No abstract

In illuminated rod outer segment membranes, GTP and guanosine 5'-[fyl-imido]triphosphate (p[NH]ppG) have reciprocal effects on cGMP phosphodiesterase (PDEase; 3':5'-cyclic-nucleotide 5'-nucleotidohydrolase, EC 3.1.4.17) activity and cGMP binding to noncatalytic-sites on that enzyme. Two micromolar p[NH]ppG increased PDEase activity more than 2-fold while inhibiting cGMP binding more than 40%. Reduction of noncatalytic cGMP binding, which followed addition of p[NH]ppG, was not a result of PDEase activation. Both effects of p[NH]ppG were completely dependent on the presence of bleached rhodopsin. A heat-stable factor has been found to inhibit PDEase activity and also to stimulate cGMP binding to noncatalytic cGMP binding sites. Addition of p[NH]ppG reversed the effects of this factor on both PDEase activity and cGMP binding. During purification of this material, the activity peaks for both PDEase inhibition and activation of noncatalytic cGMP binding comigrated on both Blue Sepharose CL-6B column chromatography and sucrose density gradients centrifugation, suggesting that the same factor could be responsible for both inhibition of PDEase activity and enhancement of noncatalytic cGMP binding. Limited tryptic proteolysis of PDEase, which markedly reduced cGMP binding to the noncatalytic sites, and experiments using highly purified cAMP (free of cGMP) as substrate for PDEase showed that the binding of cGMP to noncatalytic sites was not required for the heat-stable inhibitory factor to inhibit PDEase activity. We discuss possible relationships between the regulation of PDEase and the binding of cGMP to noncatalytic sites.

This paper introduces a methodology for neural network global optimization. The aim is the simultaneous optimization of multilayer perceptron (MLP) network weights and architectures, in order to generate topologies with few connections and high classification performance for any data sets. The approach combines the advantages of simulated annealing, tabu search and the backpropagation training algorithm in order to generate an automatic process for producing networks with high classification performance and low complexity. Experimental results obtained with four classification problems and one prediction problem has shown to be better than those obtained by the most commonly used optimization techniques.

Guanylate cyclase from rod photoreceptors of amphibian (toad, Bufo marinus, and frog, Rana catesbiana) and bovine retinas was solubilized and purified by a single chromatography step on a GTP-agarose column. Silver staining ofpurified amphibian enzymes in SDS/polyacrylamide gels disclosed a doublet band (110 and 115 kDa), while the bovine enzyme appeared as a singlet band (110 kDa). The identification of these guanylate cyclases was confirmed using three chromatography systems with the purified enzymes. Specific binding to Con A-Sepharose suggested that rod guanylate cyclase is a glycoprotein. Two-dimensional gel electrophoresis of purified toad, frog, and bovine enzymes resolved two, three, and five variants, respectively, that differed in isoelectric point. Two variants of toad guanylate cyclase showed differences in various characterizations. These data suggest multiple mech-

Retinal guanylyl cyclase-1 (retGC-1), a key enzyme in phototransduction, is activated by guanylyl cyclase-activating proteins (GCAPs) if [Ca 2؉ ] is less than 300 nM. The activation is believed to be essential for the recovery of photoreceptors to the dark state; however, the molecular mechanism of the activation is unknown. Here, we report that dimerization of retGC-1 is involved in its activation by GCAPs. The GC activity and the formation of a 210-kDa cross-linked product of retGC-1 were monitored in bovine rod outer segment homogenates, GCAPs-free bovine rod outer segment membranes and recombinant bovine retGC-1 expressed in COS-7 cells. In addition to recombinant bovine GCAPs, constitutively active mutants of GCAPs that activate retGC-1 in a [Ca 2؉ ]-independent manner and bovine brain S100b that activates retGC-1 in the presence of ϳ10 M [Ca 2؉ ] were used to investigate whether these activations take place through a similar mechanism, and whether [Ca 2؉ ] is directly involved in the dimerization. We found that a monomeric form of retGC-1 (ϳ110 kDa) was mainly observed whenever GC activity was at basal or low levels. However, the 210-kDa product was increased whenever the GC activity was stimulated by any Ca 2؉ -binding proteins used. We also found that [Ca 2؉ ] did not directly regulate the formation of the 210-kDa product. The 210-kDa product was detected in a purified GC preparation and did not contain GCAPs even when the formation of the 210-kDa product was stimulated by GCAPs. These data strongly suggest that the 210-kDa cross-linked product is a homodimer of retGC-1. We conclude that inactive retGC-1 is predominantly a monomeric form, and that dimerization of retGC-1 may be an essential step for its activation by active forms of GCAPs.