Chemotactic Activity of S100A7 (Psoriasin) Is Mediated by the Receptor for Advanced Glycation End Products and Potentiates Inflammation with Highly Homologous but Functionally Distinct S100A15¹

Human S100A7 (hS100A7, psoriasin) was first identified as a protein up-regulated in inflamed hyperplastic psoriatic skin (1). Since then, hS100A7 has been widely studied because it is differentially expressed and released during inflammation (2, 3, 4, 5). Recently, hS100A15, highly homologous to hS100A7, was also identified as up-regulated in psoriasis, thereby suggesting a role in the inflammatory phenotype (6). However, the characteristics and functional mechanisms that link both hS100A7 and hS100A15 to inflammation are not defined. Both proteins belong to the multigenic S100 family of EF-hand proteins (7, 8). Despite their small size and conserved functional domains, gene duplications and variations throughout vertebrate evolution led to an increase in number and diversity within the S100 family. Genomic analysis of the hS100A7 and hS100A15 encoding chromosomal regions (chromosome 1q21, epidermal differentiation complex) reveals that both evolved recently by gene duplications during primate evolution, diverging the least among all S100 family members (9, 10). Thus, their high homology (93% sequence identity) makes it difficult to distinguish them when co-expressed. To address this further, we have developed unique reagents and functional assays to dissect similarities and differences among these closely related homologs. This study provides insight into properties and functional mechanism of S100A7- and S100A15-induced inflammation and reveals an unexpected biological diversity that may drive their functional synergism in inflammatory diseases.

All the materials and methods used have been approved by the National Cancer Institute.

cDNA was prepared by PCR cloning of full-length transcripts (hS100A7: NM_002963; hS100A15: AY189119) isolated from human skin 6. Proteins were expressed in Escherichia coli BL21 (DE3) cells harboring plasmid His⁶-myelin basic protein (MBP)³ -tev hS100A15/-hS100A7, and proteins were purified as described (11). Purified S100 protein, a single band on Coomassie-stained gels (M_r of ∼12 kDa), was purged of endotoxin before experiments by chromatography onto Detoxigel columns (Pierce) with endotoxin levels <0.05 EU/μg protein. Endotoxin levels are below biological activity as demonstrated in vivo and in vitro inflammation and chemotaxis assays demonstrating that neutralization of S100 proteins or heating abrogated functional activity.

Normal human keratinocytes (Cascade Biologics) were cultured in keratinocyte growth medium containing insulin (5 μg/ml) and bovine pituitary extract (50 μg/ml) at 37°C in air containing 5% CO₂. Keratinocytes washed twice with PBS were harvested into lysis buffer for protein analysis (Cell Signaling Technology) as described below. Chinese hamster ovary (CHO) cells were cultured in Ham’s F12 (Invitrogen) with 10% BSA plus Zeocin (200 μg/ml) for selection of receptor for advanced glycation end products (RAGE) transfectants at 37°C in air containing 5% CO₂ (12).

Monospecific antisera to hS100A15 were prepared in rabbits by injecting a synthetic peptide that corresponds to the N-terminal amino acid sequence of the deduced hS100A15 protein (GenBank accession no. AAO40032). The Abs were affinity purified using the synthetic peptide coupled to Affigel-15 (Bio-Rad). Most of the commercial and donated hS100A7 Abs tested detected both proteins (data not shown). The monoclonal anti-hS100A7 Ab (Imgenex; Abcam) specifically detects recombinant S100A7 monomer without crossreacting with recombinant hS100A15 or several other hS100 proteins (50 ng/lane) (see Fig. 1 A). hS100A8 and hS100A10 (Novus Biologicals) were used as controls. Additionally, preadsorptions with increasing doses of the corresponding cognate proteins blocked respective S100 Ab staining (Ref. 13 and data not shown).

To measure downstream MAP kinase activity, human granulocytes from normal volunteers were resuspended in RPMI 1640 (Invitrogen) at 10 × 10⁶/ml and were stimulated with 1 μg/ml hS100A7 ± 15 nM ZnCl₂ at 37°C for 5 min before being pelleted followed by removal of media. The pellets were quick frozen in a dry ice/methanol bath. Cells were preincubated with neutralizing anti-RAGE (5 μg/ml, R&D Systems) or soluble RAGE (R&D Systems) 30 min before stimulation.

For immunoblot analysis, total cell lysates of neutrophils, CHO cells, or cultured keratinocytes (20 μg) were prepared using 1% Triton-containing lysis buffer (Cell Signaling Technologies). Proteins were separated using a 12% SDS-polyacrylamide gel, transferred to reinforced nitrocellulose membranes, and visualized by Ponceau stain. Filters were incubated with blocking buffer (TBS (pH 7.4) with 5% milk, 0.1% Tween 20) for 30 min, primary Ab (anti-hS100A15, 1 μg/ml; anti-hS100A7 Ab, 1 μg/ml; anti-phospho-ERK1/2, anti-total ERK1/2, 1:1000, Cell Signaling Technologies; anti-RAGE, 1:250, Santa Cruz Biotechnology; anti-β-actin, 1:20,000, Chemicon International) overnight, and secondary Ab for 1 h with several washes (TBS (pH 7.4), 0.1% Tween 20) between incubations.

Immunoflorescent staining was performed on serial 5-μm frozen sections of human normal and psoriatic skin fixed in acetone. The sections were treated with 96% methanol and 4% H₂O₂, blocked in 10% normal goat serum, and incubated overnight with anti-hS100A15 or anti-hS100A7 (5 μg/ml each). Experiments without primary Abs and serial dilution competition assays were performed in the absence and presence of blocking peptide to determine the optimal working concentration and specificity of the primary hS100A15 Ab. Donkey anti-rabbit Cy3 (1:250) was used as a secondary Ab to the hS100A15 or donkey anti-mouse FITC (1:250) to hS100A7 (The Jackson Laboratory). When sections were co-stained, FITC-labeled Mart-1/Hmb 45 cocktail at 1/25 dilution (monoclonal, Biocare Medical) and MHCII-FITC at 1/250 dilution (Serotec) were mixed with the primary Ab. All sections were nuclear stained with DAPI (Sigma-Aldrich) and mounted. All biopsies were taken with the patient’s informed consent and approval of the local ethics committee.

Human leukocytes from normal volunteers were Percoll purified (95%) from leukopacs provided by the National Institutes of Health blood bank (under IRB 99-CC-0168) as previously described (14). The chemoattractants hS100A7, hS100A15, and CXCL8/IL-8 (PeproTech) were diluted in RPMI 1640 containing 1% BSA and placed in the lower 25 μl chamber. hS100A7 was mixed with ZnCl₂ at least 30 min prior to being placed in a chemotaxis chamber. Cells were suspended in the above medium at 0.5–5 × 10⁶/ml and 50 μl was placed in the upper chamber, which was separated from the lower one by a polycarbonate filter (pore size pf 5 μm), which was coated with fibronectin for lymphocyte migration. In selected experiments, neutralizing anti-hRAGE (5 μg/ml), pertussis toxin (50–100 ng/ml), CXCL12 (PeproTech), PD98059 (10 μM), hS100A7, or hS100A15 was added 30 min at 37°C before the cells were added in the upper chamber. Cells were allowed to migrate for 3 h (T lymphocytes), 1 h (granulocytes), or 1.5 h (monocytes) at 37°C in a humidified, 5% CO₂ incubator. Cells attached to the lower filter surface were stained with a rapid stain (Richard-Allan Scientific), counted in six fields, and the averages were expressed as ratio (chemotactic index) of the numbers of cells in the treated sample vs control (spontaneous migration).

Recombinant hS100A7 was biotinylated using N-hydroxysuccinimide ester of biotin (NHS-PEO₄-biotin) containing a polyethylene glycol spacer arm that renders/maintains hS100A7 aqueous solubility. hS100A7 (2 mg) dissolved in PBS (0.1 M phosphate, 0.15 M NaCl (pH 7.2)) was mixed with 2 molar excess of NHS-PEO₄-biotin and incubated on ice for 2 h. Biotinylated-S100A7 (biotin-S100A7) was separated from unreacted NHS-PEO₄-biotin using Econo-Pac 10DG column (Bio-Rad) and stored at −80°C until use. Ligand binding assays were performed in 96-well polystyrene plates that were preadsorbed with purified murine recombinant soluble RAGE (100 ng/well) overnight and blocked with 0.25% casein (I-block) (Applied Biosystems). Assays were performed in the presence of the indicated concentration of biotinylated hS100A7 with/without excess of unlabeled recombinant hS100A7, hS100A15, nonimmune IgG (Jackon ImmunoResearch Laboratories), anti-mouse RAGE (R&D Systems), anti-human RAGE (R&D Systems), soluble RAGE (sRAGE; R&D Systems), CXCL8/IL-8, or I-block in PBS (1 mM KH₂PO₄, 3 mM Na₂HPO₄, 155 mM NaCl, 1.2 mM CaCl₂, 50 μM ZnCl₂) for 90 min at 37°C. Wells were washed rapidly to remove unbound ligand, then incubated with streptavidin alkaline phosphatase (1/1500 dilution) in I-block at room temperature for 60 min, and washed with PBST (PBS containing 0.05% Tween-20), distilled water, and Tris buffer (20 mM Tris, 1 mM MgCl₂ (pH 9.5)) before adding CDP-Star containing Emerald II enhancer (Applied Biosystems) and incubating at room temperature for 20 min and at 4°C overnight. Luminescence was measured using a TR717 microplate luminometer (Applied Biosystems). Binding of biotin-S100A7 to RAGE is directly proportional to the amount of luminescence signal in the well and is reported as luminescence units. High-binding polystyrene 96-well plates were obtained from Greiner Bio-One, streptavidin-alkaline phosphatase, I-block (casein), and chemiluminescent substrate (CDP-Star containing Emerald II enhancer) were obtained from Applied Biosystems, and NHS-PEO₄-biotin was from Pierce.

Control vector or RAGE-transfected CHO cells were allowed to adhere to immobilized hS100A7, hS100A15 (10 μg/ml each), and to BSA (as control) according to a previously described modified protocol (15). Briefly, microtiter plates were coated with the S100 proteins in bicarbonate buffer (pH 9.6) and blocked with 3% BSA solution. Fluorescently labeled (2′,7′-bis-(2-carboxyethyl)-5-(and 6)-carboxyfluorescein acetoxymethyl ester) CHO control or RAGE-vector-transfected cells were washed in serum-free RPMI 1640 medium and plated onto the precoated wells (10⁵/well) at 37°C for 60 min in the absence or presence of zinc (50 μM) or neutralizing anti-RAGE or sRAGE (5 μg/ml, R&D Systems). After the 60-min incubation period, the wells were washed and adherent cells were quantified by measuring absorbance at 590 nm.

C57BL/6 RAGE^−/− mice were characterized previously (15). HBSS (Invitrogen), recombinant native or heat-inactivated S100 proteins (20 μg in 1 ml HBSS/mouse) (16, 17) were administered i.p. into C57BL/6 wild-type and RAGE^−/− mice. As control, thioglycollate-induced peritonitis was performed as previously described (12). To evaluate neutrophil recruitment, mice were sacrificed at 4 h after injection. Thereafter, the peritoneal lavage was collected and the numbers of emigrated cells were analyzed by hemocytometry. There was no difference in the circulating neutrophil count of wild-type or RAGE^−/− mice. Recovered peritoneal leukocytes were incubated for 5 min with 2.3G2 anti-FcIII/II receptor, then stained at 4°C for 30 min with allophycocyanin-conjugated anti-Gr1 as a neutrophil marker. Cells were fixed with 1% paraformaldehyde, and cytometry was done with a FACSCalibur cytometer (BD Biosciences) and results were analyzed with Diva software (BD Biosciences). Similar results were obtained by analyzing leukocyte populations with conventional smears. All Abs were purchased from BD Pharmingen.

The amino acid sequences of the human S100 proteins were aligned using the program T-Coffee (version 2.11).

Data were analyzed for statistical significance by ANOVA with post hoc Bonferroni analysis using SPSS (version 15.0), with p ≤ 0.05 being considered significant.

Because hS100A7 and hS100A15 are almost identical in sequence (93%), distinguishing their expression has been difficult (Fig. 1,A). To address this further, Abs were generated in rabbits to a unique N-terminal sequence in hS100A15. Immunoblotting revealed a single monomer band of recombinant hS100A15 distinct from other S100 proteins and native low- and high-molecular weight bands in keratinocyte lysates, as is typical for many S100 proteins 4 (Fig. 1 A). The hS100A15 Ab did not detect related recombinant hS100 proteins, particularly hS100A7. Similarly, the monoclonal hS100A7 Ab revealed specific staining of the hS100A7 monomer along with high-molecular weight forms in keratinocyte lysates that were distinct from those detected with the hS100A15 Ab.

Using these specific Abs, the expression and distribution of the corresponding hS100A7 and hS100A15 proteins in human skin were studied. In normal epidermis, hS100A7 expression was confined to the granular/cornified layer only (Fig. 1,B, upper left panel). Human S100A15 colocalized there, but unlike hS100A7, it was expressed by epidermal basal cells and dendritic-shaped cells (Fig. 1,B, upper right panel). By co-staining, these hS100A15-expressing dendritic cells were identified as melanocytes (MART-1 positive), whereas MART-1-negative, hS100A15-positive cells were identified as epidermal Langerhans cells (Fig. 1,B, lower left panel) and dermal dendritic cells (MHCII-positive, not shown). In the dermis, hS100A15 was further expressed by endothelial cells interior to vascular smooth muscle cells (smooth muscle actin-positive) (Fig. 1,B, lower right panel) and other cell types, where hS100A7 and hS100A15 were differentially expressed (summarized in Fig. 1 C).

Since psoriasis is the prototype of inflammatory disease where hS100A7 and hS100A15 were first identified, our unique hS100A15 Ab provided an opportunity to discriminate the expression of these proteins in psoriatic skin (Fig. 1 D). Compared with normal skin, both hS100A7 and hS100A15 are up-regulated in inflamed lesional psoriatic skin and co-expressed by the epidermal suprabasal compartments. Additionally, hS100A15 is highly expressed by basal psoriatic keratinocytes at the epidermal-stroma junction, as well as by dendritic and stromal cells in the dermis. Their differential expression in multiple cell types in different compartments of the skin might reflect their divergence and functional synergism, with implications for disease pathogenesis.

Both human S100A7 and S100A15 were identified in psoriasis and up-regulated in psoriasis, suggesting involvement in the inflammatory phenotype (1, 6). To investigate the possibility that these proteins attracted some of the inflammatory cells, we analyzed the chemotactic ability of hS100A7 and hS100A15 in a dual chamber assay (Fig. 2,A). hS100A15 was identified as chemotactic for both granulocytes and monocytes, but not for lymphocytes. In contrast, hS100A7 attracted all three leukocyte populations at similar concentrations. In both cases, chemotaxis diminished at 100 ng/ml, developing an overall bell-shaped response that is typical for chemokines (18). To examine the role for a Gi protein-coupled receptor (GiPCR) in the chemotactic response, the assay was performed in the presence of pertussis toxin (Fig. 2 B). hS100A15-mediated chemotaxis was attenuated upon pertussis toxin treatment, indicating signaling through a GiPCR. In contrast, pertussis toxin did not affect the hS100A7-induced neutrophil migration. As a positive control for the assay, IL-8 chemotaxis, mediated by the GiPCR CXCR1/2, was reduced by the pertussis toxin treatment.

A series of additional studies were performed to further elucidate the nature of the chemotactic receptors (Fig. 2,C). hS100A7-mediated chemotaxis could be inhibited by the specific anti-hS100A7 Ab but could not be desensitized through activation of the classical pertussis toxin-sensitive interaction of CXCL12 with CXCR4. In contrast, blocking of RAGE by neutralizing Abs abolished the hS100A7-mediated chemotaxis, but not the IL-8-mediated effect. To confirm a role of RAGE in hS100A7-mediated chemotaxis, hS100A7 was placed in the bottom chamber and normal RAGE-bearing granulocytes were added to the upper chamber (Fig. 2 D). When neutralizing anti-RAGE was added to the cells, hS100A7-mediated cell migration was inhibited, excluding undirected chemokinesis. In contrast, cell migration toward hS100A15 placed in the lower chamber was not affected by blocking RAGE on granulocytes in the upper chamber. Further, checkerboard analysis showed that hS100A7 and hS100A15 pro-migratory effects are mediated through chemotaxis (data not shown).

To address the nature of the pertussis toxin-sensitive hS100A15 receptor, desensitization experiments with ligands were performed that signal through GiPCR. Pretreatment of neutrophils with the ligands RANTES/CCL5, SDF-1/CXCL12, MCP-1/CCL2, and IL-8/CXCL8 did not attenuate the hS100A15-mediated chemotaxis, as opposed to corresponding ligand controls (data not shown). Due to the myeloid preference of hS100A15 and inability to attract mature or immature dendritic cells and lymphocytes, chemokine receptors that are preferentially expressed on lymphocytes or dendritic cells were excluded as potential targets for hS100A15 (CXCR 3, 5, and 6; CCR 4 and 6–10). BRAK/CXCL14 ligand (18, 19), which binds to an as yet unknown chemokine receptor(s), was able to desensitize hS100A15-mediated monocyte chemotaxis, whereas hS100A15 did not desensitize the chemotactic response to CXCL14 (data not shown). This unidirectional desensitization is probably based on the greater potency of CXCL14 than hS100A15.

To directly test the concept of hS100A7 and hS100A15 binding to different target receptors, ligand binding studies were performed (Fig. 3 A). Biotinylated hS100A7 bound to immobilized purified RAGE, and binding could be blocked by increasing concentrations of unlabeled hS100A7. Biotinylated hS100A7 binding to RAGE substrate was inhibited in the presence of excess sRAGE, a truncated form of the receptor spanning the extracellular domain, and by anti-human and anti-mouse RAGE-IgG as well as anti-hS100A7. Unrelated casein or nonimmune IgG was without effect, as was competition with increasing concentrations of the hS100A15 paralog or IL-8.

Structural and biochemical data indicate that the S100A7 homodimer structure is altered through the binding and release of zinc, an important mediator in inflammation (20, 21). To test if zinc affects the hS100A7-RAGE interaction and function, hS100A7-mediated ligand binding, signaling, and chemotaxis were analyzed. Ligand binding experiments with RAGE in a natural cellular environment showed that adhesion of RAGE-transfected CHO cells to hS100A7-coated wells required zinc (Fig. 3,B). The specificity of RAGE-hS100A7 interaction was shown by inhibition in the presence of excess sRAGE or neutralizing anti-RAGE IgG. Furthermore, the presence of zinc did not affect the lack of binding of RAGE-bearing cells toward plates coated with hS100A15. RAGE-deficient CHO cells transfected with a vector control did not adhere to plates coated with either S100 protein. Human neutrophils exposed to a gradient of recombinant hS100A7 protein in the absence of zinc showed no significant chemotactic response. The ability of hS100A7 to attract neutrophils was restored upon zinc replenishment (Fig. 3,C). In line with zinc-dependent hS100A7-RAGE interaction and chemotaxis, neutrophils stimulated with hS100A7 showed RAGE-dependent phosphorylation of the migration-related MAPK ERK1/2 pathway (22) only in the presence of zinc (Fig. 3,D). hS100A7-induced ERK1/2 phosphorylation was blocked by sRAGE (not shown) or neutralizing anti-RAGE IgG. The hS100A7-mediated chemotaxis was attenuated when neutrophils were pretreated with an ERK1/2 inhibitor, similar to IL-8 (Fig. 3 E), indicating the requirement for ERK1/2 signaling for chemotactic response to hS100A7.

To test the concept that both hS100A7 and hS100A15 are chemoattractants in vivo, corresponding recombinant proteins were injected i.p. into C57BL/6 mice. Proinflammatory response was evaluated by analyzing cell count and cell type attracted into the peritoneal cavity. Both S100 proteins provoked an inflammatory response that was dose- and time-dependent and was absent when heat-inactivated S100 proteins were injected (data not shown). In C57BL/6 wild-type mice, a response was detected by 4 h for both proteins attracting Gr-1-positive cells compared with control (Fig. 4 A). The hS100A7-induced inflammation was absent in RAGE^−/− mice, whereas the hS100A15-mediated inflammatory response remained intact. As a positive control, i.p. injected MIP-2/mCXCL2 provoked an inflammation in both wild-type and RAGE^−/− mice.

Since we found that hS100A7 and hS100A15 have distinct proinflammatory functions mediated through distinct mechanisms but are expressed in different compartments in the same inflammatory states, we tested for their combined effect as chemoattractants (Fig. 4,B). Although i.p. application of either hS100A7 or hS100A15 provoked an inflammatory response (compare Fig. 4,A), this was amplified substantially when both hS100A7 and hs100A15 were injected together. Thus, hS100A7 and hS100A15 act synergistically when co-expressed (Fig. 1 D), suggesting that both proteins might contribute to the amplified inflammatory phenotype in psoriasis and other inflammatory diseases.

Originally identified as a marker for psoriasis, hS100A7 (psoriasin) (1) and recently hS100A156 were found overexpressed in other inflammatory diseases (23, 24, 25). Physiologically, hS100A7 and hS100A15 may work together as part of an innate host defense controlling microbial growth on normal and inflamed skin (13, 26, 27, 28). Our current data indicate another function as chemoattractants for leukocytes. hS100A7 (13) and hS100A15 (25) are both regulated by proinflammatory cytokines that are predominant in inflammatory diseases, such as psoriasis (29, 30). That they are co-evolved, co-expressed, and co-regulated in identical pathological states begs the question of the biological significance of the evolutionary selection for these almost identical proteins.

Our data suggest a joint biological impact of hS100A7 and hS100A15 through evolutionary diversification. Both proteins are distinct in expression pattern, chemotactic activity, and functional mechanism. While each of them induces a proinflammatory response by itself, their proinflammatory activity is potentiated when acting together, indicating that they contribute independently but synergistically to inflammation that may be fundamental to the disease phenotype. We provide mechanistic insights into hS100A7 and hS100A15 synergism by showing their biological diversity in expression in distinct cells in both normal and inflamed tissue, in receptor activation, and in chemotactic function. The documented up-regulation of hS100A7 in epithelial cancers (31) and similar findings for hS100A15 (data not shown) suggest that combined expression of these chemotactic proteins may contribute to other diseases as well. Having established specific Abs for each protein, we are now in a position to address this question. Furthermore, parallel targeting of hS100A7 and hS100A15 would allow a novel approach to treat associated diseases based on distinct mechanisms with possibly enhanced therapeutic efficacy (32, 33).

With the discovery of hS100A7 in 1991, structural and functional data have stimulated the quest for mechanistic clarification of its extracellular action (21, 34). The multiligand receptor RAGE is implicated in inflammatory processes, including leukocyte migration (12, 15, 35, 36). RAGE is expressed at low levels in normal tissues and becomes up-regulated wherever its ligands accumulate. Ligands initiate a sustained cellular activation through MAP kinases, culminating in the activation of NFκB (37, 38). Through recognition of β-sheet fibrillar structures or proinflammatory cytokine-like mediators of the S100/calgranulin family or high mobility group box-1 (HMGB-1), RAGE participates in the phenotype of inflammatory skin diseases, diabetes, amyloidosis, and promotes tumor progression (37, 39, 40, 41, 42, 43).

The identification of RAGE as a receptor for hS100A7 allows for specific targeting of hS100A7-mediated effects, such as cell migration in inflammation. Furthermore, with participation of RAGE in tumor growth and metastasis (43), hS100A7-RAGE interaction might be important to understand the role of secreted hS100A7 in epithelial tumorigenesis.

In contrast to hS100A7, hS100A15 preferentially attracts myeloid cells through a pertussis toxin-sensitive mechanism, suggesting signaling through a GiPCR. Two other chemotactic S100 proteins, S100-L and CP-10, bind to pertussis toxin-sensitive receptors, which, however, are yet to be identified (44, 45). Similar to CP-10, calcium mobilization through hS100A15 could not be detected (data not shown). This observation is consistent with the idea that S100 proteins are less potent than the classical chemoattractants. Homologous desensitization experiments with chemokine receptor ligands suggest that receptors preferentially active on myeloid cells (CXCR1/2, CCR1) or widespread receptors (CXCR4; CCR2, 3, and 5) are not involved in hS100A15-mediated chemotaxis. Collectively, our data suggest that hS100A15 binds to an unidentified GiPCR that also interacts with CXCL14.

hS100A15 is functionally active in the presence of calcium alone compared with hS100A7, suggesting that additional factors are important for hS100A7 chemotactic activity. Besides calcium, many S100 proteins bind zinc with high affinity, implying that divalent zinc is a major regulator of their extracellular function (13, 20, 21, 46). In hS100A7 (21) and others (37, 46, 47), the location of the zinc binding site is near one of the proposed target protein binding sites; however, mechanistic and functional consequences are undefined. We demonstrate that zinc is required for interaction and signaling of hS100A7 with the target protein RAGE and that it is functionally important for hS100A7-RAGE-mediated chemotaxis. Thus, our data indicate that zinc is a necessary co-factor for hS100A7-mediated proinflammatory activity. Together with previous studies (13, 26, 27, 28) showing that the antimicrobial action of hS100A7 was through zinc sequestration, we suggest that zinc may be involved in regulation of hS100A7 dual proinflammatory and antimicrobial functions.

For S100 proteins, individual physical and biochemical characteristics including the spatial structure of the S100 dimer/multimers determine the specificity of the target interaction (7, 48, 49). We have data suggesting that sequences and secondary structures of the hS100A7 and hS100A15 monomers are alike (6, 11). However, the quaternary structures of both proteins are distinct and could account for binding to different target receptors (B. Ahvazi et al., unpublished data). Furthermore, two of the amino acids divergent in hS100A7 and hS100A15 (Thr⁵²Ile, Asn⁵³His) appear to reside in a region that was shown to be a chemotactic domain for mouse S100A8 (50). If functionally conserved in hS100A7 and hS100A15, these differences might contribute as well to the distinct chemotactic properties of these highly homologous proteins.

Understanding the diversity and functional synergism of disease-associated proteins is crucial for developing therapeutic interventions based on distinct mechanisms with possibly enhanced therapeutic efficiency in S100A7- and S100A15-mediated pathological conditions. The unexpected different biology of almost identical proteins suggests that S100A7 and hS100A15 functional synergisms through biological diversification might underlie their biological impact, and thus their evolutionary selection.

We are grateful to Stephen Wincovitch and Susan Garfield for skillful assistance with confocal microscopy and Barbara Taylor for advice and help with the flow cytometry. We thank William Gillette and Jim Hartley for recombinant proteins, Roberta Smith for quantitative cell analysis, and Helen Rager for endotoxin analysis. We thank Christophe Cataisson for skillful help with the in vivo experiments and critical reading of the manuscript. We thank Alif Dharamsi for competent technical assistance, and Peter Nawroth and Angelika Bierhaus for providing RAGE^−/− mice.

The authors have no financial conflicts of interest.