Calmodulin (CaM)-kinase II holoenzymes composed of either alpha or beta subunits were analyzed using light scattering to determine a mechanism for self-association. Under identical reaction conditions, only alphaCaM-kinase II holoenzymes self-associated. Self-association was detected at a remarkably low enzyme concentration (0.14 microM or 7 microg/mL). Light scattering revealed two phases of self-association: a rapid rise that peaked, followed by a slower decrease that stabilized after 2-3 min. Electron microscopy identified that the rapid rise in scattering was due to the formation of loosely packed clusters of holoenzymes that undergo further association into large complexes of several microns in diameter over time. Self-association required activation by Ca(2+)/CaM and was strongly dependent on pH. Self-association was not detected at pH 7.5, however, the extent of this process increased as reaction pH decreased below 7.0. A peptide substrate (autocamtide-2) and inhibitor (AIP) designed from the autoregulatory domain of CaM-kinase II potently prevented self-association, whereas the peptide substrate syntide-2 did not. Thus, CaM-kinase II self-association is isoform specific, regulated by the conditions of activation, and is inhibited by peptides that bind to the catalytic domain likely via their autoregulatory-like sequence. A model for CaM-kinase II self-association is presented whereby catalytic domains in one holoenzyme interact with the regulatory domains in neighboring holoenzymes. These intersubunit-interholoenzyme autoinhibitory interactions could contribute to both the translocation and inactivation of CaM-kinase II previously reported in models of ischemia.