Endocytic Properties of the M-type 180-kDa Receptor for Secretory Phospholipases A 2 *

Endocytic properties of the M-type 180-kDa receptor for secretory phospholipases A 2 (sPLA 2 ) were first inves- tigated in rabbit myocytes that express it at high levels. Internalization of the receptor was shown to be clath- rin-coated pit-mediated, rapid ( k e (cid:53) 0.1 min (cid:50) 1 ), and li-gand-independent. The signal sequence for internaliza- tion was then identified upon transient and stable expression of various receptor constructs with mutated cytoplasmic sequences. Analysis of the internalization efficiency of the mutants suggested that the NSYY motif encodes the major endocytic signal, with the distal tyrosine residue playing the key role. Amino acid substi- tutions at the putative casein kinase II phosphorylation site of the receptor did not affect internalization. A chi- meric protein composed of the extracellular and transmembrane domains of the rabbit sPLA 2 receptor and of the cytoplasmic domain of the structurally homologous human macrophage mannose receptor retained the high affinity for sPLA 2 and was internalization competent, exhibiting 50% endocytic activity of the M-type sPLA 2 receptor. The results indicate the compatibility of the structural domains of the two parent proteins and pro- vide evidence for the interchangeable character of their internalization signals.

Endocytic properties of the M-type 180-kDa receptor for secretory phospholipases A 2 (sPLA 2 ) were first investigated in rabbit myocytes that express it at high levels. Internalization of the receptor was shown to be clathrin-coated pit-mediated, rapid (k e ‫؍‬ 0.1 min ؊1 ), and ligand-independent. The signal sequence for internalization was then identified upon transient and stable expression of various receptor constructs with mutated cytoplasmic sequences. Analysis of the internalization efficiency of the mutants suggested that the NSYY motif encodes the major endocytic signal, with the distal tyrosine residue playing the key role. Amino acid substitutions at the putative casein kinase II phosphorylation site of the receptor did not affect internalization. A chimeric protein composed of the extracellular and transmembrane domains of the rabbit sPLA 2 receptor and of the cytoplasmic domain of the structurally homologous human macrophage mannose receptor retained the high affinity for sPLA 2 and was internalization competent, exhibiting 50% endocytic activity of the M-type sPLA 2 receptor. The results indicate the compatibility of the structural domains of the two parent proteins and provide evidence for the interchangeable character of their internalization signals.
Secretory phospholipases A 2 (sPLA 2 s) 1 are implicated in a number of crucial physiological and pathological processes in mammals (for recent reviews, see Refs. [1][2][3]. The pancreatictype sPLA 2 (group I), besides its well known digestive function, has been implicated in fertilization (4), cell proliferation (5), and contraction of vascular and airway smooth muscles (6,7). Elevated levels of the inflammatory-type sPLA 2 (group II) in serum, bronchial lavage, and synovial fluids are associated with propagation of inflammation and hypersensibilization of the organism (for a recent review, see Ref. 8).
Different types of membrane receptors for sPLA 2 s have been recently identified using various types of venom sPLA 2 s (9 -12). The M-type receptor is a 180-kDa protein that was first identified in rabbit muscle cells in culture (10). It binds some venom sPLA 2 s like OS 1 , one of the sPLA 2 s from the Taipan snake venom with a K d value of 38 pM, but it does not bind bee venom sPLA 2 (10). The N-type receptor, which is abundant in the brain (9) and is present in most tissues (12), does not bind OS 1 , but it associates very tightly with bee venom sPLA 2 (K d ϭ 100 pM). It is made of polypeptide units of about 40 -50 and 85 kDa (9). The M-type receptor binds the pancreatic-type sPLA 2 as well as the inflammatory-type sPLA 2 with K d values of about 1-10 nM (13,14). These sPLA 2 s are probably the normal endogenous ligands of the M-type 180 kDa receptor. Experiments with pancreatic-type sPLA 2 suggested some of its effects, i.e. induction of DNA synthesis and cell proliferation (5) contraction (6), cell migration (15), and eicosanoid production (16) to be mediated by the M-type receptor. However, the intracellular signal transduction mechanisms linking these multiple effects to the event of ligand-receptor interaction still remain obscure.
The high molecular weight M-type sPLA 2 receptors were cloned from rabbit (13), bovine (17), mouse (18), and human (19) tissues. It is a single subunit, type I plasma membrane glycoprotein of 180 -200 kDa. The molecule is made of a large highly glycosylated extracellular N-terminal portion, comprising a N-terminal cystein-rich region, a fibronectin-like type II domain, and a repeat of eight carbohydrate-recognition domains in tandem followed by a single transmembrane domain and a short intracellular C-terminal tail. The overall molecular organization of the M-type receptor is similar to that of the macrophage mannose receptor that mediates the uptake of mannose-glycosylated ligands and the phagocytosis of parasitic microorganisms (20,21) and to the recently discovered DEC-205 membrane protein, which is implicated in antigen processing (22). The structural similarity is especially pronounced in the extracellular and membrane-spanning protein portions, while the C-terminal cytoplasmic domains seem to be more specific for each protein.
The mannose receptor and DEC-205 protein are known to be internalized with high efficiency (22)(23)(24). Endocytosis is intimately linked to their function as integral parts of the defense and clearance systems. Internalization of the M-type sPLA 2 receptor has also been observed (19,25), and it could play an important role in receptor function. The purpose of this work is to analyze in details the endocytic properties of the M-type sPLA 2 receptor by determining its rate of internalization, the utilized endocytic mode (ligand-induced or constitutive), as well as the structural signals promoting the endocytic process.

EXPERIMENTAL PROCEDURES
Materials-OS 1 , one of the Oxyuranus scutellatus scutellatus venom sPLA 2 was prepared and radioactively labeled using Na 125 I (Amersham Corp.) as described previously (10). The plasmid construct containing the full-length human macrophage mannose receptor cDNA was a kind gift of Dr. R. A. B. Ezekowitz (Harvard Medical School, Boston, MA). Guinea pig anti-sPLA 2 receptor antibodies were raised against the purified rabbit M-type sPLA 2 receptor (10). Porcine pancreatic PLA 2 was from Boehringer Mannheim. Oligonucleotide primers were from Genosys Biotechnologies, Inc., Cambridge, United Kingdom. Hybond nitrocellulose (pore size, 0.45 m) was from Amersham Corp.
Plasmid Constructions-Preparation of the rabbit M-type sPLA 2 receptor cDNA in the pRc/CMV vector (Invitrogen) has been described previously (13). M-type sPLA 2 receptor mutants were prepared by mu-tagenesis on the double-stranded plasmid constructs, containing the entire sPLA 2 receptor cDNA, using the Transformer site-directed mutagenesis kit (Clontech). Sequences of mutagenic oligonucleotide primers are listed in Table I. The selection primer (5Ј-GATATCAAG-GTTAACGATACCGTC) was designed to mutate the unique ClaI site in the polylinker region of the plasmid construct to a HpaI site. The efficiency of mutagenesis was typically 10 -20%. The chimeric sPLA 2 receptor/mannose receptor cDNA was prepared using a combination of the "sticky feet" technique (26) and of the Transformer-based sitedirected mutagenesis. The 3Ј cDNA region, coding for the entire cytoplasmic portion of the macrophage mannose receptor (amino acid residues 1410 -1455), was amplified by polymerase chain reaction using the oligonucleotide primers indicated in Table I and the full-length mannose receptor cDNA as a template. The heat-denatured antisense strand of the purified polymerase chain reaction product was then used as a megaprimer in the primer extension reaction on the singlestranded sPLA 2 receptor cDNA plasmid construct. All mutations were verified by DNA sequencing. Plasmid constructs for transfection experiments were purified on Qiagen columns (Qiagen).
Cell Culture and Transfection-Primary cultures of rabbit skeletal muscle cells were prepared and handled as described previously (10). COS cells were grown in Dulbecco's modified Eagle's medium, Life Technologies, Inc., supplemented with 10% fetal calf serum and antibiotics (200 units/ml penicillin, 50 g/ml streptomycin). 1 day before transfection, cells were plated at a density of 7 ϫ 10 5 cells onto 85-mm Petri dishes. The cells were transfected by a modification of the DEAEdextran/chloroquine method (27) using 30 g of supercoiled DNA per plate. 16 h after transfection, the cells were trypsinized and subcultured into 24-or 6-well cluster plates. The expression of the constructs was analyzed 72 h after transfection.
For stable expression, HEK 293 cells (ATCC) were transfected with mutant and wild-type sPLA 2 receptor cDNA plasmid constructs using the calcium phosphate precipitation method (28). Cell clones were selected based on acquired resistance to G418 (1 mg/ml final concentration). Levels of M-type sPLA 2 receptor expression were estimated by Scatchard plot analysis of sPLA 2 binding to cell membrane preparations. When used in ligand binding and internalization experiments, HEK 293 cells were cultured on polylysine-precoated plates.
Receptor Endocytosis-Internalization was followed using a method described by Wiley and Cunningham (29) previously. Confluent COS cells in 24-well plates were switched to serum-free medium containing 20 mM Hepes, pH 7.4, 0.1% bovine serum albumin (this medium will hereafter be called DHB) at least 3 h before measurements. Internalization was initiated by addition of 125 I-OS 1 (60 pM in DHB) at indicated time points. At the end of the incubation, the plates were rapidly transferred on ice and rinsed 4 times with ice-cold DHB. Cell surfacebound radioactivity was stripped off by treatment with 1 ml/well of ice-cold 50 mM glycine-HCl medium at pH 3.0, containing 100 mM NaCl, 2 mg/ml polyvinylpyrrolidone, 2 M urea for 10 min (30). Stripping efficiencies for the 125 I-OS 1 were about 95%. The intracellular radioactivity was assessed by dissolving the acid-rinsed cells in 1 N NaOH (1 ml/well) for 10 min at 37°C. Nonspecific binding (typically less than 5% total binding) was determined using mock-transfected COS cells and subtracted. All experimental determinations were performed at least in triplicate. Internalization experiments on cells in hypertonic medium were conducted similarly, except that the cells were pretreated (30 min, 37°C) and further processed in DHB, containing 0.45 M sucrose. The data were analyzed using the plots representing a ratio of internalized to surface bound ligand as a function of time (29). The internalization rate constant (k e ) was determined by linear regression analysis. The correlation coefficients were generally Ͼ0.98.
Binding Experiments and Membrane Preparations-Binding assays were performed as described previously (10) using 125 I-OS 1 as a ligand. Nonspecific binding in control samples was estimated in the presence of an excess of unlabeled OS 1 (100 nM). Binding on cultured cells was assessed after 3 h of incubation with the ligand at 4°C. Binding on total cell lysates was performed at 20°C after 90 min of incubation with 125 I-OS 1 . Exposure of confluent rabbit skeletal muscle cells to 100 nM porcine pancreatic sPLA 2 was performed for indicated time intervals at 37°C. The plates were then quickly transferred on ice, the incubation medium was withdrawn, and the cells were rinsed 4 times with ice-cold ligand-free DHB prior to 125 I-OS 1 addition in DHB. In experiments concerning ligand-free receptor internalization, rabbit skeletal muscle cells were pretreated with 10 M monensin (31) in complete medium for 2 h at 37°C. Monensin was then included in all experimental solutions.
Cell lysates were prepared by sonicating cell suspensions in 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride (lysis buffer) as described previously (10). Membrane fractions were prepared by several cycles of freezing, thawing, and centrifugation of cells in lysis buffer (32). Protein concentration was determined by the Bio-Rad adaptation of Bradford's dye-binding assay.
Immunoblot Analyses-Protein samples were analyzed by SDS-polyacrylamide gel electrophoresis. After electrophoresis, proteins were transferred to nitrocellulose membranes for 1 h in 25 mM Tris, 0.2 M glycine buffer at 300 mA. The blots were probed with guinea pig anti-PLA 2 receptor immune serum (1:5,000), and a secondary goat antiguinea pig antibody (1:10,000) was conjugated to horseradish peroxidase (Cappel, Organon Teknika). The immune complexes were visualized using an enhanced luminescence Western blotting system (ECL, Amersham Corp.). Filters were exposed to Kodak XAR films.
Indirect Immunofluorescence Microscopy-This was performed essentially as described previously (32). Guinea pig anti-sPLA 2 receptor immune serum (1:200 dilution) and fluorescein isothiocyanate-conjugated goat anti-guinea pig immunoglobulins, whole molecules (1:100, Sigma) were used for intracellular localization of receptor molecules. The cells were mounted in Mowiol 4 -88 (Calbiochem-Novabiochem Corp) and viewed in a Leitz Aristoplan microscope (Wild Leitz) using an interference blue (fluorescein isothiocyanate) filter and a 40ϫ or oilimmersion 100ϫ lens. 2 Receptor-The expression level and internalization properties of the endogenous M-type sPLA 2 receptor were assessed on the primary culture of confluent rabbit skeletal muscle cells. The radioactively labeled ligand used in binding and internalization experiments was a snake venom sPLA 2 , OS 1 (10). The maximal binding capacity value (B max ) estimated on total cell lysates was 0.35 pmol/mg of protein, while the number of binding sites expressed on the cell surface was 0.12 pmol/mg of protein (data not shown). The K d value for the OS 1 -receptor complex was 30 -50 pM as reported previously (10). Fig. 1 shows the results of a Western blotting of the rabbit skeletal muscle cells membrane proteins and the indirect immunofluorescence microscopy of cells using anti-sPLA 2 receptor antibodies. In the blots, the antibodies recognized a single protein band corresponding to a molecular mass of about 180 kDa. This value is in a good agreement with earlier estimations of the molecular mass of the mature highly glycosylated M-type sPLA 2 receptor (10,11). In immunofluorescence experiments, the antibodies revealed an intensive cell-surface staining typical for protein expressed at the plasma membrane. Fig. 2A shows the kinetics of 125 I-OS 1 internalization during the first 15 min after ligand addition. The internalization rate constant (k e ) was 0.09 Ϯ 0.005 min Ϫ1 (n ϭ 5). Exposure of cells to a hypertonic medium causes disassembly of membrane-associated clathrin lattices and, thus, abolishes internalization mediated by clathrin-coated pits (33). Pretreatment of rabbit skeletal muscle cells with a hypertonic sucrose-containing me-TABLE I Synthetic oligonucleotides used for mutagenesis Mutated nucleotides are in boldface. a) and b), coding and anti-sense "chimeric" PCR primers encompassing the regions of complementarity to the sPLA 2 receptor cDNA (underlined) and to the macrophage mannose receptor cDNA (italics).

Mutant
Mutagenic oligonucleotide dium resulted in a complete blockade of the receptor-mediated 125 I-OS 1 internalization ( Fig. 2A). The observed blockade was obviously not due to the modified ligand binding properties of the receptor since the levels of total cell-associated radioactivity were the same in control and sucrose-treated samples (not shown). These results suggest that endocytosis of the sPLA 2 receptor occurs via clathrin-coated pits.
The effect of the long term exposure to the ligand was investigated on cells pretreated for 30 min, 1, 2, and 24 h with 100 nM porcine pancreatic sPLA 2 (K d ϭ 10 nM) (13). Following preincubation, the cells were chilled on ice, washed with ligand-free medium, and subjected to binding analysis using 125 I-OS 1 (K d ϭ 38 pM (10) to evaluate the number of remaining sPLA 2 binding sites. The large difference in affinity of these respective ligands for the M-type receptor ensured the quantitative substitution of the pancreatic enzyme from the cell surface exposed ligand-receptor complex by OS 1 . At all time points tested, the number of cell-surface ligand binding sites in the sPLA 2 -pretreated samples was 0.1-0.13 pmol/mg of total protein, essentially the same as in controls (data not shown). The results, thus, argue against receptor sequestration and/or down-regulation upon prolonged exposure to the pancreatic enzyme.
The possibility of a sequential uptake and recycling of the unoccupied M-type sPLA 2 receptor was analyzed by comparing the quantity of the cell-surface PLA 2 binding sites in controls and in cells pretreated with a carboxylic ionophore monensin, an inhibitor of trans-Golgi endosomal trafficking (34). Fig. 2B shows that monensin-pretreatment caused a substantial, up to 57%, decrease in the number of M-type sPLA 2 receptor molecules exposed at the cell surface. The affinity for the ligand and the B max values estimated on total cell lysates were unaffected by the inhibitor (not shown), ruling out any nonspecific inhibitory or modifying effect of monensin on the sPLA 2 binding.
Mapping of the Encoded Internalization Signals-The signals governing membrane receptor internalization are typically located in the cytoplasmic portions of proteins. The 42amino acid sequence of the cytoplasmically exposed C-terminal tail of the rabbit M-type sPLA 2 receptor is presented in Fig. 3. Visual inspection of the sequence and analysis of its secondary structure predicted by Chou-Fasman algorithm outlined three peptide regions as potential internalization signals. The first region, NKGFF (residues 4 -8 in the cytoplasmic tail) is predicted to have a ␤-turn conformation and resembles the QQGFF internalization motif of the epidermal growth factor receptor (35). The second region, NSYY (residues 16 -19), also has a high propensity for ␤-turn formation and is closely related to the canonical NPXY consensus sequence found in a number of endocytic receptors (36,37). The third site located in the distal part of the tail (residues 34 -35) is represented by a LI dipeptide. Of these three regions, the two latter are highly conserved in M-type sPLA 2 receptors from human, bovine,  (13,(17)(18)(19). The identical amino acid residues are shaded. The cytoplasmic sequences of the rabbit, human, and bovine M-type sPLA 2 receptors are given in full. *, the cytoplasmic portion of the mouse receptor has 15 additional C-terminal residues that are not shown. rabbit, and mouse (Fig. 3D) and were then assumed to be the most plausible candidates for directing internalization.
Three mutants were first designed for a preliminary mapping of endocytic signals (Fig. 3A). The mutant TR2 lacks all but two residues of the cytoplasmic portion. The extramembrane dipeptide was left in the sequence to ensure the anchoring of the expressed protein in the lipid bilayer. In the mutant TR11, the carboxyl-terminal truncation was introduced three amino acid residues downstream from the putative internalization signal NKGFF. The deletion mutant ⌬16 -22 lacks the peptide sequence NSYYPTT of the receptor tail.
Wild-type and mutant M-type sPLA 2 receptors were transiently expressed in COS cells. Cell surface expression of each construct was analyzed by indirect immunofluorescence microscopy (Fig. 4). All constructs exhibited an evenly dispersed pattern of staining with high fluorescence intensity at cell borders characteristic for a plasma membrane-delivered protein. Western blotting analysis of membrane proteins (Fig. 5A) revealed similar levels of receptor expression for all constructs. The receptors were recognized by antibodies as single protein bands of the expected electrophoretic mobilities (apparent molecular mass, 170 -180 kDa). No sign for proteolytic degradation of the expressed mutants was detected. The expression level of receptor constructs (B max ) estimated on total cell ly-sates was typically 1-2 pmol/mg of protein, while control COS cells had no OS 1 binding sites (not shown). The affinity of 125 I-OS 1 for the different mutant receptors was essentially the same as that of the wild-type receptor (K d ϭ 40 Ϯ 10 pM) and of the endogenous receptor in rabbit skeletal muscle cells. Taken together, all of these results indicate that deletions in the receptor cytoplasmic domain do not affect the plasma membrane delivery and the binding properties of the M-type sPLA 2 receptor. Fig. 5B represents a diagram of specific internalization rate constants (k e ) for the wild-type and mutant M-type sPLA 2 receptors. The 125 I-OS 1 internalization by the wild-type receptor was 0.02 Ϯ 0.003 min Ϫ1 . The ⌬16 -22 mutant internalized the ligand with a 33% efficiency as compared with the wild-type receptor (k e ϭ 0.006 Ϯ 0.002 min Ϫ1 ). The TR11 and TR2 receptor mutants were essentially internalization deficient (k e ϭ 0.001 Ϯ 0.001 min Ϫ1 and 0.0003 Ϯ 0.0002 min Ϫ1 , respectively). The significantly lowered internalization rate constant of the ⌬16 -22 mutant indicated receptor endocytosis to be mainly dependent on the deleted NSYYPTT sequence.
A series of point amino acid substitutions in the putative signaling region was then introduced to define more precisely the determinants required for efficient M-type sPLA 2 receptor endocytosis. The list of residues mutated in the region NSYYPTT is shown in Fig. 3B. Western blotting and immunofluorescence analyses revealed that all mutants were expressed, processed, and delivered to the plasma membrane (not shown). The affinity of all the mutants for 125 I-OS 1 was essentially the same as that of the wild-type receptor. A diagrammatic representation of internalization rate constants for the wild-type and mutant receptors is shown in Fig. 6. Substitution of any of the two tyrosine residues in the NSYYPTT sequence by an alanine (mutants Y18A and Y19A) dramatically impaired receptor internalization. Internalization rate constants (k e ϭ 0.0054 Ϯ 0.0005 min Ϫ1 and 0.002 Ϯ 0.001 min Ϫ1 , respectively) were at most 27 and 10% of those of the wild-type receptor. Substitution of Pro-20 by an alanine (P20A) had little effect on internalization (k e ϭ 0.018 Ϯ 0.003 min Ϫ1 ). Mutation of Ser-17 to a proline (S17P) increased 125 I-OS 1 internalization by 25% (k e ϭ 0.025 Ϯ 0.001 min Ϫ1 ).
The ⌬16 -22 mutant was still capable of relatively high internalization (Fig. 6). To verify that the endocytosis of the mutant remained mediated by clathrin-coated pits, transfected COS cells were incubated in the hypertonic sucrose medium 30 min prior to internalization experiments. A total blockade of 125 I-OS 1 internalization into these cells (results not shown) proved mutant internalization via clathrin-coated pits and suggested an implication of some other additional regions (presumably, LI motif) in the process. The role of the LI motif was assessed by substitution of Leu-34 by a glycine in the sequence of the wild-type and ⌬16 -22 receptors (mutants R/L34G and ⌬/L34G, respectively; Fig. 3C). The 125 I-OS 1 internalization by the ⌬/L34G receptor was practically abolished and represented only 1% of the wild-type efficiency (Fig. 6). The same amino acid substitution in the full-length receptor (R/L34G mutant) did not affect internalization (k e ϭ 0.019 Ϯ 0.002 min Ϫ1 ). These results indicate an implication of the LI motif in internalization of the ⌬16 -22 mutant receptor but argued against an analogous role of this motif in the context of the full-length receptor.
The internalization rate constant of the M-type sPLA 2 receptor was at least 4-fold higher in rabbit skeletal muscle cells than in transfected COS cells (see above), suggesting that the molecular principles and/or signaling structures governing receptor internalization in these cell lines could differ. To verify that the observed differences in the internalization efficiency of the mutants were not restricted to one particular cell line, the experiments were reproduced in HEK 293 cells stably transfected with the wild-type, ⌬16 -22, and TR11 receptors. The HEK 293 cell line, similar to COS cells, lacks endogenous M-type sPLA 2 receptor. Internalization experiments were performed on two independent cell clones for each receptor construct, expressing 0.5-2 pmol of M-type receptor/mg of total protein. Fig. 7 shows a diagram of internalization rate constants of the wild-type and mutant receptors in HEK 293 cells. The full-length receptor internalized 125 I-OS 1 as efficiently (k e ϭ 0.067 Ϯ 0.003 min Ϫ1 ) as in rabbit skeletal muscle cells. The rate constant of the ⌬16 -22 mutant was also relatively high (k e ϭ 0.03 Ϯ 0.005 min-1 ), representing 45% of the wild-type receptor efficiency. The TR11 mutant was internalization deficient (k e ϭ 0.001 Ϯ 0.002 min Ϫ1 ). All these experiments confirmed the results obtained in the corresponding studies with COS cells, although the latter have significantly lower internalization efficiency.
Mutational Analysis of the Putative Casein Kinase II Phosphorylation Site-The cytoplasmic portion of the receptor contains a consensus sequence for casein kinase II phosphorylation, with Ser-36 as a possible phosphorylation target. To investigate the possible implications of the phosphorylationdephosphorylation process in the regulation of the receptor endocytic activity, two mutants with point amino acid substitutions at Ser-36 were prepared (Fig. 3C) and transiently expressed in COS cells. Substitution of Ser-36 by an alanine (S36A) was expected to mimic a dephosphorylated state of the receptor. Mutation of Ser-36 to an aspartate (S36D mutant) was made to mimic a putative phosphorylated state. The internalization rate constants of these receptor mutants (k e ϭ 0.022 Ϯ 0.002 and 0.019 Ϯ 0.003 min Ϫ1 , respectively; Fig. 6B) were not significantly different from that of the wild-type receptor (k e ϭ 0.02 min Ϫ1 ) arguing against an implication of Ser-36 in endocytosis.
Internalization Activity of a M-type sPLA 2 /Mannose Receptor Chimera-The chimeric cDNA construct was prepared by substitution of the nucleotide sequence coding for the entire cytoplasmic domain of the rabbit M-type sPLA 2 receptor by the  (lanes 2-4). Aliquots of COS cell membrane fractions (10 g of total protein) were subjected to SDS-PAGE (6.5% polyacrylamide) followed by Western blotting using anti-sPLA 2 receptor antibodies. Indications correspond to those in corresponding region of the human macrophage mannose receptor (Fig. 8A). The junction site was at the YK peptide sequence located immediately downstream the putative intramembrane segments of both receptors. The chimera transiently expressed in COS cells did not show any structural abnormalities. Indirect immunofluorescence analysis with anti-sPLA 2 receptor antibodies revealed that it was normally delivered to the plasma membrane (not shown). Moreover, its affinity for 125 I-OS 1 was essentially the same as that of the wild-type sPLA 2 receptor (K d ϭ 30 Ϯ 5 pM). 125 I-OS 1 internalization driven by the chimera was about 50% less efficient (k e ϭ 0.01 Ϯ 0.003 min Ϫ1 ) than that of the wild-type sPLA 2 receptor (Fig. 8B). DISCUSSION This paper describes the endocytic properties of the M-type sPLA 2 receptor and characterizes its structurally encoded internalization signal(s). The endocytic receptors studied to date can be subdivided into two main groups. Receptors of the first group recycle constitutively and function as shuttles delivering their ligands to the cell interior. Their internalization is independent on ligand binding. It can be rather fast, i.e. 10 -60%/ min (corresponding to a k e of 0.1-0.6 min Ϫ1 ) for the LDL, asialoglycoprotein (38), or transferrin receptors (30) or slow, i.e. 1-2%/min for the Fc and fibronectin receptors (37,39). Internalization of the other group of receptors (such as hormone and growth factor receptors) is clearly ligand-dependent. It is negligible in the absence of their ligands and can be increased more than 50-fold (k e of 0.3-0.6 min Ϫ1 ) in the presence of corresponding ligands (40,41). In many cell types, a long term exposure of this latter group of receptors to their ligands leads to rapid intracellular sequestration and degradation (down-regulation) of the receptors.
The calculated internalization rate constant of the M-type sPLA 2 receptor in a primary culture of rabbit skeletal muscle cells (k e ϭ 0.1 min Ϫ1 ) was that of efficiently internalized receptors but was significantly lower than that reported for the structurally homologous mannose receptor (k e ϭ 4.12 min Ϫ1 ) in sinusoidal endothelial cells from rat liver (24). The blockade of 125 I-OS 1 uptake observed under hypertonic conditions, which causes a disassembly of plasma membrane-associated clathrin lattices (33), proved receptor internalization via clathrincoated pits ( Fig. 2A). Long term exposure to a high concentration of a possible endogenous ligand, pancreatic-type sPLA 2 , had no effect on the number of the cell surface-exposed 125 I-OS 1 binding sites arguing against ligand-induced sequestration and/or down-regulation of the sPLA 2 receptor in rabbit skeletal muscle cells. In the absence of the added ligand, monensin, a classical inhibitor of the trans-Golgi endosomal trafficking (31,34), produced a large, more than 57% reduction in the number of cell surface OS 1 -binding sites, indicating the rapid uptake and recycling of the ligand-free sPLA 2 receptor. Similar extent of the monensin-induced protein trapping within the cells was reported earlier for the constitutively recycling transferrin receptor (34). Taken together, the above results suggest that internalization of the M-type sPLA 2 receptor in rabbit skeletal muscle cells is rapid, clathrin-coated pit-mediated, and ligandindependent. These properties allowed us to classify the M-type sPLA 2 receptor to the group of constitutively recycling receptors, the same as for its structural homologue, the mannose receptor (20,21,23). The classification of the M-type receptor into this group is in contrast with previously suggested role of the sPLA 2 receptor in cell proliferation (5) and eicosanoid production (16). Indeed, such biological effects are generally supposed to be mediated by receptors of the other group, i.e. by ligand-activated receptors. This apparent inconsistency is presently difficult to explain. Noteworthy, the mannose receptor, which also belongs to the group of constitutively recycling receptors, was shown to be involved in cell proliferation (42).
Internalization of the membrane receptors is dependent on their cytoplasmic signal sequences selectively recognized by adaptor proteins of clathrin-coated pits (for a recent review, see Ref. 37). The internalization signals identified to date fall into three main groups. The first group is represented by four to six-amino acid sequences containing an aromatic residue, usually a tyrosine, in a tight ␤-turn structure (43)(44)(45)(46)(47). This type of signal is found in the vast majority of endocytic receptors, a typical example being a consensus NPXY motif (36), first detected in the structure of the LDL receptor (43). The dileucine (or leucine-isoleucine) dipeptide (48 -53) and the recently identified KK(X/F)(X/F) sequence (54) represent two other groups of internalization signals. Importantly, analogous motives were shown to be involved in other intracellular sorting events, such as lysosomal targeting (48,(55)(56)(57)(58)(59), transcytosis, and basolateral sorting in polarized cells (53, 60 -62), and in the endoplasmic reticulum protein retrieval (54,63,64).
The cytoplasmic portion of the rabbit sPLA 2 receptor contains three stretches of potential signal sequences (Fig. 3): (i) the tight turn-forming region NSYY; (ii) the LI dipeptide in the distal part of the tail; and (iii) the juxtamembrane NKGFF tight turn region, which resembles the internalization signal QQGFF of the epidermal growth factor receptor (35). The latter motif, being poorly conserved in mammalian sPLA 2 receptors (Fig. 3D), was a priori considered the least plausible candidate for an internalization signal. Indeed, the receptor mutants TR11 and ⌬16 -22, although retaining this peptide sequence, exhibited dramatically impaired internalization (Fig. 5). Conversely, the internalization defect of the ⌬16 -22 mutant strongly suggested the NSYY motif to be the endocytic signal of the rabbit M-type sPLA 2 receptor (Fig. 6).
The NSYY motif, although obviously related to the NPXY signal of the LDL receptor, has some peculiar features (Fig.  3D). The sequence contains two evolutionary conserved tyrosine residues instead of one, and in the rabbit and mouse M-type sPLA 2 receptors, the canonical proline residue is substituted by a serine. Alanine scanning of the region has proved Tyr-19 to be the key residue of the signal sequence. This feature clearly distinguishes the M-type sPLA 2 receptor from the structurally homologous macrophage mannose receptor, where an alanine substitution of the single cytoplasmic tyrosine impaired internalization only partially (about 50% of the wildtype efficiency) (65). The internalization defect of the Y18A mutant (25% of the wild-type rate) suggested that in the signal sequence motif of the rabbit M-type sPLA 2 receptor, the amino acid residue occupying the consensus X position can be a subject for certain structural restraints. Substitution of Ser-17 by a proline was made to reconstitute the canonical NPXY motif. The mutation increased internalization, but only slightly (25%), indicating that both serine and proline residues are permissible at this position. Indeed, both residues are known to be turn-promoting (66). However, the M-type sPLA 2 receptor is to our knowledge the first reported example of an endocytic receptor in which this substitution has naturally occurred. Alanine substitution of the highly conserved Pro-20 immediately adjacent to the critical tyrosine residues (Fig. 3D) was of no significant effect on endocytosis, suggesting that this residue is not important for receptor internalization.
Interestingly, the internalization of the ⌬16 -22 mutant, although significantly impaired, was still clathrin-coated pitmediated, and considerably higher than that of the C-terminally truncated mutants TR11 and TR2 (Fig. 5). Substitution of Leu-34 by a glycine (⌬/L34G) abolished internalization and proved the endocytosis of the ⌬16 -22 mutant to be dependent on the LI motif (residues 34 -35). In the full-length receptor, however, the same glycine substitution had no effect on internalization (R/L34G mutant, Fig. 6). These results suggest that the LI motif, although recognized as an internalization signal in the receptor mutant with the deleted NSYY region, does not play the same role in the context of the full-length sPLA 2 receptor. The data on the presence of "cryptic" internalization (sorting) signals in the cytoplasmic portions of the proteins are not unprecedented. Analogous findings were made for the lysosomal acid phosphatase, the interleukin 6 signal transducer gp130, and the cation-independent mannose 6-phosphate/insulin-like growth factor II receptor (51,57,67).
The negatively charged region surrounding the LI motif of the sPLA 2 receptor is highly conserved from mouse to human (Fig. 3D) and contains a consensus casein kinase II phosphorylation site, with Ser-36 as a putative target residue (13). Similar structures were found in the cytoplasmic domains of many recycling receptors (68) and are supposed to be implicated in intracellular sorting events, such as basolateral targeting (60), sorting into Golgi-derived clathrin-coated pits (57), and/or to regulate the efficiency of receptor internalization (69). Amino acid substitutions at Ser-36 mimicking putative phosphorylated and dephosphorylated sPLA 2 receptor states (Fig.  6B) were of no significant effect on the rate of endocytosis, arguing against implication of the region in internalization of the M-type sPLA 2 receptor.
The M-type sPLA 2 receptor and the macrophage mannose receptor are predicted to share the same structural organization (13), although the amino acid sequence identity of these two receptors is quite low (28% identity in the extracellular region and only 17% in the cytoplasmic tail). The cytoplasmic NTLY motif of the macrophage mannose receptor resembles the NSYY signal sequence of the rabbit M-type sPLA 2 receptor and was previously shown to be implicated in internalization (65). It was, thus, interesting to analyze the internalization properties of a chimera composed of the predicted extracellular and transmembrane domains of the rabbit M-type sPLA 2 receptor and of the cytoplasmic portion of the human macrophage mannose receptor. The chimeric receptor was properly delivered to the plasma membrane and displayed the normal binding affinity for 125 I-OS 1 , proving the correct assignment of the topological domains. The internalization efficiency of the chimera was only 50% less efficient than that of the wild-type sPLA 2 receptor, suggesting that the cytoplasmic domains and the encoded internalization signals of these two proteins are interchangeable.
In conclusion, this work has revealed the mechanism of endocytosis of the M-type sPLA 2 receptor and has identified the structural regions implicated in its internalization. The role of the M-type sPLA 2 receptor internalization process remains, however, obscure. Internalization of the M-type sPLA 2 receptor could first be an important component of the signal transduction system coupled to the binding of the sPLA 2 to its specific receptor. It could serve to terminate the signals produced by sPLA 2 on target cells. Alternatively, the endosomal vesicles could serve as a vehicle delivering sPLA 2 to specific intracellular compartments, where the ligand, before being degraded in lysosomes, could manifest its enzymatic activity. Another possibility is that the internalization of the M-type sPLA 2 receptor could serve a clearance function selectively withdrawing sPLA 2 from the extracellular fluid. This could be of a crucial importance at various inflammatory disease states (such as rheumatoid arthritis, acute peritonitis, septic shock) when high levels of inflammatory type II sPLA 2 are produced and secreted (8). Further investigations on the properties of the M-type sPLA 2 receptor in various tissues and species should provide more evidence on its function and the role of internalization process, in particular. ical School, Boston, MA) for the generous gift of the full-length human macrophage mannose receptor cDNA plasmid construct. We also thank Dr. J.-L. Carpentier (University of Geneva, Geneva, Switzerland) for fruitful discussions and helpful suggestions and Dr. P. Ancian, Dr. J. P. Nicolas and Dr. R. Waldmann for help at initial stages of the project and expert advice with computer programs. Finally, we thank M. M. Larroque, F. Aguila, and A. Douy for an expert technical assistance.