Cloning and expression of a mouse macrophage cDNA coding for a membrane glycoprotein of the scavenger receptor cysteine-rich domain family.

We have cloned from murine macrophages a cDNA coding for a new protein of the scavenger receptor family whose mRNA is increased very strongly by adherence and moderately by exposure to tumor necrosis factor and interferon-gamma. The nucleotidic sequence extends for 2168 bases and encodes a protein of 559 amino acids with six potential glycosylation sites. The first 100 NH2-terminal amino acids represent a single scavenger receptor cysteine-rich domain, whereas the COOH-terminal end of the molecule is compatible either with a transmembrane hydrophobic peptide followed by a very short intracytoplasmic sequence or a signal sequence for an anchoring via a glycophosphatidylinositol. The protein is highly homologous to most of the very recently identified human MAC-2-binding protein and murine cyclophilin C-associated protein.

mRNA is also expressed in celI lines of other origin, in particular fibroblasts. The protein may be the murine homologue to the very recently identified human MAC-2-binding protein (hMAC-2-BP) (8). MAC-2 is a bifunctional secreted protein (9) with a lectin domain capable of binding laminin (10) and which, in association with its ligand, MAC-2-BP, may serve as a bridge between the macrophages and the extracellular matrix, microorganisms, or other cells bearing galactosylated proteins. One of the roles of this new cell membrane protein might then be related to macrophage adhesion and migration to the site of infection. With very minor differences, this protein is also homologous to the cyclophilin C-associated protein (CyCAP) (ll), which makes it a candidate in the cascade of immunoregulatory events resuking from the absorption of the immunosuppressive drug cyclosporin A.

MATERIALS AND METHODS
cDNA Cloning and Sequencing-Poly(A+) RNA from thioglycolateinduced peritoneal adherent macrophages obtained from BALB/c mice and selected after 6 h of adherence to plastic in vitro was prepared on an oligo(dT)-Sepharose column (Pharmacia, Uppsala, Sweden) and used to construct a cDNA library in pcDNAII vector (The Librarian, Invitrogen, San Diego, CA). The library, which contained about 95,000 clones, was split into 10 sublibraries whose plasmid DNA was explored by polymerase chain reaction, using as 5'-primer oligonucleotides corresponding to the Sp6 or T7 promoters specific primer E6 (5'CACAGCGAGTTCCAAGGCC3') derived from contained in the pcDNAII vector and as the 3'-primer the genethe mouse erythropoietin sequence. Polymerase chain reaction cycles were 1 min at 94 "C, 2 min at 55 "C, and 3 min at 72 "C for a total of 40 cycles. The DNA of a sublibrary gave an amplified fragment of 1 kilobase pair using E6 and Sp6 primers. This fragment, digested with BamHI to eliminate the region between the Sp6 promoter site and the BstXI site of insertion, was used, after 32P labeling by random priming, to identify hybridizing clones from the positive sublibrary spread on agar plates. Two positive clones were identified. In order to sequence entirely the cDNA, restriction enzyme fragments were cloned in pGEM3Z (Promega, Madison, WI) or pBluescript SK(+) (Stratagene, La Jolla, CA) plasmids, and chain termination sequencing was carried out with the U. S. Biochemical Corp. sequenase procedure with Sp6, T3, or T7 promoter sequencing primers. The final sequence was determined from both strands.
Transfection Experiments-The cDNA was inserted at the X h I site of the pCAGGS expression vector containing the @actin promoter (a kind gift of J. Miyasaki) (12). Three days after transfection, COS-7 cells were lysed with 1% Nonidet P-40 in phosphate-buffered saline, nuclei eliminated by a 1000 X g centrifugation, and the postnuclear supernatants processed for SDS-polyacrylamide gel electrophoresis in reducing conditions. The gel was stained with silver nitrate.
Computer Analysis-Searches for homologies in the DNA or protein sequences were done with the aid of the Genetics Computer Group (13) and IG (Intelligenetics, Mountain View, CA) programs. Translation, predictions of eucaryotic secretory signal sequence and transmembrane region, and detection of sites and signatures in the protein sequence were done with the PCgene programs (A. Bairoch, Department of Biochemistry, CMU, University of Geneva, Switzerland).
Cell Culture-BALB/c or C3H/HeJ mice were injected 3 days before death with 1.5 ml of thioglycolate broth (Difco). Ceils of the peritoneal exudate were cultured at 10' cells/ml for 1 h in Dulbecco's modified Eagle's medium (Life Technologies, Basel, Switzerland) in plastic plates (Becton Dickinson, Lincoln Park, NJ) and washed, and the adherent cells were further cultured in Dulbecco's modified Eagle's medium complemented with insulin (15 pg/ml), transferrin (1.5 pg/ml), testosterone (2 mM), linoleic acid (1 pg/ml), bovine serum albumin (0.5 mg/ml), penicillin, and streptomycin in the presence of the different stimuli: TNF (mouse recombinant TNF, a kind gift of B. Allet, Glaxo, Geneva, Switzerland) at 100 ng/ml, lipopolysaccharide (Difco) at 0.5 pg/ml, and mouse recombinant interferon-y (Genentech, South San Francisco, CA) at 50 units/ml. RNA  SDS and two 1-h washes in 0.2% SSC, 0.2% SDS, the filters were exposed to x-ray film at -70 "C with a Dupont Cronex intensifying screen. 10 pg of genomic DNA purified from BALB/c mouse liver (16) were digested overnight with the indicated restriction enzymes and were run in 40 mM Tris acetate, 2 M EDTA, pH 8.5, agarose gel.
After denaturing transfer onto Hybond N+ (Amersham), the membrane was processed as for Northern hybridization. Peptide Synthesis, Antibody Production, and Immurwfluorescence-From the deduced amino acids sequence, we chose 4 peptides identified as immunogenic with the aid of the PCgene program. The peptides, synthesized in the laboratory of Dr. R. Frank (University of Heidelberg, Germany), were coupled to purified protein derivative (Statens seruminstitut, Copenhagen, Denmark) with glutaraldehyde (17), and a pool of 25 pg of each coupled peptide was mixed in Freund's complete adjuvant and injected into a rabbit. After repeated boosts, polyclonal antibodies were purified by affinity chromatography on a peptide-coupled CNBr-Sepharose column. Immunofluorescence studies with these antibodies were done on 24-h adherent peritoneal macrophages that were fixed with 3% paraformaldehyde and revealed with a rhodamine-conjugated goat anti-rabbit IgG. Confocal immunofluorescence analysis was done with the Bio-Rad MRCGOO coupled to a Zeiss Axiophot fluorescence microscope.

RESULTS
Cloning and Sequencing-A cDNA library was prepared in the pcDNAII vector from the poly(A+) RNA obtained from mouse thioglycolate-elicited peritoneal macrophages selected after 6 h of adherence to plastic in vitro. This library was then split into 10 sublibraries, the plasmid DNAs of which were screened by polymerase chain reaction using as primers the oligonucleotide E6 (the erythropoietin-specific primer) and oligonucleotides corresponding to either the Sp6 or the T7 promoters. With one sublibrary, an amplified fragment of 1 kilobase pair was observed, excised from agarose gel, and used as a probe for screening the positive sublibrary, thus allowing the identification of two positive clones, both with an insertion of 2.2 kilobase pairs. Fig. 1 shows the complete nucleotide sequence together with the deduced amino acid sequence.
This cDNA extends for 2168 nucleotides. There is a long  open reading frame starting from the initiation codon ATG at nucleotide positions 169-171 and ending at the termination codon TAA at position 1900-1902. The nucleotides surrounding the ATG codon match the consensus sequence for a translation initiation site (18). The next 18 amino acids have the characteristics of a signal peptide sequence (19). A putative transmembrane domain of 16 amino acids partitions the protein into a 531-amino acid extracellular domain, containing six consensus sequences for N-linked glycosylation and 15 cysteine residues and a very short intracellular domain of 12 amino acids. The COOH-terminal end sequence is also compatible with an anchoring of the protein via a glycophosphatidylinositol (see "Discussion"). The 3"noncoding region of 266 nucleotides contains one of the three possible polyadenylation signals, ATTAAA, 21 bases upstream from the poly(A) tract. We gave to this cDNA the acronym MAMA, for murine adherent macrophage cDNA. The sequence corresponding to the oligonucleotide primer used for the anchored polymerase chain reaction is located between nucleotides 1017 and 1030.

M~A C~" C A C M C % A C T~A T C C T C % C~A C~
Expression in COS-7 Cells-In order to demonstrate that this cDNA indeed directs the synthesis of a protein of the expected size, it was introduced in an eucaryotic expression vector containing the @-actin promoter (12). Two vectors, with the insert in opposite directions were used to transfect COS-7 cells. SDS-polyacrylamide gel electrophoresis analysis of the whole cell lysates after 3 days of culture showed the presence of a protein of 60-65 kDa only in the cells transfected with the vector containing the cDNA in the right direction (Fig. 2). This result is consistent with the expected molecular weight of the unglycosylated protein. bodies against different peptides synthesized from the deduced amino acid sequence were used to ascertain by immunofluorescence the presence of the MAMA protein on the surface of activated macrophages. Confocal images (Fig. 3) show that the immunofluorescence is mainly restricted to the surface of macrophages, whereas control cells with rabbit IgG in the first step were devoid of staining (not shown). It thus appears that the MAMA protein is associated with the plasma membrane, most probably by the hydrophobic sequence found in the protein.
Sequence Comparisons-Screening of data bases showed no overall homology in the total nucleotidic or amino acid sequences with other known sequences. When specific sites and signatures were screened with the Prosite program, it appeared that amino acids 6-107 of the MAMA protein match perfectly the consensus sequence of the cysteine-rich domain of the speract receptor (20), a domain also found in the macrophage scavenger receptor (7) and the T cell membrane protein CD6 (21) (Fig. 4). Another feature of the extracellular domain is found between amino acids 410 and 432 where 8 tyrosine residues are clustered. While this work was being prepared for publication, the sequences of a human protein described as the hMAC-2-BP and of the murine CyCAP were reported (8,11). Except for very minor divergences, CyCAP appears to be the same protein as MAMA. hMAC-2-BP is 69% homologous to the MAMA protein, with all the cysteines and the N-glycosylation sites conserved, SRCR domains being 82% homologous (Fig. 5), but the proteins being divergent in two domains. The tyrosine-rich domain contains 23 amino

V I I L V -G O S C A l l B C R V I I T l l~~l C R D R U D l~W W C R S W Y O~L --A V~
Do SR

Speract-1 I R L I -I I C R T E N C O S V K I Y l l A T R W~V C~l l H E N~K Q W F f f i A R --O F Y CD6-2
LRLVffiCC-hChCRVKHLEIl0EWQSVCDD~DLEDAIIWCRQW-CCwAy)hLP

4 9
I. I 1 . 1 1 1 I I I I I I I I I I I ..I I I I I I I I I I I I I I I I I I I I I I I I I I I I I . I I I I I I I I I I l I I I I I I I I I I I   I I I l I . I I I  I -I I I I .   I I I I . I I I I I I . I . I   I I I I I I I I . l I I I I I I I l I I I I I I I I . I I 1 1 . 1 1 .I I I I I I I I I I I I1III.I.III I I I I I I I I I . . .   . I I I l I I I I I I I I I I I I I I I l l I I I I I I . I   I I I I I I I I I I I I  Genomic Anulysis-Southern analysis with various restriction enzymes, including enzymes cutting in exons, suggests that the corresponding gene is present as a single copy (Fig.  6).
Modulation of MAMA mRNA Level in Macrophages and Constitutive Presence in Other Cell Types and Various Organs in Vivo-As shown in Fig. 7A, MAMA mRNA progressively accumulates in time until 24 h when thioglycolate-elicited peritoneal macrophages are cultured in vitro; this is due to adherence (which also induces the increase of some other macrophage mRNA), since when macrophages were cultured in a Teflon vial, a condition that does not allow adherence, no increase in mRNA with time was observed (not shown). Different inducers of macrophage activation were tested for their ability to modulate MAMA mRNA. Two of these, interferon-y and TNF, increased the level of the transcript. This increase was apparent after 2 h with interferon-y and after 24 h with T N F (Fig. 7 B ) . In contrast, neither lipopolysaccharide, granulocyte-macrophage colony-stimulating factor, macrophage colony-stimulating factor, nor phagocytosis of opsonized red blood cells did modify the mRNA levels (not shown). Cultures of other cell types or lines showed that MAMA mRNA is also detectable in fibroblasts, in some B cell populations, and in cells of epithelial origin from the mammary gland but not in keratinocytes (Fig. 8A). Finally, MAMA mRNA is present at variable levels in all the organs tested, with the exception of the brain (Fig. 8 B ) .

DISCUSSION
The new protein identified by the cloning and sequencing of the cDNA described in this report is a membrane glycoprotein of 559 amino acids, containing at its NH2 terminus around 100 residues that display 48% homology with the SRCR of mouse macrophage and matching the consensus sequence of the domains found in the newly recognized family of SRCR proteins. Besides the scavenger receptor itself, this growing family is composed of the sea urchin spermatozoa receptor for peptides (speract receptor) involved in the stimulation of sperm metabolism and motility (20); the CD6 (21), a . . .. of this family have at least three such domains, but the scavenger receptors have been shown to be trimeric (25). A specific function has not been linked yet to the SRCR domain. It is not involved in the main function of the scavenger receptor, which is to mediate endocytosis by macrophages, of a large number of macromolecules, including acetyl low density lipoprotein (hence its involvement in atherogenesis) (26).
We have identified a hydrophobic sequence in the MAMA protein that could serve for the anchoring to the membrane, but the protein may also be retained at the cell surface via a GPI anchor, since the MAMA protein has the required features common to the GPI-anchored proteins, namely a precursor GPI signal sequence with a short hydrophobic domain or a short or no intracellular domain (12 amino acids for MAMA with no identifiable signaling motif) and a cleavage/ attachment site located 30-40 amino acids from the COOH terminus (a tripeptide Asn-Ser-Ser is found in this region of the MAMA protein) (27-30). It should be pointed out that a transmembrane or a GPI anchoring is not mutually exclusive. Although the role of GPI anchors is not established, it may serve for a rapid and differential release of some proteins from the cell surface after stimulation (31). The possibility that the MAMA protein may use a GPI anchor deserves further investigation and is of special importance in view of the possible role of this protein, as discussed now.
Nucleotide sequence analysis shows that the MAMA protein is 69% homologous to the recently identified human MAC-2-BP (8), thus may represent its mouse equivalent, and, furthermore, is identical to the newly discovered mouse CyCAP (11). Indirect evidence exists that CyCAP is a mem- brane molecule (because cyclophilin C binds to the cell surface) (11,32), and our observation now establishes this point. Human MAC-2-BP, in contrast, has been mainly detected in extracellular locations, being present in all human fluids and extracellular mediums of a variety of cultured cells. However, in HT-29 cell lysate, this protein is found in a fully mature form and associated with its ligand MAC-2 (8), a galactosespecific lectin (33,34) that is also detectable on the cell surface of thioglycolate-induced murine macrophages (35). Thus, this provides indirect evidence that MAC-2-BP can also be found in a membrane-associated form. On the basis of what has been discussed above, the possibility that it may be released from a GPI anchor, explaining its finding as a soluble molecule, has to be considered. A soluble form may also be generated as the result of alternate exon splicing. Thus, MAMA should also exist in a soluble form in mouse fluids if, as seems likely, it is the murine form of hMAC-2-BP. In light of this possibility, an intriguing point would be that the two ligands so far identified for this MAMA/CyCAP/MAC-2-BP protein, namely the MAC-2 protein and cyclophilin C, are also extracellular proteins that nevertheless lack a signal sequence. The possibility of a role for MAMA in the transport of these ligands to the cell surface remains conjectural, but in any case MAMA would appear to be essential in holding them on the outer cell membrane.
Is there a possible relationship between the capacity of binding two apparently so disparate ligands and the strong induction of MAMA by macrophage adhesion? It has been suggested that one of the possible roles of MAC-2 is related to cell adhesion to the extracellular matrix; MAC-2 bears a lectin domain (apparently responsible for its fixation to MAC-2-BP) that can also bind laminin, a major extracellular matrix component (9,10). MAMA/MAC-2-BP could thus play the role of a loose link between the macrophage surface and the extracellular matrix, via MAC-2, being involved in macrophage migration during inflammation or, if the complex MAC-2-BP. MAC-2 is released from the cell, between the extracellular matrix and various cell or microorganism surfaces bearing glycoproteins. As for cyclophilin C, its role is presently unknown. Besides its cis/trans peptidyl prolyl isomerase activity, common to all cyclophilin and in the light of the possible involvement of MAMA in macrophage adherence and activation, it is very striking that cyclophilin A can be released in uitro by lipopolysaccharide-activated macrophage and can act, in a secreted form, as an inflammatory cytokine in vivo and as a factor chemotactic for polymorphonuclear leucocytes and monocytes in vitro (36). A comparable activity has not been tested for cyclophilin and the presence of this or other forms of cyclophilin on the surface of activated macrophages has not been explored. This suggests the very intriguing possibility that MAMA may be involved not only in macrophage attachment to the extracellular matrix or as a form of soluble adhesion molecule system in inflammation but also in focusing a peculiar form of inflammatory cytokine on the macrophage surface or in releasing, in inflammatory sites, cytokine activity under the form of a sticky complex. This might lead to a complete reappraisal in the understanding of the therapeutic action of cyclosporin, as has already been suggested (11, 32, 37), since this drug prevents the association of cyclophilin C to its binding protein, here identified as MAMA.
It is apparent from the present study that MAMA/CyCAP, while probably playing a role in macrophage adherence and migration, must be present on the surface of a variety of cells, including fibroblasts and epithelial cells, in a variety of tissues. With the antibodies used in the present study, we have, for instance, isolated from kidney lysates the MAMA protein'; it is of interest to recall in this respect that the kidney is one of the target organs of the cytotoxicity of cyclosporin. The involvement of this protein in cell adhesion in a variety of normal tissues, or in tissue remodeling, needs to be explored. Since proteins probably identical to MAC-2-BP have been detected as tumor-associated antigens in a melanoma (38) and in a lung tumor (39) cell lines, MAMA may as well play a role in tumor cell adhesion and spreading. Where the SRCR takes in place in MAMA functions remains also to be determined.