Natural Behavior Polymorphism Due to a cGMP-Dependent Protein Kinase of Drosophila

See allHide authors and affiliations

Science  08 Aug 1997:
Vol. 277, Issue 5327, pp. 834-836
DOI: 10.1126/science.277.5327.834


Naturally occuring polymorphisms in behavior are difficult to map genetically and thus are refractory to molecular characterization. An exception is the foraging gene (for), a gene that has two naturally occurring variants in Drosophila melanogaster food-search behavior: rover and sitter. Molecular mapping placed for mutations in the dg2 gene, which encodes a cyclic guanosine monophosphate (cGMP)–dependent protein kinase (PKG). Rovers had higher PKG activity than sitters, and transgenic sitters expressing a dg2 complementary DNA from rover showed transformation of behavior to rover. Thus, PKG levels affected food-search behavior, and natural variation in PKG activity accounted for a behavioral polymorphism.

The molecular identities of genetic polymorphisms in behavior have been difficult to establish because these traits are usually inherited polygenically. One example of a single gene underlying a naturally occurring polymorphism is the foraging gene (for), which is involved in food-search behavior in the fruit fly Drosophila melanogaster (1, 2). Individuals with a rover allelefor R move greater distances while feeding than do those homozygous for sitter allelesfor s (3). This difference in foraging behavior is observed during both the larval and adult stages (4). Rovers and sitters do not differ in general activity in the absence of food (4). Both rovers and sitters are wild-type forms that exist at appreciable frequencies (70% rover; 30% sitter) in natural populations (1, 5). Several mutations (6) of the locus map with the naturally occurring alleles in the 24A3-5 region of the D. melanogaster polytene chromosomes (2, 7). This region contains dg2, one of two cGMP-dependent protein kinase (PKG) genes inDrosophila (8). We report that (i) mutations infor mapped in or near dg2, (ii) excision of a P-element inserted into the dg2 gene reverted the sitter phenotype to rover, (iii) wild-type for sflies and all sitter mutants showed a decrease in PKG activity and level compared to the wild-type for R, and (iv) dg2 transgenes rescued rover larval behavior. These results demonstrate that for is dg2.

The dg2 gene has three major transcripts, T1, T2, and T3 (8), and the for mutations are localized to this region (Fig. 1) (9). The P[GAL4] (10) transposable element in 189Y was inserted in the 5′ end of the dg2 T2 transcript. This homozygous viable insertion identified a new for allele, because P-element excision reverted larval foraging behavior from a sitter to a rover phenotype (Table 1). As was the case with other sitter alleles, locomotion of the 189Y larvae was not reduced in the absence of food, indicating that the change in behavior was foraging-specific.

Figure 1

Three major dg2 transcripts are shown on the molecular map of dg2. They share a common 3′ region that encodes the kinase domain (8). Genomic clones used in the chromosome walk by D. Kalderon (8) are labeled a4, a6, c7, and RSac6, whereas those used in our walk are labeled 1I, 1A, and 1B. Mutants were mapped with restriction fragment length polymorphism analysis. Sitter for mutants cluster withindg2. for s1 andfor s2 were generated in afor R genetic background (2,4), and 189Y is a P[GAL4] enhancer trap strain with an insert in dg2.

Table 1

Foraging behavior and general locomotion offor 189Y and the excision strain E1measured as the mean ± SE (with sample size in parentheses) larval trail lengths in centimeters. The P-element in 189Y was excised with a source of transposase using the Δ2-3system (27). Excisions of the P-element were generated with theEmbedded Image

View this table:

PKG enzyme assays were performed on adult heads of wild-typefor R andfor s, and mutantfor s1 andfor s2 strains (Table2). for Rflies had significantly higher amounts of PKG enzyme activity than didfor s flies. Even greater reductions in enzymatic activity were seen in the mutantsfor s1 andfor s2. The amount of PKG in adult heads correlated with the adult foraging phenotypes of these strains (4).

Table 2

Adult heads from flies with rover and sitter alleles show significant differences in their PKG activity (measured in pmol/min per milligram protein).

View this table:

To determine whether PKG is directly responsible for the foraging polymorphism in Drosophila, we overexpressed dg2in sitter larvae. This resulted in a change of behavior to the rover phenotype. The transgenic strain contained four copies of a heat shock– driven dg2-cDNA (11). The basal level of PKG expression in this transgenic strain (Fig.2) was sufficient to rescue rover larval behavior, thus eliminating the lethal and sublethal effects of heat on the dg2-transgenic larvae (Table3). As expected, the PKG enzyme activities of the dissected larval central nervous systems (CNSs) showed that without heat shock, the dg2-cDNA transgenic strain had levels of PKG similar to those offor R and significantly higher than those of the sitter control strain (Table 3).

Figure 2

The transgenic dg2-cDNA strain that carries T2 cDNA shows some expression in the absence of heat shock and strong overexpression under heat-shock conditions relative to the control w 1;for sstrain. Heat-shock protocol is in Table 3 legend. RP49probe was used as a control (30).

Table 3

Larval foraging trail lengths (in centimeters) indg2-cDNA transgenic flies and PKG enzyme activity assay levels (in pmol/min per milligram protein) in the larval CNS. Results are given as mean ± SE, with sample size in parentheses.

View this table:

The basis for the dg2 activity difference betweenfor R andfor s was further addressed by measurement of RNA levels and PKG protein. Northern (RNA) analysis revealed that for s andfor s2 showed a small but consistent (12) reduction in the abundance of T1 RNA relative to that in for R (Fig.3A). T2 and T3 RNA were also reduced in these strains, but to a lesser extent (12). To assess protein levels, we subjected extracts of adult heads to protein immunoblot analysis by probing with an antibody to bovine PKG, or the extracts were affinity-purified by chromatography on cGMP-sepharose, labeled, and electrophoresed (13). In both experiments, a prominent band at a molecular mass of 80,000 Daltons was found. This was the only band strongly induced by heat shock in thedg2-cDNA transgenic strain, and it was less intense infor s thanfor R (Fig. 3, B and C). (This band was also somewhat less intense in for s2 and nearly absent in 189Y homozygotes). Taken together, these results argue that the difference between the naturally occurring allelesfor R andfor s is in the level of expression of the enzyme.

Figure 3

(A) Northern analysis of polyadenylated RNA from homozygous adult flies probed with a 1.5-kb fragment specific to T1 and T3 (4.6 and 3.6 kb). Thefor s naturally occurring allele and thefor s2 mutation caused a reduction in the abundance of the T1 transcript relative tofor R. RP49 probe used as a control (30). (B) Protein immunoblot of adult head extracts probed with antibody to bovine PKG (13). Lanes 1, for R; lane 2,dg2-cDNA after heat-shock; lane 3,for s; lane 4,for s 2; lane 5,w 1;for s; lane 6, 189Y. (C) SDS-PAGE of enriched PKG fraction from adult head extracts after affinity chromatography on cGMP-sepharose and labeling with 125I (13). Lanes 1,for s; lane 2,for R; lane 3, dg2-cDNA without heat shock; lane 4, dg2-cDNA after heat shock; lane 5, 189Y; lane 6, for s 2. Arrows in (B) and (C) indicate heat-inducible bands at ∼80,000 Daltons recognized by PKG antibody and purified on cGMP-sepharose. Rovers have more intense bands than sitters.

The assignment of mutations in the for gene to thedg2 locus not only establishes the identification of PKG mutations but also implicates the cGMP signal transduction pathway in the regulation of food-search behavior in D. melanogaster. Small but significant differences in the levels of this kinase affected the naturally occurring behavioral polymorphism. These small differences in PKG were even detectable in homogenates, indicating that the differences in PKG level in rovers and sitters might be larger in cells relevant to the expression of the foraging behavior. Our results suggest that the amount of kinase activity affects larval food-search behavior. Indeed, even modest quantitative changes in kinase activity affect behavior (14). Induced mutations that affect behavioral phenotypes often lie in signal transduction pathways (15). For example, the cyclic adenosine monophosphate (cAMP) system influences associative learning in flies (16, 17) and mice (18), and genetic variants in two other serine/threonine kinases, the calcium/calmodulin-dependent protein kinase II and protein kinase C, affect learning and behavioral plasticity in flies (14) and mice (19). Our finding that for encodes a PKG shows that a naturally occurring genetic polymorphism in behavior involves these pathways.

PKG has a variety of pleiotropic cellular regulatory functions (20) that are also typical of signal transduction components (15). Electrophysiological studies have shown that injected kinase affects neuronal membrane conductance in snails and mammals (21), that inhibitors of PKG block long-term potentiation in mammalian hippocampus (22), and that PKG is involved in presynaptic long-term potentiation in cultured hippocampal neurons (23). Outside the nervous system, PKG has also been implicated in controlling proliferation of smooth muscle cells (24) and neutrophil degranulation (25). Our findings assign behavioral functions to this relatively scarce member of the serine/threonine kinases and show that subtle differences in PKG can lead to naturally occurring variation in behavior.

  • * Present address: Department of Plant Science, University of British Columbia, 2357 Main Mall, Vancouver, British Columbia, Canada, V6T 1Z4.

  • Present address: Institute of Histology and Embryology, Faculty of Medicine, University of Lisbon, Avenue Professor, Egas Moniz, 1699 Lisboa Codex, Portugal.

  • Present address: The Neurosciences Institute, 10640 John Jay Hopkins Drive, San Diego, CA 92121, USA.

  • § To whom correspondence should be addressed. E-mail: mbsoko{at}


View Abstract

Navigate This Article