1. ¹Department of Internal Medicine I, University of Ulm, 89081 Ulm, Germany
2. ²Department of Surgery I, University of Halle-Wittenberg, 06097 Halle, Germany
3. ³Medizinische Klinik, Klinikum rechts der Isar, Technische Universitat Munchen, Ismaninger Str. 22, D-81675 München, Germany

Correspondence: RM Schmid, E-mail: roland.schmid@lrz.tum.de

Received 21 August 2002; Revised 19 February 2003; Accepted 19 February 2003.

Abstract

Signal transduction of the RET receptor tyrosine kinase is involved in developmental processes as well as in neoplastic transformation. Activation of RET initiates receptor autophosphorylation on specific tyrosines that act as docking sites for downstream signaling molecules. Using the cytoplasmatic part of RET as bait in a yeast two-hybrid screen, we identified a novel SH2 and SH3 domain containing adaptor protein previously termed Grap-2/Grf40/GrpL/GRID and its murine homologue as Gads/Mona, respectively. This protein, predominantly expressed in cells of hematopoietic origin, is involved in signaling downstream of the T-cell receptor and the receptor for monocyte colony-stimulating factor. Here, we show that Grap-2 is also expressed in neuroendocrine tumors and cell lines known to bear mutated forms of RET. Endogenously expressed RET and Grap-2 coimmunoprecipitate from lysates of a medullary thyroid carcinoma cell line. Grap-2 directly associates with RET in pull-down experiments using in vitro translated proteins. Overexpression of Grap-2 inhibits RET-induced NF-B activation, and cotransfection of Grap-2 significantly reduces focus formation induced by oncogenic RET in NIH 3T3 cells. Taken together, these results suggest that besides being involved in tyrosine kinase signaling in hematopoietic cells, Grap-2 plays a tissue-specific role as an inhibitor of RET mitogenic signaling.

Alterations of the RET receptor associated with human disease as well as analysis of knockout and transgenic animal models have revealed molecular mechanisms and pleiotropic effects of RET signaling (reviewed in Taraviras and Pachnis, 1999; Hansford and Mulligan, 2000; Jhiang, 2000; Schedl and Hastie, 2000). Activation of RET kinase activity is dependent on autophosphorylation of tyrosine residues located in the cytoplasmatic tail of the protein. This can be accomplished by neurotrophins belonging to the family of glial-cell line-derived neurotrophic factors (GDNF) through the formation of heterodimers between their high-affinity receptors and RET (reviewed in Baloh et al., 2000). Constitutive ligand-independent activation of RET causes different forms of human thyroid cancer (Hansford and Mulligan, 2000; Jhiang, 2000). In the dominantly inherited multiple endocrine neoplasia type2 (MEN 2A/B) and familial medullary thyroid carcinoma (FMTC) syndromes, activating point mutations of the RET gene induce malignant transformation of thyroidal C cells leading to the development of medullary thyroid carcinoma (MTC). Nearly all mutations observed in MEN 2A and FMTC result in RET dimerization because of replacement of specific cysteine residues located in the extracellular region. MEN 2B mutations mostly affect one hot spot in the kinase domain leading to changes in substrate specificity (Hansford and Mulligan, 2000).

Activation and subsequent autophosphorylation of the RET receptor is followed by the interaction with molecules containing SH2 or phosphotyrosine-binding (PTB) domains. This leads to the recruitment of multiprotein complexes at the activated receptor, and in turn, elicits a cascade of signaling events resulting in alterations of gene transcription (Hansford and Mulligan, 2000).

To identify further components of intracellular RET signaling, we used a yeast two-hybrid strategy. In our experiments performed with the cytoplasmic tail of human RET ranging from amino acids 718–1114 we identified a novel Grb2-related adaptor protein. Owing to its simultaneous identification by several groups (reviewed in Liu et al., 2001), this adaptor has acquired different names. It is here refered to as Grap-2 in accordance with the first report of the human homologue (Qiu et al., 1998). Grap-2 is distinguished from Grb2 by a proline-rich 120 amino-acid linker of an as yet unknown function located between the SH2- and carboxy-terminal SH3 domains. Expression of Grap-2 in human and mouse tissues was found to be restricted primarily to the cells of hematopoietic origin. Gene-targeting experiments (Yoder et al., 2001) and functional studies indicate involvement of Grap-2 in T-cell receptor (TCR) signaling via interaction with the adaptor proteins SLP-76 and LAT (Liu et al., 1999). In mononuclear cells, Grap-2 has been reported to negatively regulate signaling downstream of the macrophage colony-stimulating factor receptor (M-CSFR) (Bourette et al., 1998).

To verify Grap-2 expression in tissues or cell lines endogenously expressing RET, we first performed Northern blot analyses using the full-length human Grap-2 cDNA as a probe. In addition to Grap-2 RNA expression in Jurkat T cells, we obtained specific signals of approximately 4.2 and 1.8 kb in the MTC-derived human cell line, TT having a MEN 2A-type RET mutation at codon 634 (Carlomagno et al., 1995) (Figure 1a). To determine Grap-2 protein expression, a rabbit antiserum was generated using bacterially expressed and affinity-purified human Grap-2 as antigen. As shown in Figure 1b, a Grap-2 specific band of approximately 40 kDa can be detected in lysates of BON- and LCC18-cell lines derived from human gastropancreatic neuroendocrine tumors (Lundqvist et al., 1991; Evers et al., 1994) and in different human MTC specimens. This indicates that Grap-2 protein expression is not restricted to the hematopoietic cell lineage, and suggests an additional role for this adaptor in neuroendocrine-differentiated cells.

Figure 1.

Analysis of Grap-2 RNA- and protein expression in cell lines and tissues. (a) Per lane 30 g of total cellular RNA derived from the indicated cell lines was hybridized with an -[³²P]dCTP-labeled full-length Grap-2 cDNA probe. RNA was prepared by CsCl-gradient purification, resolved on 1.2% morpholino-propanesulfonic acid (MOPS)/formaldehyde agarose gels, transferred onto GeneScreen nylon membranes (NEN™Life Science Products) and crosslinked by UV irradiation. Equal loading was confirmed by Methyleneblue staining of membranes after blotting. Arrowheads indicate the two Grap-2-specific RNA bands of 4.2 and 1.8 kb. (b) For Grap-2 immunoblots, 15 g of whole cellular protein extracts derived from human neuroendocrine cell lines LCC18 (colonic carcinoid), BON (pancreatic carcinoid), TT (MTC) and different human MTC specimen (MTC1-3) were loaded per lane. The 40 kDa Grap-2 band is indicated by an arrowhead. A Grap-2-specific rabbit antiserum was generated by immunization with a bacterially expressed His-Grap-2 fusion protein, affinity-purified by NI-NTA/FPLC. NIH 3T3 mouse fibroblasts, as well as immunoblots with the preimmune sera (not shown) served as negative controls. BON- and LCC18-tumor cell lines were grown in DNM or in DMEM, respectively. Media were supplemented with 10% (v/v) heat-inactivated fetal calf serum and 1% (w/v) penicillin/streptomycin. TT-, HEK 293-, NIH 3T3- and Jurkat T cells were described previously (Ludwig et al., 2001). MTC tissue was obtained from patients surgically treated at the University Hospital of Halle, after informed consent. Studies involving the use of human tissue were approved by the Ethical Committee of the University of Halle

Full figure and legend (73K)

To further characterize the RET/Grap-2 interaction, we performed pull-down assays using in vitro translated, metabolically labeled RET proteins and recombinant, affinity-purified Grap-2. As shown in Figure 2a, His-Grap-2 fusion proteins coimmunoprecipitated with both MEN 2A (C634R)- and MEN 2B (M918T)-mutant RET, indicating that this interaction is direct. Autophosphorylation of in vitro translated RET was confirmed by antiphosphotyrosine immunoblotting (Figure 2b). To evaluate the physiological role of the RET/Grap-2 interaction, we aimed to prove their association in untransfected cells. As shown in Figure 2c, coimmunoprecipitation experiments performed in TT cells confirmed the interaction of RET and Grap-2 in vivo. To further examine cellular localization of both proteins, we used red and green fluorescent protein (RFP/GFP) fusion constructs. In transfected 293 cells, the MEN 2A-RET-RFP fusion proteins displayed the expected accumulation at the plasma membrane (Figure 3d). GFP-tagged Grap-2 was distributed throughout the cytosol in large condensations (Figure 3c). Cotransfection of GFP-Grap-2 either with RFP-tagged (Figure 3e, f) or untagged (not shown) MEN 2A-RET resulted in a shift of Grap-2 toward the plasma membrane. Thus, expression of activated RET relocates Grap-2 to the plasma membrane.

Figure 2.

Characterization of RET/Grap-2 interaction. To obtain a His-tagged Grap-2 fusion construct for bacterial expression, the Grap-2 encoding cDNA lacking the first four codons including ATG was subcloned from the yeast two-hybrid vector pGAD424 (Clontech) into pQE 32 (Qiagen), using the linker-derived SalI sites. (a) Pull-down assay of cell-free synthesized RET proteins with bacterially expressed Grap-2. [³⁵S]methionine-labeled MEN 2A (RET C634R) and MEN 2B (RET M918T) mutant RET proteins were generated from pcDNA3RET vectors described previously (Ludwig et al., 2001) using TNT™T7 Coupled Reticulocyte Lysate System (Promega) according to the instructions of the manufacturer. Briefly, HIS-Grap-2 fusion proteins were precipitated from bacterial expression cultures with anti-His monoclonal antibody and protein A/G sepharose. Precipitates from cultures transformed with empty vector served as controls (control). Equal amounts of the precipitates were incubated with in vitro translated RET proteins for 1 h in a buffer containing 140 mM NaCl, 20 mM HEPES pH 7.4, 2 mM EDTA, 1% Triton X-100, 10 mM -mercaptoethanol and 1 mM sodium orthovanadate, supplemented with Complete™ protease inhibitor cocktail (Roche). Subsequently, the precipitates were washed four times in the same buffer and subjected to SDS–PAGE on 8% gels. To demonstrate equal efficiency of methionine incorporation aliquots of the respective translates were loaded (input). After electrophoresis, the upper part of the gel was dried and autoradiographed. Precipitation efficiency was assessed on the lower part of the same gel by anti-Grap-2 immunoblotting. (b) Tyrosine phosphorylation of translated RET proteins was analysed by antiphosphotyrosine immunoblotting. (c) Coimmunoprecipitation of endogenously expressed RET and Grap-2. Lysates derived from 2 10⁷ TT medullary carcinoma cells were precipitated with anti-RET antibody or with protein A/G sepharose alone (IP). Western blotting with anti-RET and antiGrap-2 as indicated (WB). Western blots were performed as published (Ludwig et al., 2001), using anti-RET C-19 (Santa Cruz), anti-phosphotyrosine 4G10 (Upstate) and anti-His RGS (Qiagen) antibodies

Full figure and legend (74K)

Figure 3.

Colocalization of GFP-Grap-2 and RET-RFP fusion proteins. HEK 293 cells were transiently transfected by the calcium phosphate precipitation method. Cells were washed 24 h or 48 h post-transfection with phosphate-buffered saline. Red and green fluorescence of unfixed cells was visualized using a confocal laser scanning microscope (Leica, TCS 4D). Transfection of empty vectors encoding red- or green-fluorescent proteins yields homogeneously stained cells (a, b). Expression of either fusion protein alone leads to Grap-2 accumulation in the cytoplasm (c) and RET localization to the plasma membrane (d). Coexpression of both proteins (e+f) alters the Grap-2 expression. As seen in (e) Grap-2 colocalizes with RET at the cytoplasma membrane. Plasmid constructs were generated as follows: GFP-tagged Grap-2 was obtained through subcloning of the Grap-2 cDNA into BamHI and XbaI sites of pEGFP-C1 (Clontech). For RET C634R-RFP, a HindIII/HpaI fragment derived from pcDNA3RET C634R, encoding amino acids 1–1114, was inserted into HindIII and SmaI sites of pDsRed1-N1 (Clontech)

Full figure and legend (406K)

As we have reported previously, MEN 2A- and MEN 2B-RET proteins activate NF-B-dependent transcription (Ludwig et al., 2001). To evaluate a possible role for Grap-2 in RET-mediated NF-B activation, we performed cotransfection experiments in NIH 3T3 cells. In these assays, Grap-2 expression resulted in a dose-dependent reduction of RET-induced NF-B reporter activity (Figure 4a and b). RET C634R and RET M918T slightly differ in their ability to induce NF-B-dependent transcription (Ludwig et al., 2001). As shown here, NF-B reporter activity induced by RET C634R was influenced significantly stronger than that induced by RET M918T, reflecting different mechanisms in signal transmission between the oncogenic RET proteins. Nevertheless, these results indicate that Grap-2 is involved in NF-B activation mediated by both MEN 2A- as well as MEN 2B-RET proteins.

Figure 4.

Oncogenic RET induces NF-B-dependent transcription. NIH 3T3 cells were transiently transfected with 1 g of NF-B Luc-reporter along with 200 ng of expression vectors encoding oncogenic RET mutants C634R (a) or M918T (b) and increasing doses of Grap-2 expression constructs (0.5, 1 or 3 g, respectively). At 48 h post-transfection, cells were lysed for luciferase assay. Bars express fold induction of luciferase activity relative to the reporter alone and represent means s.e.m. of six independent experiments performed in triplicate. Transfection was performed using FUGENE™ (Roche), total amounts of DNA were equalized with empty pcDNA3. Luciferase assays were performed as described previously (Ludwig et al., 2001). Grap-2 inhibits RET-induced transformation of NIH 3T3 cells. NIH 3T3 cells were seeded at a density of 1.5 10⁵ per 100-mm dish 12 h prior to transfection. Equal amounts of expression constructs of Grap-2 and RET C634R (c), or Grap-2 and RET M918T (d) were stably transfected either using FUGENE™ or the calcium phosphate precipitation method with 20 g high-molecular genomic DNA derived from NIH 3T3 cells as carrier. The medium was changed 12 h post-transfection to remove the precipitate, and after an additional 24 h, the medium was replaced by DMEM supplemented with 5% newborn calf serum. The media were changed every 5 days. Transformed foci were counted by phase contrast microscopy 14–21 days after transfection. Values represent meanss.e.m. of at least three independent experiments

Full figure and legend (50K)

In our previous work, we demonstrated that the transforming activity of RET in eukaryotic cells was dependent on NF-B activation (Ludwig et al., 2001). To test whether the interaction between RET and Grap-2 influenced the capacity of RET in establishing a malignant phenotype, we performed focus forming assays in NIH 3T3 cells. Cotransfection of Grap-2 together with oncogenic RET significantly reduced the numbers of transformed colonies in this assay (Figure 4c and d). In accordance with the results obtained for the B-dependent luciferase activity, the effect of Grap-2 on MEN 2B-RET was weaker than on MEN 2A-RET. These results confirm the importance of NF-B in the context of RET-induced transformation, and indicate that Grap-2 plays an important role in modulating RET mitogenic signaling.

Taken together, the presented data extend the knowledge of Grap-2 expression to cells of neuroendocrine origin, including medullary thyroid carcinoma and cell lines derived from gastropancreatic neuroendocrine tumors, suggesting an additional neuroendocrine specific role for Grap-2. The LCC18 and BON cell lines (Figure 1b) show expression of Grap-2 protein, but do not express detectable levels of RET on immunoblots (data not shown). This implies that Grap-2 is involved in signal transduction of as yet undefined (receptor) tyrosine kinases in these cells. Grap-2 and Grb2 share the same docking sites on several proteins including M-CSFR (Bourette et al., 1998; Liu et al., 2001), which raises the possibility that Grap-2 effects could be achieved at least in part through competition with Grb2. Consistent with this, addition of Grb2 in cotransfection experiments was able to revert partially the Grap-2 effect on RET-induced NF-B activity (see supplementary material). This evokes a model according to which Grap-2 and Grb2 modulate, dependent on their concentrations, the affinity of an adaptor complex recruited to the activated receptor and thereby alter the transmitted signal.

Several studies indicate that despite their interaction with identical phosphotyrosine motifs, Grap-2 and Grb2 specify for different signals mediated through distinct downstream targets (Liu et al., 2001). In accordance with this, Grap-2 and Grb2 have been reported to recruit different proteins via their SH3 domains in T-cell signaling (Liu et al., 1999). Although we were not able to identify a unique binding motif for Grap-2 in RET, dephosphorylation of in vitro translated RET proteins abolished the RET/Grap-2 interaction (data not shown), suggesting that binding to RET is mediated by the Grap-2 SH2 domain. Thus, interaction of Grap-2 SH3 domains with downstream acting molecules and their contribution to RET signaling has to be clarified by further investigation.

The presented data support the hypothesis that Grap-2 is able to influence NF-B activation through interaction with activated, membrane-associated RET. As we demonstrated in our previous work, suppression of NF-B leads to cell death in MTC cells expressing activated RET. NF-B inhibition mediated by Grap-2 could, therefore, be employed in a therapeutical strategy in neuroendocrine tumor cells transformed by oncogenic RET.