Host cell factor-1 interacts with and antagonizes transactivation by the cell cycle regulatory factor Miz-1.

Human host cell factor-1 (HCF-1) is essential for cell cycle progression and is required, in conjunction with the herpes simplex virus transactivator VP16, for induction of viral immediate-early gene expression. We show here that HCF-1 directly binds to the Myc-interacting protein Miz-1, a transcription factor that induces cell cycle arrest at G(1), in part by directly stimulating expression of the cyclin-dependent kinase inhibitor p15(INK4b). A domain encompassing amino acids 750-836, contained within a subregion of HCF-1 required for cell cycle progression, was sufficient to bind Miz-1. Conversely, HCF-1 interacted with two separate regions in Miz-1: the N-terminal POZ domain and a C-terminal domain (residues 637-803) previously shown to harbor determinants for interaction with c-Myc and the coactivator p300. The latter functioned as a potent transactivation domain when tethered to DNA, indicating that HCF-1 targets a transactivation function in Miz-1. HCF-1 or a Miz-1-binding fragment of HCF-1 repressed transactivation by Gal4-Miz-1 in transfection assays. Moreover, HCF-1 repressed Miz-1-mediated transactivation of a reporter gene linked to the p15(INK4b) promoter. Protein/protein interaction studies and transient transfection assays demonstrated that HCF-1 interferes with recruitment of p300 to Miz-1, similar to what has been reported with c-Myc. Our findings identify Miz-1 as a novel HCF-1-interacting partner and illustrate cross-talk between these two proteins that may be of consequence to their respective functions in gene regulation and their opposing effects on the cell cycle.

Human host cell factor-1 (HCF-1) 1 is a transcriptional regulatory protein that was originally identified as an accessory factor required for induction of herpes simplex virus immediate-early genes by the viral transactivator VP16 (1)(2)(3)(4)(5). Subsequently, findings demonstrated that HCF-1 is an essential cellular protein that is required for cell proliferation (6). HCF-1 binds directly to VP16 and, in conjunction with the cellular octamer-binding transcription factor Oct-1 promotes the cooperative assembly and stability of a multicomponent protein⅐DNA transcription complex termed the VP16-induced complex (VIC) on regulatory elements present in the promoter regions of the herpes simplex virus immediate-early genes (reviewed in Refs. 1 and 7). Numerous studies of VP16 and its association with Oct-1, HCF-1, and DNA have provided fundamental insights into mechanisms of transcriptional activation and how combinatorial networks of protein/protein and protein/DNA interactions underpin complexity, specificity, and diversity in transcriptional regulation (1,7).
Human HCF-1 is a ubiquitously expressed and evolutionarily conserved protein with a number of unusual properties and features (3,5,8,9). HCF-1 is synthesized as a 2035-amino acid long precursor protein that is autocatalytically processed at centrally reiterated 26-amino acid repeat elements (HCF PRO repeats) to generate a family of N-and C-terminal polypeptides that remain tightly, but noncovalently, associated with each other (10 -13). VP16 associates with a discrete 380-residue N-terminal modular domain referred to as HCF VIC (also called the kelch domain), which is composed of six repeated copies of a kelch-like sequence proposed to form a barrel-like six-bladed ␤-propeller (14). The HCF VIC domain is necessary and sufficient to bind VP16, to stabilize VIC, and to promote transactivation (14).
In addition to subserving a role in viral gene expression, HCF-1 is essential for normal cell cycle progression (6). This finding arose from studies of tsBN67 cells, a temperaturesensitive hamster cell line that reversibly arrests at the G 0 /G 1 decision point of the cell cycle at the nonpermissive temperature. The defect in tsBN67 cells is due to a single proline-toserine missense mutation (P134S) at position 134 in HCF-1 (6). Interestingly, HCF-1(P134S) also fails to bind to VP16 (14), foretelling the existence of cellular factors that mimic VP16 in their interaction with HCF-1. The first such HCF-1-interacting cellular factor identified was LZIP (also called Luman) (15)(16)(17), a basic leucine zipper protein belonging to the cAMP-responsive element-binding protein/activating transcription factor family of transcription factors. Like VP16, LZIP and a related protein called Zhangfei (18) target determinants in the HCF VIC domain. More recently, the transcription factors GA-binding protein (19) and Sp1 (20), the nuclear hormone receptor coregulatory factor PGC-1 (21), and a protein phosphatase (22) have been shown to associate with HCF-1.
The foregoing adds to growing evidence that HCF-1 is an essential, multifunctional, co-regulatory protein that plays a global role in coordinating viral and cellular gene regulation and cell proliferation. However, the cellular role and mechanisms of action of HCF-1 and the identity of its cellular gene targets are essentially unknown. Recent findings show that HCF-1 is chromatin-associated, and it has been postulated that HCF-1 is recruited to DNA through its association with sequence-specific DNA-binding proteins, analogous to what occurs with VP16 and Oct-1 (23). The HCF VIC domain is neces-sary and sufficient for chromatin association, and the P134S mutation renders this association temperature-sensitive, suggesting that this detachment is responsible for the growth arrest phenotype in tsBN67 cells (23). However, the minimal region capable of rescuing tsBN67 cells encompasses residues 1-902, which include the HCF VIC domain and a downstream region rich in basic amino acids (6,14). Moreover, the binding of the cellular proteins LZIP and Zhangfei to HCF-1 is not required to rescue tsBN67 cells and to promote cell cycle progression (23,24). This indicates that other cellular factors important for cell cycle control that target the N-terminal domain and/or the basic region of HCF-1 may exist. The importance of the basic region in cell proliferation is underscored by recent studies with the HCF-1 family member HCF-2 and the related Caenorhabditis elegans homolog (25,26). HCF-1, HCF-2, and C. elegans HCF share conserved N-and C-terminal domains; however, HCF-2 and C. elegans HCF lack the basic region and the central HCF PRO repeats. Although HCF-2 and C. elegans HCF are able to support VIC formation, these proteins are unable to rescue the temperature-sensitive cell cycle defect in tsBN67 cells (26). Coexpression of HCF-2 inhibits rescue by HCF-1, however, suggesting that the two factors share a common interacting partner(s) (25). The foregoing indicates that the basic region of HCF-1 provides an additional function that is required, in conjunction with the N-terminal proximal region, to promote cell cycle progression.
To shed light on the cellular and functional roles of HCF-1, we carried out yeast two-hybrid interaction cloning with the HCF-1 basic region as bait to identify putative cellular factors that target this region. We show here that HCF-1 interacts physically and functionally with Miz-1, a recently identified cell cycle regulatory factor that was originally isolated by virtue of its ability to interact with the cellular oncoprotein c-Myc (27)(28)(29). Miz-1 is an 803-amino acid long POZ domain/zinc finger transcription factor that modulates transcription by binding directly to initiator elements of target genes (27,30,31). In contrast to the stimulation of cell cycle progression ascribed to HCF-1, Miz-1 causes cell cycle arrest at G 1 (27). This is manifested in part by Miz-1-mediated activation of the gene encoding the key cell cycle inhibitor p15 INK4b , a potent inhibitor of cyclin-dependent kinases (30). Accumulation of p15 INK4b at G 1 results in cell cycle arrest due to a reduction in cyclin D1-associated kinase activity. Recent studies have shown that transcriptional activation of p15 INK4b and the subsequent cell cycle arrest are potentiated by the antimitogenic cytokine transforming growth factor-␤ and antagonized by the cell proliferation mediator c-Myc through opposing pathways that directly converge on Miz-1 (30,33). Thus, c-Myc binds directly to Miz-1 on the p15 INK4b initiator element and represses Miz-1 transactivation by abrogating recruitment of the coactivator p300. Conversely, transforming growth factor-␤ activates SMAD proteins, which relieve c-Myc-mediated repression by causing the dissociation of c-Myc from Miz-1 and which directly transactivate the p15 INK4b promoter in cooperation with Miz-1. Miz-1 is thus at the nexus of reciprocal growth regulatory signaling pathways that link p15 INK4b to cell cycle control (32).
We demonstrate here that Miz-1/HCF-1 association is manifested through the basic domain of HCF-1 and by two independent regions of Miz-1: the N-terminal POZ domain and a C-terminal transactivation domain. We further show that HCF-1 represses Miz-1-mediated transactivation of a reporter gene linked to the p15 INK4b promoter, likely as a result of HCF-1 interfering with the recruitment of p300 to Miz-1. Thus, HCF-1 functions in a manner analogous to c-Myc in modulating Miz-1 function. Our findings point to a convergence between two regulatory proteins that have opposing roles in cell proliferation and may thus be relevant to the mechanisms by which HCF-1 promotes cell cycle progression.
GST Pull-down Assays-Protein binding assays with GST fusion proteins and [ 35 S]methionine-radiolabeled proteins synthesized in vitro using a coupled rabbit reticulocyte transcription/translation system (Promega) were carried out as described previously (38). Briefly, E. coli BL21 cells harboring expression vectors for GST-Miz-1, GST-HCF-1, or GST alone were grown to an A 600 nm of 0.6 -0.8 and induced with isopropyl-␤-D-thiogalactopyranoside (Gibco BRL) for 3 h. Bacteria were collected by centrifugation and resuspended in buffer containing 0.5% Nonidet P-40, 1 mM EDTA, 20 mM Tris-Cl (pH 8.0), 100 mM NaCl, and one tablet of mini C protease inhibitor (Roche Molecular Biochemicals)/25 ml of buffer, and cell extracts were prepared by sonication. 50 l of a 50:50 slurry of glutathione-Sepharose 4B was incubated with clarified cell extracts containing GST or GST fusion protein for 1 h at 4°C. Beads were collected by centrifugation and washed twice with phosphate-buffered saline, and beads containing equivalent amounts of bound protein (as determined by Coomassie Blue staining of SDSpolyacrylamide gels) were incubated with 10 -20 l of reticulocyte lysate containing radiolabeled translated protein in buffer A (150 mM KCl, 0.02 mg/ml bovine serum albumin, 0.1% Triton X-100, 0.1% Nonidet P-40, 5 mM MgCl 2 , and 20 mM Hepes (pH 7.9)) for 2-3 h at 4°C. Beads were washed extensively with buffer A lacking bovine serum albumin, and bound radiolabeled proteins were eluted from the beads by boiling in SDS sample buffer and analyzed by SDS-PAGE. Competition assays were carried out with unlabeled competitor protein synthesized in vitro as detailed in the figure legends. The total amount of rabbit reticulocyte lysate was kept constant by adding unprogrammed lysate to the binding reactions as required.
Cell Culture, Transfections, and Luciferase Assays-COS-1 cells were cultured at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 1% L-glutamine, and 1% penicillin/streptomycin. Cells were seeded in six-well plates at a concentration of 2-3 ϫ 10 5 cells/well 1 day prior to transfection to achieve 70 -80% confluence. Transfections were carried out using LipofectAMINE reagent (4 -8 l/well; Invitrogen) as outlined by the manufacturer. Unless indicated otherwise in the figure legends, transfections typically contained 0.5 g of luciferase reporter plasmid (pGal4(5X)luc or p15 INK4b luc) and 0.05-0.5 g of various effector plasmids (HCF-1, Miz-1, c-Myc, or derivatives thereof). Total DNA and promoter dosage was kept constant with the appropriate amounts of corresponding empty vectors. Luciferase activity was assayed in cell lysates prepared 48 h post-transfection and normalized to protein concentration.
Co-immunoprecipitation Assay-Transfections were carried out as described above using expression plasmids (0.5 g/well) for V5-Miz-1FL and/or FLAG-HCF-1FL. Lysates were prepared 48 h post-transfection using Nonidet P-40 lysis buffer (150 mM NaCl, 50 mM Tris (pH 8.0), 1% Nonidet P-40, and 0.2 mM phenylmethylsulfonyl fluoride). 1 ml of clarified supernatant, normalized for protein concentration, was precleared with 50 l of protein G-Sepharose (Roche Molecular Biochemicals) and incubated with 1 g of anti-FLAG antibody (Upstate Biotechnology, Inc.) and 50 l of protein G-Sepharose for 2 h at 4°C. Immune complexes were collected, extensively washed, suspended in 50 l of SDSpolyacrylamide gel sample buffer, and subjected to PAGE. Proteins were transferred to Hybond TM -C pure nitrocellulose membrane (Amersham Biosciences) and probed using mouse anti-V5 monoclonal antibody (Invitrogen) as the primary antibody, followed by horseradish peroxidase-coupled sheep anti-mouse polyclonal antibody (Amersham Biosciences) as the secondary antibody. Proteins were detected by enhanced chemiluminescence with a commercially available kit (ECL, Amersham Biosciences) according to the manufacturer's instructions.
Immunoblotting-COS-1 cells were transfected as described above, except that 1 g of the various Gal4-Miz-1 expression plasmids was used. Preparation of cell extracts for immunoblotting using Hybond TM -C pure nitrocellulose membrane was carried out as described (36,38). Gal4 fusion proteins were detected by ECL as described above using mouse anti-Gal4 DNA-binding domain monoclonal antibody (Santa Cruz Biotechnology) as the primary antibody and horseradish peroxidase-coupled sheep anti-mouse polyclonal antibody as the secondary antibody.

HCF-1 Interacts with Miz-1 via Determinants in the Basic
Domain-We carried out yeast two-hybrid screens to identify novel HCF-1-interacting proteins that may provide insights into the cellular functions, targets, and mechanisms of action of HCF-1. Similar approaches by others (15,16) have used the N-terminal domain (residues 1-450) of HCF-1 as bait because this region is known to be sufficient for interaction with VP16 and for promoting VIC formation (14). We focused our attention on a separate region of HCF-1 (residues 450 -1439) that encompasses the basic domain (Fig. 1A) because this region has been shown to be required, along with the HCF VIC domain, to rescue the temperature-sensitive block to cell cycle progression in tsBN67 cells (14). The rationale was that the basic region may target novel HCF-1-interacting proteins that subserve a role in cell cycle regulation and/or other functions of HCF-1 that are independent of the HCF VIC region.
Yeast two-hybrid screens using a Gal4-DBD-HCF-1-(450 -1439) fusion protein as bait resulted in the identification of several novel HCF-1-interacting proteins from a HeLa cell Gal4 AD cDNA library. Among the clones identified were isolates containing partial cDNAs encoding a zinc finger transcription factor called Miz-1, a recently described protein that was originally isolated by virtue of its ability to interact with the cellular oncoprotein c-Myc (27,28). Miz-1 is an 803-amino acid long protein that contains 13 zinc fingers of the C 2 H 2 class and an N-terminal POZ domain (see Fig. 3A). The longest Miz-1 cDNA clone that we isolated from the two-hybrid screen encoded residues 269 -803 of Miz-1 and is thus missing the Nterminal POZ domain. As shown in Fig. 1B, HCF-1 bound specifically to Miz-1, as ␤-galactosidase activity was observed only in the presence of both Gal4-DBD-HCF-1 bait and Gal4-AD-Miz-1 prey plasmids. Significantly, Miz-1 failed to interact with HCF-1 residues 1-380 (Fig. 1B), a domain that is sufficient for interaction with VP16 and the cellular factors LZIP and Zhangfei (14 -17).
To determine whether Miz-1 and HCF-1 associate in vivo in mammalian cells, we carried out co-immunoprecipitation experiments with cells cotransfected with expression vectors for V5-tagged full-length Miz-1 and FLAG-tagged full-length HCF-1. Cell extracts were immunoprecipitated with anti-FLAG antibody and probed with anti-V5 antibody. As shown in Fig. 1C, full-length Miz-1 was present in immune complexes precipitated with anti-FLAG antibody from cells that had been transfected with both FLAG-HCF-1FL and V5-Miz-1FL expression vectors, but not from control cells that were transfected with V5-Miz-1FL alone. Thus, Miz-1 and HCF-1 form a stable complex in vivo in mammalian cells.
To map more precisely the region in HCF-1 that is targeted by Miz-1, we constructed a series of deletions within Gal4-DBD-HCF-1-(450 -1439) and tested these for interaction by two-hybrid analysis with Gal4-AD-Miz-1-(269 -803). As shown in Fig. 2, a fusion protein containing HCF-1 residues 750 -836 was sufficient to bind to Miz-1. Residues between 836 and 1439, which include the HCF PRO repeats, were not necessary for interaction. However, the strongest binding, as reflected by ␤-galactosidase activity, was observed with HCF-1-(750 -902), suggesting that residues between 836 and 902 may contribute to more robust binding and/or stability of the Miz-1⅐HCF-1 complex in vivo. A fragment spanning residues 836 -902 was, however, insufficient to bind with Miz-1. Miz-1 bound directly to HCF-1, as determined by reciprocal GST pull-down assays with radiolabeled proteins synthesized in vitro (Fig. 3B). As shown in Fig. 3B, radiolabeled full-length Miz-1 bound to GST-HCF-1-(750 -902), but not to GST alone. Thus, Miz-1 targets part of the basic region of HCF-1 that is required for cell cycle progression (14), with amino acid residues between 750 and 836 being of particular importance for interaction.
The Miz-1 fragment isolated in the two-hybrid screen was missing N-terminal residues 1-269, which include the POZ domain (residues 1-108). This indicates that the POZ domain is not required for interaction with Miz-1, at least in the presence of the C-terminal interacting domain. However, because POZ domains have been shown to provide protein interaction interfaces (39 -41), we tested whether this region may bind independently to HCF-1. As shown in Fig. 3B, reciprocal GST binding experiments with Miz-1-(1-108) demonstrated that the POZ domain itself was capable of interacting with HCF-1.
Thus, as summarized in Fig. 3C, Miz-1 contains at least two modular subdomains that can independently bind to HCF-1. These regions are distinct from those that mediate Miz-1 interaction with its other known interacting partners. The foregoing does not preclude the possibility that subregions of Miz-1  2. Miz-1 targets the basic region of HCF-1. N-and C-terminal deletions of HCF-1 between residues 450 and 1439, as indicated, were cloned into the Gal4-DBD plasmid and tested for interaction with Miz-1 (residues 269 -803) by two-hybrid analysis in yeast. Specific ␤-galactosidase activity was determined by the liquid quantitative assay. The values shown represent the mean ␤-galactosidase (␤-GAL) activity Ϯ S.D. from three independent transformants assayed in duplicate. located between residues 108 and 637, which include the zinc finger cluster, are also important for or influence interaction with HCF-1.

HCF-1 Targets a Transactivation Function in Miz-1-Most of the HCF-1-interacting factors identified to date, including
Miz-1 described herein, are themselves transcription factors, underscoring the growing evidence that HCF-1 serves as a global transcriptional co-regulatory factor. Recent findings have shown that, in some cases, HCF-1 is recruited directly to the transcriptional activation domains of its interacting partners, including those present in LZIP (42) and GA-binding protein (19), suggesting that one role of HCF-1 may be to directly modulate the activation potential of its interacting partners.
Because a Miz-1 transcriptional activation domain had not been identified at the time, we sought to determine whether the HCF-1-binding interface at the C terminus of Miz-1 could serve as a transactivation domain when tethered to DNA. For this purpose, we tested a series of Gal4-Miz-1 fusion proteins by transient transfection assays in mammalian cells using a luciferase reporter gene driven by a promoter containing multiple upstream Gal4-binding sites. As shown in Fig. 4A, full-length Miz-1 (Gal4-Miz-1FL), as well as a derivative lacking the Nterminal POZ domain (Gal4-Miz-1⌬POZ), resulted in 13-and 31-fold induction of reporter gene activity, respectively, relative to the activity obtained with the Gal4 DBD alone. Thus, full-length Miz-1, as well as the ⌬POZ derivative, activates transcription when tethered directly to DNA.
To determine whether the transactivation function maps to regions required for interaction with HCF-1, we tested the activity of a Gal4 fusion with residues 637-803 of Miz-1. As shown in Fig. 4A, Gal4-Miz-1-(637-803) potently activated transcription, resulting in a 253-fold activation of reporter gene expression relative to the Gal4 DBD alone. Interestingly, transactivation by this derivative was significantly higher than that observed with the full-length protein or the ⌬POZ derivative, a difference that could not be attributed to differences in expression levels of the respective proteins as determined by Western blot analysis (Fig. 4B, lanes 1, 2, and 4, respectively). This suggests that the more robust activation is an inherent property of this isolated domain and that the full activation potential of this region may be masked in the context of the intact protein.
Overlapping subfragments spanning residues 637-718 and 701-803 were marginally active (4 -5-fold induction over the control), whereas a fragment containing residues 701-738 was inactive. These subfragments were also unable to bind to HCF-1. As a control, we also tested residues 176 -284, a region that encompasses binding determinants for c-Myc (see below) and that has been predicted to harbor an activation function because of its high content of acidic amino acids (28). This region was, however, unable to stimulate expression of the reporter gene.
Thus, the C-terminal HCF-1-interacting domain of Miz-1 harbors an autonomous activation function. The potency of this region in transactivation correlates with its ability to interact with HCF-1. These findings add to the growing evidence that HCF-1 targets activation domains in its interacting partners.
HCF-1 Represses Gal4-Miz-1-mediated Transcriptional Activation-To begin to probe the functional consequences of Miz-1/HCF-1 interaction, we carried out transient transfection assays with various Gal4-Miz-1 fusion proteins in the presence or absence of an expression vector for HCF-1 to determine whether HCF-1 modulates transactivation by Miz-1. As shown in Fig. 5A, cotransfection of full-length HCF-1 led to a dose-dependent repression of transactivation mediated by Gal4-Miz-1FL (Fig. 5A). Similarly, HCF-1 inhibited transactivation by a Miz-1 derivative lacking the POZ domain or a derivative containing only the C-terminal transactivation domain (residues 637-803). In each case, cotransfection of an HCF-1 expression vector led to ϳ70 -80% inhibition of activity at the highest concentrations of HCF-1 used. The repressive effect was specific to Gal4-Miz-1 derivatives because similar levels of HCF-1 had no effect on transactivation mediated by Gal4-VP16AAD, a Gal4 derivative linked to the potent acidic transactivation domain of VP16. Thus, the repression observed in Miz-1 transactivation is not due to a generalized titration of co-regulatory factors by HCF-1. The observation that HCF-1 repressed Gal4-Miz-1FL and Gal4-Miz-1⌬POZ to a similar degree suggests that inhibition by HCF-1 is independent of the POZ domain.
HCF-1 Represses Miz-1 Activation of the p15 INK4b Promoter-Having shown a repressive effect of HCF-1 on the transcriptional activity of Gal4-Miz-1 fusion proteins, we sought to determine whether a similar effect could be observed with native Miz-1 in the context of a natural Miz-1 target gene. Miz-1 has recently been shown to stimulate transcription of the cyclin-dependent kinase inhibitor p15 INK4b through direct binding to the initiator element in this promoter (30). To de-termine whether HCF-1 modulates Miz-1-mediated activation of p15 INK4b expression, we performed cotransfection assays with a luciferase reporter gene linked to the p15 INK4b promoter. As shown in Fig. 6, transfection of an expression vector for full-length Miz-1 led to a 7-fold induction of p15 INK4b reporter gene activity, consistent with recent findings reported by others (30). In contrast, HCF-1 expression had no effect on reporter gene activity. However, coexpression of HCF-1 with Miz-1 inhibited Miz-1 mediated transactivation by ϳ70%.
Miz-1 activation of p15 INK4b expression is also inhibited by interaction with c-Myc (30). Consistent with this, transfection of a c-Myc expression vector, although having no effect on its own, repressed Miz-1-dependent activation of the reporter gene (Fig. 6). The extent of repression by c-Myc was comparable to that observed with HCF-1. Interestingly, coexpression of both HCF-1 and c-Myc led to an additive effect on inhibition of Miz-1 activity, reducing activity to the basal levels observed with the reporter gene alone. Thus, HCF-1, like c-Myc, antagonizes Miz-1-dependent activation of the native p15 INK4b promoter.
HCF-1 Interferes with Recruitment of p300 to Miz-1-c-Myc has been shown to inhibit transactivation by Miz-1 by interfering with recruitment of the p300 coactivator (30). Because HCF-1 targets a region of Miz-1 that is also involved in p300 association, we speculated that HCF-1 might also function in this manner. To determine whether this is the case, we carried out a mammalian two-hybrid assay using Gal4-Miz-1FL and chimeric p300-VP16. As shown in Fig. 7A, transfection of cells with the p300-VP16 expression vector on its own had no effect on reporter gene activity. However, cotransfection expression vectors encoding p300-VP16 and Gal4-Miz-1 led to a 2-fold increase in reporter gene activity compared with cells transfected with Gal4-Miz-1 alone, similar to what has been previously observed (30). Inclusion of an expression vector encoding full-length HCF-1 abrogated the stimulatory effect of p300- VP16 and further repressed Gal4-Miz-1FL transactivation potential. To determine whether this is the result of HCF-1 interfering with the recruitment of p300 to Miz-1, we used the GST pull-down competition assay with a C-terminal fragment of p300 (residues 1572-237) that has been shown to interact with Miz-1 (30) fused to GST along with radiolabeled fulllength Miz-1 and unlabeled competitor proteins synthesized in vitro. As shown in Fig. 7B, full-length Miz-1 bound to GST-p300-(1572-2371) as expected. Inclusion of equivalent amounts of unlabeled HCF-1- (1-902), c-Myc, or Miz-1 (as a positive control) reduced binding of radiolabeled Miz-1 by 60 -70% (HCF-1 did not bind to GST-p300; data not shown). In contrast, inclusion of unlabeled luciferase (negative control) had no effect on binding. Thus, HCF-1 competes with p300 for interaction with Miz-1.
The Causative Mutation in HCF-1 That Leads to Cell Cycle Arrest in tsBN67 Cells Does Not Affect Binding to Miz-1-A subfragment of HCF-1 spanning residues 1-902, including the basic region that is targeted by Miz-1, is necessary and sufficient to rescue the temperature-sensitive cell cycle defect in tsBN67 cells (14). The causative mutation (P134S) in HCF-1 also abrogates interaction with cellular proteins such as LZIP FIG. 5. HCF-1 antagonizes transactivation by Gal4-Miz-1. A, COS-1 cells were cotransfected with 0.5 g of pGal4(5X)luc reporter plasmid and the indicated Gal4-Miz-1 fusion expression vector (50 ng) or Gal4-VP16AAD (50 ng) in the absence or presence of increasing amounts of an expression vector for full-length HCF-1 as indicated. Extracts were prepared 40 -48 h later and assayed for luciferase activity. Values shown represent the mean Ϯ S.D. from three independent transfections carried out in duplicate and are normalized to the value obtained from the respective Gal4 fusion effector plasmid alone, which was taken as 100% for each case. B, the minimal Miz-1-binding domain of HCF-1 was sufficient to inhibit transactivation by Gal4-Miz-1. Transfections and measurement of luciferase activity were carried out in an identical manner as described for A, except that the expression vector for HCF-1-(750 -902) was used in place of full-length HCF-1. (15,17); however, this is incidental to the ability of HCF-1 to promote cell cycle progression (24). As we have shown, the N-terminal domain of HCF-1 (residues 1-380) is not necessary for Miz-1 interaction in vitro or in vivo. Nevertheless, the possibility remains that, in the context of the native protein or larger subfragments of HCF-1, the P134S point mutation may induce long-range perturbations that could modulate Miz-1 binding and/or function. To test this directly, we used GST pull-down assays with GST-Miz-1-(637-803) and in vitro synthesized, radiolabeled, wild-type HCF-1-(1-902) or the corresponding HCF-1 fragment harboring the P134S point mutation. As expected, radiolabeled HCF-1-(1-902) bound to GST-Miz-1-(637-803), but not to GST alone (Fig. 8A). HCF-1-(1-902)(P134S) bound to GST-Miz-1-(637-803) as efficiently as did wild-type HCF-1. As a positive control, we tested binding to GST-VP16-(1-404), a GST fusion protein linked to residues 1-404 of VP16 that contains determinants necessary for interaction with HCF-1 (34). As expected, HCF-1-(1-902), but not HCF-1-(1-902)(P134S), bound to GST-VP16- .
To determine whether the P134S mutation affects the ability of HCF-1 to attenuate Miz-1-mediated transactivation, we carried out transient transfection experiments as described for Fig. 5. As shown in Fig. 8B, full-length HCF-1 harboring the P134S mutation retained the ability to inhibit transactivation by Miz-1. Thus, under these experimental conditions, the causative point mutation in HCF-1 that is responsible for the temperature-sensitive cell cycle defect in tsBN67 cells does not interfere with association with Miz-1 or inhibit transactivation by Miz-1. DISCUSSION HCF-1 is a complex, multifunctional protein that is essential for normal cell cycle progression through the G 0 /G 1 transition (6); however, its cellular roles and mechanisms of action remain poorly understood. Our findings that HCF-1 physically and functionally interacts with Miz-1, a cell cycle regulator that causes cell cycle arrest at G 1 (27,30), and antagonizes Miz-1-dependent activation of the gene encoding a key cell cycle regulatory protein provide a provocative new dimension to the activities of these two proteins that may be of consequence to their respective roles in transcription and their reciprocal effects on cell cycle progression.
Miz-1 targets determinants present in the basic region of HCF-1, a domain that, in conjunction with the N-terminal region, is able to rescue the cell cycle defect in tsBN67 cells (14). Whether Miz-1 is an important effector in this process remains to be determined. Herr and co-workers (14) demonstrated that the minimal fragment of HCF-1 capable of rescuing the cell cycle defect in tsBN67 cells is residues 1-902 and that further C-terminal deletion of this fragment to position 836 abrogates rescue. The minimal Miz-1-binding interface mapped to residues 750 -836, and although residues between 836 and 902 enhance binding with Miz-1, they are neither necessary nor sufficient for interaction. Moreover, the causative mutation (P134S) in HCF-1 leading to cell cycle arrest does not affect interaction with or transactivation by Miz-1. The foregoing implies that Miz-1 may not be necessary for HCF-1mediated cell cycle control and that the phenotypic basis of the P134S mutation may be related to as yet unidentified proteins that target the N-terminal and/or basic regions. However, this does not necessarily negate a role for Miz-1 because cell cycle control by HCF-1 is likely a highly complex process that involves the cooperative interplay with many effector targets that recognize both the N-terminal HCF VIC domain and the basic region. Recent studies have shown that the basic region of HCF-1 is targeted by several other transcription factors, including GA-binding protein (19) and Sp1 (20). More recently, we have identified a novel zinc finger transcription factor that, like Miz-1, binds to residues 750 -902 of HCF-1. 2 Thus, the basic region of HCF-1 is targeted by at least four distinct transcription factors. It will be of interest to unravel the combinatorial interplay of these factors with HCF-1 and their potential relationships to the function of HCF-1 in cell cycle regulation and/or transcriptional control.
HCF-1 targets two independent regions of Miz-1: the Nterminal POZ domain and the C-terminal region spanning residues 637-803. The latter region also harbors overlapping but distinguishable determinants for interaction with c-Myc and the coactivator p300, suggesting the possibility of functional interplay among these factors in their interactions with Miz-1 (27,30). Interestingly, region 637-803 functions as a potent autonomous transactivation domain, and preliminary analysis indicates that the determinants required for binding to HCF-1 and for transactivation are overlapping and/or intimately linked. Thus, as has been shown to be the case with several of its other interacting partners (19,42), HCF-1 targets a transactivation function in Miz-1. This finding underscores a potential general functional property of HCF-1 whereby it modulates the activity of its partner proteins by directly interacting with their respective transactivation domains (see below). In-terestingly, transcriptional activation mediated by the isolated C-terminal domain was significantly more robust vis-à -vis fulllength Miz-1 or the Miz-1⌬POZ derivative, suggesting that, in the context of the full-length protein, other regions of Miz-1 lying between residues 109 and 637 may serve to dampen the activation potential of the C-terminal transactivation domain. The full transcriptional activation potential of Miz-1 may thus be triggered only at times when it has been relieved of this attenuation.
Overexpression of Miz-1 results in G 1 arrest via a process mediated in part by induction of the gene for the cyclin-dependent kinase inhibitor p15 INK4b at G 1 (30), which leads in turn to a reduction in cyclin D1-associated kinase activity. c-Myc binds directly to Miz-1 and inhibits Miz-1 activation of p15 INK4b expression, consequently relieving cell cycle arrest and allowing progression through the cell cycle. Conversely, p15 INK4b expression is stimulated by transforming growth factor-␤, which activates SMAD proteins, which prevent recruitment of c-Myc to Miz-1 and which directly cooperate with Miz-1 in transactivating the p15 INK4b promoter (33). Thus, growth inhibition pathways stimulated by transforming growth factor-␤ and growth stimulation pathways induced by c-Myc converge through Miz-1 in the regulation of p15 INK4b gene transcription (32).
Functional analysis indicated that HCF-1 repressed transactivation by Miz-1 both on the natural p15 INK4b promoter and on artificial Gal4-responsive promoters in a manner that was associated with recruitment of p300 to Miz-1. This may point to a potential mechanism by which HCF-1 may stimulate cell cycle progression. Thus, as observed with c-Myc, inhibition of Miz-1 activation of p15 INK4b by HCF-1 may be expected to result in increased activity of cyclin D1/CDK4 and thus progression through the G 1 /S restriction point. Interestingly, repression of p15 INK4b expression was potentiated in the presence of both HCF-1 and c-Myc, perhaps reflecting an aggregate response due to the fact that both c-Myc and HCF-1 inhibit recruitment of p300 to Miz-1.
Interference with p300 recruitment is one of several possible mechanisms by which HCF-1 inhibits Miz-1 activity because HCF-1 also independently associates with the Miz-1 POZ domain. POZ domains are conserved protein/protein interaction motifs and, when found as part of transcription factors, usually function as repressive domains, in part through recruitment of corepressors such as N-CoR (nuclear receptor corepressor) and SMRT (silencing mediator for retinoid and thyroid hormone receptors) (43,44). The Miz-1 POZ domain is thought to be in a latent repressive state, and it has been postulated that c-Myc serves to convert the Miz-1 POZ domain into an active repressive domain (27). Our findings demonstrate that HCF-1, as well as the minimal Miz-1-interacting subregion (residues 750 -902), inhibits transactivation of both Gal4-Miz-1FL and Gal4-Miz-1⌬POZ fusion proteins to a similar degree, suggesting that the POZ domain is not involved in HCF-1-mediated repression. The relevance of HCF-1 interaction with the isolated Miz-1 POZ domain is unclear at present; however, it may be related to results of a recent report showing that Miz-1 is present in the cytoplasm in association with microtubules (31).
In the presence of drugs that induce expression of the low density lipoprotein receptor, Miz-1 translocates to the nucleus, where it binds to and activates transcription of the low density lipoprotein receptor gene. Translocation under these circumstances requires the integrity of the POZ domain, suggesting that the POZ domain may be required for nuclear import of Miz-1 under certain conditions. HCF-1 has been shown to serve as a nuclear import factor for VP16 (45) and thus could potentially play a similar role with Miz-1. Native HCF-1 is found 2 H. H. L. Wong, D. Piluso, and J. P. Capone, unpublished data. almost exclusively in the nucleus; however, discrete N-terminal subfragments of HCF-1 have been shown to accumulate in the cytoplasm at G 0 (46). It is interesting to speculate that these subfragments may modulate nuclear translocation of Miz-1, thereby potentiating the repressive effects on Miz-1 activity.
In summary, we have identified Miz-1 as a novel interacting partner for HCF-1 and illustrate an intriguing new pathway of regulation that potentially links the opposing effects of these two proteins on cell cycle control. Our findings also represent the first example of HCF-1 inhibiting transactivation of an associated transcription factor. Recently, a novel HCF-1-related protein called HCF-like protein-1 has been identified and shown to inhibit transactivation by LZIP (47). Thus, HCF-1 and its family members may function as both coactivators and corepressors of transcription. We believe that this dual nature can serve HCF-1 well in cell cycle control because HCF-1 could function as a coactivator of genes that are required for cell cycle progression while also functioning as a corepressor of genes (such as p15 INK4b ) that serve to inhibit cell cycle progression. Further studies to elucidate the physiological importance of Miz-1/HCF-1 interaction and functional interplay with other factors involved in cell cycle control and gene regulation will be of interest.