ArfGAP3 is a component of the photoreceptor synaptic ribbon complex and forms an NAD(H)-regulated, redox-sensitive complex with RIBEYE that is important for endocytosis

Abstract

Ribbon synapses are tonically active synapses in the retina and inner ear with intense vesicle traffic. How this traffic is organized and regulated is still unknown. Synaptic ribbons, large presynaptic structures associated with numerous synaptic vesicles, appear to be essential for this process. The base of the synaptic ribbon is anchored at the active zone and is a hotspot of exocytosis. The synaptic ribbon complex is also important for vesicle replenishment. RIBEYE is a unique and major component of synaptic ribbons. It consists of a unique A-domain and an NAD(H)-binding, C-terminal B-domain. In the present study, we show that the Arf-GTPase activating protein-3 (ArfGAP3), a well characterized regulator of vesicle formation at the Golgi apparatus, is also a component of the synaptic ribbon complex in photoreceptor synapses of the mouse retina and interacts with RIBEYE as shown by multiple, independent approaches. ArfGAP3 binds to RIBEYE(B)-domain in an NAD(H)-dependent manner. The interaction is redox sensitive because NADH is more efficient than the oxidized NAD(+) in promoting ArfGAP3-RIBEYE interaction. RIBEYE competes with the GTP-binding protein Arf1 for binding to ArfGAP3. Thus, binding of RIBEYE(B) to ArfGAP3 could prevent inactivation of Arf1 by ArfGAP3 and provides the synaptic ribbon with the possibility to control Arf1 function. The interaction is relevant for endocytic vesicle trafficking because overexpression of ArfGAP3 in photoreceptors strongly inhibited endocytotic uptake of FM1-43.

Keywords: ArfGAP3; RIBEYE; endocytosis; photoreceptor synapse; ribbon synapse; synaptic ribbon.

Figure 1.

RIBEYE(B) and RIBEYE(AB) interact with ArfGAP3 in the YTH system. Summary plates of YTH analyses obtained with the indicated bait and prey plasmids. For convenience, experimental bait–prey pairs are underlayered in color (green in case of interacting bait–prey pairs; control matings are noncolored). RIBEYE(B) and also full-length RIBEYE [RIBEYE(AB)] interact with ArfGAP3 in the YTH system (matings 1 and 2). Mating 11 denotes an unrelated positive control (Magupalli et al., 2008). pSE1111 is an irrelevant prey vector and pSE1112 is an irrelevant bait vector (Tai et al., 1999; Magupalli et al., 2008). Negative control matings of the ArfGAP3 prey clone with empty bait clones (mating 7) or irrelevant bait clones (mating 8) demonstrate that the ArfGAP3 clone is not auto activating. The other matings represent negative control matings for the RIBEYE bait clones (matings 3–6) or RIBEYE prey clones (matings 9 and 10), demonstrating that these constructs are also not auto activating in the YTH system. RE(AB), full-length RIBEYE, containing both RIBEYE(A)- and RIBEYE(B)-domain; RE(A), RIBEYE(A)-domain; RE(B), RIBEYE(B)-domain; AGD, ArfGAP-domain of ArfGAP3.

Figure 2.

Summary tables of YTH matings. A, RIBEYE interacts with the ArfGAP-domain (AGD) of ArfGAP3. RIBEYE(B) and also full-length RIBEYE [RIBEYE(AB)] interact with the ArfGAP-domain (AGD) of ArfGAP3 in the YTH system (matings 1 and 2). Mating 9 denotes an unrelated positive control (Magupalli et al., 2008). The other matings are auto-activation controls. None of the yeast constructs is auto activating. B, The NAD(H)-binding subdomain of RIBEYE(B) interacts with ArfGAP3. The NAD(H)-binding subdomain of RIBEYE(B), the NBD, interacts with ArfGAP3 (mating 1), but not the substrate-binding subdomain of RIBEYE(B), the SBD (mating 5). Matings 2–4 are positive control matings–(Bassoon for SBD-RE(B); tom Dieck et al., 2005), Munc119 for NBD-RE(B) (Alpadi et al., 2008), RE(A) for RE(B) (Magupalli et al., 2008)–and matings 6–11 are negative controls (auto-activation controls). C, Schematic domain structures of ArfGAP3. D, An NAD(H)-binding-deficient mutant of RIBEYE(B), RE(B)G730A, does not interact with ArfGAP3 (mating 1), while wild-type RIBEYE does (mating 6). Mating 7 is a positive control mating for RE(B)G730A. E, Arf1 interacts with ArfGAP3 (mating 1). All constructs are non-auto activating as demonstrated by the negative control matings (2–5); mating 6 is a positive control mating. RE(A), RIBEYE(A)-domain; RE(B), RIBEYE(B)-domain; RE(B)G730A, RIBEYE(B)G730A; GAP-dom, GAP-domain (AGD) of ArfGAP3; NBD, NAD(H)-binding subdomain of RIBEYE(B); SBD, substrate-binding subdomain of RIBEYE(B).

Figure 3.

RIBEYE(B) specifically interacts with ArfGAP3 in fusion protein pull-down assays (SDS-PAGE analyses). Pull-down analyses of RIBEYE(B)/ArfGAP3 complexes were analyzed by Coomassie blue-stained polyacrylamide gel after SDS-PAGE. Lanes 1–4 show the indicated purified fusion proteins (input fractions). All input lanes represent 30% of the input fraction. Input proteins were loaded in separate lanes to demonstrate that the input fusion proteins display only a single, main protein band. In lanes 5–8, 100% of the pull-down reactions were loaded. MBP-tagged fusion proteins were used as immobilized bait proteins and GST-tagged proteins as soluble prey proteins. Only ArfGAP3-MBP pulled down RE(B)-GST (lane 5, arrowhead) but not MBP alone (lane 7). Neither MBP alone nor ArfGAP3-MBP pulled down GST alone (lanes 6 and 8). SDS-PAGE demonstrated that ArfGAP3-MBP specifically pulled down RIBEYE(B)-GST, demonstrating interaction of the two proteins in this assay system.

Figure 4.

RIBEYE(B) specifically interacts with ArfGAP3 in fusion protein pull-down assays (Western blot analyses). To exclude that the tag has an importance for the pull-down results and to further exclude that any prey protein is unspecifically pulled down by bait-GST, we also analyzed the results of the pull-down assays by Western blotting with anti-MBP and anti-GST antibodies. The reaction buffer used for these experiments contained 1 mm βME. GST-tagged fusion proteins were used as immobilized bait proteins and eluted MBP-tagged proteins as soluble prey proteins. Similar to the experiments described in Figure 3, only RIBEYE(B)-MBP (lane 5) and not MBP alone (lane 6) is pulled down by the ArfGAP3(AGD)-GST. GST alone (lane 7) and MBP alone (lane 8) do not pull down RIBEYE(B)-MBP, as shown by Western blotting with antibodies against MBP (Fig. 4A), demonstrating the specificity of the interaction. The Western blot data fully confirm the results shown in Figure 3 that were obtained by SDS-PAGE analyses. In Figure 4B, the same blot as analyzed in Figure 4A was reprobed (after stripping of the blot) with antibodies against GST to show equal loading of the bait proteins. RE(B)-MBP, RIBEYE(B)-MBP; ArfGAP3(AGD)-MBP, MBP-tagged ArfGAP-domain (AGD) of ArfGAP3.

Figure 5.

RIBEYE(B) is recruited by ArfGAP3 into a Golgi-like distribution in transfected COS cells. COS7 cells were transfected with the indicated mcherry-tagged ArfGAP3 or EGFP-tagged RIBEYE(B) constructs. Transfected cells were analyzed for the intracellular distribution of the respective proteins via direct epifluorescence microscopy. Cells transfected with ArfGAP3-mcherry alone show the typical enrichment at the Golgi apparatus in a perinuclear localization (A, B), as already previously shown (Dogic et al., 1999; Eugster et al., 2000; Lewis et al., 2004; Watson et al., 2004; Frigerio et al., 2007; Kartberg et al., 2010; Yu et al., 2012). In contrast, RIBEYE(B) is diffusely distributed in single-transfected cells (C; Schmitz et al., 2000). If RIBEYE(B)-EGFP is cotransfected with ArfGAP3-mcherry, RIBEYE(B) virtually completely redistributed from a diffuse distribution into the Golgi-like, perinuclear localization indicating interaction between RIBEYE(B) and ArfGAP3 (D). n, nucleus. The arrow in D points to the Golgi-like localization to which the RIBEYE(B)-EGFP signal is translocated in ArfGAP3-mcherry-transfected cells. Scale bars: A–D, 10 μm.

Figure 6.

The binding between RIBEYE(B)-domain and the ArfGAP-domain (AGD) of ArfGAP3 is stimulated by NAD(H) in a redox-sensitive manner. A, B, Pull-down experiments were performed as in Figure 3 both in the absence (lanes 3—10; A, B) or presence (lane 11; A, B) of 1 mm βME. RIBEYE binds to the ArfGAP-domain (AGD) of ArfGAP3 only in the presence of 1 mm βME (lane 11; A, B) but not in the absence of 1 mm βME (lane 3; A, B). Lanes 1 and 2 show the respective input fractions (30% input). We tested whether addition of NADH (A) or NAD⁺ (B) could substitute for the presence of 1 mm βME in promoting ArfGAP3/RIBEYE(B) interaction. As a matter of fact, increasing concentrations of both NADH (A) as well as NAD⁺ (B) could promote binding of RIBEYE(B)-GST to the AGD of ArfGAP3 even in the absence of 1 mm βME (lanes 4—10; A, B). The reduced form of the dinucleotide, NADH (A), was more effective than the oxidized form, NAD⁺ (B), to promote RIBEYE(B)/ArfGAP3 interaction. Semiquantitative evaluation of the binding experiments (n = 4) demonstrated that ∼450 nm NADH and ∼700 nm NAD⁺ are promoting half-maximal binding of ArfGAP3 to RIBEYE(B) in the pull-down assays.

Figure 7.

Western blot analyses of two antibodies that were generated against the C terminus of ArfGAP3. A, Schematic drawing denotes the areas against which the two polyclonal ArfGAP3 antibodies (Cterm2 and Cterm3) were generated. B–D, In Western blot analyses, both antibodies (Cterm2 and Cterm3) detected a single band at the expected running position of ArfGAP3 at ∼55 kDa. This ArfGAP3 Western blot band could be blocked by pre-absorption of the antibody with the respective ArfGAP3-GST fusion protein (lane 2; C; data not shown) but not by pre-incubation with GST alone (C, lane 3; data not shown). Cb, Loading control (immunoblotting of the same blot as shown in Ca) after stripping of the blot and reprobing with an antibody against GCAP2 (Venkatesan et al., 2010), demonstrating equal protein loading.

Figure 8.

Coimmunoprecipitation of RIBEYE and ArfGAP3 from the bovine retina (Western blot analyses). A, ArfGAP3 immune serum (lane 3) and ArfGAP3 pre-immune serum (lane 2) were tested for their capability to coimmunoprecipitate RIBEYE from the bovine retina. Lane 1 shows the input fraction (1% of total input). RIBEYE is coimmunoprecipitated by ArfGAP3 immune serum (Cterm3 antiserum; lane 3) but not by ArfGAP3 pre-immune serum (lane 2). B, Shows the same blot as in A but reprobed with anti-ArfGAP3 antibodies (after stripping of the blot). This blot shows that ArfGAP3 was successfully immunoprecipitated by the immune serum (lane 3) but not by the control pre-immune serum (lane 2). HC and LC indicate the Ig heavy and light chains, respectively.

Figure 9.

ArfGAP3 is strongly enriched at synaptic ribbons of photoreceptor synapses in situ (conventional imaging). ArfGAP3 colocalizes with synaptic ribbons. The 0.5-μm-thick retinal sections are double immunolabeled with antibodies against ArfGAP3 and monoclonal antibodies against RIBEYE(B)/CtBP2 (A, B). ArfGAP3 Cterm3-antibody was used in A, and ArfGAP3 Cterm2 was used for immunolabeling of ArfGAP3 in B. Strong ArfGAP3 immunosignals were found in an identical manner with both ArfGAP3 antibodies at the RIBEYE-immunolabeled synaptic ribbons and in close vicinity to synaptic ribbons. The dashed circles denote single immunolabeled photoreceptor presynaptic terminals. Arrows in B and C point to single immunolabeled synaptic ribbons. No immunosignals were observed in the presynaptic terminals if pre-immune serum was used (data not shown). A and B were obtained by conventional imaging. OPL, Outer plexiform layer. Scale bars: A, 1 μm; B, 5 μm.

Figure 10.

A, B, Pre-absorption control experiments for the immunolabeling analyses shown in Figure 9A. Double immunolabeling of 0.5-μm-thick mouse retinal sections with ArfGAP3 (Cterm3) antibody pre-absorbed either with ArfGAP3-GST-fusion protein (B) or with GST alone (A). To visualize ribbon synapses, sections were coimmunolabeled with mouse monoclonal antibodies against RIBEYE(B)-domain/CtBP2. Pre-absorption with ArfGAP3Cterm3-GST fusion protein completely blocked the ArfGAP3 immunosignals at the synaptic ribbon (B), whereas GST alone had no influence on the ArfGAP3 immunosignals (A), demonstrating the specificity of the previous immunolabeling results. C, D, Pre-absorption control experiments for the immunolabeling analyses shown in Figure 9B. Double immunolabeling of 0.5-μm-thick mouse retinal sections with ArfGAP3 (Cterm2) antibody pre-absorbed with either ArfGAP3Cterm2-GST-fusion protein (D) or with GST alone (C). Synaptic ribbons were coimmunolabeled with mouse monoclonal antibodies against RIBEYE(B)-domain/CtBP2. Pre-absorption with the specific fusion protein completely blocked the ArfGAP3 immunosignals at the synaptic ribbon (D), whereas GST alone had no influence on the ArfGAP3 immunosignals (C), demonstrating the specificity of the previous immunolabeling results. A–D were obtained by conventional imaging. OPL, Outer plexiform layer; INL, inner nuclear layer. Scale bars: A–D, 5 μm.

Figure 11.

ArfGAP3 is strongly enriched at the synaptic ribbon of photoreceptor synapses in situ (SR-SIM imaging with ArfGAP3 Cterm3 antibody). A, 0.5-μm-thick retinal sections double immunolabeled with affinity-purified rabbit antibodies against ArfGAP3 (ArfGAP3 Cterm3 antibody) and mouse monoclonal antibodies against RIBEYE(B)/CtBP2. B, Shows 0.5-μm-thick retinal sections double immunolabeled with affinity-purified rabbit antibodies against ArfGAP3 (Cterm3 antibody) and mouse monoclonal antibodies against the active zone protein bassoon. Arrows in A and B point to single immunolabeled synaptic ribbons. Arrowheads denote ArfGAP3 immunoreactivity at the synaptic ribbon. A and B were obtained by SR-SIM imaging. ONL, Outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer. Scale bars: A, B, 1 μm.

Figure 12.

Localization of ArfGAP3 in the presynaptic photoreceptor terminal in relation to other presynaptic proteins. The 0.5-μm-thick retinal sections were triple immunolabeled with affinity-purified rabbit polyclonal antibodies against ArfGAP3 (Cterm3), mouse monoclonal antibodies against dynamin (hudy-1), and DyLight 650 directly labeled primary antibodies against RIBEYE(B)/CtBP2. ArfGAP3 and RIBEYE are located very close to each other (A, B). The ArfGAP3 immunosignal is located within the ring-like dynamin immunosignal that demarcates the presynaptic plasma membrane of the periactive zone that surrounds the synaptic ribbon (Wahl et al., 2013). A was obtained by conventional imaging; B is a micrograph obtained by SR-SIM imaging. Arrows in A and B point to immunolabeled single synaptic ribbons. Arrowheads indicate ArfGAP3 immunoreactivity at the synaptic ribbon. OPL, Outer plexiform layer. Scale bars: A, B, 1 μm.

Figure 13.

The ArfGAP3 effector Arf1 is enriched at the synaptic ribbon complex of photoreceptor synapses. A, B, Double immunolabeling experiments of 0.5-μm-thick retinal sections with mouse monoclonal antibodies against Arf1 and rabbit polyclonal antibodies against RIBEYE (U2656) demonstrated a close enrichment of Arf1 around the synaptic ribbon complex. ONL, Outer nuclear layer; OPL, outer plexiform layer. Scale bars: A, B, 1 μm. (C) Western blot analyses demonstrated that Arf1 is strongly expressed in the mouse retina.

Figure 14.

RIBEYE(B) and Arf1 compete with each other for binding to ArfGAP3. We tested in fusion protein pull-down experiments whether Arf1 and RIBEYE(B) can bind simultaneously to ArfGAP3 or whether they compete with each other for ArfGAP3-binding. A, B, Show representative Westerns blots incubated with the indicated antibodies to test for the binding of the respective fusion proteins. After detection of the GST-tagged protein (A1, B1), blots were stripped and re-incubated with antibodies against MBP (A2, B2). A3, B3, SNAP-tagged immobilized ArfGAP3 bait protein was visualized with SNAP-Vista Green (NEB). A, We tested whether increasing concentrations of Arf1 added to a fixed concentration of immobilized ArfGAP3 would inhibit binding of RIBEYE(B) to ArfGAP3. RIBEYE(B) was kept at a constant concentration in these experiments. B, We tested whether increasing concentrations of RIBEYE(B) added to a fixed concentration of immobilized ArfGAP3 would inhibit binding of Arf1 to ArfGAP3. Arf1 was kept at a constant concentration in these latter experiments. In both sets of experiments, we observed a competitive behavior between Arf1 and RIBEYE(B) in binding to ArfGAP3. These data demonstrate that RIBEYE(B) competes with Arf1 for binding to ArfGAP3. indicating that binding of RIBEYE(B) and Arf1 to ArfGAP3 is mutually exclusive. Abbreviations: AGD*, extended ArfGAP-domain of ArfGAP3.

Figure 15.

Overexpression of ArfGAP3 in mouse photoreceptors inhibits endocytic uptake of FM1–43. FM1–43 was used to compare endocytic uptake in photoreceptors that were either electroporated with mcherry alone (A, B) or ArfGAP3-mcherry (C, D). In mcherry-electroporated photoreceptors, there was an intense uptake of FM1–43 in the synaptic terminals (A, B). The uptake of FM1–43 in mcherry-transfected photoreceptors was similar to the FM1–43 uptake in nontransfected photoreceptors (data not shown). In contrast to mcherry-transfected photoreceptors, ArfGAP3-mcherry-overexpressing photoreceptors showed a strong inhibition of FM1–43 uptake in the synaptic terminals (C, D), indicating that ArfGAP3 is essential involved in endocytosis at the photoreceptor synapse. OS, Outer segment; IS, inner segment; Scale bars: A–C, 1 μm; D, 0.75 μm.