Identification and investigation of Drosophila postsynaptic density homologs.

AMPA receptors are responsible for fast excitatory transmission in the CNS and the trafficking of these receptors has been implicated in LTP and learning and memory. These receptors reside in the postsynaptic density, a network of proteins that links the receptors to downstream signaling components and to the neuronal cytoskeleton. To determine whether the fruit fly, Drosophila melanogaster, possesses a similar array of proteins as are found at the mammalian PSD, we identified Drosophila homologs of 95.8% of mammalian PSD proteins. We investigated, for the first time, the role of one of these PSD proteins, Pod1 in GluR cluster formation at the Drosophila neuromuscular junction and found that mutations in pod1 resulted in a specific loss of A-type receptors at the synapse.


Introduction
The majority of neurotransmission in the mammalian central nervous system uses glutamate as a neurotransmitter. One type of ionotropic glutamate receptor, AMPA receptors (AMPARs), is responsible for fast excitatory transmission in the CNS. The regulated delivery and insertion of AMPARs receptors has been implicated in long term potentiation (LTP, for review see Malinow and Malenka, 2002) and contextual fear learning (Hu et al. 2007;Matsuo et al. 2008). Therefore, the mechanisms that govern AMPAR expression and traffi cking are of considerable interest.
AMPARs are tetramers composed of GluR1-4 (Hollmann and Heinemann, 1994;Monoghan and Wenthold, 1997;Gereau and Swanson, 2008). Although AMPARs may be synthesized in dendrites (Ju et al. 2004), most AMPAR mRNA is located in the neuronal cell body suggesting that AMPARs must be transported to their synaptic destinations (Esteban, 2003). There is some evidence that kinesins mediate the cellular traffi cking of AMPAR-containing vesicles along the microtubule cytoskeleton. The heavy chain of kinesin directly interacts with GRIP (Setou et al. 2002), which binds to the AMPAR subunits GluR2 and GluR3 Srivastava et al. 1998). GluR2 and GRIP also associate with liprin-α (Wyszynski et al. 2002), which interacts with KIF1 . Vesicles containing AMPARs must be transferred from microtubules to actin fi laments before their fi nal delivery into dendritic spines. This process may be mediated by the motor protein, myosin Vb . Traffi cking of receptors to the synapse is mediated by a family of transmembrane regulator proteins (TARPs) Tomita et al. 2004;Tomita et al. 2005;Nicoll et al. 2006;Ziff, 2007) that may also infl uence AMPAR kinetics (Milstein et al. 2007).
AMPARs are dynamically regulated at the synapse. For example, transient stimulation of NMDA receptors suffi cient to produce LTP results in the rapid insertion of AMPARs into the postsynaptic membrane (Liao et al. 1995;Liao et al. 1999;Liao et al. 2001;Poncer and Malinow, 2001) possibly from recycling endosomes ). This de novo insertion of receptors is dependent upon the interaction between the AMPAR subunit, GluR1 and the scaffolding protein, SAP97 (Hayashi et al. 2000). At synapses, AMPARs are part of dense protein networks called postsynaptic densities (PSD), which are located opposite from presynaptic release sites. The molecular composition of the PSD has been characterized using biochemical approaches, mass spectrometry, and proteomics (Kennedy, 1998;Husi and Grant, 2001;Jordan et al. 2004;Peng et al. 2004;Boeckers, 2006;Collins et al. 2006;Dosemeci et al. 2007) revealing a complex structure composed of hundreds of proteins. The complexity of the interactions between proteins suggests that perturbations of many PSD proteins could affect AMPAR traffi cking or localization.
We sought to determine whether the fruit fl y, Drosophila melanogaster, possesses a similar array of proteins as are found at the mammalian glutamatergic PSD. The Drosophila genome encodes 21 putative ionotropic glutamate receptor subunits, including homologs of mammalian NMDA, AMPA, kainate, and delta receptor subunits (Sprengel et al. 2001). The Drosophila neuromuscular junction (NMJ) is glutamatergic making it similar in composition and function to mammalian central synapses (Collins and DiAntonio, 2007). The receptors at the NMJ are classifi ed non-NMDA receptors. Similar to their mammalian homologs, Drosophila GluRs are tetramers that contain three essential subunits including GluRIIC (Marrus and DiAntonio, 2004), GluRIID , and GluRIIE ) along with either GluRIIA (Schuster et al. 1991) or GluRIIB (Petersen et al. 1997). These two receptor types, A-type (which contain GluRIIA, -IIC, -IID, and -IIE but not -IIB) or B-type (which contain GluRIIB, -IIC, -IID, and -IIE but not -IIA), are differentially expressed and clustered (Marrus and DiAntonio, 2004;Schmid et al. 2008) and interact with distinct components of postsynaptic density (Chen and Featherstone, 2005;. As in mammals, Drosophila glutamate receptors form postsynaptic tetramers that mediate fast synaptic transmission (DiAntonio, 2006), and NMDA receptors are required for learning (Xia et al. 2005, Lin, 2005Wu et al. 2007). This suggests that glutamate receptor (GluR) function may be largely conserved, but it remains unknown whether mechanisms of glutamate receptor trafficking and anchoring are also conserved. The use of an evolutionarily simpler system could facilitate the understanding of molecular functions and relationships between proteins involved in GluR traffi cking. We found that 95.8% of mammalian PSD proteins have Drosophila homologs. We investigated, for the fi rst time, the role of one of these PSD proteins, Pod1, in GluR cluster formation at the NMJ and found that mutations in pod1 resulted in a specifi c loss of A-type receptors at the synapse.

Bioinformatics
We searched the literature for proteins that regulate AMPAR, KARs, or reside in the PSD. Mammalian protein sequences were extracted from the National Center for Biotechnology Information (http://www. ncbi.nlm.nih.gov/). The mammalian sequences used were either mouse, rat, or human. The amino acid sequence obtained was compared with annotated proteins in Drosophila using FlyBase's BLAST (http://flybase.bio.indiana.edu/blast/). Gene expression patterns were retrieved from the Berkeley Drosophila Genome Project Expression Pattern database (http://www.fruitfl y.org/cgi-bin/ ex/insitu.pl).

Antibodies and immunocytochemistry
For immunocytochemistry and microscopy, animals were dissected and fi xed for 30-60 min in either Bouin's fi xative (when GluR antibodies were used), or 4% paraformaldehyde in PBS (for Pod1 labeling). Third instar larvae were dissected and fi llet preparations were pinned down in Sylgard lined Petri dishes. All dissections were done in Drosophila standard saline (135 mM NaCl, 5 mM KCl, 4 mM MgCl, 1.8 mM CaCl, 5 mM TES, 72 mM sucrose) at RT. Mouse monoclonal anti-GluRIIA (Iowa Developmental Studies Hybridoma Bank, Iowa City, IA) was used at 1:100. Rabbit polyclonal anti-GluRIIB and anti-GluRIIC were gifts from Aaron DiAntonio (Washington University, St. Louis, MO) and were used at 1:2000 and 1:5000, respectively. Guinea pig polyclonal anti-Pod1 was a gift from Yuh-Nung Jan (University of California, San Francisco) and was used at 1:1000. Fluorescently conjugated anti-HRP (Jackson Immunoresearch Labs, West Grove, PA) was used at 1:100. Goat anti-rabbit, goat anti-mouse, or goat anti-guinea pig fl uorescent (FITC or TRITC) secondary antibodies (Jackson Immunoresearch Labs, West Grove, PA) were used at 1:400. The 6/7 NMJ of abdominal hemisegments A3 or A4 were used for all studies. Confocal images were obtained using an Olympus FV500 laser-scanning confocal microscope. Image analysis and quantifi cation was performed using ImageJ and Adobe Photoshop software.

Electrophysiology
All electrophysiology was performed on the ventral body wall muscle 6. Larval recordings were performed on third instar larvae 110-120 hr AEL. Muscle 6 was voltage-clamped at −60 mV. Standard two-electrode voltage clamp techniques were used, as previously described . Data were acquired and analyzed using a Gene clamp 500 amplifi er and pClamp9 (Axon Instruments, Union City, CA). All dissections and recordings were done in standard Drosophila saline at 19°C.

Fly stocks
All animals were raised at 25°C in standard fl y vials with corn meal molasses medium. Pod1 stocks were gifts from Yuh-Nung Jan (University of California, San Francisco). Control animals used were w 1118 .

Data acquisition and statistics
GluR clusters were measured manually by outlining GluR clusters using NIH Image J software as previously described (Featherstone et al. 2002;Chen and Featherstone, 2005;Rasse et al. 2005). Total GluR fl uorescence was quantifi ed by measuring fl uorescence intensity at the synapse and subtracting background/muscle fl uorescence intensity using Adobe Photoshop CS2. Statistics were performed using GraphPad Prism (v. 4.01). Statistical comparisons were made using unpaired students t-tests or, for distributions, Kolmogorov-Smirnov tests. Statistical signifi cance in fi gures is represented as follows: * = p Ͻ 0.05, ** = p Ͻ 0.001, and *** = p Ͻ 0.0001. All error bars represent S.E.M.

Most PSD proteins have Drosophila homologs
To assess the similarity by which mammalian and fl y non-NMDA receptors might be traffi cked and anchored to the synapse, we searched the literature for proteins that interact with AMPARs or KARs. Of the 40 proteins we found that regulate AMPARs or KARs, 38 (95%) have Drosophila homologs (Table 1). If these Drosophila homologs function similarly to regulate GluR traffi cking and localization at the glutamatergic Drosophila NMJ, we would expect them to be expressed in neurons, muscle, or both. Therefore, we used the Berkeley Drosophila Genome Project (BDGP) Gene Expression Database (http://www.fruitfl y.org/cgibin/ex/insitu.pl) to examine the expression patterns of these genes. The expression patterns for 14 of these genes are documented. Of these, 5 are expressed in muscle, 6 are expressed in neurons, 2 are expressed ubiquitously, and one is expressed in other tissue. In other words, of the 15 genes with documented expression patterns, 93% are expressed in tissues consistent with conserved function.
Some mammalian GluRs are embedded within the PSD, a specialized protein network that allows postsynaptic cells to receive information. We extended our search of the literature to include proteins that make up the PSD. Of the 199 proteins we found that are localized to the PSD, 191 (96.0%) have Drosophila homologs (Supplemental Table 1). 21 of the Drosophila genes are homologous for more than one mammalian PSD protein, consistent with the recent confi rmation that families of genes expanded between fl y and mouse (Emes et al. 2008). The BDGP has documented the expression pattern for 63 of these genes. Of these, 18 are expressed in muscle, 29 are expressed in neurons, 4 are expressed in both neurons and muscle, 7 are expressed ubiquitously, and 5 are expressed in other tissues. Thus, 92% of Drosophila proteins homologous to mammalian PSD proteins are expressed in tissues consistent with conserved function. We conclude from these data that the signaling machinery surrounding Drosophila GluRs is likely to be similar to that found in the mammalian PSD.

Mutations in pod1 reduce GluRIIA cluster sizes
To test whether one of the Drosophila genes listed in Supplemental Table 1 plays a role in GluR cluster formation, we examined the NMJ of pod1 mutants. pod1 is one of two coronin family members in Drosophila and has been shown to crosslink actin and microtubules in cultured S2 cells (Rothenberg et al. 2003). We selected pod1 for further study because the literature suggests a number of cytoskeletal proteins are part of the PSD (40 of the 199 PSD proteins in Supplemental Table 1) and pod1 is expressed in both neurons and muscle. We fi rst wanted to confi rm that pod1 is localized to NMJs by examining its immunoreactivity ( Fig. 1) and found that Pod1 immunoreactivity (which is eliminated in pod1 mutants; data not shown) is enriched at the NMJ suggesting Pod1 may function at the NMJ.
To determine whether the loss of A-type GluRs affects the synaptic function of the NMJ, we performed two-electrode voltage clamp. Muscle 6 was voltage clamped at -60 mV and spontaneous miniature excitatory junction currents (sEJCs or 'minis') were recorded. The frequency of minis is significantly reduced in pod1 mutant animals (Fig. 2C, D; w 1118 = 2.7 ± 0.23 Hz, n = 10; pod1 P1 = 1.34 ± 0.12 Hz, n = 8, p = 0.0002; pod1 ∆17 = 0.95 ± 0.14 Hz, n = 7, p Ͻ 0.0001). This reduction may represent changes in presynaptic function (Rothenberg et al. 2003) as well as minis being lost in baseline noise. Consistent with this and the reduction in GluRIIA staining, sEJC amplitudes are also signifi cantly reduced in pod1 mutants ( Fig. 2C; pod1 P1 K-S statistic = 0.957, p Ͻ 0.0001; pod1 ∆17 K-S statistic = 0.977, p Ͻ 0.0001). The smaller mini amplitudes taken together with the immunocytochemical data indicate that pod1 mutants contain fewer A-type receptors. In agreement with this, we found that the sEJC decay time was signifi cantly reduced in pod1 mutants (data not shown, w 1118 = 12.20 ± 0.25 ms, n = 10; pod1 P1 = 9.96 ± 0.29 ms, n = 8, p Ͻ 0.0001; pod1 ∆17 = 10.76 ± 0.25 ms, n = 7, p Ͻ 0.0001). Shorter decay times are associated with specifi c loss of A-type GluRs (DiAntonio et al. 1999;Schmid et al. 2008). We conclude from these data that pod1 plays a role in the expression or localization of A-type, but not B-type GluRs.

Discussion
Synaptic plasticity and memory rely on the traffi cking and proper localization of postsynaptic GluRs. Although a number of studies address the subunit-specifi c traffi cking of AMPARs at the synapse (for reviews see Malinow and Malenka, 2002;Derkach, Oh et al. 2007;Greger et al. 2007), relatively little is known about how the receptors get transported to the synapse and anchored in the proper locations. The Drosophila genome encodes homologs of mammalian NMDA, AMPA, kainate, and delta receptor subunits (Sprengel et al. 2001). Therefore, an evolutionarily simpler system such as Drosophila could be used to dissect the function of genes and proteins that regulate GluR traffi cking. We searched the literature for proteins that regulate AMPARs or KARs and proteins that are found within the PSD. 95.8% of these proteins have Drosophila homologs. No homologs were found for 11 mammalian proteins. Interestingly, this included the scaffolding proteins Bassoon (Takao-Rikitsu, 2004) andAKAP 79/150 (Dell-Acqua et al. 2006). This may be due to the reduced complexity of the fl y NMJ (see below).
Several lines of evidence suggest these Drosophila homologs may have conserved functions. First, of the homologs we examined with documented expression patterns, 92.2% are found in neurons, muscle, or both, consistent with conserved function. Further, 31 of these homologs have been reported at the Drosophila NMJ, which is a glutamatergic synapse. Second, 29 of the homologs were recently identifi ed by mass spectrometry as members of a protein complex associated with the Drosophila NR2 GluR subunit (Emes et al. 2008). Third, two of the Drosophila homologs have been shown to regulate GluRs. Pak positively regulates GluR cluster formation at the NMJ when it is downstream of Dock (Albin and Davis 2004). Coracle, the Drosophila homolog of the mammalian 4.1 N protein (see Table 1), interacts with GluRIIA subunits and anchors A-type receptors to the actin cytoskeleton . Finally, four of the Drosophila homologs, Didum (Myosin Va), l(1)G0003 (Rab11 family interacting protein), Pnut (Cdc10 and Septin 7), and Polo (Polo-like kinase) were identifi ed in a forward genetic screen for genes that regulate GluR cluster formation (Liebl and Featherstone, 2005) at the Drosophila NMJ. We present evidence here that indicates that Pod1, the Drosophila homolog of Coronin 7 (see Supplemental Table 1), also regulates GluR cluster formation at the Drosophila NMJ.
The Coronins are an evolutionarily conserved family of proteins that regulate the actin cytoskeleton and vesicle transport (for reviews see Rybakin and Clemen, 2005;Uetrecht and Bear, 2006). Mammalian Coronins 1a (Collins et al. 2006), 1b, 1c (Peng et al. 2004;Collins et al. 2006), and 2b (Jordan et al. 2004;Collins et al. 2006) were identifi ed as components of the PSD via mass spectrometry. Coronin 7 is localized to the cis-Golgi and cytoplasmic vesicles (Rybakin et al. 2004). There are two Drosophila Coronin homologs. Coro is most similar to Coronins 1a, 1b, 1c, and 2b while Pod1 is most similar to Coronin 7. None of these proteins have been previously linked to GluRs. Previous studies in Drosophila (Rothenberg et al. 2003;Bharathi et al. 2004) and mammals (Rybakin and Clemen, 2005;Uetrecht and Bear, 2006), however, indicate that the coronins are expressed in the nervous system and/or muscle. This, coupled with their role in cytoskeleton remodeling, suggests they may be involved in GluR cluster formation. Consistent with this, we found Pod1 present at the NMJ (Fig. 1). It has also been shown to be localized in the tips of growing motor neuron axons during embryogenesis in Drosophila (Rothenberg et al. 2003).
We tested our hypothesis that Pod1 is involved in GluR cluster formation by examining pod1 mutant synapses. The loss of pod1 led to a reduction in the size of GluRIIA-containing clusters as   (Osterweil et al. 2005) and forms a complex with GluR1 and SAP-97 (Wu et al. 2002) Jar 53.2/71.5 (Continued)  well as a signifi cant reduction in synaptic GluRIIA immunoreactivity. Interestingly, the GluR cluster sizes determined microscopically do not differ between pod1 P1 and pod1 ∆17 despite the fact that mini amplitudes in pod1 ∆17 null mutants are much lower. Surface expression of some GluRIIA may therefore be supported in pod1 P1 mutants even when total synaptic GluRIIA is severely reduced. A-type receptors are linked to the actin cytoskeleton via their interaction with coracle . This raises the possibility that the loss of GluRIIA is specifi c to the synapse. In this scenario, A-type receptors would be traffi cked to the synapse but not properly anchored to the synapse in pod1 mutants. Alternatively, pod1 could be required for transport of GluRIIA-containing receptors from the cis Golgi to the synapse. Further studies will be required to determine how the loss of pod1 affects A-type receptor traffi cking. There was no signifi cant reduction in the sizes of GluRIIB or GluRIIC clusters. This is likely because B-type receptors are anchored to the cellular cytoskeleton in a different, unknown way. These data are consistent with the role of the coronins in mammals where they are known to regulate the actin cytoskeleton (Cai et al. 2008; for reviews see Rybakin and Clemen, 2005;Uetrecht and Bear, 2006) and suggests Coronin 7 may also participate in actin regulation. Although both A-and B-type receptors at the Drosophila NMJ are linked to microtubules , only A-type receptors depend on the integrity of the actin cytoskeleton .
There exist a number of important differences between mammalian central synapses and Drosophila NMJ synapses. First, the Drosophila NMJ is a single cell in vivo system where a single presynaptic motor neuron synapses on a single postsynaptic muscle cell. It is estimated that mammalian CNS neurons synapse with as many as 10,000 other neurons. Therefore, the Drosophila NMJ is a simple model system lacking the complexity found in mammalian CNS synapses. This could partly account for the small percentage of mammalian proteins with no Drosophila homologs. Second, because the postsynaptic cell at the NMJ is a muscle cell, Drosophila NMJs lack dendritic spines but extend fi lopodia to contact presynaptic motor neurons during embryonic development (Ritzenthaler et al. 2000;Ritzenthaler et al. 2003). Thus, proteins and mechanisms specifi c to dendritic spines are probably not included at the fl y NMJ. The NMJ, however, represents only a small percentage of fl y glutamatergic synapses. Most fl y glutamatergic synapses are found in the larval and adult CNS (Daniels et al. 2008). Consistent with this, many of the putative fl y PSD proteins identified here are expressed in the fly CNS. Glutamate receptors and PSD proteins in the fl y CNS probably function as in mammals. For example, similar to mammalian studies, central NMDA receptors are required for fl y learning (Glanzman, 2005;Lin, 2005;Xia et al. 2005;Wu et al. 2007). It is currently unknown whether fl y central synapses exhibit plasticity, but the NMJ exhibits post tetanic potentiation (Kuromi and Kidokoro, 2003;Cheung et al. 2006) and LTD (Guo and Zhong, 2006).
In conclusion, we have shown that most mammalian PSD proteins have Drosophila homologs and that these homologs are likely to have conserved functions. Therefore, the analysis of mutant phenotypes in Drosophila could enhance our understanding of GluR cluster formation and the PSD. Consistent with this, we have shown for the first time that the Drosophila homolog of Coronin 7, Pod1, is involved in the formation of GluRIIA containing GluR clusters possibly by regulating the actin cytoskeleton.