RasGRP, a Ras Guanyl Nucleotide- Releasing Protein with Calcium- and Diacylglycerol-Binding Motifs

See allHide authors and affiliations

Science  15 May 1998:
Vol. 280, Issue 5366, pp. 1082-1086
DOI: 10.1126/science.280.5366.1082


RasGRP, a guanyl nucleotide–releasing protein for the small guanosine triphosphatase Ras, was characterized. Besides the catalytic domain, RasGRP has an atypical pair of “EF hands” that bind calcium and a diacylglycerol (DAG)-binding domain. RasGRP activated Ras and caused transformation in fibroblasts. A DAG analog caused sustained activation of Ras-Erk signaling and changes in cell morphology. Signaling was associated with partitioning of RasGRP protein into the membrane fraction. Sustained ligand-induced signaling and membrane partitioning were absent when the DAG-binding domain was deleted. RasGRP is expressed in the nervous system, where it may couple changes in DAG and possibly calcium concentrations to Ras activation.

The cellular properties of neurons are modulated by a number of extrinsic signals, including synaptic activity, neurotrophic factors, and hormones. These signaling systems alter the intracellular concentrations of second messengers such as calcium and cyclic nucleotides, and these small molecules can regulate the activities of protein kinases (1). As the mechanisms linking Ras signaling to nerve function are not completely understood, we developed a cDNA cloning approach to identify proteins that enhance Ras signaling in the brain. From rat brain mRNA, we derived cDNAs that could complement a transformation-defective allele of v-H-ras in a fibroblast transformation assay 2).

One cloned cDNA, rbc7 (rat brain cDNA 7), was weakly transforming in rat2 fibroblasts in the absence of the transformation-defective allele of v-H-ras. However, the rbc7 product did appear to function in the Ras pathway. The morphology of transformed foci was similar to that obtained with an activated version of Ras. Compared with control cells, pooled populations of rbc7-expressing cells exhibited a higher saturation density, some anchorage-independent growth, and a tumorigenic phenotype. More pronounced morphological transformation was exhibited by rat2 cells expressing rbc7 and overexpressing c-H-ras. We also observed strong morphological transformation when rbc7 was expressed in rv68BUR, a somatic mutant that is hypersensitive to transformation by Ras. This rat2 derivative is heterozygous for an activating mutation inMek1, a kinase that functions downstream of Ras in cell transformation (3, 4).

Analysis of the sequence of rbc7 and the deduced protein product indicated that the cDNA is a 5′ and 3′ truncated version of a larger normal transcript (Fig. 1A). The sequence of the normal version of rbc7, which we refer to as RasGRP (Ras guanyl-releasing protein), was determined from rat brain cDNAs isolated from a phage library and from polymerase chain reaction (PCR) products (5). Within the predicted rbc7 and RasGRP polypeptides, we identified several putative functional regions (Fig. 1). The catalytic region includes the CDC25 box, named for the prototypic Ras activator from Saccharomyces cerevisiae (6), and the Ras exchange motif (REM) typical of guanyl nucleotide–releasing factors that interact with Ras and its closest relatives (7). From the analysis of conserved sequences within the CDC25 box, RasGRP exhibits about 50% similarity to RasGRF1, a brain-specific Ras activator that is regulated by calcium-bound calmodulin (8,9). RasGRP also exhibits about 50% similarity to SOS1 (10), a Ras activator that links receptor tyrosine kinases to Ras in a variety of cell types (11).

Figure 1

Structures of rbc7 and RasGRP proteins. (A) Schematic domain structure map of RasGRP/rbc7. The REM box, CDC25 box, EF hands, and DAG-binding domain (DG) are shown as open boxes. The regions found in RasGRP but missing in rbc7 are shown as hatched boxes. (B) The sequence of RasGRP was deduced from the sequence of the rbc7 isolate and from overlapping cDNA clones. The entire coding sequence was confirmed from PCR products recovered by reverse transcription PCR from rat brain RNA. RasGRP and rbc7 differ at two internal sites: Asp404 is Asn in rbc7 and Gly576 is Glu in rbc7. The Met and Pro residues underlined represent the first and last residues of the deduced rbc7 product. rbc7 encodes a 550-residue, 63.4-kD product. RasGRP encodes a 795-residue, 90.3-kD product (Genbank accession number AF060819). (C) Sequence comparisons of the deduced protein primary structures. The REM box and the three structurally conserved regions (SCRs) within the CDC25 box (29) are aligned with the corresponding regions of mouse SOS1 (10) and rat RasGRF1 (8). The EF hands are aligned with those of calcineurin B (Cal B) (30). The DAG-binding domain of RasGRP is aligned with that of PKCδ (31). In all alignments, residues that are identical or chemically similar are in bold; those that are identical are marked with an asterisk. In the alignments of the REM and CDC25 boxes, residues that are underlined in RasGRP are identical or similar to either SOS1 or RasGRF1. In the alignment of the EF hands, the underlined residues are those that interact with calcium. In the alignment of the DAG domain of RasGRP to that of PKCδ,the underlined residues are those that coordinate with zinc atoms and are conserved in all DAG-binding proteins (32).

Besides the catalytic region, RasGRP has a structure resembling a pair of calcium-binding “EF hands” (12). This calcium-binding module differs from the typical paired EF hand structure in that the region between the calcium-binding loops consists of only 15 residues rather than the 20 to 30 residues typically found. RasGRP also has a diacylglycerol (DAG)–binding domain. On the basis of the sequence analysis of the CDC25 box and the presence of the EF hands and the DAG-binding domain, RasGRP appears to be a third type of mammalian Ras guanyl nucleotide–releasing factor in the CDC25 family.

We used bacterial systems to express RasGRP sequences, and the expressed proteins were tested for the relevant biochemical activities. A protein consisting of the catalytic region enhanced dissociation of the Ras–guanosine diphosphate (GDP) complex and the association of Ras with guanosine triphosphate (GTP) (Fig.2, A and B) (13). No activity was observed when either R-Ras or RhoA was used as a substrate (14). Recombinant RasGRP protein and Ras also formed a stable complex in vitro (15). To examine binding of calcium to the EF hands, we resolved glutathione S-transferase (GST)–RasGRP fusion proteins expressed in Escherichia coli by electrophoresis, transferred them to a nitrocellulose filter, and probed them with 45Ca (Fig. 2C). GST-GRP bound calcium, as did GST-EF1, which contains alanine substitutions in the first EF hand (16). In contrast, GST-EF2 and GST-EF1EF2, which contain substitutions in the second EF hand or both EF hands, respectively, did not bind calcium. Thus, the second EF hand is apparently the higher affinity site. A fusion protein consisting of GST and the DAG-binding domain bound to [3H]phorbol 12,13-dibutyrate (PDBu), a DAG analog (Fig. 2D) (17).

Figure 2

Biochemical analysis of RasGRP proteins. (A) Dissociation of Ras-GDP promoted by RasGRP. Release of [3H]GDP from Ras was followed with an immunoprecipitation procedure (13). Reactions included buffer (Buff.), RasGRP (catalytic region) (GRP), or p30GRF1 (GRF1), as indicated. Values represent the proportion of GDP that was released from the Ras-GDP complex and are the average of three determinations with the standard error of the mean indicated. (B) Association of Ras and GTP promoted by RasGRP. Ras was incubated with [α32-P]GTP, and the formation of Ras-GTP was monitored (13). Reactions included buffer, RasGRP (catalytic region), or p30GRF1, as indicated. The amount of complex that formed is expressed as a proportion of that observed under saturating conditions. (C) Binding of calcium by GST-RasGRP proteins. Proteins consisting of the rbc7HA product fused to GST were resolved by SDS-PAGE, blotted to nitrocellulose, and then probed with45Ca. Lane 1, wild type; lane 2, GST-EF1 (the first EF hand mutated); lane 3, GST-EF2 (the second EF hand mutated); and lane 4, GST-EF1EF2 (both EF hands mutated) (16). (D) Interaction of the DAG-binding domain with a DAG analog. Test samples were incubated with labeled PDBu, and ligand complexed with protein was assessed by filter binding. Column 1, buffer control; column 2, 50 μg of soluble mouse brain protein (including PKC); column 3, 50 μg of GST protein; and column 4, 50 μg of GST-DG protein (17). Values are the average of three determinations with the standard error of the mean indicated.

To determine how the domains of RasGRP might contribute to Ras function, we engineered fibroblasts to express various forms of the protein. A hemagglutinin (HA) epitope–tagged version of the transforming sequence, rbc7HA, directed expression of a 66-kD protein (18). Relative to rat2 cells expressing the empty vector, cells expressing rbc7HA exhibited an increased amount of Ras-GTP (Fig.3A) (19). When cells expressing rbc7HA or full-length RasGRP were treated with phorbol 12-myristate 13-acetate (PMA), a DAG analog, the amount of Ras-GTP increased. To examine the response to more physiologic stimuli, we treated cells expressing RasGRP with endothelin-1. In parental rat2 cells, this peptide growth factor results in weak activation of the mitogen-activated protein kinases, Erk1 and Erk2, probably by tyrosine kinase–dependent pathways (20). Endothelin-1 also stimulates phospholipid breakdown in rat fibroblasts (21), and this process increases the concentrations of membrane DAG and free cytoplasmic calcium. Rat2 cells engineered to express full-length RasGRP exhibited an increased level of Ras-GTP that increased further when cells were treated with endothelin-1 (Fig. 3B).

Figure 3

Facilitation by RasGRP proteins of Ras signaling initiated by PMA and endothelin-1. (A) Cells were incubated with 32Pi for 4 hours to label GTP and GDP pools. Rat2 cells expressing the empty vector (1, 2) or the rbc7-expressing vector (3, 4) were either untreated (1,3) or treated with PMA for 2 min (2,4). Guanyl nucleotides associated with Ras were immunoprecipitated and analyzed. The values represent the amounts of GTP expressed as a percentage of total guanyl nucleotide (19). 7HA, rbc7HA. (B) Rat2 cells expressing empty vector (Puro) or RasGRP (GRP) were either left untreated or treated with endothelin-1 (Endo-1) (100 nM) for 10 min. Postnuclear cell lysates were prepared and then incubated with either GST-RBD (the Ras-binding domain of Raf) (+) or GST alone (−). After precipitation, 21-kD Ras was detected with an immunoblot method. Cells also overexpressed c-H-Ras. The total amounts of Ras in the lysates were very similar. (C) The activation state of Erk was assessed with an SDS-PAGE mobility shift assay after various periods of PMA treatment. Lanes 1 to 8, untreated or treated for 2, 5, 10, or 60 min or 4, 6, or 8 hours. Rat2 cells expressed empty vector (top), rbc7HA (middle), or ΔDG (bottom). The arrows indicate the positions of the phosphorylated pp42 species of Erk2. Erk1 behaved similarly but was only visualized upon longer exposure to film. (D) Cell morphology was studied after exposure to PMA (100 nM) for 40 hours. Rat2 cells expressing rbc7HA were exposed to solvent control (panel 1), PMA (10% serum) (panel 2), or PMA (0.5% serum) (panel 3). Rat2 cells expressing ΔDG were exposed to solvent control (panel 4), PMA (10% serum) (panel 5), or PMA (0.5% serum) (panel 6). (E) Rat2 cells that expressed either rbc7HA or ΔDG were labeled with [35S]methionine overnight, treated with either solvent or PMA for 2 min, and then disrupted in a Tenbroeck-style homogenizer. After a low-speed centrifugation to remove unbroken cells and nuclei, the lysates were centrifuged at 100,000g to prepare particulate (P) and soluble (S) fractions. Proteins in each fraction were solubilized in buffer containing Triton X-100, and RasGRP proteins were precipitated with an antibody to HA, resolved by SDS-PAGE, and visualized by fluorography.

Treatment of parental rat2 cells with PMA resulted in transient and incomplete activation of Erk (Fig. 3C). This process may have involved the direct activation of the Raf-Mek-Erk protein kinase cascade by protein kinase C (PKC) (22), and it did not involve an increase in Ras-GTP (Fig. 3A). When cells expressed full-length RasGRP or the truncated rbc7HA protein, PMA-induced activation of Erk was strong and sustained. When cells expressed ΔDG, a version of rbc7HA missing the DAG-binding domain (Fig. 1A), Erk activation was similar to that seen with the empty vector (Fig. 3C).

When subjected to prolonged exposure to PMA, rat2 cells expressing rbc7HA assumed a transformed morphology (Fig. 3D). Reducing the concentration of serum in the culture medium from 10 to 0.5% exaggerated this effect of PMA on morphology, although serum reduction alone was without effect. The ΔDG protein did not facilitate a substantial PMA-induced change in cell morphology, confirming the importance of the DAG-binding domain for RasGRP signaling.

To question whether DAG signaling might serve to recruit RasGRP to the plasma membrane, where it can interact efficiently with Ras, we monitored subcellular fractionation of RasGRP. Less than half of the rbc7HA protein was found in the membrane preparation from untreated cells. More of the protein was associated with the membrane fraction after PMA treatment (Fig. 3E). This increase was not seen with the ΔDG protein.

A single 5.6-kb RasGRP RNA species was detected in brain but not in other tissues (Fig. 4A) (23). Expression of the normal RasGRP transcript was widespread in the brain with the highest grain densities in the hippocampal CA1 and CA3 fields (Fig. 4B) (24). The majority of the grains were located within the pyramidal cell layers, probably over the pyramidal cell bodies. In contrast, we did not detect RasGRP mRNA in the granule cells of the adjacent dentate gyrus.

Figure 4

Expression of RasGRP RNA in the nervous system. (A) RNA blot hybridization of RasGRP transcripts from transformed cells and various adult rat tissues. Lane 1, rat2 cells engineered to express rbc7HA; lane 2, brain; lane 3, liver; lane 4, lung; and lane 5, intestine. In other blots, heart, muscle, spleen, and uninfected rat2 fibroblasts were negative. Exposure was for 4 days. The agarose gel was stained with ethidium bromide before blot transfer to check that the RNA was intact and that each lane contained similar amounts of RNA. Numbers at left are molecular masses given in kilobases. (B) Expression of RasGRP mRNA in the adult hippocampus. (a) Film autoradiograph of coronal section showing localization of RasGRP mRNA in CA1 and CA3. Label in the dentate gyrus (DG) is not above background levels. (b) Emulsion autoradiography showing the section outlined by the square in (a). RasGRP is restricted to the regions over the pyramidal cell bodies in these sections.

Ras-GTP is thought to stimulate at least two effector pathways: the Raf-Mek-Erk kinase cascade and the phosphoinositide 3-OH kinase pathway, which can result in activation of the protein kinase Akt/PKB (25). In neurons, signaling in the Raf-Mek-Erk pathway promotes differentiation, axonal growth, and synaptic plasticity (26), whereas signaling through Akt/PKB is associated with cell survival (27). Our characterization of RasGRP suggests that it is a Ras activator that links DAG and possibly calcium second messengers to Ras output in neurons. Both calcium and DAG can arise from several sources in neurons, and they could affect RasGRP independently. Alternatively, receptors linked to phospholipase C stimulate cleavage of phosphatidylinositol bis-phosphate, and the resulting calcium and DAG signals could coordinately regulate RasGRP (28).

  • * To whom correspondence should be addressed. E-mail: jim.stone{at}


Stay Connected to Science

Navigate This Article