Role of TFG Sequences Outside the Coiled-Coil Domain in TRK-T3 Oncogenic Activation

Role of TFG sequences outside the coiled-coil domain in TRK-T3 oncogenic activation

Emanuela Roccato1, Sonia Pagliardini1, Loredana Cleris2, Silvana Canevari3, Franca Formelli2, Marco A Pierotti1,4 and Angela Greco*,1,4

1Operative Unit #3, Department of Experimental Oncology, Istituto Nazionale Tumori, Via G. Venezian, 1 20133 Milano, Italy; 2Operative Unit #13, Department of Experimental Oncology, Istituto Nazionale Tumori, Via G. Venezian, 1 20133 Milano, Italy; 3Operative Unit #11, Department of Experimental Oncology, Istituto Nazionale Tumori, Via G. Venezian, 1 20133 Milano, Italy

The TRK-T3 oncoprotein, isolated from a human sequences from different activating genes (Pierotti papillary thyroid tumor, arises from the fusion between et al., 1996). Several thyroid TRK oncogenes have been the N-terminal domain of the TFG gene and the tyrosine isolated in our laboratory and named TRK, TRK-T1, kinase domain of the NTRK1 receptor. The 68 kDa TRK- TRK-T2 and TRK-T3. TRK oncogene is activated by T3 oncoprotein displays a constitutive tyrosine kinase TPM3, encoding the nonmuscle tropomyosin (Martin- activity resulting in its capability to transform NIH3T3 Zanca et al., 1986); TRK-T1 and TRK-T2 are both cells. The TFG portion of TRK-T3 contains a coiled-coil activated by TPR gene, encoding a filamentous protein domain, which mediates protein oligomerization essential of the nuclear pore complex (Byrd et al., 1994); in TRK- for the oncogene constitutive activation, and several T3 oncogene, the activating portion is contributed by consensus sites for protein interaction. In this study, we TFG, a novel gene on chromosome 3 encoding a protein investigate the role of TFG sequences outside the coiled- of yet unknown function (Greco et al., 1995). All the coil domain on TRK-T3 activation, We constructed four activating sequences contain dimerization/oligomeriza- mutants carrying different deletions of TFG sequences tion domains playing a crucial role in oncogenic and expressed them in mammalian cells. By performing activation by inducing constitutive, ligand-independent biochemical and biological assays we demonstrated that tyrosine kinase activity. Analysis of signal transduction all the deleted regions are required for TRK-T3 activa- demonstrated that Shc is activated by TRK oncopro- tion, as they are involved in different mechanisms such as teins and therefore capable to recruit Grb2 (Borrello protein processing, formation of stable and/or functional et al., 1994). We have more recently also determined the complexes, and possible interaction with other proteins. activation of FRS2, FRS3, PLCg, ERK1/2 and JNK By constructing site-specific mutants, we demonstrated a MAP kinases in cells transformed by TRK-T3 (Greco crucial role for a PB1 domain and a considerable et al., manuscript in preparation), as well as the contribution of an SH2-binding motif in TRK-T3 recruitment of IRS1 and IRS2 by TRK-T1 oncoprotein oncogenic activation. This workestablishes an important (Miranda et al., 2001). role for TFG sequences outside the coiled-coil domain in The thyroid TRK-T3 oncogene encodes a cytoplasmic the activation of the thyroid TRK-T3 oncogene. protein made of 593 amino acids, 193 of which are Oncogene (2003) 22, 807–818. doi:10.1038/sj.onc.1206189 contributed by TFG (TRK Fused Gene), a novel gene first discovered in the rearranged form, encoding a Keywords: TRK-T3; coiled-coil domain; TFG; protein whose major characteristic is the presence of a NTRK1; oncogenic activation coiled-coil domain (Greco et al., 1995). Recently, the normal TFG counterpart was cloned and characterized. TFG gene is ubiquitously expressed in human adult tissues and it is conserved among several species, Introduction including Caenorhabditis elegans. In addition to the coiled-coil domain, the TFG protein also contains TRK oncogenes, created by structural rearrangements putative phosphorylation sites for PKC and CK2, of the NTRK1/NGF receptor gene (Kaplan et al., 1991; glycosylation sites, as well SH2- and SH3-binding sites Klein et al., 1991), are associated with a consistent (Mencinger et al., 1997). Several of these sites are fraction of human papillary thyroid carcinoma. The identical in TFG proteins from different species, rearrangements produce chimeric proteins containing indicating that the protein might be involved in basic the NTRK1 tyrosine kinase domain preceded by cell processes (Mencinger and Aman, 1999). Interest- ingly, most of these sites are present in the TFG portion *Correspondence: A Greco; E-mail: [email protected] contained in TRK-T3. 4Senior co-authors The coiled-coil structural motif comprises two or Received 19 June 2002; revised 22 October 2002; accepted 24 more right-handed a-helices wrapped around each other October 2002 with a small, left-handed super helical twist. This Mechanism of activation of TRK-T3 oncogene E Roccato et al 808 domain confers to the proteins the capability to fold investigations. On the whole, our results clearly demon- into coiled-coils, thus mediating oligomerization (Lupas strate that TFG activating sequences regulate the TRK- et al., 1991; Lupas, 1996). Recent studies based on gel T3 biological activity not only by promoting its filtration chromatography have shown that proteins constitutive oligomerization. bearing coiled-coil domains form complexes containing a variable molecule number, ranging from two to six (Matthews et al., 2001). Coiled-coil domains are present in almost all the sequences activating oncogenes derived Results from receptor and nonreceptor tyrosine kinases (Ro- drigues and Park, 1994), and their importance in Construction and biochemical analysis of TRK-T3 mediating oligomerization leading to constitutive acti- deletion mutants vation has been documented in several cases. These To define the role of the TFG regions outside the coiled- include the MET (Rodrigues and Park, 1993), the RET/ coil domain in TRK-T3 oncogenic activation, we have PTC2 (Durick et al., 1995), the RET/PTC 1 (Tong et al., constructed the four deletion mutants shown in Figure 1. 1997), the TRK-T3 (Greco et al., 1998), the RET/PTC3 The mutants were constructed by PCR amplification of (Monaco et al., 2001), the Bcr-Abl (McWhirter et al., TRK-T3-specific regions, and in all of them the TFG 1993) and the PML-RAR (Grignani et al., 1999) initiation codon was maintained (see Materials and oncogenes. However, also other regions, different from methods). The T3/L1 mutant, already analysed in a the coiled-coil domain, might be important for onco- previous study (Greco et al., 1998), carries the deletion protein activation. This has been demonstrated for Bcr- of the TFG portion preceding the coiled-coil domain; in Abl, whose activation requires also a region containing T3/L2, the sequences preceding and following the an SH2 domain (McWhirter et al., 1993), and for TRK- coiled-coil motif have been removed; T3/L3 carries a T3, whose transforming activity requires the N-terminal small N-terminal deletion (aa 2–27); in T3/L4 the region region of TFG (Greco et al., 1998). following the coiled-coil domain has been removed. The The TFG portion contained in TRK-T3 oncoprotein mutants were inserted into the pIRESneo bicistronic can be divided into three regions: the N-terminal region expression vector, which carries the neomycin resistance (aa 1–96); the coiled-coil domain (aa 97–124); the region gene as selectable marker. between the coiled-coil domain and the junction to TRK-T3 constructs were transiently transfected into NTRK1 (aa 125–193). In a previous work, we have 293T cells, and cell extracts were immunoprecipitated investigated the contribution of TFG coiled-coil domain with anti-TRK antibodies. Western blot analysis with to TRK-T3 oncogenic activation. By constructing antiphosphotyrosine and anti-TRK antibodies was several mutants, we have shown that the TRK-T3 performed (Figure 2a). All the mutants expressed activity depends on oligomerization mediated by the coiled-coil domain. However, as mentioned above, we have also shown that the TFG N-terminal portion, although not interfering with protein self-association, is required for TRK-T3 fully transforming activity (Greco et al., 1998), suggesting that the regions outside the coiled-coil domain are also important for oncogenic activation. In the present work, we have further extended this concept by investigating the role of TFG regions outside the coiled-coil domain in TRK-T3 oncogenic activity. This issue is particularly interesting, based on the evidence that TFG contains a single coiled-coil domain, but also displays consensus sites for interaction with other proteins. This raises the question whether or not a coiled-coil domain is sufficient to activate RTK oncogenes. We have constructed four deletion mutants, lacking different portions of TFG sequences outside the coiled-coil domain, and we have assayed their Figure 1 Schematic representation of TRK-T3 constructs used in biochemical and biological activities. We provide this study. The portions contributed by TFG and NTRK1 are evidence that the different deletions interfere with shown. The coiled-coil (CC), transmembrane (TM) and tyrosine protein activation, processing and capability to form kinase (TK) domains are indicated. Putative phosphorylation sites for PKC and CK2 are indicated with circles and ‘v’, respectively. functional complexes. To unveil the specific TFG Asterisks indicate the glycosylation site. SH2- and SH3-binding elements responsible for the TRK-T3 regulation, we motifs are indicated by empty circles and arrowheads, respectively. abrogated consensus sites for protein interaction such as The PB1 domain is indicated by a dotted underline. Tyrosine PB1, SH2- and SH3-binding motifs. Our data showed residues involved in Shc and FRS2 interaction (Y291), PLCg interaction (Y586), and tyrosines of the activation loop (Y470, that the PB1 and SH2-binding motifs give an important Y475 and Y476) are indicated. The structure of the wild-type TRK- contribution to TRK-T3 oncogenic activation, whereas T3 protein (T3/WT) is shown at the top. In the mutated proteins the role of the SH3-binding domain requires further the deleted regions are indicated by bridging lines

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 809

Figure 2 Biochemical analysis of TRK-T3 wild-type and mutated proteins. (a) The TRK-T3 constructs and the pIRES empty vector were transfected into 293T cells; cell extracts from the same transfection were used for all the Western blots. Samples were immunoprecipitated (IP) with anti-TRK, anti-PLCg, anti-Shc antibodies or pulled-down with GST-FRS2(PTB) fusion proteins and immunoblotted with antiphosphotyrosine (pTyr), anti-TRK, anti-PLCg or anti-Shc antibodies. The TRK-T3 proteins, the PLCg protein and the three Shc isoforms (p46, p52, and p66) are indicated, (b) Immunokinase assay of TRK-T3 wild-type and mutated proteins. Cell extracts from 293T cells transiently transfected with TRK-T3 constructs or the pIRES empty vector were immunoprecipitated with anti-TRK antibodies and subjected to a kinase assay (upper panel) as described in Materials and methods. TRK-T3 protein expression levels, detected by using 125I-labeled protein A (Amersham), are shown in the lower panel. The relative levels of TRK-T3 proteins and their kinase activity were visualized and quantiﬁed by using Phosphorlmager apparatus (Typhoon 8600, Amersham). Ratios of autophosphorylation to TRK-T3 proteins were calculated and are presented as means 7 s.d. of three different experiments. A TRK-T3 kinase dead mutant is shown as control phosphorylated proteins whose molecular weight was namely Shc, PLCg and FRS2 (Figure 2a). Hybridization estimated as 68 kDa for T3/WT, 55 kDa for T3/L1, of PLCg and Shc immunocomplexes with antipho- 45 kDa for T3/L2, 63 kDa for T3/L3 and 53 kDa for T3/ sphotyrosine antibodies showed that all the TRK-T3 L4. It is important to note that T3/L4 presented an proteins co-precipitate with the two adaptors. PLCg electrophoretic migration faster than T3/L1, although it and Shc were phosphorylated, and the extent of carries a shorter deletion; this discrepancy might be activation correlated with the level of TRK-T3 proteins related to the effect of deletion on protein structure. phosphorylation. Similarly, pull-down experiments With respect to tyrosyl phosphorylation, it was reduced with GST-FRS2(PTB) fusion protein, followed by in T3/L1 and T3/L3, and drastically reduced in T3/L2; Western blot analysis with antiphosphotyrosine anti- whereas the T3/L4 mutant showed a phosphorylation bodies, showed that the TRK-T3 mutated proteins level slightly increased with respect to wild type. retain the capability to activate FRS2, although to We next analysed the effect of TFG deletions on different extents. In particular, the T3/L3 protein, TRK-T3 ability to recruit several signal transducers, slightly visible in Figure 2a (last panel), was clearly

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 810 detected after a longer exposure (data not shown). activity of TRK-T3 proteins on the exogenous substrate Altogether these results show that TRK-T3 mutants MBP (data not shown). The differences in kinase retain the capability to recruit/activate PLCg, Shc and activity are comparable to those in steady-state phos- FRS2 signal transducers. phorylation shown in Figure 2a. To determine the effect of TFG deletions on TRK-T3 By immunoﬂuorescence of transiently transfected intrinsic kinase activity, we performed immunokinase NIH3T3 cells, we investigated the cellular localization assay (Figure 2b). Cell extracts from 293T cells of mutated TRK-T3 proteins. All of them showed transfected with TRK-T3 wild-type or deleted mutants cytoplasmic localization, similarly to the wild type were immunoprecipitated with anti-TRK antibody. The (Figure 3a). Interestingly, T3/L2-expressing cells immunocomplexes were incubated with [g-32P]ATP in showed dispersed punctate cytoplasmic distribution the presence of exogenous substrate myelin basic protein concentrate in visible granules. This is reminiscent of (MBP), or subjected to Western blot hybridization with aggresomes, intracellular structures formed by the anti-TRK antibodies, to determine the protein level. The aggregation of misfolded, insoluble proteins (Johnston quantiﬁcation of relative kinase activity resulted from et al., 1998), and suggests that the T3/L2 mutation the ratio between the TRK-T3 autophosphorylation might interfere with the oncoprotein processing. Indeed, level with respect to protein expression is shown in the Western blot analysis of transiently transfected 293T bar graph. Autokinase activity was reduced by 30% in cells showed that T3/L2 is present in both soluble and T3/L1 and T3/L3, and was almost undetectable for T3/ insoluble fractions, but slightly more abundant in the L2, whereas it was increased in T3/L4 (Figure 2b). insoluble one, at variance of T3/WT and T3/L1 proteins Similar results were obtained analysing the kinase (Figure 3b).

Figure 3 Cellular localization of TRK-T3 proteins, (A) NIH3T3 cells transiently transfected with TRK-T3 wild type (a) or mutated proteins T3/L1 (b), T3/L2 (c), T3/L3 (d). T3/L4 (e) were stained with anti-TRK antibodies/labeled secondary antibody. Images were obtained with a confocal microscope. Scale bar ¼ 15 mm. (B) Western blot analysis of 293T cells transfected with TRK-T3 constructs. Soluble and insoluble proteins were extracted as described in Materials and methods and analysed by immunoblotting with anti-TRK antibodies. Arrowheads indicate the wild-type and mutated TRK-T3 proteins

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 811 Self-association and complex formation of wild-type and that TRK-T3 forms high molecular weight complexes deleted TRK-T3 oncoproteins (>220 kDa) (Greco et al., 1995). To better deﬁne TRK- T3 complexes we performed HPLC size-exclusion The TFG portion of TRK-T3 contains a coiled-coil chromatography (Superdex 200 HR10/30; Pharmacia). domain driving the oncoprotein self-association impor- Extracts prepared from 293T cells expressing the tant for full oncogenic activity. To determine if the different TRK-T3 constructs were fractioned by size- regions outside the coiled-coil domain contribute to the exclusion chromatography, and samples of each fraction TRK-T3 self-association, we performed in vitro binding were immunoblotted with anti-TRK antibodies assays with a GST/T3 fusion protein, made of the (Figure 4b). The T3/WT protein was eluted in fractions glutathione-S-transferase fused in frame with TRK-T3 corresponding to high molecular masses of approxi- protein (from aa 23 to C-terminus) (Greco et al., 1998). mately 400–480 kDa. Since the molecular weight of The mutated proteins were translated in vitro in the monomer TRK-T3 wild type is 68 kDa, such complexes presence of 35S-labeled methionine (Figure 4a, left might contain six molecules; alternatively, they might panel). Samples containing approximately the same contain other proteins in addition to TRK-T3. amount of each TRK-T3 protein were incubated with Although the complex composition remains to be the GST/T3 fusion protein conjugated to glutathione- determined, the relevance of this study is the demonstra- sepharose. All the labeled TRK-T3 proteins were eluted tion that the wild-type TRK-T3 protein is part of the from the GST complexes (Figure 4a, right panel), high molecular weight complexes. Based on these indicating that the capability to associate in vitro with evidences we next investigated the effect of mutations TRK-T3 wild type was not abrogated by the deletions. on complex formation by performing size-exclusion Computer-assisted analysis with the MultiCoil score chromatography and Western blot analysis with anti- program indicated that the TFG coiled-coil domain has TRK antibodies for all the mutants, as described above high probability to fold into trimers (data not shown). (Figure 4b). T3/L1 protein showed a peak at 400– In a previous work, based on the sedimentation proﬁle 480 kDa, but was also detected in fractions correspond- on a sucrose density gradient, we have demonstrated ing to lower molecular weights. T3/L2 protein, which is

Figure 4 Self-association and complex formation of wild-type and mutated TRK-T3 proteins, (a) In vitro self-association assay. cDNA constructs were transcribed and translated in vitro as described in Materials and methods. The in vitro translated proteins were electrophoresed before (left) and after (right) incubation with the sepharose-bound GST/T3 fusion protein (see Materials and methods), (b) Lysates of 293T cells expressing TRK-T3 proteins were fractionated by size-exclusion chromatography, and samples from fractions were resolved by SDS–PAGE and immunoblotted using anti-TRK antibodies. The TRK-T3 wild-type complexes elute in fractions 18 and 19, which correspond to molecular masses of approximately 480–400 kDa. The fraction numbers are indicated. The fractions corresponding to molecular size standards thyroglobulin (670 kDa), gamma globulin (158 kDa), and ovalbumin (44 kDa) are indicated. The molecular weight of TRK-T3 proteins is indicated

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 812 deﬁnitely smaller than wild-type, was eluted in the same high molecular weight fractions as the wild-type protein. This pattern is compatible with the formation of abnormal complexes, containing either an aberrant number of TRK-T3 molecules or proteins not recruited by wild-type TRK-T3, such as proteins of the degradation pathway. The T3/L3 protein, with molecular weight similar to the wild type (63 kDa), showed a more distributed elution pattern, ranging from 70 to 480 kDa, indicating a reduced capability to form functional complexes. T3/L4 protein showed an elution pattern similar to T3/WT and T3/L1, indicating no effect of the deletion in complex formations.

Differentiating activity is retained by TRK-T3 deletion mutants Similarly to the NGF-stimulated NTRK1 receptor, the TRK oncogenes are capable of inducing differentiation Figure 5 Quantification of differentiating activity of TRK-T3 wild of PC12 cells. To test the effect of TFG deletions on this type and mutants. PC12 cells were cotransfected with TRK-T3 activity, PC12 cells were cotransfected with the different constructs, VGF8-luc, and Renilla plasmids, as described in TRK-T3 constructs and the VGF8-luc reporter plasmid: Materials and methods. Both luciferase activities were measured the latter is made of the luciferase reporter gene driven using the Dual-Luciferase reported assay system (Promega). The values were expressed relative to Renilla luciferase activity for by the promoter of the vgf, a gene induced by NGF and normalization. The results presented are an average of four considered a differentiation marker (Possenti et al., experiments 1992; Canu et al., 1997). All the mutated proteins were able to induce formation of neurites (data not shown). However, PC12 cells expressing T3/L2 and T3/L3 domain may have a crucial role in TRK-T3 oncoprotein proteins displayed shorter neurites with respect to T3/ transforming activity, although they abrogate neither WT-, T3/L1- and T3/L4-expressing cells. To quantify the constitutive oncogene activation, nor its differentiat- the differentiating activity, we measured the capability ing capability. of TRK-T3 proteins to induce the expression of the We have previously observed that the expression of a VGF8-luc reporter gene by performing luciferase assays transfected gene carried by a mammalian expression on the extracts of cotransfected PC12 cells described vector increases if transfected cells are cultured for 10– above. As shown in Figure 5, the luciferase activity was 15 days in the presence of the antibiotic whose resistance slightly increased in cells expressing T3/L1; in cells gene is on the same vector (Roccato et al., 2002). We transfected with T3/L2 and T3/L3 constructs, a decrease were interested in determining whether the TRK-T3 of luciferase activity (54 and 45%, respectively) was mutants transforming activity might be modulated by observed; a minor reduction (20%) was detected in cells the oncoprotein expression level. To this aim, trans- expressing T3/L4. These data suggest that all TRK-T3 fected NIH3T3 cells were first selected in a medium deleted mutants retain the capability to promote PC12 containing G418 for 2 weeks, then shifted to 5% serum cells differentiation, although to a different extent. medium, and scored for foci 2 weeks later (see Materials and methods). As reported in Table 2 (third column), by Transforming activity of mutated TRK-T3 proteins applying G418 preselection foci were induced not only by T3/WT and T3/L1, but also by mutants T3/L2, T3/ To determine the effect of TFG deletions on TRK-T3 L3 and T3/L4, which were scored negative in the oncogenic activity, we analysed the capability of TRK- previous analysis (Tables 1 and 2, first column). On the T3 mutants to transform NIH3T3 cells. Transfected contrary, no foci arose in NIH3T3 cells transfected with cells were selected in the presence of G418, to determine the empty vector, and no recovery of transforming the transfection efficiency, or in 5% serum medium, to activity was observed for mutants T3/L2. T3/L3 and T3/ determine the transforming activity. The results of a L4 by simply extending to 4 weeks the selection time in representative focus formation experiment are reported 5% serum, without G418 preselection (Table 2, second in Table 1. The T3/L1 mutant induced foci with column). Altogether, these data suggest that transform- efficiency reduced by 80% with respect to the wild type. ing activity is not completely abrogated in the TRK-T3 In a previous study, with a different expression vector, mutants since it can be recovered by enhancing the the T3/L1 mutant failed to induce foci formation (Greco TRK-T3 proteins expression level. Thus, the achieve- et al., 1998); this discrepancy is most likely related to ment of a threshold oncoprotein level is necessary to differences in expression level, as it will be discussed induce the transformed phenotype. In keeping with later. On the contrary, no transforming activity was these results, NIH3T3 clones, selected and cultured in detected for mutants T3/L2, T3/L3 and T3/L4. These the presence of G418, and expressing considerable data suggest that the regions outside the coiled-coil amount of TRK-T3 mutated proteins, displayed trans-

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 813 Table 1 Effect of TFG deletions on TRK-T3 transforming activity Transfected plasmid G418R colonies (N/mgDNA) Â 102 Foci (N/mgDNA) Â 102 Foci/G418R colonies Â 10À2 Relative transforming activity T3/WT 19.33 9.92 50 1 T3/L1 27.25 2.53 10 0.2 T3/L2 18.03 o0.01 o0.01 o0.01 T3/L3 11.27 o0.01 o0.01 o0.01 T3/L4 25.33 0.4 1.5 0.03 pIRES 34.00 o0.01 o0.01 o0.01

(NIH3T3 cells were transfected with wild-type or mutated TRK-T3 plasmids or empty vector (pIRES), and a focus formation assay was performed. Number of foci, number of G418 resistant colonies and relative transforming activity are shown)

Table 2 Effect of G418 preselection on TRK-T3 deleted mutants injecting 103 cells: T3/L2 cell line failed to induce tumor transforming activity. formation, whereas the other cell lines induced tumor 10% serum+G418 formation in 3/3 mice, with latency longer than wild Transfected 5% serum 5% serum (2 weeks) 5% type. The tumor weight rate was comparable for all the Plasmid 2 weeks 4 weeks serum (2 weeks) cell lines. Since no differences in growth rate were T3/WT + + + detected by in vitro proliferation assay (data not shown), T3/L1 + + + the differences in tumor latency were most likely related T3/L2 — — + to the intrinsic tumorigenicity of the cell lines. These T3/L3 — — + data indicate that all the TFG deletions interfere with T3/L4 — — + PIRES — — — TRK-T3 tumorigenicity, in agreement with the in vitro NIH3T3 transformation results. NIH3T3-transfected cells were selected in 5% serum medium for 2 or 4 weeks (first and second column, respectively), or selected in G418 for 2 weeks, then shifted in 5% serum medium for other 2 weeks (third Construction and analysis of site-specific TRK-T3 column). Afterwards, selection plates were scored for foci. + and À mutants indicate the foci presence and absence, respectively The data reported in the previous paragraphs showed that the transforming activity was reduced in all the formed phenotype indistinguishable from the wild type. TRK-T3 deletion mutants. Whereas the defects of T3/ In fact, they were spindle shaped, less adherent than L2 and T3/L3 mutants can be ascribed to impairments untransfected NIH3T3 cells, and showed a complete of processing and complex formation, respectively, such disorganization of the actin stress fibers (data not properties are unaffected in T3/L1 and T3/L4 mutants. shown). Interestingly, some important sites, suggesting possible To get more insight into the transforming potential of interaction with other proteins, are present in the TFG the TRK-T3 mutants, we analysed their capability to region. A computer-assisted search with the SMART induce tumor formation in vivo. The NIH3T3 cell lines program revealed the presence of a PB1 domain (aa 10– expressing TRK-T3 proteins and showing transformed 91), presumably involved in protein–protein interaction phenotype described above were injected into nude (Ito et al., 2001). Such domain is deleted completely or mice. For each cell line, two different doses (104 and partially in T3/L1 and T3/L3 mutant, respectively. 103 cells) were used. By injecting 104 cells per animal, all Moreover, a putative SH2-binding site is removed by the cell lines induced tumor formation in 3/3 mice, the T3/L1 deletion, and a putative SH3-binding site is except T3/L2, which induced tumor formation in 2/3 contiguous to the coiled-coil domain (Figure 1). To mice. They differed in the latency, which was delayed of determine the role of these specific domains in TRK-T3 3 days for T3/L1, 5 days for T3/L3, 13 days for T3/L4 oncogenic activation, we constructed three site-specific and 16 days for T3/L2 with respect to T3/WT (Table 3). mutants (Figure 6a). T3/K14A carries the mutation of The difference in tumorigenicity was emphasized by the conserved Lys residue within PB1 domain; such

Table 3 In vivo tumor growth induced by TRK-T3 mutants Cell line 104 cells 103 cells Tumor latency Tumor weight Tumor latency Tumor weight Tumor-bearing (days at 0. 1 g) (days at 1 g) Tumor-bearing (days at 0. 1 g) (days at 1 g) mice (mean7s.d.) (mean7s.d.) mice (mean7s.d.) (mean7s.d.) T3/WT 3/3 1371.0 1871.0 3/3 2273.0 2873.1 T3/L1 3/3 1670.6 2471.5 3/3 2772.9 39.574.0 T3/L2 2/3 2970.7 3572.1 0/3 / / T3/L3 3/3 1871.5 2571.1 3/3 2872.0 3472.9 T3/L4 3/3 2672.0 3876.1 3/3 3576.6 4878.5

Mice were injected s.c. with 104 or 103 NIH3T3-transfected cells. Mice were checked for tumor latency and tumor growth as described in Materials and methods. Results derived from a representative experiment are shown.

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 814

Figure 6 Biochemical and biological analysis of site-speciﬁc TRK-T3 mutants. (a) The amino-acids sequence of the TFG portion contained in TRK-T3 is shown. PB1 domain is indicated by a dotted underline; coiled-coil domain is boxed; SH2-binding motif is written in italic; SH3-binding motif is indicated by a double-dashed underline. Boldface letters indicate the residues mutated in T3/ K14A, T3/Y33F and deleted in T3/nPro proteins. (b) Phosphorylation and expression of TRK-T3 mutated proteins. The TRK-T3 constructs and the pIRES empty vector were transfected into 293T cells. Cell extracts were immunoprecipitated with anti-TRK antibodies and then subjected to Western blot with antiphosphotyrosine antibodies (pTyr) (upper panel) or anti-TRK antibodies (lower panel). (c) Transforming activity of TRK-T3 mutated proteins. NIH3T3 cells were transfected with wild-type or mutated TRK- T3 plasmids or empty vector (pIRES), and a focus formation assay was performed. Plates derived from a representative focus assay are shown. For each mutant the relative transforming activity (RTA), calculated as the ratio between number of foci and number of G418- resistant colonies, is reported. Transforming activity of T3/WT was set at 100%.

residue in Bem1p has been shown to be important for proteins recruited by TFG may be involved in this the interaction with the cognate PC domain of Cdc24p process. (Ito et al., 2001; Terasawa et al., 2001). T3/Y33F carries the mutation of the Tyr residue within the putative SH2- binding domain. In T3/DPro the putative SH3-recogni- tion motif (PPGEPGP) was deleted. The mutants were Discussion constructed by site-directed mutagenesis using the TRK- T3 wild-type cDNA inserted into the pIRESneo The thyroid TRK-T3 oncoprotein is composed of the expression vector as template (see Materials and NTRK1 tyrosine kinase domain fused in frame with the methods). The mutated constructs were transiently N-terminal portion of TFG, a novel protein containing transfected into 293T cells; Western blot analysis a coiled-coil domain, putative phosphorylation sites for showed that they are properly expressed and phos- PKC and CK2, SH2- and SH3-binding consensus sites phorylated (Figure 6b). The effect of the site-specific (Mencinger et al., 1997) and a PB1 motif identified in mutations on TRK-T3 activity was investigated by this study. NIH3T3 transfection/focus formation assay. The results Coiled-coil domains promote oncoproteins multimer- are reported in Figure 6c. T3/K14A mutant displayed a ization that triggers constitutive transautophosphoryla- very low transforming activity, comparable to the tion and oncogenic activation. We have previously background. T3/Y33F transforming activity was re- shown that the TFG coiled-coil domain mediates the duced by 80% with respect to wild type and equivalent capability of TRK-T3 to form complexes and therefore to T3/L1. T3/DPro induced transformation with effi- is essential for its oncogenic activation (Greco et al., ciency slightly reduced with respect to wild type. These 1998). data demonstrate that the mutation of the specific An attractive hypothesis is the possibility that fusion interaction motifs analysed affects strongly or at least in partners may contribute to the activation of rearranged part the TRK-T3 oncogenic activation, suggesting that tyrosine kinase oncogenes with additional functions,

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 815 beside oligomerization. Therefore, we have recently although it is smaller and, presumably, unable to focused our attention on the TFG regions outside the interact with putative proteins recruited by the deleted coiled-coil domain and analysed their role in TRK-T3 sequences. The T3/L2 sedimentation profile may be due oncogene activating mechanism. To this aim, we either to the formation of homocomplexes with an constructed four deletion mutants lacking different aberrant molecule number, or to the presence of portions of TFG sequences external to the coiled-coil proteins not recruited by wild-type TRK-T3, such as domain. The mutated proteins, transiently expressed proteins of the degradation machinery. into 293T cells, displayed different kinase activity. T3/ Analysis of TRK-T3 mutants biological activity L4 showed an increased activity with respect to wild showed apparently different results in PC12 and type; in the other mutants, the kinase activity was NIH3T3 cells. In fact, by transfection of PC12 cells reduced to a different extent. All the mutated proteins we showed that all the mutants maintain the differ- are able to recruit and activate PLCg, Shc and FRS2 entiating activity, although a reduction for T3/L2 and pathways, and such capability correlates, in general, T3/L3 was observed. On the contrary, NIH3T3 trans- with their phosphorylation level. fection experiments showed that the T3/L1 capability to Immunofluorescence experiments of transfected cells form foci was reduced by 80% with respect to the wild showed that TRK-T3 wild-type and mutated proteins type, whereas no transforming capability was observed are located in the cytoplasm. The peculiar punctuate for the other mutants. However, we detected transform- cellular distribution concentrate in visible granules of ing activity of TRK-T3 mutants by enhancing the T3/L2 protein suggested a defect in protein processing, protein expression level with G418 preselection of further confirmed by the detection of a considerable transfected NIH3T3 cells before foci selection. This amount of T3/L2 in the insoluble protein pool of procedure allows the achievement of a threshold transiently transfected cells. oncoprotein level necessary to induce the transformed The TFG coiled-coil domain is predicted to fold into phenotype. Transformation induced by the mutated trimers, and we have previously shown that it triggers proteins was morphologically indistinguishable from the TRK-T3 multimerization (Greco et al., 1998). By wild type. However, NIH3T3 cell lines transformed by removing TFG sequences outside the coiled-coil do- TRK-T3 mutants differed from wild type in in vivo main, TRK-T3 oncogene maintains its in vitro self- tumorigenicity, since they showed a significant delay in association capability, as demonstrated by pull-down tumor formation. experiments with the GST/T3 fusion protein. By size- TFG contains several putative sites for interaction exclusion chromatography we have determined that with other proteins. The presence of SH2- and SH3- wild-type TRK-T3 forms in vivo high molecular weight binding motifs has been previously reported (Mencinger complexes of approximately 400–480 kDa. Considering and Aman, 1999). A computer analysis revealed the that the molecular weight of TRK-T3 monomer is presence of a putative PB1 domain, a novel protein 68 kDa, such complexes would include six oncoprotein module capable of binding to target proteins that molecules. This hexameric organization might result contain PC motifs (Ito et al., 2001). To unveil the from the assembly of two TRK-T3 trimers, as described elements playing a crucial role in TRK-T3 oncogenic for RSV F protein (Matthews et al., 2001). However, the activation, we constructed mutants in which the above complexes might contain other proteins, in addition to mentioned domains, presumably involved in protein TRK-T3. We excluded the presence of adaptor proteins interaction, were ablated. We constructed three mutants for the following reasons: (i) the same chromatography carrying, respectively, the mutation of the Lys residue of pattern of T3/WT was obtained for a TRK-T3 kinase- the PB1 domain shown to be important for the binding dead mutant or T3/WT treated with the specific of Bem1p to the PC domain of Cdc24p; the mutation of inhibitor K252a (data not shown); (ii) Western blot the Tyr residue within the SH2-binding domain; and the analysis failed to detect Shc and PLCg in the chromato- deletion of the SH3-binding domain. Biological analysis graphy’s fractions containing TRK-T3, whereas Shc was indicated that the PB1 and SH2-binding domains give detected in low molecular weight fractions (data not an important contribution to TRK-T3 biological shown). The complexes might also include the wild-type activity. PB1 domain seems to be indispensable, since TFG protein; however, given the high expression level of its mutation completely abrogates TRK-T3 transform- TRK-T3, the wild-type TFG contribution to the ing activity. This suggests that interaction with protein complex formation might be irrelevant. recruited by the PB1 domain plays a crucial role. On the The chromatography patterns of the mutated proteins other hand, the mutation of the SH2-binding motif showed that T3/L1 and T3/L4 proteins form complexes significantly reduced the oncogenic activity, indicating similar to T3/WT, with a prevalent distribution in an important role for such domain. Interestingly, the fractions of high molecular weight; T3/L3, with a SH2-binding motif is specific for the tyrosine phospha- molecular weight similar to the wild type, shows a large tase SHP-1, proposed to play a regulatory role in many distribution ranging from 70 to 480 kDa and corre- cellular signaling processes (Fujioka et al., 1996; sponding to different status of aggregation. This Mencinger and Aman, 1999). With respect to the SH3- indicates that the deleted sequences in this mutant binding domain, the slight reduction of transforming interfere with the formation and/or stability of func- activity observed in this study does not allow speculat- tional TRK-T3 complexes. The T3/L2 protein is ing about its precise role in TRK-T3 activation; this recovered in the same fractions as the wild type, topic will require more investigations.

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 816 In this paper, we have shown that the TFG coiled-coil introduced into pIRESneo expression vector linearized with domain is necessary, but not sufficient for the activation NotI and BamHI/blunt. The resulting construct encodes the of the thyroid TRK-T3 oncogene. Using both deletion T3/L3 protein in which the TFG initiation codon is followed and site-specific mutants we have demonstrated that the by Asn 28. The T3/L4 mutant was constructed by PCR amplification regions outside the coiled-coil domain play a crucial role 0 in TRK-T3 transforming activity. In fact, such se- followed by DNA ligation. The 5 moiety was amplified using oligonucleotides AG3055 (50-ATC CTG GAG TCC ACC quences regulate protein processing, complex formation ATG-30) and AG3047 (50-TAA TAG GAT CCA ATA TTG and possible interaction with other proteins. This latter GTG GAA GGT-30). The 30 moiety was amplified using possibility is strongly supported by the constructs oligonucleotides AG3048 (50-ATT ATG GAT CCC CGG carrying mutations of the PB1 and SH2-binding TGG-30) and AG2964 (50-TTT TTT TTC AAG GGA TAA domains, in which a single amino acid change strongly TAA ATA TAA TTG CT-30). The amplified fragments of 423 affected the transforming activity. The search for and 1388 bp, were cloned into pGEMT vector to produce the proteins interacting with PB1 and SH2 domains and pGL4/50 and pGL4/30 plasmids, respectively, and subjected to the analysis of such interactions, as well as the analysis nucleotide sequence. The 50 fragment was excised from plasmid pGL4/50 by BamHI and SalI/blunt digestion and was inserted of other consensus sites will help in elucidating the 0 mechanisms by which the TFG sequences regulate into the pGL4/3 plasmid after linearizing with BamHI and SphI/blunt. The resulting plasmid pGL4 carried the deletion of TRK-T3 activation. nucleotides 424–621. After nucleotide sequence the deleted TRK-T3 cDNA was excised from plasmid pGL4 by ApaI/NotI digestion, blunted, and inserted into the pIRESneo vector linearized with EcoRV, to produce T3/L4. Materials and methods All the site-specific mutants were constructed by the Gene Editort in Vitro Site-directed Mutagenesis System (Promega), Plasmid construction according to the manufacturer’s instructions, using as template The TRK-T3 cDNA inserted into the pRC/CMV expression the T3/WT plasmid. T3/K14A, carrying the mutation of the vector previously reported (Greco et al., 1995) was excised by Lysine 14 to Alanine, was constructed using the oligonucleo- HindIII/XbaI digestion, blunted, and inserted into the pIR- tide 50-AAGCTAATCGTCGCAGCTCAACTTGGG-30 (the ESneo vector linearized with NotI and blunted, to produce T3/ mutated nucleotides are indicated in boldface type). T3/Y33F, WT. The T3/L1 mutant was constructed by PCR amplification carrying the mutation of the Tyrosine 33 to Phenylalanine, was (Greco et al., 1998). The oligonucleotides utilized were constructed using the oligonucleotide 50-GATATTACTTTT- AG3046 (50-ATC CTG GAG TCC ACC ATG AAT GGC GATGAATTAG-30 (the mutated nucleotides is indicated in CAG CCA AGA CC-30) and AG2964 (50-TTT TTT TTC boldface type). T3/nPro, carrying the deletion of the SH3- AAG GGA TAA TAA ATA TAA TTG CT-30). The amplified binding motif (nucleotides 391–411), was constructed using the 1740 bp fragment was inserted into pGEMT plasmid vector, oligonucleotide 50-TTGGATAGCTTGGAATCCACCAA- subjected to nucleotide sequence, excised by NotI and ApaI/ TATTCC-30. Clones carrying the K14A or Y33F mutations blunt digestion and introduced into pIRESneo expression were identified by sequence analysis. In the case of T3/DPro vector linearized with NotI and BamHI/blunt. The resulting mutant, a fragment spanning the mutation was amplified by construct encodes the T3/L1 protein in which the TFG PCR. Mutant fragments were identified by faster electro- initiation codon is followed by Asn 91. phoretic mobility with respect to wild type, owing to the 21 The T3/L2 mutant was constructed by PCR amplification bases deletion. All of the mutant clones were subjected to followed by DNA ligation. The 50 moiety was amplified using nucleotide sequence to exclude possible additional mutations oligonucleotides AG3046 (50-ATC CTG GAG TCC ACC accidentally occurring during the mutagenesis reaction. ATG AAT GGC CAG CCA AGA CC-30) and AG3047 (50- The TRK-T3 kinase dead mutant, used as control, carries TAA TAG GAT CCA ATA TTG GTG GAA GGT-30). The the mutation of the ATP binding site Lysine 339 to Alanine 30 moiety was amplified using oligonucleotides AG3048 (50- (Greco et al., 1998). ATT ATG GAT CCC CGG TGG-30) and AG2964 (50-TTT The VGF8-luc is a reporter plasmid made of vgf promoter TTT TTC AAG GGA TAA TAA ATA TAA TTG CT-30). sequence fused to the luciferase reporter gene (a kind gift from The amplified fragments of 153 and 1389 bp, respectively, were Dr R Possenti). cloned into pGEMT vector to produce the pGL2/50 and The pRL-CMV vector, containing the coding region of the pGL2/30 plasmids, respectively, and subjected to nucleotide Renilla luciferase (RL) gene, was provided by Promega. sequence. The 50 fragment was excised from plasmid pGL2/50 by BamHI/NotI digestion and was inserted into the pGL2/30 Cell culture and transfection plasmid linearized with the same restriction enzymes. The resulting plasmid pGL2 carried the deletion of all TFG Mouse NIH3T3 fibroblasts were cultured in Dulbecco’s sequence except the coiled-coil domain. After nucleotide modified Eagle’s medium (DMEM) supplemented with 10% sequence, the deleted TRK-T3 cDNA was excised from calf serum; human kidney 293T cells were cultured in DMEM plasmid pGL2 by NotI digestion and inserted into the supplemented with 10% fetal calf serum. PC12 cells were pIRESneo vector linearized with the same restriction enzymes, grown in RPMI-1640 medium supplemented with 5% fetal calf to produce T3/L2. serum and 10% horse serum. The T3/L3 mutant was constructed by PCR amplification. The NIH3T3 cells (8 Â 104/60 mm plate) were transfected by 0 The oligonucleotides utilized were AG3514 (5 -ATC CTG the CaPO4 method using 250 ng of expression plasmid DNA GAG TCC ACC ATG AAT GAA GAT ATT ACT-30) and together with 15 mg of mouse carrier DNA. Transfected cells AG2964 (50-TTT TTT TTC AAG GGA TAA TAA ATA were selected in DMEM supplemented with 10% serum in the TAA TTG CT-30). The amplified 1929 bp fragment was presence of G418 antibiotic (400 mg/ml), to determine the inserted into pGEMT plasmid vector, subjected to nucleotide transfection efficiency, and in medium containing 5% serum to sequence, excised by NotI and ApaI/blunt digestion, and determine the transforming activity. In some experiments, a

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 817 G418 preselection was applied. Transfected NIH3T3 cells were myelin basic protein (MBP). After washing with RIPA buffer, firstly selected for 2 weeks in DMEM supplemented with 10% proteins were eluted and subjected to SDS–PAGE, and the serum in the presence of G418; subconfluent cell cultures were gel was dried. 32P-labeled proteins were revealed and quantified then shifted to a medium containing 5% serum for additional 2 by using PhosphorImager apparatus (Typhoon 8600, Amer- weeks. At the end of selection, G418-resistant colonies and sham). A second set of immunoprecipitates was used for transformed foci were either fixed or isolated for further protein-level determination by Western blotting with anti- studies. TRK antibodies. Immunoreactive bands were detected by I25 The 293T cells were transiently transfected using the CaPO4 using I-labeled protein A (Amersham) and visualized/ method with 5 mg of plasmid DNA. One day after transfection, quantified by using PhosphorImager apparatus (Typhoon cells were kept in 10% serum medium for 6–7 h, serum-starved 8600, Amersham). 20 h in DMEM containing 0.5% FCS, and then processed for protein extraction. Immunofluorescence assay PC12 cells were transfected using Cellfectin (Life Technol- ogies, Inc.). Cells (2 Â 105) were seeded on collagen-coated 12- Transiently transfected NIH3T3 cells were plated on glass multiwell plates and transfected with 50 ng of specific plasmid coverslips, were fixed with 3% paraformaldehyde plus 2% DNA together with 150 ng of VGF8-luc plasmid, 50 ng of sucrose, and stained with anti-TRK antibodies (Santa Cruz pRL-CMV plasmid (Promega) and 350 ng of plasmid carrier Biotech.) and Alexa-546-conjugated secondary antibody DNA. Cells were incubated with the reagent for 7 h and scored (1 : 500; Alexa Fluor, Molecular Probes). Stained cells were for neurites outgrowth 2 days later. observed with confocal microscope BioRad Radiance 2100 equipped with HeNe laser 546 nm and mounted on a Nikon Eclipse 300 fitted with a Plan Apo 60 Â /1.4 lens. The images Luciferase activity assay were digitalized at a resolution of 8 bits onto an array of Both luciferase activities of VGF8-luc and pRL-CMV genes in 512 Â 512 pixel. PC12 cells lysates (5 mg) were measured using the Dual- luciferase reporter assay system (Promega), according to the In vitro translation manufacturer’s recommendations. Light emission was measured using a TD-20/20 Luminometer. The TRK-T3 wild-type and mutated cDNAs cloned into pGEMT vectors were subjected to in vitro transcription– translation using the TNT-coupled reticulocyte lysate system Immunoprecipitation, pull-down and Western Blot analyses (Promega), according to the manufacturer’s specification. The 35 Cells were lysed with PLCLB buffer (50 mm HEPES, 150 mm proteins were translated in the presence of S-labeled NaCl, 10% glycerol, 1% Triton X-100, 1.5 mm MgCl ,1mm methionine; reactions were performed in a volume of 25 ml, 2 using 1 mg of plasmid DNA as template. EGTA, 10 mm Na4P2O7, 100 mm NaF) supplemented with aprotinin, pepstatin, leupeptin, PMSF, and Na3VO4. Insoluble material was recovered by centrifugation at 13 000 g for 15 min In vitro binding assay using GST/T3 fusion protein m and solubilized in 50 ml10m Tris-HCl, 1% SDS for 10 min at Expression and purification of GST/T3 fusion protein were room temperature. After addition of 200 ml PLCLB, samples performed as described elsewhere (Greco et al., 1998). In vitro were sonicated for 20 s with a tip sonicator. translated TRK-T3 proteins were incubated with the glu- One milligram of the cell extracts was precipitated with the tathione-sepharose-conjugated GST/T3 fusion protein at 41C appropriated antibodies, or with GST-FRS2(PTB) fusion for 2 h. The bound proteins were eluted from the complexes by proteins conjugated to glutathione-sepharose. The precipitates boiling in Laemmli buffer and resolved on 8.5% polyacryla- were washed three times with HNTG buffer (20 mm HEPES, m mide gel. After fixing in 30% methanol, 7% acetic acid, the gel 150 m NaCl, 0.1% Triton X-100, 10% glycerol) and boiled in was treated with 1 m salicylic acid, sodium salt for 30 min and Laemmli sample buffer. Protein samples were electrophoresed dried. Autoradiography was performed after overnight ex- on an 8.5% SDS–PAGE, transferred onto nitrocellulose filters posure to hyperfilm. and immunoblotted with the appropriated antibodies. The immunoreactive bands were visualized using horseradish peroxidase-conjugated secondary antibody and enhanced Size-exclusion chromatography chemiluminescence (Amersham). The anti-TRK and anti- Nondenatured cell extracts from 293T cells transiently PLCg antibodies were from Santa Cruz Biotech., Inc.; the transfected with TRK-T3 wild type or mutants were lysed in anti-Shc, anti-phosphotyrosine antibodies were from Upstate 200 ml of NP-40 lysis buffer (1% NP-40, 10% glycerol, 140 mm Biotechnology. Inc. NaCl, 10 mm Tris) supplemented with the protease inhibitors described above. After 0.45 mm filtration the samples were Immunokinase assay applied to a size-exclusion column (Superdex 200 HR10/30; Pharmacia) and eluted at a flow rate of 0.5 ml/min. Fractions 293T cells transiently expressing wild-type and deleted TRK- were collected at 1 min intervals and analysed by Western blot T3 proteins were lysed in radioimmune precipitation buffer with anti-TRK antibodies. (RIPA) (20 mm Tris-HCl, pH 7.5, 150 mm NaCl, 5 mm EDTA, 1% NP-40) supplemented with protease inhibitors. TRK-T3 In vivo studies proteins were immunoprecipitated with anti-TRK antibodies, adsorbed on protein A-Sepharose beads, and washed twice Animal studies were reviewed and approved by the Ethics with RIPA buffer. One set of immunoprecipitates was washed Committee for Animal Experimentation of the Istituto once with incubation buffer (50 mm HEPES, 20 mm MnCl2, Nazionale Tumori (Milan, Italy) and are in accordance with 5mM MgCl2,1mm dithiothreitol), afterwards the samples the guidelines of the UK Coordinating Committee for Cancer were incubated for 10 min at 301Cin50ml of the same buffer Research. Female CD-1 nu/nu mice (8 to 9 weeks old) (Charles containing 10 mm ATP and [g-32P]ATP (5000 Ci/mmol; Amer- River, Calco, Italy) were injected s.c. with NIH3T3-transfected sham Biosciences) in the presence of exogenous substrate cells; for each animal 104 or 103 cells in 0.2 ml PBS were used.

Oncogene Mechanism of activation of TRK-T3 oncogene E Roccato et al 818 Three mice per sample were used for each experiment. Tumor Acknowledgements growth of injected animals was assesed by evaluating tumor We thank Miss Cristina Mazzadi for secretarial assistance. We latency, that is, days to reach 0.1 g, and by monitoring tumor thank Prof. Marchisio for advises on immunoﬂuorescence weight (TW) twice a week. Tumor weight in grams was procedure, Elena Luison for chromatography experiments and estimated by the formula TW ¼ d 2 Â D/2, where d and D are, Dr Cristiano Rumio for immunoﬂuorescence analysis by respectively, the shortest and longest diameters of the tumor, confocal microscopy. This work was supported by AIRC measured in centimeters. (Italian Association for Cancer Research).

References

Borrello MG, Pelicci G, Arighi E, De Filippis L, Greco A, Martin-Zanca D, Hughes SH and Barbacid M. (1986). Nature, Bongarzone I, Rizzetti MG, Pelicci PG and Pierotti MA. 319,743–748. (1994). Oncogene, 9, 1661–1668. Matthews JM, Young TF, Tucker SP and Mackay JP. (2001). Byrd DA, Sweet DJ, Pante’ N, Konstantinov KN, Guan T, J. Virol., 74, 5911–5920. Shapire ACS, Mitchell PJ, Cooper CS, Aebi U and Gerace McWhirter JR, Galasso DL and Wang JY. (1993). Mol. Cell. L. (1994). J. Cell Biol., 127, 1515–1526. Biol., 13, 7587–7595. Canu N, Possenti R, Rinaldi AM, Trani E and Levi A. (1997). Mencinger M and Aman P. (1999). Biochem. Biophys. Res. J. Neurochem., 68, 1390–1399. Commun., 257, 67–73. Durick K, Yao VJ, Borrello MG, Bongarzone I, Pierotti MA Mencinger M, Panagopoulos I, Andreasson P, Lassen C, and Taylor SS. (1995). J. Biol. Chem., 270, 24642–24645. Mitelman F and Aman P. (1997). Genomics, 41, Fujioka Y, Matozaki T, Noguchi T, Iwamatsu A, Yamao T, 372–331. Takahashi N, Tsuda M, Takada T and Kasuga M. (1996). Miranda C, Greco A, Miele C, Pierotti MA and Van Mol. Cell Biol., 16, 6887–6899. Obberghen E. (2001). J. Cell. Physiol, 186, 35–46. Greco A, Fusetti L, Miranda C, Villa R, Zanotti S, Pagliardini Monaco C, Visconti R, Barone MV, Pierantoni GM, S and Pierotti MA. (1998). Oncogene, 16, 809–816. Berlingieri MT, De Lorenzo C, Mineo A, Vecchio G, Fusco Greco A, Mariani C, Miranda C, Lupas A, Pagliardini S, A and Santoro M. (2001). Oncogene, 20, 599–608. Pomati M and Pierotti MA. (1995). Mol. Cell. Biol., 15, Pierotti MA, Bongarzone I, Borrello MG, Greco A, Pilotti S 6118–6127. and Sozzi G. (1996). Genes Chromosones Cancer, 16, Grignani F, Gelmetti V, Fanelli M, Rogaia D, De Matteis S, 1–14. Ferrara FF, Bonci D, Nervi C and Pelicci PG. (1999). Possenti R, Di Rocco G, Nasi S and Levi A. (1992). Proc. Oncogene, 18, 6313–6321. Natl. Acad. Sci. USA, 89, 3815–3819. Ito T, Matsui Y, Ago T, Ota K and Sumimoto H. (2001). Roccato E, Miranda C, Ranzi V, Gishizky ML, Pierotti MA EMBOJ., 20, 3938–3946. and Greco A. (2002). Br. J. Cancer, 87, 645–653. Johnston JA, Ward CL and Kopito RR. (1998). J. Cell. Biol., Rodrigues GA and Park M. (1993). Mol. Cell. Biol., 13, 143, 1883–1898. 6711–6722. Kaplan DR, Martin-Zanca D and Parada LF. (1991). Nature, Rodrigues GA and Park M. (1994). Curr. Opin. Genet. 350, 158–160. Develop., 13, 6711–6722. Klein R, Jing S, Nanduri V, O’Rourke E and Barbacid M. Terasawa H, Noda Y, Ito T, Hatanaka H, Ichikawa S, Ogura (1991). Cell, 65, 189–197. K, Sumimoto H and Inagaki F. (2001). Embo J., 20, Lupas A. (1996). Trends Biochem. Sci., 21, 375–382. 3947–3956. Lupas A, Van Dyke M and Stock J. (1991). Science, 252, Tong Q, Xing S and Jhiang SM. (1997). J. Biol. Chem., 272, 1162–1164. 9043–9047.

Oncogene