Oct-1 Counteracts Autoinhibition of Runx2 DNA Binding To Form a Novel Runx2/Oct-1 Complex on the Promoter of the Mammary Gland-Specific Gene β-casein

ImmunoblottingNuclear extracts were prepared as previously described (25). Equal amounts of nuclear extracts were electrophoresed on a sodium dodecyl sulfate (SDS)-12% polyacrylamide gel and then transferred to a nitrocellulose membrane. The membrane was incubated with either a polyclonal anti-Runx2 antibody (Oncogene Research Products) or a mouse antihemagglutinin (HA) antibody for the detection of HA-Oct-1; the secondary antibody used was goat anti-rabbit antibody (BD Biosciences Pharmingen) or goat anti-mouse antibody (Transduction Laboratories), respectively. Immunocomplexes were detected by using Supersignal West Dura extended-duration substrate (Pierce) and visualized by using a Bio-Rad Fluor-S multi-imager.

EMSAs.Oligonucleotides were radiolabeled with [α-³²P]dCTP by using the Klenow fragment according to standard protocols (50). The following oligonucleotide sequences were used to investigate the binding of Runx2 and Oct-1 to the β-casein promoter: Runx, 5′-CTAGAACTGACCGCAGCTGGCCGT-3′ (7); Runxβ-casein, 5′-CTAGTTACAAACCACAAAATTA-3′; MutRunxβ-casein, 5′-CTAGTTACAAAGGAGAAAATTA-3′; octamer, 5′-GATCCTTAATAATTTGCATACCCTCA-3′; and Runx/octamer+10 bp, 5′-CATGACAAACCACAATGACTTACGAAATTAGCATGTCA-3′. The sequences of the other oligonucleotides used in electrophoretic mobility shift assays (EMSAs) are denoted in the figures. EMSAs were performed as previously described (25).

Plasmid constructs.The β-casein promoter was amplified from genomic DNA derived from HC11 cells by nested PCR with the following oligonucleotides. First-round amplification primers were mouse β-casein P1 (5′-ATCTTTGATGTTTCCATATC-3′) and mouse β-casein P2 (5′-CTTTAGTGGAGGACAAGAGA-3′), and second-round amplification primers were mouse β-casein P3 (5′-GATCGAGCTCTGCTTCATAACTGAGGTTAAAG-3′) and mouse β-casein P4 (5′-GATCAAGCTTCTGAAAGGATGATGTCCTATC-3′) (66). The resulting 360-bp fragment was ligated to the SacI and HindIII sites of the pGL3-Basic firefly luciferase reporter vector (Promega). The Runx-binding site in the β-casein promoter was specifically mutated by using a QuikChange mutagenesis kit (Stratagene) according to the manufacturer's protocol with the oligonucleotides mousePMTOP (5′-GAGTATCTTACAAAGGAGAAAATTAGCATGTC-3′) and mousePMBOT (5′-GACATGCTAATTTTCTCCTTTGTAAGATACTC-3′). The octamer-binding site in the β-casein promoter was specifically mutated by using oligonucleotides β-caseinOctMut P1 (5′-CAAACCACAAAATTCGCCGGTCATTAAGTG-3′) and β-caseinOctMut P2 (5′-CACTTAATGACCGGCGAATTTTGTGGTTTG-3′).

The pRunx2i expression vector was previously described (25). Runx2 and AML-1/ETO were expressed by using pCMV-OSF2 and pCMVAML-1/ETO, respectively (16, 38). Oct-1 and POU were expressed by using pCDNA3-HA-Oct-1 and pCDNA3-HA-POU, respectively (55). The Runx2 deletions were created by PCR with pRK-Flag-Cbfa1 as a template. Details of the oligonucleotides used can be provided on request (1). The PCR products were ligated to the EcoRI and XbaI sites of pRK5. The integrity of the plasmid sequences was confirmed by sequencing.

Protein production.Wild-type and truncated human Oct-1-glutathione S-transferase (GST) fusion proteins were created by PCR with pCDNA3-HA-Oct-1 as a template. The products were ligated to the XhoI and XbaI sites of pGEX-KG. Proteins were expressed by using a TNT kit with SP6 polymerase according to the manufacturer's instructions (Promega). GST protein purification and pull-down assays were performed as described previously (56). The Runt domain was expressed in Escherichia coli as a GST fusion protein from plasmid pGEX-AML-1Runt and purified as described previously (7, 67a).

Cell cultures and transfections.HC11 and UMR106 cells were maintained as previously described (25). HC11 cell lines were transfected in either 6- or 24-well plates by using Lipofectamine 2000 transfection reagent (Invitrogen) and a total of 2 or 0.8 μg of DNA, respectively, as previously described (25). Transfection of HC11 cells that were cultured in the presence of lactogenic hormones was performed as follows. HC11 cells (1.5 × 10⁵) were placed in a 24-well dish. After 16 h, RPMI medium containing 10% fetal bovine serum (FBS), gentamicin (50 μg/ml), bovine insulin (5 μg/ml), prolactin (5 μg/ml), and dexamethasone (0.1 μM) was added to the cells. After 24 h, the cells were transfected with a total of 800 ng of DNA containing 360 ng of the reporter plasmid being investigated and 40 ng of pRLSV40. At 5 h after transfection, the transfection mixture was removed and replaced with RPMI medium containing 10% FBS, gentamicin (50 μg/ml), bovine insulin (5 μg/ml), prolactin (5 μg/ml), and dexamethasone (0.1 μM). At 54 h after transfection, the cells were lysed, and luciferase activity was determined by using a dual-luciferase reporter assay (Promega).

Transfection of HC11 cells with pRunx2i was performed as follows. HC11 cells (3 × 10⁵) were placed in a 24-well dish and cultured with RPMI medium containing 10% FBS, gentamicin (50 μg/ml), bovine insulin (5 μg/ml), prolactin (5 μg/ml), and dexamethasone (0.1 μM). After 24 h, the cells were transfected with a total of 800 ng of either pSUPER or pRunx2i (8, 25). At 5 h after transfection, the transfection mixture was removed and replaced with RPMI medium containing 10% FBS, gentamicin (50 μg/ml), bovine insulin (5 μg/ml), prolactin (5 μg/ml), and dexamethasone (0.1 μM). At 48 h after transfection, the cells were transfected for a second time with a total of 1 μg of DNA. Each transfection mixture contained 360 ng of the reporter plasmid being investigated, 40 ng of pRLSV40, and 600 ng of either pSUPER or pRunx2i. The cells were lysed at 48 h after the second transfection, and luciferase activity was determined.

All transfections were performed in triplicate, and data are presented as means and standard deviations (SDs). Values are relative to the luciferase activity of pGL3-Basic.

To obtain nuclear extracts from cells that had been transfected with either pSUPER or pRunx2i, the first transfection of the HC11 cells was the same as that described above. At 48 h after transfection, the cells were transfected for a second time with a total of 1 μg of DNA. Each transfection mixture contained 400 ng of pGFP and 600 ng of either pSUPER or pRunx2i. At 24 h after the second transfection, the cells were sorted by fluorescence-activated cell sorting; viable green fluorescent protein-expressing cells were collected and used to prepare nuclear extracts.

ChIP assays.Chromatin immunoprecipitation (ChIP) assays were performed by using a ChIP assay kit (Upstate Biotechnology) according to the manufacturer's protocol, except that formaldehyde cross-linking was performed for 30 min. Complexes were immunoprecipitated with 0.8 μg of anti-Runx2 antibody (Oncogene), anti-Oct-1 antibody (Santa Cruz), or anti-Flag antibody (Sigma). The sections of the β-casein promoter containing the Runx- and octamer-binding sites were detected by PCR with the primers mouse β-casein P5 (5′-TAGAATTTCTTGGGAAAGAC-3′) and mouse β-casein P2 (see above).

Runx2 is required for the activation of the β-casein promoter.It was previously established that mutations in block C abolish the activation of the β-casein promoter by lactogenic hormones (49). These mutations centered on two adjacent sites (Fig. 1A). One of these sites was identified as an octamer element that can be bound by Oct-1, but the identity of the factor binding to the adjacent site was not established. When we visually examined block C, we identified the site adjacent to the octamer element as a consensus Runx-binding site (Fig. 1A). To determine whether this sequence could indeed recruit Runx transcription factors, a series of EMSAs were performed with a double-stranded radiolabeled oligonucleotide that encompassed the Runx consensus sequence from the β-casein promoter (Runxβ-casein). When this sequence was incubated with a purified recombinant Runx DNA-binding domain (Runt), a complex was retarded in the gel (Fig. 1A, lane 2). This complex was not observed in the presence of a 100-fold molar excess of unlabeled wild-type DNA and was almost completely abolished by incubation with a nonradiolabeled Runx oligonucleotide that contained a known Runx-binding site (Fig. 1A, lane 4). In contrast, competition with the β-casein oligonucleotide that contained a mutation in the Runx-binding site (MutRunxβ-casein) had no effect on the formation of the retarded complex (Fig. 1A, lane 5). The results formally demonstrate that the β-casein promoter contains a bona fide Runx-binding site. Block C is therefore a composite Runx/octamer element. A database search for Runx/octamer elements of the type found in the β-casein promoter demonstrated that it is present in the promoters of β-, α-, and γ-caseins from a wide range of species (Fig. 1B).

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG. 1.

The β-casein promoter contains conserved Runx and octamer consensus sequences. (A) (Upper panel) Diagrammatic representation of the β-casein promoter and the three highly conserved regions, blocks A, B, and C. (Lower panel) EMSA demonstrating specific DNA binding of the purified Runt domain to the Runx-binding site in the β-casein promoter. The Runt domain was incubated with radiolabeled oligonucleotide Runxβ-casein in the presence or absence of a 100-fold molar excess of competitor DNA, as indicated above the lanes. The Runx competitor is a known Runx-binding site. (B) Sequence alignment of the Runx/octamer elements in the promoters of β-, α-, and γ-caseins from several species. (C) AML-1/ETO represses β-casein promoter activity. HC11 cells were transfected with 360 ng of either β-caseinWt or β-caseinMRunx in the presence or absence of 400 ng of a plasmid expressing AML-1/ETO. Cells were cultured in the presence of dexamethasone, insulin, and prolactin. (D) EMSA demonstrating that endogenous Runx2 from HC11 cells binds to the Runx2-binding site in the β-casein promoter. HC11 nuclear extracts were incubated with radiolabeled oligonucleotide Runxβ-casein in the presence of Runx-specific antibodies (Ab). The Runx2 complex and the supershifted complex are indicated. (E) Runx2 RNAi inhibits β-casein promoter activity. HC11 cells were transfected with 360 ng of either β-caseinWt or β-caseinMRunx in the presence or absence of pRunx2i. In the absence of pRunx2i, HC11 cells were transfected with the parent plasmid, pSUPER. Cells were cultured in the presence of dexamethasone, insulin, and prolactin. (F) EMSA with nuclear extracts from HC11 cells transfected with either pSUPER or pRunx2i. Nuclear extracts were incubated with a radiolabeled oligonucleotide containing a Runx-binding site in the presence or absence of anti-Runx2 antibodies. Nuclear extracts from osteoblast-like cell line UMR106 were used as a positive control for the anti-Runx2 antibodies (lanes 5 and 6). All transfections were performed in triplicate; luciferase activities are presented as means and SDs. All values are relative to the activity of the pGL3-Basic reporter.

We next investigated the effect of a dominant-negative Runx protein on the transcriptional activity of the β-casein promoter. Transient transfections were performed with mammary epithelial cell line HC11, which maintain the ability to express β-casein when cultured in the presence of lactogenic hormones (2). HC11 cells were transfected with either wild-type or mutant Runx-binding site β-casein reporters and a plasmid expressing the dominant-negative Runx protein AML-1/ETO (Fig. 1C). The activity of the wild-type β-casein reporter decreased by more than 50% in the presence of AML-1/ETO (Fig. 1C). Mutation of the Runx-binding site reduced the activity of the promoter by more than 75%, and AML-1/ETO had no effect on this mutant reporter (Fig. 1C). The effect of mutating the Runx-binding site is in agreement with the previous finding that the Runx sequence in block C is essential for the activation of the β-casein promoter in mammary epithelial cells (49). In addition, the ability of AML-1/ETO to suppress the activity of the β-casein promoter suggests that the Runx-binding site can recruit Runx factors in vivo.

To determine which Runx factor(s) binds to the β-casein promoter, nuclear extracts were incubated with the Runx-binding site in the presence of antibodies specific for individual Runx factors (Fig. 1D). Nuclear extracts produced two bands when analyzed by EMSAs. In the presence of anti-Runx2 antibodies, the lower of the two bands was abolished and a supershifted complex was formed (Fig. 1D, lane 3). The Runx1- and Runx3-specific antibodies did not result in a specific supershift The data demonstrate that endogenously expressed Runx2 is able to bind to the Runx-binding site in the β-casein promoter.

To establish whether endogenous Runx2 is required for the activity of the β-casein promoter, cells were transfected with either the wild-type β-casein reporter or the Runx-binding site mutant reporter in the presence of pRunx2i, which expresses Runx2-specific small interfering RNA. When cotransfected with pRunx2i, the activity of the wild-type reporter plasmid was reduced by more than 50% (Fig. 1E). In contrast, pRunx2i had no effect on the activity of the Runx-binding site mutant reporter plasmid. The amount of Runx2 was significantly reduced in cells transfected with pRunx2i, as determined by supershift analysis with Runx2-specific antibodies (Fig. 1F). Taken together, the data demonstrate that both the intact Runx-binding site and Runx2 are required for full transcriptional activation of the β-casein promoter.

Runx2 and Oct-1 form a complex on the Runx/octamer element.Having established that block C is a composite Runx/octamer element and that Runx2 is required for the activation of the β-casein promoter, we next sought to determine whether a Runx2/Oct-1 complex forms on the β-casein promoter. The Runx/octamer element was incubated with nuclear extracts in the presence or absence of competitor DNA (Fig. 2). In the absence of competitor DNA, three complexes were observed: C1, C2, and C3 (Fig. 2B, lane 1). Complexes C1 and C2 were abolished in the presence of unlabeled octamer DNA, indicating that these complexes contain an octamer-binding protein (Fig. 2B, lane 2). When the Runx sequence was used as a competitor, complexes C1 and C3 were abolished, indicating that both of these complexes contain a Runx protein (Fig. 2B, lane 3). All three complexes were absent in the presence of unlabeled Runx/octamer DNA (Fig. 2B, lane 4). This specific pattern of competition indicates that complex C2 contains an octamer-binding protein, C3 contains a Runx factor, and C1 contains both an octamer-binding protein and a Runx factor.

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG. 2.

Endogenously expressed Runx2 and Oct-1 bind to the β-casein promoter. (A) Sequences of the β-casein Runx/octamer element and the known Runx- and octamer-binding sites used in the EMSAs. (B) EMSA demonstrating the specific binding of factors to the Runx/octamer element present in the β-casein promoter. Nuclear extracts from HC11 cells were incubated with radiolabeled Runx/octamer DNA in the presence or absence of a 100-fold molar excess of competitor DNA, as indicated above the lanes. (C) EMSA in which nuclear extracts from HC11 cells were incubated with radiolabeled Runx/octamer DNA in the presence or absence of specific antibodies (Ab), as indicated above the lanes. NS, nonspecific control antibody (anti-Flag). The identity of each complex is indicated. (D) ChIP assay demonstrating the binding of endogenous Runx2 and Oct-1 to the endogenous β-casein promoter. Chromatin was obtained from HC11 cells and immunoprecipitated with the antibodies indicated above the lanes. Lane 1 contains input chromatin that was not immunoprecipitated. (E) Activation of the β-casein promoter by Runx2 and Oct-1. HC11 cells were transfected with 900 ng of β-caseinWt, β-caseinMRunx, or β-caseinMOctamer in the presence or absence of Runx2 (333 ng) or Oct-1 (333 ng) expression plasmids. The total amount of DNA used in each transfection was adjusted to 2 μg by using pCMV5 and pCDNA3. All transfections were performed in triplicate; luciferase activities are presented as means and SDs. All values are relative to the activity of the pGL3-Basic reporter.

The identity of the proteins in each of the complexes was established by using anti-Runx2 and anti-Oct-1 antibodies in supershift assays (Fig. 2C). Incubation with Runx2-specific antibodies demonstrated that complexes C1 and C3 contain Runx2, whereas incubation with Oct-1-specific antibodies demonstrated that complexes C1 and C2 contain Oct-1. These data demonstrate that the C1 complex consists of Oct-1 and Runx2 specifically bound to the Runx/octamer element of the β-casein promoter.

To confirm that the endogenous β-casein promoter is bound by endogenous Runx2 and Oct-1, ChIP assays were performed with Runx2- and Oct-1-specific antibodies. The β-casein promoter could be amplified from immunoprecipitated complexes obtained with either anti-Runx2 or anti-Oct-1 antibodies but not with irrelevant anti-Flag antibodies (Fig. 2D). Runx2 and Oct-1 are therefore bound to the endogenous β-casein promoter.

Overexpression of Runx2 and Oct-1 resulted in the activation of the β-casein promoter to a greater extent than did that of either protein alone (Fig. 2E). The magnitude of activation was dependent upon both Runx and octamer sequences, since mutation of either sequence significantly inhibited promoter activity. Importantly, Oct-1 was still able to activate the promoter when the octamer sequence was intact and the Runx sequence was mutated, but maximum activity was reduced by more than 50%. In contrast, Runx2 was unable to activate the promoter when the Runx sequence was intact and the octamer sequence was mutated. Runx2 therefore requires an intact octamer element to contribute to the activity of the β-casein promoter. Taken together, these findings suggest that Oct-1 can bind to and activate the promoter in the absence of Runx2 but that it cooperates with Runx2 to achieve maximum activation of the promoter. Runx2 therefore boosts the level of β-casein promoter activity in an Oct-1-dependent manner.

Oct-1 stimulates the binding of Runx2 to the Runx/octamer element.The Runx2/Oct-1 complex was investigated further by using in vitro-translated Runx2 and Oct-1 in EMSAs (Fig. 3). Both proteins formed distinct complexes when incubated individually with the Runx/octamer element (Fig. 3A, lanes 2 and 3). In contrast, when both proteins were incubated together, they preferentially formed a distinct ternary complex (Runx2/Oct-1) in the presence of excess Runx/octamer DNA (Fig. 3A, lane 4). The formation of this ternary complex was dependent on the presence of both the Runx and the octamer sequences, since it was completely abolished when either sequence was mutated (Fig. 3A, lanes 8 and 12). This experiment also indicated that Oct-1 stimulates the recruitment of Runx2 to DNA. We therefore performed a titration experiment in which increasing amounts of Oct-1 were incubated with constant amounts of Runx2 and the Runx/octamer element (Fig. 3B). The amount of the ternary complex increased with increasing amounts of Oct-1 to a level greater than that observed with Runx2 alone (Fig. 3B). Hence, Oct-1 stimulates Runx2 recruitment to the Runx/octamer element. Increasing amounts of Runx2 also resulted in increasing amounts of the ternary complex (Fig. 4A). However, in this experiment, the amount of the ternary complex was only slightly larger than that observed with Oct-1 alone, demonstrating that Runx2 has only a minor effect on the binding of Oct-1 (Fig. 4A, compare lanes 2 and 7).

• Open in new tab
• Download powerpoint

FIG. 3.

Oct-1 stimulates the binding of Runx2 to DNA. (A) (Upper panel) Sequences of the mutant Runx/octamer elements used in the EMSAs. Mutations in either the Runx- or the octamer-binding site are indicated by asterisks. (Lower panel) EMSA showing the formation of a Runx2/Oct-1 complex on the Runx/octamer element with in vitro-translated Runx2 and Oct-1. Proteins were incubated with radiolabeled Runx/octamer, RunxMut/octamer, or Runx/octamerMut, as indicated above the lanes. (B) Titration of increasing amounts of Oct-1 in the presence of a constant amount of Runx2. Runx2 and Oct-1 were incubated separately or together with radiolabeled Runx/octamer DNA. Lane 1 contains unprogrammed lysate (3 μl), lane 2 contains Oct-1 (3 μl), and lane 3 contains Runx2 (3 μl). In lanes 4 to 8, a constant amount of Runx2 (3 μl) was incubated with increasing amounts of Oct-1 (0.5, 1.0, 3.0, 5.0, and 8.0 μl). The identity of each complex is indicated.

• Open in new tab
• Download powerpoint

FIG. 4.

Juxtaposition of the Runx- and octamer-binding sites is important for the efficient formation of a Runx2/Oct-1 complex. Runx2 and Oct-1 were incubated separately or together with radiolabeled Runx/octamer (A), Runx/octamer A, in which the sites were separated by an additional 5 bp (B), or Runx/octamer B, in which the sites were separated by an additional 10 bp (C). Increasing amounts of Runx2 were added in the presence of a constant amount of Oct-1. Lane 1 contains Runx2 (3 μl), and lane 2 contains Oct-1 (3 μl). In lanes 3 to 7, a constant amount of Oct-1 (3 μl) was incubated with increasing amounts of Runx2 (0.5, 1.0, 3.0, 5.0, and 8.0 μl). The intensity of the complexes was quantified, and the data are represented graphically below each EMSA.

The Runx and octamer sequences in the Runx/octamer elements of the casein promoters are separated, almost invariably, by 2 bp (Fig. 1B). To establish whether this rigorous conservation of spacing between the sequences was important for the formation of the Runx2/Oct-1 complex, we tested the effect of increasing the spacing between the sequences. When increasing amounts of Runx2 were titrated with a fixed amount of Oct-1 and a mutant Runx/octamer element in which the sequences were separated by an additional 5 or 10 bp, there was a significant decrease in ternary complex formation compared to that seen with the wild-type Runx/octamer element (Fig. 4). Thus, the juxtaposition of the Runx and octamer sequences is critical for the efficient formation of a Runx2/Oct-1 complex.

Runx2 contains N-terminal and C-terminal autoinhibitory domains for DNA binding.To establish whether Oct-1 stimulates the recruitment of Runx2 via a direct interaction, we initially deleted the residues C terminal to the Runt domain and tested the ability of Oct-1 to stimulate the recruitment of the truncated Runx2 protein (Fig. 5A and B). Deletion of the C-terminal region to residue 292 did not prevent the efficient formation of a ternary complex with Oct-1. However, this truncated protein had a significantly increased ability to bind autonomously to DNA and formed a binary complex as efficiently as the ternary complex (Fig. 5B, lanes 6 and 7). In contrast, a smaller deletion to residue 462 did not affect the ability of Runx2 to bind autonomously to DNA; it was still able to form a stimulated Runx2/Oct-1 complex. These findings suggested that the C-terminal region of Runx2 contains an autoinhibitory domain for DNA binding between amino acids 292 and 462. More precise mapping of the autoinhibitory region established that truncation to residue 424 resulted in an observable increase in autonomous DNA binding, which was further enhanced upon deletion of additional C-terminal residues to amino acid 240 (Fig. 5C, even-numbered lanes). In contrast to their differential abilities to bind autonomously to DNA, all deletion proteins were able to form a ternary complex with Oct-1 as efficiently as wild-type Runx2 (Fig. 5C, odd-numbered lanes). Thus, the C-terminal region of Runx2 inhibits its DNA-binding function and is required for Oct-1-dependent recruitment to the Runx/octamer element.

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG. 5.

The C-terminal region of Runx2 is essential for its Oct-1-dependent recruitment to the Runx/octamer element. (A) Diagrammatic representation of the Runx2 C-terminal deletions. The panel below the diagram shows in vitro translations of the C-terminal deletions of Runx2 on an SDS-12% polyacrylamide gel. (B) EMSA showing wild-type Runx2 (C-528) and C-terminally truncated proteins C-468 and C-292 binding to the Runx/octamer element in the presence or absence of Oct-1. Equal amounts of Runx2 truncated proteins and Oct-1 were added. The identities of the complexes are indicated. (C) Mapping of the minimal C-terminal region required for Oct-1-dependent recruitment of Runx2. An EMSA shows the binding of Runx2 C-terminal truncations to the Runx/octamer element in the presence or absence of Oct-1. Equal amounts of Runx2 truncated proteins and Oct-1 were added.

The presence of two translated Runx2 protein products in the EMSAs prevented clear mapping of the autoinhibitory domain. We predicted that the two bands arose from full-length Runx2 and an N-terminal truncation starting from the internal methionine at position 39. We therefore mapped the C-terminal autoinhibitory domain in the context of N-terminal deletions (Fig. 6). Two N-terminal deletions, N-41 and N-108, in which the first 40 and 107 amino acids were deleted, respectively, were made. Comparison of the relative binding of wild-type Runx2 with the binding of the N-41 deletion demonstrated that residues within the first 40 amino acids inhibited DNA binding (Fig. 6B, lanes 2 and 3). Further deletion to the Runt domain (N-108) resulted in a more modest increase in binding (Fig. 6B, lane 5). The more modest binding seen with the larger deletion may reflect a role for residues between 40 and 107 in enhancing DNA binding; alternatively, a deletion so close to the start of the Runt domain may have a slight negative effect on its ability to bind to DNA. When the C-terminal region of the N-terminally truncated proteins was deleted, further significant enhancement of DNA binding was observed (Fig. 6B, lanes 4 and 6). These observations indicated that Runx2 DNA binding is autoinhibited by regions located both N and C terminal to the Runt domain.

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG. 6.

The N-terminal and C-terminal regions of Runx2 contain autoinhibitory domains for DNA binding. (A) Diagrammatic representation of the Runx2 deletions; the arrows indicate the N termini of the two series of truncated proteins (N-41 and N-108). The panel below the diagram shows in vitro translations of the N-41 series of C-terminal deletions on an SDS-12% polyacrylamide gel. Lane L contains unprogrammed lysate. (B) EMSA showing that both the N-terminal and the C-terminal regions contain autoinhibitory domains. Equal amounts of proteins were used in the reactions. The identities of the proteins are shown above the lanes. (C) EMSA mapping of the C-terminal autoinhibitory domain. Equal amounts of in vitro-translated N-41 series of Runx2 truncations were incubated with the Runx/octamer element. The identities of the proteins are shown above the lanes. (D) Oct-1-dependent recruitment of Runx2 does not require the N-terminal region of Runx2. (Upper panel) In vitro translations of the N-108 series of Runx2 proteins on an SDS-12% polyacrylamide gel. (Lower panel) EMSA showing binding of the N-108 series of Runx2 truncations to the Runx/octamer element in the presence or absence of Oct-1. Equal amounts of Runx2 proteins were used in the reactions.

More precise mapping of the C-terminal autoinhibitory domain was carried out by generating a series of C-terminal deletions in the context of the N-41 truncation. Analysis of the autonomous DNA-binding activities of these proteins demonstrated that the major autoinhibitory domain was located between residues 424 and 396 (Fig. 6C). C-terminal Runx2 deletions in which the N terminus also was truncated were also tested for their abilities to form a ternary complex with Oct-1. Oct-1 formed a ternary complex to the same extents with all of the truncated proteins (Fig. 6D). However, stimulated recruitment of Runx2 was not observed in proteins with deletions up to and extending beyond residue 396. The N terminus of Runx2 is therefore not essential for the Oct-1-stimulated recruitment of Runx2.

Taken together, these observations demonstrate that the C-terminal region of Runx2 is required for its Oct-1-dependent recruitment to the Runx/octamer element and that the major autoinhibitory domain for DNA binding resides in this region.

The C-terminal region of Runx2 interacts with Oct-1.To investigate whether Runx2 and Oct-1 interact, coimmunoprecipitation experiments were performed. We expressed the proteins in HeLa cells by using plasmids expressing Runx2 and HA-Oct-1. Runx2 could be coimmunoprecipitated with the anti-HA antibody, suggesting that Runx2 and Oct-1 interact directly (Fig. 7A). Further analysis of this interaction with GST pull-down assays demonstrated that wild-type Runx2 interacted with GST-Oct-1 (Fig. 7B). The interaction diminished progressively when GST-Oct-1 was incubated with a series of C-terminally truncated Runx2 proteins. Deletion of residues from the C terminus to amino acid 462 significantly reduced binding to Oct-1, and the interaction was almost completely abolished when the deletion in Runx2 was extended to amino acid 343 (C-343). Taken together, the data demonstrate that Runx2 interacts with Oct-1 predominantly via the C-terminal region of Runx2. While we could not define a discrete Oct-1 interaction domain, at least part of it resides within the same region as the autoinhibitory domain for DNA binding.

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG. 7.

The C-terminal region of Runx2 is required for its interaction with Oct-1. (A) Coimmunoprecipitation of Runx2 and Oct-1. HeLa cells were transfected with Runx2 and Oct-1-HA expression plasmids as indicated above the lanes. Immunoprecipitation was performed with an anti-HA antibody (Ab), and the membrane was incubated with an anti-Runx2 antibody. (B) GST pull-down assay showing that the C-terminal region of Runx2 is required for its interaction with Oct-1. A GST-Oct-1 fusion protein was incubated with in vitro-translated Runx2 and its C-terminal truncations as indicated above the lanes. (C) Runx2 interacts with the POU domain of Oct-1. A diagrammatic representation shows various truncated forms of Oct-1 used in pull-down assays. In the panels below the diagram, GST fusion proteins were incubated with in vitro-translated Runx2 and visualized on an SDS-polyacrylamide gel. Lanes 1 and 7 contained 10% the input of in vitro-translated Runx2.

To map the Runx2 interaction domain in Oct-1, GST pull-down assays were performed with various deletion derivatives of GST-Oct-1. Deletion of the N-terminal region to the POU domain (Oct-1 amino acids 1 to 444) did not affect the interaction with Runx2 (Fig. 7C, lane 4). In contrast, deletion to amino acid 274, which removes the entire POU domain, abolished the interaction (Fig. 7C, lane 3). Moreover, incubation of Runx2 with just the POU domain of Oct-1 fused to GST demonstrated that Runx2 interacts with Oct-1 via the POU domain (Fig. 7C, lane 8).

Model of Runx2/Oct-1 complex formation.The crystal structures of the POU domain/DNA complex and the Runt domain/DNA complex were used to model the POU domain and the Runt domain bound to the β-casein Runx/octamer element (Fig. 8). The model clearly shows that the two domains can bind to the adjacent sequences without contacting each other. The structure of the C terminus has not been resolved, and it is depicted as interacting with the Runt domain to block DNA binding. We propose that the inhibition of DNA binding is relieved by interaction of the Runx2 C-terminal region with the POU domain of DNA-bound Oct-1 (Fig. 8).

• Open in new tab
• Download powerpoint

FIG. 8.

Model illustrating the proposed mechanism by which Oct-1 stimulates the recruitment of Runx2 to the Runx/octamer element. The structures of the Runt domain and POU domains bound to DNA were modeled on the Runx/octamer element (11, 60). The C-terminal autoinhibitory region (C) is depicted as interacting with the DNA-binding face of the Runt domain. Upon recruitment of Runx2 to the POU/DNA complex, the C-terminal region is depicted as interacting with the POU domain, thereby unmasking the DNA-binding face of the Runt domain and stimulating its recruitment to the Runx/octamer element. The N-terminal domain is omitted for clarity.

A role for Runx2 in mammary gland-specific gene expression.In this study, we have demonstrated that Runx2 is an essential component of the transcriptional regulatory complex that is required for full transcriptional activation of the β-casein gene. The main significance of this finding is that it demonstrates a role for the osteoblast transcription factor Runx2 in the regulation of a mammary gland-specific gene. Other studies have also revealed molecular connections between breast and bone. For example, genes important in the control of bone remodeling are expressed in breast cancer and in the lactating breast; these include the genes for RANK, RANKL, vitamin D, bone sialoprotein, osteopontin, and calcitonin (reviewed in reference 36). It is also intriguing to note that osteocalcin, a Runx2 target in osteoblasts, and β-casein, a Runx2 target in mammary epithelial cells, are both calcium phosphate-binding proteins (22a, 61). These data may reflect a conserved functional role for Runx2 in the physiological regulation of calcium. Moreover, given the precedent for Runx transcription factors being essential for the development of a number of tissues, it is possible that Runx2 has a more extensive role in the mammary gland.

Regulation of β-casein gene expression by the Runx2/Oct-1 complex.It is well established that the primary mechanism of activation of the β-casein gene is via STAT5 and GR but it is also clear that the Runx2/Oct-1 complex is required (2, 9, 13, 19, 34, 49, 63, 64). We have not yet investigated the regulation of the Runx2/Oct-1 complex by signal transduction pathways but it is possible that both the formation and activity of the complex are regulated throughout the pregnancy cycle. Indeed, the activity of both proteins can be regulated by phosphorylation, which may be important in the context of β-casein transcription (17, 20). Another potential role for the Runx2/Oct-1 complex is to act as a “landing pad” for other transcription factors. For example, both proteins have been shown to interact with GR, suggesting that the Runx2/Oct-1 complex may recruit GR (42, 48). Oct-1 was recently shown to interact with STAT5; if this interaction occurs on the β-casein promoter, then the Runx2/Oct1 complex may be important in maintaining promoter architecture during activation by interacting with STAT5 bound at blocks A and B (35). Runx2 is also a nuclear matrix-binding protein and may serve to recruit the β-casein promoter to transcriptionally active nuclear subdomains via its interaction with Oct-1 on the Runx/octamer element (reviewed in reference 58).

The Runx2/Oct-1 complex is reminiscent of the Runx1/Ets-1 complex in that both proteins in these complexes bind to adjacent sites, and the formation of the complex is stimulated by relieving autoinhibition of DNA binding as a result of their interaction (18, 22, 26, 31). In the Runx1/Ets-1 complex, the two proteins reciprocally stimulate their DNA binding by interacting via their respective autoinhibitory domains, resulting in truly cooperative binding. The Runx2/Oct-1 complex differs slightly, in that Runx2 binding is stimulated by Oct-1, but Oct-1 binding is only slightly enhanced by the presence of Runx2. The importance of the juxtaposition of the two sites also suggests that there is little flexibility in the protein-protein interaction between the C-terminal region of Runx2 and the POU domain. In addition, while the model depicted indicates that the two DNA-binding domains do not interact directly, it is possible that the DNA is deformed and that the Runt and POU domains are in contact when bound to the DNA (Fig. 8).

Autoinhibition of Runx2 DNA binding.Autoinhibition of DNA binding was not previously reported for Runx2, although it has been observed that a GST fusion protein containing the isolated Runt domain has a greater affinity for DNA than GST-Runx2 (15). The intramolecular mechanism of autoinhibition is likely to be similar to that of Runx1, as the domains map to similar regions (18, 22, 26, 31).

Regions located both N terminal and C terminal to the Runt domain have the capacity to autoinhibit DNA binding, although the C-terminal region possesses the major autoinhibitory function. The N-terminal autoinhibitory domain is located within the first 40 amino acids of Runx2. Deletion of the first 40 amino acids of Runx1 also resulted in an increase in DNA binding (22). Both proteins contain a block of 19 amino acids that is highly conserved and is therefore likely to contain the N-terminal autoinhibitory region. This region is also conserved in Runx3, suggesting that N-terminally mediated autoinhibition of DNA binding is a feature of all three mammalian Runt domain transcription factors. The major autoinhibitory region in the C-terminal region of Runx2 mapped to between residues 396 and 462. This region contains the well-defined nuclear matrix targeting signal, and it will be important to determine whether inhibition of DNA binding and association with the nuclear matrix are related or distinct activities (67).

The finding that Runx2 contains an autoinhibitory domain for DNA binding is also likely to be an important feature of its function in osteoblasts. Runx2 DNA-binding activity is stimulated in response to activation by the mitogen-activated protein kinase and protein kinase C pathways (17). Upon stimulation of these pathways, the C-terminal PST region of Runx2 becomes phosphorylated; this region encompasses the autoinhibitory domain, and we hypothesize that phophorylation stimulates DNA binding by relieving autoinhibition mediated by the C-terminal region.

In summary, we have shown that Runx2 forms a novel complex with Oct-1 on the highly conserved Runx2/Oct-1 element and contributes to the expression of the β-casein gene. This is the first evidence that Runx2 is involved in the regulation of a mammary gland-specific gene. Given the importance of Runx transcription factors in tissue development, it is possible that Runx2 has a more extensive role in the mammary gland. It will therefore be important to address the role of Runx2 in vivo by generating mice in which the Runx2 gene has been specifically deleted in the mammary gland.