ELKS, a protein structurally related to the active zone protein CAST, is involved in Ca2+-dependent exocytosis from PC12 cells

Authors


  • Communicated by: Shuh Narumiya

* Correspondence: E-mail: tohtsuka@med.u-toyama.ac.jp

Abstract

The active zone protein CAST binds directly to the other active zone proteins RIM, Bassoon and Piccolo, and it has been suggested that these protein–protein interactions play an important role in neurotransmitter release. To further elucidate the molecular mechanism, we attempted to examine the function of CAST using PC12 cells as a model system. Although PC12 cells do not express CAST, they do express ELKS, a protein structurally related to CAST. Endogenous and exogenously expressed ELKS, RIM2 and Bassoon were colocalized in punctate signals in PC12 cells. Over-expression of full-length ELKS resulted in a significant increase in stimulated exocytosis of human growth hormone (hGH) from PC12 cells, similar to the effect of full-length RIM2. This increase was not observed following over-expression of deletion constructs of ELKS that lacked either the last three amino acids (IWA) required for binding to RIM2 or a central region necessary for binding to Bassoon. Moreover, over-expression of the NH2-terminal RIM2-binding domain of Munc13-1, which is known to inhibit the binding between RIM and Munc13-1, inhibited the stimulated increase in hGH secretion by full-length RIM2. Furthermore, this construct also inhibited the stimulated increase in hGH secretion induced by full-length ELKS. These results suggest that ELKS is involved in Ca2+-dependent exocytosis from PC12 cells at least partly via the RIM2-Munc13-1 pathway.

Introduction

Neurotransmitter release occurs at a specialized membrane domain in nerve terminals, designated the active zone (Landis et al. 1988). The meshwork or cytomatrix at the active zone (CAZ) extends from the presynaptic plasma membrane into the synaptic bouton where it is associated with synaptic vesicles. The CAZ has been believed to play an organizational role in defining the active zone and a functional role in neurotransmitter release (Südhof 1995; Garner et al. 2000; Dresbach et al. 2001).

Although the full molecular composition of the CAZ is currently unknown, several CAZ-specific proteins have been identified, including RIMs (Wang et al. 1997, 2000), Munc13-1 (Brose et al. 1995), Bassoon (tom Dieck et al. 1998), Piccolo/Aczonin (Cases-Langhoff et al. 1996; Wang et al. 1999; Fenster et al. 2000), CAST (also known as ERC2), and ELKS (also known as CAST2/ERC1) (Ohtsuka et al. 2002; Wang et al. 2002; Deguchi-Tawarada et al. 2004). CAST and ELKS are structurally related CAZ proteins that consist of several coiled-coil domains and a unique COOH-terminal amino acid (aa) sequence, IWA, which is required for binding to the PDZ domains of RIMs (Ohtsuka et al. 2002; Wang et al. 2002; Deguchi-Tawarada et al. 2004). CAST has been shown to form a large molecular complex in the active zone by directly binding to RIMs, Bassoon, Piccolo and ELKS and by indirectly binding to Munc13-1 (Ohtsuka et al. 2002; Deguchi-Tawarada et al. 2004; Takao-Rikitsu et al. 2004). Indeed, Bassoon and Piccolo directly bind to the second coiled-coil region of CAST. Disruption of the interaction between CAST and Bassoon or CAST and RIM1 significantly reduces neurotransmission in cultured neurons. Recently, experiments using a freeze-fracture replica labeling method have revealed that the concentration of CAST is more than 100-fold higher in the active zone than in the extrasynaptic area surrounding the active zone (Hagiwara et al. 2005). Therefore, we speculate that the CAST/ELKS-dependent large molecular complex may be the molecular basis for the slight electron density observed in the presynaptic active zone and play a role in neurotransmitter release.

ELKS was originally identified as a gene with its 5′ terminus fused to the RET tyrosine kinase oncogene in a papillary thyroid carcinoma (Nakata et al. 1999). Recently, ELKS has been demonstrated to function in insulin exocytosis from pancreatic beta cells, since introduction of the Bassoon-binding region of ELKS or suppression of ELKS expression by a siRNA reduced the fusion of insulin-containing vesicles (Ohara-Imaizumi et al. 2005). In addition, ELKS has been identified as Rab6-interacting protein2 and implicated in membrane trafficking in several tissues (Monier et al. 2002). Finally, as described above, we previously identified ELKS as a novel CAZ protein in neurons (Deguchi-Tawarada et al. 2004). It is also noteworthy that ELKS has several ubiquitously expressed splicing isoforms outside of the brain, which lack the COOH-terminal IWA sequence (Monier et al. 2002; Nakata et al. 2002; Wang et al. 2002; Deguchi-Tawarada et al. 2004).

In the present study, we used PC12 cells as a model system to further elucidate the function of the ELKS/CAST family in exocytosis. In PC12 cells, ELKS, but not CAST, was expressed and co-localized with the other CAZ proteins Bassoon and RIM2. Furthermore, using various constructs of ELKS, we found that ELKS is involved in Ca2+-dependent exocytosis from PC12 cells at least partly through the RIM2-Munc13-1 pathway.

Results

Expression and localization of CAST and ELKS in PC12 cells

We summarize the structure of the ELKS/CAST family (Fig. 1A) (Monier et al. 2002; Nakata et al. 2002; Wang et al. 2002; Deguchi-Tawarada et al. 2004). We then examined whether PC12 cells expressed CAST and/or ELKS. An anti-CAST antibody (Ab) did not reveal the presence of a band of ∼120 kDa (CAST), whereas an anti-ELKS Ab detected a protein band of ∼120 kDa in PC 12 cell lysates (Fig. 1B). The protein band of ELKS in PC12 cells showed a similar mobility on SDS-PAGE to that of Myc-ELKS expressed in HEK293 cells, which differed from that of Myc-ELKSɛ (a ubiquitous isoform of ELKS) (Nakata et al. 2002; Deguchi-Tawarada et al. 2004). These results indicate that ELKS, but not CAST, is expressed in PC12 cells at the protein level. To confirm this, we examined the localization of ELKS in PC12 cells. ELKS showed punctate signals, which were co-localized with those of Bassoon at the basal membrane (Fig. 1C; upper panel). Furthermore, Bassoon was co-localized with another CAZ protein RIM2 (Fig. 1C; middle panel). It should be noted that RIM2, but not RIM1, has been shown to be expressed in PC12 cells (unpublished observations; Ozaki et al. 2000). In addition, ELKS was partially co-localized with the synaptic vesicle protein synaptotagmin (Fig. 1C; lower panel). These results indicate that ELKS is associated with the exocytotic machinery together with the other CAZ proteins Bassoon and RIM2.

Figure 1.

Expression and localization of ELKS in PC12 cells. (A) Comparison of the structures of CAST and ELKS. (B) Endogenous expression of the CAST family in PC12 cells. HEK293 cells, which were transfected with Myc-CAST, Myc-ELKS or Myc-ELKSɛ, and intact PC12 cells were extracted and aliquots (10 µL) of the extracts were subjected to SDS-PAGE, followed by Western blotting using anti-CAST and anti-ELKS Abs. (C) Localization of ELKS in PC12 cells. PC12 cells were fixed and double-stained with anti-ELKS or anti-RIM2 Abs and anti-Bassoon or anti-synaptotagmin Abs. Bar = 10 µm. The results are representative of three independent experiments.

Involvement of ELKS in Ca2+-dependent exocytosis

To examine the involvement of ELKS in Ca2+-dependent exocytosis, we performed hGH assays using PC12 cells. In these assays, a plasmid encoding hGH or the protein to be examined is co-transfected into PC12 cells and the cells are then stimulated by a high K+ concentration in the presence of Ca2+. The high K+ concentration causes membrane depolarization, followed by opening of the voltage-gated Ca2+ channels and entry of Ca2+, which results in vesicle fusion and hGH release (Sugita et al. 1999). As control experiments, we confirmed the high K+-dependent effects of Rab3A Q81L (50% decrease) and RIM2 (50% increase) (see Fig. 4) on the stimulated hGH release: it is well established that both proteins regulate Ca2+-dependent exocytosis from PC12 cells (Ozaki et al. 2000; Schluter et al. 2002). Under the same conditions, over-expressed Myc-ELKS (full-length) significantly enhanced the stimulated hGH release (Fig. 2A). Next, we examined the localization of exogenously expressed ELKS in comparison with the localizations of endogenous RIM2, Bassoon and synaptotagmin. To achieve this, we used enhanced green fluorescent protein (EGFP)-tagged ELKS, since it was somewhat difficult to detect the Myc signals under our experimental conditions. Similar to Myc-ELKS, EGFP-ELKS also enhanced the stimulated hGH release under the same conditions (our unpublished observations; see Fig. 8). Exogenously expressed ELKS showed punctate signals, which were consistent with those of endogenous ELKS (Fig. 1C) and co-localized with those of endogenous RIM2, Bassoon and synaptotagmin at the basal membrane (Fig. 2B).

Figure 4.

Involvement of the RIM2–ELKS interaction in Ca2+-dependent exocytosis from PC12 cells. (A) Importance of the interaction between RIM2 and ELKS for Ca2+-dependent exocytosis. (a) Binding assay of RIM2ΔPDZ. Expression plasmids for EGFP-RIM2 or EGFP-RIM2ΔPDZ and Myc-ELKS were co-transfected into HEK293 cells, followed by immunoprecipitation using an anti-Myc Ab. The immunoprecipitates were then analyzed by Western blotting using the indicated Abs. (b) Effects of RIM2 and the RIM2ΔPDZ mutant on the high K+-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. *P < 0.01. (B) Localization of EGFP-RIM2 in PC12 cells. PC12 cells were transfected with EGFP-RIM2 or EGFP-RIM2ΔPDZ, and then stained with the anti-RIM2 or anti-Bassoon Ab. (a) Localization of EGFP-RIM2. (b) Localization of EGFP-RIM2ΔPDZ. Bar = 10 µm. The results are representative of three independent experiments.

Figure 2.

Effect of ELKS on Ca2+-dependent exocytosis from PC12 cells. (A) Effects of ELKS on low K+ (open bars)- or high K+ (closed bars)-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. *P < 0.01. (B) Localization of EGFP-ELKS in PC12 cells. PC12 cells were transfected with EGFP-ELKS, and then stained with an anti-RIM2, anti-Bassoon or anti-synaptotagmin Ab. Bar = 10 µm. The results are representative of three independent experiments.

Figure 8.

Localization mechanism of ELKS in PC12 cells. (A) Alignment of the sequences of the conserved region in the ELKS/CAST family. (B) Effect of ELKSΔ161-200 on Ca2+-dependent exocytosis from PC12 cells. (a) EGFP construct of ELKSΔ161-200. (b) Effects of ELKS and ELKSΔ161-200 on the high K+-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. (C) Localization of EGFP-ELKS and EGFP-ELKSΔ161–200 in PC12 cells. PC12 cells were transfected with EGFP-ELKS or EGFP-ELKSΔ161–200, and then stained with the anti-RIM2 Ab. Bar = 10 µm. The results are representative of three independent experiments.

Previously, we have shown that the unique COOH-terminal IWA sequence of CAST and ELKS is required for binding to RIM1 and that disruption of the CAST and RIM1 interaction significantly impairs neurotransmission in neuron (Ohtsuka et al. 2002; Deguchi-Tawarada et al. 2004; Takao-Rikitsu et al. 2004). These observations led us to hypothesize that the IWA sequence of ELKS plays an important role in exocytosis from PC12 cells. To clarify this assumption, we prepared a deletion mutant lacking the IWA sequence (Myc-ELKSΔ3 amino acids). Consistently, Myc-ELKSΔ3 amino acids did not bind to EGFP-RIM2 (Fig. 3Aa). Next, we over-expressed the full-length and mutant ELKS constructs in PC12 cells. Under the conditions in which Myc-ELKS enhanced the stimulated hGH release, Myc-ELKSΔ3 amino acids had no effect (Fig. 3Ab). Exogenously expressed ELKSΔ3 amino acids showed punctate signals that were co-localized with those of endogenous RIM2 and Bassoon (Fig. 3B). These results suggest that ELKS is involved in Ca2+-dependent exocytosis from PC12 cells, and this function probably requires its interaction with RIM2.

Figure 3.

Involvement of the ELKS-RIM2 interaction in Ca2+-dependent exocytosis from PC12 cells. (A) Importance of the interaction between ELKS and RIM2 for Ca2+-dependent exocytosis. (a) Binding assay of ELKSΔ3 amino acids. Expression plasmids for Myc-ELKS or Myc-ELKSΔ3 amino acids and EGFP-RIM2 were co-transfected into HEK293 cells, followed by immunoprecipitation using the anti-GFP Ab. The immunoprecipitates were then analyzed by Western blotting using the indicated Abs. (b) Effects of ELKS and the ELKSΔ3 amino acids mutant on the high K+-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. *P < 0.01. (B) Localization of EGFP-ELKSΔ3 amino acids in PC12 cells. PC12 cells were transfected with EGFP-ELKSΔ3 amino acids, and then stained with the anti-RIM2 or anti-Bassoon Ab. Bar = 10 µm. The results are representative of three independent experiments.

Involvement of the interaction between ELKS and RIM2 in Ca2+-dependent exocytosis

To confirm the above hypothesis, we prepared a deletion mutant of RIM2 that lacked the PDZ domain (EGFP-RIM2ΔPDZ). This domain is required for its binding to CAST and ELKS (Ohtsuka et al. 2002; Wang et al. 2002; Deguchi-Tawarada et al. 2004). Consistent with previous observations, EGFP-RIM2ΔPDZ did not bind to Myc-ELKS under conditions in which EGFP-RIM2 bound to Myc-ELKS, as assessed by immunoprecipitation assays (Fig. 4Aa). Next, we performed hGH assays using these constructs. Similar to RIM1 (Wang et al. 1997), EGFP-RIM2 significantly enhanced the stimulated hGH release, while EGFP-RIM2ΔPDZ also enhanced the stimulated hGH release, but to a lesser degree (∼50% of the effect of EGFP-RIM2; Fig. 4Ab). These results suggest that binding to ELKS may play an important role in RIM2 function during Ca2+-dependent exocytosis and that in the absence of binding to ELKS, RIM2 does not appear to exert its full effect on Ca2+-dependent exocytosis from PC12 cells.

Next, we analyzed the localization of exogenously expressed RIM2. Immunofluorescence microscopic analyses revealed that, similar to endogenous RIM2 (see Fig. 1C), EGFP-RIM2 showed punctate signals that were well co-localized with endogenous ELKS and Bassoon (Fig. 4Ba). On the other hand, EGFP-RIM2ΔPDZ was diffusely distributed throughout the cells under the conditions in which endogenous ELKS and CAST showed punctate signals (Fig. 4Bb). These results suggest that ELKS may play crucial roles in the localization of RIM2 in PC12 cell.

Complex formation of ELKS, RIM2, and Munc13-1 in Ca2+-dependent exocytosis

CAST has been reported to form a ternary complex with RIM1 and Munc13-1 (Ohtsuka et al. 2002). Specifically, CAST binds directly to RIM1 and in turn RIM1 binds directly to Munc13-1 (see Fig. 9). These findings allowed us to speculate that a similar complex formation may occur in Ca2+-dependent exocytosis from PC12 cells. We separately transfected the individual expression plasmids for full-length ELKS, RIM2 and Munc13-1 into HEK293 cells, extracted the proteins and mixed them in various combinations, followed by immunoprecipitation using an anti-Myc Ab for Munc13-1. When Munc13-1 was immunoprecipitated, ELKS was co-immunoprecipitated with Munc13-1 only in the presence of RIM2 (Fig. 5Aa), indicating that ELKS forms a ternary complex with RIM2 and Munc13-1. The direct interaction between RIM1 and Munc13-1 was previously reported to be involved in the regulation of synaptic vesicle priming (Betz et al. 2001). Furthermore, the RIM-binding domain (RIMBD; aa1–317) of Munc13-1 was shown to function in a dominant-negative manner in the priming step (Betz et al. 2001). Thus, we used the RIMBD of Munc13-1 to examine the relationship of Munc13-1 with ELKS and RIM2 in Ca2+-dependent exocytosis from PC12 cells. Similar to the case for RIM1 and Munc13-1, the RIMBD of Munc13-1 was also able to bind directly to RIM2 (Fig. 5Ab). Next, we performed hGH assays in the presence of RIMBD. EGFP-RIM2 significantly enhanced the stimulated hGH release, and this effect was almost completely inhibited in the presence of Myc-Munc13-1 RIMBD (Fig. 5Ba). Similarly, HA-ELKS also significantly enhanced the stimulated hGH release, an effect that was almost completely inhibited by the presence of Myc-Munc13-1 RIMBD (Fig. 5Bb). These results suggest that ELKS is involved in Ca2+-dependent exocytosis from PC12 cells at least partly through the RIM2-Munc13-1 pathway.

Figure 9.

Schematic diagram of the structure and interactions of CAZ proteins in neuron and PC12 cell. In neurons, CAST forms a large molecular complex at the active zone by directly binding to RIM1, Bassoon, Piccolo, and ELKS, and by indirectly binding to Munc13-1. CAST and ELKS form a homo- and/or hetero-oligomerization with each other (Deguchi-Tawarada et al. 2004). In PC12 cells, ELKS, but not CAST, is expressed and binds to Bassoon, Piccolo, and RIM2 at basal membrane. ELKS appears to form a homooligomer there. The conserved NH2-terminal region (aa161–200) of ELKS is required for localization to the basal membrane in PC12 cells. However, this localization mechanism is currently unknown.

Figure 5.

Involvement of the ELKS-RIM2-Munc13-1 interaction in Ca2+-dependent exocytosis from PC12 cells. (A) Biochemical analysis of the ELKS-RIM2-Munc13-1 complex. (a) Ternary complex formation of ELKS, RIM2, and Munc13-1. The indicated expression vectors were transfected into HEK293 cells. The expressed proteins were extracted and mixed in the indicated combinations, followed by immunoprecipitation using the anti-Myc Ab. The immunoprecipitates were then analyzed by Western blotting using the indicated Abs. (b) Binding assay of the Munc13-1 RIMBD. Expression plasmids for EGFP-RIM2 and/or Myc-Munc13-1RIMBD were transfected into HEK293 cells, followed by immunoprecipitation using the anti-Myc Ab. Arrowhead indicates the band corresponding to heavy chain of the anti-Myc Ab. (B) Effects of the Munc13-1 RIMBD on the RIM2- or ELKS-mediated enhancement of hGH secretion from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. (a) Effect on the RIM2-mediated enhancement. (b) Effect on the ELKS-mediated enhancement. *P < 0.01.

Complex formation of ELKS and Bassoon in Ca2+-dependent exocytosis

Previously, it has been reported that the interaction between Bassoon and CAST plays an important role in neurotransmitter release (Takao-Rikitsu et al. 2004). Thus, we examined the function of the Bassoon-binding domain (BsnBD) of ELKS in Ca2+-dependent exocytosis from PC12 cells. To achieve this, we prepared a deletion mutant of ELKS, which lacked the BsnBD (Myc-ELKSΔBsnBD). Immunoprecipitation assays revealed that Myc-ELKS bound to Bassoon, whereas Myc-ELKSΔBsnBD did not (Fig. 6Aa). However, Myc-ELKS and Myc-ELKSΔBsnBD both bound to RIM2 (Fig. 6Ab), since the binding sites for Bassoon and RIM2 are different (Ohtsuka et al. 2002; Deguchi-Tawarada et al. 2004; Takao-Rikitsu et al. 2004; Ohara-Imaizumi et al. 2005). When these constructs were over-expressed, Myc-ELKSΔBsnBD had no effect on the stimulated hGH release under the conditions in which Myc-ELKS significantly enhanced the stimulated hGH release (Fig. 6Ac). In addition, exogenously expressed ELKSΔBsnBD showed punctate signals, which were co-localized with those of endogenous RIM2 and Bassoon (Fig. 6B). These results suggest that the BsnBD of ELKS is involved in Ca2+-dependent exocytosis from PC12 cells.

Figure 6.

Involvement of the ELKS-Bassoon interaction in Ca2+-dependent exocytosis from PC12 cells. (A) Importance of the interaction between ELKS and Bassoon for Ca2+-dependent exocytosis. (a) Binding assay of ELKSΔBsnBD and Bassoon. Expression plasmids for Myc-ELKS or Myc-ELKSΔBsnBD and EGFP-Bassoon were co-transfected into HEK293 cells, followed by immunoprecipitation using the anti-GFP Ab. (b) Binding assay of ELKSΔBsnBD and RIM2. Expression plasmids for EGFP-RIM2 and/or Myc-ELKSΔBsnBD were transfected into HEK293 cells, followed by immunoprecipitation using the anti-Myc Ab. The immunoprecipitates were then analyzed by Western blotting using the indicated Abs. (c) Effects of ELKS and ELKSΔBsnBD on the high K+-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. Data are expressed as the percentage of hGH secretion relative to the total cellular content. *P < 0.01. (B) Localization of EGFP-ELKSΔBsnBD in PC12 cells. PC12 cells were transfected with EGFP-ELKSΔBsnBD, and then stained with the anti-RIM2 or anti-Bassoon Ab. Bar = 10 µm. The results are representative of three independent experiments.

Functional domains of ELKS in Ca2+-dependent exocytosis

In a final set of experiments, we systematically investigated which regions of ELKS are involved in Ca2+-dependent exocytosis. To this end, we prepared various constructs of ELKS (Fig. 7Aa), and then transfected each expression plasmid into PC12 cells, followed by hGH assays. Among the constructs, Myc-ELKS-1 (full-length) enhanced the stimulated hGH release, whereas the other constructs either had no effect or slightly inhibited the stimulated hGH release (Fig. 7Ab). These results suggest that, in addition to the BsnBD and the COOH-terminal IWA sequence, the NH2-terminal region of ELKS (aa1–120) is also required for the full effect of ELKS on Ca2+-dependent exocytosis from PC12 cells. Next, we examined the localization of the exogenously expressed constructs in PC12 cells. Under the conditions in which ELKS-1 (full-length) was co-localized with endogenous RIM2 (data not shown), ELKS-2 and ELKS-8 were also co-localized with endogenous RIM2 at the basal membrane (Fig. 7B). The other constructs, including ELKS-3 and ELKS-6 (Fig. 7B) and ELKS-4, ELKS-5 and ELKS-7 (data not shown), were diffusely distributed throughout the cells. These results suggest that the NH2-teminal region of ELKS containing the first coiled-coil region may be essential for its clustering with RIM2 and Bassoon.

Figure 7.

Domains of ELKS required for its localization and function in PC12 cells. (A) Domains of ELKS required for Ca2+-dependent exocytosis. (a) Various EGFP-tagged constructs of ELKS. (b) Effects of the various ELKS constructs on the high K+-induced secretion of hGH from PC12 cells. Secretion of hGH was determined by the amount released into the medium relative to the total cellular content. High K+-induced secretion of hGH in empty vector-transfected PC12 cells was used as a control (100%). Data are expressed as the percentage of hGH secretion relative to the control. *P < 0.01. (B) Localization of the various EGFP-ELKS constructs in PC12 cells. PC12 cells were transfected with the various EGFP-ELKS constructs, and then stained with the anti-RIM2 Ab. Bar = 10 µm. The results are representative of three independent experiments.

We compared the NH2-terminal regions of CAST and ELKS from different species including rat, Caenorhabditis elegans, and Drosophila. The alignment revealed that the region from aa161–200 within the coiled-coil domain was well conserved during evolution (Fig. 8A). To examine the importance of this region, we prepared an EGFP-tagged deletion construct of ELKS lacking aa161–200 (EGFP-ELKSΔ161–200; Fig. 8Ba). Under the same conditions in which EGFP-ELKS significantly enhanced the stimulated hGH release, EGFP-ELKSΔ161–200 also enhanced the release, although the degree of enhancement was slightly lower than that of EGFP-ELKS (Fig. 8Bb). When we transfected each plasmid into PC12 cells, ELKS showed punctate signals that were well co-localized with those of endogenous RIM2, whereas ELKSΔ161–200 was diffusely distributed throughout the cell (Fig. 8C). These results suggest that the NH2-terminal conserved region of ELKS is required not only for its function in Ca2+-dependent exocytosis but also for its localization at sites where endogenous RIM2 is localized.

Discussion

In this paper, we have attempted to reveal a possible role for ELKS, a protein structurally related to CAST, in exocytosis from PC12 cells, mainly by using the well-established hGH assay. Although we did not examine directly the effect of the suppression of endogenous ELKS, we would like to postulate that endogenous ELKS is involved in Ca2+-dependent exocytosis from PC12 cells according to the following several lines of evidence:

  • 1In PC12 cells, ELKS, but not CAST, is expressed and clustered with RIM2 and Bassoon at the basal membrane.
  • 2Over-expression of the COOH-terminal region of ELKS significantly inhibits the stimulated hGH release (see Fig. 7ab).
  • 3This effect is not observed using the region which lacks IWA sequence. IWA is required for ELKS to bind to RIM2 (see Fig. 7ab).
  • 4This COOH region is almost identical to that of CAST.

It has been shown that over-expression of the COOH region of CAST inhibits neurotransmitter release in cultured hippocampal neurons, by inhibiting its interaction with RIM1 (Takao-Rikitsu et al. 2004). Therefore, it is likely that the COOH-terminal region functions in a dominant negative manner and then inhibits the function of endogenous ELKS, resulting in the decrease in the stimulated hGH release by RIM2.

To date, ELKS has been shown to be involved in various cellular functions. Specifically, ELKS has been identified as a gene fused to RET tyrosine kinase in thyroid carcinomas (Nakata et al. 1999), it may serve as a novel regulatory factor in the NF-κB signaling pathway (Ducut Sigala et al. 2004), it has been identified as Rab6 small G protein-interacting protein2 and implicated in Golgi transport (Monier et al. 2002) and, similarly to CAST, it has been shown to be a novel component of the presynaptic active zone in conventional and ribbon synapses (Deguchi-Tawarada et al. 2004, 2006). This accumulating evidence suggests that, in addition to its important function in the brain, ELKS may be involved in various kinds of cellular signaling pathways and vesicular trafficking outside of the brain. Moreover, it is noteworthy that ELKS, but not CAST, is expressed in pancreatic beta cells, where it regulates the exocytosis of insulin (Ohara-Imaizumi et al. 2005). The BsnBD of CAST was reported to be involved in neurotransmitter release in cultured neurons (Takao-Rikitsu et al. 2004). Consistently, the BsnBD of ELKS was also found to play an important role in insulin exocytosis from pancreatic beta cells (Ohara-Imaizumi et al. 2005). Thus, together with these observations, our present data suggest that ELKS may be involved in general exocytosis in both the brain and other tissues.

The CAZ proteins RIM1 and Munc13-1 interact directly with each other and this interaction plays a crucial role in a late step of neurotransmitter release, namely the vesicle priming step (Betz et al. 2001). Genetic knockout studies support this notion (Augustin et al. 1999; Castillo et al. 2002; Schoch et al. 2002). The finding of complex formation by CAST, RIM1 and Munc13-1 in the brain suggests that CAST may regulate the same signaling pathway during neurotransmitter release, which is regulated by RIM1 and Munc13-1. Indeed, disruption of the binding between CAST and RIM1 significantly impaired neurotransmission in primary cultured neurons (Takao-Rikitsu et al. 2004). Similar to the case in the brain, our current results suggest that ELKS regulates exocytosis at least in part through binding to RIM2. It is also interesting that the ELKS mutant lacking the IWA sequence had no effect on the stimulated increase in hGH release, whereas the RIM2 mutant, which lacks the PDZ domain, slightly enhanced the stimulated increase in hGH release. These observations can be explained by the possibility that the diffusely distributed RIM2ΔPDZ still binds to Munc13-1. Consistently, expression of the zinc finger domain of RIM1, which shows a diffuse distribution, was able to enhance exocytosis from chromaffin cells (Sun et al. 2001). Moreover, the RIMBD of Munc13-1 inhibited the effect of ELKS on the stimulated increase in hGH release. This could provide another line of evidence for our notion that the CAST-RIM1-Munc13-1 complex is involved in neurotransmitter release (Ohtsuka et al. 2002; Takao-Rikitsu et al. 2004). However, a previous analysis of ELKS knockout in C. elegans appears to be against this idea, since deletion of the elks gene in C. elegans did not affect the localization of RIM or show a structural or functional phenotype (Deken et al. 2005), which is surprising since C. elegans ELKS binds directly to the PDZ domain of RIM, similar to mammalian CAST and ELKS. Currently, we cannot provide a suitable explanation for this discrepancy, although it may simply be due to a difference between species.

Our primary goal is to understand the physiological functions of CAST and ELKS in the formation, maintenance, and disruption of the presynaptic active zone in the central nervous system. To achieve this, we will have to elucidate the molecular mechanisms underlying the localization of CAST and ELKS at the active zone. Although CAST is not expressed in PC12 cells, ELKS is expressed and co-localized with the other CAZ proteins Bassoon and RIM2. Therefore, hGH assays provide us with a tool not only for analyzing the functional domains of ELKS for hGH release but also for examining the localization mechanism of ELKS at clusters containing CAZ proteins, which in turn should provide us with more insights into the localization mechanisms of CAST and ELKS in the brain (Fig. 9). In the present study, we have shown that an ELKS mutant lacking the NH2-terminal region (aa161–200) is diffusely distributed in PC12 cells under the conditions in which wild-type ELKS is co-localized with RIM2. In a previous study, we showed that the NH2-terminal half region of CAST is required for its synaptic localization in neurons (Ohtsuka et al. 2002). Thus, the data in the present paper are consistent with this previous observation. One possible scenario is that this evolutionarily conserved region may bind directly to a transmembrane protein at the basal membrane of PC12 cells. However, this region is located within the first coiled-coil region, which is required for homo- and/or hetero-oligomerization of CAST and ELKS (Deguchi-Tawarada et al. 2004). Thus, we cannot exclude the possibility that the phenotype of this deletion mutant may simply reflect a lack of oligomerization of ELKS. To address this issue, identification of the binding partners for this region is currently in progress.

Experimental procedures

cDNA cloning

ELKSɛ cDNAs were cloned from mouse brain total RNA by RT-PCR using an RNA PCR kit (TaKaRa) according to the manufacturer's instructions.

Antibodies

The rabbit polyclonal anti-CAST2/ELKS Ab was obtained as previously described (Deguchi-Tawarada et al. 2004, 2006). The mouse monoclonal anti-Myc (9E10; Roche), mouse monoclonal anti-GFP (Roche), rat monoclonal anti-HA (Roche), mouse monoclonal anti-Bassoon (StressGen), mouse monoclonal anti-synaptotagmin (Chemicon), and rabbit polyclonal anti-RIM2 (Synaptic Systems) Abs were purchased from commercial sources.

Expression vectors

Expression vectors were constructed in pCIneo-Myc (Ohtsuka et al. 1998) and pEGFPC1 (Clontech Laboratories) using standard molecular biological methods. The ELKS constructs used contained the following amino acid residues: ELKSΔ3 amino acids (aa1–945), ELKSΔBsnBD (aa1–444, 605–stop), ELKS-1 (aa1–stop), ELKS-2 (aa121–stop), ELKS-3 (aa201–stop), ELKS-4 (aa401–stop), ELKS-5 (aa605–stop), ELKS-6 (aa765–stop), ELKS-7 (aa765–945), and ELKS-8 (aa1–444). The expression vector containing the mouse RIM2 cDNA (pCMV-HA-RIM2) was kindly supplied by Dr S. Seino (Chiba University, Chiba, Japan). RIM2ΔPDZ contained aa1–685, 757–stop. Munc13-1RBD contained aa1–317.

PC12 cell culture, transfection, and immunocytochemistry

PC12 cells were maintained in DME medium containing 10% fetal bovine serum and 5% horse serum. For Western blot analysis of ELKS, proteins were extracted from PC12 cells (2 × 107 cells) with 1 mL of Triton X-100 lysis buffer (20 mm Tris-HCl, pH 7.5, 150 mm NaCl, 0.5 mm EDTA, 1 mm DTT, and 1% (w/v) Triton X-100). The sample was then centrifuged at 10 000 g at 4 °C for 20 min, and an aliquot (10 µL) of the supernatant was analyzed by Western blotting. For immunocytochemistry, PC12 cells were transfected with the indicated expression vectors using the Lipofectamine plus reagent (Invitrogen) according to the manufacturer's instructions. The transfected cells were fixed with 2% paraformaldehyde and 4% sucrose in PBS for 20 min at room temperature, and then permeabilized with 0.1% (w/v) Triton X-100 in PBS for 15 min. Nonspecific binding was blocked with 4% Block Ace for 2 h, and the cells were then incubated with a primary Ab diluted in 0.4% Block Ace in PBS for 2 h. After five washes in PBS for 5 min each, the cells were further incubated with an appropriate secondary Ab, and then rinsed again with PBS. Coverslips were mounted on the slides using a ProLong Antifade Kit. Fluorescence images were acquired by confocal laser microscopy using a 63× or 100× oil immersion objective lens (LSM510; Carl Zeiss Microimaging Inc.).

Measurement of hGH release from PC12 cells

PC12 cells were plated at a density of 5 × 105 cells/35 mm dish and incubated for 18–24 h. Next, the cells were transfected using the Lipofectamine plus reagent (Invitrogen) as described above. hGH release experiments were performed at 48 h after the transfection. The transfected cells were washed twice in a physiological salt solution (PSS; 140 mm NaCl, 4.7 mm KCl, 2.5 mm CaCl2, 1.2 mm MgSO4, 1.2 mm KH2PO4, 20 mm HEPES, pH 7.4, 11 mm glucose) and then incubated for 20 min in a low K+ solution (PSS containing 4.7 mm KCl and 140 mm NaCl) or a high K+ solution (PSS containing 60 mm KCl and 85 mm NaCl). The supernatants were collected and the cells were harvested. The amounts of hGH secreted into the medium and retained in the cells were measured using an hGH ELISA kit (Roche).

Immunoprecipitation from HEK293 cells

Immunoprecipitation of HEK293 cell lysates was performed as previously described (Ohtsuka et al. 2002). Briefly, each expression plasmid was transfected into HEK293 cells, and the expressed proteins were extracted with the Triton X-100 lysis buffer and mixed in various combinations. After incubation for 2 h, 1.0 µg of the indicated Ab was added to the sample, followed by incubation at 4 °C for 1 h. Next, 20 µL of Protein A-Sepharose beads was added and the sample was further incubated for 1 h. The beads were extensively washed with the Triton X-100 lysis buffer and the bound proteins were eluted by boiling in SDS sample buffer. The eluted samples were analyzed by Western blotting.

Acknowledgements

We thank Dr S. Seino (Chiba University, Japan) for providing the RIM2 cDNA.

Ancillary