Nuclear Localization of Peptidylarginine Deiminase V and Histone Deimination in Granulocytes*

Peptidylarginine deiminase (PAD) deiminates arginine residues in proteins to citrulline residues Ca2+ dependently. There are four types of PADs, I, II, III, and V, in humans. We studied the subcellular distribution of PAD V in HL-60 granulocytes and peripheral blood granulocytes. Expression of green fluorescent protein-tagged PADs in HeLa cells revealed that PAD V is localized in the nucleus, whereas PAD I, II, and III are localized in the cytoplasm. PAD V deletion mutants indicated that the sequence residues 45–74 have a nuclear localization signal (NLS). A sequence feature of this NLS is a three-lysine residue cluster preceded by a proline residue and is not found in the three other PADs. Substitution of the lysine cluster by an alanine cluster abrogated the nuclear import activity. These results suggested that the NLS is a classical monopartite NLS. HL-60 granulocytes, neutrophils, and eosinophils stained with antibody specific for PAD V exhibited distinct positive signals in the nucleus. Subcellular fractionation of HL-60 granulocytes also showed the nuclear localization of the enzyme. When neutrophils were stimulated with calcium ionophore A23187, protein deimination occurred in the nucleus. The major deiminated proteins were identified as histones H2A, H3, and H4. The implication of PAD V in histone modifications is discussed.

Peptidylarginine deiminase (PAD) deiminates arginine residues in proteins to citrulline residues Ca 2؉ dependently. There are four types of PADs, I, II, III, and V, in humans. We studied the subcellular distribution of PAD V in HL-60 granulocytes and peripheral blood granulocytes. Expression of green fluorescent proteintagged PADs in HeLa cells revealed that PAD V is localized in the nucleus, whereas PAD I, II, and III are localized in the cytoplasm. PAD V deletion mutants indicated that the sequence residues 45-74 have a nuclear localization signal (NLS). A sequence feature of this NLS is a three-lysine residue cluster preceded by a proline residue and is not found in the three other PADs. Substitution of the lysine cluster by an alanine cluster abrogated the nuclear import activity. These results suggested that the NLS is a classical monopartite NLS. HL-60 granulocytes, neutrophils, and eosinophils stained with antibody specific for PAD V exhibited distinct positive signals in the nucleus. Subcellular fractionation of HL-60 granulocytes also showed the nuclear localization of the enzyme. When neutrophils were stimulated with calcium ionophore A23187, protein deimination occurred in the nucleus. The major deiminated proteins were identified as histones H2A, H3, and H4. The implication of PAD V in histone modifications is discussed.
A family of peptidylarginine deiminases (PAD) 1 (EC 3.5.3.15) catalyzes the conversion of arginine residues in proteins into citrulline residues in the presence of calcium ion. Four types of rodent PADs, I, II, III, and IV, and of human PADs, I, II, III, and V, are known (1)(2)(3)(4)(5)(6)(7)(8). Rodent PAD IV is most closely related to human PAD V. The structures of these PADs are relatively conserved in the C-terminal two-thirds of the sequence and diverge more distantly in the N-terminal one-third of the se-quence. Rodent enzymes have different substrate specificities toward synthetic substrates and different tissue distributions (9,10). Biochemical and immunocytochemical studies have suggested the involvement of PAD I in the terminal differentiation of epidermis (11)(12)(13), that of PAD II in myelination and demyelination of central nerve axons (14, 15), and that of PAD III in the keratinization of hair follicles (8,16,17). Information on the biological functions of mouse PAD IV and human PAD V is rather limited. In addition, increased amounts of deiminated myelin basic proteins are in the afflicted area of brains of patients with multiple sclerosis, and prevalent autoantibodies recognizing citrulline residues as an epitope of autoantigen in patients with rheumatoid arthritis have suggested the pathological involvement of PAD in diseases (15, 18 -21).
PAD V was first found in HL-60 cells when cells were induced to differentiate into granulocytes (HL-60 granulocytes) by all-trans-retinoic acid (RA) and differentiate into monocytes by 1␣,25-dihydroxyvitamin D 3 (7). PAD V was also found in peripheral blood granulocytes (22). PAD V in HL-60 cells can be activated to deiminate nuclear proteins of nucleophosmin/B23 and histones by stimulation with calcium ionophore (23). This has suggested the location of PAD V in the nucleus. The locations of PADs in cells are important for understanding a role of PADs in cellular functions, but have not yet been studied comprehensively (15,22,24). Here we show the nuclear localization of PAD V in granulocytes compared with three other PADs and discuss a role of PAD V in histone modification.

EXPERIMENTAL PROCEDURES
Cell Culture and Preparation of HL-60 and Blood Granulocytes-HL-60 and HeLa cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen) (25). HL-60 granulocytes were produced by culturing HL-60 cells (3 ϫ 10 5 cells/ml) with 1 M all-trans-retinoic acid for 3 days (7). Human granulocytes were prepared from heparinized blood of healthy donors with Polymorphprep TM (AXIS-SHIELD) according to the manufacturer's instructions. Neutrophils and eosinophils were separated in a Percoll density reagent (Amersham Biosciences) (22).
Construction of Expression Plasmids-A whole PAD V encoding region in pGEX-PAD V was subcloned into a pEGFP vector (Clontech) (7). A human PAD I cDNA (KAT12008) was purchased from Takara Bio. A human PAD II cDNA (KIAA 0994) was from Kazusa DNA Research Institute (26). A whole coding region of PAD II was amplified with PCR using a pair of primers attached with an EcoRI linker. PAD I was amplified with PCR using a pair of primers attached with a SalI linker. PAD III was amplified with PCR using pKKhPAD3 as a template and a pair of 5Ј and 3Ј primers attached with an EcoRI linker and a SalI linker, respectively. The amplified cDNAs were subcloned into a pEGFP vector. PAD V deletion mutants 1-262 and 262-663 were prepared by digestion of a PAD V cDNA with SmaI, and 1-394 was prepared by digestion with XhoI. Deletion mutants 58 -663, 182-663, 1-104, 35-104, and 45-74 were prepared with PCR using a pGEX-PAD V as a template and a pair of 5Ј and 3Ј primers attached with an EcoRI linker. Alanine substitution mutant 1-104 (K59A/K60A/K61A) for three lysine residues (59 -61) in a wild type was constructed by overlap extension (27). Sequences in constructs prepared with PCR were verified by sequencing. * This work was supported in part by grants-in-aids for Cancer Research from the Ministry of Health, Welfare and Labour, for Promotion of Research at Yokohama City University, and the Japan Foundation for Applied Enzymology Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Transfection-Growing HeLa cells were collected by trypsinization, washed with RPMI 1640, and resuspended in RPMI 1640 medium without fetal bovine serum and antibiotics. The cells (4 ϫ 10 6 cells) were mixed with 30 g of the CsCl-purified plasmid in 0.5 ml of medium and subjected to electroporation at 200 V, 1180 F, and low ohm with a Cell-Porator (Invitrogen). The cells were incubated on ice for 10 min and then cultured in the complete medium for 24 h.
Preparation of Recombinant PADs-Escherichia coli BL-21 carrying PAD I, II, and V cDNAs in a pGEX-6P vector were cultured with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside at 25°C for 16 h (7). From their extracts GST-fused PADs were affinity purified by glutathione-Sepharose 4B chromatography and digested with Prescission protease (Amersham Biosciences), and excised PADs were recovered in a flowthrough fraction through a glutathione-Sepharose 4B column. Recombinant PAD III was a gift from Hidenari Takahara, Ibaraki University (8). The purified recombinant PADs I, II, III, and V showed specific activities of 165, 37, 120, and 224, respectively, when determined with N ␣ -benzoyl-L-arginine ethyl ester under the conditions described previously (7).
Purification of Anti-PAD V(N)-The previously described rabbit antiserum against GST-PAD V was affinity chromatographed on an Nterminal PAD V fragment-(1-262) bound column (7). The serum (2 ml) was fractionated with 17% sodium sulfate, and precipitated IgG was dialyzed, passed through a DEAE-Sephacel column equilibrated with 17.5 mM phosphate buffer, pH 6.3, and then passed through GST-bound Sepharose. The unabsorbed fraction was bound to GST-PAD V-(1-262)bound Sepharose (a 0.6-ml column containing 2.5 mg of protein) and the column was washed with 20 mM Tris-HCl, pH 7.4, 0.5 M NaCl. The bound IgG was eluted with 0.1 M glycine-HCl, pH 3.0, with a yield of about 200 g.
Immunoblotting-Sample proteins were subjected to SDS-10% PAGE by the method of Laemmli (28) and the resolved proteins were electrotransferred to a nitrocellulose membrane. Blots were probed with anti-PAD V(N) (0.15 g/ml) or rabbit anti-GFP antiserum (Clontech) (1:5,000), and then bound IgG was detected with a horseradish peroxidase conjugate of goat anti-rabbit IgG (1:5,000) (Bio-Rad) using a chemiluminescence reagent kit, Renaissance (PerkinElmer Life Sciences). Blots of deiminated proteins were treated with the medium for chemically modifying citrulline residues at 37°C for 3 h and then modified citrulline residues were detected with a rabbit monospecific antibody to modified citrulline (anti-MC, 0.125 g/ml) (29). The citrulline modifying medium consisted of 1 part of reagent solution A containing 0.0416% FeCl 3 -6H 2 O, 4.6 M H 2 SO 4 , and 3.0 M H 3 PO 4 and 1 part of reagent solution B containing 0.5% diacetylmonoxime and 0.25% antipyrine. The chemiluminescent signal was captured and recorded in a Fluor-S MAX MultiImager and the signal intensities were determined with Quantity One software (Bio-Rad).
Double Immunofluorescence Cytostaining-HL-60 cells and blood granulocytes in suspension were washed with Hank's balanced salt solution free of calcium chloride and magnesium chloride and were fixed with ice-cold 4% paraformaldehyde in PBS(Ϫ) for 30 min. The fixed cells were washed, suspended in ice-cold PBS(Ϫ), and spread on glass slides by cytospinning at 1,600 rpm for 6 min in a Cytofuge TM 2 (StatSpin Inc). The cells were post-fixed with 4% paraformaldehyde for 15 min at room temperature. Cytospin preparations of HL-60 cells and blood eosinophils were heated to boiling in 10 mM citrate buffer, pH 7.0, under a microwave at 550 W for 10 min and cooled to room temperature (30). This antigen retrieval procedure was omitted for neutrophils. The fixed cells were incubated with 2 M Tris-HCl, pH 7.4, for 15 min, permeabilized with 0.1% Triton X-100 in PBS(Ϫ) for 10 min, and then blocked with 2% normal goat serum containing 2% bovine serum albumin in PBS(Ϫ). HL-60 cells were double-stained with a mixture of rabbit anti-PAD V(N) (1.5 g/ml) and mouse anti-MPO monoclonal antibody 3-2H3 (0.5 g/ml) (31). The bound IgGs were detected with a biotin-labeled goat anti-rabbit IgG (Wako) (1:200), a streptavidin-Cy3 and a fluorescein isothiocyanate-labeled goat anti-mouse IgG (Jackson ImmunoResearch Inc.) (1:200). Neutrophils were double-stained with anti-PAD V(N) and anti-MPO, and eosinophils with anti-PAD V(N) and mouse anti-MBP monoclonal antibody BMK-13 (Monosan) (1:40 dilution). The bound IgGs were detected using a fluorescein isothiocyanatelabeled goat anti-rabbit IgG (Wako) (1:200) and a tetramethylrhodamine isothiocyanate (TRITC)-labeled goat anti-mouse IgG (Sigma) (1:200). For preabsorption, 0.15 g of either anti-PAD V(N) or nonimmune IgG and 3 g of recombinant PAD V protein were mixed, incubated for 30 min at room temperature, and centrifuged at 15,000 ϫ g for 20 min. The supernatant was used for the immunoreaction. Cells were stained for nuclear DNA with DAPI and mounted in a solution containing 0.1 M Tris-HCl, pH 9.0, 50% glycerol, 0.1% NaN 3 , and 2.5% 1,4-diazabicyclo[2,2,2]octane. Subcellular Fractionation-HL-60 granulocytes were suspended in buffer A (20 mM Tris-HCl, pH 7.6, 5 mM MgCl 2 , 1.5 mM KCl, 1 mM phenylmethanesulfonyl fluoride, 2 mM dithiothreitol, and 0.1% Nonidet P-40) and disrupted with a Dounce homogenizer (25). The homogenate was centrifuged at 760 ϫ g at 4°C for 10 min to separate it into supernatant (cytoplasmic) and pellet (nuclear) fractions. The supernatant and pellet fractions were subjected to immunoblotting.
Immunocytochemistry of Deiminated Proteins-Blood granulocytes were suspended at 2 ϫ 10 6 cells/ml in Locke's solution, stimulated with 1 M calcium ionophore A23187 for 5 min at 37°C, and fixed with ice-cold 4% paraformaldehyde for 30 min. The cytospin preparations were incubated with medium for chemically modifying citrulline residues for 3 h at 37°C and then were incubated successively with anti-MC (0.625 g/ml), goat anti-rabbit IgG-biotin conjugate, and streptavidin-peroxidase conjugate (11,23). The medium consisted of 1 part of solution A (0.0416% FeCl 3 -6H 2 O, 4.6 M H 2 SO 4 , and 3.0 M H 3 PO 4 ) and 1 part of solution B (1% diacetylmonoxime, 0.5% antipyrine, and 0.5 M acetic acid). Signals for bound peroxidase were developed with 3,3Јdiaminobenzidine as a substrate and the cells were stained with Giemsa.
Histone Extraction-Histones were extracted from a nuclear fraction with 0.4 N H 2 SO 4 and precipitated with acetone, and the precipitates were subjected to SDS-15% PAGE and immunoblotting as described previously (23).

Localization of GFP-tagged PAD V in HeLa Cells and Its
Comparison with Those of PADs I, II, and III-To investigate the subcellular localization of PAD by fluorescence microscopy, we prepared plasmid constructs containing PADs I, II, III, and V cDNA fused with a 3Ј end of the EGFP gene and used them to transfect HeLa cells for their expression. After 24 h in culture, their expressions were confirmed by immunoblotting using anti-GFP serum (Fig. 1A). GFP-tagged PADs with about 98 kDa were detected in GFP-PAD gene-transfected cells, but not in control cells. The enzymatic activities of these GFP-PAD fusion proteins were confirmed (Fig. 1B). On incubation of lysates of GFP-PAD gene-transfected cells with Ca 2ϩ , but without Ca 2ϩ , a number of deiminated proteins with a wide range of molecular weights were detected with anti-MC (lanes 3-10). The control cell lysate formed no deiminated protein with or without CaCl 2 (lanes 1 and 2). Panel C of Fig. 1 shows fluorescence micrographs of these cells in parallel cultures. The control EGFP was located throughout the cell (panels a and f). GFP-tagged PAD V was located exclusively in the nucleus stained with DAPI (panels e and j), whereas GFP-tagged PADs I, II, and III were located only in the cytoplasm (panels b-d and g-i). These results suggested a unique nuclear location of PAD V in the cells.
NLS of PAD V-To identify an NLS of PAD V, we prepared N-and C-terminal deletion mutants that were fused with the EGFP gene ( Fig. 2A). These constructs were expressed in HeLa cells for 24 h. Expressions of all GFP-tagged deletion mutant proteins were confirmed by immunoblotting using anti-GFP serum on the basis of their expected sizes (data not shown). No isolated deletion mutants had PAD activity (data not shown). As shown in Fig. 2B, the fluorescence micrographs show localizations of these deletion mutants in these cells. The control GFP was distributed throughout the cells (panel a). The Nterminal deletion mutants 58 -663, 182-663, and 261-663 were distributed throughout the cells, although more densely in the cytoplasm than in the nucleus (panels b-d). However, the C-terminal deletion mutants 1-394, 1-262, and 1-104 and deletion mutant 35-104 were located in the nucleus (panels e-h). Deletion mutant 45-74 as short as 30 amino acids was also located in the nucleus (panel i), indicating the presence of an NLS. This fragment has three consecutive lysine residues, 59 -61, preceded by proline residue 57 (Fig. 2C). These are features of a basic type NLS motif (Fig. 2D). But that was not found in the three other PADs (Fig. 2C).
To determine the contribution of the three lysine residues to the NLS activity, we prepared a wild type 1-104 and its mutant 1-104 (K59A/K60A/K61A) which has three alanine (59 -61) instead of the three lysine residues and fused them with the GFP-GST gene (Fig. 3A). These constructs were transfected to HeLa cells for expression. Expressions of GFP-GST-tagged wild type and mutant proteins in 24-h culture cells were confirmed by immunoblotting using anti-GFP antiserum (data not shown). Panel B of Fig. 3 shows the fluorescence micrographs of these cells. The wild type GFP-GST-PAD-(1-104) was located in the nucleus, whereas the mutant GFP-GST-PAD-(1-104) was located in the cytoplasm, and the control GFP-GST was found throughout the cell. These results suggested that the KKK motif in this position might be essential for the nuclear import of PAD V.
Localizations of PAD V in HL-60 Granulocytes and Blood Granulocytes-The above results indicate the ability of recombinant PAD V to become located in the nucleus. We next de-termined the locations of native PAD V in HL-60 and blood granulocytes by immunocytochemistry. We affinity purified an antibody specific for PAD V from antiserum against PAD V with its N-terminal segment 1-261-conjugated Sepharose beads. It was termed anti-PAD V(N). The antibody was first tested with recombinant human PADs I, II, III, and V by immunoblotting (Fig. 4A). It reacted with PAD V, but not with similar amounts of PADs I, II, and III. This antibody was also tested on lysates of HL-60 cells (Fig. 4B). HL-60 granulocytes expressing PAD V were prepared by culturing HL-60 cells with RA. These cells gave one distinct band of 67 kDa PAD on a blot (lane 2), whereas control HL-60 cells gave no signal (lane 1).
For immunocytochemical staining, cytospin preparations of paraformaldehyde-fixed HL-60 cells were first immersed in a boiling citrate buffer and then double-stained with a mixture of anti-PAD V(N) and anti-MPO. MPO was used as a marker specific for a neutrophil cytoplasmic granule. This antigen retrieval procedure was essential for visualization of PAD V staining. On staining of HL-60 granulocytes, the distinct signals for PAD V were detected in the nucleus, but not in the cytoplasm (Fig. 4C, panel e). In control HL-60 cells no signal was detected in the nucleus, but a faint signal was detected in the cytoplasm (panel b). Signals for MPO were consistently found in the cytoplasm of control HL-60 cells and HL-60 granulocytes (panels a and d). These results indicated that HL-60 PAD V is localized in the nucleus.
We next examined PAD V in peripheral blood granulocytes. Granulocytes gave only a single band with 67 kDa on blots and its signal intensity increased with increase in the cell number from 1 ϫ 10 4 to 4 ϫ 10 4 (Fig. 5A, lanes 5-7). The cellular content of PAD V was estimated to be ϳ1.9 ϫ 10 6 molecules per cell using recombinant PAD V as a standard (lanes 1-4). Neutrophils were double-stained with a mixture of anti-PAD V(N) and anti-MPO (Fig. 5B) without boiling in citrate buffer. Distinctive signals of anti-PAD V were confined to narrow diffuse DAPI-staining regions located along segmented forms of the nucleus (panels b and c and e and f). This is more clearly shown at higher magnification (panel d-f). The PAD-positive cells were identified as neutrophils by their nuclear morphology and by the cytoplasmic location of the MPO signal (panels a, d, g, and j). When nonimmune IgG was used as a control, no signals were detected in the nucleus, whereas the cytoplasm was stained faintly (panel h). Preabsorption of the antibody with recombinant PAD V completely abolished the signal in the nucleus, although the faint cytoplasmic signals remained (panel k). Preabsorption of nonimmune IgG also showed only the faint cytoplasmic signals (data not shown). These results indicated that neutrophil PAD V is localized in the nucleus.
An eosinophil fraction prepared in a Percoll density gradient was Ͼ50% in purity. Eosinophil cytospin preparations were treated by microwave to boiling and double stained with a mixture of anti-PAD V(N) and anti-eosinophil MBP, as a marker specific for a cytoplasmic granule. Panel C of Fig. 3 shows the fluorescence micrographs. The major signal of PAD was found overlapping with the nucleus stained with DAPI, whereas a weaker signal was found in the cytoplasm overlapped with the MBP signal (panel a-c). When cells were stained with nonimmune IgG, no nuclear signal was detected, but the cytoplasmic signal was observed again (d-f). These cytoplasmic signals could be accounted for by nonspecific IgG binding to the cytopolasm. These results indicate that eosinophil PAD is localized in the nucleus.
Subcellular Fractionation and Solubilization of Nuclear PAD V-To substantiate the PAD V nuclear location shown above, we examined PAD V of HL-60 granulocytes by fractionation of their lysates into nuclear and cytoplasmic fractions. Immunoblotting of these fractions indicated that PAD V was present only in the nuclear fraction, and ␣-tubulin was mostly in the cytoplasmic fraction (Fig. 6A, upper panel). Protein staining of a gel showed that all core histones were recovered in the nuclear fraction (lower panel). The virtual absence of ␣-tubulin and histones in the nuclear and cytoplasmic fractions, respectively, showed negligible cross-contamination between these fractions. The amounts of lactate dehydrogenase activity, PAD activity, and DNA determined were 99% of the total in the cytoplasmic fraction, and 97 and 100% in the nuclear fraction, respectively. These results also indicated the nuclear localization of PAD V. To examine the solubilization of PAD V from the nucleus, we extracted the nuclear fraction with increasing concentrations of NaCl with or without 0.5% CHAPS (Fig. 6B). PAD was almost fully solubilized at 0.2 M NaCl without CHAPS and addition of 0.5% CHAPS solubilized PAD at 0.1 M NaCl (upper panel). Core histones remained insoluble under all the conditions employed (lower panel).
Occurrence of Protein Deimination in the Nucleus in Neutrophils-To investigate where protein deimination occurs in blood granulocytes, we suspended cells in medium containing 2 mM CaCl 2 and gave a 5-min stimulus with 0.5 M calcium ionophore A23187. These cells were stained for deiminated proteins with anti-MC. Panel A of Fig. 7 shows the micrographs of these cells. The positive signals were confined to segmented forms of the nucleus in neutrophils with the stimulus (panel b), but not without the stimulus (panel a). The number of positive cells amounted to about 50% of the total cells within 5 min and reached almost all cells by 15 min. Next, we identified deiminated proteins in cells by immunoblotting using anti-MC (Fig.  7B). The unstimulated cells gave no bands on blots (lanes 1 and  5), whereas the stimulated cells gave three major distinct bands with ϳ14, 17, and 18 kDa and several other weak bands with higher molecular weight (lanes 2-4 and 5-8). The signal intensities of these bands (14 -18 kDa) increased with increase in the A23187 concentration from 0.25 to 1 M (lanes 2-4) and reached a plateau at 4 M (data not shown), and reached a maximum at 5 min during a 15-min period with 1 M A23187 (lanes 5-8). These deiminated proteins were all recovered in the nuclear fraction. These 14-, 17-, and 18-kDa proteins were identified as histones H4, H2A, and H3, respectively, by SDS-PAGE of nuclear proteins extracted with H 2 SO 4 and by twodimensional gel electrophoresis specialized for histones (data not shown), as found in HL-60 granulocytes (23). These results indicated that protein deimination occurs in the nucleus.

DISCUSSION
This work presents various lines of evidence for the nuclear localization of PAD V in HL-60 granulocytes and peripheral blood granulocytes. First, immunocytochemical staining of HL-60 granulocytes, neutrophils, and eosinophils using anti-PAD V(N) revealed distinct signals of PAD V in the nucleus, but no significant signals in the cytoplasm. These nuclear signals in neutrophils were completely abolished by incubation of the antibody with an excess amount of recombinant PAD V. These results indicate the nuclear localization of most of PAD V in these cells. Previously, in immunostaining of neutrophils and eosinophils using an anti-PAD V, we observed the positive signals in the cytoplasm, but no signals in the nucleus and these observations led to the misstatement "PAD V localizes in the cytoplasmic granules" (22). By investigation of the discrepancy in the previous and present work, we found that in neutrophils fixed under the previous conditions, the nuclear signals were decreased actually to an undetectable level. The previous conditions include steps of the Ϫ20°C storage of fixed cytospin preparations and the second fixation before use. This may lead to overfixation and result in masking of nuclear antigen. In fact, about 3 h prolonged fixation of neutrophils completely blocked the nuclear staining, but boiling of this preparation resumed the distinct positive nuclear staining (data not shown). In immunostaining of eosinophils, omission of boiling of the preparation gave the intense cytoplasmic signals to the nuclear signals, like in the previous work (data not shown). This made the latter signal invisible. Boiling of this preparation improved the visualization of the nuclear positive signal as shown in this work. High nonspecific binding of IgG to the cytoplasm in eosinophils is also observed by others (32). Nuclear antigen masking in neutrophils and eosinophils and cytoplasmic high nonspecific binding of IgG in eosinophils probably made it difficult to distinguish a difference in signals between nonimmune IgG and immune IgG in the previous work.
Another new aspect of immunocytostaining presented here is that the nuclear PAD signal in neutrophils was confined to a euchromatin region that is distinguishable from a heterochromatin region by diffuse DAPI staining (Fig. 5B, panels e and f). Whether such a distinctive intranuclear distribution of PAD V is associated with the degree of granulocyte maturation and with nuclear function remains unclear.
Second, fluorescence microscopy of HeLa cells expressing EGFP-labeled PADs more clearly revealed the nuclear location of recombinant PAD V without the antigen blocking and nonspecific IgG binding problem discussed above. We identified an NLS of PAD V as a 30-amino acid segment (residues 45-74).
This segment has an NLS motif ( 56 PPAKKKST 63 ) that is similar to monopartite basic type NLSs of SV40 T-antigen, c-Myc, and RanBP3 (33)(34)(35). This sequence feature is also found in mouse and rat PAD IVs, but not in human and rodent PADs I, II, and III (Fig. 2, C and D). The locations of these PAD IV in cells remains to be determined. Nuclear import of this type of NLS is mediated by an adaptor importin ␣ and a receptor importin ␤. There are several types of importin ␣ specifically interacting with NLSs (36 -38). What types of importin ␣ and importin ␤ mediate the nuclear import of PAD V will be studied.
Third, immunocytochemical staining of deiminated proteins revealed protein deimination occurring in the nucleus when blood granulocytes were stimulated with the calcium ionophore A23187. We identified these deiminated proteins as histones H3, H2A, and H4. The occurrence of protein deimination in the nucleus is also found in HL-60 granulocytes (23). These results indicate that PAD V resides with substrate proteins in the same compartment. The proteins known to be deiminated in vivo are all the cytoplasmic or non-nuclear proteins and seem to be deiminated by cytoplasmic PADs: deiminations of keratin and fillagrin by PAD I in epidermis (11,12); deimination of trichohyalin by PAD III in hair follicles (8,16,17); and deimination of myelin basic proteins by PAD II in brain oligodendrocytes (14, 15,39). These findings are consistent with a notion that PAD V is the only enzyme in the nucleus.
The nuclear localization of PAD V suggests that PAD V may be a new factor modulating a variety of the nuclear function depending on chromatin structure. Chromatin structure and function are most often modulated by post-translational modifications of acetylation, phosphorylation, methylation, and ubiquitination. Most of these modifications interdependently  5-8). After the stimulation, the cells were precipitated with ice-cold 10% trichloroacetic acid, washed, and dried with acetone, and the residues were dissolved for SDS-15% PAGE. The resolved proteins were transferred to a polyvinylidene difluoride membrane and detected with anti-MC. Lanes 1-4, cells treated for 5 min with 0, 0.25, 0.5, and 1.0 M A23187, respectively; lanes 5-8, cells treated with 1 M A23187 for 0, 5, 10, and 15 min, respectively. occur in histone tails protruding from nucleosome core particles and cause changes in their interaction with neighboring nucleosomes and nonhistone proteins and in higher-order chromatin structure (40,41). Arginine methylation in N-terminal histone tails appears to facilitate transcriptional activation (42)(43)(44). It is possible that arginine deimination also modulates chromatin structure by decreasing positive charges in the histone tails. Determination of the arginine deimination sites on histones is in progress. The consequences of deiminations of histones and other nuclear proteins in vivo await future investigation.
Nothing is known about a role of the PAD in granulocyte differentiation and granulocyte function. Neutrophils are activated by priming of the cells with cytokines and chemokines released during inflammation (45). Besides, neutrophils are apoptosis-prone cells having a banded or segmented nucleus occupied with large extents of highly condensed chromatin (46 -48). Future studies on whether or not PAD V is associated with a priming process and apoptosis and on what kind of calcium signaling is essential to activate nuclear PAD V are needed.