Alterations in the ErbB family of growth factor receptors, their signaling components, and mutational activation of Ras proteins are major contributors to malignant transformation. Recently, mutant Ras was shown to be capable of activating ErbB receptors in a ligand-independent manner. Furthermore, it was observed that nucleolin, a transcriptional regulator and ribosome biogenesis factor, can bind both K-Ras and the cytoplasmic tail of ErbB receptors to enhance ErbB receptor activation. However, the functional significance of these interactions to cancer pathogenesis has not been probed. Here, we show that endogenous nucleolin interacts simultaneously in vivo with endogenous Ras and ErbB1 (EGFR) in cancer cells. The C-terminal 212 amino acids of nucleolin were determined to be sufficient to interact with ErbB1 and all Ras protein isoforms (H-, N-, and K-Ras). Nucleolin partially colocalizes with Ras at the plasma membrane. Moreover, activated but not wild-type Ras facilitates nucleolin interaction with ErbB1 and stabilizes ErbB1 receptor levels. Most importantly, these three oncogenes synergistically facilitate anchorage-independent cell growth in vitro and tumor growth in vivo. Our findings suggest strategies to target nucleolin as a general approach to inhibiting ErbB- and Ras-driven cancers. Cancer Res; 71(6); 2140–51. ©2011 AACR.
Receptor tyrosine kinases (RTK) regulate cell proliferation and differentiation when activated upon binding to their respective growth factor ligands, and when deregulated may lead to the development of cancer (1, 2). The ErbB subfamily of RTKs contains 4 members: ErbB1 (EGFR, HER1), ErbB2 (Neu, Her2), ErbB3 (HER3), and ErbB4 (HER4), which have been strongly implicated in the development of human cancers (2). Activation of ErbB receptors induces receptor dimerization, which transmits a signal across the plasma membrane and activates the intracellular tyrosine kinase domain (3, 4). This in turn leads to transautophosphorylation on tyrosine residues that function as docking sites for signal transducing molecules (5). Among the most prominent signaling pathways activated by ErbB receptors is the Ras/MAPK (mitogen activated protein kinase) cascade (6). Activated Ras can induce the release of growth factors including EGF-like ligands, such as amphiregulin, TGFα, HB-EGF, and NRG (7, 8), thereby inducing receptor phosphorylation via an autocrine or paracrine loop. Constitutively activated Ras can also induce ErbB phosphorylation in a ligand-independent manner (9). Ras signaling is aberrantly activated in a large fraction of human tumors (1, 10). Because Ras signaling represents a convergence point for numerous diverse extracellular signals, Ras and its effectors are appropriate targets for therapeutic intervention. Abnormal function of the ErbB receptors and their ligands is also involved in human cancer, and these receptors are currently targets of an increasing number of cancer drugs (2, 11).
Recently, we have shown that ErbB1 interacts with the multifunctional protein nucleolin (12), a ubiquitously expressed acidic phosphoprotein with key functions both in transcription and in the synthesis and maturation of ribosomes (13, 14). Nucleolin is thus involved in critical aspects of transcriptional regulation, cell proliferation, and cell growth (15, 16). Nucleolin was originally identified as a nuclear protein localizing primarily to the nucleoli but is now appreciated to undergo nuclear-cytoplasmic shuttling and to also be present on the cell surface of some types of cells (14, 17, 18). Inhibition of cell-surface nucleolin and nucleolin activities suppresses cell and tumor growth (13, 19–21) in breast, prostate, and glioma cell lines, which also express high levels of ErbB receptors and/or activated Ras protein. An aptamer-targeting nucleolin, AS1411, is in phase II clinical trial for relapsed/refractory acute myeloid leukemia (22). We recently identified non-nuclear nucleolin as an ErbB receptor–interacting protein (12). This interaction leads to receptor dimerization and activation as well as to colony growth in soft agar. The ErbB1 juxtamembrane region containing the nuclear localization sequence (NLS), which is important for ErbB1 kinase activation (23), is also important both for nucleolin–ErbB1 interaction and for nucleolin-induced ErbB1 dimerization and activation (24). In addition, we have shown that this effect is mediated by the C-terminal 212 amino acids of nucleolin (24). Thus, it was suggested that the cross-talk between nucleolin and ErbB proteins might be related to tumor growth.
In this study, we show a cross-talk between nucleolin, ErbB1, and Ras proteins. All 3 typical Ras isoforms (H-, K- and N-Ras) interact with nucleolin. The C-terminal 212 amino acids of nucleolin are sufficient for this interaction and colocalize with Ras at the plasma membrane. Moreover, we show that GTP-bound Ras strongly enhances nucleolin–ErbB1 interaction. Finally, we show that the expression of activated H-Ras(G12V), nucleolin, and ErbB1 synergistically enhances cell transformation, as evidenced by increased colony formation in soft agar and increased tumor volume in nude mice.
Materials and Methods
Materials and buffers
Human recombinant EGF was purchased from Boehringer Mannheim. Polyclonal anti-ErbB1 antibodies (Abs), monoclonal anti-phosphotyrosine, anti-GFP, and anti-nucleolin were purchased from Santa Cruz Biotechnology. Monoclonal anti-pan-Ras and polyclonal anti-Ras were purchased from Calbiochem. Monoclonal anti p-Erk were purchased from New England Biolabs Inc. Anti-β-actin and anti-H-Ras monoclonal antibodies were from Santa Cruz Biotechnology. Monoclonal anti-β-tubulin and all other reagents were from Sigma. GRO (AS1411) aptamer and CRO (inactive oligomer) were synthesized as previously described (13, 22). Salirasib (FTS), a farnesylcysteine mimetic, was a gift from Concordia Pharmaceuticals and was prepared for experiments as described previously (25). HNTG buffer contained 20 mmol/L HEPES (pH 7.5), 150 mmol/L NaCl, 0.1% Triton X-100, and 10% glycerol. Solubilization buffer contained 50 mmol/L HEPES (pH 7.5), 150 mmol/L NaCl, 1% Triton X-100, 1 mmol/L EGTA, 1 mmol/L EDTA, 1.5 mmol/L MgCl2, 10% glycerol, 2 mmol/L sodium vanadate, 1 mmol/L phenylmethylsulfonylfluoride, 10 mg/mL aprotinin, and 10 mg/mL leupeptin.
HEK-293, HEK-293T, MDCK, and U87 cell lines were grown in Dulbecco's modified Eagle's medium (DMEM) and DLD1 cell line was grown in RPMI-1640, supplemented with 10% FBS. Cells were transfected using jetPEI (Polyplus Transfection) or using calcium phosphate (CaPO4) precipitation. Cell lysates were prepared 48 hours following transfection as described. Preparation of HEK-293 cells stably expressing the ErbB-1 receptor was previously described (24). Expression vectors GFP-ErbB1, GFP-H-Ras (12V) or pcDNA3-H-Ras(12V), and DsRed-nucleolin were introduced by JetPeI transfection into MDCK cells either alone or together. The neomycin (G418)-resistant colonies were examined for ErbB-1, Ras, or nucleolin expression, and several colonies (3 clones from each combination) were selected for further analysis. NIH3T3 cells were obtained from American Type Culture Collection and routinely grown in a humidified incubator at 37°C with 5% CO2. Cells were grown in DMEM supplemented with 10% bovine serum (BS; Life Technologies No. 16170078), 62.5 μg/mL penicillin, and 100 μg/mL streptomycin (Hyclone Laboratories). NIH3T3 cells were transfected with the vector pZIP, or pZIP-H-Ras (12V), using Lipofectamine 2000 (Invitrogen), according to the manufacturer's directions, and geneticin-resistant cells were selected to be examined by Western blotting.
GFP-ErbB1 was provided by Marcello Erlich, Tel Aviv University, Israel. Expression vectors for pEGFP-nucleolin, DsRed-nucleolin (DsRed-NCL), and DsRed-nucleolin mutants were previously described (12, 24). Flag-nucleolin expression vectors were generously donated by both James A. Borowiec (New York University, NY) and Paula J. Bates (University of Louisville, Louisville, KY). N-Ras and N-Ras(13V) expression vectors were previously described (26). GFP-K-Ras, GFP-K-Ras(12V), GFP-H-Ras, and GFP-H-Ras(12V) were previously described (27).
Lysate preparation, immunoprecipitation, and immunoblotting
Cells were exposed to the indicated stimuli and solubilized in solubilization buffer. Lysates were cleared by centrifugation. For direct electrophoretic analysis, boiling gel sample buffer was added to cell lysates. For other experiments, Abs were first coupled to anti-mouse IgG agarose for 1 hour at room temperature; alternatively, anti-Flag M2 agarose beads (Sigma) were used. Proteins in the lysate supernatant were precipitated with aliquots of the beads–antibody complexes for 2 hours at 4°C. The immunoprecipitates were washed 3 times with HNTG, resolved by SDS-PAGE (28), and electrophoretically transferred to nitrocellulose membrane. Membranes were blocked for 1 hour in TBST buffer (0.02 mol/L Tris-HCl, pH 7.5, 0.15 mol/L NaCl, and 0.05% Tween 20) containing 6% milk, blotted with 1 μg/mL primary Abs for 2 hours, followed by 0.5 μg/mL secondary antibody linked to horseradish peroxidase. Immunoreactive bands were detected with the enhanced chemiluminescence reagent (Amersham Corp).
HEK-293T cells were plated on glass coverslips placed in 6-well plates at a density of 2 × 105 cells per well. Cells were transfected with or without pCDNA3-ErbB1 with GFP-K-Ras (12V) or GFP-H-Ras (12V) and DsRed, DsRed-NCL, DsRed-N-ter, or DsRed-212 nucleolin expression vectors. After 48 hours, cells were fixed in 4% paraformaldehyde at room temperature for 30 minutes. For triple staining, fixed slides were immunostained with rabbit anti-ErbB1 antibodies (1:100) followed by Alexa Fluor 405–labeled donkey anti-rabbit secondary antibodies (1:1,000; Invitrogen) as previously described (29). Cells were examined by fluorescent microscopy at 63× magnification with Zeiss 510 META confocal microscope.
Soft agar assay
Cells were seeded at a density of 1,000 cells per well in 96-well plates in DMEM containing 10% FBS. The cells were mixed with 0.05 mL (per well) of 0.33% Noble agar, and the mixture was poured onto a layer of 0.05 mL 1% Noble agar in DMEM containing 10% FBS. The upper layer of the agar was covered with 0.1 mL of culture medium. Assays were carried out at least 6 times. The number of colonies with size above 0.01 mm2 was estimated on day 8, using a binocular and light microscopy with the image analyzer program Image Pro-Plus. Results shown are the mean ± SD of 6 determinations from each of the clones used.
Tumor growth in nude mice
MDCK cells harboring nucleolin, ErbB1, H-Ras(12V), and their combinations were implanted into nude nice. Nude CD1-Nu mice (25–30 g) were housed in barrier facilities on a 12-hour light/dark cycle. Food and water were supplied ad libitum. On day 0, a total of 2 × 106 cells of each clone in 0.1 mL of PBS were implanted subcutaneously just above the right femoral joint. Tumor volumes were then monitored on the indicated days as described previously (30).
Analysis of nucleolin–Ras interactions
In our previous reports, we have shown that nucleolin interacts with each of the 4 ErbB receptors and that this interaction has functional consequences, as it induces ligand-independent receptor activation (12). In related studies, we found that constitutively active Ras also induces receptor phosphorylation in a ligand-independent manner (9). It was not known whether the 2 phenomena are related to each other, a possibility that might have a strong impact on cell transformation. We therefore investigated whether Ras can affect nucleolin–ErbB interaction. We first examined the ability of nucleolin to coprecipitate with Ras isoforms. Cells coexpressing Flag-nucleolin and GFP-tagged wild-type (wt) or constitutively active H-Ras or K-Ras were lysed and subjected to immunoprecipitation of nucleolin using anti-Flag Abs, followed by immunoblotting for Ras with anti-GFP Abs (Fig. 1A). As a control, cells were transfected with GFP-H-Ras(12V) and lysates were precipitated with anti-Flag Abs; no Ras precipitation was observed (Fig. 1A). Similarly, cells expressing wt or constitutively active N-Ras were immunoprecipitated with antinucleolin Abs to pull down the endogenous nucleolin. The blots were then exposed to anti-Ras Abs (Fig. 1B). As a control, lysates of cells expressing N-Ras(13V) were precipitated with beads in the absence of anti-nucleolin Abs (Fig. 1B). As shown in Figure 1A and B, all 3 Ras isoforms both in their wt and active forms were effectively coprecipitated with nucleolin, indicating that nucleolin can bind all Ras isoforms and that these interactions are independent of the GDP/GTP-bound states of Ras.
We next postulated that if nucleolin mediates the effects of Ras on ErbB receptors, then nucleolin will precipitate not only Ras but also ErbB. To this end, we examined HEK-293 cells stably expressing ErbB1 and transiently expressing each of the wt or active Ras isoforms. Cell lysates were subjected to immunoprecipitation with anti-nucleolin Abs and immunoblotted with anti-GFP, anti Ras, or anti ErbB1 Abs as indicated. As a control, cells were transfected with GFP only. We found that nucleolin precipitated each of the 3 Ras isoforms and also ErbB1 (Fig. 2A and B). Importantly, nucleolin precipitated the ErbB1 receptor more effectively when coexpressed with each of the constitutively active Ras isoforms than with the wt proteins (1.38 ± 0.03, 1.57 ± 0.24, and 1.88 ± 0.23-fold for K-Ras, H-Ras, and N-Ras, respectively; Fig. 2A and B). These results indicate that GTP-bound Ras increases the interaction between nucleolin and ErbB1.
To further study the interaction between the 3 proteins, cell lysates from HEK-293 cells stably expressing ErbB1 and transiently expressing the K-Ras(12V) isoform were subjected to immunoprecipitation with anti-Ras Abs and immunoblotted with anti-ErbB1, anti-Ras, or anti-nucleolin Abs (Fig. 2C). As shown, Ras precipitated nucleolin and ErbB1. Taken together, these results indicate that Ras interacts with nucleolin and ErbB1 and nucleolin interacts with Ras and ErbB1. Therefore, it seems that there is a cross-talk between ErbB1, nucleolin, and Ras, either directly or indirectly. We do not know whether these interactions occur simultaneously.
We then examined the interaction between ErbB1, nucleolin, and Ras in human cancer cell lines without ectopic expression of any of these proteins. We used 2 cell lines: the colon cancer cell line DLD-1, which endogenously expresses ErbB1, nucleolin, and mutant K-Ras(12V), and the glioblastoma cell line U87, which overexpresses ErbB1 and exhibits high Ras-GTP levels. Cell lysates were subjected to immunoprecipitation, using anti-nucleolin Abs, and immunoblotted with either anti-ErbB1 or anti-Ras Abs. As a negative control, lysates were precipitated with anti-GFP Abs. In both cell lines, nucleolin precipitated ErbB1 and Ras proteins (Fig. 2D). These results support the conclusion that nucleolin interactions with ErbB1 and Ras occur in cancer cells expressing these proteins endogenously.
The C-terminal 212-amino acid region of nucleolin (212C-ter) interacts with Ras at the plasma membrane and is required for nucleolin–Ras interactions.
We have previously shown that 212C-ter (scheme in Fig. 3A), which contains the RNA-binding domain 4 (RBD4, residues 498–645) and the glycine–arginine-rich C-terminal region (GAR, residues 645–710), is both required and sufficient for nucleolin interaction with ErbB1 (24). To further examine the interaction between Ras and nucleolin, we carried out coimmunoprecipitation experiments on HEK-293T cells cotransfected with activated Ras isoforms and GFP-nucleolin deletion mutants (Fig. 3B–D). K-Ras(12V), H-Ras(12V), and N-Ras(13V) each immunoprecipitated effectively GFP-nucleolin 212C-ter and less effectively the individual GFP-tagged-GAR or -RBD4 domains. In contrast, the GFP-tagged-N-terminal mutant of nucleolin lacking RBD4/GAR (residues 1–498) did not precipitate with any of the Ras isoforms. These results suggest that the C-terminal 212 amino acid residues of nucleolin are both essential and sufficient for nucleolin–Ras interaction, as they are for nucleolin–ErbB1 interaction.
Next we examined whether nucleolin or its isolated C-terminal 212 amino acids colocalize with Ras at the plasma membrane, a major site of the interaction between nucleolin and ErbB1 (12) and of Ras (27). To this end, we cotransfected HEK 293T cells with GFP-K-Ras(12V) or GFP-H-Ras(12V) and pDsRed empty vector, pDsRed-nucleolin, pDsRed-212C-ter, or pDsRed-N-ter (negative control). The subcellular localization of the labeled proteins was studied by dual fluorescent confocal microscopy. Consistent with our earlier studies (27), GFP-K-Ras(12V) was localized mainly to the plasma membrane and also to cytoplasmic vesicular compartments (Fig. 4). Full-length pDsRed-nucleolin (RFP-NCL) was mainly localized to the nucleus and also to the cytosol and plasma membrane, as expected (Fig. 4). We observed partially overlapping localization of GFP-K-Ras and pDsRed-nucleolin at the plasma membrane. As predicted from its ability to coimmunoprecipitate Ras (Fig. 3), a portion of DsRed-212C-ter overlapped with GFP-K-Ras at the plasma membrane (Fig. 4). Consistent with the fact that the NLSs of nucleolin reside in its N-terminus, the majority of DsRed-212C-ter was not preferentially localized to the nucleus but was distributed throughout the nucleus and the cytoplasm whereas DsRed-N-ter was localized exclusively to the nucleus, where no GFP-K-Ras(12V) was detected. Thus, GFP-K-Ras(12V) and DsRed-N-ter showed no colocalization (Fig. 4), as predicted by their lack of interaction when the cells were lysed (Fig. 3). We saw similar overlapping localization of nucleolin with H-Ras (Fig. 4). Next we examined the colocalization of the 3 proteins by conducting a triple-staining experiment. To this end, we cotransfected HEK-293T cells with GFP-K-Ras(12V), pCDNA3-ErbB1, and pDsRed empty vector, pDsRed-nucleolin, pDsRed-212C-ter, or pDsRed-N-ter. The subcellular localization of the labeled proteins was studied by triple fluorescent confocal microscopy. As shown, the ErbB1 (blue) colocalizes with GFP-K-Ras(12V) and DsRed-nucleolin or DsRed-212C-ter at the plasma membrane but not with pDsRed-N-ter (Fig. 4, bottom panel). Thus, we conclude the 3 proteins colocalize at the plasma membrane, strengthening the notion of a cross-talk between them. These results also indicate that nucleolin 212C-ter are important for the interaction with both ErbB1 and Ras.
All Ras isoforms reduce EGF-induced reduction at the levels of ErbB1
Previously, we have shown that overexpression of nucleolin reduces EGF-mediated ErbB1 degradation (12). It is not known whether active Ras similarly causes such a reduction. The strong association between nucleolin–ErbB1, and nucleolin–Ras, prompted us to examine possible relationships between Ras and the levels of ErbB1. We thus examined next the effect of EGF on receptor levels in ErbB1-expressing HEK-293 cells that were transfected with either GFP (control) or one of the activated Ras proteins: GFP-H-Ras(12V), GFP-K-Ras(12V), or N-Ras(13V). The transfectants were stimulated with EGF (100 ng/mL) for the indicated periods of time, and cell extracts were subjected to immunoblot analysis with anti-ErbB1 Abs. In agreement with previous reports, EGF induced in the control cells (GFP transfectants) a time-dependent decrease in ErbB1 levels (Fig. 5A). Each of the 3 activated Ras isoforms inhibited the reduction in ErbB1 receptor levels seen after EGF stimulation (Fig. 5A). These results indicate that activated Ras isoforms affect receptor levels, either by reducing receptor degradation or by increasing receptor synthesis/transport.
Next we examined the effect of Ras or nucleolin inhibition on ErbB1 receptor levels. DLD1 cells were treated with FTS (Salirasib) or GRO (AS1411) or both for 48 hours and then treated with EGF for the indicated time periods (Fig. 5B). As shown, inhibition of Ras with FTS and nucleolin with GRO resulted each in enhanced receptor degradation (Fig. 5B). EGF-induced receptor degradation was enhanced even more in cells treated with both drugs. Similarly, these combined effects of the drugs were reflected in the downstream signaling to Erk (Fig. 5B). Namely, inhibition of Ras and nucleolin together inhibited EGF-induced Erk activation even more than each of the treatments alone. The cross-talk between Ras and nucleolin was also observed in cells expressing H-Ras (12V). Unlike growth of control cells, growth of H-Ras (12V) expressing cells was strongly inhibited by GRO (AS1411) but not by the control oligonucleotide (CRO; Fig. 5C).
Coexpression of ErbB1, Ras, and nucleolin facilitates cell transformation and tumorigenicity
We next examined whether the impact of nucleolin and Ras on ErbB1 is manifested in mediating cell transformation. To enable transformation and in vivo experiments using stable cell lines, we prepared MDCK cells stably expressing either empty vector, nucleolin alone, nucleolin and ErbB1, H-Ras (12V) alone, nucleolin and H-Ras(12V), ErbB1 and H-Ras(12V), or the combination of all 3 proteins and selected 3 individual clones from each transfection for further use. Levels of expression of each of the proteins in typical clones are shown in Figure 6A. We subsequently examined the possibility that the 3 proteins synergistically enhance cell transformation by soft agar assays using the MDCK clones described above. Cells were plated in soft agar and maintained in culture for 8 days before quantitating the number and size of colonies able to grow in an anchorage-independent manner. Results of a typical experiment (Fig. 6A) showed that under the conditions tested, neither nucleolin nor H-Ras(12V) alone, or nucleolin and ErbB1 together, induced the formation of colonies significantly higher than the background level of the small colonies seen with parental MDCK cells. In contrast, expression of nucleolin and H-Ras(12V) or ErbB1 and H-Ras(12V) together induced the formation of much larger colonies than those formed by cells expressing each of the proteins alone (Fig. 6, P < 0.001 and P < 0.01, respectively). Most importantly, MDCK cells overexpressing all 3 proteins formed a dramatically higher number of large colonies in soft agar than clones expressing both nucleolin and H-Ras(12V) or ErbB1 and H-Ras(12V; Fig. 6, P < 0.001).
We then examined whether these effects are mediated by nucleolin–212C-ter region which is important for the interaction with ErbB1 and Ras. We prepared clones that express stably empty vector, nucleolin–212C-ter alone, nucleolin–212C-ter and ErbB1, H-Ras(12V) alone, nucleolin–212C-ter and H-Ras(12V), ErbB1 and H-Ras(12V), or the combination of all 3 proteins, and 3 individual clones from each transfection were selected (typical clones are shown in Fig. 6B). As shown, 212C-ter was sufficient to mediate the anchorage-independent growth effect when coexpressed with active Ras and ErbB1 (Fig. 6B). These results strongly suggest that the combined expression of the 3 proteins synergistically enhances cell transformation and that 212C-ter of nucleolin are sufficient.
To gain additional support for the possibility that nucleolin, H-Ras(12V), and ErbB1 synergistically enhance transformation, we carried out experiments in nude mice. The mice were xenografted with each of the MDCK cell lines expressing the various combinations of the proteins. The following groups of mice (6 per group) received 2 × 106 cells subcutaneously in the flank: Mock; nucleolin alone; H-Ras(12V) alone; nucleolin + ErbB1; nucleolin + H-Ras-(12V); or nucleolin, ErbB1, and H-Ras(12V). Tumor volumes were monitored for 25 days following injection (Fig. 7). Mock, nucleolin, or nucleolin + ErbB1 transfectants formed tumors that were only barely detectable, and H-Ras(12V) or nucleolin + H-Ras(12V) transfectants formed relatively small tumors. In contrast, cells expressing nucleolin, ErbB1, and H-Ras(12V) formed large tumors (Fig. 7, P < 0.01), supporting the conclusion that they synergize in induction of cell transformation and that all 3 proteins are required for maximal transforming activity.
Nucleolin is a major component of the nucleolus and is also found in other cellular compartments (31). This key protein is involved in various aspects of ribosome biogenesis, from transcriptional regulation to the assembly of preribosomal particles. One study described interactions between nucleolin and K-Ras in the nucleolus (32). Other reports suggest that it could also play an important role in nonnucleolar functions (14). Recently, we described a novel interaction between nucleolin and ErbB receptors at the plasma membrane and that this interaction enhances ErbB-mediated responses (12). In other experiments, we showed that constitutively active Ras isoforms induce ErbB4 receptor phosphorylation in a ligand-independent manner, suggesting a direct, intracellular, positive control of Ras or its signals over ErbB4 (9). Together, these studies raised the possibility of functional interactions between nucleolin, ErbB receptors, and Ras proteins. Here we show for the first time that cross-talk between the 3 proteins contributes to cell transformation.
Our experiments show that nucleolin is capable of interacting with all 3 Ras isoforms (H-Ras, K-Ras, and N-Ras). Previously, we found that H-Ras(12V) protein interacts specifically with galectin-1 (25, 33) and that K-Ras(12V) protein interacts specifically with galectin-3 (34). We also showed that these interactions are the driving forces that promote nanoclustering of both active H-Ras (35) and active K-Ras (36) proteins in the inner leaflet of the plasma membrane, processes known to promote robust digital signaling (37, 38). However, no escort proteins of N-Ras were reported. In this study, we have identified for the first time nucleolin as such an escort protein of N-Ras. Nucleolin is however less selective than galectins, first, because it binds all Ras isoforms and, second, because it binds both the GDP- and GTP-bound forms of Ras. Thus, nucleolin–Ras interactions are independent of the GDP/GTP conformations of Ras proteins. Clearly, then the interactions of Ras isoforms with nucleolin differ from those of H-Ras and K-Ras proteins with galectins because only the latter depend on Ras-GTP (33, 34). However, the GTP-binding state of Ras is important for the interactions of ErbB1 and nucleolin, as discussed below.
The 3 highly homologous Ras proteins differ mainly in their hypervariable (hvr) region upstream of the common C-terminal farnesyl cysteine carboxymethylester (39–42). Because all 3 interact with nucleolin, it is reasonable to assume that the C-terminal farnesylcysteine and/or the domains upstream of the hvr that they share in common may be involved in nucleolin–Ras interactions. Nucleolin was reported previously to interact with K-Ras4B but not with H-Ras in the nucleolus (32). In a study published during the final preparation of our manuscript, it was reported that nucleolin from postnuclear supernatants of BHK cells interacts with K-Ras but not with H-Ras (43). The basis for this discrepancy is currently unclear. The latter study showed that a GFP-fusion protein containing only the hvr domain of K-Ras4B was sufficient for interaction with nucleolin, consistent with their and our findings that nucleolin–Ras interactions are not affected by the GTP-binding status of Ras. Additional experiments are required to determine which regions in Ras are critical for the interactions with nucleolin.
Using mutational analysis, we found that the interaction between nucleolin and Ras is mediated specifically by the 212C-ter of nucleolin and not by the N-terminal domain of the protein. This interaction seems to occur predominantly at the plasma membrane since the nucleolin–212C-ter mutant overlaps there both with K-Ras(12V) and with H-Ras(12V). Interestingly, this very same domain is also important for nucleolin–ErbB1 interaction (24), suggesting that the nucleolin–212C-ter region is involved in the interactions of nucleolin with ErbB1 and Ras to enhance cell transformation and cancer development.
Nucleolin is a major nucleolar phosphoprotein of exponentially growing eukaryotic cells (14), and it is overexpressed in highly proliferating cells (44). Nucleolin was originally identified as a nucleolar protein but is now considered to be a shuttle protein present in the nucleolus and cytoplasm and at the cell surface (14, 17, 18). Interestingly, it is also found on the cell surface of a wide range of cancer cells and is used as a diagnostic marker for cancer and for the development of anticancer drugs (18, 20, 44).
The ErbB subfamily of RTK has been implicated, more than other growth factor receptors, in the development of several human cancers (2). Elevated levels of the ErbB1 (EGFR) and/or its cognate ligands have been identified as a common component of multiple cancer types and seem to promote tumor growth (45). Among the signaling pathways activated by the receptors is the prominent Ras/MAP kinase signaling cascade. Our study shows a new link between nucleolin, ErbB1, and Ras proteins, which may contribute to human cancer. Our results are consistent with a model whereby a cross-talk between the 3 proteins may contribute to human malignancies. Our model is based on several lines of evidence: (i) ErbB1 pulls down nucleolin (12), (ii) nucleolin pulls down both Ras and ErbB1, (iii) Ras pulls down ErbB1 and nucleolin, (iv) nucleolin, ErbB1, and Ras coimmunoprecipitate from lysates of the human cancer cell lines DLD1 and U87 that endogenously express all 3 proteins, and (v) coexpression of nucleolin, ErbB1, and oncogenic Ras enhances cell transformation, as evidenced by anchorage-independent growth in soft agar and by tumor growth in nude mice.
Whereas the interactions between nucleolin and Ras are independent of the Ras nucleotide-binding state, the interaction between nucleolin and ErbB1 is dependent on the GDP/GTP state of Ras, as shown by the fact that the interaction is strongly increased by Ras-GTP compared with Ras-WT. These novel findings of cooperation between 3 oncogenes are clearly manifested in cell transformation and are highly relevant to human malignancies.
As discussed above, the ErbB receptors are involved in many human cancers and are therefore attractive candidates for targeted therapy. Two major classes of anti-ErbB therapeutics are already in clinical use or are in advanced development. The first is composed of several Abs directed against the extracellular domain of ErbBs and the second of small-molecule tyrosine kinases inhibitors (11). Ras-targeted therapy is at a less advanced stage. One class of putative Ras inhibitors is the farnesyltransferase inhibitors which have exhibited limited antitumor efficacy (46). A second class consists of farnesylcysteine mimetics including Salirasib (46) which showed significant efficacy in phase I/II trials (http://www.concordiapharma.com). Finally, nucleolin binds G-rich oligonucleotide (GRO) aptamers that induce antiproliferative responses. One of the aptamers named AS1411 (formerly AGRO100 or GRO26B-OH) is currently being tested as an anticancer agent in phase II clinical trials (22). Using Salirasib and GRO to inhibit Ras and nucleolin, respectively, we show that ErbB1 levels and EGF-mediated Erk activation are reduced. These observations further support the notion that nucleolin and Ras can affect receptor activities.
Our results showing oncogenic cooperation between nucleolin, ErbB1, and Ras suggest that combined treatment may have stronger inhibitory effects than individual inhibitors. Perhaps, even more important is the finding that responsiveness to ErbB-targeted therapy varies between patients, with no clear explanations for the basis of these differential responses. Interestingly, more recent experiments have shown that approximately 30% of colon cancer patients do not respond to ErbB1 inhibition, especially when they express mutated K-Ras (47, 48). It is not yet completely known why colon cancer patients harboring mutant K-Ras do not respond to Erbitux or Tarceva, although constitutive activation of downstream survival pathways is almost certainly one important contributor. It is tempting to also speculate that a robust nucleolin-mediated ErbB1 activation by mutant K-Ras may reduce or prevent the inhibitory effect of the drug. Whether or not this is the case, our results provide a strong rationale to consider a strategy of treatment of colon (or lung) cancer patients expressing mutant K-Ras with combinations of Ras and ErbB1 inhibitors, nucleolin and ErbB1 inhibitors, or Ras and nucleolin inhibitors.
Disclosure of Potential Conflicts of Interest
P.J. Bates has financial interests (shareholder) in Antisoma PLC. The other authors disclosed no potential conflicts of interest.
We thank Dr. Geoffrey J. Clark, University of Louisville, for providing the pZIP and pZIP-H-Ras (V12) vectors.
This work was supported by the Cancer Biology Research Center of Tel Aviv University, Israel (R. Pinkas-Kramarski), by the Kauffman Prostate Cancer Research Fund (R. Pinkas-Kramarski), by the Yosef Sagol Fellowship for brain research in Tel Aviv University (K. Farin), by Grant 2005344 from the United States-Israel Binational Science Foundation (Y. Kloog and A.D. Cox), by the Prajs-Drimmer Institute for The Development of Anti-degenerative Drugs (Y. Kloog), and NIH grant R01CA122383 to P.J. Bates. Y. Kloog is an incumbent of The Jack H. Skirball Chair in Applied Neurobiology.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.