This application is a continuation-in-part of Tedder, U.S. Patent Application Serial No. 07/313,109, filed February 21, 1989 the whole of which is hereby incorporated by reference herein.

Part of the work leading to this invention was made with United States Government funds. The U.S. Government has certain rights in this invention.

This invention relates to human leukocyte-associated cell surface proteins.

Background of the Invention Genes exclusively expressed by one cell lineage, but not by others, often define the function of that cell population. The generation of genes by the assembly of functionally independent domains has occurred frequently as new genes have evolved to encode proteins with new functions. An inducible endothelial-leukocyte adhesion molecule (ELAM-1) , having several functionally independent domains, is expressed on the surface of cytokine-treated endothelial cells. This molecule is thought to be responsible for the accumulation of blood leukocytes at sites of inflammation by mediating the adhesion of cells to the vascular lining (Bevilacqua et al., Proc. Natl. Acad.

Sci. USA 84.J9238 (1987)). A granule membrane protein found in platelets and endothelial cells, termed GMP-140, has been cloned and is homologous with ELAM-1 (Johnston et al.. Blood Suppl. 1 7 :327A (1988)).

Summary of the Invention The invention generally features a leukocyte- associated cell surface protein LAM-1 (leukocyte adhesion molecule-l) , which contains domains homologous with binding domains of animal lectins, growth factors, and C3/C4 binding proteins; the specific domains of the LAM-1 protein; and the genomic DNA sequences encoding the LAM-1 protein and the specific domains of LAM-1.

Preferred embodiments of the invention include essentially purified proteins comprising sequences of amino acids having 90% or greater homology with the amino acid residues of specific domains of human leukocyte-associated cell surface protein LAM-1 given in Fig. 2, i.e., the lectin domain given by residues 42-170, the EGF-like domain given by residues 171-206, the short consensus repeat unit I domain given by residues 207-269, the short consensus repeat unit II domain given by residues 270-331, the leader domain given by residues 15-41, the transmembrane domain given by residues 332-373, and the phosphorylation domain given by residues 374-380.

The invention further features purified LAM-1 or a domain thereof covalently bonded to an immunoglobulin heavy chain constant region.

In another aspect the invention features an essentially purified human genomic fragment of approximently 30 kb comprising the restriction sites as indicated in Fig. 4B.

In another aspect the invention features methods of treating a patient suffering from a leukocyte-mobilizing condition that includes administering to the patient a therapeutic agent including a therapeutic amount of the LAM- 1 protein or a domain thereof, or of an antagonist to the LAM-1 protein or domain thereof, or of a fusion protein

including the LAM-1 protein or a domain thereof covalently bonded to an im unoglobulin heavy chain constant region. The therapeutic agent is administered in a pharmaceutically acceptable carrier substance. in preferred embodiments of the methods the patient is suffering from tissue damage, an autoimmune disease, or cancer; or is an organ or tissue transplant recipient.

In another aspect the invention features using the LAM-1 protein or domain thereof to identify a ligand that binds to the protein or to a molecule that is specifically associated with the protein, or fragment thereof, to generate a functional molecule. Ligands so identified can also be used in the methods of the invention described above. As used herein the term "antagonist to LAM-l" includes any agent which interacts with LAM-l and interferes with its function, e.g., antibody reactive with LAM-1 or any ligand which binds to LAM-l. The term "identify" is intended to include other activities that require identification of an entity, such as isolation or purification. The term "essentially purified" refers to a protein or nucleic acid sequence that has been separated or isolated from the environment in which it was prepared or in which it naturally occurs. Leukocyte-associated cell surface protein LAM-l plays an important role in leukocyte-endothelial cell interactions, especially selective cell trafficking to sites of inflammation. The LAM-l protein or domains thereof, or other molecules that interfere with leukocyte adhesion and function, can be used therapeutically to inhibit the inflammatory response and to treat such conditions as tissue damage and metastasis of cancer cells.

Other features and advantages of the invention will be apparent from the following description of the preferred embodiments thereof and from the claims.

Description of the Preferred Embodiments In the drawings. Figs. 1A and IB show the structure of LAM-l cDNA clone.

Fig. 2 shows the cDNA nucleotide sequence and also shows the amino acid sequence of LAM-l.

Figs. 3A, 3B, and 3C show the homologies of LAM-l with other proteins.

Figs. 4A, 4B, and 4C show the restriction map and the exon-intron organization of the Iyam-1 gene.

Fig. 5 shows the nucleotide sequence of exons II through X of the lyam-1 gene. The leukocyte adhesion molecule-1 (LAM-l) , also called lymphocyte-associated molecule-1, is expressed by human lymphocytes, neutrophils, monocytes and their precursors and is a member of the selectin family of cellular adhesion/homing receptors which play important roles in leukocyte-endothelial cell interactions, especially selective cell trafficking to sites of inflammation. LAM-l combines previously unrelated domains found in three distinct families of molecules: animal lectins, growth factors, and C3/C4 binding proteins. cDNA encoding the LAM-l protein was initially identified as follows. B cell-specific cDNAs were isolated from a human tonsil cDNA library (ATCC #37546) using differential hybridization with labeled cDNAs derived fro* either B cell (RAJI) RNA or T cell (HSB-2) RNA (Tedder et al., Proc. Natl. Acad. Sci. USA 85:208-212 (1988)).

Positive plaques were isolated and cloned, and the cDNA inserts were subcloned into the plasmid pSP65 (Promega, Madison, WI) . Nucleotide sequences were determined using

the method of Maxam and Gilbert (Meth. Enzymol. 65:499 (1980)). Gap penalties of -1 were assessed during homology analysis for each nucleotide or amino acid in the sequence where a gap or deletion occurred. One of the 261 RAJI+ HSB2- cDNA clones isolated, B125, contained a 1.90 kb cDNA insert that hybridized with a 2.4 kb RNA species found in several B cell lines (Tedder et al., supra) . However, B125 did not hybridize with any of the other RAJI+ RSB2- clones or with mRNA from several T cell lines. The B125 cDNA clone was characterized by restriction mapping and nucleotide sequence determination. A near-full-length 2.3 kb cDNA that hybridized with B125 was isolated, sequenced, and termed pLAM-1.

The expression of LAM-l mRNA by cell lines of lymphoid and non-ly phoid origin was examined. Northern blot analysis revealed that LAM-l cDNA hybridized strongly to a 2.6 kb RNA species and weakly to a 1.7 kb RNA species in poly(A)+ RNA isolated from the B cell lines Raji, SB, Laz-509, and GK-5. However, RNA isolated from two pre-B cell lines (Nalm-6, PB-697) , three B cell lines (Namalwa, Daudi, BJAB) , five T cell lines (CEM, Hut-78, HSB-2, Molt- 15, Molt-3), a myelomonocytic cell line (U937 and U937 cultured with LPS) and erythroleukemic K-562 cell line did not hybridize with LAM-l cDNA suggesting that expression of this gene was preferentially associated with B lymphocytes. Neutrophils expressed LAM-l mRNA but had a relatively lower amount of transcript among total mRNA when compared with the Raji cell line or blood T lymphocytes. LAM-l cDNA has also been used to transfer expression of LAM-l to cells that do not express the gene.

As shown in Fig. 1A, a restriction map was constructed by the standard single, double or triple digestions of pLAM-1. The coding region is shown in black.

Arrows indicate the direction and extent of nucleotide sequence determination and the open circles indicate 5'-end labeling. In Fig. IB, a schematic model of the structure of the LAM-l mRNA is shown. Thin lines indicate 5' and 3' untranslated sequences (UT) , while the thick bar indicates the translated region. The boxes represent the lectin-like and epidermal growth factor (EGF)-like domains and the two short consensus repeat (SCR) units. The open box indicates the transmembrane (TM) region. pLAM-l contains an open reading frame that could encode a protein of 372 amino acids as shown in Fig. 2. The numbers shown above the amino acid sequence designate amino acid residue positions. The numbers to the right indicate nucleotide residue positions. Amino acids are designated by the single-letter code, and * indicates the termination codon. The boxed sequences identify possible N-linked glycosylation sites. Hydrophobic regions that may identify signal and transmembrane peptides are underlined. The amino acid sequence of LAM-l indicates a structure typical of a membrane glycoprotein. The mature LAM-l protein has an extracellular region of about 294 amino acids containing 7 potential N-linked carbohydrate attachment sites. LAM-l has a cytoplasmic tail of 17 amino acids containing 8 basic and 1 acidic residues. The processed LAM-l protein has a Mr of at least 50,000 and can be isolated by conventional techniques, such as affinity column chromatography with antibody or ligand, from cell lines that normally express this receptor or from transfected cell lines. Or the protein can be synthesized by in vitro translation of the LAM-l cDNA.

LAM-l combines domains homologous to domains found in three distinct families of molecules: animal lectins, growth factors, and C3/C4 binding proteins. The

extracellular region of LAM-l contains a high number of Cys residues (7%) with a general structure as diagrammed in Fig. IB. As indicated in Fig. 3, segments of homologous proteins are shown with the amino acid residue numbers at each end. Homologous amino acids are shown in boxes. Gaps (-) have been inserted in the sequences to maximize homologies. The first 157 amino acids of the protein (Fig. 3A) were homologous with the low-affinity receptor for IgE (Kikutani et al.. Cell 4_7:657 (1986)), the asialoglycoprotein receptor (Spiess et al., Proc. Natl. Acad. Sci. USA 8^:6465 (1985)) and several other carbohydrate-binding proteins (Drickamer et al., J. Biol. Chem. 2j56:5827 (1981) ; Ezekowitz et al., J. Exp. Med. 167:1034 (1988); Krusius et al., J. Biol. Chem 162:13120-13125 (1987); and Takahashi et al., J. Biol. Chem. 260:12228 (1985)). The amino acids conserved among all animal-lectin carbohydrate recognition domains are indicated (*) . Although the sequence homologies were less than 30%, all the invariant residues found in animal lectin carbohydrate-recognition domains were conserved (Drickamer, J. Biol. Chem. 263:9557 (1988)). The lectin domain included amino acid residues 42-170 given in Fig. 2.

The next domain of 36 amino acids, at residues 171- 206 shown in Fig. 2, was homologous (36-39%) with epidermal growth factor (EGF) (Gregory, Nature 25_7:325 (1975)) and the EGF-like repeat units found in Factor IX (Yoshitake et al., Biochem. £5.:3736 (1985)) and fibroblast proteoglycan core protein (Krusius et al., supra) (Fig. 3B) .

Immediately following these domains were two tandem domains of 62 amino acids each, given by residues 207-269 and 270-331 of Fig. 2, that were homologous with the short consensus repeat unit (SCR) that comprises the IL-2 receptor (Leonard et al.. Nature 311:626 (1984)), Factor XIII (Ichinose et al., Biochem. 2J5:4633 (1986)) and many C3/C4

binding proteins (Klickstein et al., J. Exp. Med. 16_5_:1095 (1987); and Morley et al., EMBO J. 2:153 (1984)). In contrast with all of the previously described SCR that contain four conserved Cys residues, these two SCR possessed six Cys residues. The four conserved Cys residues found in all SCR are indicated in Fig. 3C by (*) ; the additional conserved Cys found in LAM-l are indicated by (+) . Of the multiple SCR present in each of these proteins, the SCR with the highest homology to LAM-l is diagrammed (Fig. 3C) . A 15 amino acid spacer followed the short consensus repeat units, preceding the transmembrane domain.

The structure of the lyam-1 gene, which encodes the LAM-l protein, was determined by isolating overlapping genomic DNA clones that hybridized with a LAM-l cDNA probe. The Iyam-1 gene spans greater than 30 kilo base pairs (kb) of DNA and is composed of at least 10 exons. The 5* end of the LAM-l mRNA was mapped by primer extension analysis, revealing a single initiation region for transcription. Exons II through X contain translated sequences; exon II encodes the translation initiation codon, residue 14 shown in Fig. 2; exon III encodes the leader peptide domain, residues 15-41; exon IV encodes the lectin-like domain, residues 42-170; exon V encodes the epidermal growth factor¬ like domain, residues 171-206; exons VI and VII encodes the short consensus repeat unit domains, residues 207-269 and 270-331; exon VIII encodes the transmembrane region, residues 332-373; exon IX encodes seven amino acids containing a potential phosphorylation site, residues 374- 380; and exon X encodes the five remaining amino acids of the cytoplasmic tail and the long 3' untranslated region. The pLAM-l cDNA was labeled with P ³² and used as a probe to isolate hybridizing DNAs from a human leukocyte

genomic DNA library. Approximately 1 X 10 ⁶ plaques were screened, and 13 plaques that hybridized with the cDNA probe were identified and isolated. Seven of these clones were found to contain inserts with unique restriction enzyme maps representing overlapping genomic fragments spanning at least 30 kb. These inserts, LAMG-17, -19, -20, -28, -35, -37, and -47, were further digested and subcloned into plasmids. Detailed restriction maps of these subclones were made and compared to those of intact inserts to determine their correct locations (Figs. 4A and 4B) .

The correctness of the restriction map was verified with Southern blot analysis. DNA isolated from two B cell lines, BL and BJAB, and one T cell line, HSB-2, was digested to completion with Bam HI, Bal II, or Pvu II, size- fractionated, and transferred onto nitrocellulose. This filter was probed with the LAM-l cDNA clone, pLAM-1. All genomic fragments derived from endonuclease digested DNA hybridized with a cDNA probe to generate hybridizing bands of the appropriate size. The pLAM-1 cDNA clone encodes an 85-bp 5* untranslated region. An oligonucleotide homologous with the 5' sequence of the pLAM-l cDNA was used as a probe for primer extension analysis. This oligonucleotide was hybridized with poly (A ⁺ ) RNA isolated from the human B cell line Raji, the LAM-l negative human B cell line Namalwa, the mouse pre-B cell line A20, and yeast tRNA as a control. Complementary DNA was synthesized by extending the primer with reverse transcriptase. The major primer extension product obtained using the human LAM-l positive B cell line RNA was extended 126 nucleotides beyond the translation initiation site. There was a single cluster of transcription initiation sites for the Iya.n-2 gene apparent

in the reaction with Raji RNA that was not found with the LAM-l negative Namalwa RNA. Several primer extension products of size similar to those of the human B cell line RNA were obtained with mouse B cell RNA. Therefore, murine B cells may express an RNA species that cross-hybridizes with the oligonucleotide probe used. No primer extension products were obtained in the yeast tRNA control reactions. The relationship of the primer extension results to the cloned LAM-l cDNAs and the most 5' exon of lyam-1 isolated was used to determine the nucleotide sequence of the exon that encodes the translation initiation AUG condon. This exon ends immediately after the site that encodes the translation initiation codon (Fig. 5) and overlaps precisely with the pLAM-1 cDNA sequence. The length of the cDNA clone obtained by Bowen et al. , J. Cell Biol. 109:421-427 (1989) agrees precisely with the primer extension results except for two nucleotides. However, 15 nucleotides before the 5' end of the cDNA the sequence diverges from the genomic sequence at a site homologous with the 3 ^» splice acceptor site consensus sequence. Therefore, it is most likely that this 15-bp region is derived from the exon upstream of this potential splice site. Thus, the primer extension results indicate that exon I would most likely be composed of 15 or fewer base pairs. A 15-bp oligonucleotide homologous with the 5' nucleotides present in the cDNA clone of Bowen et al., supra but not encoded by exon II, was used to probe the 10 kb of cloned DNAs 5* of exon II; however, specific hybridization was not detected by Southern blot analysis. Under the conditions necessary for hybridization of this oligonucleotide, significant cross-hybridization occurred with λ-DNA, making it difficult to use this oligonucleotide to isolate the first exon from a λ-based genomic library.

These results suggest that the exon which encodes the translation initiation site is the second exon of the lyam-1 gene (Fig. 4C) . Consistent with this, the 900 bp upstream of exon II did not contain any apparent "TATA" or "CCAAT" sequences frequently found in promoter regions of eukaryotic genes (Fig. 5) . Therefore, it is likely that the transcription initiation region and exon I are further than 10 kb upstream from exon II of the Iyam-2 gene. SI nuclease protection analysis was carried out using the 5' region of exon II as a labeled probe for hybridization with poly (A ⁺ ) RNA from Raji, Namalwa, and A20 cells. Two mRNA species were protected in the Raji mRNA, while no SI protection was provided by the other RNAs. The length of these fragments was consistent with differential splicing at the two potential CAG/N splice cites located within the potential splice acceptor site in exon II. It is therefore likely that the transcription initiation region has not been identified.

The majority of the exons were localized by comparison of the restriction enzyme maps of the genomic clones and the pLAM-1 cDNA. In cases where this method did not provide definitive results, subcloned DNA fragments were digested with selected restriction enzymes, electrophoresed through agarose gels, and transferred to nitrocellulose. Fragments containing exons were identified by Southern blot analysis using labeled cDNA or oligonucleotide probes. The exon that encodes the 3' untranslated region of the LAM-l cDNA was not contained within the 30 kb of isolated DNA fragments. Therefore, a labeled 0.9-kb Dra I fragment containing most of the 3 ¹ untranslated region of the pLAM-1 cDNA was used as a probe to identify a homologous 3.2-kb fragment generated by complete Eco RI digestion of genomic

DNA. Eco RI-digested genomic DNA fragments of this size were used to make a partial λ-gtll genomic library from which the 3.2-kb Eco RI fragment was cloned. This 3.2-kb fragment did not overlap with the previously isolated genomic DNAs.

The exact boundaries of the exons were determined by nucletide sequence analysis. From this analysis, nine exons were identified which make up the entire pLAM-1 cDNA. Exon II encodes the translation initiation codon, and exon III encodes the leader domain of the LAM-l protein (Fig. 5) . Each of the lectin-like, epidermal growth factor-like, transmembrane, and short consensus repeat domains was encoded by a separate exon. The smallest exon, IX, is 19 bp in length and may encode a carboxyl-terminal phosphorylation cassette. The last 5 amino acids of the LAM-l protein and the 3 ¹ untranslated region which includes the poly(A) attachment site, AATAAA, are encoded by exon X as shown in Fig. 5. The nine exons which encode pLAM-1 were split inside codons in all cases, except the junction between exons II and III. In each instance, the consensus sequences of 5' donor splice sites and 3' acceptor splice sites were adhered to. Nucleotide sequence polymorphisms within the coding region were observed between the genomic clones containing exon V that encoded SCR I and the pLAM-1 clone at cDNA nucleotide positions 741 and 747 (A to G) , leading to a coding change from Asn to Ser in both cases, and at position 816 (A to G) changing the Glu to a Gly. Use

As leukocyte migration and infiltration into areas of tissue damage or injury or tissue transplant can cause or increase pathology, agents that impede these processes can be used for therapeutic treatment. Specifically, leukocyte- mediated inflammation is involved in a number of human

clinical manifestations, including the adult respiratory distress syndrome, multi-organ failure and reperfusion injury. One way of inhibiting this type of inflammatory response would be to block competitively the adhesive interactions between leukocytes and the endothelium adjacent to the inflamed region. As LAM-l mediates the migration and adhesion of blood leukocytes, treatment of a patient in shock, e.g., from a serious injury, with an antagonist to cell surface LAM-l function can result in the reduction of leukocyte migration to a level manageable by the target endothelial cells and the subsequent dramatic recovery of the patient.

Individual domains of LAM-l or the entire LAM-l protein can be used for therapeutic treatment. In addition, the LAM-l protein or a specific domain can be joined to a carrier protein to increase the serum half-life of the therapeutic agent. For example, a LAM-l fusion protein with human IgGl heavy chain, an antibody-like immunoglobulin chimera, was produced as follows: An altered fragment of LAM-l cDNA was produced that generated a Ban II endonuclease cleavage site within the exon VIII encoded domain by PCR using an antisense primer with the sequence GTTATAATCGGGCTCCTTAATC. This generated a Ban II sequence at nucleotide positions 1073-1078 (Fig. 2) . The LAM-l cDNA fragment encoding the leader, lectin, EGF-like and SCR domains plus this altered spacer domain were fused at this Ban II site to the Ban II site within the hinge region of a human IgGl heavy chain cDNA. This generated a cDNA encoding the LAM-l extracellular domains fused with the hinge, CH2, and CH3 domains of IgGl. This cDNA was subcloned into a mammalian expression vector and was subsequently transfected into COS-7 cells. The soluble dimeric LAM-l/IgGl protein product secreted into the medium retained all of the

antigenic epitopes of the LAM-l molecule defined by 16 anti- LAM-1 monoclonal antibodies. In a similar fashion a soluble immunoglobulin chimera can be obtained for each specific exon-encoded domain of LAM-l, or fragment thereof. The immunoglobulin chimera are easily purified through IgG- binding protein A-Sepharose chromatography. The chimera have the ability to form an immunoglobulin-like dimer with the concomitant higher avidity and serum half life.

Other agents can also be joined to the LAM-l protein or to a specific domain to form a useful product. For example, the LAM-l lectin domain can be combined with the toxic portion of a cytotoxin to produce a fusion protein. Also, a LAM-l protein domain can be combined with an imaging agent, e.g., a fluorescent agent, producing a fusion molecule useful for imaging sites of inflammation. The fusion proteins can be transcribed from a cDNA hybrid molecule, as described above, or the agent may be covalently bonded to the LAM-l protein or domain by routine procedures. In addition, subpopulations of malignant cells that express the LAM-l receptor protein would allow the receptor to function in metastasis of tumor cells. Agents developed to block receptor function can inhibit the metastasis and homing of malignant cells.

The therapeutic agents may be administered orally, parenterally, or topically by routine methods in pharmaceutically acceptable inert carrier substances. Optimal dosage and modes of administration can readily be determined by conventional protocols.

The lyam-1 gene itself can also be used in genetic therapy. Individuals having a genetic defect in the lyam-1 gene would be unable to produce a fully active LAM-l protein leukocyte "homing" receptor and thus would be unable to mobilize sufficient leukocytes to a site of inflammation.

Individuals suspected to having a congenital defect in the lyam-l gene could be screened for this genetic disorder using the sequence and structural information described. Treatment of affected individuals would then be possible using the lyam-1 gene or fragments thereof.

The normal regulation of the lyam-1 gene, as evidenced by the appearance and disappearance of the LAM-l protein on the surface of a specific leukocyte sub- population can be monitored to test the effects of drugs or specific therapies that would alter gene expression.

Other embodiments are within the following claims.

What is claimed is: