Molecular Characterization of Peptidylarginine Deiminase in HL-60 Cells Induced by Retinoic Acid and 1α,25-Dihydroxyvitamin D3 *

Three types of peptidylarginine deiminase (PAD), which converts a protein arginine residue to a citrulline residue, are widely distributed in animal tissues. Little is known about PAD of hemopoietic cells. We found that PAD activity in human myeloid leukemia HL-60 cells was induced with the granulocyte-inducing agents retinoic acid and dimethyl sulfoxide and with the monocyte-inducing agent 1α,25-dihydroxyvitamin D3. We cloned and characterized a PAD cDNA from retinoic acid-induced cells. The cDNA was 2,238 base pairs long and encoded a 663-amino acid polypeptide. The HL-60 PAD had 50–55% amino acid sequence identities with the three known enzymes and 73% identity with the recently cloned keratinocyte PAD. The recombinant enzyme differs in kinetic properties from the known enzymes. Immunoblotting and Northern blotting with an antiserum against the enzyme and the cDNA, respectively, showed that a protein of approximately 67 kDa increased concomitantly with increase of mRNA of approximately 2.6 kilobases during granulocyte differentiation. During monocyte differentiation the same mRNA and protein increased as in granulocyte differentiation. Neither the enzyme activity nor the protein was found in macrophage-induced cells. These results suggested that expression of the PAD gene is tightly linked to myeloid differentiation.

Peptidylarginine deiminases (PADs) 1 (protein-arginine deiminase, protein L-arginine iminohydrolase, EC 3.5.3.15) are a family of post-translational modification enzymes which convert arginine residues to citrulline residues in the presence of calcium ion. Enzymatic deimination in vitro changes the functional properties of various proteins and alters their secondary and tertiary structures (1)(2)(3)(4). Deimination of keratins, filaggrin, and trichohyalin is involved in the process of keratinization of skin and hair (4 -9). Deiminated keratins and filaggrin are found in the cornified layer of the epidermis and deiminated trichohyalin is localized in the medulla of hair and the inner root sheath of hair follicles and these modifications are tightly linked to cell-specific stages of epidermis differentiation and hair follicle development (5)(6)(7)(8)(9). Extensively deiminated forms of myelin basic protein are also found in normal infant brain and in demyelinated areas of brain with multiple sclerosis, and this deimination is thought to be associated with immature myelination (10,11). We reported a correlation between deimination of vimentin in mouse peritoneal macrophages and ionomycin-induced apoptosis (12). Deimination of a 70-kDa nuclear protein in cultured keratinocytes associated with apoptosis was also reported recently (13). All these findings suggest involvements of PAD in biological as well as pathological processes. There are at least three types of PAD in various rodent tissues which seem to be cell type specific (3, 14 -16). Their substrate specificities for BAEE and Bz-L-Arg and their antigenic properties are different. PAD type II purified from rat muscle has been well characterized. It is also present in the brain, spinal cord, and some secretory tissues. PAD types I and III are mainly present in the epidermis and uterus and in hair follicles, respectively. PAD cDNAs for types I, II, and III have been isolated from rat, mouse, and sheep, but not from humans (9,(17)(18)(19). Their amino acid sequences constituting 662 to 673 amino acid residues have been deduced. Recently, a novel PAD cDNA named type IV was isolated from a keratinocyte cell line from a newborn rat and rat epidermis, but the distribution of the enzyme in cells and tissues is not yet known (20,21).
PAD activities in rat granulocytes and mouse peritoneal macrophages have been reported, but nothing is known about the enzyme properties or structures of the enzymes (22). We studied PAD in human myeloid leukemia HL-60 cells, which can be induced to differentiate into granulocytes by retinoic acid and into monocyte/macrophages by 1␣,25-(OH) 2 D 3 or TPA (23). We report here the molecular characterization of HL-60 cell PAD induced by retinoic acid and regulation of its expression in myeloid differentiation.

EXPERIMENTAL PROCEDURES
Chemicals-Gigapack III Gold packaging extract and ZAP II/EcoRI/ calf intestine alkaline phosphatase-treated vector were from Strat-* This work was supported in part by a Sasakawa Scientific Research Grant from The Japan Science Society (to K. N.), grants-in-aid from the Ministry of Education, Science, Sports and Culture of Japan (to T. S.), and a grant for Promotion of Research at Yokohama City University (to M. Y.). 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.
The agene. Hybond-N nylon membranes, cyanogen bromide-activated Sepharose 4B, a GST expression system, and PreScission TM protease of a 3C protease were from Amersham Pharmacia Biotech. pCRII and a Fast Track mRNA isolation kit were from Invitrogen. SequeTherm TM and Long-Read TM Cycle Sequencing kits were from Epicentre Technologies. SuperScript II RT was from Life Technologies, Inc. Expand Taq DNA polymerase was from Roche Molecular Biochemicals. BAEE was from the Peptide Institute, Inc. Bz-L-Arg and RA were from Sigma. 1␣,25-(OH) 2 D 3 was from Wako Pure Chemicals Co. TPA was from Midland Corp. Rat muscle PAD type II (15), rat recombinant PAD type IV (20), rat PAD type II cDNA (17), rabbit anti-rat PAD type II serum (15), rabbit anti-modified citrulline IgG (24), rabbit-anti MPO serum (25), and MPO cDNA (26) were described previously.
Cell Culture-HL-60 cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (General Scientific Laboratories) and 50 g/ml kanamycin sulfate. For granulocyte differentiation, the cells were seeded at a density of 3 ϫ 10 5 cells/ml and cultured in the presence of 1 M RA or 1.25% Me 2 SO (25). For monocyte/ macrophage differentiation, the cells were cultured in the presence of 0.1 M 1␣,25-(OH) 2 D 3 or 10 ng/ml TPA (26). For TPA treatment, the cells were seeded at a density of 9 ϫ 10 5 cells/ml.
Assay of PAD Activity-PAD activity was determined using BAEE as a substrate as described previously (15). Harvested HL-60 cells were resuspended at 2 ϫ 10 8 cells/ml in a lysis buffer containing 20 mM Tris-HCl (pH 7.6), 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1% Triton X-100 and ruptured by freeze-thawing 3 times. The reaction mixture (50 l) containing 0.1 M Tris-HCl (pH 7.6), 10 mM CaCl 2 , 5 mM dithiothreitol, 10 mM BAEE, and 25 l of the cell lysate was incubated at 50°C for 1 h. Then the reaction was stopped by adding 12.5 l of 5 M perchloric acid. The perchloric acid-soluble fraction was subjected to a colorimetric reaction with citrulline as a standard. The reaction was linear with time up to 3 h with cell lysates and recombinant enzyme under the assay conditions. One unit of the enzyme was defined as the amount of enzyme catalyzing the formation of 1 mol of citrulline derivative in 1 h under the assay conditions. Kinetic parameters for BAEE and Bz-L-Arg were estimated from the activities assayed at 37°C for 1 h from Lineweaver-Burk plots. Protein concentrations were determined by the method of Bradford with bovine serum albumin as a standard (27).
Construction of a cDNA Library from RA-treated HL-60 Cells-Poly(A) ϩ RNA was prepared from HL-60 cells treated with 1 M RA for 3 days using a Fast Track mRNA isolation kit. cDNA was synthesized from 5 g of poly(A) ϩ RNA with oligo(dT) 25 (dA/C/G) as a primer using Moloney murine leukemia virus reverse transcriptase, followed by addition of EcoRI-NotI-SalI adaptor and phosphorylation of the 5Ј end using a Great Lengths cDNA synthesis kit (CLONTECH) according to the supplier's manual. The cDNA was ligated into an EcoRI site of ZAP II vector and then packaged at 22°C for 2 h using Gigapack III GOLD phage extract.
Screening of the cDNA Library-Approximately 5 ϫ 10 5 plaques were screened by plaque hybridization with a 32 P-labeled rat PAD type II cDNA probe prepared by the random oligoprimer DNA labeling method (28). Hybridization was carried out in a solution containing 5 ϫ SSPE, 5 ϫ Denhardt's solution, 50% formamide, 10% dextran sulfate, 1% SDS, and the probe (8 ϫ 10 7 cpm/5 ng/ml) at 45°C overnight. The membranes were washed twice with 2 ϫ SSC, 0.1% SDS at room temperature and 1 ϫ SSC, 0.1% SDS at 65°C for 15 min. They were exposed to x-ray film at Ϫ80°C (29). Positive cDNA clones were characterized by restriction enzyme digestions. Two PAD cDNA clones, 7-2 and 13-2, were chosen, subcloned into plasmids and sequenced. Clones with cDNAs for the 5Ј-end of the PAD were isolated by the 5Ј-RACE method (30). A sample of 1 g of poly(A) ϩ RNA isolated from RA-treated HL-60 cells was reverse-transcribed with SuperScript II TM using an antisense primer (nt 360 -342); 5Ј-CGGTGAGGTAGAGTAGAGC-3Ј. The first strand cDNA synthesized was polyguanylated with terminal deoxynucleotidyl transferase. The second strand cDNA was synthesized with Expand Taq DNA polymerase using the polyguanylated cDNA as a template and a C primer; 5Ј-GGCCCGACGTCGCATGAATTCGCCCCCCCCCC-CC-3Ј and then the cDNA was amplified by PCR using an ApaI primer; 5Ј-GGGCCCGACGTCGCATG-3Ј and a nested antisense primer (nt 329 -310); 5Ј-AGTCTTGGGTCCGTAGTATG-3Ј. The PCR product was subcloned into pCR II and sequenced.
DNA Sequencing-The cycle sequencing reaction was performed using an IRD41-labeled primer and SequeTherm DNA polymerase by the chain termination method (29). Nucleotide sequences were determined with an Li-COR DNA sequencer, model 4000L. The current nucleotide sequence and protein sequence data bases were searched with a BLAST program (31).
Preparation of a Recombinant GST-hPAD-The entire coding sequence of PAD cDNA was constructed from a 5Ј-RACE cDNA and 7-2 cDNA by overhang extension by PCR. Briefly, a 5Ј-RACE cDNA was amplified using an M13 p8 primer (TOYOBO) and the antisense primer (nt 329 -310) described above and then treated with T 4 DNA polymerase to excise the 3Ј extruded portion. The PCR product, whose 3Ј end overlaps the 5Ј end (nt 246 -329) of the 7-2 cDNA sequence, was annealed with KpnI-cut 7-2 cDNA and elongated at 68°C. The elongated product was amplified with a sense primer (hPAD-ex1: 27-mer) consisting of a 5Ј EcoRI site (underlined) and a 19-nt sequence (nt 27-45): 5Ј-CCGAATTCATGGCCCAGGGGACATTGA-3Ј, an antisense primer (hPAD-ex2: 35-mer) consisting of a 5Ј EcoRI-NotI site (underlined) and a 19-nt sequence (nt 2,093-2,075): 5Ј-CCGAATTCGCGGCCGCGAGC-TCTTGCTTGCCACAC-3Ј and Expand Taq DNA polymerase. The amplified cDNA was digested with EcoRI and subcloned into an EcoRI site of pGEX 4T-1 containing a thrombin site and named pGEX-hPAD. The hPAD cDNA was also subcloned into an EcoRI site of pGEX 6P-1 containing a 3C protease site. BL-21 cells transformed with pGEX-hPAD were grown in 2 ϫ YT medium at 25°C to a cell density of 1.0 at 600 nm and then after addition of 0.1 mM isopropyl-␤-D-thiogalactopyranoside for a further 5 h. The cells were resuspended in a lysis buffer containing 20 mM Tris-HCl (pH 7.6), 1 mM EDTA, and 0.1% Triton X-100 and disrupted by 2-3 passages through a French press. The cell lysate was brought to a concentration of 1 M NaCl and centrifuged at 15,000 ϫ g for 30 min. The supernatant was loaded on a glutathione-Sepharose 4B column and the column was thoroughly washed with lysis buffer containing 0.1 M NaCl. The recombinant fusion protein was eluted with a solution of 10 mM glutathione in 50 mM Tris-HCl (pH 8.0), 0.1 M NaCl, and 0.1% Triton X-100. The yield of enzyme activity was about 26%.
Preparation of Antiserum against PAD-Purified GST-hPAD (360 g) in complete Freund's adjuvant was injected into rabbits and then they were given a booster injection of the same antigen in incomplete Freund's adjuvant. Anti-PAD serum was applied to a GST-Sepharose column. The unabsorbed fraction contained anti-PAD activity. An aliquot was diluted 100-fold with PBS(Ϫ) and then incubated with 280 g/ml recombinant GST at room temperature for 20 min before use for immunoblotting. This preincubation was necessary for bleaching a nonspecific band of about 70 kDa.
SDS-PAGE and Immunoblotting-Sample proteins were subjected to SDS-10% PAGE by the method of Laemmli (33) and then transferred to a nitrocellulose membrane. For immunostaining of deiminated proteins, the membrane was treated at 37°C for 3 h with the medium for chemically modifying citrulline residues and then modified citrulline residues were detected by coupled immunoreactions with rabbit antimodified citrulline IgG (0.125 g/ml) for 1 h and horseradish peroxidase conjugate of goat anti-rabbit IgG (1:5,000) for 1 h by a reported method (7,24) with slight modification. Immunoblotting of PAD was performed using anti-GST-hPAD serum (1:3,000) or anti-rat type II PAD and 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 (NEN Life Science Products). The blot was reprobed with anti-MPO serum (1:3,000) after deprobing with a solution of 2% SDS, 62.5 mM Tris-HCl (pH 6.5), 0.1 M 2-mercaptoethanol as described (34).

Expressions of PAD Activity in HL-60 Cells Induced to Differentiate into Granulocytes and Monocytes-When
HL-60 cells were grown in the presence of RA, a granulocyte inducing agent, their PAD activity increased in the exponential phase of cell growth and reached a plateau in the stationary phase. No activity was detected in the absence of RA throughout the 3-day culture period (Fig. 1, A and B). During cell growth in the presence of RA, the MPO activity of the cells rapidly decreased to about 10% of that of control cells, indicating differentiation of the HL-60 cells into granulocytes (Fig. 1C), as reported previously (25). Various compounds are known to induce differentiation of HL-60 cells into granulocytes, monocytes, or macrophages (23). After additions of these compounds, the cells were examined for expression of PAD. Table I summarizes the effects of various differentiation inducers on the expressions of PAD in cells cultured for 2 days. Like RA, another granulocyte inducing agent, Me 2 SO also caused increase in PAD activity. The monocyte inducing agent 1␣,25-(OH) 2 D 3 increased PAD activity of the cells, while cells cultured with the macrophage inducing agent TPA, like control cells, did not show induction of PAD activity. The PAD in cells cultured with RA showed a ratio of activities to Bz-L-Arg and BAEE of about 1.5. The PAD in cells cultured with 1␣,25-(OH) 2 D 3 also showed similar activities to Bz-L-Arg and BAEE (data not shown). The ratios of the two HL-60 PADs differed from those of four known rat enzymes (1.0, 0.2, 0.2, and 0.2 for type I, II, III, and IV, respectively) (15,16,20).
We then examined whether PADs produced in HL-60 cells also act on cellular proteins ( Fig. 2A). Lysates of cells cultured with RA or 1␣,25-(OH) 2 D 3 for 3 days were incubated at 37°C for 1 h with or without 10 mM CaCl 2 and then subjected to SDS-PAGE. Deiminated proteins in the protein blots were probed with anti-modified citrulline IgG. On incubation with Ca 2ϩ , both the cell lysates showed numerous deiminated proteins migrating in a wide molecular weight range (lanes 5 and 6), but on incubation without Ca 2ϩ no deiminated proteins were detected (lanes 2 and 3). Untreated cell lysates did not show any deiminated proteins, regardless of the presence or absence of Ca 2ϩ (Fig. 2A, lanes 1 and 4). These results indicate that PADs in the RA cell lysates and 1␣,25-(OH) 2 D 3 cell lysates can deiminate various cellular proteins in the presence of Ca 2ϩ . In addition, the absence of detectable deiminated proteins in the intact cells suggested that a few proteins might be targeted slightly under in vivo conditions. Immunostaining of similar protein blots loaded with RA-and 1␣,25-(OH) 2 D 3 cell lysates containing 7 milliunits of PAD with anti-rat PAD type II IgG did not give any positive signals, although 2.8 and 14 milliunits of PAD of rat muscle PAD type II gave bands of about 72 kDa (Fig. 2C). This also suggested that the HL-60 PADs produced in cells cultured with RA or 1␣,25-(OH) 2 D 3 differ from the type II enzyme.
Cloning and Characterization of a Human PAD cDNA-To isolate and characterize the HL-60 PAD, we used a cDNA cloning strategy. We constructed a cDNA library in ZAP II    Fig. 1. The cells were harvested and disrupted by freeze-thawing. A, protein deiminating activity. Cell lysates equivalent to 2 ϫ 10 6 cells were incubated in the absence and presence of 10 mM CaCl 2 at 37°C for 1 h and then subjected to SDS-PAGE. The protein blots were probed with antimodified citrulline IgG using horseradish peroxidase-goat anti-rabbit IgG as a secondary antibody as described under "Experimental Procedures." Lanes 1-3, untreated and RA-and 1␣,25-(OH) 2 D 3 -treated cell lysates, respectively, incubated in the absence of Ca 2ϩ ; lanes 4 -6, the same as lanes 1-3, respectively, except that the lysates were incubated with Ca 2ϩ . B, protein blots stained with Amido Black 10B. Lanes 1-3, untreated cell lysate, and RA-and 1␣,25-(OH) 2 D 3 -treated lysates, respectively. C, immunoblotting of HL-60 PAD. Protein blots containing 7 milliunits of PAD activity of RA-and 1␣,25-(OH) 2 D 3 -treated cell lysates were probed with anti-rat type II PAD IgG as described above. Lanes 1 and 2, 2.8 and 14 milliunits of rat PAD II, respectively; lanes 3 and 4, 7 milliunits of PAD of RA-cell lysates and 7 milliunits of PAD of 1␣,25-(OH) 2 D 3 -cell lysates, respectively. from HL-60 cells treated with RA for 3 days, and then screened the library by plaque hybridization with rat PAD type II cDNA as a probe. Two positive cDNA clones, 7-2 and 13-2, were selected and sequenced. Their sequences overlapped, but a sequence for a 5Ј portion of PAD mRNA was missing. Thus, a 5Ј portion of PAD cDNA was prepared by the 5Ј-RACE method. Several 5Ј-RACE cDNAs were obtained and sequenced. They had the same sequence. Three overlapping cDNAs were 5Ј-RACE cDNA (nt 1 to 329), 7-2 cDNA (nt 246 to 2, 286), and 13-2 cDNA (nt 1,374 to 2,286). Alignment of the 5Ј-RACE cDNA and the 7-2 and 13-2 cDNAs gave a full-length cDNA named human PAD V cDNA (hPAD V cDNA). The cDNA was 2,286 bp long, and consisted of a 5Ј-untranslated region of 26 bp, a coding region of 1,992 bp, a 3Ј-untranslated region of 268 bp including a polyadenylation signal, AATAAA (nt 2,236 to 2,241), and a poly(A) tail (nt 2,264 to 2,286). The coding sequence encoded a polypeptide of 663 amino acid residues with a calculated molecular mass of 74,100 Da. The calculated pI of the protein was 6.12. The deduced amino acid sequence showed 55, 50, and 55% identities with those of rat PAD types I, II, and III, respectively, and 73% identity with rat keratinocyte PAD type IV, whose distribution in cells and tissues is not yet known. The carboxyl two-thirds of the sequences were relatively conserved, while the sequences of their amino-terminal one-thirds were more divergent (data not shown).
Expression and Characterization of a Recombinant HL-60 PAD-To express the above cloned PAD cDNA as a GST fusion protein in E. coli, we constructed the entire coding sequence (nt 27 to 2,093) of PAD from a 5Ј-RACE cDNA and a 7-2 cDNA by overhang extension and PCR and inserted it into the pGEX 4T-1 vector. An isolated construct of pGEX-hPAD contained one base substitution of G for A at nt 1,367, which did not result in any change in the amino acid sequence encoded by the original PAD cDNA. Cells transformed with pGEX-hPAD showed high PAD activity (specific activity 18.3 with BAEE as a substrate), but cells transformed with pGEX-hPAD␣ containing the PAD cDNA in the reverse direction had no activity (data not shown). Most of the enzyme activity in cell extracts was recovered in a soluble fraction and then was affinity purified on a GSH-Sepharose column with a yield of 26%. The preparation gave a single major band of approximately 97 kDa on SDS-PAGE (Fig. 3). Its specific activity (units/mg) was about 399, which was close to that of a homogeneous preparation of PAD type II purified from rat muscle. A GST-hPAD fusion protein was digested with PreScission 3C protease and then the recombinant enzyme was isolated. The activities of this enzyme on the synthetic substrates BAEE and Bz-L-Arg were studied at 37°C. The kinetic parameters V max , K m , K cat , and K cat /K m for these substrates were estimated from Lineweaver-Burk plots (Table II). The K cat value for BAEE was the same as that for Bz-L-Arg. The K m for BAEE was larger than that for Bz-L-Arg. The K cat /K m ratio for Bz-L-Arg was 1.5 times that for BAEE. The K m value for Bz-L-Arg with lysates of cells cultured with 1␣,25-(OH) 2 D 3 was estimated to be 0.7 mM and was similar to the value of 0.9 mM of that with the recombinant enzyme. The other kinetic properties of the recombinant enzyme also appeared to reflect the properties of PADs in cells cultured with RA and 1␣,25-(OH) 2 D 3 , which had relatively higher activity for Bz-L-Arg than for BAEE, as mentioned above.
The action of the recombinant PAD on cellular proteins in uninduced HL-60 cell lysates containing no endogenous PAD was also examined. The recombinant enzyme was incubated with an uninduced cell lysate with various concentrations of CaCl 2 of up to 10 mM and the resulting deiminated proteins were analyzed by immunoblotting with anti-modified citrulline IgG (Fig. 4). The deiminated proteins with a large range of molecular weights increased in a CaCl 2 concentration-dependent manner (lanes 1-7). Without CaCl 2 , no deiminated protein was detected (lane 1). With 10 M CaCl 2 , a faint signal at the dye front was detected (lane 2) and its intensity increased with increase in CaCl 2 concentration. With 0.5 mM CaCl 2 , four strong signal bands of 33, 34, 40, and 50 kDa besides the front band were seen (lane 4). With over 1 mM CaCl 2 , numerous cellular proteins were deiminated (lanes 5-10). These results indicated the absolute requirement of Ca 2ϩ for PAD action and the preference of PAD for some cellular proteins at a limited concentration of CaCl 2 . The different patterns of proteins deiminated by endogenous PAD in cell lysates ( Fig. 2A) and by recombinant enzyme added to cell lysates (Fig. 4) might be due to different subcellular localizations of the endogenous cellular PAD and recombinant enzyme. 2 Regulation of PAD Gene Expression in HL-60 Cells during Granulocyte and Monocyte/Macrophage Differentiations-We studied the dynamic nature of PAD expression in HL-60 cells by immunoblotting and Northern blotting. First, antiserum to a purified GST-hPAD protein was raised and its specificity for cellular proteins was studied by immunoblotting (Fig. 5A). Uninduced cells gave no signal (lane 1). Cells grown in the presence of RA or 1␣,25-(OH) 2 D 3 for 3 days gave a band of approximately 67 kDa (lanes 2 and 3). A partial thrombin digest of purified GST-hPAD gave two bands: one at about 67 kDa, nearly equivalent to the cellular 67-kDa band and the other at about 97 kDa, equal in size to the undigested GST-hPAD (lane 4). Addition of excess GST-hPAD to the serum completely eliminated the signals of cellular PADs from the two types of cells and GST-hPAD digests (lanes 5-8). Addition of an equivalent molar amount of recombinant GST did not affect the signals (lanes 1-4). Mixtures of the RA-cell lysates and recombinant enzyme gave no distinguishable 67-kDa band (data not shown). These results proved that the antiserum is specific for cellular 67-kDa PAD. We also examined the tempo-  ral changes of PAD expression in HL-60 cells cultured for 5 days with or without granulocyte-and monocyte/macrophageinducing agents in the same way (Fig. 5B, upper panel). During granulocyte differentiation induced by RA or Me 2 SO, a 67-kDa band became detectable on day 2 of culture and its signal intensity gradually increased during culture for 5 days (lanes 4 -11). Similary, during monocyte differentiation induced by 1␣,25-(OH) 2 D 3 , a band of 67 kDa became detectable on day 2 of culture and its intensity increased until the end of the culture period (lanes [12][13][14][15]. Untreated cultures gave no bands ( lanes  1-3). Moreover, surprisingly, differentiated macrophages induced by TPA gave no bands (lanes 16 -19). Rapid decrease in the amount of precursor MPO and progressive decrease in the amount of MPO were observed on the same blot, confirming the differentiation of HL-60 cells into granulocytes, monocytes, and macrophages under these culture conditions reported previously (23, 25, 26) (Fig. 5B, lower panel). These results indicated that the same 67-kDa PAD is produced in RA-and Me 2 SOinduced granulocytes and also in 1␣,25-(OH) 2 D 3 -induced monocytes, but is not produced in TPA-induced macrophages. Next, we examined the amount of PAD mRNA in similarly induced HL-60 cells using the above cloned cDNA as a probe for Northern blotting (Fig. 6, upper panel). The cells cultured with RA, Me 2 SO, and 1␣,25-(OH) 2 D 3 for 2 days gave a major band (about 2.6 kb) and a minor band (about 3.2 kb) of mRNA (lanes 2-4). Untreated cells gave no band (lane 1). Rehybridization of the blot with a 5Ј portion-specific probe (a 1.2-kbp sequence upstream of the XhoI site), the portion of which diverges in PADs, also gave bands of 3.2-and 2.6-kb mRNA (data not shown), suggesting that the 3.2-kb species was closely related to 2.6-kb mRNA. Decreasing intensities of MPO mRNA signals and similar intensities of glyceraldehyde-3-phosphate dehydrogenase mRNA signals during these treatments served as a cell differentiation marker and internal control, respectively. These results of Northern blotting and immunoblotting suggested that expression of the PAD gene is linked with granulocyte and monocyte differentiations of HL-60 cells and is regulated at the transcriptional level. DISCUSSION In this work we characterized a novel PAD induced in HL-60 cells during myeloid differentiation. This is the first characterization of PAD from cells of hematopoietic origin and human origin. The PAD activity in HL-60 cells was induced during RAand Me 2 SO-induced granulocyte differentiation and also during 1␣,25-(OH) 2 D 3 -induced monocyte differentiation, but not in TPA-induced macrophage differentiation. Expression of the PAD appears to be closely linked to cell-specific stages of myeloid differentiation of HL-60 cells. Mouse peritoneal macrophages and mouse macrophage-like cell lines show PAD activities (12,22). Although HL-60 cells can differentiate during normal myeloid differentiation in vivo and express many differentiation-associated markers, not all the phenotypes expressed in normal mature cells are always induced during HL-60 cell differentiations (23). Studies are needed on the dynamic expression of PAD during maturation of HL-60 monocytes into macrophages and in normal human monocytes and macrophages. The PAD of HL-60 monocytes appears to be identical with that of HL-60 granulocytes: both have similar activity on Bz-L-Arg and BAEE. The sizes of PADs detected by FIG. 4. Action of recombinant PAD on HL-60 cellular proteins. The standard reaction mixture for the PAD reaction was described under "Experimental Procedures," except that a mixture of 0.01 unit of purified GST-hPAD enzyme and untreated HL-60 cell lysate (50 g) as protein substrate was incubated in the absence and presence of various concentrations of CaCl 2 at 37°C for 1 h. The resulting deiminated proteins in 10 g of cell lysates were analyzed by immunoblotting with anti-modified citrulline IgG as described for Fig. 2A. immunoblotting in HL-60 granulocytes and monocytes were the same and the sizes of the mRNAs in the two types of cells were also the same. Several of the same phenotypes are known often to be induced in both granulocyte and monocyte differentiation of HL-60 cells (23). Since both PAD mRNAs and proteins were simultaneously detectable during HL-60 cell differentiation, the expression of PAD is regulated at a transcriptional level. HL-60 cells have RA and D 3 receptors (36,37). It is still unknown whether these receptors can activate PAD gene expression during HL-60 cell differentiation.
The molecular mass of the recombinant PAD was calculated to be 74,635 kDa including a molecular mass of 535 for the NH 2 -terminal 5-amino acid extension derived from a GSTlinker portion. Cellular PADs display a band of about 67 kDa on SDS-PAGE and recombinant PAD also shows similar mobility without an appreciable effect of the extension. The mobility of HL-60 PADs is significantly more than the expected mobility of the calculated molecular mass, owing to the basic nature of the protein. The mobility of HL-60 PAD is different from those of rat PAD types I, II, and IV, which display 72-kDa bands (data not shown). Recombinant HL-60 PAD showed similar relative activities toward BAEE and Bz-L-Arg, while PAD type IV has higher relative activity toward BAEE than to Bz-L-Arg (20). Thus HL-60 PAD seems to differ from rat type IV judging from its mobility on SDS-PAGE, its kinetic properties, and its amino acid sequence. Type IV cDNA has been cloned from a rat keratinocyte cell line and epidermis (20,21). But its occurrence in cells and tissues has not yet been demonstrated. The amino acid sequence of HL-60 PADs was compared with those of four known rat enzymes (20). HL-60 PAD showed 73% amino acid sequence homology with that of rat keratinocyte PAD IV and 50 -55% homologies with those of rat PAD types I, II, and III (17)(18)(19)(20)(21). The carboxyl two-thirds of the sequences of PADs are relatively well conserved, but their amino-terminal one-third portions are divergent. Interestingly, seven highly conserved sequences each consisting of 6 to 9 amino acid residues are located in PAD sequences and of these some are thought to be a Ca 2ϩ -binding site, a substrate recognition site or a catalytic site. However, by comparison of its sequences, it is hard to determine these sites. Homology search also revealed no Ca 2ϩ binding motif such as an EF-hand and C 2 motif in PAD. For understanding these functional sites, studies are required on the relationship of enzyme structures and their functions.
The biological role(s) of PAD in myeloid differentiation and in mature granulocytes, monocytes, and macrophages is entirely unknown. No deiminated cellular proteins were found in intact RA-and 1␣,25-(OH) 2 D 3 -induced HL-60 cells as shown in Fig. 2, although their cell lysates could deiminate cellular proteins on addition of Ca 2ϩ . These results suggest that PADs in intact cells are activated by external signals. When HL-60 granulocytes and monocytes are stimulated by the chemotactic factors fMet-Leu-Phe and leukotriene B 4 , the cytosolic free Ca 2ϩ concentration of a few micromolar is transiently elevated through calcium influx, and Ca 2ϩ ionophore stimulation also elevates the cytosolic concentration to 10 M (38 -40). Recently we reported that mouse peritoneal macrophages selectively deiminate vimentin when stimulated by Ca 2ϩ ionophore and that deiminated vimentin is accumulated in the periphery of nuclei. These events are considered to cause early changes in nuclear morphology with simultaneous apoptosis. PAD in cells is considered to be involved in a degenerative process through deimination of intermediate filaments such as vimentin and keratins. Interestingly, citrulline residues of deiminated filaggrin are constituents of epitopes recognized by autoantibodies in patients with rheumatoid arthritis (41)(42). The function of polymorphonuclear leukocytes and macrophages infiltrating into the synovial cavity of patients with rheumatoid arthritis is considered to be associated with inflammation and immune responses elicited by autoantibodies. The role of PAD in granulocytes/polymorphonuclear leukocytes and macrophages may be induced by external Ca 2ϩ stimuli generated in host defense responses of inflammation and immune responses. The PAD cDNA from hemopoietic cells and antiserum reported in this work should aid in studies on PAD in various stages of cells during granulocyte and monocyte/macrophage development.