Major histocompatibility complex (MHC) class II molecules have been implicated in cell adhesion in two ways. In addition to the well-established role of class II antigens in low-affinity adhesion provided by interactions between class II and CD4, recent data indicated that class II may also induce adhesion between T and B cells by activating the CD18/CD11a (LFA-1) adhesion pathway. Here we report that monoclonal antibodies (mAb) against HLA-DR (L243, p4.1, HB10a, VI15) and certain broad class II reacting mAb (TU35, TU39), but not anti-DQ (TU22, Leu-10) mAb, induced homotypic aggregation of human class II-positive monocytic (I937) and T leukemic (HUT78) tumor cell lines and Epstein-Barr virus (EBV) transformed B-lymphoid cell lines (EBV-LCL). Class II-negative cell lines (U-937 and the EBV-LCL mutant line 616) were not induced to aggregate. An HLA-G-transfected EBV-LCL, 221-AGN, but not the class I-negative parental line, 221, showed homotypic aggregation in response to an HLA-G specific mAb (87G) and a broad reacting class I-specific mAb (IOT2). Both cell lines responded with aggregation to anti-class II mAb (TU35). The anti-class I mAb, W6/32, had no effect on all cell lines tested and two anti-beta 2-microglobulin mAb had variable, weak effects. The aggregation response was an active, temperature-sensitive process which was almost totally abrogated by azide and by cytochalasins B and E, but unaffected by colchicine, EDTA, aphidicolin, actinomycin D and protein tyrosine kinase inhibitors (genistein, herbimycin A). Serine/threonine protein kinase inhibitors (staurosporin, H7) partly inhibited the aggregation responses. There was no strict correlation between induction of aggregation and epitope density. FcR were not involved in the aggregation response, since F(ab')2 fragments of anti-DR mAb, L243, were as effective as the whole antibody. The aggregation was not influenced by mAb against accessory molecules previously shown to be involved directly or indirectly in homotypic aggregation [CD11a (LFA-1)/CD18/CD54 (ICAM-1), CD58 (LFA-3)/CD2, BB1/CD28, CD43, and CD44]. In conclusion, these data provide further evidence that HLA molecules are implicated in a novel, cellular aggregation phenomenon involving the cytoskeleton.