S100A12 Is Expressed Exclusively by Granulocytes and Acts Independently from MRP8 and MRP14*

Changes in cytosolic calcium concentrations regulate a wide variety of cellular processes, and calcium-binding proteins are the key molecules in signal transduction, differentiation, and cell cycle control. S100A12, a recently described member of the S100 protein family, has been shown to be coexpressed in granulocytes and monocytes together with two other S100 proteins, MRP8 (S100A8) and MRP14 (S100A9), and a functional relationship between these three S100 proteins has been suggested. Using Western blotting, calcium overlays, intracellular flow cytometry, and cytospin preparations, we demonstrate that S100A12 expression in leukocytes is specifically restricted to granulocytes and that S100A12 represents one of the major calcium-binding proteins in these cells. S100A12, MRP8, and MRP14 translocate simultaneously from the cytosol to cytoskeletal and membrane structures in a calcium-dependent manner. However, no evidence for direct protein-protein interactions of S100A12 with either MRP8 or MRP14 or the heterodimer was found by chemical cross-linking, density gradient centrifugation, mass spectrometric measurements, or yeast two hybrid detection. Thus, S100A12 acts individually during calcium-dependent signaling, independent of MRP8, MRP14, and the heterodimer MRP8/MRP14. This granulocyte-specific signal transduction pathway may offer attractive targets for therapeutic intervention with exaggerated granulocyte activity in pathological states.

Granulocytes and monocytes are major effector cells during inflammatory processes. Undue activation of these cells is a pathophysiological factor in many diseases, e.g. rheumatoid arthritis and chronic inflammatory bowel disease (1)(2)(3)(4). In monocytes and granulocytes, intracellular Ca 2ϩ regulates various acute response activities, such as respiratory burst, phagocytosis, degranulation, and release of degrading enzymes (5)(6)(7)(8). One molecular pathway of calcium signal transduction is calcium-binding proteins of the S100 multigene family, which comprises a group of small, acidic proteins with a tissue-and cell cycle-specific expression (9 -11). S100 proteins contain two calcium-binding sites per molecule (12). Two members of this family have been found in human granulocytes and monocytes, called macrophage migration inhibitory factor-related protein 8 (MRP8) 1 (S100A8) and MRP14 (S100A9), which represent up to 40% of the calcium binding capacity in monocytes (13,14). Both proteins form noncovalently associated complexes in a calciumdependent manner (15,16). These complexes translocate from cytoplasm to membranes, as well as to intermediate filaments, after elevation of intracellular Ca 2ϩ levels, and this correlates with the induction of inflammatory actions of granulocytes and monocytes. Thus, MRP8 and MRP14 seem to be important regulators of cytoskeletal/membrane interactions during phagocyte activation (17)(18)(19).
The calcium-induced change in the complex pattern of these two proteins has been considered to be important for their biological function (15,19). For example, only the complex of MRP8 and MRP14 has been described to inhibit casein kinases I and II, two enzymes involved in regulation of gene expression via mediation of RNA polymerase activity (20). Recently, a third S100 protein (S100A12) was identified in human and porcine granulocytes (21)(22)(23). The primary structure of human S100A12 consists of 91 amino acids with a molecular mass of 10,444 Da confirmed by ESI-MS (21). S100A12 has been also found in small amounts in lysates of monocytes (24) and lymphocytes (23), but the amounts of S100A12 detected in these cells were in the range of contaminations by granulocytes in these mononuclear preparations. Synonyms of S100A12 are calgranulin C, calgranulin-related protein (22), and calcium-binding protein in amniotic fluid-1 (25,26). Human S100A12 shows highest homologies to MRP14 (46%) and to MRP8 (40%) (21). Upon calcium binding, S100A12 from porcine granulocytes undergoes a gross conformational change, supporting the idea that this protein is involved in Ca 2ϩ -dependent signal transduction events (23).
The parallel expression of MRP8, MRP14, and S100A12 in myeloid cells (24), the copurification of these three S100 proteins (27), and the general finding of the regulatory role of protein-protein interactions on S100-protein functions (28,29) has led to the assumption that these three calcium-binding proteins are involved in a common signal transduction pathway. However, there are no published data available analyzing possible interactions of S100A12 with MRP8 or MRP14. We now demonstrate on the single cell level that expression of S100A12 in leukocytes is specifically restricted to granulocytes. Furthermore, we found no evidence for interaction of S100A12 with MRP8, MRP14, and/or their heterocomplexes, either in the presence or the absence of calcium, as shown by density gradient centrifugation, chemical cross-linking, UV-MALDI-MS, and a yeast two hybrid system. Therefore, we conclude that S100A12 carries out its function exclusively in granulocytes without physically interacting with MRP8 or MRP14. 45 CaCl 2 (1 mCi/ml) was obtained from Amersham Pharmacia Biotech. Electrophoresis reagents were purchased from Bio-Rad. Polyvinylidene difluoride and nitrocellulose membranes (pore size, 0.2 m) were from Schleicher and Schuell. Bis-(sulfosuccinimidyl)suberate (BS 3 ) was from Pierce. Molecular mass markers were from Amersham Pharmacia Biotech. All other solvents and reagents were purchased from Sigma and were at least of analytical grade.
Antibodies-MRP8 and MRP14 were detected by non-cross-reactive affinity-purified rabbit antisera (a-MRP8 and a-MRP14). The monospecifity was evaluated by immunoreactivity against recombinant and native proteins, as well as against transfected cell lines, as described previously (13,17). The polyclonal affinity-purified rabbit antisera against S100A12 (a-S100A12) were made by immunization of rabbits with purified S100A12. The specificity of a-S100A12 was ascertained by immunoreactivity against purified material and cell lysates on Western blots. a-S100A12 showed a slightly cross-reactivity against MRP8. For detection of MRP8/MRP14 complexes, a monoclonal antibody 27E10 (19) was employed recognizing only the MRP8/MRP14 heterodimer but not single monomers of MRP8 and MRP14. As specific markers for granulocytes, monocytes or lymphocytes monoclonal antibodies directed against CD3, CD14, CD15, and CD16 (Dianova, Hamburg, Germany) were used. For controls, polyclonal rabbit IgG (Amersham Pharmacia Biotech) and monoclonal mouse IgG 1 (Dianova) of irrelevant specificity were employed throughout all investigations.
Flow Cytometry and Immunocytochemistry-Monocytes, granulocytes, and lymphocytes were separately analyzed by flow cytometry employing a FACScan (Becton Dickinson, Heidelberg, Germany) equipped with Lysis II software. Peripheral blood leukocytes were gated by the side scatter and forward scatter filters, and the purity of each cell group was controlled by parallel stainings for CD3, CD14, CD15, and CD16. After isolation, cells were stained using an indirect immunofluorescence method as described earlier (19). For intracellular staining, cells were fixed and permeabilized using a cell permeabilization kit according to the instructions of the company (Biozol, Eching, Germany).
In addition, cytospin preparations of monocytes, granulocytes, and HL-60 cells were analyzed for expression of S100A12, MRP8, and MRP14 by indirect immunocytochemistry using antisera against S100A12, MRP8, and MRP14 followed by peroxidase-conjugated second stage goat anti-rabbit antibodies (Dianova) for detection.
Subcellular Fractionation-Granulocytes were fractionated in subcellular compartments as described previously (17,37). Briefly, cells were suspended in 20 mM HEPES, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, pH 7.4, containing either 0.1 mM Ca 2ϩ or 1 mM EGTA. After sonification, lysates were subjected to differential centrifugation maintaining the calcium and EGTA concentrations and a temperature of 4°C throughout the whole procedure. Cytoskeletal fractions were obtained by centrifugation at 1000 ϫ g for 15 min (Kontron, model T1065). The pellet (cytoskeleton) was washed three times in 20 mM HEPES, 1% Nonidet P-40 including sonification for three cycles. The 1000 ϫ g supernatant was centrifuged again at 10,000 ϫ g in order to discard cellular debris and aggregated membrane sheets. The 10,000 ϫ g supernatant was centrifuged at 100,000 ϫ g for 90 min. The pellet (membrane fraction) was washed twice in 5 mM HEPES and centrifuged at 100,000 ϫ g for 60 min to release cytosolic proteins trapped in membrane vesicles by osmotic shocks. The 100,000 ϫ g supernatant was considered as the cytosolic fraction. The purity of subcellular fractions was established by measuring lactate dehydrogenase activity in the different fractions during the final washing step (38). The lactate dehydrogenase activity in the cytoskeletal and membrane fraction was less than 0.1% of that in the cytosolic fraction. Western blots of the different fractions were stained with a-S100A12, a-MRP8, and a-MRP14 and peroxidase-conjugated second stage goat anti-rabbit antibody (17).
Ca 2ϩ Overlay-Ca 2ϩ binding of proteins was detected by the method of Murayama et al. (39). Proteins of MonoQ fractions were separated by 15% SDS-PAGE and electroblotted onto nitrocellulose membranes using 50 mM Tris borate buffer adjusted to pH 8.3. After blotting, the membranes were incubated three times in washing buffer (10 mM imidazole, 60 mM KCl, 5 mM MgCl 2 , pH 6.8) for 10 min. Subsequently, membranes were incubated with washing buffer containing 0.1 mCi 45 CaCl 2 /100 ml for 15 min. Membranes were washed three times for 3 min in doubly distilled water to remove unbound calcium, dried, and analyzed by autoradiography. Spots were evaluated densitometrically (Personal Densitometer, Molecular Dynamics, Sunnyvale, CA). Ca 2ϩ binding was corrected to the different concentrations of S100A12, MRP8, and MRP14 in granulocytes.
Chemical Cross-linking-Chemical cross-linking of S100A12, MRP8, and MRP14 was carried out using BS 3 according to the method of Staros and Kakkad (40). Protein samples from MonoQ fractions containing either MRP8 and MRP14 or S100A12, MRP8, and MRP14 were dialyzed against 20 mM HEPES, 150 mM NaCl, pH 7.3, and adjusted to a concentration of 0.5 mg of protein ml Ϫ1 and 2 mM BS 3 . The reaction was quenched after 30 min by addition of 1 M Tris to a final concentration of 20 mM. Samples cross-linked with BS 3 were incubated with 1% ␤-mercaptoethanol prior to separation by 15% SDS-PAGE.
Density Gradient Centrifugation-To analyze the influence of calcium binding on a complex formation of S100A12, MRP8, and MRP14, density gradient centrifugation was employed. A glycerol gradient ranging from 15 to 30% in 50 mM HEPES, 140 mM NaCl, pH 7.3, was used in the presence of either 0.1 mM Ca 2ϩ or 1 mM EGTA.
Gradients were loaded with samples containing S100A12, MRP8, and MRP14 and dialyzed against the buffer described above adjusted to 10% glycerol and 1% ␤-mercaptoethanol. Samples were centrifuged for 20 h at 20°C and 150,000 ϫ g. After centrifugation, the gradient was fractionated into 20 aliquots. Proteins in each fraction were precipitated with trichloroacetic acid and analyzed by 15% SDS-PAGE under reducing conditions.
UV-MALDI-MS-The principle of UV-MALDI-MS is described in the literature (41,42). In our investigations, a home-built time-of-flight system equipped with both a linear and a reflectron port was used. A frequency-tripled Nd-YAG-laser (JK-Laser Ltd., Rugby, United Kingdom) at 355 nm with a pulse duration of 15 ns served for the desorption/ ionization of molecules. A conventional secondary electron multiplier (EMI 9643, Electron Tube Ltd., Middlesex, UK) plus conversion dynode (R2362, Hamamatsu Photonics, Hamamatsu, Japan) in front of it was used as detector system. A transient recorder (LeCroy 9354, LeCroy, Chestnut Ridge, NY) and a personal computer were used for data accumulation and data processing. Only a limited number of matrix/ laser wavelengths combinations preserved complexes throughout the whole procedure, and most importantly, only spectra recorded from the top layer of the samples showed pronounced signals of the complexes under such conditions. This latter observation is known as the "first shot phenomenon" (43,44). The matrix employed was 2,6-dihydroxyacetophenone (Aldrich, Steinheim, Germany), which was prepared as a saturated solution in a mixture of acetonitrile/water (1/3; v/v). The actual UV-MALDI-samples were prepared directly on the stainless steel MALDI target by mixing 0.5 l of the protein solution (1 mg/ml) with 0.5 l of either doubly distilled water or a 10 mM CaCl 2 solution and 2.5 l of matrix solution. All samples were allowed to dry in a stream of cold air.
Yeast Two-hybrid System-A yeast two-hybrid system according to Fields and Song (45) was used for in vivo interaction studies of either MRP8 or MRP14 with S100A12, as well as studies of the homodimer-ization propensity of S100A12 as described previously (46). Briefly, polymerase chain reaction-amplified cDNA fragments encoding the entire human S100A12 were fused to either the GAL4 DNA-binding domain of the yeast plasmid pAS2-1 or the GAL4 activation domain of pACT2, respectively, according to the instructions of the company (CLONTECH Laboratories Inc., Palo Alto, CA). pAS2-1 encodes the N-terminal GAL4 DNA-binding domain (amino acids 1-147). pACT2 encodes the C-terminal acidic GAL4 activation domain (amino acids 768 -881). For interaction studies with MRP8 or MRP14, the cDNA of S100A12 was fused in frame to the GAL4 activation domain, and the cDNAs of either MRP8 or MRP14, respectively, were fused to the GAL4 DNA-binding domain. Furthermore, the cDNA of S100A12 was cloned in frame with the DNA-binding domain of pAS2-1 in order to investigate homodimerization formation of S100A12. The cDNA fusions were performed by ligation of cDNA of the desired S100 protein that had been amplified with polymerase chain reaction using oligonucleotide linkers to allow the in frame insertion into the yeast expression vectors. All constructs were sequenced in order to ensure that the intended correct insertion was successful. Transformations of pairwise vector combinations were done into the yeast strain Y190 of Saccharomyces cerevisiae (47).
The phenotypical detection of protein/protein interactions was investigated by using a sensitive and rapid X-gal filter lift assay for the detection of ␤-galactosidase activity followed by a photometric O-nitrophenyl ␤-D-galactopyranoside test for the quantification of ␤-galactosidase units (CLONTECH Laboratories Inc.).

RESULTS AND DISCUSSION
Calcium-dependent signaling is a major pathway leading to activation of granulocytes and monocytes during inflammatory situations (48,49). Analysis of such signal pathways is important, because uncontrolled activation of these cell populations is a major pathomechanism in many inflammatory diseases. Three calcium-binding proteins of the S100 family, MRP8, MRP14, and S100A12, have been characterized in myeloid cells; the relation of these proteins, however, is unclear (18,50). S100A12 Is Specifically Expressed by Granulocytes-To investigate the expression pattern of S100A12 in leukocytes, lysates of granulocytes, monocytes, and lymphocytes, as well as of HL-60 cells stimulated with Me 2 SO (differentiation to granulocytes), 1␣,25-dihydroxy-vitamin D 3 (monocytes), or phorbol 12-myristate 13-acetate (mature macrophages), were analyzed by Western blot technique. We found a strong expression of S100A12 in granulocytes (Fig. 1A), whereas we did not observe S100A12 expression in either HL-60 cell lysates or in lymphocyte lysates. Monocyte lysates exhibit a clear albeit weak immunoreactivity for S100A12. However, a low level immunoreactivity, as found by us and others (23, 24) may well be due to contamination of the mononuclear cell fractions with granulocytes. To address this possibility, we performed intracellular flow cytometry staining for S100A12 and for cell type-specific markers simultaneously. We detected S100A12 expression ex-FIG. 2. Calcium-dependent translocation of S100A12 in granulocytes. Cells were lysed in the presence of either 0.1 mM Ca 2ϩ or 1 mM EGTA and subsequently separated into cytosolic, membrane, and cytoskeletal fractions by differential centrifugation. Proteins of each different fraction were separated by gel electrophoresis, and S100A12 was detected by Western blot analysis. S100A12 was located solely in the cytoplasm of the cells after chelating of cellular calcium by 1 mM EGTA. In the presence of calcium, the majority of S100A12 was found in the membrane and cytoskeletal fractions.

FIG. 1. Expression pattern of S100A12 in granulocytes, monocytes, lymphocytes, and HL-60 cells. HL-60 cells were stimulated with
Me 2 SO (1% v/v), 1␣,25-dihydroxy-vitamin D 3 (100 nM), or phorbol 12-myristate 13-acetate (20 ng/ml) for 3 days. A, after cell lysis, lysates (20 g for granulocytes, 100 g for all others) were separated by 15% SDS-PAGE under reducing conditions, electroblotted, and subsequently analyzed by polyclonal antibodies against S100A12. Leukocytes: L, lymphocytes; M, monocytes; G, granulocytes. HL-60 cells: C, nonstimulated control cells; D, Me 2 SO; V, 1␣,25-dihydroxy-vitamin D 3 ; P, phorbol 12-myristate 13-acetate. In granulocytes, a strong immunoreactivity, corresponding to S100A12, can be detected, whereas other lysates show only weak (monocytes) or no staining. B, histograms obtained from intracellular flow cytometry of granulocytes, monocytes, and lymphocytes. In each histogram, the open profile shows leukocytes stained with a control antiserum of irrelevant specificity, whereas the shaded profile represents a-S100A12 reactivity. Only in granulocytes was a significant shift in fluorescence intensity observed, whereas no S100A12 expression was detected in monocytes or lymphocytes. C, cytospin preparations of granulocytes (G) and monocytes (M) stained with a-S100A12 by indirect immunoperoxidase technique. Only polymorphnuclear granulocytes reacted with a-S100A12, whereas monocytes were negative. clusively in granulocytes, but not in monocytes or in lymphocytes (Fig. 1B). Likewise, using cytospin preparations of monocytes and granulocytes, we found S100A12 expression exclusively in granulocytes (Fig. 1C). We found no S100A12 expression in cytospin preparations of lymphocytes or HL-60 cells (data not shown). Thus, our data demonstrate that expression of S100A12 is specifically restricted to granulocytes, and detection of S100A12 in lysates of monocytes or lymphocytes as described earlier is due to contaminations by granulocytes.
Calcium Induces Translocation of S100A12-As shown earlier, MRP8 and MRP14 translocate from the cytosol to membrane and cytoskeletal structures in a calcium-dependent manner (17,18). To determine the subcellular distribution of S100A12, purified granulocytes were lysed in the presence and absence of calcium and separated into cytosolic, cytoskeletal, and membranous fractions by differential centrifugation. In the absence of calcium, S100A12 was found predominantly in the cytosol, whereas addition of calcium induced a strong translocation to both membrane and cytoskeletal components (Fig.  2), closely resembling the distribution pattern of MRP8 and MRP14 (17,18,19). This finding is consistent with a report in which opsonized zymosan, a stimulus known to act via elevation of intracellular Ca 2ϩ levels, has been shown earlier to induce a translocation of S100A12, MRP8, and MRP14 from the cytosol to the submembranous cytoskeleton (24). S100A12 Is a Major Ca 2ϩ -binding Protein in Granulocytes-Calcium binding properties of S100A12, MRP8, and MRP14 analyzed by 45 Ca 2ϩ overlays are shown in Fig. 3A. The relative amount of calcium ions bound to individual proteins was calculated after densitometrically scanning autoradiographic bands and correcting for the individual abundance of S100A12, MRP8, and MRP14 (Fig. 3A). The level of calcium binding of MRP8 was set to 1, and calcium binding of S100A12 was approximately 18 times higher than that of MRP8 and comparable to MRP14. We conclude that S100A12 represents a major calcium-binding protein in granulocytes. This is particular interesting in the light of granulocyte-specific expression of S100A12. Calcium signaling is involved in receptor-mediated signal transduction pathways common to both types of phagocytes, monocytes and granulocytes. Some receptors are specifically expressed by either cell type and hence facilitate cell type-specific responses, and these include CC and CXC chemokine receptors (51)(52)(53)(54)(55). Other receptors, such as CCR1, are expressed by both cell types and elicit calcium signals, yet their activation can lead to different responses in each cell type (56). This strongly suggests differences in the downstream signaling cascades, and these may well be caused by specifically expressed calcium-binding proteins, such as S100A12. S100A12 Does Not Complex with MRP8 and MRP14 -MRP8 and MRP14 form noncovalently associated protein complexes in a calcium-dependent manner (15,16). The structural homologies of S100A12 to MRP8 and MRP14, the parallel expression in granulocytes, the copurification of all three proteins, the identical calcium-dependent translocation, and the common finding of complex formation in the S100 family suggest that S100A12 might be exhibit its intracellular functions via interaction with MRP8 and/or MRP14. In the first step, we employed density gradient centrifugation for the analysis of complex formation patterns (Fig. 3B). In EGTA-containing samples, S100A12, MRP8, and MRP14 were found in the same fractions of the glycerol gradient (18 Ϯ 2%, Fig. 3B, bottom  panel). In the presence of calcium, MRP8 and MRP14 shifted to fractions of significantly higher glycerol concentrations (24 Ϯ 3%), reflecting calcium-induced complex formation (Fig. 3B, top  panel). In contrast, after addition of calcium, only a minor shift and a clear separation from MRP8/MRP14-containing fractions was observed for S100A12 (20.3 Ϯ 3%), indicating no interaction of S100A12 with MRP8 and MRP14. The question whether this minor shift of S100A12 is due to homodimerization or conformational changes of monomers cannot be answered by this method. In addition, covalent cross-linking of fractions containing exclusively MRP8 and MRP14 with BS 3 led to a similar complex pattern as described earlier (Fig. 3C, lane 1) (15). In fractions containing additional S100A12, two more bands, at about 6 and 22 kDa, were found, probably representing the monomer and homodimer of S100A12 (Fig. 3C, lane 2). However, due to a slight cross-reactivity of our a-S100A12 with MRP8, Western blot analysis could not finally exclude the possibility of complex formation of S100A12 with MRP8 (data not shown).
Analysis of S100A12, MRP8, and MRP14 Interactions by UV-MALDI-MS-In a recent report we have shown, that MRP8 and MRP14 form noncovalently associated tetramers in the presence of calcium ions that are stable under UV-MALDI-MS conditions (16). We therefore analyzed formation of possible FIG. 3. Calcium binding properties and complex formation of S100A12. A, proteins were separated by 15% SDS-PAGE, blotted onto nitrocellulose membranes, and stained with Amido Black B10 (lane 1). Prior to protein staining, blots were incubated with 45 Ca 2ϩ , washed, dried, and autoradiographed (lane 2). Bands were quantified by densitometrical analysis, and calcium binding was related to the relative protein amounts of the three molecules (plot). Calcium binding of MRP8 was set equal to 1. B, fractions containing S100A12, MRP8, and MRP14 were loaded on a glycerol gradient in the presence of either 0.1 mM calcium or 1 mM EGTA and centrifuged, and their distribution in the gradient was subsequently analyzed by 15% SDS-PAGE under reducing conditions. In the presence of EGTA, S100A12, MRP8, and MRP14 showed an almost identical distribution centered in the low density fractions of the gradient. Addition of calcium induced a marked shift for MRP8 and MRP14 to higher glycerol densities, whereas for S100A12, only a minor shift was observed. C, complex formation pattern of S100A12, MRP8, and MRP14 after cross-linking with BS 3 . Fractions containing MRP8 and MRP14 exclusively or containing all three proteins, S100A12, MRP8, and MRP14, were cross-linked with BS 3 and subsequently separated using 15% SDS-PAGE under reducing conditions. Protein bands were detected by Coomassie Blue staining. Lane 1, MRP8 and MRP14 ϩ BS 3 ; lane 2, S100A12, MRP8, and MRP14ϩ BS 3 ; lane 3, S100A12, MRP8, and MRP14 without BS 3 .
homo-and heterocomplexes of fractions of isolations containing S100A12, MRP8, and MRP14 by UV-MALDI-MS, a method that allows definitive discrimination of different complex forms by exact molecular weight determination. Using UV-MALDI-MS, we detected only the monomeric masses of S100A12, MRP8, and the two known isoforms of MRP14 in the absence of calcium ( Fig. 4 and Table I). Peaks for S100A12 and MRP8 at 10,444 Ϯ 9 and 10,842 Ϯ 3 Da are in agreement with the expected masses derived from the amino acid sequences, indicating no posttranslational modifications ( Fig. 4A and Table I). The signal at 13,157 Ϯ 3 Da is in agreement with the expected mass of MRP14, which misses the N-terminal methionine and has a acetyl group at the remaining N terminus. MRP14*, an isoform resulting from a translation start at amino acid 5, is represented by a peak at 12,693 Ϯ 3 Da, with posttranslational modifications identical to those observed for MRP14 (15,57). The absence of intense signals in the mass range 21-26 kDa in the calcium-free preparation excludes artificial formation of heterodimers or homodimers of S100A12, MRP8, and MRP14 under these conditions, especially by disulfide bridge formation.
In the presence of calcium, however, we observed the formation of protein complexes; the molecular masses at 47.4, 47.8, and 48.3 kDa are in accordance with tetramers consisting of two molecules of MRP8 and two molecules of MRP14 and/or MRP14* as described recently (16). However, we did not detect S100A12-homodimers or any interaction of S100A12 with MRP8 or MRP14 (Fig. 4B).
A second group of peaks is found at a molecular mass of around 24 kDa. This signal may indicate doubly charged tetramers (T 2ϩ ) or singly charged heterodimers of MRP8 and MRP14/MRP14*. The triplet structure of this peak, however, is identical to that found in the singly charged tetramer (T ϩ ) at 48 kDa (for more details, see Ref. 16). The peak at m/z 16 kDa (T 3ϩ ) strongly supports this conclusion, because no other oligomer matches this mass. We therefore conclude that the observed calcium-induced changes in the S100A12, MRP8, and MRP14 preparation correspond solely to tetramers of MRP8 and MRP14/MRP14*, i.e. (MRP8/MRP14*) 2 , (MRP8/MRP14*) (MRP8/MRP14), and (MRP8/MRP14) 2 , whereas for S100A12, only the molecular mass for the monomer was found.
Comparative analysis of the monomeric signals in the presence and absence of calcium allows us to calculate the number of calcium ions stably bound (Table I). MRP8 monomers display a significantly weaker calcium binding than MRP14, MRP14*, and S100A12, and this is in agreement with results obtained by calcium overlays. S100A12 Forms Homodimers in Vivo-Weak protein-protein interactions might be destroyed and therefore not detected under UV-MALDI-MS-conditions. We therefore utilized a yeast two-hybrid system that allows for analyzing protein-protein interactions in a eukaryotic organism under in vivo conditions. Testing S100A12 against either MRP8 or MRP14, we found no interaction between these proteins, neither by means of the X-gal assay (Fig. 5) nor by the quantitative photometric Onitrophenyl ␤-D-galactopyranoside test (data not shown). Accordingly, screening a leukocyte cDNA library against MRP8 or MRP14, we identified the complementary protein, i.e. MRP8 or FIG. 5. Interactions of S100A12, MRP8, and MRP14 in the two hybrid system. The filter lift assay shows the combinations S100A12-S100A12, MRP8-MRP14, S100A12-MRP8, and S100A12-MRP14 coexpressed in the yeast two hybrid system. Yeast cells containing the plasmids were grown on nitrocellulose filter. Only in the case of strong interactions were blue yeast cells obtained, resulting from a enzymatic conversion of X-gal by ␤-galactosidase. Yeast cells without galactosidase activity remained colorless. No interactions were detected between MRP8 and S100A12 or between MRP14 and S100A12, whereas heterodimerization was found for MRP8 and MRP14, and a weak homodimerization was found for S100A12.  4. Mass spectrometric analysis of S100A12, MRP8, and MRP14 interactions. UV-MALDI-mass spectra of a fraction containing MRP8, MRP14, and S100A12 obtained under reducing conditions (1 mM dithiothreitol). The matrix used was 2,6-dihydroxyacetophenone. First shots of given spots are accumulated only. A, in the absence of calcium, only the monomeric species of MRP8, MRP14, MRP14*, and S100A12 were detected. B, calcium-induced formation of noncovalently associated tetramers (T ϩ ) composed of two molecules of MRP8 and two molecules of MRP14 and/or MRP14*. Neither homodimers of S100A12 nor interactions of S100A12 with MRP8 and/or MRP14 were found. A final calcium concentration of 1 mM was employed.
MRP14, but not S100A12 (data not shown). Analysis of S100A12-homodimerization revealed a ␤-galactosidase activity of approximately 14 Ϯ 5% (n ϭ 5) in relation to that found for the heterodimeric interaction of MRP8 with MRP14 (set as 100%), which indicates a weak but significant homodimer formation (Fig. 5). We therefore conclude that S100A12 does not interact with either MRP8 or MRP14. In the light of these findings, we conclude that the previously observed copurification of S100A12, MRP8, and MRP14 might be due to similar biochemical properties of these three molecules and most likely does not indicate physical interaction between MRP8/MRP14 and S100A12. Homodimer formation of S100A12 leads to the additional band in the cross-linking experiments and may also be at least partly responsible for the minor shift of S100A12 during density gradient centrifugation. The weak S100A12 homodimer interaction found in the two hybrid system was probably destroyed under UV-MALDI-MS conditions, and this is in contrast to the more stable MRP8/MRP14 complexes. Thus, only the combination of conventional biochemical methods, i.e. density centrifugation and cross-linking, with UV-MALDI-MS and the two hybrid system has allowed us to clearly define the interaction of S100A12, MRP8 and MRP14. The accuracy of mass determination by UV-MALDI-MS is useful in the identification of isoform patterns and the degree of calcium ion binding to single monomers and noncovalently associated complexes; however, the analysis occurs under conditions harsh enough to destroy weaker interactions. The two hybrid system as a eukaryotic expression system, in contrast, allows detection of weak and/or transient interactions under in vivo conditions. Similar approaches may be useful for other members of the S100 family to differentiate physiological protein-protein interactions from artifacts due to the preparation procedure that are not of biological relevance.
Conclusions-Analogous to both MRP8 and MRP14, S100A12 seems to play an important role during calcium-dependent activation of granulocytes, as indicated by its high calcium binding capacity. It has been previously shown that MRP8 and MRP14 modulate the activity of target proteins by binding to them in a calcium-dependent manner and, because S100A12 is also translocated from cytoplasm to cytoskeleton and membrane structures in response to elevated calcium concentrations, it seems likely that S100A12 exerts its effects through a similar mechanism. However, in contrast to MRP8 and MRP14, S100A12 is expressed only by granulocytes, and thus it is tempting to speculate that S100A12 transduces signals specific to granulocytes or modulates common stimuli for granulocytes and monocytes to granulocyte-specific responses (58,59). Furthermore, because we did not find evidence for S100A12 to form complexes with MRP8 and MRP14, it seems likely that S100A12 transduces calcium signals separately and that S100A12 might be aimed at targets different from those of MRP8 and MRP14. In brief, we suggest that S100A12 is an important calcium signal transducer in granulocytes. An understanding of such granulocyte-specific transduction pathways will offer specific targets to analyze and modulate undesired activities of these cells during inflammatory diseases.