Cell lines were obtained from American Type Culture Collection (ATCC) and maintained in Dulbecco's Modified Eagle Medium (DMEM; Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Gemini Bio-Products, West Sacramento, CA, USA) and 1% penicillin/streptomycin/glutamine mixture (Invitrogen, Carlsbad, CA, USA). Culture medium for the HS578T cell line was supplemented with 0.01 mg/ml bovine insulin (Invitrogen, Carlsbad, CA, USA).

The University of California Santa Cruz (UCSC) genome browser 36 was used to determine which nucleotides in the HP1α promoter region are conserved between human, mouse, rat and chimpanzee. The sequence of the HP1α gene promoter region was entered into the TESS database to identify transcription factor binding motifs 35. Motifs with high similarity to a consensus binding site and high sequence conservation between species were noted.

Portions of the HP1α gene promoter region were PCR-amplified from a BAC clone [GenBank:AC078778] containing a portion of human chromosome 12. The PCR-amplified fragments were sequenced and ligated into the pGL3 luciferase reporter vector (Promega, Madison, WI, USA) digested with HindIII and XhoI. Primer sequences are available upon request.

The QuikChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA, USA) was used according to the manufacturer's instructions. Mutations in transcription factor binding sites were as follows: site YY1.1 was changed from AAATGG to AAGCTT, site YY1.2 was changed from AAAATGGCG to AAAGCTTCG, the E-box site was changed from CACGTG to CGATCG, and site NRF-1.3 was changed from TGCGCAGGCGCA to TGCGCATATGCA. In each case, nucleotides reported to be critical for factor binding were changed. The YY1.2 mutation abolished YY1 binding (see text), and the NRF-1.3 mutation abolished NRF-1 binding as determined by electrophoretic mobility shift assay (EMSA) (data not shown).

RT-PCR was performed using the Roche LightCycler instrument. Reactions were carried out in LightCycler capillaries (20 μl) (Roche, Indianapolis, IN, USA) and contained SYBR. Green Taq ReadyMix (Sigma, St. Louis, MO, USA), Enhanced avian reverse transcriptase (Sigma, St. Louis, MO, USA) (2 units/reaction), and 250 nM forward and reverse primers. Reactions contained 400 ng of template RNA with the exception of the 28S-specific reactions, which contained 5 ng of template RNA. Reverse transcription was performed at 61°C followed by a 95°C denaturation step. The annealing temperature for all reactions was 60°C.

Concentrations of protein samples were quantified using Bradford reagent (BioRad, Hercules, CA, USA). Samples were run on gels containing 8 or 10% acrylamide (National Diagnostics, Atlanta, GA, USA) and then transferred to nitrocellulose membranes, which were blocked with tris-buffered saline with Tween (TBST) containing 5% milk. The membranes were probed with antibodies against HP1α (Millipore #07-346, Billerica, MA, USA), YY1 (Santa Cruz Biotechnology sc-7341X, Santa Cruz, CA, USA), hnRPA1 (Abcam ab5832, Cambridge, MA, USA), or β-tubulin (Covance TU27 MMS410P, Princeton, NJ, USA). Signal was detected using horseradish peroxidase-conjugated secondary antibody and developed using the BioRad Immun-Star reagents (Hercules, CA, USA).

Cells were plated in 35 mm culture dishes (10⁵ cells/dish). On the following day cells were transfected with plasmid DNA (0.5 μg total) using 1.5 μl TransIT-LT1 reagent (Mirus, Madison, WI, USA) per plate. Each plate was transfected with 0.25 μg of luciferase reporter plasmid and 0.25 μg of the CMV-β-galactosidase plasmid. The final concentration of plasmid DNA was 0.5 ng/μl. After 6 to 8 hours cells were washed with phosphate-buffered saline (PBS) and re-fed. The next day plates were washed with PBS and cells were harvested using 0.4 ml Triton/Gly-Gly lysis buffer (25 mM Gly-Gly, pH 7.8; 15 mM MgSO₄; 4 mM ethylene glycol tetraacetic acid (EGTA); 1% Triton X-100; 1 mM dithiothreitol (DTT)). Luciferase assays were performed by adding 50 μl of cell lysate to 300 μl of luciferase reaction solution (27 mM Gly-Gly, pH 7.8; 16 mM MgSO₄; 0.1 mg/ml BSA; 1 mM DTT; 1 mM ATP). The sample was put in a luminometer (EG&G Berthold Lumat LB 9507, Oak Ridge, TN, USA), which injected 100 μl of 1 mM D-luciferin into each sample and measured the light emission for 10 seconds. β-galactosidase assays were carried out by adding 50 μl of cell lysate to 500 μl of Lac-Z reaction buffer (100 mM sodium phosphate, pH 6.95; 10 mM KCl; 1 mM MgSO₄; 0.21% β-mercaptoethanol), followed by addition of 100 μl of o-nitrophenyl-β-D-galactopyranoside (ONPG) (4 mg/ml). The reactions were stopped by addition of 1 M Na₂CO₃ and quantified by measuring the absorbance at 420 nm. Transfection and reporter assays were performed at least three times and the result of a representative experiment is shown.

MCF7 cells were plated in 60 mm dishes (10⁵ cells/dish). The following day the cells were transfected with siRNA using Oligofectamine reagent (Invitrogen, Carlsbad, CA, USA). siRNA oligos were diluted to 500 nM with Opti-MEM (Invitrogen, Carlsbad, CA, USA). Oligofectamine reagent (2.8 μl/plate) was diluted approximately 1:100 in Opti-MEM and allowed to stand at room temperature for five minutes. siRNA and Opti-MEM dilutions were then mixed at a 1:1 ratio and allowed to stand at room temperature for 20 minutes before adding to plates of cells. The final concentration of siRNA in each plate was 50 nM. The following day the plates were washed with PBS and re-fed with DMEM containing 10% fetal bovine serum. Five days after transfection RNA and protein were isolated using TRI Reagent (Sigma, St. Louis, MO, USA) according to the manufacturer's instructions. siRNA oligos were purchased from Dharmacon and were specific for luciferase (sense strand 5'-CUUACGCUGAGUACUUCGA-3') or YY1 (oligo 59: sense strand 5'-AAGAUGAUGCUCCAAGAAC-3'; oligo OT: sense strand 5'-CAUAAAGGCUGCACAAAGA-3').

Probes were prepared by end-labeling each oligo (wild type probe: 5'-GCGCAAAACTCGCCATTTTACTACACG-3' and its complementary sequence; YY1 mutant probe: 5'-GCGCAAAACTCGAAGCTTTACTACACG-3' and its complementary sequence) with γ-³²P ATP using T4 polynucleotide kinase (Promega, Madison, WI, USA). Radiolabeled probes were purified using Quick Spin Columns (Roche, Indianapolis, IN, USA). A 6% TBE (tris, boric acid, edta)-polyacrylamide gel was poured and allowed to polymerize overnight. Nuclear extract (25 μg) was incubated on ice for 15 minutes in EMSA buffer (6 mM Tris, pH 8.0; 6 mM MgCl₂; 150 mM NaCl; 1 mM DTT) containing 10 μg/ml BSA, 10 μg/ml poly (dI-dC), 50 μg/ml salmon sperm DNA (Invitrogen, Carlsbad, CA, USA). One μg of antibody against AcK9H3 (Upstate #06-599) or YY1 (Santa Cruz sc-7341X) was then added to the appropriate samples, and all samples were incubated at room temperature for 30 minutes. One μl of the appropriate radiolabeled probe was then added and the samples (10 μl total volume) were incubated at room temperature for 30 minutes. Samples were run on a 6% polyacrylamide gel (pre-run for 30 minutes) at 80V for approximately 1.5 hours. The gel was dried and exposed to film.

The chromatin immunoprecipitation (ChIP) assay was carried out as previously described 37 with the exception of the sonication step. The chromatin was sonicated in a dry ice/ethanol bath for 10 minutes at amplitude of 40% to generate DNA fragments between 150 and 350 bp in length. Samples included either 2 μg normal mouse IgG, 0.2 μl nuclear respiratory factor-1 (NRF-1) antibody 38 or 4 μg YY1 antibody (Santa Cruz sc-1703). PCR was performed using 5 μl of the ChIP-enriched DNA in a 25 μl reaction. The reaction was carried out for 34 cycles with an annealing temperature of 60°C. PCR products were stained with SYBR Green I nucleic acid gel stain (Invitrogen, Carlsbad, CA, USA) and run on a 2% low-melt agarose gel. The gel was visualized using a Typhoon scanner (GE Healthcare, Piscataway, NJ, USA) and the intensity of each band was quantified using ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA). Primer sequences are available on request.

HS578T cells were split 1:6 and transfected the next day with 6 μg of either an empty vector (CMV-empty) or a YY1 overexpression vector (pCDNA3 HA-YY1) using TransIT-LT1 reagent (Mirus, Madison, WI, USA). After 72 hours the cells were treated with Versene-EDTA (Cambrex Bio Science, East Rutherford, NJ, USA) and were resuspended in migration buffer (DMEM without phenol red (Mediatech, Manassas, VA, USA), 1% BSA, 1 μM MgCl₂, 0.2 μM MnCl₂). Migration transwells (Corning, Lowell, MA, USA) and matrigel invasion chambers (BD Biosciences, Franklin Lakes, NJ, USA) were prepared according to the manufacturer's protocols. 10⁵ cells were used for each migration assay and 5 × 10⁴ cells for each invasion assay. Twenty-four hours later the migration and invasion chambers were cleaned with cotton-tip applicators and stained for 15 minutes in crystal violet solution (diluted 1:5 in dH₂O from a stock of 2 mg/ml in methanol). The membranes were destained in dH₂O and allowed to dry overnight. The stain was eluted from the migration membranes using 10% acetic acid, and the eluate was read at OD 600 nm in a Versamax microplate reader (Molecular Devices, Sunnyvale, CA, USA). The invasion membranes were removed, and mounted on slides using Permount (Fisher Scientific, Pittsburgh, PA, USA). The cells in 8 to 10 high-power fields (200× magnification) were counted and averaged for each membrane.

We cloned the upstream sequence of the human gene encoding HP1α, which shares a 603 bp intergenic region with the gene encoding heterogeneous nuclear ribonucleoprotein A1 (HNRPA1; Figure 1a). We used the UCSC genome browser 36 to look for areas of sequence conservation between species in the region between the HP1α and HNRPA1 gene transcriptional start sites. Part of the sequence of the intergenic region is depicted in Figure 1b. Nucleotides that are conserved between human, mouse, rat, and chimpanzee are indicated in red. Note the high degree of sequence conservation in the intergenic region. Next, we used the TESS database to identify potential transcription factor binding sites in this region (Figure 1b) 35. The HP1α gene promoter has been shown to be a target of E2F1 and E2F4 as well as NRF-1 3739. It is therefore not surprising that we identified a highly conserved E2F binding motif, as well as three highly conserved NRF-1 binding motifs. We also noted the presence of an E-box site, which has been suggested to be involved in differential expression of HP1α between MCF7 and MDA-MB-231 cells 40, and two highly conserved YY1 binding motifs. We performed quantitative RT-PCR analysis to determine relative mRNA levels of HP1α and HNRPA1 in MCF7 and HS578T cells. In agreement with previously published data 2829, expression of HP1α mRNA and protein is lower in HS578T and MDA-MB-231 cells than in MCF7 cells (Figures 1c, d; data not shown). In contrast, we found that the expression of HNRPA1 mRNA and protein does not differ between these two cell lines (Figures 1c, d).

The relative difference in HP1α mRNA level between non-invasive and invasive breast cancer cell lines can be attributed to a difference in transcriptional activity or a difference in mRNA stability. To test if it is due to a difference in mRNA stability, we determined the mRNA half-life in MCF7 and MDA-MB-231 cells using actinomycin D treatment followed by quantitative RT-PCR. We found that the half-life of HP1α mRNA is approximately 12.5 hours in both cell lines (data not shown). Therefore, as a whole, our data suggest that there is a difference in transcriptional activity at the HP1α gene promoter between different cell lines.

To determine the level of transcriptional activity at the HP1α gene promoter we sub-cloned the promoter region into a luciferase reporter construct. We made several luciferase constructs, one containing the entire bidirectional promoter region (CbxD98) and others containing truncations of the promoter at the 5' end (CbxU283 and CbxU146; Figures 1b and 2a). We also explored the possibility that sequence downstream of the start site may be involved in differential expression and found the sequence to have no effect on promoter activity (data not shown). Because it is difficult to directly compare luciferase activity levels between two cell lines, we decided instead to observe differences in the trend of luciferase activity of different constructs in each cell line. Therefore, the luciferase activity of each construct in Figure 2a is represented as a percentage of the activity of the CbxD98 construct. Truncating a portion of the promoter between -594 and -283 resulted in increased luciferase activity in both cell lines (compare CbxD98 and CbxU283 in Figure 2a). Truncation of the sequence between -283 and -146 caused a reduction in luciferase activity. Interestingly, this reduction in luciferase activity was more pronounced in MCF7 cells than in HS578T cells (compare CbxU283 and CbxU146 in Figure 2a). The results suggest that the region between -283 and -146 may be important for higher expression of HP1α in MCF7 cells as observed in Figure 1c.

To determine which binding motifs in the region between -283 and -146 may be important for differential promoter activity, we introduced mutations into four putative transcription factor binding sites in the region in the CbxU283 construct (Figure 1b). These binding sites included an NRF-1 site, two YY1 sites, and one E-box site. In Figure 2b, the luciferase activity of each mutant construct is displayed as a percentage of the activity of the wild-type CbxU283 construct. Each mutation had an effect on the promoter activity that was comparable in both cell lines; however, the only mutation that consistently had a differential effect between the two cell lines was the mutation in the YY1.2 site (the distal YY1 site, construct M2). The YY1.2 mutation caused a more pronounced reduction in promoter activity in MCF7 cells than in HS578T cells similar to what was observed for CbxU146 (Figure 2b). These results suggest that YY1 may play a role in the differential expression of HP1α between MCF7 and HS578T cells.

To determine what role YY1 plays in HP1α expression we performed siRNA knockdown experiments. MCF7 cells were transfected with siRNAs against YY1 or, as a control, an siRNA against luciferase. Two different YY1-specific siRNA oligos (59 and OT) were used to account for possible off-target effects. Five days after transfection with siRNA, protein and RNA were collected. Both of the YY1-specific siRNAs were effective in reducing the level of YY1 protein (Figure 3a). Quantitiative RT-PCR was used to determine the effect of YY1 knockdown on the expression of several mRNAs (Figure 3b). The level of HP1α mRNA was reduced significantly following YY1 knockdown. In contrast, the level of HNRPA1 mRNA was unaffected by YY1 knockdown. In addition, the levels of HP1β and HP1γ mRNAs were relatively unaffected by YY1 knockdown. The fact that YY1 knockdown had a specific negative effect on HP1α expression led to the conclusion that YY1 is a positive regulator of HP1α expression.

The results of the luciferase experiments indicated that YY1 might play a role in the differential expression of HP1α in MCF7 and HS578T cells. An examination of the YY1 mRNA and protein levels showed that they are both reduced in HS578T cells compared with MCF7 cells (Figure 1c, d). This is significant, because our knockdown experiment indicated that YY1 is a positive regulator of HP1α expression. We used gel shift experiments to determine if a lower level of YY1 expression in HS578T cells resulted in decreased binding of YY1 to the YY1.2 site in vitro. Labeled probes containing the wild type or mutant YY1.2 sequence were incubated with nuclear extract from MCF7 or HS578T cells. The YY1-specific shift was only visible when we used MCF7 nuclear extract, indicating that there was not likely to be enough YY1 in the HS578T extract to produce a visible shift (Figure 4, compare lanes 2 and 5). The identity of the YY1-shifted species was confirmed by super-shift with an antibody against YY1 (Figure 4, lane 4 asterisk). Note the presence of a nonspecific band that was not shifted by the YY1-specific antibody. An unrelated antibody (against Ac H3K9) did not produce a super-shift (Figure 4, lane 3). The shifted band representing the YY1 protein-DNA complex was present only when the wild type probe was used, indicating that the mutation introduced into the YY1.2 site effectively disrupted YY1 binding (Figure 4, compare lanes 2 and 9). When more nuclear extract from HS578T cells was used, we observed a faint YY1 shift in the HS578T samples consistent with a low level of YY1 protein in these cells (data not shown).

The gel shift experiment demonstrated that reduced expression of YY1 in HS578T cells results in reduced binding of YY1 to its DNA binding motif in vitro. Next, we wished to determine if this phenomenon also occurs at the endogenous HP1α gene promoter using the ChIP assay. Chromatin prepared from MCF7 and HS578T cells was immunoprecipitated using specific antibodies against YY1 or NRF-1. The HP1α gene promoter was previously identified as an NRF-1 target gene; thus, it served as a positive control 39. NRF-1 occupancy was detected at HP1α gene promoter in both cell lines (Figure 5a). In contrast, YY1 occupancy at the HP1α gene promoter could only be detected in MCF7 cells. As a positive control for YY1 occupancy, we amplified a region in the glucocorticoid receptor gene promoter (located 2.2 kb upstream from the transcriptional start site), which contains three YY1 consensus binding motifs 41. YY1 occupancy was detected at this region in both cell lines, indicating that YY1 DNA binding activity is not entirely absent in HS578T cells (Figure 5a). As a negative control, we amplified a region 22 kb downstream of the HP1α gene transcriptional start site. As expected, only background signal was detected in this region. We quantified the ChIP data from several experiments and calculated the occupancy of each transcription factor as a percentage of input (Figure 5b). From our ChIP experiments we conclude that the lower level of YY1 in HS578T cells results in reduced occupancy of YY1 at the HP1α gene promoter region. These studies indicate that the reduced occupancy of YY1 at the HP1α gene promoter in HS578T cells contributes to lower HP1α expression.

HP1α has been identified as a suppressor of breast cancer cell invasion in vitro 29. We have shown that YY1 is a positive regulator of HP1α expression. Next, we wanted to know if YY1 expression correlates with breast cancer cell invasiveness. The invasion of metastatic cancer cells in vitro requires two distinct cellular phenotypes: migration toward a chemoattractant and invasion through a matrigel matrix. These assays represent a simplified model of the metastatic process in vivo. We tested the effect of YY1 overexpression on both migration and invasion of the invasive breast cancer cell line HS578T. Cells were transiently transfected with either an empty vector or a YY1 expression construct and then subjected to in vitro migration and invasion assays. As shown in Figure 6, overexpression of YY1 caused about a 50% reduction in the migration of HS578T cells (Figure 6a). YY1 overexpression also led to a decrease in invasive potential of these cells (Figure 6b). The reduction in invasion was similar to the reduction in migration, indicating that the decreased invasion is due to the decreased migration. Interestingly, the decrease in migration/invasion occurred in the absence of detectable increase in HP1α expression (Figure 6c). We conclude that YY1 overexpression suppresses the migration of HS578T cells in a manner independent of HP1α expression.