Keywords

1 What Is the RB Neuron?

Our skin allows for the perception of various sensations, including thermal, chemical, and mechanical sensations, via the activity of sensory neurons that project peripheral terminals into the skin. In vertebrates, cell bodies of somatosensory neurons are found in either the trigeminal ganglion (TG) or the dorsal root ganglion (DRG). The TG sensory neurons innervate the head skin, whereas the DRG sensory neurons are involved in the reception of stimulation to the rest of the body, such as the trunk and limbs. Both the TG and DRG neurons express many types of sensory receptor channels that receive stimulus information, such as thermosensitive, chemosensitive, and mechanosensitive channels (McKemy et al. 2002; Lumpkin and Caterina 2007; Coste et al. 2010; Geffeney and Goodman 2012; Li et al. 2016).

In addition to these sensory neuronal populations, the larvae of anamniote vertebrates (lampreys, teleosts, and amphibians) have an additional population of sensory neurons, termed Rohon-Beard (RB) neurons after the scientists who first described them in the nineteenth century. RB neurons are pseudo-unipolar neurons which have two types of axons, namely, central and peripheral axons (Bernhardt et al. 1990; Metcalfe et al. 1990) (Fig. 4.1a). The RB neurons share following morphological features between the anamniote vertebrates (Roberts 2000). Large spherical cell bodies of RB neurons (diameter 10–20 μm) reside in the dorsal spinal cord as longitudinal columns. From the cell body, the central axon extends in the rostral-caudal direction, and the ascending branch extends to the hindbrain, whereas the peripheral axon projects toward the skin to innervate dermal sensory structures or terminate as free endings in the dermis or epidermis (Fig. 4.1b). These axons never cross the midline, thus the elongation is restricted to the ipsilateral side (Bernhardt et al. 1990; Metcalfe et al. 1990; Halloran et al. 1999). The sensory input from RB neurons to the CNS activates a neural network that triggers an escape response to move away from the stimulus (Metcalfe et al. 1990). The stereotyped dorsal position and morphological features including large spherical cell body help us to identify RB neurons in the spinal cord. In addition, the large cell body makes electrophysiological recording of neural activity easier.

Fig. 4.1
figure 1

Schematic diagram of RB neuron

(a) Morphology of RB neuron. Central axon bifurcate and extend in both the rostral and caudal directions, whereas the peripheral axon is highly branched to innervate skin broadly. (b) The cell bodies of RB neurons are arranged as two lines of longitudinal columns in the dorsal spinal cord. Both the central and peripheral axons grow only in the ipsilateral side. The central axons elongate in the rostral or caudal direction in the spinal cord, whereas the peripheral axons elongate to the outside of the spinal cord and innervate the skin

SC spinal cord, NC notochord, AC abdominal cavity

Full size image

2 Why Zebrafish Is Widely Used for Investigation of RB Neuron

Using zebrafish embryo as experimental model, many papers according RB neuron have been published. Zebrafish (Danio rerio) has been widely used as an excellent vertebrate model to study biological science, including developmental biology and neuroscience. Zebrafish is a small tropical fish; therefore, we can easily breed them in laboratory. Adult female lay 100–200 eggs during mating with adult male twice a week throughout the year. The fertilized egg rapidly develops to larva whose body is optically transparent. In optimum temperature, 28.5 °C, the larvae hatch out in 3 days after fertilization (dpf), and they start to swim and take food in 4–5 dpf (Kimmel et al. 1995). The rapid development and the optical transparency allow us to observe developmental morphological changes that occur in the larval body. Zebrafish typically attain sexual maturity approximately 3 months post fertilization. This short generation time is advantageous for genetic analysis and manipulation, and many transgenic lines and mutants including gene knockout lines have been established. Transgenic lines that express fluorescent protein in specific cell type help the developmental observation. To label RB neurons, many transgenic zebrafish that express fluorescent proteins in RB neurons have been established (Asakawa et al. 2008; Palanca et al. 2013). These advantages also facilitate neuroscience study using electrophysiological recording of neuronal activity in fluorescence protein labeled neuron and manipulation of neuronal activity by optogenetic methods in developing zebrafish embryos in either wild-type background or mutant background. In addition to the genetic labeling, RB neuron cell body can also be labeled by immunostaining (Table 4.1). The immunostaining method allows us to investigate subcellular distribution of proteins in RB neuron. Consequently, zebrafish have come to be broadly used as an animal model to investigate RB neurons. In addition to zebrafish, Medaka (Oryzias latipes), Japanese small fresh water fish, has also been used as a model vertebrate for biological investigations. Medaka has similar advantages to zebrafish, short generation time (2–3 months), rapid development, and optical transparency of larval body (Wittbrodt et al. 2002). However, the Medaka model has not been used to RB neuron study, unlike zebrafish.

Table 4.1 Antibodies that can label the cell body of RB neuron
Full size table

Larvae of lamprey and amphibian have also RB neuron. However, lamprey and Xenopus are less advantageous compared to zebrafish. In lamprey, it is very difficult to correct the embryos, because breeding procedure for lamprey has not been established. In addition, generation time of lamprey is 5–7 years (McCauley et al. 2015). On the other hand, Xenopus laevis are easily maintained in laboratory. However, this frog has some undesirable features for genetic manipulation. The generation time is longer than 1 year and furthermore the genome is allotetraploid (Hirsch et al. 2002). These features hamper generation of genetically engineered frogs, because all four alleles have to be mutated to generate homozygous mutants through time-consuming crossbreeding programs. In addition, body of their larvae is opacity, because each cell contains yolk. In spite of these disadvantages, some studies have examined the development of RB neurons in Xenopus laevis (Patterson and Krieg 1999; Fujita et al. 2000; Coen et al. 2001; Rossi et al. 2008; Park et al. 2012). To overcome the disadvantages in Xenopus laevis, Xenopus tropicalis has been introduced as a new model amphibian (Hirsch et al. 2002). This frog has a smaller diploid genome than Xenopus laevis, and the generation time of 4–6 month, which is shorter than that of Xenopus laevis. These features help to generate transgenic or gene knockout animal. In fact, a high efficient transgenesis method and a CRISPR/Cas9 mediated targeted mutagenesis method have been applicated to Xenopus tropicalis (Ogino et al. 2006; Nakayama et al. 2013). The new model frog might contribute to RB neuron study as a useful model animal.

3 Development of the RB Cell

Zebrafish develop rapidly at 28.5 °C. During the developmental process, a fish-shaped embryo with a distinguishable head, trunk, and tail is formed within the first 24 hours post-fertilization (hpf). During early development, zebrafish larvae begin to display two touch-evoked escape responses, contraction and swimming (Saint-Amant and Drapeau 1998). The touch-evoked contractions are rapid, alternating contractions of the trunk, which begin by 21 hpf. On the other hand, the touch-evoked swimming begins by 28 hpf. This tactile sensitivity of larval zebrafish is conferred by two sensory neuron populations, TG and RB neurons, as mentioned above. Thus, these sensory neurons are becoming functional and are incorporated into the neural pathway that induces the escape response by 21 hpf.

CNS development begins with the formation of the neural plate, an ectoderm derivative on the dorsal side of the embryo. In zebrafish, the formation of the neural plate begins as dorsal epiblast thickening at 9 hpf (Kimmel et al. 1995). By 19 hpf, the neural plate is progressively transformed into the neural tube, i.e., the precursor of the vertebrate CNS, through dynamic cellular movement. During this process, the medial region of the neural plate forms the ventral side of the neural tube, whereas the lateral region forms the dorsal side of the neural tube. The process of neural tube formation involves two intermediate stages termed the neural keel and neural rod that are present at 13 hpf and 16 hpf, respectively. During neural tube formation, primary spinal neurons begin to extend axons toward neuronal and muscle targets and the larval zebrafish develop locomotion.

RB neurons and trunk neural crest cells (NCCs) are derived from the same region of the lateral neural plate (Blader et al. 1997; Lewis and Eisen 2003). Defective mutations in BMP signaling lead to a loss or dramatic reduction in both RB neurons and NCCs (Nguyen et al. 1998; Nguyen et al. 2000). In addition, the development of RB neuron and NCCs were impaired also in zebrafish lacking functional prdm1, transcription factor expressing at the lateral neural plate where RB neuron and NCCs are being specified (Hernandez-Lagunas et al. 2005). These data suggest that RB neuron and NCCs share some developmental molecular mechanisms. The segregation of RB neurons and NCCs is dependent on Delta-Notch signaling. At around 12 hpf, some of the progenitor cells express several delta homologs that activate Notch receptors on adjacent cells, which then differentiate into NCCs (Haddon et al. 1998; Cornell and Eisen 2000). Stimulation of Notch signaling reduces neural plate expression of ngn1 (Blader et al. 1997). Expression of ngn1 is detected in the neural plate from the 3-somite stage (10-11 hpf), and then the hign-level ngn1 expressing cells form solid clusters that are srrounded by low-level ngn1 expressing cells. Functional analysis by over-expression or knockdown of the ngn1 gene demonstrated that ngn1 is necessary for RB neuron development (Blader et al. 1997; Andermann et al. 2002). Ngn1 induces expression of downstream basic helix-loop-helix (bHLH) genes, such as neuroD, to promote neural differentiation from proliferative neural precursor cells to post-mitotic neurons (Kanungo et al. 2009).

RB neurons begin to extend their central axons in both the rostral and the caudal direction at 16 hpf (Kuwada et al. 1990; Liu and Halloran 2005). The two central axons initiate outgrowth simultaneously from opposite ends at the basal (ventral) side of the cell body and then elongate to the opposite direction (Andersen et al. 2011). These central axons from RB neurons of the ipsilateral side grow longitudinally together in the spinal cord, forming the dorsal longitudinal fasciculus (DLF), which is a relatively loose fascicle (Liu and Halloran 2005). Similarly, the peripheral axons begin to extend from the cell body and either emerge as a branch from one of the central axons or directly emerge from the cell body at 17–18 hpf. After the emergence, the peripheral axon extends toward the outside of the spinal cord to innervate the skin (Kuwada et al. 1990; Liu and Halloran 2005; Andersen et al. 2011). After leaving the spinal cord, the peripheral axons bifurcate several times to form a highly branched structure that densely covers the surface of the skin. Thus, the central axon elongation is fascicular, whereas the peripheral axon elongation is repulsive. The differences in the projection patterns suggest that different mechanisms guide RB neuron processes.

LIM homeodomain transcription factors (LIM-HD) and their cofactors (CLIMs) are required for the elongation and branching of the peripheral axon (Segawa et al. 2001; Becker et al. 2002; Andersen et al. 2011; Tanaka et al. 2011). In contrast to the peripheral axon, the growth rate of the central axon is increased in LIM-HD activity disrupted RB neuron. The opposite effect of the LIM-HD activity indicates that the growth of the central and the peripheral axon is differentially regulated. Recently, it was reported that Calsyntenin-1 (Clstn-1) is a critical regulator for growth and branching of the peripheral axons (Ponomareva et al. 2014). Clstn-1 is a kinesin adaptor and is required for rapid movement of Rab5 containing endosomes along axons. Ponomareva et al. suggested that the Rab5-containing endosomes may deliver important molecules for the peripheral axon branching process to the branching point in the axon. Dpysl3 and PlexinA4 are downstream genes of Isl1 and Isl2a, respectively (Miyashita et al. 2004; Tanaka et al. 2011). Dpysl3 is a member of cytosolic phosphoproteins that mediate semaphorin signaling. Knockdown experiments using antisense morpholino for Dpysl3 or Sema3D indicate that Dpysl3 cooperates with Sema3D for RB neuron peripheral axon outgrowth (Tanaka et al. 2011). After leaving the spinal cord, peripheral axons are attracted to molecular cues from the skin. This axon guidance requires the leukocyte antigen-related (LAR) family of receptor tyrosine phosphatases ptprfa and ptprfb (Wang et al. 2012). LAR receptors are expressed in RB neurons, and LAR-deficient neurons show a defect in peripheral axon guidance. In addition, heparin sulfate proteoglycans (HSPGs), a direct ligand for LAR receptors, are enriched in the skin; thus, HSPGs may be involved in the attraction of peripheral axons to the skin. The branching of peripheral axons requires Slit2 and PlexinA4, which are commonly known as the receptors for semaphorins.

RB neurons normally innervate ipsilateral skin, as the peripheral axons do not cross the dorsal midline, in which Sema3D is expressed (Halloran et al. 1999). Sema3D deficiency or knockdown studies show that Sema3D propels elongating peripheral axons. More specifically, Sema3D propels the peripheral axons away from the spinal cord through repulsion process. However, the Sema3D does not influence central axon extension. In contrast to Sema3D, transient axonal glycoprotein-1 (TAG-1) affects central axon projection but not peripheral axon projection. TAG-1 is a glycosyl-phosphatidylinositol-linked (GPI-linked) membrane protein that is expressed in RB neurons during axon elongation and promotes the extension of the central axons of RB neurons (Warren et al. 1999; Liu and Halloran 2005). In this manner, central and peripheral axons elongate to different directions and follow different guidance cues.

4 Escape Response Evoked by RB Neuron Activation

After the 21 hpf stage, tactile stimulations on the trunk and tail are received by RB neurons and transmitted to the CNS to induce the touch-evoked response (Saint-Amant and Drapeau 1998; Low et al. 2012). Although tactile sensitivity has been well-established in vertebrates, the mechanosensory molecules for the sensation were not identified until recently. To identify the mechanically activated channel, Coste et al. screened several mouse and rat cell lines by applying force to the cell surface while patch-clamp recording, they identified two mechanically activated cation channels, Piezo1 and Piezo2, in mouse neuroblastoma cell line Neuro2A (Coste et al. 2010). Both Piezo1 and 2 are expressed in some mechano-sensitive tissues related to visceral pain, such as the bladder, colon, and lung. In the aspect of cutaneous sensation, Piezo2 was required for mechanically activated currents in subset of DRG neurons, whereas the expression levels of Piezo1 in DRG neurons were very low. A part of the Piezo2 expressing DRG neurons also expresses transient receptor potential cation channel subfamily V member (TRPV1) channels, which are activated by heating and exposure to capsaicin, suggesting a potential role of Piezo2 in noxious mechanosensation (Coste et al. 2010). Coste et al. also showed that piezo proteins are conserved in many animals, plants, and single cell organisms, such as mycetozoa and ciliophoran (Coste et al. 2010). The zebrafish genome has three piezo homologs, piezo1, piezo2a, and piezo2b. The expression of piezo2b is specific to TG and RB neurons in larvae at 24 hpf, whereas piezo1 and piezo2a are not expressed in these neurons (Faucherre et al. 2013). Morpholino oligonucleotide-mediated piezo2b knockdown shows that piezo2b is essential for tactile sensation in zebrafish larvae (Faucherre et al. 2013). A subset of the Piezo2b expressing RB neurons also express TRPA1b, which is activated by exposure to mustard oil, a noxious stimulating compound. The gene expression profile is comparable with mouse DRG neurons. It is generally considered that the touch-evoked response is mediated by reticulospinal (RS) neurons in the hindbrain.

Reticulospinal (RS) neurons in the hindbrain are involved in the initiation and regulation of C-start escape from various stimuli, including sound, visual, and tactile stimulation (O’Malley et al. 1996; Liu and Fetcho 1999; Weiss et al. 2006). Among RS neurons, Mauthner cells and two of its serial homologs, MiD2cm and MiD3cm, have been well-studied for their involvement in the initiation and regulation of the escape response (Lee et al. 1991; O’Malley et al. 1996; Nakayama and Oda 2004; Korn and Faber 2005). These neurons are thought to constitute a parallel pathway that is collectively termed the “Mauthner series (M-series)” (Lee et al. 1991). Mauthner cells are a pair of giant RS neurons located in rhombomere 4 of the hindbrain in teleosts and amphibians. This neuron projects large diameter axons toward the contralateral spinal motor neurons. Following stimulation, the excited Mauthner cell emits a single action potential which is sufficient to evoke a C-start escape response via contraction of the contralateral skeletal muscle. On the other hand, MiD2cm and MiD3cm reside in rhombomeres 5 and 6, respectively. Similar to Mauthner cells, these Mauthner homologs extend axons to the contralateral side of spinal cord and make synaptic contact with contralateral motor neurons. These Mauthner homologs can initiate the escape response with similar kinetics to that initiated by Mauthner cell. However, the latency of the Mauthner homolog-initiated escape is longer than Mauthner cell-initiated escape (Kohashi and Oda 2008). Tactile stimulation of the head induces the largest amount of bending, and the turning angle reduces as the stimulated position moves toward a posterior region (Saint-Amant and Drapeau 1998; Umeda et al. 2016). A correlation between the stimulated position and the bending angle is required for survival in response to a threat in the environment. For forward-facing threats, fish should substantially change the direction of their swimming in the opposite direction. In contrast, for rear-facing threats, fish should rapidly move forward but not in the opposite direction. The combination of activated M-series neurons probably affects C-bend turning angles for the escape. When only Mauthner cells are activated by electrical stimulation, C-start escape with small turn angle is initiated (Nissanov et al. 1990). Similarly, caudal tactile stimuli activate only the Mauthner cell, leading to the induction of a small turning angle, whereas stimulation to the rostral trunk region activates Mauthner cells and the homologous M-series neurons and then an escape response with larger turn angle is initiated (O’Malley et al. 1996). How is A-P information transmitted to higher brain regions and processed to regulate escape behavior? The cell bodies of RB neurons are arranged in a pair of longitudinal columns on the dorsal side of the spinal cord. The longitudinal arrangement of RB neurons is suitable for detecting the anterior-posterior (A-P) positional information of stimuli. The A-P positional information of stimuli affects the bending angle. Although all RB neurons have an ascending central axon, most of the ascending axons terminate in the spinal cord, while a few RB axons enter the hindbrain. Interestingly, Mauthner dendrite-contacting RB neurons are more abundant in the rostral spinal cord, whereas many of the central axon of caudal RB neurons does not reach the hindbrain (Umeda et al. 2016). The sensory information from the caudal RB neuron may be relayed by Commissural Primary Ascending (CoPA) neurons. CoPA neurons reside in the dorsal spinal cord, and their dendrites receive glutamatergic synaptic input from RB neurons (Gleason et al. 2003; Pietri et al. 2009). In addition, the axon of CoPA neurons ascends in the contralateral dorsal spinal cord to the diencephalon, thus the M-series neurons could be activated by this RB-CoPA pathway (Hale et al. 2001; Pietri et al. 2009; Umeda and Shoji 2017). Meanwhile, the escape response could also be initiated via an intra-spinal circuit but not M-series neurons. Previous studies show that disconnecting the spinal cord and hindbrain at somite 2 did not interfere with the touch response. In contrast, disconnection at somite-10 completely abolishes the touch response; thus, the neural circuitry that is sufficient to generate the touch response resides in the spinal cord between somites 2 and 10 (Downes and Granato 2006; Pietri et al. 2009). Anatomical data suggest that the CoPA neurons also contact descending interneurons, such as the circumferential ipsilateral descending (CiD) neurons and ipsilateral projecting (IC) neurons in the rostral spinal cord (Pietri et al. 2009). These interneurons provide excitatory input to motor neurons through both glutamatergic synapses and gap junction mediated electrical synapses (Saint-Amant and Drapeau 2001; Knogler and Drapeau 2014). However, additional functional studies using electrophysiology or ablation experiments are necessary to verify that CoPA neurons make functional synaptic contact with CiD and IC neurons. The larval zebrafish reflex arc for the touch response has been thought to consist of RB neurons, RS neurons, and motor neurons. However, this circuit could be more complex, and the details have not been elucidated.

5 Other Sensations Received by RB Neurons

In addition to mechanical stimulus, it has been suggested that RB neurons are also activated by thermal or chemical stimulus (Prober et al. 2008; Low et al. 2010a; Gau et al. 2013; Ogino et al. 2015).

Previous studies have identified five heat-activated channels (TRPV1-4, TRPM3) and two cold-activated channels (TRPM8 and TRPA1) in the rodent genome. These seven thermosensitive TRPs are expressed in sensory neurons in the TG and DRG and are activated at different temperatures (Dhaka et al. 2006; Vriens et al. 2011). TRPV1 and 2 respond to painful heating, activating at temperatures >42 °C and >52 °C, respectively. In contrast, TRPV3 and 4 are activated by non-painful warming (TRPV3: >33 °C; TRPV4: <27 °C to 42 °C). TRPM3 responds to noxious heating (>42 °C), similar to TRPV1. TRPV1 is also activated by capsaicin, a pungent compound from chili peppers, therefore capsaicin gives us burning sensation when the compound contacts to mucous membrane, such as in the oral cavity, that are innervated by TRPV1-expressing sensory neurons. Similarly, menthol provides a cooling sensation due to activation of the cool-activated channel TRPM8 (Bautista et al. 2007). In human sensory system, TRPM8 is activated at <25 °C, whereas another cool-activated channel, TRPA1, is activated at <17 °C. In addition to cooling stimuli, TRPA1 is also activated in response to chemical stimulants such as allyl isothiocyanate (mustard oil), cinnamaldehyde (a pungent component in cinnamon), diallyl disulfide (a pungent component in garlic), acrolein (a toxic component in tear gas and vehicle exhaust), and 4-hydroxynonenal (an endogenous compound that is produced in response to tissue injury, inflammation, and oxidative stress) (Prober et al. 2008).

Zebrafish is a freshwater tropical fish that is broadly distributed across parts of India, Bangladesh, Nepal, Burma, and Pakistan (Lawrence 2007). Based on the habitat condition, 28.5 °C is recommended as optimum temperature for rearing in laboratory (Kimmel et al. 1995). As expected from their preference temperature, larval zebrafish robustly avoid noxious hot (36 °C) or cold (10 °C) temperatures (Prober et al. 2008; Gau et al. 2013). Gau et al. conducted a gene knockdown experiment using antisense morpholino for the trpv1 gene and revealed that TRPV1 is required for the heat avoiding response but not required for the response to cold stimulus. The heat avoiding behavior is probably evoked by TG and RB neurons activation, which express transient receptor potential (TRP) channels, trpv1 and trpa1b (Prober et al. 2008; Gau et al. 2013). In contrast to TRPV1, TRPA1b is not a thermal sensing channel but a chemical sensing channel that is activated by mustard oil (Prober et al. 2008). The activation of RB neurons by mustard oil evokes an escape response (Low et al. 2010a; Ogino et al. 2015). The TRP1b expressing RB neurons also express Piezo2b which is required for a tactile stimulus-evoked escape response (Faucherre et al. 2013). On the other hand, piezo2b-positive/trpa1b-negative RB neurons have also been observed, thus trpa1b is expressed in a subset of piezo2b expressing RB neurons (Faucherre et al. 2013). Gene expression analysis revealed that some subunits of P2X, ATP-gated ion channel, are expressed in RB neurons (Kucenas et al. 2003; Palanca et al. 2013). The zebrafish genome contains nine genes encoding a subunit of P2X. Among them, four genes, p2rx3a, p2rx3b, p2rx4.1, and p2rx514, were expressed in RB neurons. The expression of the p2rx3a and p2rx3b genes were detected in the majority of RB neurons, whereas the expression of p2rx4.1 and p2rx514 were detected only in a small proportion of RB neurons in 24 hpf larvae (Kucenas et al. 2003). The p2rx3a and p2rx3b genes are homologous to the mammalian p2rx3 gene, which is expressed in sensory neurons for pain sensation. The P2X channel function has both homomeric and heteromeric channels of the subunits with different kinetics (Kucenas et al. 2003; Roberts et al. 2006). The diversities of the P2X subunits may provide a physiological diversity to RB neurons. Above data of gene expression pattern suggests that RB neurons are a much-diverged population. In addition, an RB neuron subtype that simultaneously expresses trpa1b, p2rx3a, protein kinase C alpha (PKCα) was identified by gene expression analysis (Slatter et al. 2005; Palanca et al. 2013). The simultaneous expression of different modal receptors in the RB neuron indicates that RB neurons are multimodal sensory neurons. Physiological analysis of the touché mutant, a light-touch unresponsiveness mutant, revealed that RB neurons can be classified into two groups based on their electrical responses to subthreshold tactile stimuli (Low et al. 2010a, 2011). That is, subthreshold tactile stimuli induce generator potentials in type I but not in type II RB neurons. The voltage of the generator potentials increases with the intensity of the subthreshold tactile stimuli. When the voltage reaches a sufficient amplitude, the generator potentials trigger action potentials. In the touché mutant, type II RB neurons have completely disappeared. Investigation of the touché mutant suggests that type II RB neurons transmit light-touch sensations (Low et al. 2010a).

Unlike the heat-sensing TRP channels, the equivalent cold-sensing TRPs have not been identified in zebrafish RB neurons. As mentioned above, two cold-activated TRP channels, TRPA1 and TRPM8, were identified in mammalian genomes. In zebrafish, TRPA1 is required for chemical sensations although trpa1 knockout larvae responded to cooling stimuli in a manner similar to that of wildtype larvae (Prober et al. 2008). No homolog of trpm8 has been identified in comprehensive genome-wide studies in zebrafish (Saito and Shingai 2006). Genome analysis has also revealed the absence of TRPM8 in other teleost from 10 different orders (Gracheva and Bagriantsev 2015); thus, the loss of the trpm8 gene may have occurred in the common ancestor of the teleost. In the zebrafish genome, 28 homologs of TRP channels have been identified, 3 trpv homologs, 2 trpa1 homologs, 12 trpc homologs, and 11 trpm homologs (Saito and Shingai 2006; Kastenhuber et al. 2013; Von Niederhausern et al. 2013). The physiological function of the TRP channels can be diverse among teleosts and mammals. For example, mammalian TRPV1 is not only activated in response to heat at >42 °C but also in response to a low pH and capsaicin. In contrast, the zebrafish TRPV1 is activated at 32 °C or in response to low pH but not in response to capsaicin (Gau et al. 2013). Further research with new genetic techniques such as genome editing or new in vivo imaging techniques could contribute to the discovery of the genes for cold sensation. In addition to above TRP channels, the transcripts of trpm7 and trpc4 have been detected in RB neurons (Prober et al. 2008; Gau et al. 2013; Low et al. 2011; Von Niederhausern et al. 2013). However, the functional investigation thereto has not been conducted. Future investigations may also possibly shed light on the physiological significance of the RB neurons.

6 Apoptosis of RB Cells

RB neurons have been considered to be almost completely removed via caspase activity-dependent programmed cell death (apoptosis), and then the primary sensory function of RB neurons is replaced by DRG neurons (Williams et al. 2000; Svoboda et al. 2001; Reyes et al. 2004). The apoptotic RB neurons in larval zebrafish can be detected using transferase-mediated dUTP nick end-labeling (TUNEL) (Cole and Ross 2001). The apoptosis is first observed at 24 hpf in the rostral region of the spinal cord. During development, the region of apoptotic RB neurons moves caudally toward the end of the spinal cord. By 48 hpf, most apoptotic RB neurons are concentrated in the caudal region. Simultaneously with the active apoptosis of RB neurons, the peripheral processes of DRG neurons form. Apoptotic activity peaks between 36 and 48 hpf and rapidly decrease by 72 hpf. This correlation suggests a link between the death of RB neurons and the birth of DRG neurons. However, RB cell degeneration occurs normally in mutants lacking DRG (colorless mutant). Thus, RB apoptosis and DRG development may be separate processes.

TrkC1, a receptor for the neurotrophin NT-3, is expressed in subpopulations of RB neurons. RB neuronal apoptosis is initiated only in TrkC1-negative neurons, suggesting that TrkC1 and NT-3 protect RB neurons from apoptosis (Williams et al. 2000). In support of this hypothesis, antibodies that deplete NT-3 induce RB neuronal apoptosis, while exogenous application of NT-3 reduces RB neuronal apoptosis. In addition to TrKC1-signaling, cyclin-dependent kinase 5 (Cdk5) may be involved in the regulation of RB neuronal apoptosis. SiRNA-mediated knockdown of Cdk5 promotes RB neuronal cell death, whereas overexpression of Cdk5 decreases RB neuronal cell death (Kanungo et al. 2006). One of the major apoptosis regulating mechanism is the mitochondrial pathway in which cytochrome c that released from mitochondria promote caspase activation (Czabotar et al. 2014). This process is regulated by bcl-2 proteins. In Xenopus laevis, RB neuron survival rate was significantly increased by overexpression of Bcl-XL, one of the anti-apoptotic bcl-2 protein, in the nervous system (Coen et al. 2001). In cultured mammalian cell, the expression level of Bcl-XL was reduced by depression of cdk5 activity (Brinkkoetter et al. 2009). Thus, the apoptosis of RB neuron might be regulated by the cdk5-Bcl-XL regulated mitochondria pathway.

Voltage-gated sodium channel (Nav) mediated electrical activity is also likely required for the initiation of RB neuronal apoptosis, because pharmacological blockade of the Nav current reduced RB neuron death (Svoboda et al. 2001). This hypothesis is supported by the observation that RB neuron death was also reduced in a mutant whose RB neurons cannot generate action potential due to loss of Nav on the surface (Nakano et al. 2010). Nav is a large protein (~260 kDa) with 24 transmembrane domains and is responsible for the rising phase of the action potential in all neurons and muscles (Cantrell and Catterall 2001). Among nine Nav subtypes (Nav1.1–1.9), RB neuron express Nav1.1 and Nav1.6 (Novak et al. 2006). The Nav1.6 encoded by the scn8aa gene is the dominant subtype in RB neurons, as the morphants and mutants of scn8aa show a significant reduction in the peak amplitude of Nav in RB neurons and are unresponsive to tactile stimuli (Pineda et al. 2006; Low et al. 2010b). Antisense morpholino-mediated knockdown of the Nav1.6 also reduced the RB neuron death (Pineda et al. 2006). However, the relationship between the mitochondria pathway and Nav1.6 mediated electrical activity has not been elucidated yet.

Many studies including the above-mentioned have argued that most, if not all, RB neurons are lost by 4 dpf. However, recent studies revealed that 30–40% of RB neurons that express PKCα persist until at least 16 dpf (Slatter et al. 2005; Patten et al. 2007; Palanca et al. 2013). The existence of the long-lived RB neurons suggests that the removing of RB neurons in zebrafish may be much more of a restricted phenomenon than previously thought.

7 Mutants That Contributed to Investigate Nav Synthesis and Transport Mechanisms

As mentioned above, RB neurons require Nav to generate action potential. In excitable cells, such as neuron and muscle, Navs are transported to the specific regions of the cell membrane to form clusters, after the synthesis in the endoplasmic reticulum (ER). The Nav clusters have been identified in the axon initial segment and nodes of Ranvier of neurons. These clusters are molecular machinery for the generation and propagation of action potentials. Although the molecular basis of the Nav clustering at these sites has been extensively investigated (Rasband 2010), the molecular mechanisms that govern synthesis and transport of Nav has been less explored.

Six mutant zebrafish with reduced touch response have been identified in a large mutagenesis screening (Granato et al. 1996). RB neurons of four mutants of them (alligator, macho, steiffer, crocodile) showed reduction in Nav current amplitude, leading to a defect in the generation of action potentials and becoming touch-insensitive (Ribera and Nüsslein-Volhard 1998; Carmean and Ribera 2010). At present, responsible genes of alligator and macho have been identified (Carmean et al. 2015) (Ogino et al. 2015). Detailed analysis of these mutant provided insights into the molecular mechanisms of Nav synthesis and transport.

The mutation responsible for the alligator mutant is a nonsense mutation at leucine 39 (L39X) in a gene encoding the really interesting new gene (RING), finger protein 121 (RNF121) (Ogino et al. 2015). The nonsense mutation resulted in premature stop codon before the first transmembrane domain, thus the mutated gene appeared to be a null allele. RNF 121 is an E3-ubiquitin ligase with 6-transmembrane domain present in the endoplasmic reticulum (ER) and cis-Golgi compartments, and the catalytic activity is involved in ER-associated degradation (ERAD) (Darom et al. 2010; Araki and Nagata 2011). Some of the nascent protein that is synthesized in the ER would be misfolded or unfolded. In healthy cells, the inadequately folded proteins are removed by ERAD to prevent accumulation of the aberrant protein. In RB neurons of the mutant alligator, Navs accumulated in ER and cis-Golgi, instead of being transported to the cell surface, thus, RNF121 activity is essential for the proper intracellular trafficking of Nav in RB neurons, probably through removing of the misfolded Nav. Therefore, the study in alligator mutant indicated that RNF121 participates in the quality control of Nav during their synthesis (Ogino et al. 2015).

A missense mutation in the start codon of the pigk gene has been identified as the mutation responsible for macho mutant (Carmean et al. 2015). The missense mutation would result in a loss of function of the PigK protein, which is involved in addition of glycophospatidylinositol (GPI) to immature protein in ER. The GPI residue serves as an anchor for protein to bind to the cell membrane; thus, hypofunction of the complex that catalyzes GPI attachment leads to the mislocalization of GPI-anchored proteins. In addition to the macho mutant, the reduction of Nav current in RB neurons was recorded in a novel touch-insensitive mutant, mi310 (Nakano et al. 2010). In RB neurons of a mi310 mutant, the responsible mutation was identified in a gene encoding the PigU subunit of the GPI transamidase complex. This mutation was a missense mutation that abolishes the enzyme activity of PigU. These studies revealed that RB neurons require GPI transamidase activity for the proper Nav current and also for touch sensitivity. However, it remains to be elucidated how GPI transamidase affects to the Nav current, because Nav is not GPI-anchored protein. Unidentified GPI-anchored protein possibly participates to establish the proper Nav current.

8 Tools for Manipulating RB Neurons

Many genetic encoding tools have been introduced to zebrafish biology. As aforementioned, zebrafish embryos are suitable for in vivo imaging and manipulation of neural activity due to the transparency of their body.

Intracellular Ca2+ concentration transiently increases during the generation of action potential via Ca2+ influx through voltage-gated Ca2+ channels. The Ca2+ transients in electrically stimulated RB neurons were visualized in cameleon-expressing zebrafish as the first successfully genetically encoded calcium indicator (GECI) (Higashijima et al. 2003). Cameleon is a FRET (fluorescence resonance energy transfer)-based GECI in which two different fluorescent proteins are linked by calmodulin (CaM) and CaM binding M13 peptide (Miyawaki et al. 1997). Binding of Ca2+ to CaM induces conformational change that increases the FRET efficiency between the fluorescent proteins. Thus, increases in intracellular Ca2+ concentration are visualized as a change in the ratio of fluorescence intensities between these fluorescent proteins. Therefore, cameleon serves as a ratiometric calcium indicator. After the development of cameleon, Nakai et al. produced a high Ca2+ affinity GECI composed of a single GFP, named GCaMP (Nakai et al. 2001). GCaMPs are circularly permuted (cp) EGFP that are fused to CaM at its C terminus and the M13 peptide at its N terminus (Nakai et al. 2001). Their fluorescence intensity is enhanced through the conformational change that is induced by Ca2+ binding to the calmodulin domain; therefore, GCaMPs visualize the Ca2+ transient as an alteration of the fluorescent intensity. In many model animals including zebrafish, GCaMPs were widely used for measuring neuronal activity in large populations of neurons because of their high Ca2+ affinity, which results in a high signal to noise ratio (Akerboom et al. 2012; Muto et al. 2011; Muto et al. 2013; Marsden and Granato 2015; Warp et al. 2012). After their development, to improve sensitivity, GCaMP was mutated to produce variants (Ohkura et al. 2005; Tallini et al. 2006; Tian et al. 2009; Akerboom et al. 2012; Chen et al. 2013; Muto et al. 2011). Fluorescence changes in GCaMP7a, the latest version of GCaMP, are approximately 3.2-fold greater than those of GCaMP-HS (Muto et al. 2013). Using the improved GCaMP expressing zebrafish larvae, spontaneous neuronal activities were visualized in the tectum, habenula, and hindbrain at single-cell resolution (Muto et al. 2013). To expand the GFP-based single color palette, blue fluorescent GECI (B-GECO) and red fluorescent GECI (R-GECO) were developed through mutating the cpGFP and replacing the cpGFP with cpRFP, respectively (Zhao et al. 2011). These GECIs enabled multicolor Ca2+ imaging in zebrafish (Walker et al. 2013). Subcellular localization of GECIs can be modulated with the addition of a signal peptide or a protein that is localized to a relevant subcellular domain. Presynaptic terminal localized GCaMPs (SyGCaMPs) and SyRGECO were developed via fusion with synaptophysin, a transmembrane protein in synaptic vesicles (Dreosti et al. 2009; Nikolaou et al. 2012; Akerboom et al. 2012; Walker et al. 2013). Similarly, following the development of targeted GCaMPs, postsynaptic targeted GCaMP was developed via fusion of the postsynaptic matrix protein homer to the N terminus of GCaMP (Pech et al. 2015), and cell membrane targeted GCaMP was developed via fusion of the CAAX peptide to the C terminus of GCaMP (Tsai et al. 2014) or fusion of the LCK peptide to the N terminus of GCaMP (Shigetomi et al. 2010). In addition, mitochondrion-targeted GCaMP was developed via fusion of cytochrome C oxidase 4 to the N terminus of GCaMP (Park et al. 2010), and nuclear localized GCaMP was developed via fusion of histone H2B to the N terminus of GCaMP (Freeman et al. 2014) or fusion of a nuclear localization signal to the N terminus or C terminus of GCaMP (Kim et al. 2014). The combination of these targeting techniques and the multi-color palette enable multi-color Ca2+ imaging in a single cell.

Channelrhodopsin-2 (ChR2) and halorhodopsin (NpHR) enabled light-stimulated non-invasive manipulation of neural activity in zebrafish larvae. ChR2 is a light-gated channel that permeates a wide range of mono- and divalent cations, upon blue-light absorption. This cation flow depolarizes the membrane potential with high temporal precision, as the ChR2-mediated depolarizing currents reach a maximal rise rate within 2.3 ± 1.1 m sec after a blue-light pulse onset (Boyden et al. 2005). Single action potentials are reliably elicited in ChR2 expressing neurons after the light emission (Boyden et al. 2005). The first application of ChR2 in zebrafish was an experiment in which ChR2 was expressed in trigeminal and RB neurons under the control of the isl1 promoter (Douglass et al. 2008). After stimulating the somatosensory neurons using a standard dissecting microscope and light pulses, single spikes from a single somatosensory neuron could drive escape behavior. Recently, Umeda et al. (2016) identified a novel escape response that is induced after single RB neuron excitation using an improved ChR2, ChRWR (channelrhodopsin-wide receiver), and laser microscopy to perform more precise single cell stimulation (Umeda et al. 2016). ChRWR is a chimeric protein of ChR1 and ChR2, and this improved ChR2 has a higher efficiency in plasma-membrane expression and photocurrents with little desensitization (Wang et al. 2009). NpHR is a fast light-activated chloride pump, thus NpHR can reversibly silence neural firing, contrast to ChR2. For reliable silencing, substantial membrane expression of NpHR is required. However, NpHR forms aggregates that are toxic to cells at high expression levels. This toxicity may result from the retention of NpHR in the ER (Gradinaru et al. 2007; Zhao et al. 2008). In addition, analysis of the amino acid sequence of NpHR revealed that it does not contain a typical signal peptide sequence (Zhao et al. 2008). To overcome this problem, the membrane expression efficiency of NpHR was improved through replacing the first 27 amino acids of NpHR with the signal peptide of the β subunit of nAchR and adding an ER export signal from Kir2.1 to the C terminus of NpHR (Gradinaru et al. 2007; Zhao et al. 2008). The new enhanced NpHR (eNpHR) does not form aggregates and its membrane expression level is dramatically increased (Gradinaru et al. 2007; Zhao et al. 2008). An eNpHR expressing transgenic zebrafish line has been established, and neural firing in this model is effectively suppressed by light stimulation (Arrenberg et al. 2009). In addition, using this transgenic line in combination with ChR2 identified swim command neurons in the zebrafish hindbrain. The swimming behaviors that were induced in response to stimulating ChR2 expressing hindbrain neurons were blocked via the activation of NpHR in the same hindbrain neurons (Arrenberg et al. 2009).

Kaede is a photo-convertible fluorescent protein that changes from green to red after irradiation with ultra violet (UV) or violet light (Ando et al. 2002). Therefore, specific Kaede-expressing cells can be labeled after photo-conversion. Additionally, cells can be labeled with different colors after differentiation. UV irradiation to Kaede expressing cells at early stages of development labels early differentiated cells with red fluorescence, while only more recently differentiated cells are green.

9 Conclusion

Genetic manipulation using forward-genetics approaches have generated many mutant zebrafish lines, some of which are useful as human disease models. In zebrafish larvae, RB neurons are easily identifiable, and their relatively large cell body is suitable for electrophysiological recording of neural activity. Moreover, RB neurons are an attractive model for investigating the molecular mechanisms of axonal elongation. Many enhancer/promoter sequences that drive gene expression in all or a subset of RB neurons have been isolated (Table 4.1). With these enhancer/promoter sequences, any gene of interest can be expressed in RB neurons to investigate its function. Moreover, the function of the genes of interest can be investigated in human disease models. Zebrafish are a highly fertile animal and their larvae are easily drug-treatable via the addition of a compound into the breeding media. In this manner, experiments can be performed to investigate interactions among a diseased state and gene activity, the pharmacological effects of drugs, or new therapeutic agents using high throughput screening. The activation of RB neurons evokes a stereotyped escape response; thus, RB neuronal activity can be explored as a behavioral response. As a result, the RB neurons of zebrafish are an attractive experimental model. However, by 4 dpf, most RB neurons undergo apoptosis; therefore, their short lifespan may be disadvantageous for some experiments. However, long-lived RB neurons that express PKCα may be investigated to overcome this disadvantage. Experiments using the RB neurons of mutant zebrafish have revealed the molecular mechanisms that are essential for the proper intracellular trafficking of nascent protein. The RB neuron experimental model will continue to contribute insight for both basic and applied science.