Fiz1, a Novel Zinc Finger Protein Interacting with the Receptor Tyrosine Kinase Flt3*

The receptor tyrosine kinase Flt3 has been shown to play a role in proliferation and survival of hematopoietic progenitor cells as well as differentiation of early B lymphoid progenitors. However, the signaling events that control growth or differentiation are not completely understood. In order to identify new signaling molecules interacting with the cytoplasmic domain of Flt3, we performed a yeast two-hybrid screen. In addition to several SH2 domain-containing proteins, we have isolated a novel Flt3interacting zinc finger protein (Fiz1) with 11 C2H2-type zinc fingers. Fiz1 binds to the catalytic domain of Flt3 but not to the structurally related receptor tyrosine kinases Kit, Fms, and platelet-derived growth factor receptor. This association is independent of kinase activity. The interaction between Flt3 and Fiz1 detected in yeast was confirmed by in vitro and in vivo coprecipitation assays. Fiz1 mRNA is expressed in all murine cell lines and tissues tested. Anti-Fiz1 antibodies recognize a 60-kDa protein, which is localized in the nucleus as well as in the cytoplasm. Together, these results identified a novel class of interaction between a receptor tyrosine kinase and a signaling molecule which is independent of the well established SH2 domain/phosphotyrosine binding.


From the Fred Hutchinson Cancer Research Center, Division of Basic Sciences, Seattle, Washington 98109-1024
The receptor tyrosine kinase Flt3 has been shown to play a role in proliferation and survival of hematopoietic progenitor cells as well as differentiation of early B lymphoid progenitors. However, the signaling events that control growth or differentiation are not completely understood. In order to identify new signaling molecules interacting with the cytoplasmic domain of Flt3, we performed a yeast two-hybrid screen. In addition to several SH2 domain-containing proteins, we have isolated a novel Flt3 interacting zinc finger protein (Fiz1) with 11 C 2 H 2 -type zinc fingers. Fiz1 binds to the catalytic domain of Flt3 but not to the structurally related receptor tyrosine kinases Kit, Fms, and plateletderived growth factor receptor. This association is independent of kinase activity. The interaction between Flt3 and Fiz1 detected in yeast was confirmed by in vitro and in vivo coprecipitation assays. Fiz1 mRNA is expressed in all murine cell lines and tissues tested. Anti-Fiz1 antibodies recognize a 60-kDa protein, which is localized in the nucleus as well as in the cytoplasm. Together, these results identified a novel class of interaction between a receptor tyrosine kinase and a signaling molecule which is independent of the well established SH2 domain/phosphotyrosine binding.
Flt3 receptor tyrosine kinase (RTK) 1 is a member of the class III RTKs (1)(2)(3)(4)(5). Common structural features include the extracellular region composed of five immunoglobulin-like domains and an intracellular tyrosine kinase made up of an ATP-binding loop and a catalytic domain separated by a kinase insert domain. Additional members of this receptor family are the receptors for macrophage-colony-stimulating factor, and steel factor, encoded by the FMS (6 -8) and KIT (9, 10) protoonco-genes, respectively, and the receptors for ␣and ␤-plateletderived growth factors (PDGFRA and -B). The RTKs Flt3, Fms, and Kit play a key role in hematopoiesis by stimulating proliferation and/or differentiation of various hematopoietic cell types (11,12). Mice lacking a functional Flt3 receptor have normal mature hematopoietic populations; however, they exhibit reduced numbers of early B cell precursors and multipotent stem cells (13). The recently cloned Flt3 ligand (FL) (14 -16), in combination with other cytokines, has been shown to stimulate proliferation of human and murine hematopoietic progenitor/stem cells in vitro as well as in vivo (14 -20). FL also promotes growth of early B cell progenitor cells in combination with IL-7 (21,22) and was shown to induce adhesion of the precursor B cell line BaF3/Flt3 to fibronectin by activating the fibronectin receptors VLA-4 and VLA-5 integrins (23).
Little is known about the signaling pathways of Flt3. By using a chimeric Fms/Flt3 receptor, several groups have shown phosphorylation of and/or association with phospholipase C-␥1, Ras GTPase-activating protein, Vav, Shc, the p85 subunit of phosphatidylinositol 3Ј-kinase, and Grb2 in fibroblasts and pro-B cells (24,25) upon receptor activation. The site of p85 interaction with murine Flt3 was mapped to tyrosine 958 (YQNM) in the C-terminal tail of Flt3 (24,26,27) which fits the consensus sequence favored by the SH2 domains of p85 (28). In contrast, p85 does not interact with human Flt3 (29) which is lacking a potential p85-binding site in the C terminus. In addition, ligand stimulation of the human Flt3 receptor results in phosphorylation of SHP-2, SHIP, and a 100-kDa protein in monocytic THP-1 cells (29), phosphorylation of SHC and CBL in myeloid cells, and SHC and CBLB in pro-B cells (30,31).
In order to understand the specific function of Flt3, it is necessary to understand its signaling pathways in more detail. Therefore, we used the cytoplasmic domain of Flt3 as bait in a two-hybrid screen to identify new signaling molecules that interact with this RTK. This technology has been used successfully to isolate novel substrates of various other RTKs, such as the association of SH2 domains of phospholipase C-␥2 and Mona with Fms (32,33), the SH2 domain of the adaptor molecule APS with Kit (34), the PTB domain of Dok-R with Tek/ Tie2 (35), and the zinc finger protein ZPR1 with the epidermal growth factor receptor (EGFR) (36). In this report we describe the isolation of a novel C 2 H 2 zinc finger protein and its specific interaction with the catalytic domain of Flt3 in a kinase-independent manner.

EXPERIMENTAL PROCEDURES
All recombinant DNA work was done using standard techniques (37). Additional details of plasmid constructions and oligonucleotide sequences are available upon request.
Cells and Culture-The murine hematopoietic cell lines BaF3, BaF3/ Flt3, and BaF3/MT-Fiz1/Flt3 were grown in RPMI 1640 supplemented with 5% fetal calf serum (HyClone), 5% iron-supplemented calf serum (HyClone), and 0.2% conditioned medium of X63-IL-3 cells expressing recombinant IL-3 (38). The BaF3/Flt3 cell line was generated by infection of BaF3 cells with the pJZen2-Flt3 retroviral vector. Infected cells * This work was supported by a Public Health Service Grant CA40987 (to L. R. R.) and by a "Deutsche Forschungsgemeinschaft" postdoctoral fellowship (to I. W.). Biotechnology and Biocomputing Shared Resources at the Fred Hutchinson Cancer Research Center also contributed significantly to this work. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
were sorted for receptor expression by flow cytometry using a polyclonal antibody (antibody 8580) against the extracellular domain of Flt3. The BaF3/MT-Fiz1/Flt3 cell line was generated by electroporating the expression plasmid Rc/CMV-MT-Fiz1 into the parental cell line BaF3 and selecting for G418-resistant clones. After screening for MT-Fiz1 expression, cells were infected with Flt3 as described above. HEK293T and NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium containing 5% fetal calf serum and 5% iron-supplemented calf serum.
Antibodies-The antibody 8580 against the extracellular domain of Flt3 was obtained by immunizing rabbits with Rab9 cells expressing the murine Flt3 receptor. The antibodies 270J and 9767I against the intracellular domain of Flt3 were generated by immunizing rabbits with a bacterially expressed 6ϫHis fusion protein containing residues 788 -1000 of the murine Flt3 receptor which was purified as inclusion bodies and solubilized in 0.1% SDS with boiling. The antibody 9704 against Fiz1 was obtained by immunizing rabbits with a bacterially expressed and affinity purified glutathione S-transferase (GST) fusion protein containing residues 254 -500 (ZF7-11) of the murine Fiz1 protein. Anti-Fiz1 serum was affinity purified by passage through a column of covalently linked GST protein. The flow-through was collected and bound to a column of covalently linked GST-Fiz1/ZF7-11 fusion protein. Affinity purified antibodies were eluted with low pH.
Yeast Two-hybrid Screening-The PCR-amplified cytoplasmic domain of the wild-type murine Flt3 receptor (aa 565-1000) was cloned in frame into the LexA-DNA binding domain vector pBTM116 (39) using the BamHI site. The two-hybrid screen was performed as described previously (40). Briefly, Saccharomyces cerevisiae L40 was sequentially transformed with the LexA-Flt3 bait plasmid and 400 g of a cDNA library, which was made from EML-C1 cells fused to the VP16 transcriptional activation domain (41). A total of 9 ϫ 10 6 transformed colonies were screened on plates lacking tryptophan, histidine, uracil, leucine, and lysine (-WHULK) but containing 50 mM 3-amino-1,2,4triazole (Sigma) and X-gal. Specificity of positive interactors was tested by mating assays (40). Plasmid DNA was isolated from five clones, which were positive for ␤-galactosidase activity yeast, and the VP16 library inserts were PCR-amplified. PCR products were gel-purified and directly sequenced with Applied Biosystems PRISM TM terminator (Perkin-Elmer). Obtained sequences were analyzed using the BLAST program on the World Wide Web.
cDNA Cloning-An EML-C1 cDNA library constructed with the ZAP-cDNA synthesis kit (Stratagene) (41) was used to isolate overlapping Fiz1 cDNA clones which were sequenced on both strands. The 5Ј-UTR was cloned by PCR using a FDC-P1/Mac11 cDNA library as template, T3 as 5Ј primer annealing to vector sequence, and oligonucleotide 3ЈZF4 (5Ј-TGTGTGAATTCATCACACGTTGCAACATACTGA-GC-3Ј) as 3Ј primer annealing to Fiz1 cDNA.
Plasmid Construction and Interaction Trap in Yeast-A BamHI fragment encoding the whole ORF of Fiz1 was cloned into the BglII site of the pCS3ϩMT vector (kindly provided by J. A. Cooper, Fred Hutchinson Cancer Research Center, Seattle, WA) (42) resulting in 6ϫ Myc epitopetagged Fiz1 fusion protein. pCS3ϩMT-Fiz1 was digested with HindIII/ XbaI, and the fragment containing MT-Fiz1 was cloned into the expression plasmid pRc/CMV (Invitrogen).
Construction of the catalytic domain swap mutants utilized overlap extension PCR mutagenesis (43). Codons 784 and 785 of murine Flt3 and codons 771 and 772 of the murine Fms receptor were changed to introduce an NcoI restriction site. Codons 955 and 956 of murine Flt3 and codons 917 and 918 of the murine Fms receptor were mutated to introduce a KpnI restriction site. Mutated fragments were cloned into pBTM116 using the BamHI restriction site, generating the plasmids LexA-Flt3-NK (N784H, V785G, E955V, and A956P) and LexA-Fms-NK (R771H, P772G, D917V, and Q918P). By using the NcoI and KpnI restriction sites, the swap mutants LexA-Flt3/TKII-Fms and LexA-Fms/TKII-Flt3, respectively, were cloned by standard restriction digest and ligation. All constructs were sequenced to confirm correct mutagenesis.
To test for interaction, LexA plasmids were transformed into yeast strain AMR70. VP16 plasmids were transformed into yeast strain L40. Mating assays were performed as described previously (40), and interaction of various mutants (Fig. 3) was visualized by spotting cells onto -WHULK plates containing 50 mM 3-amino-1,2,4-triazole and X-gal. Color of colonies was recorded after 3 days.
[ 35 S]Methionine-labeled Flt3 was obtained by cloning the cytoplasmic domain into the vector pET14b (Novagen), and in vitro transcription/translation was done using the STP2, T7 system (Novagen). Lysates were precleared by incubating with GST protein bound to beads and then used for binding reactions with 1 g of GST or 1 g of GST-Fiz1 protein in IVT-150 buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 M ZnCl 2 , 1% Nonidet P-40). Interacting complexes were washed 3 times with IVT-750 buffer (IVT buffer with 750 mM NaCl) and eluted by boiling in 2ϫ Laemmli buffer. Proteins were separated on a 10% SDS-polyacrylamide gel, and bound proteins were visualized by autoradiography.
1.3 ϫ 10 7 cells of the BaF3 or BaF3/Flt3 cell line were lysed by sonication in IVT-150 buffer (supplemented with 1 mM PMSF, 20 g/ml aprotinin). Lysates were cleared by centrifugation and precleared by incubation with GST protein bound to beads. Supernatants were incubated with 5 g of GST or GST-Fiz1 protein in IVT-150 buffer, washed 3 times with IVT-150 buffer, and eluted by boiling in 2ϫ Laemmli buffer. Proteins were separated on a 7.5% SDS-polyacrylamide gel, transferred to nitrocellulose membrane, and blotted with anti-Flt3 antiserum 270J.
Immunoprecipitation and Immunoblotting-BaF3/MT-Fiz1/Flt3 cells (2.5 ϫ 10 7 /sample) were starved in RPMI, 1% FBS for 3 h, washed in PBS, stimulated with 1 g/ml Flt3 ligand (FL, a kind gift of Immunex, Seattle, WA) for 5 min at 37°C, and lysed by sonication in Nonidet P-40 lysis buffer (50 mM NaCl, 50 mM Tris-HCl, pH 7.3, 30 mM Na 4 P 2 O 7 , 50 mM NaF, 5 M ZnCl 2 , 0.2% Nonidet P-40, 1 mM PMSF, 20 g/ml aprotinin, 2 mM orthovanadate). Lysates were cleared of cell debris by centrifugation and subsequently precleared for 30 min at 4°C with 5 l of preimmune serum and 20 l protein A-Sepharose (Amersham Pharmacia Biotech). Supernatants (equalized for protein amount) were used for immunoprecipitation (IP) and incubated overnight with 20 l of affinity purified anti-Fiz1 antibody and 20 l of protein A-Sepharose. IPs were washed 3 times with lysis buffer and eluted by boiling in 2ϫ Laemmli buffer. Proteins were separated on a 6.5% SDS-polyacrylamide gel and transferred to nitrocellulose membrane on a semi-dry blotting apparatus (Ellard Instrumentation Ltd., Seattle, WA). After probing with antibodies as indicated, immunocomplexes were visualized using enhanced chemiluminescence (NEN Life Science Products).
Cell Fractionation-Fractionation was done as described previously (44). In brief, cells (2.5 ϫ 10 7 /fractionation) were washed once in ice-cold PBS and once in hypotonic buffer (10 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 1 mM PMSF, 20 g of aprotinin). Cells were lysed in hypotonic buffer using a Dounce homogenizer (tight fitting pestle, 20 -30 strokes). Nuclei were spun down at 800 ϫ g, and the nuclear pellet was washed twice in hypotonic buffer and further purified by resuspending in 0.4 ml of hypotonic buffer containing 0.5 M sucrose, 0.1% Triton X-100 and centrifuging through a cushion of 1 ml of hypotonic buffer containing 1 M sucrose, 0.1% Triton X-100. Nuclei were lysed by sonication in Nonidet P-40 lysis buffer (0.5% Nonidet P-40). Membranes were separated from the cytoplasmic fraction by high speed centrifugation (100,000 ϫ g), and the pellet containing the membranes was resuspended in Nonidet P-40 lysis buffer. Fractions were analyzed by IP/Western blot.
Northern Blot Hybridization-Total RNA of various cell lines or mouse tissues was extracted with guanidinium isothiocyanate and purified by centrifugation through a 5.7 M CsCl gradient (45). By using oligo(dT)-cellulose (New England Biolabs) affinity purification, poly(A) ϩ RNA was isolated from mouse tissue total RNA. 20 g of total RNA or 3 g of poly(A) ϩ RNA were separated on 1.2% agarose, 0.6 M formaldehyde gels, transferred to GeneScreen membrane (NEN Life Science Products), hybridized, and washed as recommended by the manufacturer.

RESULTS
Isolation of Fiz1, a Novel Zinc Finger Protein-To identify novel signaling intermediates of the receptor tyrosine kinase Flt3, we employed the yeast two-hybrid system. The hybrid bait consisted of the DNA binding/dimerization domain of LexA and the cytoplasmic region of murine Flt3 (4,5). A screen of 9 ϫ 10 6 primary transformants of an EML-C1 cDNA expression library (41) yielded five potential positive clones detected by strong ␤-galactosidase activity. We obtained known SH2 domain-containing proteins including the p85 subunit of phosphatidylinositol 3Ј-kinase, 3BP-2, 3 Nck, and ShcB, and a novel protein we named Fiz1 (Flt3 interacting zinc finger protein 1).
Several overlapping cDNA clones for Fiz1 were isolated by screening an EML-C1 cDNA library. To ensure the complete ORF was represented, two different 5Ј-untranslated regions (5Ј-UTR) were cloned by PCR using an FDC-P1/Mac11 cDNA library (41) as template. The two transcripts, 2,587 and 2,595 base pairs in length, contained a single open reading frame (ORF) encoding a protein of 500 aa (Fig. 1A) with a calculated molecular mass of 52 kDa. The predicted initiator methionine codon is contained within the context of a consensus translational initiation sequence defined for higher eukaryotes (47). Upstream of the common first ATG codon we found an in-frame stop codon in both 5Ј-UTRs. Sequence comparison revealed that the initial 1.6-kilobase pair Fiz1 cDNA clone (EML-33) obtained from the two-hybrid library contained the full-length ORF. Data base searches indicated that Fiz1 encodes a novel protein with homology to numerous zinc finger proteins. As shown in Fig. 1, Fiz1 contains 11 putative zinc finger domains (ZF1-ZF11) matching the consensus C 2 H 2 zinc finger motif (48) (Fig. 1B). The only exceptions from the consensus are seen in ZF7, which shows a separation of the two conserved cysteine residues with one instead of two amino acids, and in ZF8 which has a conservative Phe 3 Tyr substitution. The short sequence between the ZF domains follows, to a high degree, the defined sequence of the H/C link (49). The zinc fingers are distributed throughout the protein and occur in clusters of four, two, two, and three (Fig. 1C).
Expression of Fiz1 in Cells and Tissues-The pattern of Fiz1 expression was examined by Northern blot analysis of RNA from various mouse cell lines and tissues (Fig. 2). A single RNA species of 2.6 kilobase pairs was detected in all hematopoietic and non-hematopoietic cell lines tested but to a lower degree in Bac.1, Raw264.7, and WEHI cells. In addition, it was found to be expressed constitutively in all mouse tissues examined (brain, heart, kidney, liver, lung, skeletal muscle, pancreas, spleen, thymus, and uterus).

Fiz1 Binds Specifically to the Catalytic Kinase Domain of
Flt3-To analyze the interaction of Flt3 with Fiz1 in more detail, different Flt3 mutants were tested for interaction in the yeast two-hybrid system (Fig. 3). Replacing the lysine 645 within the ATP-binding site of the kinase domain with an alanine (K645A) results in a kinase-defective Flt3 mutant that binds Fiz1 with the same strength as wild-type Flt3 (Fig. 3B). In contrast, Fiz1 does not bind to the cytoplasmic domain of the structurally related receptor tyrosine kinases Fms, Kit, and PDGFR (Fig. 3B). Therefore, Fiz1 binds specifically to Flt3 in a kinase-independent manner.
To determine what region of the Flt3 receptor is bound by Fiz1, several Flt3 deletion mutants were examined (Fig. 3B). Deletion of the kinase insert domain or the C terminus (CT) had no effect on binding of Fiz1 to Flt3. However, deletion of the juxtamembrane domain (JM) completely eliminated the interaction. Expression of JM alone or JM plus tyrosine kinase domain I (TKI, ATP-binding loop) failed to restore Fiz1 interaction. These results suggest that Fiz1 does not bind the JM domain of Flt3 directly, but a deletion of JM changes the conformation of the catalytic domain (TKII) and therefore interferes with Fiz1 binding. In fact, deletion of JM results in a kinase-inactive receptor molecule indicating a role of this region in the regulation of kinase activity (Fig. 3B). The impact of JM on modulating the kinase activity of receptors has been previously reported. The Y569A mutation in JM of c-Fms (50) as well as the mutations Y579F and Y581F in the JM of the PDGFR (51) result in kinase-inactive receptors. The deletion of seven aa in JM of c-Kit constitutively activates the receptor (52). Moreover, Fiz1 did not bind to a bait expressing just TKII and CT (Fig. 3B) most likely because this truncated mutant is not folded in the correct conformation. To test the hypothesis that Fiz1 binds to TKII of Flt3, we generated TKII swap mutants. As shown in Fig. 3C, Fiz1 bound to wild-type Flt3 but not to wild-type Fms. Exchanging TKII of Flt3 with TKII of Fms resulted in loss of binding (Flt3/TKII-Fms, Fig. 3C), whereas cloning of TKII of Flt3 into Fms resulted in gain of binding (Fms/TKII-Flt3, Fig. 3C). All constructs were tyrosine-kinase functional in the two-hybrid assay, as shown by positive interaction with the SH2 domain of p85. These results lead us to conclude that TKII indeed is the site of Fiz1 interaction.
Fiz1 Interacts with Flt3 in Vitro-To confirm the specific interaction between Fiz1 and Flt3 observed in yeast cells, we examined the potential direct interaction of Fiz1 with Flt3 in vitro. Full-length Fiz1 was expressed in E. coli as a GST fusion protein. Fusion proteins were immobilized on glutathione-agarose beads and incubated with in vitro translated, [ 35 S]methionine-labeled, cytoplasmic domain of Flt3 (aa 565-1000). As shown in Fig. 4A, Flt3 specifically associated with GST-Fiz1 but not with GST alone.
To confirm this result, GST-Fiz1 fusion protein was incubated with total cell lysates prepared from the pre-B cell line BaF3 or a subline BaF3/Flt3 expressing the wild-type Flt3 receptor. As shown in Fig. 4B, GST-Fiz1 interacted with the immature as well as the mature glycosylated Flt3 receptor molecule. GST alone did not interact with any of the receptor isoforms. These results verify that Fiz1 is able to associate in vitro with Flt3 without receptor activation.
In Vivo Association of Fiz1 and Flt3-To examine the interaction between Fiz1 and Flt3 in mammalian cells, we generated a rabbit polyclonal antiserum reactive against the Cterminal half of Fiz1 (ZF7-ZF11, Fig. 1C). The antiserum was affinity purified using the antigen. This antiserum detected a 60-kDa protein in HEK 293T cells transfected with an expression vector encoding full-length Fiz1 (Fig. 5A). By using whole cell extracts from BaF3 cells, a single protein band of the same size was detected suggesting that the product encoded by the isolated cDNA corresponds to the endogenous protein (Fig. 5A).
To confirm the interaction between Flt3 and Fiz1 in vivo, coimmunoprecipitations were performed. BaF3 cells expressing Myc-tagged murine Fiz1 together with wild-type murine Flt3 (BaF3/MT-Fiz1/Flt3) were starved of growth factor and either unstimulated or stimulated with Flt3 ligand (FL, 5 min . B, yeast mating assays between S. cerevisiae L40 containing VP16-Fiz1 and S. cerevisiae AMR70 expressing wild-type or mutant Flt3 cytoplasmic domains, or wild-type Kit, Fms, or PDGFR cytoplasmic domains. Interactions were monitored by blue color after spotting yeast on plates lacking WHULK but containing 50 mM 3-amino-1,2,4-triazole and X-gal. Kinase activity of each bait was monitored by Western blotting yeast total cell lysates and probing with a ␣-phosphotyrosine antibody (4G10). C, yeast mating assays between S. cerevisiae L40 containing VP16-Fiz1 or VP16 fused to the C-terminal SH2 domain of p85, as a positive control, and S. cerevisiae AMR70 expressing cytoplasmic domains of wild-type Flt3, wild-type Fms, or the catalytic swap mutants Flt3/TKII-Fms and Fms/TKII-Flt3.

FIG. 4. In vitro interaction between Flt3 and Fiz1
. GST or GST-Fiz1 fusion proteins were expressed in E. coli BL21 (DE3), purified, and immobilized on glutathione-agarose beads. A, in vitro translated, [ 35 S]methionine-labeled Flt3 (aa 565-1000) was incubated with 1 g of GST or GST-Fiz1 protein. The beads were washed, and bound proteins were eluted, separated by SDS-PAGE (6.5%), and visualized by autoradiography. B, total cell lysate isolated from BaF3 or BaF3/Flt3 cells were incubated with 1 g of GST or GST-Fiz1 protein. Bound proteins were run on a 6.5% polyacrylamide gel, blotted, and probed with a ␣-Flt3 antibody. at 37°C). Cell extracts were incubated with affinity purified ␣-Fiz1 antibody, and immunoprecipitated proteins were analyzed by Western blotting with an ␣-Flt3 antibody. As shown in Fig. 5B, the glycosylated as well as the unmodified form of Flt3 specifically coprecipitated with Fiz1 in extracts from stimulated, but not unstimulated, cells. The blot was reprobed with ␣-Fiz1 antibody to show that similar levels of Fiz1 were present in each IP. This result confirmed the in vivo association of Fiz1 with Flt3 following receptor stimulation.
Interaction between RTKs and their signaling molecules, including SH2 domain-containing proteins, often results in tyrosine phosphorylation of these molecules. We could not detect any tyrosine phosphorylation of Fiz1 following receptor activation (data not shown). Furthermore, forced overexpression of Fiz1 in the cell line BaF3/MT-Fiz1/Flt3 had no effect on proliferation rate as well as factor-dependent growth of the cells. Likewise, Flt3 kinase activity was not changed in these cells (data not shown).
Fiz1 Is a Nuclear and Cytoplasmic Protein-Many C 2 H 2 zinc finger proteins are found in the nucleus. However, Fiz1 localization exclusively in the nucleus would pose a problem with regard to its potential interaction with cytoplasmic Flt3. To test the subcellular localization of Fiz1, we purified nuclear, cytoplasmic, and membrane fractions of BaF3/Flt3 cells. The distribution of the Fiz1 protein was determined by IP/Western analyses using the ␣-Fiz1 antibody. As shown in Fig. 6A, Fiz1 is localized mainly in the nuclear and cytoplasmic compartment of the cell; very little Fiz1 protein was detected in the membrane fraction.
To confirm the subcellular localization of Fiz1 direct immunofluorescence studies were carried out. Immunofluorescence staining of NIH3T3 fibroblasts expressing endogenous Fiz1 demonstrated the presence of Fiz1 protein in the nuclear as well as the cytoplasmic compartment (Fig. 6B, panels a and b). The cytoplasmic staining appeared in speckles suggesting the possibility of localization to endosome compartments of the cell. However, treatment of the cells with 0.03% saponin (53) prior to fixation resulted in the loss of about 80% of the cytoplasmic Fiz1 staining, showing that Fiz1 seems not to be associated with cellular structures like endosomes. Specificity of ␣-Fiz1 antibody was shown by incubating the primary antibody with the antigen GST-Fiz1/ZF7-11 that was used to generate the antiserum. No immunofluorescence was detectable in those cells (Fig. 6B, panels c and d). DISCUSSION In this study we describe the identification of a novel C 2 H 2 zinc finger protein Fiz1 via its interaction with the receptor tyrosine kinase Flt3. The association found in yeast was verified with in vitro and in vivo coprecipitation assays.
Fiz1 protein is composed of 11 C 2 H 2 zinc finger domains spread over the entire protein in clusters of two, three, or four fingers. The classical C 2 H 2 zinc finger domains were first identified in the transcription factors TFIIIA and Krü ppel and were shown to be DNA binding domains (54). However, recent stud- A, BaF3 cells expressing wild-type Flt3 were separated into cytoplasmic (C), membrane (M), and nuclear (N) fractions. Distribution of Fiz1 protein was determined by analyzing the fractions by immunoprecipitation using an ␣-Fiz1 antibody. Precipitated proteins were run on a 6.5% polyacrylamide gel and blotted with an ␣-Fiz1 antibody. B, NIH3T3 cells were plated on fibronectin-coated coverslips, fixed with 4% paraformaldehyde, stained using an affinity purified ␣-Fiz1 antibody (panels a and c), and visualized with a fluorescein isothiocyanate-conjugated anti-rabbit antibody. ␣-Fiz1 antibody was blocked specifically by the antigen GST-Fiz1/ZF7-11 (panel c). Nuclei were stained with 4,6-diamidino-2-phenylindole reagent (panels b and d), and stains were examined with a Deltavision microscope.
ies elucidated the involvement of C 2 H 2 zinc finger domains in protein-protein interactions. Homodimerization of the multifinger protein Ikaros is mediated by its two C-terminal fingers (55). Likewise, Friend-of-GATA binds GATA-1 with its sixth Krü ppel-like finger (56), and the interaction between early endosomal autoantigen EEA1 and the small GTPase Rab5 involves the C 2 H 2 zinc finger motif found in the N terminus of EEA1 (57). More recently, different types of zinc finger proteins were found to be an important class of receptor-interacting molecules and to be involved in receptor signaling. For example, Enigma, a LIM domain protein, associates with the insulin receptor (58); TRAF, a ring finger protein, binds the tumor necrosis factor receptor (59); and the CCCC zinc finger motif of ZPR1 interacts with the EGFR tyrosine kinase (36). Fiz1 is the first example of a C 2 H 2 zinc finger protein binding to a receptor tyrosine kinase. Furthermore, we found that the N-terminal four (ZF1-4) or the C-terminal three (ZF9 -11) zinc finger domains of Fiz1, respectively, are sufficient to interact independently with Flt3 (data not shown).
The interaction of Fiz1 was mapped to the catalytic domain of Flt3. This situation is equivalent to that shown for the zinc finger protein ZPR1, which binds residues 908 -958 of the EGFR, corresponding to subdomains X and XI of the tyrosine kinase domain (36). Upon receptor stimulation and tyrosine phosphorylation, ZPR1 is released from EGFR and translocates into the nucleus. We have investigated whether a similar scenario is seen for Fiz1. However, Fiz1 binding to Flt3 occurs after ligand stimulation of the receptor (Fig. 5). In addition, there is no detectable change in Fiz1 protein distribution within the cell after receptor stimulation (data not shown). Therefore, initial Fiz1 localization to the nucleus is either independent of Flt3 suggesting a separate nuclear function of Fiz1 which is not effected by Flt3. Alternatively, Fiz1 translocation to the nucleus might need additional signals like cell contact or adhesion which is supported by the finding that densely grown NIH3T3 cells show Fiz1 staining that is restricted to the nucleus (data not shown). Thus, although Fiz1 and ZPR1 bind to the catalytic domains of Flt3 and EGFR, respectively, receptor stimulation has a different impact on the binding of each zinc finger protein to its respective receptor and resultant cellular distribution of each protein.
The interaction of Fiz1 is not dependent on the tyrosine kinase activity of the receptor, as the kinase dead Flt3 mutant (K645A) interacts with Fiz1 with the same strength as wildtype Flt3. In addition, Fiz1 binds in vitro to the non-phosphorylated mature receptor expressed on the surface of unstimulated BaF3/Flt3 cells as well as to the immature receptorprecursor. However, as shown in Fig. 5, in vivo binding of Fiz1 to Flt3 occurs after stimulation of Flt3 with its ligand FL. Thus, ligand-induced relocalization of the mature Flt3 receptor to the cytoplasm as a result of internalization (60) might be necessary for in vivo association of the mature Flt3 receptor and Fiz1. Other possibilities might be that ligand stimulation of the receptor enhances binding affinity of Fiz1 and Flt3 in mammalian cells, eventually by receptor oligomerization or even receptor phosphorylation. Receptor dimerization is known to occur in the yeast system due to the dimerization of the LexA-DNA binding domain. However, receptor oligomerization might not be relevant in the in vitro system. Thus, Flt3/Fiz1 binding in the yeast and in vitro systems might be facilitated by higher protein concentrations and therefore independent of receptor stimulation. Further investigation needs to be done to clarify the exact mechanism of this interaction.
As shown in Fig. 3B, Fiz1 interacts specifically with Flt3. This specificity is surprising with respect to the expression pattern of Fiz1. Flt3 expression is restricted to bone marrow and early hematopoietic progenitor cells, thymus, spleen, lymph nodes, placenta, brain and gonads (4,5,46). In contrast, Fiz1 is expressed ubiquitously, suggesting that Fiz1 may interact with other proteins besides Flt3 and proteins different from class III RTKs. To address this question, a yeast two-hybrid screen using Fiz1 as bait has been performed. The additional Fiz1-interacting proteins will be reported elsewhere.
In conclusion, we have identified a novel C 2 H 2 zinc finger protein Fiz1 by its interaction with the catalytic domain of the receptor tyrosine kinase Flt3. Although the physiological function remains to be clarified, Fiz1 represents a new class of receptor-interacting molecules.