R esting natural killer (NK) cells express the p75 chain of the IL-2 receptor (IL-2R beta) and most NK cells express the CD2 (erythrocyte rosette) receptor. The cell adhesion molecule, LFA-3, is a natural co-ligand for CD2. Tac antigen (IL-2R alpha), a p55 IL-2R subunit, can be expressed after NK activation and may play a role in IL-2-induced NK proliferation. Little is known of the molecular mechanisms underlying cytokine production in NK cells. We investigated the roles of IL-2R alpha, IL-2R beta, and CD2/LFA-3 in the molecular regulation of NK cell granulocyte/macrophage-colony-stimulating factor (GM-CSF) production. Enriched populations of peripheral blood NK cells were separated into CD16-positive and CD16-negative fractions by flow cytometry; positively selected cells were greater than 97% positive for CD16 (the FcIII receptor for IgG which is present on almost all NK cells), less than 1% positive for the T cell antigen CD3, and did not demonstrate rearrangement of the T cell receptor beta chain gene by Southern blot. NK cell supernatants were harvested after 3-4 d of incubation with 0-100 U/ml IL-2, or after incubation with anti-CD2 (T11(3] MAb and sheep red blood cells (SRBC are a homologue for LFA-3). Parallel cell aliquots were harvested at 3-16 h for transcriptional run-on assays, S1 nuclease assays, and actinomycin D mRNA t1/2 determinations. IL-2-activated NK supernatants contained large amounts of GM-CSF (178 +/- 35 pg/ml) by ELISA as did supernatants from CD2-activated NK cells (T11(3) MAb + SRBC: 212 +/- 42) vs. less than 20 pg/ml for NK cells incubated alone or with either SRBC or T11(3) MAb alone. Sepharose-linked anti-CD3 MAb did not induce GM-CSF release from NK cells. By S1 analysis, both IL-2 and CD2 stimulation markedly augmented GM-CSF mRNA expression but with very different latencies of onset. IL-2R beta MAb inhibited greater than 85% of GM-CSF release from IL-2-activated NK cells and markedly suppressed IL-2-induced GM-CSF mRNA expression, whereas IL-2R alpha MAb even at 2,000-fold molar excess of IL-2 had little effect (less than 10%) on either GM-CSF release or mRNA expression. Run-on assays showed that GM-CSF is constitutively transcribed in NK cells and that IL-2 and CD2-activated cells had a three- to fourfold increased rate of GM-CSF transcription compared to nonstimulated cells. The t1/2 of GM-CSF mRNA in IL-2-activated NK cells was identical to that of unstimulated NK cells (15 min), whereas GM-CSF mRNA t1/2 in CD2-activated NK cells was increased 2.5-fold. We conclude that GM-CSF production in NK cells is regulated by both the IL-2Rbeta and the CD2 receptor but not by IL-2Ralpha, that both transcriptional and posttranscriptional signals act together to modulate the level of GM-CSF mRNA in NK cells, and that the molecular mechanisms underlying NK cell GM-CSF production are dependent in part on differential surface receptor activation.