It has become axiomatic that, in contrast to other protein kinases, cAMP-dependent protein kinase (cAPK) is activated only when its catalytic (C) and regulatory (RII2) subunits dissociate. To directly evaluate this postulation, the ability of cAMP to dissociate the holoenzyme form of cAPK was examined by measuring the rotational mobility of the carboxyfluorescein-labeled C subunit (CFC) complexed to the dimeric RII2 regulatory subunit under equilibrium conditions. The rotational mobility was determined from an analysis of the time-resolved emission anisotropy of the CFC subunit. The time-resolved anisotropy decays were best fitted by a sum of two exponentials for both the free CFC subunit and the RII2CFC2 complex (holoenzyme). In the absence of cAMP, the two rotational correlation times (phi F and phi S) were 1.7 +/- 0.3 and 18.3 +/- 0.7 ns for the free CFC subunit and 2.3 +/- 0.1 and 93 +/- 2 ns for the RII2CFC2 complex, respectively. The faster rotational correlation times can be attributed to the localized rotations of the label and the slower rotational correlation times to the global rotations of the entire molecule. The addition of cAMP had no significant effect on either the fast or the slow rotational correlation time of the RII2CFC2 complex (phi F = 2.0 +/- 0.2 ns and phi S = 93 +/- 9 ns). Control experiments established that the RII2CFC2 complex was fully activated by cAMP at the same concentrations (0.2-0.4 microM) used for the anisotropy measurements. Together, the results demonstrate (1) that cAMP can induce the catalytic activity of cAPK without subunit dissociation and (2) that cAMP binding to holoenzyme is insufficient to explain its in vivo dissociation.