Cloning and Expression of Bovine Neutrophil β-Defensins

β-Defensins are microbicidal peptides implicated in host defense functions of phagocytic leukocytes and certain surface epithelial cells. Here we investigated the genetic structures and cellular expression of BNBD-4, -12, and -13, three prototypic bovine neutrophil β-defensins. Characterization of the corresponding cDNAs indicated that BNBD-4 (41 residues) derives from a 63-amino acid prepropeptide and that BNBD-12 (38 residues) and BNBD-13 (42 residues) derive from a common 60-amino acid precursor (BNBD-12/13). The peptides were found to be encoded by two-exon genes that are closely related to bovine epithelial β-defensin genes. BNBD-4 and BNBD-12/13 mRNAs were most abundant in bone marrow, but were expressed differentially in certain non-myeloid tissues. In situ hybridization and immunohistochemical studies demonstrated that BNBD-4 synthesis is completed early in myelopoiesis. BNBD-12 was localized exclusively to the novel dense granules, organelles that also contain precursors of cathelicidins, antimicrobial peptides that undergo proteolytic processing during phagocytosis. In contrast to cathelicidins, Western blot analyses revealed that mature β-defensins are the predominant organellar form in myeloid cells. Stimulation of neutrophils with phorbol myristate acetate induced secretion of BNBD-12, indicating that it is co-secreted with pro-cathelicidins. The exocytosis of BNBD-12 by activated neutrophils reveals different mobilization pathways for myeloid α- and β-defensins.

␤-Defensins are microbicidal peptides implicated in host defense functions of phagocytic leukocytes and certain surface epithelial cells. Here we investigated the genetic structures and cellular expression of BNBD-4, -12, and -13, three prototypic bovine neutrophil ␤-defensins. Characterization of the corresponding cDNAs indicated that BNBD-4 (41 residues) derives from a 63amino acid prepropeptide and that BNBD-12 (38 residues) and BNBD-13 (42 residues) derive from a common 60-amino acid precursor (BNBD-12/13). The peptides were found to be encoded by two-exon genes that are closely related to bovine epithelial ␤-defensin genes. BNBD-4 and BNBD-12/13 mRNAs were most abundant in bone marrow, but were expressed differentially in certain non-myeloid tissues. In situ hybridization and immunohistochemical studies demonstrated that BNBD-4 synthesis is completed early in myelopoiesis. BNBD-12 was localized exclusively to the novel dense granules, organelles that also contain precursors of cathelicidins, antimicrobial peptides that undergo proteolytic processing during phagocytosis. In contrast to cathelicidins, Western blot analyses revealed that mature ␤-defensins are the predominant organellar form in myeloid cells. Stimulation of neutrophils with phorbol myristate acetate induced secretion of BNBD-12, indicating that it is co-secreted with pro-cathelicidins. The exocytosis of BNBD-12 by activated neutrophils reveals different mobilization pathways for myeloid ␣and ␤-defensins.
Polymorphonuclear leukocytes (PMN 1 ; neutrophils) provide one of the first lines of host defense against invading microorganisms. In addition to their production of microbicidal reactive oxygen intermediates, PMNs also inactivate microorganisms by exposing them to antimicrobial peptides and proteins in the phagolysosomal vacuole. In recent years several classes of phagocyte-derived antimicrobial peptides have been purified from mammalian phagocytes (reviewed in Refs. [1][2][3]. Similar molecules have also been isolated from specialized epithelia (4 -7), suggesting that antimicrobial peptides may play a role in the intrinsic resistance of tissues to microbial invasion.
The cytoplasm of ruminant neutrophils is populated with unique, dense (so-called large) granules that were reported to contain most of the cellular antimicrobial protein activity (8). Romeo, Gennaro, and co-workers purified from dense granules three cationic antimicrobial peptides termed bactenecins, peptides now known to be members of the cathelicidin antimicrobial peptide family (9 -11). Bovine neutrophil granules also contain ␤-defensins, a family represented by 13 cationic, tridisulfide-containing peptides containing 38 -42 residues (12). Although ␤-defensins are similar in conformation to ␣-defensins, the disulfide bonding patterns of the two defensin families are distinct (13,14) (Fig. 1). Neutrophil ␤-defensins have potent antibacterial activities against Staphylococcus aureus and Escherichia coli in vitro (12). In addition to their abundant expression in neutrophils, related ␤-defensins are expressed in the epithelium of bovine trachea (tracheal antimicrobial peptide, or TAP; Ref. 4), tongue (lingual antimicrobial peptide, or LAP; Ref. 7), and intestine (enteric ␤-defensin, or EBD; Ref. 15). Several homologous peptides were also purified from chicken and turkey leukocytes (16,17), and members of the ␤-defensin gene family from sheep (18), pig (19), and mouse (20,21) have been characterized at the DNA level. Two human ␤-defensins were recently described. The first, hBD-1, was isolated from plasma ultrafiltrate (22) and from urine, plasma, and vaginal secretions (23), and hBD-1 mRNA was detected in several epithelial tissues (24) including airway epithelium (25). hBD-2, isolated from psoriatic skin, was microbicidal or microbistatic against bacteria and yeast in vitro (26), and at least two studies have demonstrated that hBD-2 expression is elevated in inflammatory states (27,28). Investigations of the functions of mouse ␤-defensins are likely to provide new insights into the role of these peptides in host defense (20,29,30).
Here we report studies on neutrophil ␤-defensin genes, their expression during myelopoiesis, and the accumulation of ␤-defensin peptide in cytoplasmic organelles. The genes and cDNAs encoding three prototypical ␤-defensins, BNBD-4, -12, and -13, were cloned and sequenced, demonstrating a high degree of similarity between bovine neutrophil and epithelial ␤-defensins at the mRNA and protein levels. Expression of ␤-defensins in various tissues was assessed by Northern blot analysis, and the pattern of expression during myelopoietic differentiation was determined by in situ hybridization and immunohistochemical staining of neutrophils and neutrophil progenitors. In addition, the organellar site of ␤-defensin stores in mature neutrophils was established by immunogold electron microscopy. The extent of post-translational processing of the peptide precursor in immature and mature neutrophilic cell populations was determined, and studies were performed to establish the fate of ␤-defensin peptide during stimulation with phorbol myristate acetate.

EXPERIMENTAL PROCEDURES
Tissue Isolation-Neutrophils were obtained from anticoagulated whole blood of dairy cows as described previously (12). RNA for cDNA library construction was obtained from calf bone marrow curetted from femurs and flash frozen in liquid N 2 within 2 h of sacrifice. Northern blot, primer extension, and RACE analyses were carried out using bone marrow isolated from adults animals as described (31). In situ hybridization studies were performed on calf bone marrow cells. Leukocyte DNA was used for genomic Southern hybridizations.
General Methods for Cloning and DNA Characterization-Oligonucleotide probes were end-labeled to a specific activity of ϳ10 7 dpm/pmol using [␥-32 P]ATP as described (31). Purified plasmid DNA was sequenced from both strands using the dideoxy-chain termination method (32). PCR products were incubated in a standard fill-in reaction and then subcloned as described (33).
Genomic Cloning-A bovine genomic library in EMBL-3 (CLON-TECH, Palo Alto, CA) was screened with 32 P-labeled TAP48a (5Ј-CCA-AGCAGACAGGACCAGGAAGAGGAGCGCGAGGAGCAGGTGATGG-AGCCTCAT) as described (31). TAP-48a-positive plaques were rescreened at lower phage density through tertiary screens until pure. Phage DNA was isolated from liquid culture (36), and EcoRI restriction enzyme fragments of phage insert DNA were subcloned by ligation into the multiple cloning site of pBluescript II SKϩ plasmid DNA (Stratagene). Preliminary sequence analysis indicated that two of the phage clones encoded BNBD-4 and BNBD-12/13, and these were selected for further analysis.
Sequence Alignments-Sequence data were analyzed using MacVector 6.0 (IBI, New Haven, CT) or Geneworks 2.4 (IntelliGenetics, Mountain View, CA) software packages. cDNA and predicted amino acid sequences were compared within and between species using the sequence alignment function of Geneworks 2.4. cDNA and genomic sequences and the predicted cDNA translation product were compared with the GenBank™ data base using BLAST (37,38).
Northern Blot Analysis-Total RNA was extracted and subjected to Northern blot analysis as described (31). The ␤-defensin probes were BNBD-4/251a (see above) and BNBD-12/239a (GTTCTGTCGAAGGG-CACAGTTTCTGTCTCCGCGTAGG). Labeled probes were hybridized overnight to immobilized RNA in 37.5% (v/v) formamide, 5ϫ SSC, 5ϫ Denhardt's, 1% (w/v) SDS at 42°C, and then washed at high stringency in 0.1ϫ SSC, 0.1% SDS at 57°C for 30 min (36). A bovine ␣-tubulin cDNA probe was used as a control for RNA integrity (31). The filter was stripped of oligonucleotide label by incubation in 0.1ϫ SSC, 0.1% SDS at 70°C for 30 min and exposed to film to assure removal of probe prior to hybridization with another probe.
In Situ Hybridization-Heparinized bovine bone marrow cells were cytocentrifuged onto poly-L-lysine-coated slides, fixed, and dehydrated as described previously (34). UTP-digoxigenin-labeled sense and antisense riboprobes were transcribed from linearized plasmid containing the full-length BNBD-4 cDNA using T7 or Sp6 RNA polymerases (40), and hybridized to fixed cells as described previously (34). RNA-RNA hybrids were detected with alkaline phosphatase-conjugated antidigoxigenin antibody and developed with nitro blue tetrazolium (40).
Antibody Production-Antibodies were prepared against synthetic linearized (reduced and carboxymethylated) BNBD-12, and separately against folded and oxidized peptide. Details of the BNBD-12 synthesis will be described elsewhere. Briefly, the peptide chain was assembled using automated Fmoc (N-(9-fluorenyl)methoxycarbonyl) solid phase synthesis, and following acidolytic cleavage and deprotection of the peptide resin, the linear peptide was purified by reversed phase HPLC. Five mg of reduced peptide was oxidized by stirring a solution containing 100 g/ml peptide in 0.1 M Tris acetate, pH 8.0, for 48 h in room air. Following oxidative folding, the peptide was purified to homogeneity by RP-HPLC. The purified peptide was shown to be identical to natural BNBD-12 by acid-urea PAGE, RP-HPLC, and matrix-assisted laser desorption ionization/time of flight mass spectroscopy. Linearized BNBD-12 was prepared by reducing 1 mg of the purified, oxidized/ folded peptide with DTT, followed by alkylation with iodoacetic acid as described previously (41). The linearized, S-carboxymethylated peptide was dialyzed exhaustively against 5% acetic and lyophilized. Folded/ oxidized (2.0 mg) and linearized (1.0 mg) BNBD-12 preparations were individually conjugated to an equal weight of crystallized ovalbumin (Sigma) in 20 mM sodium phosphate buffer, pH 7.4, containing 6.6 mM glutaraldehyde. The conjugation mixtures were stirred for 4 h at room temperature, then quenched by addition of glycine to 150 mM final concentration, dialyzed against water, and lyophilized. New Zealand White rabbits were separately immunized with the resulting conjugates to generate anti-native BNBD-12 or anti-linear BNBD-12 antibodies. Antibody titers obtained from each rabbit were greater than 1:10,000 as determined by enzyme-linked immunosorbent assay. The IgG fraction of each antiserum preparation was generated using DEAE Econo columns (Bio-Rad) per manufacturer's instructions. Antisera or IgG preparations were used for immunocytochemistry, immunogold electron microscopy, and Western blotting as indicated.
Immunohistochemistry-Cytospin slides were prepared using suspensions of peripheral blood leukocytes and suspensions of calf bone marrow cells. Slides were blocked for 30 min with 5% normal goat

FIG. 1. Covalent structural comparison of ␣and ␤-defensins.
The schematic shows the similar size and dissimilar cystine motifs of the two myeloid defensin families. Single letters indicate highly conserved amino acids. X denotes positions that are more variable. The disulfide connectivities in the ␣ family are 1-6, 2-4, and 3-5; in the ␤ family they are 1-5, 2-4, and 3-6. serum and incubated with anti-native BNBD-12 antibody at the indicated dilutions at 8°C for 18 h. Negative control incubations were performed using preimmune serum or peptide-preabsorbed antiserum. Slides were washed and developed using biotinylated goat anti-rabbit IgG using a VectaStain avidin-biotin-glucose oxidase kit (Vector Laboratories, Burlingame, CA) per the manufacturer's recommendations. Nuclear Fast Red was used as counterstain.
Electron Microscopy-Peripheral blood neutrophils were prepared for transmission electron microscopy using standard methods. Briefly, cells were fixed in 2% glutaraldehyde, 0.1 M sodium cacodylate, pH 7.4, post-fixed in 1% osmium tetroxide, and counterstained with uranyl acetate. Samples were then dehydrated in a series of graded ethanols, immersed in propylene oxide, and embedded in Epon 812/araldite resin. Sections (600 -900 Å) were prepared using an Ultracut E microtome (Reichert) and evaluated with a Zeiss 10A transmission electron microscope. Cells were stained for peroxidase prior to transmission electron microscopy. For immunogold labeling, leukocytes were suspended in PBS and fixed by adding an equal volume of 8% paraformaldehyde, 10% sucrose, in PBS, pH 7.4. After 1 h at room temperature, the suspension was washed three times with PBS containing 5% sucrose, and then embedded in 10% gelatin, 5% sucrose in PBS by centrifugation at 200 ϫ g. The gelatin plug was cut into small pieces, infiltrated with 2.3 M sucrose/PBS overnight at 4°C, mounted on grids, frozen, and stored in liquid nitrogen. Thin sections were cut as described previously (42), and incubated for 1 h with 1:10 anti-native BNBD-12 IgG as primary antibody. The secondary antibody was goat anti-rabbit IgG conjugated to 10 nm gold beads (Zymed Laboratories Inc.) and was used at 1:50 dilution. Preimmune rabbit serum was used in the primary incubation as a negative control.
Western Blot Analysis-Peripheral blood leukocytes and fresh bone marrow cells were suspended in Hanks' balanced salt solution containing 2.5 mM MgCl 2 . Cell suspensions were counted, pelleted by centrifugation, and snap-frozen. Some cell preparations were treated for 10 min at 0°C with 2 mM diisopropyl fluorophosphate (DFP), a cell-permeable serine protease inhibitor, as described previously (43) prior to freezing. Extracts of DFP-treated or untreated cells were obtained by one of the following methods. (i) A frozen cell pellet was immersed in a boiling water bath. Simultaneously, 250 l of preheated (ϳ95°C), DTTcontaining SDS-PAGE sample buffer (44) was added, the suspension was mixed and boiled for 5 min. DNA in some samples was sheared with a 22-gauge needle, and these samples were boiled once again for 5 min.
(ii) Alternatively, the cell pellet was suspended in 5-10 ml of ice-cold 10% acetic acid, briefly sonicated, and stirred for 18 h at 0°C. The suspension was clarified by centrifugation at 27,000 ϫ g for 20 min at 4°C, and the supernatant was lyophilized in aliquots of 0.5-1.0 ϫ 10 7 cell eq/tube. Lyophilized material was dissolved in DTT-containing SDS-PAGE sample buffer, and boiled for 5 min. Immediately prior to electrophoresis, DTT-reduced samples were alkylated with a 3 M excess of iodoacetic acid in the dark for 5-20 min, and the alkylation reaction was quenched by addition of excess DTT. Samples were electrophoresed on Tricine-SDS gels containing 15% acrylamide (44), and electroblotted onto 0.22-m nitrocellulose membranes (MSI, Westborough, MA) using a semidry blotting apparatus. Lyophilized samples obtained by acid extraction were also analyzed by acid-urea (AU)-PAGE on 12.5% gels and transferred to nitrocellulose membranes (45). For immunodetection of blots from SDS gels, the primary incubation employed 1:1000 or 1:5000 dilutions of anti-linear BNBD-12 IgG; for blots of AU gels, 1:3000 diluted anti-native BNBD-12 antiserum was used. In all cases, the second antibody was 1:100,000 goat anti-rabbit IgG antiserum conjugated to horseradish peroxidase. Blots were developed by exposing x-ray film to the chemiluminescence signal obtained using Ultra or Super Signal substrates (Pierce).
Neutrophil Activation-Freshly prepared neutrophils (Ͼ95%) were suspended in Hanks' balanced salt solution, pH 7.4, without calcium, 2.5 mM MgCl 2 to a cell density of 5 ϫ 10 7 cells/ml. Suspensions were activated by addition of phorbol myristate acetate (PMA) in 0.1% Me 2 SO to a final concentration of 30 or 100 nM, and incubated with gentle tumbling for 30 and 60 min at 37°C. At the specified intervals, cells were tested for viability by trypan blue exclusion (Ͼ90% at each time point), and the incubation mixtures were centrifuged at 234 ϫ g for 15 min at 4°C. The supernatant was then removed, placed on ice, acidified with 10% acetic acid, and lyophilized. Cell pellets were suspended in ice-cold 10% acetic acid and sonicated for 15 s on ice. Sonicates were stirred overnight at 8°C, clarified by centrifugation, and the  (12). The locations of the spliced introns are indicated with vertical arrows. The predicted translation start site (67) and the polyadenylation addition signal (68) are in bold. A, BNBD-4. The sequence represents a composite from the sequence of a gt10 cDNA clone and four clones from 5Ј-RACE PCR analysis. The gt10 clone contained the sequence spanning nucleotides 117 to 452. Two of the RACE clones extend from the primer site (391) to the 5Ј-most nucleotide (ϩ1), and the other two suggest a start at ϩ3. The GenBank™ accession number for the BNBD-4 cDNA sequence is U36200. B, BNBD-12/13. The sequence represents a composite from the genomic EMBL3 clone and the reverse transcription-PCR product from bone marrow derived mRNA. The inverted brackets indicate the sequence obtained from the cDNA sequence, which spans the splice junction site. The first four underlined residues (SGIS) correspond to the amino-terminal tetrapeptide of BNBD-13; the vertical line at Gly-23 indicates the amino-terminal residue of BNBD-12.

Characterization of cDNA and Genomic
Clones-To facilitate studies on the temporal and tissue-specific patterns of neutrophil ␤-defensin expression, we initially isolated and characterized the cDNA encoding BNBD-4. Six cDNA clones encoding BNBD-4 were isolated from a bone marrow library screened with an oligonucleotide corresponding to a 10-amino acid sequence conserved in ␤-defensins BNBD-2-6 and -10 -12/13 (12). The longest of these clones extended 19 nucleotides upstream of the putative initiation methionine, and the 125-nucleotide 3Ј-untranslated region contained a polyadenylation consensus signal 35 nucleotides upstream of the poly(A) tail (Fig. 2). The cDNA encoding BNBD-12/13 was obtained following reverse transcription-PCR of bone marrow mRNA (see "Experimental Procedures") and confirmed by comparison with the sequence of the corresponding genomic clone (see below).
The BNBD-4 and -12/13 cDNAs predict precursors of 63 and 60 amino acids, respectively, each containing an 18 -22-residue segment containing a signal sequence. The mature BNBD-4, -12, and -13 peptides are composed of the carboxyl-terminal 41, 38, and 42 residues of the respective precursors (Figs. 2 and 3). For both peptide precursors, the presumed translation start was implied by the presence of a Kozak consensus sequence just proximal to the AUG (Fig. 2) and coding sequences for signal peptides, which are highly similar to those in other epithelial ␤-defensin precursors (Fig. 3). The 4-residue (SGIS) "pro-segment" in BNBD-12 is identical to that of the first 4 amino acids of BNBD-13 previously isolated from peripheral blood neutrophils, suggesting that BNBD-12 and -13 derive from alternate processing of a common precursor.
Comparison of the bovine ␤-defensin precursors demonstrates that leukocyte and epithelial peptides (EBD, TAP, and LAP) are similar in size and sequence, with the greatest degree of sequence conservation being in the signal peptide regions (Fig. 3). There is slightly more sequence similarity among the respective myeloid and epithelial ␤-defensins than between them. On the other hand, hBD-2, a human epithelial defensin, differs substantially from the bovine peptides along the entire primary sequence (Fig. 3) BNBD-4 and BNBD-12/13 genomic clones were isolated by screening a genomic library to identify TAP-related sequences (4). Plasmid subclones from two of the isolated phage clones were isolated, subcloned, and sequenced, revealing that they encoded BNBD-4 and BNBD-12/13. Comparison of the BNBD-4 genomic clone with the corresponding cDNA sequence indicated that the gene consists of two exons flanking an intron of 1486 nucleotides (Fig. 4, A and B; GenBank™ accession number AF008307). 5Ј-RACE analyses established two possible sites of transcription initiation for BNBD-4, either 133 or 135 nucleotides upstream of the initiation methionine. Analysis of the 5Ј-flanking region revealed a number of eukaryotic consensus sequences for putative cis-acting regulatory elements (46,47). A TATA box was identified at Ϫ22, and putative CAAT enhancer-binding protein (C/EBP; Ϫ147; Ref. 46) and polyoma enhancer binding protein-2/core binding factor (PEBP2/CBF; Ϫ100; Ref. 47) response elements were also localized to this region.
Analysis of the BNBD-12/13 genomic clone confirmed the cDNA analysis indicating that a single genomic sequence encodes both BNBD-12 and BNBD-13, and that it is very similar to BNBD-4 in organization and sequence (Fig. 4C). The genes were 88% identical across coding and adjacent flanking regions. Interestingly, although the size of the intron of the BNBD-12/13 gene (approximately 1.5 kilobases as estimated by restriction map analysis) was similar to BNBD-4 gene, and despite highly similar sequence surrounding the 3Ј-splice acceptor site, the site of splicing differed by 12 nucleotides, corresponding exactly to the four additional residues (QRVR) present in the pro-segment of BNBD-4, which are lacking in BNBD-12/13 (Figs. 3 and 4).
Southern analysis was performed to estimate the copy number of the BNBD-4 gene. Under conditions of low stringency, Southern blots probed with a ␤-defensin-specific oligonucleotide (BNBD-4 251a) revealed multiple bands (data not shown), which likely indicate cross-hybridization to other ␤-defensin genes. At higher stringency, single bands were detected in lanes corresponding to each restriction digest (Fig. 5). While not conclusive, these findings suggest that BNBD-4 is encoded by a single copy gene among a family of closely related genomic sequences.
Tissue Distribution of BNBD-4 and BNBD-12 mRNAs-Myeloid ␤-defensins were initially isolated from bovine peripheral blood neutrophils, but other possible sites of expression had not been investigated. BNBD-4 and -12 mRNAs from various bovine tissues were analyzed by Northern blot hybridization using gene-specific probes designed as described under "Experimental Procedures." Abundant mRNAs of ϳ0.5 kilobase encoding both peptides were identified in bovine bone marrow (Fig. 6). BNBD-4 mRNA was also detected at lower levels in distal small intestine, as well as trachea, lung, and spleen. The tissue specificity of BNBD-12 hybridizing signal was similar to that of BNBD-4, except for the absence of BNBD-12 mRNA in lung and spleen. These data indicate that neutrophil ␤-defensin genes are expressed in myeloid and non-myeloid tissues.
BNBD-4 Gene Expression during Neutrophilic Myelopoiesis-To determine the myelopoietic stage at which ␤-defensin transcripts first appear, we performed in situ hybridization experiments on bone marrow cell cytocentrifuge preparations using an antisense riboprobe corresponding to the full-length BNBD-4 cDNA. The probe hybridized strongly to promyelocytes and myelocytes, the earliest recognizable stages of neutrophilic myelopoiesis (Fig. 7, A-C). Faint hybridization signals were detected in myeloblasts, and low levels of BNBD-4 mRNA were detectable in metamyelocytes, but no signal was detectable in neutrophilic bands or mature cells. Parallel hybridizations with sense riboprobes were negative (Fig. 7D). These data indicate that the BNBD-4 gene is selectively transcribed during the early stages of granulocyte differentiation, but not at the terminal stages of neutrophilic myelopoiesis or in mature neutrophils.
BNBD-12 Peptide Accumulation during Neutrophilic Maturation-Rabbit anti-BNBD-12 antibody was used to detect ␤-defensin peptide in fixed cytospins of peripheral blood leukocytes. As shown in Fig. 8A, the antibody stained the cytoplasm of mature neutrophils with a granular pattern. BNBD-12 immunoreactivity was not detected in eosinophils or mononuclear cells (Fig. 8A), and preimmune serum was completely nonreactive with all cell types (Fig. 8B). Bone marrow cells were similarly stained with anti-BNBD-12 antibody. In these experiments, all identifiable neutrophil progenitors were immunopositive, but the granularity of staining was much finer than observed with peripheral mature cells (Fig. 8C). Cells stained with BNBD-12-preabsorbed antibody were negative (Fig. 8D). These data demonstrate that peptide of native conformation is detectable at all stages of neutrophil differentiation. Further, it reveals a close correlation between ␤-defensin gene transcription, assessed by in situ hybridization, and the cytoplasmic accumulation of the translated product.
Localization of ␤-Defensins to Neutrophil Dense Granules-A novel feature of ruminant neutrophils is the presence of at least three types of cytoplasmic granules (48), one of which, the large dense granule (Fig. 9A), has been shown previously to contain propeptide forms of cathelicidins (49). To establish the organellar distribution of ␤-defensin in bovine neutrophils, we performed immunogold localization studies on thin sections of cells reacted with anti-BNBD-12 antibody. As shown in Fig. 9, gold particles localized exclusively over dense granules, but were absent from sections over the other two granule types, the azurophil and specific granules. The preimmune serum control incubation showed no immunoreactivity (Fig. 9D).
Molecular Form of ␤-Defensins in Immature and Mature Leukocytes-Zanetti and co-workers (49) demonstrated that the cathelicidins Bac5 and Bac7 are stored in bovine neutrophil dense granules as inactive propeptides, and that their conver-sion to mature, microbicidal peptides requires proteolytic processing. To determine the molecular form of ␤-defensins in neutrophil dense granules, we analyzed extracts of neutrophils and bone marrow cells by Western blotting. Extracts were produced under various conditions to address and exclude the possibility of proteolytic degradation during extraction. To this end we used boiling-SDS or 10% acetic acid to extract snap-frozen cell pellets that had been pretreated with DFP and compared these extracts to those generated without DFP treatment. In SDS-PAGE Western blots of DFP-treated or acid-extracted neutrophils, anti-linear BNBD-12 IgG detected a single band that comigrated with S-carboxymethylated BNBD-12 (Fig. 10A). The immunoreactivity was independent of the protocol employed for extraction, indicating that mature BNBD-12 is the predominant peptide form in neutrophils. An identical result was obtained when bone marrow cell lysates were analyzed by Western blot. As shown in Fig. 10B, anti-linear BNBD-12 antibody detected only a single band that comigrated with mature BNBD-12. As observed with neutrophil lysates, this pattern was independent of the method of cell extraction. Because of their similar sizes, BNBD-12 and BNBD-13 are not easily resolved on SDS-PAGE. However, because all bovine native ␤-defensins migrate with unique R F values in AU-PAGE (50), we utilized this electrophoretic system to further analyze BNBD-12 immunoreactivity in neutrophil and bone marrow lysates. As shown in Fig. 10C, anti-native BNBD-12 antibody FIG. 7. In situ hybridization analysis of bovine bone marrow cells. Bone marrow cell cytospin slides were prepared for in situ hybridization as described under "Experimental Procedures." Bone marrow cells were identified by standard criteria using bright field and phase contrast microscopy (70) and are identified as follows: promyelocyte (PM) and myelocyte (MC). Bar corresponds to 20 M. Hybridization of bovine bone marrow preparations was performed using antisense (A-C) and sense (D; negative control) riboprobes corresponding to the full-length BNBD-4 cDNA. Positive hybridization was detected with alkaline phosphataseconjugated anti-digoxigenin antibody developed with nitro blue tetrazolium and counterstained with nuclear fast red (40). detected both BNBD-12 and -13 standards, which were readily resolved in AU-PAGE. Analysis of PMN and bone marrow cell acid extracts revealed the presence of only BNBD-12, irrespective of whether the preparations were first treated with DFP. Taken together, these data indicate that BNBD-12 is the predominant molecular form both in biosynthetically quiescent neutrophils and in biosynthetically active bone marrow.
Exocytosis of BNBD-12 by Activated Neutrophils-Neutrophils were stimulated with 30 or 100 nM PMA, a potent secretagogue that selectively mobilizes secretory granules and vesicles. Western blot analysis was then used to determine whether BNBD-12 exocytosis occurs under these conditions, ones previously shown to induce the secretion of Bac5 and Bac7 (51) (Fig. 11). No BNBD-12 was detected in any incubation supernatant at the 30-min time point, and control cells (no PMA) released no BNBD-12 throughout the 60-min experiment (n ϭ 5). After 60 min, BNBD-12 was detected in supernatants at both PMA concentrations. Based on analysis of the Western blot signals, we estimate that 15-20% of the cellular BNDD-12 is released and freely soluble. DISCUSSION ␤-Defensins are the second family of tridisulfide host defense peptides found in mammalian cells (52). Peptides comprising the first such family, ␣-defensins, are genetically distinct from ␤-defensins, and the covalent structures of the two peptide families, including their disulfide motifs, are dissimilar (Fig. 1). The precursor structures of ␣-defensins and ␤-defensins are also quite different. Myeloid ␣-defensin precursors are 87-97 amino acids in length (34). A 19-residue signal sequence is followed by a 39 -45-residue polyanionic pro-segment, and the carboxyl-terminal 29 -34 residues correspond to the mature defensin. While signal sequences occur in each of the known ␤-defensin precursors (Fig. 3), an acidic pro-segment is absent. Instead, a stretch of 0 -4 neutral amino acids separates the signal and mature peptide sequences (Fig. 3). The difference in ␣and ␤-defensin precursor structures is of interest, as it has been suggested that the anionic pro-segment in ␣-defensins neutralizes the cationic charge of the mature peptide during biosynthesis and organellar packaging to protect the producing cell from autocytotoxicity (53,54). It is evident that charge balance in the ␤-defensin precursor is not required since the net charge of the initial translation product, as well as the mature peptides, is strongly positive (Fig. 3). The structural differences of the two defensin families suggest that they may be sorted and packaged by distinct intracellular pathways.
Although the genetic organization and primary and disulfide structures of the ␣and ␤-defensins differ (Fig. 1), Zimmermann et al. (14) reported that the peptide fold of BNBD-12 is very similar to that of the ␣-defensins. The possible ancestral relationship of the two defensin families was suggested by recent gene mapping data, which place ␣and ␤-defensin genes within a 150-kilobase segment on human chromosome 8 (55). An additional clue may be inferred from the isolation and structure determination of a "big defensin" from horseshoe crab hemocytes (56). This peptide possesses the ␤-defensin disulfide bonding pattern, but certain other sequence features are more similar to the rat neutrophil ␣-defensin NP-2. Possibly this invertebrate peptide is related to an ancestral peptide from which the two mammalian defensin families have evolved.
As expected, BNBD-4 and -12/13 mRNAs were abundant in bone marrow (Fig. 6), the predominant site of neutrophil production in adult animals. In situ hybridization of bone marrow cells demonstrated that BNBD-4 mRNA levels were highest in promyelocytes and myelocytes, the myelopoietic stages during which dense granule formation occurs (48). Lower BNBD-4 mRNA levels were detectable in myeloblasts, and trace level signals were observed in some metamyelocytes. However, no hybridization was observed in bands or mature neutrophils, indicating that ␤-defensin synthesis is complete at the time of release of the mature neutrophil into the peripheral circulation. The overall profile of myeloid ␤-defensin mRNA synthesis resembles the pattern of gene transcription observed for neutrophil ␣-defensins (34, 57) except that peak ␤-defensin gene expression appears to be slightly shifted toward more mature bone marrow elements. This is consistent with the sequence of granulogenesis in ruminant neutrophils, in which the appearance of azurophil granules (the site of ␣-defensin storage in non-ruminant PMN) is followed closely by the formation of ␤-defensin-containing dense granules (48).
With rare exceptions, myeloid ␣-defensin expression has only been noted in neutrophil granulocyte precursors of several mammals, or in rabbit pulmonary macrophages. On the other hand, we detected mRNAs for BNBD-4 and/or BNBD-12/13 in trachea, lung, spleen, and intestine (Fig. 6). Recently, Tarver et al. (15) cloned the cDNA for BNBD-4 from bovine small intestine, and Ryan et al. (58) demonstrated that bovine alveolar macrophages express BNBD-4 and -5, as well as two "epithelial" ␤-defensins, TAP and EBD. Since mature neutrophils lack ␤-defensin mRNAs, the mRNA levels in these tissues most likely represent endogenous gene expression in these organs (15). However, it is possible that neutrophils might be activated to express one or more ␤-defensin in certain tissues.
Comparison of the gene sequences of the epithelial (TAP, EBD) and myeloid ␤-defensin genes may facilitate studies aimed at understanding their regulation and tissue-specific expression. Although the 5Ј-flanking sequences (ϳ750 nucleotides) of BNBD-4, TAP, and EBD are 85% identical, analysis of this region for known transcription factor-binding consensus sequences failed to identify motifs that differentiate BNBD-4 from the epithelial ␤-defensins. Previous studies have demonstrated induction of epithelial ␤-defensin expression in re- FIG. 10. Western blot analysis of BNBD-12 in neutrophils and bone marrow cells. A, immunoblot of PMN extracts. A 15% Tricine-SDS gel was loaded with 0.5 g each of reduced, carboxymethylated BNBD-12 and -13 as standards, and four lanes were each loaded with 2 ϫ 10 6 cell eq of PMN extract. Cells were either extracted directly by the method indicated, or first treated with DFP as described under "Experimental Procedures." Anti-linear BNBD-12 IgG was the primary antibody. The slower migrating bands in the BNBD-12 and -13 lanes are peptide dimers that are sometimes generated during the alkylation step. The band at ϳ15 kDa is unrelated to BNBD-12 since it also appears in blots developed with only the secondary antibody. Molecular size markers are aprotinin (6.5 kDa) and ␣-lactalbumin (14.4 kDa). B, immunoblot of bone marrow cells. Bone marrow cells were prepared, electrophoresed, and tested for BNBD-12 immunoreactivity exactly as in panel A. Each lane was loaded with 1 ϫ 10 6 cell eq. BNBD-12 and -13 standards were 0.25 g/lane. C, acid-urea PAGE of PMN and bone marrow cell extracts. PMNs or bone marrow cells (4 ϫ 10 6 cell eq/lane) were extracted with 10% acetic acid directly or following treatment with DFP as indicated. Immunodetection was with anti-native BNBD-12 antiserum. Standards were native BNBD-13 (2.0 g) and BNBD-12 (0.5 g).
FIG. 11. Exocytosis of BNBD-12 from activated neutrophils. Freshly obtained PMN were stimulated with PMA as described under "Experimental Procedures." Supernatant samples corresponding to 4 ϫ 10 6 cells from each incubation were reduced, alkylated, and electrophoresed on a 15% Tricine-SDS gel. Pelleted cells from the corresponding incubations were extracted, and 1 ϫ 10 6 cell eq of each sample were loaded. Immunoblotting was performed with 1:3000 dilution of antilinear BNBD-12 IgG as described in the legend to Fig. 10. sponse to extracellular stimuli (7,15,59,60). Increased TAP mRNA levels are induced by exposure to endotoxin (59) and tumor necrosis factor (61) and may be mediated in part by an NF-B recognition site in the TAP gene promoter region. Neither BNBD-4 nor EBD have an NF-B consensus binding sequence. However, all three of these ␤-defensin genes contain putative C/EBP␤ recognition sequences (46). C/EBP␤, also known as NF-IL-6 (62), is a member of the basic region-leucine zipper (␤-zip) family of transcriptional activators and functions to transactivate numerous genes in response to lipopolysaccharide, interleukin-1␤, and tumor necrosis factor (63). Additionally, the BNBD-4 promoter region contains a PEBP2/CBF recognition sequence (47) that is also present in the 5Ј-flanking sequence of the EBD gene and in the intron of the TAP, EBD, and BNBD-4 genes. PEBP2/CBF has been implicated in the regulation of other myeloid-specific genes. MyNF-1, a putative homolog of PEBP2/CBF␣, binds to and transactivates a PEBP2/CBF response element in the myeloperoxidase and neutrophil elastase genes of murine promyelocytic cells (47). Since BNBD-4 expression is tightly linked to the phase of cellular differentiation (Fig. 7), we speculate that the PEBP2/CBF response element may participate in the regulation of ␤-defensin gene expression during specific stages of neutrophilic myelopoiesis. Future studies will be necessary to characterize the function of these putative regulatory elements and will likely depend on the availability of ␤-defensin-expressing myeloid cell lines.
It was previously shown that the bovine neutrophil dense granules, organelles that distinguish ruminant neutrophils from leukocytes of non-ruminant mammals, contain cationic antimicrobial peptides and proteins not present in azurophil or specific granules (64). Conversely, dense granules lack azurophil or specific granule markers, with the exception of lactoferrin (64). Gennaro and co-workers (10,49,51,65) demonstrated that dense granule antimicrobial factors include peptides termed bactenecins that are members of the cathelicidin family. Data presented here (Fig. 9) demonstrate that ␤-defensins are also stored in dense granules. Consistent with the immunolocalization studies, we purified several known ␤-defensins from preparations of purified dense granules, but detected none in mixed organelles depleted of dense granules (data not shown). Therefore, since ␤-defensins and bactenecins are stored together in dense granules, it is likely that they are discharged simultaneously during neutrophil activation.
Although co-packaged in the dense granules, cathelicidins, but not ␤-defensins, are stored as inactive propeptides (49). Upon neutrophil stimulation, the cathelicidins Bac5 and Bac7 are cleaved from their respective propeptides and released extracellularly (51). Conversely, data presented here indicate that ␤-defensins exist as fully processed peptides in bovine neutrophils since anti-BNBD-12 antibody detected only mature BNBD-12 in cell lysates (Fig. 10). This result was obtained irrespective of the manner in which the cells were extracted, i.e. with SDS versus 10% acetic acid, with or without DFP pretreatment. This finding is consistent with the 10:1 abundance ratio of BNBD-12 to BNBD-13 in similarly extracted neutrophils (12). Western blot analyses also indicate that the transformation of the primary translation product to mature ␤-defensin occurs rapidly in the bone marrow, since we detected no bands larger than that of the mature peptide in lysates rich in neutrophil progenitors (Fig. 10, B and C). This suggests that the ␤-defensin precursor has a very short halflife, a feature that differentiates ␤-defensin propeptide processing from Bac5 and Bac7, which reside within the dense granules exclusively as propeptides (49).
The demonstration that members of the cathelicidin and ␤-defensin families are packaged in the same granule raises the possibility that there may be important synergistic or even regulatory interactions that may occur as these antimicrobial molecules are discharged from this novel organelle. Activation of neutrophils with PMA (Fig. 11) demonstrated that BNBD-12 is exocytosed under conditions that induce secretion of Bac5 and Bac7. In contrast, Ganz (66) reported that human neutrophils secrete very little ␣-defensin under similar conditions, indicating that ␣and ␤-defensins are mobilized differently from their respective granules. Although we observed that the fraction of BNBD-12 that was recovered from supernatants was less than the 50% secretion of Bac5 and Bac7 previously reported (51), this difference may be ascribed to differences in binding of the respective peptides to the neutrophil surface after exocytosis, or to subtle differences in experimental conditions. Nevertheless, the co-mobilization of these two peptide families suggests that studies on their potential interactions, both within the phagosome and extracellularly, are merited.