MEK and ERK Activation in Ras-Disabled RBL-2H3 Mast Cells and Novel Roles for Geranylgeranylated and Farnesylated Proteins in FcεRI-Mediated Signaling

In rat basophilic leukemia cells (RBL-2H3), cross-linking the high affinity receptor for IgE (FcεRI) activates the receptor-associated tyrosine kinases, Lyn and Syk, as well as Bruton’s tyrosine kinase (1, 2, 3), and causes the tyrosine phosphorylation of multiple substrates, including immunoreceptor tyrosine-based activation motifs in the β and γ subunits of the heterotrimeric (αβγ₂) FcεRI itself (4); two phospholipase Cγ (PLCγ)⁴ isoforms, PLCγ1 and PLCγ2 (5); the guanine nucleotide exchange factor Vav (6); phosphatidylinositol 3-kinase (7, 8); the adaptor protein Grb2 (9); and others. Protein-tyrosine phosphorylation in turn activates a signaling cascade leading to inositol-1,4,5-trisphosphate (Ins(1, 4, 5)P₃) production (10, 11), Ca²⁺ mobilization (12, 13), Ras activation (14), and the activation of the ERK and JNK MAP kinases (8, 15, 16, 17). These biochemical and ionic responses lead to functional responses, including secretion, actin polymerization, membrane ruffling, the assembly of actin plaques implicated in cell spreading, and increased cytokine production (reviewed in 18 .

Previously, we used lovastatin to inhibit 3-hydroxy 3-methylglutaryl coenzyme A reductase, the rate-limiting enzyme in the pathway to isoprenoid and cholesterol biosynthesis (19, 20). This treatment inhibits Ag-stimulated ⁴⁵Ca²⁺ influx, Ins(1, 4, 5) P₃ production, secretion, ruffling, and spreading in RBL-2H3 cells. The signaling pathway was restored by the addition of mevalonate, which is a precursor of the farnesyl and geranylgeranyl pyrophosphate substrates of protein prenyltransferases. Neither dolichol nor cholesterol, downstream of isoprenoid synthesis and metabolism in the cholesterol biosynthetic pathway, restored signaling responses. These studies identified critical roles for isoprenylated proteins in the coupling of FcεRI cross-linking to biochemical and functional responses. However, they could not determine whether the active species belonged to the relatively limited group of proteins with C-terminal CAAX motifs that are modified by farnesylation (mostly Ras isoforms but also Rap2, RhoB, the γ subunit of transducin, type 1 Ins(1, 4, 5)P₃ 5-phosphatase, and others) (21, 22, 23) or to the larger group of proteins with C-terminal CAAX, CC, or CXC motifs that are modified by geranylgeranylation (most of the Rho, Rab, and Ral family members; reviewed in 24 .

Here, we use the benzodiazepine-based farnesyltransferase inhibitor BZA-5B described by James et al. (25) to analyze the contributions of farnesylated proteins to FcεRI signaling and, by comparison with lovastatin, to identify roles for geranylgeranylated proteins in the signaling pathway. BZA-5B is a CAAX peptidomimetic that competes for the farnesyltransferase linking a C15 farnesyl group to a cysteine within carboxyl-terminal CAAX sequences on specific proteins. Like lovastatin, BZA-5B inhibits the farnesylation and Ag-induced activation of most Ras species (26, 27). Unlike lovastatin, BZA-5B and other CAAX peptidomimetics have little or no inhibitory activity toward type I geranylgeranyltransferase, which catalyzes the posttranslational geranylgeranylation of proteins, especially Rho and Ral family members, the CAAX box of which usually ends in a C-terminal leucine residue. They are also inactive against type II geranylgeranyltransferase, which acts on the Rab proteins that have C-terminal CC or CXC sequences (reviewed in Refs. 28 and 29).

Our results show that lovastatin and BZA-5B prevent both the farnesylation and Ag-induced activation of the Ras isoforms found in RBL-2H3 cells. Neither inhibitor substantially affects FcεRI-mediated tyrosine kinase activation, the earliest event in the FcεRI-coupled signaling pathway, or the activation of MEK and ERK1/ERK2, which are downstream of Ras in many cell types. Both inhibitors block cell division. However, a series of lovastatin-sensitive responses to FcεRI cross-linking were either unaffected or enhanced by BZA-5B. Analysis of these similarities and differences suggests Ras-independent pathways to MEK and ERK activation in RBL-2H3 cells and both predicted and novel roles for farnesylated and geranylgeranylated proteins in FcεRI-coupled signaling.

The farnesyltransferase inhibitor BZA-5B and its inactive analogue BZA-7B were generous gifts of Dr. James Marsters (Genentech, San Francisco, CA). Stock solutions (100 mM) of BZA-5B and BZA-7B were prepared in DMSO and stored at −20°C. For use, BZA-5B and BZA-7B were diluted into normal saline containing 100 mM reduced glutathione (used to maintain BZA-5B in its reduced form) as carrier and then further diluted in culture medium to final concentrations of 1 mM reduced glutathione and 0.1% DMSO. Lovastatin was solubilized as described in Kita et al. (30). Mouse monoclonal anti-DNP IgE was prepared as described in Liu et al. (31). The MEK inhibitor PD98058 was from BioMol (Plymouth Meeting, PA).

RBL-2H3 cells were grown as adherent monolayers on tissue culture flasks in MEM (Life Technologies, Grand Island, NY) supplemented with 15% FCS, penicillin-streptomycin, and l-glutamine. For microscopy, cell monolayers were cultured on 13-mm glass coverslips. For secretion assays, cell monolayers were grown in 24-well tissue culture plates. Unless otherwise noted, incubations were for 24 h with 10 μM lovastatin or for 72 h with 10 or 100 μM BZA-5B; control cells were incubated with carrier alone. In most experiments, IgE receptors were primed by the addition of 1 μg/ml anti-DNP-IgE during the last 12–14 h of incubation. After excess IgE was washed away, cells were activated with 0.01–1.0 μg/ml DNP-BSA (Molecular Probes, Eugene, OR).

For Northern blot detection of ras expression, mRNA was prepared from ∼50 × 10⁶ RBL-2H3 cells using the MicroFastTrack mRNA isolation kit (Invitrogen, San Diego, CA). mRNA (1.75 μg) was size fractionated by electrophoresis in 1.2% agarose formaldehyde gels and transferred to GeneScreenPlus (NEN-Dupont Research Products, Boston, MA). v-H-ras, v-K-ras, and N-ras human cDNA probes (Oncor, Gaithersburg, MD) were radiolabeled by random priming (Prime-It II kit, Stratagene, La Jolla, CA). Hybridizations followed the membrane manufacturer’s instructions. Blots were exposed to X-OMAT AR film (Eastman Kodak, Rochester, NY) at −70°C using an intensifying screen.

For Southern blot analysis of genomic ras, DNA was isolated from supernatants remaining after mRNA isolation by the MicroFastTrack kit as follows: the EDTA concentration of the supernatant was brought to 20 mM, and proteinase K was added to a final concentration of 0.5 mg/ml; the mixture was incubated overnight at 45°C with shaking; and DNA was then extracted with phenol/chloroform and precipitated with ethanol. DNA was also isolated from the liver of a Wistar rat (Harlan Sprague-Dawley, Indianapolis, IN). DNA samples (3 μg each) were digested to completion with the restriction endonucleases BamHI, EcoRI, and HindIII (Life Technologies), size fractionated in 0.8% agarose, and transferred to GeneScreenPlus. DNA blots were probed with radiolabeled v-H-ras, v-K-ras, and N-ras probes as described above.

To detect alternatively spliced forms of K-ras, mRNA prepared from adherent RBL-2H3 cells (20 × 10⁶/cells) was reverse transcribed to cDNA using the RiboClone System (Invitrogen, San Diego, CA). This cDNA, as well as cDNA from Wistar rat brain (a gift of Dr. K. K. Caldwell, University of New Mexico), was denatured and amplified by PCR using a single sense primer from exon 1 (5′-TGTGGTAGTTGGAGCTGGTGG-3′) and antisense primers reflecting either the exon 4A alternative splice form (5′-AATTTTCACACAGCCAGGAGT-3′) or the exon 4B alternative splice form (5′-GTACACCTTGTCCTTGACTTC-3′). As a positive control, glyceraldehyde 3-phosphate dehydrogenase (G3PDH) was amplified from the same cDNA using commercial primers (Clontech Laboratories, Palo Alto, CA). Amplified products were size fractionated by electrophoresis in a 1% agarose formaldehyde gel containing ethidium bromide (0.5 μg/ml). Sizes of DNA bands were estimated based on the mobilities of DNA standards (Life Technologies).

To detect transforming mutations in H-ras species expressed in RBL-2H3 cells, cDNA was prepared from RBL-2H3 mRNA, and exons 1 and 2 of the H-ras gene were amplified by PCR using the primers of Wang et al. (32). The amplified material was cloned into the Bluescript vector (Stratagene) and used to transform Escherichia coli SURE cells (Stratagene); colonies were screened using a radiolabeled v-H-ras probe (Oncor). Nucleotide sequence was determined using a Sequenase version 2.0 sequencing kit (United States Biochemical, Cleveland, OH) with the vector-specific primers recommended by the manufacturer. Exons 1 and 2 of the K-ras gene were similarly amplified using previously described primers (33, 34, 35) and sequenced.

RBL-2H3 cells were incubated for 2 h with BZA-5B or carrier. The medium was then supplemented with 10 μM lovastatin to inhibit the endogenous production of mevalonic acid, plus 10 μCi/ml [³H]mevalonolactone (40–60 Ci/mmol; American Radiolabeled Chemical, St. Louis, MO), and incubation continued for 20 h. Cells were lysed in 1 ml of lysis buffer A (10 mM Tris, pH 7.2; 50 mM NaCl; 20 mM sodium pyrophosphate; 50 mM NaF; 2 mM iodoacetamide; 5 μM ZnCl; 0.5% Triton X-100; 100 μM NaVO₄; 0.1% BSA; 10 μg/ml aprotinin and leupeptin; and 1 mM PMSF) and Ras proteins immunoprecipitated from the clarified lysates with the anti-Ras mAb Y13-259 prebound to protein A-Sepharose (Pharmacia, Piscataway, NJ). Beads were washed with buffer A. Immunoprecipitated proteins were separated by SDS-PAGE on 25-cm 12–15% gradient gels, and the gels were incubated with a fluorographic enhancer (Amersham, Arlington Heights, IL) and exposed for 20 days to X-OMAT MR film at −70°C for fluororadiography. Alternatively, the incorporation of [³H]mevalonolactone product into total immunoprecipitated protein was determined by boiling the beads for 5 min in 500 μl of 1 N HCl and measuring radioactivity from Ras-bound isoprene groups by liquid scintillation counting.

Drug-treated RBL-2H3 cells were transferred to inorganic phosphate-free medium, with 10% dialyzed FCS plus IgE and drugs and/or carrier for 2 h, followed by a 2-h incubation in inorganic phosphate-free medium supplemented with 333 μCi/ml [³²P]orthophosphate (Amersham). Cells were activated for 2 min with 1 μg/ml DNP-BSA and washed with ice-cold PBS, and 1 ml of lysis buffer B (50 mM Tris, pH 7.4; 10 mM MgCl₂; 500 mM NaCl; 1% Triton X-100; 0.5% sodium deoxycholate; and 0.05% sodium dodecyl sulfate) was added. Ras proteins were immunoprecipitated using mAb Y13-259 and washed with modified buffer B (0.1% Triton X-100 and 0.005% SDS) followed by 5 mM Tris-phosphate (pH 7.4). Radiolabeled guanine nucleotides were eluted from the immunoprecipitated proteins with 5 mM Tris-phosphate, 2 mM EDTA, and 2 mM DTT, pH 7.0. Carrier nucleotides were added to the eluate and the mixtures separated by one-dimensional TLC on polyethyleneimine-cellulose plates (36). Guanine nucleotides (GMP, GDP, and GTP) on dried plates were localized under UV light. Dried plates were exposed to phosphor screens (Molecular Dynamics, Sunnyvale, CA) for 3–4 days and scanned, and radioactivity in GTP and GDP spots was quantified using ImageQuant software (Molecular Dynamics). Data were expressed as the percentage total Ras in the GTP-bound form using the following equation: % GTP-Ras = cpm in GTP/cpm in (GTP + 1.5 GDP).

RBL-2H3 cells were radiolabeled with 100–200 μCi/ml [³²P]orthophosphate, activated with DNP-BSA, and lysed as described for Ras activation. Clarified lysates were incubated overnight with 1 μg of affinity-purified, polyclonal anti-phosphotyrosine Ab (generated by J. Potter and G. Deanin, University of New Mexico, as described in 37 precoupled to protein A-Sepharose beads (Pharmacia). Phosphoproteins were eluted from the beads with 1 mM phenylphosphate in the presence of 0.01% OVA and the protease inhibitors described above. Phosphoproteins were separated by 10% SDS-PAGE and detected by autoradiography.

RBL-2H3 cells were lysed with 1 ml of ice-cold lysis buffer C (25 mM HEPES, pH 7.5; 150 mM NaCl; 0.5% Triton X-100; 0.5% Brij-96; 0.1 mM EGTA; 1 mM NaVO₃; and protease inhibitors). Kinases were immunoprecipitated from clarified lysates using kinase-specific polyclonal Abs from Santa Cruz Biotechnology (Santa Cruz, CA; C-12 for Raf-1, C-18 for MEK; C-4 for ERK1/ERK2), all prebound to protein A-Sepharose beads. Beads were washed three times in ice-cold buffer C, twice more in the same buffer but containing 0.05% Triton X-100 and 0.05% Brij-96, and once in 25 mM HEPES, pH 7.5. Kinase reactions were initiated by adding 40 μl per sample of reaction buffer (30 mM Tris-Cl, pH 7.5; 15 mM MgCl₂; and 0.1 mM EGTA) containing 3 μg per sample kinase-dead MEK (glutathione S-transferase (GST)-MEK; K97A; Upstate Biotechnology, Lake Placid, NY) for Raf-1, kinase-dead ERK (GST-ERK; K91A; a kind gift of Dr. Alan Saltiel, Parke-Davis, Ann Arbor, MI) for MEK, myelin basic protein (MBP; Sigma, St. Louis, MO) for ERK, and 10 μCi per sample [γ-³²P]ATP (Redivue; Amersham). After 20 min at 30°C, reactions were terminated by the addition of 10 μl of 8× Laemmli buffer and boiling for 10 min. Proteins were separated by 12% SDS-PAGE and phosphoproteins visualized by autoradiography. Data were quantified using a PhosphorImager (Molecular Dynamics).

Levels of Ins(1, 4, 5) P₃ in TCA extracts of activated cells were determined using the Ins(1, 4, 5) P₃-specific radioreceptor assay of Challiss et al. (38) with modifications described in Ref. 8.

RBL-2H3 cell monolayers were fixed in 2% glutaraldehyde and dehydrated and carbon coated as previously described (39). Surface topography was observed with a Hitachi S800 scanning electron microscope.

RBL-2H3 cells (∼50 × 10⁶ cells/condition) were incubated for 2 h without or with 1 μg/ml DNP-BSA and then lysed, and mRNA was isolated as above using the FastTrack mRNA isolation kit. cDNA was prepared by reverse transcription, denatured, and amplified for 35 cycles (1 min at 94°C, 1 min at 53°C, and 2 min at 72°C) in the presence of PCR primers for IL-4 (sense, 5′-TTTAGGCTTTCCAGGAAGT-3′; antisense, 5′-GAGATCATCAACACTTTGAAC-3′) or G3PDH (control). Amplified products were size fractionated by 1% agarose gel electrophoresis in the presence of 0.5 μg/ml ethidium bromide. Sizes of DNA bands were estimated based on the mobilities of DNA standards (Life Technologies); the predicted product for IL-4 is ∼300 bp.

RBL-2H3 cell monolayers in 24-well plates (∼2 × 10⁵ cells per well) were loaded overnight with [³H]serotonin (400 nCi/ml; NEN-Dupont). Secretion was measured from the release of this preloaded mediator as described (40). Results are reported as percentage of total [³H]serotonin released in 20 min by duplicate samples and are corrected for spontaneous release from unstimulated cells in the same set.

[Ca²⁺]_i mobilization was measured by ratio imaging microscopy of fura-2-loaded cells as previously described (13). Each experiment provided time-resolved analyses of Ca²⁺ levels for 6–10 individual cells within a single field of view.

Effects of prenylation inhibitors on cell growth were determined by [³H]thymidine incorporation assays. Cells were harvested at 24-h intervals and counted, and 10⁵ cells were incubated in suspension for 1 h at 37°C with 3.5 μCi of nonmethylated 6-[³H]thymidine (Amersham) in a total of 0.35 ml. The cells were collected by centrifugation and washed, and the final pellet was solubilized in 0.25 ml 10% SDS plus 0.25 ml 1 N NaOH. Radioactivity in aliquots of the solubilized material was determined by liquid scintillation counting. Cell viability was determined in the same cell preparations by addition of trypan blue to a separate sample of each culture and observation of dye exclusion in a light microscope. Apoptosis was measured by staining cells in Krishan buffer (0.1% sodium citrate, pH 7.4, containing 0.3% Nonidet P-40, 0.005% propidium iodide, and 0.02 mg/ml ribonuclease A). Fluorescence was measured using a Becton Dickinson FACSscan flow cytometer with CellFIT software. BD Modfit software was used to identify uniform sub-G₀/G₁ peaks representing apoptotic cells.

Fig. 1 A shows the results of Northern blot analyses to detect ras species in RBL-2H3 mRNA preparations. A single band with apparent molecular mass of ∼1.5 kDa was detected when mRNA blots were hybridized with a radiolabeled H-ras-directed probe. A K-ras-directed probe revealed a single band with apparent molecular mass of ∼1.8 kDa. No bands were detected with a N-ras-directed probe (data not shown).

In Fig. 1,B, PCR amplification of cDNA was used to detect K-ras splice variants in mRNA from RBL-2H3 cells and Wistar rat brain. RBL-2H3 cells contain both K-ras4a and K-ras4b, whereas rat brain contains only K-ras4b. Fig. 1 C shows Southern blot analysis of H- and K-ras genomic DNA from RBL-2H3 cells, derived originally from a Wistar rat mastocytoma, and Wistar rat liver. Restriction fragment maps generated from genomic DNA from both sources are identical, suggesting that no major mutation or rearrangement of H- and K-ras occurred in the cell line. Exons 1 and 2 of oncogenic ras genes typically encode proteins with amino acid substitutions at residues 12, 13, 59, or 61 (41, 42). These mutant Ras proteins are constitutively maintained in the GTP-bound (active) state. Direct nucleotide sequencing of these conserved sites established that neither H-ras nor K-ras has transforming mutations in RBL-2H3 cells (data not shown).

Newly prenylated Ras species were detected from the incorporation of [³H]mevalonolactone into Y13-259-precipitable proteins in the presence of 10 μM lovastatin, which inhibits the de novo production of isoprene precursors. mAb Y13-259 immunoprecipitated a doublet of newly synthesized, prenylated Ras from [³H]mevalonolactone-labeled control cells (Fig. 1,D); separate experiments (not shown) using isoform-specific Abs established that the upper band is H-Ras, and the lower band is K-Ras. BZA-5B (100 μM) blocked incorporation of [³H]mevalonolactone products into this doublet of Ras proteins by greater than 70% (Fig. 1 D; compare lanes 1 and 2). Y13-259 Western blots of replicate Y13-259 immune complexes (not shown) detected very similar levels of H- and K-Ras protein in control and BZA-5B-treated cells. Thus, the reduced signal from [³H]mevalonolactone labeling is due to reduced protein prenylation, not to reduced Ras protein concentrations. Indeed, BZA-5B-treated cells probably overexpress Ras proteins. This was indicated by the presence in anti-Ras immunoblots of BZA-5B-treated cells of an additional higher molecular mass band that was completely nonprenylated but showed cross-reactivity with anti-ubiquitin Abs. This band, representing perhaps 10% of the total signal in anti-Ras blots, presumably represents newly synthesized Ras that was marked for degradation. As an independent measurement, we used liquid scintillation counting to quantify the effect of BZA-5B on [³H]mevalonolactone incorporation into Ras immunoprecipitates. These experiments confirmed that 100 μM BZA-5B inhibits the prenylation of Ras species by more than 70%. In contrast, 1 μM and 10 μM concentrations of BZA-5B had no measurable effect on Ras prenylation (data not shown).

Ag-induced Ras activation was determined from the increase in the percentage of Ras in the GTP-bound (active) form in Y13-259 immunoprecipitates prepared from cell lysates (Fig. 2). In control cells, incubation with Ag caused an increase in GTP-Ras to approximately twice resting levels within 2 min. Cells treated for 72 h with 100 μM BZA-5B or for 24 h with 10 μM lovastatin responded to FcεRI cross-linking with either no change or a small decrease in the percentage of Ras in the GTP-bound form.

In many cells, activated Ras couples directly to the serine/threonine kinase Raf-1. Raf-1 activity was determined from the ability of immune complexes prepared with the Raf-1-specific Ab, C-12, to phosphorylate kinase-dead GST-MEK. The results in Fig. 3 A (top) show that FcεRI-induced Raf-1 activity peaked within 5 min and remained elevated for at least 10 min. Raf-1 activation was abolished in cells exposed for 24 h to 10 μM lovastatin. In contrast, 72-h incubation with 100 μM BZA-5B resulted in only a modest reduction in Raf-1 activation.

MEK and ERK activities toward kinase-dead GST-ERK and MBP, respectively, were similarly measured in immune complex kinase assays. FcεRI cross-linking activates MEK (Fig. 3,A, middle) and ERK1/ERK-2 (Fig. 3 A, bottom). Activation was maximal within 5 min after cross-linking and persisted for at least 10 min. The rate and extent of MEK and ERK1/ERK2 activation were very similar in control, BZA-5B-treated, and lovastatin-treated cells.

RBL-2H3 cells were also treated with the MEK inhibitor PD98059 (43). In contrast with the prenylation inhibitors, 1-h treatment with 100 μM PD98059 inhibited the FcεRI-induced activation of MEK (Fig. 3,B, top) and of ERK1/ERK2 (Fig. 3 B, bottom).

Previously, Shakarjian et al. (44) reported a small inhibition of Ag-induced protein-tyrosine phosphorylation in lovastatin-treated cells. Fig. 4 shows the results of similar studies in our laboratory; arrows indicate bands identified in previous studies (2, 40) as the FcεRI-associated tyrosine kinases Lyn and Syk, which are implicated in signal initiation and signal propagation, respectively (40). Lovastatin pretreatment had no effect on the overall pattern of protein-tyrosine phosphorylation in Ag-stimulated cells (Fig. 4; compare lanes 2 and 5). Incubation with BZA-5B also had no effect on the pattern of basal and Ag-induced protein-tyrosine phosphorylation (Fig. 4; compare lanes 1 and 2 with lanes 3, 4, and 6). Lovastatin, but not BZA-5B, modestly reduced the extent of Ag-induced phosphorylation. These results reveal no substantial role for protein prenylation in the processes of tyrosine kinase activation that initiate the FcεRI-coupled signaling cascade.

A series of lovastatin-sensitive responses to FcεRI cross-linking were unaffected by the farnesyl-specific inhibitor BZA-5B. Assays of Ag-induced Ag-activated Ins(1, 4, 5)P₃ synthesis are shown in Fig. 5. As previously reported, lovastatin inhibited Ag-stimulated Ins(1, 4, 5)P₃ production (20). Ag-induced Ins(1, 4, 5)P₃ production was the same in control cells and in cells treated with 100 μM BZA-5B.

Assays of Ag-induced membrane ruffling and spreading are shown in Fig. 6. Unstimulated RBL-2H3 cells adhered to glass or plastic surfaces and showed a modest spreading response (Fig. 6,A). Cross-linking the FcεRI transformed the dorsal surface from a microvillous to a lamellar topography and caused a dramatic increase in cell spreading and adhesion (Fig. 6,B). As reported previously (20), cells treated for 24 h with 10 μM lovastatin became rounded and poorly adherent (Fig. 6,C) and no longer showed ruffling and spreading responses to Ag (Fig. 6,D). In contrast, 72-h incubation with 100 μM BZA-5B did not alter the morphology of resting cells (Fig. 6,E). Furthermore, BZA-5B-treated cells responded to Ag by vigorous ruffling and spreading (Fig. 6 F).

The production of cytokines, including IL-4, is a late (2–4 h) response to FcεRI cross-linking that depends on Ca²⁺ mobilization, signaling to the nucleus, and transcriptional activation (reviewed in 45 . The results in Fig. 7 show that Ag-induced IL-4 message production was the same in control and BZA-5B (100 μM)-treated cells. Lovastatin inhibited the Ag-induced production of IL-4 mRNA in RBL-2H3 cells. IL-4 production induced by the Ca²⁺ ionophore ionomycin was not inhibited by lovastatin.

Remarkably, two lovastatin-sensitive responses to FcεRI cross-linking, secretion, and the release of Ca²⁺ from stores, were enhanced in BZA-5B-treated cells. Secretion data are shown in Fig. 8. As reported previously, 24-h incubation of RBL-2H3 cells with 10 μM lovastatin blocked the Ag-induced release of [³H]serotonin (20). Treatment for 1 h with 1 μM or 10 μM concentrations of the MEK inhibitor PD98059 also inhibited FcεRI-induced secretion. In contrast, 72 h incubation of RBL-2H3 cells with BZA-5B increased the Ag-stimulated secretion of [³H]serotonin. Secretion was also induced in RBL-2H3 cells by ionomycin. Like Ag-induced secretion, ionomycin-induced secretion was inhibited by lovastatin and PD98059 and enhanced by BZA-5B.

The Ca²⁺ mobilization response induced by Ag in a typical IgE-primed control cell is illustrated in Fig. 9 A. Before stimulation, cells maintained a low basal concentration of free Ca²⁺. Addition of 0.1 μg/ml DNP-BSA resulted, after a characteristic delay, in an abrupt elevation in [Ca²⁺]_i that was maintained for at least 10 min. Previous work showed that the Ca²⁺ spike resulted from Ca²⁺ release from intracellular stores and that the maintained elevation required Ca²⁺ influx (12, 13).

Lovastatin-treated cells also released Ca²⁺ stores in response to Ag (Fig. 9,B). Consistent with previous evidence that lovastatin inhibits Ag-induced [⁴⁵Ca²⁺] uptake, the maintained component of the Ca²⁺ mobilization response was significantly smaller in lovastatin-treated cells than in control cells. Furthermore, in approximately one-third of cells (17 of 49) the Ca²⁺ response consisted of fairly regular oscillations, interpreted as cycles of stores release and uptake, superimposed on a slowly rising baseline (as in Fig. 9 B). Comparable oscillations were not observed in any of 44 Ag-stimulated control cells.

Fig. 9, C and D, shows that the typical Ag-stimulated Ca²⁺ responses of cells treated with 10 and 100 μM BZA-5B is nonoscillatory and appears very similar to that of control cells. However, 16% (23 of 140 cells) of cells treated with BZA-5B, including the one illustrated in Fig. 9 D, showed Ca²⁺ oscillations prior to FcεRI cross-linking. Spontaneous oscillations were not observed in any of 62 control cells. Because spontaneous oscillations in some cells were observed in the absence of [Ca²⁺] (data not shown), they apparently result from the release and reuptake of intracellular Ca²⁺ stores. The observations suggest that a farnesylated protein acts in both resting and Ag-stimulated cells to suppress Ca²⁺ stores release.

Protocols described in Lee and Oliver (13) were used to measure lag time, defined as the time from Ag addition to a Ca²⁺ spike in cells stimulated in the absence of extracellular Ca²⁺, and response magnitude, defined as the area under the [Ca²⁺]_i curve integrated for 4 min following the onset of the Ca²⁺ response in cells stimulated in the continual presence of extracellular Ca²⁺. Results are given in Table I. In control cells, the average lag time from Ag addition to the release of intracellular Ca²⁺ stores was 81 s. The magnitude of the Ca²⁺ response was 59.8 nM · s. The average lag time from Ag addition to a Ca²⁺ response in lovastatin-treated cells was 58 s, slightly (but not significantly) less than the 81-s lag time of control cells. However, lovastatin reduced the magnitude of the Ca²⁺ response by almost 50%. The average lag times to Ca²⁺ response in cells treated with 10 and 100 μM BZA-5B were 47 s and 40 s, respectively, significantly less than the 81-s average lag time to Ca²⁺ response in Ag-stimulated control cells. In BZA-5B-treated cells, the maintained Ca²⁺ elevation was nonoscillatory and either was of comparable magnitude with that of control cells (cells treated with 10 μM BZA-5B) or was modestly reduced in comparison with that of control cells (cells treated with 100 μM BZA-5B).

Incubation of RBL-2H3 cells with BZA-5B slowed but did not abolish the incorporation of [³H]thymidine into DNA (Fig. 10,A). The reduced rate of thymidine incorporation was apparent within 24 h of BZA-5B addition and was maximal by 48 h. Growth inhibition was seen at 10 μM BZA-5B, at which most newly synthesized Ras protein could still be farnesylated (data not shown). Even at the highest dose of drug (100 μM), BZA-5B-treated cells remained viable beyond 96 h of treatment, as assessed by trypan blue exclusion (not shown). In contrast, lovastatin impaired proliferation within 24 h (Fig. 10 A), and cell viability, assessed by trypan blue exclusion, was reduced within 48 h.

Apoptosis was measured by flow cytometry of propidium iodide-treated cells. When RBL-2H3 cells were cultured in the presence of carrier alone, 100% of cells were distributed within the G₀/G₁, S, and G₂/M phases of the cell cycle (Fig. 10,B, panel 1). Twenty-four hours after heat shock (90 min at 43°C), all of the cells were apoptotic, as indicated by a broad peak of fragmented DNA (Fig. 10,B, panel 2). No apoptotic cells were detected after 72 h of BZA-5B treatment (Fig. 10,B, panel 3). There was also very little apoptosis in cultures exposed to lovastatin for 24 h (Fig. 10,B, panel 4). Results in Fig. 10 B, panel 5, show that 48 h of lovastatin treatment leads to apoptosis.

We demonstrated previously that FcεRI-stimulated signaling responses are inhibited by lovastatin, a cholesterol biosynthetic pathway inhibitor that blocks the synthesis of the farnesyl and geranylgeranyl pyrophosphates needed to isoprenylate a set of proteins recognizable by their C-terminal CAAX, CXC, or CC motifs (19, 20, 24). Signaling responses could be restored by adding mevalonic acid, which is a direct precursor of these isoprenoid derivatives, but not by dolichol and cholesterol, which are downstream of isoprenoid metabolism in the cholesterol biosynthetic pathway. These results indicated that lovastatin inactivates isoprenylated proteins required for FcεRI-stimulated signaling, presumably by preventing their association with membranes. Here, we used the farnesyl-specific inhibitor BZA-5B to analyze the contributions of farnesylated proteins to FcεRI signaling and, by comparison with lovastatin, to identify roles for geranylgeranylated proteins in the signaling pathway. We focused attention on Ras isoforms because these are the principal farnesylated species implicated in receptor-mediated signaling pathways. A role for Ras in FcεRI-coupled signaling was suggested previously by evidence that FcεRI cross-linking stimulates the tyrosine phosphorylation of Vav, a GDP-GTP exchange factor with activity toward Ras and other GTPases (6, 46), as well as by reports that FcεRI cross-linking stimulates assembly of the Ras-stimulatory Shc-Grb2-SoS complex (9, 14).

We began by identifying the ras isoforms in RBL-2H3 mast cells and demonstrating their activation by FcεRI cross-linking. Four principal Ras species, H-ras, N-ras, K-rasA, and K-rasB, have been described in animal cells (41). RBL-2H3 cells express all of these isoforms except for N-ras. Ras proteins are not amplified or expressed in oncogenic forms in RBL-2H3 cells. Ras-GTP levels increase in Ag-stimulated cells, confirming that FcεRI cross-linking activates Ras. Importantly, BZA-5B and lovastatin abolish the Ag-induced increase in Ras-GTP levels, indicating that both drugs inhibit Ras activation. Although no Ras activation could be measured, BZA-5B caused only a 70% inhibition of Ras prenylation. Others have shown that the K-Ras4B isoform (47, 27), as well as N-ras and K-Ras4A (17), can serve as an in vitro substrate for geranylgeranyl transferase-1. Thus, it is possible that some of the residual prenylated Ras reflects a pool of BZA-5B-resistant geranylgeranylated K-Ras. Because Ag-induced Ras activation measured by assays of total Ras-GTP levels is completely inhibited in BZA-5B-treated cells, it is likely that this residual prenylated Ras is not activated by FcεRI cross-linking. We recognize, however, that our assay may fail to detect an increase in GTP bound to a minor isoform of Ras.

The FcεRI-coupled signaling cascade is initiated by the activation of two receptor-associated tyrosine kinases, Lyn and Syk, which in turn phosphorylate multiple substrates (reviewed in 48 . We found that BZA-5B does not alter the Ag-stimulated tyrosine phosphorylation of RBL-2H3 proteins, including the FcεRI-associated kinases Lyn and Syk, and we confirmed previous evidence (44) that Ag-stimulated protein tyrosine phosphorylation is also not substantially inhibited by lovastatin. Thus, it is unlikely that prenylation inhibitors target the earliest events in the FcεRI signaling cascade.

In many cell systems, receptor-mediated tyrosine kinase activation results in Grb2/SoS-mediated Ras activation that couples directly to the serine/threonine kinase, Raf-1, resulting in Raf-1 recruitment to the plasma membrane and activation. Activated Raf-1 in turn phosphorylates the dual-specificity kinases MEK-1 and MEK-2, which finally phosphorylate and activate the MAP kinases ERK-1 and ERK-2, which are implicated in transcriptional activation (reviewed in Refs. 49–51). We confirmed that Raf-1 shows increased activity after FcεRI cross-linking. We also showed increased MEK and ERK1/ERK2 activities in Ag-treated cells. Nevertheless, we failed to obtain support for the classical (Ras → Raf → MEK → ERK) pathway of ERK1/ERK2 activation in RBL-2H3 cells. First, only lovastatin, and not BZA-5B, strongly inhibits Ag-stimulated Raf-1 activation. The persistent Raf-1 activation in BZA-5B-treated cells could be mediated by a small amount of BZA-5B-resistant geranylgeranylated K-RasB, which was not detected in our assay for Ras-GTP levels. Alternatively, it is possible that Ag-stimulated Raf-1 activation occurs in RBL-2H3 cells by a pathway that, instead of requiring Ras, requires the activation of members of the Rho family of geranylgeranylated GTPases (52, 53) and so is more sensitive to inhibition by lovastatin than by BZA-5B. Unexpectedly, neither BZA-5B nor lovastatin has any effect on MEK1 or ERK1/ERK2 activation. These data clearly establish the presence of Raf-1-independent pathways to MEK activation in RBL-2H3 cells. In contrast with lovastatin and BZA-5B, the MEK inhibitor PD98058 prevented Ag-induced ERK1/ERK2 activation. These results locate MEK conventionally upstream of the ERK MAP kinases in RBL-2H3 cells.

Why do mast cells use apparently Ras- and Raf-1-independent pathways to FcεRI-mediated ERK1/ERK2 activation when they clearly contain the elements of the classical Grb2/SoS/Ras/Raf/MEK pathway defined in fibroblasts? One explanation is that hematopoietic cells may use individual variations on the general signaling sequences established in fibroblasts. Consistent with this, preliminary evidence that BZA-5B reduces both basal and Ag-stimulated JNK activity toward its substrate, GST-c-Jun, in RBL-2H3 cells raises the possibility that mast cell Ras might activate a non-ERK member of the MAP kinase family (T.E.G. and B.S.W., unpublished results). Another explanation is that Ag-stimulated mast cells may activate MAP kinases by pathways that are independent of Ras-related GTPases. In other cells, protein kinase C isozymes appear to link certain receptors directly to Raf and MEK isoforms, bypassing Ras (53). Importantly for our studies, James et al. (26) reported that BZA-5B blocks epidermal growth factor-stimulated ERK activation in H-Ras-transformed Rat-1 fibroblasts, but not in untransformed cells. One explanation is that the untransformed cells contained a small (undetectable in biochemical assays) amount of BZA-5B-resistant Ras. An alternative explanation is that Rat-1 cells lacking oncogenic Ras, like RBL-2H3 cells, favor a Ras-independent pathway to receptor-mediated ERK activation.

We showed previously that lovastatin inhibits a series of FcεRI-mediated responses, including Ag-stimulated Ins(1, 4, 5)P₃ production, a measure of PLCγ activation; Ca²⁺ influx, attributed in large part to the coupling of Ca²⁺ stores release to capacitative Ca²⁺ entry (13); and a series of functional responses including secretion, ruffling, spreading, and IL-4 production. In the present study, BZA-5B failed to inhibit any of these signaling responses. These data implicate geranylgeranylated (lovastatin-sensitive, BZA-5B-insensitive) proteins acting downstream of Ag-stimulated protein-tyrosine phosphorylation in the regulation of a diverse array of signaling responses.

The inhibition of ruffling, spreading, and actin plaque assembly by lovastatin, which inhibits all protein prenylation, but not by BZA-5B, which spares protein geranylation, was predictable based on evidence from Hall that different members of the geranylgeranylated Rho family of GTPases control the formation of filopodia (CDC42), ruffles (Rac) and adhesion and spreading (Rho) (reviewed in 54 . Indeed, Guillemot and colleagues (55) recently reported that expressing dominant negative mutant forms of CDC42 in RBL-2H3 cells decreases FcεRI-induced adhesion and actin plaque assembly, while expressing dominant negative Rac1 abolishes ruffling. The selective inhibition by lovastatin of secretion was also predictable based on evidence from Prepens and colleagues (56) that Clostridium difficile toxin B, which targets the geranylgeranylated GTPases RhoA and CDC42, blocks Ag-induced secretion from RBL-2H3 cells. Other geranylgeranylated GTPases, including Rab3B and Rab3D cloned from RBL-2H3 cells (57) and primary rat mast cells (58), have also been implicated in vesicular trafficking and secretion (59).

Our data also localize geranylgeranylated proteins to sites in the FcεRI-coupled signaling cascade that have not previously been described. First, we found that lovastatin, but not BZA-5B, inhibits Ag-stimulated Ins(1, 4, 5)P₃ synthesis. This result implicates a geranylgeranylated protein in the control of Ag-induced PLCγ activation. We cannot as yet define the protein. However, recent studies have shown that the recruitment and activation of PLCγ isoforms by FcεRI cross-linking in RBL-2H3 cells is a complex process requiring not only tyrosine phosphorylation but also interactions with at least one other enzyme, phosphatidylinositol 3-kinase (5, 8). Other investigators have shown that Rac and CDC42 associate with phosphatidylinositol 3-kinase (60, 61). Thus, it is possible that a geranylgeranylated GTPase of the Rho family is involved in the regulation of PLCγ activation by inositol phospholipids.

Second, we found that Ag-stimulated Ca²⁺ influx is inhibited by lovastatin but not by BZA-5B. One explanation is simply that the reduced Ins(1, 4, 5)P₃ levels in lovastatin-treated cells are sufficient to support the initial Ca²⁺ stores release response but too low to maintain the stores in an empty state, as needed for capacitative influx. However, we showed previously that vigorous Ca²⁺ stores release and influx can be supported at low Ag concentrations that induce less Ins(1, 4, 5)P₃ synthesis than occurs in lovastatin-treated cells (13). Alternatively, previous studies by Wilson et al. (11) and Fasolato et al. (62) have implicated GTP-binding proteins in the coupling of empty Ca²⁺ stores to capacitative Ca²⁺ influx. The inhibition of Ca²⁺ influx by lovastatin but not by BZA-5B now suggests that the putative coupling protein belongs to the family of geranylgeranylated GTPases. Because lovastatin-treated cells synthesize IL-4 mRNA in response to ionomycin but not to Ag, we speculate that the Ag-induced production of IL-4 depends largely on this lovastatin-sensitive Ca²⁺ influx pathway. Secretion also requires Ca²⁺ influx (10, 12). However, lovastatin inhibits ionomycin-induced secretion, indicating roles for additional geranylgeranylated proteins downstream of Ca²⁺ influx in the pathway linking FcεRI cross-linking to degranulation.

Most interestingly, BZA-5B enhances Ag-induced secretion and reduces the lag time from Ag addition to Ca²⁺ stores release. The BZA-5B-induced potentiation of secretion and of Ca²⁺ stores release was observed at BZA-5B concentrations too low to affect Ras farnesylation and activation. We suggest, therefore, that these changes reflect the inactivation by BZA-5B of a non-Ras-farnesylated protein that is normally involved in the suppression of Ca²⁺ stores release and secretion. Properties of this protein that can be inferred from our data include its relatively slow turnover rate (since 72-h incubation with BZA-5B is required to observe potentiation) and its sensitivity to unusually low concentrations of farnesylation inhibitors. There is precedent for inhibition of farnesylation of at least one non-Ras protein, nuclear lamin B, at BZA-5B concentrations too low to inhibit Ras prenylation (63). We speculate that the target protein may be the type I Ins(1, 4, 5)P₃ 5-phosphatase. This enzyme associates with membranes via its farnesylated C-terminal CAAX motif (22, 64). Its substrates are phosphatidylinositol 4,5-bisphosphate and Ins(1, 4, 5)P₃. In particular, it converts the Ca²⁺-mobilizing metabolite, Ins(1, 4, 5)P₃, to the inactive metabolite, Ins(1, 4)P₂. In its nonfarnesylated form, type I Ins(1, 4, 5)P₃ 5-phosphatase would be expected to be soluble rather than membrane associated, reducing its access to Ins(1, 4, 5)P₃ interacting with its receptors at the endoplasmic reticulum membrane and so increasing Ins(1, 4, 5)P₃-mediated Ca²⁺ store release and secretion.

In course of these studies, we observed that the MEK inhibitor PD98059 potently inhibits secretion. This result supports previous evidence that ERK activation is required for Ag-stimulated secretion in RBL-2H3 cells (16). Nevertheless, since MEK and ERK activation occur normally in lovastatin-treated cells, the inhibition of secretion by lovastatin is not explained by this pathway.

Ras activation plays a pivotal role in the pathways coupling growth factor receptors to signal transduction pathways, at least in fibroblast cell lines. It was therefore surprising that our studies failed to reveal any clear role for Ras in FcεRI-mediated signaling in RBL-2H3 cells. We considered the possibility that the principal role of Ras is in mast cell growth regulation rather than acute aspects of FcεRI signaling. Consistent with this hypothesis, both lovastatin and BZA-5B blocked DNA synthesis within 24 h. BZA-5B concentrations that abolish Ras activation did not decrease cell viability or induce apoptosis, at least over a 96-h period. In contrast, lovastatin impaired cell viability and induced apoptosis within 48 h, presumably reflecting important roles for geranylgeranylated (lovastatin-sensitive, BZA-5B-insensitive) proteins in the pathways that protect against apoptosis. Relatively low concentrations of BZA-5B that only modestly reduced Ras prenylation also arrested cell growth. One explanation is that a non-Ras-farnesylated protein with greater sensitivity to BZA-5B is essential for RBL-2H3 cell proliferation. Alternatively, if a Ras pathway contributes to mast cell proliferation, even a small amount of nonfarnesylated Ras may block cell proliferation by trapping Raf or another effector molecule in an inactive cytoplasmic complex. Previously, Lerner et al. (65) invoked the induction of signal curtailing Ras-Raf complexes to explain how another farnesyl-specific inhibitor, FTI-277, may block the growth of Ras-transformed cells at concentrations 100-fold lower than those required to abolish Ras farnesylation.

Previous investigators have shown that in general BZA-5B and other farnesylation inhibitors do not significantly inhibit the proliferation of non-ras-transformed tumor cells and tissue culture lines (25, 26, 66). However, a screen by Sepp-Lorenzino et al. (67) of a large panel of cell lines lacking Ras mutations for sensitivity to the farnesyltransferase inhibitor FTI L-744832 revealed that all seven of the hematopoietic cell lines in the panel were sensitive. This suggests that the proliferation of hematopoietic system cells expressing nononcogenic Ras isoforms may be unusually sensitive to farnesyltransferase inhibitors. Nevertheless, BZA-5B-treated RBL-2H3 cells remain viable for extended periods. Thus, growth arrest of hematopoietic cells may not compromise the use of farnesyltransferase inhibitors in cancer therapy.

We thank Dr. J. Marsters (Genentech) for supplying BZA-5B and BZA-7B and Dr. R. Larson (University of New Mexico) for assistance with apoptosis assays. Cytometry and microscopy experiments were performed using shared facilities supported in part by the University of New Mexico Cancer Research and Treatment Center.