RACK1 regulates Ki-Ras-mediated signaling and morphological transformation of NIH 3T3 cells

Abstract

Activating Ras mutations are involved in a significant fraction of human tumors. A suppressor screen using a retroviral mouse fibroblast cDNA library was performed to identify novel factors in Ras-mediated transformation. We identified a novel potent inhibitor of Ras-mediated morphological transformation encoded by a truncated version of the receptor for activated C-kinase (RACK1). The truncated protein, designated RACK1ΔWD1, lacked the N-terminal 49 amino acids encoding the first of the 7 WD40 repeats in RACK1. RACK1ΔWD1 expression restored contact inhibition, stress fiber formation and reduced ERK phosphorylation in Ki-Ras transformed NIH 3T3 cells. We demonstrate that truncated RACK1 is involved in complexes consisting of wild-type RACK1 and protein kinase C isoforms α, βI and δ, compromising the transduction of an activated Ras signal to the Raf-MEK-ERK pathway. The cellular localization of RACK1ΔWD1 differed from wtRACK1, indicating that signaling complexes containing the truncated version of RACK1 are incorrectly localized. Notably, 12-O-tetradecanoyl-13-phorbol acetate (TPA) mediated intracellular translocation of RACK1-interacting PKC α and δ was abrogated in RACK1ΔWD1-expressing cells. Our data support a model where RACK1 acts as a key factor in Ki-Ras-mediated morphological transformation. © 2006 Wiley-Liss, Inc.

Members of the ras gene family (Ki-ras, Ha-ras and N-ras) are structurally related and encode proteins (p21) known to play an important role in the regulation of normal signal transduction and cell growth. A range of ras mutations are present with high frequency in different human tumors. The mechanism by which mutated Ras contributes to cancer development and morphological transformation of cells is still not completely understood. Ras generates downstream effects through Raf, Rac and Rho affecting cell growth, lamellipodia formation and stress fiber formation, respectively.1, 2, 3 Activated Ras recruits Raf-1 from the cytosol to the cell membrane, where Raf-1 activation takes place through dephosphorylation of inhibitory sites by protein phosphatase 2A (PP2A) as well as the phosphorylation of activating sites by a range of kinases. Activated Raf-1 phosphorylates and activates MEK (ERK/MAPK kinase), which in turn phosphorylates and activates extracellular-signal-regulated kinase (ERK/MAPK). Activated ERK has many substrates in the cytosol, and in the nucleus it controls gene expression by phosphorylating transcription factors such as Elk-1 and other Ets-family proteins.4, 5, 6 The activation of the ERK/MAPK pathway is central in Ras signaling, and expression of active ERK is sufficient to transform 3T3 cells.7 Importantly, mutations in B-Raf are frequently observed in human cancers (melanomas). In addition, the c-Jun N-terminal kinase (JNK)/MAPK pathway,8, 9 the PI3-K/Akt10, 11, 12, 13 and Src14, 15 pathways have been linked to cellular transformation, while the p38 MAPK pathway provides negative feedback of Ras transformation by blocking activation of JNK.16

Oncogenic Ras signaling can be blocked by inhibiting Ras palmitoylation and farnesylation.17 Dominant-negative mutations of Raf-118 and MEK19 effectively block Ras oncogenicity. Similarly, the membrane-anchored glycoprotein RECK can function downstream of Ras to block Ras-mediated morphological transformation.20, 21 Spry2 of the Sprouty family of human homologs to drosophila Sprouty was shown to block Ras signaling by inhibiting Raf activation.22 In an attempt to identify and characterize novel factors inhibiting the transforming signal of mutated Ki-Ras, we used a retroviral cDNA expression screen in NIH 3T3 cells. Here, we show that a truncated version of receptor for activated C-kinase (RACK1) is able to block the Ki-Ras transforming signal.

The importance of correct spatial and temporal organization of the individual components in signal complexes is increasingly recognized. Based on homology to the G_β-subunit of heterotrimeric G-proteins, the 7 WD40 repeats of RACK1 are postulated to form a propeller structure, where the blades potentially bind different proteins.23, 24 This will provide a platform for functional specificity and bring together components of one or several signaling pathways. It has recently been shown that dimerized RACK1 binds activated Ras and affects Ras-dependent signaling.25 Importantly, RACK1 binds a subset of protein kinase C (PKC) isoforms in their activated form and is thus essential in increasing the PKC substrate specificity.26, 27, 28

ERK/MAPK, extracellular signal-regulated kinase/mitogen-activated protein kinase; JNK, c-Jun N-terminal kinase; MEK, mitogen-activated protein/extracellular signal-regulated kinase; pERK, phosphorylated ERK; PI3-K, phosphatidylinositol-3-OH kinase; PKC, protein kinase C; PLD, phospholipase D; RACK1, receptor for activated C-kinase; RT-PCR, reverse transcriptase polymerase chain reaction; TPA, 12-O-tetradecanyl phorbol-13-acetate (phorbol ester); SDS-PAGE, SDS-polyacrylamid gel electrophoresis.

Here, we demonstrate that the truncated form of RACK1 lacking the first WD40 repeat is an effective inhibitor of Ki-Ras-mediated transformation of 3T3 cells and restores cellular contact inhibition and actin fiber formation. Our studies show that the MAP kinase pathway is important in Ki-Ras-mediated morphological transformation, and that the activation of ERK by mutated Ki-Ras is significantly decreased in cells coexpressing N-terminally truncated RACK1. In addition, translocation of PKC, ERK phosphorylation and filopodia-formation in response to 12-O-tetradecanoyl-13-phorbol acetate (TPA) treatment are inhibited by truncated RACK1. We further demonstrate that RACK1ΔWD1 interacts with wtRACK1 and PKC isoforms and inhibits the cellular translocation of PKC in response to TPA. Our present findings designate RACK1 as an important PKC-dependent factor in Ras signaling. Our results are consistent with a model in which RACK1–PKC interaction is involved in modulating the Ras-MAPK signaling.

Material and methods

Recombinant cDNA library and plasmids

A mouse fibroblast fragment cDNA library in the murine retroviral vector pBabe X was used for cell morphology screening, and a Jurkat cDNA library in pBabe M was used to clone the full-length cRACK1.29 The vector MutK pBabepuro contained the murine Ki-ras gene mutated in codon 11 (A to V). MSCV lacZ was used to estimate the level of transfection. The env, gag and pol genes from Moloney murine leukemia virus (MuLV) were expressed from pRSV GP and pCMV Env. Candidate genes were tagged with a C-terminal V5-epitope from the paramyxovirus, SV5 (Invitrogen), and were cloned into the retroviral vector wzlHYGRO.

Materials and antibodies

12-O-tetradecanyl phorbol-13-acetate (TPA) from Sigma was dissolved in acetone. Wortmannin and PD 98059 were from Sigma. Monoclonal anti-RACK1 IgM was from Affinity Research Products and anti-V5 was from Invitrogen. Rabbit polyclonal antibodies against ERK1 (H-8), PKCα, βI and δ were from Santa Cruz Biotechnologies. Monoclonal anti-phospho-p44/42 MAP kinase (E10, pERK1/2) was from Cell Signaling. FITC-conjugated goat-anti-rabbit IgG was from Southern Biotechnology and Alexa568-conjugated goat-anti-mouse IgG was from Molecular Probes. Horseradish peroxidase linked anti-mouse and anti-rabbit antibodies were from Amersham.

Cell culture and virus production

NIH 3T3 cells (American type culture collection, ATCC) and derived clones were cultured in Dulbecco's Modified Eagle's Medium supplemented with 10% calf serum. The 293T cells were maintained in Dulbecco's Modified Eagle's Medium supplemented with 10% heat-inactivated fetal calf serum. Replication defective virus was produced as described by Swift et al.,30 using 293T cells transfected with 10 μg library in pBabe X, 1 μg MSCV lacZ, 8 μg pRSV GP and 8 μg pCMV Env, or with 10 μg MutK pBabe puro, 1 μg MSCV lacZ, 8 μg pRSV GP and 8 μg pCMV Env for the challenge experiment. The supernatant was removed after 48 hr and was sterile filtered through a 0.45-μm filter. 3T3 cells were infected with virus representing the mouse fragment cDNA plasmid library. Virus-library-carrying 3T3 cells were challenged with MutK-ras virus and were selected for puromycin resistance. Retrovirus rescue experiments were done by transfecting virus-library-carrying 3T3 cells with the gag-pol and env genes. For growth rate experiments, 3T3 cells were seeded onto 6-cm dishes, 5,000 cells per dish, and were medium changed every 4 to 5 days. The total number of cells per dish was determined every second or third day. For foci staining, cells were grown for 16–20 days and were stained with Gimsa (Sigma).

RNA purification and analysis

Total cellular RNA was isolated by the acid guanidinium-thiocyanate method.31 Total RNA (20 μg) was used in Northern blot experiments. The electrophoresis, blotting and hybridization conditions were as described in Helleland et al.32 The specific activities of the DNA probes were approximately 1 × 10⁹ cpm/μg DNA. RT-PCR primers were 5′-CCC TTT TTC TGG AGA CTA-3′ (52°C), 5′-GAT CCT CCC TTT ATC CAG-3′ (56°C) and 5′-CAG GTG GGG TCT TTC ATT CC-3′ (62°C). Automated DNA sequencing was performed using the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit and the ABI Prism 377 XL.

Immunoblotting and immunofluorescence assay

SDS-polyacrylamid gel electrophoresis (SDS-PAGE) and immunoblots were performed as described.33 For anti-pERK1/2-immunoblots, cells were serum starved for 16 hr prior to treatment and harvesting. Immunofluorescent staining was performed as described.34 For translocation studies, cells were treated with high dose (1.7 μM) or low dose (2 nM) TPA for the appropriate time prior to fixing in 4% formaldehyde in C-buffer (137 mM NaCl, 5 mM KCl, 1.1 mM Na₂HPO₄^.2H₂O, 0.4 mM KH₂PO₄, 5.5 mM glucose, 4 mM NaHCO₃, MES, pH 6.1, 2 mM EGTA and 2 mM MgCl₂). For actin staining, cells were incubated for 1 hr in 1:40 Rhodamine phalloidin (Molecular Probes) dissolved in 1% BSA/PBS solution. Cells were mounted on SlowFade (Molecular Probes). Confocal laser scanning microscopy was carried out using the Leica TCS Confocal System attached to a Leica DM RXA microscope with a 40× oil immersion objective and with Leica PowerScan software. The figures were created using Adobe Photoshop version 7.0.

Immunoprecipitation

Cells were treated with 1 μM TPA for 10 min, harvested and lysed in 500 μl lysis buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.5% NP-40, 5 mM EDTA, 1mM Na₃VO₄, 1 mM Pefablock (Roche)). Approximately 1 × 10⁷ cells were used. The lysate was added 30 μl of Protein A/G Plus-Agarose (Santa Cruz) and incubated for 1 hr at 4°C. After centrifugation at 5,000 rpm for 3 min, the supernatants were collected and incubated for 4 hr at 4°C with specific antibody (2.5 μg), and 50 μl of Protein A/G Plus-Agarose blocked with 1% BSA was added and incubated for 16 hr at 4°C. The samples were centrifuged as mentioned earlier, washed 4 times with 1× PBS and subjected to SDS-PAGE and immunoblotting.

Results

Suppressor screen for Ki-Ras-mediated morphological transformation

To identify proteins that regulate Ras-mediated morphological transformation, we performed a retroviral suppressor screen. 3T3 cells were infected with a retroviral mouse fibroblast cDNA fragment library followed by transduction with a retroviral vector carrying activated Ki-ras and the puromycin resistance gene. Activated Ki-ras expression in 3T3 cells induces a morphological transformation characterized by cell rounding and foci formation. The dual-infected cells were screened for colonies with normal fibroblastic morphology (flat, spread) as identified by light microscopy (Fig. 1a). These clones were presumed to express cDNAs that could oppose the Ras-signaling pathway that leads to transformation. Several independent subclones were established (designated 3T3 K1-70/Ki-Ras). One cell clone (3T3 K66/Ki-Ras) was chosen for further study. The K66 cDNA sequence from the murine cDNA library showed high similarity to the mouse G_β gene and to the G_β-homologue receptor for activated protein C-kinase (RACK1). Our sequence alignments identified K66 as murine RACK135 presenting a 150 nucleotide deletion of the 5′-end. The gene fragment encoded a potential protein product that constituted amino acids 50–317 of RACK1, thus lacking the first WD40 repeat, and displayed complete sequence identity to both human and rat RACK1 (Fig. 1b). In the following, we therefore refer to this gene fragment as RACK1ΔWD1. Northern blot showed that the RACK1ΔWD1 cell clone expressed a specific RACK1ΔWD1 band not found in other cell clones, in addition to a wild type RACK1-specific band (Fig. 1c).

Figure 1. Identification of a truncated RACK1 mRNA from retroviral cDNA-library-expressing 3T3 cells challenged with Ki-ras. (a) Light microscopy photographs of living 3T3 cells infected with retrovirus carrying lacZ (vector/lacZ), RACK1ΔWD1, RACK1ΔWD1/Ki-ras, or Ki-ras. Pictures were captured using a Leica FFF inverted microscope (×10 objective). (b) Alignment of the amino acid sequences of murine RACK1ΔWD1, human RACK1 (HuRACK1) and rat RACK1. The sequenced PCR product of RACK1ΔWD1 was translated in all reading frames, identified using pBLAST and aligned to related sequences using ClustalX. (c) RNA purified from 3T3, 3T3 lacZ (vector/lacZ), 3T3 RACK1ΔWD1/Ki-Ras, 3T3 K57/Ki-Ras and 3T3 K37/Ki-Ras cells was subjected to gel electrophoresis and RNA blot using a ³²P-labelled antisense RACK1ΔWD1 probe. (d) Growth curves of 3T3, 3T3 RACK1ΔWD1/Ki-Ras, 3T3 RACK1ΔWD1-2/Ki-Ras and 3T3 Ki-Ras cells. 5,000 cells of each cell line were seeded onto 6-cm dishes at day 0. Cells from 2 parallel dishes were trypsinized, and the number of cells was determined at the indicated number of days. Cells were medium changed every 4th day. The experiment was repeated 3 times and a representative experiment is shown.

To demonstrate that the reverted 3T3 RACK1ΔWD1/Ki-Ras phenotype was due to expression of the RACK1ΔWD1 cDNA, RACK1ΔWD1 cells were transfected with gag-pol and env expression plasmids, and 3T3 cells were infected with the rescued RACK1ΔWD1 and Ki-ras/puro retroviruses. The resulting puromycin-selected cells (RACK1ΔWD1-2/Ki-Ras) displayed a flat, nontransformed morphology. Growth rate experiments showed that the cell doubling time was similar in control and Ki-Ras transformed cells, but that the growth properties and number of cells at confluence was markedly affected by Ki-Ras (Figs. 1a and 1d). At high cell density, cells coexpressing RACK1ΔWD1/Ki-Ras or RACK1ΔWD1-2/Ki-Ras showed growth arrest similar to control cells (Fig. 1d). 3T3 cells stably transfected with wild type RACK1 (wtRACK1) showed no sign of morphological transformation but cultures maintained a higher cell density (˜60%) at confluence, when compared with untransfected 3T3 cell (data not shown). Overexpressed wtRACK1 was not able to induce growth arrest in Ki-Ras transformed cells (data not shown). These results demonstrate that RACK1ΔWD1-expression restores contact inhibition in activated Ki-Ras-expressing cells.

To evaluate specific protein expression of RACK1ΔWD1 and wtRACK1, the respective sequences were C-terminally tagged with the V5 epitope. Using antibodies to this epitope, we showed that 3T3 cells transduced with RACK1ΔWD1 expressed a 31.5-kD protein product (Fig. 2a, upper panel). The lower panel in Figure 2a shows the expression of endogenous RACK1 and RACK1-V5 using monoclonal antibodies to RACK1. The RACK1 antibody did not detect RACK1ΔWD1. To determine the effect of RACK1ΔWD1-expression on cell–cell contact, inhibition and foci formation, stably transfected 3T3 cell lines were cultured for 16 days and were stained with Gimsa. Representative images of the transformation experiments are shown in Figure 2b. No foci formation was observed in untransfected cell cultures or in cells expressing exogenous RACK1 or RACK1ΔWD1, while an average of 30.3 foci per dish were found in control Ki-Ras-expressing 3T3 cell cultures (Fig. 2c). In Ki-Ras cells coexpressing RACK1ΔWD1, the number of foci was reduced to an average of 4.6 per dish in 3 individual experiments. wtRACK1/Ki-Ras cells showed an intermediate level of foci formation. Taken together, these results indicate that RACK1ΔWD1 is a negative regulator of Ki-Ras signaling and transformation.

Figure 2. Effect of RACK1ΔWD1 and RACK1 expression on foci formation in 3T3 cells transformed with mutated Ki-ras. Stable cell lines coexpressing Ki-Ras with the carrier vector wzlHygro (vector), RACK1ΔWD1-V5 or RACK1-V5 were seeded on 6-well dishes, 5,000 cells per dish. At day 16, cells from 1 dish were lysed and equal amounts of protein were subjected to SDS-PAGE and immunoblotting using an antibody against the V5-epitope or against wtRACK1 (M_w 36 kDa) (a). The molecular weight of RACK1-V5 and RACK1ΔWD1-V5 were predicted to be 37.3 and 31.5 kDa, respectively. Four dishes in parallel were Gimsa-stained, pictures were taken using a Leica M420 microscope at 35 × 1.0 resolution (b), and the number of foci per dish was determined (c). The results are representative of 3 experiments.

RACK1ΔWD1 prevents actin depolymerization in Ki-Ras-expressing cells

The change in cell morphology of transformed 3T3 Ki-Ras cells was accompanied by loss of actin fibers compared with control cells (Fig. 3, panels a and b). 3T3 cells stably expressing RACK1ΔWD1 alone maintained a flat shape with well-defined actin fibers (Fig. 3, panel c). Strikingly, in 3T3 RACK1ΔWD1/Ki-Ras cells, well-defined actin stress fibers were restored concomitant with increased cell spreading (Fig. 3, panel d).

Figure 3. Coexpression of RACK1ΔWD1 restores actin stress fibers in Ki-Ras transformed 3T3 cells. Control 3T3 cells (a), Ki-Ras (b), RACK1ΔWD1 (c), and RACK-1ΔWD1/Ki-Ras-expressing 3T3 cells (d) were grown on coverslips, fixed in 4% formaldehyde in C-buffer, and analyzed by confocal microscopy using Rhodamin-falloidin to visualize the actin fibers. The inset in panel B shows the actin pattern and morphology changes in 3T3 Ki-Ras cells after 16 hr treatment with 50 μM PD98059; inset is with 1/2 magnification. Similar results were obtained in 4 independent experiments.

It has previously been shown that Ki-Ras recruits Raf to the plasma membrane,13 and thus, it is believed that Ki-Ras predominantly activates the Raf-MEK-ERK (MAPK) pathway. To test this, we treated 3T3 Ki-Ras cells with PD98059, a MEK inhibitor. The inhibitor restored both the cell morphology and actin fiber structure (Fig. 3, inset in panel b). In contrast, the PI3-K inhibitor Wortmannin neither counteracted the Ki-Ras induced loss of F-actin nor reverted cells to nontransformed morphology (data not shown). This shows that loss of stress fibers in cells expressing mutated Ki-Ras is linked to MAPK pathway activation, but not the PI3K pathway. Therefore, it is possible that RACK1ΔWD1 restores actin fibers by inhibiting MAPK pathway signaling.

Ki-Ras-induced ERK activity in confluent cells is inhibited by RACK1ΔWD1

As we had shown that inhibition of the MAPK pathway could counteract Ki-Ras induced morphological transformation, we were interested in whether the expression of truncated RACK1 influenced ERK activation. To test this, we determined the phophorylation status of ERK1/2 using antibodies specific for its phosphorylated forms (pERK1/2). In cells grown to confluence and subsequently serum starved for 16 hr, pERK levels were substantially reduced in RACK1ΔWD1/Ki-Ras cells compared with Ki-Ras cells (Fig. 4a, upper panel, lanes 2 and 4). In addition, the basal (serum induced) pERK level in control 3T3 cells was reduced by expressing RACK1ΔWD1 (upper panel, lane 1 and 3). Staining with antibodies to total ERK1 showed that equal amount of protein was loaded in each well (Fig. 4a, middle panel). The expression of RACK1ΔWD1 and Ki-Ras protein was confirmed by staining with anti-V5 (Fig. 4a, lower panel). Thus, expression of N-truncated RACK1 inhibited Ki-Ras-mediated phosphorylation of ERK. This was also demonstrated by immunofluorescent staining of serum-deprived cells using the pERK-specific antibody (Fig. 4b). In Ki-Ras transformed cells, a high pERK level was detected in the cytoplasm and nucleus compared with that of control cells (Fig. 4b, panels A and E). In RACK1ΔWD1 cells, a weak pERK signal was observed (panel B). In the cells coexpressing RACK1ΔWD1 and Ki-Ras, the nuclear pERK signal was completely blocked and the cytoplasmic signal was strongly reduced compared with Ki-Ras transformed cells (panel E and G).

Figure 4. The effect of RACK1ΔWD1 on basal ERK activity in 3T3 cells. (a) 3T3 cells with retroviral carrier vector, 3T3 Ki-Ras, 3T3 RACK1ΔWD1 and 3T3 RACK1ΔWD1/Ki-Ras grown to confluence were serum starved for 16 hr and lysed. Equal amounts of protein were subjected to SDS-PAGE and immunoblotting was performed using anti-pERK1/2 (upper panel) combined with peroxidase-labelled anti-mouse IgG and ECL-detection, followed by stripping and probing for total ERK1 and 2 using anti-ERK1 (middle panel). Incubation with anti-V5 in a parallel experiment confirmed the expression of the Ki-Ras and RACK1ΔWD1 protein products (lower panel). Results representative of 3 experiments are shown. (b) 3T3 cells with retroviral carrier vector (panels A and B), 3T3 RACK1ΔWD1 (panels C and D), 3T3 Ki-Ras (panels E and F) and 3T3 RACK1ΔWD1/Ki-Ras (panels G and H) were grown on coverslips for 24 hr, serum starved for 16 hr and treated with acetone (Control; A, C, E and G) or with 2 nM TPA for 10 min (B, D, F and H). Cells were fixed and immunofluorescence was performed using anti-pERK1/2 combined with Alexa568-labelled anti-mouse IgG. Confocal laser scanning microscopy was carried out using the Leica TCS Confocal System attached to a Leica DM RXA microscope with a ×40 oil immersion objective and with Leica PowerScan software.

PKC, one of the main RACK1 interaction partners, is implicated in the activation of the ERK/MAPK pathway.36 To further study the effect of RACK1ΔWD1 on ERK activation, cells were treated with PKC-activating phorbol ester (TPA). Treating 3T3 cells for 10 min with 2 nM TPA induced a strong pERK nuclear signal in control cells (Fig. 4b, panels A and B). Expression of RACK1ΔWD1 abolished this activation (Fig. 4b, panels C and D). Thus, RACK1ΔWD1 substantially inhibited PKC-mediated ERK activation in response to TPA treatment. In Ki-Ras cells, TPA-treatment led to additional ERK activation as seen by the strong nuclear staining (Fig. 4b, panels E and F). In RACK1ΔWD1/Ki-Ras cells, some cytoplasmic ERK phosphorylation was observed but no translocation to the nucleus took place during the test period (Fig. 4b, panels G and H). These experiments demonstrate that RACK1ΔWD1 interferes with both mutated Ki-Ras- and TPA-induced activation of the MAPK pathway and blocks the nuclear accumulation of pERK.

RACK1 and RACK1ΔWD1 display different cellular localization patterns but are able to form similar protein-protein interactions

RACK1 is involved in facilitating specific subcellular localization of a range of proteins involved in signal transduction.24 It has previously been shown that during stimulation of CHO cells, RACK1 interacts with PKCβII followed by transport of the complex to the plasma membrane.37 In agreement with these results, we observed that in RACK1ΔWD1-expressing cells, endogenous wtRACK1 translocated towards the plasma membrane in response to 40 min high dose (1.7 μM) TPA treatment (Fig. 5a, panels A and B). Also, wtRACK1 accumulated in the typical cell protrusions seen in response to TPA treatment (Fig. 5a, panel B). Interestingly, with low dose (6 nM) TPA treatment, morphological cell change appeared much later in RACK1ΔWD1 cells than in control cells (at approx. 40 min compared with 20 min in controls, data not shown). N-terminally truncated RACK1 protein was diffusely distributed in the cytoplasm with weak nuclear staining, indicating that it was not anchored to cytoskeleton or membranous structures to the same degree as wtRACK1 (Fig. 5a, panel C). Importantly, no change in RACK1ΔWD1 localization was seen in response to TPA (Fig. 5a, panel D). This suggests that correct localization and translocation of RACK1 is dependent on the first WD repeat.

Figure 5. (a) Localization of endogenous RACK1 and RACK1ΔWD1 in RACK1ΔWD1-expressing 3T3 cells. 3T3 RACK1ΔWD1 cells were grown on coverslips. Cells were serum starved for 16 hr prior to 30 min stimulation with 1.7 μM TPA and were fixed in 4% formaldehyde in C-buffer. Immunofluorescent staining was performed using anti-RACK1 or anti-V5 combined with Alexa568-anti-mouse secondary antibodies. Confocal laser scanning microscopy was carried out using the Leica TCS confocal system attached to a Leica DM RXA microscope with a ×40 oil immersion objective, and with Leica PowerScan software. (b) Coimmunoprecipitation of RACK1ΔWD1 with RACK1, PKCα, βI and δ. Immunoprecipitation (IP) was performed on cell lysate from 3T3 cells with retroviral carrier vector (vector), RACK1ΔWD1-expressing 3T3 cells (RACK1ΔWD1), and RACK1-expressing 3T3 cells (RACK1) using anti-V5 and agarose A/G beads. Cells were serum starved for 16 hr and treated with 1.0 μM TPA for 10 min prior to harvesting. Immunoprecipitated proteins were detected by immunoblotting (IB) with a combination of anti-V5 and anti-RACK1 to detect both wtRACK1 and the V5-tagged proteins (upper panel), or with anti-PKCα, βI or δ (panel 2, 3 and 4), respectively.

To determine whether truncated RACK1 was able to make the same protein–protein interactions as wtRACK1, we expressed the V5-tagged proteins in 3T3 cells and used the V5-antibody to isolate the interaction complexes by immunoprecipitation. The experiments indicated that there was an interaction between RACK1 and RACK1ΔWD1, because antibodies to V5-tagged RACK1ΔWD1 and RACK1 coimmunoprecipitated endogenous wtRACK1 (Fig. 5b, upper panel). RACK1 has previously been reported to form dimers through a postulated dimerization site in WD4,38 and Chu and coworkers demonstrated that RACK1 dimerizes upon reaction with GTP-K(B)-Ras(Q61K).25 The coimmunoprecipitation could also reflect that both proteins are able to bind components of a larger multiprotein complex. The majority of known RACK1-interactions take place through WD40 repeats3–7, and these could still be structurally intact in the N-terminally truncated protein. As shown in the 3 lower panels of Figure 5b, the PKC isoforms α, βI and δ were coimmunoprecipitated with both RACK1-V5 and RACK1ΔWD1-V5, indicating that the RACK1 deletion mutant is able to bind to RACK1 interaction partners.

The first panel of Figure 5b shows that the anti-V5 antibody pulled down more wtRACK1-V5 than RACKΔWD1-V5 because of the difference in expression levels. Thus, the high relative level of PKCβ1 and δ immunoprecipitated by RACKΔWD1, compared with wtRACK1, could indicate that the truncated protein has a preference for these isoforms. Specific inhibitors to PKCα, βI or δ did not allow identification of the PKC isoform that contributes most to the ERK activation (data not shown). Thus, PKCα, βI and δ may each have an important contribution to the activation of ERK in Ki-Ras transformed cells.

RACK1ΔWD1 blocks the translocation of RACK1 interacting proteins

We have shown that both wtRACK1 and truncated RACK1 can take part in signal transduction complexes containing different PKC isoforms (Fig. 5b). In addition, the localization of truncated RACK1 was shown to be severely disrupted (Fig. 5a). Thus, the inhibition of Ki-Ras mediated signaling and of ERK activation in RACK1ΔWD1-expressing cells could be caused by defects in the correct localization of RACK1 interacting proteins. To investigate this further, we assayed the cellular localization of endogenous PKCα, βI and δ in the presence of RACK1ΔWD1 in cells treated with 6 nM TPA. Control cells transduced with the retroviral carrier vector accumulated PKCα at the plasma membrane, especially at lamellipodia, and formed typical TPA-induced cell protrusions (Fig. 6a, panels A and B). In RACK1ΔWD1 cells, the translocation of PKCα to the plasma membrane and lamellipodia was reduced, and the morphological change in response to TPA was inhibited (Fig. 6a, panels C and D). PKCδ showed a distinct translocation to the nucleus in response to TPA, and this nuclear accumulation was significantly reduced in RACK1ΔWD1 cells (Fig. 6b). In wild type cells, no clear PKCβI-translocation was observed in response to TPA, and thus the effect of RACK1ΔWD1 could not be studied (data not shown). These results indicate that expression of N-terminally truncated RACK1 affects translocation of activated PKCα and δ and has an impact on their downstream effects, as demonstrated by the lack of TPA-induced change in cellular morphology. The reduction in PKC translocation by RACK1ΔWD1-expression is one possible explanation to its inhibitory effect on Ki-Ras induced ERK activity and cellular transformation.

Figure 6. Translocation of PKCα and δ in response to TPA. 3T3 stable cell lines with retroviral carrier vector or expressing RACK1ΔWD1 were grown on coverslips. Cells were serum starved for 16 hr prior to treatment with 6 nM TPA for 20 min, and subsequent fixation in 4% formaldehyde in C-buffer. (a) Immunofluorescent staining was performed using anti-PKCα combined with FITC-anti-rabbit secondary antibodies. Arrows indicate examples of PKCα-accumulation in cell–cell contact areas and lamellipodia in TPA-treated cells. (b) Immunofluorescent staining using antibodies to PKCδ combined with FITC-anti-rabbit secondary antibodies. Arrows indicate the localization of nuclei in the cells. Confocal laser scanning microscopy was carried out using the Leica TCS Confocal System attached to a Leica DM RXA microscope with a ×40 oil immersion objective and with Leica PowerScan software.

Discussion

Using a retroviral cDNA screen, we identified a cDNA fragment encoding an inhibitor of activated Ki-ras-mediated morphological cell transformation. Alignment analyses identified the cDNA fragment as a 5′-truncated version of RACK1.27, 39 This suggests that RACK1 may be part of a Ras-mediated signal leading to morphological cell transformation. Many of the known Ras effectors have been isolated based on their direct interaction with Ras by the yeast two-hybrid system. The retroviral cDNA screen is not restricted to proteins or polypeptides that interact directly with Ras. Our approach therefore provides us with a very powerful tool for the identification of gene products or parts of gene products that interfere negatively with activated Ki-Ras-mediated morphological transformation.

The identification of a RACK1 fragment as a possible inhibitor of Ki-Ras-mediated morphological transformation is intriguing. RACK1 has a central position as a scaffold protein and is implicated at several levels of signal transduction ranging from events at the plasma membrane receptors40, 41, 42, 43, 44 to interactions with transcription factors45 and regulated translation of mRNA.46, 47, 48 Our present findings place RACK1 as an important factor in a Ras-initiated signal regulating cell morphology and actin organization. The MEK inhibitor PD98059 restored stress fiber formation in Ki-Ras transformed cells, demonstrating the involvement of MAPK signaling in Ki-Ras-induced loss of actin fibers (Fig. 3). RACK1ΔWD1-expression attenuated Ki-Ras induced ERK phosphorylation and rescued actin fibers in Ki-Ras transformed cells similar to the MEK-inhibitor. This suggests that the observed effect on cell morphology by RACK1ΔWD1 is generated at least in part through inhibition of the ERK/MAPK pathway. Figure 4b indicates that RACK1ΔWD1 in particular inhibits the nuclear entry of pERK. Ki-Ras is the dominating Ras-oncogene in colon cancer.49 It has been suggested that higher activity of the MAPK pathway requires elevated levels of RACK1, which could explain the upregulation of RACK1 expression in colon cancer.50, 51 Our results showed that RACK1 upregulation in itself was not sufficient to increase MAPK activation.

The first propeller of the postulated RACK1 structure has so far only been implicated in the interaction with the transducin heterotrimeric G-protein. Transducin G_αG_βG_γ and G_βG_γ are believed to interact with sequences in RACK1 propeller 1 and 2 through WD–WD binding to the G_β subunit.52 A splice variant of Ha-Ras with unknown function, p19, has been shown to interact with RACK1.53 And recently, an interaction between RACK1 and S-ras(Q61K) of the shrimp Penaeus japonicus was shown, where RACK1 dimers reacted with GTP-K(B)-Ras, but not with GDP-K(B)-Ras.25 Thus, it is intriguing to speculate that RACK1ΔWD1 is unable to take part in a direct interaction with Ki-Ras, and that a RACK1/RACK1ΔWD1 heterodimer could inhibit the Ki-Ras signal.

Although most of the domains of wtRACK1 known to be important for protein interaction are theoretically intact in the RACK1ΔWD1 protein, the localization pattern is severely disrupted (Fig. 5a). This could potentially sequester a subpopulation of RACK1-interaction partners at wrong locations in the cell. As RACK1 has a central role in regulating cPKC biological activity,28, 54, 55 it is likely to be involved in PKC-dependent signal transduction during Ras-signaling. To further investigate this possibility, we studied the translocation of known RACK1-interacting PKC isoforms in response to TPA. We demonstrated that RACK1ΔWD1 interacted with complexes containing the PKC isozymes α, βI and δ (Fig. 5b), and had a negative effect on PKCα and δ translocation in response to TPA (Fig. 6). This suggested that RACK1ΔWD1 could be linked to disrupted or aberrant PKC activity. In this context, it cannot be excluded that RACK1ΔWD1 may have a preserved specificity for interaction with the Ki-Ras-Raf-PKC pathway.

The relatively low expression level of RACK1ΔWD1 was not expected to block all RACK1 interactions (see Fig. 2a), and immunofluorescence showed that expression of the deletion mutant had no visible effect on the translocation of wtRACK1 in response to TPA (Fig. 5a, panel B). Nevertheless, RACK1ΔWD1 inhibited both mutated Ki-Ras and low dose TPA-induced ERK activation. Thus, RACK1ΔWD1 is likely to give its effect through an important subset of molecular events organized at RACK1-containing complexes. It is possible that the presence of truncated RACK1 in signaling complexes could divert the signal to an alternative, growth-inhibiting pathway. For instance, JNK interacts with Raf and MEK in a positive feedback loop on the oncogenic Ras signaling pathway56 and RACK1 serves as an adaptor for PKC-mediated JNK activation.57 Thus, it is evident that the possibility for substantial cross talk between different transforming pathways exists, with RACK1 as a major link.

In summary, the N-terminally truncated form of RACK1 attenuates Ki-Ras signaling. We cannot conclude whether this is due to an effect as a gain-of-function or a dominant negative regulator. However, our data support the notion that RACK1 is an important organizing link between different signaling pathways. We have previously shown that interaction between RACK1 and PKCα is necessary for phospholipase D (PLD) activity, since a peptide inhibitor of RACK1-PKCα binding inhibits PLD activity in response to platelet-derived growth factor (PDGF).28, 58 PLD facilitates the degradation of phosphatidylcholine (PC) to phosphatidic acid (PA), which is necessary for the membrane binding of Raf during Ras-mediated activation. We here show that a complex containing RACK1, truncated RACK1 and the PKC isoforms α, β1 and δ may be less efficient in transferring an activated Ras signal to the Raf-MEK-ERK pathway (Fig. 7). Important is the observed inhibition of PKC translocation in RACK1ΔWD1-expressing cells. Thus, a number of the components believed to contribute to morphological cell transformation physically interact with RACK1 and may transiently exist as one functional complex in which 2 RACK1 molecules are present.

Figure 7. Proposed model of Ki-Ras signaling through ERK requiring RACK1 and PKC; the possible inhibitory role of RACK1ΔWD1. We have previously shown that interaction between PKC and RACK1 is essential for PLD activity in mouse fibroblasts. When activated, PLD cleaves phosphatidylcholine (PC) to phosphatidic acid (PA), which facilitates Raf binding to the membrane. We envision a structural change in the RACK1 complex where PLD is either repositioned or released, manifesting an accessible PA site in the plasma membrane. RACK1 dimers have been shown to bind to activated Ras, and Ras will then recruit Raf to the PA-site in the plasma membrane. As shown by others and us, the activation of the Raf-MEK-ERK pathway by Ras is dependent on PKC activity. We propose that inhibition of Ki-Ras signaling by truncated RACK1 (RACK1ΔWD1) is attained by either direct interference with the function of the described complexes at the plasma membrane or activation of alternative, growth inhibiting pathways perhaps due to disrupted, cytoplasmic localization of PKC signaling. Both scenarios result in the block of Ki-Ras/PKC-induced phosphorylation and nuclear entry of ERK and thus lead to inhibition of cell growth.

Acknowledgements

The authors wish to thank Hildegard Kanestrøm, Ingeborg Hageberg and Bjørg Flatekvål for their devoted technical assistance.