Mast Cell and Monocyte Recruitment by S100A12 and Its Hinge Domain*

S100A12 is expressed at sites of acute, chronic, and allergic inflammation. S100 proteins have regions of high sequence homology, but the “hinge” region between the conserved calcium binding domains is structurally and functionally divergent. Because the murine S100A8 hinge domain (mS100A842–55) is a monocyte chemoattractant whereas the human sequence (hS100A843–56) is inactive, we postulated that common hydrophobic amino acids within the S100A12 hinge sequence may be functional. The hinge domain, S100A1238–53, was chemotactic for human monocytes and murine mast cells in vitro. S100A1238–53 provoked an acute inflammatory response similar to that elicited by S100A12 in vivo and caused edema and leukocyte and mast cell recruitment. Circular dichroism studies showed that S100A1238–53 had increased helical structure in hydrophobic environments. Mutations in S100A1238–53 produced using an alanine scan confirmed that specific hydrophobic residues (I44A, I47A, and I53A) on the same face of the helix were critical for monocyte chemotaxis in vitro and generation of edema in vivo. In a hydrophobic environment such as the cell membrane, these critical residues would likely align on one face of an α-helix to facilitate receptor interaction. Interaction is unlikely to occur via the receptor for advanced glycation end products but, rather, via a G-protein-coupled mechanism.

S100A12 is a proinflammatory protein that is chemotactic for monocytes (1,2) and promotes neutrophil adhesion and mobilization of neutrophils from the bone marrow (3). S100A12 in synovial fluid from rheumatoid arthritis patients, (2) lungs of patients with asthma (4), and cystic fibrosis (5) and in chronic inflammatory bowel diseases (for review, see Ref. 6) may contribute to leukocyte recruitment. S100A12 (also known as CAAF1, calgranulin C and EN-RAGE 4 (extracellular newly identified receptor for advanced glycation end products)) and the two other inflammation-associated calgranulins, S100A8 (migration inhibitory factor-related protein (MRP8)) and S100A9 (MRP14), are located within an S100 gene cluster on human chromosome 1q21.2-q22 (7). Human S100A12 shares 40 and 46% amino acid identity with S100A8 and S100A9, respectively, and is constitutively expressed with these in neutrophils. S100A8/S100A9 are more widely expressed in a variety of cell types associated with active inflammation (8) and depend on hetero-complex formation for many functions (9), whereas S100A12 does not interact with other S100 proteins (10). On the basis of structural and distributional differences, S100A12 is considered functionally distinct (8,11), and there is no S100A12 in the rodent genome (7,8). S100A12 is up-regulated in monocytes by pro-inflammatory mediators (2,12) and is expressed by eosinophils and macrophages in asthmatic airways in regions where mast cells (MC) accumulate. It provokes mild degranulation and potentiates IgE-mediated MC activation (4) of human and rodent mucosal and tissue MC in vitro and induces pro-inflammatory cytokines, particularly interleukin-6 and -8. S100A12-induced changes in the microcirculation including edema, increased leukocyte rolling, adhesion, and transmigration are MCdependent (4). Leukocyte recruitment is a characteristic of asthma, and MC numbers infiltrating bronchial smooth muscle are greater than in normal subjects (13) and are linked to hyperresponsiveness. Stem cell factor may be MC-specific, but effects on MC migration and activation are weak (14), and several chemokines, adenosine nucleotides, and IgE, in the presence or absence of allergen, may contribute to MC sequestration (15)(16)(17).
Here we show that S100A12 promotes migration of MC in vitro and in vivo and that S100A12 38 -53 , a region containing the hinge domain, was chemotactic for murine bone marrow-derived mast cells (BMMC), rat peritoneal mast cells (PMC), and human monocytes in vitro. Like S100A12, S100A12 38 -53 provoked mild mast cell degranulation in vitro and leukocyte recruitment in vitro and in vivo and induced a Ca 2ϩ influx in mouse BMMC. Amino acid residues important for activity were identified; those essential for MC activating activity and edema were hydrophobic (Leu 40 , Ile 44 , Ile 47 , and Ile 53 ). Dose response curves indicated a single receptor on MC and high and low affinity receptors on THP-1 monocytoid cells that were inhibited by pertussis toxin, suggesting involvement of G-protein-coupled receptors in signal transduction. Results strengthen our proposal for a role for S100A12 in allergic inflammation and may lead to rational design of S100A12 antagonists.

EXPERIMENTAL PROCEDURES
Human Subjects and Animals-Peripheral blood from normal donors was obtained with approval of the Human Care and Ethics Committee of the University of New South Wales with written informed consent. All animals were specific pathogenfree; BALB/c mice were used as a source of BMMC; Quackenbush Swiss mice were used to study responses to S100A12 in vivo, and Sprague-Dawley rats (10 weeks old) were the source of PMC. Experiments were approved by the Animal Care and Ethics Committee of the University of New South Wales.
Protein and Peptides-Recombinant human S100A12 was expressed and purified as described (2). Chain assembly of S100A12 38 -53 (KELANTIKNIKDKAVI) and its alanine scan mutants (24) was performed manually as described (25). Endotoxin levels in all preparations were Ͻ10 pg/10 g using the chromogenic limulus amebocyte lysate assay (Cape Cod Associates, Woods Hole, MA).
Electrospray-Ionization Mass Spectrometry-Electrosprayionization mass spectrometry was performed on an LCT timeof-flight mass spectrometer (MicroMass, Manchester, UK). Samples were injected into a moving solvent (100 l/min; 70% aqueous acetonitrile ϩ 0.1% formic acid) and entered the analyzer through an interface plate. Full scan mass spectra were acquired over the mass range 500 -2000 Da. Molecular masses were calculated from the observed m/z value using Mass Lynx software (MicroMass).
Circular Dichroism (CD) Spectroscopy-CD spectra were recorded on a Jasco J-730 spectropolarimeter. Samples (59 M) were dissolved in 50 mM phosphate buffer, pH 7.2, and data between 190 and 260 nm were recorded.
Peripheral blood mononuclear cells from buffy coats of citrated blood of normal donors were isolated using Ficoll-Paque TM Plus, (Amersham Biosciences), washed 3 times with Ca 2ϩ , Mg 2ϩ -free Dulbecco's phosphate-buffered saline (Invitrogen), and resuspended in 0.2% bovine calf serum/RPMI. BMMC were differentiated from murine bone marrow stem cells cultured in conditioned media consisting of supernatants from WEHI-3B (source of interleukin-3) and NIH-3T3 fibroblasts (source of stem cell factor) and 10% bovine calf serum in RPMI 1640 as described (26). Cells were passaged weekly and differentiated until Ͼ98% MC (ϳ3 weeks), assessed by metachromatic staining of granules with toluidine blue (TB).
PMC harvested from rats by peritoneal lavage were isolated as described (27). Purity by TB staining was ϳ99% MC, and viability was Ͼ98%. Cells were suspended in 0.2% bovine serum albumin (Sigma)/RPMI.
Chemotaxis and Inhibition Studies-Chemotaxis of calcein (Molecular Probes Inc., Eugene, OR)-loaded cells was performed as described (2) using 96-well chambers separated by a polycarbonate membrane (NeuroProbe Inc., Bethesda, MD) with pores of 5 m for peripheral blood mononuclear cells, 8 m for BMMC, and 10 m for THP-1, with six replicates per sample. C5a (Sigma; 10 Ϫ8 M) was employed as a positive control. Chambers were incubated at 37°C in 5% CO 2 in air for 1.5 h (peripheral blood mononuclear cells), 2 h (THP-1), or 4 h (BMMC).
F-actin Polymerization and Cytosolic Ca 2ϩ Flux-Cells (0.5 ϫ 10 6 /ml) in Hanks' balanced salt solution and 0.2% bovine serum albumin were prewarmed at 37°C in 5% CO 2 before the addition of S100A12 or S100A12 38 -53 diluted in the same buffer at 37°C (2). Cells (2.5 ϫ 10 4 ) in 96-well plates cells were mixed with 50 l of stimulants to yield final concentrations between 10 Ϫ9 and 10 Ϫ12 M. Cells were permeabilized and stained for F-actin for 60 min with a mixture of 0.05% (w/v) digitonin (Calbiochem) and 0.5 M fluorescein isothiocyanatephalloidin (Sigma). fMLP (10 Ϫ9 M) was used as positive control.
In Vivo Studies-To assess inflammatory responses in vivo, mice were injected with 10 g of S100A12 or S100A12 38 -53 in 1 ml of Hanks' balanced salt solution intraperitoneally. Peritoneal cavities were lavaged (5 ϫ 5 ml of citrated Hanks' balanced salt solution). 8 and 24 h later cells were harvested by centrifugation, and total MC numbers were determined using cytospin preparations after TB staining. Alternatively, S100A12, S100A12 38 -53 , or mutant peptides (2 g/30 l of Hanks' balanced salt solution/site) were injected into the footpad. Footpad thickness was measured with an electronic caliper (Mitutoyo, Japan) before and 20 and 30 min and 1 and 2 h after injections by two blinded investigators. Edema was defined as positive if the difference in footpad thickness 60 min post-injection was 0.1 mm greater than the pre-injection reading. To determine changes in vascular permeability, Evan's blue (1% in saline; 0.05 ml/mouse) was injected intravenously 30 min before footpad injection. The assessment was expressed as a clinical score: 0, no blue, no edema; 1, edema, no blue; 2, edema, light blue; 3, edema, blue; 4, edema, dark blue. For some samples footpads were harvested 2 and 8 h post-injection, and leukocyte infiltration was assessed on formalin-fixed sections stained with hematoxylin and eosin or TB to detect MC. Macrophages were identified with anti-Mac-3 mAb (BD Biosciences Pharmingen), and reactivity was detected with biotinconjugated secondary antibody and streptavidin-horseradish peroxidase with 3,3Ј-diaminobenzidine (DAKO, Denmark).
Statistical Analysis-Data are expressed as the means Ϯ S.D. Data from two groups were assessed using the Student's t test and multiple groups by analysis of variance followed by the Tukey-Kramar or Bonferroni test, and p values Ͻ 0.05 were considered significant.

Human A12 Is a Mast Cell Chemoattractant in Vitro and in
Vivo-Because S100A12 activates MC, and as MC numbers increase in asthmatic tissue, their chemotactic response was tested. S100A12 stimulated BMMC migration in a dosedependent manner that was optimal at 10 Ϫ11 M ( Fig. 1; 2.74 Ϯ 0.19 ϫ 10 4 cells migrated compared with control, 2.02 Ϯ 0.17 ϫ 10 4 cells, p Ͻ 0.05). Directional, rather than random migration, was confirmed when the gradient was negated by equivalent amounts of S100A12 in the upper and lower chambers ( Fig. 1; p Ͻ 0.01 relative to maximal chemotaxis at 10 Ϫ11 M S100A12in the lower chamber only). Human cord blood MC migration was also significantly greater than control values with 10 Ϫ9 -10 Ϫ10 M S100A12 (p ϭ 0.02 compared with control; not shown).
Responses of both types of MC to S100A12 were equivalent or greater than those provoked by a pre-optimized concentration of C5a. Because BMMC are responsive to S100A12 (Ref. 4 and Fig. 1), these were routinely used as they are more rapidly differentiated in vitro at significantly less cost.
Chemotactic Activity of S100A12 38 -53 -To determine whether the hinge domain of S100A12 could provoke a chemotactic response, migration of THP-1 monocytoid cells, peripheral blood mononuclear cells (S100A12 is only chemotactic for monocytes) (2), and mouse BMMC were tested. S100A12 consistently provoked THP-1 cell migration ( Fig. 2A) in a biphasic manner with two prominent peaks at 10 Ϫ9 and 10 Ϫ12 M (both p Ͻ 0.001 compared with negative control). S100A12 38 -53 was also chemotactic at these concentrations (p Ͻ 0.05 compared with negative control), although the peptide was significantly less potent than the full-length protein. When equivalent levels of S100A12 38 -53 were included in the upper and lower chambers, basal migration did not significantly increase ( Fig. 2A), confirming directional migration. Similar dose response optima were obtained for migration of blood monocytes in response to S100A12, although the peptide was not active at 10 Ϫ9 M (Fig. 2B). Chemoattractants cause polymerization of actin, a major component of microfilaments important for cell migration. S100A12 (10 Ϫ11 M) increased F-actin content in THP-1 cells within 20 s, which plateaued then increased and was sustained over 2 min. fMLP caused similar increases but was less sustained. S100A12 38 -53 provoked a somewhat slower response with a second more intense rise in F-actin content apparent after 60 s, which returned to base line within 120 s (Fig. 2C).

TABLE 1 Leukocyte recruitment after S100A12 injection
Inflammatory cells in footpads of mice 8 h post-injection of S100A12 or S100A12 38 -53 . Cell numbers were quantitated as described under "Experimental Procedures." Data are expressed as the means Ϯ S.D./field, 3 fields/section from each of 4 mice/group.

Cell type
Vehicle S100A12 S100A12 38  tide, and L40A, I44A, I47A, and I53A were essential for edema and optimal chemotactic responses. Structural Analysis-The secondary structure of S100A12 38 -53 was examined by CD spectroscopy in aqueous and hydrophobic environments (Fig. 7, A-D). It adopted a random structure in aqueous buffer and had a higher propensity to form an ␣-helical conformation, as indicated by characteristic minima at A 208 and A 222 nm , with increasing solvent hydrophobicity. Each of the inactive mutants had CD spectra identical to native S100A12 38 -53 (not shown).
Helical wheel analysis showed that residues important for chemotactic activity of the hinge domain occur on one side of the helix (Fig. 7E). These include Asn-42, Ile-44, Asn046, Ile-47, and Ile-53. Residues essential for edema were also hydrophobic (Leu-40, Ile-44, Ile-47, and Ile-53) Alignments of S100A12, mS100A8, and hS100A8 are given (Fig. 7F); identical amino acids are indicated in red. Asn-46 and Ile-47 were critical amino acids for S100A12 chemotaxis, and these are also located in the hinge domain of mS100A8 but not in hS100A8.

DISCUSSION
Mast cells are important effector cells in allergic inflammation, infection, and in some chronic inflammatory conditions and are widely distributed in vascularized tissues and certain epithelia. Their localized accumulation may be due to redistribution of neighboring MC in response to factors such as stem cell factor, transforming growth factor-␤, C5a, CXC chemokines (30), and interleukin-15 (31). Our recent studies showed S100A12 to be a potent activator of MC, and its expression in macrophages in the vicinity of tryptase-positive MC, in eosinophils infiltrating the airways of patients with asthma, and elevated levels in the sputum of patients with eosinophilic asthma implicate it in pathogenesis (4). S100A12-activated MC produce chemokines that could contribute to the high numbers of infiltrating monocytes found in the airways of patients with allergic asthma (32). However, S100A12 is also a potent chemoattractant for monocytes, (2), and here we demonstrate its chemotactic activity for MC, with optimal activity in the low nanomolar range and potency greater than C5a. These properties strongly support a role for this protein in allergic inflammation.
In contrast to monocytes, BMMC responded to native S100A12 and to S100A12 38 -53 with equal magnitude, with a bell-shaped dose response over a narrow optimal concentration range around 10 Ϫ11 M, and both induced a Ca 2ϩ influx of a magnitude comparable with that generated by C5a. Prominent MC degranulation was evident after footpad injection of S100A12 38 -53 , and vascular permeability and leukocyte numbers, including macrophages and MC, increased. The hinge region peptide consistently generated edema and leukocyte recruitment after footpad injection into mice.
The C-terminal and hinge domains of S100 proteins can have distinct functions. Studies with S100〉 confirmed that at least two sites interact with effector proteins and suggest that the C-terminal domain may be common to interactions with a number of effector proteins, whereas that containing the hinge domain may be target-specific (36). Comparisons of the crystal structure of human S100A9, which has high sequence homology with S100A12, indicate that the hinge region is a targetbinding site in these proteins (37), a proposal supported by the crystal structure of S100A12 (38). CD spectroscopy confirmed that the S100A12 hinge domain adopted a random structure in aqueous buffer, with increasing hydrophobicity resulting in a higher propensity to form an ␣-helical conformation. Helical wheel analysis suggests that in a hydrophobic environment such as a cell membrane, critical residues in this region would align on one face of the helix to interact with a receptor. The hydrophobic residues Asn-42, Ile-44, Ile-47, and Ile-53 were essential for monocyte chemotaxis, MC activation, and edema in vivo; Asn-46 was also critical for chemotaxis. Similarly, the hinge region of murine S100A8 has considerable secondary structure in a hydrophobic environment (39). Upon Ca 2ϩ bind-

Activity of hinge mutants in vivo and in vitro
Clinical scores of edema in mouse footpads (n ϭ 4) injected with 2 g of S100A12 or hinge peptide mutants assessed as detailed under "Experimental Procedures." THP-1 cell numbers (ϫ10 4 ; means Ϯ S.D.) attracted by 10 Ϫ9 or 10 Ϫ12 M preparations are from 3 experiments (6 wells/assay). Basal migration, 6.6 Ϯ 0.5 ϫ 10 4 cells; C5a, 10 Ϫ8 M, 11.7 Ϯ 0.7 ϫ 10 4 cells. The activity of full-length S100A12 is provided for comparison. Mutants (italics) with significantly reduced activity are shown in bold. The native peptide is represented by A41A and A51A is in bold italics.
ing, dimeric S100 proteins undergo a large conformational change, exposing a hydrophobic cleft defined by residues in the hinge region, the C terminus, and regions of helix 3, such as those important for interaction of S100A1 and its target peptide (TRTK-12) derived from the actin-capping protein CapZ (40). Binding of S100B to p53 down-regulates p53 tumor suppressor activity in cancer cells. An NMR-docked model of an inhibitor of this response, pentamidine, bound Ca 2ϩ -loaded S100B within helices 3 and 4 and the hinge region (41). Thus, calcium binding may expose the hinge domain (residues 38 -53) of S100A12 and facilitate interactions with lipid environments within the surface membrane. Interestingly, Asn-46 and Ile-47 were critical amino acids for S100A12 chemotaxis, and these are also located within the hinge domain of murine S100A8 but not in human S100A8 and may account for the lack of chemotactic activity of the human peptide (21). Dose response curves indicated high and low affinity receptors for S100A12 on THP-1 monocytoid cells, and the Ala scan supported this notion as Asn-42 and Ile-44 were essential for the chemotactic response provoked by 10 Ϫ12 M S100A12 38 -53 , whereas mutation of Lys-38, Leu-40, Asn-46, Ile-47, Asp-49, Lys-50, and Ile-53 all significantly reduced migration provoked by 10 Ϫ9 M S100A12 38 -53 , suggesting involvement of additional charged interactions. Identification of S100 receptors has been elusive. RAGE is proposed as a pan S100 receptor, although there are clearly additional/alternate receptors (4,42). S100A12 was the first of this class of protein found to bind RAGE (1). However, we recently showed that although human MC respond to S100A12, RAGE mRNA/protein expression was not found in these cells. Studies using RAGE-null mice show that RAGE plays little role in adaptive immune responses (43), and although a soluble RAGE antagonist (sRAGE) reduced delayedtype hypersensitivity responses proposed to be mediated by S100A12 (1), this was similarly suppressive in RAGE-null mice, and undefined effects other than simply blocking cell-surface RAGE function are proposed (43). sRAGE inhibited the ability of S100A12 to promote migration of THP-1 cells at 10 Ϫ9 M, but this was not affected by inhibiting ERK1/2 even though RAGE is expressed by these cells (4). Attempts to neutralize RAGE with appropriate antibodies were not successful (not shown). Furthermore, as for murine S100A8 (20), both peaks of S100A12induced migration of THP-1 cells were pertussis toxin-sensitive, suggesting a G-protein-coupled mechanism. Although these parameters were not tested using MC, we showed that human cord blood-derived MC do not express RAGE (4), and it is possible that MC and THP-1 cells share a common, as yet unidentified receptor. Because S100A12 binds the two C-type domains of RAGE (44), we hypothesized that an unidentified receptor may have sequence similarity to this region. However, data base searches (BLASTP) yielded no homology matches to G-protein-coupled receptors when we searched with both C-type domains together or separately. We are currently investigating the nature of S100A12 interactions with MC.
This study indicates that S00A12 may contribute to monocyte and MC migration and MC activation and identifies the FIGURE 7. Structural analysis. CD spectroscopy for S100A12 38 -53 displayed in increasingly hydrophobic environments (uncorrected for peptide concentration) (A) of acetonitrile (B), trifluoroethanol (TFE; C), and sodium dodecyl sulfate (SDS; D); overlays of each of the maximum concentrations of hydrophobic buffer. E, helical wheel analysis revealed a helical structure consisting of seven spokes corresponding to two full turns of an ␣-helix. Blue residues located on one side of the wheel had some contribution, and red residues on the other side strongly contributed to activity. F, alignments of hinge domains of mS100A8, hS100A8, and S100A12; identical amino acids shown in red.
hinge region as a functional domain. It reinforces a potential role for S100A12 in allergic inflammation and forms the basis for the design of antagonists that may have useful therapeutic applications. Delineation of the structural motifs that mediate S100A12-target interactions and their relevance in vivo are required to fully understand its role in the pathogenesis of inflammatory diseases.