Identification of a Region in the Vitamin D-binding Protein that Mediates Its C5a Chemotactic Cofactor Function*

The vitamin D-binding protein (DBP), also known as group-specific component or Gc-globulin, is a multifunctional plasma protein that can significantly enhance the leukocyte chemotactic activity to C5a and C5a des-Arg. DBP is a member of the albumin gene family and has a triple domain modular structure with extensive disulfide bonding that is characteristic of this protein family. The goal of this study was to identify a region in DBP that mediates the chemotactic cofactor function for C5a. Full-length and truncated versions of DBP (Gc-2 allele) were expressed in Escherichia coli using a glutathione S-transferase fusion protein expression system. The structure of the expressed proteins was confirmed by SDS-PAGE and immunoblotting, whereas protein function was verified by quantitating the binding of [3H]vitamin D. Dibutyryl cAMP-differentiated HL-60 cells were utilized to test purified natural DBP and recombinant expressed DBP (reDBP) for their ability to enhance chemotaxis and intracellular Ca2+ flux to C5a. Natural and full-length reDBP (458 amino acid residues) as well as truncated reDBPs that contained the N-terminal domain I (domains I and II, residues 1–378; domain I, residues 1–191) significantly enhanced both cell movement and intracellular Ca2+ concentrations in response to C5a. Progressive truncation of DBP domain I localized the chemotactic enhancing region between residues 126–175. Overlapping peptides corresponding to this region were synthesized, and results indicate that a 20-amino-acid sequence (residues 130–149, 5′-EAFRKDPKEYANQFMWEYST-3′) in domain I of DBP is essential for its C5a chemotactic cofactor function.

Vitamin D-binding protein (DBP) 1 is a multifunctional and highly polymorphic plasma protein synthesized primarily in the liver (1,2). DBP is expressed as a single polypeptide chain with a molecular mass of ϳ56 kDa and circulates in plasma at 6 -7 M (1, 2). Due to its extensive polymorphisms, DBP initially was named the group-specific component of serum, later shortened to Gc-globulin (3). DBP is a member of the albumin (ALB), ␣-fetoprotein (AFP), and ␣-albumin/afamin (AFM) gene family and thus has the characteristic multiple disulfidebonded, triple domain modular structure (1). Besides functioning as a circulating vitamin D transport protein, it has been demonstrated that plasma DBP effectively scavenges G-actin released at sites of necrotic cell death and prevents polymerization of actin in the circulation (1,2). Distinct binding regions within the 458-amino-acid sequence of DBP have been identified: a vitamin D sterol binding segment in the N-terminal domain (amino acids 35-49) and a G-actin binding region in the C-terminal domain (amino acids 373-403) (4,5). More recent work on the crystal structure of DBP (bound to either vitamin D 3 or actin) has confirmed the vitamin D sterol binding site but has demonstrated that actin interacts with distinct amino acid sequences in all three DBP domains (6 -9).
Complement C5a is a 74-amino-acid peptide generated by limited proteolytic cleavage of C5 during complement activation (10). C5a is a very potent chemotactic factor for all leukocytes as well as several other cell types, and the peptide has several other proinflammatory functions as well (10). C5a exerts these activities by binding to its high affinity receptor (C5aR or CD88) on the plasma membrane of target cells (11). Several groups have demonstrated that purified DBP can significantly enhance the neutrophil chemotactic activity (i.e. cochemotatic activity) of C5a and its stable breakdown product C5a des-Arg (12)(13)(14)(15)(16)(17). In addition to neutrophils, DBP can also augment the C5a chemotactic activity for monocytes and fibroblasts (18,19). The chemotactic enhancing properties of DBP appear to be restricted to C5a/C5a des-Arg since this protein cannot enhance the chemotactic activity of formylated peptides, IL-8, leukotriene B 4 , or platelet-activating factor (12,13,16,18). The mechanisms by which DBP acts as a chemotactic cofactor for C5a are not known. The goal of this study was to determine whether a cochemotactic region could be located within DBP. The results demonstrate for the first time that DBP enhances both C5a-mediated chemotaxis and Ca 2ϩ influx in differentiated HL-60 cells. In addition, a 20-amino-acid sequence within the N-terminal domain I of DBP (residues 130 -149) was found to possess the C5a chemotactic cofactor activity.
Cells and Cell Culture-Human promyelocytic cell line HL-60 was obtained from ATCC (Rockville, MD) and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum. Differentiation of HL-60 cells was initiated with 250 M Bt 2 cAMP for 48 h.
Construction of Full-length and Truncated Recombinant Expressed DBP (reDBP)-To construct GST fusion proteins, DNA sequences corresponding to the indicated regions of full-length human DBP or DBP domains were amplified by PCR, designed to generate products with 5Ј BamHI and the 3Ј XhoI restriction site. The Escherichia coli-expressed proteins were designated as follows: full-length reDBP (residues 1-458), DBP domains I and II (residues 1-378), DBP domains II and III  (residues 192-458), DBP domain I (residues 1-191), DBP domain II  (residues 192-378), DBP domain III (residues 379 -458) were cloned into the corresponding site using pGEX-4T-2 expression plasmid as GST fusion partners in E. coli (sequences of all primers available from the authors upon request). E. coli strain XL-1 Blue (Stratagene, CA) served as host for DNA manipulation, and the E. coli strain BL-21 served the host for protein expression. Truncated versions of domain I (residues 1-175, 1-150, 1-125, and 1-112) also were expressed in E. coli as described above. Each construct was confirmed by DNA sequencing.
Expression and Purification of Full-length and Truncated reDBP-DBP was expressed in E. coli according to the protocol described by Swamy et al. (21). BL21 cells carrying pGEX-4T-2 plasmids expressed fusion protein GST linked to the DBP constructs. E. coli were grown at room temperature (22-24°C) in Luria broth (LB) containing 100 g/ml ampicillin and 50 g/ml chloramphenicol until absorbance at 600 nm (A 600 ) was 0.4 -0.6. The expression of fusion proteins was induced by the addition of 0.5-1.0 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 -8 h. Cells were collected by centrifugation at 5000 ϫ g. The E. coli pellets (from 1 liter) were resuspended in 40 ml of Tris-buffered saline (50 mM Tris-HCl, pH 8.3, 150 mM NaCl) containing 0.2 mg/ml lysozyme, 0.1% Triton X-100, and 2 ml of a protease inhibitor mixture. The cells were disrupted by sonication, and insoluble material was removed by centrifugation at 12,000 ϫ g for 20 min. The detergent-soluble cell lysate containing the expressed fusion protein was mixed with 1 ml of a 50% slurry of glutathione-agarose, preequilibrated in Tris-buffered saline. After several washes with Tris-buffered saline, the bound GST⅐DBP was eluted with 20 mM oxidized glutathione in 100 mM Tris buffer, pH 8.3. Fusion protein was cleaved with thrombin and dialyzed against PBS to remove oxidized glutathione, and GST was separated from reDBP using a glutathione-Sepharose affinity column. Sequencing fidelities of the reDBPs were confirmed by N-terminal sequencing.
Vitamin D 3 Binding Assay-Competitive binding assays of natural DBP full-length and truncated reDBP with 1% Triton X-100) were incubated at 4°C for 20 h followed by treatment with 100 l of 2.5% ice-cold dextran-coated charcoal and centrifugation at 5000 ϫ g at 4°C. Clear supernatants from the centrifuged samples were mixed with the scintillation mixture and counted for radioactivity. Each sample was assayed in triplicate.
Ligand-induced Calcium Mobilization-Calcium mobilization studies were performed using the Fluo-3 AM probe as described previously (22). Differentiated HL-60 cells (1 ϫ 10 7 cells/ml) were resuspended in HBSS, 1% BSA containing 2 M Fluo-3 AM (Molecular Probes, Eugene, OR) and incubated at 37°C for 40 min. Cells incubated without the dye were used as a control to measure autofluorescence (F min ). Following the dye uptake, cells were washed twice and then suspended at 5 ϫ 10 6 cells/ml in HBSS containing 1% BSA. Half of the cell suspension was treated with 50 nM DBP for 30 min at room temperature, and the other half served as the untreated control. For each measurement, 400 l of cell suspension (with or without DBP) was added to a cuvette and stimulated with various concentrations of purified C5a at room temperature (22-24°C). Fluorescence was then measured immediately using a PerkinElmer Life Sciences LS-5 fluorometer at 505-nm excitation, 526-nm emission for Fluo-3 AM. F max was measured by treating labeled cells with 60 M digitonin. Intracellular free calcium concentrations were calculated using the following formula: where K d ϭ 325 nM for Fluo-3 according to the manufacturer (Molecular Probes).
Chemotaxis Assay-Chemotaxis was performed as described previously (23). In brief, 35 l of C5a (0.01-1 nM) in the chemotaxis buffer (HBSS with 1% BSA, 10 mM HEPES) was placed in duplicate in the lower chamber of a 48-well chamber (Neuroprobe, Gaithersburg, MD). The lower compartments were covered with a 5-m pore cellulose nitrate filter, and then 50 l of the cell suspension (2.5 ϫ 10 5 /well) was pipetted into the upper compartments. The chemotactic chamber was incubated for 1 h at 37°in a humidified incubator with 5% CO 2 . Following incubation, the filter was fixed, stained, and mounted on a microscope slide. Cell movement was measured as the distance in microns that the leading front of the cells had migrated into the filter (24).
Data Analysis and Statistics-Results of several experiments were analyzed for significant differences among group means using analysis of variance followed by Newman-Keul's multiple comparisons post-test utilizing the statistical software program InStat (GraphPad Software, San Diego, CA).

DBP Enhances C5a-induced Chemotaxis and Ca 2ϩ Influx in
Differentiated HL-60 Cells-Previous published work in our laboratory has demonstrated that neutrophils co-incubated with DBP display significant increased movement to suboptimal concentrations (10 -100 pM) of C5a, i.e. the cochemotactic effect (12,23,25). However, neutrophils are short-lived, terminally differentiated cells and cannot be genetically manipulated. A cell culture model for neutrophils is the promyelocytic cell line HL-60 that can be induced to differentiate into neutrophil-like cells using agents such as Me 2 SO or Bt 2 cAMP. Differentiation of HL-60 cells for 48 h using 250 M Bt 2 cAMP induced expression of the C5a receptor and both permitted chemotaxis to C5a alone and significantly enhanced movement to C5a in the presence of purified natural DBP (Fig. 1A). These results are consistent with our previous reports using neutrophils and indicate that differentiated HL-60 cells will serve as a good cell culture model to investigate the cochemotactic function of DBP.
Intracellular calcium mobilization is a rapid and well characterized event in response to C5a binding to its receptor. Fig.  1B shows that cells pretreated with DBP for 30 min display a significantly enhanced intracellular calcium influx in response to C5a. In contrast, untreated cells that had DBP and C5a added simultaneously did not show augmented calcium influx (data not shown), supporting previous studies showing that DBP needs to bind to the cell surface for at least 15 min before enhanced chemotaxis is observed (25). Nevertheless, the results presented in Fig. 1 show that DBP can enhance C5ainduced chemotaxis and calcium mobilization in differentiated HL-60 cells and indicate that this cell line will serve as a good cell culture model to investigate the cochemotactic function of DBP.
Analysis of E. coli-expressed DBP by SDS-PAGE and Western Blotting-The commercially available (Invitrogen) 1374-nucleotide DBP cDNA, Gc2 allele, was expressed in E. coli BL-21 in the plasmid vector pGEX-4T-2 fused to GST. Upon isopropyl-1-thio-␤-D-galactopyranoside induction, the GST⅐DBP fusion protein was expressed, as judged by SDS-PAGE ( Fig. 2A). SDS-PAGE of the elution protein revealed an ϳ80-kDa band corresponding to the molecular mass of full-length DBP fused to GST. The bands at ϳ65, 55, and 45 kDa correspond to the molecular mass of GST with truncated forms of DBP. All lanes have a small amount of a 30-kDa band. This probably represents a degradation product of GST or the unfinished product of the protein synthesis. Thrombin cleavage of the purified fusion protein resulted in separation of GST from DBP. The SDS-PAGE and immunoblotting after cleavage and separation of GST revealed a single protein band of 56 kDa for full-length DBP, 40 kDa for domains I and II, 30 kDa for domains II and III, and 20 kDa for domain I or domain II (Fig. 2B).
Functional Characterization of reDBP-The capacity of reDBP to bind 25(OH)-vitamin D 3 was measured to determine whether the expressed proteins could functionally bind a physiological ligand. Competitive binding assays of reDBP with a fixed amount of [ 3 H]25(OH)-D 3 and various amounts of unlabeled 25(OH)-D 3 demonstrate that all reDBP could displace the radiolabel in a similar dose-dependent manner (Fig. 3), indicating that the vitamin D sterol binding site of the reDBPs is similar to the purified, natural DBP. In addition, full-length reDBP-bound G-actin in a 1:1 molar complex as detected by non-denaturing PAGE (data not shown). The ability of reDBPs to enhance C5a-mediated chemotaxis and Ca 2ϩ flux in differentiated HL-60 cells was determined next. Fig. 4 demonstrates that reDBPs containing the N-terminal domain I have the capacity to enhance chemotaxis (Fig. 4A) and intracellular Ca 2ϩ flux (Fig. 4B) in response to a C5a stimulus. Cells treated with DBP alone showed no response (data not shown). In addition, undifferentiated HL-60 cells, which express very little C5a receptor, did not react to C5a or C5a plus DBP (data not shown).
Previous results clearly demonstrate that the C5a cochemotactic function of DBP resides in the N-terminal domain. Therefore, to identify the cochemotactic sequence within this region, a series of truncated versions of domain I were generated. Initially, constructs containing either the N-terminal activity. Fig. 5 shows the analysis of domain I truncations by SDS-PAGE (Fig. 5A) and immunoblotting (Fig. 5B). Fig. 6 demonstrates that domain I truncations 1-112 and 1-125 lack enhancing activity, whereas truncations 1-150 and 1-175 possess almost the same level of activity as full-length natural DBP for both chemotaxis (Fig. 6A) and Ca 2ϩ flux (Fig. 6B).
The results from Fig. 6 indicate that a region within domain I of DBP, from amino acids 126 to 175, confers the C5a cochemotactic function. Two overlapping peptides within this region of DBP were next synthesized to determine more precisely a cochemotactic sequence. Peptide 1 (residues 130 -152, EAFRK-DPKEYANQFMWEYSTNYG) and peptide 2 (residues 150 -172, NYGQAPLSLLVSYTKSYLSMVGS), by themselves or mixed together, could not enhance C5a-mediated chemotaxis (data not shown). Consequently, an alternative approach of using these peptides to block the cochemotactic function of full-length natural DBP by pretreating cells with each peptide was employed. The results from Fig. 7 clearly show that peptide 1-(130 -152), but not peptide 2- (150 -172), could completely block the cochemotactic function of purified full-length natural DBP (C5a ϩ DBP). In addition, peptide 1-(130 -152) also could eliminate the HL-60 cell chemotactic response to complementactivated serum, indicating that this peptide can function to block a potent chemotactic signal in a diverse protein mixture (Fig. 7). DISCUSSION This study describes several novel findings. It is the first report of differentiated HL-60 cells displaying increased C5amediated chemotaxis to DBP, the first report of a DBP-mediated enhancement of C5a-induced Ca 2ϩ influx in any cell type, and the first report to identify a sequence in DBP that mediates the C5a chemotactic cofactor function. Differentiated HL-60 cells are the second myeloid cell line to show enhanced C5a chemotaxis to DBP. Recently, we have demonstrated that undifferentiated U937 cells transfected with the C5a receptor can also be utilized as a cell culture model to investigate the cochemotactic function of DBP (20). Both cell lines will be useful to further elucidate the mechanisms by which DBP augments chemotaxis to C5a. In addition, the demonstration of a DBPmediated increase in calcium flux to C5a is particularly important because it will permit dissection of intracellular signaling pathways triggered by DBP bound to the cell surface. It is interesting to note that HL-60 cells needed to be pretreated with purified DBP for at least 15 min prior to C5a addition to detect an increased calcium signal (Fig. 1B), perhaps indicating the formation of a DBP signaling complex on the plasma membrane as suggested previously (20). However, the major finding in this report is the identification of a 20-amino-acid sequence in the N-terminal domain of DBP that blocks C5a cochemotaxis to full-length natural DBP.
DBP contains 14 disulfide bonds and, therefore, presents a challenge for expressing the protein in a functional form. Swamy et al. (21) previously reported expression of functional recombinant DBP in E. coli using GST as a fusion partner. Using their approach, we generated a series of functional truncated proteins that localized the C5a cochemotactic function to a 50-amino-acid region (residues 126 -175) in the C-terminal portion of domain I. Two overlapping peptides covering this region were synthesized and clearly demonstrated that the cochemotactic function was confined to the 20-amino-acid sequence 5Ј-EAFRKDPKEYANQFMWEYST-3Ј (residues 130 -149). This sequence is identical among the three major allelic forms of human DBP (Gc-1F, Gc-1S, Gc-2) and correlates with the fact that there is no difference in the C5a cochemotactic function between these DBP isoforms (16). BLAST search of this sequence produced no other match, besides DBP, in any eukaryote, indicating that this region is unique to DBP and that it probably does not function by mimicking a similar sequence in a signaling molecule. In addition, alignment of this human peptide with the corresponding sequences in rat (26), mouse (27), and rabbit DBP (28), the only other mammalian sequences currently described, show a very high degree of amino acid similarity; 78% of the residues are identical, and the remainder are conservative or homologous substitutions. Thus, it is reasonable to speculate that this region in domain I of DBP functions as the cochemotactic sequence in all mammals.
Several recent reports have described the crystal structure of DBP, either unligated or bound to G-actin or vitamin D (6 -9). Analysis of the structure has revealed that the protein is comprised of a series of ␣-helices, much like albumin (29), but in contrast to albumin, DBP folds into a hook-like structure that serves as the G-actin binding site (30). The cochemotactic sequence in DBP (residues 130 -149) is located partly in ␣-helix number 7 (residues 125-134) but mostly in ␣-helix number 8 (residues 136 -150) in domain I (6 -9). Three-dimensional analysis of DBP crystal structure using the NIH NCBI software program Cn3D (version 4.1) indicates that this region is not blocked when DBP binds G-actin or vitamin D sterols and is accessible to interact with cells. This structural analysis correlates well with our recent functional studies that demonstrated that ligation of DBP with G-actin, 25-OH vitamin D 3 , or both did not alter the C5a cochemotactic activity of DBP. 2 Therefore, this cochemotactic peptide is a distinct functional sequence and constitutes the forth such region (the others are G-actin, vitamin D, and the polysaccharide structure of DBP-MAF (macrophage-activating factor) to be defined in the plieotropic DBP.
The binding of DBP to cells is required for the protein to mediate its numerous functions: a chemotactic cofactor for C5a, a macrophage or osteoclast-activating factor, clearance of DBPactin complexes by the liver, and delivery of vitamin D sterols and free fatty acids to cells (1,2). The cell surface DBP binding site is only partially characterized, but this essential link in DBP physiology is poorly understood. DBP appears to bind with relatively low affinity to multiple cell surface ligands including chondroitin sulfate proteoglycans (31), low density lipoprotein receptor family members megalin and cubulin (32,33), and possibly CD36 (34). Recently, we have shown that the platelet-derived adhesive glycoprotein thrombospondin-1 (a known ligand for CD36) is required for the full C5a cochemotactic activity of DBP and may be involved in the formation or regulation of the putative DBP binding site complex (20). The results presented in this report do not answer the question of whether there is a distinct cell binding and cochemotactic re-FIG. 6. Effect of truncated reDBP of domain I on C5a-mediated responses in differentiated HL-60 cells. HL-60 cells were differentiated using 250 M bt 2 cAMP for 48 h. A, cells (6 ϫ 10 6 /ml) were resuspended in chemotaxis buffer allowed to respond to 100 pM C5a alone (Control) or C5a plus 50 nM each indicated DBP for 60 min at 37°C. Numbers represent net cell movement in 60 min, mean Ϯ S.E. of 3-5 experiments. B, cells loaded with Fluo-3 AM dye were resuspended in HBSS ϩ 1% BSA at 5 ϫ 10 6 /ml with or without 50 nM indicated DBP and then incubated at room temperature for 30 min. The increase in intracellular calcium after stimulation with 100 pM C5a was then measured. Data are expressed as the maximal nM change in the intracellular Ca 2ϩ level. Results are the mean Ϯ S.E. of 3 separate experiments. Asterisks denote that the sample is significantly greater (p Ͻ 0.01) than control value. gion in DBP or whether the sequence identified (residues 130 -149) performs both functions. DBP has been shown to bind to chondroitin sulfate proteoglycans, and this protein has several glycosaminoglycan consensus binding sequences (K/R-rich regions) in domains II and III. In particular, an 11-amino-acid sequence (403-413) in domain III (5Ј-KKKLAERLKAK-3Ј) is a very basic region that may be a putative glycosaminoglycan binding site. However, the cell binding site on DBP remains to be identified. The results also do not preclude the possibility that there are other sequences in DBP that may be involved in its cochemotactic function. The cochemotactic sequence 5Ј-EAFRKDPKEYANQFMWEYST-3Ј (residues 130 -149) can effectively block the capacity of full-length natural DBP to enhance chemotaxis to C5a, but the sequence alone cannot augment chemotaxis, possibly suggesting that other regions of DBP may be involved in this function. Nevertheless, the results reported herein provide strong evidence that amino acids 130 -149 play an essential role in the C5a chemotactic cofactor function of DBP and, furthermore, could signify that compounds derived from this sequence may have therapeutic potential to limit C5a-mediated tissue injury.