Ultrastructural evidence that the granules of human natural killer cell clones store membrane in a nonbilayer phase.

Electron-microscopic examination of five LGL clones, JT3, JTB18, CNK6, CNK7, and CNK10, expressing natural killer activity and T11 and NKH1 phenotype, showed that three of the clones, JT3, CNK6, and CNK7, had crystalline structures in their densest granules. These structures generally consisted of hexagonally packed lattices with a 6.9-nm point-to-point spacing. JTB18 and CNK10 had no structures in their granules. The attack of one clone, JT3, on two resistant target tumor cell lines, KG1 and Laz509, was also examined under three conditions. First, JT3 cells and targets were incubated together. There was little adherence, degranulation, or killing. Second, cells were incubated with anti-T11(2) and T11(3), antibodies against the E-rosette receptor/antigen complex, which activate resting T cells and enhance cytolytic activity of NK clones and CTL. JT3 cells adhered to the targets, formed zones of apposition between NK and target cell membranes, polarized, and degranulated into the space between the two cells, killing the targets. Third, cells were incubated with both anti-T11(2/3) and anti-LFA-1, an antibody that inhibits adherence. The JT3 cells did not form zones of apposition with the targets, but degranulated in discrete areas on their own surface. In all cases, discharged crystalline granules transformed to sheets of membrane and vesicles. These studies suggest that phospholipids are packed in hexagonal lattices in the granules of the resting cells and transform to bilayer structures during exocytosis. The crystalline nature of the granule may immobilize lytic molecules and protect the resting cell from lysis. Further, the vesicles may serve to transport the lytic molecules from the effector to the target cell. Anti-LFA-1 does not inhibit target recognition or exocytosis, but instead blocks membrane interactions of the effector cell with its target.