Human PLU-1 Has Transcriptional Repression Properties and Interacts with the Developmental Transcription Factors BF-1 and PAX9*

PLU-1 is a large (1544 amino acids) nuclear protein that is highly expressed in breast cancers and is proposed to function as a regulator of gene expression. A yeast two-hybrid screen using PLU-1 as bait has identified two unrelated PLU-1 interacting proteins, namely brain factor-1 (BF-1) and paired box 9 (PAX9), both of which are developmental transcription factors. BF-1 and PAX9 interact with PLU-1 via a novel conserved sequence motif (Ala- X -Ala-Ala- X -Val-Pro- X 4 -Val-Pro- X 8 - Pro, termed the VP motif), because deletion or site-di-rected mutagenesis of this motif in either protein abolishes PLU-1 interaction in vivo . In a reporter assay system, PLU-1 has potent transcriptional repression activity. BF-1 and PAX9 also represses transcription in the same assay, but co-expression of PLU-1 with BF-1 or PAX9 significantly enhances this repression. Mutation of the PLU-1 binding motifs in BF-1 and PAX9 abolishes the observed PLU-1 co-repression activity. These data support a role for PLU-1

The human PLU-1 gene was identified from a differential screen looking for genes regulated by c-ErbB2 signaling using the ce-1 cell line, a human epithelial cell line overexpressing c-ErbB2, and the antibody 4D5/Herceptin (an inhibitor of c-ErbB2 signaling) (1). PLU-1 encodes for a 1544 amino acid multidomain protein that is localized exclusively to the nucleus and is constitutively expressed in 90% of breast cancers (1,2). Since the initial identification of PLU-1, two independently reported splice variants coding for PLU-1, RBP2-H1 1 (3) and RBBP2H1A (4), have been confirmed (2). However, little is known of the function of PLU-1, although there are several well defined domains that provide some functional clues, including the dri (dead ringer)/ARID (AT-rich interactive domain) DNA binding domain (5,6), three plant homeodomain/leukemia-associated protein domains (7,8), five putative nuclear localization signals, and two putative nuclear binding motifs. In addition, there are also several novel, uncharacterized domains including an N-terminal Jumonji domain (9), a Trp/Tyr/Phe/ Cys domain (1), recently renamed the PLU-domain (10), and a small Cys/His-rich region, which, together with the PLU-domain, are conserved in several genes (1). Recent bioinformatic analyses have revealed a shorter sequence within the PLUdomain that is found in at least 110 different proteins and has been termed the C-terminal Jumonji domain (9,11).
The PLU-domain, together with the ARID DNA binding motif, has high sequence homology with several other proteins, across a variety of species including Rum1 (regulator Ustilago maydis 1) (10), RBP2 (human) (12), and LID (little imaginal discs; Drosophila melanogaster) (13). Several recent studies on some of these homologues have provided insights into potential PLU-1 functions. RBP2, for example, has been reported (14) to act as a co-activator of nuclear receptors by enhancing receptormediated transcription. The Drosophila protein LID defines a new class of homeobox proteins that may be involved in maintaining both transcriptionally active and inactive chromatin states (13), whereas the repression of a specific set of genes by Rum1 is crucial for sporulation in U. maydis (10). Thus the presence of these conserved domains in PLU-1 suggests that PLU-1 is involved in transcriptional regulation and that overexpression in breast cancers may relate to the transformation process (1). Continuing studies on endogenous PLU-1 have confirmed high levels of expression of the protein in breast cancers and breast cancer cell lines and restricted expression in normal adult tissues with the exception of testis, suggesting that PLU-1 could belong to the class of testis/cancer antigens (2).
To determine possible molecular functions for PLU-1, we have carried out a yeast two-hybrid screen using PLU-1 as bait and report the findings here. We have identified two unrelated human transcription factors that interact with PLU-1 in vivo, namely brain factor-1 (BF-1, also known as FOXG1b) and paired box 9 (PAX9). We show that PLU-1 acts as a specific transcriptional co-repressor of BF-1 and PAX9 in reporter gene assays and that PLU-1 co-repression activity is mediated by an interaction with BF-1 and PAX9 via a conserved sequence motif present in both transcription factors.
Yeast Two-hybrid Screens and Interaction Assays-The constructs pAS2-1 PLU-1 and pAS2-1 PLU-domain were used as bait to screen a human fetal brain library (Clontech) in a yeast two-hybrid screen according to the protocols of Clontech except that S. cerevisiae PJ69 -4A (15) was used. Colonies containing interacting proteins were initially identified by growth on minimal medium lacking histidine, tryptophan, leucine, and adenine. For analysis of ␤-galactosidase expression, colony filter lift assays (Clontech) were used. 170 clones were obtained using full-length PLU-1 as bait and 230 clones using the PLU-domain as bait. Three clones containing fragments of BF-1 (GenBank TM accession number NM_005249) and two clones containing fragments of PAX9 (Gen-Bank TM accession number NM_006194) were identified. All other clones were either unknown or commonly observed false positives such as collagen, NADH dehydrogenase, or ferritin. Interactions were initially confirmed in a yeast two-hybrid assay using full-length BF-1 and PAX9. Full-length RBP2 cloned into pAS2-1 was used to test whether RBP2 would interact with BF-1 or PAX9.
To confirm that the putative PLU-1 interaction motif was required for PAX9-BF-1 interaction, deletion constructs were made by PCR and verified by sequencing. Deletions of PAX9 (cDNA coding for amino acids 3-194 and 3-166) and BF-1 (cDNA coding for amino acids 2-404 and 2-383) in pACT2 were used with full-length PLU-1 and the PLUdomain in pAS2-1 in a yeast two-hybrid assay. Interactions were identified as above, and ␤-galactosidase activity was measured as described by Clontech using chlorophenol red-␤-D-galactopyranoside as the chromogenic substrate. The interaction of the pACT2 BF-1/PAX9 mutants with PLU-1 in pAS2-1 was tested as described above.
Tissue Culture-HEK293 and HT1080 cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum and 0.3 g/ml L-glutamine at 37°C with 5% CO 2 . 1-7HB2 cells were grown in the above stated culture medium with the addition of 5 g/ml hydrocortisone and 10 g/ml insulin.
Repression Assays-The reporter constructs were made by inserting the major late promoter of adenovirus (and also 5GAL4 binding sites in the case of the pGL3-Basic/GAL4bs/ad.prom. construct) from pGL5 luc (Promega) into the multiple cloning site of pGL3-Basic (Promega), using either the NheI/BstBI or KpnI/HindIII restriction sites. Constructs were verified by sequencing.
Varying amounts of pBIND PLU-1, wild-type, or mutant pBIND BF-1/PAX9 with reporter construct were transfected into 70 -90% confluent HEK293 cells. Wild-type and mutant pBIND constructs were also co-transfected with pcDNA3.1(Ϫ)/myc-HisA PLU-1 for co-repression studies. All transfections followed the LipofectAMINE (Invitrogen) protocol according to the manufacturer's instructions. Cells were harvested after 24 h, and luciferase assays were performed using the Dual luciferase reporter assay system (Promega). The empty pBIND vector was co-transfected with the reporter constructs as a control. The polycomb repressor HPC3 (17) was cloned into the pBIND vector, and the resulting GAL4DBD-HPC3 fusion was used as a positive control. Negative controls for these experiments included the independent co-transfection of pGL3-Basic/ad.prom. reporter construct and the inclusion of pcDNA3.1(Ϫ)/myc-HisA (for PLU-1 co-repression studies). Data were normalized by assaying Renilla luciferase activity (pBIND). Luciferase activity from empty pBIND vector was arbitrarily set to 100%. All other measurements are expressed relative to this value. All data shown are the average of at least three independent experiments.

RESULTS
Identification of BF-1 and PAX9 as PLU-1 Interacting Proteins-To identify potential PLU-1 interacting partners, a yeast two-hybrid interaction assay was carried out. A human fetal brain library was screened using either full-length PLU-1 or the shorter PLU-domain (amino acids 368 -643) as bait ( Fig.  1). A number of positive PLU-1 interacting clones were identified although many were false positives (see "Experimental Procedures"). However, three clones containing fragments of BF-1 (NM_005249) and two clones of PAX9 (NM_006194) were identified in both screens indicating that specific PLU-1 interacting clones had been isolated. PCR, cloning, and sequence verification of both full-length BF-1 and PAX9 allowed further yeast two-hybrid assays (Table I). Using both full-length PLU-1 and the PLU-domain, these data confirmed our initial observations, with ␤-galactosidase activities for BF-1 and PAX9 between 8 and 20% of the positive control (Table I). The interactions with the PLU-domain (residues 368 -643) appeared 2-fold stronger (Table I), although this could reflect the levels of expressed full-length PLU-1 in the yeast system. To investigate the specificity of the PLU-1-BF-1/PAX9 interactions, a close homologue of PLU-1, namely RBP2 (84% identical), was tested for interactions with BF-1 and PAX9 (Table I). Under identical experimental conditions, full-length RBP2 did not interact with BF-1 and PAX9, suggesting that the PLU-1-BF-1/PAX9 interactions in yeast are specific.
A Novel PLU-1 Interacting Motif in BF-1 and PAX9 -The BF-1 and PAX9 clones isolated from the yeast two-hybrid library both contained 3Ј sequences that coded for C-terminal regions within their respective protein sequences. Although a sequence alignment of the obtained BF-1 and PAX9 sequences showed no overall similarity, a small region of ϳ20 amino acids was identified in both sequences (amino acids 383-404 in BF-1; 168 -189 in PAX9) that contained a conserved motif (Ala-X-Ala-Ala-X-Val-Pro-X 4 -Val-Pro-X 8 -Pro). The motif, which we term the VP motif ( Fig. 2A), suggests a possible common PLU-1 interaction site in BF-1 and PAX9. To investigate whether the VP motif mediated the PLU-1 interaction, a series of deletion constructs were made in BF-1 and PAX9 that removed either sequences C-terminal to the motif or the motif itself (Fig. 2, B and C). These constructs were then tested for PLU-1 interaction in yeast (Table II). Deletion constructs that contained the motif (BF-1 2-404; PAX9 3-194) all interact with PLU-1 similar to full-length proteins (Table II). However, removal of the motif (BF-1 2-383; PAX9 3-166) abrogated the PLU-1 interaction (Table II). To confirm the validity of the PLU-1 interacting motif in the context of full-length BF-1/PAX9, site-directed mutagenesis of each of the conserved residues was carried out ( Fig.  2A). Mutation of either Val-Pro (to Gly-Ala) or single Pro (to Ala) completely abolishes PLU-1 interaction (Table II), establishing that both the conserved residues and structural integrity of the BF-1/PAX9 motif are required for PLU-1 interaction.
BF-1 and PAX9 Interact with PLU-1 in Vivo-To confirm that BF-1 and PAX9 interact with PLU-1 in vivo, a series of co-immunoprecipitation (co-IP) experiments were carried out using either tagged constructs or cell lines expressing endogenous components. Fig. 3A shows the co-IP results of transiently expressed GAL4DBD-BF-1 and PLU-1-myc in HT1080 cells.
PLU-1 Has Transcriptional Repression Properties-Our PLU-1 interaction data establish that PLU-1 can interact in vivo with a subset of transcription factors through a novel sequence motif. We therefore wanted to test whether PLU-1 alone has an effect on transcription by utilizing a luciferase reporter gene assay in mammalian cells (Fig. 4A). A GAL4DBD-PLU-1 fusion construct was tested, and even at sub-picomolar amounts of transfected plasmid, PLU-1 dramatically reduces the basal expression of luciferase (Fig. 4B). To assess the significance of this repression effect, we carried out a similar experiment with the unrelated but established transcriptional repressor HPC3, a component of the Polycomb repression complex (17). The results reveal a similar trend of repression activity for both PLU-1 and HPC3 (Fig. 4, B and C), illustrating the transcriptional repression potency of PLU-1 under these experimental conditions.
Effect of PLU-1 Co-expression on the Repression Effect of BF-1 and PAX9 -It has been demonstrated previously (19,20) that both BF-1 and PAX9 have transcriptional repression properties in reporter gene assays. To test the effects of PLU-1 on BF-1/PAX9 repression activity, co-transfection experiments of GAL4DBD-BF-1/PAX9 and PLU-1-myc were performed (Fig. 5).
Expression of BF-1 shows a marked decrease in luciferase activity (ϳ3-fold) compared with empty vector (Fig. 5A) in agreement with previous studies. However, co-expression with PLU-1-myc (equimolar amounts) reproducibly shows a greater repression activity (ϳ25%) than BF-1 alone (Fig. 5A, compare white with striped black columns). As controls, PLU-1-myc was co-expressed with GAL4DBD ( Fig. 5A; striped gray column), and GAL4DBD (gray column) or GAL4DBD-BF-1 (white columns) were co-expressed with empty myc vector. These results clearly show that PLU-1 can act as a co-repressor of BF-1 under these experimental conditions.
The repression activity of PAX9 in these assays is less marked than with BF-1 (Fig. 5, A and B, compare white col-  Ϫ The pVA3 (SV40 large T antigen) and pTD-1 p53 plasmids from CLONTECH were used as a positive control. The ␤-galactosidase activity for the positive control was arbitrarily set at 100%. ND, not determined. umns); however, there is a consistent and reproducible decrease in reporter expression with increasing amounts of PAX9 (Fig. 5B, compare white with gray column). Co-expression of equimolar amounts of GAL4DBD-PAX9 with control plasmid shows an ϳ2-fold decrease in luciferase activity (0.25 pmol; see Fig. 5B). Similar to BF-1, co-expression with PLU-1-myc shows an enhanced repression activity for GAL4DBD-PAX9 (Fig. 5B, compare white with striped black columns). The same controls were performed as those for the BF-1 experiments. Together these data show that PLU-1 can act as a co-repressor of both BF-1 and PAX9 under these experimental conditions. PLU-1 Expression Does Not Affect the Repression Activity of HPC3-HPC3 is a known human polycomb group protein with established transcriptional repression activity in reporter gene assays (17). We therefore used HPC3 as a control in the PLU-1 co-expression measurements. Experiments were performed as

TABLE II The interaction between PLU-1 and BF-1/PAX9 is dependent on conserved amino acids in the VP motif
No interaction is observed between PLU-1 and BF-1/PAX9 in the absence of the VP motif or if the conserved residues within the motif are mutated. Ϫ ND BF-1 (2-404) V394G, P395A Ϫ Ϫ ND a The pVA3 (SV40 large T antigen) and pTD-1 p53 plasmids from CLONTECH were used as a positive control. The ␤-galactosidase activity for the positive control was arbitrarily set at 100%. ND, not determined.
for BF-1 and PAX9 but using a GAL4DBD-HPC3 construct. HPC3 repression activity is shown in Fig. 5C (solid white  columns) and is similar to other experiments (compare with Fig. 4C). However, unlike BF-1 and PAX9, co-expression of PLU-1-myc has no affect on HPC3 repression activity (Fig. 5C, compare white and striped black columns). The same controls were performed as those for the BF-1 and PAX9 experiments. Together, these data show that the co-repression activity of PLU-1 is specific to BF-1 and PAX9. (Table II) show that deletion and/or mutation of the VP motif in both BF-1 and PAX9 abrogates PLU-1 interaction. To further test the specificity of the VP motif for PLU-1 interaction in vivo, a series of co-IPs using GAL4DBD-BF-1 VP mutants and PLU-1-myc were carried out in HEK293 cells (Fig. 6). Co-expressions of PLU-1-myc with wild-type BF-1 and BF-1 VP mutants are shown in Fig. 6i (lanes 2-5, respectively). The results of co-IPs using the 9E10 myc antibody are shown in Fig. 6ii with only the wild-type BF-1 pulled down by PLU -1-myc (lane 7). These data support the yeast interaction experiments and provide compelling evidence that the interaction between PLU-1 and BF-1 in vivo is mediated specifically by the VP motif. Similar data was also obtained for PAX9 (data not shown).

PLU-1 Interaction with BF-1 and PAX9 Is Mediated by a Novel VP Motif-Our yeast interaction experiments
Mutation of Conserved Residues in the VP Motif Abolishes Co-repression by PLU-1-To test the functional relevance of the VP motif, further repression assays were performed. The following BF-1 VP mutants V388G,P389A, V394G,P395A, and P404A were tested either alone or co-expressed with PLU-1myc in luciferase reporter gene assays (Fig. 7A). Because both mutant and wild-type BF-1 have similar luciferase repression activities (data not shown), all the data have been normalized to wild-type BF-1 (Fig. 7A, compare gray columns with white column). Co-expression of PLU-1-myc only results in decreased reporter activity for wild-type BF-1 (WT; compare white column with black striped column), whereas the activities of the BF-1 VP mutants are unaffected (compare gray with gray/black striped columns). The same controls as described previously were used. These data show that under these experimental conditions, the VP motif mediates the PLU-1 co-repression activity.
A similar set of VP mutants for PAX9 was also tested (V173G,P174A, V179G,P180A, and P189A) (Fig. 7B). Similar to BF-1, only co-expression of PLU-1-myc with wild-type PAX9 results in decreased luciferase activity (WT; compare white column with black striped column). Repression activities for the PAX9 VP mutants are unaffected by PLU-1-myc co-expression (Fig. 7B, compare gray with gray/black striped columns). The same controls as described previously were used. Together these data provide compelling evidence that the PLU-1 co-repression activity of BF-1 and PAX9 is mediated by the VP interaction motif. DISCUSSION PLU-1 was first identified as being significantly overexpressed in breast cancers (1). Given that PLU-1 is a nuclear   FIG. 4. Transcriptional repression properties of PLU-1. A, schematic representation of the pGL3 reporter plasmid used in all the repression assays. This plasmid contains 5GAL4 DNA-binding sites and the major adenovirus late promoter regulating the expression of luciferase. B, co-transfection of pGL3 reporter plasmid with increasing amounts of GAL4DBD-PLU-1 (white columns). All values are normalized to the control (gray column). C, co-transfection pGL3 reporter plasmid with increasing amounts of GAL4DBD-HPC3 (white columns). All values are normalized to the control (gray column). HPC3 is a known transcriptional repressor and acts as a positive control in these assays.
FIG. 5. Transcriptional co-repression of BF-1 and PAX9 by PLU-1. The ability of PLU-1 to alter the repression activities of BF-1, PAX9, and HPC3 in luciferase reporter assays are shown. HEK293 cells were transfected with the indicated plasmids. A, repression activity of increasing amounts of GAL4DBD-BF-1 (white columns) compared with co-expression with PLU-1-myc (black striped columns). All values were normalized against the empty GAL4DBD and empty myc vectors (gray column). Co-expression of PLU-1-myc is also shown (gray striped column). B, repression activity of increasing amounts of GAL4DBD-PAX9 (white columns) compared with co-expression with PLU-1-myc (black striped columns). All values were normalized against the empty GAL4DBD and empty myc vectors (gray column). Co-expression of PLU-1-myc is also shown (gray striped column). C, repression activity of increasing amounts of GAL4DBD-HPC3 (white columns) compared with co-expression with PLU-1-myc (black striped columns). All values were normalized against the empty GAL4DBD and empty myc vectors (gray column). Co-expression of PLU-1-myc is also shown (gray striped column). Unlike BF-1 and PAX9, the expression of PLU-1-myc has no effect on the observed HPC3 repression activity under these experimental conditions. protein and contains a number of known domains found in other transcriptional regulators, we also proposed that PLU-1 plays a role in the regulation of gene expression. In this study, we have further characterized the function of PLU-1 by identifying potential interacting partner proteins and have explored the functional significance of these interactions by using a transcriptional reporter assay.
PLU-1 Interacts with Two Developmental Transcription Factors via a Novel Sequence Motif-From a yeast two-hybrid screen, we identified two specific PLU-1 interacting proteins, namely human BF-1 (also known as FOXG1b) and human PAX9 and showed that both interactions can occur in vivo by co-immunoprecipitation experiments. The two identified PLU-1 interacting proteins are unrelated DNA-binding transcription factors; BF-1 belongs to the forkhead family whereas PAX9 belongs to the paired box family. Interestingly, the murine homologues of BF-1 (Bf-1) and PAX9 (Pax9) are critical for the development of several structures in the mouse embryo. Bf-1 is required for the development of the brain and parts of the optic tract and is involved in neural cell proliferation (21)(22)(23)(24) whereas Pax9 is required for the development of some craniofacial features, limbs, teeth, and thymus (25)(26)(27)(28). The significance of the PLU-1-PAX9/BF-1 interactions are further underlined by the recent studies on murine Plu-1 (29). During mouse embryogenesis, expression of Plu-1 is found to overlap both temporally and spatially with Pax9 and Bf-1 (29) suggesting that any Plu-1-Pax9/Bf-1 protein interaction is of functional importance during early embryonic development. In addition to this data, preliminary comparative studies of gene expression levels in breast cancers versus normal lactating mammary gland have revealed high expression levels of PLU-1 and PAX9 in the tumor and lymph node containing malignant breast cells (data not shown). Together these data suggest that any Plu-1-Bf-1/Pax9 protein interaction is of functional importance during embryogenesis and that any PLU-1-PAX9 interactions could play an important role in the development of the breast cancer malignancy.
The interactions between PLU-1 and BF-1/PAX9 are mediated by a novel conserved PLU-1 interacting motif (Ala-X-Ala-Ala-X-Val-Pro-X 4 -Val-Pro-X 8 -Pro; termed the VP motif) that is found in both transcription factors. The motif is found C-terminal to the defined DNA binding domains of both BF-1 and PAX9 and constitutes the only sequence similarity between both proteins. Database searches of SwissProt and TrEMBL reveals that the motif is primarily confined to BF-1 or PAX9 homologues. Our data using the closest human PLU-1 homologue RBP2 suggests that the motif is specific for PLU-1 interaction, because both PAX9 and BF-1 do not interact with RBP2. It is possible, however, that variations of the VP motif may be required for interactions with closely related PLU-1 homo-  Fig. 2. A, repression activities of wild-type GAL4DBD-BF-1 (white column) and VP mutants (gray columns) compared with co-expression with PLU-1-myc (black and gray/black striped columns, respectively). All values are normalized against the wild-type GAL4DBD-BF-1 (white column). B, repression activities of wild-type GAL4DBD-PAX9 (white column) and VP mutants (gray columns) compared with co-expression with PLU-1-myc (black and gray/ black striped columns, respectively). All values are normalized against the wild-type GAL4DBD-PAX9 (white column). In all cases mutation of the VP motif abolishes the observed PLU-1 co-repression activity. logues in different species or even tissues. Indeed zebrafish Pax9 contains the motif with a conservative substitution, Val to Met, in the second VP, which may mediate binding to zebrafish PLU-1. Studies are underway to further investigate the specificity and structural basis of the PLU-1 interacting motif.
PLU-1 Has Transcriptional Co-repression Properties Mediated by a Novel Motif-Using a well defined reporter assay system, our data clearly show that PLU-1 has significant transcriptional repression properties, similar in effect to HPC3, a known transcriptional repressor and member of the polycomb group of proteins involved in maintaining gene silencing (17). Our data also confirm that both BF-1 and PAX9 have transcriptional repression activities in agreement with previous studies (19,20). Because we have clearly shown that the repression activity by BF-1 and PAX9 is enhanced by co-expression of PLU-1 it is plausible that PLU-1 acts as a co-repressor of BF-1/PAX9 in regulating targeted gene expression in vivo. Our data clearly show that a novel PLU-1 interacting sequence, termed the VP motif, found in both BF-1 and PAX9 mediates PLU-1 co-repression activity. Single amino acids changes within this motif are sufficient to abrogate PLU-1 interaction and corepression activity indicating that the interaction is highly specific. Together these data support a role for PLU-1 in regulating both BF-1 and PAX9 transcriptional activity in vivo. It will now be important to identify the target genes of BF-1 and PAX9 and elucidate the role of PLU-1 in regulating these genes.
Functional Implications of a PLU-1-BF-1/PAX9 Interaction-Previous studies on the avian sarcoma virus 31 oncoprotein Qin (the viral homologue of BF-1) showed that Qin binds the same DNA consensus sequence as BF-1 and represses gene transcription (30). The major transcriptional repression domain maps to the C-terminal region of Qin (30), where interestingly Qin also contains the VP motif described in our study. This transcriptional repression activity directly correlates with the oncogenic transformation potential of Qin (31). Further studies have shown that Qin transformation potential does indeed require an intact C-terminal region of the protein (32), of particular significance, because it is proximal (by eight residues) to the VP motif. It is tempting to speculate that the oncogenic potential of Qin has dependence on a PLU-1 interaction and that loss of such an interaction prevents transformation. In considering which genes might be affected by BF-1, Qin, or PAX9, it is interesting to note that the promoter of the human cyclin-dependent kinase inhibitor p27 Kip1 contains an exact sequence match to the Qin binding site (31), and a subsequent report has shown that p27 Kip1 is regulated by a forkhead transcription factor in response to interleukin (33). Further evidence of a link is that XBF-1 is known to affect p27 XIC1 expression in a dose-dependent manner (34). BF-1 also inhibits TGF-␤-mediated growth inhibition and transcriptional activation by associating with SMAD proteins (19,35). There are currently no genes identified known to be directly controlled by PAX9, but a consensus DNA binding sequence is known (20).
Another potential functional connection is the recent observation (36) that BF-1 is part of the groucho repression complex. Groucho proteins are recruited via multiprotein complexes to specific regions of the nucleus and act as potent co-repressors of transcription by establishing a repressive chromatin structure (36 -41). PAX9 has not been shown to bind groucho proteins directly but perhaps significantly contains an octapeptide motif required for transcriptional activity, and which, in other PAX proteins, has been shown to interact with the groucho-related protein GrG4 (42). Although there is at present no evidence to indicate that PLU-1 is involved in groucho-mediated transcriptional repression, it is tempting to speculate that through interactions with BF-1 and/or PAX9, PLU-1 influences groucho-mediated repression. Interestingly, the PAX9 octapeptide is within six amino acids of the VP motif, so the interaction between PAX9 and PLU-1 would most likely exclude a PAX9groucho protein interaction. It is possible that PLU-1 could therefore compete for PAX9 binding. Because both BF-1 and PAX proteins interact with members of the groucho family of corepressors it is plausible that PLU-1 has a role in either competing with groucho proteins for PAX9 or BF-1 binding or is involved directly in groucho-mediated transcriptional repression.