RasGRP4, a New Mast Cell-restricted Ras Guanine Nucleotide- releasing Protein with Calcium- and Diacylglycerol-binding Motifs IDENTIFICATION OF DEFECTIVE VARIANTS OF THIS SIGNALING PROTEIN IN ASTHMA, MASTOCYTOSIS, AND MAST CELL LEUKEMIA PATIENTS AND DEMONSTRATION OF THE IMPORTANCE OF RasGRP4 IN MAST

From the ‡Department of Medicine, Brigham and Women’s Hospital, and Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, the Department of Medicine, University of New South Wales, and Department of Immunology, Allergy, and Infectious Disease, St. George Hospital, Kogarah, New South Wales 2217, Australia, and **Laboratories of Molecular Biophysics, Pels Family Center for Biochemistry and Structural Biology, Rockefeller University, New York, New York 10021

A cDNA was isolated from interleukin 3-developed, mouse bone marrow-derived mast cells (MCs) that contained an insert (designated mRasGRP4) that had not been identified in any species at the gene, mRNA, or protein level. By using a homology-based cloning approach, the ϳ2.6-kb hRasGRP4 transcript was also isolated from the mononuclear progenitors residing in the peripheral blood of normal individuals. This transcript information was then used to locate the RasGRP4 gene in the mouse and human genomes, to deduce its exon/ intron organization, and then to identify 10 single nucleotide polymorphisms in the human gene that result in 5 amino acid differences. The >15-kb hRasGRP4 gene consists of 18 exons and resides on a region of chromosome 19q13.1 that had not been sequenced by the Human Genome Project. Human and mouse MCs and their progenitors selectively express RasGRP4, and this new intracellular protein contains all of the domains present in the RasGRP family of guanine nucleotide exchange factors even though it is <50% identical to its closest homolog. Recombinant RasGRP4 can activate H-Ras in a cation-dependent manner. Transfection experiments also suggest that RasGRP4 is a diacylglycerol/phorbol ester receptor. Transcript analysis of an asthma patient, a mastocytosis patient, and the HMC-1 cell line derived from a MC leukemia patient revealed the presence of substantial amounts of non-functional forms of hRas-GRP4 due to an inability to remove intron 5 in the precursor transcript. Because only abnormal forms of hRas-GRP4 were identified in the HMC-1 cell line, this immature MC progenitor was used to address the func-tion of RasGRP4 in MCs. HMC-1 leukemia cells differentiated and underwent granule maturation when induced to express a normal form of RasGRP4. Thus, RasGRP4 plays an important role in the final stages of MC development.
Mast cells (MCs) 1 release a diverse array of biologically active molecules (including cytokines, chemokines, leukotrienes, prostaglandins, amines, proteoglycans, and proteases) when activated via their high affinity IgE (Fc⑀RI), complement, or protease-activated receptors. MCs play important roles in bacteria infections (1-3) due, in part, to their release of tumor necrosis factor-␣ (4) and varied granule tryptases (5,6) that work in concert to induce the extravasation of neutrophils into tissues that kill bacteria. Although these and other findings document that MCs play beneficial immunosurveillance and effector roles in the body, it has been known for some time that MCs also exhibit adverse roles in numerous inflammatory disorders including asthma, chronic urticaria, and systemic anaphylaxis. Recent studies (7)(8)(9) have revealed that these effector cells also participate in the pathogenesis of AIDS.
Because MCs play such prominent roles in inflammation, a major effort has been made to identify the regulatory proteins that act relatively early in the varied intracellular signaling pathways of this cell to dampen MC development and/or function. Substantial progress has been made in our understanding of how different populations of MCs signal through their Ig, complement, and cytokine receptors, even though no signal transduction protein has been identified to date that is relatively specific to this cell type. The Ras superfamily of small GTP-binding proteins plays pivotal roles in virtually every cell. GTP activates the varied Ras family members, whereas GDP inactivates them. Guanine nucleotide exchange factors (GEFs) activate these signaling proteins by dissociating bound GDP. All GEFs that have been identified to date in MCs (e.g. the Ras GEF Sos (10) and the Rac GEF Vav (11,12)) are expressed in many other cells. Nevertheless, the fact that a few GEFs that are more restricted in their cellular expression have been identified during the last couple of years has raised the possibility that an undiscovered, more restricted GEF might be present in MCs. Three such relatively restricted regulatory proteins are Ras guanine nucleotide releasing protein (RasGRP) 1 (also known as CalDAG-GEFII), RasGRP2 (also known as CalDAG-GEFI), and RasGRP3 (also known as KIAA0846 and CalDAG-GEFIII) (13)(14)(15)(16)(17)(18)(19)(20)(21). There is considerable interest in RasGRP1, RasGRP2, and RasGRP3, in part because these homologous GEFs have additional Ca 2ϩ and phorbol ester/diacylglycerol (DAG)-binding motifs in their C-terminal domains that implicate their involvement in multiple signaling pathways. T cells express RasGRP1 and targeted disrupted of this GEF in mice results in a marked deficiency of mature CD4 ϩ /CD8 Ϫ and CD4 Ϫ /CD8 ϩ T cells (22). Thus, at least one RasGRP is of critical importance in cellular differentiation and/or maturation.
Mouse bone marrow-derived MCs (mBMMCs) developed with interleukin (IL) 3 (23)(24)(25) are the non-transformed cells that have been most widely used to understand at the molecular level the factors and mechanisms that regulate the proliferation, differentiation, maturation, adherence, cellular senescence, and function of mammalian MCs. For example, mBMMC cDNA libraries have been used by us to clone mouse MC protease-5, mouse MC protease-7, transmembrane tryptase, and serglycin. We also have used mBMMCs to identify some of the key cells and cytokines that regulate the development and function of MCs.
We now describe the cloning of a new GEF from human and mouse MCs and their progenitors. This intracellular protein is designated as RasGRP4 (or CalDAG-GEFIV based on the alternative nomenclature for these proteins) because it is the fourth member of its family. MCs express H-Ras (26), and we show that recombinant RasGRP4 can activate this Ras isoform in vitro in a Ca 2ϩ -and Mg 2ϩ -regulated manner. The cloning of the mouse and human RasGRP4 cDNAs and genes now provides investigators the opportunity to identify defects in this new signaling protein in MC-dependent diseases. In this regard, we additionally describe single nucleotide/amino acid polymorphisms of the hRasGRP4 gene, transcript, and protein.
We also describe splice variants of the hRasGRP4 transcript in a mastocytosis patient, an asthma patient, and an immature MC line derived from a MC leukemia patient (27) that encode substantially altered forms of hRasGRP4. Finally, using the latter leukemia MC line, we show that RasGRP4 is a signal transduction protein that participates in the final stages of MC development, most likely by acting downstream of c-kit.

EXPERIMENTAL PROCEDURES
Cloning of the mRasGRP4 and hRasGRP4 Transcripts and Genes and Chromosomal Location of the Human Gene-Greater than 2000 clones were arbitrarily isolated and sequenced from an mBMMC cDNA library (28) using standard molecular biology procedures. The BALB/c mBMMCs used to create the library had been cultured for 6 weeks to ensure that no contaminating cell types were present. As assessed by RNA blot analysis, one of the isolated clones corresponded to all but a few hundred nucleotides of the major mRasGRP4 transcript present in mBMMCs. Thus, a "rapid amplification of cDNA ends" (RACE) approach was carried out with an RLM-RACE kit from Ambion (Austin, TX) to deduce the nucleotide sequence of the missing 5Ј portion of the mRasGRP4 transcript. BALB/c mBMMCs, generated as described previously (25), were the cellular source of the mRNA used in this 5Ј-RACE reaction. The first PCR was carried out using the outer adapter 5Ј-G-CUGAUGGCGAUGAAUCAACACUGCGUUUGCUGGCUUUGAUGA-AA-3Ј and the mRasGRP4-specific antisense oligonucleotide 5Ј-CGGA-ACTCCCAGGTAGTGA-3Ј. The second nested PCR was carried out using the inner adapter 5Ј-CGCGGATCCGACACTCGTTTGCTGGCT-TTGATGAAA-3Ј and the mRasGRP4-specific antisense oligonucleotide 5Ј-CAGGACTTAGCAGGCTGGAG-3Ј. Each of the 30 cycles in the PCR consisted of a 30-s denaturing step at 94°C, a 30-s annealing step at 55°C, and a 2.5-min extension step at 72°C. Multiple amplified products were subcloned into pCR2.1-TOPO (Invitrogen, Carlsbad, CA), and their inserts were sequenced in both directions using an ABI-377 sequencer and standard methods to deduce the nucleotide sequences of full-length transcripts in mBMMCs.
The site where the hRasGRP4 gene resides on human chromosome 19q13.1 had not been sequenced in its entirety before we released our mRasGRP4 and hRasGRP4 nucleotide sequence data to the public at our GenBank TM sites AF331457, AY040628, AY048119, AY048120, and AY048121. Nevertheless, a genomic fragment was identified near the chromosome 19q13.1 gap site that was homologous to the nucleotide sequence that corresponds to exons 1-9 of the mRasGRP4 gene. By using primers that correspond to sequences in this human genomic fragment, a 900-bp cDNA was isolated from the mononuclear cells in human peripheral blood that we concluded probably encoded the corresponding portion of the mRasGRP4 transcript. To deduce the nucleotide sequence of a full-length 2.6-kb hRasGRP4 transcript, a RACE approach was then carried out with an RLM-RACE kit from Ambion (Austin, TX), and total RNA was obtained from peripheral blood leukocytes. After the mononuclear cells were enriched from peripheral blood using Ficoll-Paque (Amersham Biosciences AB), they were lysed, and total RNA was extracted and purified using Trizol reagent (Invitrogen). Human leukocyte Marathon-Ready TM cDNAs (CLONTECH, Palo Alto, CA) also were used in the identification of the major hRasGRP4 expressed in the human population. The RACE was carried out using the 5Ј-RACE primer 5Ј-ACCTGTCGGGCTGTGCCTCA-3Ј and the 3Ј-RACE primer 5Ј-CAGCACCAAGGCCCTCCTGGAGT-3Ј. Each of the 30 cycles in the PCR consisted of a 30-s denaturing step at 94°C, a 30-s annealing step at 55°C, and a 2.5-min extension step at 72°C. Multiple amplified products were subcloned into pcDNA3.1/Directional/V5-His-TOPO (Invitrogen, Carlsbad, CA), and their inserts were sequenced in both directions using an ABI-377 sequencer.
Tissue Distribution of the hRasGRP4 Transcript-Human cDNA libraries (CLONTECH) from adult heart, brain, placenta, lung, liver, skeletal muscle, kidney, pancreas, spleen, thymus, prostate, testis, ovary, small intestine, colon, and leukocytes; from fetal heart, brain, lung, liver, kidney, spleen, thymus, and skeletal muscle; and from peripheral blood mononuclear cells, CD4 ϩ T lymphocytes, CD8 ϩ T lymphocytes, and CD19 ϩ B lymphocytes (before and after lectin activation) were used to evaluate the relative distribution of the hRasGRP4 transcript in varied adult and fetal human tissues and in varied peripheral blood cells. Each library was created by CLONTECH using pooled poly(A) ϩ RNA from a large number of people. For example, the leukocyte cDNA library used in this study to clone hRasGRP4 cDNAs was generated using mRNA pooled from 550 individuals. A blot (CLON-TECH) containing ϳ2 g of poly(A) ϩ RNA from varied human tissues also was probed under conditions of high stringency with a radiolabeled 303 bp-probe that corresponds to residues 329 -631 in an hRasGRP4 cDNA.
Four apparently normal individuals, a patient with asthma, a patient with systemic mastocytosis, and a MC line derived from a MC leukemia patient (27) were evaluated individually for the presence of abnormal hRasGRP4 transcripts. The systemic mastocytosis patient used in our study was a 53-year-old male that had the disease for ϳ25 years. He presented in 1997 with marked abdominal swelling and extensive lymphadenopathy on a background of long term severe flushing and diarrhea. Physical examination demonstrated extensive skin involvement with pigmentation and erythematous areas on his face trunk and limbs. He had marked ascites hepatomegaly down to the right iliac fossa and marked axillary, inguinal, cervical, and epitrochlear lymphadenopathy. He also had mild splenomegaly. Bone marrow trephine carried out July 1997 demonstrated extensive fibrosis and abnormal megakaryocytes, moderate eosinophilia, and reduced numbers of erythrocytes and granulocytes. Approximately 3% of the cells in the bone marrow biopsy of the patient were MCs that possessed an abnormal morphology. The marrow was hypercellular with moderate infiltration of abnormal primitive cells. These cells did not express CD3, CD15, CD20, CD34, CD45, CD45RO, and or light chains; they also failed to stain when incubated with periodic acid-Schiff reagent. However, they expressed CD43 and CD117. Analysis of the cells in the bone marrow of the patient on November 1997 revealed that ϳ30% of them were blast cells, consistent with the development of an acute myeloid leukemia. The patient was given chemotherapy. Although the initial response was good, he had a stormy course over the next 3 years. Eventually, the patient died due to the complications of the leukemia. When the patient was in clinical remission, peripheral blood leukocytes were obtained and frozen for future study. RNA, purified from the isolated cells, was used to evaluate hRasGRP4 expression.
FIG. 1. Cloning of the mRasGRP4 cDNA. Depicted is the nucleotide sequence of the cDNA that corresponds to the predominant ϳ2.3-kb mRasGRP4 transcript in BALB/c mBMMCs, as well as the predicted amino acid sequence of its translated product. An alternate form of the mRasGRP4 transcript also was identified in mBMMCs that lacks the 15-nucleotide sequence that encodes the "VSTGP" sequence in the DAG-binding domain at residues 561-565 of the protein. The initial 163 nucleotides in the depicted cDNA correspond to the 5Ј-untranslated region and the first eight amino acids in the translated product. Whether or not the 5Ј-untranslated region is derived from multiple exons remains to be determined. However, preliminary analysis of the mRasGRP4 gene indicates that the 3Ј end of the first coding exon corresponds to residues 118 -140 in the depicted cDNA. The subsequent 16  The mRasGRP4 cDNA lacks a classical "AATAAA" or "ATTAAA" polyadenylation regulatory site 10 -30 residues upstream of its 3Ј poly(A) tract. Nevertheless, because "AAAAAA" has weak polyadenylation promoting activity (67), it is presumed that nucleotides 2272-2277 control the polyadenylation of the mRasGRP4 transcript.
The asthma patient used in our study was a 32-year-old Caucasian male with a history of asthma, allergic rhinitis, conjunctivitis, and atopic dermatitis. The initial diagnosis of asthma was made at age 7. He used inhaled steroids from age 7 to 12. From age 18 to today, he uses inhaled ␤2 agonists and inhaled steroids twice per day. He has been hospitalized three times due to the acute exacerbations of severe asthma. The patient has a history of allergic reactions to house dust mites and cats; he often gets asthmatic attacks during upper respiratory tract infections.
RNA, isolated from the HMC-1 cell line and the mononuclear cells of four normal individuals and the above two patients, was converted into cDNA with a reverse transcription (RT) system from Promega (Madison, WI). PCRs were carried out using 5-l portions of each cDNA preparation (corresponding to ϳ1 ng cDNA) and 0.4 M of the sense oligonucleotide 5Ј-AATGCACCGGAAAAACAGGA-3Ј and 0.4 M of the antisense oligonucleotide 5Ј-TGAGTCTGGAGATGGCACTG-3Ј to generate the 900-bp product that corresponded to exons 1-9 in the hRasGRP4 transcript. In all instances, the 30 cycles of each PCR consisted of a 30-s denaturing step at 94°C, a 30-s annealing step at 55°C, and a 1-min extension step at 72°C. As an internal control, samples also were evaluated for the presence of glyceraldehyde-3phosphate dehydrogenase with the sense and antisense oligonucleotides 5Ј-TGAAGGTCGGAGTCAACGGATTTGGT-3Ј and 5Ј-CATGTGG-GCCATGAGGTCCACCAC-3Ј. Additional PCRs were carried out with the sense oligonucleotide 5Ј-TGCAGATCTGTCACCTGGTC-3Ј and the antisense oligonucleotide 5Ј-CGGAACTCCAGGTAGGTGAG-3Ј and the sense oligonucleotide 5Ј-CTTCTGACCTCCCAGGCCTG-3Ј and the antisense oligonucleotide 5Ј-GTAGCGGGCGTAGTTGTTGT-3Ј to evalu-ate the expression of the variant 1 and 2 transcripts, respectively, in the region that corresponds to exons 5-6 of the normal hRasGRP4 transcript.
hRasGRP4 Immunohistochemistry-Analysis of the primary amino acid sequences of the varied RasGRP family members revealed that the N terminus is poorly conserved in this family of GEFs. A computer search also failed to reveal any amino acid sequence in the GenBank TM protein data bases that resembled the 14-mer peptide Met-Asn-Arg-Lys-Asp-Ser-Lys-Arg-Lys-Ser-His-Gln-Glu-Cys residing at residues 1-14 in hRasGRP4. Thus, an anti-peptide approach was used to obtain rabbit antibodies that specifically recognize the N terminus of hRas-GRP4. The above synthetic peptide was generated by Affinity Bioreagents (Golden, CO) and coupled to keyhole limpet hemocyanin through the succinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate linker with the thiol group of the C-terminal Cys in the peptide. A rabbit was immunized four times with the peptide conjugate (ϳ0.5 mg/immunization) over a 60-day period. The resulting anti-hRasGRP4 antibodies were then purified using a standard peptide affinity chromatography approach.
Immunohistochemistry was carried out on 4% paraformaldehyde/ PBS-fixed paraffin-embedded human breast and stomach tissue sections obtained from Imgenex (San Diego, CA) using standard methodologies. Tryptases and chymases are stored in abundance in the secretory granules of human MCs, and the levels of these granule proteins greatly exceed that of any intracellular signaling protein in a MC. Whereas double-staining approaches often are used to identify human MCs in tissues that coexpress these two families of serine proteases (29), such an immunohistochemical approach cannot be used

FIG. 1-continued
RasGRP4 effectively to identify MCs that express tryptase and hRasGRP4 due to the substantial differences in the levels of these two proteins. Thus, a more reliable serial section approach (30) was used to identify those cells in human breast and stomach that express hRasGRP4.
Generation of Recombinant mRasGRP4 and hRasGR4 -A fulllength mRasGRP4 cDNA was constructed using the ϳ2.1-kb cDNA clone isolated from the mBMMC library and the subsequent 565-bp product generated using the 5Ј-RACE approach. The latter product was liberated from its vector with EcoRI, and then a PCR approach was carried out with the sense oligonucleotide 5Ј-CACCATGAACCGGAAA-GACATCAAA-3Ј and the antisense oligonucleotide 5Ј-CCAAAAC CGG-CTTATGACTG-3Ј to create a product that corresponds to residues 141-590 in the mRasGRP4 transcript. The larger fragment was liberated with HindIII and XhoI. A PCR was carried out with the sense oligonucleotide 5Ј-CATGAATCTGGGAGTGCTGA-3Ј and the antisense oligonucleotide 5Ј-GACGCTGGGCTTCGAGGAAGC-3Ј. The resulting two PCR products were mixed and then used as templates to generate a single product that corresponds to the entire coding domain (i.e. residues 141-2174) of the mRasGRP4 cDNA. Primers used in this final PCR were the sense oligonucleotide 5Ј-CACCATGAACCGGAAAGACA-TCAAA-3Ј and the antisense oligonucleotide 5Ј-GACGCTGGGCTTCG-AGGAAGC-3Ј. Each PCR consisted of a 30-s denaturing step at 94°C, a 30-s annealing step at 55°C, and a 1-min extension step at 72°C. Only 25 cycles were carried out in these PCRs to minimize the generation of point mutations. In order to detect the recombinant protein and to purify it from the lysates of transfected COS-7 cells and 3T3 fibroblasts, the final PCR product was placed into the mammalian expression vector pcDNA3.1/Directional/V5-His-TOPO (Invitrogen) upstream of the sequences that encode the V5 and His 6 peptides. By using a similar approach with the sense oligonucleotide 5Ј-CACCATGAACAG-AAAAGACAGTAAGAGG-3Ј and the antisense oligonucleotide 5Ј-GGA-ATCCGGCTTGGAGGATGCAGT-3Ј, the entire coding domain of an hRasGRP4 cDNA (generated from one of the 2.1-kb cDNAs we isolated from the human leukocyte library) was subcloned into pcDNA3.1/ Directional/V5-His-TOPO.
Vector lacking an RasGRP4 insert was used as a negative control in these expression experiments. African green monkey, SV40-transformed kidney COS-7 cells (line CRL-1651, ATTC), and 3T3 fibroblasts were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Transient transfections were performed with Su-perFect (Qiagen, Valencia, CA) according to the manufacturer's instructions. Cells were plated at a density of 2 ϫ 10 5 cells per well in a 6-well plate 24 h prior to transfection. They were transfected for 2-3 h, trypsinized, and then replated into parallel plates for both immunofluorescence (24-well plates containing 11-mm coverslips) and SDS-PAGE/ immunoblot analysis (12-well plates). Conditioned media and cells were collected 24 h post-transfection. mRasGRP4-and hRasGRP4-expressing fibroblasts also were obtained by transfecting Swiss Albino mouse 3T3 fibroblasts with the above expression plasmids. The resulting fibroblasts were cultured in enriched media supplemented with 200 -500 ng/ml G418 to increase the percentage of RasGRP4-expressing cells in the culture.
The presence of recombinant protein was evaluated by SDS-PAGE/ immunoblotting with anti-V5 antibody (Invitrogen) or the above hRas-GRP4-specific antibody. For immunodetection of the recombinant protein, cell, or tissue, lysates were boiled in SDS sample buffer containing ␤-mercaptoethanol, as were samples of the conditioned media. The resulting soluble proteins were resolved on a 12% polyacrylamide gel (Bio-Rad) and blotted onto polyvinylidene difluoride membranes (Bio-Rad). They were then exposed to Tris-buffered saline containing 0.1% Tween 20, 5% non-fat milk, and 0.5% goat serum to minimize nonspecific binding of the relevant antibody. Treated lysates were exposed to a 5000-fold dilution of a stock solution of mouse anti-V5 antibody or rabbit anti-hRasGRP4 antibody in Tris-buffered saline and 0.1% Tween 20, followed by horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit IgG secondary antibody (Bio-Rad). Immunoreactive proteins were detected using BioMax MR film (Eastman Kodak) and chemiluminescence kits from Pierce. Subcellular fractionation studies were carried out to locate where recombinant RasGRP4 resides in the COS-7 cell transfectants. For these experiments, 1 ϫ 10 7 transfectants were washed three times with Hanks'-buffered salt solution, suspended in 1.4 ml of MIC buffer (4 mM HEPES, pH 7.0, containing 50 mM sucrose, 0.4 mM EDTA, and 0.2 mM dithiothreitol), and sonicated on ice. The resulting lysates were centrifuged at 4°C for 10 min at ϳ1,500 ϫ g and then at ϳ10,000 ϫ g for another 10 min to remove nuclei and other organelles. The soluble fraction at this step was centrifuged for an additional 30 min at ϳ100,000 ϫ g, and the resulting supernatant (cytosolic fraction) was subjected to SDS-PAGE/immunoblot analysis. The resulting plasma membrane-enriched microsomal fraction was washed once with MIC buffer and then placed in 1 ml of MIC buffer supplemented with 10% deoxycholic acid, 10% Triton X-100, and 30% glycerol. After a 30-min incubation at 4°C, the detergent-extracted membrane fraction was centrifuged at ϳ100,000 ϫ g for another 30 min. A sample of this supernatant (detergent-extracted microsomal fraction) also was subjected to SDS-PAGE/immunoblot analysis.
To evaluate its guanine nucleotide exchange activity inside a living cell and to address whether or not mouse and human RasGRP4 is also a likely phorbol ester receptor, RasGRP4 Ϫ , mRasGRP4 ϩ , and hRas-GRP4 ϩ fibroblasts were placed on top of 11-mm glass coverslips in 24-well culture dishes. The attached cells were then exposed to 10 nM phorbol 12-myristate 13-acetate (PMA) (Calbiochem) for 15-30 min at 37°C. After the coverslips were washed, they were incubated for 10 min in PBS supplemented with 4% paraformaldehyde. The fixative was removed, and the treated cells were immersed in Ϫ20°C methanol for 10 min, rinsed in PBS, and incubated in 5% normal horse serum in PBS for 30 min before the addition of mouse anti-V5 antibody (Invitrogen) and rabbit anti-actin antibody (Sigma). The latter antibody recognizes

FIG. 2. Evaluation of hRasGRP4 mRNA levels in varied fetal and adult tissues and in different populations of peripheral blood mononuclear cells.
Quantitative RNA blot (a) and semi-quantitative RT-PCR (b-d) approaches were used to evaluate hRasGRP4 mRNA levels in the indicated adult (a, b, and d) and fetal (c) human tissues and cells. The 303-bp probe that corresponds to residues 329 -631 in the hRasGRP4 cDNA was used in the more quantitative RNA blot analysis. The primers used in RT-PCR analyses correspond to sequences residing in exons 1 and 9 of the hRasGRP4 gene. The expected 900-bp product (arrow) is indicated. Because the hRasGRP4 transcript was abundant in the leukocyte preparation (a and b), another experiment was carried out in which pooled peripheral blood mononuclear cells from 4 to 36 individuals were sorted by CLONTECH based on their expression of CD4, CD8, and CD19 (d). Samples of the resulting cell populations were evaluated for their expression of hRasGRP4 mRNA immediately after their isolation (resting) or after their subsequent exposure to phytohemagglutinin, pokeweed mitogen, and/or concanavalin A (lectin-activated). hRasGRP4 mRNA was detected in unfractionated peripheral blood mononuclear cells but not in T and B lymphocytes. Glyceraldehyde-3-phosphate dehydrogenase-specific primers were used in a similar analysis to confirm the presence of intact RNA in all samples (data not shown).
the common C-terminal domain of the varied isoforms of actin.
Anti-V5 antibody was used to identify epitope-tagged RasGRP4. Cells were incubated with a 100 -500-fold dilution of each primary antibody for 2 h, washed several times in PBS, and then exposed for 1 h to the relevant secondary antibodies (a 200-fold dilution of Cy2-antimouse antibody and a 2,000-fold dilution of Cy3-anti-rabbit antibody; Jackson ImmunoResearch, West Grove, PA) in blocking solution supplemented with Hoechst dye 33258 at 50 ng/ml (Sigma). Stained cells were washed extensively with PBS and mounted. The resulting cells were viewed using a Nikon Eclipse 800 microscope; images were digitally captured using a CCD-SPOT RT digital camera and compiled using Adobe Photoshop software (version 5.5).
Guanine Nucleotide Exchange Assay-ProBond TM nickel-chelating resin (Invitrogen) was used to purify His 6 -tagged recombinant hRas-GRP4 from the transfectants. As recommended by the manufacturer, ϳ1 ϫ 10 7 transfectants were placed in 4 ml of 20 mM sodium phosphate, pH 7.4, buffer containing 500 mM sodium chloride and multiple protease inhibitors (Roche Diagnostics). Each cell suspension was lysed by two freeze-thaw cycles using liquid nitrogen and a 42°C water bath. Liberated nuclear DNA was sheared by passing the resulting preparation through an 18-gauge needle four times, and then the cellular debris was removed by a 5-min centrifugation step at 4°C and at ϳ14,000 ϫ g. The resulting hRasGRP4-enriched supernatant was incubated with nickel-charged agarose resin for 1 h at 4°C to ensure efficient binding of the His 6 -tagged recombinant protein. After the equilibration step, the resin was centrifuged in a "Spin" column at 800 ϫ g for 2 min. The non-bound material was discarded, and the column was washed extensively with 500 mM sodium chloride and 20 mM sodium phosphate, pH 6.0, to remove weakly associated protein. Five ml of 50 mM imidazole elution buffer was applied, and the resulting hRasGRP4-enriched eluate was concentrated to ϳ0.5 ml with a Centriplus-50 (Millipore, Bedford, MA) filtering device having a 50-kDa cut-off membrane.
A modification of the guanine nucleotide exchange assay described by Zheng et al. (31) was used to evaluate the predicted function of hRas-GRP4. In this assay, 2 g of recombinant human H-Ras (Oxford Biomedical Research, Oxford, MI) was placed in 60 l of 100 mM NaCl, 2 mM EDTA, 0.2 mM dithiothreitol, 20 mM Tris-HCl, pH 8.0, supplemented with 100 M AMP-PNP tetralithium salt (Roche Diagnostics) and 10 M GDP (Sigma). After a 5-min incubation at room temperature, MgCl 2 was added to achieve a final concentration of 5 mM. The solution was then incubated for another 15 min at room temperature to load H-Ras with non-radiolabeled GDP. Twenty l of the resulting solution was added to 75 l of reaction buffer (100 mM NaCl, 10 mM MgCl 2 , 20 mM Tris-HCl, pH 8.0, supplemented with 100 M AMP-PNP, 0.5 mg/ml bovine serum albumin, and 2 M guanosine 5Ј-␥- 35 S-triphosphate (ϳ11,000 cpm/pmol; Amersham Biosciences)), followed by 5 l of purified recombinant hRasGRP4 in the above buffer. The resulting samples were incubated at room temperature generally for 20 min. In a kinetic study with H-Ras, 15-l aliquots were removed every 5 min and placed in 4 ml of ice-cold termination buffer (100 mM NaCl, 10 mM MgCl 2 , and 20 mM Tris-HCl, pH 8.0). Non-bound radioactivity was removed by filtering each reaction mixture through a 25-mm cellulose nitrate membrane possessing 0.45-m pores (Whatman). After each filter was washed with 10 ml of ice-cold termination buffer, it was placed in 5 ml of Filtron-X scintillation fluid (National Diagnostics, Atlanta, GA), and the amount of bound radioactivity was quantitated using a Beckman LS5801 machine. For a negative control, purified recombinant hRas-GRP4 was boiled for 5 min and then sonicated for 5 min before the guanine nucleotide exchange activity of the denatured protein was assessed.
The ability of CaCl 2 (0.2-10 mM) and PMA (3-200 M) to inhibit the guanine nucleotide exchange activity of recombinant hRasGRP4 also was evaluated. In these competition assays, recombinant H-Ras was preloaded with GDP. CaCl 2 or PMA was then added, followed by hRas-GRP4 and radiolabeled GTP. The reaction mixtures were then incubated for 20 min at room temperature.
Comparative Protein Structure Modeling-Three-dimensional models of residues 34 -445 and 541-597 of normal and variant 2 hRasGRP4 were built using MODELLER (32,33). Residues 34 -445 contain the putative REM and CDC25-like catalytic domains and correspond to 61% of the translated protein. This portion of hRasGRP4 was modeled based on the crystallographic structure of hSos1 (residues 568 -1032 of GenBank TM accession number A37488) complexed to H-Ras (Protein Data Bank code 1BKD) (34). Residues 541-597 in hRasGRP4 correspond to its putative DAG-binding domain. This portion of the GEF was modeled based on the NMR structure of the Cys 2 domain in rat protein kinase C-␥ (Protein Data Bank code 1TBN) (35). The regions are ϳ20 and ϳ33% identical to the corresponding regions in Sos1 and protein kinase C-␥, respectively.
Importance of Normal Isoforms of RasGRP4 in the Differentiation and Granule Maturation of Human MCs-As noted under "Results," the hRasGRP4 gene was transcribed, but the resulting transcript was not processed correctly in HMC-1 cells. Because a functional form of the signaling protein was not expressed in HMC-1 cells, this MC leukemia cell line was used in an attempt to deduce the function of RasGRP4 in MCs. As also noted under "Results," we identified a number of single nucleotide polymorphisms in the hRasGRP4 transcript in the human population that, in turn, result in five amino differences in the translated products. Four of these amino acid polymorphisms are non-conservative changes. There is only one mRasGRP4 allele in the inbred BALB/c mouse. Because it was not obvious which hRasGRP4 allele should be used in our initial attempt to correct the developmental problem in the HMC-1 cell line, the BALB/c mouse-derived RasGRP4 described under "Results" was used in the transfection experiments. Recombinant RasGRP4 possesses the V5 epitope at its C terminus. Thus, anti-V5 antibody was used to confirm that a functional, nontruncated version of the GEF was expressed in the transfectants. Immunohistochemistry and/or SDS-PAGE/immunoblot approaches were then used with anti-tryptase and anti-chymase antibodies (Chemicon International, Temecula, CA) to evaluate the levels of these neutral proteases in the two populations of cells. A pancreatic carboxypeptidase A (CPA)-derived antibody (Sigma) that also recognizes MC CPA was used to evaluate the levels of this exopeptidase. Finally, standard electron microscopic methodologies were carried out to evaluate the ultrastructure of the HMC-1 cell line before and after transfection.

RESULTS
Cloning of the mRasGRP4 Transcript and Gene-Sequence analysis of Ͼ2000 clones arbitrarily selected from a BALB/c mBMMC cDNA library resulted in the identification of a clone that contained a 2.1-kb insert whose nucleotide sequence did not match any sequence in GenBank TM genome and expressed sequence tag data bases. A search of Celera's mouse nucleotide data bases with the resulting product also failed to reveal any intact, completely sequenced gene or transcript that corresponded to the mBMMC-derived cDNA. The fact that the unknown transcript had not been identified in another cell type, coupled with the finding that it was a low abundant transcript in mBMMCs, raised the possibility that it encoded a novel MC-restricted regulatory protein. Thus, a 5Ј-RACE approach was carried out to deduce the nucleotide sequence of the missing 227-bp portion of the full-length ϳ2.3-kb cDNA (Fig. 1). Another closely related cDNA was subsequently isolated in this screening approach that was missing 15 nucleotides. Analysis of the exon/intron organization of the mRasGRP4 gene revealed that the alternate mRasGRP4 transcript was caused by differentiation splicing of the precursor transcript. The Gen-Bank TM accession numbers for the two isoforms of mRasGRP4 cloned in this study are AF331457 and AY040628. Preliminary analysis of its gene (see Fig. 1 legend) revealed that the coding portion of the mRasGRP4 transcript is derived from 17 exons. The similarity of the exon/intron organizations of the mRas-GRP4 and hRasGRP2 genes supported the nucleotide sequence data that initially suggested mRasGRP4 is a new member of the RasGRP family of GEFs.
Cloning of the hRasGRP4 Transcript, Evaluation of its Expression, Chromosomal Location of its Gene, and Identification of Single Nucleotide Polymorphisms of Its Gene-An extensive search of GenBank TM and Celera's nucleotide and protein data bases failed to reveal any complete human cDNA, gene, or protein that closely resembles that of the mRasGRP4 transcript, gene, and/or protein. Nevertheless, the 11.5-kb nucleotide sequence residing at one end of the human chromosome 19q13.1-derived BAC clone AC011469 closely resembled that of the 5Ј-half (i.e. exons 1-9 and introns 1-8) of the mRasGRP4 gene. Although the nucleotide sequence of an overlapping BAC genomic clone corresponding to the missing 3Ј-half of the hRas-GRP4 gene had not been deposited in GenBank TM , these pre-liminary data raised the possibility that a functional hRas-GRP4 gene exists in the human genome. Based on the chromosomal assignment of BAC clone AC011469, we also concluded that this new gene probably resides ϳ8 kb downstream of the ryanodine receptor 1 (RYR1) gene (GenBank TM Lo-cusLink accession number 6261) within a small region of human chromosome 19q13.1 that had not been sequenced by the Human Genome Project before the release of our nucleotide sequence data (e.g. GenBank TM accession number AY048119) to the public on August 19, 2001. Primers corresponding to the nucleotide sequences residing in exons 1 and 9 of the suspected hRasGRP4 gene were used in RT-PCR (Fig. 2, b-d) approaches to determine whether or not this putative new human gene is transcribed in vivo. Because the expected transcript was detected in nearly every examined human fetal tissue, the novel human gene we identified on chromosome 19q13.1 is transcribed in vivo. In addition, it is transcribed relatively early in human development. The level of the hRasGRP4 transcript was more abundant in fetal lung than in heart, brain, liver, kidney, spleen, thymus, or muscle. Although the expected 900-bp PCR product was detected in cDNA libraries generated from a number of adult human tissues, the amount of hRasGRP4 mRNA in these tissue samples was considerably less than that in fetal lung (Fig. 2c).
Different sized RT-PCR fragments were occasionally seen in the tissue and blood samples pooled from many individuals. As noted below, sequence analysis of the corresponding transcripts in the mononuclear cells of a systemic mastocytosis patient and an asthma patient revealed that these larger and smaller RT-PCR products are the result of differential splicing of the precursor transcript. The levels of hRasGRP4 mRNA were below detection in resting or lectin-activated CD19 ϩ B lymphocytes, CD4 ϩ T lymphocytes, or CD8 ϩ T lymphocytes (Fig. 2d). The levels of hRasGRP4 mRNA also were below detection in lectin-treated mononuclear cells. Nevertheless, the steady-state levels of the hRasGRP4 transcript were sufficiently high enough in the non-lectin-treated mononuclear leukocytes to detect the transcript by routine blot analysis (Fig. 2a).
By using pooled mRNA obtained in varied 3Ј-and 5Ј-RACE approaches, eventually we were able to deduce the nucleotide sequence of the primary full-length hRasGRP4 transcript in the mononuclear leukocytes of most normal individuals (Fig.   3). The GenBank TM accession number for the major hRasGRP4 cDNA present in the human population is AY048119. Sequence analysis of multiple subcloned hRasGRP4 cDNAs revealed 10 single nucleotide differences in the eight analyzed clones that ultimately resulted in five amino acid differences in the translated products. It is presumed that these nucleotide differences are the result of minor allelic polymorphisms of a single hRas-GRP4 gene in the human genome rather than distinct genes that are Ͼ99% identical. The primary hRasGRP4 transcript is ϳ300 bp larger than the primary mRasGRP4 transcript due to a somewhat longer 3Ј-untranslated region. However, the rest of the hRasGRP4 transcript is Ͼ80% identical to that of the mRasGRP4 transcript.
No genomic fragment was detected in the GenBank TM ge-nome data base that corresponded to the 3Ј end of the hRas-GRP4 cDNA. Probing of Celera's human genome data base with the full-length hRasGRP4 cDNA also failed to reveal an Evaluation of the Structure of RasGRP4 Protein-As assessed immunohistochemically using antibodies directed against the predicted N terminus of the translated product, the hRasGRP4 transcript was selectively converted into protein in the tryptase ϩ MCs that reside in varied normal human tissues (Fig. 4). Every tryptase ϩ MC in the interlobular connective tissue of human breast and the mucosa layer of human stomach contained hRasGRP4 protein. Similar immunohistochemical data were obtained in the mouse with a different antibody. Thus, RasGRP4 is selectively expressed in mature MCs and their progenitors in humans and mice.
The nucleotide sequences of the isolated cDNAs indicate that mouse and human RasGRP4 exist as 673-678-residue, ϳ75-kDa proteins. Mouse and human RasGRP4 have a Ͻ50% amino acid sequence identity with mouse, rat, and human RasGRP1, RasGRP2, and RasGRP3 (Fig. 5). As found for other cytosolic proteins, RasGRP4 lacks a hydrophobic signal peptide at its N terminus. A Kyte-Doolittle hydropathy analysis also failed to FIG. 6. Three-dimensional models of domains in hRasGRP4. a, schematic representation of the three-dimensional model of residues 34 -445 of hRasGRP4 (gray) bound to H-Ras (blue). Shown in ball and stick (green) representation are the two residues (Val 359 and Glu 363 ) in hRasGRP4 corresponding to the residues in hSos-1 that are absolutely essential for its interaction with H-Ras. The 14-residue peptide that is lost in the variant 2 transcript isolated from the asthma patient is highlighted (red). Four allelic differences in hRasGRP4 have been noted at residues 120, 145, 261, and 335 (yellow). The fifth allelic difference occurs outside of the modeled region at residue 18. b, schematic representation of the three-dimensional model of the DAG-binding domain of hRasGRP4 (residues 537-590). The conserved His and Cys residues in DAG-binding proteins are indicated (blue). All representations were rendered using Molscript (68).

FIG. 7. Generation of recombinant hRasGRP4 in COS-7 cells and fibroblasts
, and evaluation of its guanine nucleotide exchange activity. a, purified native (f) and heat-denatured (q) recombinant hRasGRP4 were evaluated for their ability to transfer radiolabeled GTP to GDP-loaded H-Ras in a kinetic manner at room temperature. The level of immunoreactive hRasGRP4 in the conditioned media of the expressing cells always was below detection (data not shown). The inset shows the SDS-PAGE/immunoblot analysis of lysates of fibroblasts transfected with the hRasGRP4 construct; the depicted blot was probed with the anti-V5 antibody that recognizes the C-terminal epitope tag. Molecular weight markers are shown on the left. Recombinant hRasGRP4 is ϳ10 kDa larger than the native protein due to the epitope tag. b, the ability of purified recombinant hRasGRP4 to transfer radiolabeled GTP to GDP-loaded recombinant H-Ras was then compared after a 20-min incubation at room temperature in the presence of the indicated amounts of CaCl 2 .
reveal an extended hydrophobic domain in the primary amino acid sequence of the protein. Immunoreactive mRasGRP4 and hRasGRP4 were not constitutively secreted from the COS-7 cell and fibroblast transfectants, and immunoreactive mRasGRP4 was not detected in extracellular matrices adjacent to tissue MCs. Thus, RasGRP4 is an intracellular protein.
A dendrogram comparison of the nine most related mouse proteins in this superfamily revealed that mRasGRP4 is slightly more similar to mRasGRP2 than to mRasGRP1 (Fig.  5a). The amino acid sequence of mRasGRP3 has not yet been deposited in GenBank TM . Nevertheless, mRasGRP4 is only distantly related to hRasGRP3. mRasGRP4 also lacks the extended C-terminal domains present in mRasGRP1, rRasGRP1, hRasGRP1, and hRasGRP3 (Fig. 5b). The putative REM, CDC25-like catalytic, Ca 2ϩ -binding EF hands, and phorbol ester/DAG-binding domains of the varied RasGRPs are highlighted in Fig. 5b. All of these domains are present in mouse and human RasGRP4 at the expected locations in its primary amino acid sequence. The most conserved region is the CDC25like catalytic domain. For example, residues 405-433 in the CDC25-like catalytic domains of mRasGRP4 and mRasGRP1 differ only in two amino acid residues, and even these differences are minimal (i.e. Leu 3 Val and Phe 3 Tyr). The amino acid sequence of hRasGRP4 is ϳ85% identical to that of mRas-GRP4 (Fig. 5c). Except for the C-terminal 21 amino acids where the degree of sequence identity drops to 52%, the high degree of conservation extends throughout the entire length of the protein.
Residues 34 -445 of hRasGRP4 (Fig. 6a) are predicted to resemble the two ␣-helical domains of hSos-1 with 6 and 11 helices in the first and second domains, respectively. Mimicking hSos-1, the eighth ␣-helix in the CDC25-like catalytic domain of hRasGRP4 is predicted to interact with H-Ras. The residues involved in H-Ras interaction are generally conserved between hRasGRP4 and hSos1. The three-dimensional model of residues 537-590 of hRasGRP4 (Fig. 6b) also closely resembles that of the putative DAG-binding domain of protein kinase C␥ despite the poor degree of primary amino acid sequence identity. The minor allelic differences identified so far in hRas-GRP4 are not expected to grossly alter its three-dimensional structure, nor are they likely to affect its interaction with H-Ras or DAG greatly.
Guanine Nucleotide Exchange Activity of Recombinant Ras-GRP4 and Increased Sensitivity of RasGRP4-expressing Fibroblasts to PMA-Bioengineered forms of mouse and human Ras-GRP4 were expressed in COS-7 cells and fibroblasts that contained the V5 and His 6 peptides at their C termini. The latter epitope tag was added so that each recombinant protein could be identified and purified. Although substantial amounts of immunoreactive RasGRP4 were always recovered in the soluble cytosolic portion of the lysates of the transfectants, a portion of the recombinant protein was consistently recovered in the microsomal fractions. Thus, RasGRP4 does not reside in a single intracellular compartment in either transfected cell type.
As the three-dimensional model of residues 34 -445 predicted, purified recombinant hRasGRP4 was able to transfer ␥-35 S-GTP to GDP-loaded H-Ras in a catalytic manner (Fig.  7a). The first EF hand in rRasGRP1 functions as the primary Ca 2ϩ -binding site of this protein (14), and the critical residues that form the "regulatory EF hand" in rRasGRP1 and other Ca 2ϩ -binding proteins (36) are present in mouse and human RasGRP4 at the expected locations. Whether or not Ca 2ϩ binds specifically to this domain remains to be determined using a site-directed mutagenesis approach. Nevertheless, 1 mM Ca 2ϩ was sufficient to inhibit significantly the ability of recombinant hRasGRP4 to transfer GTP to GDP-loaded H-Ras even if 5 mM Mg 2ϩ was present in the reaction buffer (Fig. 7b).
The phorbol ester/DAG-binding, C1 domain in protein kinase C is ϳ50 residues in length and possesses the motif of HX 12 CX 2 CX 13/14 CX 2 CX 4 HX 2 CX 7 C, where H is His, C is Cys, and X is any other amino acid. Because these residues are present in mouse and human RasGRP4 and because the threedimensional model predicts that residues 537-590 resemble a C1-like domain, the possibility that RasGRP4 is a phorbol ester/DAG receptor also was tested experimentally. As noted in Fig. 8, RasGRP4-expressing fibroblasts underwent dramatic morphologic changes when exposed to low levels of PMA for only 15 min.  level (a, e, and i), all populations of fibroblasts are extremely adherent to plastic culture dishes before PMA treatment and form prominent focal adhesions and extended membrane projections. RasGRP4 Ϫ fibroblasts do not undergo noticeable morphologic changes when exposed to low levels of PMA for a brief period (c). However, many of the mRasGRP4 ϩ fibroblasts (arrows in g) and hRasGRP4 ϩ fibroblasts (arrows in j) quickly round up when similarly treated. As assessed immunohistochemically, actin (red) redistributes in those mRasGRP4 ϩ fibroblasts that have been exposed to PMA (h). DNA stains blue in these immunohistochemical assays. Approximately 80% of the fibroblasts shown in panel g express mRasGRP4 as assessed immunohistochemically with anti-V5 antibody (data not shown). Similar findings were obtained in a second experiment with mRasGRP4.

Isolation of Abnormal hRasGRP4 Transcripts in an Asthma
domains of the forms of hRasGRP4 that are expressed in the MC-committed progenitors residing in the peripheral blood of an asthma patient. Five of the eight arbitrarily subcloned cDNAs from this patient corresponded, with minor differences, to the normal hRasGRP4 cDNA depicted in Fig. 3. However, two of the cDNAs (designated variant 1) were 117 nucleotides FIG. 9. Isolation of two aberrant hRasGRP4 cDNAs from an asthma patient and a mastocytosis patient that encode truncated proteins. a, aberrant hRasGRP4 cDNA (designated variant 1) was isolated from an asthma patient and a mastocytosis patient that contained a 117-bp insertion. Sequence analysis of the cDNA and the hRasGRP4 gene revealed variant 1 was caused by a failure of the hRasGRP4-expressing cell to remove intron 5 (shaded region) in the precursor transcript. The 117-bp insertion (shaded sequence) causes an early translation-termination codon (*). One of the two sequenced variant 1 transcripts isolated from the asthma patient possessed an additional problem at the intron 7/exon 8 splice site that resulted in the loss of the indicated 2 nucleotides (dashes at residue 1168) in the newly formed 3Ј-untranslated region. b, a different aberrant hRasGRP4 cDNA (designated variant 2) was isolated from the same asthma patient. Variant 2 encodes a truncated form of hRasGRP4 that lacks the Leu-Ser-Pro-Gly-Gly-Pro-Gly-Pro-Pro-Leu-Pro-Met-Ser-Ser sequence that precedes the CDC25-like catalytic domain in the normal protein. Analysis of the hRas-GRP4 gene revealed that variant 2 is caused by a failure of the hRasGRP4expressing cell to use the normal intron 5/exon 6 splice site to remove intron 5 in the precursor transcript. The use of a cryptic splice site in the middle of exon 6 to eliminate the intron and a portion of exon 6 results in the production of a truncated hRasGRP4. The area of the transcript and protein that is affected by this post-transcriptional splicing event is indicated (#). The minor nucleotide differences noted in the variant 1 and 2 cDNAs relative to the hRasGRP4 cDNA shown in Fig. 3 are in boldface. The 5Јuntranslated regions of the variant 1 and 2 cDNAs were not deduced. Nevertheless, it is assumed that the initial 215 nucleotides in these transcripts correspond to those in the normal hRasGRP4 transcript. c, RT-PCR analyses were carried out using different primer sets to evaluate the expression of the variant 1 (top panel) and variant 2 (bottom panel) forms of the hRasGRP4 transcript in a pooled leukocyte preparation derived from 550 individuals and in the MC progenitors residing in the blood of four normal individuals, a patient with asthma, and a patient with systemic mastocytosis. d, shown is a schematic representation of the splicing events that result in the generation of the normal hRasGRP4 transcript and its two abnormal variants in the asthma patient. larger in size. Sequence analysis revealed that variant 1 was created by a failure of the hRasGRP4-expressing cell to remove intron 5 from the precursor transcript (Fig. 9, a and d). The failure to remove this single intron results in the formation of a premature translation-termination codon in the expressed transcript. Assuming the normal translation-initiation site is used, the resulting 170-residue protein will not be able to activate H-Ras because it lacks ϳ75% of its primary amino acid sequence, including the entire CDC25-like catalytic domain.
One of the eight arbitrarily subcloned cDNAs from the asthma patient possessed a 42-residue deletion (Fig. 9, b and d) rather than a 117-bp insertion. This isoform (designated variant 2) also was caused by a failure of the hRasGRP4-expressing cell in this patient to properly remove intron 5. Because the normal intron 5/exon 6 splice site was not used in the maturation of the precursor human transcript, a cryptic splice site residing 42 nucleotides within exon 6 was employed to remove intron 5 during the maturation of the variant 2 transcript. The RasGRP4 open reading frame of the truncated variant 2 hRasGRP4 transcript remains intact. However, the resulting protein lacks the 14-mer sequence that links the two major helical domains within the N-terminal segment (Fig. 6a). The GenBank TM accession numbers for the variant 1 and 2 hRasGRP4 transcripts are AY048120 and AY048121, respectively.
An RT-PCR approach was next used to evaluate the prevalence of both abnormal hRasGRP4 transcripts in the general population and in an additional mastocytosis patient. As noted in Fig. 9c using different primer sets, the levels of the variant 1 and 2 transcripts were below detection in the MC progenitors residing in the blood of four normal individuals. Nevertheless, aberrant hRasGRP4 transcripts were detected in the pooled leukocyte preparation derived from 550 individuals. Variant 1 mRNA was found in the mastocytosis patient, but the level of the variant 2 transcript was below detection.
None of the eight hRasGRP4 cDNAs isolated and sequenced from the asthma patient and none of the hRasGRP4 cDNAs isolated and sequenced from the leukocyte preparation generated from multiple individuals lacked an entire coding exon. These preliminary data suggest that the loss of an entire coding exon in the hRasGRP4 transcript is a rare event, if it ever occurs, in the MC-committed progenitors circulating in the blood of normal individuals. Nevertheless, removal of any combination of exons 7-16 (except exons 14 and 15) will result in a truncated transcript that remains in the correct reading frame relative to the normal translation-termination codon. Thus, further studies are needed to address whether or not differential exon splicing of the precursor hRasGRP4 transcript occurs in the mature MCs that develop in different tissue sites of normal individuals.

Evaluation of hRasGRP4 Expression and Function in the HMC-1 Cell Line-
The immature HMC-1 cell line was derived by Butterfield et al. (27) in 1988 from a patient with a MC leukemia. The identification of defective variants of hRasGRP4 mRNA in a mastocytosis patient (Fig. 9c) raised the possibility that the HMC-1 cell line might also express abnormal variants of hRasGRP4. Thus, hRasGRP4 expression at the mRNA level was next evaluated in this cell line. By using various primers sets, we discovered that the hRasGRP4 gene is transcribed in HMC-1 cells (Fig. 10). However, only abnormal variants of hRasGRP4 are expressed in this transformed cell line. As noted in Fig. 10, a new isoform of hRasGRP4 (designated variant 3) that is closely related to the variant 1 isoform was isolated from HMC-1 cells. Sequence analysis of this RT-PCR product revealed that the larger sized transcript was caused by a failure to remove intron 5 and additionally intron 3 in the precursor transcript. The failure to remove these two introns results in a translation-termination codon that occurs earlier in variant 3 than that in variant 1. Thus, even if translated, variant 3 also would be unable to activate any Ras family member. A premature translation-termination codon at a similar location in a tryptase transcript results in its rapid degradation in C57BL/6 mouse mBMMCs by the nonsense-mediated pathway (37). If the defective hRasGRP4 transcript is being catabolized nearly as fast as it is being generated as we suspect, this would account for its lower levels in the HMC-1 cell line.
HMC-1 cells are poorly granulated, blast-like leukemia cells (Fig. 11i) that fail to express detectable amounts of MC chymase (Fig. 11e) or CPA (Fig. 11g) protein. Because HMC-1 cells only express small amounts of ␤-tryptase (Fig. 11, b and c), this cell line became an attractive MC-committed progenitor to  RasGRP4 begin to address the function of RasGRP4 in a more natural setting than occurs in a transfected fibroblast line. Thus, using an expression/transfection approach, HMC-1 cells were induced to express a normal, biologically active form of mRas-GRP4 (Fig. 11a). As noted in Fig. 11, the resulting transfectants underwent dramatic morphologic changes (Fig. 11j) and increased their levels of tryptase substantially (Fig. 11, b and  d). The transfectants also began expressing MC chymase (Fig.  11f) and CPA (Fig. 11h). DISCUSSION We describe a new mouse and human cation-dependent, GEF/phorbol ester receptor (designated RasGRP4) that is selectively expressed in MCs and their progenitors. The hRas-GRP4 gene is Ͼ15 kb in size, consists of 18 exons, and resides on chromosome 19q13.1. Because the same four RasGRP genes are present in the mouse and human genomes, the varied members of its family apparently evolved Ͼ100 million years ago before the divergence of mice and humans. The hRasGRP1, hRasGRP2, and hRasGRP3 genes reside on chromosomes 15, 15, and 2, respectively (17,20). It now appears that an ancestral RasGRP-like gene duplicated twice, and the resulting three genes translocated to distinct chromosomes. The new gene that segregated to the genomic fragment that eventually became human chromosome 19 failed to duplicate again and became RasGRP4 in both species. A similar situation occurred for the RasGRP3 gene that eventually developed on human chromosome 2. However, the ancestral RasGRP-like gene that segregated to the genomic fragment that eventually became human chromosome 15 duplicated one more time to become the RasGRP1 and RasGRP2 genes. The resulting four genes then underwent substantial amino acid divergence to create the final RasGRP1, RasGRP2, RasGRP3, and RasGRP4 genes in both species. All four genes were then maintained throughout the evolution of mice and humans. The biological significance of these evolutionary events is that most RasGRP data obtained in mice should be relevant to humans. In addition, mRasGRP4 should be active in human MCs and hRasGRP4 should be active in mouse MCs.
As assessed by RNA blot (Fig. 2a), RT-PCR (Fig. 2, b-d), and immunohistochemical ( Fig. 4) analyses, RasGRP4 is an ϳ75-kDa intracellular protein that is selectively expressed in mature MCs and their progenitors. When the four RasGRPs were compared (Fig. 5), the least conserved regions were found to be the N and C termini. The N terminus of hRasGRP2 is myristoylated and palmitoylated at Gly 2 and Cys 7 , respectively (19).
Although RasGRP4 has a conserved Cys at residue 14, residues 2 and 7 in this GEF are Asn and Lys, respectively (Figs. 1, 3, and 5). Not only does the N terminus of RasGRP4 fail to resemble that of RasGRP2, its amino acid sequence also does not resemble that found in other palmitoylated and myristoylated G proteins (38,39). Thus, RasGRP4 and RasGRP2 appear to differ in at least one important structural feature.
The Saccharomyces cerevisiae cell division cycle 25 (CDC25) protein is required for the function of adenylate cyclase in yeast, and CDC25 promotes the exchange of guanine nucleotides bound to Ras (40,41). Yeast replication is dependent on As assessed immunohistochemically (c-h) and by SDS-PAGE-immunoblot analysis (b), the mRasGRP4-expressing transfectants (؉) contained substantially more tryptase (b-d), chymase (e and f), and CPA (g and h) in their granules than the non-transfectants (Ϫ). Although granules are rarely found in HMC-1 cells (i), many granules that contain electron dense material are found in the transfectants (j). Shown in the inset in the lower right j is a higher magnification of two typical granules in the transfectants. Although the transfectant depicted in the EM shown in j is more typical of the cells in the mRas-GRP4-expressing cultures, ϳ10% of the transfectants contain granules that are nearly completely filled with electron dense material (data not shown).
CDC25, and inactivation of this GEF results in cell cycle arrest in G 1 . The most conserved region within RasGRP1 and Ras-GRP4 corresponds to the catalytic domain of CDC25 (Fig. 5). A comparative three-dimensional model of residues 34 -445 of hRasGRP4 (Fig. 6a) predicted that the overall structure of this region resembles that of hSos1 even though the sequence identity is only ϳ20%. The key residues in hSos1 needed to interact with H-Ras are all present in RasGRP4 when it is folded. These data therefore suggested that RasGRP4 probably functions in MCs as a GEF. As noted in Fig. 7, recombinant RasGRP4 is indeed able to activate H-Ras in vitro. MCs express varied members of the Ras superfamily of GTP-binding proteins (12,26,(42)(43)(44)(45)(46)(47). However, because recombinant hRasGRP3 is able to transfer GTP to GDP-loaded H-Ras, R-Ras, and Rap1 in vitro (18), it remains to be determined which Ras family members RasGRP4 prefers to interact with inside a living MC.
Immunoelectron microscopy revealed that a substantial portion of the RasGRP4 present in the splenic mouse MCs resides on the cytosolic side of the plasma membrane of the cell. Although this finding strongly implies that RasGRP4 participates in early signaling events at the plasma membrane of the MC, immunoreactive mRasGRP4 also was found in the cytoplasm. Subcellular fractionation studies of the mRasGRP4 and hRasGRP4 transfectants confirmed these ultrastructural findings. Based on these data, some unknown intracellular factor or post-translational modification event must regulate the movement of RasGRP4 from the cytoplasm to the inner leaflet of the plasma membrane of the MC. Activation of MCs via Fc⑀RI or c-kit results in the rapid generation of DAG, and the generation of this lipid has been linked to the morphological changes that occur in MCs. Mouse, rat, and human RasGRP1, RasGRP2, and RasGRP3 have C1-like domains, and recombinant hRasGRP3 expressed in E. coli can bind 12-[ 3 H]phorbol 13-dibutyrate efficiently in the presence of phospholipids (20). We are unaware of any group that has expressed a truncated RasGRP lacking its C1 domain to map precisely the location of its phorbol ester/DAG-binding site. Nevertheless, the data reported in numerous studies (19,20,48,49) have led to the conclusion that DAG is required for the efficient movement of the varied RasGRPs from their cytosolic compartment to the plasma membrane.
The primary consequence of increased levels of DAG in MCs is thought to be activation of varied protein kinase C isoforms (50). However, the facts that RasGRP4 contains a potential DAG-binding domain (Fig. 6b) that is relatively conserved (Fig.  5) and that PMA treatment of RasGRP4-expressing fibroblasts results in dramatic morphologic changes (Fig. 8) now suggest that DAG and varied phorbol esters also regulate the movement of RasGRP4 to the plasma membrane and that this translocation process ultimately contributes to the membrane ruffling and spreading seen in activated MCs. Although the putative DAG-binding domain in RasGRP4 appears to be functionally important in the context of a living cell (Fig. 8), PMA does not enhance or suppress the guanine nucleotide exchange activity of the major form of RasGRP4 expressed in MCs, at least in one in vitro assay. A similar finding has been reported for recombinant hRasGRP2 (19).
The specific receptor-mediated signaling pathway(s) in MCs that depends on RasGRP4 remains to be determined experimentally. All mature MCs appear to express Fc⑀RI, c-kit, and RasGRP4. Although RasGRP4 could participate in Fc⑀RI-mediated signaling pathways, our data suggest an important role for our intracellular protein in c-kit-mediated signaling. c-kit is the only cytokine and adherence receptor that is absolutely essential for the development of mature MCs (51, 52), and MCs undergo substantial morphological changes when they bind to c-kit ligand ϩ mesenchymal cells (53). The finding that Ras-GRP4-expressing fibroblasts undergo dramatic morphologic changes when exposed to PMA (Fig. 8) now suggests that RasGRP4 helps regulate the morphological changes that occur when MCs are activated via c-kit. W/W v mice are MC-deficient (51) because of a genetic abnormality in c-kit (54). This mouse strain contains normal numbers of MC-committed progenitors in its bone marrow, and these progenitors readily target to connective tissue sites. However, the committed progenitors cannot develop into mature MCs because of the genetic defect in the tyrosine kinase domain of c-kit that resides inside MCs and participates in signal transduction pathways. Interestingly, Gordon and Galli (55) noted that substantial numbers of mature MCs develop in the skin of W/W v mice following exposure to PMA. This finding suggested that an undefined PMAdependent signaling protein acts downstream of c-kit in mouse MC development.
Human asthma is a polygenic disorder that is additionally influenced by environmental factors. It is generally accepted that pulmonary MCs play an important role in the initiation and/or progression of this disease. Thus, an intense effort has been made during the last decade to understand the varied signaling pathways that regulate the development and function of pulmonary MCs. We discovered that the hRasGRP4 gene resides at chromosome 19q13.1. Interestingly, gene-linkage studies carried out by others have revealed a human asthma susceptibility locus (p ϭ 0.0013) at chromosome 19q13.1 in Caucasians (56). The fact that many cytokine precursor transcripts undergo alternative splicing to produce receptor antagonists rather than receptor activators (57-60) documents the importance of post-transcriptional mechanisms in allergic asthma. Thus, we looked for the expression of altered forms of hRasGRP4 that would be caused by differential splicing of the precursor transcript. Although some of the single amino acid polymorphisms noted in Figs. 3 and 5c could result in forms of the GEF that differ slightly in their ability to activate H-Ras, we searched for the expression of hRasGRP4 transcripts in an asthma patient that encodes more severely altered proteins. Sequence analysis of eight arbitrarily subcloned hRasGRP4 cDNAs from an asthma patient revealed the presence of two transcripts that contained a 117-nucleotide insertion due to a failure of the hRasGRP4-expressing MC progenitor to remove intron 5 from the precursor transcript ( Fig. 9, a and d). No point mutation was noted at the exon 5/intron 5 or intron 5/exon 6 boundaries of the gene. Thus, the variant 1 transcript appears to be caused by an unprecedented post-transcriptional mechanism. Although the mechanism that hinders removal of intron 5 in this patient remains to be determined, the functional consequences of the post-transcriptional event are clear. The aberrant splicing event results in the expression of a 170-residue, non-functional protein.
Another abnormal hRasGRP4 cDNA was identified in the asthma patient that also was caused by a failure to properly remove intron 5. In contrast to the variant 1 transcript, the variant 2 transcript possessed a 42-nucleotide deletion (Fig. 9,  b and d). When translated, variant 2 should encode a 659residue protein that lacks the Pro-rich, 14-mer sequence immediately preceding the CDC25-like catalytic domain. The three-dimensional model (Fig. 6a) predicts that the deleted sequence is an extended loop in the CDC25-like catalytic domain opposite the face that interacts with H-Ras, linking one helical domain with another. The loss of this "rigid" Pro-rich, 14-mer sequence undoubtedly will affect the interaction of the two helical domains within the N-terminal segment of hRas-GRP4. Because the first helical region is postulated to stabilize the second region that interacts with H-Ras (34), disruption of the structure of the loop linking the two domains should adversely affect the ability of hRasGRP4 to transfer GTP to H-Ras. As noted in Fig. 9c, these two aberrantly spliced transcripts were not found in the MC progenitors in the blood of four normal individuals. The mouse counterparts of these two abnormal forms of hRasGRP4 also were not found in BALB/c mBMMCs. Nevertheless, preliminary RT-PCR analysis of 12 other asthma patients revealed that the expression of the nonfunctional, variant 1 form of hRasGRP4 is a common occurrence in this patient group. 2 Surprisingly, variant 1 is more prevalent in the human population than variant 2 (Figs. 9c and 10) even though the former is the more severely altered form of hRasGRP4. Although it is possible that the expression of nonfunctional forms of hRasGRP4 in multiple asthma patients is coincidental, the gene-linkage studies are suggestive of an adverse role for hRasGRP4 in the development of asthma in some patients.
The inbred BXH2 and AKXD13 mouse strains are particularly susceptible to leukemia viruses, and Li et al. (61) noted that spontaneous retroviral insertion into the RasGRP1 gene in these strains often results in myeloid, B cell, and T cell leukemia. The hRasGRP4 gene resides at chromosome 19q13.1. As noted at the "Mitelman Data base of Chromosome Aberrations in Cancer" at the NCI web site (cgap.nci.nih.gov/ Chromosome/Mitelman), breakpoint alterations at chromosome 19q13.1 often lead to leukemia. Systemic mastocytosis and MC leukemia are heterogeneous disorders that result in the production of excessive numbers of MCs. The genetic abnormality that occurs in most systemic mastocytosis patients has not been deduced. Nevertheless, many mastocytosis patients possess a mutation in the intracellular domain of c-kit that causes their tissue MCs to be in a heightened state of activation (62,63). Interestingly, this gain-in-function mutation was first described in the HMC-1 cell line (64) established in 1988 from a patient with a MC leukemia (27). The identification of transcripts that encode substantially altered forms of hRasGRP4 in the HMC-1 cell line (Fig. 10) and in a patient with systemic mastocytosis (Fig. 9) could be another coincidence. However, we recently identified the same defect in two other systemic mastocytosis patients. 3 Interestingly, one of these patients only produces variant 1. The cumulative data therefore raise the additional possibility that the expression and intracellular accumulation of substantial amounts of aberrant forms of hRasGRP4 (particularly variants 1 and 3) in human MCs and their progenitors is a contributing factor in the development of systemic mastocytosis and/or MC leukemia.
Its precise role, if any, in systemic mastocytosis and MC leukemia remain to be determined experimentally. Nevertheless, the fact that HMC-1 cells do not express a normal, biologically active form of hRasGRP4 (Fig. 10) implies that this GEF is not essential for the early stages of MC commitment. Despite this finding, we were intrigued by the fact that HMC-1 cells are so immature that they do not even express the high affinity IgE receptor on their surfaces (27). Unlike cutaneous MCs, HMC-1 cells also lack electron dense granules and do not contain detectable amounts of MC chymase or CPA protein (Fig. 11, e, g, and i). Although HMC-1 cells expresses a ␤-tryptase (65), the amount of this neutral protease in the secretory granules of the cell is extremely low (Fig. 11, b and c) relative to a normal tissue MC (66) and the immature MC populations that we and others routinely generate in our varied culture systems.
Mature MCs are rarely seen in the tissues of W/W v and Sl/Sl d mice that possess genetic defects in c-kit and its ligand, respectively. Thus, c-kit-dependent signaling events are essential for the final stages of MC development. Despite the fact that HMC-1 cells possess a gain-in-function mutation in c-kit, these cells exhibit a blast-like appearance. To explain this inconsistency, we concluded that HMC-1 cells probably have a defect downstream of c-kit that prevents these cells from differentiating and maturing. Because HMC-1 cells express only abnormal forms of hRasGRP4, we concluded that this cell line might be useful for deducing the function of RasGRP4 in MCs. When HMC-1 cells were induced to express a normal form of mRasGRP4, the resulting transfectants contained many electron dense granules (Fig. 11j). Because the transfectants began expressing MC chymase (Fig. 11f) and CPA (Fig. 11h) and also increased their granule tryptase content dramatically (Fig. 11,  b and d), RasGRP4 appears to play a very important role in the final stages of MC differentiation and maturation.