N-methyl-D-aspartate (NMDA) receptors are one of three pharmacologically distinct subtypes of ionotropic receptors that mediate a majority of excitatory neurotransmission in the brain via the endogenous amino acid, L-glutamate. NMDARs form heteromeric complexes usually comprised of the essential NR1 subunit and various NR2 subunits. Molecular and electrophysiological properties of NMDARs suggest that they may be the Hebbian 'coincidence detectors' hypothesized to underlie associative learning. Because of the nonspecificity of drugs that modulate NMDAR function or the relatively chronic genetic manipulations of various NMDAR subunits from mammalian studies, conclusive evidence for such an acute role for NMDARs in adult behavioral plasticity, however, is lacking. Moreover, a role for NMDARs in memory consolidation remains controversial (Xia, 2005; full text of article).

The Drosophila genome encodes two NMDAR homologs, dNR1 and dNR2. When coexpressed in Xenopus oocytes or Drosophila S2 cells, dNR1 and dNR2 form functional NMDARs with several of the distinguishing molecular properties observed for vertebrate NMDARs, including voltage/Mg²⁺-dependent activation by glutamate. Both proteins are weakly expressed throughout the entire brain but show preferential expression in several neurons surrounding the dendritic region of the mushroom bodies. Hypomorphic mutations of the essential dNR1 gene disrupt olfactory learning, and this learning defect is rescued with wild-type transgenes. Importantly, Pavlovian learning is disrupted in adults within 15 hr after transient induction of a dNR1 antisense RNA transgene. Extended training is sufficient to overcome this initial learning defect, but long-term memory (LTM) specifically is abolished under these training conditions. In conclusion, this study uses a combination of molecular-genetic tools to (1) generate genomic mutations of the dNR1 gene, (2) rescue the accompanying learning deficit with a dNR1⁺ transgene, and (3) rapidly and transiently knockdown dNR1⁺ expression in adults, thereby demonstrating an evolutionarily conserved role for the acute involvement of NMDARs in associative learning and memory (Xia, 2005).

The NMDAR channel is highly permeable to Ca²⁺ and Na⁺, and its opening requires simultaneous binding of glutamate and postsynaptic membrane depolarization. Once activated, the NMDAR channel allows calcium influx into the postsynaptic cell where calcium triggers a cascade of biochemical events resulting in synaptic changes (Xia, 2005).

Cellular studies have suggested that NMDAR is involved in several forms of synaptic plasticity, including long-term potentiation and long-term depression. The NMDAR possesses an interesting molecular property, namely, a voltage-dependent blockade of glutamate-induced calcium flux. This suggests NMDAR's role as the above mentioned 'Hebbian coincidence detector' underlying associative learning. Additional, non-Hebbian cellular mechanisms appear necessary, however, to model associative learning adequately. To that end, behavioral studies attempting to demonstrate an acute role for mammalian NMDARs in associative learning and/or memory have been limited by (1) the nonspecificity of drugs that modulate NMDAR function or (2) the relatively chronic genetic manipulations of various NMDAR subunits. Whether NMDARs also are involved with memory consolidation is even more controversial (Xia, 2005).

In invertebrates, pharmacological manipulations have suggested that NMDA-like receptors mediate associative learning in Aplysia (Roberts, 2003) and memory recall in honeybee (Si, 2004), and the function of an NR1 homolog, NMR-1, has been characterized in C. elegans (Brockie, 2001). These studies did not determine which potential NMDAR homologs form functional NMDARs, (Dingledine, 1999), however, direct demonstrations of roles for specific NMDAR genes in behavioral plasticity still are lacking in these model systems. Therefore molecular, genetic, electrophysiological, and behavioral experiments were persued on the Drosophila NMDAR subunit genes, dNR1 (Ultsch, 1993) and dNR2, which together establish an acute role for NMDAR in associative learning and in long-term memory consolidation (Xia, 2005).

Homology searches of the Drosophila genome database and cloning suggest dNR1 is the only gene bearing high amino acid sequence similarity to the mammalian NMDA receptor subunit NR1. Compared with its vertebrate counterpart, dNR1 shows high homology with respect to its entire size, domain structures, and active physiological sites. dNR2 appears to be the sole gene encoding the Drosophila homolog of mammalian NR2, although there are four NR2 family members in vertebrates (Yamakura, 1999). dNR2 undergoes alternative splicing, however, to generate eight different transcripts and three protein variants. The domain structures of dNR2 show high homology to vertebrate NR2, but its entire size, active physiological sites, and molecular function are only moderately conserved from its mammalian counterparts (Xia, 2005).

The dNR1 transcript is highly regulated during development and is expressed at high levels in late embryos when the larval nervous system is formed, in late pupae when the adult central nervous system develops, and in adult head. Western blots confirmed that both proteins are expressed at a high level in adult head but not in the body. Immunostaining also indicates that they may be expressed throughout the whole brain and at especially high levels in several neurons surrounding the calyx of the MBs. The interpretation of generally weak expression of dNR1 and dNR2 is further supported by Western blots showing a detectable band from single-head preparations. Thus, dNR1 and dNR2 likely function together in most places, which is in agreement with functional analyses. In contrast, dNR1 appears to have a broader pattern of preferential expression than dNR2 in adult brain, suggesting alternative associations with other endogenous glutamate receptors. Alternatively, dNR1 alone may form functional NMDAR channels in vivo, given its weak but significant NMDA-selective response in Xenopus oocytes. It might be noted, however, that functional NMDA receptors can be formed by expression of NR1 alone in Xenopus oocytes but not in mammalian cell lines. Finally, dNR1 has an RSS (Retention Signal Sequence) motif at its C terminus, similar to its mammalian homolog, suggesting that dNR1, when not associated with dNR2 or other glutamate receptors, may be retained in the ER rather than inserted in the cell membrane (Xia, 2005).

Coexpression of dNR1 and dNR2-2 in Xenopus oocytes generated NMDA-selective responses. Similarly, functional homomeric receptors can be formed within the AMPA and kainate subunit families but probably not for NMDA receptors in vertebrates, and highly active NMDAR channels are only formed when the NR1 subunit is expressed in combination with one of the four NR2 subunits (Dingledine, 1999; Mori, 1995). Pharmacological, anatomical, biochemical, and immunological studies also have established heteromeric, but not homomeric, assembly of NMDAR channel subunits in vivo (Mori, 1995). The physiological features that distinguish NMDAR from other ionotropic glutamate receptors are (1) high permeability to Ca²⁺, (2) selective activation by NMDA and L-asparate, (3) modulation by glycine as the coagonist for glutamate, and (4) voltage-dependent blockade by Mg²⁺ (Dingledine, 1999). The electrophysiological profile of dNR1 and dNR2 coexpressed in Xenopus oocytes or Drosophila S2 cells reveals that the functional NMDARs produce most of these distinguishing characteristics including selective activation by NMDA and L-asparate, modulation by glycine as the coagonist for glutamate, and voltage- and Mg²⁺-dependent conductance. Thus, Drosophila likely has functional NMDARs consisting of two subunits, dNR1 and dNR2 (Xia, 2005).

The NMDA-selective conductance was sensitive to Mg²⁺ blockade only in Drosophila S2 cells but not in Xenopus oocytes up to 10 mM, which is highly reminiscent of NMDA receptors in C. elegans (Brockie, 2001). Proper external ionic conditions for oocytes and insect cells are remarkably different. The endogenous Mg²⁺ concentration for fly muscle, for instance, is about ten times higher than that for oocytes, suggesting that invertebrate NMDA receptors have evolved to be less sensitive to Mg²⁺. Molecular evidence exists in support of this conclusion. Replacement of the asparagine residue in the pore-forming TM2 domain reduces but does not abolish Mg²⁺ block for mammalian NR receptors (Dingledine, 1999). This crucial asparagine residue in dNR2 subunits is replaced by glutamine. In addition, TM1, TM4, and the short linker between TM2 and TM3 domains also are critical determinants for Mg²⁺ block (Kuner, 1996). Although the linker appears conserved in dNR2, TM1 and TM4 are not (Xia, 2005).

Fly NMDA receptors have been shown to regulate the larval locomotor rhythm (Cattaert, 2001). This effect can be blocked completely by MK801, requiring binding to the same asparagine residue to execute its antagonist effect (Ferrer-Montiel, 1995). MK801 also suppresses NMDAR-mediated juvenile hormone biosynethesis in cockroach (Chiang, 2002) (Xia, 2005).

This study provided the first demonstration that NMDARs are required acutely for associative learning in Drosophila. Pavlovian task is a form of fear conditioning which uses well-defined odors as conditioned stimuli (CSs) and footshock as an unconditioned stimulus (US). When tested immediately after Pavlovian conditioning (one training session), flies homozygous for either of two different hypomorphic mutations performed poorly in this task, although they seem to grow normally, do not show any obvious behavioral abnormalities, and most importantly, show normal sensorimotor responses to the stimuli used for this task. The learning deficit in dNR1 mutants can be rescued fully in transgenic flies carrying either of two different genomic constructs containing the dNR transcription unit, which constitutes definitive proof that this transcription unit is responsible for the phenotypic defect observed in these mutants (Xia, 2005).

dNR1 is acutely required for associative learning. Disruption of dNR1, with an hs-GAL4 driver to induce expression of a dNR1 antisense message, yielded a learning deficit specifically and transiently. These results rule out any potential developmental explanation for the adult learning defect. The data extend to insects similar findings from pharmacological and genetic studies in mammals and provide the strongest argument to date that adult learning and memory depend on proper NMDA receptor function (Xia, 2005).

Acute disruption of dNR1 also disrupts 1-day memory after spaced training, without affecting 1-day memory after massed training. The specific abolition of LTM, without affecting 1-day memory after massed training, is similar to that produced by induced expression of a CREB-repressor transgene and indicates a specific disruption of cycloheximde-sensitive LTM with no effect on cycloheximide-insensitive ARM. Hence, CREB-dependent LTM formation appears to depend on normal NMDA receptor function. The cAMP/PKA/CREB signaling pathway has been shown to be involved in diverse processes ranging from hippocampal LTP and barrel formation to learning and memory in mammals. In most of these experimental contexts, activation of NMDARs is required for LTM formation (Riedel, 2003). Recent experiments in mammals also have revealed NMDAR-dependent activation of CREB during LTP and LTM in both amygdala and hippocampus (Schulz, 1999; Cammarota, 2000). Interestingly, two functionally distinct NMDA receptor signaling complexes have been identified: synaptic and extrasynaptic (Hardingham, 2002). Synaptic NMDARs can cause sustained CREB phosphorylation and CRE-mediated gene expression, whereas extrasynaptic NMDARs actively suppress CREB activity via an as yet unknown mechanism. Hence, it seems likely that synaptic NMDAR complexes regulate memory formation by controlling nuclear signaling to CREB (Xia, 2005).

This characterization of a role for NMDA receptors in behavioral plasticity of Drosophila again reinforces the notion that the functional homologies among various model systems is appreciable. Many intracellular signaling proteins are known to be physically associated with vertebrate NMDA receptors (Husi, 2000). The newly identified NMDAR complex consist of more than 80 different proteins, organized into receptor, adaptor, signaling, cytoskeletal, cell adhesion, and novel proteins (Husi, 2000). Genetic and pharmacological disruptions of several components of the NMDAR complex produce learning impairments in rodents. Obvious Drosophila homologs can be identified for a majority of these 80 proteins. Among of them are NR1, PKA subunits, PKC isoforms, and NF1. Disruptions of these genes yield associative learning deficits in flies (Xia, 2005).

The conservation of NMDA-dependent behavioral plasticity in invertebrates further demonstrates that a unified mechanism underlies associative learning and memory. Because behavioral plasticity is tightly associated with synaptic plasticity, it is speculated that similar cellular mechanisms of NMDAR-mediated long-term changes, such as LTP and LTD, may also exist in the adult insect brain. Drosophila genetics now can be applied to discover additional genes and signaling pathways important for NMDAR-dependent plasticity (Xia, 2005).

In conclusion, this study has established that Drosophila likely has functional NMDARs consisting of two subunits, dNR1 and dNR2. Combined expression of both dNR1 and dNR2 generated NMDA-selective responses, whereas expression of either of them individually no significant NMDA-dependent responses in oocytes. The eletrophysiological profile of dNR1 and dNR2 coexpressed in Xenopus oocytes or Drosophila S2 cells reveals that the functional NMDARs produce most of these distinguishing properties specific to mammalian counterparts including selective activation by NMDA and L-asparate, modulation by glycine as the coagonist for glutamate, and voltage- and Mg²⁺-dependent conductance (Xia, 2005).

This study also demonstrates that NMDARs not only are involved acutely for associative learning but also are required for LTM consolidation. Genomic mutations of the essential dNR1 gene yield defects in a Pavlovian olfactory learning task, and these learning defects are fully rescued by two different genomic transgenes containing the dNR1⁺ coding sequence. Importantly, it was shown that Pavlovian learning is disrupted within 15 hr via transient induction in adults of a dNR1 antisense RNA transgene. Finally, the transient knockdown of dNR1 also specifically abolishes the consolidation of protein synthesis- and CREB-dependent LTM (Xia, 2005).

GENE STRUCTURE

cDNA clone length - 4186 bp (Nmdar1) and 4180 (Nmdar2-RC)

Bases in 5' UTR - 236 (Nmdar1) and 601 (Nmdar2-RC)

Exons - 15 for Nmdar1 and 12 for Nmdar2-RC

Bases in 3' UTR - 956 (Nmdar1) and 327 (Nmdar2-RC)

PROTEIN STRUCTURE

Amino Acids - 997 (Nmdar1) and 1083 (Nmdar2 isoform C)

Structural Domains

The dNR1 gene is a large gene, containing 15 exons (Ultsch, 1993). Exon 1 (noncoding) undergoes alternative splicing, giving rise to two different transcripts, which contain the same coding sequence but which differ in the 5′ untranslated region. The putative dNR1 protein from these splice forms faithfully maintains all the major structural features of NR1 receptor. The protein contains one hydrophobic region at the amino terminus supposedly as the signal peptide, three hydrophobic transmembrane regions (TM1), a hydrophobic pore-forming segment in the carboxyl terminal half, and two ligand binding domains (S1–S2) with high homology to bacterial amino acid binding proteins. dNR1 also has a potential type II PDZ domain binding motif at its C terminus (X-Ψ-X-Ψ, where Ψ is a hydrophobic amino acid), suggesting interactions with other PDZ domain-containing proteins (Sheng, 2001). Most of the important amino acid residues for ligand binding are conserved in dNR1. A key asparagine residue (N631) is present in the TM2 domain and presumably controls the Ca²⁺ permeability and voltage-dependent Mg²⁺ blockade (Burnashev, 1992) (Xia, 2005).

dNR2, as confirmed by complete cloning, appears to be the only gene encoding the fly NR2 homolog, whereas there are four mammalian members in the NR2 subfamily (Dingledine, 1999; Yamakura, 1999). dNR2 undergoes alternative splicing, mostly at the 5′ untranslated region, generating eight different transcripts that may encode three different proteins. Full-length cDNAs have been isolated for all eight variants. Six of them contain the same coding sequence but differ from each other at the 5′ untranslated region, with five of them containing a separate noncoding exon 1. All three deduced NR2 proteins bear highest homology to NMR-2 in C. elegans, rat NR2D and NR2B, with respect to their overall sequence or their ligand binding and pore-forming transmembrane domains. Several anti-peptide monoclonal or polyclonal anti-dNR2 antibodies have been generated that specifically recognized two different bands on Westerns. Because two of the putative dNR2 peptides were predicted to have similar molecular weight, it is still unclear whether the two bands in fact contained all three protein variants (Xia, 2005).

The domain structures of NR2 receptors are largely conserved in dNR2, but its general sequence homology and the active physiological sites only moderately mimic its mammalian counterparts. The protein contains four hydrophobic regions (TM1–TM4) in the carboxyl terminal half that align perfectly with the three hydrophobic transmembrane regions and a hydrophobic pore-forming segment (TM2) in other ionotropic glutamate receptors (Dingledine, 1999). Like its rat counterpart, dNR2 has conserved major determinants of glutamate binding in the N-terminal ligand binding domain (S1) preceding transmembrane segment TM1 and the loop (S2) between TM3 and TM4 (Dingledine, 1999). The two asparagine residues, which are present in the TM2 domain of NMDA receptors and control the Ca²⁺ permeability and voltage-dependent Mg²⁺ blockade (Dingledine, 1999), however, are not conserved in dNR2. Finally, the type I PDZ binding motif (X-S/T-X-V) is not present in dNR2, whereas it is well conserved in all vertebrate NR2 homologs (Sheng, 2001). Thus, Drosophila NMDA receptors may physically interact with PDZ domain-containing proteins through dNR1 but not dNR2, which is usually the case in vertebrates (Xia, 2005).

date revised: 5 February 2007

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.