The AP2 clathrin adaptor protein complex regulates the abundance of GLR-1 glutamate receptors in the ventral nerve cord of Caenorhabditis elegans

Regulation of glutamate receptor trafficking controls synaptic strength and plasticity. This study takes advantage of viable, null mutations in subunits of the clathrin adaptor protein 2 (AP2) complex in Caenorhabditis elegans to reveal a novel and unexpected AP2-dependent trafficking step for glutamate receptors early in the secretory pathway.

Abstract

Regulation of glutamate receptor (GluR) abundance at synapses by clathrin-mediated endocytosis can control synaptic strength and plasticity. We take advantage of viable, null mutations in subunits of the clathrin adaptor protein 2 (AP2) complex in Caenorhabditis elegans to characterize the in vivo role of AP2 in GluR trafficking. In contrast to our predictions for an endocytic adaptor, we found that levels of the GluR GLR-1 are decreased at synapses in the ventral nerve cord (VNC) of animals with mutations in the AP2 subunits APM-2/μ2, APA-2/α, or APS-2/σ2. Rescue experiments indicate that APM-2/μ2 functions in glr-1–expressing interneurons and the mature nervous system to promote GLR-1 levels in the VNC. Genetic analyses suggest that APM-2/μ2 acts upstream of GLR-1 endocytosis in the VNC. Consistent with this, GLR-1 accumulates in cell bodies of apm-2 mutants. However, GLR-1 does not appear to accumulate at the plasma membrane of the cell body as expected, but instead accumulates in intracellular compartments including Syntaxin-13– and RAB-14–labeled endosomes. This study reveals a novel role for the AP2 clathrin adaptor in promoting the abundance of GluRs at synapses in vivo, and implicates AP2 in the regulation of GluR trafficking at an early step in the secretory pathway.

INTRODUCTION

Alterations in glutamate receptor (GluR) levels at the synapse by activity-dependent exo- and endocytosis can alter synaptic strength and affect learning and memory (Shepherd and Huganir, 2007). The adaptor protein 2 (AP2) complex and clathrin function together at synapses to mediate activity-dependent endocytosis of mammalian AMPA-type GluRs (AMPARs) (Carroll et al., 1999; Lüscher et al., 1999; Lin et al., 2000; Man et al., 2000; Wang and Linden, 2000; Lee et al., 2002). This AP2-dependent internalization of AMPARs is a critical mechanism underlying the induction of some forms of synaptic plasticity such as long-term depression (LTD) in hippocampal cultures (Man et al., 2000; Lee et al., 2002; Palmer et al., 2005; Unoki et al., 2012; Matsuda et al., 2013) and in vivo (Griffiths et al., 2008; Yoon et al., 2009).

Membrane protein sorting and trafficking between various intracellular compartments and the plasma membrane is mediated by several distinct adaptor protein (AP) complexes (Owen et al., 2004; Robinson, 2004). Five AP complexes, AP1–AP5, have been identified in mammals (Keen, 1987; Simpson et al., 1997; Dell'Angelica et al., 1999; Hirst et al., 2011), whereas the Caenorhabditis elegans genome encodes two alternative AP1 complexes, one AP2 complex and one AP3 complex (Lee et al., 1994; Grant and Hirsh, 1999; Shim et al., 2000; Shim and Lee, 2005). Each AP complex interacts with specific membrane phospholipids and protein cargoes via distinct sorting motifs to mediate vesicle formation at various intracellular membranes (Owen et al., 2004). AP2 associates with the plasma membrane in part via its interaction with the plasma membrane–enriched phospholipid, phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P₂) (Owen et al., 2004) and regulates clathrin-mediated endocytosis (CME; McMahon and Boucrot, 2011). Similar to other AP complexes, AP2 is a heterotetramer consisting of two large subunits, α and β2; one medium subunit, μ2; and one small subunit, σ2. In rodent neurons, the μ2 subunit of AP2 binds to the cytosolic tail of GluR2 subunits and promotes the CME of AMPARs at synapses (Man et al., 2000; Lee et al., 2002; Kastning et al., 2007). Acute treatment of neuronal cultures with peptides that interfere with AP2 binding to GluR2 can block AMPAR endocytosis and LTD (Lee et al., 2002; Kastning et al., 2007; Griffiths et al., 2008; Yoon et al., 2009). However, because knockout of AP2 subunits in mice and flies results in embryonic lethality (González-Gaitán and Jäckle, 1997; Mitsunari et al., 2005), the in vivo role of AP2 in AMPAR trafficking has not been investigated. In this study, we take advantage of viable, null mutations and strong loss-of-function mutations in several AP2 subunits in C. elegans to investigate the role of AP2 in AMPAR trafficking in vivo.

The C. elegans AMPAR GLR-1 is expressed in interneurons, where it localizes to sensory-interneuron and interneuron-interneuron synapses (Hart et al., 1995; Maricq et al., 1995; Rongo et al., 1998; Brockie et al., 2001; Burbea et al., 2002) and is required for several glutamate-dependent behaviors (Hart et al., 1995; Maricq et al., 1995; Zheng et al., 1999; Mellem et al., 2002; Chao et al., 2004). In this study, we identify an unexpected role for the AP2 complex in GLR-1 trafficking. We find that in contrast to increased GLR-1 at synapses in the ventral nerve cord (VNC), as has been observed in other endocytic mutants such as unc-11/AP180 clathrin adaptin mutants (Burbea et al., 2002), GLR-1 levels are reduced in the VNC of several AP2 subunit mutants. Although genetic analyses indicate that AP2 functions upstream of GLR-1 endocytosis in the VNC, the receptor does not accumulate at the plasma membrane of interneuron cell bodies as might be expected if CME were disrupted, but instead accumulates in intracellular compartments including Syntaxin-13– and RAB-14–labeled endosomes. This study characterizes, for the first time, the effect of AP2-null subunit mutations on GluR trafficking in vivo and reveals a novel role for AP2 in regulating GluR trafficking early in the secretory pathway.

RESULTS

AP2 functions in interneurons to promote GLR-1 abundance in the VNC

We use the genetic model organism C. elegans to identify genes and mechanisms that regulate AMPAR trafficking in vivo. We analyze the abundance of the AMPAR GLR-1 at synapses by measuring the distribution of a green fluorescent protein (GFP)-tagged version of GLR-1 (GLR-1::GFP). When expressed under the glr-1 promoter, GLR-1::GFP localizes in a punctate pattern within VNC interneurons (Rongo et al., 1998). More than 80% of these GLR-1::GFP puncta are closely apposed by presynaptic markers, suggesting that GLR-1::GFP is localized to postsynaptic sites (Rongo et al., 1998; Burbea et al., 2002). Additionally, expression of GLR-1::GFP under the glr-1 promoter rescues the behavioral defects of glr-1–null mutants, indicating that GLR-1::GFP forms a functional receptor (Rongo et al., 1998).

The abundance of mammalian AMPARs at the postsynaptic membrane is regulated by the clathrin adaptor AP2 and CME (Carroll et al., 1999; Lüscher et al., 1999; Lin et al., 2000; Man et al., 2000; Wang and Linden, 2000; Lee et al., 2002; Kastning et al., 2007). In C. elegans, disruption of the gene encoding the clathrin adaptin unc-11/AP180 causes defects in CME (Zhang et al., 1998; Nonet et al., 1999) and, subsequently, increased levels of GLR-1::GFP in the VNC (Burbea et al., 2002). We tested whether the clathrin adaptor AP2 was also involved in endocytosis of GLR-1 in the VNC by analyzing the abundance of GLR-1::GFP in animals with loss-of-function mutations in the μ2 subunit of AP2, apm-2 (also known as dpy-23). Surprisingly, we found that GLR-1::GFP puncta fluorescence intensity decreases by 26% (p < 0.001) in apm-2(gm17)/μ2 loss-of-function mutants compared with wild-type controls (Figure 1, A, B, and H; see Materials and Methods for quantification). The apm-2 mutation had a similar effect on GLR-1::GFP in a glr-1(n2461)–null mutant background (unpublished data). Histogram analysis shows that GLR-1::GFP puncta intensities decrease in apm-2 mutants across the entire population of VNC puncta analyzed (Figure 1I). In addition, mutations in a second independent null allele of apm-2(e840)/μ2 exhibited a 30% decrease in GLR-1::GFP in the VNC (p < 0.001; Figure 1, A, D, and H). Expression of mCherry-tagged apm-2 or untagged apm-2 cDNA under control of the glr-1 promoter corrects the GLR-1::GFP reduction observed in apm-2(gm17) or apm-2(e840) mutants, respectively (Figure 1, C, E, H, and I). These unexpected results led us to test whether the APM-2/μ2 subunit was functioning independently of the AP2 complex to regulate GLR-1 or whether other subunits of the AP2 complex contributed to this process. We found that loss-of-function mutants for the α subunit, apa-2(ox422), or the σ2 subunit, aps-2(tm2912), also had decreased GLR-1::GFP levels in the VNC (p < 0.001), and this effect was similar in magnitude (32–40%) to that observed in apm-2 mutants (Figure 1). We did not analyze AP2 β subunit (apb-1) mutants, because this subunit is shared between the AP1 and AP2 complexes in C. elegans (Shim and Lee, 2000; Boehm and Bonifacino, 2001). We also found that loss-of-function mutations in other adaptor proteins that act early in the secretory pathway, including the Golgi-localized AP1 subunit apm-1/μ1 (Shafaq-Zadah et al., 2012) or the sole C. elegans GGA (Golgi-localized, gamma adaptin ear–containing, ARF-binding) adaptor protein apt-9/GGA (Boehm and Bonifacino, 2001) had no effect (p > 0.05) on GLR-1::GFP puncta intensities in the VNC (Supplemental Figure S1). Taken together, these results indicate that AP2 functions in glr-1–expressing interneurons to promote GLR-1 abundance in the VNC.

FIGURE 1:

The clathrin adaptor AP2 promotes the abundance of GLR-1 in the VNC. (A–G) Representative images of GLR-1::GFP puncta in the anterior VNC of L4 wild-type (A), apm-2(gm17) (B), Pglr-1::mCherry::apm-2 rescued apm-2(gm17) (C), apm-2(e840) (D), Pglr-1::apm-2 rescued apm-2(e840) (E), apa-2(ox422) (F), and aps-2(tm2912) (G) animals expressing integrated GLR-1::GFP (nuIs24). All animals are oriented with anterior to the left and ventral to the top in these and all subsequent images (unless indicated otherwise). (H) Quantification of GLR-1::GFP puncta intensities (Norm., normalized) for the strains pictured in A–G. Shown are the means and SEMs for n = 121 wild-type, n = 57 apm-2(gm17), n = 21 Pglr-1::mCherry::apm-2 rescued apm-2(gm17) animals, n = 60 apm-2(e840), n = 27 Pglr-1::apm-2 rescued apm-2(e840), n = 25 apa-2(ox422), and n = 31 aps-2(tm2912). Values that differ significantly from wild type are indicated by asterisks above each bar, whereas other comparisons are marked by brackets (**, p < 0.001, Tukey-Kramer test). (I) Cumulative probability histogram of GLR-1::GFP puncta intensities (based on individual puncta intensities) for wild-type (thick black line), apm-2(gm17) (gray line), and Pglr-1::mCherry::apm-2 rescued apm-2(gm17) (Rescue) (thin black line) animals (p < 0.001 for wild type vs. apm-2(gm17); p < 0.001 for apm-2(gm17) vs. Rescue; p > 0.05 for wild type vs. Rescue, Kolmogorov-Smirnov test).

We tested whether the effects of AP2 mutation on GLR-1 were specific or whether AP2 mutation also reduced the abundance of other neurotransmitter receptors at synapses, such as the AChR α7 subunit ACR-16 (Francis et al., 2005) and the GABAR subunit UNC-49 (Bamber et al., 1999) at the neuromuscular junction (Bamber et al., 1999). We found that the fluorescence intensity of GFP-tagged ACR-16 and GFP-tagged UNC-49 increased at the neuromuscular junction of apm-2 mutants (Supplemental Figure S2). These results are consistent with the predicted role of AP2 in endocytosis at the synaptic plasma membrane and suggest that the ability of AP2 to promote GLR-1 levels in the VNC may be relatively specific.

glr-1 transcript levels are not reduced in apm-2 mutants

We tested whether the reduction in GLR-1::GFP observed in the VNC of AP2 subunit mutants was due to decreased transcription of glr-1. We measured the amount of glr-1 mRNA relative to act-1 (actin) mRNA by real-time PCR in wild type and apm-2(gm17) mutants. Instead of a decrease in glr-1 transcript levels, we observed a 2.3-fold increase in glr-1 mRNA in apm-2(gm17) mutants (p < 0.01) compared with wild-type controls (Figure 2). The increase in glr-1 mRNA may suggest a potential feedback mechanism in which reductions in synaptic GLR-1 trigger increases in glr-1 transcription. Nevertheless, this result indicates that the decrease in GLR-1::GFP in the VNC of apm-2 mutants is not due to a reduction in glr-1 transcript levels.

FIGURE 2:

GLR-1 mRNA levels are not reduced in apm-2 mutants. Real-time PCR analysis of mRNA levels in mixed-stage populations of wild-type and apm-2(gm17) animals expressing GLR-1::GFP (nuIs24). glr-1 and act-1 mRNA were detected by linear-range PCR amplification of total cDNA from each strain using gene-specific primers. Graph depicts the mean glr-1 to act-1 ratio (n = 9 replicates from three independent mRNA preps for each genotype) (Norm., normalized). Error bars show SEM. Values that differ significantly from wild type are indicated by an asterisk (*, p < 0.01, Student's t test).

The decrease in GLR-1 in the VNC of apm-2 mutants is not due to presynaptic changes in the number of synaptic vesicles

AP2 regulates synaptic vesicle (SV) endocytosis and recycling in neurons of C. elegans, Drosophila, and mammals (González-Gaitán and Jäckle, 1997; Gu et al., 2008, 2013; Dittman and Ryan, 2009). In C. elegans, the number of SVs at presynaptic nerve terminals in motor neurons is greatly reduced in apm-2 mutants (Gu et al., 2008, 2013). Because alterations in presynaptic activity can lead to changes in postsynaptic GLR-1 levels (Grunwald et al., 2004), we conducted two experiments to test whether the effects of apm-2 on SV numbers could indirectly affect GLR-1 in the VNC. First, we measured the abundance of GLR-1::GFP in the VNC of animals containing mutations in the C. elegans Stonin orthologue unc-41 (Mullen et al., 2012). Stonin family proteins interact with endocytic machinery and have been implicated in mediating SV endocytosis and recycling in C. elegans, Drosophila, and mammals (Fergestad et al., 1999; Martina et al., 2001; Walther et al., 2004; Kelly and Phillips, 2005; Diril et al., 2006). In C. elegans, unc-41 mutants exhibit reductions in SV numbers at presynaptic sites that are almost identical to apm-2 mutants (Gu et al., 2008; Mullen et al., 2012). Furthermore, unc-41; apm-2 double mutants have reductions in SV numbers that are indistinguishable from unc-41 single mutants (Mullen et al., 2012), indicating that unc-41 and apm-2 function in the same genetic pathway to regulate SV numbers. If the decrease in GLR-1::GFP in the VNC of apm-2 mutants was due to a postsynaptic response to reduced SVs, then we would expect unc-41 mutants to have a decrease in GLR-1::GFP in the VNC similar to that of apm-2 mutants. In contrast, we found that GLR-1::GFP puncta intensities were unaltered in the VNC of unc-41 mutants (p > 0.05) compared with wild-type controls (Supplemental Figure S3, A–C). These data suggest that the effect of AP2 on SV recycling cannot explain the decrease in GLR-1::GFP abundance in the VNC of apm-2 mutants. We also measured the distribution of GLR-1::GFP in the VNC of unc-104/KIF1A mutants. UNC-104 is required to transport SVs to presynaptic terminals in C. elegans neurons (Hall and Hedgecock, 1991), and the SV marker SNB-1::GFP is absent from synapses in the VNC of unc-104 mutants (Rongo et al., 1998). We found no difference in the fluorescence intensity of GLR-1::GFP in the VNC of wild type and unc-104 mutants (Supplemental Figure S3, D–F), consistent with findings from a previous study (Rongo et al., 1998). Together, these data suggest that the decrease in GLR-1 in the VNC of apm-2 is unlikely a secondary consequence of the presynaptic effects of AP2 on SV numbers.

APM-2 is not required during early development to regulate GLR-1 in the VNC

We next tested whether the decrease in GLR-1 abundance in the VNC of AP2 mutants was due to defects in early development. To test this, we expressed apm-2 cDNA under a heat shock– inducible promoter (hsp16.2) in apm-2(gm17) mutants expressing GLR-1::GFP. We raised animals at 20°C until the larval stage 4 (L4) of development and then heat shocked them at 30°C for 2.5 h to induce expression of apm-2 cDNA. Animals were subsequently shifted to 20°C for 17.5 h to recover before imaging of GLR-1::GFP in the VNC of adult animals. Under this temperature-shift paradigm, we found that apm-2(gm17) mutants exhibited a 30% decrease (p < 0.001) in GLR-1::GFP puncta intensities in the VNC compared with wild-type controls (Figure 3, A, B, and D), which is similar to that observed in non heat-shocked apm-2 mutants (Figure 1). In contrast, heat shock of apm-2(gm17) mutants expressing apm-2 (under control of the heat shock–inducible promoter) exhibited increased GLR-1::GFP puncta intensities in the VNC (p < 0.01; Figure 3, A, C, and D). These results suggest that APM-2 is not required during early development to regulate GLR-1. These data also indicate that APM-2 can function in the adult nervous system to promote and maintain GLR-1 levels in the VNC.

FIGURE 3:

APM-2 can act in the mature nervous system to regulate the abundance of GLR-1 in the VNC. (A–C) Representative images of the anterior VNC of adult wild-type (A), apm-2(gm17) (B), and heat-shock rescued apm-2(gm17) (C) animals expressing GLR-1::GFP (nuIs24) after 2.5 h of heat shock and 17.5 h of recovery. Heat-shock rescued apm-2 mutant animals express wild-type apm-2 cDNA under control of a heat shock–inducible promoter (pzEx259). (D) Quantification of GLR-1::GFP puncta intensities (Norm., normalized) for the strains pictured in A–C. Graph depicts mean and SEMs for n = 29 wild-type, n = 24 apm-2(gm17), and n = 27 heat shock–rescued apm-2(gm17) animals. Values that differ significantly from wild type are indicated by asterisks above each bar, whereas other comparisons are marked by brackets (*, p < 0.01, **, p < 0.001, Tukey-Kramer test).

APM-2 functions upstream of GLR-1 endocytosis in the VNC

GluRs are trafficked from the cell body to the VNC via vesicle-mediated transport and are subsequently delivered to the plasma membrane via exocytosis. At the synapse, GluRs are internalized via CME followed by either recycling back to the plasma membrane or degradation in the lysosome (Ehlers, 2000; Lin et al., 2000; Shepherd and Huganir, 2007). The reduction in GLR-1 abundance in the VNC of AP2 mutants could arise from a decrease in anterograde trafficking of GLR-1 from the cell body to the VNC, a decrease in postendocytic receptor recycling, or an increase in postendocytic receptor degradation. We tested whether AP2 functions upstream of GLR-1 endocytosis in the VNC by measuring GLR-1::GFP distribution in the VNC of apm-2; unc-11 double mutants. Mutations in the clathrin adaptin unc-11/AP180 block CME (Zhang et al., 1998; Nonet et al., 1999) and result in increased GLR-1::GFP abundance in the VNC (Burbea et al., 2002). If AP2 functions upstream of GLR-1 endocytosis in the VNC, then we would expect apm-2; unc-11 double mutants to have a similar reduction in GLR-1::GFP puncta intensities in the VNC, as observed in apm-2 single mutants. Conversely, if AP2 functions downstream of GLR-1 endocytosis in the VNC (i.e., to promote postendocytic recycling or inhibit degradation of GLR-1), then we would expect GLR-1::GFP to accumulate in the VNC of apm-2; unc-11 double mutants to an extent similar to that found in unc-11 single mutants. We found that apm-2; unc-11 double mutants exhibit a decrease in GLR-1::GFP puncta intensities in the VNC similar to apm-2 single mutants (p = 0.6; Figure 4, A–E). This result suggests that AP2 functions upstream of UNC-11/AP180–dependent endocytosis of GLR-1 in the VNC.

FIGURE 4:

APM-2 functions upstream of GLR-1 endocytosis in the VNC. (A–D) Representative images of GLR-1::GFP puncta in the anterior VNC of L4 wild-type (A), unc-11(e47) (B), apm-2(gm17) (C), and apm-2(gm17); unc-11(e47) (D) animals expressing GLR-1::GFP (nuIs24). (E) Quantification of GLR-1::GFP puncta intensities (normalized) for the strains pictured in A–D. Shown are the means and SEMs for n = 49 wild-type, n = 22 unc-11(e47), n = 57 apm-2(gm17), and n = 26 apm-2(gm17); unc-11(e47) animals. Values that differ significantly from wild type are indicated by asterisks above each bar, whereas other comparisons are marked by brackets (*, p < 0.01, Tukey-Kramer test). N.S. denotes no significant difference (p > 0.05). (F and G) Representative images of the anterior VNC of L4 wild-type (F) and apm-2(gm17) (G) animals expressing integrated GLR-1(4KR)::GFP (nuIs108). (H) Quantification of GLR-1(4KR)::GFP puncta intensities (Norm., normalized) for the strains pictured in F and G. Shown are the means and SEMs for n = 27 wild-type and n = 38 apm-2(gm17) animals. (I–K) Representative images of PVC interneuron cell bodies of L4 wild-type (I), apm-2(gm17) (J), and apm-2 rescued (K) animals expressing GLR-1::GFP (nuIs24). (L) Quantification of GLR-1::GFP mean fluorescence intensity (Norm., normalized) for the strains pictured in I–K. Shown are the means and SEM for n = 27 wild-type, n = 24 apm-2(gm17), and n = 23 apm-2 rescued animals. Values that differ significantly from wild type are indicated by asterisks above each bar (*, p < 0.01, **, p < 0.001, Tukey-Kramer test).

To further confirm that AP2 functions upstream of GLR-1 endocytosis and degradation in the VNC, we tested whether the levels of a nonubiquitinatable form of GFP-tagged GLR-1 (GLR-1(4KR)::GFP) are reduced in the VNC of apm-2 mutants. Ubiquitination of GLR-1 promotes CME and degradation of the receptor (Burbea et al., 2002). Mutation of all four lysines to arginines in the GLR-1 cytoplasmic tail prevents ubiquitination and results in the accumulation of GLR-1(4KR)::GFP in the VNC (Burbea et al., 2002). We found that GLR-1(4KR)::GFP fluorescence puncta intensities were decreased by 41% (p < 0.001) in the VNC of apm-2 mutants compared with wild-type controls (Figure 4, F–H). This result is similar to the decrease of GLR-1::GFP observed in the VNC of apm-2 mutants (Figure 1, A, B, D, and H). Taken together, these results suggest that AP2 functions upstream of GLR-1 endocytosis and postendocytic degradation in the VNC, consistent with a role for AP2 at an earlier trafficking step in the cell body.

GLR-1 accumulates in the neuronal cell bodies of APM-2 mutants

If AP2 regulates GLR-1 trafficking from the cell body to the synapse, then we would expect GLR-1 to accumulate in the interneuron cell bodies of AP2 subunit mutants. We tested this prediction by measuring the abundance of GLR-1::GFP in PVC interneuron cell bodies of apa-2 and apm-2 mutants. We found that GLR-1::GFP cell body fluorescence increased by 71% in apa-2(ox422) mutants (unpublished data) and by 65% in apm-2 (gm17) mutants compared with wild-type controls (p < 0.001; Figure 4, I, J, and L). This accumulation was partially rescued by expression of wild-type apm-2 cDNA (under control of the glr-1 promoter) in apm-2(gm17) mutant neurons (Figure 4, I–L). These results indicate that GLR-1 abundance is increased in the cell body of AP2 subunit mutants, consistent with a role for AP2 in regulating an early step in GLR-1 trafficking from the cell body to the VNC.

GLR-1 does not accumulate at the plasma membrane of the cell body in apm-2 mutants

Because GLR-1 decreases in the VNC and increases in the cell bodies of apm-2 mutants, we tested whether AP2 mediates GLR-1 endocytosis at the plasma membrane of the cell body. In this model, GLR-1 would traffic from the Golgi to the plasma membrane of the cell body and be subsequently internalized via an AP2-dependent step before being trafficked to synapses in the VNC. In support of this idea, MHC class II receptors and lysosome-associated membrane proteins can be delivered to the plasma membrane before AP2-dependent endocytosis and subsequent trafficking to their target endosome and lysosome compartments, respectively (Dugast et al., 2005; Janvier and Bonifacino, 2005; McCormick et al., 2005). In addition, GLR-1 has been detected on the surface of interneuron cell bodies (Zheng et al., 2004), but whether this plasma membrane pool of receptors are trafficked to the VNC in an AP2-dependent manner has not been tested.

To test whether APM-2 regulates GLR-1 levels at the surface of the cell body, we cultured neurons expressing GLR-1::GFP tagged with an extracellular hemagglutinin (HA) epitope (HA::GLR-1::GFP; Zheng et al., 2004) isolated from wild-type and apm-2(gm17) mutant animals, as previously described (Christensen et al., 2002; Zheng et al., 2004; Strange et al., 2007; see Materials and Methods). We measured surface HA::GLR-1::GFP levels with Alexa Fluor 594–conjugated anti-HA antibody under permeabilized and nonpermeabilized conditions. In control neurons, anti-HA–Alexa Fluor 594 antibody labeled only cell surface HA::GLR-1::GFP under nonpermeabilized conditions (Figure 5B), whereas the antibody labeled both surface and internal HA::GLR-1::GFP under permeabilized conditions (Figure 5A). Similar to the accumulation of GLR-1::GFP observed in apm-2 mutant cell bodies in vivo (Figure 4, I–L), we found that HA::GLR-1::GFP abundance increased in the cell body of apm-2(gm17) mutant neurons in culture (Figure 5, B and C; Norm. wild type: 1.00 ± 0.08 SEM [n = 30], apm-2(gm17): 1.32 ± 0.11 SEM [n = 30], p < 0.05). We also found that surface HA::GLR-1::GFP levels at puncta in the neuronal processes of nonpermeablized neurons were reduced in apm-2 mutants (average Alexa Fluor 594 puncta fluorescence intensity [Norm.]: wild type: 1.0 ± 0.09 [n = 22], apm-2: 0.72 ± 0.06 [n = 20], p < 0.01; unpublished data). These data suggest that apm-2 mutant neurons in culture exhibit decreased surface GLR-1 along neuronal processes and increased accumulation of GLR-1 in cell bodies, recapitulating our in vivo data (Figures 1 and 4). Under nonpermeabilized conditions, we found that the average Alexa Fluor 594 fluorescence intensity was unaltered at the cell surface of apm-2 mutants compared with wild-type controls (p = 0.8; Figure 5, B–D). Consistent with this result, we found that HA::GLR-1::GFP appears to accumulate in an internal compartment in apm-2 mutant cell bodies (Figure 5C, arrowhead), resulting in a decreased surface to total HA::GLR-1::GFP ratio (average Alexa Fluor 594 fluorescence intensity/GFP intensity [Norm.]: wild type: 1.0 ± 0.06; apm-2: 0.69 ± 0.04, p < 0.001). These results indicate that GLR-1 levels at the plasma membrane of the cell body are unaltered in apm-2 mutants, suggesting that AP2 does not function in GLR-1 endocytosis at the cell body surface.

FIGURE 5:

GLR-1 does not accumulate in the plasma membrane of the cell body of apm-2 mutants. (A) Representative images of dissociated neurons under permeabilized conditions isolated from wild-type worms expressing an integrated extracellular HA-tagged GLR-1::GFP receptor (HA::GLR-1::GFP) (pzIs12). Shown are the GFP (left panel), Alexa Fluor 594 (middle panel), and merged (right panel) channels. (B and C) Representative images of dissociated neurons under nonpermeabilized conditions isolated from wild-type (B) and apm-2(gm17) (C) worms expressing HA::GLR-1::GFP (pzIs12). Shown are the GFP (left panel), Alexa Fluor 594 (middle panel), and merged (right panel) channels. (D) Quantification of Alexa Fluor 594 mean puncta fluorescence intensity (Norm., normalized) from the cell bodies of dissociated neurons for wild type and apm-2(gm17) mutants. Shown are the means and SEMs for n = 30 wild-type and n = 30 apm-2(gm17) animals. Arrows mark red, surface HA::GLR-1::GFP signal; the arrowhead marks green, internal HA::GLR-1::GFP. N.S. denotes no significant difference (p > 0.05, Student's t test).

GLR-1 partially accumulates in endosomes of apm-2 mutants

Our data indicate that GLR-1::GFP accumulates in the cell bodies of apm-2 mutants (Figures 4, I–L, and 5C), but not at the cell surface (Figure 5, B–D), suggesting that GLR-1 accumulates in an internal compartment of apm-2 mutants. To identify this compartment, we analyzed the colocalization of GLR-1::GFP with a panel of red fluorescent protein (RFP)-tagged, compartment-specific markers using confocal microscopy (Chun et al., 2008). These markers have previously been used to study GLR-1 trafficking through intracellular compartments (Chun et al., 2008). We observed little colocalization between GLR-1::GFP and RFP-tagged compartment-specific SNARES, RFP::BET-1 (ERGIC), RFP::Syntaxin-5 (cis-Golgi), RFP::GOS-28 (Golgi), or RFP::Syntaxin-16 (trans-Golgi) in apm-2 mutants (Figure 6, A–H). In contrast, we found strong colocalization between GLR-1::GFP and the early/recycling endosome markers RFP::Syntaxin-13 (Prekeris et al., 1998; Figure 6, I and J) and mCherry::RAB-14 (Junutula et al., 2004; Kelly et al., 2010; Figure 6, K and L) in apm-2 mutants. Although quantification revealed that there was no change in the percentage of GLR-1::GFP pixels colocalized with RFP::Syntaxin-13 or mCherry::RAB-14 in apm-2 mutant cell bodies compared with wild-type controls (Figure 6, I–M; see Materials and Methods), we found that the mean intensity of GLR-1::GFP pixels colocalized with either RFP::Syntaxin-13– or mCherry::RAB-14–labeled endosomes increased by ∼90% in apm-2 mutants (p < 0.001) compared with wild-type controls (Figure 6N; see Materials and Methods). In contrast, we found no change in GLR-1::GFP pixel intensity in RFP::BET-1– or RFP::Syntaxin-16–labeled compartments in apm-2 mutants (Figure 6N). Notably, we observed no change in the size of RFP::Syntaxin-13– (mean RFP::Syntaxin-13 pixel area [μm²]: wild-type: 3.37 ± 0.23, apm-2: 4.21 ± 0.37, p > 0.05) or mCherry::RAB-14–labeled endosomes (mean mCherry::RAB-14 pixel area [μm²]: wild type: 4.48 ± 0.56, apm-2: 3.83 ± 0.41, p > 0.05) in apm-2 mutants compared with controls. The increase in GLR-1::GFP intensity in Syntaxin-13– and RAB-14–labeled compartments suggests that GLR-1 accumulates in endosomes in apm-2 mutant cell bodies. In addition, we observed a decrease in the percentage of GLR-1::GFP pixels colocalized with the trans-Golgi network (TGN) marker RFP::Syntaxin-16 in apm-2 mutants versus wild-type controls (p < 0.01; Figure 6, G, H, and M), whereas there was no change in the percentage of GLR-1::GFP colocalized with the ERGIC marker RFP::BET-1 (p > 0.05; Figure 6, A, B, and M). It should be noted that, in addition to the accumulation of GLR-1::GFP in Syntaxin-13– and RAB-14-labeled endosomes, we also observed an accumulation of GLR-1::GFP in another compartment that was not labeled by our panel of organelle markers and thus remains to be identified. Nevertheless, these results indicate that, in the absence of apm-2, GLR-1 accumulates in the cell body in internal compartments, including Syntaxin-13– and RAB-14–labeled endosomes.

FIGURE 6:

GLR-1 partially accumulates in Syntaxin-13– and RAB-14–positive compartments in the cell body of apm-2 mutants. (A–L) Representative images of PVC interneuron cell bodies of L4 wild-type (A, C, E, G, I, and K) and apm-2(gm17) (B, D, F, H, J, and L) animals expressing GLR-1::GFP (nuIs24) and RFP-tagged markers for specific organelles including RFP::BET-1 (ERGIC) (A and B), RFP::Syntaxin-5 (cis-Golgi) (C and D), RFP::GOS-28 (Golgi) (E and F), RFP::Syntaxin-16 (trans-Golgi) (G and H), RFP::Syntaxin-13 (early/recycling endosomes) (I and J), and mCherry::RAB-14 (early/recycling endosomes) (K and L). (M) Quantification of the percentage of GLR-1::GFP pixels colocalized with RFP::BET-1, RFP::Syntaxin-16, RFP::Syntaxin-13 and mCherry::RAB-14 in wild-type and apm-2(gm17) animals. Graph depicts mean and SEM values for n = 20 wild type and n = 24 apm-2 (gm17) for RFP::BET-1, n = 23 wild type and n = 24 apm-2(gm17) for RFP::Syntaxin-16, n = 26 wild type and n = 33 apm-2 (gm17) for RFP::Syntaxin-13, and n = 20 wild type and n = 25 apm-2(gm17) for mCherry::RAB-14. (N) Quantification of the mean intensity of GLR-1::GFP pixels colocalized with RFP::BET-1, RFP::Syntaxin-16, RFP::Syntaxin-13 and mCherry::RAB-14 (Norm., normalized) in wild-type and apm-2(gm17) animals. Graph depicts mean and SEM values for the n values indicated for M. Arrowheads indicate strong colocalization. Values that differ significantly from wild type are indicated by asterisks above each bar (*, p < 0.01, **, p < 0.001, Student's t test).

Because the μ2 subunit of AP2 has been shown to interact with mammalian GluR2 receptors in vitro (Lee et al., 2002; Kastning et al., 2007), we tested whether APM-2::mCherry could colocalize with GLR-1::GFP in vivo. We found that APM-2::mCherry localized to puncta in interneuron cell bodies, and some of these puncta colocalized with GLR-1::GFP (Supplemental Figure S4). The accumulation of GLR-1 in intracellular compartments in apm-2 mutants prompted us to test whether APM-2::GFP could localize to intracellular compartments labeled by our panel of RFP-tagged organelle markers described above. We found that APM-2::GFP partially colocalized with RFP::Syntaxin-13– (Figure 7A) and RFP::Syntaxin-16– (Figure 7B) labeled compartments, but not with RFP::Syntaxin-5– (Figure 7C) or RFP::GOS-28– (Figure 7D) labeled compartments. These data suggest that APM-2 can localize to intracellular compartments, where it may function to regulate GLR-1 trafficking.

FIGURE 7:

APM-2 localizes to intracellular compartments. (A–D) Representative confocal images of PVC interneuron cell bodies in L4 wild-type animals expressing APM-2::GFP and RFP-tagged markers for specific organelles including RFP::Syntaxin-13 (early and recycling endosomes) (A), RFP::Syntaxin-16 (trans-Golgi) (B), RFP::Syntaxin-5 (cis-Golgi) (C), and RFP::GOS-28 (Golgi) (D). Green channel (left panel), red channel (middle panel), and merged (right panel) images of cell bodies are shown for each pair of markers. Arrowheads indicate colocalization.

The μ2 subunit of AP2 interacts in vitro with a stretch of basic amino acids in the cytoplasmic tail of rat GluR2, and this interaction is completely abolished by mutation of a single conserved basic residue, Lys-844 to alanine (Kastning et al., 2007). Because this basic residue is conserved in the cytoplasmic tail of GLR-1 as Arg-887, we mutated this arginine to alanine (GLR-1(R887A)) to disrupt the predicted binding to μ2 and tested whether this point mutation would affect the trafficking of the receptor in wild-type animals. For this experiment, we generated transgenic lines expressing either GLR-1(WT)::GFP or GLR-1(R887A)::GFP at comparable levels as determined by Western blotting using anti-GFP antibodies on total worm lysates (Figure 8, A and B). To isolate the effects of the R887A mutation, we conducted our analysis in the background of a glr-1; glr-2 double mutant, because endogenous GluR subunits have been shown to oligomerize with other mutant versions of GLR-1 and traffic them to the VNC (Chang and Rongo, 2005). We found that GLR-1(R887A)::GFP fluorescence decreased at puncta in the VNC (Figure 8, C and D; average puncta fluorescence [Norm.]: GLR-1(WT)::GFP: 1.0 ± 0.02 [n = 30], GLR-1(R887A)::GFP: 0.8 ± 0.04 [n = 26], p ≤ 0.01), and increased in the cell body (Figure 8, E and F; average cell body fluorescence [Norm.]: GLR-1(WT)::GFP: 1.0 ± 0.04 [n = 30], GLR-1(R887A)::GFP: 1.3 ± 0.08 [n = 26], p < 0.01) of wild-type neurons relative to GLR-1(WT)::GFP. These effects were similar to those observed for wild-type GLR-1::GFP in the VNC and cell body of apm-2 mutants (Figures 1 and 4). Furthermore, we compared the relative distributions of GLR-1(WT)::GFP and GLR-1(R887A)::GFP by calculating the ratio of GFP fluorescence in the VNC to the cell body to internally control for any potential differences in expression between the two transgenes. We found that the VNC to cell body ratio of GLR-1(R887A)::GFP fluorescence was reduced by 37% compared with the ratio of GLR-1(WT)::GFP (Figure 8G). These data are consistent with a model in which APM-2 functions at an internal compartment such as endosomes to promote GLR-1 trafficking from the cell body to the VNC.

FIGURE 8:

GLR-1(R887A)::GFP accumulates in the cell body and decreases in the VNC. (A) Western blot probed with anti-GFP antibodies (top gel) and antitubulin antibodies (bottom gel) of total worm lysates from the indicated strains showing comparable levels of transgene expression. The ratio of each GLR-1::GFP transgene to tubulin is indicated at the bottom of the gel and graphed in B from three independent lysates of each strain. The ratio for nuIs24 was normalized (Norm.) to one. (C and D) Representative images of the anterior VNC of L4 glr-1(ky176) glr-2(ak10) animals expressing GLR-1(WT)::GFP (pzEx223) (C) or GLR-1(R887A)::GFP (pzEx214) (D). (E and F) Representative images of PVC interneuron cell bodies of L4 glr-1(ky176) glr-2(ak10) animals expressing GLR-1(WT)::GFP (pzEx223) (E) or GLR-1(R887A)::GFP (pzEx214) (F). (G) Quantification of the ratio of GLR-1::GFP mean fluorescence intensity in the VNC to the cell body for the strains pictured in C–F. Shown are the normalized (Norm.) ratios and SEM for n = 30 GLR-1(WT)::GFP and n = 26 GLR-1(R887A)::GFP–expressing animals. Values that differ significantly are indicated by an asterisk (*, p < 0.01).

DISCUSSION

Our data show that AP2 functions in interneurons to promote the abundance of GLR-1 at synapses in the VNC. We found that loss-of-function mutations in the AP2 subunits apm-2/μ2, apa-2/α, or aps-2/σ2 result in decreased levels of GLR-1::GFP in the VNC (Figure 1). The defect observed in apm-2 mutants can be rescued by expression of wild-type apm-2 cDNA in interneurons using the glr-1 promoter (Figure 1). In addition, inducible expression of apm-2 in the mature nervous system can rescue the reduction in GLR-1::GFP in the VNC (Figure 3), suggesting that APM-2 can function in adults to maintain GLR-1 levels. Genetic double-mutant analyses with the clathrin adaptin unc-11/AP180 suggest that AP2 acts upstream of GLR-1 endocytosis in the VNC. We found that mutations in apm-2/μ2 block the accumulation of GLR-1 observed in the VNC of unc-11/AP180 endocytic mutants (Figure 4, A–E; Burbea et al., 2002). Similarly, the VNC accumulation of a nonubiquitinatable version of GLR-1, GLR-1(4KR)::GFP (Burbea et al., 2002), is reduced in apm-2 mutants (Figure 4, F–H), suggesting that apm-2 does not act at a postendocytic step to prevent GLR-1 degradation. These data suggest that AP2 acts upstream of GLR-1 endocytosis and postendocytic degradation in the VNC. Finally, GLR-1::GFP accumulates in cell bodies of apm-2 mutants (Figure 4, I–L), consistent with the idea that AP2 regulates a trafficking step early in the secretory pathway to promote GLR-1 trafficking.

Our results were unexpected, because previous work using rodent neurons in culture demonstrated that AP2 functions at synapses to regulate CME of AMPARs (Lee et al., 2002; Kastning et al., 2007; Matsuda et al., 2013). These studies utilized strategies that blocked but did not ablate AP2 function, including the use of inhibitory peptides to disrupt AP2 binding to AMPAR subunits (Lee et al., 2002; Kastning et al., 2007) or small interfering RNA to knock down specific AP2 subunits (Matsuda et al., 2013). The effect of AP2 subunit mutants on GluR trafficking has not been studied in vivo, because null mutations in AP2 subunits in mice result in embryonic lethality (Mitsunari et al., 2005). Thus, a role for AP2 in regulating GluR trafficking earlier in the secretory pathway in mammals may have been missed. In C. elegans, AP2 may also act at synapses to regulate GLR-1 endocytosis, but this role of AP2 may be masked in our subunit mutants, because AP2 is required at an earlier trafficking step in the cell body.

The μ2 subunit of AP2 can bind directly to an atypical motif in the cytoplasmic tail of mammalian GluR2 (Kastning et al., 2007), and this motif is conserved in GLR-1 (Burbea et al., 2002). Because AP2 is known to associate with the plasma membrane and regulate CME (Owen et al., 2004; McMahon and Boucrot, 2011), and our data show that GLR-1 accumulates in the cell body of apm-2 mutants (Figures 4–6), we tested whether GLR-1 accumulates at the plasma membrane in apm-2 mutant cell bodies. In this model, GLR-1 would traffic from the Golgi to the surface of the cell body before being internalized by an AP2-dependent step and trafficked to the VNC. Indeed, several studies show that membrane proteins can traffic via the plasma membrane en route to their ultimate destination (Dugast et al., 2005; Janvier and Bonifacino, 2005; McCormick et al., 2005; Lasiecka and Winckler, 2011; Kaplan et al., 2012). However, we did not observe an increase in HA-tagged GLR-1::GFP levels at the cell surface of apm-2 mutant cell bodies (Figure 5). Instead, we found that GLR-1 accumulates in intracellular compartments in apm-2 mutant cell bodies. Colocalization experiments with GLR-1::GFP and a panel of RFP-tagged organelle markers (Chun et al., 2008) indicate that GLR-1 accumulates, in part, in Syntaxin-13– and RAB-14–labeled endosomes (Figure 6). These results illustrate that, in the absence of apm-2, GLR-1 is not efficiently trafficked to the VNC and accumulates in endosomes in the cell body.

Although AP2 does not appear to regulate surface levels of GLR-1 at the cell body, AP2 could still function in a canonical manner at the plasma membrane to indirectly regulate GLR-1 by controlling the internalization of another transmembrane protein required for forward GLR-1 trafficking. Indeed, AP2-dependent endocytosis of the Wnt secretion factor MIG-14/Wntless followed by retromer-dependent trafficking back to the Golgi is required for the forward trafficking of Wnt to the cell periphery in C. elegans (Bänziger et al., 2006; Pan et al., 2008), and retromer is required for GLR-1 trafficking (Zhang et al., 2012). However, our data support a more direct role for AP2 in regulating GLR-1 trafficking. We found that APM-2 is localized to puncta that partially colocalize with GLR-1 (Supplemental Figure S4), the endosomal marker Syntaxin-13, and the TGN marker Syntaxin-16 (Figure 7). Furthermore, the distribution of GLR-1(R887A)::GFP, which contains a point mutation predicted to disrupt binding to μ2, phenocopies the distribution of wild-type GLR-1::GFP in apm-2 mutants (i.e., decreases in the VNC and accumulates in the cell body of wild-type neurons; Figure 8). These data suggest that APM-2 can localize to intracellular compartments in the cell body, where it may function in a noncanonical manner to regulate some aspect of GLR-1 trafficking between the TGN and endosomal compartments.

We tested whether clathrin might function together with AP2 to regulate GLR-1 by expressing a dominant-negative (dn) version of clathrin heavy chain (clathrin hub mutant) that is known to disrupt clathrin function (Liu et al., 1998). In contrast to AP2 subunit mutants, expression of dn-clathrin results in increased levels of GLR-1::GFP in the VNC (Supplemental Figure S5). These data could suggest that AP2 regulation of GLR-1 trafficking does not require clathrin, since AP2 subunit mutants and dn-clathrin expression result in distinct phenotypes. However, we favor an alternative interpretation, wherein clathrin functions together with UNC-101/μ1 (one of two alternative μ1 subunits of AP1) to regulate GLR-1 trafficking at a step before AP2 function. A previous study showed that GLR-1::GFP mislocalizes to presynaptic regions of the polarized neuron RIA in unc-101/μ1 mutants (Margeta et al., 2009). We found that GLR-1::GFP increases in the VNC of unc-101/μ1 mutants similar to dn-clathrin–expressing animals (Supplemental Figure S5). Because the VNC interneurons are unipolar and have pre- and postsynaptic regions along the VNC, we speculate that GLR-1::GFP is mislocalized, perhaps to presynaptic regions in the VNC of unc-101/μ1 and dn-clathrin–expressing animals. Genetic double-mutant analysis revealed that expression of dn-clathrin or mutation of unc-101/μ1 completely block the reduction of GLR-1 observed in the VNC of apm-2 single mutants (Supplemental Figure S5). These data suggest that unc-101/μ1 and clathrin function upstream of AP2 to regulate GLR-1 trafficking and are consistent with a model whereby disruption of clathrin or unc-101/μ1 results in mislocalization of GLR-1. This model is supported by several other studies demonstrating that, in the absence of AP1 function, target cargoes can become rerouted to alternate cellular destinations. For example, in C. elegans, mutations in unc-101/μ1 result in missorting of the ODR-10 odorant receptor and transient receptor potential channels from neuronal sensory cilia to axons of chemosensory neurons (Dwyer et al., 2001; Bae et al., 2006). Similarly, in mammalian hippocampal neurons, disruption of AP1 results in the mislocalization of N-methyl-d-aspartate receptor subunits to axons, and disruption of AP1 and clathrin result in the misincorporation of transferrin receptors into axonal carriers and inappropriate trafficking to axons (Farias et al., 2012). Clathrin may still function with AP2 at an intracellular compartment to regulate GLR-1 trafficking; however, the requirement for clathrin and AP1 at an earlier trafficking step precludes our ability to test this idea.

While localization of AP2 to the plasma membrane is largely dependent on its interaction with the phosphoinositide PtdIns(4,5)P₂, which is enriched in the plasma membrane (Di Paolo and De Camilli, 2006), several studies suggest that PtdIns(4,5)P₂ and AP2 may also function on intracellular compartments. For example, low levels of PtdIns(4,5)P₂ can be detected on several organelles, including the Golgi, endosomes, and lysosomes (Traub et al., 1996; Arneson et al., 1999; Watt et al., 2002), and the kinases and phosphatases that regulate PtdIns(4,5)P₂ turnover have been found at the Golgi (Godi et al., 1999; Jones et al., 2000; De Matteis et al., 2002; Choudhury et al., 2005; Hyvola et al., 2006). Interestingly, several noncanonical roles for AP2 have begun to emerge. A recent study in C. elegans showed that AP2 and clathrin function in a noncanonical manner to regulate both the phagocytosis of apoptotic cells and the maturation of phagosomes (Chen et al., 2013). Another recent study showed that AP2 and AP4 can localize to autolysosomes during macroautophagy. Phosphatidylinositol 5-kinase generates microdomains of PtdIns(4,5)P₂ on autolysosomes to recruit AP2 and AP4 during lysosome re-formation after macroautophagy (Rong et al., 2012). Although macroautophagy is often associated with cellular stress or starvation, increasing evidence suggests that macroautophagy also occurs under normal growth conditions (Ravikumar et al., 2010; Maday et al., 2012). Because the C. elegans genome does not encode an AP4 complex, and AMPARs mislocalize to axons and accumulate in intracellular compartments resembling autophagosomes in AP4-deficient hippocampal neurons (Matsuda et al., 2008), further study will be needed to determine whether AP2 regulation of macroauto­phagy contributes to GLR-1 trafficking in vivo.

In summary, the analysis of viable null and loss-of-function mutants in AP2 subunits in this study revealed a novel function for AP2 in regulating GluR trafficking at an early step in the secretory pathway. It will be interesting in the future to test whether this AP2-dependent trafficking step is regulated by neuronal activity as a mechanism to control synaptic GluR levels.

MATERIALS AND METHODS

Strains

The following strains were used in this study: N2 (Bristol), nuIs24 (Pglr-1::glr-1::gfp), nuIs25 (Pglr-1::glr-1::gfp), nuIs108 (Pglr-1::glr-1(4kr)::gfp), pzIs12 (Pglr-1::ha::glr-1::gfp), oxIs22 (Punc-49::unc-49::gfp), ufIs8 (Pmyo-3::acr-16::gfp) (kindly provided by M. Francis [University of Massachusetts Medical School, Worcester, MA]), pzEx74 (Pglr-1::apm-2), pzEx201 (Pglr-1::apm-2::mcherry), pzEx214 (Pglr-1::glr-1(r887a)::gfp), pzEx223 (Pglr-1::glr-1::gfp), pzEx235 (Pglr-1::apm-2::mcherry), pzEx259 (Phsp-16.2::apm-2), pzEx278 (Pvha-6::apm-1::mcherry), pzEx283 (Pglr-1::dn-clathrin), pzEx347 (Pglr-1::mcherry::rab-14), nuEx1330 (Pglr-1::rfp::bet-1), nuEx1331 (Pglr-1::rfp::gos-28), nuEx1332 (Pglr-1::rfp::syntaxin-16), nuEx1337 (Pglr-1::rfp::syntaxin-13), nuEx1340 (Pglr-1::rfp::syntaxin-5), apa-2 (ox422) (kindly provided by E. Jorgensen [University of Utah, Salt Lake City, UT]), apm-1 (ok2578), apm-2 (gm17), apm-2 (e840), aps-2 (tm2912), apt-9 (tm3776), glr-1(n2461), glr-1 (ky176) and glr-2 (ak10) (kindly provided by A. V. Maricq [University of Utah, Salt Lake City, UT]), unc-11 (e47), unc-41 (e268), unc-101 (m1), and unc-104 (e1265). All strains were maintained at 20°C as previously described (Brenner, 1974).

Constructs, transgenes, and germ-line transformation

Plasmids were generated using standard recombinant DNA techniques, and transgenic strains were created by plasmid microinjection. Details are available upon request. nuIs24, nuIs25, nuIs108, nuEx1329, nuEx1330, nuEx1331, nuEx1332, nuEx1337, and nuEx1340 (kindly provided by Josh Kaplan [Massachusetts General Hospital, Boston, MA]; Rongo et al., 1998; Burbea et al., 2002; Chun et al., 2008), oxIs22 (Bamber et al., 1999), and pzIs12 (Kowalski et al., 2011) have been previously described. Pglr-1::apm-2 (FJ#122) was made by PCR amplifying the apm-2 coding sequence from N2 cDNA, and inserting into the pV6 expression vector using NheI/KpnI restriction sites to create plasmid FJ#122. pzEx74 was generated by injecting Pglr-1::apm-2 at 25 ng/μl along with the coinjection marker Pmyo2::nls::mcherry at 10 ng/μl. Pglr-1::apm-2::mcherry (FJ#123) was created by subcloning mCherry flanked by NotI restriction sites into a plasmid containing Pglr-1::apm-2 with a NotI restriction site engineered immediately before the apm-2 stop codon. pzEx235 was generated by injecting Pglr-1::apm-2::mcherry (FJ#123) at 50 ng/μl alongside the coinjection marker Pmyo2::nls::mcherry at 10 ng/μl. Phsp-16.2::apm-2 (FJ#124) was generated by subcloning the apm-2 coding sequence from Pglr-1::apm-2 into the NheI/KpnI restriction sites of the pPD49.78 expression vector, which contains the hsp-16.2 heat shock–inducible promoter (gift from Andrew Fire, Stanford University, Palo Alto, CA). pzEx259 was generated by injecting Phsp-16.2::apm-2 at 25 ng/μl alongside the coinjection marker Pmyo2::nls::mcherry at 10 ng/μl. Pvha-6::apm-1::mcherry (FJ#125) was generated by subcloning a 2.9-kb of sequence immediately upstream of the start codon of vha-6 (Pvha-6) into the pPD49.26 expression vector using HindIII/XbaI restriction sites. Apm-1 coding sequence was then PCR amplified from N2 cDNA and inserted into the pPD49.26 expression vector immediately after Pvha-6 using the NheI/KpnI restriction sites. mCherry flanked by NotI restriction sites was then subcloned into the pPD49.26 expression vector containing Pvha-6 and apm-1 coding sequence immediately before the apm-1 stop codon. pzEx278 was generated by injecting Pvha-6::apm-1::mcherry at 25 ng/μl alongside the coinjection marker Pmyo2::nls::mcherry at 10 ng/μl. Pglr-1::glr-1(r887a)::gfp (FJ#127) was made by PCR amplifying a plasmid containing Pglr-1::glr-1::gfp (FJ#126) with primers containing a “GCG” sequence at the site corresponding to glr-1 R887 (AGG), using a QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). The resulting plasmid was then digested with NcoI, and the ∼2.8-kb fragment containing sequence encoding R887A was then subcloned into the NcoI digested Pglr-1::glr-1::gfp plasmid and subsequently sequenced. pzEx214 was made by injecting Pglr-1::glr-1(r887a)::gfp at 50 ng/μl. pzEx223 was made by injecting Pglr-1::glr-1::gfp at 50 ng/μl. The glr-1 promoter used in pzEx214 and pzEx223 consists of a ∼2.3-kb fragment which starts immediately 5′ from the start codon and ends just before a HindIII site. Pglr-1::dn-clathrin (chc-hub) was made by PCR amplifying a region of the chc-1 (clathrin heavy chain) coding region corresponding to residues 1075–1681 (which are homologous to those reported in Liu et al., 1998) from N2 cDNA, and subcloning the fragment under the control of the glr-1 promoter in pV6 using the NheI/KpnI restriction sites. pzEx283 was generated by injecting Pglr-1::dn-clathrin at 25 ng/μl alongside the coinjection marker Pmyo2::nls::cherry at 10 ng/μl. pzEx347 was made by injecting Pglr-1:: mcherry::rab-14 (KP#1541, gift from Josh Kaplan) at 25 ng/μl alongside the coinjection marker ttx-3::dsred at 75 ng/μl.

Fluorescence imaging

All VNC imaging experiments were performed with a Carl Zeiss AxioImager M1 microscope (Oberkochen, Germany) with a 100× Plan-Apochromat objective (1.4 numerical aperture) with GFP and RFP filter cubes. Images were acquired with an Orca-ER charge-coupled device camera (Hamamatsu, Hamamatsu, Japan), using MetaMorph, version 7.1 software (Molecular Devices, Sunnyvale, CA). All L4 animals were immobilized with 30 mg/ml 2,3-butanedione monoxamine (Sigma-Aldrich, St. Louis, MO) for 6–8 min before imaging. For heat-shock rescue experiments, L4 animals were shifted from 20ºC to 30ºC for 2.5 h. After heat shock, animals recovered at 20ºC for 17.5 h before imaging. All images were acquired from the anterior portion of the VNC, posterior to RIG and AVG neuronal cell bodies. To quantitate fluorescent puncta in the VNC, we made maximum-intensity projections of z-series stacks (1-μm depth). Exposure settings and gain were optimized to fill the 12-bit dynamic range without saturation and were set identically for all images acquired within an experiment with a given fluorescent marker. Line scans of VNC puncta were made using MetaMorph, version 6.0 software. Line-scan data were analyzed with Igor Pro, versions 5 and 6 (Wavemetrics, Portland, OR), using custom-written software (gift from Jeremy Dittman, Weill Cornell Medical College, New York) as described previously (Burbea et al., 2002). For normalization of arc lamp output, the fluorescence intensity of 0.5-μm FluoSphere beads (Invitrogen, Carlsbad, CA) was measured for each day of imaging. Puncta intensities were normalized to the average bead intensity for the corresponding day. Graphs of puncta intensities display intensity values normalized to wild-type controls. Puncta widths were calculated by measuring the width of each punctum at 50% of the maximal peak fluorescence intensity. Puncta densities were measured by calculating the average number of puncta per 10 μm along the VNC. All imaging quantification in the VNC is reported with average ± SEM values, and statistical significance was determined using Student's t test for experiments with two genotypes or the Tukey-Kramer test for experiments comparing more than two genotypes. For quantification of the amount of total GLR-1::GFP or surface HA::GLR-1::GFP in neuronal cell bodies, images were taken of PVC cell bodies with GFP or RFP filters, and maximum-intensity projections were made from z-series stacks (2-μm depth) as previously described (Juo et al., 2007; Kowalski et al., 2011; Monteiro et al., 2012). The average pixel intensities of three distinct regions of each cell body were measured using MetaMorph, version 6.0 software. All imaging quantification in PVC cell bodies is reported with average ± SEM values, and statistical significance was determined using Student's t test. To quantify the amount of total or surface HA::GLR-1::GFP in neuronal processes in C. elegans neuronal cultures, we first identified neurons with distinct, well-formed processes. Line scans of puncta in these processes were then generated and analyzed as described above.

Fluorescence imaging of GLR-1::GFP and the various RFP-tagged compartment markers or APM-2::mCherry in PVC interneuron cell bodies was performed using a Zeiss LSM510 confocal microscope with a 63× objective (NA 1.4). L4 animals were immobilized with 30 mg/ml 2,3-butanedione monoxamine as described above. Fluorescent puncta were quantitated from a single representative plane containing both GFP and RFP signals using Zeiss LSM, version 4.2 software. Zeiss LSM software was used to quantitate the percent colocalized and the mean intensity of GLR-1::GFP. Background thresholds were first applied to each image using the threshold function to eliminate background fluorescence before analysis of an image region of interest (ROI) drawn around the perimeter of the cell body. The percentage and mean intensity of GLR-1::GFP pixels colocalized with specific organelle markers within the image ROI were calculated in LSM software for each image. Mean percentage and intensity of colocalized GLR-1::GFP pixels were then calculated for each strain by averaging the percentage and mean intensity values determined for n ≥ 20 animals. All imaging quantification in PVC cell bodies is reported as average ± SEM values, and statistical significance was determined using Student's t test.

Real-time PCR

Total RNA was isolated from mixed-stage wild-type (nuIs24) and apm-2 mutant (nuIs24; apm-2 (gm17)) strains using the RNeasy Fibrous Tissue Mini kit (Qiagen, Valencia, CA) as previously described (Kowalski et al., 2011; Monteiro et al., 2012). Three independent RNA preparations were isolated for each genotype. cDNA from these RNA preps was synthesized using Superscript III Reverse Transcriptase (Invitrogen). Real-time PCR was performed on the MX3000P real-time PCR machine (Tufts Center for Neuroscience Research, Boston, MA) using the Brilliant SYBR Green QPCR Master Mix and SureStart Taq (Stratagene, Santa Clara, CA). The relative amount of glr-1 mRNA compared with act-1 mRNA was determined as previously described (Pfaffl, 2001), after generating standard curves to calculate the efficiency of glr-1 and act-1 primers. The ratio of glr-1 mRNA to act-1 mRNA in apm-2 (gm17) mutants was normalized to that of wild-type animals with average ± SEM values reported. Statistical significance was determined using the Student's t test (n = 3 replicates of three independent cDNA preps for each genotype).

Primary culture of C. elegans neurons

Neurons were isolated from wild type and apm-2 mutants expressing Pglr-1::ha::glr-1::gfp (pzIs12) and cultured as previously described (Christensen et al., 2002; Strange et al., 2007). Briefly, egg-bearing adult worms were collected from age-synchronized plates and washed with M9 buffer to remove bacteria. Worms were lysed in egg isolation solution (2N NaOH, 20% bleach) for 5 min and washed several times with egg buffer (118 mM NaCl, 48 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 25 mM HEPES, pH 7.3). Eggs were harvested by sucrose float and digested under sterile conditions using chitinase (Sigma-Aldrich). Digested eggs were then resuspended in sterile L-15 medium (Invitrogen) containing 10% (vol/vol) heat-inactivated fetal bovine serum (FBS; Invitrogen), 50 U/ml penicillin (Invitrogen), and 50 μg/ml streptomycin (Invitrogen). Cells were then gently dissociated from the digested eggs by pipetting the cell suspension against the inside wall of a sterile microcentrifuge tube and were then filtered using a 5-μm Durapore syringe filter (Millipore, Bedford, MA). Cells were seeded onto 18-mm glass coverslips (Thermo Fisher Scientific, Waltham, MA) coated with peanut-lectin (Sigma-Aldrich) and grown in L-15 media with FBS, penicillin, and streptomycin. Cell cultures were maintained in humidified boxes at 20°C, and immunolabeling experiments were conducted 3–4 d postplating.

Immunolabeling

Primary cultured neurons were fixed and stained as previously described (Zheng et al., 2004). Briefly, neurons were fixed in 3% paraformaldehyde at room temperature for 30 min. For labeling of surface HA::GLR-1::GFP receptors, cells were blocked in 0.2% bovine serum albumin for 20 min and then incubated with 0.5 μg/ml anti-HA–Alexa Fluor 594 antibodies (16B12; Invitrogen) for 45 min. For labeling of all HA::GLR-1::GFP receptors, neurons were permeabilized in 0.2% Triton X-100 for 2 min before blocking. Cells were then washed in phosphate-buffered saline and mounted onto glass slides for imaging.

Abbreviations used:

AMPARs

AMPA-type GluRs

AP2

adaptor protein 2

CME

clathrin-mediated endocytosis

dn

dominant-negative

FBS

fetal bovine serum

GFP

green fluorescent protein

GGA

Golgi-localized, gamma adaptin ear–containing, ARF-binding

GluR

glutamate receptor

HA

hemagglutinin

LTD

long-term depression

RFP

red fluorescent protein

ROI

region of interest

SV

synaptic vesicle

TGN

trans-Golgi network

VNC

ventral nerve cord.