Defective learning in mutants of the Drosophila gene for a regulatory subunit of cAMP-dependent protein kinase

Abstract

Disruptions of a Drosophila gene encoding a regulatory subunit of cAMP-dependent protein kinase homologous to mammalian RIbeta (dPKA-RI) were targeted to the first (noncoding) exon of dPKA-RI via site-selected P element mutagenesis. Flies homozygous for either of two mutant alleles showed specific defects in olfactory learning but not in subsequent memory decay. In contrast, olfactory acuity and shock reactivity, component behaviors required for normal odor avoidance learning, were normal in these mutants. Northern and Western blot analyses of mRNA and protein extracted from adult heads have revealed a complex lesion of the PKA-RI locus, including expression of a novel product and over- or underexpression of wild-type products in mutants. Western blot analysis revealed reductions in RI protein in mutants. PKA activity in the absence of exogenous cAMP also was significantly higher than normal in homogenates from mutant adult heads. These two mutant alleles failed to complement each other for each of these phenotypic defects, eliminating second-site mutations as a possible explanation. These results establish a role for an RI regulatory subunit of PKA in Pavlovian olfactory conditioning.

Fig. 2.

Structure of PKA-RI gene and its relation to the P element insertional mutations. A, Structure of the PKA-RI gene (after Kalderon and Rubin, 1988). λgS8 is a DNA clone recovered from an EMBL3 genomic library, which was used to characterize mutants ofPKA-RI. pcD6BO is a plasmid recovered from a cDNA library, which corresponds to a class Ia transcript and which was used as a probe to recognize all known transcripts ofPKA-RI (see text). Locus depicts the intron and exon (black boxes) structure of thePKA-RI locus. Arrows above exons I, II, and IV indicate alternative transcription start sites for class I, II, and IV messages. The start site for the class III message has not yet been determined. Ia, Ib,Ic, II, III, andIV represent exon maps of the six knownPKA-RI transcripts. AUG indicates the presumed translation start sites for each. B, Detail of first exon of PKA-RI. Two identical P elements (triangle) inserted independently near the 5′ end of exon I (black box) in a region of heterogeneous transcription start sites (stippled box) upstream of the presumed translation start site for class I RI isoforms.77F3, P_L,P_R, 77F1, and77F2 represent PCR primers used to identify and characterize the PKA-RI mutants (see Materials and Methods). 77F1/77F3 represents a PCR product used as a probe for class I-specific transcripts (see Fig.4A).

Fig. 3.

A, Learning and memory retention of a conditioned odor avoidance response in wild-type flies, homozygous7I5 or 11D4 mutants, and heteroallelic7I5/11D4 mutants. The PI is a function of the percentage of flies that avoided the shock-associated odor versus a control odor in a T maze. A PI of 0 represents a 50:50 distribution (no learning); a PI of 100 represents a 0:100 distribution (strong learning). Planned comparisons between group means at each retention time after a two-way ANOVA (see Materials and Methods) revealed significant differences between mutant (11D4, 7I5, and7I5/11D4) versus wild-type (Can-S) flies (all p values < 0.001) but no differences among the mutants (all p values > 0.04). For each group, n = 6 PIs, exceptn = 12 PIs for Can-S at 0, 15, 30, 60, and 180 min retention. Error bars indicate SEM. B, Olfactory acuity in untrained wild-type flies and in homozygous7I5 or 11D4 mutants. The ability of flies to smell the odors used during conditioning experiments was assayed by giving naive flies a choice between each odor (OCT orMCH) versus air in the T maze. A PI was calculated in a manner similar to that described above. Planned comparisons between group means for each odor after a two-way ANOVA (see Materials and Methods) revealed no significant differences between mutant (11D4 or 7I5) versus wild-type (Can-S) flies (all pvalues > 0.15). For each group, n = 16 PIs. Error bars indicate SEM. C, Shock reactivity in untrained wild-type flies and in homozygous 7I5 or11D4 mutants. The ability of flies to sense the electroshock used during conditioning experiments and to escape from it was assayed by giving naive flies a choice between an electrified arm versus an unelectrified arm in the T maze. A PI was calculated in a manner similar to that described above. Planned comparisons between group means for each odor after a one-way ANOVA (see Materials and Methods) revealed no significant differences between mutant (11D4 or 7I5) versus wild-type (Can-S) flies (both p values > 0.43). For each group, n = 8 PIs. Error bars indicate SEM.

Fig. 1.

In situ hybridization toPKA-RI transcripts in the adult brain (200× magnification). Twelve micrometer cryostat frontal sections of wild-type (Canton-S) adult heads (male and females; no differences were observed) probed with pcD6BO (see Fig. 2). A–C, Antisense probe showing PKA-RI expression in three serial sections of dorsal brain via mushroom bodies (c, calyx of mushroom body, which is surrounded by Kenyon cell bodies;OL, optic lobe; CB, central brain).D, Sense probe showing no nonspecific signal.

Fig. 4.

Aberrant gene expression in PKA-RImutants. A, Northern blot analysis of poly(A⁺) RNA extracted from heads of Canton-S (Can-S) flies, homozygousRI^7I5 andRI^11D4 mutants, and heteroallelicRI^7I5/RI^11D4mutants and probed with a PCR product specific to class I transcripts (see Fig. 2). Signals of rp49 were used for normalization; PKA-RI band intensities were quantified by phosphorimage analysis, indicating a 150% net increase in class I transcripts in mutants (see text). A novel 4.8 kb RNA species also was detected in mutant extracts. B, P element-specific probe hybridized only to the novel 4.8 kb transcript in mutants on a Northern blot similar to that described in A, indicating transcription of an aberrant PKA-RI message from within the P element. C, Western blot analysis of total protein extracted from heads of Canton-S (Can-S) flies, homozygous RI^7I5 andRI^11D4 mutants, and heteroallelicRI^7I5/RI^11D4mutants and probed with a rabbit polyclonal antibody to bacterially expressed class I RI protein. Although cross-reacting with several proteins, this anti-RI recognized three proteins at 50, 48, and ∼40 kDa, which were expressed at lower-than-normal levels in mutants. In contrast, a fourth protein was recognized near 84 kDa, which was expressed at a higher-than-normal level in mutants. Also apparent were several other cross-reacting proteins, which were expressed at similar levels among wild-type and mutant flies. Anti-α-tubulin (α-tub) was used as a loading control for this experiment.

Fig. 5.

A neuronal model for olfactory associative learning in Drosophila, involving the cAMP signal transduction pathway. Sensory input from olfactory cues produces neural activity in mushroom body neurons (MBNs), producing an increase in intracellular calcium. Sensory input from footshock also activates MBNs via a modulatory neuron, which releases dopamine (DA) or serotonin (5HT) synthesized by the Dopa decarboxylase(Ddc) gene. This monoamine neurotransmitter binds to its postsynaptic receptor (R?) on the MBNsand activates a rutabaga (rut)-encoded adenylyl cyclase (AC) via a stimulatory subunit of G-protein (Gs), which itself is encoded by theDrosophila Gsα (dGsα) gene. Coincident activation of AC by calcium and G-protein leads to a synergistic increase in intracellular cAMP, which is hydrolyzed by dunce (dnc)-encoded phosphodiesterase (PDE). Increased cAMP levels may be involved in several intracellular effects (?), but nevertheless cAMP binds to an RI regulatory subunit of PKA (RI), encoded by the Drosophila PKA-RI (dPKA-RI) gene. cAMP binding causes RI dimers to disassociate from dimers of the PKA catalytic subunit (PKA), encoded by DC0. FreePKA then is able to phosphorylate many cytoplasmic targets (?), one of which is a potassium channel (K⁺ channel) composed of Shaker subunits. Free PKAalso is translocated to the nuclei of MBNs, where it phosphorylates the CREB transcription factor (CREB) encoded by dCREB2. Phosphorylated CREBthen initiates a cascade of gene expression that produces gene products (?) involved with long-term functional and structural changes at MBN synapses. The neuronal effect of these biochemical changes is to increase transmitter efficacy between MBNs and their follower neurons, which mediate the motor output responsible for conditioned odor avoidance responses. This model assumes MBNs to be sights of associative learning because chemical ablation of MBNs or molecular disruption of G-protein function in MBNs completely abolishes olfactory associative learning. Protein components of the signal transduction pathway are labeled only after disruptions of identified genes are shown to produce defects in olfactory associative learning that cannot be attributed to nonspecific effects on the sensory or motor responses required to perform this task. Other neural substrates or signal transduction pathways will be added to this working model as they are revealed in future studies (see text for more details).