Rab3A Effector Domain Peptides Induce Insulin Exocytosis via a Specific Interaction with a Cytosolic Protein Doublet*

A key protein involved in the regulated exocytotic mechanism in neuroendocrine cells is the GTP-binding protein, RabSk Rab3A is thought to mediate exocytosis by an interaction of its effector domain with a putative effector protein. We demonstrate here that Rab3A effec- tor domain peptides specifically stimulated insulin exocytosis in electroporated insulin-secreting cells ac- tivation, 6-8 p ~ ) in a Ca2+-independent manner, although in the presence of Caw insulin exocytosis was further potentiated. By using a ‘%radiolabeled photoactivated cross-linking Rab3A effector domain peptide, we identi- fied a cytosolic protein doublet (REEP-1 and REEP-21, which specifically interacted with the Rab3A effector domain. Competitive inhibition studies revealed this protein-protein interaction to be at a concentration equivalent to that required for Rab3A effector domain peptides to trigger insulin exocytosis (&, 6-8 JIM). Fur-thermore, under basal secretory conditions REEP-1 and -2 were membrane-associated, but upon stimulation of exocytosis they were released into a cytosolic fraction. Our results suggest that REEP-1 and -2 are part of at 4 (Beckman SW 55 rotor). The supernatant was removed as a soluble "cytosolic" fraction from the "membrane" pellet, lyophilized, and resuspended in 50 pl of cross- link buffer described above. The pellet was directly resuspended in 50 pl of cross-link buffer. Proteins that associate with the Rab3A effector domain peptides were detected in these islet subcellular fractions by photoactivated cross-linking of an excess of 1261-Rab3AL-X (2 x IO8 cpm) as described above.

[Ca2+], are necessary to mediate glucose-stimulated insulin release (5). Among these additional factors, protein kinases (31, GTP, and low molecular weight GTP-binding proteins (4, 6-8) have been implicated, but the molecular mechanism as to how these factors promote insulin exocytosis is not known.
Guanine nucleotide-binding proteins of the Ras superfamily function as molecular switches for a wide variety of cellular functions (9). A sub-branch of the Ras family, the Rab class of GTP-binding proteins, is implicated in directing vesicular transport in eukaryotic cells (10). One member, Rab3A, has been specifically implicated in controlling mammalian regulated exocytosis. It has significant homology to the yeast Sec4 protein, which is required for vesicular transport from the trans-Golgi network to the plasma membrane (10). Rab3A is specifically expressed in neuroendocrine cells (11) and is a cytosolic protein of 25 kDa that is mostly located on the cytoplasmic face of secretory granules (12) or synaptic vesicles (13). Peptides that mimic the effector domain of Rab3A induce regulated exocytosis in cells (1&17), including pancreatic P-cells (8). This has led to the proposal that Rab3A mediates neuroendocrine exocytosis via an interaction of its effector domain with a putative effector protein (15,161, but such a "Rab3A effector protein" has not yet been identified. However, in this study we have used a synthetic photoactivatable cross-linking Rab3A effector domain peptide to identify a protein doublet that specifically associates with the Rab3A effector domain. This protein-protein interaction could play a role in the regulation of insulin exocytosis. EXPERIMENTAL PROCEDURES Materi~Zs-Na[~~~III was from Amersham Corp. t-ButoxycarbonyV benzyl-protected amino acids for peptide synthesis were from Applied Biosystems, except t-butoxycarbonyl benzoylphenylalanine (Bpa),' which was purchased from Bachem Inc. NycodenzTM was from Nycomed Pharma (Oslo, Norway). Unless otherwise indicated, all other chemicals were purchased from Sigma or Fisher and were of the highest grade/ purity available.

Rab3A Effector Domain's Interaction in Insulin Exocytosis
hexadecapeptide of the sequence DVYQEPTDPKE'PQQWY that had little similarity to the Rab3A effector domain, named "nonsense," was used as an additional control peptide. All synthetic peptides Were HPLC-purified as described previously (18), and amino acid analyses were as predicted.
Electroporation and Incubation of Semi-permeabilized HITTI5 Cells-A T-75 flask of confluent insulin producing HIT-TI5 cells was released from the flask with trypsin, washed in phosphate-buffered saline, and then washed twice in a buffer consisting of 270 m~ mannitol, present. HIT cells in +Ca2+ or -Ca2+ buffers were divided into 10 samples on ice. Each sample was washed and then resuspended in 250 p1 of the same +Ca2+ or -Ca2+ incubation buffer containing either no addition, or GTP# (50 p~) , or Rab effector domain peptide (1-100 p~) . The cells were then incubated on ice for 10 min, followed by 15 min at 37 "C. The incubation was stopped by placing the cells on icelwater. The cells were pelleted by centrifugation (1000 x g, 5 min, 4 "C), where the supernatant was retained for assessment of insulin release and the pellet for assessment of insulin content by radioimmunoassay (20). The semi-permeabilization of HIT cells by electroporation was assessed by significant 12'I-Rab3AL-X peptide and trypan blue uptake (4) compared with non-electroporated cells. In addition, negligible lactate dehydrogenase release was detected from electroporated HIT-T15 cells during a 15-min incubation at 37 "C, suggesting that there was minimal leakage of HIT cell cytosolic proteins.
Tissue-Insulinoma tissue (22) propagated in NEDH (New England Deaconess Hospital strain) rats was used as an abundant source of pancreatic pcell tissue. Rat insulinoma subcellular fractions, enriched in plasma membrane, insulin secretory granules, and cytosol, were prepared by differential and NycodenzTM density gradient centrifugation and characterized by marker enzyme analysis as described previously (23, 24).
Rat Pancreatic Islet Incubation and Fractionation-Rat pancreatic islets were isolated as described (25). Samples of 150 islets were preincubated at 37 "C for 30 min in 200 pl of Krebs-Ringer Hepes buffer pH 7.4 containing 0.5% (w/v) BSA and a basal 2.8 m M glucose as previously outlined (25). The islets were washed once in the same buffer and then incubated for a further 30 min at 37 "C in the same buffer containing either basal 2.8 m~ glucose or a stimulatory mixture of 16.7 nm glucose, 10 p~ forskolin, and 30 m~ KC1 that was intended to achieve a maximal rate of insulin exocytosis (1). Islets were pelleted by centrifugation (2 min at 800 x g), and the media were removed for radioimmunoassay of insulin release (20). Islets were resuspended in 1 ml of ice-cold 10 nm ammonium bicarbonate (pH 9.01, disrupted with 3-4 strokes of a Potter homogenizer at 400 rpm, and incubated on ice for 30 min. The osmotically lysed islet homogenate was centrifuged at 1500 x g for 5 min to remove unlysed cells, and the supernatant was removed and then centrifuged at 200,000 x g for 30 min at 4 "C (Beckman SW 55 rotor). The supernatant was removed as a soluble "cytosolic" fraction from the "membrane" pellet, lyophilized, and then resuspended in 50 pl of crosslink buffer described above. The membrane pellet was directly resuspended in 50 pl of cross-link buffer. Proteins that associate with the Rab3A effector domain peptides were detected in these islet subcellular fractions by photoactivated cross-linking of an excess of 1261-Rab3AL-X (2 x IO8 cpm) as described above.
Rab3A Detection-Analysis of Rab3A present in rat pancreatic islet fractions was by immunoprecipitation with specific Rab3A antisera (non-immune sera was used as a control), followed by GTPg6S binding to the immunoprecipitate as described previously (26).
Other Procedures-Protein was determined by using the BCA (bicinchoninic acid) method (Pierce) with BSAas standard. Free Ca" concentration was estimated from reference stability constants as described (27).
An analog of the Rab3A effector domain peptide (Rab3AL-X) was synthesized with the photoactivatable amino acid analog, Bpa, substituted for phenylalanine in the peptide at position -8. The Rab3AL-X peptide induced insulin exocytosis (Fig. lA 1, indicating that the Bpa for phenylalanine substitution did not adversely affect its biological activity. When exposed to ultraviolet radiation, Bpa-containing peptides cross-link covalently to the specific proteins with which they interact (21). The ' "1-Rab3AL-X peptide was used as a probe to identify specific p-cell proteins that interact with the Rab3A effector domain. In a rat insulinoma homogenate, two proteins were found to specifically cross-link to '2sI-Rab3AL-X that we have tentatively named REEP-1 and REEP-2 (for Bab Exocytotic Effector Proteins (Fig.  2 A , lane A). The electrophoretic mobility of REEP-1 and -2 suggested apparent molecular masses of 20 and 17 kDa. However, these proteins are covalently cross-linked to the 2.8-kDa "'1-Rab3AL-X so that representative molecular masses of REEP-1 and -2 were more likely 17 and 14 kDa, respectively. In the presence of 30 p~ control peptides, nonsense, Rab2, Rab4, and Rab5, no competitive inhibition of lZsI-Rab3AL-X crosslinking to REEP-1 and -2 was observed ( Fig. 2 A , lanes B, C, F, and G, respectively). In contrast, in the presence of 30 J~M Rab3A or Rab3AL peptides, cross-linking of '2sI-Rab3AL-X to REEP-1 and -2 was inhibited ( Fig. 2 A , lanes C and D). The Rab effector domain peptides are hydrophobic and may be prone to nonspecific interaction with proteins. However, REEP-1 and -2 cross-linking to ""I-Rab3AL-X was not competed by Rab2, Rab4, or Rab5 peptides, which have equivalent hydrophobicity to Rab3A and Rab3AL peptides. Thus, REEP-1 and -2 appear to specifically interact with the Rab3A effector domain. Other detectable proteins that cross-linked to the lZsI-Rab3AL-X were considered nonspecific, since cross-linking was not blocked by Rab3A or Rab3AL peptides. The protein of around 66-68 kDa that nonspecifically cross-links to "'1-Rab3AL-X was likely to be BSA, which was used as a "carrier" for trichloroacetic acid precipitation after the cross-linking reaction. Specific inhibition of lzSI-Rab3AL-X cross-linking to REEP-1 and -2 by Rab3AL peptide was dose-independent (Fig. 2B). The calculated mean 2 S.E. (n 2 4) K, for REEP-1 = 8.2 2 0.9 J~M and for REEP-2 = 6.1 2 0.8 VM (Fig. 2B). Thus, the specific interaction of REEP-1 and -2 with Rab3A effector domain peptides (Fig.   2B was at a concentration of these peptides equivalent to that which specifically evoked insulin exocytosis (Fig. 1, B and C).
Cross-linking of lZsI-Rab3AL-X to rat insulinoma subcellular fractions suggested REEP-1 and -2 to be cytosolic proteins in pancreatic p-cells (Fig. 3). However, in isolated rat islets the intracellular location of REEP-1 and -2 was dependent upon the regulatory state of p-cell exocytosis. At basal insulin exocytosis, about 70% of REEP-1 and -2 were membrane-associated and 30% were cytosolic (Fig. 4, A and B ). However, upon a maximal stimulation of insulin release (Fig. a), REEP-1 and -2 redistributed so that about 25% were membrane-associated and 75% were cytosolic (Fig. 4, A and B ) . An excess of "' 1-Rab3A.L-X was used in these experiments to ensure that all of REEP-1 and -2 present in the islet fractions could be accounted for. In addition, it can be noted that cytosolic REEP-1 and -2 run at an apparent lower molecular weight than "membraneassociated" REEP-1 and -2. However, this was not due to a modification of membrane-associated REEP-1 and -2 per se; rather it was due to composition of the samples that gave the cytosolic preparation a greater electrophoretic mobility. In parallel islet experiments, using specific Rab3A antisera as de- scribed previously (261, it was found that Rab3A protein remained in the membrane fraction during stimulation of insulin exocytosis (Fig. 4C) as previously indicated (26). This suggested that Rab3A remained attached to secretory granule membranes during exocytosis. However, it should be realized that Rab3A associates with most secretory granules in a cell (12) and that only a minor proportion of the insulin secretory granule population undergoes exocytosis from islet P-cells, even under stimulatory circumstances (<5% of insulin content released in 30 min (28)). Translocation of Rab3A upon exocytosis would be difficult to observe against a massive background of Rab3A associated with an insulin secretory granule storage pool. Thus, local redistribution of Rab3A at the site of exocytosis should not be ruled out (29).

Rab3A Effector Domain's Interaction in Insulin Exocytosis
Several proteins associate with Rab3A that are generally related to GTP/GDP cycling. GDP dissociation inhibitor binds to the C terminus of Rab3A (30), so REEP-1 and -2 are not GDP dissociation inhibitors as they associate with the Rab3A effector domain. Rab3A-specific GAP (GTPase-accelerating protein) has a high molecular weight (31), and point mutations to the effector domain of Rab3A do not generally affect GAP sensitivity (19). Thus, it is unlikely that REEP-1 and -2 are GAPS. In contrast to GAP, mutations to the effector domain of Rab3A affect GRF (guanine nucleotide-releasing factor) sensitivity (191, but Rab3A GRF has a high molecular mass (311, which implies that either REEP-1 and -2 are subunits of a Rab3A-GRF complex or not necessarily GRFs. However, in yeast a Sec4-specific GRF of 17 kDa (named Dss4-1) has been isolated (32) that has a mammalian equivalent (named Mss4) of 14 kDa (33). Although Mss4 was rather effective a t promoting GDP release from yeast Sec4 protein it was relatively poor a t releasing GDP from Rab3A. Because of their similar molecular weight, however, it may be possible that REEP-1 and -2 are members of a Mss4-related GRF family that is specific for Rab3A (33). Soluble cross-linking techniques have unveiled an 85-kDa protein, named rabphilin-3A, that associates with the GTP-bound form of Rab3A (34,35). The molecular weight difference between REEP-1 and -2 and rabphilin-3A suggests these proteins are not related. Furthermore, because the Rab3AL-X peptide did not specifically detect an 85-kDa protein, it suggests that rabphilin-3A interaction with Rab3A is not via Rab3A's effector domain. Although the identity of REEP-1 and -2 has yet to be established, for the moment they are likely novel candidates involved in regulated exocytotic mechanism, if not novel proteins themselves.
Our results also raise the question as to the role of REEP-1 and -2 in regulated exocytosis. Experimental evidence suggests that a pre-exocytotic protein complex is formed (36, 37) among secretory granule (or synaptic vesicle), cytosolic membrane, and plasma membrane proteins (of which Rab3A is a likely member (10,37,38)). Recently, Rab3A has been implicated in recruiting synaptic vesicles for docking with the plasma membrane, in a step prior to triggering exocytosis (391, and may therefore play a role in the formation of pre-exocytotic protein complexes. It has been proposed that dissociation of this preexocytotic complex is key to promoting fusion between the secretory granule and plasma membranes for regulated exocytosis (38,40). Rab3A has been demonstrated as an inhibitor of regulated hormone release, and it is alleviation of the Rab3A inhibition that may, in part, be key to triggering exocytosis (41, 42). It is possible that REEP-1 and -2 are inhibitory proteins that are released from the Rab3A effector domain upon stimulation of exocytosis. The membrane association of REEP-1 and -2 with the Rab3A's effector domain in the basal state may be consistent with these proteins also being members of a preexocytotic protein complex. Redistribution of REEP-1 and -2 to a cytosolic fraction upon stimulation of exocytosis could be symptomatic of a dissociation of the pre-exocytotic protein complex (38,40,41). It follows that the introduction of Rab3A effector domain peptides into semi-permeabilized cells would shift the equilibrium of REEP-1 and -2 associating with endogenous Rab3A effector domain (presumably on the secretory granule surface in a pre-exocytotic fusion complex) toward REEP-1 and -2 associating with Rab3A effector domain peptides in the cytosol. Hence, REEP-1 and -2 come out of the pre-exocytotic protein complex promoting its dissociation that, in turn, initiates exocytosis in a Ca2+-independent fashion. I t follows that further detailed characterization of specific protein-protein interactions within the pre-exocytotic complex could identify the key regulatory events that trigger exocytosis in vivo.