TNF-α-Converting Enzyme Cleaves the Macrophage Colony-Stimulating Factor Receptor in Macrophages Undergoing Activation¹

The M-CSF (or CSF1) controls the survival, proliferation, and differentiation of mononuclear phagocytes (1). Macrophage functions at a site of inflammation are apparently regulated by the balance between the responses to M-CSF and to macrophage activators. We previously showed that in murine macrophages various types of macrophage activators, including LPS, IL-2, and IL-4, rapidly down-modulate M-CSFR from cell surface, and that this transmodulation is mediated via a common mechanism involving phospholipase C (PLC)³ and protein kinase C (PKC). We also showed that no quantitative relationship exists between the down-modulation of M-CSFR and the effects of activators on the M-CSF-dependent macrophage proliferation, suggesting that the down-modulation regulates macrophage function, rather than controlling proliferation (2, 3, 4). The mechanism of M-CSFR down-modulation downstream the involvement of PLC and PKC remained to be elucidated.

The treatment of NIH-3T3 murine fibroblasts, ectopically expressing M-CSFR, with the PKC activator tetradecanoylphorbol myristate acetate (TPA) was shown to determine the release into culture medium of a portion of M-CSFR that maintains its ligand-binding ability (5). This fact, taken together with our findings, led us to hypothesize that macrophage activators recruit, via intracellular PLC/PKC-dependent signals, an extracellularly active endoprotease that cleaves cell surface M-CSFR (4). A PKC-mediated ectodomain shedding was, in fact, shown for a number of cell surface proteins, including pro-TNF-α (TNF), pro-TGF-α, prostem cell factor, CD14, CD16, CD43, CD44, and L-selectin, and the receptors for TNF and IL-6 (6, 7, 8). However, as the cleavage sites of these proteins share only little, if any, sequence similarity, they are believed to be shed by a number of different proteases. In particular, no information was available about the identity of the protease responsible for the down-modulation of M-CSFR in macrophages undergoing activation.

We report in this study a stepwise approach to the identification of M-CSFR-cleaving protease in macrophages undergoing activation. Using a murine macrophage cell line, we found that the enzyme responsible for the down-modulation of M-CSFR by either LPS or TPA is a transmembrane metalloprotease. M-CSFR cleavage was abolished by pretreating cells in vivo with an Ab directed against the extracellular, catalytic domain of pro-TNF-converting enzyme (TACE). Monocytes obtained from TACE-negative mice were unable to down-modulate M-CSFR in response to TPA. These data indicated that TACE-dependent cleavage is the mechanism responsible for M-CSFR down-modulation in activated macrophages.

The BAC-1.2F5 (BAC) cell line was derived from murine (BALB/c × A.CA)F₁ adherent spleen cells immortalized by transfection with replication-deficient SV40 DNA. BAC cells are strictly dependent on M-CSF for survival and proliferation in culture and retain a large number of the phenotypical and functional characters of normal macrophages (9). BAC cells therefore represent a homogeneous population, free of cells usually contaminating primary macrophage preparations and potential source of interfering cytokines, and were used for all previous studies on M-CSFR down-modulation in macrophages (2, 3, 4). BAC cells were cultured in DMEM (EuroClone, Cramlington, Northumberland, U.K.) supplemented with 2 mM glutamine, 10% heat-inactivated FBS (EuroClone; catalogue ECS-0180), and murine rM-CSF (6 ng/ml). M-CSF was bacteria expressed, HIS-TAG conjugated, and affinity purified on Ni²⁺-NTA-agarose columns (Qiagen, Hilden, Germany). Incubation was conducted at 37°C in 95% air, 5% CO₂, water-saturated atmosphere. Culture medium was completely renewed every 2 days. BAC cells were passaged by scraping at subconfluency (80% saturated growth surface, i.e., approximately 5 × 10⁴ cells/cm²), corresponding to a density of 3 × 10⁵ cells/ml.

The Dexter-ras-myc (DRM) cell line derived from long-term cultures established with bone marrow cells of (C57BL/6 × 129)F₁ mice, following immortalization of nonadherent cells by infection with a ras/myc-encoding retrovirus. DRM cells are strictly dependent upon GM-CSF for growth in culture and exhibit a monocytic cell surface phenotype (10). DRM cells, homozygous for a TACE mutation (TACE^ΔZn/ΔZn) which deletes the Zn-binding domain, thereby abolishing metalloproteinase activity (TACE-negative DRM cells), were obtained from chimeric mice generated with C57BL/6 cells and 129-derived TACE-negative cells as above, except for the treatment of long-term cultures with G418 to select against TACE^+/+ cells (10). Wild-type (TACE^+/+) and TACE-negative DRM cells were cultured in RPMI 1640 (EuroClone), supplemented with 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 μM 2-ME, 10% FBS, 0.5 ng/ml murine rGM-CSF, and 10 ng/ml murine rIL-3 (PeproTech, Rocky Hill, NJ; catalogue 315-03 and 213-13, respectively) (11). Incubation was conducted as above. DRM cells are loosely adherent and were passaged by gentle scraping, after reaching a density of approximately 2 × 10⁶ cells/ml.

Subconfluent BAC cell cultures were incubated for 16 h in DMEM supplemented with 10% FBS, in the absence of M-CSF, to up-regulate M-CSFR expression. Cells were then treated for 60 min with 1 μM TPA (Sigma, St. Louis, MO; catalogue P-8139) or 10 ng/ml LPS (from Escherichia coli; Sigma; catalogue L-4391). In some experiments, cell monolayers were pretreated with an acid buffer (125 mM NaCl, 50 mM glycine-HCl, pH 3) or with a neutral buffer (same composition, pH 7), washed three times with PBS, and then stimulated for 60 min with 1 μM TPA in FBS-free DMEM. At the end of treatments, cell monolayers were washed three times with PBS at 0–2°C, and cells were removed by scraping in PBS (1 ml/plate), transferred into Eppendorf tubes, centrifuged, and lysed in 50 μl/tube of an ice-cold buffer (pH 7.1) containing 10 mM Tris-HCl, 50 mM NaCl, 1% Triton X-100, 5 mM disodium EDTA, 1 mM PMSF (Merck, Darmstadt, Germany; catalogue 7349), 10 μg/ml N-tosyl-l-phenylalanine chloromethyl ketone (TPCK) (Sigma; catalogue T-4376), 0.1 U/ml aprotinin (Sigma; catalogue A-1153), and 4 μg/ml pepstatin-A (Sigma; catalogue P-4265). Insoluble material was removed by centrifugation (18,000 × g, 30 min, 4°C), and the protein concentration of supernatants was determined by the Bradford method (12).

1,10-Phenanthroline, 5 mM (Sigma; catalogue P-9375); IC-3, TNF protease inhibitor, 200 μM (Immunex, Seattle, WA); BB3103, 10 μM (chosen as a representative of a group including BB2116, BB2275, and BB2284; British Biotech, Oxford, U.K.); EDTA, 10 mM (Sigma; catalogue E-1644); aprotinin, 0.1 U/ml; PMSF, 1 mM; calpain inhibitor-I, 20 μM (Boehringer Mannheim, Mannheim, Germany; catalogue 1086 090); pepstatin-A, 6 μM; α₂-macroglobulin, 7.5 mg/ml (Boehringer Mannheim; catalogue 602 442); dec-RVKR-cmk, 50 μM (Alexis, Läufelfingen, Switzerland; catalogue 260-022-M001); phalloidin, 10 μM (Sigma; catalogue P-2141); phosphoinositide-specific PLC (PI-PLC), 100 μmol/ml (Boehringer Mannheim; catalogue 1 143 069).

Anti-murine M-CSFR rabbit polyclonal Ab (anti-M-CSFR), raised against the bacteria-expressed, affinity-purified extracellular domain of murine M-CSFR (NH₂-terminal aa 1–311), was kindly provided by Dr. Manuela Baccarini, Institut für Mikrobiologie und Genetik, Wiener Biozentrum (Vienna, Austria). Anti-TACE mouse mAb Tc3-7.49 (anti-TACE), raised against the catalytic domain of rTACE, was kindly provided by Dr. Marcia Moss, GlaxoWellcome (Durham, NC) (13). Anti-TNF mouse mAb (anti-TNF; PeproTech; catalogue 500-P64), which was effective in inhibiting TNF activity, was assessed by an HLA induction expression assay in murine melanoma cells (14).

Subconfluent BAC cell cultures were incubated for 16 h in M-CSF-free medium (see above), washed three times with sterile PBS, and then treated for 30 min with 1 μM TPA in DMEM. Culture medium (about 3 ml/dish) was recovered, and 5 mM EDTA, 1 mM PMSF, and 10 μg/ml TPCK were added before centrifugation to remove cells and dialysis through a 10-kDa cutoff membrane in 1 L of Tris-HCl, 500 μM (pH 7.4; two changes of buffer in 24 h). Dialyzed medium was lyophilized and then dissolved in 50 μl of cell lysis buffer (see above), and the protein concentration was determined by the Bradford method. Samples were then subjected to SDS-PAGE and electroblotting, as described above.

Volumes of cell lysate or culture medium containing equal amounts of protein were processed for SDS-PAGE by adding appropriate volumes of a 4-fold concentrated Laemmli buffer and 2-ME solution (100 mM final concentration) and boiling for 7 min. Proteins were separated in 7.6% polyacrylamide, 5-cm-long, 0.75-mm-thick minigels (200 V, 60 min) in a buffer (pH 8.1–8.4) containing 25 mM Tris, 192 mM glycine, 0.5% SDS, and then transferred onto nitrocellulose membranes by electroblotting (100 V, 90 min) in a buffer (pH 8.1–8.4) containing 25 mM Tris, 192 mM glycine, and 10% methanol.

To estimate M-CSFR expression, nitrocellulose membranes were incubated (3 h at room temperature (RT)) in PBS containing 0.1% Tween 20 (TPBS), with the addition of 5% BSA (Sigma; catalogue A-3059) to saturate unspecific protein binding sites. Membranes were then washed in TPBS, incubated (6 h at RT) in a 1/500 dilution of anti-M-CSFR in TPBS containing 5% BSA, washed in TPBS again, and incubated (1 h at 4°C) in TPBS containing 2% BSA and a 1/5000 dilution of a HRP-conjugated anti-rabbit IgG Ab (Sigma; catalogue A-6154). After a final wash in TPBS, membranes were incubated (1 min at RT) in a chemiluminescent reagent (ECL protein detection system; Amersham International, Little Chalfont, U.K.), and HRP-coated protein bands were visualized on Hyperfilm-ECL (Amersham) after a 2- to 5-min exposure. To estimate TACE expression, nitrocellulose membranes were blocked as above, washed in PBS, incubated (overnight at 4°C) in a 1/5 dilution of anti-TACE in PBS containing 5% BSA, washed in PBS again, and incubated (1 h at 4°C) in PBS containing 5% BSA and a 1/1000 dilution of an HRP-conjugated anti-mouse IgG Ab (Sigma; catalogue A-4416). After a final wash in PBS, proteins were revealed by ECL as above.

Cells were treated with biotinylated human M-CSF, followed by streptavidin-conjugated PE (PharMingen, San Diego, CA; catalogue 13025D), and processed as described (15). Cells were counted with a FACScan (Becton Dickinson, Mountain View, CA). M-CSF-untreated/PE-treated cells were used for zeroing the analyzer.

Fig. 1 shows that a short treatment of BAC macrophages with TPA resulted in the disappearance from cell lysates of M-CSFR, as detected by immunoblotting with αM-CSFR, raised against the extracellular portion of murine M-CSFR (Fig. 1,A). This was paralleled by the release into culture medium of a soluble 90- to 100-kDa M-CSFR detectable using the same Ab (Fig. 1 B). Thus, TPA activated an endoprotease cleaving M-CSFR in macrophages, as expected (5).

Macrophages produce a wide range of proteases involved in the breakdown of extracellular matrix molecules. As a first step in the identification of the M-CSFR-cleaving protease, macrophages were treated, before activation, with several classes of protease inhibitors: α₂-macroglobulin (all protease families), PMSF and aprotinin (serinoproteases), calpain-inhibitor-I (cysteinoproteases), pepstatin-A (aspartatoproteases), and the cation chelators EDTA and 1,10-phenanthroline (metalloproteases). Cation chelators prevented TPA or LPS from down-modulating M-CSFR (Fig. 2, A and B). This effect was abolished in the presence of zinc chloride at concentrations saturating ion-chelating capacity (Fig. 2,B). All other families of inhibitors were ineffective (not shown). No protein band other than M-CSFR, cross-reacting with anti-M-CSFR antiserum, was altered by the treatment with TPA or LPS. The effects of competitive inhibitors of metalloproteases such as IC-3 and BB3103 (16, 17, 18) were then tested. These compounds also completely inhibited the TPA- or LPS-induced down-modulation of M-CSFR (Fig. 2, C and D), an effect abolished by an extensive wash of cells before the treatment with the activator (not shown), which indicates that the inhibition was reversible. None of metalloprotease inhibitors used had any effect on basal M-CSFR expression or was toxic for cells (not shown). These results indicated that the M-CSFR-cleaving protease is a metalloprotease.

The location of the M-CSFR-cleaving protease with respect to plasma membrane was then investigated. Cell surface proteins comprise 1) transmembrane, or 2) GPI-anchored endogenous polypeptides, or 3) soluble, cell membrane-associated serum or endogenous polypeptides. We tried first to prevent protease activity by prewashing cell monolayers with a pH 3 isotonic buffer, a procedure that removes membrane-associated, soluble proteins. Cell wash did not interfere with the down-modulation of M-CSFR induced by either TPA (Fig. 3,A) or LPS (not shown) in the absence of serum. To rule out the possibility that the M-CSFR-cleaving protease was an endogenous, soluble protease rapidly secreted following the acid wash and upon activation, cells were treated with phalloidin, an inhibitor of exocytosis (19), before and immediately after the acid wash and then throughout the stimulation with TPA. Phalloidin did not interfere with the down-modulation of M-CSFR, indicating that the M-CSFR-cleaving protease is not a rapidly secreted, soluble protein (Fig. 3,A). On the other hand, the treatment of macrophages with PI-PLC, before and throughout the stimulation with TPA, did not prevent the down-modulation of M-CSFR, indicating that the M-CSFR-cleaving protease is not a GPI-anchored protein (Fig. 3 B). No effect on M-CSFR expression was determined by phalloidin or PI-PLC alone (not shown). The above results, taken together, indicated that the M-CSFR-cleaving protease is a transmembrane metalloprotease.

Transmembrane metalloproteases of both the membrane-type matrix metalloprotease (MT-MMP) and the disintegrin/metalloprotease (ADAM) families (20, 21, 22) are converted to their active forms by intracellular serine endoproteases of the paired basic amino acid-cleaving enzymes (PACE) family (23, 24, 25, 26). Processing by PACE is restricted to pro-proteins containing the consensus sequence Arg-X-Lys/Arg-Arg at the junction between the pro- and catalytic domains, so that peptidylchloromethylketones containing this sequence behave as specific, competitive inhibitors of PACE activity (25). The most effective of these compounds, dec-RVKR-cmk, was therefore tested for its capacity of interfering with the TPA-induced down-modulation of M-CSFR. Fig. 4 shows that in cells pretreated with dec-RVKR-cmk, the effect of TPA was inhibited, indicating that the M-CSFR-cleaving protease requires PACE-dependent processing for its activation. These results confirmed that the M-CSFR-cleaving protease is a transmembrane metalloprotease.

The ADAM transmembrane metalloproteases, differently from the enzymes of the MT-MMP family, exhibit a large (120–130 aa) cytoplasmic portion, which makes the ADAM enzymes a more likely target for signals generated intracellularly by macrophage activators. One of the best-characterized members of the ADAM family is TACE (ADAM-17), responsible for the ectodomain shedding of receptors for TNF and other cytokines (11, 21). Using anti-TACE, which recognizes the extracellular, catalytic domain of TACE, we found that this enzyme is expressed in BAC macrophages and that its expression is inhibited by the pretreatment of cells with dec-RVKR-cmk (Fig. 5,A). When cells were pretreated in vivo with anti-TACE, the down-modulation of M-CSFR induced by TPA was inhibited (Fig. 5 B). The treatment with anti-TACE alone had no effect on M-CSFR expression, nor did the treatment with an unrelated, control Ab interfere with the TPA-induced down-modulation of M-CSFR (not shown). These data indicated that TACE is involved in M-CSFR cleavage.

To more directly demonstrate that TACE is responsible for M-CSFR down-modulation in mononuclear phagocytes undergoing activation, we compared the effects of TPA on M-CSFR expression in wild-type or TACE-negative DRM monocytes (Fig. 6). Flow cytometry using biotinylated M-CSF to label M-CSFR showed that the treatment with TPA markedly reduced the percentage of M-CSF/M-CSFR-positive, wild-type DRM cells. On the contrary, in TACE-negative cells, TPA was completely ineffective, indicating that the TPA-dependent down-modulation of M-CSFR requires TACE expression. To further characterize the importance of this requirement, we addressed the question of whether the involvement of TACE in M-CSFR down-modulation may be indirect, being mediated by TACE-dependent release of TNF from cell surface, a possibility suggested by the previous observation that exogenous TNF rapidly reduces M-CSFR expression (27, 28). The pretreatment of cells with anti-TNF at dilutions (10 μg/ml) preliminarily assayed to block TNF activity in vivo (not shown) did not interfere with the down-modulation of M-CSFR induced by either TPA or LPS (not shown), indicating that the TACE-dependent cleavage of M-CSFR is not mediated by autocrine TNF.

Fig. 6 shows that under basal conditions (i.e., without TPA treatment), wild-type DRM monocytes exhibit a higher variability and lower average values of M-CSFR fluorescence intensity than TACE-negative cells. This feature suggested that in the absence of macrophage activators, low levels of M-CSFR shedding are driven by constitutive TACE activity. This hypothesis was tested by treating BAC macrophages, in the absence of macrophage activators, with BB3103 or IC-3 at the same concentrations used above. Fig. 7 shows that the inhibitors enhanced basal M-CSFR expression (by 2.15-fold, on the average, as determined by band densitometry of immunoblots). This enhancement is apparently due to the block of TACE activity, and of the related basal M-CSFR shedding shown in Fig. 1,B (left lane). Taken together, the data shown in Figs. 1, 6, and 7 demonstrated an either constitutive or macrophage activator-inducible TACE-dependent shedding of M-CSFR in mononuclear phagocytes.

The results reported in this work provide strong evidence that TACE is responsible for the down-modulation of M-CSFR in macrophages undergoing activation. At the beginning of this study, no information was available on the protease responsible for M-CSFR shedding. A stepwise approach to the identification of this protease was therefore undertaken. We demonstrated first that the M-CSFR-cleaving protease is a metalloprotease, showing that M-CSFR down-modulation is inhibited not only by cation chelators, but also by hydroxamic acid-derived, competitive inhibitors of metalloproteases (Fig. 2). The latter result ruled out the possibility that chelators were toxic for cells or aspecifically effective due to depletion of intracellular calcium, and led to add M-CSFR to the group of receptors that, following TPA stimulation, are cleaved by a metalloprotease sensitive to hydroxamate inhibitors. The group includes the receptors for TNF (p55–60 and p75–80), IL-6, IL-2, nerve growth factor (low affinity), and thyroid-stimulating hormone (8, 29, 30, 31). These receptors are cleaved at an extracellular site, a fact in keeping with the release of soluble M-CSFR in culture medium (Fig. 1,B). The sensitivity of M-CSFR shedding to metalloprotease inhibitors, however, could not be considered a demonstration of the extracellular nature of the enzyme catalytic activity, as all the inhibitors we used are cell diffusible. For the same reason, nothing more could be inferred from the removal of inhibition by an extensive cell wash following the treatment with the inhibitor (Fig. 3,A). We were, on the other hand, able to surely exclude that the M-CSFR-cleaving protease is a soluble membrane-associated or a GPI-anchored protein (Fig. 3,B) and, therefore, to indirectly conclude that the enzyme is a transmembrane protein. This conclusion is supported by the ineffectiveness of 1,10-phenanthroline in the presence of zinc chloride (Fig. 2,B) and by the results of Fig. 1 B, since transmembrane metalloproteases are preferentially zinc metalloproteases (26) and their catalytic domain is extracellular.

To confirm that the M-CSFR-cleaving protease was a transmembrane metalloprotease, we tried to inhibit M-CSFR down-modulation by interfering with the expression of this type of proteases. While enzymatically active, soluble matrix metalloproteases (MMP) are usually generated via the extracellular processing of proenzymes, transmembrane metalloproteases are cleaved to their active form intracellularly, within the secretory pathway, by furin-type proteases of the PACE family (24, 25). The furin-recognition motif (Arg-X-Lys/Arg-Arg), necessary for processing by PACE (26), is present in all transmembrane metalloproteases, but not in MMP, the only exception being stromelysin-3/MMP-11 (24, 26). As the latter enzyme is, like any other cell surface-associated soluble protein, sensitive to the acid wash of cells (Fig. 3,A), the effectiveness of dec-RVKR-cmk in preventing M-CSFR down-modulation (Fig. 4) indicated that the M-CSFR-cleaving protease belongs to either the MT-MMP or the ADAM family of transmembrane metalloproteases.

The facts that the enzymes of the MT-MMP family exhibit a very short cytoplasmic domain, unlikely target of intracellular activatory signals triggered by macrophage activators, and that only one member of this family (MT4-MMP) is expressed in leukocytes (32), directed our attention to the ADAM proteases. The main result obtained using anti-TACE, reactive with the catalytic domain of TACE, was the inhibition of the TPA-induced cleavage of M-CSFR by pretreating cells with the Ab in vivo, apparently due to the block of TACE catalytic activity (Fig. 5,B). These experiments demonstrated that TACE is responsible for M-CSFR cleavage and definitively confirmed that the catalytic site of the enzyme, as well as the cleavage site of M-CSFR, is extracellular. An even stronger indication of the role of TACE in M-CSFR cleavage was obtained by using TACE-negative monocytes, in which TPA was completely ineffective in down-modulating M-CSFR expression (Fig. 6). An interesting side-result of these experiments was the discovery of a sizeable TACE-dependent shedding of M-CSFR in the absence of macrophage activators (Figs. 1 B, 6, and 7), pointing to a constitutive activity of TACE in controlling M-CSFR density on cell surface. A soluble M-CSFR fragment, including the ligand-binding domain, is likely to behave as a decoy receptor and to interfere with the M-CSF concentration active on cells (33). The physiological role of a constitutive M-CSFR shedding from macrophage surface is worthy of further investigation.

A question we needed to address was whether TACE is directly or indirectly responsible for M-CSFR down-modulation. The fact that TNF had been shown to down-modulate M-CSFR (27, 28) suggested the possibility that an M-CSFR-cleaving protease different from TACE was activated following a TACE-dependent release of TNF. This possibility resulted very unlikely, as an in vivo treatment with αTNF was completely ineffective in preventing M-CSFR shedding. On the other hand, it is worth pointing out that, on the basis of the results of Figs. 2–4, TNF, to determine M-CSFR cleavage, would need to activate in any case a transmembrane metalloprotease, or a cascade including at least one enzyme of this type. Thus, the most straightforward conclusion we could derive from all above is that TACE is the protease cleaving M-CSFR in mononuclear phagocytes undergoing activation. As for the down-modulation of M-CSFR by TNF, its mechanism has not been elucidated yet. Our results are compatible with the hypothesis that TNF actually activate (yet being not the only activator of) TACE, which would, in parallel, cause direct M-CSFR shedding and establish a positive feedback of TNF release. In conclusion, on the basis of all above, M-CSFR can be added to the list of TACE substrates, such as β-amyloid protein precursor and members of the epidermal growth factor receptor family (34, 35, 36).

The conclusion that TACE is responsible for M-CSFR down-modulation was reached by inducing receptor shedding with TPA. Part of the results presented (those of Figs. 2 and 3), however, were obtained with either TPA or LPS, suggesting that the effects of the two compounds, as far as M-CSFR shedding is concerned, are largely overlapping. Indeed, all the physiological macrophage activators tested to date, including LPS, IL-2, and IL-4, trigger M-CSFR down-modulation via the PLC→PKC pathway (2, 3, 4), while TPA interacts with this process downstream, directly activating PKC. Further work is necessary to elucidate how the activation of PKC is linked to that of TACE.

The identification of the protease responsible for M-CSFR shedding in mononuclear phagocytes undergoing activation extends the characterization of M-CSFR down-modulation vs down-regulation. M-CSF-dependent down-regulation is based on the internalization of ligand-receptor complexes in peripheral endosomes and their targeting to esoproteases of the lysosomal system (37), whereas the trans/down-modulation induced by macrophage activators is operated by endoproteases active on cell surface (5). Down-regulation, differently from down-modulation, requires tyrosine kinase-active M-CSFR and is PKC independent (5). Down-regulation involves a progressive decrease of M-CSFR levels in individual cells, synchronous for all cells within a population. On the other hand, down-modulation is driven via an all-or-nothing response of individual cells, progressively extended to the entire population. Once a signal threshold is reached in an individual cell, the response is triggered and all receptors from cell surface are rapidly removed (38). Our results indicated that this signal is represented by TACE activation.

The question of the physiological meaning of the rapid M-CSFR down-modulation induced by macrophage activators requires further investigation. We know that this down-modulation occurs in any sort of mononuclear phagocytes, including monocyte or macrophage cell lines and peritoneal or bone marrow-derived primary macrophages (2 , this study), is triggered by different types of macrophage activators (2, 3, 4), and it regulates macrophage function by blocking some nonproliferative effects of M-CSF (4). M-CSF has been actually shown to inhibit the induction by IFN-γ of class II MHC molecules, thereby suppressing macrophage functions in the immune response (39). On the other hand, macrophages and dendritic cells are capable of converting into one another until the late stages of their respective maturation processes, enabling tissue-resident macrophages to convert to dendritic cells upon appropriate signals (40). It is interesting to note in this work that these signals are mainly represented by IL-4 and GM-CSF, both able to down-modulate M-CSFR (4, 41). We (4) and others (38) hypothesized that the induction of a fully activated (in particular, the APC) phenotype in mononuclear phagocytes requires the rapid block of M-CSF-dependent signals. Investigations on this hypothesis will take advantage of the findings reported in this study, as they will enable to restore, by preventing M-CSFR down-modulation, the sensitivity of macrophages to M-CSF while they respond to the activators.

We thank Professor Massimo Olivotto (Department of Experimental Pathology and Oncology, University of Florence) for moral and material support to this work.