Profiling Excitatory and Inhibitory Neurotransmission with Multiplexed AOC Imaging

The Molecular Logic of Excitatory and Inhibitory Balance

Every computation the brain performs depends on the interplay between excitation and inhibition. Glutamatergic neurons drive excitatory transmission, depolarising their targets through ionotropic receptors at synapses distributed across cortical, hippocampal, and subcortical circuits. GABAergic interneurons counter this drive, providing the inhibitory tone that shapes signal timing, prevents runaway excitation, and enables the selective activation of neuronal ensembles. The ratio between these opposing forces, commonly referred to as excitatory–inhibitory (E/I) balance, is not a fixed set point but a dynamically regulated parameter that varies across brain regions, cell types, and developmental stages.

The concept that disrupted E/I balance contributes to neurological and psychiatric disease has become a unifying framework in neuroscience. Rubenstein and Merzenich first proposed that an elevated excitation-to-inhibition ratio in cortical circuits could account for core features of autism spectrum disorder¹. Since then, the E/I imbalance hypothesis has been extended to epilepsy, schizophrenia, Alzheimer's disease, and other conditions². Recent work combining imaging, transcriptomics, and genetics has confirmed that glutamate and GABA gene sets contribute to both symptom severity and cortical structure in autism³.

Yet testing E/I balance at the molecular level — determining which excitatory and inhibitory receptors are expressed in which cells, in what ratios, and how these ratios change in disease — requires spatial resolution that bulk assays cannot provide. The Allen Institute's whole-brain cell atlas, which classified over five thousand neuronal clusters using single-cell RNA sequencing and MERFISH spatial transcriptomics, revealed that excitatory and inhibitory neuron subtypes are defined by distinct combinations of neurotransmitter receptors and transporters⁴. Translating these transcriptomic signatures to the protein level is where spatial proteomics, named Method of the Year 2024 by Nature Methods⁵, and multiplexed antibody-oligonucleotide conjugate (AOC) imaging become essential tools.

This article examines the major molecular components of excitatory and inhibitory neurotransmission, reviews their spatial expression patterns across brain circuits, and considers how multiplexed AOC imaging enables researchers to profile the E/I landscape in intact tissue at cellular resolution.

AMPA Receptors: Fast Excitatory Transmission and Synaptic Strength

AMPA receptors mediate the majority of fast excitatory synaptic transmission in the brain. These ligand-gated ion channels open within milliseconds of glutamate binding, producing the rapid postsynaptic depolarisation that initiates information transfer at excitatory synapses. AMPA receptors assemble as tetramers from four subunits, GluA1 through GluA4, and the specific subunit composition of a receptor determines its trafficking, conductance, and calcium permeability⁶.

GluA1 (encoded by GRIA1) is a principal determinant of synaptic plasticity. GluA1-containing AMPA receptors are inserted into synapses during long-term potentiation (LTP), the cellular correlate of learning, and their ubiquitin-mediated removal underlies long-term depression and memory flexibility⁷. Bhatt and colleagues demonstrated that virally induced reduction of GluA1 across hippocampal subfields impairs spatial working memory, while GluA1 ablation specifically from excitatory cells of hippocampal CA2 disrupts short-term social memory⁸. These subfield-specific roles illustrate why spatial resolution matters: GluA1 function depends on where in the circuit it is expressed. Recent cryo-electron microscopy structures of the GluA1 homotetramer have revealed conformational dynamics that explain how this subunit tunes signalling at individual synapses⁹.

GluA2 (encoded by GRIA2) plays a fundamentally different role. Nearly all GluA2 subunits in the adult brain undergo RNA editing at the Q/R site, where a genomically encoded glutamine is converted to arginine by the enzyme ADAR2. This single amino acid change renders GluA2-containing AMPA receptors impermeable to calcium, providing a critical neuroprotective mechanism¹⁰. When GluA2 editing fails or GluA2 expression is reduced, as occurs following ischaemia, in epilepsy, and in amyotrophic lateral sclerosis, calcium-permeable AMPA receptors accumulate, triggering excitotoxic neuronal death¹¹. A 2023 study demonstrated that GluA2 Q/R site editing is impaired in Alzheimer's disease and that unedited GluA2-containing receptors act as an epigenetic switch driving dendritic spine loss, neurodegeneration, and cognitive decline¹². Loss-of-function variants in GRIA2 itself cause neurodevelopmental disorders characterised by intellectual disability, speech delay, and seizures¹³.

The ratio of GluA1 to GluA2 at a synapse is therefore a molecular readout of both synaptic strength and vulnerability. Synapses with a high proportion of GluA1-containing, GluA2-lacking receptors are calcium-permeable and highly plastic; those dominated by GluA2-containing heteromers are calcium-impermeable and stable. Mapping both subunits simultaneously across tissue sections reveals how this balance is distributed across cell types and brain regions, information that cannot be extracted from single-target immunostaining.

NMDA Receptors: Coincidence Detection and Synaptic Plasticity

While AMPA receptors carry the fast component of excitatory transmission, NMDA receptors serve as coincidence detectors that gate synaptic plasticity. NMDA receptors require simultaneous glutamate binding and postsynaptic depolarisation to open, a property that makes them sensitive to the temporal correlation between presynaptic and postsynaptic activity. This coincidence detection mechanism is the molecular basis for Hebbian learning: the principle that neurons that fire together wire together.

GluN1, encoded by GRIN1, is the obligatory subunit of all NMDA receptors. Every functional NMDA receptor contains two GluN1 subunits paired with two GluN2 (A–D) or GluN3 (A–B) subunits, and there is no known functional homologue that can substitute for GluN1¹⁴. This makes GluN1 a universal marker for NMDA receptor-bearing neurons. Alternative splicing of GRIN1 generates eight variants with distinct effects on receptor kinetics and synaptic accumulation; mice expressing the GluN1a splice form show enhanced hippocampal LTP and superior spatial memory compared to those expressing GluN1b¹⁵.

The functional consequences of NMDA receptor disruption are severe. Disease-associated variants in GRIN1 produce neurodevelopmental disorders characterised by intellectual disability, developmental delay, and epilepsy through distinct channel gating pathomechanisms¹⁴. At the circuit level, loss of GluN2A-containing NMDA receptors produces brain-region-specific changes in both excitatory neurons and glia, with downstream dysregulation of dopamine signalling that provides a molecular link between glutamatergic dysfunction and psychiatric phenotypes¹⁶. Because GluN1 is expressed in virtually every neuron, its spatial signal in a multiplexed panel provides a reference channel against which other markers, both excitatory and inhibitory, can be normalised to assess relative receptor density on a per-cell basis.

VGlut1: Identifying Excitatory Neurons at the Presynaptic Terminal

Glutamate receptors mark the postsynaptic side of excitatory synapses. On the presynaptic side, the vesicular glutamate transporters (VGluts) load glutamate into synaptic vesicles and serve as definitive molecular identifiers of glutamatergic neurons. The discovery of VGlut1 and VGlut2 made it possible, for the first time, to specifically identify and target neurons that use glutamate as their neurotransmitter¹⁷.

VGlut1 (encoded by SLC17A7) and VGlut2 (encoded by SLC17A6) exhibit a striking complementary distribution in the adult brain. VGlut1 is the predominant transporter in the cerebral cortex, hippocampus, and cerebellar cortex, regions where glutamatergic neurons display a low probability of transmitter release and use-dependent facilitation. VGlut2 predominates in the thalamus, brainstem, and deep cerebellar nuclei, where synapses show higher release probability¹⁸. This complementary pattern means that VGlut1 effectively labels corticocortical, hippocampal, and cerebellar excitatory terminals, while VGlut2 labels thalamocortical and subcortical inputs.

In spatial transcriptomics and single-cell RNA sequencing datasets, SLC17A7 (VGlut1) expression is one of the primary classifiers used to assign neurons to the excitatory class⁴. A recent integrated single-nucleus and spatial transcriptomics atlas of the human hippocampus used VGlut1 expression to define excitatory neuron populations and map the spatial organisation of excitatory postsynaptic specialisations across hippocampal subfields¹⁹. Including VGlut1 in a multiplexed AOC panel therefore provides a presynaptic anchor point for excitatory neuron identification, complementing the postsynaptic information provided by AMPA and NMDA receptor markers.

VGlut1 expression is also clinically relevant. Altered VGlut1 levels have been reported in Parkinson's disease, where reduced cortical VGlut1 immunoreactivity may reflect the loss of corticostriatal glutamatergic input that contributes to motor and cognitive symptoms²⁰. In Alzheimer's disease models, VGlut1 loss in the hippocampus correlates with synaptic degeneration and precedes overt neuronal death¹⁷.

GABA-B Receptors: Slow Inhibitory Modulation Across Neural Circuits

The inhibitory counterpart to glutamatergic excitation is GABAergic inhibition, and the GABA-B receptor represents its slow, modulatory arm. Unlike ionotropic GABA-A receptors, which open chloride channels within milliseconds to produce fast inhibitory postsynaptic potentials, GABA-B receptors are G protein-coupled receptors that exert their effects over hundreds of milliseconds through second messenger cascades²¹.

GABA-B-R1 (encoded by GABBR1) is an essential component of the obligate heterodimer that forms the functional GABA-B receptor. The receptor requires both GABBR1 and GABBR2 subunits: GABBR1 binds GABA via its venus flytrap domain, while GABBR2 couples to the G protein machinery²¹. The GABBR1 subunit exists as two major splice variants, GABBR1a and GABBR1b, that confer distinct subcellular localisations. GABBR1a contains a pair of sushi domains that direct the receptor to axon terminals, where it functions as a presynaptic autoreceptor and heteroreceptor. GABBR1b lacks these domains and localises preferentially to the somatodendritic compartment, where it mediates postsynaptic inhibition²².

At presynaptic terminals, GABA-B receptor activation inhibits neurotransmitter release through multiple parallel mechanisms. Recent work has dissected these pathways, showing that the inhibition of spontaneous glutamate release requires Gβγ interaction with the C terminus of SNAP25, while inhibition of spontaneous GABA release operates through alternative Gβγ targets, demonstrating that GABA-B receptors regulate excitatory and inhibitory release through distinct molecular mechanisms at the same terminal²³. This dual regulatory role positions GABA-B-R1 as a critical node for understanding how inhibitory tone is maintained across circuits.

GABA-B receptor expression is widespread but regionally heterogeneous, with particularly high densities in the cerebellum, hippocampus, thalamus, and certain cortical layers²⁴. Reduced GABBR1 and GABBR2 levels have been reported in the cerebella of individuals with bipolar disorder, schizophrenia, and major depressive disorder, suggesting that GABA-B receptor deficits contribute to the inhibitory dysfunction observed across psychiatric conditions²⁴. Mapping GABA-B-R1 alongside excitatory markers in the same tissue section provides a direct spatial readout of the molecular components underlying E/I balance, the ratio that defines circuit function and dysfunction.

Excitatory–Inhibitory Imbalance in Neurological Disease

The proteins described above are not merely markers of synaptic identity; they are causal participants in the disease mechanisms linked to E/I imbalance. In autism spectrum disorder, quantitative proteomics of the dorsolateral prefrontal cortex has revealed widespread downregulation of postsynaptic density proteins, including AMPA and NMDA receptor subunits, scaffolding proteins (DLG4, Shank1–3, Homer1), and cell adhesion molecules (NRXN1, NLGN2)²⁵. This synaptic dysmaturation affects both excitatory and inhibitory compartments, consistent with a model in which the developmental trajectory of E/I balance is disrupted rather than simply shifted.

In Alzheimer's disease, progressive disruption of E/I balance, measured both within and between brain regions, correlates with the transition from mild cognitive impairment to dementia¹². The molecular basis includes GluA2 editing failure (producing calcium-permeable AMPA receptors), NMDA receptor hypofunction, and reduced GABAergic interneuron activity. Each of these changes alters the spatial distribution of specific receptor proteins, making them detectable targets for multiplexed spatial profiling. The ability to measure GluA1, GluA2, GluN1, VGlut1, and GABA-B-R1 simultaneously in disease tissue provides a five-dimensional molecular portrait of E/I balance that no single-marker approach can achieve.

Multiplexed Profiling with Antibody-Oligonucleotide Conjugates

Profiling excitatory and inhibitory markers in the same tissue section demands a technology that can resolve multiple targets without spectral crosstalk or tissue degradation. As described in our companion article on ion channel mapping, antibody-oligonucleotide conjugates (AOCs) achieve this through DNA-barcoded detection: each antibody carries a unique oligonucleotide tag, and targets are revealed sequentially through cycles of fluorescent reporter hybridisation, imaging, and dehybridisation^26,27. For a comprehensive overview of the scientific techniques enabled by antibody-oligonucleotide conjugates, our technical reference covers the full range of detection platforms.

For E/I profiling, this approach offers specific advantages. Neurotransmitter receptors can be expressed at modest levels relative to abundant structural proteins, and their detection benefits from the signal amplification strategies available within AOC-based workflows. Platforms such as SABER-IMC provide enzymatic-free signal amplification that boosts sensitivity for low-abundance targets without introducing the artefacts of enzymatic amplification. Meanwhile, the iterative nature of cyclic immunofluorescence with DNA-barcoded antibodies means that the five E/I markers described in this article can be combined with dozens of additional targets, including the twelve voltage-gated ion channels from our companion panel, in a single experiment.

The practical consequence is that researchers can simultaneously ask which cells express excitatory receptors, which express inhibitory receptors, in what proportions, and how those proportions relate to cell type, laminar position, and proximity to pathological features, all from one tissue section.

Designing an Excitatory–Inhibitory Panel for Spatial Neuroscience

An effective E/I profiling panel pairs the five neurotransmitter markers described in this article with cell-type identifiers and contextual markers. The five core targets (GluA1, GluA2, GluN1, VGlut1, and GABA-B-R1) capture both sides of the E/I equation across presynaptic and postsynaptic compartments.

Adding cell-type markers strengthens the analysis. Pairing excitatory markers with GFAP and Aldh1L1 for astrocyte identification is particularly informative, since astrocytes actively regulate extracellular glutamate and GABA concentrations through transporter-mediated uptake and can themselves release glutamate via VGlut-dependent mechanisms. Including calbindin identifies specific interneuron and Purkinje cell populations, adding inhibitory cell-type resolution to the receptor-level E/I data.

For disease studies, the panel can be extended with ion channel markers to link E/I receptor balance to electrical phenotype. The combination of GluA1/GluA2 ratios with Nav1.1 (PV+ interneuron marker) and Kv3.1B (fast-spiking phenotype marker) provides a direct molecular bridge between receptor composition and intrinsic excitability, a combined view that has been technically out of reach until multiplexed AOC panels made it possible.

As with any multiplexed panel, conjugation quality matters. For guidance on optimising the degree of oligonucleotide labelling and minimising background, our technical articles on fine-tuning AOC parameters and eliminating non-specific binding provide detailed recommendations. Our immunofluorescence protocol for antibody-oligo conjugates covers tissue preparation and staining workflows applicable to neurotransmitter receptor targets.

Looking Forward

Excitatory–inhibitory balance is one of the most consequential parameters in brain function, yet it has been remarkably difficult to measure at molecular resolution in intact tissue. The convergence of multiplexed AOC imaging with a growing catalogue of validated conjugates for glutamate receptors, vesicular transporters, and GABA receptors now makes it possible to profile the E/I landscape cell by cell, synapse by synapse. As these panels expand to incorporate ion channels, scaffolding proteins, and additional receptor subtypes, the spatial portrait of neurotransmission will deepen from a two-colour map of excitation versus inhibition into a high-dimensional molecular atlas of synaptic identity.

References