Epigenetic mechanisms modulate differences in Drosophila foraging behavior

Ina Anreiter; Jamie M. Kramer; Marla B. Sokolowski

doi:10.1073/pnas.1710770114

New Research In

Edited by Gene E. Robinson, University of Illinois at Urbana–Champaign, Urbana, IL, and approved September 26, 2017 (received for review June 14, 2017)

This article has a Correction. Please see:

Correction for Anreiter et al., Epigenetic mechanisms modulate differences in Drosophila foraging behavior - April 15, 2019

See related content:

Epigenetics and Drosophila foraging
- Nov 07, 2017

Significance

Individuals of the same species display remarkable variation in behavior even in identical contexts, but the molecular mechanisms that underlie this variation are still poorly understood. Here we present important findings on the regulation of behavioral variation. We show that epigenetic regulation interacts with genetic variation, and provide causal evidence that this mechanism underlies distinct foraging behavioral strategies. More globally, our findings show that individuals within a species may behave differently due to the epigenetic control of gene expression.

Abstract

Little is known about how genetic variation and epigenetic marks interact to shape differences in behavior. The foraging (for) gene regulates behavioral differences between the rover and sitter Drosophila melanogaster strains, but the molecular mechanisms through which it does so have remained elusive. We show that the epigenetic regulator G9a interacts with for to regulate strain-specific adult foraging behavior through allele-specific histone methylation of a for promoter (pr4). Rovers have higher pr4 H3K9me dimethylation, lower pr4 RNA expression, and higher foraging scores than sitters. The rover–sitter differences disappear in the presence of G9a null mutant alleles, showing that G9a is necessary for these differences. Furthermore, rover foraging scores can be phenocopied by transgenically reducing pr4 expression in sitters. This compelling evidence shows that genetic variation can interact with an epigenetic modifier to produce differences in gene expression, establishing a behavioral polymorphism in Drosophila.

Although it has been shown that variation in human and animal behavior correlates with genetic polymorphisms and epigenetic regulation (1, 2), causal links among genetic variation, epigenetic regulation, and behavior have not been established.

The foraging (for) gene in Drosophila melanogaster is a complex gene that encodes several different isoforms of a cGMP-dependent protein kinase (3). for regulates various behavioral and physiological phenotypes in the fly and other organisms, including humans (4 ⇓⇓⇓⇓–9). Importantly, this gene, with its rover and sitter allelic variants, is known to give rise to naturally occurring behavioral variations in D. melanogaster. Larvae with the rover allele move longer distances while foraging than those with sitter alleles, and the rover allele exhibits genetic dominance over the sitter in this trait (3). for also affects rover–sitter differences in foraging behavior in adult flies (10, 11). These rover–sitter behavioral differences have been shown to arise from genetic variation in the for gene, but until now the molecular mechanisms underlying rover–sitter differences have remained elusive.

for has four separate transcription start sites and one transcription termination site encoding at least 21 different transcripts, which cluster into four transcript classes of similar coding sequences according to promoter: pr1, pr2, pr3, and pr4 (4). This complexity could contain the key to understanding the regulation of rover–sitter behavioral differences as well as for’s pleiotropic functions. The function of each of for’s transcripts and how they are regulated are currently unknown.

Epigenetic modifiers play an important role in depositing marks that recruit transcription factors and regulate expression. Drosophila G9a (dG9a, EHMT) is an epigenetic modifier known to methylate the regions of the for promoters (12). G9a is one of three histone methyltransferases that catalyze H3K9me2 in flies. While the other two H3K9me2 methyltransferases, egg and Su(var)3–9, are involved mainly in the formation of heterochromatin, G9a acts predominantly on the 5′ of transcription start sites in euchromatic DNA. This pattern of methylation is usually repressive and associated with transcriptional plasticity of actively transcribed genes (12).

Here we show that G9a regulation of for is responsible for rover–sitter differences in adult foraging behavior. Our results demonstrate that allele-specific histone methylation drives differences in behavior, a mechanism that has not been addressed experimentally.

Results

for Interacts with G9a to Mediate Rover–Sitter Differences in Adult Foraging Behavior.

A schematic representation of the for gene, including transcription start sites and promoter areas methylated by G9a, is shown in Fig. 1A. We used rover and sitter flies with [G9a wild type (WT)] and without (G9a null) functional G9a alleles to test for a possible interaction of G9a and for on the rover–sitter difference in adult foraging behavior. Flies were tested in a foraging arena in which individuals are given the opportunity to search for and ingest sucrose drops (Fig. 1B). Rovers bearing G9a WT alleles, have significantly greater foraging success (i.e., number of food drops found and ingested) compared with sitters (P < 0.001; Fig. 1C), but this rover–sitter difference disappears when rovers and sitters carry G9a null alleles (Fig. 1 D and E; P = 0.751). Thus, G9a is required for this rover–sitter behavioral difference. Notably, the G9a null mutation significantly increases foraging behavior in sitters (P < 0.001), but not in rovers (P = 0.285) (Fig. 1 D and E), indicating a selective interaction of the sitter for alleles with the G9a null alleles. The higher sensitivity of sitters to the loss of G9a is also seen in starvation resistance, a trait correlated with foraging behavior (Fig. S1). Both rovers and sitters with the G9a mutation survive longer under starvation conditions than rovers and sitters with WT G9a; however, the increase in starvation resistance is much greater in sitters.

Fig. 1.

foraging (for) interacts with G9a to mediate differences in adult feeding behavior. (A) The for gene model. The four transcription start sites (pr1–4) are marked with green arrows, exons are in gray boxes, and introns are shown as black lines. G9a-associated methylation sites are shown in purple. (B) Foraging arena schematic. (C) Rovers find significantly more sucrose drops than sitters in 5 min (t₅₈ = −6.829; P < 0.001) and 10 min (t₅₈ = −8.523; P < 0.001). (D) Rovers with G9a WT find significantly more sucrose drops than sitters with G9a WT during a 5-min test [F_(3,116) = 10.58; P < 0.001], but this difference disappears in G9a null flies (P = 0.751). (E) Rovers with G9a WT find significantly more sucrose drops than sitters with G9a WT during a 10-min test [F_(3,116) = 10.88; P = 0.004], but this difference disappears in G9a null flies (P = 0.678). n = 20/strain.

Fig. S1.

G9a affects starvation resistance significantly more in sitters than in rovers. Here 5- to 6-d-old adult females were transferred to agar vials with a water source in groups of 10, and the number of dead flies was scored every 6 h until the last fly died. While G9a nulls show increased starvation resistance in both rovers and sitters (Kaplan–Meier survival statistic, S₃ = 214.225; P < 0.001), the G9a-mediated response in starvation resistance is significantly larger in sitters (S₃ = 116.167; P = 0.00) than in rovers (S₁ = 62.133; P = 1.3E-14); n = 10.

Differential Expression and Methylation of pr4 Correlates with Differences in Adult Foraging Behavior.

We next used qRT-PCR to quantify expression levels from each of the for promoter-specific transcript groups (Fig. 2 A–D). Rovers bearing the G9a WT alleles have significantly lower expression than sitters for pr2 (Fig. 2B) and pr4 (Fig. 2D); however, the G9a null mutation eliminated the rover–sitter expression difference at pr4 (Fig. 2D). Thus, G9a is required for the rover–sitter expression difference of pr4.

Fig. 2.

The for promoters are differentially expressed and show G9a-dependent methylation differences. (A) Rovers and sitters do not differ in for pr1 expression [F_(3,11) = 3.96; P = 0.053]. (B) Rovers with G9a WT have significantly less pr2 expression than sitters with G9a WT [F_(3,11) = 42.39; P < 0.001], and this difference is not G9a-dependent as it is maintained in G9a nulls (P < 0.001). (C) Rovers and sitters do not differ in for P1/3 expression [F_(3,11) = 2.24; P = 0.161]. (D) Rovers with G9a WT have significantly less pr4 expression than sitters with G9a WT [F_(3,11) = 10.62; P < 0.003], and this difference is G9a-dependent, as it disappears in G9a nulls (P = 0.2). (E) Rovers with G9a WT have significantly more for pr1 H3K9me2 than sitters with G9a WT [F_(3,15) = 6.59; P = 0.007]. (F) Rovers with G9a WT have significantly less pr2 H3K9me2 than sitters with G9a WT [F_(3,15) = 68.86; P < 0.001], and this difference is not G9a-dependent, as it is maintained in G9a nulls (P < 0.001). (G) Rovers and sitters do not differ in for pr3 H3K9me2. (H) Rovers with G9a WT have significantly more pr4 H3K9me2 than sitters with G9a WT [F_(3,15) = 8.62; P = 0.002], and this difference is G9a-dependent, as it disappears in G9a nulls (P = 0.476). n = 3 for qRT-PCR and n = 4 for ChIP-qPCR with 20 adult mated females/biological replicate. *0.05 > P > 0.01; **0.01 > P > 0.001; ***P > 0.001.

To further explore the interaction between G9a and allele-specific for expression, we performed chromatin immunoprecipitation-qPCR to assess H3K9me2 levels at the for promoters (Fig. 2 E–H). pr4 shows a G9a-mediated rover–sitter methylation difference (Fig. 2H) that agrees with the lower expression of pr4 in G9a WT rovers. Rovers have significantly higher pr4 H3K9me2 levels than sitters (P = 0.002), but, importantly, this difference disappears in the presence of the G9a null allele (P = 0.476). This provides further evidence that the rover–sitter difference depends on G9a. The fact that some H3K9me2 methylation remains in the G9a mutant demonstrates compensation by another H3K9me2 methyltransferase when G9a is lost. However, this other H3K9me2 methyltransferase does not discriminate between the rover and sitter alleles, resulting in no difference in H3K9me2 methylation of pr4 in the G9a null mutants.

Rovers and sitters also have differing H3K9me2 levels at pr1 (Fig. 2E); however, this does not alter the expression of pr1 (Fig. 2A). Although histone methylation marks are generally associated with either repression or activation of nearby genes, the relationship between these marks and expression is not necessarily causal or linear. Specifically, H3K9me2 is associated with a wide range of gene repression patterns, and can be found at both active and repressed genes (13). Considering a chain of effect in which methylation regulates expression and altered gene expression regulates behavior, the H3K9me2 methylation difference at pr1 cannot be responsible for the rover–sitter behavioral differences described here, because it does not alter pr1 gene expression.

Differences in H3K9me2 methylation cannot be explained by genetic variation within G9a, because our rover and sitter strains were constructed to share identical G9a WT alleles (Materials and Methods). Furthermore, G9a expression levels do not differ in rovers and sitters (Fig. S2A). Consequently, the rover–sitter difference in pr4 methylation does not arise from G9a expression differences in these strains. To rule out involvement of egg and SU(VAR)3–9, the only other H3K9me methylases in Drosophila, we assessed the expression of egg and SU(VAR)3–9 in rovers and sitters. Like G9a, egg expression does not differ between rovers and sitters (Fig. S2B). On the other hand, SU(VAR)3–9 is more highly expressed in sitters (Fig. S2C). However, since sitters have lower H3K9me2 levels, SU(VAR)3–9 expression it is not responsible for the for pr4 methylation pattern. This further supports the hypothesis that G9a mediates the rover–sitter difference in pr4 expression; G9a targets pr4 differently in rovers compared with sitters.

Fig. S2.

Relative expression (qRT-PCR) of the three H3K9me2 histone methyltransferases in Drosophila, G9a, egg, and SU(VAR)3–9. (A) G9a expression does not differ between rovers and sitters. (B) egg expression shows no significant differences between strains and thus no association with G9a or for. (C) SU(VAR)3–9 expression differs between strains [F_(3,11) = 17.91; P = 0.001], in a pattern suggesting that for and G9a interact to regulate SU(VAR)3–9 expression, as opposed to for being regulated by SU(VAR)3–9. n = 3 with 20 adult mated females/biological replicate. *0.05 > P > 0.01; **0.01 > P > 0.001; ***P > 0.001.

Rover Foraging Behavior Can Be Phenocopied in Sitters by Transgenically Reducing pr4 Expression.

To establish a causal relationship between for pr4 expression and foraging behavior, we designed an RNAi construct that specifically targets pr4 transcripts (Fig. S3 B–F). Before knocking down for, we show that foraging success is significantly greater in rover–sitter heterozygotes than in sitters (P = 0.017) and does not differ significantly from that in rovers (P = 0.138) (Fig. 3B). Correspondingly, rovers, sitters, and rover–sitter heterozygotes with G9a null alleles do not differ in their foraging success (Fig. 3B). The fact that the foraging behavior of rover heterozygotes is comparable to that of the rover homozygotes allowed us to perform RNAi experiments in heterozygotes. Since sitters have higher pr4 expression and lower foraging scores than rovers, we predicted that knockdown of pr4 in sitters would result in an increase in foraging scores. As predicted, pr4 knockdown in sitters increases foraging relative to the transgenic controls [Fig. 3C; F_(4,94) = 13.487; Gal4 control, P = 0.002; UAS control, P = 0.014], while further knockdown in rovers has no effect on foraging. It is possible that foraging behavior might not have increased in rovers because of a ceiling effect. This could be because rovers naturally forage at a physiological maximum, or because there is a limiting step in the activation or repression of a downstream target of for. Nevertheless, our finding that a knockdown of pr4 transcripts in sitters results in an increase in foraging success conclusively demonstrates a causal relationship between pr4 expression levels and differences in rover and sitter adult foraging behavior. The differences in adult foraging behavior are also reflected in the proportion of time spent in areas containing food (Fig. 3 A and D), but not in overall distance traveled during the test (Fig. 3E).

Fig. 3.

Rover–sitter foraging behavior is directly regulated by pr4 expression levels. (A) Representative images of rover and sitter foraging paths with position coordinates over 10 min plotted as a scatterplot. (B) There are significant differences in foraging behavior among rovers, sitters, and rover–sitter heterozygotes [F_(4,119) = 11.09; P < 0.001], driven by G9a. Rovers with WT G9a forage significantly more in 10 min than sitters with WT G9a (P < 0.001). rover–sitter heterozygotes with WT G9a forage significantly more than sitters with WT G9a (P = 0.017), and are not significantly different from rovers with WT G9a (P = 0.138). Rovers, sitters, and rover–sitter heterozygotes with G9a null show no differences in foraging behavior. (C) Reducing pr4 expression by driving pr4-RNAi with the da-GAL4 driver significantly affects foraging behavior [F_(4,94) = 13.49; P < 0.001]. pr4 RNAi expression significantly increases sitter foraging behavior compared with controls (P = 0.002 compared with UAS control and P = 0.014 compared with GAL4 control), and does not significantly alter rover behavior (P = 0.719). (D) Reducing pr4 expression by driving pr4-RNAi with the da-GAL4 driver significantly affects the proportion of time spent in the interior food-containing area of the arena [F_(4,70) = 6.88; P < 0.001]. (E) Reducing pr4 expression by driving pr4-RNAi with the da-GAL4 driver does not affect the total distance traveled during foraging [F_(4,68) = 4.16; P = 0.005]. n = 20 for all tests. *0.05 > P > 0.01; **0.01 > P > 0.001.

Fig. S3.

Relative expression (qRT-PCR) of overall for expression and of RNAi knockdown using the for-RNAi for pr4 in whole flies. (A) There are no significant genotype differences in overall for expression in whole flies [F_(3,11) = 0.57; P = 0.653]. (B–F) Our RNAi line generated to target pr4 transcripts knocks down only pr4 transcripts. (B) Expression of this RNAi has no significant effect on overall for expression in rovers and sitters compared with controls. (C) Expression of this RNAi has no significant effect on for pr1 expression in rovers and sitters compared with controls. (D) Expression of this RNAi has no significant effect on for pr2 expression in rovers and sitters compared with controls. (E) Expression of this RNAi has no significant effect on expression of the P1/3 class of transcripts. (F) Expression of this RNAi has a significant effect on overall for pr4 expression in rovers and sitters compared with controls [t₃ = 14.339, P < 0.001, for rovers compared with rover control; F_(2,11) = 6.087, P = 0.021, for sitters compared with sitter controls]. n = 4, with 20 adult virgin females/biological replicate.

The Rover–Sitter for pr4 Difference Is Tissue-Specific.

Because all of the for promoters showed expression in whole adult flies (Fig. 2), we dissected candidate tissues and assessed pr4 expression in rover and sitter flies with WT or null G9a alleles (Fig. 4). Notably, highly significant differences between rover and sitter pr4 expression in the brain (P < 0.001) disappear in the G9a null background (P = 0.190) (Fig. 4). Ovaries have a smaller but also significant rover–sitter difference in pr4 expression that also disappears in the G9a null background. These findings suggest that rover and sitter adult foraging behavior is likely driven by pr4 expression differences in the brain and the ovaries.

Fig. 4.

The rover–sitter difference in for pr4 expression arises from the brain and ovaries. (A) Rovers with G9a WT have significantly less pr4 expression in the brain than sitters with G9a WT [F_(3,11) = 32; P < 0.001], and this difference is G9a-dependent, as it disappears in G9a nulls (P = 0.19). (B) Rovers and sitters do not differ in for pr4 expression in the gut [F_(3,11) = 0.13; P = 0.937]. (C) Rovers with G9a WT have significantly less pr4 expression in the ovaries than sitters with G9a WT [F_(3,11) = 10.34; P = 0.004], and this difference is G9a-dependent, as it disappears in G9a nulls (P = 0.19). (D) Rovers and sitters with G9a WT do not differ in for pr4 expression in the carcass [F_(3,11) = 8.03; P = 0.242], but rovers with G9a null have significantly less pr4 expression in the carcass than sitters with G9a WT (P = 0.014). (E) Rovers and sitters differ in one SNP in a 0.1-kb region upstream of the pr4 transcription start site. For qRT-PCR, n = 3, with 20 adult mated female tissues/biological replicate for all tissues. *0.05 > P > 0.01; **0.01 > P > 0.001; ***P > 0.001.

Rover and Sitter for Promoter DNA Sequences Are Polymorphic.

Since the difference in methylation at pr4 does not originate with G9a itself, the most likely explanation for the differences in expression and methylation of pr4 in rovers and sitters are DNA single nucleotide polymorphisms (SNPs) that can affect the recruitment of G9a to the promoter region. To address this, we sequenced the rover and sitter for alleles and found several SNPs, mostly in the noncoding region; one of these SNPs was in pr4 (Fig. 4E and Table S1). We then searched the sequence of pr4 for predicted transcription factor binding sites that coincide with the single SNP found in this region. The pr4 region had predicted transcription factor binding sites for six different factors/classes (Mad, GAGA factor, T11, Prd, Dfd, and FTZ), with the highest confidence for three predicted mad sites, one of which falls on the single SNP found in pr4 (Fig. 4B). This SNP coincides with a site within the mad binding sequence that does not allow substitutions, most likely resulting in no binding of this factor at this site in the rover strain, but not in the sitter strain.

View this table:

Table S1.

SNPs between D. melanogaster rover and sitter DNA sequences

Discussion

We show that rovers and sitters have a natural difference in adult foraging behavior that is caused by differences in G9a-dependent expression of the for pr4 transcripts. pr4 is differentially methylated by G9a in rovers and sitters, and we demonstrate that pr4 is solely responsible for the rover–sitter behavioral polymorphism in adult foraging behavior. Nevertheless, G9a is not the sole transcriptional regulator, or the sole H3K9 methyltransferase, regulating pr4 expression. Our results show that the loss of G9a can result in more or less H3K9me2 at pr4, depending on the for allele present. This dual function of G9a has been previously shown in mice, where G9a is able to both repress and activate gene expression through interactions with other proteins in its regulatory complex (14). While pr4 is responsible for regulating the rover–sitter difference in adult foraging behavior, other for promoters likely regulate other for-related phenotypes. In fact, our expression data show that other for promoters are differentially expressed in rovers and sitters. For example, pr2 is highly expressed in sitters and not expressed at all in rovers (Fig. 2B). The pr2 expression difference also correlates with H3K9me2, but cannot be explained solely by G9a. pr2 and pr4 transcribe different isoforms of for (P1 and P4, respectively) that might differ in function. Our results suggest that the expression of for’s four promoters might be regulated by distinct regulatory complexes, and that each promoter might influence distinct behavioral phenotypes.

We also found that the difference in pr4 expression is tissue-specific, being driven by the brain and ovaries. The central nervous system and ovaries might be linked in regulating feeding behavior, since reproduction constitutes the major energy expenditure of female flies, and sex peptide signaling in the reproductive organs affects the feeding behavior of female flies (15). Our work highlights the complex epigenetic architecture that underlies behavioral regulation.

The lack of a DNA-binding domain suggests that G9a is targeted to specific DNA regions through interactions with DNA-binding proteins, such as transcription factors. SNPs in the promoter region could lead to differential binding of DNA-binding proteins that recruit G9a. For instance, the SNP in pr4 lies within a conserved site of a putative mad binding motif, and potentially could affect mad binding. If mad is one of the elements in the G9a complex, then less binding of mad in the rover strain (which would be predicted from the SNP) potentially could explain the lower pr4 H3K9me2 levels in rovers. Like G9a, mad has been shown to act as both a repressor and an activator of gene transcription, depending on context (16). Although mad is best known for its role in development (17), some studies suggest that it might have regulatory functions in the mature nervous system (18).

In conclusion, the mechanisms by which epigenetic regulation influences behavioral differences are poorly understood. Epigenetic regulation has been shown to be a mechanism through which animals adjust their behavior and physiology to the environment in which they live. Not all individuals respond similarly to the same environmental cue, however. In this case, epigenetic-by-genetic interaction would be an important but neglected component of gene-by-environment interactions. The deposition of epigenetic marks can depend on underlying genetic differences (19), and genetic variation likely plays an important role in moderating epigenetic differences between individuals. Importantly, epigenetic-by-genetic interactions present an avenue through which genetic variation outside of gene coding regions can modulate phenotypic variability. Two other noteworthy studies in humans and prairie voles have reported associations among genetic variation, DNA methylation, and behavior (1, 2). Here we used the fruit fly to establish molecular causality, and provide definitive evidence for how the complex interactions among genetics, epigenetics, and isoform-specific gene regulation causes variation in naturally occurring behavioral polymorphisms.

Materials and Methods

Fly Strains and Rearing.

All flies were reared on a standard cornmeal-molasses medium at 25 °C on a 12-h light/dark cycle with lights on at 0800 h. The rover (for) and sitter (for^s) strains (10) have reisogenized for^R or for^s second chromosomes and share reisogenized X and third chromosomes. The G9a null and its corresponding G9a WT allele were originally designated as EHMT^DD1 and EHMT⁺ (12). The daughterless-GAL4 (da-GAL4) driver was a gift from Tony Harris, Department of Cell & Systems Biology, University of Toronto, Toronto. The foraging pr4 RNAi line was generated in the M.B.S. laboratory, and the UAS-Dcr line was acquired from the Bloomington Drosophila Stock Center (24651). A more detailed description of the strains is provided in the SI Materials and Methods.

Genomic Sequencing of the for^R and for^s Lines.

Full genomic sequencing was done on the for^R and for^s lines. DNA was extracted from 50 males and 50 females of each strain using a Qiagen DNeasy Blood and Tissue Kit (catalog no. 69504), following the manufacturer’s instructions. TruSeq gDNA library preparation and paired-end 100-bp sequencing on the Illumina HiSeq platform was done at the McGill University and Génome Québec Innovation Centre. For reference-guided assembly, the reads were mapped to the D. melanogaster reference genome (release 5.57) using the default parameters in bwa v. 0.6.0-r85 (20), and consensus sequences for each line were generated with Samtools v. 0.1.18 (21). Consensus sequences for each chromosome were deposited in GenBank (sitter accession nos. CP023329–CP023334; rover accession nos. CP023335–CP023340). Consensus sequences were aligned and annotated in Geneious v. R10.0.5 (22). Putative transcription factor-binding sites were assessed by submitting 100 bp of the DNA sequence immediately upstream of the pr4 transcription start site to PROMO v. 3.0.2 (23), specifying both species and factor as D. melanogaster.

Generation of the pr4 RNAi Line.

For the transgenic knockdown of foraging pr4 the foraging pr4 RNAi line was generated using the pWIZ RNAi cloning vector (24). A region complementary to the 3′ end of exon 7 was used to amplify a pr4 isoform-specific region of 723 bp. The primers (Table S2) included an NheI restriction site (underscored in the table), which was used to clone the 5′-3′ fragment into the NheI site of pWIZ, and the 3′-5′ fragment into the AvrII (which has complementary sticky ends with NheI) site of pWIZ. P-element injections into w1118, performed by BestGene, resulted in insertion of the transgene on the second chromosome.

View this table:

Table S2.

D. melanogaster primer sequences

Adult Foraging Assay.

The adult foraging assay (AFA) has been described in detail by Hughson et al. (11). In brief, females were collected at eclosion and housed in groups of 20 females and 10 males. Mated 5- to 6-d-old females were food deprived with a water source for 24 ± 0.5 h before being tested in the AFA. Foraging tests were performed in the afternoon to avoid circadian effects on feeding. A more detailed description of the AFA setup is provided in SI Materials and Methods.

qRT-PCR.

RNA of whole flies or tissues was extracted using the RNeasy Mini Kit (catalog no. 74104; Qiagen) with RNase-Free DNase (catalog no. 79254; Qiagen). RNA integrity was assessed, and cDNA was synthesized from 1 μg of tRNA with the iScript Advanced cDNA Synthesis Kit for qRT-PCR (catalog no. 1725037; Bio-Rad). qRT-PCR was performed on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad), using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and gene-specific primers (Table S2). Target gene expression was standardized to three reference genes (α-tub, act5c, and 1433ε), and fold change values (2^−ΔΔCt) were determined to quantify relative expression differences between genotypes. More details are provided in SI Materials and Methods.

Chromatin Immunoprecipitation-qPCR.

Crude fly extract was obtained by homogenizing 20 flies in PBS, followed by cross-linking with 1% formaldehyde for 30 min. Nuclei were isolated, and cross-linked chromatin was fragmented by sonicating on ice for 60 cycles (high power, 30 s on/off). Chromatin immunoprecipitation was performed with anti-H3K9me2 antibodies (07-441; Upstate Biotechnology), and Protein A/G beads (Santa Cruz Biotechnology) were used to capture antibody-bound chromatin. Chromatin immunoprecipitated DNA was isolated by phenol/chloroform extraction and ethanol precipitation. qPCR on chipped and input (not chipped chromatin) was performed with primers targeting foraging promoter areas (Table S2), and methylation levels were accessed as %input. moca-cyp, used as a negative methylation control, showed low methylation (<3%) in all strains (Fig. S4), and 2cta and 2chi, used as positive methylation controls, showed high methylation (40–50%) in all strains, with no significant differences among strains (Fig. S4).

Fig. S4.

ChIP-qPCR for high and low H3K9me2-methylated control genes. (A) moca-cyp, used as a negative control, showed low methylation (<3%) in all strains. (B and C) 2cta (B) and 2chi (C), used as positive controls, show high methylation (40–50%) in all strains, with no significant differences among strains. n = 4 with 20 adult mated females/biological replicate.

Statistical Analysis.

All statistical analyses were performed in SigmaPlot 11.0. Data were tested for normality and equal variance, and one- or two-way ANOVA was performed to test for the effects of strain and treatment and their interactions. Post hoc pairwise multiple comparison procedures were done using the Holm–Sidak method. Kaplan–Meier survival analysis (log-rank) with post hoc multiple comparisons by the Holm–Sidak method were performed on the starvation resistance data.

SI Materials and Methods

Fly Strains.

We isogenized the for rover (for^R) and sitter (for^s) strains from our laboratory, starting with the sitter line using single females crossed to balancer males. To make the rover line, we isogenized the rover’s second chromosome, containing the foraging locus, and combined it with the X and third chromosomes from the isogenized sitter line. Consequently, the rover and sitter lines have for^R or for^s second chromosomes and share isogenized X and third chromosomes from the sitter strain. We fully sequenced these lines (see below).

To create our G9a test strains, we used the G9a null and G9a WT control alleles, originally identified as EHMT^DD1 and EHMT⁺, respectively (12). These alleles were generated by inserting a P element in G9a and then creating a precise (control allele) and imprecise (mutant allele) excision of that P element. Using this method, the mutant and control were generated in the same genetic background. We substituted the X chromosomes in our rover and sitter lines for the G9a null and G9a WT X chromosomes. The final rover and sitter lines used in our experiments have the same G9a null and G9a WT X chromosomes, for^R or for^s second chromosomes, and the same third chromosomes.

The da-GAL4 driver, a kind gift from Tony Harris, Department of Cell & Systems Biology, University of Toronto, Toronto, was backcrossed for six generations into the isogenized sitter third chromosome and crossed into the G9a WT rover and sitter lines to produce strains G9a⁺;for^s;da-GAL4 and G9a⁺;for^R;da-GAL4.

The foraging pr4 RNAi line was generated in the M.B.S. laboratory (Materials and Methods), backcrossed for 10 generations into the sitter second chromosome, and combined with the G9a⁺ X chromosome and nine-generation backcrossed UAS-Dcr (Bloomington Drosophila Stock Center; 24651, third chromosome) to generate the G9a⁺;for-RNAi^pr4,UAS -dcr^s line.

AFA.

The AFA was adapted from Hughson et al. (11). Flies were tested in a round arena, 15 cm in diameter and 1.5 cm high, with an acclimatization chamber underneath and an elevator shaft through which the test animal was allowed to enter the arena at the center. The arena was divided into four quadrants, each with a 2-cm circle in the center. A 0.2-µL drop of 10% sucrose dyed blue with 0.1% erioglaucine was placed in the center of each circle. The rim around the edge of the arena was filled with water to prevent the flies from climbing up the walls, and a lid was placed on top of the arena to allow for uniform illumination.

For testing, a single fly was introduced into the acclimatization chamber underneath the arena using an aspirator. After a 2-min acclimatization period, the barrier to the elevator shaft was removed, and the fly was allowed to climb into the arena. The test period started when the fly entered the arena, and the number of sucrose drops that the fly consumed was scored at 5 and 10 min. The fly’s locomotion was recorded using a Microsoft LifeCam Studio webcam and a custom tracker developed in the public domain JavaGrinders (11). Python 2.7 analysis scripts were used to extract the movement parameters (11). Tests were conducted between 1,300 and 1,700 h to control for circadian influences on foraging behavior.

qRT-PCR.

For whole flies, total RNA was extracted from 5- to 6-d-old mated females (with exception of the pr4 RNAi verification, for which virgin females were used) using the RNeasy Mini Kit (catalog no. 74104; Qiagen) and the corresponding RNase-Free DNase Set (catalog no. 79254; Qiagen), following the manufacturer’s instructions for purification of total RNA from animal tissues. RNA was extracted from three biological replicates with 20 females per replicate. For tissue-specific qRT-PCR, brains, ovaries, guts and carcasses were dissected in ice-cold Schneider’s medium and immediately frozen on dry ice. Twenty tissue specimens were dissected per biological replicate, and RNA was extracted as above. Following extraction, RNA was quantified using a Nanodrop 2000c spectrophotometer (Thermo Fisher Scientific), and RNA integrity was accessed by gel electrophoresis.

cDNA was synthesized with the iScript Advanced cDNA Synthesis Kit for qRT-PCR (catalog no. 1725037; Bio-Rad), using 1 μg of tRNA per sample, following the manufacturer’s instructions. qRT-PCR was performed on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad), using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and gene-specific primers (Table S2).

Since pr3 is embedded in an exon that is shared by the P1 (made by pr1 and pr2) and P3 (made by pr3) protein isoforms, and thus cannot be amplified separately, the primer pair designed for pr3 was designated P1/3. Primer efficiency was calculated, and only primers with an efficiency of 100–110% were used. Cycling conditions followed the manufacturer’s protocol. Target gene expression was standardized to three reference genes (α-tub, act5c, and 1433ε) with robust stability values (mean coefficient of variance <0.05; mean M value <0.1). Fold change values (2^−ΔΔCt) were determined to quantify relative expression differences between genotypes.

Acknowledgments

We thank Tony Harris (University of Toronto) for the da-GAL4 driver; Ian Dworkin and Thomas Braukmann for help with the genome assemblies; the PNAS editor and reviewers for comments that greatly improved the manuscript; and Joel Levine, Jeff Dason, and Maria Aristizabal for comments on an early draft. This research was supported by a Natural Sciences and Engineering Council of Canada and Canadian Institute for Advanced Research grant (to M.B.S.), and an Ontario Graduate Scholarship (to I.A.). The stocks used in this study were obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537).

Footnotes

↵¹To whom correspondence should be addressed. Email: marla.sokolowski{at}utoronto.ca.

Author contributions: I.A., J.M.K., and M.B.S. designed research; I.A. performed research; J.M.K. and M.B.S. contributed new reagents/analytic tools; I.A. analyzed data; and I.A. and M.B.S. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The sequences reported in this paper have been deposited in the GenBank database [accession nos. CP023329–CP023334 (sitter) and CP023335–CP023340 (rover)].
See Commentary on page 12365.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1710770114/-/DCSupplemental.

Published under the PNAS license.

View Abstract

References

↵
1. Klengel T, et al.
(2013) Allele-specific FKBP5 DNA demethylation mediates gene-childhood trauma interactions. Nat Neurosci 16:33–41.
OpenUrl CrossRef PubMed
↵
1. Okhovat M,
2. Berrio A,
3. Wallace G,
4. Ophir AG,
5. Phelps SM
(2015) Sexual fidelity trade-offs promote regulatory variation in the prairie vole brain. Science 350:1371–1374.
OpenUrl Abstract/FREE Full Text
↵
1. Osborne KA, et al.
(1997) Natural behavior polymorphism due to a cGMP-dependent protein kinase of Drosophila. Science 277:834–836.
OpenUrl Abstract/FREE Full Text
↵
1. Allen AM,
2. Anreiter I,
3. Neville MC,
4. Sokolowski MB
(2017) Feeding-related traits are affected by dosage of the foraging gene in Drosophila melanogaster. Genetics 205:761–773.
OpenUrl Abstract/FREE Full Text
↵
1. Ben-Shahar Y,
2. Leung HT,
3. Pak WL,
4. Sokolowski MB,
5. Robinson GE
(2003) cGMP-dependent changes in phototaxis: A possible role for the foraging gene in honey bee division of labor. J Exp Biol 206:2507–2515.
OpenUrl Abstract/FREE Full Text
↵
1. Ingram KK, et al.
(2016) Context-dependent expression of the foraging gene in field colonies of ants: The interacting roles of age, environment and task. Proc Biol Sci 283:20160841.
OpenUrl CrossRef PubMed
↵
1. Cho CE,
2. Brueggemann C,
3. L’Etoile ND,
4. Bargmann CI
(2016) Parallel encoding of sensory history and behavioral preference during Caenorhabditis elegans olfactory learning. Elife 5:e14000.
OpenUrl CrossRef
↵
1. Feil R, et al.
(2003) Impairment of LTD and cerebellar learning by Purkinje cell-specific ablation of cGMP-dependent protein kinase I. J Cell Biol 163:295–302.
OpenUrl Abstract/FREE Full Text
↵
1. Sokolowski HM, et al.
(2017) The Drosophila foraging gene human orthologue PRKG1 predicts individual differences in the effects of early adversity on maternal sensitivity. Cogn Dev 42:62–73.
OpenUrl CrossRef
↵
1. Pereira HS,
2. Sokolowski MB
(1993) Mutations in the larval foraging gene affect adult locomotory behavior after feeding in Drosophila melanogaster. Proc Natl Acad Sci USA 90:5044–5046.
OpenUrl Abstract/FREE Full Text
↵
1. Hughson BN, et al.
(August 30, 2017) The adult foraging assay (AFA) detects strain and food-deprivation effects in feeding-related traits of Drosophila melanogaster. J Insect Physiol doi:10.1016/j.jinsphys.2017.08.011.
OpenUrl CrossRef
↵
1. Kramer JM, et al.
(2011) Epigenetic regulation of learning and memory by Drosophila EHMT/G9a. PLoS Biol 9:e1000569.
OpenUrl CrossRef PubMed
↵
1. Filion GJ, et al.
(2010) Systematic protein location mapping reveals five principal chromatin types in Drosophila cells. Cell 143:212–224.
OpenUrl CrossRef PubMed
↵
1. Bittencourt D, et al.
(2012) G9a functions as a molecular scaffold for assembly of transcriptional coactivators on a subset of glucocorticoid receptor target genes. Proc Natl Acad Sci USA 109:19673–19678.
OpenUrl Abstract/FREE Full Text
↵
1. Carvalho GB,
2. Kapahi P,
3. Anderson DJ,
4. Benzer S
(2006) Allocrine modulation of feeding behavior by the sex peptide of Drosophila. Curr Biol 16:692–696.
OpenUrl CrossRef PubMed
↵
1. Grandori C,
2. Cowley SM,
3. James LP,
4. Eisenman RN
(2000) The Myc/Max/Mad network and the transcriptional control of cell behavior. Annu Rev Cell Dev Biol 16:653–699.
OpenUrl CrossRef PubMed
↵
1. Charng W-L,
2. Yamamoto S,
3. Bellen HJ
(2014) Shared mechanisms between Drosophila peripheral nervous system development and human neurodegenerative diseases. Curr Opin Neurobiol 27:158–164.
OpenUrl CrossRef PubMed
↵
1. Sulkowski MJ, et al.
(2016) A novel, noncanonical BMP pathway modulates synapse maturation at the Drosophila neuromuscular junction. PLoS Genet 12:e1005810.
OpenUrl CrossRef PubMed
↵
1. Bell JT, et al.
(2011) DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol 12:R10.
OpenUrl CrossRef PubMed
↵
1. Li H,
2. Durbin R
(2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25:1754–1760.
OpenUrl CrossRef PubMed
↵
1. Li H, et al., 1000 Genome Project Data Processing Subgroup
(2009) The sequence alignment/map format and SAMtools. Bioinformatics 25:2078–2079.
OpenUrl CrossRef PubMed
↵
1. Kearse M, et al.
(2012) Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28:1647–1649.
OpenUrl CrossRef PubMed
↵
1. Farré D, et al.
(2003) Identification of patterns in biological sequences at the ALGGEN server: PROMO and MALGEN. Nucleic Acids Res 31:3651–3653.
OpenUrl CrossRef PubMed
↵
1. Lee YS,
2. Carthew RW
(2003) Making a better RNAi vector for Drosophila: Use of intron spacers. Methods 30:322–329.
OpenUrl CrossRef PubMed

Article Alerts

Email Article

Citation Tools

Request Permissions

Proceedings of the National Academy of Sciences: 114 (47)

Table of Contents

Submit

Article Classifications

Biological Sciences
Genetics

[1] ↵
Klengel T, et al.
(2013) Allele-specific FKBP5 DNA demethylation mediates gene-childhood trauma interactions. Nat Neurosci 16:33–41.
OpenUrl CrossRef PubMed

[2] Klengel T, et al.

[3] ↵
Okhovat M,
Berrio A,
Wallace G,
Ophir AG,
Phelps SM
(2015) Sexual fidelity trade-offs promote regulatory variation in the prairie vole brain. Science 350:1371–1374.
OpenUrl Abstract/FREE Full Text

[4] Okhovat M,

[5] Berrio A,

[6] Wallace G,

[7] Ophir AG,

[8] Phelps SM

[9] ↵
Osborne KA, et al.
(1997) Natural behavior polymorphism due to a cGMP-dependent protein kinase of Drosophila. Science 277:834–836.
OpenUrl Abstract/FREE Full Text

[10] Osborne KA, et al.

[11] ↵
Allen AM,
Anreiter I,
Neville MC,
Sokolowski MB
(2017) Feeding-related traits are affected by dosage of the foraging gene in Drosophila melanogaster. Genetics 205:761–773.
OpenUrl Abstract/FREE Full Text

[12] Allen AM,

[13] Anreiter I,

[14] Neville MC,

[15] Sokolowski MB

[16] ↵
Ben-Shahar Y,
Leung HT,
Pak WL,
Sokolowski MB,
Robinson GE
(2003) cGMP-dependent changes in phototaxis: A possible role for the foraging gene in honey bee division of labor. J Exp Biol 206:2507–2515.
OpenUrl Abstract/FREE Full Text

[17] Ben-Shahar Y,

[18] Leung HT,

[19] Pak WL,

[20] Sokolowski MB,

[21] Robinson GE

[22] ↵
Ingram KK, et al.
(2016) Context-dependent expression of the foraging gene in field colonies of ants: The interacting roles of age, environment and task. Proc Biol Sci 283:20160841.
OpenUrl CrossRef PubMed

[23] Ingram KK, et al.

[24] ↵
Cho CE,
Brueggemann C,
L’Etoile ND,
Bargmann CI
(2016) Parallel encoding of sensory history and behavioral preference during Caenorhabditis elegans olfactory learning. Elife 5:e14000.
OpenUrl CrossRef

[25] Cho CE,

[26] Brueggemann C,

[27] L’Etoile ND,

[28] Bargmann CI

[29] ↵
Feil R, et al.
(2003) Impairment of LTD and cerebellar learning by Purkinje cell-specific ablation of cGMP-dependent protein kinase I. J Cell Biol 163:295–302.
OpenUrl Abstract/FREE Full Text

[30] Feil R, et al.

[31] ↵
Sokolowski HM, et al.
(2017) The Drosophila foraging gene human orthologue PRKG1 predicts individual differences in the effects of early adversity on maternal sensitivity. Cogn Dev 42:62–73.
OpenUrl CrossRef

[32] Sokolowski HM, et al.

[33] ↵
Pereira HS,
Sokolowski MB
(1993) Mutations in the larval foraging gene affect adult locomotory behavior after feeding in Drosophila melanogaster. Proc Natl Acad Sci USA 90:5044–5046.
OpenUrl Abstract/FREE Full Text

[34] Pereira HS,

[35] Sokolowski MB

[36] ↵
Hughson BN, et al.
(August 30, 2017) The adult foraging assay (AFA) detects strain and food-deprivation effects in feeding-related traits of Drosophila melanogaster. J Insect Physiol doi:10.1016/j.jinsphys.2017.08.011.
OpenUrl CrossRef

[37] Hughson BN, et al.

[38] ↵
Kramer JM, et al.
(2011) Epigenetic regulation of learning and memory by Drosophila EHMT/G9a. PLoS Biol 9:e1000569.
OpenUrl CrossRef PubMed

[39] Kramer JM, et al.

[40] ↵
Filion GJ, et al.
(2010) Systematic protein location mapping reveals five principal chromatin types in Drosophila cells. Cell 143:212–224.
OpenUrl CrossRef PubMed

[41] Filion GJ, et al.

[42] ↵
Bittencourt D, et al.
(2012) G9a functions as a molecular scaffold for assembly of transcriptional coactivators on a subset of glucocorticoid receptor target genes. Proc Natl Acad Sci USA 109:19673–19678.
OpenUrl Abstract/FREE Full Text

[43] Bittencourt D, et al.

[44] ↵
Carvalho GB,
Kapahi P,
Anderson DJ,
Benzer S
(2006) Allocrine modulation of feeding behavior by the sex peptide of Drosophila. Curr Biol 16:692–696.
OpenUrl CrossRef PubMed

[45] Carvalho GB,

[46] Kapahi P,

[47] Anderson DJ,

[48] Benzer S

[49] ↵
Grandori C,
Cowley SM,
James LP,
Eisenman RN
(2000) The Myc/Max/Mad network and the transcriptional control of cell behavior. Annu Rev Cell Dev Biol 16:653–699.
OpenUrl CrossRef PubMed

[50] Grandori C,

[51] Cowley SM,

[52] James LP,

[53] Eisenman RN

[54] ↵
Charng W-L,
Yamamoto S,
Bellen HJ
(2014) Shared mechanisms between Drosophila peripheral nervous system development and human neurodegenerative diseases. Curr Opin Neurobiol 27:158–164.
OpenUrl CrossRef PubMed

[55] Charng W-L,

[56] Yamamoto S,

[57] Bellen HJ

[58] ↵
Sulkowski MJ, et al.
(2016) A novel, noncanonical BMP pathway modulates synapse maturation at the Drosophila neuromuscular junction. PLoS Genet 12:e1005810.
OpenUrl CrossRef PubMed

[59] Sulkowski MJ, et al.

[60] ↵
Bell JT, et al.
(2011) DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol 12:R10.
OpenUrl CrossRef PubMed

[61] Bell JT, et al.

[62] ↵
Li H,
Durbin R
(2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25:1754–1760.
OpenUrl CrossRef PubMed

[63] Li H,

[64] Durbin R

[65] ↵
Li H, et al., 1000 Genome Project Data Processing Subgroup
(2009) The sequence alignment/map format and SAMtools. Bioinformatics 25:2078–2079.
OpenUrl CrossRef PubMed

[66] Li H, et al., 1000 Genome Project Data Processing Subgroup

[67] ↵
Kearse M, et al.
(2012) Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28:1647–1649.
OpenUrl CrossRef PubMed

[68] Kearse M, et al.

[69] ↵
Farré D, et al.
(2003) Identification of patterns in biological sequences at the ALGGEN server: PROMO and MALGEN. Nucleic Acids Res 31:3651–3653.
OpenUrl CrossRef PubMed

[70] Farré D, et al.

[71] ↵
Lee YS,
Carthew RW
(2003) Making a better RNAi vector for Drosophila: Use of intron spacers. Methods 30:322–329.
OpenUrl CrossRef PubMed

[72] Lee YS,

[73] Carthew RW

Main menu

User menu

Search

New Research In

Epigenetic mechanisms modulate differences in Drosophila foraging behavior