Identification and Characterization of Rain, a Novel Ras-interacting Protein with a Unique Subcellular Localization*

The Ras small GTPase functions as a signaling node and is activated by extracellular stimuli. Upon activation, Ras interacts with a spectrum of functionally diverse downstream effectors and stimulates multiple cytoplasmic signaling cascades that regulate cellular proliferation, differentiation, and apoptosis. In addition to the association of Ras with the plasma membrane, recent studies have established an association of Ras with Golgi membranes. Whereas the effectors of signal transduction by activated, plasma membrane-localized Ras are well characterized, very little is known about the effectors used by Golgi-localized Ras. In this study, we report the identification of a novel Ras-interacting protein, Rain, that may serve as an effector for endomembrane-associated Ras. Rain does not share significant sequence similarity with any known mammalian proteins, but contains a Ras-associating domain that is found in RalGDS, AF-6, and other characterized Ras effectors. Rain interacts with Ras in a GTP-dependent manner in vitro and in vivo, requires an intact Ras core effector-binding domain for this interaction, and thus fits the definition of a Ras effector. Unlike other Ras effectors, however, Rain is localized to perinuclear, juxta-Golgi vesicles in intact cells and is recruited to the Golgi by activated Ras. Finally, we found that Rain cooperates with activated Raf and causes synergistic transformation of NIH3T3 cells. Taken together, these observations support a role for Rain as a novel protein that can serve as an effector of endomembrane-localized Ras.

Ras is a plasma membrane-localized protein that cycles between GDP-bound (inactive) and GTP-bound (active) states. After activation by external stimuli, Ras-GTP binds with multiple downstream effectors through the core Ras effector-binding domain (residues [32][33][34][35][36][37][38][39][40], which results in the stimulation of signaling cascades that regulate cytoplasmic and nuclear events. A growing number of Ras effectors have been identified over the years, with the Raf serine/threonine kinases, the phosphatidylinositol 3-kinase (PI3K) 1 lipid kinases, and the Ral guanine nucleotide exchange factors among the best characterized (3)(4)(5)(6). Ras activation of effector function is mediated, in part, by the recruitment of these cytoplasmic proteins to the plasma membrane.
Recent studies involving live-cell imaging, electron microscopy, and fluorescence resonance energy transfer have shown that in addition to the plasma membrane, H-Ras and N-Ras, but not K-Ras, localize to intracellular membranes of the endoplasmic reticulum and Golgi (7). Philips and colleagues (8) have shown that Golgi-localized Ras is activated through a Src/phospholipase C␥1/Ras guanyl nucleotide-releasing protein pathway that is distinct from the classic protein receptor tyrosine kinase/Grb2/SOS pathway that activates Ras at the plasma membrane. Furthermore, they demonstrated that endomembrane-associated Ras is biologically active because silencing of endogenous Ras guanyl nucleotide-releasing protein-mediated activation of Ras completely blocks T-cell receptor-stimulated Ras activation in Jurkat cells and neuronal differentiation in PC-12 cells. In NIH3T3 transformation assays, a constitutively active, palmitoylation-deficient H-Ras mutant that is localized and activated on endomembranes exhibits transforming activity comparable with that of wildtype activated H-Ras (9). Whether the endomembrane and plasma membrane pools of biologically active Ras use the same set of effector molecules remains unknown.
Genetically engineered variants of Raf, PI3K, and Ral guanine nucleotide exchange factors that are targeted to the plasma membrane in the absence of Ras activation are constitutively active as signaling molecules (10 -13), establishing these proteins as bona fide effectors of plasma membranelocalized Ras. The identification of the effectors of endomembrane-associated Ras is the next challenge in understanding the role and importance of Ras signaling originating from endomembranes. Herein, we report the identification of Rain, a novel Ras-interacting protein that displays the characteristics of an effector of endomembrane-localized Ras. First, Rain pos-sesses a Ras-associating (RA) domain homologous to the RA domains of other Ras effectors, including the Ral guanine nucleotide exchange factors, AF-6, Rin1, and phospholipase C⑀ (14). Second, like Raf and all other known Ras effectors, Rain preferentially binds to the GTP-loaded form of Ras in vitro and in vivo, and the interaction of Rain with Ras is abolished by mutations in the core effector domain of Ras. Third, we show that Rain localizes to a perinuclear, juxta-Golgi region in intact cells and is recruited to the Golgi by active Ras. Finally, we determined that co-expression of Rain together with activated Raf causes synergistic transformation of NIH3T3 cells. These data support a role for Rain as a novel effector for the activated pool of Ras localized to endomembranes in mammalian cells.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Assays-Individual Gal4 DNA binding domain (DB)-Ras fusion constructs (DB-Ras G12V,C186S; DB-Ras G12V,T35S,C186S; DB-Ras G12V,E37G,C186S; and DB-Ras G12V,Y40C,C186S) were generated by standard procedures. Full-length cDNA sequences encoding effector domain mutants of activated H-Ras G12V were amplified by PCR using primers to introduce 5Ј-BamHI and 3Ј-EcoRI sites and to alter the coding sequence to contain a C186S substitution of the cysteine residue of the H-Ras CAAX prenylation signaling motif. The resulting PCR products were cloned into the yeast expression vector pAS2-1 (BD Biosciences Clontech). Two hybrid screens were performed in Saccharomyces cerevisiae strain Y190 using RasG12V,C186S-Gal4 DB as the bait and a human skeletal muscle cDNA library cloned into pACT2 (BD Biosciences Clontech). Transformants were plated on Leu Ϫ Trp Ϫ His Ϫ selection, and large colonies were assayed for ␤-galactosidase activity as described previously (15). The pACT2 plasmid was recovered from each ␤-galactosidase-positive colony and propagated in Escherichia coli strain DH5. All positive clones were sequenced and characterized further as described in the text.
A Gal4 DB-Rain (amino acids 79 -447) yeast expression construct in pAS2-1 was generated using available NcoI sites. Interactions between Rain and Ras family members were tested by co-transformation of S. cerevisiae strain L40 (gift from A. Vojtek) with pACT2 Rain 79 -447 and individual small GTPases expressed in yeast as LexA DB fusion proteins (gift from A. Vojtek and M. Hansen) (16). Transformants were selected and liquid cultures assayed for ␤-galactosidase activity as described above.
Isolation of the Full-length Rain cDNA-Rain 79 -962 (clone C15) was excised from the pACT2 plasmid by EcoRI/XhoI digestion and cloned into pBS-KS. Expressed sequence tag clone AI928221 was obtained from the American Type Culture Collection and used to isolate the cDNA sequence encoding the N terminus of Rain. A 600-bp fragment (encoding amino acids 1-180 of Rain) was isolated by EcoRI/SacI digestion and subcloned into the pBS-KS plasmid (designated Rain 1-180/pBS). pBS Rain 79 -962 was digested with SacI, and the isolated 2.2-kb fragment was inserted into the SacI site of pBS Rain 1-180. A plasmid with a confirmed orientation of the SacI fragment then was digested with EcoRI/HincII to excise a 1.1-kb fragment encoding the Rain amino terminus, and this fragment was subcloned into pBS Rain 79 -962 digested with EcoRI/HincII to generate a full-length Rain cDNA. This full-length cDNA was subcloned into the pcDNA, pcDNA3 HA, and pBabe-puro mammalian expression vectors. An expression vector encoding Rain tagged at the amino acid terminus with green fluorescent protein Rain (GFP-Rain) was generated by PCR amplification of the enhanced GFP sequence and insertion into pcDNA3 Rain. The same approach was used to generate an expression vector encoding cyan fluorescent protein-tagged Rain (CFP-Rain). The human Rain cDNA sequence reported here has been entered into the GenBank data base under accession number AY378097.
Cell Culture and Transient Transfection Assays-NIH3T3 mouse fibroblasts were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. HEK293 cells were maintained in high glucose Dulbecco's modified Eagle's medium, 10% fetal bovine serum, 1 mM sodium pyrophosphate, and antibiotics. COS-1 cells were cultured in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and antibiotics.
NIH3T3 Transformation Assays-We employed a cooperation focus formation assay to assess the contribution of Rain to transformation through Ras signaling pathways. NIH3T3 cells were plated at 1 ϫ 10 5 cells per 60-mm dish and transfected the next day using Lipo-fectAMINE Plus (Invitrogen) and procedures recommended by the manufacturer. DNA constructs used for transfection were pBabe-raf-Raf22W (50 ng/plate) (17), pcDNA3HA RhoA63L (1 g/plate) (18), and pcDNA3 Rain (1 g/plate). Raf22W is an amino-terminally truncated, constitutively activated variant of human Raf-1. RhoA(63L) is a GTPase-deficient, constitutively activated mutant of human RhoA. Cells were fed growth medium 3 h after transfection and every 48 h thereafter. The appearance of transformed foci was monitored for 16 days and quantified visually using phase contrast microscopy. Data are expressed as number of foci per plate, averaged from three plates per group from three independent transfection experiments. For documentation, the cultures were fixed and stained with 0.4% crystal violet.
Rain Expression Analyses-To evaluate Rain mRNA expression in the mouse, we used total RNA isolated from adult mouse tissues using TRIzol reagent (Invitrogen) and reverse transcription-PCR. The primers for the mouse Rain cDNA were as follows: 5Ј-CTACTCTGGGTGT-GTTCCAGGC-3Ј (forward), 5Ј-TGCGTCATCTGTCACAGGGC-3Ј (reverse) and were designed to amplify a 550-bp fragment at the 3Ј end of the Rain open reading frame. Two micrograms of total RNA from each tissue were reverse-transcribed using random hexamers and Moloney murine leukemia virus reverse transcriptase for 1 h, and 1 l of each reaction was amplified by PCR with Taq polymerase for 35 cycles. For human Rain mRNA expression analysis, a multiple-tissue human mRNA blot (BD Biosciences Clontech) was probed with a randomprimed, 32 P-labeled 2.1-kb cDNA fragment of Rain (nt 1167-3230). Blots were hybridized at 65°C in rapid-hybridization buffer (Amersham Biosciences) for 2 h, washed, and exposed to x-ray film. To evaluate Rain protein expression, mouse tissues isolated from an adult FVB mouse were lysed by homogenizing in radioimmunoprecipitation assay buffer containing protease inhibitors. Extracts were cleared by centrifugation and protein concentration determined by Bradford assay (Bio-Rad, Hercules, CA). Equal amounts of extract were separated by SDS-polyacrylamide gel, transferred to nitrocellulose membrane, and immunoblotted using the polyclonal anti-Rain antiserum described above.
Interaction between Rain and Small GTPases-Bacterially expressed glutathione S-transferase (GST) fusion proteins containing the isolated Ras-binding domain of c-Raf-1 (amino acids 51-131; Raf-RBD) and a region of Rain containing the RA domain (amino acids 121-245; Rain-RA) were expressed and purified from Rosetta E. coli strain (Novagen, San Diego, CA). Bacterially expressed and purified H-Ras protein (Pan-Vera, Madison, WI) was pre-loaded with 2 mM ␤,␥-imido-GTP or GDP by incubating for 10 min at 37°C in 40 l of loading buffer (20 mM Tris, pH 7.5, 10 mM EDTA, 5 mM MgCl 2 and 1 mM dithiothreitol). Reactions were cooled on ice, the MgCl 2 concentration adjusted to 15 mM, and the samples diluted to 300 l with binding buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM MgCl 2 , and 0.1% Triton-X100) supplemented with 0.2% bovine serum albumin and 25 M concentration of the corresponding nucleotide before incubation with 5 g of GST-RA-Rain or GST-Raf-RBD proteins for 2 h at 4°C. Protein complexes bound to glutathione beads were washed three times with binding buffer, eluted in SDS sample buffer, and resolved by SDS-PAGE. The amount of H-Ras protein bound to RA-Rain or Raf-RBD was detected by immunoblotting using a pan-Ras antibody (Ab-3; Oncogene, San Diego, CA).
To determine the ability of Rain to bind to other small GTPases, HEK293 cells were transiently transfected with 2 g of plasmid DNA expressing Flag epitope-tagged versions of activated H-Ras, Rap1A, R-Ras, or M-Ras/R-Ras3 (a gift of L. Quilliam). Forty-eight hours after transfection, cells were lysed in buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, and 1% Triton X-100) containing protease inhibitors and 250 g of each cell lysate were incubated for 1.5 h at 4°C with 5 g of GST, GST-Raf-RBD, or GST-Rain-RA proteins immobilized on glutathione beads. Protein complexes were washed three times, eluted with SDS sample buffer, resolved by SDS-PAGE, transferred to nitrocellulose membrane, and immunoblotted using an anti-Flag antibody (M2; Sigma).
Co-immunoprecipitation Analyses-NIH3T3 cells stably expressing H-Ras61L alone, or together with HA-Rain, were grown to subconflu-ence and lysed in buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 0.1 mM MgCl 2 , 10% glycerol, and 0.5% Nonidet P-40) containing protease inhibitors. Five hundred micrograms of total protein extract in a volume of 1 ml were incubated overnight at 4°C with 50 l of anti-HA antibody coupled to agarose beads (clone 3F10; Roche Applied Science). Protein complexes were washed three times with lysis buffer, eluted from the beads with SDS sample buffer, and resolved by 12.5% SDS-PAGE. Bound proteins were detected by immunoblot analysis using anti-HA and anti-Ras antibodies.
Immunofluorescence Analyses-COS-1 cells were plated at 3 ϫ 10 5 cells per 60-mm plate and transfected with 2 g of HA-Ras61L, 3 g of GFP-Rain, or both constructs using Superfect transfection reagent (QIAGEN). Forty-eight hours after transfection, cells were fixed with 4% paraformaldehyde and stained using an anti-HA antibody (HA.11; Covance Inc., Princeton, NJ) as described previously (19). Staining was visualized using an Olympus fluorescence microscope with a 60ϫ objective. To confirm Rain localization at the Golgi, 1 ϫ 10 5 COS-1 cells were plated onto 35-mm MatTek dishes containing number 1.5 glass slides on the bottom (MatTek Corp., Ashland, MA) and transfected as described above with 1 g of CFP-Rain and 0.3 g of YFP-GalT (a gift from M. Philips) in the presence or absence of 0.5 g of H-Ras61L-Flag. Twenty-four hours after transfection, cells were fixed in 4% paraformaldehyde or viewed live using a Zeiss 510 LSM confocal microscope.

Identification of a Novel Ras-interacting Protein-To identify
additional Ras effectors, a yeast two-hybrid screen was performed using a Gal4-DB-H-Ras G12V fusion protein as the bait and an adult human skeletal muscle cDNA library as the source of interacting proteins. From 1.8 ϫ 10 7 independent transformants, 46 clones were selected that displayed interaction dependent growth on His Ϫ agar and high levels of ␤-galactosidase activity. The specificity of the interaction with Ras G12V was confirmed by retesting each clone with a variety of control baits (data not shown). Sequence analysis of the clones revealed that 41 represented partial cDNA sequences for the known Ras effectors, Rin1, RalGDS, and RalGDS-like 2 protein. The remaining five clones represented cDNAs encoding the same protein. All of these cDNAs were 2.9 kb long and contained an open reading frame of 883 amino acids, a 284-bp 3Ј untranslated region, and a poly(A) tail. A search of the GenBank nucleotide data base with the sequence of a representative clone (C15) revealed no significant homology with any known genes but identified several expressed sequence tag clones identical to regions of the C15 cDNA. One expressed sequence tag (GenBank accession number AI928221) contained information that was used to extend the C15 cDNA by 0.3 kb in the 5Ј direction and to incorporate an in-frame initiator ATG codon (Fig. 1A). The completed cDNA (3.2 kb) is identical to a recently released IMAGE clone (GenBank accession number BC028614) and contains an open reading frame for a 962amino acid, highly basic (pI ϭ 8.7) polypeptide with a predicted molecular mass of 104 kDa. Because the protein encoded by this cDNA binds to Ras, we named this protein Rain, for Rasinteracting protein.
Rain contains a single RA (RalGDS/AF-6) domain (14) in the N terminus (Fig. 1A). This domain (amino acids 144 -247) has the highest homology to the RA domains of human AF-6 ( Fig.  2), a protein first identified as a 3Ј portion of a translocation product in some human leukemias (20) and subsequently as a putative Ras/Rap effector (21). An amino acid sequence alignment of the RA domain of Rain with the RA domains of other known Ras-interacting proteins is shown in Fig. 1B. Rain possesses the conserved basic residue (Arg182) found to be critical for the interaction of other effectors with Ras [Arg-89 in c-Raf-1 and Lys-685 in RGL (22,23)]. In addition to the RA domain, Rain contains a proline-rich region (amino acids 59 -136) that is similar to the Src homology 3 domain binding motifs found in many intracellular proteins. Rain also contains a putative dilute (DIL) domain (amino acids 769 -877) found in the globular tail of the myosin V proteins (24). In myosins, the DIL domain may be involved in binding to cargo during vesicle trafficking (25,26). It is interesting that, outside of the myosin V family, AF-6 is the only protein found to contain a DIL domain. However, the role of this domain in AF-6 function is unknown.
When the amino acid sequence of Rain was used to search the NCBI protein data base, we identified a protein sequence that is likely to represent a rat homolog of Rain (accession number XP_214916). In addition, this search identified a second human and mouse protein pair (accession number NP_060529 and NP_84887, respectively) that has the same domain architecture as Rain. These proteins, along with the AF-6 and its orthologs, are likely to comprise a new family of proteins with a unique combination of protein domains (Fig. 2). No Rain orthologs have been identified in invertebrates.
Our search of the human high throughput genomic sequence data base identified two genomic contigs containing the Rain gene (GenBank TM accession numbers AC009002 and AC008888). The contigs and the hypothetical protein they encode (FLJ20401) map to chromosome 19q13.33. Alignment of the Rain cDNA with its genomic sequence revealed that the Rain gene consists of 12 exons and 11 introns, 10 of which interrupt the protein-coding region (Fig. 1C). Expression analysis of the human RAIN gene revealed a single transcript (ϳ3.5 kb) that closely matched the size of the cloned Rain cDNA in all human tissues examined, with highest levels found in heart (Fig. 3A). Reverse transcription-PCR analysis of RNA isolated from adult mouse tissues confirmed that the Rain gene is transcriptionally active in the majority of samples examined, with the highest level of Rain mRNA detected in mouse lung (Fig. 3B).
To evaluate the expression of the Rain protein, we used a carboxyl-terminal peptide from human Rain to generate a rabbit polyclonal antiserum. Immunoblot analysis of protein extracts prepared from mouse tissues with the anti-Rain antiserum revealed two peptides of 115 and 85 kDa in lung and a single 85-kDa peptide in spleen (Fig. 3C). To test the specificity of the Rain antiserum, immunoblots were performed with protein extracts isolated from mouse lung or from NIH3T3 cells expressing HA-tagged Rain. The anti-HA antibody detected the 115-kDa HA-Rain protein expressed in the transfected cells (Fig. 3D). When the same extracts were immunoblotted using anti-Rain antibodies, 115-and 85-kDa peptides were detected in both samples. Pre-incubation of the anti-Rain antibodies with the Rain peptide used to generate the antiserum completely abolished detection of the 115-kDa protein and had no effect on the detection of the 85-kDa peptide. No proteins of 115 kDa were observed when pre-immune rabbit serum was used to probe the same samples (data not shown). Thus, together with the predicted size of Rain, these analyses indicate that the 115-kDa band corresponds to endogenous mouse Rain protein.
Rain Binds to Ras in a GTP-dependent Manner-Ras effectors preferentially bind to the activated (GTP-bound) form of Ras through a region referred to as the core effector domain (Ras residues 32-40). Yeast two-hybrid assays were used to evaluate Rain interaction with different Ras effector domain mutants. These results show that an intact Ras effector domain is necessary for interaction with Rain. Although Rain showed a strong interaction with E37G H-Ras effector domain variant, no interaction was detected using the T35S or Y40C variants (Fig. 4A).
We next determined whether the Rain-Ras interaction is influenced by the activation state of Ras. Recombinant Ras protein was loaded with GDP or ␤,␥-imido-GTP (a non-hydrolyzable GTP analog) and incubated with a GST fusion protein containing the isolated RA domain of Rain (amino acids 121-245). The RBD of Raf-1 (Raf-RBD) was used as a positive control. The amount of Ras protein bound was determined by immunoblotting using an anti-Ras antibody. As shown in Fig.   4B, Rain exhibited a higher affinity for Ras-GTP than Ras-GDP. As expected, a similar binding preference was observed for the Raf-RBD. When we evaluated the ability of Ras to bind to Rain peptides lacking the RA domain, no interactions were detected (data not shown), indicating that the RA domain is necessary and sufficient for interaction with Ras. Taken to-gether with the yeast two-hybrid studies, these results show that Rain meets the criteria of an effector for Ras.
We next tested whether the Rain-Ras interaction occurs in

FIG. 2. Schematic diagram of Rain and proteins with a similar domain organization.
Rain family members were identified by searching both protein and nucleotide databases. RA, Ras-associating, RalGDS/AF-6-like domain (14); DIL, myosin V-like motif (24); PDZ, PSD-95/Dlg/ZO-1 domain (also called DHR or GLGF) (48). Hypothetical proteins are identified by accession numbers. Mouse Rain sequence was compiled from two nucleotide sequences, BC028483 and NM_027253, which were translated and compared with the human Rain sequence. The percentage similarity between each Rain domain and the respective domains in other family members is indicated. Percentage similarity was calculated based on the alignment generated using the ClustalW algorithm. The XP_221937 sequence is incomplete at the amino terminus. h, human; m, mouse; r, rat; d, D. melanogaster; c, Caenorhabditis elegans. mammalian cells. For these studies, NIH3T3 cells were stably transfected with an expression vector for activated H-Ras (Ras61L) and either a control plasmid or a plasmid expressing full-length, HA-tagged Rain. HA-Rain was immunoprecipitated from whole cell lysates using an anti-HA antibody, and the protein complexes were resolved by SDS-PAGE and immunoblotted using an anti-Ras antibody. Results revealed that the Ras protein is detected only in precipitates from HA-Rainexpressing cells (Fig. 4C). These results demonstrate that Rain-Ras interaction occurs in mammalian cells and supports a role for Rain as a physiologically relevant effector of Ras.
Rain Binds to Other Ras Family Members-Because many of the Ras effectors characterized to date (e.g. Raf, RalGDS, AF-6, Nore1) interact with other Ras-related proteins in addition to Ras, we used yeast two-hybrid assays to test the ability of the Rain RA domain to bind to other members of the Ras superfamily of GTPases. In agreement with our initial studies with the Gal4 DB-Ras G12V bait, Rain interacted with the LexA DB-H-Ras G12V fusion protein (Fig. 4D). Rain also interacted strongly with Rap2 and with activated R-Ras, Rap1A and TC21/R-Ras2 proteins. No interaction was detected between Rain and a constitutively active form of RhoA.
To extend these observations, we examined the ability of the Rain RA domain to bind to various Ras family proteins expressed in mammalian cells. Constitutively active forms of H-Ras, Rap1A, R-Ras, and M-Ras/R-Ras3 were expressed transiently in HEK293 cells and cell lysates were incubated with GST-Rain-RA or with GST-Raf-RBD and GST as controls. In agreement with previous observations, GST-Raf-RBD interacted strongly with the activated H-Ras, Rap1A, and R-Ras proteins and less avidly with M-Ras/R-Ras3 (27)(28)(29)(30). When GST-Rain-RA was tested, interaction was detected with activated H-Ras, Rap1A, and, to a lesser degree, R-Ras. GST-Rain-RA did not interact with M-Ras/R-Ras3 (Fig. 4E), and no interaction was detected between the control GST protein and any of the Ras-related proteins. These results demonstrate that the RA domain of Rain interacts strongly with both Ras and Rap proteins and suggests that Rain may not be an exclusive effector for Ras-mediated signaling in cells.
Rain Co-localizes with Activated Ras-The majority of Ras effectors characterized to date are cytosolic and are dynamically recruited to plasma membrane upon Ras activation. To determine the subcellular location of Rain, we transiently transfected COS-1 cells with an expression vector encoding GFP-Rain alone or together with a vector expressing HA-Ras61L. When expressed alone, GFP-Rain showed no plasma FIG. 4. Rain interacts with Ras-GTP in vitro and in vivo. A, single point mutations in the Ras effector binding domain block Ras-Rain interaction. Protein interaction between the RA domain of Rain fused to the Gal4 transcription activation domain (TAD) and H-Ras G12V or the indicated H-Ras G12V effector domain mutants fused to the Gal4 DNA binding domain (DB) was evaluated by a yeast two-hybrid assay. The strength of interaction was determined by liquid ␤-galactosidase assays. The average activity was calculated for each protein pair using a minimum of three independent yeast cotransformants. Bars denote S.D. B, Rain RA domain binds preferentially to Ras-GTP. H-Ras protein was loaded with ␤,␥imido-GTP or with GDP and incubated with a GST-Rain-RA or GST-Raf-RBD fusion proteins immobilized on glutathione beads. The presence of bound Ras was detected using a pan-Ras antibody. C, Ras-GTP and Rain associate in vivo. Lysates from NIH3T3 cells stably expressing H-Ras61L or H-Ras61L and full-length HA-Rain were immunoprecipitated using an anti-HA antibody coupled to beads. The presence of Ras and HA-Rain in the immune complexes was detected by immunoblotting with a pan-Ras antibody and anti-HA antibody, respectively. Molecular mass markers (kilodaltons) are indicated on the left. D, Rain interacts with other small GTPases. Protein interaction between Gal4 TAD Rain-RA and the indicated small GTPases fused to LexA DB was tested using a yeast two-hybrid assay as described in A. E, Rain interacts with other small GTPases. HEK293 cells were transiently transfected with expression vectors for the activated, Flag-tagged Ras, Rap, R-Ras, and M-Ras/R-Ras3 proteins. Cell lysates were prepared and the interaction of each protein with GST-, GST-Raf-RBD, or GST-Rain-RA proteins was assessed using standard pull-down assays. Immunoblotting with an anti-Flag antibody was used to detect bound Ras family members. The asterisk marks a nonspecific band that appears in all reactions with GST-Rain-RA. Similar results were obtained in three independent experiments. membrane association and seemed to be located in a perinuclear region of the cell (Fig. 5A). As expected, when activated Ras61L was expressed alone, the protein was localized to the plasma membrane as well as to a perinuclear region (Fig. 5A). It is interesting that when HA-Ras61L was co-expressed with GFP-Rain, we observed a pool of Ras co-localized with GFP-Rain to a perinuclear region within the cells and no detectable GFP-Rain fluorescence at the plasma membrane.
The perinuclear site of accumulation of Ras has been identified previously as Golgi (7,9). To confirm that the site of Rain and Ras co-localization was the Golgi, we co-expressed CFP-Rain and a trans-Golgi marker, YFP-galactosyl transferase (GalT), in the absence or presence of several H-Ras mutants. As shown in Fig. 5C, in the absence of activated H-Ras61L, CFP-Rain was detected in perinuclear, juxta-Golgi, and cytoplasmic vesicles and did not co-localize with YFP-GalT. However, in the presence of overexpressed wild-type H-Ras, some CFP-Raincontaining vesicles were found at the Golgi and the fluorescent signal overlapped partially with YFP-GalT. In cells expressing a constitutively active mutant of Ras, association at the Golgi was increased, and we observed an overlap of CFP-Rain and YFP-GalT fluorescence. In contrast, CFP-Rain retained perinuclear, juxta-Golgi localization and was not recruited to the Golgi when co-expressed with dominant-negative, GDP-bound Ras (RasN17). These data indicate that Rain is present in juxta-Golgi vesicles in intact cells and is recruited to Golgi membranes in the presence of active Ras. This, in turn, strongly suggests that Rain is an effector for activated Ras at Live cells were imaged 20 h after transfection using a Zeiss LSM 510 confocal microscope. In these images, the YFP channel was assigned red, the CFP channel was assigned green, and co-localized signals appear yellow. the endomembrane, but not at the plasma membrane.
Rain Cooperates with Activated Raf to Cause Synergistic Transformation-The potent transforming activity of oncogenic Ras is a caused by the combined activation of multiple, effector-mediated signaling pathways (3)(4)(5)(6). One approach for assessing the importance of Ras effector proteins to Ras function is the use of cooperation focus formation assays (31)(32)(33). For example, whereas activated RalGDS or PI3K alone do not cause transformation of NIH3T3 cells, co-expression of these effectors with activated Raf results in a synergistic enhancement of transforming activity (12,33).
Before testing the synergistic transforming activity of Rain, NIH3T3 cells were transfected with an expression vector for Rain. Although the protein was expressed by the cells, no foci were formed (data not shown). In contrast, NIH3T3 cells transfected with an expression vector for activated Raf (Raf22W) displayed the expected level of focus formation (Fig. 6, A and  B). When Rain and Raf22W were co-expressed in NIH3T3 cells, a 2-fold increase in focus formation was observed. This level of transformation was comparable with that obtained with RhoA(63L) and Raf22W, which are know to act synergistically in this assay (32). These data support the possibility that Rain, when activated in the context of other Ras effectors, can promote transformation. DISCUSSION Recent studies demonstrated that, in addition to the plasma membrane, H-Ras also localizes to endomembranes of the endoplasmic reticulum and Golgi, where it is activated in response to growth factor stimulation (7)(8)(9). It is becoming clear now that signal output from endomembrane-localized Ras is quantitatively different from plasma membrane-associated Ras (9; reviewed in Ref. 34), although questions remain as to what role(s) endomembrane-localized Ras plays in cells and which signaling pathways it engages. Most of the Ras effectors identified to date are cytosolic and are recruited dynamically to the plasma membrane upon Ras activation. This suggests that Golgi-localized Ras may engage new, as-yet-unidentified, effectors to transmit signals. Identification of such effectors will be a critical step in our understanding of endomembrane-localized Ras signaling.
In the present study, we report the identification of a putative Ras effector, Rain. Similar to all known effectors of Ras, Rain associates preferentially with activated Ras-GTP, and this interaction is dependent on an intact Ras effector-binding domain. However, in contrast to Raf and other Ras effectors, Rain does not seem to be a cytosolic protein and does not transit to the plasma membrane upon Ras activation. Instead, we found that Rain was located in a perinuclear, juxta-Golgi compartment and, upon Ras activation, is recruited to the Golgi but not to the plasma membrane. Most importantly, we found that Rain cooperates with Raf in cellular transformation assays, suggesting that Rain may serve as an effector for Ras localized to activated endomembrane but not plasma membrane.
We identified Rain in a yeast two-hybrid library screen using constitutively active H-Ras as a bait. Similar to the established effectors of Ras, Rain contains an RA (RalGDS/AF-6) domain in the amino terminus. The RA domain of Rain binds preferentially to Ras-GTP through an intact Ras effector domain. The strong sequence identity between the core effector domains of several members of the Ras superfamily of proteins (35) permits the interaction of Ras effectors with other small GTPases. For example, Raf binds to Ras, Rap1, and R-Ras (29), and RalGDS binds to Ras, Rap1A, Rap2, R-Ras, and TC21 (36 -39). In this study, we demonstrate that the Rain RA domain strongly interacts with Rap1A and Rap2 and shows a weaker interaction with activated TC21 and R-Ras. This profile of interaction is similar to that observed for AF-6 (40), and recent evidence has demonstrated that these AF-6 interactions are functionally significant. AF-6 can serve as a negative regulator of Ras-mediated activation of Erk (41) but also functions as a Rap effector during dorsal closure of the Drosophila melanogaster embryo (42) and as a component of the Rap-GTPase activating protein/Rap-GTP complex regulating Rap activation during ␤ 1 integrin-mediated cell adhesion (43). We have not examined the role of Rain in Rap-mediated signaling. However, given the fact that Rap is found on endomembranes and is activated in that location after exposure of cells to growth factors (9), there is a strong possibility that Rain functions as a physiological effector of both Ras and Rap proteins in cells.
Our analyses with Ras effector domain variants determined that Rain efficiently binds to the E37G but not to the T35S or Y40C mutants of activated H-Ras (12V). These Ras effector domain variants exhibit differential impairment of binding to the three key effectors important for Ras transformation (31). T35S retains binding to Raf but not to PI3K or RalGDS, E37G binds RalGDS but not Raf or PI3K, and Y40C binds PI3K but not Raf or RalGDS. It is interesting to note that although T35S is competent to transform NIH3T3 mouse fibroblasts, E37G shows the greatest transforming activity when assayed in human fibroblasts or epithelial cells (17,44,45). We found that FIG. 6. Rain cooperates with Raf to transform NIH3T3 cells. NIH3T3 cells were transfected with plasmids expressing activated Raf22W alone or together with plasmids encoding activated RhoA63L, Rain, or an empty vector control and scored for focus formation after 16 days. A, quantification of foci using phase contrast microscopy. The average number of foci per plate for each experimental group is represented (see ''Experimental Procedures'' for details). Bars denote standard error. B, for documentation, the plates from A were stained with 0.4% crystal violet and photographed. co-expression of Rain synergistically enhanced Raf transforming activity, which is consistent with the effector domain interaction data. Because E37G transformation of human cells cannot be attributed solely to interaction with RalGDS, it is likely that Rain contributes in some way to the transforming actions seen with this effector domain mutant of Ras.
Perhaps our most intriguing observation is the demonstration that ectopically expressed Rain localizes to a perinuclear, juxta-Golgi region and becomes recruited/relocalized to the trans-Golgi region after expression of activated Ras. This suggests that Rain can serve as an effector of Golgi-localized Ras because its cellular localization is influenced by the activation status of Ras. We have confirmed this striking pattern of Rain localization in a variety of mammalian cell lines (COS-7, HEK293, NIH3T3) using other forms of full-length, tagged Rain protein (GFP, HA, and Flag). In addition, we were able to rule out the possibility of artifacts caused by cell fixation by comparing the localization of Rain in both fixed and live cells. Unfortunately, although Rain transcripts are detected in the majority of tissue samples examined, our anti-Rain antiserum has not detected endogenous Rain in situ by immunostaining or in tissue or cell line extracts (other than lung) (Fig. 3) by immunoblotting. As a result, at this time we are unable to confirm the localization of endogenous Rain in cells. As additional antibodies are being generated, we will continue to screen cell lines with our available antiserum in an effort to identify cell types that mirror the high level of Rain expression detected in mouse lung tissue.
Similarities between the Rain and AF-6 protein families (Fig. 2) extend past the RA domains to a second motif referred to as a DIL domain, which is a characteristic feature of the myosin V family of proteins (24). In myosins, the DIL domain is located in the globular carboxyl-terminal extension and may be involved in binding to cargo during vesicle trafficking (25,26). Takai and colleagues (46,47) have reported that Afadin, the rat homolog of AF-6, localizes to the Golgi in addition to its primary location at cell-cell junctions. Our preliminary data indicate that both the amino-(RA domain) and the carboxyl-(DIL domain) termini of Rain are required for its subcellular localization to a perinuclear, juxta-Golgi region (data not shown). Thus, it is tempting to speculate that Rain binds to Ras-GTP through its amino terminus and perhaps facilitates movement of Ras to the plasma membrane via contact with vesicles through its carboxyl terminus.
In summary, we have identified Rain as a putative effector of Ras, and possibly Rap, function in mammalian cells. In contrast to other Ras effectors, Rain does not associate with activated Ras at the plasma membrane. Thus, Rain may serve as an effector specific for endomembrane-associated Ras and facilitate a unique function for endomembrane-localized Ras that distinguishes it from the function of Ras at the plasma membrane. Our future studies will include an evaluation of the functional relevance of the Ras-Rain interaction and its role in cellular proliferation and differentiation.