The existence of a second receptor binding site on human prolactin (hPRL) was investigated by site-directed mutagenesis. First, 12 residues of helices 1 and 3 were mutated to alanine. Since none of the resulting mutants exhibit reduced bioactivity in the Nb2 cell proliferation bioassay, the mutated residues do not appear to be functionally necessary. Next, small residues surrounding the helix 1-helix 3 interface were replaced with Arg and/or Trp, the aim being to sterically hinder the second binding site. Several of these mutants exhibit only weak agonistic properties, supporting our hypothesis that the channel between helices 1 and 3 is involved in a second receptor binding site. We then analyzed the antagonistic and self-antagonistic properties of native hPRL and of several hPRLs analogs altered at binding site 1 or 2. Even at high concentrations (approximately 10 microM), no self-inhibition was observed with native hPRL; site 2 hPRL mutants self-antagonized while site 1 mutants did not. From these data, we propose a model of hPRL-PRL receptor interaction which slightly differs from that proposed earlier for the homologous human growth hormone (hGH) (Fuh, G., Cunningham, B. C., Fukunaga, R., Nagata, S., and Goeddel, D. V., and Well, J. A. (1992) Science 256, 1677-1680). Like hGH, hPRL would bind sequentially to two receptor molecules, first through site 1, then through site 2, but we would expect the two sites of hPRL to display, unlike the two binding sites of hGH, about the same binding affinity, thus preventing self-antagonism at high concentrations.