cbl-3: a new mammalian cbl family protein - Nature

Oncogene (1999) 18, 3365 ± 3375 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

cbl-3: a new mammalian cbl family protein Maccon M Keane1, Seth A Ettenberg1, Marion M Nau1, Priya Banerjee1, Mauricio Cuello1, Josef Penninger2 and Stan Lipkowitz*,1 1

Genetics Department, Medicine Branch, National Cancer Institute, Bethesda Naval Hospital, Bethesda, Maryland 20889, USA; Amgen Institute/Ontario Cancer Institute, Departments of Medical Biophysics and Immunology, University of Toronto, Toronto/ Ontario, Canada

2

We have cloned a new human gene, cbl-3, which encodes a protein with marked homology to the cbl family of proteins. The predicted protein encoded by this gene retains the conserved phosphotyrosine binding domain (PTB) in the N-terminal and the zinc ®nger but is signi®cantly shorter (MW 52.5 kDa) than the other mammalian cbl proteins. The protein lacks the extensive proline rich domain and leucine zipper seen in c-cbl and cbl-b and structurally most resembles the C. elegans and Drosophila cbl proteins. The gene is ubiquitously expressed with highest expression in the aerodigestive tract, prostate, adrenal gland, and salivary gland. The protein is phosphorylated and recruited to the EGFR upon EGF stimulation and inhibits EGF stimulated MAP kinase activation. In comparison to the other mammalian cbl proteins (e.g. cbl-b), cbl-3 interacts with a restricted range of proteins containing Src Homology 3 regions. An alternatively spliced form of the cbl-3 protein was also identi®ed which deletes a critical region of the PTB domain and which does not interact with the EGFR nor inhibit EGF stimulated MAP kinase activation. These data demonstrate that cbl-3, a novel mammalian cbl protein, is a regulator of EGFR mediated signal transduction. Keywords: cbl proteins; EGF receptor; signal transduction

Introduction The cbl family of proteins is found in metazoans from nematodes to vertebrates and the proteins have several highly conserved domains including a novel N-terminal phosphotyrosine binding (PTB) motif and a zinc ®nger (Blake et al., 1991; Hime et al., 1997; Keane et al., 1995; Lupher et al., 1997; Meisner et al., 1997; Yoon et al., 1995). The cbl proteins are phosphorylated upon activation of a variety of receptors which signal via protein tyrosine kinases and they associate with many proteins containing Src Homology 2 (SH2) and Src Homology 3 (SH3) domains (reviewed in Miyake et al., 1997; Smit and Borst, 1997). These diverse interactions modulate signaling through many pathways (Miyake et al., 1997; Smit and Borst, 1997). For example, in mast

*Correspondence: S Lipkowitz, Genetics Department, Medicine Branch, National Cancer Institute, Building 8, Room 5101, National Naval Medical Center, Bethesda, MD 20889-5101, USA Received 4 August 1998; revised 6 January 1999; accepted 7 January 1999

cells overexpression of c-cbl inhibited FceRI induced activation of Syk kinase and downstream serotonin release (Ota and Samelson, 1997), in ®broblasts overexpression of c-cbl inhibited and antisense repression of c-cbl enhanced EGF induced phosphorylation of the epidermal growth factor receptor (EGFR) and activation of the Jak-Stat pathway (Ueno et al., 1997), and overexpression of cbl-b in Cos cells inhibited Vavinduced Jun Kinase activation (Bustelo et al., 1997). Recent data show that c-cbl de®cient mice have hyperplastic hematopoietic and breast tissue consistent with a negative regulatory role in cellular proliferation for c-cbl (Murphy et al., 1998). Together these data indicate that the cbl proteins are important regulators of intracellular signaling and consequently of cell function and development. A role for the cbl proteins in EGFR signaling was ®rst demonstrated in C. elegans by genetic studies that show that sli-1 is a negative regulator of the Let-23 receptor tyrosine kinase (the EGFR homolog) in vulva development (Jongeward et al., 1995; Yoon et al., 1995). These developmental eects have been extended to Drosophila where D-cbl has been shown to associate with the Drosophila EGFR and overexpression of Dcbl in the eye of Drosophila embryos inhibits EGFR dependent photoreceptor cell development (Hime et al., 1997; Meisner et al., 1997). In mammalian cells, several studies have shown that c-cbl becomes phosphorylated and recruited to the EGFR upon stimulation and can alter signaling (reviewed in Miyake et al., 1997; Smit and Borst, 1997). Here, we describe a new mammalian cbl gene, cbl-3, and begin to characterize its role in EGFR mediated signal transduction.

Results Identi®cation and cloning of cbl-3 A cDNA clone (I.M.A.G.E. Consortium CloneID 526956), which appeared to be a novel member of the cbl family, was identi®ed by a BLAST search of the National Center for Biotechnology Information (NCBI) human EST data base. This cDNA clone, derived from a cDNA library prepared from the Stratagene human pancreas adenocarcinoma cell line CFPAC-1 (Schoumacher et al., 1990), was obtained from ATCC. The 753 bp clone was sequenced and nucleotide and predicted amino comparisons revealed 60 ± 70% homology to either c-cbl (Blake et al., 1991) or cbl-b (Keane et al., 1995) over the ®rst *450 bp. Due to the homology to the two known mammalian genes, we called this new gene cbl-3.

cbl-3; a new cbl protein MM Keane et al

3366

Northern analysis, using the above cDNA insert as a probe, demonstrated expression of a 1.7 ± 2.0 kb mRNA at high levels in liver, pancreas, small intestine, colon, prostate and trachea (Figure 1). Longer exposure of the Northern blot revealed low levels of expression in the stomach, lung and thyroid gland (data not shown). No expression of cbl-3 was seen in hematopoietic tissues (i.e., PBL, spleen, thymus lymph node, and bone marrow). The size of the mRNA was distinct from those reported for either ccbl (Langdon et al., 1989a) or cbl-b (Keane et al., 1995). The pancreatic adenocarcinoma cell line, CFPAC-1, and the breast carcinoma cell line, ZR751, also expressed high levels of the mRNA for cbl-3. Further analysis of expression using a mRNA dot blot revealed that while cbl-3 was expressed ubiquitously at low levels, the highest levels of expression were in the tissues described above and additionally in the adrenal gland and salivary gland. In order to obtain the full length cDNA for cbl-3, a cDNA library made from the CFPAC-1 cell line was screened using the 753 bp EST described above as a probe. Seven independent overlapping cDNA clones were obtained by screening 1.06106 recombinant phage plaques. Additional sequence of the 5' end of the transcript was obtained from multiple cDNAs generated by the 5' RACE procedure (Frohman et al., 1988). The composite nucleotide sequence and predicted protein is shown in Figure 2. The cDNA contains 1584 bp up to the poly(A) tail. The composite cDNA contains a consensus polyadenylation signal (Wilusz et al., 1989) in the 3' untranslated region (UTR) 23 bp upstream to the poly(A) tail. The short 3' UTR was con®rmed by RT ± PCR amplification of the 3' end of cbl-3 from RNA from the CFPAC-1 and ZR75-1 cell lines utilizing an oligo dT primer (to both synthesize the ®rst strand cDNA and amplify the cDNA) and a sense primer within the coding region of the cDNA. The composite cDNA encodes a predicted protein of 474 amino acids (MW 52.5 kDa) beginning with a putative translation initiation codon at nucleotide 64 and ending with a termination codon at nucleo-

tide 1486. The indicated translation initiation codon is not in good agreement with the Kozak consensus sequence (Kozak, 1986) but it is the most 5' ATG. There is a downstream ATG at nucleotide 133 that is a reasonable Kozak consensus sequence that would result in a protein *2.5 kDa shorter. Both potential initiation codons are 5' to the regions of homology with c-cbl and cbl-b (see Figure 4a) and thus either is likely to serve as the translation initiation codon. Interestingly, cbl-b does not have a Kozak consensus translation initiation ATG (Keane et al., 1995). There are no stop codons in the 5' UTR and thus it is possible that there is further sequence missing from the 5' end that contains a more upstream translation initiation codon. We think that this is unlikely for several reasons. Of the 16 5' RACE cDNA clones that were sequenced, eight began at the ®rst nucleotide in the composite sequence and no longer 5' RACE clones were generated in multiple experiments. This transcript initiation site was found in RACE cDNA clones from the CFPAC-1 pancreas adenocarcinoma cell line, the ZR75-1 breast carcinoma cell line, and from normal pancreas. The sequence after the ®rst nucleotide is in good agreement with the consensus sequence of a transcript initiation site ([A/G]YYYYY) (Corden et al., 1980). Finally, the length of the consensus sequence (1584 nucleotides) is sucient to account for the 1.8 ± 2.0 kb mRNA seen on Northern blots when polyadenylation is taken into account. One of seven cDNA clones that was isolated from the CFPAC-1 cDNA library had a deletion of 138 nucleotides within the coding region (shown by arrowheads in Figure 2) resulting in a predicted protein that is 46 amino acids shorter (MW 47.5 kDa). This deletion contains a highly conserved region of the cbl proteins which has been demonstrated to be important to the phosphoprotein binding ability of the cbl proteins (reviewed in Miyake et al., 1997; Smit and Borst, 1997). Using RT ± PCR, both the full length long form (cbl-34) and the deleted short form (cbl-35) of the mRNA were detected in the CFPAC-1 pancreas carcinoma cell line, the ZR75-1 breast carcinoma cell line, and normal pancreas (Figure 3). The PCR products were cloned and sequenced to con®rm their

Figure 1 Expression of cbl-3 in normal tissues. Two mg of poly(A)+ RNA from normal human tissue were probed with a cbl-3 cDNA probe (top panel) or b-actin (bottom panel) as a loading control. Size in kb is shown along the left of the top panel. pbl, peripheral blood mononuclear cells


3367

Figure 2 sequences deleted in (Genbank

Composite nucleotide sequence and predicted protein of cbl-3. The nucleotide sequence represents the composite of several overlapping cDNA clones and the products of 5' RACE experiments. The arrowheads indicate the region the alternatively spliced form. The zinc ®nger is double underlined and the polyadenylation signal is single underlined. accession numbers: cbl-3L, AF117646; cbl-3S, AF117647)

indicates that the short form represents a normally occurring alternatively spliced mRNA. The PCR product indicated by the asterisk in Figure 3 was found, upon sequencing, to be an alternatively spliced form of the mRNA that contained a Line element insert which resulted in a truncated protein (*24 kDa). This form was not characterized further and most likely represents an incompletely spliced mRNA. cbl-3 shares high homology with the other cbl family proteins

Figure 3 RT ± PCR demonstrating the expression of alternatively spliced forms of cbl-3 cDNA in normal pancreas, the ZR75-1 breast carcinoma cell line and CFPAC-1 pancreatic adenocarcinoma cell line. PCR products were separated on a 1.2% agarose gel containing ethidium bromide to visualize the DNA. The location of the long and short forms of cbl-3 and their size in bp are indicated along the right side of the gel. The asterisk represents an additional alternatively spliced form seen in all three samples (see text for discussion). The size standard in bp is shown along the left hand side of the gel

identity. The presence of the short form of the mRNA in RNA isolated from normal tissue (i.e. pancreas) and malignant cell lines (i.e. CFPAC-1 and ZR75-1)

The predicted cbl-3 protein contains a PTB domain (aa 36 ± 350) and a C3HC4 zinc ®nger motif (aa 351 ± 390) (Lovering et al., 1993) which are conserved within all of the cbl proteins (Figure 4a). This region shows the highest homology to the other mammalian cbl proteins (66 ± 68%), the Drosophila protein (65%), and C. elegans cbl protein (59%). The cbl-3 protein ends 84 amino acids after the zinc ®nger motif and structurally looks most similar to Sli-1 and D-cbl, the C. elegans and Drosophila cbl proteins respectively (Hime et al., 1997; Meisner et al., 1997; Yoon et al., 1995). A phylogenetic comparison of the cbl proteins con®rms that cbl-3 is more closely related to the more primitive cbl proteins than it is to the other mammalian cbl proteins (Figure 4b). Like Sli-1, cbl-3 has a short proline rich region near the C-terminus which could


3368

serve as a binding site for SH3 proteins (Cohen et al., 1995) but it lacks the more extensive proline rich domain seen in c-cbl and cbl-b. cbl-3 also lacks the putative leucine zipper motif found in c-cbl and cbl-b. The region deleted in the alternatively spliced short form of cbl-3 (indicated by the arrowheads in Figure 4a) is in the highly conserved PTB region and contains the glycine residue (indicated by an asterisk in Figure 4a) that was originally found to be the site of an inactivating mutation (glycine to glutamic acid) in the a

Sli-1 protein (Yoon et al., 1995). Mutation of this glycine to glutamic acid has been shown to abrogate binding of cbl proteins to phosphorylated protein tyrosine kinases (Lupher et al., 1996; Meisner et al., 1997; Miyake et al., 1997; Smit and Borst, 1997). Chromosomal localization of cbl-3 cbl-3 was localized to chromosome 19 by analysis of DNA from a panel of somatic cell hybrid cells. The


location was con®rmed and further delineated by analysis of the RH01.02 Stanford G3 Radiation Hybrid Panel. Cbl-3 was linked to marker SHGC11815 with a LOD score of 8.3 which maps to chromosome 19q13.2. The c-cbl and cbl-b genes have previously been localized to chromosome 11 (Savage et al., 1991) and chromosome 3 (Keane et al., 1995), respectively. cbl-3 associates with SH3 proteins c-cbl and cbl-b have previously been shown to interact with a variety of SH3 proteins both in vitro (Keane et al., 1995; Rivero-Lezcano et al., 1994) and in vivo (reviewed in Miyake et al., 1997; Smit and Borst, 1997). The short proline rich sequence in cbl-3 near the C-terminus (aa 434 ± 447) has homology to c-cbl and cbl-b (Figure 4a) and contains several possible PXXP motifs. This motif has been found in high anity SH3 ligands (Cohen et al., 1995). To test possible association of cbl-3 with SH3 proteins, HAepitope tagged cbl-3L was transfected into human embryonic kidney cells expressing the SV40 large Tantigen (293T) and the lysate from these cells was incubated with equal amounts of various Gst fusion proteins containing SH3 domains (Figure 5). The cbl-3 that was bound to each Gst fusion protein was determined by precipitating the protein with glutathione-agarose beads, fractionating the precipitated protein by SDS ± PAGE, and analysing the gels by immunoblotting with an anti-HA antibody. The long form of cbl-3 (cbl-3L) associated with fusion proteins containing the SH3 domains of Lyn and Crk but not with any of the other SH3 proteins (Figure 5). The aberrant size of the cbl-3 protein seen in the Lyn lane is due to distortion caused by the Gst-Lyn fusion protein which migrates in the gel at the same size as cbl-3. The short form of cbl-3 (cbl-3S) produced identical results (data not shown). In contrast, cbl-b associated strongly with three and weakly with four of the nine proteins tested.

3369

cbl-3L is phosphorylated and recruited to the epidermal growth factor receptor upon stimulation The cbl proteins from C. elegans, Drosophila, and mammals have been shown to be phosphorylated and to associate with the EGFR upon EGF stimulation (reviewed in Miyake et al., 1997; Smit and Borst, 1997). To investigate possible interaction of cbl-3L and cbl-3S with the EGFR, HA-epitope tagged cbl-3L or cbl-3S was transfected along with the EGFR into 293T cells. The cells were then serum starved and stimulated with EGF and the resultant lysates used to compare the interactions between the EGFR and each cbl protein. These lysates were immunoblotted and probed with anti-phosphotyrosine (anti-pty), anti-HA, and anti-EGFR (Figure 6a, left panels). Lysates immunoblotted with anti-HA showed cbl-3 proteins that were at the predicted sizes for cbl-3L (*52 kDA) and cbl-3S (*48 kDa) (Figure 6a, bottom left panel). Lysates from the transfected cells showed a prominent EGFinduced tyrosine phosphoprotein at *180 kDa

Figure 5 cbl-3 interacts with SH3 proteins. Cell lysates containing cbl-3L (top panel) or cbl-b (bottom panel) were precipitated with 5 mg of GST fusion proteins of the SH3 domains for the proteins indicated along the top of the ®gure as described in Materials and methods, and the presence of the cbl protein in the precipitate determined by immunoblotting with the anti-HA antibody. GST was included as a negative control in lane 2. An aliquot of the cell lysate for each protein is shown in the ®rst lane (lysate). Size in kDa is shown along the left side of the ®gure. Similar results were obtained using cbl-3S (data not shown)

b

Figure 4 (a) Comparison of the predicted amino acid sequences of the cbl family of proteins. Top, schematic diagram; homology is indicated by black bars. The PTB, zinc fnger, the proline rich region (Prolines), and the leucine zipper are indicated above the schematic diagram. The per cent identity and homology to cbl-3 of the region of highest homology containing the PTB and zinc ®nger are shown above each protein (identity/homology). The number of amino acids in each protein is shown at the right. Bottom, amino acid sequence, homology is indicated by capital letters. The zinc ®nger region is double underlined and the region deleted in the alternatively spliced form is demarcated by arrowheads. The asterisk represents the conserved glycine residue previously shown to be important for binding to receptor tyrosine kinases. Comparisons were performed using the MACAW program (National Center of Biotechnology Information). (b) Phylogenetic relationships between the cbl proteins. The analysis was performed using the Clustal program and PC/Gene


3370

a

b

Figure 6 cbl-3L is phosphorylated and recruited to the EGFR upon EGF simulation. (a) 293T cells were transiently transfected with either HA tagged cbl-3L (7 mg) and the EGFR (1 mg) (cbl-3L+EGFR), with HA tagged cbl-3S (7 mg) and the EGFR (1 mg) (cbl-3S+EGFR), or with vector alone (8 mg) (V). The cells were starved for 4 h, incubated with or without EGF (100 ng/ ml) for 10 min, and protein lysates were prepared. Lysates (left panels) or anti-HA precipitated cbl-3L and cbl-3S (right panels) were run on parallel gels, transferred, and immunoblotted with the antibodies shown to the left of the blots. (b) Lysates from EGF starved or stimulated 293T cells transiently transfected with HA tagged cbl-3L+EGFR were immunoprecipitated with the anti-HA antibody as above, an aliquot was taken from the precipitate (Ip-1), and then the immunoprecipitated complexes were denatured and re-immunoprecipitated (Ip-2). The immunoprecipitates were immunoblotted with the anti-pty antibody (top panel) and stripped and reprobed with the anti-HA antibody (bottom panel). The positions of the cbl-3 and the EGFR proteins are indicated by the arrows on the right side of the ®gure. Molecular weight standards (in kDa) are shown to the left of the blots


a

b

(corresponding to the position of the EGFR) as well as several other prominent phosphoproteins (Figure 6a, top left panel). We are unable to see clear evidence of phosphorylation of the cbl-3 proteins in the lysate. These lysates were used to immunoprecipitate the cbl3 proteins (using an anti-HA antibody) to assess phosphorylation and association with the EGFR (Figure 6a, right panels). When the immunoprecipitated proteins were probed with anti-pty, EGF-induced phosphorylation of the cbl-3L protein, but not the cbl3S protein, was observed (Figure 6a, right top panel). A phosphoprotein that migrates at the size of the EGFR was co-immunoprecipitated with cbl-3L but not with cbl-3S. Probing the precipitated protein with antiEGFR con®rmed that, upon EGF stimulation, the EGFR was co-immunoprecipitated with cbl-3L but not with cbl-3S (Figure 6a, right middle panel). Probing the precipitated proteins with anti-HA demonstrated equal precipitation of the cbl-3L and cbl-3S proteins. These results show that cbl-3L but not cbl-3S is phosphorylated and recruited to the EGFR upon EGF stimulation. The major sites of phosphorylation of c-cbl are in the C-terminus half of the protein that is absent in cbl3L (Miyake et al., 1997). To con®rm that cbl-3L was phosphorylated upon EGF stimulation, we immunoprecipitated the cbl-3L protein with the anti-HA antibody (Figure 6b, Ip 1), denatured the immunoprecipitated complex (Chen et al., 1995), and reimmunoprecipitated the cbl-3L protein with the antiHA antibody (Figure 6b, Ip 2). Both Ip 1 and Ip 2 had a phosphoprotein that migrated at the same size as cbl3L in the EGF stimulated samples (Figure 6b, top panel). In contrast, the co-precipitating EGFR in the ®rst immunoprecipitation (Figure 6b, Ip 1) is not seen when the complex is denatured prior to re-precipitation (Figure 6b, Ip 2). This result con®rms that cbl-3L is phosphorylated upon EGF stimulation. cbl-3L inhibits EGF induced MAP kinase (MAPK) activation

Figure 7 cbl-3L inhibits EGF stimulated MAPK activation. (a) EGF stimulated erk-2 activation was assayed in 293T cells transiently transfected with HA-epitope tagged erk-2 (0.1 mg) and either HA-epitope tagged cbl-3L (7 mg), cbl-3S (7 mg), or vector control (7 mg). The cells were starved and then stimulated for the indicated times with EGF (10 ng/ml). Cells were lysed, the transfected proteins immunoprecipitated with the anti-HA antibody, and erk-2 activation was assayed by immunoblotting with an antibody speci®c for activated phosphorylated erk-2 (phosphoerk-2). Total erk-2 and cbl-3 protein was assessed by stripping and re-probing the ®lter with the anti-HA antibody. The positions of the cbl-3 and the erk-2 proteins are indicated by the arrows on the right side of the ®gure. Molecular weight standards (in kDa) are shown to the left of the blots. (b) HeLa cells were cotransfected with reporter plasmids for MAPK transcriptional activation along with increasing concentrations of cbl-3L or cbl3S plasmids (as described in Materials and methods). EGF stimulation of reporter activity (in fold stimulation) is shown along the Y axis and the amount of cbl-3L or cbl-3S plasmid transfected is shown along the X axis. The amount of DNA was kept constant by the addition of vector DNA. The numbers represent the average+s.d. for triplicate measurements from a representative experiment. The expression of the transfected cbl-3 protein is shown in the immunoblot beneath the graph

Upon stimulation of growth factor receptors, MAPK is activated by phosphorylation of speci®c tyrosine and threonine residues (Payne et al., 1991). To investigate whether the interaction of cbl-3L with the EGFR aects MAPK activation, HA-epitope tagged erk-2 was cotransfected along with vector, HA-epitope tagged cbl-3L, or cbl-3S into 293T cells. The cells were then serum starved and stimulated with EGF and the transfected proteins were immunoprecipitated with the anti-HA antibody. Activation of erk-2 was assessed by immunoblotting with a phospho-speci®c antibody that recognizes activated erk-2. EGF stimulation induced rapid activation (within 5 min) of erk-2 in cells expressing cbl-3L, cbl-3S and the vector controls (Figure 7a). However, the duration of erk-2 activation was shorter in cells expressing cbl-3L compared to either those expressing cbl-3S or the vector control. Activated MAPK in turn phosphorylates and activates the transcription factor Elk-1 (Janknecht et al., 1993). To con®rm that cbl-3L inhibits the activation of MAPK, we used transient expression of reporter plasmids speci®c for the activated Elk-1 in HeLa cells (Figure 7b). cbl-3L inhibited EGF induced stimulation of transcription by MAPK in a dose dependent fashion. In contrast, the short form of cbl-

3371


3372

3, which does not get phosphorylated upon EGF stimulation and does not associate with the EGFR, did not aect MAPK dependent transcriptional activation by the EGFR. The EGF induced stimulation in the cells transfected with both 0.5 and 1.5 mg of cbl-3L was statistically dierent from the vector control (P=0.048 and P=0.0005 respectively) while the EGF induced stimulation in the cells transfected with either 0.5 or 1.5 mg of cbl-3S was not dierent from the control (P=0.70 and P=0.48 respectively). Similar results were obtained in NIH3T3 cells (data not shown). Discussion The cbl-3 gene encodes a new protein with marked homology to the other members of the cbl family (Figure 4). There is conservation of the N-terminal PTB domain and zinc ®nger and a short proline rich region. Structurally, the protein seems most similar to the C. elegans and Drosophila cbl proteins, sli-1 and Dcbl respectively (Figure 4). The lack of an extensive proline rich region in the cbl-3 protein suggests that it should have a much more restricted ability to bind to SH3 proteins compared to c-cbl and cbl-b and indeed cbl-3 bound to only two of the nine SH3 proteins tested compared to cbl-b which bound to seven of the nine SH3 proteins (Figure 5). The signi®cance of the SH3 interaction of cbl-3 with Lyn and Crk remains to be determined. Cbl-3 mRNA is ubiquitously expressed at low levels but the highest expression is in the aerodigestive tract (i.e., stomach, liver, pancreas, small intestine, colon, trachea, and lung), prostate, thyroid gland, adrenal gland, and salivary gland (Figure 1). This pattern of expression is distinct from c-cbl and cbl-b. Both c-cbl and cbl-b are expressed widely with the highest expression of c-cbl in hematopoietic cells and the testis (Langdon et al., 1989b; Smit and Borst, 1997) and the highest expression of cbl-b in testis, ovary, breast cancer cell lines, and fetal brain (Bustelo et al., 1997; Keane et al., 1995). c-cbl and cbl-b have not been shown to be highly expressed in the gastrointestinal tract or the other tissues that express the highest levels of cbl-3 (Bustelo et al., 1997; Keane et al., 1995; Langdon et al., 1989b; Smit and Borst, 1997). The high expression of cbl-3 in the tissues described above suggest that it will have an important role in these tissues. The long form of cbl-3 is phosphorylated and recruited to the EGFR upon EGF stimulation (Figure 6) and this results in a decrease in EGF induced MAPK activation (Figure 7). While the major sites of phosphorylation of c-cbl have been mapped to the Cterminus region of the protein that is absent in cbl-3 (Miyake et al., 1997), our data clearly demonstrate phosphorylation of cbl-3 upon EGF stimulation (Figure 6). The Drosophila cbl protein, which similarly lacks the C-terminus phosphorylation sites of c-cbl, has also been shown to be phosphorylated upon EGF stimulation (Hime et al., 1997; Meisner et al., 1997). Data are not available on the phosphorylation of the C. elegans cbl protein, Sli-1, and the sites of phosphorylation in the Drosophila protein are not known. The mechanism by which cbl-3L associates with the EGFR remains to be determined. Proteins in the cbl family have been shown to interact with the EGFR

through both direct binding of the PTB domain and through adaptor mediated binding via the Grb2 protein (Miyake et al., 1997; Smit and Borst, 1997). The failure of cbl-3 to interact with the SH3 domain of Grb2 in vitro (Figure 5) suggests that it is likely to directly bind via its N-terminus PTB domain. D-cbl and v-cbl have similarly been shown to bind to the EGFR despite their inability to bind to the Grb2 adaptor protein (Hime et al., 1997; Meisner et al., 1997; Thien and Langdon, 1997). The failure of the alternatively spliced short form of cbl-3 to bind to the EGFR (Figure 6) is consistent with direct binding via the PTB domain. The cbl-3S form deletes a region of cbl-3 that contains the glycine residue (glycine 276) that has been shown to be critical to the PTB function of this region (Figure 4) (Lupher et al., 1996; Meisner et al., 1997; Thien and Langdon, 1997; Yoon et al., 1995). Mutation of this glycine to glutamic acid has been shown to abrogate binding of cbl proteins to phosphorylated protein tyrosine kinases (Lupher et al., 1996; Meisner et al., 1997; Thien and Langdon, 1997). The role of the cbl-3S protein is unknown. One possibility is that it acts as an inhibitory molecule by binding to proteins normally recruited to the EGFR via their interaction with the cbl-3L protein. The interaction of cbl-3L with the EGFR shortens the duration of EGF stimulated MAPK activation (Figure 7a) and subsequently decreases MAPK dependent transcriptional activation (Figure 7b). We have observed a similar eect of cbl-b on MAPK activation (Ettenberg et al., 1999). The decrease in duration of MAPK activation by cbl-3L and cbl-b suggest that the cbl proteins function to enhance activation dependent mechanisms of inhibition. Recent work has demonstrated that c-cbl can enhance growth factor induced ubiquitination and degradation of the receptors (Levkowitz et al., 1998; Miyake et al., 1998; Wang et al., 1996). Thus, a possible mechanism for the inhibition we observed is that cbl-3L, upon EGF stimulation, binds to the EGFR and enhances the degradation of the activated receptor. Our data demonstrate that cbl-3 is a novel mammalian cbl protein which shares structural and functional homology to the other cbl family members, in particular to the more primitive species of cbl proteins. Materials and methods cDNA screening An oligo(dT) primed cDNA library in the Uni-ZapTM XR vector prepared from the human pancreatic adenocarcinoma cell line CFPAC-1 (ATCC CRL-1918) was purchased from Stratagene and screened with an a-32P-dCTP (Amersham) labeled cDNA probe isolated from an IMAGE consortium EST(#526956). cDNA inserts from this library were subcloned into the pBluescript plasmid vector by in vivo excision (Short et al., 1988). Additional 5' sequence of the cbl-3 cDNA was obtained by the RACE (rapid ampli®cation of cDNA ends) technique (Frohman et al., 1988) using the 5' Race System kit (GIBCO BRL). The procedure was carried out according to the instruction manual. First-strand cDNA synthesis was performed with 2.0 mg and 1.0 mg respectively of total cellular RNA from the CFPAC-1 cell line and normal pancreas or 1.0 mg of poly(A)+ RNA from the ZR751 breast cancer cell line using an antisense oligonucleotide


primer from the 5' end of cbl-3 cDNA clones isolated from the phage libraries (5'-(CAU)4GGTCGACGCATTGCTCTTCTA-3'). The puri®ed and oligo (dC) tailed cDNA was PCR ampli®ed using an antisense oligonucleotide primer from cbl-3 which was internal to the primer used in the ®rst strand cDNA synthesis (5'-(CAU)4TCCTGACTGCCCGGCCCAGGG-3') and the 5' universal primer supplied with the kit as described in the instruction manual. To enhance the product yield, 1 or 2 ml of the ampli®ed material was reampli®ed using a nested antisense oligonucleotide primer from cbl-3 (5'-(CAU)4TTCCCACTGTCGCCCCCACGG-3') and the 5' RACE ampli®cation primer supplied with the kit. The CAU repeats were added to the primers to facilitate cloning the PCR fragments. One tenth of each reaction was assessed by fractionation on an agarose gel. The ®nal PCR products were cloned into pAMP1 DNA using the CloneAmp system (GIBCO-BRL). DNA sequencing and sequence analysis DNA sequencing was performed on both strands by the dideoxy chain termination method (Sanger et al., 1977) using the T7 Sequenase 7-deaza-dGTP Sequencing kit (Amersham). Nucleotide sequences were analysed using MacVector (Eastman Kodak) and compared with those in the GenBank data base at the National Center of Biotechnology Information (NCBI) using the BLAST network service. Predicted protein sequence comparisons were performed using the MACAW Program (NCBI). RT ± PCR for alternatively spliced forms of the mRNA First strand cDNA was synthesized from 2 mg of total cellular RNA from the CFPAC-1 cell line and normal pancreas or 1 mg of poly(A)+ RNA from the ZR75-1 breast cancer cell line using the protocol described in the 3' RACE System for Rapid Ampli®cation of cDNA Ends kit (GIBCOBRL). 2 ml of the ®rst strand cDNA was ampli®ed using Amplitaq DNA polymerase (Perkin-Elmer) in 50 ml reaction volume under standard conditions recommended by the supplier. The mixture was denatured for 3 min at 958C followed by 35 thermal cycles (1 min at 948C, 1 min at 558C and 2 min at 728C). The cbl-3 speci®c primers used in the initial PCR were: sense primer 5'-CGGCGGCTCTGGGGACTTTCT-3' and antisense primer 5'-GAGCAGCACATCTGGGTGCCC-3'. To enhance the product yield, 2 ml of a 1/100 dilution of the ®rst PCR ampli®cation was reampli®ed using nested primers. The primers used in the nested reaction were: sense primer 5'-(CUA)4AAGCTGGCCATCATCTTCAGC-3' and antisense primer 5'-(CAU)4GCCTTCCTGGTCACTGCTGTT-3'. The primers for this experiment were chosen in the regions common to both spliced forms. The CUA and CAU repeats on the 5' ends of the sense and antisense primers were added to facilitate cloning the ampli®ed products. The PCR products were analysed by gel electrophoresis on a 1.2% agarose gel. The ®nal PCR products were cloned into pAMP1 DNA using the CloneAmp system (GIBCO-BRL) and sequenced as above. Isolation and analysis of RNA RNA was extracted from cells as previously described (Chirgwin et al., 1979) and poly(A)+ RNA was isolated by oligo(dT) cellulose chromatography (Aviv and Leder, 1972). Multiple tissue poly(A)+ Northern and dot blots were purchased from Clontech. Samples of total cytoplasmic RNA (15 mg) or of poly(A)+ RNA (1 ± 2 mg) were size fractionated on 1% agarose-formaldehyde gels and transferred to nitrocellulose or nylon membranes (Davis et al., 1986). Hybridization to a-32P-dCTP labeled DNA probes (speci®c activity 500 ± 2000 c.p.m./pg) were performed in 46 SSC, 40% formamide, 0.86 Denhardt's solution, 20 mg/ml

sheared herring sperm DNA, 10% dextran sulfate, 1% SDS, and 7 mM Tris-HCl overnight at 428C. Filters were washed in 0.16 SSC-0.1% sodium dodecyl sulfate at 588C and visualized by autoradiography. Chromosomal localization Localization of cbl-3 to a speci®c chromosome was performed by PCR analysis of DNA from a panel of somatic cell hybrid cell lines (NIGMS Panel #2) using cbl-3 speci®c primers (sense 5'-AAGCTGGCCATCATCTTCAGC3'; anstisense 5'-GTGAAGACGTCGAACTCGAAG-3'). Localization to a region of chromosome 19 was performed by PCR analysis of the RH01.02 Stanford G3 Radiation Hybrid Panel (Research Genetics). The PCR products were size fractionated on agarose gels, transferred to nitrocellulose membranes (Southern, 1975), and hybridized to a cbl-3 speci®c a-32P-dCTP labeled oligonucleotide probe (Davis et al., 1986). The Radiation Hybrid data were submitted to the Stanford Radiation Hybrid e-mail server for analysis. SH3 interactions Plasmids containing constructs for GST fusion proteins for the SH3 domains of Grb2, Gap, Fyn, Fgr, Lyn, Crk, Eps8, and a-Spectrin, and phospholipase Cg1 were obtained from Keith Robbins. Fusion proteins were prepared as previously described (Rivero-Lezcano and Robbins, 1994; RiveroLezcano et al., 1994). HA-epitope tagged cbl-3 or cbl-b protein was obtained from lysates of 293T cells transfected with either cbl-3L, cbl-3S or cbl-b constructs. To investigate protein interactions 500 mg of cell lysate containing cbl-3 or cbl-b protein were incubated in solution with 5 mg of puri®ed GST-fusion proteins pre-bound to agarose-GSH beads (Pharmacia) for 1 h with tumbling at 48C, washed ®ve times, boiled in sample buer, fractionated by 10% SDS ± PAGE, and transferred to PVDF membranes. Precipitated cbl protein was visualized by immunoblotting with an antiHA antibody as described below. Cell culture and transfections A nine amino acid epitope tag from the in¯uenza virus hemagglutin protein (HA) (Wilson et al., 1984) was added to the N-terminus of the full length long and short forms of cbl3 by PCR and the cDNAs were cloned into pCEFL, a mammalian expression vector with the elongation factor promoter and a neomycin selectable marker (provided by Dr Silvio Gutkind). The construct was sequenced to verify that there were no mutations introduced. The HA-epitope tagged erk-2 cDNA in the pcDNA3 vector was provided by Dr Silvio Gutkind and the EGFR cDNA in the LTR2 vector (Di Fiore et al., 1987) was provided by Dr Jacalyn Pierce. CFPAC-1 cells were obtained from ATCC and maintained in culture as recommended. The cells were harvested at midlog phase to isolate RNA as described above. 293 cells transfected with the SV40 large T antigen (293T) (provided by Mike Erdos) were maintained in culture in DMEM supplemented with 10% fetal calf serum (FCS) and 1% Penicillin-Streptomycin (Pen-Strep) and were transfected with various constructs using calcium phosphate (5 Prime?3 Prime, Inc.) according to the protocol included with the reagents. To measure the eects of EGF stimulation, 293T cells were grown to 70% con¯uence and serum starved in DMEM supplemented with 0.1% bovine serum albumin (BSA) and 1% Pen-Strep for 4 h. One-hundred ng/ml of recombinant human EGF (Collaborative Biomedical Products) was added for 10 min, the cells were washed two times in ice-cold PBS containing 0.2 mM sodium orthovanadate and the cells were lysed in ice-cold lysis buer (10 mM Tris HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X100, 10% glycerol, 1 mM AEBSF, 20 mg/ml leupeptin, 20 mg/ml

3373


3374

aprotinin, 10 mg/ml pepstatin, 2 mM sodium orthovanadate). The lysates were cleared of debris by centrifugation at 16 000 g for 15 min at 48C. HeLa cells (obtained from ATCC) were maintained in culture in DMEM supplemented with 10% FCS and 1% Pen-Strep and were then transfected with various constructs using Lipofectamine (GIBCO BRL) according to the method accompanying the reagent. Experiments to measure MAPK dependent transcriptional activation were performed by a modi®cation of the procedure described in the PathDetect manual (Stratagene). Cells were seeded in 6-well plates at 26105 cells/well and allowed to adhere overnight. Cells were transfected with 1 mg of the pFR-Luc luciferase reporter plasmid, 50 ng of the pFA2-Elk-1 plasmid, 400 ng of the EGFR plasmid, and the indicated amounts of the cbl-3 long form or cbl-3 short form. The total amount of DNA was kept constant by the addition of empty vector. Each transfection was done in six identical wells. After a 6 h incubation, the transfection medium was replaced with medium containing 0.5% FCS and the cells were allowed to grow overnight. Twenty-four hours after the transfection began, EGF (10 ng/ml) was added to three of each set of six wells and the cells were incubated for an additional 24 h. The cells were washed two times with ice-cold PBS and lysed in Cell Culture Lysis Reagent (Promega). The luciferase activity in each sample was determined using Luciferase Assay Reagent (Promega) and a Monolight 2010 luminometer (Analytical Luminescence Laboratory). The luciferase activity was corrected for the protein concentration in each sample. EGF stimulation (fold stimulation) was calculated as the luciferase activity in EGF stimulated cells/luciferase activity in unstimulated cells. The values for fold stimulation were compared using a two tailed Student's t-test.

Immunoprecipitation and immunoblotting Mouse monoclonal anti-HA (12CA5; Boehringer Mannheim) was used for immunoprecipitation. Rabbit polyclonal antiHA antibody (Y-11, Santa Cruz Biotechnology), anti-EGFR (1005, Santa Cruz Biotechnology), and phospho-speci®c antiMAPK antibody (9101, New England Biolabs) were used for immunoblotting. Horseradish peroxidase linked anti-phosphotyrosine (4G10; Upstate Biotechnology Inc.) was used for immunoblotting. Horseradish peroxidase linked donkey antirabbit Ig (Amersham) was used along with ECL detection reagent (Super Signal; Pierce) to visualize immunoblots. Immunoblotting was performed as previously described (Ausubel et al., 1994). For immunoprecipitation, protein from total cell lysate was incubated for 30 min on ice with antibody. Immune complexes were recovered by incubation with protein A/G+ agarose beads (Santa Cruz Biotechnology) at 48C for 1 h with tumbling. Immune complexes were washed ®ve times in cold lysis buer, resuspended in 26 loading buer (Promega), boiled for 5 min, and then resolved by 10% SDS ± PAGE. The gels were transferred to nitrocellulose membranes (Schleicher and Schuell) or to PVDF membranes (Immobilon P, Millipore). Denaturation of immunoprecipitated complexes and reprecipitatation of the cbl-3 protein was performed as previously described (Chen et al., 1995). Brie¯y the ®rst immunoprecipitation was performed as described above. The precipitated complex was resuspended and boiled for 5 min in 40 ml of denaturation buer (20 mM Tris-HCl pH 7.4, 50 mM NaCl, 5 mM DTT, 1% SDS, and 1 mM sodium orthovanadate), the denatured sample was diluted to 800 ml with cell lysis buer, and the second immunoprecipitation was again performed as described above.

References Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA and Struhl K (eds). (1994). Current procotols in molecular biology. Greene Publishing Associates, Inc. and John Wiley & Sons Inc. Aviv H and Leder P. (1972). Proc. Natl. Acad. Sci. USA, 69, 1408 ± 1412. Blake TJ, Shapiro M, Morse III HC and Langdon WY. (1991). Oncogene, 6, 653 ± 657. Bustelo XR, Crespo P, Lopez-Barahona M, Gutkind JS and Barbacid M. (1997). Oncogene, 15, 2511 ± 2520. Chen Y, Chen CF, Riley DJ, Allred DC, Chen PL, Von Ho D, Osborne CK and Lee WH. (1995). Science, 270, 789 ± 791. Chirgwin JM, Przybyla AE, MacDonald RJ and Rutter WJ. (1979). Biochemistry, 18, 5294 ± 5299. Cohen GB, Ren R and Baltimore D. (1995). Cell, 80, 237 ± 248. Corden J, Wasylyk B, Buchwalder A, Sassone-Corsi P, Kedinger C and Chambon P. (1980). Science, 209, 1406 ± 1414. Davis LG, Dibner MD and Battey JF. (1986). In: Basic Methods in Molecular Biology. 1st edn. Elsevier Science Publishing Co., New York. Di Fiore PP, Pierce JH, Kraus MH, Segatto O, King CR and Aaronson SA. (1987). Science, 237, 178 ± 182. Ettenberg SA, Keane MM, Nau MN, Frankel M, Wang LM, Pierce JH and Lipkowitz S. (1999). Oncogene, 18, 1855 ± 1866. Frohman MA, Dush MK and Martin GR. (1988). Proc. Natl. Acad. Sci. USA, 85, 8998 ± 9002. Hime GR, Dhungat MP, Ng A and Bowtell DD. (1997). Oncogene, 14, 2709 ± 2719. Janknecht R, Ernst WH, Pingoud V and Nordheim A. (1993). EMBO J., 12, 5097 ± 5104.

Jongeward GD, Clandinin TR and Sternberg PW. (1995). Genetics, 139, 1553 ± 1566. Keane MM, Rivero-Lezcano OM, Mitchell JA, Robbins KC and Kipkowtiz S. (1995). Oncogene, 10, 2367 ± 2377. Kozak M. (1986). Cell, 44, 283 ± 292. Langdon WY, Hartley JW, Klinken SP, Ruscetti SK and Morse III HC. (1989a). Proc. Natl. Acad. Sci. USA, 86, 1168 ± 1172. Langdon WY, Hyland CD, Grumont RJ and Morse III HC. (1989b). J. Virol., 63, 5420 ± 5424. Levkowitz G, Waterman H, Zamir E, Kam Z, Oved S, Langdon WY, Beguinot L, Geiger B and Yarden Y. (1998). Genes Dev., 12, 3663 ± 3674. Lovering R, Hanson IM, Borden KL, Martin S, O'Reilly NJ, Evan GI, Rahman D, Pappin DJ, Trowsdale J and Freemont PS. (1993). Proc. Natl. Acad. Sci. USA, 90, 2112 ± 2116. Lupher Jr ML, Reedquist KA, Miyake S, Langdon WY and Band H. (1996). J. Biol. Chem., 271, 24063 ± 24068. Lupher Jr ML, Songyang Z, Shoelson SE, Cantley LC and Band H. (1997). J. Biol. Chem., 272, 33140 ± 33144. Meisner H, Daga A, Buxton J, Fernandez B, Chawla A, Banerjee U and Czech MP. (1997). Mol. Cell. Biol., 17, 2217 ± 2225. Miyake S, Lupher Jr ML, Andoniou CE, Lill NL, Ota S, Douillard P, Rao N and Band H. (1997). Crit. Rev. Onco., 8, 189 ± 218. Miyake S, Lupher Jr ML, Druker B and Band H. (1998). Proc. Natl. Acad. Sci. USA, 95, 7927 ± 7932. Murphy MA, Schnall RG, Venter DJ, Barnett L, Bertoncello I, Thien CBF, Langdon WY and Bowtell DDL. (1998). Mol. Cell. Biol., 18, 4872 ± 4882. Ota Y and Samelson LE. (1997). Science, 276, 418 ± 420.


Payne DM, Rossomando AJ, Martino P, Erickson AK, Her JH, Shabanowitz J, Hunt DF, Weber MJ and Sturgill TW. (1991). Embo J., 10, 885 ± 892. Rivero-Lezcano OM and Robbins KC. (1994). FEBS Lett., 338, 183 ± 186. Rivero-Lezcano OM, Sameshima JH, Marcilla A and Robbins KC. (1994). J. Biol. Chem., 269, 17363 ± 17366. Sanger F, Nicklen S and Coulson AR. (1997). Proc. Natl. Acad. Sci. USA, 74, 5463 ± 5467. Savage PD, Shapiro M, Langdon WY, Geurts van Kessel AD, Seuanez HN, Akao Y, Croce C, Morse HD and Kersey JH. (1991). Cytogenet. Cell. Genet., 56, 112 ± 115. Schoumacher RA, Ram J, Iannuzzi MC, Bradbury NA, Wallace RW, Hon CT, Kelly DR, Schmid SM, Gelder FB, Rado TA and Frizzell RA. (1990). Proc. Natl. Acad. Sci. USA, 87, 4012 ± 4016. Short JM, Fernandez JM, Sorge JA and Huse WD. (1988). Nucleic Acids Res., 16, 7583 ± 7600.

Smit L and Borst J. (1997). Crit. Rev. Oncog., 8, 359 ± 379. Southern EM. (1975). J. Mol. Biol., 98, 503 ± 517. Thien CB and Langdon WY. (1997). Oncogene, 14, 2239 ± 2249. Ueno H, Sasaki K, Miyagawa K, Honda H, Mitani K, Yazaki Y and Hirai H. (1997). J. Biol. Chem., 272, 8739 ± 8743. Wang Y, Yeung YG, Langdon WY and Stanley ER. (1996). J. Biol. Chem., 271, 17 ± 20. Wilson IA, Niman HL, Houghton RA, Cherenson AR, Connolly ML and Lerner RA. (1984). Cell, 37, 767 ± 778. Wilusz J, Pettine SM and Shenk T. (1989). Nucleic Acids Res., 17, 3899 ± 3908. Yoon CH, Lee J, Jongeward GD and Sternberg PW. (1995). Science, 269, 1102 ± 1105.

3375