Post-translational Modification of Nuclear Co-repressor Receptor-interacting Protein 140 by Acetylation*

Receptor-interacting protein 140 (RIP140) is a versatile co-regulator for nuclear receptors and many transcription factors and contains several autonomous repressive domains. RIP140 can be acetylated, and acetylation affects its biological activity. In this study, a comprehensive proteomic analysis using liquid chromatography-tandem mass spectroscopy was conducted to identify the in vivo acetylation sites on RIP140 purified from Sf21 insect cells. Eight acetylation sites were found within the amino-terminal and the central regions, including Lys111, Lys158, Lys287, Lys311, Lys482, Lys529, Lys607, and Lys932. Reporter assays were conducted to examine the effects of acetylation on various domains of RIP140. Green fluorescent protein-tagged fusion proteins were used to demonstrate the effect on nuclear translocation of these domains. A general inhibitor of reversible protein deacetylation was used to enrich the acetylated population of RIP140. The amino-terminal region (amino acids (aa) 1–495) was more repressive and accumulated more in the nuclei under hyperacetylated conditions, whereas hyperacetylation reduced the repressive activity and nuclear translocation of the central region (aa 336–1006). The deacetylase inhibitor had no effect on the carboxyl-terminal region (aa 977–1161) where no acetylation sites were found. Hyperacetylation also enhanced the repressive activity of the full-length protein but triggered its export into the cytosol in a small population of cells. This study revealed differential effects of post-translational modification on various domains of RIP140 through acetylation, including its effects on repressive activity and nuclear translocation of the full-length protein and its subdomains.

Environmental factors in the extracellular milieu utilize signal transduction pathways to propagate their cues into gene expression (1)(2)(3). Often the proteins involved in such signaling pathways undergo post-translational modification. A variety of post-translational modifications, including phosphorylation, acetylation, methylation, and glycosylation, regulate protein functions (4 -7). The study of protein function by identifi-cation of proteins along with their post-translational modification has been referred to as "functional proteomics" and is an important step in delineating signal transduction pathways. One major challenge is to identify post-translational modifications on these proteins in vivo (8).
Receptor-interacting protein 140 (RIP140) 1 is a co-regulator for many transcription factors (9). Nuclear receptors represent the largest group of transcription factors that interact with RIP140 (10 -13). Human RIP140 was initially characterized as a ligand-dependent co-activator for a chimeric estrogen receptor (10). However, the mouse RIP140 cloned in our laboratory with the ligand-binding domain of an orphan nuclear receptor TR2 as the bait was shown to be a potent corepressor for TR2 in the absence of putative ligand (14). Later many researchers including our group reported RIP140 as a suppressor for nuclear hormone receptors and many other transcription factors (15)(16)(17). RIP140 is recruited to nuclear receptors through its nine LXXLL motifs and a modified motif of LXXML where X can be any amino acid (14,18). RIP140 also contains four autonomous repressive domains (RDs). RD1 is located in the amino-terminal region (aa 1-495), RD2 and RD3 are located in the central portion (aa 336 -1006), and RD4 is located in the carboxyl-terminal region (aa 977-1161). These domains function through various mechanisms. The amino-terminal region recruits histone deacetylases (HDACs) (17), whereas the central region interacts with carboxyl-terminal binding proteins (CtBP1 and CtBP2) (19). In terms of its physiological action, RIP140-null mice showed female reproductive defects (9). Studies of these animals revealed that RIP140 could play an important role in the regulation of fat accumulation in adipose tissues (20).
Post-translational modifications of transcription factors including nuclear receptors and their co-regulators have attracted much attention (21)(22)(23)(24)(25)(26). However, few studies have examined modifications of these proteins in vivo due to technical difficulties in expressing and purifying these nuclear proteins from eukaryotic cells. Recently we were able to ex-press and purify RIP140 from insect cells, a model system widely used for expressing eukaryotic proteins. We found 10 phosphorylation sites on RIP140 purified from insect cells (27) that appeared to play a role in its repressive activity (not shown). Although RIP140 could be acetylated (28), this modification was identified on the in vitro modified protein at Lys 446 using p300 and p300/CBP-associating factor/p300 as the enzyme. However, point mutation of this particular residue could not prevent RIP140 acetylation, suggesting that RIP140 could also be acetylated at other residues. We therefore conducted a comprehensive proteomic analysis of the acetylation pattern of eukaryotically expressed RIP140. For comparison, RIP140 expressed in Escherichia coli was examined in parallel. LC-ESI-MS/MS technique identified eight acetylated residues in the amino-terminal and the central regions of RIP140 from insect cells.
We also determined the effects of acetylation on the biological activity of RIP140 and explored the mechanism by which such activity was initiated. Reporter assays were conducted for each dissected domain by using a general inhibitor of reversible protein deacetylation. In addition GFP-tagged proteins were used to examine the effect of acetylation on the nuclear translocation of each domain. It was found that hyperacetylation enhanced both the repressive activity and the nuclear translocation of the amino-terminal region. In contrast, hyperacetylation abolished the repressive activity of the central region and diminished its nuclear translocation. No effects were found in the carboxyl-terminal region where no acetylation sites were identified.

EXPERIMENTAL PROCEDURES
Construction of His-tagged RIP140 and Purification of RIP140 from Sf21 Insect Cells-The full-length mouse RIP140 (14,27) was tagged with a His epitope by replacing the carboxyl-terminal domain of RIP140 with a PCR-amplified His-tagged fragment. The full-length His-tagged RIP140 cDNA was cloned into an insect expression vector, pVL1392 (Invitrogen), at the BglII and KpnI sites. The expression and purification of RIP140 was performed as described previously (27). Briefly Sf21 insect cells were infected with the recombinant baculovirus vector. RIP140 was purified from the nuclear extract by affinity chromatography on Talon resin. The protein was concentrated through a 10-kDa Centricon filter. The glutathione S-transferase (GST)-RIP140 protein from E. coli was prepared as described previously (14,27).
Mass Spectrometric Analysis of RIP140 -Detailed experimental procedures for mass spectral analysis of RIP140 protein samples were described previously (27). Purified His-RIP140 protein from insect cells and GST-RIP140 from E. coli were resolved by SDS-PAGE. Gel slices containing RIP140 were subjected to overnight in-gel tryptic digestion (29,30). The samples were analyzed by MALDI-TOF MS (QSTAR XL, Applied Biosystems, Inc., Foster City, CA) using ␣-cyano-4-hydroxycinnamic acid as a matrix in a positive ion reflection mode. For LC-MS, an LC Packings (Dionex, Sunnyvale, CA) Famos autosampler and an LC Packings Switchos pump were used to concentrate and desalt the sample on an LC Packings C 18 nanoprecolumn. The precolumn was connected in-line with a capillary column (100-m inner diameter, packed with 5-m, 200-Å pore size C 18 particles), and peptides were eluted using an LC Packings Ultimate LC system.
The LC system was on line with an Applied Biosystems QSTAR Pulsar quadrupole TOF mass spectrometer equipped with a Protana nanoelectrospray source. As peptides were eluted from the column, they were focused into the mass spectrometer. Information-dependent acquisition (IDA) was used to acquire MS/MS. The IDA mode was set to measure continuous cycles of three full scan TOF MS from 400 -550, 550 -750, and 750 -1200 m/z plus three product ion scans from 50 to 4000 m/z. Data from the IDA experiments were analyzed using the MASCOT MS/MS data search (www.matrixscience.com) of the NCBI data bank. The mass tolerance of both precursor ions and the MS/MS fragment ions was set at Ϯ0.1 Da, and carbamidomethylcysteine was specified as a static modification. Acetylated lysine and oxidized methionines were specified as variable modifications. All MS/MS spectra were manually checked to verify sequence assignments. Peaks with a minimum height of 3% relative to the base peak were considered, and a 100-ppm tolerance was used to establish matches with the theoretical b and y ions that were predicted with the help of Bioanalyst software (Applied Biosystems).
Cell Culture, Transfection, and Reporter and Translocation Assays-For reporter gene assays, the amino-terminal region (aa 1-495, for RD1), central region (aa 336 -1006, for RD2 and RD3), carboxylterminal region (aa 977-1161, for RD4), and the full-length RIP140 (aa 1-1161) were cloned into pBD-GAL4 vector (Stratagene, La Jolla, CA) as described previously (14). The luciferase reporter construct for the trans-repressive activity was made by placing five copies of the GAL4 binding sites (5Ј-CGGAGGACAGTACTCCG-3Ј) upstream of the thymidine kinase-luciferase (TK-luc) reporter (14). GFP-tagged fusion proteins of the repressive domains and full-length RIP140 were also constructed by placing each domain (14) into a pEGFP-C1 vector (Clontech). Both the amino-terminal and the central regions encode a nuclear localization signal (NLS). The carboxyl-terminal region, however, is devoid of a NLS. Therefore, a NLS derived from TR2 nuclear receptor (31) was fused to its 3Ј-end prior to fusion with GFP, ensuring its nuclear accumulation for subsequent assays on the effects of inhibitory drugs. The assays were conducted in COS-1 cells maintained at 37°C in a CO 2 incubator in Dulbecco's modified essential medium supplemented with 10% fetal bovine serum. Transfection experiments and luciferase and ␤-galactosidase gene (lacZ) assays were done as described previously (14). Sodium butyrate (Upstate Biotechnology, Lake Placid, NY) was used as a general protein deacetylase inhibitor.

Expression, Purification, and Mass Spectral Analysis of
Tryptic RIP140 -Prior studies validated the effectiveness of expressing recombinant proteins in insect cells to generate large quantities of low abundance proteins such as transcription factors (32,33). Although stoichiometry appears variable, post-translational site occupancy of such proteins expressed in insect cells is the same as that seen in proteins from mammalian cells (34). Therefore, we expressed RIP140 protein in insect cells for further purification and analyses. To confirm that modification by acetylation occurs specifically in RIP140 expressed in insect cells, we also expressed and purified GST-tagged RIP140 from E. coli and analyzed them in parallel by mass spectroscopy. Using affinity chromatography under denaturing conditions, we were able to purify the recombinant protein to over 95% homogeneity (27). The proteins were subjected to trypsin digestion, and the tryptic peptides were first analyzed by MALDI-TOF MS to identify RIP140. The MALDI-TOF MS data were subjected to a MAS-COT search at the NCBI data bank. The MASCOT search results confirmed the identity of the protein with sequence coverage to over 60% of the total protein (data not shown). LC-ESI-MS/MS was used to identify post-translational modification sites on RIP140. We recorded three independent full scan ion chromatograms from m/z 400 to 550, 550 to 750, and 750 to 1200. IDA was used to acquire MS/MS data. IDA analyses were performed on the tryptic digests of RIP140 expressed in E. coli and insect cells. The data from IDA experiments were searched using a MASCOT search. The mass tolerance of both precursor ion and the MS/MS fragment ions was set at Ϯ0.1 Da, carbamidomethylcysteine was specified as a static modification, and acetylation at lysine was specified as a variable modification. The results revealed eight tryptic acetylated peptides from RIP140 expressed in insect cells (Table I). No acetylated peptide was detected on RIP140 expressed in E. coli.
Mapping of Acetylation Sites on RIP140 -The MS/MS data were analyzed manually to map the acetylation sites on RIP140. A 42-amu shift was considered an indication of acetylation. As both acetylation and trimethylation render the same integral mass shift, the immonium ion at m/z 126 specific for acetylated lysine was monitored (35,36). All the peptides examined demonstrated the presence of this ion (Fig. 1, A-H). In addition, the absence of marker ions for trimethylated lysine generated by the loss of trimethylamine from b or y ions as [M Ϫ 59] ϩ was checked carefully. Similarly a 42-amu negative ⌬mass shift due to the loss of an acetyl group from y or b ions was also considered in the assignment of the acetylation sites (35,36). Initial analysis of mass data showed some peptides displaying a 43-amu shift instead of the 42 units typical for acetylation. This raised the possibility of carbamylation of lysine residues rather than acetylation in the modified peptides. However, careful analysis of the MS/MS data confirmed that this additional 1-unit ⌬mass shift was due to the deamidation of either glutamine or asparagine residues present in the acetylated peptides. The MS/MS spectra of the modified peptides were always compared with the unmodified peptides (data not shown) for fingerprinting purposes.
The peptide spanning residues 101-112 (precursor molecular weight, 1371.75) displayed a doubly charged ion at m/z 686.88 in the total ion chromatogram (Fig. 1A). The precursor molecular mass showed a 43-amu shift due to acetylation of a lysine and deamidation of an asparagine residue (Table I) The triply charged precursor ion at m/z 620.98 of the modified peptide spanning residues 155-170 (molecular weight, 1859.91) showed a 43-amu shift that could be accounted for by acetylation at the lysine residue and deamidation of a glutamine residue (Table I). The singly charged y ions from y1 to y9 and b ions from b1 to b3 were identical to those of the unmodified peptide (Fig. 1B). The doubly charged y12 ion at m/z 681.3 corresponded to the unmodified peptide deamidated at Gln 160 . The singly charged b5 ion at m/z 628.4 and the b6 and b7 ions were attributed to the acetylated Lys 158 .
The MS/MS spectrum of the modified peptide spanning residues 283-298 exhibited a y9 ion at m/z 1011.5 and a b4 ion at m/z 451.2, identical to the unmodified peptide (Fig. 1C). However, the intense b5 ion at m/z 621.3 displayed a 42-amu shift, suggesting acetylation at Lys 287 . Similarly a 42-amu shift in the molecular mass was observed for the doubly charged precursor ion at m/z 880.93 of the peptide spanning residues Tryptic digests of RIP140 protein were subjected to LC-ESI-MS/MS. Three independent full scan ion chromatograms from m/z 400 to 550, 550 to 750, and 750 to1200 were recorded in an IDA mode to acquire MS/MS data. The IDA data were analyzed on line with a MASCOT (www.matrixscience.com) MS/MS data search of the NCBI data bank. The MS/MS data were analyzed manually to confirm the sequences of the modified and unmodified forms of the same peptide identified by the data bank search. The full scan chromatograms were analyzed to assign the charged state, retention time, and intensities of the peptides. Bold K indicates acetylated lysine.

Residues
Sequence  305-320 (molecular weight, 1759.84) ( Table I). In the MS/MS spectrum (Fig. 1D), the y9 ion at m/z 920.5 corresponded to that seen in the unmodified peptide. The b5 ion at m/z 542.3 and the b6 o ion at m/z 652.30 were also identical in the native peptide. Taken together, these data suggested the acetylation at Lys 311 , located between b6 and y9. The presence of this acetylation site was substantiated by the presence of doubly charged y16 ion at m/z 859.4 caused by loss of an acetyl group.
The MS/MS spectrum of the modified peptide spanning residues 476 -492 showed a b11 ion at m/z 1252.6, consisting of an intact acetyl moiety (Fig. 1E). The presence of an acetylated lysine residue in the peptide was demonstrated by the presence of the immonium marker ion at m/z 126 and the doubly charged y16 and y17 ions appearing at m/z 867.4 and 923.9, respectively, due to loss of an acetyl group. The y10 ion at m/z 1124.5, b5 ion at m/z 482.3, and the a6 ion at m/z 567.4 were identical to those of the native peptide. This provided evidence that Lys 482 was the acetylation site.
The 43-amu shift observed for the precursor ion at m/z 880.93 of the peptide spanning residues 517-537 (molecular weight, 2405.11) could be attributed to an acetylation along with a deamidation (Table I). The y8 ion at m/z 1012.5 and b4 ion at m/z 370.2 (Fig. 1F) were identical to those of the unmodified peptide. However, the doubly charged b11 ion at m/z 562.3 and the singly charged b5, b6, b9, and b10 ions at m/z 499.2, 614.3, 951.4, and 1066.5, respectively, corresponded to the unmodified peptide deamidated at Gln 521 . This suggested an acetylation site between b11 and y8. Thus, the acetylation was assigned to Lys 529 .
The modified peptide spanning residues 606 -630 (precursor molecular weight, 2549.19) appeared as a triply charged ion at m/z 850.74 (Fig. 1G), whereas that of the unmodified peptide appeared at m/z 836.4 (precursor molecular weight, 2506.17). The 43-amu difference between the two indicated the modification by acetylation along with deamidation of either a glutamine or an asparagine residue. The singly charged y12 ion at m/z 1168.6 and the doubly charged y21 and y22 ions at m/z 1053.5 and m/z 1097.0, respectively, of the modified peptide corresponded to those seen in the unmodified peptide. This indicated an acetylation site at Lys 607 . This was confirmed by the observation of the singly charged b2 ion at m/z 228.1, the b3 ion at m/z 357.2, and the b4 ion at m/z 444.2.
The modified peptide spanning residues 930 -938 showed a 43-amu shift (Table I). This indicated that the peptide was modified either by acetylation along with deamidation or carbamylation. However, the intense immonium ion signal at m/z 126 indicated the shift was more likely due to an acetylated lysine (Fig. 1H). The modified peptide showed a y6 ion at m/z 707.4, corresponding to deamidation of Asn 953 of the unmodified peptide. The y7 ion at m/z 877.5 exhibited a 43-amu shift, suggesting acetylation of Lys 932 . This was supported by the a3 ion at m/z 359.2 and the b4 ion at m/z 474.2, both of which showed a 42-unit mass shift due to acetylation on Lys 932 .
Role of Acetylation on Repressive Activity of RIP140 -A trans-repressive assay was conducted using the GAL4 reporter system in COS-1 cells. However, RIP140 contains multiple RDs, and multiple acetylation sites were found in these various domains. These experiments were therefore performed on both the full-length protein and its subdomains fused to the same DNA-binding domain of GAL4. Hyperacetylation was induced by blocking protein deacetylases with sodium butyrate. In the presence of this deacetylase inhibitor, the amino-terminal region (aa 1-495) was more repressive than the untreated control (Fig. 2). In contrast, the repressive activity of the central region (aa 336 -1006) was significantly reduced under the same conditions. Deacetylase inhibitors exerted no effect on the carboxyl-terminal region (aa 977-1161) of RIP140 where no acetylation was identified. As observed for the amino-terminal region, full-length RIP140 had greater repressive activity when it was hyperacetylated, suggesting that the repressive activity of the amino-terminal domain probably dominated over regulation by the rest of this molecule.

FIG. 1-continued
Effect of Acetylation on Translocation of RIP140 -Transcription activities of nuclear receptors and transcription factors are partially regulated by their nuclear translocation. We determined whether acetylation affected the nuclear translocation of RIP140 and if the effect was the same for different domains of RIP140. We used a GFP tag strategy to follow the cellular distribution of RIP140 domains under normal and hyperacetylated conditions. According to amino acid sequences, both the amino-terminal and the central regions encode a NLS and therefore could be enriched in the nuclei. The carboxyl-terminal domain contained no NLS but could be accumulated in the nuclei by fusing to a NLS (Fig. 3c). As expected, the full-length protein is exclusively nuclear (Fig.  3d). Treatment with deacetylase inhibitor enhanced nuclear localization of the amino-terminal region (Fig. 3, compare a  and e). In control cells, it was restricted to the nuclei of ϳ26% of cells, whereas hyperacetylation dramatically enhanced its translocation to the nucleus in 74% of cells (Fig. 3i). In contrast, sodium butyrate caused the central region to be evenly distributed between the cytoplasm and the nucleus, suggesting a reduction in its nuclear accumulation (Fig. 3f). Approximately 73% of treated cells exhibited an even distribution pattern in the cytoplasm and the nucleus, whereas only 38% of the untreated cells had the even distribution pattern (Fig.  3i). No significant effect on translocation was detected for the carboxyl-terminal domain of RIP140 under the same conditions (Fig. 3, c and g). This is consistent with the results of the reporter assays where sodium butyrate had no effect on the repressive activity of this region. Interestingly although hyper-acetylation rendered the full-length RIP140 more repressive in the reporter assay (Fig. 2), this condition triggered the fulllength RIP140 to be translocated to the cytosol in a portion of the cell population. This suggests that effects of acetylation extend to some as yet undetermined function of RIP140 that contributes to its repressive activity in addition to its nuclear accumulation.
In summary, acetylation exerted differential effects on the various domains of RIP140 with respect to their contribution to the overall biological activity of RIP140 as a transcription co-repressor. Nuclear translocation of both the amino-terminal and the central regions was affected by acetylation of lysine residues, which correlated with the effects on their repressive activity. The activity of the carboxyl-terminal re-tivity of the carboxyl-terminal region, which contained no acetylation sites.
Because acetylation has a potential role in protein translocation (39,42,43), we used GFP fusion proteins to dissect the effect of acetylation on nuclear translocation of these individual domains. Hyperacetylation facilitated the translocation of the amino-terminal region to the nucleus. In contrast, the central region became evenly distributed in the cytosol and nucleus. This correlated well with the differential effects of the inhibitor on these regions as determined by the trans-repressive assays using the GAL4 system. Interestingly sodium butyrate triggered translocation of the full-length RIP140 to the cytosol in a portion of cells despite higher levels of repressive activity detected in inhibitor-treated cells. This would suggest that other, as yet unidentified, repressive mechanisms were also affected by acetylation. Nevertheless both trans-repressive and GFP-tagged nuclear translocation assays suggested that acetylation differentially affected the various domains of RIP140. The ultimate effect of acetylation on the full-length RIP140 likely depends upon its stoichiometry and the relative abundance of other involved components in the cellular environment (such as HDACs and CtBP) and the nuclear translocation machinery under specific conditions.
Previously we reported the complete map of phosphorylation sites on RIP140 (27) and found phosphorylation of RIP140 also affected its repressive activity. It is likely the regulatory activities of RIP140 are subjected to modulation by extensive protein modification. Indeed in the complex cellular environment, a particular protein might be subjected to a wide range of post-translational modifications, producing a heterogeneous population. Such a protein might well be diversified to interact with many different factors. It is tempting to speculate that RIP140 modification by acetylation at different lysine residues could contribute to its varied effects on the regulation of gene expression. The current study represents the first step in the detailed mapping of potential acetylation sites of RIP140. Such a map will be useful in the generation of acetylated peptide-specific antibodies, which may then be used to map acetylation sites on RIP140 in mammalian cells. Future studies are needed to determine the specific residues responsible for the effects of RIP140 on the expression of various genes.