At the end of the nineteenth century, the classic Golgi stain, which is still widely used to study the development of neurons, was formulated. The principle of sparse labeling is that the dark brown precipitation generated by the reaction between the potassium bichromate solution and silver ions randomly labels a relatively small number of neurons, particularly the dendrites. Using Golgi staining, researchers have recorded the morphological changes to dendrites occurring in the early developmental stages (Ivy and Killackey, 1981; Niwa et al., 2010). The use of hematoxylin and eosin (HE) stains the cell nuclei within brain tissue blue with hematoxylin, and proteins are stained pink with eosin (Fischer et al., 2008; Picut et al., 2021). HE staining can detect morphological changes to neurons in model mice during their early developmental stages (Zhou et al., 2018). The Nissl staining method uses basic dyes like cresyl violet acetate or toluidine blue to permanently stain nucleic acids within neurons in the brain (Paul et al., 2008; Smiley et al., 2023) and is used for detecting the morphology and pathology of neuronal tissues and understanding the cytoarchitecture of different brain areas (Kawase-Koga et al., 2009). Immunohistochemical staining uses antibodies to target specific antigens in order to show morphological changes to neurons at different developmental stages (Lee et al., 1998). Nevertheless, these classic staining methods are laborious. They are limited in use to fixed brain samples, instead of live ones.

In the second half of the twentieth century, many traditional neural tracers were developed that can label neurons in vivo. Horseradish peroxidase (HRP) is one of the earliest markers used in the morphological study of neuronal axons and is still being used in labeling neonatal axons (van der Togt and Feirabend, 1990; Sohur et al., 2014; Harb et al., 2016; De León Reyes et al., 2019; Saleeba et al., 2019). Phytohemagglutinin (Phal) (Fish et al., 1991) is a long-lasting tracer employed for labeling neonatal axons. Biocytin and neurobiotin can also clearly label the dendritic and axonal structure of neurons (Staiger et al., 2004; Moreno-Velasquez et al., 2020). Biotinylated dextran amine (BDA) can be used to detect the normal morphological development of axons and dendrites in neonates (Ding and Elberger, 2001; Sugar and Witter, 2016; Kroon et al., 2019) and abnormal morphology in disease models (van Velthoven et al., 2012). Fluorescent dyes, such as Fluro Gold (Hartung et al., 2016) and Fast Blue (Lopez-Roman et al., 1993), are absorbed by axon terminals and accumulate in the soma through retrograde transport, thus they highlight the dendrites of neurons involved in specific loop connections. Unfortunately, high doses of dyes like Fluro Gold may lead to the death of young mice (Hu H. et al., 2021).

Injecting chemical markers into the ventricle can label massive progenitor cells, with some of the markers being passed onto the progeny. Thymidine analogs like bromodeoxyuridine (BrdU) are commonly used markers (Miller and Nowakowski, 1988) that can be inserted into the double-strand DNA of progenitors during S-phase (Inta et al., 2008). However, they label neurons that have already been in the migration state, which means that intact migration and differentiation pathways cannot be fully traced. To overcome this problem, researchers developed Flash-tag (FT) technology. After injection into the ventricles, carboxyl fluorescein (CFSE) can bind to the proteins in M phase progenitors and their progeny. The specificity of CFSE for M-phase progenitors is the result of the soma of these cells being transiently exposed to the CFSE injected into the cerebrospinal fluid (Telley et al., 2016; Govindan et al., 2018; Yoshinaga et al., 2021). These chemical markers combine with specific molecules in progenitor neurons but are diluted as the cells undergo rapid division.

Single-cell injection of traditional chemical tracers can achieve the goal of targeting specific individual neurons in neonatal mouse brains (Larsen and Callaway, 2006). Anterograde tracers such as biocytin (Moreno-Velasquez et al., 2020), biotinylated glucosamine (Cunningham et al., 2002; Staiger et al., 2004), and carbocyanine dyes (red fluorescent Dil and green fluorescent Dio) (Lee et al., 2005; Lickiss et al., 2012; Sohur et al., 2014) can mark the axons and dendritic structures of developing neurons. By utilizing physical methods like single-neuron injection, researchers can locate specific neurons within a targeted area. As it is challenging to perform single-neuron injection, more labeling methods targeting specific neurons are needed.

Chemical tracers are diverse, highly sensitive, and can be transported retrograde or anterograde by being attached to different vectors at different rates (Table 1). However, some types of traditional chemical tracers are toxic to neurons, prone to dye leakage and the mislabeling of passing fibers, and fail to identify neurons that express specific molecular markers. These limitations may reduce their reliability and accuracy for labeling neurons in neonatal mice.

Neurotropic viruses are widely used vectors for labeling neurons in young mice. As the first viral tracers used to label the soma of the neonatal mouse, retroviruses can permanently integrate the genome they carry into host neurons after mitosis, achieving a long-term and stable fluorescent signal (Price et al., 1987; Badea et al., 2003; Yu et al., 2009). Despite this, the infection efficiency of retroviruses is relatively low. In addition, gene insertion by retroviruses is random, which may lead to the mutation of endogenous genes, resulting in unexpected abnormal changes to neurons.

Adeno-associated viruses (AAV) have the advantages of versatility and multiple serotypes, rendering them a popular tool in the field of neuroanatomy (Castle et al., 2016; Hammond et al., 2017). A viral tracer can be injected into neonatal mouse brains, facilitating the visualization of neurons in about 2 weeks (Cheetham et al., 2015; Chen et al., 2018). Through the implementation of an exceptional extrinsic transgenic system and the utilization of various viral tracers, one can investigate specific morphological transformations in distinct neuron types, such as pyramidal neurons and interneurons, in the postnatal period (Kim et al., 2013). Furthermore, AAV can carry a supernova system (Luo et al., 2016) that makes use of the low leakage characteristics of the tetracycline response element for select marking (Figure 3C). The clustered regularly interspersed short palindromic repeats (CRISPR)-CRISPR associated (Cas)9 system can be used to integrate specific genes and viral tracers into the genomes of target neurons with unprecedented speed and precision, allowing researchers to visualize developing neurons by identifying the specific proteins they express (Uemura et al., 2016). In addition, diluted viruses such as AAV (Gibson and Ma, 2011; Kim et al., 2013), Sindbis virus (Lendvai et al., 2000), and retroviruses (Kerloch et al., 2019) may be used to sparsely label neurons in neonatal mice.

Viral tracers are a popular tool in the field of neurodevelopment. Nevertheless, there are still several limitations associated with the application of viral tracers in neonatal mouse brains. Only a few studies have revealed details of the morphological development of neurons during the early postnatal period using viral tracers (Pilpel et al., 2009), as it generally takes about 2 weeks for viral tracers to express detectable signals in mouse brains. There is some debate as to whether the use of viral tracers causes the inflammation and gene-coding-related abnormal growth of neurons (Uyaniker et al., 2019). In addition, the standard procedure used for the stereotactic injection of viruses into adult mice may not be suitable for use on neonatal mice. It is more complicated to anesthetize, immobilize, locate brain regions, and precisely inject viruses into neonatal mice (Ho et al., 2020; Olivetti et al., 2020).

Transgenic mice provide an option for non-invasive labeling. A wide variety of transgenic mice can be generated by prokaryotic injection, bacterial artificial chromosomes (BACs), or knock-in techniques. In the study of neuron lineage tracing, transgenic mice have been generated to target specific types of neurons that develop after embryonic development (Gao et al., 2014; Huang, 2014; He et al., 2016).

In the case of transgenic mice, different insertion sites and copies of foreign genes can be used to enable the sparse and random labeling of neurons. Driven by specific promoters, genes in these transgenic mice express site-specific recombinases that identify specific DNA sequences and activate the expression of fluorescent proteins (Huang et al., 2002). For instance, the dendrites of neurons in the cortex and hippocampus have been visualized in transgenic mice that express fluorescent proteins regulated by the Thy1-promotor (Feng et al., 2000). Mosaic analysis with double markers (MADM) is a genetic system that allows for the fluorescent labeling of neurons in vivo and has been used in the tracing of neuron lineages (Espinosa and Luo, 2008; Figure 3A). In MADM, two chimeric marker genes interrupted by a loxP-containing intron are located on homologous chromosomes, and the expression of the fluorescent proteins requires cyclization recombinase (Cre)-mediated interchromosomal recombination. Transgenic mice used for lineage tracing with the MADM (Figure 3B) system can also be used to visualize partial neuronal dendrites in neonatal mice brains (Yu et al., 2009; Tasic et al., 2012). Cyclization recombinase estrogen receptor (Cre-ER) transgenic mice, which express fusion proteins containing estrogen receptors and Cyclization recombinase (Cre), have neurons that are sparsely labeled under the induction of tamoxifen (Badea et al., 2003). The Flp/FRT recombinase system is based on the same strategy as the Cre/loxp system, but it has only been applied in a few cases for visualizing the neuronal morphology in neonatal mice brains (Nern et al., 2011). Mosaicism with repeat frameshift (MORF) allows a single Bacterial artificial chromosome (BAC) transgene to sparsely label neurons in mice. A mononucleotide G22 repeat is inserted, and only neurons with the G22 to G3n frameshift express the fluorescent signals. With the aid of in vivo imaging techniques, transgenic mice facilitate the dynamic long-term tracking of development of dendrites (Portera-Cailliau et al., 2005; Zuo et al., 2005).

Despite the advantages, labeling signals in transgenic mice maybe unstable in the early developmental stages. Transgenic mice like mGFP, L21, may express relatively weak signals during the early postnatal period (Porrero et al., 2010). The temporal expression of these signals may be caused by the gradual promotion of the expression of genes, such as the Thy-1 promotor. Or it could be attributed to the insertion of foreign genes and the timing of the expression of specific markers in various neuron types. For example, only the dendrites and soma-proximal axons of dopamine neurons can be observed in the early postnatal period of Mosaicism with repeat frameshift (MORF) transgenic mice (Lu and Yang, 2017). BAC transgenic mice, which are based on homologous recombination, show weak signals for labeled GABAergic neurons (GABA: Gamma-aminobutyric acid) during the early postnatal period (Chattopadhyaya et al., 2004). CreER transgenic mice, which need to be treated with tamoxifen before they activate the expression of specific genes, do not display strong signals in their cholinergic neurons until 2 or 3 weeks after birth (Rotolo et al., 2008). While the combination of transgenesis and viral tracers increases the flexibility of sparse labeling (Le Magueresse et al., 2011; Zhang et al., 2017), the problem of the delayed expression of genetic tags remains unresolved. Further improvements are required to optimize the labeling efficiency of transgenic tools.

In utero electroporation (IUE) is a popular labeling method that has been in development since 2001 (Saito and Nakatsuji, 2001). Gene vectors can be injected into the ventricle as early as the embryonic stage. Then, electrical pulses induced by the electrodes on both sides of the embryo’s head help transduce the vectors into the progenitor neuron population within a specific brain region, such as the cortex (Niwa et al., 2010) or hippocampus (Navarro-Quiroga et al., 2007). As the progenitor neurons continue to divide, foreign genes are passed to daughter neurons, which differentiate and migrate to targeted regions. For mice, the optimal surgical period for IUE is between 10.5 and 16.5 embryonic days (Yoshida et al., 2010). In this regard, IUE can alleviate the issue of delayed gene expression in the postnatal period. The vectors carrying fluorescent protein genes can be transfected into specific neurons during the embryonic period to enable the tracking of axon fiber bundles during development (Wang et al., 2007; Chen et al., 2008; Ka and Kim, 2016; Kerloch et al., 2019). However, the transfection range of IUE is limited, and electrical impulses may affect the normal development of neurons (Azzarelli et al., 2017).

A notable advantage of transgenic labeling technology lies in its ability to precisely detect molecular markers in distinct types of neurons. This allows for controlled detection signals to be expressed solely in targeted neuron subtypes. IUE provides a reliable protocol for delivering genetic tags to specific types of neurons. By selectively targeting neural progenitors across distinct developmental stages, IUE can facilitate the investigation of unique groups of newborn neurons that will subsequently migrate to various brain regions (Mizuno et al., 2010). Minimal amounts of plasmid (Pacary et al., 2012; Sugiyama et al., 2020) are needed, and single-neuron electroporation can be used to label a relatively small number of neurons (Uesaka et al., 2005). Vectors carrying the supernova system can be delivered to specific neurons for morphological labeling (Mizuno et al., 2018). Moreover, using CRISPR/Cas9 combined with piggyBac transposase technology, researchers can track the development of specific neocortical progenitors with IUE (Chen et al., 2015). The promoter-assisted spares-neuron multiple-gene labeling using in utero electroporation (PASME) system (Ako et al., 2011) utilizes the Thy1S promoter, the Cre-loxP system, and IUE. The Thy1S promoter is highly selectively expressed in the postnatal neocortex and can induce GFP expression in only a limited number of neurons (Figure 3D).

It is vital that we discover what links the different types of morphological changes that occur in the basic unit of signal transmission in the brain, the neuron, with the connections and functions of complex neural circuits. The development of genetic engineering technology provides abundant options for scientists to track changes to the morphology of neurons during the developmental period (Table 2). Genetic labeling methods can be used to target different types of neurons, but it is difficult to target single neurons by physical methods, and there is a lack of stable signal during the early developing period. Consequently, to obtain a continuous and comprehensive map of the morphological development of various neuron types, further research is required.