific interactions of densin. Immunogold electron microscopy on hippocampal cultures shows intense labeling for densin at postsynaptic densities (PSDs), but also some labeling at extrasynaptic plasma membranes of soma and dendrites and endoplasmic reticulum. At the PSD, the median distance of label from the postsynaptic membrane was ~27 nm, with the majority of label (90%) confined within 40 nm from the postsynaptic membrane, indicating predominant localization of densin at the PSD core. Depolarization (90 mM K+ for 2 min) promoted a slight shift of densin label within the PSD complex resulting in 77% of label remaining within 40 nm from the postsynaptic membrane. Densin molecules firmly embedded within the PSD may target a minor pool of CaMKII to substrates at the PSD core.

Stimulation-induced structural changes at the nucleus, endoplasmic reticulum and mitochondria of hippocampal neurons.

Neurons exhibit stimulation-induced ultrastructural changes such as increase of thickness and curvature of the postsynaptic density, decrease in contact area between subsurface cistern and plasma membrane, and formation of CaMKII clusters and synaptic spinules. These structural characteristics help in identifying the activity state of the neuron and should be taken into consideration when interpreting ultrastructural features of the neurons. Here in organotypic hippocampal slice cultures where experimental conditions can be easily manipulated, two additional features are documented in forebrain neurons as reliable benchmarks for stimulation-induced structural changes: (1) The neuronal nucleus showed conspicuous clustering of dark chromatin, and (2) the endoplasmic reticulum formed stacks with a uniform gap of ~ 13 nm filled with dark materials. Both structural changes progressed with time and were reversible upon returning the slice cultures to control medium. These stimulation-induced structural changes were also verified in dissociated hippocampal neuronal cultures and perfusion-fixed brains. In hippocampal slice cultures, the neuronal chromatin clustering was detectable within 30 s of depolarization with high K+ (90 mM) or treatment with NMDA (50 µM). In contrast, the formation of ER cisternal stacks did not become apparent for another 30 s. Importantly, in dissociated neuronal cultures, when the extracellular calcium was chelated by EGTA, treatment with high K+ no longer induced these changes. These results indicate that the stimulation-induced chromatin clustering and formation of ER stacks in neurons are calcium-dependent. Additionally, mitochondria in neuronal somas of tissue culture samples consistently became swollen upon stimulation. However, swollen mitochondria were also present in some neurons of control samples, but could be eliminated by blocking basal activity or calcium influx. This calcium-dependent structural change of mitochondria is specific to neurons. These structural changes may bring insights to the neuron's response to intracellular calcium rise upon stimulation.

Activity-dependent decrease in contact areas between subsurface cisterns and plasma membrane of hippocampal neurons.

Subsurface cistern (SSC) in neuronal soma and primary dendrites is a specialized compartment of endoplasmic reticulum (ER) that is in close apposition (10 nm) with the plasma membrane (PM). ER-PM contact areas are thought to be involved in intracellular calcium regulation. Here, structural changes of SSC in hippocampal neurons were examined by electron microscopy upon depolarization with high K+ (90 mM) or application of NMDA (50 µM) in rat dissociated cultures as well as organotypic slice cultures. The number and average length of SSC-PM contact areas in neuronal somas significantly decreased within 30 s under excitatory condition. This decrease in SSC-PM contact area progressed with time and was reversible. These results demonstrate a structural decoupling between the SSC and the PM upon stimulation, suggesting that there may be a functional decoupling of the calcium regulation. Because SSC-PM contact areas may mediate calcium influx, the decrease in contact area may protect neurons from calcium overload upon heightened stimulation.

IRSp53 accumulates at the postsynaptic density under excitatory conditions.

IRSp53 (BAIAP2) is an abundant protein at the postsynaptic density (PSD) that binds to major PSD scaffolds, PSD-95 and Shanks, as well as to F-actin. The distribution of IRSp53 at the PSD in cultured hippocampal neurons was examined under basal and excitatory conditions by immuno-electron microscopy. Under basal conditions, label for IRSp53 is concentrated at the PSD. Upon depolarization by application of a medium containing 90 mM K+, the intensity of IRSp53 label at the PSD increased by 36±7%. Application of NMDA (50 µM) yielded 53±1% increase in the intensity of IRSp53 label at the PSD compared to controls treated with APV, an NMDA antagonist. The accumulation of IRSp53 label upon application of high K+ or NMDA was prominent at the deeper region of the PSD (the PSD pallium, lying 40-120 nm from the postsynaptic plasma membrane). IRSp53 molecules that accumulate at the distal region of the PSD pallium under excitatory conditions are too far from the plasma membrane to fulfill the generally recognized role of the protein as an effector of membrane-bound small GTPases. Instead, these IRSp53 molecules may have a structural role organizing the Shank scaffold and/or linking the PSD to the actin cytoskeleton.

A novel synaptic junction preparation for the identification and characterization of cleft proteins.

Identification of synaptic cleft components has been hampered by the lack of a suitable preparation enriched in synaptic junctions devoid of adjoining peripheral membranes. Prior strategies for the isolation of synaptic junctions, relying on detergents for the removal of peripheral membranes, resulted in substantial loss of membranes lining the cleft. Here, a novel, detergent-free method is described for the preparation of a synaptic junction (SJ) fraction, using phospholipase A2. Limited digestion of synaptic plasma membrane (SPM) fraction with phospholipase A2 followed by centrifugation over a sucrose cushion results in selective removal of membranes peripheral to the cleft while junctional membranes remain relatively intact as observed by electron microscopy. Enrichment in synaptic junctional structures and loss of membranes peripheral to the junctional area are further verified by demonstrating enrichment in PSD-95 and loss in mGluR5, respectively. The SJ fraction is enriched in neuroligins and neurexins, in agreement with immuno-electron microscopy data showing their selective localization to the junctional area. Among additional cell adhesion molecules tested, N-cadherin and specific isoforms of the SynCAM and SALM families also show marked enrichment in the SJ fraction, suggesting preferential localization at the synaptic cleft while others show little enrichment or decrease, suggesting that they are not restricted to or concentrated at the synaptic cleft. Treatment of the SJ fraction with glycosidases results in electrophoretic mobility shifts of all cell adhesion molecules tested, indicating glycosylation at the synaptic cleft. Biochemical and ultrastructural data presented indicate that the novel synaptic junction preparation can be used as a predictive tool for the identification and characterization of the components of the synaptic cleft.

The Postsynaptic Density: There Is More than Meets the Eye.

The postsynaptic density (PSD), apparent in electron micrographs as a dense lamina just beneath the postsynaptic membrane, includes a deeper layer, the "pallium", containing a scaffold of Shank and Homer proteins. Though poorly defined in traditionally prepared thin-section electron micrographs, the pallium becomes denser and more conspicuous during intense synaptic activity, due to the reversible addition of CaMKII and other proteins. In this Perspective article, we review the significance of CaMKII-mediated recruitment of proteins to the pallium with respect to both the trafficking of receptors and the remodeling of spine shape that follow synaptic stimulation. We suggest that the level and duration of CaMKII translocation and activation in the pallium will shape activity-induced changes in the spine.

CaMKII-mediated displacement of AIDA-1 out of the postsynaptic density core.

Ankyrin repeat and sterile alpha motif domain-containing protein 1B (ANKS1B, also known as AIDA-1) is a major component of the postsynaptic density (PSD) in excitatory neurons where it concentrates at the electron-dense core under basal conditions and moves out during activity. This study investigates the molecular mechanism underlying activity-induced displacement of AIDA-1. Experiments with PSD fractions from brain indicate phosphorylation of AIDA-1 upon activation of endogenous CaMKII. Immuno-electron microscopy studies show that treatment of hippocampal neurons with NMDA results in an ~ 30 nm shift in the median distance of the AIDA-1 label from the postsynaptic membrane, an effect that is blocked by the CaMKII inhibitor tatCN21. CaMKII-mediated redistribution of AIDA-1 is similar to that observed for SynGAP. CaMKII-mediated removal of two abundant PSD-95-binding proteins from the PSD core during activity is expected to initiate a molecular reorganization at the PSD.

Zinc Stabilizes Shank3 at the Postsynaptic Density of Hippocampal Synapses.

Shank3 is a postsynaptic density (PSD) scaffold protein of the Shank family. Here we use pre-embedding immunogold electron microscopy to investigate factors influencing the distribution of Shank3 at the PSD. In dissociated rat hippocampal cultures under basal conditions, label for Shank3 was concentrated in a broad layer of the PSD, ~20-80 nm from the postsynaptic membrane. Upon depolarization with high K+ (90 mM, 2 min), or application of NMDA (50 µM, 2 min), both the labeling intensity at the PSD and the median distance of label from the postsynaptic membrane increased significantly, indicating that Shank3 molecules are preferentially recruited to the distal layer of the PSD. Incubation in medium supplemented with zinc (50 µM ZnCl2, 1 hr) also significantly increased labeling intensity for Shank3 at the PSD, but this addition of Shank3 was not preferential to the distal layer. When cells were incubated with zinc and then treated with NMDA, labeling intensity of Shank3 became higher than with either treatment alone and manifested a preference for the distal layer of the PSD. Without zinc supplementation, NMDA-induced accumulation of Shank3 at the PSD was transient, reversing within 30 min after return to control medium. However, when zinc was included in culture media throughout the experiment, the NMDA-induced accumulation of Shank3 was largely retained, including Shank3 molecules recruited to the distal layer of the PSD. These results demonstrate that activity induces accumulation of Shank3 at the PSD and that zinc stabilizes PSD-associated Shank3, possibly through strengthening of Shank-Shank association.

Depolarization of Hippocampal Neurons Induces Formation of Nonsynaptic NMDA Receptor Islands Resembling Nascent Postsynaptic Densities.

Depolarization of neurons in 3-week-old rat hippocampal cultures promotes a rapid increase in the density of surface NMDA receptors (NRs), accompanied by transient formation of nonsynaptic NMDA receptor clusters or NR islands. Islands exhibit cytoplasmic dense material resembling that at postsynaptic densities (PSDs), and contain typical PSD components, including MAGUKS (membrane-associated guanylate kinases), GKAP, Shank, Homer, and CaMKII detected by pre-embedding immunogold electron microscopy. In contrast to mature PSDs, islands contain more NMDA than AMPA receptors, and more SAP102 than PSD-95, features that are shared with nascent PSDs in developing synapses. Islands do not appear to be exocytosed or endocytosed directly as preformed packages because neurons lacked intracellular vacuoles containing island-like structures. Islands form and disassemble upon depolarization of neurons on a time scale of 2-3 min, perhaps representing an initial stage in synaptogenesis.

AIDA-1 Moves out of the Postsynaptic Density Core under Excitatory Conditions.

AIDA-1 is highly enriched in postsynaptic density (PSD) fractions and is considered a major component of the PSD complex. In the present study, immunogold electron microscopy was applied to determine localization as well as the activity-induced redistribution of AIDA-1 at the PSD using two antibodies that recognize two different epitopes. In cultured rat hippocampal neurons under basal conditions, immunogold label for AIDA-1 is mostly located within the dense core of the PSD, with a median distance of ~30 nm from the postsynaptic membrane. Under excitatory conditions, such as depolarization with high K+ (90 mM, 2 min) or application of NMDA (50 µM, 2 min), AIDA-1 label density at the PSD core is reduced to 40% of controls and the median distance of label from the postsynaptic membrane increases to ~55 nm. The effect of excitatory conditions on the postsynaptic distribution of AIDA-1 is reversed within 30 minutes after returning to control conditions. The reversible removal of AIDA-1 from the PSD core under excitatory conditions is similar to the redistribution of another abundant PSD protein, SynGAP. Both SynGAP-alpha1 and AIDA-1 are known to bind PSD-95. Activity-induced transient translocation of these abundant proteins from the PSD core could promote structural flexibility, vacate sites on PSD-95 for the insertion of other components and thus may create a window for synaptic modification.

A negative feedback loop controls NMDA receptor function in cortical interneurons via neuregulin 2/ErbB4 signalling.

The neuregulin receptor ErbB4 is an important modulator of GABAergic interneurons and neural network synchronization. However, little is known about the endogenous ligands that engage ErbB4, the neural processes that activate them or their direct downstream targets. Here we demonstrate, in cultured neurons and in acute slices, that the NMDA receptor is both effector and target of neuregulin 2 (NRG2)/ErbB4 signalling in cortical interneurons. Interneurons co-express ErbB4 and NRG2, and pro-NRG2 accumulates on cell bodies atop subsurface cisternae. NMDA receptor activation rapidly triggers shedding of the signalling-competent NRG2 extracellular domain. In turn, NRG2 promotes ErbB4 association with GluN2B-containing NMDA receptors, followed by rapid internalization of surface receptors and potent downregulation of NMDA but not AMPA receptor currents. These effects occur selectively in ErbB4-positive interneurons and not in ErbB4-negative pyramidal neurons. Our findings reveal an intimate reciprocal relationship between ErbB4 and NMDA receptors with possible implications for the modulation of cortical microcircuits associated with cognitive deficits in psychiatric disorders.