Transforming Properties of the Huntingtin Interacting Protein 1/ Platelet-derived Growth Factor β Receptor Fusion Protein*

We have previously reported that the Huntingtin interacting protein 1 (HIP1) gene is fused to the platelet-derived growth factor β receptor (PDGFβR) gene in a patient with chronic myelomonocytic leukemia. We now show that HIP1/PDGFβR oligomerizes, is constitutively tyrosine-phosphorylated, and transforms the murine hematopoietic cell line, Ba/F3, to interleukin-3-independent growth. A kinase-inactive mutant is neither tyrosine-phosphorylated nor able to transform Ba/F3 cells. Oligomerization and kinase activation required the 55-amino acid carboxyl-terminal TALIN homology region but not the leucine zipper domain. Tyrosine phosphorylation of a 130-kDa protein and STAT5 correlates with transformation in cells expressing HIP1/PDGFβR and related mutants. A deletion mutant fusion protein that contains only the TALIN homology region of HIP1 fused to PDGFβR is incapable of transforming Ba/F3 cells and does not tyrosine-phosphorylate p130 or STAT5, although it is itself constitutively tyrosine-phosphorylated. We have also analyzed cells expressing Tyr → Phe mutants of HIP1/PDGFβR in the known PDGFβR SH2 docking sites and report that none of these sites are necessary for STAT5 activation, p130 phosphorylation, or Ba/F3 transformation. The correlation of factor-independent growth of hematopoietic cells with p130 and STAT5 phosphorylation/activation in both the HIP1/PDGFβR Tyr → Phe and deletion mutational variants suggests that both STAT5 and p130 are important for transformation mediated by HIP1/PDGFβR.

normality associated with CMML and results in expression of a fusion protein, TEL/PDGF␤R, containing the amino-terminal portion of TEL (ETV6) and the transmembrane and tyrosine kinase domains of platelet-derived growth factor ␤ receptor (PDGF␤R) (1).
The TEL/PDGF␤R fusion has constitutive kinase activity mediated by the TEL pointed (PNT) self-association motif, transforms the murine hematopoietic cell line Ba/F3 to IL-3independent growth and induces hematopoietic malignancies in murine models of leukemia (2)(3)(4)(5). The kinase activity, as well as the PNT oligomerization motif, is necessary for transformation. These structure-function relationships are similar to those observed in other transforming tyrosine kinase fusions such as TPR/MET (6), BCR/ABL (7), TEL/ABL (2), and TEL/ JAK2 (8,9). Each of these require an oligomerization domain to constitutively activate the partner tyrosine kinase and hence transform cells.
We have identified another cytogenetic abnormality associated with the clinical phenotype of CMML, t(5;7)(q33;q11.2), that results in expression of the transforming fusion protein HIP1/PDGF␤R (10). Huntingtin interacting protein 1 (HIP1) is 116-kDa protein that binds Huntingtin, the protein product of the gene mutated in Huntington disease (11,12). All but the 18 carboxyl-terminal amino acids of HIP1 are fused to the transmembrane and tyrosine kinase domains of the PDGF␤R in this novel fusion protein.
The function of HIP1 is unknown, but it contains evolutionarily conserved sequences. The yeast homologue, SLA2P, is essential for yeast growth as well as assembly and function of the cortical cytoskeleton (13). There is also a C. elegans homologue of unknown function. 2 Finally, HIP1 has a leucine zipper motif and the carboxyl terminus is homologous to TALIN, a cytoskeletal protein implicated in cell-substratum as well as cell-cell interactions (15).
Several signaling pathways are employed by the activated native PDGF␤R kinase. The signaling intermediates, including SRC family members, phosphatidylinositol 3-kinase (PtdIns 3-kinase), RasGTPase-activating protein, and phospholipase C␥ (PLC␥), may also be activated by HIP1/PDGF␤R. Certain tyrosine residues when phosphorylated are designated SH2 docking sites, since they are binding sites for these SH2 domain-containing signaling proteins. Two tyrosine phosphorylation sites (Tyr 579 and Tyr 581 ) ( Fig. 1) associate with the SRC family tyrosine kinases (SRC, Fyn, and Yes) (16,17), and this association leads to mitogenesis in NIH 3T3 cells. The 85-kDa regulatory subunit of PtdIns 3-kinase binds to phosphorylated Tyr 740 and Tyr 751 in the kinase insert domain of the PDGF␤R, and PLC␥ binds to phosphorlyated Tyr 1009 and Tyr 1021 (18). Both are required for full mitogenic activity in HepG2 cells (19).
Here we have dissected the role of HIP1 in HIP1/PDGF␤R transformation by deletion mutational analysis of the HIP1 moiety. We have also determined the relevance of the mitogenic pathways known to be activated by native PDGF␤R by testing the effect of key tyrosine to phenylalanine substitutions in the PDGF␤R in the context of the HIP1/PDGF␤R fusion.

EXPERIMENTAL PROCEDURES
Reconstruction of the Fusion cDNA for Expression Experiments-The breakpoint was amplified from patient material using primers HIP1301F (5Ј-CCTGAAACTGCTAAGAACCA-3Ј) and PDGF␤R 1806R, and the product was digested with BglI and NheI. The BglI-SacII fragment of the PDGF␤R was isolated after BglI and SacII digestion of the PDGF␤R cDNA and ligated to the NheI-BglI breakpoint fragment. This ligation reaction was amplified with primers containing the NheI and SacII sites (5Ј-AAATTGCTGCTAGCACAGCCCAGCTTG-3Ј and 5Ј-CTGGTCCCGCGGCAGCTCCCACGTGGA-3Ј, respectively), digested with SacII, and ligated with the 3Ј-end of PDGF␤R or the kinaseinactive point mutant (PDGF␤R (R634K) (18)). This ligation reaction was then digested with NheI and ligated with the 5Ј-end of HIP1 (the HIP1 cDNA was the kind gift of M. Hayden, University of British Columbia, Vancouver) via the unique NheI site. The region amplified by PCR was confirmed to be void of PCR-generated mutations by sequence analysis. The resultant 4.6-kilobase inserts (kinase-active and -inactive) were subcloned into the appropriate vectors for use in the expression experiments.
Construction of Mutants of HIP/PDGF␤R-The reconstructed cDNAs for HIP1/PDGF␤R and H/P(KI) were subcloned into pcDNA3 (Invitrogen). H/P(del)3 was generated by digestion of pcDNA3-HIP1/ PDGF␤R with PflMI, blunting with T4 DNA polymerase, digestion with EcoRI, and isolation of the insert. The insert was ligated with EcoRI/ EcoRV-cut pcDNA3.
The 23-amino acid leucine zipper was deleted by amplifying nucleotides 1534 -2641 of HIP1 with an XbaI site engineered onto the 5Ј-end of the amplicon. The resulting PCR product was then digested with XbaI and NheI and ligated into XbaI/NheI-digested HIP1/PDGF␤R in the vector pBluescript KS(ϩ). This construct was then digested with HpaI and NheI, and the insert was ligated into HpaI/NheI-digested pcDNA3-HIP1/PDGF␤R.
In addition, HIP1 amino-terminal deletion mutants were constructed. The H/P(RI) deletion mutant was prepared by digestion of the wild type HIP1/PDGF␤R construct in pcDNA3 with EcoRI followed by religation. H/P(NI) was made by digesting HIP1/PDGF␤R in pcDNA3 with EcoRI and NheI, filling in the sticky ends with T4 DNA polymerase followed by religation.
The resultant PCR products were digested with EcoRI and SacII and ligated with EcoRI/SacII-digested H/P(RI) pcDNA3 vector.
To construct the F8 mutant of HIP1/PDGF␤R, the wild type PDGF␤R F8 mutant was used and has been described previously (19). An F6 mutant bearing mutations in Tyr 716 , Tyr 740 , Tyr 751 , Tyr 771 , Tyr 1009 , and Tyr 1021 was constructed by digesting the mutated F8 PDGF␤R in pcDNA3 (kindly provided by Andrius Kazlauskas) with SacII and EcoRI and replacing the released fragment with the HIP1/PDGF␤R EcoRI/SacII 2-kilobase fragment (Fig. 1). To make the most membraneproximate mutations that would convert this F6 mutant to HIP1/ PDGF␤R(F8), an antisense oligonucleotide was designed to encode Tyr 3 Phe mutants at positions Tyr 579 and Tyr 581 of the PDGF␤R. This oligonucleotide also contained the unique SacII site in the PDGF␤R. This 3Ј-primer was then used with a sense oligonucleotide just 5Ј of the unique NheI site of the HIP1 cDNA (nucleotides 2264 -2284 of the HIP1 sequence) to amplify a PCR product using the "wild type" fusion cDNA as a template. The product was digested with SacII and NheI and subcloned into a NheI/SacII-digested HIP1/PDGF␤R-bearing pcDNA3. For the immunoprecipitations, one-half of the reaction mixture was incubated with antiserum to the carboxyl terminus of the human PDGF␤R (UBI), and immune complexes were collected with protein G-Sepharose. After fixing the gels containing the [ 35 S]methionine-labeled and SDS-PAGE-separated proteins, the gels were treated with Amplify (Amersham Pharmacia Biotech), dried, and exposed to film for 2 h. Predicted electrophoretic mobilities for H/P and related mutants were based on analysis of electrophoretic mobilities of native PDGF␤R and TEL/PDGF␤R, which migrate with apparent molecular masses of approximately 20 -30 kDa larger than predicted by primary sequence.
The resultant mutant (H/P(F2)) was then digested with SacII and RI, and the fragment that was released was used to replace the wild type sequence in the H/P(F6)MSCVneo mutant as well as to replace the wild type sequence in the H/P.MSCVneo construct. Briefly, the nonmutated RI/SacII fragment was excised, and the vectors were ligated with the RI/SacII Tyr 579 -and Tyr 581 -mutated fragment. The amplified fragment was sequenced.
In Vitro Transcription/Translation-In vitro transcription/translation was performed using a rabbit reticulocyte lysate kit (TNT system; Promega) according to the manufacturer's specifications. Proteins were labeled with [ 35 S]methionine. One-half of the reaction was removed, and 2 l of IgG specific for the C terminus of PDGF␤R was used to immunoprecipitate translated products. The immunocomplexes were precipitated with protein G-Sepharose and washed three times with 1 ml of lysis buffer (150 mM NaCl, 20 mM Tris⅐HCl, pH 7.4, 1% Triton X-100). Total reaction mixtures and immunoprecipitates were resolved by SDS-PAGE, and the separated proteins were visualized by fluorography using Amplify (Amersham Pharmacia Biotech) according to the manufacturer's instructions.
Immunofluorescence-COS-1 cells were transfected with HIP1/ PDGF␤R using the DEAE-dextran method (20). The transfected COS-1 cells were subjected to indirect immunofluorescence antibody staining 48 h after transfection. At 24 h, cells were replated at 400,000 cells/ml onto glass coverslips. At 48 h, cells were fixed with 4% formaldehyde in phosphate-buffered saline and blocked in 5% milk in phosphate-buffered saline/1% Triton X-100 for 1 h, and then a 1:100 dilution of antiserum to the cytoplasmic tail of the PDGF␤R (Pharmingen) was added for 1 h. After washing the cells, secondary fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Vector Laboratories, Burlingame, CA) was added for 30 min. The cells were then washed, incubated with 1 g/ml Hoechst dye 33258 for 1 min, and washed again, and then the coverslips were mounted on glass slides. The cells were visualized with an Olympus fluorescence microscope and photographed with Kodak Royal Gold II 400 print film.
The extracts were either subjected to immunoprecipitation with antiserum to the cytoplasmic tail of the PDGF␤R as described under the in vitro transcription/translation section and/or directly separated with 7% SDS-PAGE, transferred to nitrocellulose, and detected with the indicated antibodies as described (4). When analyzing associated proteins, the immunoprecipitates were washed with Tris-buffered saline containing 0.5% Triton X-100 rather than the lysis buffer (150 mM NaCl, 20 mM Tris⅐HCl, pH 7.4, 1% Triton X-100).
When the Western blot was probed twice with anti-phosphotyrosine antibodies (4G10) followed by anti-PDGF␤R tail antibodies, the blot was stripped of the 4G10 antibody by incubating it at 50°C for 30 min  The blot was then washed with phosphate-buffered saline four times for 10 min prior to staining with the second antibody. Electromobility Shift Assays (EMSAs)-EMSAs were performed as described previously (22). In brief, nuclear extracts (6 mg) from control cells or cells expressing H/P and the various mutant fusion proteins were incubated (20 min) with a 32 P-labeled and purified doublestranded oligonucleotide probe recognized by activated signal transducer and activator of transcription (STAT) protein complexes (␥-interferon-activated sequence based on the FcgRI gene promoter). For supershift analysis, anti-STAT antibodies (purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA); catalogue numbers as follows: STAT1, SC-346X; STAT3, SC-482X; STAT5b, SC-835X) were added. The DNA-protein complexes were resolved on 4% TRIS, acetate, EDTApolyacrylamide gels and detected by autoradiography.

Oligomerization of HIP1/PDGF␤R and Related Mutants-
Full-length HIP1/PDGF␤R (10) was used to construct related mutants including H/P(KI), which contains a point mutation, R634K, known to inactivate PDGF␤R kinase activity; H/P(RI)(del)LZ, which contains a 23-amino acid deletion of the leucine zipper motif as well as deletion of amino-terminal sequences up to the EcoRI site; a series of HIP1 amino-terminal deletion mutants; and the Tyr 3 Phe mutations in the PDGF␤R portion of the fusion (Fig. 1). The N-terminal deletion mutants were constructed by using the unique restriction enzyme sites EcoRI (H/P(RI)) and NheI (H/P(NI)) to excise HIP1 5Ј sequences or amplifying HIP1 and PDGF␤R sequences of the fusion protein by PCR. All mutants were confirmed by sequence analysis.
When selected constructs were translated in vitro, HIP1/ PDGF␤R and related mutants migrated at the expected sizes and were recognized by an antibody to the carboxyl terminus of the PDGF␤R (Fig. 2). Utilization of the first ATG of the fulllength HIP1/PDGF␤R and truncated H/P(RI)(del)LZ yielded protein products with expected electrophoretic mobilities of 220 and 170 kDa, respectively (see legend to Fig. 2). H/P(NI) and H/P(del)H migrated as a 100-kDa doublet and a 90-kDa protein, respectively. The 100-kDa doublet is consistent with the presence of two ATG start sites in HIP1 after the NheI site at methionines 965 and 980 (Fig. 1).
Oligomerization of HIP1/PDGF␤R and related mutants was tested by in vitro translation and co-immunoprecipitation experiments as described previously (4). In brief, an antibody to the carboxyl-terminal epitope of the PDGF␤R was used to immunoprecipitate the protein complexes. The epitope recognized by this antibody is not present in the H/P(del)3 construct. Hence, the antibody can only immunoprecipitate the H/P(del)3 protein if there is association with the fusion proteins that contain this epitope. The H/P, H/P(RI)(del)LZ, H/P(KI), and H/P(NI) constructs all co-immunoprecipitated the H/P(del)3 protein, demonstrating that the leucine zipper motif was not necessary for oligomerization in this assay. In contrast, the deletion mutant without any HIP1 sequences (H/P(del)H) did not interact with the H/P(del)3 proteins (Fig. 2). Taken together, these data demonstrate that the carboxyl-terminal 55 amino acids of HIP1 (TALIN homology region) in the fusion protein are sufficient for oligomerization of HIP1/PDGF␤R.
Expression and Tyrosine Phosphorylation of HIP1/PDGF␤R and Related Mutants-Expression and transforming properties of HIP1/PDGF␤R and selected mutants were tested in the IL-3-dependent, murine hematopoietic cell line, Ba/F3. The HIP1/PDGF␤R constructs were subcloned into the murine ecotropic retroviral vector MSCVneo. Viral supernatants were prepared by transient transfection of 293T cells and used to infect Ba/F3 cells (21).
Expression of the wild type and selected mutant proteins under control of the viral long terminal repeat was confirmed with antibodies directed against the PDGF␤R carboxyl terminus (in cytoplasmic domain) (Fig. 3). HIP1/PDGF␤R and all of the mutants were constitutively tyrosine-phosphorylated in stable Ba/F3 infectants, except for the kinase-inactive point mutant H/P(KI) (Fig. 3). HIP1/PDGF␤R, H/P(RI)(del)LZ and H/P(NI) had predicted electrophoretic mobilities of 220, 170, and 100 kDa, respectively. The phosphorylated band at 180 kDa in the HIP1/PDGF␤R and H/P(RI)(del)LZ lanes is probably the result of an alternative start site at ATG497, the first ATG after the EcoRI site of HIP1, because on longer exposures this band is recognized by the anti-PDGF␤R antibodies (data not shown and Fig. 1).
In contrast, the H/P(NI) mutant, which also oligomerizes and is constitutively tyrosine-phosphorylated, was incapable of conferring IL-3-independent growth. For control experiments, the kinase-inactive mutant H/P(KI) and insert-free MSCVneo did not confer IL-3-independent growth (Table I).
Tyrosine Phosphorylation of Proteins in Transformed Cell Extracts-Western blots of whole cell lysates of the stable cell lines with anti-phosphotyrosine antibodies identified a consistent and prominent 130-kDa phosphoprotein in the H/P-and H/P(RI)(del)LZ-transformed cells but not the vector, H/P(KI), or H/P(NI) nontransformed cell lines (Fig. 3, right panel). There were no other consistent differences in the phosphorylation patterns of transformed cells compared with nontransformed cells other than autophosphorylated HIP1/PDGF␤R and the related mutant fusion proteins.
To further characterize HIP1 sequences required for p130 tyrosine phosphorylation, mutants were constructed with further amino-terminal truncations of H/P between the EcoRI and NheI sites of the HIP1 sequence (Fig. 1). These mutants were then tested for transforming ability (Table II) and ability to associate with and phosphorylate p130 (Fig. 4). Extracts were immunoprecipitated with antiserum against the PDGF␤R tail, followed by staining with anti-phosphotyrosine antibodies. The results showed that the prominent 130-kDa phosphoprotein associated exclusively with the fusion proteins that transform the Ba/F3 cell line (Fig. 4). Of particular note, p130 was not detected and did not associate with the nontransforming H/P variants: H/P(ATG752), H/P(ATG830), or H/P(NI) (Figs. 3 and 4 and data not shown).
Since the only consistent difference between transformed and nontransformed cell extracts was an increase in tyrosine phosphorylation of p130, p130 tyrosine phosphorylation may be relevant to transformation. To determine the identity of p130, proteins involved in PDGF␤R or IL-3 signal transduction that migrated in the 110 -140-kDa molecular mass range were targeted. Extracts were immunoprecipitated with specific antibodies, and the immunoprecipitates and supernatants were blotted first with anti-phosphotyrosine antibodies, stripped, and reprobed with the specific antibodies. p130 was not detected by the specific antibodies that efficiently cleared the extracts of the immunogens. This strategy was used to exclude CBL, RAS-GAP, CAS, JAK1, JAK2, JAK3, TYK2, IL3␤R, PLC␥, PtdIns 3-kinase, and focal adhesion kinase as candidates for p130 (data not shown).

PtdIns 3-Kinase and PLC␥ Phosphorylations Are Not Required for Ba/F3 Cell IL-3-independent Growth-Since
PtdIns 3-kinase and PLC␥ activation are part of the mitogenic signal transduction of the native PDGF␤R, we assessed their role in H/P transformation. To do this, Tyr 3 Phe point mutants in the known SH2 docking sites of the PDGF␤R (Fig. 1) were constructed and then tested in the Ba/F3 cell system. Table II and Fig. 4 show that the H/P(F2)-and H/P(F8)-infected Ba/F3 cells are transformed and have hyperphosphorylated p130. Hence, none of the known SH2 docking sites are required for transformation or p130 phosphorylation.
The F8 mutant remains tyrosine-phosphorylated despite abrogation of the major SH2 docking sites by tyrosine to phenylalanine mutations. These data demonstrate that there are additional sites of tyrosine phosphorylation on either the HIP1 or the PDGF␤R moieties. Phosphorylation of tyrosine residues has also been demonstrated in comparable Tyr 3 Phe mutants in the native PDGF␤R (18). The lack of phosphorylated SH2 interaction sites may lead to loss of association with tyrosine phosphatases, with concomitant increase in tyrosine phosphorylation at the residual sites.
To probe the role of PtdIns 3-kinase and PLC␥ in more detail, tyrosine phosphorylation of the 85-kDa subunit of PtdIns 3-kinase and PLC␥ were measured in the H/P(F8)-transformed cells. It was hypothesized that by analogy to the native PDGF␤R (19), the H/P(F8) mutant would not bind PtdIns 3-kinase or PLC␥. Ba/F3 cell lysates were subjected to immunoprecipitation with anti-PtdIns 3-kinase and anti-PLC␥ antibodies and then immunoblotted with antibody against phosphotyrosine or the immunoprecipitating antibody (Fig. 5, A and B). Cells expressing the wild-type H/P or the H/P(F2) mutant had increased phosphorylation of both proteins that were expressed at similar levels in each of these cell lines. As hypothesized, the H/P(F8) mutant, which has Tyr 3 Phe point mutations in all of the known SH2 docking sites has lost its ability to associate with and phosphorylate PtdIns 3-kinase and phospholipase C␥. This finding together with the fact that the H/P(F8) variant with mutations in all known SH2 docking sites is transforming, suggests that signaling via the inositol phosphate-containing intermediates is unnecessary for Ba/F3 transformation.
Constitutive STAT5 Activation Is Associated with H/P Transformation-Since the mutant fusion protein H/P(F8) transforms Ba/F3 cells but lacks all of the known SH2 binding sites and is incapable of activating PtdIns 3-kinase or PLC␥ (Table II and Figs. 4 and 5), we analyzed other signal transduction pathways that might be activated by the PDGF␤R portion of the fusion protein. In addition to PtdIns 3-kinase and PLC␥, STATs are reported to be activated by native PDGF␤R (23). Hence, STAT1, -3, and -5, which are highly expressed in Ba/F3 cells, were analyzed for activation in transformed cells.
Electromobility shift assay using a ␥-interferon-activated sequence probe that binds to multiple activated STAT proteins demonstrated constitutive STAT5 activation in the transformed cells by both H/P and H/P(F8) (Fig. 6A). A shifted FIG. 4. H/P and various transforming mutants associate with the 130-kDa phosphoprotein. Lysates were immunoprecipitated with anti-PDGF␤R tail antibody (Pharmingen) and washed with Tris-buffered saline containing 0.5% Triton X-100, separated on 7% PAGE, and blotted onto nitrocellulose. Proteins were detected with horseradish peroxidaseconjugated anti-phosphotyrosine 4G10 monoclonal antibody or anti-PDGF␤R antibody as labeled. A, lanes 1-4 contain H/P, H/P(F2), H/P(F8), and H/P(KI), respectively. The first three cell lines were grown the in the absence of IL-3 and in the presence of G418 (lanes 1-3). The last cell line was grown in the presence of IL-3 and G418. B, lanes 1-7 in each blot contain H/P(ATG497), H/P(ATG599), H/P(A-TG690), H/P(ATG752), H/P(ATG830), H/P(NI), and insert-free MSCVneo cells, respectively. The first three cell lines were grown in the absence of IL-3 and presence of G418 (lanes 1-3). The last three cell lines (lanes 4 -6) were grown in the presence of IL-3 and G418.
doublet was detected using nuclear extracts from the various stably infected cell lines. The upper band of the doublet was constitutively supershifted with STAT5 antibody that recognizes both STAT5a and STAT5b. No supershift was detected when STAT1 or STAT3 antibodies were added to the binding reaction (data not shown). STAT5 was constitutively tyrosinephosphorylated in transformed cells, consistent with the electromobility shift assay results (Fig. 6B). These data demonstrate that although the transforming H/P(F8) mutant does not activate PtdIns 3-kinase or PLC␥, it is a potent activator of STAT5.
Finally, immunofluorescent staining with the anti-PDGF␤R antibody of HIP1/PDGF␤R transiently transfected COS-1 cells showed the fusion protein was localized to the cytoplasm (Fig.  7). Similar data were obtained for various HIP1 truncation mutants and in stably transfected Ba/F3 cells (data not shown).
In summary, the H/P protein oligomerizes, has constitutive kinase activity, transforms Ba/F3 cells to IL-3-independent growth, and is localized to the cytoplasm. The properties of oligomerization, constitutive kinase activity, and transformation do not require the leucine zipper motif as a deletion mutant of the leucine zipper retains these properties. In contrast, a mutant containing only the 55 C-terminal amino acids of HIP1 fused to PDGF␤R kinase (H/P(NI)) retains only the oligomerization and constitutive tyrosine kinase activity. When these 55 amino acids are deleted (H/P(del)H) oligomerization is abrogated. Taken together, these data demonstrate that the 55 carboxyl-terminal amino acids of HIP1 are sufficient for oligomerization and autophosphorylation but not for transformation of hematopoietic cells.
In addition, transformation by HIP1/PDGF␤R and related mutants correlates with constitutive STAT5 activation and tyrosine phosphorylation of p130. These three properties require HIP1 amino acids 690 -752 but do not require any of the known SH2 signal transducing molecule docking sites in the PDGF␤R portion of the fusion. This suggests that in addition to tyrosine kinase activation, HIP1 sequences that function in ways other than self-association may be relevant for transformation of Ba/F3 cells to IL-3-independent growth and that both p130 and the STAT5 are important targets of HIP1/PDGF␤R in mediating transformation. DISCUSSION HIP1/PDGF␤R is a novel fusion protein associated with CMML and t(5;7)(q33;q11.2). HIP1/PDGF␤R oligomerizes, is constitutively tyrosine-phosphorylated, and transforms Ba/F3 cells to IL-3-independent growth. The transformed cells have a hyperphosphorylated p130 protein and constitutively activate STAT5. The kinase-inactive mutant H/P(KI) does not transform cells. As expected in H/P(KI) infected cells, neither it, p130, nor STAT5 are constitutively tyrosine-phosphorylated.
The contribution of HIP1 sequences to transformation and

FIG. 5. PtdIns 3-kinase and PLC␥ phosphorylation and association with H/P and Tyr to Phe mutants of H/P in Ba/F3 cells.
A, lysates (500 g) were immunoprecipitated with rabbit anti-PtdIns 3-kinase antibody (UBI) and immunoblotted as noted. B, lysates (500 g) were immunoprecipitated with anti-bovine PLC␥ mixed monoclonal IgG (UBI) and immunoblotted as noted. The immunoprecipitates were washed three times with Tris-buffered saline containing 0.5% Triton X-100, separated on 7% PAGE, and blotted onto nitrocellulose. Proteins were detected by Western blot first with anti-phosphphotyrosine 4G10 antibodies and then stripped and reprobed with the specific antibodies to control for relative expression and efficiency of immunoprecipitation. kinase activation of HIP1/PDGF␤R is more complex. To assess the role of HIP1 in these functions, deletion mutants of HIP1 were prepared, including H/P(RI), H/P(RI)(del)LZ, H/P(NI), stepwise deletions between the EcoRI and NheI sites of the H/P fusion cDNA, and H/P(del)H. It was expected that the leucine zipper motif would be required for oligomerization and kinase activation by analogy with other transforming fusion proteins (2, 4, 6 -9). In each of these examples, kinase activation is mediated by an oligomerization motif contributed by the nontyrosine kinase fusion partner and is requisite for kinase activation and transformation. In the case of TPR/MET, the oligomerization motif is, in fact, a leucine zipper domain (6). Hence, it was surprising that deletion of the leucine zipper motif in HIP1/PDGF␤R had no effect on oligomerization, kinase activation or transformation.
The H/P(RI), H/P(RI)(del)LZ, H/P(NI), and H/P(del)H deletion mutants were then constructed to identify HIP1 sequences sufficient for oligomerization, tyrosine kinase activation, and transformation. The H/P(NI) mutant contains only the 55 car-boxyl-terminal amino acids of HIP1, including the TALIN homology region. H/P(NI) oligomerized and had tyrosine kinase activity, whereas the H/P(del)H mutant, which lacks any HIP1 sequences, did not oligomerize. These findings demonstrate that the carboxyl-terminal 55-amino acid domain containing the TALIN homology region are sufficient for oligomerization and tyrosine kinase activation. Self-association is a function that, to our knowledge, has not previously been attributed to TALIN.
Surprisingly, we found that although H/P(NI) oligomerizes and has tyrosine kinase activity, it is incapable of transforming cells. In light of this, we made more refined HIP1 aminoterminal deletions between the EcoRI and the NheI sites. These data suggest a critical role in transformation and p130 tyrosine phosphorylation for HIP1 amino acids 690 -752. Unfortunately, there is no amino acid sequence homology to known functional motifs in this amino acid 690 -752 polypeptide to suggest a biological role.
There are several potential explanations for the lack of transformation but retention of constitutive tyrosine kinase activation by the H/P mutants that do not contain aminoterminal amino acids up to amino acid 752. One possible explanation for the lack of transformation by these H/P mutants is that there are subtle differences in tyrosine kinase specific activity that cannot be appreciated by assessment of autophosphorylation with anti-phosphotyrosine antibodies.
Another possibility is that there is an altered subcellular localization of the nontransforming HIP1 deletion mutants compared with HIP1/PDGF␤R, leading to sequestration of the constitutive tyrosine kinase from substrates whose phosphorylation is necessary for transformation. HIP1/PDGF␤R is predominantly localized to the cytoplasm of transiently transfected COS-1 cells (Fig. 7) and Ba/F3 cells (data not shown). Because Ba/F3 cells have scant cytoplasm, it is not possible to exclude association with the plasma membrane or cytoskeletal elements. Further analysis using confocal microscopy of hematopoietic cells stably transfected with HIP1/PDGF␤R, the nontransforming mutants, and mutants containing myristoylation signals or nuclear localization signals may definitively characterize the role of subcellular localization in transformation of relevant cell types by the fusion protein.
FIG. 6. STAT5 is activated in H/P-transformed cells. A, electromobility shift assays were performed on nuclear extracts of Ba/F3 cells infected with a variety of constructs that were either growing constitutively in the absence of IL-3 (lanes 1-3, 5, 7, 8 -10, 12, and 14) or in the presence of IL-3 (lanes 4, 6, 11, and 13). The MSCVneo and the H/P(NI) cell lines were "starved" of IL-3 for 6 h. Extracts were incubated with radiolabeled ␥-interferon-activated sequence probe and separated on 6% PAGE. B, Ba/F3 cells were "starved" of IL-3 for 6 h (lane 1), maintained in 0.5 ng/ml IL-3 (lane 2), or constitutively grown in media lacking IL-3 (lanes 3-6). Cells were lysed and assayed for phosphorylation of STAT5 by immunoprecipitation of STAT5 and Western blot with 4G10 anti-phosphotyrosine antibodies (upper panel) or anti-STAT5 antibodies (lower panel). The third possibility is that sequences amino-terminal to the TALIN homology region could, in addition to relocalization, mediate interaction with heterologous proteins involved in transformation. Candidates for the heterologous interaction include Huntingtin, HIP1 itself, and the unidentified 130-kDa tyrosine-phosphorylated protein found in transformed cells. Since most candidate proteins in the 110 -140 kDa molecular mass range have been excluded, identification of the 130-kDa protein will probably require biochemical purification and sequence analysis.
Tyr to Phe mutational analysis of the PDGF␤R portion of H/P demonstrates that PtdIns 3-kinase and PLC␥ activation are dispensable for transformation in the Ba/F3 assay. These data contrast with the function of native receptor in HepG2 cells in which there is an absolute dependence on PLC␥ and PtdIns 3-kinase activation for mitogenic signaling. STAT5 is activated by the native PDGF␤R, is known to transmit mitogenic signals, and is activated by IL-3 stimulation of Ba/F3 cells. We have demonstrated that although H/P(F8) does not activate PLC␥ or PtdIns 3-kinase, it retains the ability to activate STAT5 as assessed by Western blot analysis for phosphotyrosine and electromobility shift assay analysis. These data, together with the known activation of STAT5 in hematopoietic cells by BCR/ABL (22,24,25) and TEL/JAK (8,14), suggest that STAT5 activation is a common pathway of transformation by tyrosine kinase-containing fusion proteins.
In summary, we have completed a detailed analysis of the transforming properties of H/P. HIP1 contributes at least two functional domains necessary for transformation: first, a critical oligomerization motif within the TALIN homology region required for kinase activation and transformation of hematopoietic cells and, second, a domain from amino acid 690 to 752 that is required for transformation and association and phosphorylation of p130. Furthermore, p130 phosphorylation and association correlates with transforming ability of H/P. PDGF␤R contributes tyrosine kinase activity, which is necessary but not sufficient for transformation. Although PtdIns 3-kinase and PLC␥ activation are dispensable for transformation, H/P shares a feature common to other tyrosine kinase fusion in its ability to activate STAT5. Identification of p130 and STAT5 target genes should provide further insight into the pathogenesis of CMML.