Modulation of Histone Deacetylase 6 (HDAC6) Nuclear Import and Tubulin Deacetylase Activity through Acetylation*

Background: HDAC6 is an important deacetylase for cytoplasmic and nuclear function. However, how it is regulated is unknown. Results: HDAC6 acetylation sites were determined, and acetylation affected HDAC6 enzymatic activity and nuclear transport. Conclusion: Acetylation is an important post-translational modification that regulates HDAC6 function. Significance: This study provides insight into regulation of HDAC6 function. The reversible acetylation of histones and non-histone proteins by histone acetyltransferases and deacetylases (HDACs) plays a critical role in many cellular processes in eukaryotic cells. HDAC6 is a unique histone deacetylase with two deacetylase domains and a C-terminal zinc finger domain. HDAC6 resides mainly in the cytoplasm and regulates many important biological processes, including cell migration and degradation of misfold proteins. HDAC6 has also been shown to localize in the nucleus to regulate transcription. However, how HDAC6 shuttles between the nucleus and cytoplasm is largely unknown. In addition, it is not clear how HDAC6 enzymatic activity is modulated. Here, we show that HDAC6 can be acetylated by p300 on five clusters of lysine residues. One cluster (site B) of acetylated lysine is in the N-terminal nuclear localization signal region. These lysine residues in site B were converted to glutamine to mimic acetylated lysines. The mutations significantly reduced HDAC6 tubulin deacetylase activity and further impaired cell motility, but had no effect on histone deacetylase activity. More interestingly, these mutations retained HDAC6 in the cytoplasm by blocking the interaction with the nuclear import protein importin-α. The retention of HDAC6 in the cytoplasm by acetylation eventually affects histone deacetylation. Thus, we conclude that acetylation is an important post-translational modification that regulates HDAC6 tubulin deacetylase activity and nuclear import.

Histone deacetylases (HDACs) 2 catalyze the removal of acetyl groups from the ⑀-amino groups of lysine residues from histone and a variety of non-histone proteins. Mammalian HDACs are classified into four classes (I-IV) based on the sequence homology of the yeast histone deacetylases Rpd3 (reduced potassium dependence 3), Hda1 (histone deacetylase 1), and Sir2 (silent information regulator 2). Class II HDACs are further divided into two subclasses (IIa and IIb) based on the similarity of protein sequence and function. Class IIa contains HDAC4, HDAC5, HDAC7, and HDAC9, whereas Class IIb contains HDAC6 and HDAC10 (reviewed in Ref. 1).
The Class IIb member HDAC6 possesses two deacetylase domains and a zinc finger motif (ZnF-UBP, zinc-finger ubiquitin binding domain) (2). HDAC6 deacetylates ␣-tubulin, cortactin, and Hsp90 to regulate cell motility, cilium assembly, cell adhesion, immune synapses, macropinocytosis, maturation of the glucocorticoid receptor, and activation of some protein kinases (3)(4)(5)(6). Depending on its availability in the nucleus, HDAC6 may also deacetylate histones (7). In addition to its deacetylase domains, HDAC6 possesses a ZnF-UBP finger that binds to ubiquitin and is involved in ubiquitin-dependent aggresome formation and cellular clearance of misfolded proteins (3,8). Both the deacetylase and ubiquitin-binding activities of HDAC6 are required for these processes. Therefore, HDAC6 regulates various processes in the cytoplasm. Several lines of evidence suggest that HDAC6 can also reside in the nucleus and interact with nuclear proteins, including HDAC11 (9), sumoylated p300 (10), the transcriptional corepressor LCoR, and transcription factors such as NF-B and Runx2 (11). It has been suggested that nuclear import and export signals within HDAC6 regulate the cellular localization of HDAC6 (12,13). However, how the shuttling of HDAC6 between the nucleus and cytoplasm is regulated remains largely unknown. In addition, it is not clear how HDAC6 activity is modulated.
In eukaryotic cells, acetylation is among the most common covalent modifications. Deacetylation of non-histone protein plays a diverse role in the regulation of all aspects of cellular processes (14). Acetylation of non-histone protein affects a broad spectrum of protein function, including signaling, enzymatic activity, protein-protein or protein-DNA interaction, protein stability, and protein localization (reviewed in Ref. 15). Acetylation modification has been found in several deacetylases. It had been shown that HDAC1 can be acetylated at the catalytic domain and the C-terminal region, resulting in a dramatic reduction in enzymatic activity (16). Other deacetylases such as HDAC6 and Sirt2 can also be acetylated by p300; how-ever, the function of the modulation is largely unknown (17,18).
Here, we show that HDAC6 is acetylated by p300 on five clusters of lysine residues. One cluster (site B) of acetylated lysine is in the N-terminal nuclear localization signal region. Acetylation of site B lysine residues significantly reduced HDAC6 tubulin deacetylase activity and further impaired cell motility, but had no effect on histone deacetylase activity. More interestingly, these mutations retained HDAC6 in the cytoplasm by blocking the interaction with the nuclear import protein importin-␣ and eventually affected histone deacetylation. Thus, our results indicate that acetylation is an important posttranslational modification that regulates HDAC6 tubulin deacetylase activity and nuclear import.
Cell Culture and Transient Transfections with siRNA and Plasmid DNA-293T, HeLa, and HCT116 cells were grown in DMEM supplemented with 10% fetal bovine serum (HyClone). A549 cells stably expressing siRNA for HDAC6 were established as described previously (19) and maintained with 0.2 mg/ml G418 in DMEM with 10% fetal bovine serum. A549-HDAC6 knockdown cells stably overexpressing GFP-tagged WT HDAC6 and mutant ABCDE were obtained by transient transfection, followed by flow cytometry sorting with anti-GFP antibody. For both cell lines, Ͼ80% of the cells were GFP-positive (mutant ABCDE, 87.3%; and WT, 86.9%). Jurkat and K562 cells were grown in RPMI 1640 medium with 10% fetal bovine serum. Transient transfection was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
Identification of Acetylated Lysine Residues by Mass Spectrometry-Gel slices containing acetylated HDAC6 were prepared and subjected to tandem mass spectrometry analysis as described (4,16).
Fluorescence Microscopy-NIH-3T3 cells were grown on 12-mm coverslips and transfected with GFP HDAC6 constructs. At 2 days after transfection, the cells were washed with 10 mM MES, 10 mM NaCl, 1.5 mM MgCl 2 , and 2.5% glycerol and fixed with 4% paraformaldehyde for 10 min at room temperature, followed by permeabilization with 0.5% Triton X-100 for 15 min on ice. Cells were then mounted after a 30-min incubation with DAPI. All imaging experiments were carried on a Zeiss Axioplan 2 fluorescence microscope.
Recombinant Protein Purification-Expression and purification of FLAG-tagged HDAC6 and mutants from baculovirus-infected insect cells were performed as described previously (16).
In Vitro Tubulin Deacetylation Assay-Microtubule-associated protein-rich tubulin (ML113, Cytoskeleton, Inc.) was polymerized in general tubulin buffer (80 mM PIPES (pH 6.9), 2 mM MgCl 2 , and 0.5 mM EGTA) with 1 mM GTP and 10 M Taxol (Sigma-Aldrich) at a concentration of 1 mg/ml. 200 ng of polymerized microtubules was incubated with HDAC6 and mutants in tubulin deacetylase buffer (20 mM Tris-Cl (pH 8.0) and 60 mM NaCl) at 37°C for 2 h and then placed on ice for 15 min. The resulting products were subjected to SDS-PAGE and Western blotting.
Cell Migration Assay-The Transwell cell migration assay was performed. The 8-m Transwell chambers were purchased from Corning. Briefly, 1.2 ϫ 10 5 A549 cells were suspended in 100 l of culture medium and plated in the upper chamber. 600 l of medium was added to the lower chamber of the migration plate. After 1 or 2 days of incubation, non-migrating cells (located inside the upper chambers) were gently removed by cotton swabs. The live cells located in the bottom chambers and in the exterior of the upper chambers were quantified by MTS assay (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; Promega) with a plate reader.

HDAC6 Is Acetylated by p300 on Five Clusters of Lysine
Residues-It has been reported that HDAC6 can interact with p300 (10) and be acetylated by p300 in vivo (18). However, the function of acetylated HDAC6 has not been studied. We first wanted to identify the acetylated lysine residues of HDAC6. Recombinant human HDAC6 purified from baculovirus-infected insect cells was incubated with p300 in vitro, and the reaction resulted in strong acetylation of HDAC6 (Fig. 1A). The acetylated lysine residues of HDAC6 were determined by mass spectrometry. HDAC6 was acetylated at multiple lysine residues in five clusters ( Fig. 1B and supplemental Fig. S1). Sites A and B are in the nuclear localization domain. Site C is located in the second deacetylase domain, and sites D and E are located at the junction between the second deacetylase domain and the SE14 domain. These acetylation sites are conserved throughout mammalian species (supplemental Fig. S2), indicating the conserved function in HDAC6 acetylation. Each cluster of lysines was mutated to glutamines to mimic the acetylated form of lysine. The recombinant wild-type and mutant HDAC6 proteins were purified from baculovirus-infected insect cells and tested for acetylation modification. No single-cluster mutation affected HDAC6 acetylation. The mutations of sites A-D combined partially affected the acetylation level, and mutations of all five clusters abolished HDAC6 acetylation (Fig. 1C), confirming that all five sites are subject to acetylation.
HDAC6 Acetylation Affects Tubulin Deacetylase Activity and Cell Mobility-To test whether the mutations affect HDAC6 tubulin deacetylase activity, GFP HDAC6 and mutant constructs were transfected into 293T cells, and tubulin acetylation levels were determined. The results indicate that all single-site Lys-to-Gln mutations, except mutant B (mutB), did not affect tubulin deacetylase activity ( Fig. 2A). Mutations of site B decreased the tubulin deacetylase activity of HDAC6. Mutant ABCDE (mutABCDE) and mutB were also transfected into A549-HDAC6 knockdown cells (19). The overexpression of these mutants reduced tubulin deacetylase activity (Fig. 2B). To further confirm that the reduction in tubulin deacetylase activity directly resulted from the mutation of the protein, an in vitro assay was also performed with recombinant HDAC6 mutants and polymerized microtubules. The results show that, consistent with the in vivo results, mutB and mutABCDE had reduced tubulin deacetylase activity (Fig. 2C). Thus, the acetylationmimicking mutations of HDAC6 affect tubulin deacetylase activity in vitro and in vivo.
Because it has been shown that HDAC6 can positively regulate cell mobility through deacetylation of tubulin (22,23), we rationalized that the reduction in tubulin deacetylase activity through acetylation of HDAC6 will result in a change in cell mobility. To test this, GFP-HDAC6 and GFP-mutABCDE were stably transfected into A549-HDAC6 knockdown cells. Cell mobility was tested in these two cell lines. Cells expressing mutABCDE had lower mobility than cells overexpressing wildtype HDAC6 (Fig. 2C). This result indicates that acetylation of HDAC6 reduces tubulin deacetylase activity and subsequently affects cell mobility.
HDAC6 Acetylation Does Not Affect Histone Deacetylase Activity-Because HDAC6 can also deacetylate histone, we next tested whether acetylation of HDAC6 affects its histone deacetylase activity. Recombinant HDAC6 was acetylated in vitro; tritium-labeled acetylated histone was then added to test the deacetylase activity of HDAC6. Acetylation did not affect HDAC6 histone deacetylase activity (Fig. 3A). The mutations that mimicked acetylation also did not affect the histone deacetylase activity (Fig. 3B). The mutation of each site did not affect histone deacetylase activity as well (supplemental Fig.  S3). These results indicate that HDAC6 acetylation affects tubulin deacetylase activity, but not histone deacetylase activity.
Acetylation Retains HDAC6 in the Cytoplasm by Blocking the Interaction with Importin-␣-Because HDAC6 acetylation sites A and B are located in the N-terminal nuclear localization domain (Fig. 4A), we next wanted to investigate whether acetylation affects HDAC6 cellular localization. HDAC6 possesses one nuclear localization signal and two nuclear export signals (Fig. 4A) (12,13). The nuclear export signals may play a dominant role because the majority of HDAC6 is localized in the cytoplasm. To investigate the function of acetylation of the nuclear localization domain, we created a construct that contained the N-terminal 66 amino acids, including a nuclear localization domain, but not nuclear export signals (13). When transfected into 3T3 cells, GFP-HDAC6(1-66) was mostly in the nucleus (Fig. 4B). This agrees with a previous observation showing that this region is important for nuclear import (13). The site A acetylation-mimicking mutant was partially localized in the nucleus. Interestingly, the site B and site AB mutants were localized mostly in the cytoplasm (Fig. 4B). These results indicate that acetylation of site B abolishes the nuclear localization signal, resulting in cytoplasm retention of HDAC6. A, FLAG-tagged recombinant HDAC1 and HDAC6 were incubated with or without p300 and CoA. The products were subjected to immunoblotting with anti-acetyl-lysine antibody. B, schematic representation of HDAC6 functional domains and clusters of acetylated lysine residues. Five clusters of lysine residues (sites A-E) were identified as acetylated sites by mass spectrometry. NLS, nuclear localization signal; NES, nuclear export signal. C, recombinant HDAC6 and Lys-to-Gln (K to Q) mutants were subjected to acetylation reactions with p300 and [ 3 H]acetyl-CoA. The resulting products were separated by SDS-PAGE and subjected to autoradiography. HDAC6 proteins were stained with Coomassie Blue.
Although HDAC6 resides mainly in the cytoplasm (3), it has been reported that HDAC6 is capable of shuttling between the nucleus and the cytoplasm (12,24). It has been show that HDAC6 can partially reside in the nucleus and interact with nuclear proteins in a human T cell leukemia cell line (Jurkat cells) (25). We then further investigated whether leukemia cells contain more nuclear HDAC6 than other cell types. The results show that, in addition to Jurkat cells, the human leukemia cell line K562 also has a higher level of nuclear HDAC6 compared with non-hematopoietic cell lines such as HCT116 and 293T (Fig. 4C). To study whether HDAC6 acetylation affects nuclear localization of HDAC6, GFP wild-type HDAC6, mutB, and mutABCDE were transfected into K562 cells, and their cellular localization was examined. The expression levels of GFP HDAC6 and mutants were comparable in the cell extract; however, wild-type HDAC6, but not mutB or mutABCDE, was detected in the nuclear extract (Fig. 4D). These results indicate that acetylation of HDAC6 affects its nuclear localization signal and results in cytoplasmic retention of HDAC6. Acetylations impair HDAC6 tubulin deacetylase activity and decrease cell mobility. A, 293T cells were transfected with GFP-tagged HDAC6 and mutants. HDAC6 and acetylated tubulin (Ac-tubulin) levels were examined by immunoblotting. The ␣-tubulin levels were included as a loading control. Acetylated tubulin levels were quantified and expressed as a ratio of the acetylated tubulin level versus the total tubulin level. B, A549-HDAC6 knockdown (KD) cells were transfected with GFP HDAC6, mutB, or mutABCDE. HDAC6, acetylated tubulin, and acetylated histone levels were examined by immunoblotting. The ␣-tubulin levels were included as a loading control. Acetylated tubulin levels were quantified and expressed as a ratio of the acetylated tubulin level versus the total tubulin level. C, the tubulin deacetylase activities of recombinant wild-type HDAC6 and mutants were determined in vitro by deacetylation of polymerized microtubules. D, cell migration assay performed with A549-HDAC6 knockdown cells stably overexpressing GFP-tagged wild-type HDAC6 and mutABCDE. The number of migrated cells was determined by absorbance at 490 nm. * Significant difference compared with 1d (Student's t-test, p Ͻ 0.01). . Acetylation does not affect HDAC6 histone deacetylase activity in vitro. A, FLAG-tagged recombinant HDAC6 was first acetylated by p300, and the reaction was then subsequently diluted with deacetylase buffer and incubated with [ 3 H]acetyl-histone to examine the histone deacetylase activity. No significant deacetylase activity change was observed (p Ͼ 0.05, Student's t test). B, the histone deacetylase activities of recombinant wildtype HDAC6 and mutABCDE were determined by deacetylation of [ 3 H]acetyl-histone.
Next, we investigated whether the loss of the nuclear localization signal by acetylation is due to the loss of interaction with the nuclear import protein. FLAG-tagged recombinant HDAC6 and mutB or mutABCDE were immobilized on M2-agarose beads and incubated with whole cell extracts from K562 cells. HDAC6 pulled down a significant amount of impor- tin-␣. However, the interaction between importin-␣ and mutB or mutABCDE was significantly reduced (Fig. 5). These results suggest that importin-␣ is important for HDAC6 nuclear import and that acetylation of HDAC6 reduces the interaction with importin-␣. Thus, acetylated HDAC6 is unable to be imported into the nucleus, resulting in a reduction in nuclear HDAC6 levels.
Retention of HDAC6 in the Cytoplasm by Acetylation Affects Histone Deacetylation-Although acetylation of HDAC6 did not affect its histone deacetylase activity (Fig. 3), the reduction in the nuclear fraction of HDAC6 protein by acetylation may eventually affect the overall histone deacetylase activity in the nucleus. To test this, GFP-HDAC6 and GFP-HDAC6 mutants mimicking acetylated or catalytically inactive HDAC6 were overexpressed at a high level into K562 cells, and the nuclear localization of HDAC6 proteins and histone deacetylase activity were examined. Consistent with the results in Fig. 4, a portion of GFP wild-type HDAC6, but not mutB or mutABCDE, localized in the nucleus. The HDAC6(H216A/H611A) mutant, which is a catalytically defective mutant, was also detected in the nucleus (Fig. 6). However, the histone acetylation level in the nucleus was reduced only in cells overexpressing GFPtagged wild-type HDAC6 (Fig. 6). These results indicate that nuclear HDAC6 can play an important role in deacetylation of histone. In addition, acetylation of HDAC6, which inhibits HDAC6 nuclear localization, prevents histone deacetylation by HDAC6.

HDAC6 Is Acetylated on Five Clusters of Lysine Residues-
Here, we have shown that HDAC6 can be acetylated by p300 on five clusters of lysine residues. Sites A and B are localized within the N-terminal nuclear localization domain, and mutations that mimic acetylation abolish nuclear import of HDAC6. Site C is located in the second deacetylase domain and is conserved only within the second deacetylase domain of mammalian HDAC6, suggesting a unique role for this deacetylase domain. However, we do not anticipate that acetylation of this site will affect deacetylase activity. The function of acetylation of this site is still under investigation. Sites D and E are located at the junction between the second deacetylase domain and the SE14 domain and are less conserved. Mutation of these two sites also does not affect enzymatic activity. However, because HDAC6 has diverse cytoplasmic and nuclear functions, whether mutation of these sites affects HDAC6 function other than deacetylase activity is under investigation.
Tubulin and Histone Deacetylase Activities May Be Regulated through Different Mechanisms-Acetylation of HDAC6 affects tubulin deacetylase activity, but not histone deacetylase activity. Although it is debatable whether both deacetylase domains contribute equally to the deacetylase activities of tubulin and histone (5, 22, 26 -28), our data suggest that histone and tubulin deacetylase activities may be regulated differently, as acetylation of HDAC6 affects only tubulin deacetylase activity, but not histone deacetylase activity. Acetylation of site B seems to play a major role in down-regulation of tubulin deacetylase activity (Fig. 2, A and B). Interestingly, site B is localized outside of the catalytic domain. Similarly, HDAC1 acetylation sites also localize outside of catalytic domain, and acetylation of HDAC1 attenuates its deacetylase activity (16). Thus, it may be common for HDACs that their activities can be regulated through a regulatory domain that is outside of the core deacetylase domain.
Acetylation of HDAC6 Affects Nuclear Shuttling-HDAC6 has a strong nuclear export signal, which results in cytoplasmic retention of HDAC6. In addition, another region of human HDAC6, the SE14 domain, also contributes to the strong anchorage of HDAC6 in the cytoplasm (13). However, we and others have shown that a portion of HDAC6 can localized in the nucleus (12,25). Some cells, especially hematopoietic cells, have significant amounts of nuclear HDAC6. It has been shown that HDAC6 can be recruited to gene promoters and gene bodies, suggesting a role for histone deacetylation and transcription regulation (25). Others have also shown that HDAC6 can interact with a number of nuclear factors and modulate their activity (9 -11, 29, 30). The localization of HDAC6 in the nucleus is mainly because HDAC6 has a nuclear localization signal located in the N-terminal region. This region can interact with importin-␣ and be shuttled into the nucleus. Interestingly, this region is heavily acetylated, and acetylation blocks the interaction with importin-␣ and results in a reduction in nuclear HDAC6. In turn, the reduction in nuclear HDAC6 affects his- FIGURE 5. Acetylation retains HDAC6 in the cytoplasm by deficient nuclear import. FLAG-tagged wild-type HDAC6 (HD6WT-Flag) and mutants were expressed in and purified from baculovirus-infected insect cells and immobilized on M2-agarose beads. Beads conjugated with FLAG peptide, beads alone, or beads incubated with uninfected insect cell extracts were also prepared as controls. The beads were subsequently incubated with K562 nuclear extracts, and the resulting complexes were subjected to Western blotting with antibodies as indicated. FIGURE 6. HDAC6 acetylation affects histone deacetylase activity in vivo through cytoplasmic retention. Cellular and nuclear extracts from K562 cells overexpressing GFP alone or conjugated with wild-type HDAC6 (HD6) or mutants were subjected to Western blot analysis with antibodies as indicated. H216/611A, H216A/H611A; Ac-H3, acetyl-histone H3.
tone deacetylation and subsequently affects gene function. Therefore, acetylation is an important post-translational modification that regulates both the cytoplasm and nuclear function of HDAC6.