Division of Molecular and Developmental Biology, National Institute of Genetics, Department of Genetics, Graduate University for Advanced Studies (SOKENDAI), Mishima, Shizuoka, Japan
Division of Molecular and Developmental Biology, National Institute of Genetics, Department of Genetics, Graduate University for Advanced Studies (SOKENDAI), 1111 Yata, Mishima, Shizuoka 411-8540, Japan
The developing nervous system comprises a variety of neuronal types, which confers a considerable complexity in structure and function on neural networks (Jessell,2000; Guthrie,2007). Because such complexity arises, at least in part, from a hierarchical interplay among stimulatory and inhibitory transcription factors that impose developmental restrictions on progenitor cells, the study of spatial and temporal expression patterns of such transcription factors has provided important insight into the sub-structures of the developing nervous systems (Jessell,2000). Among vertebrate model organisms, the zebrafish (Danio rerio) possesses several advantages for the visualization of specific neuronal types. First, its body is transparent during embryonic and larval development, and specific neuronal types are thus easily visualized and identified by their position and morphology. Second, it is possible to construct transgenic fish that express marker genes in specific neuronal types under the control of cis-regulatory elements of neuronal type-specific transcription factors (Higashijima et al.,2000).
The mnx class of homeodomain protein genes have been shown to encode transcription factors that regulate the differentiation of many classes of motor neurons, including the spinal motor neurons and the abducens motor neurons (i.e. the VIth cranial nerve; Tanabe et al.,1998; Arber et al.,1999; Thaler et al.,1999). The cis-regulatory elements of the mnx1/Hb9 gene has been identified and used for labeling motor neurons in the zebrafish, chick, and mouse (Arber et al.,1999; Lee et al.,2004; Flanagan-Steet et al.,2005; Nakano et al.,2005; Boon et al.,2009; Zelenchuk and Bruses,2011). In the zebrafish, transgenic lines carrying the 3 kbp upstream sequence of mnx1/hlxb9 linked to a fluorescent marker protein gene have been developed for the study of the mechanisms that control the axonal guidance of spinal motor neurons and the formation of neuromuscular junctions (Flanagan-Steet et al.,2005; Jing et al.,2009). The use of the mnx1 transgenic fish, however, is limited mainly to the study of early developmental stages, because in this transgenic fish green fluorescent protein (GFP) is expressed in various spinal neurons in addition to motor neurons in the later developmental stages (Zelenchuk and Bruses,2011).
In the present study, we aimed to isolate transgenic markers of specific neuronal types by enhancer trapping. Enhancer trapping is an effective approach to the development of transgenic zebrafish in which specific cell types are labeled (Parinov et al.,2004; Ellingsen et al.,2005; Nagayoshi et al.,2008). By using the in vivoTol2 transposition system (Urasaki et al.,2008), we established the transgenic line HGj4A, which carries an enhancer trap construct in the vicinity of the mnr2b/hlxb9lb gene, which encodes a transcription factor belonging to the mnx class of homeodomain proteins (Wendik et al.,2004). We report here the characterization of HGj4A transgenic fish expressing GFP in the spinal and abducens motor neurons.
RESULTS AND DISCUSSION
To obtain transgenic zebrafish in which specific neuronal types are visualized, we performed an enhancer trap screen by using the in vivoTol2 transposition system (Urasaki et al.,2008). Adult male fish carrying both the donor transposon insertion HG6D, which directs GFP expression in the notochord (Nagayoshi et al.,2008), and the Fuji transgene, which encodes the cDNA for the Tol2 transposase downstream of the hsp70 promoter, were treated once with a heat shock to induce transposition and then crossed with wild-type females. By screening the resulting F1 embryos, we identified an embryo exhibiting a distinct GFP expression pattern in the spinal cord, but not in the notochord, at 22 hours postfertilization (hpf; Fig. 1B). Due to the high degree of mosaicism present in the germ line of the heat-shocked founder male (Urasaki et al.,2008), we identified only one F1 embryo with the same GFP expression pattern in the spinal cord out of 924 F1 embryos from the founder male fish. Therefore, the single GFP-positive embryo had to be carefully raised to adulthood to obtain the F2 progeny.
We then analyzed the fin clip DNA of the F1 fish by Southern blot hybridization and identified a single transposon insertion that was responsible for the GFP expression pattern. The insertion and the transgenic fish were both designated HGj4A. We performed linker-mediated polymerase chain reaction (PCR), and found that the HGj4A insertion was located 2,154 bp upstream of the coding sequence of the mnr2b gene (Fig. 1A). To determine the transcription start site of mnr2b, 5′ rapid amplification of cDNA ends (RACE) analysis was performed. The 5′RACE products contained the 258 bp 5′ untranslated region (UTR) sequence of mnr2b, which was longer than the previously annotated 5′UTR of mnr2b (60 bp; Wendik et al.,2004; Fig. 1A).
To characterize the GFP expression pattern in the HGj4A line in more detail, we observed the HGj4A embryos by confocal microscopy. We found that, at 22 hpf, the GFP signal was restricted to the ventral side of the spinal cord (Fig. 1B). We crossed the HGj4A fish with SAIG213A;UAS:RFP double transgenic fish, which expresses red fluorescent protein (RFP) in the caudal primary motor neurons (CaP; Muto et al.,2011). In the HGj4A; SAIG213A;UAS:RFP triple transgenic embryos, the GFP and RFP signals were overlapped in the CaP motor axon extending ventrally, indicating that the HGj4A embryo expressed GFP in the CaP motor neurons (Fig. 1E–G). Zebrafish embryos have another type of primary motor neurons, middle primary neurons (MiP), which exhibit axonal extension to the dorsal skeletal muscles. In the HGj4A embryo, the GFP signal was also detected in MiP neurons (Fig. 1E). In addition, at 22 hpf, the HGj4A embryos expressed GFP in the pharyngeal and gut endodermal cells (Fig. 1C–E). As judged from co-labeling using the Tg(gata1:mRFP) fish (Kitaguchi et al.,2009), GFP signal was also detected in the intermediate cell mass (Fig. 1H–J). Furthermore, GFP expression was also detected in the tail bud cells (Fig. 1B). The expression pattern of GFP in the HGj4A embryos was consistent with that of the mnr2b gene identified by mRNA in situ hybridization (Wendik et al.,2004), indicating that HGj4A enhancer trap insertion captured the activity of the mnr2b enhancers.
GFP expression in the spinal cord was observable beyond the early embryonic stage. Until 5 days postfertilization (dpf), GFP-positive cells were densely aligned on the ventral side of the spinal cord in HGj4A larvae. The axons emanating from these GFP-positive cells and innervating the skeletal muscles were clearly observable from 48 hpf to 5 dpf (Fig. 2A–D), indicating that these GFP-positive cells include the spinal motor neurons. In addition, a small population of cells that was slightly dorsal to the dense GFP-positive population was sparsely labeled with GFP in HGj4A embryos and larvae (Fig. 2C,D). We found that these sparse GFP-positive cells also included the motor neuron, as judged from the single-cell labeling of the spinal motor neuron (Supp. Fig. S1, which is available online). To test whether the GFP-positive cells contain other types of neurons than the spinal motor neurons, we performed several double staining experiments. First, to examine whether the GFP-positive cells include glutamatergic neurons, we crossed HGj4A fish with Tg(vglut2a: loxPDsRed-GFP) line (Kimura et al.,2006; Bae et al.,2009) and investigated the resulting HGj4A; Tg(vglut2a: loxPDsRed-GFP) double transgenic offspring. We found that, in the double transgenic offspring, the GFP signal did not overlap with the red fluorescence at 48 hpf and at 5 dpf (Fig. 2E,F; 300 GFP-positive cells from five embryos and data not shown), indicating that the GFP-positive neurons are not glutamatergic. Next, to ask whether the GFP-positive cells include inhibitory interneuron, we performed immunohistochemistry for Pax2, which is expressed in most glycinergic and many GABAergic spinal interneurons (Batista and Lewis,2008). We found that the strong Pax2 signal was restricted within the dorsal half of the spinal cord and did not overlap with GFP-positive cells at 48 hpf (Fig. 2G,H, 62 Pax2+ cells from five embryos), suggesting that the most GFP-positive cells are not inhibitory interneurons. Consistent with this observation, the GFP-positive cells hardly overlapped with the fluorescent marker expressed by the cis-regulatory element of the glycine transporter-2 gene (Supp. Fig. S2). Altogether, these observations suggest that most, if not all, GFP-positive neurons in the spinal cord are the motor neurons in HGj4A embryos and larvae. Aside from the spinal cord, strong GFP expression was also observed in the lens and posterior gut (Fig. 2B). A relatively weaker GFP signal was observed in the heart and skeletal muscles (Fig. 2B), which is most likely due to the basal activity of the hsp70 promoter in the enhancer trap construct (Nagayoshi et al.,2008).
In addition to the GFP expression in the spinal cord, by 48 hpf, two distinct bilateral clusters were found to express GFP at the midline in the rhombomere 5 (r5) and r6 in the hindbrain (Fig. 3A). From these GFP-positive clusters, the axon-like processes proceeded first ventrolaterally, and then turned in an anterior direction (Fig. 3A). We hypothesized that the GFP-positive clusters included cranial nerve VI (the abducens motor neurons), which is known to locate in the r5 and r6 regions, and innervates the lateral rectus muscles, which control abduction of the eye. To confirm this, we tested whether the axon-like GFP-positive processes in the HGj4A larvae form neuromuscular junction with the lateral rectus muscles. For this purpose, we searched our collection of gene trap and enhancer trap fish in our laboratory (Kawakami et al.,2010) and identified hspGFFDMC131B fish, which carries a single copy insertion of a Gal4FF enhancer trap construct and expresses the Gal4FF transcription factor (Asakawa et al.,2008) in the extraocular muscles, including the lateral rectus muscle (Fig. 3C). We crossed the hspGFFDMC131B fish with UAS:RFP reporter fish, and then analyzed RFP expression in the hspGFFDMC131B;UAS:RFP double transgenic embryos. While in the double transgenic larvae the RFP signal was not obvious in the primordium of the lateral rectus muscle before 66 hpf (Easter and Nicola,1996; data not shown), the RFP expression became evident in the fully extended lateral rectus muscle at 72 hpf (Fig. 3C) and persisted at least until day7 (data not shown). We then crossed the hspGFFDMC131B;UAS:RFP fish with the HGj4A fish and analyzed the GFP and RFP expression. We found that the GFP-positive axon-like processes projected to the RFP-positive lateral rectus muscles by 72 hpf (Fig. 3B,C). The GFP-positive processes extensively contacted with the lateral rectus muscles mainly at the region 30–75 μm away from the medial end of the lateral rectus muscles (Fig. 3D), where a large clusters of acetylcholine receptors (AChRs) were detected (Fig. 3E,F), indicating that the neuromuscular junctions were formed. The punctate AChR clusters were also detected on more lateral side of the lateral rectus muscles (Fig. 3E,F, brackets). From these observations, we concluded that the GFP-positive clusters in the hindbrain include the abducens motor neurons. Between 48 and 72 hpf, the GFP-positive cluster in each hemi-segment at r5 or r6 began to segregate dorsoventrally, and it formed two essentially distinct clusters along the dorsoventral axis by 120 hpf (Fig. 3G–I). We also noted that the ventral population of these GFP-positive clusters had dendrites which were extensively developed toward the lateral side, while the dorsal clusters did not have such dendritic structures (Fig. 3J,K), suggesting that the GFP-positive clusters in the hindbrain contain several distinct cell types.
By enhancer trapping, we successfully established the HGj4A line, which labels the spinal motor neurons during the embryonic and larval stage. As the HGj4A line does not contain the Gal4 gene, this line is useful for monitoring spinal motor neurons in combination with the Gal4-UAS system, which can label the Gal4-expressing cells with a reporter distinguishable from GFP fluorescence, as we demonstrated in the present study. Another advantage of the HGj4A line is visualization of the abducens motor neurons. Among the currently available transgenic markers for the abducens motor neurons, the olig2 enhance/promoter is used to express a reporter gene, but innervation of the lateral rectus muscles by the abducens motor neurons has not been demonstrated anatomically (Shin et al.,2003; Kucenas et al.,2008; Schoonheim et al.,2010). We have demonstrated here that the abducens motor neurons come into contact with the lateral rectus muscles by 72 hpf. This timing is in good correlation with the observation that the time period of the developmental onset of the optkinetic response in zebrafish is 73 hpf (Easter and Nicola,1996). Thus, we propose that the transgenic zebrafish we described here, the hspGFFDMC131B and HGj4A lines, would provide a powerful system for the elucidation of the cellular and molecular mechanisms of axonal guidance and neuromuscular junction formation between the abducens motor neurons and the lateral rectus muscles. An intriguing observation is that each GFP cluster in the hindbrain hemisphere segregates dorsoventrally during the second day of development and the cells in these segregated sub-clusters exhibit a distinct morphology. Identification of the cell types included in these sub-clusters should provide novel insights into the development of the neural circuits controlling eye movement by means of the abducens nerve. In conclusion, the present study demonstrated that enhancer trapping in zebrafish is a powerful approach for dissecting the sub-structures of the nervous system in vertebrates. We are generating more transgenic markers for specific neuronal types in our ongoing enhancer trap and gene trap screens.
Southern Blot Hybridization, Linker-Mediated PCR, and 5′ RACE Analysis
Southern blot hybridization, inverse PCR, and linker-mediated PCR were carried out as described previously (Asakawa and Kawakami,2009). Either 32P-labeled enhanced GFP (EGFP) or Gal4FF probes were used to detect the HGj4A insertion and the hspGFFDMC131B insertion, respectively. To perform 5′RACE, total RNA was prepared from 50 GFP-positive embryos at 24 hpf by homogenizing the embryos in 1 ml of Trizol Reagent (Life Technologies). Five micrograms of the total RNA were used for cDNA synthesis using a primer in the mnr2b coding sequence mnr2b-r1 (5′- CAA TGC TGA AAT GGG GTA CAT-3′) and the 5′RACE System for Rapid Amplification of cDNA Ends version 2.0 (Invitrogen). Using the cDNA as a template, 35 cycles of a first round of PCR (94°C for 30 sec; 55°C for 30 sec; 72°C for 2 min) was carried out with the primers mnr2b-r2 (5′-GTA CAT GCC TGG GAT AGA AGT GAG-3′) and Abridged Anchor Primer (AAP: 5′-GGC CAC GCG TCG ACT AGT ACG GGI IGG GII GGG IIG-3′) followed by a second round of PCR using mnr2b-r3 (5′-CCT GAA GTT CTT TGA CTT TTC CAT-3′) and Abridged Universal Amplification Primer (AUAP: 5′-GGC CAC GCG TCG ACT AGT AC-3′). The 5′RACE products were cloned with a TA cloning kit (Invitrogen) and sequenced.
Immunohistochemistry and AChR Staining
For Pax2 staining, HGj4A embryos at 48 hpf were dechorionated and fixed for 2 hr with PBS (pH 7.4) containing 4.0% paraformaldehyde. Rabbit Pax-2 polyclonal antibody (1/300, COVANCE, PRB-276P) and Alexa Fluor 633 (1/1,000) were used as primary and secondary antibody, respectively. For AChR staining, HGj4A larvae at 120 hpf were fixed for 3 hr with PBS (pH 7.4) containing 4.0% paraformaldehyde. The larvae were treated with rabbit polyclonal anti-GFP antibody (1:500 dilution; Invitrogen A6455), then with a mixture of goat anti-rabbit Alexa Fluor 488 (1:1,000, Molecular Probes) and α-Bungarotoxin Alexa Fluor 647 conjugate (10 μg/ml, B35450, Molecular Probes). For simultaneous staining of AChR and the lateral rectus muscle, hspGFFDMC131B; UAS:RFP double transgenic larvae at 120 hpf were treated as above, except that rat anti-RFP antibody (5F8, chromotek) and goat anti-rat Alexa Fluor 546 (1:1,000, Molecular Probes) were used instead as primary and secondary antibody, respectively.
BAC Transgenic Line
The transgenic line Tg(vglut2a: loxPDsRed-GFP) was constructed using a bacterial artificial chromosome (DKEY-145P24) by a previously described method (Kimura et al.,2006). The details for generating this line will be described elsewhere.
Gal4FF Enhancer Trap Screening
The enhancer trap screening was performed as described (Asakawa et al.,2008; Asakawa and Kawakami,2009). The enhancer trap construct T2KhspGFFDMC contains a 638-bp DNA fragment of the hsp70 promoter region, the Gal4FF gene and the SV40 polyA signal between the essential cis-sequences of Tol2. The detailed structure of T2KhspGFFDMC will be described elsewhere. A plasmid DNA carrying T2KhspGFFDMC was injected into fertilized eggs with Tol2 transposase mRNA synthesized in vitro. The injected fish were raised to adulthood and crossed with homozygous UAS:GFP reporter fish. One hundred thirty-one injected fish were examined and in total 89 GFP expression patterns were identified in the F1 offspring. The F1 fish that displayed GFP in the lateral rectus muscle were analyzed by Southern bolt hybridization, and the fish that carried a single insertion were identified and designated hspGFFDMC131B.
To observe the GFP signals, we used male fish carrying the HGj4A insertion as the parental fish, because female HGj4A fish lay eggs with maternally-expressed GFP due to hsp70 promoter activity (Nagayoshi et al.,2008). A fluorescence stereomicroscope (MZ16FA, Leica) equipped with a CCD camera (DFC300FX, Leica) was used to observe and take images of GFP-expressing embryos. A live embryo or larva was embedded in 1% low-melting agarose (NuSieve GTG Agarose, Lonza) on a Glass Base dish (IWAKI, 3010-035) and subjected to confocal microscopy using an Olympus FV-1000D laser confocal microscope. Images of live embryos and larvae were acquired as serial sections along the z-axis and processed using the Olympus Fluoview Ver2.1b Viewer and Adobe Photoshop CS5.
We thank A. Urasaki for the isolation and initial characterization of HGj4A, Dr. Y. Kimura for Tg(vglut2a: loxPDsRed-GFP), and A. Ito, M. Suzuki, M. Mizushina, N. Mouri, and Y. Kanebako for fish maintenance. K.A. and K.K. were supported by Grants-in-Aid for Scientific Research, K.A. was funded by The Uehara Memorial Foundation and The Kao Foundation for Arts and Sciences, and K.K. was funded by the National BioResource Project from the Ministry of education, Culture, Sports, Science, and Technology of Japan.