Regulation of Transcription Factor YY1 by Acetylation and Deacetylation

Plasmids.

Glutathione S-transferase (GST) was expressed from pGSTag (51). GST-YY1 (aa 1 to 414) [or, simply, GST-YY1(1–414)] was expressed from pGST-YY1 (73), and deletion constructs of GST-YY1 were generated by restriction enzyme digestion and religations of pGST-YY1. GST-p53 expression plasmid has been described (26). GST-p300 and GST-PCAF were expressed from plasmids pGEX2T-p300 (aa 1195 to 1810) and pGEX5X-PCAF (aa 352 to 832), respectively (9).

Gal4-YY1 was expressed from pM1-YY1, which was constructed by inserting the full-length YY1 cDNA in frame with the Gal4 DNA-binding domain in pM1 (52). pM1-YY1 (K170-R200) was prepared using adapter oligonucleotides containing AAG (lysine) to AGG (arginine) mutations within YY1 aa 170 to 200 of pM1-YY1. pG5CAT-Control was constructed by inserting five Gal4 DNA-binding sites into the BglII site of pCAT-Control (Promega), which contains the chloramphenicol acetyltransferase (CAT) gene downstream of the simian virus 40 promoter and enhancer sequences. Flag-tagged YY1 (F-YY1) was expressed from pCEP4F-YY1, which was generated by inserting the full-length YY1 cDNA into the pCEP4F vector (80). Serial Flag-YY1 deletion constructs were made by restriction enzyme digestion and religations of pCEP4F-YY1. pET15b-YY1, which expressed nontagged full-length YY1 in Escherichia coli in an inducible system, was constructed by cloning the full-length YY1 cDNA into pET15b (Novagen). pGEM7Zf3X-HD1, which was used to generate in vitro-translated HDAC1, was made by subcloning full-length HDAC1 into the pGEM7Zf3X vector, which is derived from pGEM7Zf(+) (Promega).

The following plasmids have been previously described: pBJ5-HD1F (64), which expresses HDAC1 C-terminally tagged with a Flag epitope; pBJ5.1-HD1F (H199F), HDAC1 point mutant (22); and pME18S-FLAG-HDAC2, which expresses Flag-tagged HDAC2 (35).

Recombinant proteins.

The GST fusion constructs of p300, PCAF, p53, and YY1 deletion mutants were expressed in E. coli DH5α, bound to glutathione-agarose beads (Sigma), washed extensively in phosphate-buffered saline (PBS), and eluted with 25 mM reduced glutathione. The eluate was then dialyzed against a buffer containing 20 mM Tris-HCl (pH 8), 0.5 mM EDTA, 100 mM KCl, 20% glycerol, 0.5 mM dithiothreitol (DTT), and 1 mM phenylmethylsulfonyl fluoride.

Non-tagged YY1 was expressed in E. coli BL21(DE3), induced with 0.2 mM isoprophyl-β-d-thiogalactopyranoside (IPTG), and captured by Ni²⁺-immobilized metal affinity chromatography (Invitrogen). Unbound bacterial proteins were removed with 50 mM imidazole, and bound YY1 was eluted with 500 mM imidazole. F-YY1 used in electrophoretic mobility shift assays (EMSA) was purified on an anti-Flag column (Sigma) under stringent conditions following the manufacturer's suggestions.

In vitro acetylation reactions.

Purified GST-YY1 and the serial deletion proteins were incubated at 30°C for 30 min with 0.25 μCi of [³H]acetyl coenzyme A ([³H]acetyl-CoA) (Amersham) and either purified GST-p300 (0.2 μg) or GST-PCAF (1 μg) in 30 μl of acetylation buffer containing 50 mM Tris-HCl (pH 8.0), 5% glycerol, 0.1 mM EDTA, 50 mM KCl, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride and 10 mM sodium butyrate. Proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The gels were fixed by Coomassie blue staining and subjected to signal amplification (Amplify; Amersham) prior to exposing to X-ray film.

For subsequent mass spectrometry analyses, 0.5 pmol of the YY1 peptide (GRVKKGGGKKSGKKSYLSGGAGAAGGRGADP) was acetylated in acetylation buffer for 2 h with CAT-assay grade acetyl-CoA (Amersham).

Cell culture, transfection, and CAT assays.

HeLa cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and penicillin-streptomycin. 10⁶ HeLa cells were seeded into 60-mm-diameter tissue culture dishes. Sixteen hours later, 10 μg of plasmids (for CAT assays, 5 μg of pG5CAT-Control plus 5 μg of pM1 effector plasmids) were transfected into cells using the calcium phosphate coprecipitation method (17). Forty-eight hours after transfection, cells were harvested by scraping and lysed by repeated freeze-thawing, and extracts were assayed for CAT activity by thin-layer chromatography (16).

Immunoprecipitation and Western blot analysis.

Immunoprecipitation of Flag-YY1 deletion proteins was done using anti-Flag M2 affinity gel (Sigma) following the manufacturer's suggestions.

Western blot analyses were performed using standard protocols (21). Immunoprecipitated proteins were detected with diluted primary antibodies (1:1,000 dilution of anti-acetyl-lysine [Upstate]; 1:5,000 dilution of anti-Flag M2 [Sigma]; 1:1,000 dilution of anti-HDAC1, HDAC2, or HDAC3 rabbit anti-serum [35, 66, 72]) followed by 1:7,500-diluted alkaline phosphatase-conjugated secondary antibodies (Promega). The blots were subsequently developed with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium (Promega).

In vitro protein-protein interaction assays.

³⁵S-labeled HDAC1 was generated from pGEM7Zf3X-HD1 using T7 RNA polymerase and the TNT Reticulocyte Lysate System (Promega). GST-YY1 (aa 170 to 200) was either acetylated with cold acetyl-CoA or mock acetylated and was then captured onto glutathione-agarose beads. In vitro-translated HDAC1 (5 μl) was mixed with the beads in the presence of PBS plus 0.2% NP-40 at 4°C for 1 h. Beads were washed extensively in PBS plus 0.2% NP-40. Bound proteins were eluted by boiling in Laemmli sample buffer, separated by SDS-PAGE, and detected by Coomassie blue staining and autoradiography.

Chemical labeling of peptides with [³H]acetate.

Peptides were labeled chemically according to the protocol described (64) with minor modifications. The peptides (0.4 mg) were labeled with 5 mCi of [³H]sodium acetate (2 to 5 Ci/mmol; New England Nuclear) in the presence of 0.24 M benzotriazol-1-gloxytrif-(dimethylamino)-phosphonium hexafluorophosphate and 0.2 M triethylamine (Aldrich). Labeled peptides were purified on a Microcon-SCX column (Millipore).

Histone deacetylation assays.

HeLa cells were transfected with 10 μg of pCEP4F-YY1 deletion plasmids by the calcium phosphate coprecipitation method (17). Forty-eight hours after transfection, cells were harvested, lysed in PBS plus 0.1% NP-40, and immunoprecipitated with anti-Flag M2 affinity gel (Sigma). HDAC activities of the immunoprecipitated Flag-YY1 were determined using a peptide corresponding to residues 2 to 24 of histone H4 as described (64) except that incubation was performed at room temperature overnight. Trichostatin A (TSA) (400 nM [final concentration]; Sigma) or a five-column volume of Flag peptide (Sigma) was added to the immunoprecipitate 30 min prior to addition of the H4 substrate peptide, if appropriate.

Immunofluorescence analysis.

HeLa cells were grown on chamber slides (Nalge Nunc International) for about 24 h and transfected with 10 μg of F-YY1 deletion constructs. Two days later, cells were washed with ice-cold PBS, fixed with 4% paraformaldehyde for 10 min, rinsed again with PBS, covered with 400 μl of 1% bovine serum albumin in PBS for 1 h at room temperature, washed again in PBS, and then treated with 1:200 dilution of anti-Flag fluorescein isothiocyanate (FITC) conjugate antibody (Upstate) for 1 h at room temperature. Subsequently, cells were subjected to extensive washing with PBS and coverslips were applied with one drop of antifade mounting medium with DAPI (4′,6′-diamidino-2-phenylindole) (Vector) before analysis under a fluorescence microscope.

EMSAs.

Purified YY1 proteins and GST-p53 were first acetylated or mock acetylated. Single-stranded oligodeoxynucleotides corresponding to a consensus YY1-binding site (56) or a p53 cognate sequence (18) were labeled individually with [γ³²-P]ATP and T4 polynucleotide kinase, heated together at 65°C, and allowed to anneal by slow cooling to room temperature. Binding reactions were performed in a 12-μl reaction volume containing 12 mM HEPES (pH 7.9), 10% glycerol, 5 mM MgCl₂, 60 mM KCl, 1 mM DTT, 0.5 mM EDTA, bovine serum albumin (50 mg/ml), 0.05% NP-40, 0.1 mg of poly(dI-dC), approximately 1 ng of proteins, and 5 fmol of radiolabeled DNA. Reaction mixtures were incubated for 10 min at room temperature and separated on 4% nondenaturing polyacrylamide gels. The gels were then dried and exposed to film.

YY1 is acetylated by p300 and PCAF.

Acetylation has been shown to regulate the activity of many transcription factors, including sequence-specific DNA-binding factors p53 (18), GATA-1 (5, 27), E2F (43, 44), and MyoD (53). Analysis of the YY1 amino acid sequence reveals that YY1 contains multiple lysine residues that are potential substrates for acetylation. Interestingly, within YY1's HDAC interaction domain (residues 170 to 200), there are six lysines arranged in pairs (Fig. 1). As described previously, YY1 interacts with the HAT p300 (1, 36), which interacts with another HAT, PCAF (76). Remarkably, both p300 and PCAF catalyze the acetylation of transcription factors (reviewed in reference 63). To determine if p300 or PCAF acetylated YY1, we performed in vitro acetylation reactions using GST-tagged p300 and PCAF purified from E. coli as enzymes. YY1 serial deletion constructs were similarly purified from E. coli as GST fusion proteins and used as substrates. GST-PCAF acetylated various deletion constructs of YY1 (Fig. 2A, lanes 1 to 5, 8 to 11, 13, 15, and 16) but not GST alone (lanes 12 and 17). Unlike PCAF, GST-p300 efficiently acetylated GST-YY1(170–200) as well as GST-YY1(1–200) (Fig. 2B, lanes 3, 12, and 15). GST alone was not acetylated by p300, showing that the in vitro acetylation was specific to YY1 (Fig. 2B, lane 1). Moreover, acetylation of YY1 was dependent on p300 or PCAF and was not due to YY1 auto-acetylation (Fig. 2C).

The arrangement of the six lysines within the p300 acetylation domain of YY1 closely resembles the sequence of the p300-acetylated region of histone H2B (63), suggesting that YY1 residues 170 to 200 contain an authentic p300 acetylation domain. Furthermore, the PCAF-interacting domain of p300 overlaps with its YY1-interacting domain (1, 76), which suggests that acetylation of YY1 by PCAF versus by p300 might be either cooperative or mutually exclusive in vivo, depending on whether or not YY1 binding to p300 can be competed by PCAF. Our data that full-length GST-YY1 was acetylated by PCAF and not by p300 (Fig. 2A, lanes 1, 11, and 13, and Fig. 2B, lanes 2, 7, and 19) further strengthen the suggestion that the conformation of YY1, perhaps affected by selective interaction with p300 and PCAF, is important for YY1 acetylation. Taken together, we found that PCAF acetylated YY1 at residues 170 to 200 and at its C terminus. p300, in contrast, efficiently acetylated YY1 only at residues 170 to 200 and only when YY1 was in a C-terminal truncated form, implying that p300-mediated acetylation of YY1 is dependent on the conformation of YY1.

To determine which region of YY1 was acetylated in vivo, F-YY1 deletion constructs were transiently transfected into HeLa cells and immunoprecipitated with anti-Flag antibody. Acetylated forms of F-YY1 were detected by Western blotting with anti-acetyl lysine antibody (Fig. 2D, top panel). To ensure that all F-YY1 deletions were expressed, blots were also probed with anti-Flag antibody (Fig. 2D, bottom panel). The results of these experiments show that YY1 is acetylated in vivo at residues 170 to 200 as well as in the C-terminal residues 261 to 414. Taken together, our in vitro and in vivo acetylation results suggest that there are two acetylation domains on YY1: residues 170 to 200, which are acetylated by both p300 and PCAF, and the C terminus, which is acetylated by PCAF only.

Because both p300 and PCAF acetylate YY1 at residues 170 to 200, we were interested in determining the specificity of acetylation by these two enzymes. Using GST-p300 or GST-PCAF, we in vitro acetylated a synthetic peptide corresponding to YY1 residues 170 to 200. We then compared the mass spectra produced by mass spectrometry. Figure 2E shows that mock-acetylated YY1 peptide had an M_r of 2,861 (left, 2861 m/z). When this peptide was acetylated with PCAF, another peak emerged with an M_r of 2,903 (right, 2903 m/z). Compared to the mock-acetylated peptide, this additional peak had a mass corresponding to one additional acetyl group (2903 − 2861 = 42), suggesting that the YY1 peptide was acetylated once by PCAF. Interestingly, when we compared the spectrum from the p300-acetylated peptide with that of the mock-acetylated peptide, we found three additional peaks at 2903, 2945, and 2987 m/z, which contained one, two, and three additional acetyl groups, respectively (Fig. 2F). This result strongly suggests that YY1 can be acetylated by p300 at three different lysines between residues 170 and 200.

Lysine-to-arginine mutations within YY1 residues 170 to 200 significantly reduce the transcriptional repression activity of YY1.

To understand the effect of acetylation on the transcriptional activity of YY1, we mutated the six lysines within YY1 residues 170 to 200 to arginines. Arginine substitutions preserve the charges of the affected amino acid residues but prevent acetylation in vivo by histone acetyltransferases. Both wild-type and mutant YY1 were fused to a Gal4 DNA-binding domain and transfected into HeLa cells in combination with a CAT reporter driven by the SV40 promoter containing five Gal4 binding sites. Consistent with previous findings (58, 73), wild-type Gal4-YY1 was a potent transcriptional repressor (Fig. 3, top panel, lanes 2 and 5; right panel). However, when the six lysines were mutated and no longer able to be acetylated, YY1 lost much of its repression activity (Fig. 3, top panel, lanes 3 and 6; right panel). This finding suggests that acetylation of YY1 is necessary for the maximum transcriptional repression activity of YY1. Western blot analysis showed that the loss of repression was not due to lack of expression of the mutant YY1 proteins (Fig. 3, bottom panel, compare lane 3 to lane 2 and lane 6 to lane 5). It is possible that lysine-to-arginine mutations result in a conformational change in YY1, which causes a lowered transcriptional repression activity. However, this would still suggest that the lysine residues within YY1 residues 170 to 200 are important for the full repression activity of YY1.

Acetylation of YY1 residues 170 to 200 increases YY1 binding to HDACs.

Using GST pull-down assays, we previously demonstrated that HDACs interact with YY1 residues 170 to 200 (73). Therefore, we asked if acetylation of YY1 in this region would affect YY1's interaction with HDACs. Figure 4 shows that [³⁵S]methionine-labeled HDAC1 bound more efficiently to PCAF-acetylated and p300-acetylated GST-YY1(170–200) than to unacetylated GST-YY1(170–200) (compare lane 1 to lane 2 and lane 3 to lane 4). The same amount of GST-YY1(170–200) was used in all reactions as shown by Coomassie staining. Similar results were obtained using in vitro-transcribed and -translated HDAC2 (data not shown). In short, our data suggest that acetylation of YY1 at residues 170 to 200 significantly increases the binding of YY1 to HDACs.

HDAC1 and HDAC2 deacetylate YY1 residues 170 to 200 but not the C-terminal region of YY1.

It is possible that the binding of HDACs to acetylated YY1(170–200) results in the subsequent deacetylation of YY1(170–200). To test this hypothesis, a synthetic peptide corresponding to YY1 residues 170 to 200 was chemically labeled with [³H]sodium acetate and used as a substrate in deacetylation assays. Figure 5A shows that immunopurified Flag-tagged HDAC1 and HDAC2 (F-HDAC1 and F-HDAC2) deacetylated the YY1(170–200) peptide. Deacetylation of YY1(170–200) by F-HDAC1 and F-HDAC2 was abolished by treatment with tricostatin A (TSA), a potent inhibitor of HDAC1 and HDAC2. Furthermore, if F-HDAC1 and F-HDAC2 were immunoprecipitated in the presence of an excess competitor, Flag peptide, deacetylase activity was not observed on the YY1 peptide. Similarly, endogenous HDAC1 and HDAC2 immunopurified from HeLa cells (HDAC1 and HDAC2) also deacetylated the YY1(170–200) peptide.

To provide further evidence that HDACs deacetylated YY1, we used an F-HDAC1 immunoprecipitate to deacetylate full-length GST-YY1 and GST-YY1(170–200), both of which were in vitro acetylated with PCAF. As expected, YY1(170–200) was efficiently deacetylated by HDAC1 (Fig. 5B, top panel, compare lane 1 to lane 2). As a control, YY1(170–200) was not deacetylated by an HDAC1 mutant that was devoid of deacetylation activity (22) (Fig. 5B, right panel and top panel, lane 3). A Coomassie-stained gel (bottom panel) showed that the same amount of YY1(170–200) was present in each reaction. Similarly, YY1(170–200) acetylated by p300 was also deacetylated by HDAC1 (Fig. 5B, compare lane 5 to lane 6). To our surprise, and in contrast to YY1(170–200), full-length YY1 was not appreciably deacetylated by HDAC1 (Fig. 5C, lane 1). This observation suggests that deacetylases only target specific regions of YY1. In support of this argument, we also found that HDAC1 did not deacetylate two additional PCAF-acetylated GST-YY1 proteins, YY1(261–333) and YY1(261–414) (Fig. 5C, lanes 5 and 7). We conclude that only residues 170 to 200 of YY1, and not the C-terminal zinc finger region of YY1, can be deacetylated by HDAC1 and HDAC2.

HDACs bind YY1 at multiple regions.

Based on our previous result that YY1 interacts with HDACs at residues 170 to 200 in vitro (73), the inability of HDAC1 to deacetylate the C-terminal zinc finger region of YY1 may arise from a general failure of HDACs to interact with YY1 in the zinc finger region. To test this possibility, we performed coimmunoprecipitation analyses to map the interaction domain of YY1 with HDACs in vivo (Fig. 6). Surprisingly, our results showed that in addition to residues 170–200, YY1 also interacted with HDACs at residues 261 to 333 in vivo. These results, together with those of our previous in vitro acetylation deacetylation experiment, suggest that YY1 interacts with HDACs at two domains: residues 170 to 200, where bound HDACs deacetylate YY1, and the C-terminal residues 261 to 333, where bound HDACs do not result in deacetylation of YY1.

YY1 contains associated HDAC activity, which localizes to C-terminal residues 261 to 333 of YY1.

After we precisely mapped the HDAC-binding domains in YY1, we sought to determine the functional effect of this interaction. We found that immunoprecipitated endogenous YY1 from HeLa cells also contained histone deacetylase activity, which was inhibited by TSA (Fig. 7A). Using serial F-YY1 deletions, we determined the histone deacetylase activity domain of YY1, which localized to residues 261 to 333 (Fig. 7B and C). This region of YY1 was necessary and sufficient for the HDAC activity associated with YY1 (Fig. 7B and C). Most strikingly, the YY1 histone deacetylase activity domain completely overlapped with one of the HDAC-interacting domains of YY1. Furthermore, the HDAC activity associated with YY1 residues 261 to 333 was highly specific, because the activity was sensitive to TSA (Fig. 7B) and competed by excess Flag peptide (Fig. 7B). A representative Western blot shows that different F-YY1 deletion mutants expressed equally well (Fig. 7D). Immunofluorescence analysis confirmed that F-YY1 deletion proteins that did not exhibit HDAC activity localized to either the nucleus or both the nucleus and cytoplasm, ruling out the possibility that the lack of histone deacetylase activity was due to abnormal localization of the mutants (Fig. 7E). These results strongly suggest that stable interaction between HDACs and YY1 contributes to YY1's histone deacetylase activity.

Acetylation of YY1 at the C-terminal zinc finger domain decreases the DNA-binding activity of YY1.

Because the C-terminal acetylation domain (residues 261 to 333) of YY1 overlaps with the zinc finger DNA-binding domain (residues 261 to 414) of YY1, we tested whether acetylation of YY1 at the zinc finger domain would affect the DNA-binding activity of YY1. Indeed, when in vitro acetylated by PCAF, GST-YY1 bound less avidly than did unacetylated GST-YY1 to the initiator element of the adeno-associated virus P5 promoter, which contains a consensus YY1 binding site (Fig. 8A, compare lane 7 to lane 8). Similar results were obtained when a GST-YY1 deletion construct that contained only the zinc finger domain of YY1 was tested for its DNA-binding properties (Fig. 8A, compare lane 5 to lane 6). In contrast, and consistent with earlier reports, acetylated GST-p53 bound to its recognition sequence better than unacetylated GST-p53 (Fig. 8A, compare lane 3 to lane 2). Thus, the decrease in DNA-binding activity caused by acetylation is specific to YY1. To rule out the possibility that this decrease in DNA-binding activity was associated with the GST fusion constructs, we tested nontagged YY1 purified from E. coli, as well as F-YY1 purified from HeLa cells. Both forms of YY1 exhibited decreased DNA-binding activity upon acetylation (Fig. 8B), proving that the DNA-binding activity of YY1 decreases when the zinc finger domain of YY1 is acetylated.