Myeloid-related protein (MRP) 8 and MRP14, calcium-binding proteins of the S100 family, are secreted by activated monocytes via a novel, tubulin-dependent pathway.

Myeloid-related protein (MRP) 8 and MRP14, two members of the S100 family expressed in myelomonocytic cells, have been ascribed some extracellular functions, e.g. antimicrobial, cytostatic, and chemotactic activities. Since S100 proteins lack structural requirements for secretion via the classical endoplasmic reticulum/Golgi route, the process of secretion is unclear. We now demonstrate the specific, energy-dependent release of MRP8 and MRP14 by human monocytes after activation of protein kinase C. This secretory process is not blocked by inhibitors of vesicular traffic through the endoplasmic reticulum and Golgi, and comparative studies on tumor necrosis factor-alpha and interleukin-1beta indicate that MRP8 and MRP14 follow neither the classical nor the interleukin-1-like alternative route of secretion. Inhibition by microtubule-depolymerizing agents revealed that MRP8/MRP14 secretion requires an intact tubulin network. Accordingly, upon initiation of MRP8/MRP14 secretion, immunofluorescence microscopy showed a co-localization of both proteins with tubulin filaments. Release of MRP8 and MRP14 is associated with down-regulation of their de novo synthesis, suggesting that extracellular signaling via MRP8/MRP14 is restricted to distinct differentiation stages of monocytes. Our data provide evidence that the S100 proteins MRP8 and MRP14 are secreted after activation of protein kinase C via a novel pathway requiring an intact microtubule network.

Myeloid-related protein (MRP) 8 and MRP14, two members of the S100 family expressed in myelomonocytic cells, have been ascribed some extracellular functions, e.g. antimicrobial, cytostatic, and chemotactic activities. Since S100 proteins lack structural requirements for secretion via the classical endoplasmic reticulum/Golgi route, the process of secretion is unclear. We now demonstrate the specific, energy-dependent release of MRP8 and MRP14 by human monocytes after activation of protein kinase C. This secretory process is not blocked by inhibitors of vesicular traffic through the endoplasmic reticulum and Golgi, and comparative studies on tumor necrosis factor-␣ and interleukin-1␤ indicate that MRP8 and MRP14 follow neither the classical nor the interleukin-1-like alternative route of secretion. Inhibition by microtubule-depolymerizing agents revealed that MRP8/MRP14 secretion requires an intact tubulin network. Accordingly, upon initiation of MRP8/MRP14 secretion, immunofluorescence microscopy showed a co-localization of both proteins with tubulin filaments. Release of MRP8 and MRP14 is associated with down-regulation of their de novo synthesis, suggesting that extracellular signaling via MRP8/ MRP14 is restricted to distinct differentiation stages of monocytes. Our data provide evidence that the S100 proteins MRP8 and MRP14 are secreted after activation of protein kinase C via a novel pathway requiring an intact microtubule network.
Myeloid-related protein-8 (MRP8; S100A8) 1 and MRP14 (S100A9) are two calcium-binding proteins of the S100 family (1-3), which has grown to be one of the largest subfamilies of EF-hand proteins (4). Members of this family are defined by their homologies to two calcium-binding proteins highly enriched in nervous tissue, S100␣ and S100␤. S100 proteins are characterized by the presence of two calcium-binding sites of the EF-hand-type, the N-terminal of which differs from the conserved EF motive by two additional amino acids. They have a relatively small molecular mass of around 10 kDa and, in contrast to other calcium-binding proteins such as calmodulin, show tissue-specific expression patterns. S100 proteins play a role in cell differentiation, cell cycle progression, regulation of kinase activities, and cytoskeletal-membrane interactions (4,5). In addition, extracellular functions have been reported for distinct S100 proteins; S100␤, the prototypic member of this family, can be found as an extracellular protein inducing neurite extension (5,6). S100L (S100A2) is chemoattractive for eosinophils (7), whereas psoriasin (S100A7) exhibits chemotactic activity for neutrophils and CD4 ϩ lymphocytes (8).
MRP8 and MRP14 are expressed at high concentrations by granulocytes and during early differentiation stages of monocytes but are absent in lymphocytes and mature tissue macrophages (9 -12). Down-regulation of MRP8 and MRP14 expression in monocytes involves a calcium-mediated suppressor mechanism (13). Phagocytes express MRP8 and MRP14 under multiple inflammatory conditions, e.g. during rheumatoid arthritis, allograft rejections, or inflammatory bowel diseases (9,14,15). Inflammatory disorders as chronic bronchitis, cystic fibrosis, or rheumatoid arthritis are associated with elevated serum levels of MRP8 and MRP14 (16,17). The close correlation between serum levels and disease activity led to the assumption that MRP8 and MRP14 are released from leukocytes during inflammatory events (4,5,16), e.g. during transendothelial diapedesis (11). The mechanism of release, either after cell death or by an active secretory process, has yet not been elucidated. In recent years several studies reported extracellular functions of MRP8 and MRP14, including antimicrobial, cytostatic, or chemotactic activities (18 -20), thus favoring an active mechanism of secretion. However, neither MRP8 and MRP14 nor any other S100 protein has a signal sequence for secretion via the classical ER/Golgi route. They therefore resemble cytokines such as interleukin-1␤ (IL-1␤) and basic fibroblast growth factor, which are released into the extracellular space via a so-called alternative pathway of secretion (21). However, it is not known whether all nonclassically secreted proteins use the same mechanisms of release.
Since the pathway of S100 protein release has not yet been examined (4), we intended to elucidate the molecular mechanisms associated with the release of MRP8 and MRP14 by human monocytes. Here we provide evidence that MRP8 and MRP14 are secreted after activation of protein kinase C via a novel pathway requiring an intact microtubule network.

MATERIALS AND METHODS
Cells and Cell Culture-Human peripheral blood leukocytes were obtained from buffy coats which arise during preparation of packed red * This work was supported by grants Sonderforschungsbereich 293-96 and Ro 1190/2-2 from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Enzyme-linked Immunosorbent Assay (ELISA) for MRP8/MRP14, Tumor Necrosis Factor-␣ (TNF-␣), and IL-1␤-The MRP8/MRP14 content in the supernatants of cultured monocytes was quantified by a sandwich ELISA as described earlier (16,22). The ELISA was calibrated with recombinant MRP14 in concentrations ranging from 1 to 1000 ng/ml. The sensitivity was less than 2.5 ng/ml.
For detection of secreted IL-1␤ and TNF-␣, Biotrak sandwich ELISA systems were obtained from Amersham-Buchler (Braunschweig, FRG) and employed according to the manufacturer's instructions. The sensitivities were less than 5 and 1 pg/ml for TNF-␣ and IL-1␤, respectively.
Flow Cytometry-Monocytes of culture day 1 were stimulated as described above and processed for flow cytometry. Immunostaining procedures were performed as reported earlier employing mAb 27E10 (9,23). Surface expression was analyzed employing a FACScan (Becton Dickinson, Heidelberg, FRG) equipped with Lysis II software.
Metabolic Labeling and Immunoprecipitation-Monocytes of culture day 1 were harvested, washed, and preincubated in modified Eagle's medium without methionine (Life Technologies, Inc.) at a density of 5 ϫ 10 7 cells/ml. Monocytes were then labeled by adding 250 Ci/ml [ 35 S]methionine (Amersham-Buchler) to the same medium for 2 h. Medium was removed and replaced by McCoy's 5A medium supplemented with 15% fetal calf serum. Cells (2 ϫ 10 7 ) were treated for another 4 h with either medium, 10 nM PMA, or 10 nM PMA plus 20 nM staurosporine. Thereafter, supernatants were collected, and cells were washed and lysed in phosphate-buffered saline containing 1% Nonidet P-40 and 2 mM phenylmethylsulfonyl fluoride (Sigma). Supernatants and lysates were prepared for immunoprecipitation as described earlier (13) using monospecific affinity-purified antisera against MRP8 or MRP14, mAb 27E10, mAb H21 against p11, or nonspecific isotypematched antibodies as a control. Samples were separated by 15% SDSpolyacrylamide gel electrophoresis under reducing conditions. The relative amounts of MRP8 and MRP14 monomers were determined densitometrically by scanning of autoradiography bands using a Fast Scan supplied with Image Quant software (Molecular Dynamics, Sunny Vale, CA).
Immunofluorescence Microscopy-Monocytes cultured for 3 days on fibronectin (Becton Dickinson)-coated Lab-Tec chamber slides (Nunc, Wiesbaden, FRG) were either left untreated or incubated for 4 h with 10 nM PMA. In some experiments 5 g/ml colchicine, 2 g/ml nocodazole, 1 M demecolcine, or 5 g/ml cytochalasin B was added during the last hour of the incubation period. Cells were washed with phosphatebuffered saline, permeabilized by 10 mM Hepes, pH 6.8, 100 mM KCl, 3 mM MgCl 2 , 200 mM saccharose, 1 mM phenylmethylsulfonyl fluoride (CS buffer) containing 0.5% Triton X-100 for 2 min, washed twice in CS buffer for 5 min, subsequently fixed with 3.7% formaldehyde in phosphate-buffered saline for 4 min and methanol for 6 min at Ϫ20°C, and processed for single-or double-labeling immunofluorescence as described earlier (25) using a-MRP8, a-MRP14, mAb 27E10, phalloidin-FITC, mAb TUB 2.1 against ␤-tubulin, and mAb V9 against vimentin as primary antibodies. For double-labeling experiments staining of MRP8 or MRP14 with a polyclonal rabbit serum was followed by detection of cytoskeletal components employing mAbs from mice. No cross-reactivity or spillover was detected in control experiments after omitting specific antibodies or replacing them by isotype-matched control antibodies of irrelevant specifity. Fluorescence stainings were analyzed on a Zeiss photomicroscope.
Northern Blot Analysis-Total RNA of monocytes was prepared according to the guanidine hydrochloride method (13). Filters were hybridized with cDNA probes specific for MRP8 or MRP14 which were labeled with 32 P by a random primer method (Multiprime DNA labeling system; Amersham). Membranes were reprobed with cDNAs described above with stripping the blots in between employing 0.1% SDS at 95°C. Autoradiographic bands were quantified by densitometrically scanning. Data obtained from MRP8 or MRP14 mRNA bands were normalized to the corresponding 18 S rRNA bands.
Statistics-The U test according to Mann-Whitney (for values without normal distribution) was performed to determine significant differences in MRP8/MRP14 secretion. Values of p Ͼ 0.05 were considered not to be significant.

Induction of MRP8 and MRP14
Release by Monocyte-stimulating Agents-To study the regulation of MRP8 and MRP14 secretion, monocytes were cultured for 1 day and subsequently stimulated with GM-CSF, IL-1␤, IL-4, IL-6, interferon-␥, or lipopolysaccharide for 4 h. Thereafter, supernatants were analyzed for MRP8/MRP14 content by ELISA. As shown in Fig.  1A, GM-CSF, IL-1␤, and lipopolysaccharide induced significant MRP8/MRP14 release by human monocytes. MRP8 and MRP14 have earlier been demonstrated to assemble to noncovalently linked di-, tri-, and tetraheteromeric complexes (26,27). A sandwich ELISA using a mAb against MRP14 does not provide information about the stoichiometric ratio of heteromeric complexes; therefore, the data presented refer to MRP8/ MRP14 (MRP8/14). The exact molecular ratio of secreted MRP8 and MRP14 was determined by metabolic labeling (see below). To consider nonspecific MRP8 and MRP14 release due to cell lysis, lactate dehydrogenase activity in the supernatant was determined concomitantly. Specific release was then presented as the ratio of MRP:lactate dehydrogenase (nanograms/unit).
To analyze intracellular signaling pathways resulting in MRP8/MRP14 release, monocytes were exposed to PMA, Bt 2 cAMP, choleratoxin, pertussis toxin, or forskolin. ELISA assays revealed a strong induction of MRP8/MRP14 release by PMA. Modulators of intracellular cAMP had no effect (Fig. 1B). PMA effects on MRP8/MRP14 release could be inhibited by the protein kinase inhibitors H-7 (inhibitory concentration (IC) 50 ϭ 8 M), HA-1004 (IC 50 ϭ 18 M), and staurosporine (IC 50 ϭ 2 nM); IC 50 of these inhibitors resembled their K i regarding protein kinase C, thus confirming involvement of the latter (Fig.  2A). The PMA analogue 4aPMA which does not exhibit intrinsic activity regarding protein kinase C activation (28) had no effect on MRP8/MRP14 release even at 10-fold higher concen-trations (Fig. 2B). dPPA, an agonist of the protein kinase C isotype ␤ 1 (29), the most abundant protein kinase C isoform in monocytes (30), did not affect MRP8/MRP14 release, indicating a protein kinase C subtype-specific pathway of intracellular signaling. PMA-and cytokine-induced release of MRP8/MRP14 was not associated with translocation of the latter to the cell surface as determined by flow cytometry (data not shown).
Identification of MRP8 and MRP14 Subunits in Monocyte Supernatants-To study potential complex assembly of released MRP8/MRP14, supernatants of monocytes exposed to [ 35 S]methionine and stimulated with PMA were analyzed by immunoprecipitation, subsequent SDS-polyacrylamide gel electrophoresis, and autoradiography. Using a-MRP8, a-MRP14, or mAb 27E10 against the complex of MRP8/MRP14, all antibodies precipitated a similar pattern of MRP8 and MRP14, indicating that complexes of both represent the predominant extracellular form (Fig. 3, A-D). Quantification of precipitated MRP8 and MRP14 by densitometrical scanning revealed a ratio of MRP8:MRP14 of 1:3 which reflects the relative methionine content of both proteins (2 in MRP8 and 6 in MRP14). Furthermore, there was no difference between the intracellular and extracellular MRP8:MRP14 ratio, indicating that both are released at similar rates. To exclude nonspecific release of MRP8 and MRP14 into the supernatant, parallel immunoprecipitation experiments were performed for p11, another member of the S100 family expressed by myelomonocytic cells (4). As demonstrated in Fig. 3E, p11 expression was induced by exposure to PMA. In contrast to the strong [ 35 S]methionine incorporation into intracellular p11, no p11 could be detected in the supernatant of PMA-treated monocytes. This observation adds to the evidence that PMA-stimulated cells were viable and that MRP8 and MRP14 were selectively released.
Molecular Mechanisms of MRP8/MRP14 Secretion-In the next set of experiments we analyzed molecular mechanisms involved in PMA-induced MRP8/MRP14 release. To evaluate dependence of PMA-induced MRP8/MRP14 secretion on de novo protein synthesis, monocytes were exposed to cycloheximide. Treatment with this protein synthesis inhibitor did not affect the amount of MRP8 and MRP14 in the supernatant ( Fig. 4A). In contrast, concomitant incubation with PMA and either one of the two inhibitors of cellular energy metabolism, dinitrophenol or CCCP, led to a significant inhibition of MRP8/ MRP14 release, indicating MRP secretion as being an energydependent active process (Fig. 4A). The intracellular route of MRP8 and MRP14 is independent from the classical ER/Golgi pathway as demonstrated by the inability of monensin and brefeldin A to inhibit PMA-induced release of these proteins (Fig. 4A). Inhibition of tubulin polymerization by nocodazole, colchicine, or demecolcine, however, significantly suppressed release of MRP8/MRP14 into the supernatants (Fig. 4B). In contrast, perturbation of the actin filament system by cytochalasin B had no effect (Fig. 4B). To confirm the latter observations, we performed experiments using immunofluorescence microscopy. In nonstimulated cells, MRP8 and MRP14 show a diffuse staining pattern, whereas treatment with PMA resulted in a filamentous MRP8/MRP14 distribution, which resembled the tubulin network (Fig. 5, A, B, and D). As a control, a mAb against the intermediate filament vimentin was employed. Vimentin generally shows a thinner filamentous network that was most prominent perinuclear but sparse at submembraneous areas (Fig. 5C). In contrast, antibodies against tubulin as well as against MRP8 and MRP14 stained cytoskeletal filaments of larger diameter, which extended out to the cell periphery. Actin filaments were visualized by phalloidin-FITC and showed a completely different pattern (data not shown). Furthermore, double-labeling experiments were performed that revealed a clear co-localization of MRP8 and MRP14 with the tubulin network in PMA-treated monocytes (Fig. 5, E and  F). Double-labeling with vimentin revealed that the cytoskeletal network stained by a-MRP8 or a-MRP14 was stronger and clearly more extended to the cell periphery (Fig. 5, G and H). Depolymerization of microtubules by demecolcine (Fig. 6, A-C), nocodazole or colchicine (data not shown) resulted in a completely different staining pattern: MRP8 and MRP14 were then found to be diffusely dispersed over all the cytoplasm (Fig. 6A), whereas tubulin was focally condensed (Fig. 6B), implying that MRP8 and MRP14 bound preferentially to filamentous microtubules. The intermediate filament system was moderately affected by this treatment, leading to a more condensed perinu-clear pattern, but still presented as a clear filamentous network (Fig. 6C).

Effects of Brefeldin A, Monensin, and Microtubule-depolymerizing Drugs on Secretion of IL-1␤ and TNF-␣-
We then compared the mode of MRP8/MRP14 release with that of IL-1␤ and that of TNF-␣, representatives of an alternative and the classical pathways of secretion, respectively. Brefeldin A and monensin, known to block the vesicular traffic at the ER and Golgi level, significantly inhibited PMA-induced TNF-␣ secretion, whereas IL-1␤ release was up-regulated or unaffected, respectively (Fig. 7). Concomitant treatment with microtubuledepolymerizing agents, such as demecolcine and nocodazole, resulted in a decrease of PMA-induced TNF-␣ release. In contrast, IL-1␤ release was slightly increased by these agents. Thus, MRP8 and MRP14 secretion is different from both the IL-1-like alternative and from the classical pathways of secretion (Fig. 7).
Correlation of MRP8/MRP14 Secretion and mRNA Expression-Employing Northern blot technique, we analyzed regu- Monocytes either nontreated (blank columns) or exposed to 10 nM PMA (hatched columns) were concomitantly incubated with A, 10 g/ml cycloheximide (CX), 1 mM dinitrophenol (DNP), 10 M CCCP, 10 mg/ml monensin (Mon), or 0.5 g/ml brefeldin A (Bre A), or B, 2 g/ml nocodazole, 5 g/ml colchicine, 1 M demecolcine, 5 g/ml cytochalasin B (Cyt B), or medium as a control. Secreted MRP8/MRP14 was determined by ELISA. Data of five independent experiments are presented as described in Fig. 1A. lation of MRP8 and MRP14 expression at the mRNA level. Monocytes cultured for 1 day were incubated for 4 h with 10 nM PMA or medium as control. Activation of protein kinase C led to down-regulation of both MRP8 and MRP14 mRNA. The time course of MRP8 and MRP14 accumulation into the supernatant was closely paralleled by down-regulation of MRP8 and MRP14 mRNA (Fig. 8A). Accordingly, an inverse concentration-response relationship regarding MRP8/MRP14 release and mRNA expression was observed over a range of 1 to 100 nM PMA (Fig. 8B).

DISCUSSION
Earlier studies reported different extracellular functions of S100 proteins but did not provide any information on the mechanisms of their release (4,5,31). This is quite remarkable since neither MRP8 and MRP14 nor any other member of the S100 family comprise signal sequences that would determine their secretion via the classical ER/Golgi route.
We now demonstrate that MRP8 and MRP14 are released from monocytes during inflammatory activation via a novel secretory pathway. Mechanisms leading to MRP8/MRP14 release were shown to be energy-dependent and to involve protein kinase C activation. Employing metabolic labeling, both MRP8 and MRP14 were demonstrated to be secreted at similar rates and as complexes.
Potential nonspecific release of cytoplasmatic MRP8/MRP14, secondary to toxic effects of PMA, could be excluded at several levels. (i) Cell viability did not change significantly during the experimental procedures. (ii) Concomitant treatment with PMA and several potentially toxic agents inhibited release of MRP8 and MRP14, which is incompatible with a mere nonspecific release due to toxic cell damage. (iii) p11, another member of the S100 family, did not appear in the supernatant, despite intracellular up-regulation after PMA treatment.
The blockade of vesicular traffic through the ER and Golgi did not affect MRP8 and MRP14 release, thus ruling out involvement of the classical secretory pathway (21). Secretion is dependent on an intact microtubule network since disruption by depolymerizing agents inhibited MRP8 and MRP14 release. The latter observation is in accordance to morphological data demonstrating a clear co-localization of MRP8 and MRP14 with microtubules during the process of secretion.
Properties of MRP8 and MRP14 during secretion resemble in some aspects those of IL-1␤, which is supposed to be released via an alternative pathway of secretion (21). IL-1␤ secretion is not inhibited by monensin or brefeldin A as well (32). Moreover, there is no association of IL-1␤ with ER, Golgi apparatus, or secretory vesicles, whereas a co-localization with the microtubule network during the secretory process has been reported (33)(34)(35). However, secretion of IL-1␤ is not inhibited by microtubule-depolymerizing drugs (36) (Fig. 7), which is in contrast to our data for MRP8 and MRP14. Furthermore, uncouplers of oxidative phosphorylation increased levels of secreted IL-1␤ (32), whereas MRP8 and MRP14 release was significantly inhibited by CCCP and dinitrophenol. Thus, IL-1␤ and MRP8 and MRP14 display clear differences, suggesting that they do not share a common mechanism of release. MRP8 and MRP14 therefore appear to follow neither the classical nor the alternative secretory pathway of the IL-1␤-type.
Earlier reports described elevated serum levels of MRP8 and MRP14 during the course of a number of inflammatory diseases (2,16,17). Immunohistological data provided indirect evidence that monocytes release MRP8 and MRP14 during endothelial diapedesis at sites of inflammation (11). The complex of MRP8 and MRP14 shows antimicrobial activities, especially against Candida albicans (18). The MRP14 subunit seems to be responsible for this antimicrobial effect (37). Furthermore, MRP8⅐MRP14 complexes exhibit growth-inhibitory activities against murine bone marrow cells, macrophages, and mitogen-stimulated lymphocytes (38), which appears to depend on inhibition of casein kinase II (19). In addition, murine MRP8, but not MRP14, shows chemotactic activity for granulocytes (20). Another recently reported function of MRP8 and MRP14 refers to an antiinflammatory property; systemic application of MRP8/MRP14 mitigated the course of murine experimental arthritis (39). This picture of pleiotropic extracellular activities may reflect different functions of monomeric and complexed MRP8 and MRP14. Such a hypothesis is supported by the observation that MRP8 and MRP14 are differentially expressed in defined monocyte subpopulations under various inflammatory conditions (15).
Most S100 family proteins appear to play an intracellular role during calcium-dependent signaling (4,5). They interfere with cell cycle progression, inhibit phosphorylation reactions, or modulate membrane/cytoskeleton interactions. MRP8 and MRP14 are supposed to be involved in intracellular signaling pathways during calcium-dependent activation of monocytes. They assemble to noncovalently associated complexes (26,27) that are translocated to membrane structures and intermediate filaments upon elevation of intracellular calcium levels by calcium ionophore A23187 (22). The latter event correlated with inflammatory activation of monocytes and neutrophils, thus implicating a role of these proteins for membrane/cytoskeleton interactions (22,23,40). Secretion into the supernatant, however, did not coincide with the preceding translocation of MRP8/MRP14 to the cell membrane, indicating that these phenomena are independent events.
Other members of the S100 protein family exhibit both intraand extracellular functions as well. For example, S100␤ interferes with calcium-dependent modulation of cytoskeletal structures (41,42), but also functions as neurite extension factor in the extracellular space (6,43,44). S100␤ furthermore increases intracellular calcium levels and up-regulates protooncogene expression (31) and nitric oxide synthetase activity in neuronal cells (45). Thus, activities of distinct S100 proteins are not restricted to either the intra-or extracellular space. Structural properties of S100 proteins support such observations. Their N-terminal EF-hand exhibits a significantly lower affinity to calcium than the C-terminal EF domain. It has therefore been supposed that the N-terminal EF-hand binds calcium only at high calcium concentrations as they predominate in the extracellular environment (5).
Induction of MRP8 and MRP14 release is associated with down-regulation of de novo synthesis of these proteins at the mRNA level, thus limiting an extracellular function of these proteins to distinct stages of inflammatory reactions. Furthermore, secretion of MRP8/MRP14 is linked to a marked differentiation step in monocytes, since MRP8 and MRP14 account for up to 30% of the calcium-binding capacity of EF-hand proteins in these cells, whereas both molecules cannot be found after down-regulation of de novo synthesis in mature macrophages (13,46).
To date, extracellular functions have been ascribed to five members of the S100 family. We now for the first time provide data elucidating the mechanism of release of two of these proteins, MRP8 and MRP14, which follow neither the classical nor the IL-1␤-like alternative pathway of secretion. Whether such a novel route may also account for release of other S100 family members remains to be elucidated in future studies.