Neurons are the basic signaling components of the brain. Consequently, a fundamental part of any attempt to understand how the brain works as a whole is the investigation of functionally distinct neuronal cell types. Toward this end, immunohistochemical markers have emerged as one of the most valuable tools available to neuroscientists. Using antibodies against various cell components, investigators are able to identify cells expressing a neuronal phenotype and, moreover, collect information regarding their morphological characteristics and expression of specific proteins. Other approaches exists to define neuronal cell types. For example, physiological optical tagging sequencing (PhOTseq) relies on the initial GCaMP calcium imaging labeling of neurons through activation and then the permanent labeling of specific neurons with photoactivatable mCherry (PAmCherry) [6].

Table 1 lists the commonly used markers for all neurons and for neurons with specific neurotransmitters, many of which are discussed in more detail later. Separate articles address other cell types in the brain: glial cells, microglia, and glymphatic and meningeal lymphatic system.

[enlarge]

Figure 1. Adult rat C6 neuron immunolabeled for NSE (green) and counterstained with DAPI (blue). Bar = 20 µm. Reproduced from Figure 5A of [1].

Neurons are composed of four distinct morphological areas: a cell body (soma), dendrites, an axon, and presynaptic terminals. The highly specialized structure of these cells enables them to propagate electrical signals, or action potentials, that are the basis for communication between neurons. The soma is the metabolic center of the cell. It houses the nucleus containing the cell’s DNA as well as other organelles. From the soma, two types of processes arise. The first type, called a dendrite, receives incoming information whereas the second type, the axon, conveys outgoing information. Typically, a neuron will have several dendrites. Each of these dendrites can, in turn, contain hundreds to thousands of spines, which are the sites of input from the axons of other neurons. (It should be noted that axons may also synapse upon the soma or axon as well, although this is less frequent.) The axon arises out of a specialized region of the cell body called the axon hillock and is responsible for the propagation of the action potential. It eventually divides into several branches that terminate at synapses. Although the synapse is said to be the point of contact between neurons, the cells do not contact one another physically but are separated by a space termed the synaptic cleft. At the end of each axonal branch is the presynaptic terminal, which contains vesicles filled with neurotransmitter. When an action potential reaches the terminal, the vesicular contents are released into the synaptic cleft allowing chemical communication with the postsynaptic cell.

NSE, also referred to as gamma-enolase or enolase 2, is a cytosolic protein consistently expressed by mature neurons and cells of neuronal origin (Figure 1). It is a brain-specific glycolytic enzyme that plays an important role in intracellular energy metabolism [26]. During development, NSE levels are very low but then increase during the time frame associated with the morphological and functional maturation of neurons [27].

While NSE is widely used and accepted as a neuronal marker, it is important to note that it has been shown to be expressed in glial cells under certain conditions. Specifically, levels of NSE activity, protein, and mRNA have been detected in cultured oligodendrocytes similar to those observed in cultured neurons [28]. Its expression increases during oligodendrocyte differentiation but is repressed once the cells become mature. Under pathological conditions, NSE is detectable in glial neoplasms and reactive glial cells as well [29].

In 1992, Mullen et al reported the generation of monoclonal antibodies that recognized a neuronal specific nuclear protein in vertebrates (Figure 2) [30]. This protein, which they called Neuronal Nuclei (NeuN), was detected in most neuronal cell types throughout the central and peripheral nervous systems of adult mice [25]. NeuN, also called Fox-3 or RBFOX3, is involved in the regulation of mRNA splicing [31] and plays a role in regulating neural cell differentiation and nervous system development [31]. The appearance of the NeuN corresponds temporally to the withdrawal of neuronal cells from the cell cycle and/or with the initiation of terminal differentiation. The protein is detectable in both embryonic and adult neurons with the exception of cerebellar Purkinje cells, olfactory bulb mitral cells, retinal photoreceptor cells, and dopaminergic neurons in the substantia nigra [32, 33].

[enlarge]

Figure 2. Example of NeuN-labeled neurons in mouse neocortex. Reproduced from Figure 3B of [2].

Fluorescence-activated nuclear sorting of NeuN-stained (and often counterstained by propidium iodide) can isolate neuronal nuclei [34, 35] and is commonly used in single-neuron studies like snmC-seq [36] or single-nucleus RNA-seq [37].

MAP-2 is a neuron-specific cytoskeletal protein that is used as a marker of neuronal phenotype [38]. It is expressed in the vertebrate nervous system in both embryonic and adult tissues. Its expression is weak in neuronal precursors but becomes pronounced later (approximately one day after expression of neuron-specific tubulin isoform βIII) [39]. Studies in rats have indicated the existence of at least three isoforms of MAP-2 including 2a, 2b, and 2c. MAP-2c appears to be expressed exclusively during early development and only in axons. MAP-2c appears to be then replaced by MAP2a over the course of maturation. MAP2b, in contrast, has been found to be expressed throughout life.

MAP-2 protein is thought to be involved in microtubule assembly, acting to stabilize microtubule growth by crosslinking with intermediate filaments and other microtubules. Specifically, it may play a role in determining and stabilizing dendritic shape during neuronal development and is accordingly observed only within the dendritic arbor [40]. In general, its expression appears to be confined to neurons and reactive astrocytes [41]. Zullo JM et al used MAP2 as the marker of cortical neurons in the immunohistochemical labelling of coronal mouse brain sections [42]. Zeng Q et al identified differentiated neurons and astrocytes in primary cultures of mouse cortical neurons with MAP2 and GFAP labelling [25]. Velasco S et al identified neurons from brain organoids with a MAP-2 marker [9]. Dominy SS et al studied the co-localization of Porphyromonas gingivalis RgpB protein with hippocampal neurons from Alzheimer's disease patients using MAP2 as a neuronal marker [43].

Tubulin beta III (TUBB3), also called TuJ1, is a tubulin thought to be involved specifically during differentiation of neuronal cell types [44]. Tubulins are a major building block of microtubules, which are structural components of the cytoskeleton attributed roles in cell structure maintenance, mitosis, meiosis, and intracellular transport among others. Accordingly, immunohistochemical staining of TuJ1 is found in the cell bodies, dendrites, axons, and axonal terminations of immature neurons. One of the significant advantages associated with the use of immunohistochemical detection of TuJ1 is the degree to which it reveals the fine details of axons and terminations (Figure 3). In fact, it has been found to be useful in the detection of injury-related alterations in the composition of the somatic cytoskeleton [45]. Although a form of ß-tubulin is also found in glial cells, it is not recognized by antibodies against TuJ1 [45]. Gur-Cohen S et al labelled dermal neuronal fibers with a TuJ1 antibody [46], so did Correa-Gallegos D et al in their wound healing study [47]. Maniatis S et al used TUBB3 staining to outline motor neuron somata [48].

[enlarge]

Figure 3. Doublecortin-labeled cells in the dentate gyrus of adult mice. Reproduced from Figure 5 of [3].

Another tubulin gene TUBA4A has also been used as a neuronal marker. Hu CK et al identified neuronas in killifish embryos with Sigma antibody T7451 (clone 6-11B-1) against acetylated tubulin [49].

DCX is widely expressed by migrating neurons and is observed in the earliest stages of neuronal development [50] and is considered as a marker for developing and adult neurogenesis in the central nervous system [51, 52]. It is a microtubule-associated protein that is expressed in post-mitotic neurons and has a suggested role in the growth of neuronal processes at the leading edge of the cell [53]. Consistent with this role, DCX immunostaining appears most intense at extremities of neurites and continues into proximal regions of growth cones, but not the tips.

Activation of c-fos has been used as a marker for neuronal activation [13, 14]. Bai L et al observed increased Fos expression in specific brain regions after the stimulation of various vagal sensory neurons [54].

Glutamatergic neurons can be distinguished by the vesicular glutamate transporterse, SLC17A7 / vGluT1 and SLC17A6 / vGluT2, membrane proteins responsible for the uptake of glutamate into synaptic vesicles at presynaptic nerve terminals. A third transporter, SLC17A8 / vGluT3, is expressed in cholinergic and serotoninergic neurons [55]. vGluT1 is identified in cerebral and cerebellar cortices and hippocampus and in the terminals of auditory nerves [56]. vGLUT2 is mostly in diencephalon and rhombencephalon neurons. Both vGluT1 and vGLUT2 are concentrated in the synaptic vesicles in the terminals of the axon of asymmetric synapses.

Members of the glutamate-gated ion channel superfamily such as NMDAR1 / GluN1 and NMDAR2B / GRIN2B / GluN2B also serve as markers for glutamatergic neurons. Eight isoforms of the GluN1 gene have been identified, which vary in the number of N-terminal and C-terminal domains, while their patterns of expression are distinct in time and space in human and rodents [57-59]. NMDAR1 is identified in glutamatergic neurons in cortical plate [60] and dispersed heterogeneously within the cytoplasm and scattered in apical dendrites.

Mitochondrial glutaminase protein, encoded by GLS gene, catalyzes the hydrolysis of glutamine to glutamate in brain and kidney.

Transporters, receptors and enzymes for gamma-aminobutyric acid, a primary neurotransmitter, serve as markers for GABAergic neurons. Zeitler B et al used Darpp32, phosphodiesterase 10a (Pde10a), dopamine receptors D1

(Drd1a) and D2 (Drd2) as markers for medium spiny neurons [19].

GABA in synaptic cleft is eliminated by sodium- and chloride-dependent GABA transporter 1 / GAT1 / solute carrier family 6 member 1 gene /SLC6A1. The protein consists of twelve transmembrane domains, and both C- and N-terminals are intracellularly situated [61]. It is highly expressed in brain and liver.

GABA B receptor subunit 1 (GABBR1) and 2 form a heterodimer, serving as the receptor for GABA. Multiple alternatively spliced isoforms for GABBR1 exist. It is highly expressed in brain and also expressed broadly in spleen and other tissues.

The second component of GABA type B receptor is gamma-aminobutyric acid type B receptor subunit 2 protein (GBR2 / GABAB2). Its expression is restricted to brain.

There are two isoforms of functional GAD67 / GAD1, which catalyze the production of gamma-aminobutyric acid from L-glutamic acid. GAD67 is often used as a marker for GABAergic neurons along with GABA [62]. The gene is highly expressed in brain and also expressed in kidney, testis and other tissues.

At least eight more alternatively spliced variants are translated from this gene. Their distribution is region-dependent, though not all of them can be distinguished by immunohistochemistry [63].

Glutamate decarboxylase 2 / GAD65 / GAD2 catalyzes the production of gamma-aminobutyric acid from L-glutamic acid. It expression is restricted to brain.

Tyrosine hydroxylase is the rate-limiting enzyme during the synthesis of the catecholamines and is involved in the conversion of tyrosine to dopamine. The gene is highly expressed in adrenal glands. There are four major human TH isoforms; all are present in neurons [64].

Immunohistochemistry for TH has been an important tool for investigation of Parkinson’s disease, a condition characterized by movement disorder associated with the loss of the dopaminergic cells of the substantia nigra [24]. In post mortem studies, TH immunohistochemistry has been used to quantify the degree of dopaminergic cell loss in Parkinson’s patients [65]. To determine whether this type of loss might be mitigated by intervention, TH immunochemistry has been used as a means of examining the protective effects of various experimental manipulations in animal models of Parkinson’s disease [66], or confirm whether dopaminergic/adrenergic neurons were involved in specific processes [51, 67, 68]. Y Shwartz et al labelled sympathetic neurons innervating hair follicles with tyrosine hydroxylase [22].

Sodium-dependent dopamine transporter acts in presynaptic terminals, terminating dopamine effect by sodium-dependent reuptake. DAT-positive staining is predominantly associated with intracellular membranes (in perikarya and dendrites) or plasma membrane (in unmyelinated axons or medium- to small-diameter dendrites) [69]. Intracellular localization was variable, being detected in axons and varicosities or perycaria and dendrites [70].

GIRK2 (G protein-activated inward rectifier potassium channel 2 ) /KCNJ6 is a member of potassium inwardly rectifying channel subfamily J. It is highly expressed brain and also expressed in stomach and other tissues.

GIRK2 is considered as a mature DA neurons marker, as it is detected at day 50 in in vitro differentiation studies of human [71]. Almost all of TH-immunopositive neuronal cells are also GIRK2 positive in substantia nigra, but only 50-60% of these cells in ventral tegmental area are stained with GIRK2 [72].

This type of neurons is distinct phenotypically by proteins involved in serotonin (5-hydroxytryptamine) synthesis (tryptophan hydroxylase) and functioning (serononine transporter).

Tryptophan hydroxylases (tryptophan hydroxylase 1 and 2) catalyze the first and rate limiting step in the biosynthesis of serotonin.

Serotonin transporter recycles the neurotransmitter serotonin from synaptic spaces into presynaptic neurons and terminates its action of serotonin.

The protein Pet1, also known as FEV transcription factor, is exclusively expressed in serotonergic neurons in the brain. It is also expressed in duodenum, stomach and other tissues.

Nerve cells which use acetylcholine as main neurotransmitter are called cholinergic neurons. Thus, proteins applied as their markers are involved acetylcholine metabolism.

Choline acetyltransferase catalyzes the biosynthesis of the neurotransmitter acetylcholine. It is expressed in the vast majority of cholinergic neurons, ChAT immunoreactivity is used as a biomarker of cognitive decline in various neurodegenerative disorders. Its detection has been a significant tool for investigating the neurological changes typical of Alzheimer’s Disease (AD), which is associated with a substantial loss of basal forebrain cholinergic neurons. Post-mortem studies have used ChAT immunohistochemistry to reveal decreases in the density of fibers and axon varicosities of ChAT-positive neurons in the superior frontal cortex and amygdala of cognitively impaired patients [73, 74]. Furthermore, in animal models, immunohistochemistry has been used to evaluate the capacity of experimental manipulations or transgenes to alter the number of ChAT-positive neurons in the medial septum and the Broca region [75, 76] or striatum [77]. In regions including the cingulate, motor, and sensory cortices, and the basal forebrain, immunolabeling has also been applied to evaluate cholinergic neurons for disruption of the ChAT-ir fiber network and for changes in overall morphology [78, 79].

Vesicular acetylcholine transporter, also called solute carrier family 18 member A3, or SLC18A3, is the member of vesicular amine proton-dependent transporters. It transports acetylcholine into secretory vesicles for release into the extracellular space.

Acetylcholinesterase hydrolyzes acetylcholine at neuromuscular junctions and brain cholinergic synapses, and thus terminates signal transmission.

In addition to the markers that have been described above, there are a significant number of other neuron-associated markers (Table 1). Y Shwartz et al labeled the apposition between sympathetic nerve fibers and hair follicle stem cells with the pre-synaptic marker synaptotagmin, synaptophysin, or VMAT2 [22]. Zeng Q et al labeled presynaptic vesicles with synaptobrevin 1 in tissue sections of metastatic lesions and adjacent normal brain [25]. Abdo H et al detected unmyelinated nerve endings in the skin with PGP9.5 [80]. Brn2 [16, 81], Tbr1 [81, 82], and Ctip2 [81, 82] serve as cortical lineage neuronal markers. Velasco S et al marked forebrain progenitors with FOXG1, dorsal forebrain progenitors with EMX1 and PAX6, intermediate progenitor cells with TBR2, post-mitotic projection neurons with TBR1, corticofugal projection neurons with CTIP2, callosal projection neurons with SATB2 respectively, in brain organoids [9]. Kapogiannis D et al ascertained L1CAM-enriched plasma exosomes as of neuronal origin using synaptophysin and L1CAM [17]. Mauffrey P et al used polysialylated-neural cell adhesion molecule (PSA-NCAM) and internexin as markers of neural progenitors [51]. L Cantuti-Castelvetri et al detected olfactory neuronal progenitors with OLIG2 [23].

The first individual to recognize the vast diversity that exists across the neuron population was the founding father of neurobiology, Santiago Ramon y Cajal. Using Camillo Golgi’s method of silver impregnation, he observed the highly distinct morphological features of individual neurons. In doing so, he found that the cells comprising the brain ranged greatly in both size and shape. Whereas some neurons such as granule cells exhibited relatively simple processes arising from a small cell body, others like the cerebellar Purkinje cells were characterized by very florid and complex projections.

[enlarge]

Figure 5. Confocal images of cells double-stained for nestin (green) and βIII-tubulin (red) in the adult rat brain. Reproduced from Figure 6A of [5].

Since Cajal’s initial discovery, scientists have been able to distinguish hundreds of different neuronal cell types. However, a unified system of classification remains elusive. This is, in part, due to the fact that the names assigned to different cells tend to vary in their degree of descriptive accuracy. Furthermore, categories of neurons often overlap, the result being that a particular cell type may be associated with several different names. In general, though, neurons can be classified according to criteria that include morphology, projection patterns, electrophysiological properties, and biochemical characteristics.

Examples of classifications based upon morphology (e.g., size, soma shape, and dendritic branch patterns) are arguably the most commonly found in the scientific literature. While morphology provides an obvious and accessible means of characterization, the deeper rationale for its application lies in its functional significance. Because the shape of a neuron is determined by the input it receives and the targets onto which it projects, the neuronal structure is a direct reflection of the cell’s underlying connectivity, and therefore also of its function. Take, for example, the shape of climbing fibers and ascending branches of granule cell axons in the cerebellum. Here, the axonal shape is determined by specific connections to target Purkinje cell dendrites in which the axonal arbor makes numerous contacts with a single dendrite [95]. Similarly, the type of information revealed by morphology can be illustrated by examining the classical pyramidal cell. The term pyramidal cell refers broadly to cells that are characterized by a large, triangular soma with distinct sets of apical and basal dendrites. This structural organization is suggestive of the role these cells play in the incorporation of varied inputs from different cellular layers into an integrated message that is communicated to another brain region [96].

Another important structural distinction often made is based upon the axonal projections of a neuron. Cells that extend an axon beyond the discrete population of neurons to which they belong (i.e., nucleus) are referred to as projection, or principal neurons, whereas those that make synaptic contact only within their nucleus are called interneurons, or intrinsic neurons. It should be emphasized that these terms refer only to the axonal projection, and not to the source of inputs. Thus, a projection neuron may receive only local inputs and, conversely, an intrinsic neuron may receive input from another nucleus. Identification of these patterns of connectivity is important as it provides insight regarding the flow of information processing that underlies the function associated with these cell populations. Indeed, this key concept underlies the significant effort that has recently been directed toward the compilation of comprehensive maps of the neural connections in the brain, called “connectomes” [97].

Neuronal cell types may also be distinguished by their electrophysiological properties. For example, neocortical neurons, which exhibit significant differences in their action potential response patterns, are typically classified as regular-spiking (RS), fast-spiking (FS), and intrinsically bursting (IB) cells. Regular-spiking cells adapt strongly in response to maintained stimuli whereas FS cells sustain high firing frequency with little or no adaptation and IB cells generate clusters of spikes singly or repetitively [98, 99]. The diversity in intrinsic firing patterns of these cells is significant in that it represents an additional level at which the response of these cells to stimuli can be modulated.

With regard to the distinguishing criteria covered thus far, immunohistochemistry can be used to provide information about morphology and, to some extent, axonal projections of different neuron populations. However, the real strength of the methodology in the context of neuronal phenotyping lies in its potential to help distinguish cells based upon biochemical characteristics. Indeed, as will be described in the following sections, the identification of neurons categorized by an association with a particular neurotransmitter system or the expression of a specific gene or protein is a frequently employed strategy.

While Golgi staining remains a useful tool for identifying neuronal cell types, the technique is associated with certain drawbacks. Probably the most significant of these is that, even when successful, the staining tends to be variable. Although complete impregnation of the dendritic tree can be achieved, axonal branches and fine distal terminal arborizations are difficult to thoroughly label [100]. Additionally, it is estimated that, for reasons that are unknown, impregnation only occurs in 1-10% of cells [101, 102]. While this aspect of the protocol is exactly what enabled Cajal’s original discovery that neurons were in fact independent entities, it makes targeting specific cells difficult and also increases the quantity of tissue required for processing to achieve appropriate sample sizes.

With the advent of immunohistochemical strategies for identifying cell types in the 1970s, investigators were able to circumvent some of the technical issues associated with older methods such as Golgi staining. By that time, researchers had begun using immunohistochemistry to identify neurons containing the monoamine-synthesizing enzymes tyrosine hydroxylase (TH), dopamine-beta-hydroxylase (DBH), and tryptophan hydroxylase (TrH) [103, 104, 104]. Nevertheless, it was subsequent to reports describing the use of the brain endolases: neuron-specific enolase (NSE) and non-neuronal enolase (NNE), as specific markers of neuronal and glial cells, respectively, that the use of immunohistochemistry for identifying neuronal cells really began to flourish [105]. Other studies followed describing immunolabeling of a range of different antigens differentially expressed among neuronal cell types including NGF-Inducible Large External glycoprotein (NILE-GF) [106], choline acetyltransferase [107], parvalbumin [10, 62], calretinin [62, 77], somatostatin [62], neuronal peptide Y [62] and neurofilament protein [108, 109] as markers. While not all of these have continued to be widely utilized in this context, the impact of the methodological strategy has remained important.

More recently, molecular approaches have been developed that are helping to elucidate neuronal characteristics. Methods that utilize transcriptional modulators and site-specific recombinases to label specific neuronal populations have become available to investigators [110]. However, immunohistochemistry has managed to maintain its status as a staple methodology for identifying neuronal cell types because it is relatively low in cost to perform, requires very little specialized equipment, and uses reagents are that are widely available. Moreover, routine microscopic methods can be used to collect data and, depending on the method of visualization, immunolabeled tissue may be preserved for reference.