Acetylation by p300 Regulates Nuclear Localization and Function of the Transcriptional Corepressor CtBP2*

CtBP family members, CtBP1 and CtBP2, are unique transcriptional regulators that adapt a metabolic enzyme fold, and their activities are regulated by NAD(H)-binding. CtBP1 is both cytoplasmic and nuclear, and its subcellular localization is regulated by sumoylation, phosphorylation, and binding to a PDZ protein. In contrast, we showed that CtBP2 is exclusively nuclear. CtBP1 and CtBP2 are highly similar, but differ at the N-terminal 20 amino acid region. Substitution of the N-terminal domain of CtBP1 with the corresponding CtBP2 domain confers a dominant nuclear localization pattern to CtBP1. The N-terminal domain of CtBP2 contains three Lys residues. Our results show that these Lys residues are acetylated by the nuclear acetylase p300. Although all three Lys residues of CtBP2 (Lys-6, Lys-8, and Lys-10) appear to be acetylated, acetylation of Lys-10 is critical for nuclear localization. CtBP2 with a single amino acid substitution at Lys-10 (K10R) is predominantly localized in the cytoplasm. The cytoplasmic localization of the K10R mutant is correlated with enhanced nuclear export that is inhibited by leptomycin B. Furthermore, lack of acetylation at Lys-10 renders CtBP2 to be more efficient in repression of the E-cadherin promoter. Our studies have revealed the important roles of acetylation in regulating subcellular localization and transcriptional activity of CtBP2.

The CtBP (E1A C-terminal binding protein) family proteins are highly conserved in higher eukaryotes and are linked to important biological processes (1,2). The vertebrate genomes contain two different genes that code for two highly related proteins, CtBP1 (3,4) and CtBP2 (5,6), whereas genomes of invertebrates such as Drosophila and Caenorhabditis elegans contain a single Ctbp gene. Studies with mutant mice with targeted inactivation of the Ctbp genes suggest that they play important roles in multiple developmental processes during mouse development (7). In addition to CtBP1 and CtBP2, mammals also express an isoform of CtBP1 termed CtBP1-S/BARS (8) and an isoform of CtBP2 designated as RIBEYE (9) that are generated by differential RNA splicing and promoter utilization. The vertebrate CtBPs (6, 10 -13) and the Drosophila homolog, dCtBP (14,15), function as transcriptional corepressors. Consistent with the high sequence similarity between vertebrate CtBP1 and CtBP2, they exhibit overlapping transcriptional regulatory activities during mouse development (7). The CtBP corepressors are recruited by various repressors through a conserved CtBP-binding motif (PXDLS) that was first identified in adenovirus E1A proteins (4).
Unlike other transcription factors, CtBPs share a high degree of sequence homology with a group of metabolic enzymes of the 2-hydroxy acid dehydrogenase family (4). The crystal structure of CtBP1 has revealed that it is indeed a 2-hydroxy acid dehydrogenase (16,17). Consistent with the structural analyses, CtBP1 also possesses a slow dehydrogenase activity (16,18). At present, the relationship between the intrinsic dehydrogenase activity of CtBP and the transcriptional repression activity is not clear and remains controversial (6,16,19). However, binding of CtBP1 with NAD(H) promotes its oligomerization and enhances the affinity for PXDLS-containing transcription factors (18,20) and has been postulated to link the cellular metabolic status to transcriptional regulation (21). The dinucleotide binding activity of dCtBP has been shown to be critically important for the transcriptional repression activity in a native chromatin environment (22).
Although the transcriptional regulatory mechanisms of CtBPs remain to be fully elucidated, an analysis of a nuclear CtBP1 protein complex has revealed that CtBP may effect its transcriptional repression activity at least in part by recruitment of other enzymatic components, such as histone deacetylases and a histone methyl transferase (23) and a novel lysine-specific demethylase (24). CtBP1 has also been reported to inhibit the general transcriptional machinery through direct interaction with nuclear acetylases such as p300 via a PXDLS motif located within the bromodomain of these enzymes (25).
In addition to the nuclear transcriptional regulatory activities, CtBPs also have certain diverse cytoplasmic activities (26). For example, CtBP1-S, which is identical to CtBP1 except for the absence of the N-terminal 12 amino acid region, has been implicated in Golgi membrane fission in in vitro studies (8,(27)(28)(29). It appears that the membrane fission activity of CtBP1-S might be related to a slow acyltransferase activity associated with CtBP1/CtBP1-S (17,28). However, the existence of an intrinsic acyltransferase activity of CtBP1-S remains controversial (30). In the retina, the photoreceptor ribbon synaptic complex contains RIBEYE (9) and CtBP1 (31). RIBEYE and CtBP1 are important for the assembly and function of central nervous system synapses. Thus, CtBP family proteins in addition to their dominant transcriptional regulatory activity also perform diverse activities in the cytoplasmic compartment.
The regulation of subcellular localization of CtBP1 has recently been examined in some detail. CtBP1 have been reported to be a target for the p21-activated protein kinase (Pak) 1 (32). Pak1 specifically phosphorylates CtBP1 at Ser-158 resulting in CtBP1 redistribution from the nucleus to the cytoplasm, thereby blocking the corepressor function of CtBP1 under specific growth conditions. Although sequences of CtBP2 are conserved at the Pak1 phosphorylation site, it is not known whether CtBP2 is a target for Pak1. The polycomb protein Pc2 forms a specific complex with CtBP1 (13,23,33). The human Pc2 recruits CtBP1 and Ubc9 to the PcG bodies (34) resulting in sumoylation of CtBP1 at a single Lys (Lys-428) residue (34,35). The SUMO modification of CtBP1 appears to be critical for its nuclear accumulation. CtBP1 is also subject to an additional regulatory process that governs its nucleocytoplasmic distribution. The C-terminal region of CtBP1 has a PDZ-binding motif (1), which mediates interaction with a PDZ domain containing protein, neuronal nitric-oxide synthase, changing the localization pattern of CtBP1 from nuclear to cytoplasmic (36). Thus, sumoylation and PDZ binding exert opposing effects on subcellular localization of CtBP1 (35). In contrast to CtBP1, CtBP2 is not sumoylated (34,35) and does not contain the PDZ-binding motif, suggesting that the regulation of subcellular localization of CtBP might be different from that CtBP1. In this report, we show that CtBP2 has a dominant nuclear localization pattern, which is dictated by the N-terminal 20 amino acid domain. Further, we demonstrate that CtBP2 is acetylated by p300 at Lys-6, Lys-8, and Lys-10 and that acetylation of Lys-10 is critically important for nuclear retention of CtBP2.

EXPERIMENTAL PROCEDURES
Expression Constructs-The mammalian expression vector pIRES-neo3 (Clontech) was modified to express Flag-HA 2 tag at the N terminus of a protein of interest. A double-stranded oligonucleotide encoding the Flag-HA epitopes was ligated to the StuI/EcoRI sites of pIRESnes3 to generate pFH-IRESneo vector. All CtBP1 and CtBP2 expression constructs were based on pFH-IRESneo, using PCR-amplified coding sequences. All PCR-amplified regions in the clones were sequence-confirmed by automatic DNA sequence analysis.
Cell Culture and Transfection-HeLa and A549 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and penicillin/streptomycin. Transfections were per-formed with the Lipofectamine 2000 reagent (Invitrogen) following the manufacturer's recommendations. Typically, 2.5 ϫ 10 5 cells were plated in each well of a 6-well plate 1 day prior to transfection, and DNA precipitates were prepared using 3 g of a plasmid DNA and 7 l of the Lipofectamine 2000 reagent unless indicated otherwise. For siRNA experiments, HeLa cells were first transfected with SmartPool siRNAs (Dharmacon) with Dharmafect 1 (Dharmacon) and 24 h later were transfected with the CtBP2 expression plasmid using Lipofectamine 2000 reagent. Protein analyses were carried out 24 h after the second transfection. For E-cadherin promoter assays, the reporter plasmid pE-Cad-Luc and promoterless phRL-0 were cotransfected with different CtBP-expression plasmids into CtBP1 Ϫ/Ϫ /CtBP2 Ϫ/Ϫ mouse embryonic fibroblasts (immortalized with retrovirally introduced SV40 T-antigen), using the jetPEI transfection reagent (ISC BioExpress). Dual luciferase assays were performed with the Dual Luciferase assay kit (Promega) 2 days after transfection.
Immunoprecipitation and Western Blot Analyses-Cells were lysed for 30 min at 4°C in immunoprecipitation (IP) buffer containing 0.2% Nonidet P-40, 20 mM Hepes (pH 7.5), 150 mM NaCl, 10 mM KCl, 0.5 mM EDTA, 10 mM Na Butyrate, 10% glycerol, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride and the protease inhibitor mixture (Roche Applied Science). The cell lysates were clarified by centrifugation and bound to 30 l of FLAG (Sigma) antibody beads at 4°C for 1.5 h in IP buffer. After washing three times, immunoprecipitated proteins were examined by Western blot using the ECL-plus (Amersham Biosciences) detection method.
Immunofluorescence Analysis-For staining ectopically expressed proteins, transfected (transiently or stably) cells in 6-well plates were fixed with 3.7% formaldehyde/phosphate-buffered saline for 10 min at room temperature and then permeablized with methanol for 6 min at Ϫ20°C. After a brief rinse with phosphate-buffered saline, cells were incubated with Cy3-conjugated FLAG M2 antibody (Sigma) in a humidified chamber for 1 h at 37°C. Cells were washed with phosphatebuffered saline and photographed under a Nikon fluorescent microscope with a digital camera. The endogenous CtBP proteins were stained with CtBP1 or CtBP2 mAbs (Pharmingen), with a secondary goat anti-mouse antibody conjugated to rhodamine (Pierce).

RESULTS
CtBP2, But Not CtBP1, Is Localized Exclusively in the Nucleus-Because CtBP2 is not sumoylated and lacks the PDZ-binding motif, we hypothesized that subcellular localization of CtBP2 might be regulated differently from CtBP1. To investigate this aspect, we first examined the subcellular localization of CtBP2 in comparison with CtBP1 by indirect immunofluorescence analysis. The specificities of the antibodies used in these studies were confirmed by their differential reactivity toward CtBP1 or CtBP2 by Western blot analysis of N-terminally tagged versions (Flag-HA) of CtBP1 or CtBP2 immunoprecipitated from stably transfected cell lines (Fig. 1A). The CtBP1 antibody recognized only CtBP1, and the CtBP2 antibody recognized only CtBP2. A minor band revealed in the opposing lane might be because of heterodimerization of endogenous CtBP proteins with the exogenously expressed CtBP proteins. For the immunofluorescence analysis of endogenous CtBP2, two human cell lines, HeLa and A549, were stained with antibodies specific to CtBP1 or CtBP2. These studies revealed striking exclusive nuclear localization of CtBP2 in both HeLa and A549 cells (Fig. 1B), whereas CtBP1 displayed predominant nuclear staining with partial cytoplasmic staining (Fig. 1C). To substantiate the localization patterns further, we examined exogenously introduced CtBP1 and CtBP2 (tagged with FLAG and HA epitopes) in stable HeLa cell lines (Fig. 1D). Immunofluorescence analysis with the FLAG antibody revealed that the patterns of subcellular localization of the ectopically expressed CtBP1 and CtBP2 are similar to that of the endogenous CtBP1 and CtBP2 (compare Fig. 1, B and C with D). We also examined the localization patterns of transiently expressed CtBPs (Fig. 1E), which were found to be highly similar to patterns of endogenous CtBPs and the patterns of CtBPs expressed in stably transfected cell lines (Fig. 1D). Thus, our analyses using three different approaches have shown exclusive nuclear localization of  CtBP2 and nuclear and cytoplasmic localization of CtBP1. A similar localization pattern of CtBP1 has been reported by several previous studies (34,35).
Nuclear Targeting Function of the N-terminal Region of CtBP2-The four members of the CtBP family, CtBP1, CtBP2, CtBP1-S, and RIBEYE share a high degree of sequence identity except at the N-terminal region ( Fig. 2A). The N-terminal 20-amino-acid region of CtBP2 contains several basic amino acid residues that may potentially constitute a nuclear localization signal. To examine the potential nuclear localization signal function of this sequence, the N-terminal 21 residues of CtBP2 were replaced with Met-Ser-Gly-Val residues (Fig. 2B). The resulting CtBP2 mutant (⌬N-CtBP2), bears the N-terminal residues of CtBP1-S/BARS ( Fig. 2A). An immunofluorescence analysis of the mutant, ⌬N-CtBP2 revealed its distribution throughout the cells, in sharp contrast to the wt CtBP2 (Fig. 2C). A similar deletion of the N-terminal residues of CtBP1 generated a CtBP1 mutant, ⌬N-CtBP1, which is identical to CtBP1-S/ BARS (Fig. 2B). This deletion did not affect the partial nuclear and cytoplasmic localization of CtBP1 (Fig. 2C). These results suggest that the N-terminal region of CtBP2 contains a key determinant for the nuclear localization of CtBP2.
To examine whether the CtBP2 N-terminal region is the major determinant for the differential subcellular localization of CtBP1 and CtBP2, the N-terminal 21 residues of CtBP2 were substituted for the N-termi-nal 15 residues of CtBP1 (Fig. 2B). The resulting chimeric protein, N2-CtBP1, was detected exclusively in the nucleus (Fig. 2C), in a fashion identical to that of wt CtBP2. Thus, the distinct nuclear localization pattern of CtBP2 appears to be determined by its unique N-terminal sequence. Further, it could also confer a predominant nuclear localization pattern to CtBP1.
Lys-10 Is Critical for the Nuclear Retention of CtBP2-There are three Lys residues in the N-terminal region of CtBP2 (Fig. 3A). Because Lys residues are frequent targets for post-translational modifications, we decided to examine the effect of these residues in nuclear localization of CtBP2. To examine their roles, they were mutated together or singly to Arg residues (Fig. 3A) so that the overall charge of the mutant CtBP2 would be comparable with that of wt CtBP2. The mutant plasmids were transfected into HeLa cells, and the cells were stained with the FLAG antibody. As shown in Fig. 3B, although K6R and K8R mutations had no effects on the exclusive nuclear localization of CtBP2, mutation of all three Lys residues (3KR) as well as mutation of Lys-10 (K10R) caused CtBP2 to be localized predominantly in the cytoplasm. Thus, Lys-10 plays a critical role in exclusive nuclear localization of CtBP2.
The predominant cytoplasmic localization of 3KR-CtBP2 and K10R-CtBP2 (with nuclear exclusion) is in sharp contrast to the partial nuclear localization of ⌬N-CtBP2, which totally lacks the N-terminal region (Fig. 2). It appears that the mutations 3KR and K10R confer a dominant cytoplasmic localization pattern to CtBP2. This function may be because of cytoplasmic retention of 3KR-CtBP2 and K10R-CtBP2 by cytosolic factors or because of enhanced nuclear export of these mutant proteins. To examine whether nuclear export accounts for the predominant cytoplasmic localization of K10R-CtBP2, transfected cells were treated with the nuclear export inhibitor, leptomycin B, which specifically inhibits nuclear export mediated by CRM-1 (37). As shown in Fig.  3C, the leptomycin B treatment resulted in partial nuclear localization of mutants 3KR or K10R in majority of cells. A similar treatment of cells transfected with CtBP1 has no detectable effect on CtBP1 localization (not shown). These results suggest that K10R-CtBP2 is capable of nuclear localization but is efficiently exported from the nucleus.
N-terminal Lys Residues of CtBP2 Are Acetylated-Our results (Fig.  3) suggest that the Lys-10 is important for nuclear retention of CtBP2 and a mutation of this residue (K10R) does not significantly impede with nuclear import. Because the mutation (K10R) does not cause a significant charge change in CtBP2, we decided to investigate the potential acetylation of the Lys residues. For this purpose, a panel of CtBP proteins was immunoprecipitated from transfected cells and analyzed by Western blotting using a monoclonal pan Ac-Lys antibody. As shown in  ⌬N-CtBP2 (lane 5), a band of ϳ50 kDa was also present, which does not correspond to Flag-HA-tagged CtBP proteins. Its identity is not known.
To examine whether the endogenous CtBP2 was also acetylated, we made use of a polyclonal antibody generated against an N-terminal peptide of CtBP2. The specificity of this antibody was ascertained by Western blot analysis of transiently expressed (epitope-tagged) CtBP proteins (Fig. 4B). This antibody specifically recognized the Flag-HAtagged CtBP2 as well as Lys mutants of CtBP2, but not CtBP1 or ⌬N-CtBP2 (Fig. 4B, lower panel). The endogenous CtBP2 was immunoprecipitated from HeLa or A549 cells using this antibody, and the precipitated CtBP2 was detected by Western blot using a CtBP2 monoclonal antibody (Fig. 4C, lower panel) or the Ac-Lys monoclonal antibody (Fig. 4C, upper panel). The CtBP2 polyclonal antibody readily immunoprecipitated CtBP2 from both cell lines, and the immunoprecipitated CtBP2 was acetylated (lanes 3 and 4). In contrast, the preimmune serum did not immunoprecipitate CtBP2 nor did it contain acetylated CtBP2 (lanes 1 and 2). These results suggest that endogenous CtBP2, like transiently expressed CtBP2 is acetylated.
p300 Acetylates the N-terminal Lys Residues of CtBP2-Because our results suggest that acetylation is critical for nuclear retention of CtBP2, we hypothesized that CtBP2 might be a target for a nuclear acetylase. We examined the effects of three different nuclear acetylases, Tip60 (38), mGCN5 (39), and p300 (40). The acetylation levels of CtBP2 coexpressed with these enzymes were analyzed by immunoprecipitation of CtBP2 with the FLAG antibody and probing the Western blots with the Ac-Lys antibody (Fig. 5A). Coexpression of CtBP2 with p300 significantly enhanced the level of acetylation of CtBP2 (Fig. 5A, top panel,  lane 5). Expression of mGCN5 also resulted in a modest increase in acetylation of CtBP2 (lane 4), whereas expression of Tip60 had no significant effect on CtBP2 acetylation (lane 3). Western blots suggested that all acetylases were expressed at readily detectable levels (Fig. 5A,  lower panels). Thus, it appears that CtBP2 may be acetylated preferentially by p300.
Subsequently, we examined the effect of p300 on various CtBP2 mutants. Coexpression of p300 with wt CtBP2 (Fig. 5B, lanes 3 and 4) or mutant A58E (defective in PXDLS-binding) (lanes 5 and 6) resulted in significant enhancement of acetylation. In contrast, p300 did not have any effect on wt CtBP1 (lanes 1 and 2). Among the three N-terminal Lys mutants, coexpression of p300 with mutant K8R resulted in significant enhancement of acetylation (lanes 9 and 10), whereas p300 had only a small effect on K6R (lanes 7 and 8) and K10R (lanes 11 and 12). Thus, Lys-6 and Lys-10 of CtBP2 appear to be favored target sites for acetylation by p300. Furthermore, the cytoplasmic localization of the K10R-CtBP2 mutant appears to correlate with a deficiency in acetylation of Lys-10.
To examine the requirement of p300 for CtBP2 acetylation, HeLa cells were first transfected with a pool of siRNAs targeting p300 and then transfected with the plasmid expressing epitope-tagged CtBP2. The cell lysates were immunoprecipitated with the FLAG antibody and examined for CtBP2 acetylation by Western blot analysis. As shown in Fig. 5C, p300 siRNA specifically down-regulated the levels of p300 (lanes 4 and 5, third panel from top) and did not cause any change to the levels of ␣-tubulin (bottom panel). Importantly, the levels of acetylated CtBP2 were significantly reduced in cells transfected with p300 siRNA (lanes 4 and 5, top panel). In contrast, transfection with the control siRNA (lane 2) or the GCN5 siRNA (lane 3) did not affect CtBP2 acety-  Flag-mGCN5, and p300) and Flag-HA-tagged CtBP2. The cell lysates and FLAG immunoprecipitates (IP) were examined by Western blot analysis using antibodies for Ac-Lys, FLAG, Tip60, and p300. B, effect of p300 on acetylation of CtBP2 mutants. CtBP1 and CtBP2 were either expressed alone (Ϫ) or with p300 (ϩ). The CtBP proteins were immunoprecipitated with the FLAG antibody and probed with Ac-Lys antibody (top panel) or with the FLAG antibody (middle panel). p300 in cell lysates was detected by Western blot analysis using a p300 mAb. C, effect of p300 depletion on CtBP2 acetylation. HeLa cells were transfected with different pools of siRNAs for 24 h, followed by transfection with epitope-tagged CtBP2 for 24 h. Cell lysates were examined under conditions described in B. Mock, cells were treated only with the transfection reagent during siRNA transfection. Tu-␣, indicates tubulin-␣. Lanes 4 and 5 represent duplicate assays.  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 lation. These results reinforce the conclusion that p300 is required for CtBP2 acetylation in vivo.

Nuclear Acetylation of CtBP2
Regulation of CtBP2 Transcriptional Activity by Lys-10 Acetylation-The epithelial cell adhesion molecule E-cadherin plays critical roles during embryonic development and oncogenic transformation (41). Its promoter is subjected to both positive and negative regulations. The cellular repressors ZEB1, ZEB2, Snail, and Slug all bind to specific promoter elements to repress E-cadherin transcription (42,43). The function of ZEB1 and ZEB2 is mediated in large part through their interaction with CtBP through PLDLS-like domains. To examine the roles of CtBP2 acetylation in transcriptional repression, the pE-cad-Luc reporter plasmid was cotransfected with plasmids expressing the wild type CtBP2 or its Lys mutants into Ctbp1/Ctbp2 double knock-out mouse embryonic fibroblasts (7). Luciferase assays (Fig. 6) showed that both the K10R and the 3KR mutants of CtBP2 had a significantly enhanced activity in repression of the E-cadherin promoter. Interestingly, the N-terminal deletion mutant of CtBP2, ⌬N-CtBP2, and CtBP1 both had similar activity to CtBP2. Thus, lack of acetylation of the CtBP2 N-terminal domain is correlated with an enhanced corepressor function of CtBP2 (see "Discussion").

DISCUSSION
CtBP1 and CtBP2 share a high degree of sequence identity and are believed to have similar transcriptional regulatory functions (1,2). Genetic studies in the mouse have suggested that the two proteins have overlapping and unique functions during development (7). In addition to nuclear functions, CtBPs also have been postulated to have cytoplasmic functions (26). It is possible that some activities of the two homologs might be governed by differential subcellular localization. It appears that the activities of CtBP1 might be regulated by elaborate nucleocytoplasmic transport mechanisms that involve phosphorylation, sumoylation, and PDZ binding. Because CtBP2 lacks the PDZbinding motif and is not sumoylated, the subcellular localization of CtBP2 might be differently regulated than CtBP1. Our results have revealed that the endogenous and ectopically expressed CtBP2 has a striking nuclear restricted expression, whereas CtBP1 is expressed in both nuclear and cytosolic compartments (Fig. 1). We have demonstrated that the distinctive nuclear localization pattern of CtBP2 is dictated by its unique N-terminal sequences. In our studies, substitution of the N-terminal sequences of CtBP2 for the CtBP1 sequences has conferred a nuclear restricted expression to CtBP1. In contrast, the N-terminal domain of CtBP2 does not confer such a property to a heterologous protein such as green fluorescent protein (not shown), suggesting the function of the N-terminal domain of CtBP2 is context-dependent.
In contrast to the N-terminal sequences of CtBP2, the N-terminal sequences of CtBP1 may not be involved in the regulation of subcellular localization. The subcellular localization pattern of one of our CtBP1 mutants (⌬N-CtBP1) is indistinguishable from wt CtBP1 (Fig. 2). Because the N-terminal sequences of ⌬N-CtBP1 are identical to that of the isoform CtBP1-S/BARS, our results also suggest that there may not be a significant difference in the subcellular localization of the two isoforms of CtBP1. In contrast to CtBP1, the N-terminal sequences of the CtBP2 isoforms might be critically important for their differential localization. The CtBP2 isoform RIBEYE is a large (985 amino acids) cytoplasmic protein that is exclusively expressed at synaptic ribbons of the central nervous system (9,31,44). RIBEYE contains a large N-terminal domain (560 amino acids) fused to a C-terminal (425 amino acid) region of CtBP2, resulting in a precise deletion of the N-terminal 20 amino acids of CtBP2 (see Fig. 2A). Here, we have identified this 20-amino-acid region as a critical determinant for nuclear localization of CtBP2. Thus, it is possible that the lack of the N-terminal domain of CtBP2 in RIBEYE might contribute to its nuclear exclusion.
Several lines of evidence suggest that the N-terminal domain of CtBP2 serves as a nuclear retention signal. First, the N-terminal region of CtBP2 fails to confer nuclear targeting function to GFP (data not shown). Second, ⌬N-CtBP2 mutant lacking the N-terminal region localizes in the nucleus as well as in the cytoplasm in a pattern similar to that of CtBP1 (Fig. 2). In contrast, 3KR-CtBP2 and K10R-CtBP2 are excluded from the nucleus (Fig. 3), suggesting that the mutant proteins may actively be transported from the nucleus. Third, the nuclearly excluded CtBP2 mutants (3KR and K10R) are significantly retained in the nucleus in the presence of the nuclear export inhibitor leptomycin B (Fig. 3). The partial nuclear localization of 3KR-CtBP2 and K10R-CtBP2 in the presence of leptomycin B may be because of an incomplete inhibition of the nuclear export pathway. Alternatively, acetylation of Lys-10 in CtBP2 may be essential for CtBP2 to associate with a nuclear component for nuclear retention, which can be not be achieved by inhibition of nuclear export. In the absence of Lys-10 acetylation, CtBP2 localization may be regulated by nuclear export as well as interaction with other cellular proteins. Because substitution of the CtBP2 N-terminal region for that of CtBP1 is sufficient to confer exclusive nuclear localization to otherwise partially nuclear localized CtBP1, it can be hypothesized that CtBP1 is normally subject to an inefficient nuclear export. However, the possibility that CtBP2 N-terminal region also facilitates nuclear import cannot be excluded. We noticed that amino acids 21-40 of CtBP2 are rich in Pro, Leu, and Ile residues, and this region bears partial resemblance to the nuclear export sequence of HIV-1 Rev (reviewed by Pollard and Malim (45)). The functional significance of this sequence in CtBP2 nuclear localization remains to be explored.
We have demonstrated that nuclear retention of CtBP2 is mediated by acetylation of the Lys residues at the N-terminal region. Although acetylation of Lys-6 and Lys-10 is observed, acetylation of Lys-10 appears to be important for nuclear retention. Our results further suggest that the nuclear acetylase p300 is involved in acetylation of CtBP2 (Fig. 5). Consistent with the data from transfected cells, we have recently shown by an in vitro assay that purified CtBP2 wt protein can be robustly acetylated by purified p300, compared with marginal acetylation of purified 3KR-CtBP2 mutant protein by p300 (data not shown). Because acetylation of CtBP2 can be enhanced by exogenous expression of p300, it appears that the exclusive nuclear localization of CtBP2 does not require saturating level of acetylation. It is expected that CtBP2 acetylation is in a dynamic status regulated by acetylation/deacetylation. In this context, it should be noted that the CtBP1 protein complexes contain class I histone deacetylases (23,46,47). Although it is not known whether CtBP2 also complexes with histone deacetylases, because of its overlapping transcriptional regulatory activity with CtBP1, it is possible that CtBP2 might also recruit histone deacetylases. In addition, CtBP2 may form dimers or incorporate into large nuclear protein complexes in a process facilitated by acetylation of Lys-10. Once such a complex is established, deacetylation of CtBP2 may not have an impact on the nuclear localization of CtBP2. Failure to acetylate Lys-10 may exclude CtBP2 from such complexes and subject CtBP2 to nuclear export and other regulatory pathways.
In transcriptional assays, we observed that acetylation-deficient mutants of CtBP2 repress the E-cadherin promoter more efficiently than the acetylation-competent allele (Fig. 6). These results are unexpected considering that acetylation-deficient mutants are more concentrated in the cytoplasm. A possible model that could explain these results would be that CtBP2 mediates transcriptional repression by the sequestration and nuclear depletion of transcriptional activators (by non-acetylated CtBP2), in addition to other nuclear mechanisms. This model is suggested by the finding that K10R and 3KR mutants of CtBP2 were more active in repression of the E-cadherin promoter. There is a precedent for such a model, as CtBP has been reported to sequester and inhibit the transcriptional activities of positive transcriptional regulators such as p300/CBP (25,48,49) and nuclear ␤-catenin (50). Alternatively, cytoplasmically localized non-acetylated CtBP2 may help regulate cytoplasmic functions, which may have an indirect effect on E-cadherin promoter. In both cases, the transcriptional regulatory activity of CtBP2 is modulated by acetylation/deacetylation of the N-terminal domain. The acetylation/deacetylation of CtBP2 may also serve as a distributive mechanism to regulate the cytoplasmic and nuclear functions.
It has been recently reported that CtBP1 interacts with the bromodomain of p300 and inhibits the histone acetylase activity of p300 (25). The interaction between the bromodomain containing acetylases and CtBP1 has been reported to be mediated by a conserved PXDLS (CtBPbinding) motif that is embedded within the bromodomain (25). However, we have shown that coexpression of p300 and a CtBP2 mutant (A58E) deficient in PXDLS binding (17) results in efficient acetylation of CtBP2. Although we cannot fully rule out an interaction between p300 and CtBP2 via the PXDLS motif, we suggest that an alternate binding surface of CtBP2 might be involved in recruitment of p300. In CtBP1, the existence of a protein interaction surface other than the PXDLSbinding pocket has been inferred (51). Similarly, regions of p300, in addition to the bromodomain have been implicated in interaction with CtBP1 (48,49).
In summary, our studies suggest that acetylation of CtBP2 may be important for the nuclear retention of CtBP2 and the function of CtBP2 as a transcription corepressor. Lack of CtBP2 acetylation may result in cytoplasmic localization of CtBP2 and reduced ability of CtBP2 to interact with transcriptional repressors in the nucleus. However, cytoplasmically localized unacetylated CtBP2 may still repress transcription from certain promoters by potentially sequestering positive transcription factors.