Functional Cloning and Characterization of a Novel ...

3 downloads 223 Views 256KB Size Report
Feb 16, 2000 - ... Wei-Feng Shen¶, Nicolas Pineault‡, Suzan Imren‡, Ben Montpetit‡, ...... Shen, W. F., Chang, C. P., Rozenfeld, S., Sauvageau, G., Humphries, ...
THE JOURNAL

OF

BIOLOGICAL CHEMISTRY

Vol. 275, No. 34, Issue of August 25, pp. 26172–26177, 2000 Printed in U.S.A.

Functional Cloning and Characterization of a Novel Nonhomeodomain Protein That Inhibits the Binding of PBX1-HOX Complexes to DNA* Received for publication, February 16, 2000, and in revised form, April 27, 2000 Published, JBC Papers in Press, May 23, 2000, DOI 10.1074/jbc.M001323199

Carolina Abramovich‡§, Wei-Feng Shen¶, Nicolas Pineault‡, Suzan Imren‡, Ben Montpetit‡, Corey Largman¶, and R. Keith Humphries‡储** From ‡The Terry Fox Laboratory, British Columbia Cancer Agency, Vancouver, British Columbia V5Z 1L3, Canada, the ¶Department of Medicine, San Francisco Veterans Affairs Medical Center and University of California, San Francisco, California 94121, and the 储Department of Medicine, University of British Columbia, Vancouver, British Columbia V6T 2B5, Canada

PBX1 is a homeodomain protein that functions in complexes with other homeodomain-containing proteins to regulate gene expression during developmental and/or differentiation processes. A yeast two-hybrid screen of a fetal liver-hematopoietic cDNA library using PBX1a as bait led to the discovery of a novel non-homeodomaincontaining protein that interacts with PBX1 as well as PBX2 and PBX3. RNA analysis revealed it to be expressed in CD34ⴙ hematopoietic cell populations enriched in primitive progenitors, as is PBX1; search of the expressed sequence tag data base indicated that it is also expressed in other early embryonic as well as adult tissues. The full-length cDNA encodes a 731-amino acid protein that has no significant homology to known proteins. This protein that we have termed hematopoietic PBX-interacting protein (HPIP) is mainly localized in the cytosol and in small amounts in the nucleus. The region of PBX that interacts with HPIP includes both the homeodomain and immediate N-terminal flanking sequences. Strikingly, electrophoretic mobility shift assays revealed that HPIP inhibits the ability of PBX-HOX heterodimers to bind to target sequences. Moreover, HPIP strongly inhibits the transactivation activity of E2A-PBX. Together these findings suggest that HPIP is a new regulator of PBX function.

The PBC protein family (mammalian PBX, C. elegans CEH-20 and Drosophila extradenticle) make critical contributions to cell fate and segmental patterning during embryogenic development (1–3). PBX1, a member of the PBX family along with PBX2 and PBX3 (4), was initially identified as the chromosome 1 participant of the t(1;19) translocation, which occurs in 25% of pediatric pre-B cell acute lymphocytic leukemia and that creates a chimeric gene designated E2A-PBX1 (5, 6). The mechanism by which E2A-PBX1 causes leukemia is still un* This work was supported by the National Cancer Institute of Canada with funds from the Canadian Cancer Society and The Terry Fox Run, the Medical Research Council of Canada (MRC-C), National Institutes of Health Grant DK48642, and the Department of Veterans Affairs (to C. L.). 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. § Recipient of fellowships from the Leukemia Research Society of Canada and MRC-C. ** To whom correspondence should be addressed: Terry Fox Laboratory, 601 W. 10th Avenue, Vancouver, BC V5Z 1L3, Canada. Tel.: 604-877-6070 (ext. 3095); Fax: 604-877-0712; E-mail: keith@ terryfox.ubc.ca.

clear. However, the structure of the protein, in which the majority of PBX1, including the homeodomain, is fused to the transcriptional activation domain of E2A (6, 7), suggests that the oncogenic properties of E2A-PBX1 result from inappropriate regulation of target genes whose expression during hematopoiesis is normally regulated by wild type PBX proteins (8 –11). In vitro and in vivo data strongly suggest that PBX functions in combination with heterologous homeodomain proteins, including class I HOX proteins. As HOX cofactors, PBC proteins improve HOX specificity due to the increased size of the cooperative binding site and the strength of DNA binding, as well as by modulating recognition of cooperative binding sites by different groups of HOX proteins (12–14). In addition, cooperative DNA binding with PBC proteins may act to change the regulatory signal of HOX proteins, from repressors to activators (15). HOX-PBC interaction optimally requires the PBC homeodomain and a short carboxyl-terminal region, called the HOX cooperativity motif (16, 17). Although PBC proteins contribute to HOX DNA binding specificity, PBC-HOX complexes exhibit little transcriptional activity and thus alone do not account for the precise functions that these complexes execute in vivo (18). The highly conserved amino-terminal domain of PBC proteins mediates protein dimerizion and DNA binding with members of the MEIS subfamily of homeodomain proteins (MEIS-HTH and PNOX1PREP1) (19 –23). Moreover, recent data have shown that PREP1, MEIS, and HTH proteins can bind to DNA together with PBC-HOX heterodimers to generate trimeric protein complexes and that PREP1-PBX-HOX complexes are more potent transcriptional activators than PBX-HOX heterodimers (24 – 27). Unlike HOX proteins from paralog groups 1– 8, PREP1 and MEIS can interact with PBX in the absence of DNA (25, 27), suggesting that PREP1 as well as MEIS and HOX proteins from paralog groups 9 and 10 form stable PBX complexes in vivo. Indeed, DNA-independent complex formation with PREP1/MEIS/HTH can regulate the cellular localization of PBC proteins by masking the activity of a nuclear export signal, resulting in nuclear localization (28, 29). PBX proteins thus appear to function as part of large nucleoprotein complexes. The interactions within these complexes are probably decisive factors that allow the DNA binding proteins to discriminate among target regulatory elements. How these complexes are regulated during either early embryonic development or cellular differentiation of somatic cells to control gene expression is still unclear. We speculated that characterization of additional PBX-interacting proteins might shed

26172

This paper is available on line at http://www.jbc.org

Pbx-interacting Protein light on the mechanism of PBX function, and specifically, we sought to identify novel cofactors or modifiers of PBX1 using the yeast two-hybrid system. We chose to focus on the hematopoietic system as a source of putative interacting factors due to the compelling evidence showing the importance of PBX as well as PBX partners in normal hematopoiesis and in leukemia (11, 30 –32). We report here the cloning and characterization of a novel PBX-interacting protein, hematopoietic PBX1-interacting protein (HPIP). We demonstrate that HPIP interacts with PBX1 in vivo and, moreover, that it prevents the binding of PBX1-HOX complexes to consensus DNA binding sites. We also show that HPIP inhibits the transactivation activity of E2A-PBX. HPIP appears to be the first nonhomeoprotein shown to interact directly with and regulate the transcriptional factor function of PBX1. We also show that HPIP is co-expressed with PBX1 in early hematopoietic cells, and therefore, it may contribute to the function of PBX1 during hematopoiesis and leukemic transformation. EXPERIMENTAL PROCEDURES

Construct—PBX1a (herein designated as PBX1) fragments were amplified from the Sp-PBX1a plasmid (kindly provided by M. Cleary, Stanford University). PCRs1 were carried out using Platinium Taq DNA polymerase (Life Technologies, Inc.) as recommended by the manufacturer, and all DNA fragments generated were fully sequenced to rule out possible PCR-induced mutations. PBX and HPIP Expression Plasmids—cDNA fragments encoding the entire PBX1 coding sequence without the first Met (PBX1⌬Met) were amplified by PCR using specific primers and cloned into the yeast pGBT9 vector (CLONTECH), the bacterial pGX-5X-3X (Amersham Pharmacia Biotech) vector, and the mammalian pSuperCatch vector (33) to generate pGBT9-PBX, pGX-PBX, and pSC-FLAG-PBX, respectively. Different fragments of PBX1, as indicated in Fig. 6, were amplified by PCR using the appropriate combination of primers and cloned into pAC2 to create pAC2-PBX deletion mutants. An expression cassette for PBX1 tagged on the amino teminus with an HA epitope was generated by cloning a PBX1⌬Met amplified by PCR in frame with an HA epitope isolated from pCFV 3HA and inserted into Bluescript KS⫹. The HA-PBX1 cDNA was then cloned into pSuperCatch (pSC-HA-PBX1). HPIP4 cDNA was cloned into pGBT9 to make pBT9 HPIP4. HPIP5 cDNA was cloned into pSuperCatch to make pSC-HPIP5. HPIP cDNA was isolated from BS-HPIP and cloned into pSuperCatch in frame with the FLAG epitope to make pSC-HPIP. FLAG-tagged HPIP encoding cDNA was isolated from pSC-HPIP and cloned into the HpaI site of the murine retroviral vector MSCVneoEB (Robert G. Hawley, Sunnybrook Research Institute, Toronto, Canada) (34) to make MSCVNeo-FLAG-HPIP. Virus-containing supernatants were collected from BOSC cells transfected with the retroviral vector using standard protocols. HPIP cDNA was isolated from pSC-HPIP and cloned into pEGF-C1 vector (CLONTECH Laboratories, Inc.) to make pEGFP HPIP. HPIP5 cDNA was cloned into pM (CLONTECH) to make pM HPIP5. Details of all constructs are available upon request. Yeast Two-hybrid Screen and cDNA Cloning—The bait plasmid, pGBT9-PBX1, was transfected into the HF7C strain and used to screen 106 independent pAC2 activator plasmid cDNAs from human week 14 fetal liver (CLONTECH). Colony selection, isolation of plasmid DNA from positive clones, and tests of specificity were performed essentially as described (35). Positive pAC2 plasmid inserts were excised with BglII, cloned into the BamHI site of BS-KS, and sequenced. To obtain a full-length HPIP cDNA, 105 independent phages from a human bone marrow CD34⫹ cDNA library2 were screened with 32P-labeled HPIP5. In Vitro and In Vivo Binding Experiments—Bacterially expressed GST-PBX protein was purified as described (35). pSC-PBX1 was trans-

1 The abbreviations used are: PCR, polymerase chain reaction; GFP, green fluorescent protein; HPIP, hematopoietic PBX-interacting protein; SC, SuperCatch; HA, hemagglutinin; GBD, Gal4 DNA binding domain; aa, amino acid(s); EMSA, electrophoretic mobility shift assay; PAGE, polyacrylamide gel electrophoresis. 2 G. Sauvageau, P. Rosten, and R. K. Humphries, unpublished observations.

26173

fected into COS cells, and 48 –72 h later, FLAG-tagged PBX (FLAGPBX) protein was extracted using high salt buffer (1% Triton, 50 mM Hepes, 500 mM NaCl, 1 ␮g/ml aprotenin, 0.1 mM phenylmethylsulfonyl fluoride, 100 ␮g/ml leupeptin), diluted to a final concentration of 150 mM NaCl, incubated with anti-FLAG agarose-M2 beads (Kodak Scientific Imaging System) for 2 h at 4 °C and washed three times with isotonic buffer (high salt buffer with 150 mM NaCl) to give M2-PBX1 beads. As a control, proteins from nontransfected COS cells were extracted, incubated with beads, and washed as above (M2-COS beads). The HA-tagged HPIP5 (HA-HPIP5) protein was produced in vitro using the TNT Coupled Reticulocyte Lysate System (Promega). Pull-down assays were performed by incubating equal amounts of GST or GSTPBX1 immobilized on glutathione-Sepharose beads with HA-HPIP5 diluted in isotonic buffer. The mixtures were incubated for 3 h at 4 °C and washed three times with isotonic buffer, and bound proteins were eluted, separated by 10% SDS-polyacrylamide gel electrophoresis, and subjected to Western blot analysis using monoclonal anti-HA antibodies (Babco) (36). For pull-down assays with COS-produced FLAG-PBX1, equal amounts of anti FLAG-M2 beads (M2 beads), M2-PBX1 beads, or M2-COS beads were mixed with HA-HPIP5 protein and processed as above. For co-immunoprecipitation, 10 ␮g/10-cm plate of pSC-FLAGHPIP5, pSC-HA-PBX1, or both plasmids together were transfected into COS cells by the DEAE-dextran method (37); proteins were extracted after 48 h with high salt buffer, diluted to 150 mM NaCl, and incubated with anti-FLAG M2 beads for 4 h. Bound proteins were washed, separated on SDS-PAGE, and subjected to Western blots with anti-FLAG M2 monoclonal antibodies (Kodak Scientific Imaging System) or polyclonal anti-HA antibodies (Babco). Analysis of HPIP mRNA Expression—Bone marrow cells from human organ donor vertebrae were fractionated into CD34⫹ and CD34⫺ subpopulations using a monoclonal anti-CD34 antibody and fluorescence-activated cell sorting; total RNA was isolated and assayed by reverse transcription-PCR as described (38). Amplified cDNAs were separated on a 1% agarose gel, transferred to ␨ probe membrane (BioRad), and probed with randomly primed radioactive DNA ([␣-32P]ATP) fragments specific for each gene. For Northern blots, poly(A)⫹ RNA was isolated from total RNA (prepared using Trizol; Life Technologies, Inc.) using the Oligotex Direct mRNA kit (Qiagen). 2 ␮g of poly(A)⫹ RNA was subjected to formaldehyde/agarose gel electrophoresis, transferred to ␨ probe membrane and probed. Subcellular Localization and Molecular Mass of HPIP—Rat-1 cells were infected with MSCVNeo-FLAG-HPIP virus, selected with 1.5 mg/ml of G418 for 7 days and plated on glass cover slides for 48 h. Fixed cells were permeabilized with 0.1% Triton in PBS and blocked with PBS 3% bovine serum albumin, and the FLAG-HPIP fusion protein was visualized by incubation with M2 anti-FLAG antibodies (1:500), washed, and incubated with anti-mouse IgG-horseradish peroxidaseconjugated second antibody (1:50). The cover slides were mounted for examination with a Zeiss Axiphot fluorescence microscope. NIH 3T3 cells were transfected with GFP-HPIP by electroporation. Twenty-four hours after transfection, the cells were replated on glass cover slides. Twenty-four hours later, the cells were fixed, stained with 4⬘,6-diamidino-2-phenylindole, mounted, and viewed. Nuclear and cytoplasmic extracts from Rat-1 cells expressing FLAGHPIP were isolated following standard procedures (39). Extracts were subjected to SDS-PAGE and Western analysis using the anti-FLAG M2 monoclonal antibody. Electrophoretic Mobility Shift Assay—EMSA was performed using complementary oligonucleotides containing consensus binding sites for either PBX-HOXB7, PBX1-HOXA9, or PBX1-HOXA10 as described previously (13, 27). Transient Transfections and Luciferase Assay—For transient transfection, 293 cells were split into 24-well plates and 24 h later were transfected (by calcium phosphate) with a mixture of 3 ␮g of plasmid in each of triplicate wells. For a typical triplicate, 1 ␮g of pGL3-PBX reporter plasmid (kindly provided by M. Kamps, University of California, San Diego), 0.2 ␮g of MSCVpac E2A-PBX1a (40), and 1 ␮g of either pM, pM HPIP5, MSCVNeo FLAG HPIP, or MSCVNeo was used. The pM and MSCVNeo vectors were used in these experiments to balance the amount of input DNA in transfections. Cells were harvested 48 h after transfection, and luciferase activities were measured with the Luciferase Assay System (Promega). RESULTS

A Two-hybrid Screen for Partner Proteins of PBX1 Yielded a Novel Nonhomeodomain Protein—In an effort to identify proteins that interact with PBX1 in early stages of hematopoiesis,

26174

Pbx-interacting Protein

FIG. 1. Interaction of HPIP with PBX1. In vitro translated HAHPIP5 was incubated with GST-PBX1 or GST immobilized on glutathione-agarose beads (A) or with FLAG-PBX1 bound to anti-FLAG M2 beads (M2-PBX1), anti-FLAG M2 beads alone (M2), or anti-FLAG M2 beads incubated with extracts of nontransfected COS cells (M2-COS) (B). After the beads were washed, bound proteins were eluted and resolved on 10% SDS-PAGE. Western blot analysis was performed using anti-HA monoclonal antibodies. C, coimmunoprecipitation of FLAG-HPIP5 and HA-PBX1 in transfected COS cells. Total cell extracts were immunoprecipitated with anti-FLAG M2 beads (IP ⫹) or left untreated (IP ⫺). Western blot analysis was performed using antiFLAG M2 monoclonal antibodies (upper panel) or anti-HA polyclonal antibodies (lower panel).

we used the full-length PBX1a cDNA as bait to screen a library of human fetal liver cDNAs prepared at 14 weeks gestation when the liver is an active site of hematopoiesis. A yeast two-hybrid screen of 106 primary transformants yielded seven potential positive clones that showed strong interaction with PBX1 fused to the Gal4 DNA binding domain (GBD) but not with two negative control fusion proteins, GBD-lamin and GBD-rev. Sequence analysis revealed that two of the clones contained independent, partial cDNA inserts derived from the same mRNA. We named the protein encoded by this mRNA HPIP (for hematopoietic PBX1-interacting protein), and the proteins derived from the partial cDNA as HPIP4 and HPIP5, respectively, where HPIP5 cDNA has an additional 100 base pairs at the 5⬘-end (Fig. 3). HPIP Associates in Vitro and in Vivo with PBX1a—The interaction between HPIP and PBX1 observed in yeast cells was further confirmed in vitro using HA-tagged HPIP5 and immobilized GST-PBX1 or FLAG-PBX1 proteins. Immobilized GST-PBX was incubated with HA-HPIP5, and bound protein was analyzed by Western blotting. As shown in Fig. 1A, HAHPIP5 (lane 1) associated with GST-PBX1 (lane 2) but not with GST alone (lane 3). Furthermore, as shown in Fig. 1B, HAHPIP5 (lane 1) interacted with anti-FLAG M2 beads that were incubated with extracts of COS cells expressing FLAG-PBX1 (lane 3) but not with M2 beads alone (lane 2) or M2 beads incubated with extracts of mock-transfected COS cells. In vivo interaction between HPIP and PBX1 in mammalian cells was demonstrated by co-immunoprecipitation of FLAGHPIP5 and HA-tagged PBX1 expressed in COS cells (Fig. 1C). HPIP Is Expressed in Early Hematopoietic Cells—HPIP expression in hematopoietic cells was confirmed by Northern blot analysis of erythroid K562 and myeloid HL60 cells, which revealed a 5-kilobase pair transcript in both cell lines (Fig. 2). HPIP expression in primitive human hematopoietic cells was also demonstrated by reverse transcription-PCR analysis of RNA obtained from human bone marrow. Fig. 2B shows that HPIP is strongly expressed in the CD34⫹ fraction containing the hematopoietic progenitors and at lower levels in the CD34⫺ mature cell population. The same pattern of expression was

FIG. 2. Expression of HPIP mRNA. A, Northern blot analysis of poly(A)⫹ RNA (5 ␮g/lane) from the indicated cell lines; B, Southern blot analysis of globally amplified cDNA prepared from mRNA of CD34⫹ and CD34⫺ cells probed with HPIP4 cDNA.

found for PBX1, indicating that HPIP and PBX1 are co-expressed in the same hematopoietic compartment. A search of the expressed sequence tag data base using HPIP5 cDNA sequence showed that several embryonic and adult human clones encode potential proteins essentially identical to portions of HPIP, including clones isolated from heart, adult and infant brain, cerebellum, synovial sarcoma, and Jurkat T-cells. Mouse expressed sequence tags with significant matches to HPIP were identified from 14-day postcoitum embryos, irradiated colon, heart, and 2-cell embryos, indicating that human HPIP has a mouse homolog gene. The Sequence of the Full Coding Region of HPIP Revealed a Novel Protein with No Homology to Any Known Protein—Using the partial HPIP5 cDNA as a probe, a 3.5-kilobase pair clone containing a 2,193-nucleotide potential open reading frame and a 1,200-nucleotide 3⬘-untranslated region was identified from a human bone marrow CD34⫹ cDNA library (GenBankTM accession number AF221521). The open reading frame would encode a protein of 731 amino acids (aa) with a predicted molecular mass of 80 kDa (Fig. 3). Thus, HPIP4 and HPIP5 constitute the C-terminal 308 and 341 aa respectively (aa 424 –731 and 391– 731) (Fig. 3). Nucleotide data base searches showed that the sequence of HPIP cDNA, from nucleotide 1799 to nucleotide 2210, has 83% identity to a cDNA cloned from human lymphocytes (human clone A9A2BRB7, GenBankTM accession no. U00952) (41). Protein data base searches indicated that HPIP encodes a novel protein with no significant homology to any proteins of known function. Analysis using the PSORT II algorithm predicted that HPIP has a coil-coil secondary structure between aa 234 and 392 (coil regions have been implicated in protein-protein interactions) and two putative nuclear localization signals at aa 131 and 462, with the later resembling bipartite nuclear localization sequences commonly found in nuclear transcription factors. HPIP has several putative phosphorylation sites (e.g. sites for CKII, GSK3, mitogen-activated protein kinase, protein kinase A, protein kinase C, protein kinase B, protein kinase G, CaMII, p34cdc2, and p70s6k) and a leucine-rich region that resembles nuclear export signals that bind to the CRM1 export receptor (42) (Fig. 3). HPIP Is Expressed Mainly in the Cytosol—To assess the subcellular localization of HPIP, a GFP-HPIP fusion protein was expressed in NIH 3T3 cells. As shown in Fig. 4, most GFP-HPIP-expressing cells showed strong staining in the cytoplasm and more diffuse nuclear staining. Occasional cells showed stronger nuclear staining (not shown). A similar pattern of HPIP localization was observed in RAT-1 cells expressing HPIP tagged with a FLAG epitope (Fig. 4). Together, these findings indicate that HPIP is expressed mainly in the cytosol and in small amounts in the nucleus. To determine the molec-

Pbx-interacting Protein

26175

FIG. 5. Molecular mass of HPIP. Nuclear (lane 1) and cytosolic (lane 2) preparations of Rat-1 cells (5 ⫻ 106 and 106 cells, respectively) stably expressing FLAG-HPIP were resolved on a 7.5% SDS-PAGE. FLAG-HPIP was detected by immunoblotting with anti FLAG M2 antibodies. Lane 3 is from a total cell extract of nontransfected Rat-1 cells.

FIG. 3. Predicted amino acid sequence of HPIP. The beginnings of HPIP4 and HPIP5 are represented with a dotted arrow or black arrow, respectively; the nuclear localization signals are in boldface type, and the leucine-rich regions are underlined.

FIG. 4. Subcellular localization of HPIP. A, GFP-HPIP was detected by fluorescence microscopy. B, the same cells were counterstained with 4⬘,6-diamidino-2-phenylindole to reveal the nucleus. 4⬘,6Diamidino-2-phenylindole stains the nuclei of cells, of which only a subset are transfected and hence GFP-positive. C, Rat-1 cells infected with MSCV-FLAG-HPIP virus. The expressed FLAG-HPIP protein was detected by immunofluorescence microscopy using anti-FLAG M2 antibodies and Texas Red-labeled anti-mouse IgG.

ular weight of HPIP and to see whether there is a difference in the mass between the cytosolic and nuclear protein, Western blot analysis of nuclear and cytosolic cell extracts of RAT-1 cells expressing FLAG-HPIP was performed. A strong band of approximately 98 kDa was apparent in cytosolic extracts, and a fainter band of the same molecular weight was seen in nuclear extracts (Fig. 5). The HPIP Binding Domain of PBX1 Overlaps with Sequence Important for PBX1 Function—To localize the regions within PBX1 that interact with HPIP, different regions of PBX1 in frame with an HA epitope and the Gal4 activation domain were cloned. The different PBX1 plasmids were co-transfected with HPIP4 cDNA cloned in frame with the Gal4 binding domain in the pGBT9 vector, and the interaction of HPIP4 with the dif-

ferent PBX1 deletions was scored according to the color of the colonies by ␤-galactosidase filter assay. A PBX1 deletion mutant containing the N terminus plus the homeodomain (PBX1(2–290)) bound HPIP4 with the same efficiency as the wild type PBX1 (Fig. 6). Conversely, PBX1 mutants expressing either the homeodomain alone (PBX1 230 –290) or the homeodomain plus the C terminus (PBX1-(230 – 430)) did not detectably interact with HPIP4 in this assay. Interestingly, mutant PBX1-(2–230), containing the N terminus of the protein, showed very weak binding to HPIP4, and only the addition of the full homeodomain restored the strong interaction seen with the full-length PBX1. These results indicate that the sequence of PBX1 N terminus of the homeodomain is sufficient for interaction with HPIP4; however, the homeodomain is needed for strong interaction. The contributions of sequences N-terminal to the homeodomain were further mapped to sequences between aa 160 and the homeodomain (Fig. 6). This region is highly conserved in the PBX family, and as shown in Fig. 6, both PBX2 and PBX3 were able to interact with HPIP4. HPIP Blocks Cooperative PBX-HOX Protein Binding to DNA Targets—One of the major biological roles of PBX proteins is the formation of DNA binding complexes with HOX protein partners (9, 12, 13, 17, 43– 45). Since PBX proteins do not exhibit strong DNA binding in the absence of HOX partners, to assess the possible biological significance of HPIP, we studied its effect on the ability of PBX protein to form cooperative DNA binding complexes with a series of HOX proteins. We first used EMSA to analyze the influence of full-length HPIP on the formation of a heterodimeric DNA binding complex between PBX1 and HOXB7 proteins. HPIP reduced complex formation by approximately 70% (Fig. 7, compare lane 2 with lane 3). However, since the reticulocyte lysate used to synthesize the various proteins produced a nonspecific gel shift band in the same location as the PBX-HOXB7 complex (lane 1), the EMSA analysis was repeated using a different oligonucleotide target on which PBX forms cooperative gel shift bands with HOXA9 and HOXA10 but for which the lysate nonspecific binding is greatly decreased (14). In these experiments, the addition of HPIP decreased PBX-HOXA10 and PBX-HOXA9 complexes by 50 and 80%, respectively (Fig. 7, lane 3). HPIP Inhibits the Transcriptional Activation Capacity of E2A-PBX—We then investigated whether the inhibition of the DNA binding activity of PBX by HPIP, seen in the EMSA, would be reflected in the transcriptional activity of PBX. PBX, in contrast with E2A-PBX, is unable to activate transcription by itself (8, 10, 46); therefore, and based on our deletion data pointing at HPIP as a potential E2A-PBX binding protein, we studied the effect of HPIP on the ability of E2A-PBX to induce a luciferase reporter gene driven by seven copies of a E2A-PBX binding site (TGATTGAT) (47) upstream of the minimal FOS promoter. Co-expression of the E2A-PBX reporter plasmid, E2A-PBX, and HPIP decreased the E2A-PBX-mediated tran-

26176

Pbx-interacting Protein

FIG. 6. Schematic representation of PBX1 mutants and summary of their binding to HPIP. The designation of each protein is indicated at the left (the numbers indicate the first and last amino acid of the mutant PBX proteins). Dotted boxes indicate the PBX1 homeodomain; filled boxes denote the PBX1 sequence missing in the E2A-PBX fusion protein. SFY526 yeast cells were cotransfected with pGBT9HPIP4 and various deletions of PBX1 fused to the GAL-4 activating domain of pAC2. Each double transformant was plated on Trp⫺ Leu⫺ and tested for expression of lacZ by a ␤-galactosidase filter assay (⫹ ⫹, intense blue signal visible between 1 and 2 h of starting of the assay; ⫹, weak blue signal after 6 h; ⫺, absence of blue color after 36 h). Expression of HA-PBX mutants was confirmed by Western blot analysis with anti-HA antibodies (data not shown).

FIG. 7. EMSA analysis demonstrating the effects of HPIP on the capacity of PBX to form cooperative DNA binding complexes with HOX proteins. In vitro transcribed/translated PBX1a and HOXB7, HOXA9, or HOXA10 proteins were incubated with labeled oligonucleotide target containing a TGATTTAT consensus DNA binding site for PBX-HOXB7 or a TGATTTAC binding site for PBX-HOXA9 and PBX-HOXA10 in the absence (lane 2) or presence of full-length HPIP (lane 3), and the resulting mixtures were subjected to EMSA analysis. The control lane contained reticulocyte lysate (lane 1). PBX1a and HPIP did not shift either oligonucleotide, while HOXA9 and HOXA10 proteins alone gave EMSA bands in different positions (not shown) but did not show bands at the migration position of the PBX-HOX dimeric complex.

scriptional activity seen in cells expressing the reporter plasmid and E2A-PBX alone by approximately 80% (Fig. 8). This was reproducible in multiple independent experiments and using two different expression vectors for HPIP 5 and fulllength HPIP.

FIG. 8. HPIP repression of E2A-PBX mediated transactivation. 293 cells were transfected with pGL3-PBX luciferase reporter plasmid, E2A-PBX and pM HPIP5 (A) or MSCVNeo-FLAG-HPIP (B). The figure shows the relative luciferase activity over the basal activity observed with the reporter plasmid and either pM or MSCVNeo plasmids alone.

DISCUSSION

Although PBX homeodomain protein is thought to function as a transcription factor, its mechanism of action remains unknown. The only proteins known to associate with PBX are other homeodomain proteins. Using a yeast two-hybrid screen with the full-length PBX1a as bait, we cloned two partial cDNAs from a fetal hematopoietic liver cell library encoding a novel nonhomeodomain protein that binds PBX1 as well as PBX2 and PBX3. The validity of the association of HPIP with PBX1 was supported by in vitro interaction in pull-down experiments and in vivo by co-precipitation from mammalian cells transiently transfected with both cDNAs. HPIP cDNA encodes a novel protein of 731 amino acid residues containing no homology to any known protein. The predicted HPIP protein has a calculated molecular mass of 80 kDa that migrates as a 98-kDa polypeptide in SDS-PAGE. The slow migration of HPIP possibly results from either post-translational modifications or intrinsic SDS-resistant folding of the protein. HPIP is predicted to have a coil-coil domain, suggesting that it interacts with other proteins. Northern blot analysis showed that HPIP was expressed in hematopoietic cell lines of erythroid (K562) and myeloid (HL60) origin. Interestingly, PBX2 and PBX3 but not PBX1 are expressed in HL60 cells (4), suggesting that HPIP may regulate the functions of the first two proteins in this cell line. Using Southern blot analysis of total amplified RNA, we showed that the expression of HPIP, as well as the expression of PBX1, was much higher in primitive hematopoietic cells than in more mature ones. These results indicate co-expression of PBX and HPIP in primitive hematopoietic precursors and suggest a functional interaction between these two proteins during early hematopoiesis. Expressed sequence tags corresponding to portions of HPIP DNA were identified from embryonic and adult human and mouse tissues, suggesting that the function of HPIP is not restricted to the hematopoietic system and that it may be implicated in the early stages of development. HPIP has two putative nuclear localization signals and a leucine-rich region that resembles a nuclear export signal. Our subcellular localization studies show HPIP to be localized mainly in the cytoplasm, although a small amount was found in the nucleus. This finding, together with the knowledge that PBX is expressed in both nuclei and cytosol (48), implies that HPIP could in principle interact with PBX in either compartment. If the initial interaction occurs in the cytoplasm, the heterodimer would either stay in the cytosol or enter the nucleus as a

Pbx-interacting Protein complex. If the initial interaction occurs in the nucleus, HPIP would have to enter the nucleus on its own and then bind PBX. In either case, the interaction between HPIP and PBX could block the function of the latter. Using deletion mutants of PBX1 in the yeast two-hybrid system, we mapped the HPIP binding domain of PBX to the region between aa 160 and 230. This sequence, on its own, was enough to support weak binding to HPIP; however, the fulllength homeodomain was required for strong interaction. Two possibilities can account for this result; one is that the homeodomain directly contacts HPIP, and the other is that the homeodomain is needed to maintain the conformation of the N-terminal region open to bind HPIP. We support the first explanation, as we saw, using co-immunoprecipitation experiments with in vitro translated HPIP and PBX proteins, that the PBX homeodomain alone was able to bind the full-length HPIP (data not shown). Specific regions have been identified in PBX1 that mediate its functions. The PBX sequence that we found sufficient for binding to HPIP (aa 160 –230), partially overlaps a domain reported by Lu and Kamps to selectively repress Sp1-activated transcription in a DNA binding-independent manner (49). Moreover, Kamps et al. reported a region of PBX1 upstream of the homeodomain (aa 89 –232) in E2A-PBX1 as responsible for homeodomain-independent transformation of NIH 3T3 cells (50). The PBX homeodomain has been shown to be essential for binding of PBX to DNA and for interaction with HOX proteins (16, 51–53). Furthermore, the PBX homeodomain is needed in E2A-PBX for blocking myeloid differentiation as well as for E2A-PBX-mediated apoptosis of hematopoietic cells (50, 54). Interestingly, in contrast to HOX proteins, HPIP does not have either of the tryptophan-containing motifs shown to bind the tryptophan-binding pocket of PBX (13, 14, 43– 45, 52, 53, 55), suggesting that the PBX homeodomain interacts with HPIP through a novel binding mechanism. Insight into the function of the interaction between HPIP and PBX1 was obtained by electrophoretic mobility shift assays, which revealed that the full-length HPIP protein inhibited the binding of different PBX-HOX complexes to DNA. Since the binding of PBX to DNA is essential for its transcriptional activity, this finding suggests that HPIP could modulate the transcription factor function of PBX. Indeed, we showed that the transient expression of HPIP in transfected cells significantly inhibited the enhanced E2A-PBX-mediated transcription of a reporter plasmid. These data are significant, since they indicate 1) a functional in vivo interaction between HPIP and PBX and 2) that HPIP can repress the transactivation activity of E2A-PBX. In summary, we have discovered HPIP, a novel nonhomeodomain protein that is able to interact with the different members of the PBX family as well as with the PBX sequences on E2A-PBX. We have shown that HPIP is co-expressed with PBX1 in early hematopoietic precursors and that HPIP is able to block the binding of PBX1 to a PBX-HOX consensus binding site and to inhibit the transcriptional activation activity of E2A-PBX. While the functions of HPIP remain to be fully resolved, the data presented here strongly suggest that it is likely to be a regulator of PBX proteins in their activities as transcriptional regulators. Acknowledgments—We thank D. Mabbasa for technical assistance. REFERENCES 1. Ryoo, H. D., and Mann, R. S. (1999) Genes Dev. 13, 1704 –1716 2. Rauskolb, C., Smith, K. M., Peifer, M., and Wieschaus, E. (1995) Development 121, 3663–3673 3. Gonzalez-Crespo, S., and Morata, G. (1995) Development 121, 2117–2125 4. Monica, K., Galili, N., Nourse, J., Saltman, D., and Cleary, M. L. (1991) Mol. Cell. Biol. 11, 6149 – 6157 5. Kamps, M. P., Look, A. T., and Baltimore, D. (1991) Genes Dev. 5, 358 –368 6. Nourse, J., Mellentin, J. D., Galili, N., Wilkinson, J., Stanbridge, E., Smith,

26177

S. D., and Cleary, M. L. (1990) Cell 60, 535–545 7. Kamps, M. P., Murre, C., Sun, X. H., and Baltimore, D. (1990) Cell 60, 547–555 8. LeBrun, D. P., and Cleary, M. L. (1994) Oncogene 9, 1641–1647 9. Lu, Q., Knoepfler, P. S., Scheele, J., Wright, D. D., and Kamps, M. P. (1995) Mol. Cell. Biol. 15, 3786 –3795 10. Van Dijk, M. A., Voorhoeve, P. M., and Murre, C. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 6061– 6065 11. Thorsteinsdottir, U., Krosl, J., Kroon, E., Haman, A., Hoang, T., and Sauvageau, G. (1999) Mol. Cell. Biol. 19, 6355– 6366 12. Mann, R. S., and Chan, S. K. (1996) Trends Genet. 12, 258 –262 13. Shen, W. F., Chang, C. P., Rozenfeld, S., Sauvageau, G., Humphries, R. K., Lu, M., Lawrence, H. J., Cleary, M. L., and Largman, C. (1996) Nucleic Acids Res. 24, 898 –906 14. Shen, W. F., Rozenfeld, S., Lawrence, H. J., and Largman, C. (1997) J. Biol. Chem. 272, 8198 – 8206 15. Pinsonneault, J., Florence, B., Vaessin, H., and McGinnis, W. (1997) EMBO J. 16, 2032–2042 16. Green, N. C., Rambaldi, I., Teakles, J., and Featherstone, M. S. (1998) J. Biol. Chem. 273, 13273–13279 17. Chang, C. P., Brocchieri, L., Shen, W. F., Largman, C., and Cleary, M. L. (1996) Mol. Cell. Biol. 16, 1734 –1745 18. Krosl, J., Baban, S., Krosl, G., Rozenfeld, S., Largman, C., and Sauvageau, G. (1998) Oncogene 16, 3403–3412 19. Knoepfler, P. S., Calvo, K. R., Chen, H., Antonarakis, S. E., and Kamps, M. P. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 14553–14558 20. Berthelsen, J., Zappavigna, V., Mavilio, F., and Blasi, F. (1998) EMBO J. 17, 1423–1433 21. Chang, C. P., Jacobs, Y., Nakamura, T., Jenkins, N. A., Copeland, N. G., and Cleary, M. L. (1997) Mol. Cell. Biol. 17, 5679 –5687 22. Ferretti, E., Schulz, H., Talarico, D., Blasi, F., and Berthelsen, J. (1999) Mech. Dev. 83, 53– 64 23. Bischof, L. J., Kagawa, N., Moskow, J. J., Takahashi, Y., Iwamatsu, A., Buchberg, A. M., and Waterman, M. R. (1998) J. Biol. Chem. 273, 7941–7948 24. Jacobs, Y., Schnabel, C. A., and Cleary, M. L. (1999) Mol. Cell. Biol. 19, 5134 –5142 25. Berthelsen, J., Zappavigna, V., Ferretti, E., Mavilio, F., and Blasi, F. (1998) EMBO J. 17, 1434 –1445 26. Goudet, G., Delhalle, S., Biemar, F., Martial, J. A., and Peers, B. (1999) J. Biol. Chem. 274, 4067– 4073 27. Shen, W. F., Rozenfeld, S., Kwong, A., Kom ves, L. G., Lawrence, H. J., and Largman, C. (1999) Mol. Cell. Biol. 19, 3051–3061 28. Berthelsen, J., Kilstrup-Nielsen, C., Blasi, F., Mavilio, F., and Zappavigna, V. (1999) Genes Dev. 13, 946 –953 29. Abu-Shaar, M., Ryoo, H. D., and Mann, R. S. (1999) Genes Dev. 13, 935–945 30. Lawrence, H. J., Helgason, C. D., Sauvageau, G., Fong, S., Izon, D. J., Humphries, R. K., and Largman, C. (1997) Blood 89, 1922–1930 31. Shimamoto, T., Ohyashiki, K., Toyama, K., and Takeshita, K. (1998) Int. J. Hematol. 67, 339 –350 32. Thorsteinsdottir, U., Sauvageau, G., and Humphries, R. K. (1997) Hematol. Oncol. Clin. North Am. 11, 1221–1237 33. Georgiev, O., Bourquin, J. P., Gstaiger, M., Knoepfel, L., Schaffner, W., and Hovens, C. (1996) Gene (Amst.) 168, 165–167 34. Hawley, R. G., Lieu, F. H., Fong, A. Z. and Hawley, T. S. (1994) Gene Ther. 1, 136 –138 35. Abramovich, C., Yakobson, B., Chebath, J., and Revel, M. (1997) EMBO J. 16, 260 –266 36. Abramovich, C., Shulman, L. M., Ratovitski, E., Harroch, S., Tovey, M., Eid, P., and Revel, M. (1994) EMBO J. 13, 5871–5877 37. Sambrook. J, Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, pp. 16.41–16.46, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 38. Sauvageau, G., Lansdorp, P. M., Eaves, C. J., Hogge, D. E., Dragowska, W. H., Reid, D. S., Largman, C., Lawrence, H. J., and Humphries, R. K. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 12223–12227 39. Davis L., Kuehl, M., and Battey, J. (1994) Basic Methods in Molecular Biology, 2nd Ed. (Lange, A. A., Ed.) Elsevier Science, New York 40. Thorsteinsdottir, U., Krosl, J., Kroon, E., Haman, A., Hoang, T., and Sauvageau, G. (1999) Mol. Cell Biol. 19, 6355– 6366 41. Epplen, C., and Epplen, J. T. (1994) Hum. Genet. 93, 35– 41 42. Fornerod, M., Ohno, M., Yoshida, M., and Mattaj, I. W. (1997) Cell 90, 1051–1060 43. Chang, C. P., Shen, W. F., Rozenfeld, S., Lawrence, H. J., Largman, C., and Cleary, M. L. (1995) Genes Dev. 9, 663– 674 44. Neuteboom, S. T., Peltenburg, L. T., van Dijk, M. A., and Murre, C. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 9166 –9170 45. Phelan, M. L., Rambaldi, I., and Featherstone, M. S. (1995) Mol. Cell. Biol. 15, 3989 –3997 46. Lu, Q., Wright, D. D., and Kamps, M. P. (1994) Mol. Cell. Biol. 14, 3938 –3948 47. Calvo, K. R., Knoepfler, P., McGrath, S., and Kamps, M. P. (1999) Oncogene 18, 8033– 8043 48. LeBrun, D. P., Matthews, B. P., Feldman, B. J., and Cleary, M. L. (1997) Oncogene 15, 2059 –2067 49. Lu, Q., and Kamps, M. P. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 470 – 474 50. Kamps, M. P., Wright, D. D., and Lu, Q. (1996) Oncogene 12, 19 –30 51. Lu, Q., and Kamps, M. P. (1996) Mol. Cell. Biol. 16, 1632–1640 52. Passner, J. M., Ryoo, H. D., Shen, L., Mann, R. S., and Aggarwal, A. K. (1999) Nature 397, 714 –719 53. Piper, D. E., Batchelor, A. H., Chang, C. P., Cleary, M. L., and Wolberger, C. (1999) Cell 96, 587–597 54. Smith, K. S., Jacobs, Y., Chang, C. P., and Cleary, M. L. (1997) Oncogene 14, 2917–2926 55. Knoepfler, P. S., and Kamps, M. P. (1995) Mol. Cell. Biol. 15, 5811–5819