Complement- and cell-mediated killing utilize related effector proteins (C8/C9 and perforin, respectively), suggesting that proteins which protect cells against complement- and cell-mediated attack may also be similar. In homologous complement-mediated killing two protective proteins, which are anchored to the cell membrane by phosphatidylinositol glycan (PIG) tails, are known. To study whether similar PIG-tailed proteins protect against lymphocyte-mediated killing, nucleated cell lines with a mutation in the biosynthesis of the PIG anchor were used. It was found that PIG-tailed membrane proteins restrict homologous complement-mediated lysis but not three different types of cell-mediated killing or lysis by purified perforin. Furthermore, E from patients with an acquired defect in PIG tail biosynthesis did not differ from normal E in sensitivity to antibody-dependent cell-mediated cytotoxicity, in spite of their increased sensitivity to human C8 and C9.

The leukocyte adhesion molecule LFA-1 mediates a wide range of lymphocyte, monocyte, natural killer cell, and granulocyte interactions with other cells in immunity and inflammation. LFA-1 (CD11a/CD18) is a receptor for intercellular adhesion molecule 1 (ICAM-1, CD54), a surface molecule which is constitutively expressed on some tissues and induced on other in inflammation. Induction of ICAM-1 on epithelial cells, endothelial cells and fibroblasts mediates LFA-1-dependent adhesion of lymphocytes. Several lines of evidence have suggested the existence of a second LFA-1 ligand: homotypic adhesion of one cell line was inhibited by a monoclonal antibody to LFA-1, but not by one to ICAM-1; there exists an LFA-1-dependent, ICAM-1-independent pathway of adhesion to endothelial cells; and also, there are some types of target cells in which LFA-1-dependent T-lymphocyte adhesion and lysis are independent of ICAM-1. We have cloned this second ligand, designated ICAM-2, using a novel method for identifying ligands of adhesion molecules. ICAM-2 is an integral membrane protein with two immunoglobulin-like domains, whereas ICAM-1 has five. Remarkably, ICAM-2 is much more closely related to the two most N-terminal domains of ICAM-1 (34% identity) than either ICAM-1 or ICAM-2 is to other members of the immunoglobulin superfamily, demonstrating the existence of a subfamily of immunoglobulin-like ligands that bind the same integrin receptor.

Antigen-independent binding of T lymphocytes to a variety of cell types has been shown to be mediated by receptor-ligand pairs of adhesion molecules. In forms of inflammatory synovitis (including rheumatoid arthritis), T cells home to synovium, become activated, and participate in the generation of chronic synovitis. Using indirect immunofluorescence assays on synovial frozen tissue sections and on synovial fibroblast cell lines, we studied the distribution of cell adhesion molecules on components of the synovial microenvironment in inflammatory synovitis. We reasoned that analysis of the cell types within synovium that express adhesion molecules might provide clues to lymphocyte-stromal interactions that occur in inflammatory synovitis. We found that antibodies against the lymphocyte function-associated antigen 3 (LFA-3) molecule and the intercellular adhesion molecule 1 (ICAM-1) both reacted with macrophage-like type A synovial cells and synovial fibroblasts, as well as with tissue macrophages and vessel endothelium. Using flow cytometry, we found that anti-LFA-3 and anti-ICAM-1 (but not antibodies against their ligands CD2 and LFA-1) reacted with synovial fibroblast cells cultured in vitro. Thus, these data demonstrate that the ligands for lymphocyte LFA-1 molecules (ICAM-1) and for T cell CD2 molecules (LFA-3) are widely distributed among cell types of the synovial microenvironment and provide numerous cell types with which lymphocytes can interact via these 2 adhesion pathways during the course of inflammatory synovitis.

Leukocyte adhesion deficiency (LAD) is a heritable deficiency of the LFA-1, Mac-1, p150,95 family of leukocyte alpha beta heterodimers (the leukocyte integrins). We have studied the defect in patients who synthesize an aberrantly small form of the beta subunit common to all three proteins. S1 nuclease protection showed the presence of a 90-nucleotide mismatch in RNA from patients and relatives, correlating with inheritance of the disease. Use of the Taq polymerase chain reaction to amplify this region of RNA after first strand cDNA synthesis and sequencing showed an in-frame deletion of 90 nucleotides in the extracellular domain. Thus, this highly conserved region, 63% and 53% identical in amino acid sequence to two other beta subunits of the integrin family, is required for association of the beta subunit with alpha subunits. The 90-nucleotide region corresponds to a single exon present in both the normal and patient genome. The patient DNA has a single G to C substitution in the 5' splice site. This results in the direct joining of nonconsecutive exons in an unusual type of abnormal RNA splicing. A small amount of normally spliced message, detected by S1 nuclease protection and Taq polymerase chain reaction, encodes a normal sized beta subunit which is surface-expressed and accounts for the low levels of leukocyte integrin expression observed in these patients, and hence the moderate phenotype.

Granulocyte and natural killer (NK) cell Fc receptors for immunoglobulin G (CD16) differ in only a few amino acids, yet have phosphatidylinositol glycan (PIG) or polypeptide membrane anchors, respectively. Mutagenesis shows that anchoring is regulated by a serine residue near the PIG anchor attachment site in the extracellular domain. The NK cell isoform was not expressed on the surface of COS cells unless cotransfected with a subunit that was expressed in NK cells and that was identical to the gamma subunit of the high affinity IgE Fc receptor (Fc epsilon RI). However, the CD16 sequence and not expression of the gamma subunit is dominant in regulating PIG reanchoring.

CD16 is a low affinity Fc gamma R III expressed on granulocytes, macrophages and large granular lymphocytes, the mediators of antibody-dependent cellular cytotoxicity and NK. The occupancy of CD16 by aggregated IgG on large granular lymphocytes induces expression of activation markers, release of inflammatory mediators and triggering of effector functions such as antibody-dependent cellular cytotoxicity. Recently we and others described that CD16 is anchored to the membrane of granulocytes via a phosphatidylinositol glycan moiety. Here we show that the CD16 molecule expressed on NK cells, cultured monocytes, and lung macrophages is not phosphatidylinositol glycan moiety anchored. It is not released with phosphatidylinositol-specific phospholipase C, and after removal of N-linked carbohydrate is 5 to 7 kDa larger than the granulocyte CD16 molecule, strongly suggesting the presence of transmembrane and cytoplasmic protein domains. Redirected killing of hybridoma targets expressing anti-CD16 surface Ig shows that NK cell CD16 is unable to do so. These findings demonstrate that NK cell and granulocyte CD16 have different membrane anchors and indicate that the type of membrane anchor is an important biologic mechanism for regulating the functional capacity of surface receptors.

The leukocyte function-associated molecule 1 (LFA-1, CD11a/CD18) is a membrane glycoprotein which functions in cell-cell adhesion by heterophilic interaction with intercellular adhesion molecule 1 (ICAM-1). LFA-1 consists of an alpha subunit (Mr = 180,000) and a beta subunit (Mr = 95,000). We report the molecular biology and protein sequence of the alpha subunit. Overlapping cDNAs containing 5,139 nucleotides were isolated using an oligonucleotide specified by tryptic peptide sequence. The mRNA of 5.5 kb is expressed in lymphoid and myeloid cells but not in a bladder carcinoma cell line. The protein has a 1,063-amino acid extracellular domain, a 29-amino acid transmembrane region, and a 53-amino acid cytoplasmic tail. The extracellular domain contains seven repeats. Repeats V-VII are in tandem and contain putative divalent cation binding sites. LFA-1 has significant homology to the members of the integrin superfamily, having 36% identity with the Mac-1 and p150,95 alpha subunits and 28% identity with other integrin alpha subunits. An insertion of approximately 200 amino acids is present in the NH2-terminal region of LFA-1. This "inserted/interactive" or I domain is also present in the p150,95 and Mac-1 alpha subunits but is absent from other integrin alpha subunits sequenced to date. The I domain has striking homology to three repeats in human von Willebrand factor, two repeats in chicken cartilage matrix protein, and a region of complement factor B. These structural features indicate a bipartite evolution from the integrin family and from an I domain family. These features may also correspond to relevant functional domains.

Effective interaction between T cells and their targets requires that recognition of specific antigen be coordinated with increased cell-cell adhesion. We show that antigen-receptor cross-linking increases the strength of the adhesion mechanism between lymphocyte function-associated molecule-1 (LFA-1) and intercellular adhesion molecules (ICAMs), with intracellular signals transmitted from the T-cell antigen receptor to the LFA-1 adhesion molecule. The increase in avidity is rapid and transient, providing a dynamic mechanism for antigen-specific regulation of lymphocyte adhesion and de-adhesion.

The cell surface expression and function of the LFA-1 ligand, intercellular adhesion molecule 1 (ICAM-1), on epidermal keratinocytes (EK) was studied. ICAM-1 expression on the surface of cultured EK was either absent or weak, but was induced by treating EK with rIFN-gamma or TNF for 4-48 h. IFN-gamma and TNF were synergistic. IFN-gamma treatment increased T lymphoblast adhesion from less than 2% to 20-40%, with a concentration dependence similar to that seen for ICAM-1 induction. All of the adhesion to EK was inhibited by LFA-1 and ICAM-1 mAbs, but not by HLA-DR, CD2, or LFA-3 mAbs. There was no difference in the level of T lymphoblast adhesion to IFN-gamma-treated allogeneic or autologous EK. ICAM-1 purified from the HeLa epithelioid cell line and reconstituted into planar membranes also supported efficient adhesion of T lymphoblasts that was blocked by LFA-1 mAb bound to the T lymphoblasts or ICAM-1 mAb bound to the planar membranes. T lymphoblasts adherent to EK or ICAM-1 planar membranes were isolated by panning, and surface markers were analyzed by immunofluorescence flow cytometry. The adherent T cells were a phenotypically skewed subpopulation. They were enriched for CD8+ cells and expressed 1.5-2.5-fold higher LFA-1 and CD2 compared with the unseparated population.

Mutants that lack expression of phosphatidylinositol (PI)-anchored proteins were derived from the human B lymphoblastoid JY cell line. It was demonstrated that unlike wild-type cells, which normally express both a transmembrane and a PI-linked form of LFA-3 glycoprotein, the mutant cells expressed only the transmembrane form of LFA-3. [3H]Ethanolamine was not incorporated into LFA-3 of mutant cells, indicating that the anchor moiety was entirely missing. Blockade of normal biosynthesis of the PI-anchored form led to accumulation of two intermediates that may have intact and truncated polypeptide chains. The truncated LFA-3, which was not attached to the cell membrane, was secreted by mutant cells into culture supernatants. A possible division of adhesion function between the two forms of LFA-3 was studied by using the JY cell lines as targets for CTL. Wild-type and mutant JY cells formed conjugates with CTL and were subsequently lysed to a similar extent. In addition, wild-type and mutant JY cells stimulated CTL proliferation to the same extent. Antibody-blocking experiments demonstrated a predominant role for the CD2/LFA-3 pathway in interaction of both wild-type and mutant cells with CTL. Because E exclusively express only the PI-linked LFA-3 form, and this form is known to mediate cell adhesion, the present results indicate that the two distinct membrane-anchored LFA-3 forms are each capable of mediating adhesion. A possible division of signaling functions between the two forms of LFA-3 is under investigation.