We have determined the relative abilities of several members of the G protein  and ␥ subunit families to associate with each other using the yeast two-hybrid system. We show first that the mammalian 1 and ␥3 fusion proteins form a complex in yeast and that formation of the complex activates the reporter gene for -galactosidase. Second, the magnitude of reporter activity stimulated by various combinations of  and ␥ subunit types varies widely. Third, the reporter activity evoked by a particular combination of  and␥ subunit types is not correlated with the expression levels of these subunit types in the yeast cells. Finally, the reporter activity shows a direct relationship with the amount of hybrid ␥ complex formed in the cell as determined by immunoprecipitation. These results suggest that different  and ␥ subunit types interact with each other with widely varying abilities, and this in combination with the level of expression of a subunit type in a mammalian cell determines which G protein will be active in that cell. The strong preference of all ␥ subunit types for the 1 subunit type explains the preponderence of this subunit type in most G proteins.Most of the neurohormonal signaling pathways in mammals are mediated by heterotrimeric G proteins (1, 2). Most cells contain many different G protein subunit types, and yet particular agonists evoke a highly specific response in a cell by activating a defined G protein-mediated signaling pathway (3). Various mechanisms can contribute to this specificity. For instance, in a cell that contains many different ␣, , and ␥ subunit types, only certain types may be capable of forming a heterotrimeric complex because of the differences in the intrinsic affinity of these subunit types for one another. It has been shown that interactions between two different ␥ subunit types (␥1 and ␥2) and three different  subunit types (1-3) are selective, indicating that this is indeed a mechanism for achieving specificity (4, 5). However, these experiments were performed using in vivo or in vitro systems that could detect differences in the ability of these subunit types to interact but were not sufficiently sensitive to detect low level interaction between some of the subunit types. To obtain a more sensitive measure of the interaction between various members of the  and ␥ subunit families, we used the yeast two-hybrid system. In this system, the proteins with the potential to interact are expressed as hybrids of two different domains of a transcription factor. If the proteins interact with each other, the transcription factor domains are in proximity and capable of activating a promoter that controls reporter activity (6). We chose this system because it measures protein-protein interaction in a yeast cell and is therefore a reasonably accurate reflection of the ability of the interacting proteins to form a complex in a cell. Furthermore, it is highly sensitive in comparison with the methods used before to measure protein-protein interaction. For instance, in an assay of in...