Introduction

The reversible acetylation and deacetylation of specific lysine residues on core histone N-terminal tails is catalyzed by histone acetyltransferases (HATs) and histone deacetylases (HDACs), respectively. The competing action between HATs and HDACs determines histone acetylation levels that ultimately affect higher-order chromatin structure and gene activities 1, 2, 3. In general, hyperacetylated histones are associated with gene activation, whereas hypoacetylated histones are related to gene repression. After the first histone deacetylase (HDAC1), a yeast RPD3 homologue, was identified in mammals 4, a large number of HDACs have been characterized in many eukaryotes, and 18 HDAC-like genes were predicted in Arabidopsis 5. HDACs can be divided into 4 classes based on their sequence homology, substrate specificity, and cofactor requirement 6, 7. Class I and class II HDACs are homologous to the yeast RPD3 and HDA1 proteins, respectively, while class III HDACs are related to yeast SIR2 protein. The fourth class of HDACs, known as HD2, is found only in plants 8.

In yeast and mammalian cells, the Rpd3-like HDACs mediate transcriptional repression by interacting with specific DNA-binding proteins or associated with corepressors and by recruiting to target promoters 9, 10, 11. RPD3 and Sin3 homologues are associated with a large multiprotein complex that represses target genes involved in diverse processes such as meiosis, cell-type specificity, potassium transport, phosphate and phospholipid metabolism, methionine biosynthesis, and responses to stress and developmental changes 12, 13, 14, 15, 16, 17, 18.

In plants, the first RPD3 homolog was identified in maize, which complemented the phenotype of an rpd3 null mutant of yeast 19. Arabidopsis has four RPD3-like HDACs including AtHD1 (AtRPD3A, HDA19), AtHD6 (AtRPD3B), HDA7 and HDA9 5, each of which might have evolved to possess specific functions. For instance, HDA6 was associated with transgene silencing, DNA methylation, and regulation of rRNA genes 20, 21, 22, 23.

Down-regulation of AtHD1 by antisense inhibition and T-DNA insertion in Arabidopsis led to various phenotypic and developmental defects including early senescence, serrated leaves, formation of aerial rosettes, delayed flowering, and defects in floral organ identity 24, 25. Further analysis by DNA microarray showed that the down-regulation of AtHD1 induced up- and down-regulation of >7% transcriptome including the genes involved in cellular biogenesis, protein synthesis, ionic homeostasis, protein transcription, and plant hormonal regulation 26. The data suggest that AtHD1 is a global transcriptional regulator that provides reversible gene regulation in response to changes in developmental programs and physiological processes. AtHD1 may also play a role in plant defense as it can be up-regulated by wounding, infection of a pathogenic fungus, and plant hormones induced in pathogen response 27.

The histone deacetylase activity of AtHD1 is implied by the concomitant changes in histone acetylation and AtHD1 expression levels. The up- and down-regulation of AtHD1 was associated with the hypoacetylation and hyperacetylation, respectively, of histone H4 in transgenic Arabidopsis 25, 26, 27. Here, we report histone deacetylase activity of recombinant AtHD1 in vitro. Subcellular localization of AtHD1 suggests that AtHD1 is a nuclear protein and does not interact with nucleolus, suggesting a specific role of AtHD1 in global gene regulation that is different from that of AtHD2 or AtHDA6 and other AtHD1-related homologs.

Materials and Methods

Plant materials

Seeds of Arabidopsis thaliana ecotypes (Col, Ler, and Ws) were obtained from Arabidopsis Biological Resources Center (ABRC) and sown directly in vermiculite mixed with 10% soil. Seeds from transgenic Arabidopsis were sterilized in 10% sodium hypochlorite (commercial bleach), rinsed three times in sterilized water, and placed for germination in Murasige/Skoog (MS) medium (Sigma, St. Louis) containing 50 mg/L kanamycin 25. After two weeks, the seedlings were transferred to soil mix and grown in growth chamber. All plants were grown in a growth chamber with a cycle of 14 h of light at 22 °C and 10 h of dark at 18 °C.

Construction of 35S::AtHD1-GFP chimeric gene

The AtHD1 full-length cDNA (AF014824) was amplified from a plasmid clone pAtHD1-2 (Tian and Chen, 2001) by PCR using primer pair 5′-GCG TCG AC A TGG ATA CTG GGG GCA ATT C-3′ (forward) and 5′-CAT GCC ATG GCT GTT TTA GGA GGA AAC GCC TG-3′ (reverse). The resulting PCR product was digested with SalI and NcoI and subcloned into the plasmid pUC18/CaMV35S-GFP(S65T) so that the 3′-end of AtHD1 coding sequence was fused to the 5′-end of GFP in frame, and the whole AtHD1-GFP chimeric gene was under the control of the cauliflower mosaic virus (CaMV) 35S promoter. The expression cassette of 35S::AtHD1-GFP was excised from the pUC18 plasmid by XbaI and EcoRI double digestion and subcloned into the binary vector pBI101 for plant transformation. A 35S::GFP construct was used as control for localization studies.

DNA and RNA blot analyses

Genomic DNA (2 mg) from wild type and SHG transgenic plants were digested with EcoRI, separated by electrophoresis in a 1% agarose gel, and transferred to Hybond-N+ membrane (Amersham Pharmacia). A radiolabeled probe (0.7-kb) specific for the GFP sequence was used in hybridization (Figure 1A) by a random priming method (Amersham). The plasmid pUC18/35S::AtHD1-GFP digested with EcoRI was used as a positive control (lane V). DNA markers (M) used were Lamda DNA digested with EcoRI and HindIII (Promega).

Figure 1
figure 1

(A) The diagram of AtHD1-GFP chimeric gene cassette. 35SP: CaMV 35S promoter and NOS Ter: nopaline synthase terminator (NOS ter). Sizes of the AtHD1-GFP chimeric gene and recombinant protein are 2.2 kb and 83.1 kDa, respectively. The genomic fragment containing the transgene and the probe used for DNA and RNA blot analyses are shown above and below the diagram, respectively. (B) Genomic DNAs from a control plant and eight transgenic plants were digested with EcoRI, blotted, and hybridized with the probe indicated in Figure 1A. The transgene (a 1.45 kb fragment containing the GFP coding sequence) was detected in four of eight SHG T1 lines examined. (C) Northern blot analysis using the radiolabeled probe specific for the GFP sequence showed that the AtHD1-GFP transcript was present in three of four SHG T1 transgenic lines examined. The ethidium bromide-stained RNA gel depicting the rRNAs is shown as loading controls.

Full size image

Total RNA (20 mg) from wild type and SHG transgenic plants (with the presence of transgene verified by Southern blot) was separated by electrophoresis in a 1.5% agarose gel containing 2% formaldehyde and then transferred to a Hybond-N+ membrane (Amersham Pharmacia). The membrane was hybridized with a radio-labeled probe (Figure 1A), and the signals were detected as previously described 25.

Transient expression assays

The pUC18/35S::AtHD1-GFP and 35S::GFP plasmids were introduced into onion cells by particle bombardment. Fresh onion bulbs purchased from local grocery store were cut into slices of 1 cm2 and placed on the central area of a petri dish with 0.9% phytal agar as supporting medium. Tungsten particles of 1.1 mm in diameter were prepared and coated with plasmid DNA according to the manufacturer's recommendations (BioRad). Bombardments were performed using the Biolistic PDS-1000/He Particle Delivery System (BioRad) with the following parameters: holder level, 2; sample level, 4; gap distance, 0.63 cm; target distance, 6 cm; helium pressure, 1100 psi; chamber vacuum, 27 inches Hg (∼ 0.06 atm); amount of DNA/bombardment, 1mg; and amount of macrocarrier per bombardment, 500 mg of tungsten particles. Each sample received two times of bombardment, after which the samples were incubated in a growth chamber at 25 °C for 20–24 h. A single layer of epidermal cells was peeled from the onion scale leaves and examined for AtHD1-GFP expression under a fluorescence microscope.

Arabidopsis transformation and selection

Ten Arabidopsis thaliana (Columbia) plants were transformed with Agrobacteria harboring the binary vector pBI101/35S::AtHD1-GFP using the floral dip method 28. Seeds of T0 plants were selected by germination on MS medium (Sigma) with 50 mg/L kanamycin. Successful transformants (T1 plants) were regenerated and grown in soil. Transformation efficiency is about 1%. The presence of AtHD1-GFP transgene was detected in the T1 plants by PCR using the following primers flanking the GFP coding sequence: 5′-ATG GTG AGC AAG GGC GAG-3′ (forward) and 5′-TTA CTT GTA CAG CTC GTC CA-3′ (reverse).

Detection of GFP signals

Transient expression of AtHD1-GFP fusion protein in onion epidermal cells was detected with the Olympus AX7 fluorescence microscope. A single epidermal layer of the transformed onion scale leaf disc was peeled off and put on a slide for examination under 60 × or 100 × magnifications. Nuclei were stained by adding one drop of 4′,6-diamidino-2-phenylindole (DAPI) staining solution (1 μg/ml DAPI in PBS) to the onion cells, and DAPI fluorescence images were obtained by UV illumination of the specimen and a band-pass filter for DAPI. Green fluorescence signals were obtained by blue light illumination and a band-pass filter for GFP. Stable expression of AtHD1-GFP fusion protein in transgenic Arabidopsis was detected by confocal scanning microscopy. Dissected intact tissues were stained with DAPI staining solution and examined by differential interference contrast (DIC) and fluorescence microscopy sequentially using the Zeiss Axiophot microscope with 40 × or 100 × magnifications. Images of tissues at different confocal layers (Z-series) were taken with a 0.5 mm step size. GFP fluorescence images were acquired using a GFP filter set (Chroma Technologies, excitation HQ470/20, dichroic Q495LP, emission HQ525/50) and a Photometrics Coolsnap digital camera.

Production of recombinant AtHD1 and western blot analysis

The AtHD1 full-length cDNA was subcloned in frame into the pET21a expression vector (Novagen), which was then transformed into Escherichia coli (strain BL21). Recombinant AtHD1 production in bacteria was induced by isopropyl-β-D-thiogalactoside. Total protein was extracted from bacterial cells by sonication, and recombinant AtHD1 protein was purified by Ni-NTA affinity column (Invitrogen) followed by amylose affinity column (New England Biolabs). Affinity tag (6×His) was subsequently removed by tobacco etch virus protease cleavage. Recombinant AtHD1 protein was separated by 15% SDS-PAGE and transferred to PVDF membrane (Invitrogen) using a Trans-Blot SD Semi-dry Transfer Cell (Bio-Rad). Western blot analysis was performed according to the protocol of Western Breeze kit (Invitrogen). The primary antibody was anti-HD1 as previously described in Tian and Chen (2001). The secondary antibody used was an anti-rabbit antibody conjugated with alkaline phosphatase. Detection was carried out by adding a chemiluminescent substrate (CPD-star) to the PVDF membrane, and the signals were detected with Kodak X-ray film. Recombinant AtHDA6, AtHDA9, and AtHDA17 were also produced using the same procedure in E. coli cells and baculovirus expression system in SF9 insect cells (Invitrogen Inc.). The data were not shown because none of other proteins showed any measurable levels of histone deacetylase activity.

Histone deacetylase activity assays

Synthetic histone H3 and H4 peptides were acetylated and labeled according to the protocol of the HDAC Activity Assay kit (Upstate Biotechnology). Recombinant AtHD1 protein was added to the same amount of tritium-labeled acetylated histone H3 or H4 peptides to test for HDAC activity. For each test sample, two sets of duplicate assays with 50 mM sodium butyrate (SB), a specific inhibitor of HDAC, were included to demonstrate the specificity of deacetylation. HeLa nuclear extract that contains a blend of histone deacetylases was used as the positive control. The negative control was water (data not shown) and water plus SB. After 16–20 h of incubation at room temperature, the reaction mixtures were centrifuged and the radioactivity of the supernatant was measured by scintillation counting using the Beckman LS 6500 Scintillation counter.

Results

Expression of AtHD1-GFP fusion protein in transgenic Arabidopsis

A chimeric gene consisting of an engineered green fluorescent protein (GFP) sequence was fused to the C-terminus of AtHD1 cDNA (Figure 1A). Transgenic Arabidopsis plants carrying the same AtHD1-GFP chimeric gene were generated by Agrobacterium-mediated transformation 29, 30. These plants were named “SHG” for the transgene (35S::AtHD1-GFP) they contain. In the first round of selection (T0), the transformation efficiency was about 1%, which is a typical value for the floral dipping method. The seeds collected from the first generation of transformants (T0) were germinated in selective medium to produce T1 transgenic plants, which were used for the assay of AtHD1-GFP gene expression.

The results of Southern blot analysis (Figure 1B) show that the 35S::AtHD1-GFP transgene was present in four of the eight transgenic lines examined. The same probe used in Southern blot analysis was used in RNA blot analysis. The AtHD1-GFP transcripts were detected in three of four transgenic lines examined (Figure 1C). SHG7 may contain the transgene (lane 7, Figure 1B) but the RNA was degraded (Figure 1C). Alternatively, the transgene is silenced in this line. SHG lines 5 and 8 were selected for AtHD1 cellular localization assays.

Subcellular localization of AtHD1

To study the subcellular localization of AtHD1, the chimeric gene (Figure 1A) was expressed transiently in onion cells and stably in Arabidopsis. In the transient expression assay, the 35S::AtHD1-GFP chimeric gene was introduced into onion epidermal cells by particle bombardments. The cells were stained with DAPI (4′,6-diamidino-2-phenylindole) to reveal the nuclei and then examined using a fluorescence microscope. GFP signals were detected in the nuclei of onion cells (Figure 2, second panel), whereas the control (35S::GFP) displayed diffused patterns of localization (data not shown). The AtHD1-GFP chimeric gene was introduced into Arabidopsis plant genome by Agrobacterium-mediated transformation, and the transgenic plants were named SHG for the transgene (35 S ::At H D1- G FP) they contained. Different tissues of the SHG transgenic plants were dissected, stained with DAPI, and then examined by confocal fluorescence microscopy. In all tissues examined, including roots, leaves, flowers and developing seeds in the siliques (Figure 2, third-fifth panels), GFP signals were detected and localized in the nuclei, which is consistent with recent findings 27. We further analyzed AtHD1 localization patterns using high-resolution confocal image analysis. We found that AtHD1-GFP protein appeared to be excluded from the nucleolus because GFP signals were absent in the nucleolus (Figure 3). This is in contrast to the HD2-type HDACs that are localized in the nucleolus in maize 8 and Arabidopsis 31. Moreover, the localization pattern of AtHD1 is different from that of AtHDA6 that is primarily associated with rDNA and heterochromatic regions 23. To visualize detailed staining patterns, mesophyll cells at early prophase of the cell cycle (Figure 2Q-2S) are shown in Figure 3E-3G (enlarged scales). The DAPI fluorescence image (Figure 3F) shows that the chromatin began to condense into chromatids, and the nucleolus was obviously formed in the nucleus. The corresponding GFP fluorescence image (Figure 3G) shows that the AtHD1-GFP protein was absent in the nucleolus, and it aggregated into speckles in the nucleoplasm. The AtHD1-GFP protein may be associated with the condensing chromatids, suggesting a role of AtHD1 in mitosis.

Figure 2
figure 2

Localization of AtHD1-GFP fusion proteins. Transient expression of AtHD1-GFP fusion protein. (A-D) localization of AtHD1-GFP fusion protein in onion cells. (A) and (C) onion cells were stained with DAPI and viewed under the UV light. The bluish white spots are the nuclei (arrows). (B) and (D), GFP signals (arrows) detected in the same areas as shown in (A) and (C), respectively. Localization of AtHD1-GFP fusion protein in Arabidopsis transgenic plants. (E-J) expression of AtHD1-GFP fusion protein in root epidermal cells. Confocal microscopic images detected by DIC (E and H), DAPI-staining (F and I), and GFP-filter (G and J) in Arabidopsis SHG transgenic (E-G) and WT (H-J) plants, respectively. GFP signals were localized in the nuclei of the root epidermal cells in the SHG plants (G) but not in the WT plant (J). Note the auto-fluorescent interference was very low (J) because of no chloroplasts in the roots. (K-P) expression of AtHD1-GFP fusion protein in developing seed coats. Confocal microscopic images detected by DIC (K and N), DAPI-staining (L and O), and GFP-filter (M and P) in Arabidopsis SHG transgenic (K-M) and WT (N-P) plants, respectively. Nuclei are indicated by arrows. GFP signals were detected in the nuclei of SHG plant cells (M) but not in the WT plants (P). The background was caused by auto-fluorescence of chlorophylls in the chloroplasts. (Q-V), expression of AtHD1-GFP fusion protein in leaf mesophyll cells. Confocal microscopic images detected by DIC (Q and T), DAPI-staining (R and U), and GFP-filter (S and V) in Arabidopsis SHG transgenic (Q-S) and WT (T-V) plants, respectively. GFP signals were detected in the nuclei of the SHG plants (S) but not in the WT plants (V). The nuclei (Nu) and nucleoli (No) were clearly visible in the DIC, DAPI-stained, and GFP-activated image of the SHG plants.

Full size image
Figure 3
figure 3

AtHD1-GFP fusion protein is excluded from nucleoli and aggregates during the prophase of cell cycle. (A-B) confocal microscopic images of nuclei (Nu) and nucleoli (No) detected by DIC (A) and GFP-filter (B) in the root cells of SHG transgenic plants. (C-D) confocal microscopic images of nuclei (Nu) and nucleoli (No) detected by DAPI-staining (C) and GFP-filter (D) in the leaf mesophyll cells of SHG transgenic plants. (E-G) enlarged images of nuclei and nucleoli detected by DIC (E), DAPI-staining (F), and GFP-filter (G) in the mesophyll cells of SHG plants (same as in Figure 2 Q-S). (F) DAPI-stained image displayed visible organization of chromatin into chromatids in the prophase of cell division. (G) GFP signals appeared as speckles that were localized in the nucleus but excluded from the nucleolus.

Full size image

Histone deacetylase activity in the recombinant AtHD1 protein

Recombinant AtHD1 was produced by bacterial expression system for studying histone deacetylase activity. The recombinant protein consists of a hexahistidine tag (6 × His) for protein purification, a maltose binding protein (MBP) for increasing the solubility of the recombinant protein so as to prevent the formation of inclusion bodies in bacterial cells, a tobacco etch virus (TEV) cleavage site for removing the N-terminal segment, and the full-length AtHD1 polypeptide (Figure 4A). Recombinant AtHD1 was purified from the bacterial protein extract by affinity column with Ni-NTA resins specific for 6 × His tag followed by further purification by amylose-conjugated agarose resins specific for the MBP fragment. The N-terminal segment containing the 6 × His tag and MBP segment was removed from the recombinant protein by TEV protease. The identity of the AtHD1 protein was confirmed by western blot analysis using an antibody specific for the N-terminal region of AtHD1 25 (Figure 4B).

Figure 4
figure 4

Biochemical assays of recombinant AtHD1. (A) diagram of the expression cassette of recombinant AtHD1 driven by the inducible viral T7 promoter. Upstream AtHD1 there is a hexa-histidine tag (6 × His) for purification by affinity column, a maltose-binding protein (MBP) segment for increasing the solubility of the recombinant protein, and a tobacco etch virus (TEV) protease cleavage site for removal of the 6 × His-MBP segment. Bacteria containing the plasmid encoding recombinant AtHD1 were incubated at 25 °C for 16 h after the addition of IPTG. (B) SDS-PAGE and western blot analyses of recombinant AtHD1. The recombinant AtHD1 protein (rAtHD1) was purified from bacterial crude protein extracts by Ni-NTA affinity column followed by amylose affinity column. The rAtHD1 was subsequently cleaved by TEV protease to remove 6 × His-MBP. The rAtHD1 proteins before and after TEV cleavage had molecular weights of 100 kDa and 57 kDa, respectively, and they were visible as a prominent band in the SDS-PAGE gel stained with Coomassie blue and in the western blot analysis using AtHD1-C as primary antibody. (C) histone deacetylase assays of purified recombinant AtHD1. rAtHD1 was incubated with synthetic histone H3 or H4 peptides labeled with 3H-acetyl groups on specific lysine residues. The amount of 3H-acetate released due to histone deacetylase activity was measured by scintillation counting. Relative ratios were calculated based on the counts from the control (water + SB = 1). The error bars represent the average values ± S.D. of four duplicates of the assays using independent samples. SB: sodium butyrate.

Full size image

To determine histone deacetylase activity in vitro, the recombinant AtHD1 was incubated with synthetic histone H3 or H4 peptides labeled with 3H-acetyl groups on specific lysine residues. The radiolabeled histone peptides were immobilized on agarose beads so that the histone deacetylase activity of the recombinant AtHD1, if any, would be correlated with the amount of 3H-acetyl groups released from the histone peptides to the supernatant. The HeLa nuclear extract that contains a blend of HDACs was served as the positive control, while sodium butyrate that is a specific inhibitor of HDACs was added to the triplicates of each HDAC sample. The amount of 3H-acetate released to the supernatant was measured by scintillation counting as shown in Figure 4C. Compared with the negative control (water) or specificity control (sodium butyrate), the addition of purified rAtHD1 (to a final concentration of 5 mg/ml) led to an increased amount (1.5-fold) of 3H-acetate release, indicative of histone deacetylase activity. The HeLa nuclear extract gave an average of 3.5-fold increase in radioactive count per minute (cpm). The addition of sodium butyrate, a specific inhibitor of histone deacetylases, prevented the reactions containing HeLa nuclear extracts and rAtHD1. Recombinant HDACs (AtHD1, AtHDA6 and AtHDA7) produced in insect cells did not show measurable histone deacetylase activity (data not shown).

Discussion

AtHD1 and other RPD3-like HDACs are differentially expressed in Arabidopsis

Consistent with the data on AtHD1 expression analyzed by RNA gel blot experiments 32 (Plant Chromatin Database, http://chromdb.org) and RT-PCR 27, AtHD1 is expressed in all tissues examined (data not shown). The constitutive expression pattern of AtHD1 suggests that it may have a role in global regulation of plant gene expression 26. AtHD1 has been shown to regulate gene expression in house-keeping functions such as pathogen response 27 and in floral development such as the expression of SUPERMAN 25 in Arabidopsis. In contrast, the expression of HDA7, which had not been detected in previous studies, was found to express at very low levels in the flower tissues. This flower-specific expression pattern suggests that HDA7 may play a role in flowering. Interestingly, recent studies have shown that histone deacetylation and histone H3 dimethylation at lysines 9 and 27 are involved in FLC (Flowering Locus C) repression, whereas histone acetylation and histone H3 trimethylation at lysine 4 are associated with active FLC expression 33, 34, 35. Therefore, it would be interesting to determine the effects of HDA7 on the expression of flower specific genes.

AtHD1 is localized in the nucleus and excluded from the nucleolus

AtHD1 is supposed to be a nuclear protein that modifies its substrate histones in the chromatin. However, some HDACs can be shuttled between the nucleus and cytoplasm as a way to regulate gene expression. For instance, the mammalian class II HDACs, HDAC4, 5, 7 and 9, are associated with 14-3-3 proteins upon phosphorylation of conserved serine residues on their N-terminal regions, resulting in sequestration of the HDACs to the cytoplasm 36, 37, 38, 39, 40. In contrast, the class I HDACs, including HDAC1, 2 and 8, are predominantly localized in the nucleus and seem not to be regulated by subcellular localization 41. Exceptionally, class I HDAC3 is present in both the cytoplasm and the nucleus. It is translocated from the nucleus to the cytoplasm in response to interleukin-1b signaling, resulting in derepression of a specific subset of NF-kB regulated genes 42. In addition, the study of the expression profile and cellular localization of three RPD3-type HDACs in maize, termed ZmRpd3/101, ZmRpd3/102 and ZmRpd3/108, showed that they are present in both cytoplasm and nuclei during different stages of kernel, shoot and anther development 43. This suggests that the three maize class I HDACs may be regulated by nuclear cytoplasmic shuttling and involved extensively in regulation of various physiological and developmental processes 43.

We investigated subcellular localization of AtHD1 extensively by examining the expression of AtHD1-GFP fusion protein in various intact living tissues of Arabidopsis with high resolution. The results of transient and stable expression of AtHD1-GFP fusion protein indicated the localization of AtHD1 in the cell nuclei of onion epidermal cells and all Arabidopsis tissues examined, including root, leaf, flower and seed. Further examination of the AtHD1-GFP localization patterns indicates that the protein is likely excluded from the nucleolus, which suggests a role of AtHD1 in gene regulation that is different from ZmHD2, AtHD2 and AtHDA6 8, 20, 23. The GFP protein used in this experiment is an engineered one called S65T GFP that contains a mutation in the chromophore with the serine 65 being replaced by threonine, resulting in enhanced brightness, faster chromophore formation, and slower photobelaching 44. Control experiments with the expression of non-fusion GFP protein were not included in this study because the S65T and other types of GFP proteins without a targeting sequence had been shown to accumulate and distribute evenly in Arabidopsis and onion cells due to their small sizes 31, 45, 46. In fact, the size exclusion limit of the nuclear pores has been estimated to be 40 to 60 kDa 47, 48. Therefore, GFP protein having a molecular weight of 27 kDa can diffuse across the nuclear envelope, whereas AtHD1-GFP fusion protein that is 83 kDa in size should remain in the cytoplasm if AtHD1 lacks a nuclear localization signal.

The role of histone deacetylation in mitosis

Histone deacetylation was shown to correlate with chromatin condensation and subsequent chromatid segregation during mitosis 49, 50, 51. It has been demonstrated that the histone acetylation levels in several mammalian cell lines started to decrease as the cells enter prophase, then dramatically reduced as the cells progressed into metaphase through telophase, and restored to normal levels at the late telophase/early interphase boundary 49. Moreover, a temporal inhibition of HDACs by trichostatin A (TSA) in human fibroblasts entering mitosis resulted in defective chromatin condensation and impaired chromatid segregation 50. A similar study in plant also showed that TSA treatment on cultured tobacco cells led to abnormal anaphases and impaired the progression through mitosis 51. When we examined the subcellular localization of AtHD1-GFP fusion protein, we captured an image of a leaf mesophyll cell entering mitosis. The cell was at early prophase as evident by the presence of condensing chromatin and the nucleolus was still intact. The GFP signals revealed that AtHD1 aggregated as dots and speckles, suggesting that AtHD1 may be associated with the condensing chromatin during mitosis. We also observed that the GFP signals and DAPI staining did not overlap perfectly, suggesting that AtHD1 might be targeted to specific chromosomal regions.

Recombinant AtHD1 demonstrated histone deacetylase activity in vitro

The histone deacetylase activity of AtHD1 is implied from studies in which AtHD1 was either knocked out or overexpressed. Down-regulation of AtHD1 led to the accumulation of hyperacetylated histones in transgenic Arabidopsis 25, and overexpression of AtHD1 decreased levels of tetraacetylated histone 27. Here, we demonstrated that the recombinant AtHD1 produced in bacteria showed a moderate yet specific histone deacetylase activity in vitro, providing experimental evidence for enzymatic activity of a plant histone deacetylase. AtHD1 showed enzyme activity in three replicated experiments (using different protein extractions), whereas the recombinant AtHDA6, 9, and 17 did not show enzymatic activity (data not shown), suggesting that the low activity of AtHD1 is not due to contamination of bacterial proteins or artifacts.

The sub-optimal activity of the recombinant AtHD1 may suggest that it requires post-translational modifications or specific chaperons for proper folding, both of which are absent in the bacterial expression system. Phosphorylation is a major post-translation modification in HDACs. In the maize embryos, phosphorylation of HD1A causes a change in substrate specificity of the enzyme 52. Other HDACs, such as the maize HD2 8, human HDAC1, 2, 4 and 5 53, 54, 55 are subjected to phosphorylation, which can modulate their activities. Another post-translational modification that has been shown to regulate HDAC activity and function is called sumoylation, which involves the conjugation of small ubiquitin-related modifier 41, 56, 57, 58. However, sumoylation has been identified only in mammalian HDACs. Some HDACs are regulated by proteolytic processing. In maize, HDA1 is synthesized as an inactive precursor protein that is converted to the enzymatically active form by proteolytic removal of the C-terminal part of the protein 59. In an attempt to solve the problem caused by the lack of post-translational modifications, AtHD1 together with other RPD3-type HDACs in Arabidopsis including HDA6, 7 and 9, were produced by baculoviral expression system and then tested by HDAC assay. However, all recombinant HDACs tested did not show any HDAC activity (data not shown).

A second possible reason for the low activity of rAtHD1 is lack of co-factors or requirement of complex formation with other proteins. Notably, many purified recombinant HDACs are enzymatically inactive 41, with only the exception of yeast HOS3 and mammalian HDAC8 60, 61, 62. It will be interesting to further investigate AtHD1 activity by isolating interacting proteins.