Molecular characterization and embryonic expression of the family of N-methyl-D-aspartate receptor subunit genes in the zebrafish
Abstract
We present the cloning of 10 N-methyl-D-aspartate (NMDA) receptor subunits from the zebrafish. These subunits fall into five subtypes, each containing two paralogous genes. Thus, we report two NMDAR1 genes (NR1.1 and NR1.2), and eight NMDAR2 genes, designated NR2A.1 and NR2A.2, NR2B.1 and NR2B.2, NR2C.1 and NR2C.2, and NR2D.1 and NR2D.2. The predicted sequences of the NR1 paralogs display 90% identity to the human protein. The NR2 subunits show less identity, differing most at the N- and C-termini. The NR1 genes are both expressed embryonically, although in a nonidentical manner. NR1.1 is found in brain, retina, and spinal cord at 24 hours postfertilization (hpf). NR1.2 is expressed in the brain at 48 hpf but not in the spinal cord. NR2 developmental gene expression varies: both paralogs of the NR2A are expressed at 48 hpf in the retina, only one paralog of the NR2B is expressed at low levels in the heart at 48 hpf. Neither of the NR2C is expressed embryonically. Both paralogs of the NR2D are expressed: 2D.1 is in the forebrain, retina, and spinal cord at 24 hpf, whereas the 2D.2 is only found in the retina. Our findings demonstrate that the zebrafish can serve as a useful model system for investigating the role of NMDA receptors in the development of the nervous system. Developmental Dynamics 234:756–766, 2005. © 2005 Wiley-Liss, Inc.
INTRODUCTION
Glutamate is the main excitatory neurotransmitter in vertebrates, and its effects are mediated by means of two classes of receptors: G-coupled receptors (metabotropic glutamate receptors) and ligand-gated ion channels (ionotropic receptors). The latter class is responsible for fast glutamatergic signaling in the nervous system and is further subdivided in three main types according to their selective ligand: AMPA, kainate, or N-methyl-D-aspartate (NMDA; Monaghan et al., 1989). These receptors are distributed throughout the central nervous system (CNS) and have been implicated in neuronal plasticity during development (Collingridge, 1987; Krishtal et al., 1989; Debanne et al., 2003). The mammalian NMDA receptor was first cloned in 1991 (Moriyoshi et al., 1991), and its structure and function has been studied widely in mammals (Riedel et al., 2003; Mayer and Armstrong, 2004). These receptors are highly permeable to calcium and, thus, may play important regulatory roles in the response of neurons to signaling.
There are five NMDA receptor genes expressed in mammals: one encoding an NMDAR1 subunit (NR1; Moriyoshi et al., 1991) and four encoding four NR2 subunits (Ishii et al., 1993). The NR1 subunit has a widespread distribution throughout the CNS, whereas the NR2 subunits exhibit distinct regional and cell-specific expression patterns. The NR1 subunit carries the binding site for the co-agonist glycine (Danysz et al., 1998), and the asparagine residue involved in the voltage-dependent magnesium block is located in the second transmembrane domain on all subunits (Burnashev et al., 1992).
Previous studies have shown that members of the NMDA receptor family are expressed in the nervous system during development and that some subunits undergo a switching of expression as the nervous system matures (Monyer et al., 1994; Wenzel et al., 1996, 1997). The study of the developmental expression and function of these receptors in mammals has been somewhat limited to static investigation due to the in utero embryogenesis of these organisms.
We are using the zebrafish, Danio rerio, to study the role of NMDA receptors in neuronal development. Zebrafish have many advantages as a model for the study of neuronal development and the role of receptors in the formation of neuronal circuitry: they provide large numbers of offspring on a regular basis, they develop externally, which allows easy manipulation of the embryo, and the embryos are transparent for the first few weeks of life. Their development can be followed from the one-cell stage to a free swimming fish at 1 week or so and beyond (Kimmel, 1989), whereas in mammals, these developmental stages occur in utero and are difficult to study. Thus, Danio rerio (zebrafish) has become an exciting and extremely useful model for both static and dynamic studies of the developing vertebrate nervous system (Chitnis and Kuwada, 1990; Kimmel, 1993; Chandrasekhar et al., 1997). We, therefore, wish to exploit the utility of this organism as a model system for studying the role(s) of NMDA receptors in nervous system development. The first step in this process is presented in this report—the identification and characterization of the subunits making up the zebrafish NMDA receptor subunit gene family.
RESULTS AND DISCUSSION
Cloning of D. rerio NMDA Subunit cDNAs
A degenerate polymerase chain reaction (PCR) cloning approach was used to identify potential zebrafish NMDA receptor subunits expressed in either 4-day-old (larval) or adult zebrafish. Using degenerate PCR primers designed from highly conserved regions of the rat NMDA receptors, two separate reactions were carried out: the first was designed to identify cDNAs encoding the NR1 subunit, and the second was to obtain cDNAs corresponding to the NR2 subunits. From the first set of reactions, we were able to isolate two distinct cDNA fragments that encoded the same amino acid sequences but contained multiple differences in codon usage, suggesting that two unique genes for NR1 (NR1.1 and NR1.2) were expressed in the zebrafish. From the second set of reactions, we isolated partial cDNAs that fell into four obvious classes of NR2 receptor subunits that, when aligned with the rat and human NR2 proteins, supported their identification as the zebrafish orthologs of the mammalian NR2A, NR2B, NR2C, and NR2D subunits.
The partial amino acid sequences of the six NMDA subunit cDNAs were then used to search the Sanger Institute Danio rerio sequencing project database (http://www.sanger.ac.uk/Projects/D_rerio/). By identifying genomic fragments containing contiguous exons, we were able to determine that there were a total of 10 genes encoding NMDA subunits in the zebrafish. In this group of 10 were 2 paralogous genes for NR1, corresponding to the NR1.1 and NR1.2 cDNAs we had cloned, as well as eight genomic sequences that encoded members of the NR2 subunit family. By using virtual assembly of the predicted amino acid sequences for the NR2 subunits, it became clear that two paralogous genes were present for each NR2 subunit subtype. The four cDNAs obtained from the PCR were named NR2A.1, NR2B.1, NR2C.1, and NR2D.1. The remaining four genes, which were not detected in our initial PCR screen, were designated NR2A.2, NR2B.2, NR2C.2, and NR2D.2. To verify that the latter set of genes were indeed expressed in vivo, we designed gene-specific primers based on the database sequences and carried out PCR. We were able to obtain fragments corresponding to all four predicted subunits from adult but not larval fish. These results suggest that the paralogs were differentially regulated during development of the nervous system and were most likely not pseudogenes. This proposition was further supported by the results from in situ hybridization studies that are presented later in this report.
NR1 Subunit Sequence Is Highly Conserved Across Zebrafish and Humans
The complete cDNAs of NR1.1 and NR1.2 were obtained by rapid amplification of cDNA ends (RACE) PCR using primers designed from both the previous PCR-based data as well as the Sanger database sequences. The resulting full-length coding regions both encode proteins of 933 amino acids in length and are shown in Figure 1. As can be seen, the two zebrafish NMDA receptor 1 subunits are approximately 90% identical to the human protein. The region from amino acid 486 to the C-terminus shows the highest degree of identity between the two paralogs and between the paralogs and the human protein; this region contains the four putative transmembrane regions as well as the Mg++ binding domain. This finding argues that there is a low tolerance for even conserved amino acid changes in this portion of the molecule, given that fish and mammals are presumed to have diverged over hundreds of millions of years ago. The N-terminus region, which is proposed to lie extracellularly, is more divergent but is still more than 60% identical. Previous work from many investigators has shown that NMDA receptor complexes in mammals interact with various PDZ binding proteins and that such interactions play important regulatory roles (Kornau et al., 1995; Niethammer et al., 1996). In mammals, PDZ-ligand domains are not encoded in the majority of NR1 receptor transcripts, but rather by the NR2 subunits; thus, NR complexes rely on the NR2 subunits for protein–protein interactions by means of these motifs. In contrast, the overwhelming majority of NR1 transcripts in the zebrafish encode SVSTVV at their C-termini, which is a canonical ligand for binding to the PDZ domain (Niethammer et al., 1996). This fact suggests that a substantial percentage of zNR1 subunits can interact directly with PDZ proteins; the functional significance of this finding is unclear at this time.
This figure shows the protein alignment for the most common N-methyl-D-aspartate (NMDA) R1 receptor splice variant from zebrafish (zNMDA-1.1 and zNMDA-1.2) and human (hNMDA-1). Homology is designated by gray shading. The transmembrane regions are marked by boxes (TMI–TMIV). The Mg++ binding domain is the asparagine in TMII, which is designated by a black asterisk. The C-terminal PDZ binding domain is underlined. This variant of the human receptor contains Exon 21 but lacks the PDZ binding domain.
Zebrafish Genome Contains Paralogs for Each of the Four NR2 Subunits
The deduced full-length sequences for all eight zNR2 subunits were aligned with the human proteins using the Clustal program (Jeanmougin et al., 1998), and Figure 2 shows the phylogenetic tree that was obtained from the analysis. This visualization allowed us to unambiguously assign the appropriate designations to the zNR2 subunits. The compete sequences for all eight NR2 subunits are shown in Figures 3 and 4 and are subdivided into two subsets based on overall amino acid identities. Figure 3 shows the alignment for the zebrafish NR2A and NR2C subunit proteins deduced from the Sanger Institute database with the human NR2A and NR2C. Both zN2C sequences show approximately 50% identity to the human receptor overall. The middle region of the protein contains the four putative transmembrane regions and the Mg++-binding site (marked with an asterisk) and is very conserved (90%) between the zebrafish and human. Most of the sequence divergence occurs in the N-terminus preceding TM1 and that portion of the protein C-terminal to the TM4.
Phylogenetic tree of zebrafish and human N-methyl-D-aspartate (NMDA) receptors. This tree was generated using Clustal X and was based on the multiple alignment output of all 10 full-length zebrafish proteins and five human subunits, with the tree calculated and bootstrapped using the neighbor-joining method. The numbers at the branch points are the bootstrap values from 1,000 iterations.
Predicted protein alignment for the NMDA 2A and 2C receptor proteins from zebrafish (zNMDA-2A.1, 2A.2, 2C.1, 2C.2) and human (hNMDA-2A, 2C). Conserved amino acids are designated by gray shading. The transmembrane regions are marked by boxes (TMI–TMIV). The Mg++ binding domain is the asparagine in TMII, which is designated by a black asterisk. Alignment was carried out using Clustal X, and GeneDoc was used to create homology shading (http://www.psc.edu/biomed/genedoc).
Predicted protein alignment for the N-methyl-D-aspartate (NMDA) 2B and 2D receptor proteins from zebrafish (zNMDA-2B.1, 2B.2, 2D.1, 2D.2) and human (hNMDA-2B, 2D). Conserved amino acids are designated by gray shading. The transmembrane regions are marked by boxes (TMI–TMIV). The Mg++ binding domain is the asparagine in TMII, which is designated by a black asterisk. Alignment was carried out using Clustal X, and GeneDoc was used to create homology shading (http://www.psc.edu/biomed/genedoc).
Figure 4 contains the multiple alignments of the zNR2B and zNR2D sequences together with their human orthologs. Parts of these sequences are based on Sanger database, because the 3′-end of the zN2D cDNAs could not be obtained by PCR. We were not able to isolate the 5′-end of the zN2B.1 receptor by PCR, and the sequence could not be found in the Sanger assembly database; the most 5′-sequence that was found lies in the middle of exon 4, and no overlapping sequence traces could be found to extend back to the start codon. The zNR2B subunits are approximately 50% identical overall with the human protein, with the same regions of similarity and divergence as the NR2As and NR2Cs. The NR2D subunits, however, only show approximately 35% overall identity with the human 2D (see Fig. 4) The main divergence here again appears to be in the C-terminus. The zebrafish receptors have a much longer C-terminus tail compared with the human, and this tail contains several sequence repeats. Similar repeats are found in both paralogs of the zN2D C-terminus; between amino acids 1300 and 1320, there is a stretch of serines in both zebrafish paralogs, which is not found in the human receptor. Also, between amino acids 1440 and 1500, there is a stretch of lysines interspersed with arginines, which again is not found in the human protein. All the zNR2 subunits contain a PDZ binding motif, similar to what is observed in mammals.
Zebrafish NR1 Subunits Undergo Alternative Splicing
There are eight splice variants of the mammalian NMDAR1, resulting from 5′-insertion and 3′-deletions and insertions (Moriyoshi et al., 1991; Sugihara et al., 1992; Ishii et al., 1993). We sought to determine whether the zebrafish also uses alternative splicing for these subunits, and if so, whether the exons involved were the same as those involved in the rat. In the rat, the two 5′-splice variants differ in whether they contain exon 5, which encodes a cassette of 21 amino acids that is inserted in the amino terminal domain of the protein at amino acid residue 190. Figure 5A shows the results of PCR using cDNA derived from 4-day-old zebrafish mRNA using primers designed against exon 4 of zNR1.1 and ZNR1.2 and a reverse primer against exon 6 sequence. These reactions each yielded two specific fragments, ∼800 bp and ∼700 bp, which correspond to two splice variants. Sequencing these products demonstrated that the larger one contained exon 5 and the smaller one did not. The two isoforms appear to be expressed at similar levels. The amino acid sequence for the zebrafish exon 5 is 90% identical to the rat (see Fig. 3). It has been postulated that this region is involved in inhibition of NR1 activity by extracellular Zn+ and this inhibition can be reversed by tyrosine kinase phosphorylation of the NR2 subunits (Traynelis et al., 1998). The high degree of sequence conservation of this exon between two such divergent species as zebrafish and humans certainly supports its playing a potential regulatory role in receptor function.
A: Polymerase chain reaction (PCR) reaction using cDNA from 4-day-old zebrafish. The forward primer lies in exon 4, and the reverse primer in exon 6 of zNMDA R1.1 and zNMDA R1.2 The larger band represents a splice variant, which contains the exon 5 insertion, and the smaller band the splice variant, which lacks this insertion. B: PCR reaction as in A. This strategy represents the C-terminal splice variants produced using a forward primer in exon 20 and a reverse primer in exon 22 of NR1.1 and NR1.2. The larger band represents the splice variant containing exon 21 and the smaller band lacks this exon. C: A diagram representing the three splice variants of zNR1.1, which we detected in zebrafish. ZNR1.2 displays the same splice variants.
Figure 5B shows the 3′-splice variants for zNR1.1 and zNR1.2, resulting from PCR using primers targeted against sequence in exon 17 and 3′ of the stop codon in the 3′-untranslated region. The PCR resulted in two fragments of 760 bp and 650 bp, and sequencing revealed that the size difference was due to the presence or absence of exon 21. Unlike in the rat, where the most abundant 3′-splice variant contains exon 21 but lacks a canonical PDZ binding domain at the C-terminus, both zebrafish 3′splice variants contain the PDZ binding domain. Figure 5C shows a diagrammatic representation of the various splice variants found in both paralogs of the NMDAR1. The splice variant lacking exon 21 appears to be considerably more abundant than the one with it, which agrees with the results found in another teleost, the knifefish (Apteronotus; Bottai et al., 1998). There were three different C-terminus splice variants described for the knifefish NR1: one lacking exon 21, one having a 25 amino acid insert/deletion between exon 20 and 21 (C1′), and another having a 9 amino acid insertion/deletion between C1′ and exon 21 (C1″). We did not obtain any clones containing the C1′-or C1″-variant, but we did find both these splice sequences in the Sanger database residing in the intron separating exon 20 and 21 of the NR1.1 gene. Surprisingly, we could not detect any such sequences in the NR1.2 gene. Interestingly, the Fugu rubripes (pufferfish) database also contains two NR1 genes, with these insertion/deletion sequences again present only in one of the paralogs (data not shown).
Whereas the zNR2 subunits do not exhibit as high overall identity with the human subunits as do the two zNR1s, all the subunits were approximately 80% identical through the region containing the putative transmembrane domains. The N-terminus region, which lies extracellularly and contains the ligand-binding site, is somewhat more conserved than the C terminus, which is thought to lie intracellularly. The only completely conserved amino acids in the C-terminus are the last six residues, which make up the PDZ ligand.
Chromosomal Localization of NMDA Receptor Subunit Genes
The genomic locations of the 10 subunit genes were mapped by radiation hybrid analysis using the Goodfellow T51 panel (Kwok et al., 1998). Table 1 lists the primers used for the radiation hybrid mapping. The results are presented in Table 2. The two NMDAR1 genes are located on two different linkage groups, LG10 and LG5, respectively, but only the NR1.2 has synteny with the human NR1 which resides on chromosome 9 (Barbazuk et al., 2000). The two paralogs of the NMDAR2A, 2B, 2C, and 2D each map to separate linkage groups, LG1 and LG3 (2A), LG3 and LG1 (2B), LG3 and LG12 (2C), and LG9 and LG19 (2D). The N2A.2, both N2C paralogs have synteny with the corresponding human genes. Neither of the N2B or N2D paralogs shows synteny with the human genes.
| PRIMER PAIR | SEQUENCE |
|---|---|
| ZN1.1.RH.F | 5′-CCAGAAGCGGTACGAGCAGGTGTT-3′ |
| ZN1.1.RH.R | 5′-GCGGTGTAGGAGACCGGAGTGGGC-3′ |
| ZN1.2.RH.F | 5′-TGCGCCTGGTTCTGTTCGCCTTTC-3′ |
| ZN1.2.RH.R | 5′-TTCAGTATGCTCCTGACCTGAACA-3′ |
| ZN2A.1.RH.F | 5′-TGGCGTCCATATCGATATGAAGAA-3′ |
| ZN2A.1.RH.R | 5′-GGTCATTGCTTTCATGCGGACTGA-3′ |
| ZN2A.2.RH.F | 5′-GTGTGGAACGGCATGGTGGGAGAG-3′ |
| ZN2A.2.RH.R | 5′-CCAGAGGAAAGCGAGTGTGAGAGA-3′ |
| ZN2B.1.RH.F | 5′-TCATCGAGACCGCGATCAGTGTGA-3′ |
| ZN2B.1.RH.R | 5′-AAATCATTGGGATTCTGAAACTTC-3′ |
| ZN2B.2.RH.F | 5′-AATCGCTGCTTGGCAGACGGACGA-3′ |
| ZN2B.2.RH.R | 5′-GCTTTGCATCATGGATCATGGAAA-3′ |
| ZN2C.1.RH.F | 5′-GGCTTCTACAAACGTGCAGATATG-3′ |
| ZN2C.1.RH.R | 5′-CCTTCTTGTCACTAAGCCCAGACA-3′ |
| ZN2C.2.RH.F | 5′-ACAGTCTCACCTTCTGCCTTCCTG-3′ |
| ZN2C.2.RH.R | 5′-AACAACTTTAACCGTTAAAAGCTA-3′ |
| ZN2D.1.RH.F | 5′-GAGCCGTACAGTCCAGCTGTGTGG-3′ |
| ZN2D.1.RH.R | 5′-ACTCACCCGACTAAAAGCCAGTAA-3′ |
| ZN2D.2.RH.F | 5′-CGAAGCCTACAGAGCGGCAAGAGT-3′ |
| ZN2D.2.RH.R | 5′-TGAGTCCTGACACAGTGTCGATGT-3′ |
| Zf cDNA | Gene | Zebrafish | Humanb | Syntenyc |
|---|---|---|---|---|
| N1.1 | GRIN1 | LG 10 | 9q34 | no |
| N1.2 | GRIN1 | LG 5 | 9q34 | YES |
| N2A.1 | GRIN2A | LG 1 | 16p13 | no |
| N2A.2 | GRIN2A | LG 3 | 16p13 | YES |
| N2B.1 | GRIN2B | LG 3 | 12p13 | no |
| N2B.2 | GRIN2B | LG 1 | 12p13 | no |
| N2C.1 | GRIN2C | LG 3 | 17q25 | YES |
| N2C.2 | GRIN2C | LG 12 | 17q25 | YES |
| N2D.1 | GRIN2D | LG 9 | 19q13 | no |
| N2D.2 | GRIN2D | LG 19 | 19q13 | no |
- a Positions determined from panel T51 and mapped with RHMapper from the Zon Lab.
- b From the Human Genome Project database at UCSC.
- c From Barbazuk et al. Genome Res. (2000) 10:1351–1358.
Not All NR Subunit Genes Are Expressed in the Embryonic Nervous System
The developmental expression of NMDA receptor subunit genes at 24 and 48 hours postfertilization (hpf) was examined by using whole-mount in situ hybridization coupled with chromogenic detection. The expression patterns of the two NR1 paralogs are shown in Figure 6. At 24 hpf, the NMDAR1.1 is strongly expressed in the fore-, mid-, and hindbrain, the retina, and the spinal cord. By 48 hpf, the expression is even stronger (Fig. 6A). The NMDAR1.2 is not expressed in the spinal cord at either time point but shows some expression in the forebrain at 24 hpf and is seen throughout the brain at 48 hpf, albeit at lower levels than NR1.1 (Fig. 6B). This developmental profile of expression is similar to that found in the rat, where at embryonic day (E) 17–E19, the NR1 subunit is expressed in the cortex, striatum, and spinal cord of the rat (Monyer et al., 1994) and increases in intensity in cortex, striatum, hippocampus, and cerebellum postnatally.
A: Gene expression of NR1.1 in the head and spinal cord of 24 hours postfertilization (hpf) and 48 hpf. At 24 hpf, NR1.1 is highly expressed in the fore-, mid-, and hindbrain; in the retina; and in the spinal cord. The 24 hpf spinal cord is slightly flexed, presenting a more dorsal view at the rostral end (left), showing bilateral cell labeling. More caudally, the specimen shows that the positive neurons are found both ventrally (top) and dorsally (bottom) in the cord. At 48 hpf, this expression pattern increases in intensity in these same tissues. B: Gene expression of NR1.2 paralog. At 24 hpf, there is light staining in the forebrain. By 48 hpf, all three regions of the brain express NR1.2. The 48 hpf image is a dorsolateral view of the left side of the head. This paralog is not expressed in the spinal cord.
The NR2A.1 subunit is expressed in the retina at 48 hpf but not at 24 hpf (Fig. 7A). The NR2A.2 mRNA is detected in the retina at 48 hr but at a lower level than NR2A.1 (Fig. 7B). In the rat, the NR2A gene is not expressed prenatally, but starts to be expressed at low levels in the hippocampus, cerebellum, and retina by P7 (Watanabe et al., 1992, 1994; Monyer et al., 1994).
A: Gene expression of NR2A.1 in the retina of 48 hpf zebrafish. This subunit is not expressed at 24 hours postfertilization (hpf). B:Expression of NR2A.2 in the retina of 48 hpf but not 24 hpf zebrafish. C:In situ hybridization pattern of the NR2D.1 RNA in 24- and 48-hour zebrafish. At 24 hpf, this subunit is found in the forebrain, retina, and spinal cord. The orientation of the spinal cord is rostral to the left, caudal to the right, dorsal at the top, and ventral at the bottom. At 48 hpf, the expression has increased in intensity and is found in the fore-, mid-, and hindbrain, and in the spinal cord. The photo of the head at 48 hpf is a dorsolateral view. This expression pattern overlaps that of NR1.1. D: NR2D.2 is detectable in the retina at 24 and 48 hpf but is not found in the brain or spinal cord. All images were cropped and sized in Photoshop 6.0.
Neither of the zebrafish NR2B paralogs appear to be expressed in the developing nervous system, in contrast to the rat, where the NR2B is expressed prenatally at E17 in the cortex, hippocampus, hypothalamus, and somewhat in the spinal cord (Watanabe et al., 1992; Monyer et al., 1994) and in the retina by E15 (Watanabe et al., 1994). The expression in rat peaked at postnatal day (P) 1 and either stayed the same or decreased to undetectable levels by P28, depending on the area of the brain (Zhong et al., 1995). Seeber et al. reported that the rat NR2B subunit could be transiently detected in the heart by Western blot between E14 and P12 (Seeber et al., 2000). To determine whether the NR2B paralogs are expressed in the developing embryo at levels below the sensitivity of detection for in situ hybridization, we performed PCR using cDNAs prepared from 24 hpf, 48 hpf, 96 hpf, and 2-month-old fish. As can be seen in Figure 8, zNR2B.1 is not expressed at 24 hpf or 48 hpf. It is present by 96 hpf and decreases by 2 months. zNR2B.2 is expressed at a low level at 48 hr; the signal increases in intensity by 96 hpf and then decreases over the next 2 months.
Polymerase chain reaction (PCR) reaction using cDNA from 24 hpf, 48 hpf, 96 hpf, and 2-month-old zebrafish embryos. For zN2B.1, the forward primer lies in exon 12 and the reverse primer lies in exon 15, yielding a fragment of 680 bp. For zN2B.2, the forward primer lies in exon 4 and the reverse primer in exon 6, yielding a fragment of 775 bp. The N2B.1 does not appear to be expressed at 24 or 48 hpf but is seen at 96 hpf. The N2B.2 is not expressed at 24 hpf. It appears to be expressed at very low levels at 48 hpf and is strongly expressed at 96 hr. The expression has decreased by 2 months after fertilization.
The NR2C subunits are not expressed during early development in the zebrafish, which is similar to the pattern in the rat, where the NR2C does not appear until P12, where it is found in the cerebellum and is strongly expressed in the adult (Watanabe et al., 1992; Monyer et al., 1994). The NR2D.1 show a similar developmental profile to the rat NR2D; it is expressed strongly at 24 hpf in the spinal cord and somewhat in the forebrain, midbrain, hindbrain, and in the retina. At 48 hpf, it is strongly expressed in the whole brain, spinal cord, and retina (Fig. 7C). The rat NR2D is expressed at E17 in the cortex, cerebellum, striatum, and spinal cord (Monyer et al., 1992). By P7, the NR2D expression has decreased significantly and is present at very low levels in the cortex, colliculi, thalamus, and cerebellum. It is undetectable by P14 (Watanabe et al., 1992). The other zebrafish paralog, NR2D.2, is not expressed in the brain or spinal cord but is seen in the retina at both 24 and 48 hpf (Fig. 7D).
This study presents the cloning of the zebrafish homologs of the mammalian NMDA R1 and R2 families and five additional paralogs that are not found in mammals, corresponding to an additional NR1 and one each additional NR2A, 2B, 2C, and 2D subunit genes. These findings support the supposition that the zebrafish is an important platform in which to study the functional roles that these receptors play in neuronal differentiation and in the establishment of neuronal connectivity in the developing vertebrate nervous system.
EXPERIMENTAL PROCEDURES
Molecular Biology
Total RNA was obtained from 4-day-old zebrafish embryos using the single-step acid phenol method (Chomczynski and Sacchi, 1987) and cDNA prepared using the Powerscript kit (Invitrogen, Palo Alto, CA). The cDNA was used as template in polymerase chain reactions to clone the zebrafish NMDA subunits. Degenerate primers (24mers) were designed against highly conserved areas of the rat NR1 (forward primer: 5′-TGG AAY GGM TGA TGG GNG A-3′ [corresponding to aa WNGMMGE]; reverse primer: 5′-AAR GCD CCA RRT TDG CWG T-3′ [YTANLAA]) or NR2A (forward primer: 5′-TTC TGY AGY GAY ATY CTS A-3′ [FCIDILK]; reverse primer: 5′-GTA GAR GAY ACN TGY CAT G-3′ [MAGVFYM]). Products amplified from these reactions were purified and ligated into the pGEM-T vector (Promega Corporation, Madison, WI). These partial clones were sequenced using a BigDye (Perkin–Elmer) sequencing kit. By this means, five NMDA receptor subunits were identified (two NR1 subunits designated NR1.1 and NR1.2 and one each N2A, 2C, and 2D). The online zebrafish genomic sequence trace database (ENSEMBL) at the Sanger Institute was then searched using the sequences of these five subunits and partial exonic sequences from a further five subunits were identified, yielding a total of unique 10 genes. Specific primers were designed from the database sequence to clone the entire NR1.1 and NR1.2 subunits. The entire sequences of the NR2 subunits were deduced from the database, except for the 5′-and 3′-ends of the NR2Ds and NR2Bs, which were obtained using SMART RACE PCR (Clontech, Palo Alto, CA).
Radiation Hybrid Analysis
Physical mapping of the subunit genes was carried out using the T51 radiation hybrid panel (Kwok et al., 1999) purchased from Research Genetics (Huntsville, AL). PCR reactions (10-μl total volume) containing 25 ng of hybrid genomic DNA and a pair of gene specific forward and reverse primers (at 20 mM) were carried out using AmpliTaq Gold (Applied Biosystems, Foster City, CA) under the following conditions: 94°C for 30 sec, 60°C for 30 sec, 73°C for 1 min for 35 cycles, followed by 5 min at 73°C. Reactions were then electrophoresed through a 2% agarose gel and products visualized by ultraviolet illumination. All reactions were run in duplicate. Positive reactions were scored as 1, negatives as 0, and equivocals as 2. To be scored as a positive, a band had to be clearly detected in both replicate reactions. If there was an ambiguous result for any sample, a third reaction was run and scored. Once all 94 samples had been scored, the on-line RHMapper service provided by Dr. L. Zon's laboratory (http://134.174.23.167/zonrhmapper/Maps.htm) was used to determine the approximate physical location of the relevant gene. The primers used for these studies are shown in Table 1.
Maintenance of Fish
Zebrafish were raised and cared for as described by Westerfield (1993). Fish were kept at approximately 10 adults/L, on a 14-hr day, 10-hr night schedule at a constant 28.5°C, with feeding done twice daily. The age of the embryos given in this study refer to time since fertilization (hpf). Staging was carried out using the somite formation criteria of Kimmel (Kimmel et al., 1995). For in situ hybridization studies, reduced pigmentation of embryos/larvae was induced by adding phenylthiourea to the water (final concentration, 0.003%). All procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Experimental Animals and were approved by the SLU Animal Care and Use Committee.
In Situ Hybridization
Embryos/larvae were fixed in 4% paraformaldehyde for 24 hr and then stored in 100% methanol at −20°C until required. For in situ hybridization, the embryos were re-hydrated, treated with proteinase K, fixed a second time in 4% paraformaldehyde, and then prepared for in situ hybridization as per Jowett and Lettice (1994).
Plasmids were linearized with the appropriate enzyme, and cRNA preparation was performed using the AmpliScribe Transcription kit from EpiCentre. cRNAs were modified by covalent modification with d-nitrophenol (DNP) using the Mirus LabelIt kit and hybridized at 65°C in hybridization buffer (50% formamide, 5× standard saline citrate, 50 mg/ml heparin, 500 mg/ml tRNA, 9.2 mM citric acid, 0.1% Tween 20). Embryos were washed and incubated with a 1:2,500 dilution of alkaline phosphatase-conjugated anti-DNP antibody (Vector Laboratories) as described (Kucenas et al., 2003). Embryos were deyolked and mounted in glycerol on slides with coverslips, and photos were taken using a Nikon DN-100 digital camera attached to a zoom stereomicroscope or by an Olympus 35 mm camera attached to an Olympus microscope. The images were scanned as digital files and cropped and sized by using Adobe Photoshop 6.0.
Acknowledgements
We thank Tina Rose for technical help.
