The feasibility of long-term information storage by brain type II Ca2+/calmodulin-dependent protein kinase molecules is explored. Recent evidence indicates that this protein has switch-like properties. Equations are derived showing that a single kinase holoenzyme could form a bistable switch having the stability necessary to encode long-term memory, and that a group of kinase molecules, such as that contained within the postsynaptic density, could form a device capable of storing graded information.It is generally thought that long-term memory involves changes in the strength of synaptic connections, but the mechanism by which these synaptic "weights" might be stored is completely unknown. One view (1) (7)(8)(9), has shown that this kinase has properties with some important similarities to the hypothetical kinase mentioned above, but with some significant differences. The purpose of this paper is to explore models based specifically on properties of the Ca2+/calmodulin-dependent protein kinase type II. The principal conclusion is that it is theoretically possible for a single kinase molecule of this kind to store "on-off" information with the stability necessary to encode long-term memory and for a group of kinase molecules, such as that contained within the postsynaptic density, to form a memory storage organelle capable of storing graded information about synaptic weights. Miller and Kennedy (6) found that Ca2" is required only for the addition of the first two to four phosphates (the exact number being uncertain). The addition of these phosphates switches the holoenzyme into a Ca2+-independent "on" state in which both the autophosphorylation of the remaining sites and the phosphorylation of other substrates can proceed even in the absence of Ca2".Given these properties, a rise in Ca2+ caused by a key neuronal event would be expected to turn "on" kinase molecules. For the kinase to remain "on" after Ca2" returned to baseline, the kinase would have to resist two processes that would tend to reset it. The first of these is dephosphorylation of the holoenzyme by protein phosphatases. The second is protein turnover, which, by replacing phosphorylated protein with newly synthesized unphosphorylated protein, has an effect identical to that of the phosphatase. The resetting effect of the phosphatase itself could be avoided if the rate constant of Ca2+-independent autophosphorylation (kI) were significantly greater than the rate constant of phosphatase activity (k2); under these conditions, phosphates would be added as fast as they are removed and the holoenzyme would therefore stay in the fully phosphorylated "on" state. To avoid the resetting effects of protein turnover, it is necessary that newly synthesized, unphosphorylated, protein be phosphorylated ifthe protein it replaces was phosphorylated. As previously postulated (3), this could be accomplished if active kinase molecules phosphorylated newly synthesized molecules through an intermolecular reaction. Miller and Kennedy (6) did not detect significan...

The influence of voltage-dependent conductances on the receptor potential of Limulus ventral photoreceptors was investigated . During prolonged, bright illumination, the receptor potential consists of an initial transient phase followed by a smaller plateau phase. Generally, a spike appears on the rising edge of the transient phase, and often a dip occurs between the transient and plateau . Block of the rapidly inactivating outward current, iA, by 4-aminopyridine eliminates the dip under some conditions . Block of maintained outward current by internal tetraethylammonium increases the height of the plateau phase, but does not eliminate the dip. Block of the voltage-dependent Na' and Ca

Limulus ventral photoreceptors generate highly variable responses to the absorption of single photons. We have obtained data on the size distribution of these responses, derived the distribution predicted from simple transduction cascade models and compared the theory and data. In the simplest of models, the active state of the visual pigment (defined by its ability to activate G protein) is turned off in a single reaction. The output of such a cascade is predicted to be highly variable, largely because of stochastic variation in the number of G proteins activated. The exact distribution predicted is exponential, but we find that an exponential does not adequately account for the data. The data agree much better with the predictions of a cascade model in which the active state of the visual pigment is turned off by a multi-step process.

No abstract

No abstract