Autoacetylation Regulates P/CAF Nuclear Localization*

Acetylation is a posttranslational modification that alters the biological activities of proteins by affecting their association with other proteins or DNA, their catalytic activities, or their subcellular distribution. The acetyltransferase P/CAF is autoacetylated and acetylated by p300 in vivo. P/CAF autoacetylation is an intramolecular or intermolecular event. Intramolecular acetylation targets five lysines within the nuclear localization signal at the P/CAF C terminus. We analyzed how the subcellular distribution of P/CAF is regulated by intramolecular autoacetylation and found that a P/CAF mutant lacking histone acetyltransferase activity accumulated primarily in the cytoplasm. This cytoplasmic fraction of P/CAF is enriched for nonautoacetylated P/CAF. In addition, P/CAF deacetylation by HDAC3 and in a minor degree by HDAC1, HDAC2, or HDAC4 leads to cytoplasmic accumulation of P/CAF. Importantly, our data show that P/CAF accumulates in the cytoplasm during apoptosis. These results reveal the molecular mechanism of autoacetylation control of P/CAF nuclear translocation and suggest a novel pathway by which P/CAF activity is controlled in vivo.

p300/CBP-associated factor (P/CAF, also known as PCAF) was initially identified as a CBP/p300-binding protein, due to its sequence similarity to the yeast histone acetyltransferase (HAT) 2 GCN5 (yGCN5) (1). Both P/CAF and GCN5 belong to the GCN5 family of HATs and show sequence conservation in the regions responsible for HAT activity, the bromodomain and the ADA2 region (responsible for binding the yADA2 cofactor) (2,3). In addition, P/CAF activates transcription; to do that, it requires its intrinsic HAT activity (4,5). P/CAF is found in a complex comprising more than 20 polypeptides. Some of these polypeptides are similar to TATA box-binding protein-associated factors, whereas others contain a histone fold domain (6). P/CAF has an intrinsic ability to acetylate nucleosomal histones in vitro and in vivo and also acetylates nonhistone proteins, such as p53 (7), E2F (8,9), YY1 (10), NFB (11), SREBP (12), Ku70 (13), and Smad7, -2, and -3 (14,15). Acetylation affects the in vivo functions of these proteins. Moreover, P/CAF exists as an acetylated protein in vivo. P/CAF is autoacetylated and acetylated by p300 but not by CBP. P/CAF autoacetylation takes place via intramolecular and intermolecular mechanisms (16). Intramolecular acetylation targets five lysines in the nuclear localization signal (NLS) at amino acids 416 -442 of the C terminus. Autoacetylation leads to an increase in P/CAF HAT activity in vitro (16). Furthermore, P/CAF physically associates with HDAC1, SIRT1, and HDAC3 histone deacetylases (HDACs) (17)(18)(19). HDAC3 deacetylates P/CAF (17). This suggests that multiple HDACs target P/CAF and other factors that are not histones (17, 19 -22). In addition to its acetyltransferase activity, P/CAF has intrinsic ubiquitination activity, which is important in controlling Hdm2 protein levels (23). P/CAF may have a role in DNA damage response and apoptosis due to its ability to acetylate p53, E2F1, and Ku70 (8, 9, 13, 24 -27). Moreover, specific modifications of chromatin are likely to play an essential role in apoptosis induction. In support of this role, treatment with HDAC inhibitors induces apoptosis (28). Recent data have shown that during apoptosis, global changes in histone modifications occur (29 -34), including histone acetylation (34,35). At the same time, the activity of many HAT and HDAC enzymes is affected during apoptosis. For example, HAT CBP/p300 is cleaved during apoptosis in the central nervous system (36). More recently, it was shown that HDAC-4 and HDAC-3 nuclear-cytoplasmic shuttling promotes apoptosis (35,37).
Although P/CAF is a nuclear protein, its nuclear localization was reported to change in response to cellular or extracellular signals (17,38). However, the molecular mechanisms governing this transition were not known. Here, we report that P/CAF localization is regulated by intramolecular autoacetylation.
Indirect Immunofluorescence-Cells on coverslips were fixed in 4% paraformaldehyde in PBS for 30 min at room temperature and permeabilized with methanol for 10 min. After blocking with 3% bovine serum albumin in PBS and 0.1% Tween 20 for 1 h at room temperature, coverslips were incubated with a 1:200 -500 anti-P/CAF (Upstate Biotechnology, Inc.), 1:500 anti-acetylated histone H3 (Upstate Biotechnology), and/or 1:100 anti-FLAG (Upstate Biotechnology) antibodies in PBS and 3% bovine serum albumin for 2 h. This was followed by incubation for 1 h with goat anti-rabbit IgG, used at a 1:250 dilution in PBS and 3% bovine serum albumin. After each antibody incubation, the coverslips were washed four times for 10 min each time with PBS and 0.05% Tween 20 at room temperature. Cells fixed in 4% paraformaldehyde in PBS for 30 min at room temperature were incubated with fluorescent annexin V (Roche Applied Science) for 20 h. The nuclei were stained with DAPI. To distinguish endogenous and ectopic, transfected P/CAF immunofluorescence signals, different laser intensities were used at the confocal microscope. In addition, transfected cells were followed by GFP coexpression (see supplemental Fig. 1).
Cell Extract Preparation and Immunoblotting and Immunoprecipitation Assays-Total cell extracts were prepared in IPH buffer (40) by keeping the cells on ice for 20 min or for the indicated time, followed by centrifugation at 12,000 ϫ g for 10 min at 4°C. All buffers contained protease inhibitors (Roche Applied Science). Immunoblotting was performed using standard procedures and visualized using an ECL kit (Amersham Biosciences). The antibodies were acetyl-Lys antibody (Chemicon) used at a 1:1,000 dilution, anti-P/CAF antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) used at a 1:1,000 dilution, and anti-FLAG M2 antibody (Upstate Biotechnology) used at a 1:100. Immunoprecipitation analyses were performed as described elsewhere (41), except for the in vivo acetylation detection and immunoprecipitation and HAT assays, in which the IPH buffer was changed for radioimmune precipitation buffer.
HAT and Acetyltransferase Assays-In vitro HAT and acetyltransferase assays were performed as described elsewhere (4,8). P/CAF autoacetylation was performed under the same conditions as HAT assays but in the absence of histones (16).
In Vitro Translation, Recombinant Proteins, and Pull-down Assay-In vitro translations and GST pull-downs were performed essentially as described previously (41). GST and GST fusion proteins were expressed in Escherichia coli XA90 using the pGEX (Amersham Biosciences) vector system for 4 h at 30°C after 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside was added. Purification from crude bacterial lysates was performed as in Ref. 41. The buffer used for the pull-downs was a variation of ZЈ (25 mM HEPES, pH 7.5, 12.5 mM MgCl 2 , 20% glycerol, 0.1% Nonidet P-40, and 250 mM KCl).

P/CAF Lacking HAT Activity Localizes to the Cytoplasm-
Previous work has shown that amino acids 428 -442 of P/CAF contain the NLS. Mutations of lysines 428, 430, 441, and 442 to arginines leads to P/CAF cytoplasmic accumulation (16). These lysines are the targets of intramolecular autoacetylation (16). However, mutations of the NLS lysines are not direct proof of the autoacetylation requirement, since the lysines themselves are essential for NLS function. Here, we investigated the role of P/CAF intramolecular autoacetylation on P/CAF subcellular distribution using a HAT-domain mutant. P/CAF-(352-832)⌬HAT cannot be autoacetylated (Fig. 1, A and C), because it lacks the N-terminal part of the protein, the region required for intermolecular acetylation (16) and HAT catalytic activity (Fig. 1B). HAT catalytic activity is required for intramolecular autoacetylation (Fig. 1C). However, P/CAF-(352-832)⌬HAT retains the NLS (Fig. 1A). First, we analyzed the cellular localization of P/CAF-(352-832) and P/CAF-(352-832)⌬HAT. To do this, we transfected CV1COS cells with either P/CAF-(352-832) or P/CAF-(352-832)⌬HAT. The cellular distribution of P/CAF was analyzed by immunostaining using an antibody that specifically recognizes P/CAF. P/CAF-(352-832) signal was found exclusively in the nucleus, whereas P/CAF-(352-832)⌬HAT was located predominantly in the cytoplasm (Fig. 1D).
To confirm these results, we constructed a P/CAF HAT mutant (L606A) (Fig. 1A) that had only a limited ability to acetylate histones, and it is not able to autoacetylate (Fig. 1, B and C). We then tested the cellular localization of this mutant. CV1COS cells were transfected either with P/CAF-(352-832) or with P/CAF-(352-832)L606A, and their localizations were revealed by immunostaining using an antibody that specifically recognizes P/CAF. As expected, P/CAF-(352-832) was nuclear and P/CAF-(352-832)L606A was distributed in both the nucleus and the cytoplasm (Fig. 1D). These results suggest that P/CAF autoacetylation might be involved in its cellular distribution.
To determine whether this result was specific to the CV1COS cell line used in these experiments, we examined the subcellular localization of P/CAF-(352-832) and P/CAF-(352-832)⌬HAT in different cell lines. P/CAF-(352-832) was found mainly in the nuclei in all analyzed cell types (Table 1). In contrast, varying levels of the P/CAF⌬HAT mutant were distributed between the nucleus and the cytoplasm (Table 1). Taken together, these results suggest that P/CAF requires HAT activity and the NLS to localize in the nucleus.
We next tested whether intramolecular autoacetylation might regulate the localization of full-length P/CAF. To do this, we transfected CV1COS cells either with P/CAF full-length (P/CAF-FL) or with P/CAF⌬HAT-FL and revealed their localization by immunostaining using an antibody that specifically recognizes P/CAF. As expected, P/CAF-FL was mainly nuclear (68% of the analyzed cells), and P/CAF⌬HAT-FL was partially cytoplasmic (57% of the analyzed cells were cytoplasmic, 28% were distributed between nucleus, and cytoplasm and 15% were nuclear) (Fig. 1E). We next examined the subcellular localization of P/CAF-FL and P/CAF⌬HAT-FL in different cell lines (HeLa, NIH3T3, and 293T). P/CAF-FL was found mainly in the nuclei in all analyzed cell types. In contrast, P/CAF⌬HAT-FL was distributed between the nucleus and the cytoplasm at levels similar to those of P/CAF-(352-832)⌬HAT (data not shown). These results indicate that P/CAF autoacetylation might be involved in its cellular distribution. We performed some experiments using P/CAF-(352-832)⌬HAT or P/CAF⌬HAT-FL fused to GFP and FLAG epitopes at the N terminus of the protein. Surprisingly, in these cases, P/CAF-(352-832)⌬HAT was enriched in the nucleus (data not shown), probably due to (i) a conformational change in P/CAF protein induced by the epitopes or (ii) a change in the accessibility of the P/CAF NLS in the fusion proteins. P/CAF Nuclear Localization Correlates with Autoacetylation-Since HAT activity is required for P/CAF nuclear accumulation, we next sought to confirm that P/CAF nuclear localization correlates with autoacetylation. To that end, we transfected P/CAF-(352-832), P/CAF-(352-832)⌬HAT, and P/CAF-   JANUARY 16, 2009 • VOLUME 284 • NUMBER 3

JOURNAL OF BIOLOGICAL CHEMISTRY 1345
(352-832)L606A into CV1COS cells. The transfected proteins were recovered by immunoprecipitation, and their acetylation status was analyzed by immunoblot using an antibody that specifically recognizes acetylated lysines. Only nuclear P/CAF was acetylated in vivo ( Fig. 2A, lane 1). In contrast, cytoplasmic P/CAFs lacking HAT activity were not acetylated in vivo ( Fig.  2A, lanes 2 and 3). We then treated P/CAF-(352-832)-and P/CAF-(352-832)⌬HAT-transfected cells with varying amounts of the HDAC inhibitor TSA. Subsequently, the P/CAF location was analyzed by immunostaining. As expected, TSA did not modify the P/CAF cellular distribution (Fig. 2B), probably because cytoplasmic P/CAF-(352-832)⌬HAT is a nonacetylated protein and therefore is not affected by changes in HDAC activity.
Proteasome Inhibition Partially Reverses P/CAF Cytoplasmic Accumulation-Once we had confirmed that P/CAF acetylation correlates with localization to the nucleus, we sought to understand the molecular mechanisms responsible for this cellular distribution.
Autoacetylation could affect the localization of P/CAF-(352-832) by affecting nuclear import or nuclear export. The recognition of NLS-containing proteins in the cytoplasm is mediated by the heterodimeric importin-␣/␤ receptor, in which the importin ␣ subunit is the saturable component of the receptor, and the importin ␤ targets the NLS-containing protein in the nuclear pore complex (42,43). Thus, we asked whether P/CAF-(352-832) directly bound importin ␣ or ␤ and whether autoacetylation affected this interaction in vitro. As shown in Fig.  3A, GST-P/CAF-(352-832) interacted with importin-␣1 in an in vitro pull-down assay. Moreover, P/CAF-(352-832) bound to GST-importin-␤ in vitro (Fig. 3B). In addition, P/CAF-(352-832) autoacetylation decreased these interactions (Fig. 3, A  (lanes 3 and 4) and B (lane 3)). To confirm these results, we transfected CV1COS cells with P/CAF or P/CAF⌬HAT, prepared whole cell extracts from the transfected cells, and used them for pull-down experiments with GST, GST-importin ␣1, and GST-importin ␤ columns. Both P/CAF and P/CAF⌬HAT interacted efficiently with importin ␣1 and GST-importin ␤ (Fig. 3C). These results suggest that nuclear import was not the cause of the observed P/CAF cellular distribution, probably because cytoplasmic P/CAF, which recognized the het- A, immunoprecipitation (IP) of P/CAF, P/CAF⌬HAT, P/CAF-L606A, or vector alone from transfected CV1COS cells. P/CAF proteins were immunoprecipitated using an anti-P/CAF antibody, and immunocomplexes were tested for P/CAF acetylation status by immunoblot using antibodies that recognize acetylated lysines (Anti-AcLys; top blot) and P/CAF (Anti-P/CAF; bottom blot). B, CV1COS cells were transfected as in A and treated with 0, 100, or 150 nM TSA for 18 h. Twenty-four hours after transfection, the localization of the expressed proteins was visualized by immunofluorescence staining using anti-P/CAF antibody. Cells were costained with DAPI to reveal DNA. The results shown are representative of at least three independent experiments. erodimeric importin-␣/␤ receptor, was always nonacetylated, as we showed in Fig. 2.
The above results suggest that cytoplasmic P/CAF is efficiently imported into the nucleus. What happens to nonacetylated nuclear P/CAF? The results shown in Fig. 4A suggest that it is not actively exported to the cytoplasm. Thus, nuclear nonacetylated protein may be degraded by proteolytic processing. We therefore tested whether the proteasome inhibitor MG132 could reverse the cytoplasmic localization of P/CAF⌬HAT. To this end, we transfected CV1COS cells with P/CAF-(352-832) or P/CAF-(352-832)⌬HAT and then treated them with MG132, which blocks proteasomemediated degradation. The P/CAF localization was determined by immunostaining with an antibody that specifically recognizes P/CAF. Blocking protein degradation induced the accumulation of P/CAF in promyelocytic leukemia protein bodies (data not shown). However, it did not significantly alter the nuclear/cytoplasmic distribution of P/CAF (Fig. 4B). In contrast, MG132 treatment allowed a partial accumulation of P/CAF-(352-832)⌬HAT in the nucleus; the protein was distributed between the nucleus and cytoplasm in 70% of the cells (Fig.  4B). Under these conditions, nonautoacetylated P/CAF also accumulated in the promyelocytic leukemia protein and in other cell bodies (data not shown). P/CAF Deacetylation by HDAC3 Promotes P/CAF Cytoplasmic Accumulation-Our results thus far suggested that autoacetylation was important in promoting the nuclear localization of P/CAF. On the other hand, it had previously been shown that some HDACs bind P/CAF (17)(18)(19) and that HDAC3 inhibits P/CAF autoacetylation (17). Therefore, we tested whether deacetylation of P/CAF by HDAC1, HDAC2, HDAC3, HDAC4, or SIRT1 affected P/CAF nuclear localization. To do this, we transfected CV1COS cells with P/CAF-(352-832) in the presence or absence of coexpressed FLAG-HDAC1, HDAC2, FLAG-HDAC3, HDAC4, or SIRT1. The cellular localization of P/CAF was then analyzed by immunostaining with antibodies specific to P/CAF. Although HDAC1, -2, and -4 caused P/CAF localization to switch from primarily nuclear to both nuclear and cytoplasmic, only coexpression of HDAC3 efficiently promoted the cytoplasmic localization of P/CAF (Fig. 5A). The specific effect of HDAC3 was confirmed by using shRNA, which partially blocks the expression of HDAC3 (shRNA HDAC3) (Fig. 5B) and control shRNA (shRNA C). The results in Fig. 5B show that in the presence of low levels of HDAC3, P/CAF cytoplasmic accumulation decreases. These results were consistent with our idea that intramolecular acetylation is required for P/CAF nuclear localization. This transition correlated with P/CAF deacetylation (Fig. 5C, lane 2). To further confirm that the observed effect on P/CAF localization was due to deacetylation, we added the HDAC inhibitor TSA to the transfected cells 10 h before collection. We then analyzed the P/CAF localization. HDAC3 in the presence of TSA could not promote the P/CAF cytoplasmic transition (Fig. 5D). This correlated with high levels of P/CAF autoacetylation in vivo ( Fig. 5D, lane 5). Finally, we sought to establish whether cytoplasmic P/CAF was active. To that end, we transfected CV1COS cells with P/CAF-(352-832) in the presence or absence of coexpressed FLAG-HDAC3. Then we immunoprecipitated P/CAF and determined the HAT activity associated with the immunopellets in an in vitro HAT assay. The results show that both nuclear and cytoplasmic P/CAF were capable of acetylating histones, although a slight decrease in HAT activity was observed in cytoplasmic nonautoacetylated P/CAF (Fig. 5E), in agreement with previous results (16). We next tested whether HDAC3 expression might promote cytoplasmic accumulation of full-length P/CAF. To do this, we transfected CV1COS cells with P/CAF-FL and HDAC3 and revealed their localization by immunostaining using an antibody that specifically recognizes P/CAF and FLAG epitope. As expected, coexpression of HDAC3 pro-moted the cytoplasmic or nuclear and cytoplasmic localization of P/CAF (64% of the analyzed cells) (Fig. 5F). On the other hand, HDAC3 in the presence of TSA could not induce P/CAF cytoplasmic accumulation (Fig. 5F).
Of the assayed HDACs (HDAC1, HDAC2, HDAC3, HDAC4, and SIRT1), HDAC3 was the enzyme that more efficiently altered P/CAF cellular localization (Fig. 5A). It had been shown previously that HDAC3 binds P/CAF in C2C12 cells (17). We sought to establish whether the P/CAF-HDAC3 interaction also took place under our experimental conditions. Thus, we transfected CV1COS cells with P/CAF-(352-832) and/or FLAG-HDAC3 DNA plasmids and analyzed the localization of P/CAF and HDAC3 by immunostaining with antibodies against P/CAF and the FLAG epitope. The proteins colocalized when they were expressed together (Fig. 6A). To confirm this result, we immunoprecipitated FLAG-HDAC3 and analyzed for the presence of P/CAF in the immunopellet by immunoblot. The results indicate that under our experimental conditions, P/CAF and HDAC3 interacted in vivo (Fig. 6B, lane  3). The same results were obtained using P/CAF-FL (data not shown).
Apoptosis Induces P/CAF Cytoplasmic Accumulation-Although P/CAF is known to be a nuclear protein (there are, however, a few examples of P/CAF cytoplasmic localization) (see Refs. 17 and 38), we investigated whether endogenous P/CAF would localize to the cytoplasm under different conditions. It has been proposed that P/CAF may have a role in apoptosis and DNA damage response due to its ability to acetylate p53, E2F1, and Ku70 (9, 13, 24 -27). Thus, we investigated whether P/CAF could localize in the cytoplasm during apoptosis. To do this, we UV-irradiated CV1COS cells. We then determined endogenous P/CAF localization by immunostaining with an antibody that specifically recognizes P/CAF. Apoptosis induction was followed by staining with fluorescent annexin V. In apoptosis, endogenous P/CAF partially accumulated in the cytoplasm (Fig. 7A). Then we analyzed P/CAF localization throughout the apoptotic progression. To do that, endogenous P/CAF localization was analyzed at different times after UV irradiation. The results showed that 10 min after UV irradiation, P/CAF is clearly distributed between the nucleus and cytoplasm. After 30 min, P/CAF levels increased, and it was still localized between the nucleus and cytoplasm. Finally, after 1 h, P/CAF is distrib-uted between the nucleus and cytoplasm and in most condensed cells is completely cytoplasmic (Fig. 7, A and B, see  arrowheads). Similar results were obtained when apoptosis was induced by treatment with dexamethasone ( Fig. 7, C and D). These results suggest that apoptosis induces P/CAF cytoplasmic accumulation. FIGURE 5. P/CAF deacetylation by HDAC3 promotes P/CAF cytoplasmic accumulation. A, CV1COS cells were transfected with P/CAF and one of the following: FLAG-HDAC1, HDAC2, FLAG-HDAC3, HDAC4, or SIRT1. Localization of the expressed proteins was visualized by immunofluorescence staining using anti-P/CAF antibody. The bar graphs show the percentage of cells with predominantly nuclear (N) or cytoplasmic (C) localization or both (N ϩ C) for over 300 cells from four independent experiments. The expression levels of FLAG-HDAC1, HDAC2, FLAG-HDAC3, HDAC4, and SIRT1 were similar (data not shown). B, CV1COS cells were transfected with P/CAF, FLAG-HDAC3, shRNA HDAC3, or shRNA control (shRNA C). Localization of the expressed P/CAF and FLAG-HDA3 proteins was visualized by immunofluorescence staining using anti-P/CAF antibody and anti-FLAG antibodies, respectively. Cells were costained with DAPI to reveal DNA. C, CV1COS cells were transfected with P/CAF (lane 1), P/CAF and FLAG-HDAC3 (lane 2), or FLAG-HDAC3 alone (lane 3). P/CAF was immunoprecipitated (IP) using an anti-P/CAF antibody. The immunocomplexes were tested for P/CAF acetylation status by immunoblot using antibodies that recognize acetylated lysines (anti-AcLys; top blot) and P/CAF (anti-P/CAF; bottom blot). D, CV1COS cells were transfected as in B and treated with 0 or 150 nM TSA. P/CAF localization was visualized by immunofluorescence staining using anti-P/CAF antibody. TSA activity was followed by immunofluorescence staining using anti-Ac-H3 antibody (increased Ac-H3 levels are indicated by white arrowheads). Cells were costained with DAPI to reveal DNA. P/CAF proteins were immunoprecipitated using an anti P/CAF antibody. The immunocomplexes of P/CAF were tested for P/CAF acetylation status by immunoblot, using antibodies that recognize acetylated lysines (anti-Ac-Lys; top blot) and P/CAF (anti-P/CAF; bottom blot). The results shown are representative of at least three independent experiments. E, CV1COS cells were transfected with P/CAF and/or FLAG-HDAC3. P/CAF proteins were immunoprecipitated using an anti-P/CAF antibody in radioimmune precipitation buffer. The immunocomplexes were tested for P/CAF HAT activity in an in vitro HAT assay and for the presence of P/CAF protein by immunoblot analysis. F, CV1COS cells were transfected with P/CAF-FL and FLAG-HDAC3 in the presence or absence of 150 nM TSA for 12 h. Localization of the expressed P/CAF and FLAG-HDA3 proteins was visualized by immunofluorescence staining using anti-P/CAF and anti-FLAG antibodies, respectively. Cells were costained with DAPI to reveal DNA. FIGURE 6. P/CAF binds HDAC3 to promotes P/CAF deacetylation. A, CV1COS cells were transfected with P/CAF and/or FLAG-HDAC3, and the localization of the expressed proteins was analyzed by immunofluorescence staining using anti-P/CAF and anti-FLAG antibodies. Cells were costained with DAPI to visualize DNA. B, CV1COS cells were transfected as in A. P/CAF was immunoprecipitated (IP) using an anti-P/CAF antibody. The immunocomplexes were tested for the presence of FLAG-HDAC3 protein by immunoblot using antibodies against P/CAF (top blot) and the FLAG epitope (bottom blot).

DISCUSSION
The results show that P/CAF intramolecular acetylation is required for P/CAF nuclear localization. Lysines are essential components of NLS (45) and are often acetylated. It has been proposed that lysine acetylation regulates the nuclear/cyto-plasmic distribution of several proteins (for reviews, see Refs. 46 and 47). This modification could regulate NLS function by affecting NLS interactions with nuclear import machinery. Alternatively, acetylation within the NLS could induce conformational changes that affect NLS or nuclear export signal functions and alter protein subcellular distribution, as has been described for several proteins (48 -54). Our results indicate that P/CAF NLS acetylation does not increase the ability of P/CAF to recognize importin machinery. In addition, blocking nuclear export did not affect either P/CAF-(352-832) (nuclear) or P/CAF-(352-832)⌬HAT (cytoplasmic) localization. This suggests that some additional mechanisms (probably related to protein degradation) are responsible for the observed P/CAF intracellular distribution.
Although P/CAF is a histone acetyltransferase, like CBP and p300, it has also been recognized as a critical regulator of nonhistone proteins involved in many cellular processes, such as differentiation and apoptosis (55). In yeast, loss of the P/CAF homologous Gcn5l2 leads to high levels of apoptosis (56). The apoptosis regulators p53, p73, and E2F1 have been shown to be regulated by P/CAF acetylation (9, 24 -27, 57) in response to DNA damage. In addition, P/CAF associates with Ku70 in vivo, acetylates its C-terminal linker, and blocks the ability of Ku70 to suppress Bax-mediated apoptosis, indicating that acetylation negatively regulates the antiapoptotic function of Ku70 (13). Interestingly, the Ku70 pool responsible for Bax sequestration is cytoplasmic (58), suggesting that P/CAF acetylation takes place in the cytoplasm. Relocalization of P/CAF from the nucleus to the cytosol following cellular damage might be a key regulatory step in Bax-mediated apoptosis. Our results provide evidence for a molecular mechanism that regulates P/CAF subcellular redistribution by autoacetylation following DNA damage. Additional experiments will be needed to determine what signals regulate P/CAF autoacetylation FIGURE 7. Apoptosis induces P/CAF cytoplasmic accumulation. A, CV1COS cells were irradiated with 200 J/m 2 UV for 0 or 60 min. Localization of endogenous P/CAF was analyzed by immunofluorescence staining using anti-P/CAF antibody. Cells were costained with DAPI to visualize DNA and with fluorescent annexin V to follow apoptotic cells. B, cells were treated as in A, and P/CAF localization was analyzed at different times. C, CV1COS cells were treated with 2 M dexamethasone for 0 or 60 min. Localization of endogenous P/CAF was analyzed by immunofluorescence staining using anti-P/CAF antibody. Cells were costained with DAPI to visualize DNA (blue) and with fluorescent annexin V to follow apoptotic cells. D, cells were treated as in C, and P/CAF localization was analyzed at different times. The results shown are representative of at least three independent experiments. The arrowheads mark cells with complete P/CAF cytoplasmic localization. under proapoptotic conditions. It is possible that P/CAF interacts with the apoptotic regulator PKC␦, which has been shown to inhibit the HAT activity of p300 (59). It should also be noted that cytoplasmic relocalization of HDAC3 is important for apoptosis progression (35).
In addition to a possible P/CAF cytoplasmic function, the ability of cytoplasmic P/CAF to interact with the transcriptional machinery in the nucleus is reduced. It would be interesting to address whether the cytoplasmic localization of nonautoacetylated P/CAF has a function, such as repressing transcription by potentially sequestering positive transcriptional components. Alternatively, it may be part of a mechanism that prevents P/CAF outside of the nucleus from taking part in any further action, such as global histone H3 acetylation/deacetylation, after apoptosis is induced (34,35).