Continuous administration of soluble proteins, delivered over a 10-d period by a mini-osmotic pump implanted subcutaneously, induces a long-lasting inhibition of antigen-specific T cell proliferation in lymph node cells from BALB/c mice subsequently primed with antigen in adjuvant. The decreased T cell proliferative response is associated with a down-regulation of the T helper cell (Th)1 cytokines interleukin (IL)-2 and interferon (IFN)-gamma and with a strong increase in the secretion of the Th2 cytokines IL-4 and IL-5 by antigen specific CD4+ T cells. This is accompanied by predominant inhibition of antigen-specific antibody production of IgG2a and IgG2b, rather than IgG1 isotype. Interestingly, inhibition of Th1 and priming of Th2 cells is also induced in beta(2) microglobulin-deficient BALB/c mice, indicating that neither CD8+ nor CD4+ NK1.1+ T cells, respectively, are required. The polarization in Th2 cells is stably maintained by T cell lines, all composed of CD4+/CD8- cells expressing T cell receptor for antigen (TCR) alpha/beta chains, derived from BALB/c mice treated with continuous antigen administration, indicating that they originate from Th2 cells fully differentiated in vivo. This polarization is induced in BALB/c mice by continuous administration of any protein antigen tested, including soluble extracts from pathogenic microorganisms. Priming of Th2 cells is dose dependent and it is optimal for low rather than high doses of protein. Blocking endogenous IL-4 in vivo inhibits expansion of antigen-specific Th2 cells, but does not restore IFN-gamma production by T cells from mice treated with soluble antigen-specific Th2 cells, but does not restore IFN-gamma production by T cells from mice treated with soluble antigen, indicating the involvement of two independent mechanisms. Consistent with this, Th2 cell development, but not inhibition of Th1 cells, depends on non-major histocompatibility complex genetic predisposition, since the Th2 response is amplified in BALB/c as compared to DBA/2, C3H, or C57BL/6 mice whereas tested. These findings support the hypothesis that continuous release of low amounts of protein antigens from pathogenic microorganisms may polarize the immune response toward a Th2 phenotype in susceptible mouse strains.