Ets96B is a member of the PEA3 subfamily

To begin our studies we confirmed that Drosophila Ets96B was homologous to human ETV5. Using the Maximum-likelihood method, the phylogenetic relationship between the Drosophila protein Ets96B and vertebrate PEA3-family proteins was determined (Fig 1A). All major vertebrate and invertebrate model organisms, where a PEA3-like protein existed that had an E-value of at least 1.0e-45 when compared to human ETV5, were included. The mammalian ETV2 subgroup was used as an outgroup. From the tree produced it was evident that invertebrates contained only one PEA3-family member, while the vertebrate family expanded to include three members: ETV1, ETV4, and ETV5. We also searched for PEA3-family homologues in the important model organisms S. cerevisae and D. discoideum by mining the NCBI’s protein database nr using protein PSI-BLAST with Drosophila Ets96B, as well as vertebrate ETV1, ETV4, and ETV5 protein sequences as queries, but no homologous proteins were discovered. From this data we conclude that Drosophila melanogaster Ets96B is basally related to the vertebrate PEA3-subfamily of ETS transcription factors.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

Comparison of Drosophila Ets96B with mammalian PEA3-family members (A) The evolutionary relationship was inferred using the Maximum-likelihood method. The optimal tree with the sum of branch length = 3.794 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (5000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. (B) The protein sequences were aligned and edited using CLC Sequence Viewer 6. Colors correspond to amino acid conservation (black = conserved, blue = no conservation). The predicted ETS DNA binding domain is indicated by a dark blue underlining of amino acids near C-terminus. Human sumoylation sites are indicated by black underlining; predicted Drosophila sumoylation site is indicated by light blue underlining.

https://doi.org/10.1371/journal.pgen.1006104.g001

A multiple sequence alignment was performed comparing Drosophila Ets96B with human PEA3-family proteins (Fig 1B). This alignment demonstrated that the ETS domain at the C-terminus, found in all ETS-family proteins, was well conserved (Fig 1B, dark blue underlining). Although some conservation was observed in the N-terminus, the PEA3 domain, present in all mammalian PEA3 family members, was not conserved. In regards to regulatory conservation, mammalian ETV5 is sumoylated on multiple sites, necessary for the inhibition of ETV5 activity [22]. These sumoylation sites were conserved in all human PEA3-family members (Fig 1B, black underlining). Using the Abgent sumoplot (http://www.abgent.com/sumoplot), as well as GPS-Sumo [23], we determined that most of the mammalian sumoylation sites were not conserved in Ets96B. Drosophila Ets96B had a predicted site at lysine 429 (K429) with the sequence LKQD, and a site (K590) possibly conserved with human PEA3-family members, having the sequence GKGD (Fig 1B, light blue underlining). Whether or not these Ets96B sites are actually sumoylated is not known.

Differential gene expression in Ets96B knockdowns

Ets96B is highly expressed in the nervous system [19], therefore we knocked it down in all neuronal cells throughout development, using the pan-neuronal elav-GAL4 driver [24] to express UAS-Ets96B^RNAi1 (Ets96B^RNAi1) [25], and then performed SOLiD sequence analysis of the entire transcriptome. By mapping the Drosophila transcriptome to the reference genome obtained from flybase (build dmel_r5.47_FB2012_05), 15147 transcripts were identified, including expressed genes (mRNA), miRNA, snRNA, snoRNA and tRNA (S1 Dataset). The Drosophila gene Ets96B encodes two transcript variants (FBtr0084696 and FBtr0084697); however, the adult samples we sequenced expressed only FBtr0084696 (Ets96B-RC), which encodes the Ets96B-PC protein isoform. Ets96B-PC is identical to Est96B-PB except for the extreme N-terminus, where Ets96B-PB had a nine amino acid insert. In this experiment, 166 genes were differentially expressed, taking into consideration the 'false discovery rate' and correcting for it using Benjamini-Hochberg correction (S1 Dataset). We observed a complete lack of expression of Ets96B in the experimental sample (S1 Dataset). To confirm the SOLiD result we performed quantitative RT-PCR (qPCR) to measure the level of Ets96B transcript. Compared with controls, Ets96B transcript levels were significantly lower in elav-GAL4;UAS-Ets96B^RNAi1 males (P < 0.005) (Fig 2A). At this time we also checked Ets96B expression levels using a second UAS-Ets96B RNAi line (Ets96B^RNAi2) [25], transcript levels were also significantly lower in the heads of elav-GAL4;UAS-ETS96B^RNAi2 males (P < 0.005) (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Ets96B regulates genes involved in oxidative phosphorylation and redox reactions (A) Relative expression level of Ets96B in 5–7 day old control males or males where Ets96B was knocked down in the entire nervous system throughout development. This assay was repeated at least 7 times. (n = 25 males per treatment; ** P<0.005 compared with controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons). Error bars = SEM. (B) Pie chart showing KEGG classification of gene groups (percentage of total Ets96B up or down regulated genes /percentage of all genes in the genome belonging to a particular category). (C) Drosophila genome depicting the location of genes up (green) or down (red) regulated in Ets96B SOLiD sequencing of the entire Drosophila transcriptome.

https://doi.org/10.1371/journal.pgen.1006104.g002

After performing pair-wise comparisons between all sequenced strains, 61 differentially expressed genes remained significant (Table 1). Interestingly, most of these genes were up-regulated when Ets96B was knocked down in the CNS, with only 21% being down-regulated. Using DAVID and KEGG, we categorized the differentially expressed genes by function [26, 27]. The largest group was involved in oxidative phosphorylation and redox reactions, with 32.8 percent of the genes falling into this category (Fig 2B). The next largest groups were a tie between genes involved in lipid and sugar homeostasis, as well as those involved in translation, each contained 14.7% of the genes (Fig 2B). Next, we mapped the differentially expressed genes to identify if they were enriched in any particular chromosomal region (Fig 2C). From this map it was evident that genes affected by loss of Ets96B were distributed throughout the genome.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Solid sequencing data for up and down regulated genes in Ets96B^RNAi versus all controls

https://doi.org/10.1371/journal.pgen.1006104.t001

Ets96B inhibits the expression of a conserved molecular chaperone complex

Using a ±log2 change in expression of 2.5 as a cut-off, we investigated the genes most highly affected when Ets96B was knocked down in the entire nervous system throughout development (Fig 3A). We utilized STRING and BioGrid [28, 29] to find interacting proteins for these top hits, however, it was immediately evident that three of them interacted: Calcium-binding protein 1 (CaBP1), Calreticulin (Crc), and ERp60.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Ets96B regulates a conserved molecular chaperone complex.

(A) SOLiD data for elav-GAL4>Ets96B^RNAi versus controls (w, elav-GAL4>w¹¹¹⁸ and w¹¹¹⁸> Ets96B^RNAi) 5–7 day old male heads. Only genes regulated more than ±log2 2.5-fold change are included. (B) Relative expression levels of CaBP1, ERp60 and Crc in 5–7 day old control males or males where Ets96B was knocked down in the entire nervous system throughout development. This assay was repeated at least 7 times. Error bars indicate SEM. (n = 25 males per treatment; *** P<0.005 compared with elav-GAL4 controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons). (C) Biogrid and String were used to find all genes that interact with the molecular chaperone genes recovered in the Ets96B knockdown males. Genes in red were recovered in the Ets96B SOLiD sequencing. (D) Data showing human homologues of Drosophila genes recovered in Biogrid and String analysis. Genes in red were recovered in the Ets96B SOLiD sequencing. (E) Diagram depicting molecules involved in endoplasmic reticulum protein folding. These include molecular chaperones that function to prevent aggregation (Bip, HSP90B1, DNAJC3, HYOU1 and CALR), protein disulfide isomerases that catalyze disulfide bond formation, isomerization and reduction (PDIA3, PDIA3 and P4HB), and proteins involved in de- or re-glycosylation of improperly folded glycoproteins (PRKCSH and UGGT1). Drosophila homologues that were up-regulated when Ets96B was knocked down are in red, and those that were up-regulated more than 2.5-fold are underlined.

https://doi.org/10.1371/journal.pgen.1006104.g003

To confirm the SOLiD sequencing results we picked CaBP1, Crc and ERp60 and performed qPCR analysis (Fig 3B). The level of expression in elav-GAL4 heterozygous controls was set as 100%, represented as 1 on the graph (Fig 3B). Compared to either elav-GAL4^+/- or Ets96B^RNAi1+/- controls, Ets96B^RNAi1 knockdown flies had a significant increase in CaBP1 (P < 0.005), ERp60 (P < 0.005) and Crc (P < 0.005) expression (Fig 3B). This was similar to results obtained by SOLiD sequencing (Fig 3A).

It was reported that overexpression of an RNAi can lead to saturation of the RNA-induced silencing complex (RISC), leading to neuronal cell death or stress [30, 31]. Therefore, we performed a control experiment overexpressing GFP under the control of elav-GAL4 and then used two different GFP RNAi lines to inhibit GFP expression (Figs 1A and S3). We then measured the level of expression of CaBP1, ERp60 and Crc; no differences in the expression levels of CaBP1, ERp60 and Crc were observed between controls and GFP knockdown males (Figs 1B and S3).

When looking at interacting genes, we noticed that Crc networked in both STRING and BioGRID with another hit from the SOLiD sequence data, Glycoprotein 93 (Gp93, human HSP90B1), which in turn interacted with yet another hit, CG2918 (human HYOU1) (Table 1). Utilizing these genes, and all the genes with which they were predicted to interact, we employed the STRING program to make a network (Fig 3C). From this it was evident that the genes formed a network, with Gp93, Protein disulfide isomerise (Pdi), and CaBP1 identified as hubs (Fig 3C).

All of the genes used to make the interaction map had human homologues (Fig 3D). These proteins locate to the ER and are involved in the regulation of protein folding (Fig 3E). Of note, all homologous genes in this pathway that were highly overexpressed when Ets96B was knocked down in the CNS, are involved in disulphide bridge formation (Crc, ERp60 and CaBP1) (Fig 3E, proteins marked in red and underlined).

Ets96B affects starvation resistance

Given that knocking down Ets96B affected genes involved in energy homeostasis (mitochondrial electron transport chain, Table 1) and protein production (Fig 3E), as well as inhibited the expression of CG5966 (Drosophila homologue of human pancreatic lipase, PNLIP), we wanted to understand if loss of Ets96B had any effect on metabolism. Furthermore, to determine if Ets96B had a developmental phenotype, both the pan-neuronal elav-GAL4 driver and the Ets96B-GAL4 driver [32] were employed. Ets96B is located on the right arm of the third chromosome at 96A22 in the forward orientation and encodes for two isoforms (S1 Fig). Recently, a 832 bp region from the third intron of Ets96B-RB, or fourth intron of Ets96B-RC, was cloned and used to make a GAL4 transgenic line (GMR36H05-GAL4, which we refer to as Ets96B-GAL4) [32] (S1 Fig).

To begin, we performed a starvation assay utilizing the Drosophila Activity Monitoring System (DAMS), as it is able to uncover broad defects in metabolism [33]. Knocking down Ets96B in the entire CNS during development using either Ets96B^RNAi1 or Ets96B^RNAi2 made flies more susceptible to starvation (P < 0.005) (Fig 4A). A similar phenotype was observed when Ets96B was knocked down throughout development using the Ets96B-GAL4 driver (P < 0.005) (Fig 4A). To clarify if this was indeed a developmental phenotype, we used the elav-Gal4,tubGal80^ts driver [34] to knockdown Ets96B specifically in adults using either Ets96B^RNAi1 or Ets96B^RNAi2. Remarkably, this induced the opposite phenotype, where males were significantly more resistant to starvation than controls (P < 0.01) (Fig 4A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Ets96B regulates starvation resistance.

(A) We tested the effect of Ets96B-knockdown in flies using the starvation survival assay. 5–7 day old control and Ets96B knockdown males (GAL4 driver crossed to either Ets96B^RNAi1 or Ets96B^RNAi2) were placed in a vial containing 1% agarose and maintained at 25°C. DAMS was used to monitor activity. (n = 30–60 flies per genotype, one-way ANOVA with Tukey’s post hoc test for multiple comparisons) (B) To examine how starvation affects the expression level of Ets96B, RNA was extracted from normal fed flies, as well as after 24 and 48 h of starvation. (C) To examine how nutritional state affects the expression levels of Ets96B, RNA was extracted under different nutritional states. Flies fed ad lib were set as 100%, represented by 1 on the graphs (B, C: n = 10 replicates, one-way ANOVA with Tukey’s post hoc test for multiple comparisons) (D, E) Triglyceride levels were determined in male flies at 0, 12 and 24 hours of starvation. The (D) elav-Gal4 driver was used to express Ets96B RNAi (Ets96B^RNAi1 or Ets96B^RNAi2) throughout development (UAS-Ets96B^RNAi1 and UAS-Ets96B^RNAi2) and (E) the elav-GAL4,tub-GAL80^ts driver was used to knockdown Ets96B (Ets96B^RNAi1 or Ets96B^RNAi2) in adult males. (D, E: n = 30 males per treatment, assay was repeated at least 10 times for each genotype, one-way ANOVA with Tukey’s post hoc test for multiple comparisons). In all graphs significance levels are indicated: *, P<0.05; **, P<0.01; ***, P <0.005. Error bars = SEM.

https://doi.org/10.1371/journal.pgen.1006104.g004

To establish if Ets96B transcript levels in the brain were regulated by starvation, we performed qPCR analysis on heads from adult males starved for either 24h or 48h. Starving males for 24h significantly increased Ets96B expression (P < 0.01) (Fig 4B). Additionally, we determined if macronutrient content influenced Ets96B expression. Transcript levels of males fed a control diet (10 g/dl:10 g/dl sucrose:brewer’s yeast) were set as 100%, represented as 1 on the graph (Fig 4C). However, none of the diets had a significant effect on Ets96B expression.

Ets96B affects lipid storage

Given that starvation resistance was affected in Ets96B knockdown males, to understand if Ets96B regulates normal feeding behaviour a CAFE assay was performed to measure how much food flies fed ad libitum consumed during a 24h period [35]. There was no significant difference in the amount of food consumed when Ets96B was knocked down in the nervous system by expressing Ets96B^RNAi1 throughout development (S2 Fig). We also performed the CAFE assay using the elav-Gal4,tubGAL80^ts driver to knockdown Ets96B specifically in adults. Again no significant difference was observed between Ets96B knockdown flies or controls (S2 Fig). Perhaps these are not surprising results, given that our SOLiD data revealed no significant change in the expression of any gene known to directly influence feeding behaviour (Table 1 and S1 Dataset).

Next, we determined the stored lipid content before and during starvation. For this we measured triacylglyceride (TAG) levels in normal fed flies, as well as during starvation. Interestingly, the lipid content of elav-GAL4;UAS-Ets96B^RNAi1 or elav-GAL4;UAS-Ets96B^RNAi2 knockdown males was significantly higher than either control (P < 0.005) (Fig 4D). After 12h and 24h of starvation all strains had lower total TAG levels and there was no significant difference between strains (Fig 4D). This suggests that during the first 12h of starvation, Ets96B^RNAi knockdown males mobilized more of their stored lipids than control flies.

Next, using the elav-Gal4,tubGal80^ts driver, we knocked down Ets96B specifically in adults. Similar to the starvation assay, this induced the opposite phenotype, where the TAG content of Ets96B^RNAi1 or Ets96B^RNAi2 knockdown males was significantly lower than either control (Ets96B^RNAi1 P < 0.01, Ets96B^RNAi2 P < 0.05) (Fig 4E). As before, after 12h and 24h of starvation all strains had lower lipid contents and there was no significant difference between strains (Fig 4E). Contrary to knocking down Ets96B throughout development, when Ets96B was knocked down specifically in the adult nervous system, males mobilized significantly less of their lipid stores during the first 12h of starvation.

Even though knocking down a random gene, such as EGFP, in the nervous system had no effect on the Ets96B-regulated molecular chaperones, CaBP1, ERp60 or Crc, we still wanted to confirm that activating the RNAi system in the CNS didn’t induce unspecific phenotypes. Therefore, we measured triacylglyceride (TAG) levels in normal fed flies, as well as during starvation; of controls (elav-GAL4;UAS-GFP) and GFP knockdown males, no differences in TAG levels were observed (Figs 3C and S3).

Ets96B regulates startle-response and sleep

One of the molecular chaperones upregulated when Ets96B was knocked down in the nervous system, Crc, is necessary for startle-induced locomotion. Yamamoto et al. [36] reported that Crc mutants lack a startle-response. Therefore, we hypothesized that Ets96B knockdown males, where Crc is overexpressed (see Fig 3A and 3B), would have the opposite phenotype and be excessively responsive to being startled. Using the same assay [36], we found that knocking down Ets96B in the nervous system throughout development induced a heightened startle-response (P < 0.01) (Fig 5A). When Ets96B was knocked down specifically in adults the startle-response was the same as control males (P = 0.57) (Fig 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Ets96B knockdown during development induces a heightened startle-response.

(A, B) Startle-response test demonstrating Ets96B knockdown males have a hyperactive startle-response. (A) Ets96B knocked down throughout development using UAS-Ets96B^RNAi1 or UAS-Ets96B^RNAi2 crossed to the pan-neuronal driver elav-GAL4 (B) Ets96B knocked down only in adults using UAS-Ets96B^RNAi1 crossed to the pan-neuronal driver and temperature sensitive allele of the GAL4 inhibitor GAL80 elav-GAL4, tub-Gal80^ts. (A, B: n = 50 males per strain; ** P<0.01 compared with controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons). (C) The DAMS system was used to monitor locomotion for the first four hours flies were placed in the system. (D) The DAMS system was used to monitor locomotion prior to and after light stimulation. (E) The DAMS system was used to monitor locomotion prior to and after sound stimulation (65–70 dB). Only Ets96B^RNAi1 was used for this assay. (C-E: n = 30–60 males per strain; * P < 0.05, ** P<0.01, *** P < 0.005 compared with controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons) In all graphs error bars = SEM.

https://doi.org/10.1371/journal.pgen.1006104.g005

Ets96B was further assessed for behavioural defects using a DAMS movement based assay. During the first hour in the DAMS knocking down Est96B (Ets96B^RNAi1 or Ets96B^RNAi2), using either elav-GAL4 or Ets96B-GAL4, had significantly more movement than controls (P < 0.01, except Ets96B-GAL4;UAS-Ets96B^RNAi1 P < 0.005) (Fig 5C) This hyperactivity continued during the second hour in the DAMS (P < 0.05) (Fig 5C). The activity level of all genotypes was similar throughout the last two hours monitored. Furthermore, during the first 30 minutes after light shock, both elav-GAL4;UAS-Ets96B^RNAi and Ets96B-GAL4;UAS-Ets96B^RNAi knockdown males, using either Ets96B^RNAi1 or Ets96B^RNAi2, had significantly more movement than either control (P < 0.01, except elav-GAL4;UAS-Ets96B^RNAi1 P < 0.05), but not 30 minutes prior to or 30–60 minutes after light shock (Fig 5D). This indicates that knocking down Ets96B in the nervous system during development increased exploration/anxiety when the setting was novel, as well as after a light-shock stimulus.

Next, to check that the DAMS measured response was not specific for light-shock stimulus, we repeated the assay using a loud white noise (65-70dB) to induce stress. Similar to what was observed with light-stimulus, during the first 30 minutes after noise shock, elav-GAL4;UAS-Ets96B^RNAi1 knockdown males had significantly more movement than either control (P < 0.005), but not 30 minutes prior to or 30–60 minutes after light shock (Fig 5E)

Finally, to check for any general effects on the neurons due to overexpression of an RNAi, we again expressed GFP under the control of elav-GAL4 and used two different GFP RNAi lines to inhibit GFP expression. We then tested these males in the DAMS for any effect on the startle-response. Similar to what was observed in the TAG assay, general overexpression of an RNAi in the entire nervous system had no effect on the startle-response (Figs 3D and S3).

In addition, we employed the DAMS to determine if there was an effect on general activity and sleep/wake behaviour (Fig 6A). To measure locomotor activity, male flies were placed in the DAMS and activity was quantified every minute for a 24h period. When Ets96B was knocked down using either the elav-GAL4 (P < 0.01) or Ets96B-GAL4 (P < 0.01) driver, males were significantly more active (Fig 6B). When Ets96B was knocked down specifically in adults, activity was similar to controls (P = 0.58 compared to elav-Gal4,tubGal80^ts heterozygous control) (Fig 6C). We also observed that elav-GAL4 > Ets96B^RNAi knockdown flies (Ets96B^RNAi1 or Ets96B^RNAi2) slept significantly less during the night than controls (Fig 6A and 6D). When Ets96B was specifically knocked down in adults there was no difference in sleep/wake behaviour between knockdown males and controls (Fig 6E, only Ets96B^RNAi1 used for this experiment).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Ets96B regulates activity and sleep/wake behaviour.

The DAMS system was used to monitor locomotion and sleep/wake behaviour over a 48 hour period. (A) Bar diagram indicating general activity over a 48 hour period of adult flies maintained on a 12:12 hour light:dark cycle. White bar indicates lights-on and dark bar and gray highlight indicate lights-off. Representative result for one run using Ets96B^RNAi1 is shown (B) Total number a beam breaks over a 24 hour period in control flies or flies were Ets96B was knocked down (Ets96B^RNAi1 or Ets96B^RNAi2) throughout development either in the entire nervous system (elav-GAL4) or specifically in Ets96B neurons (Ets96B-GAL4). (C) Total number a beam breaks over a 24 hour period in control flies or flies were Ets96B was knocked down (Ets96B^RNAi1) specifically in adult neurons (elav-GAL4;tubGAL80^ts). (D) Total minutes flies slept per night in control flies or flies were Ets96B was knocked down (Ets96B^RNAi1 or Ets96B^RNAi2) throughout development, either in the entire nervous system (elav-GAL4) or specifically in Ets96B neurons (Ets96B-GAL4). (E) Total minutes flies slept per night in control flies or flies were Ets96B was knocked down (Ets96B^RNAi1) specifically in adult neurons (elav-GAL4;tubGAL80^ts). (B-E, n = 30–60 males per strain; * P < 0.05, ** P<0.01 compared with controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons) In all graphs error bars = SEM.

https://doi.org/10.1371/journal.pgen.1006104.g006

Drosophila Ets96B and mouse Etv5 are expressed in dopaminergic neurons

Next, to see where Ets96B could be involved in nervous system signalling we crossed Ets96B-GAL4 to UAS-GFP and mapped GFP expression in the adult brain. Since dopamine is known to be a major regulator of the startle-response, we co-stained the brains for Tyrosine hydroxylase (TH), which is necessary for the production of L-DOPA and specifically marks dopaminergic neurons [37]. Interestingly we saw that Ets96B was almost exclusively expressed in dopaminergic neurons (Fig 7A–7C). Specifically, we saw Est96B-GAL4 induced GFP expression in the suboesphageal ganglionic (SOG) region in the brain (Fig 7A and 7B). Furthermore, Est96B-GAL4 induced GFP-expression co-localized with TH in a subset of dopaminergic PPM1/2 neurons (Fig 7C). Interestingly, within the PPM1/2 neurons there were two neurons that were GFP positive but TH negative (Fig 7C, inset, white arrows). There was also a strong overlap between Ets96B-GAL4 driven GFP and TH in the eye and the ventrolateral protocerebrum (VLP) (Fig 7A). Although GAL4 driven GFP is not an entirely reliable source for mapping gene expression, this result is an indication that Ets96B may be expressed in the regions of the Drosophila brain governing locomotion, as well as odour and visual-driven memory and learning. Furthermore, the Ets96B-GAL4 driver may not recapitulate the entire Ets96B expression pattern.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Ets96B and Etv5 expressed on dopaminergic neurons.

(A-C) Ets96B-GAL4 was crossed to UAS-GFP to examine Ets96B expression in adult male brains. Adult male brains were then stained for GFP and Tyrosine hydroxylase (TH) expression. (A) Anterior section showing co-expression of GFP and TH in the eye, the ventrolateral prototcerebrum (VLP and the suboesophageal ganglion (SOG, see inset). (B) Midbrain, no co-expression was observed in the dopaminergic clusters PAM or PAL, while coexpression was observed in the SOG (see inset). (C) In the posterior brain section no co-expression was observed in dopaminergic clusters PPL1 or PPL2, but there was co-expression of GFP and TH in cluster PPM1/2 (see inset). There were also two neurons near the dopaminergic PPM1/2 cluster that were GFP specific, and had no TH expression (see inset, white arrow). In A-C size bar is equivalent to 100 μm. (D-G) Black scale bar, 1mm. (H-P) white scale bar, 0.5 mm. Bregma levels and described brain regions are according to Allen Mouse Brain Atlas. (H, I) Cortex layers 1–6. (J) basolateral amygdaloid nucleus, anterior part (BLA), basolateral amygdaloid nucleus, ventral part (BLV), basomedial amygdaloid nucleus, posterior part (BMP), central nucleus of amygdala, lateral part (CeL), dorsal endopiriform claustrum (DEn), stria medullaris (STIA), ventral endopiriform claustrum (VEn), (K) ventromedial thalamic nucleus (VM), (L) dorsomedial hypothalamic nucleus, ventral part (DMV), nigrostriatal tract (ns), ventromedial hypothalamic nucleus, central part (VMHC), ventromedial hypothalamuc nucleus, dorsomedial part (VMHDM), ventromedial hypothalamic nucleus, ventrolateral part (VMHVL), (M) Cortex layers 1–6, (N) parabrachial pigmented nucleus of the VTA (PBP), ventral tegmental area (VTA), (O) field CA1 of the hippocampus (CA1), field CA2 of the hippocampus (CA2), field CA3 of the hippocampus (CA3), lacunosum moleculare layer of the hippocampus (LMol), pyramidal layer of the hippocampus (Py), (P) crus 1 of the ansiform lobule (Crus1), lobule 4 and 5 of the cerebellar vermis (4/5Cb).

https://doi.org/10.1371/journal.pgen.1006104.g007

In order to specify if the expression pattern of Etv5 was enriched in dopaminergic-rich regions of the mouse brain, we performed detailed mRNA in situ hybridization mapping on coronal sections (Fig 7D–7G, S1 Table). Etv5 mRNA was expressed in a specific and distinguishable pattern in certain brain regions, predominantly the cerebral cortex (Fig 7H, 7I and 7M), the amygdala (Fig 7J), and the hypothalamus (Fig 7L). Within the cerebral cortex there was conspicuously high expression in cortical layers 2, 4 and 6, while expression was almost absent from layers 1, 3, and 5 (Fig 7H, 7I and 7M). Furthermore, high expression was also observed in the dorsal and ventral endopiriform claustrum (Fig 7J). Strong expression was observed in limited parts of the amygdale, including the basolateral amygdaloid nucleus, ventral part (Fig 7J). Considerably high expression was found in the bed stria terminalis and the central amygdaloid nucleus (Fig 7J). Etv5 expression was observed at low to moderate levels in the hypothalamus, whereas expression was almost absent in the thalamus (Fig 7K and 7L). In conclusion, hypothalamic expression was found in the dorsomedial hypothalamic nucleus and also the central, dorsomedial and ventrolateral part of the ventromedial hypothalamic nucleus (Fig 7L). Strong expression was also identified in the nigrostriatal tract (Fig 7L). In the mesencephalon expression was restricted to the ventral tegmental area (VTA), including the parabrachial pigmented nucleus of the VTA (Fig 7N). In the hippocampus, Etv5 was strongly expressed in the pyramidal cell layer, CA1 through CA3. Some expression was also found in the lacunosum moleculare layer (Fig 7O). Low levels of Etv5 mRNA expression were observed in the cerebellum, including the Granular, Molecular and Purkinje cell layers (Fig 7P). Other regions in the brain are listed in S1 Table.

Ets96B required in dopaminergic neurons for proper metabolism and behaviour

Given that Ets96B and mouse Etv5 seemed to be expressed in dopaminergic-rich regions, we used the ple-GAL4 driver (pale [ple] is the Drosophila tyrosine hydroxylase, necessary to convert tyrosine to L-DOPA) to knockdown Ets96B transcript specifically in dopaminergic neurons throughout development. We then examined the transcription levels of a select number of genes-ple, Vesicular monoamine transporter (Vmat), and Dopamine transporter (DAT)-known to influence dopamine production, release or reuptake (Fig 8A–8C). Transcription levels of the ple-GAL4 heterozygous controls were set at 100%, shown as 1 on the various graphs. When Ets96B expression was knocked down in dopaminergic neurons, the transcript level of ple was significantly upregulated (P < 0.005) (Fig 8A). Furthermore, the expression of Vmat, involved in the transportation of monoamines such as dopamine into the synaptic vesicles, was also increased in Ets96B knockdown males (P < 0.05) (Fig 8B). Finally, the expression of DAT, which is necessary to pump dopamine out of the synapse back into cytosol, was significantly lower when Ets96B expression was knocked down in dopaminergic neurons (P < 0.005) (Fig 8C). Increased dopamine production and decreased dopamine reuptake should lead to prolonged dopamine signalling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Ets96B required in dopaminergic neurons.

(A-C) Ets96B was knocked down throughout development (Ets96B^RNAi1 or Ets96B^RNAi2) specifically in dopaminergic neurons using the ple-GAL4 driver and the transcript level of genes (A) ple, (B) Vmat and (C) DAT was measured. RNA was collected from the heads of 5- to 7-day-old males for each genotype. qPCR was repeated at least 7 times for each transcript. (n = 25 males per treatment; * P<0.05, *** P<0.005 compared with controls, one-way ANOVA with Tukey’s post hoc test for multiple comparisons) (D) Ets96B was knocked down throughout development (Ets96B^RNAi1 or Ets96B^RNAi2) specifically in dopaminergic neurons using the ple-GAL4 driver. 5–7 day old control and Ets96B knockdown males were placed in a vial containing 1% agarose and maintained at 25°C, DAMS was used to monitor activity. (n = 30–60 flies per genotype, ** = P < 0.01, one-way ANOVA was used to determine significance with Tukey’s post hoc test for multiple comparisons). (E) The DAMS system was used to monitor locomotion prior to and after light stimulation. Only Ets96B^RNAi1 was used for this assay. (n = 30 males per strain; * P<0.05, *** P<0.005 compared with controls, Kruskal-Wallis non-parametric ANOVA was used to determine significance). In all graphs error bars = SEM.

https://doi.org/10.1371/journal.pgen.1006104.g008

Next, we asked if knocking down Ets96B specifically in dopaminergic neurons was sufficient to recapitulate the starvation and startle-response phenotypes. Similar to what was observed using the elav-GAL4 and Ets96B-GAL4 drivers, knocking down Ets96B specifically in dopaminergic neurons throughout development made adult males more susceptible to starvation (P < 0.01) (Fig 8D). Furthermore, lowering Ets96B transcript levels in dopaminergic neurons was sufficient to replicate the heightened light-induced startle-response (P < 0.005) (Fig 8E). To determine if the increased startle-response observed when Ets96B was knocked down specifically in dopaminergic neurons was due to increased dopamine signalling, we fed adult males the tyrosine hydroxylase inhibitor 3-iodo-L-tyrosine before performing the light-induced startle-response assay. In controls this was sufficient to abolish the light-induced startle-response (Fig 8E). In males where Ets96B was knocked down in dopaminergic neurons, inhibiting dopamine production reduced the startle-response to the same level as normal fed controls (Fig 8E). From this result we conclude that reducing Ets96B expression in dopaminergic neurons increases dopamine signalling.

Human ETV5 SNPs link to BMI and Bipolar disorder

The startle-response phenotype demonstrated by loss of Ets96B in the nervous system is similar to the effect of prepulse inhibition (PPI) in mammals. PPI is a neurological phenomenon where a weaker prestimulus (prepulse) inhibits the reaction to a subsequent stronger startle stimulus (pulse). In humans disrupted PPI responses are linked with bipolar disorder [38]. Using the GWAS Central database [39] we searched for ETV5 localized SNPs that had a significant link (P < 0.05) to either BMI or bipolar disorder. Employing this method we found 12 SNPs mapped to ETV5 that significantly linked to BMI (S2 Table). Interestingly, three of these SNPs were also significantly linked to bipolar disorder (S2 Table). Next, we made a map to define where these 12 SNPs localized within the ETV5 genomic region (Fig 9A). Of notable interest, the three BMI/bipolar associated SNPs all localized around ETV5 exons 6 and 7 (Fig 9A, blue highlighted SNPs). This is an interesting genetic region as there is evidence a truncated ETV5 transcript is produced that stops after exon 7, which would encode only the inhibitory N-terminal PEA3 domain [40]. Furthermore, this same region encoded a long non-coding antisense RNA (lncRNA) called ETV5-AS1 (Fig 9A). Using the NONCODE database [41] we determined that ETS5-AS1 was expressed in the adrenal glands, testis and thyroid.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Human ETV5 SNPs associated with body mass index and bipolar disorder.

(A) Localization of relevant single nucleotide polymorphisms (SNPs) in the ETV5 gene. SNP names according to NCBI dbSNP. The GWAS central database was used to link ETV5 associated SNPS to BMI. Interestingly, three SNPs were also associated with bipolar disorder (blue SNPs). The three bipolar-linked SNPs all localize around exons 6 and 7, which is also the location of the long, noncoding RNA (lncRNA) ETV5-AS1. Snap Proxy was then used to determine Pair-linked disequilibrium in various populations. (B) Linkage disequilibrium between SNPs in ETV5. Black boxes = exons, thick black lines between boxes = introns, red points = SNPs associated with BMI in GWAS, blue points = SNPs associated with bipolar disorder in GWAS. The LD between SNPs in the CEU population is expressed as R² and displayed as a color code below the plot.

https://doi.org/10.1371/journal.pgen.1006104.g009

Using Encode Roadmap {http://www.encode-roadmap.org/} (S4 Fig) and RegulomeDB [42] (Table 2) databases we determined the methylation state of the ETV5 gene in various brain regions. By means of these databases we found that SNPs rs7647305 and rs1356292 mapped to the promoter region. In fact, rs7647305 was predicted to map to the TATA-box and rs1356292 was predicted to be at the transcription start site (Table 2). It was interesting that rs4686730 was predicted to map to an insulin promoter factor 1 (IPF1) binding site (Table 2), considering that Etv5 was found to be necessary for the secretion of insulin in mice [43]. When examining the chromatin state of the three SNPs linked to BMI and BD, (rs4686730, rs7433760 and rs4686728) they were all quiescent in the brain germinal matrix (BGM) (Table 2). Furthermore, in brain regions known to control reward response, such as the anterior caudate (BAC) and midfrontal lobe (BMFL), rs7433760 mapped near a region with H3K4 single methylation, indicating an active enhancer (S4 Fig). In the RegulomeDB database, rs7433760 was predicted to map to an active enhancer in the angular gyrus (BAG), dorsolateral prefrontal cortex (BDPF) and the hippocampus middle (BHM), areas involved in decision making and memory. Finally, we made a linkage disequilibrium map that demonstrated tight linkage between all ETV5 SNPs linked to BMI and BD, except for BMI-linked rs6777779 (Fig 9B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Regulatory elements and chromatin states associated with ETV5 SNPs.

https://doi.org/10.1371/journal.pgen.1006104.t002

Next we wanted to determine if the human homologues of the Drosophila ER molecular chaperones (Fig 3D), whose expression is regulated by Ets96B, also linked to BMI and BD. Using data from multiple GWAS studies (BMI, Waist-hip ratio or bipolar disorder [8, 44–46]) we searched for SNPs localizing to the human ER molecular chaperone homologues. We found SNPs with a significant link (P < 0.05) to either BMI or BD that localized to three of the homologues (PDIA3, PDIA6 and HYOU1) (Fig 10A). We located 9 SNPs mapped to PDIA3, 5 SNPS mapped to PDIA6 and 5 SNPS mapped to HYOU1 that significantly linked to BMI. Furthermore, we found 1 SNP mapped to PDIA3, 3 SNPs mapped to PDIA6 and 3 SNPs mapped to HYOU1 that significantly linked to BD (Fig 10A). Next, we made linkage disequilibrium maps for PDIA6 and HYOU1 that demonstrated tight linkage between all PDIA6 SNPs linked to BD (Fig 10B); the same was true for HYOU1, except for BD-linked rs511134 (Fig 10C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Human ER molecular chaperon SNPs associated with body mass index and bipolar disorder.

(A) Summary figure of SNP association results from Genome Central [39] for PDIA3, PDIA6 and HYOU1 linkage to body mass index (BMI), Weight-hip ratio and bipolar disorder. Horizontal axis lists significant SNPs and genes. The red line represents P = 0.05 nominal significance level; above the line represents significant results. (B) Linkage disequilibrium between SNPs in PDIA6. Boxes = exons, lines between boxes = introns, The LD between SNPs in the CEU population is expressed as R² and displayed as a color code below the plot. (C) Linkage disequilibrium between SNPs in HYOU1. Boxes = exons, lines between boxes = introns, The LD between SNPs in the CEU population is expressed as R² and displayed as a colour code below the plot.

https://doi.org/10.1371/journal.pgen.1006104.g010

Fly stocks and maintenance

w*, P{w[+mW.hs]=GawB}elav[C155], w¹¹¹⁸;P{y[+t7.7] w[+mC] = GMR36H05-GAL4}attP2 P{GawB}elavC155 w*; P{FRT(whs)}G13 P{tubP-GAL80}LL2, and the w*; P{w[+mC] = ple-GAL4.F}3 driver lines, the w*; UAS-GFP line, as well as the Transgenic RNAi Project (TRiP) [25] RNAi lines y¹,sc*,v¹;P{TRiP.HMS00092}attP2 (Ets96B^RNAi1) and y1, v1; P{TRiP.JF02226}attP2 (Ets96B^RNAi2) and the RNAi lines w¹¹¹⁸;P{w[+mC] = UAS-GFP.dsRNA}143 (GFP^RNAi1) and w¹¹¹⁸;P{w[+mC] = UAS-GFP.dsRNA}142 (GFP^RNAi2) [74] were received from the Bloomington Stock Center (Bloomington, Indiana, USA; http://flystocks.bio.indiana.edu/). All strains were crossed into the same w¹¹¹⁸ genetic background. All flies, unless otherwise stated, were maintained on Jazz mix standard fly food (Fisher Scientific, Göteborg, Sweden), supplemented with 5% Brewer’s yeast extract (VWR, Stockholm, Sweden). Flies were maintained at 25°C in an incubator at 60% humidity on a 12:12 light:dark cycle. Flies crossed to GAL4 drivers (GAL4 drivers crossed to UAS-RNAi lines) and controls (GAL4 drivers or UAS-RNAi crossed to w¹¹¹⁸) were raised at 25oC until the adults emerged (F₁ generation); once collected adults were raised at 29oC for the appropriate times. All assays were performed at 25°C. In all assays, the GAL4 drivers and UAS transgenic flies were crossed to w¹¹¹⁸ flies and their F1 progeny used as controls.

Multiple sequence alignments and Phylogenetic analysis

The protein sequences used for the sequence alignment and phylogenetics analysis were mined using Flybase (www.flybase.org), NCBI HomoloGene (www.ncbi.nlm.nih.gov/homologene), and Hmmer (www.ebi.ac.uk/Tools/hmmer/). Protein sequences were aligned using CLC sequence Viewer 6 using default parameters. The sequence alignment used for phylogenetic analysis was exported in clustal format and viewed using MEGA5 [75]. The file was then transformed into a MEGA file for further analysis. Evolutionary history was inferred using the Maximum-likelihood method. The optimal tree with the sum of branch length = 3.794 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (5000 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Dayhoff matrix based method and are in the units of the number of amino acid substitutions per site. The rate variation among sites was modelled with a gamma distribution (shape parameter = 5). The analysis involved 35 amino acid sequences. All positions containing gaps and missing data were eliminated. There were a total of 170 positions in the final dataset. Evolutionary analyses were conducted in MEGA5 [75].

Library preparation and sequencing

RNA purification

The phenol-chloroform method was used for RNA extraction from tissue samples [79]. Fifty fly heads were homogenized with 800 μl TRIzol (ThermoFisher Scientific, MA, USA), 200 μl Chloroform (Sigma-Aldrich, MO, USA) was added and samples were centrifuged at 12000 rpm for 15 minutes at 4°C. The aqueous layer, which contained RNA, was separated and 500 μl isopropanol (Solvaco AB, Sweden) was added. The RNA was precipitated by storing the samples at -32°C for 2 hours. Samples were centrifuged at 12000 rpm for 10 minutes at 4°C, to collect the RNA pellets, which were then washed with 75% ethanol (Solvaco AB, Sweden) to remove the organic impurities. Samples were allowed to air dry to remove any traces of ethanol. Dried RNA pellets were dissolved in 21.4 μl of RNAse free water (Qiagen GmBH, Germany) and 2.6 μl of DNAse incubation buffer (Roche GmBH, Germany). The samples were incubated at 75°C for 15 minutes to ensure complete dissolution of RNA-pellets. 2 μl of DNAse I (10 U/μl, Roche GmBH, Germany) was added to each sample, and incubated at 37°C for 3 hr to remove DNA contamination. DNAse was deactivated by incubating the samples at 75°C for 15 minutes. Removal of DNA was confirmed by PCR using Taq polymerase (5U/μl, Biotools B & M Labs, Spain), followed by agarose gel electrophoresis. The RNA concentration was measured using a nanodrop ND 1000 spectrophotometer (Saveen Werner).

cDNA synthesis

cDNA was synthesized from RNA template using dNTP 20 mM (ThermoFisher Scientific, MA, USA), random hexamer primers and M-MLV Reverse Transcriptase (200 U/μl, ThermoFisher Scientific, MA, USA) by following manufactures instructions. cDNA synthesis was confirmed by PCR followed by agarose gel electrophoresis.

qRT-PCR

Relative expression levels of three housekeeping genes (EF-1, Rp49 & RpL11) and of the genes of interest were determined with quantitative RT-PCR (qPCR). Each reaction, with a total volume of 20 μl, contained 20 mM Tris/HCl pH 9.0, 50 mM KCl, 4 mM MgCl2, 0.2 mM dNTP, DMSO (1:20) and SYBR Green (1:50000). Template concentration was 5 ng/μl and the concentration of each primer was 2 pmol/μl. Primers were designed with Beacon Designer (Premier Biosoft) using the SYBR Green settings. All qPCR experiments were performed in duplicates; for each primer pair a negative control with water and a positive control with 5 ng/μl of genomic DNA were included on each plate. Amplifications were performed with 0.02 μg/ml Taq DNA polymerase (Biotools, Sweden) under the following conditions: initial denaturation at 95°C for 3 min, 50 cycles of denaturing at 95°C for 15 sec, annealing at 52.8–60.1°C for 15 sec and extension at 72°C for 30 sec. Analysis of qPCR data was performed using MyIQ 1.0 software (Bio-Rad, CA, USA) as previously reported [80]. Primer efficiencies were calculated using LinRegPCR [81] and samples were corrected for differences in primer efficiencies. The GeNorm protocol described by Vandesompele et al. [82] was used to calculate normalization factors from the expression levels of the housekeeping genes. Grubbs' test was performed to remove outliers. Differences in gene expression between groups were analyzed with ANOVA followed by Fisher's PLSD test where appropriate. P<0.05 was used as the criterion of statistical significance. The following primers were used: EF-1 F: 5´-GCGTGGGTTTGTGATCAGTT-3´, R: 5´-GATCTTCTCCTTGCCCATCC-3´; Rp49 F: 5´-CACACCAAATCTTACAAAATGTGTGA-3´, R: 5´-AATCCGGCCTTGCACATG-3´; RpL11 F: 5´-CCATCGGTATCTATGGTCTGGA-3´, R: 5´-CATCGTATTTCTGCTGGAACCA-3´, Ets96B F: 5´-CCTCCATAGACTACCATAT-3´, R: 5´-GTAACTCAACCTCAATGC-3´; CaBP1 F: 5´-GCAGCGTTAGTGCCTTCTATT-3´, R: 5´-CTTTCAGCACCTCCCGGTC-3´; ERp60 F: 5´-GACTTTGCCACCACCCTAAAA-3´, R: 5´-TACTCGGGCTTCAATCGCTTG-3´; Crc F: 5´-GAAAACTGGGAGGATACGTGG-3´, R: 5´-GAGAGGTCTGAATGCCTTTGTC-3´; ple F: 5´-CGAGGACGAGATTTTGTTGGC-3´, R: 5´-TTGAGGCGGACCACCAAAG-3´; Vmat F: 5´-CGTGACCTTCGGGACGATAG-3´, R: 5´-ACTAGAGCGGGAAAACCAGC-3´; DAT F: 5´-GCTTCAAACCATAAGTTCTAA-3´, R: 5´-TCGGACTTGATATTATCTACAA-3´

Starvation assay

Starvation resistance assay was performed in a similar fashion as Hergarden et al (2012) [33] Starvation resistance was measured by placing 30, 5 to 7 days old, male flies in individual tubes containing 1% agarose in the Drosophila activity monitoring system (DAMS) (Trikinetics, MA, USA). DAMS was performed at 25°C in an incubator, on a 12h:12h light:dark cycle.

Triacylglycerol determination

Flies (25 males) were homogenized with 100 μl of PBST buffer (1x phosphate buffer with 10% Tween20), incubated at 70°C for 5 minutes and then centrifuged at maximum speed for 10 minutes. The supernatant was transferred into a new eppendorff and used as samples. A glycerol standard was used to generate a standard curve with concentrations of 1.0, 0.8, 0.6, 0.4, and 0.2 mg/ml equivalent triolein concentration. 100 μl of free glycerol reagent was added to 10μl of blank (PBST), standards or samples and initial absorbance at 540nm was measured after incubation at 37°C for 15 minutes. The concentration of free glycerol in the samples was calculated from the standard curve generated by this initial absorbance value. Then, 20 μl of triglyceride reagent was added in each standard and sample and incubated at 37°C for 15 minutes. The final absorbance was taken at 540 nm to calculate triglyceride concentration from the generated standard curve. The protein concentration of each sample was measured by the Bio-Rad (Hercules, CA, USA) protein assay kit. The concentrations of free glycerides and Triglycerides in samples (mg/mg of protein) were calculated from five replicates.

Startle-response assay

We measured startle behaviour as described previously [50]. In brief, single 5–7 day old males were placed in a plastic culture vial and subjected to a mechanical disturbance by a gentle, yet sudden, tap of the vial. The vial was then placed horizontally on a bench top. Startle-induced locomotion was calculated as the number of seconds each fly was active within the 60 s period following the startle-induced locomotion. All tests were done at 25°C with 3 hours of lights on. The sample size for all genotypes was N = 50.

DAMS activity assays

Flies were tested using the Drosophila Activity Monitoring System (DAM2, TriKinetics, www.trikinetics.com). Flies were placed in 5mm x 65mm glass tubes during the morning, with food placed at one end of the tube. Flies were continually monitored by the use of a light beam, with the number of beam breaks continuously recorded over a 48 hour period. A light shock or sound shock stimulus was given approximately one day after the animals were first placed in the tubes, at 12 noon. The number of beam breaks during the first four hours the flies were housed in the tube (when the arena is at its most novel and anxiogenic), the half hour prior to the light shock or sound shock stimulus and the hour post light shock or sound shock stimulus were all recorded.

Drug treatment

For the pharmacological manipulations 1% agar with 5% sucrose was used as a vehicle medium. The tyrosine hydroxylase inhibitor, 3-iodo-L-tyrosine (Sigma-Aldrich, Stockholm, Sweden) was mixed into the medium to a final concentration of 3 mM. There day old male flies were transferred and kept on the media for 48 hours.

Drosophila brain immunohistochemistry

Male flies were anesthetized and decapitated, and the proboscis was removed. Heads were placed into a staining glass bowl containing 4% paraformaldehyde in 0.1 M sodium phosphate buffer (PBS) and left to fixate in the dark for 2 hr on ice. After fixation, heads were placed in a petri dish containing 4% agarose, and brains were dissected under a light microscope using fine forceps. Brains were washed four times for 15 min each with 0.1 M PBS. Tissues were blocked in 10% normal goat serum (NGS) for 1 hr. The NGS was then discarded, and tissues were incubated with primary antibodies (Millipore, Darmstadt, Germany, anti-GFP chicken polyclonal diluted 1:200 and Immunostar, Wisconsin, USA, mouse anti-Tyrosine hydroxylase 1:500), diluted in 0.01 M PBS containing 0.25% Triton X-100 (PBX) for 2 days at 4°C. Bowls were sealed with parafilm and aluminium foil. Following 48 hr of incubation, brains were washed four times for 15 min each with 0.01 M PBX and incubated with secondary antibody (ThermoFisher Scientific, MA, USA, goat anti-chicken Alexa Fluor 488 and goat anti-mouse Alexa Fluor 594 goat anti-rabbit), diluted 1:1000 in 0.01 M PBX, overnight. Bowls were sealed with parafilm and aluminum foil. Tissues were washed once with 0.01 M PBX for 15 min and twice with 0.01 M PBS for 15 min. Samples were mounted in Vectashield (Vector Laboratories, Peterborough, UK). Images were captured on a Zeiss LSM 510 confocal microscope and visualized using ImageJ.

In situ hybridization

Ethical statement

All animal procedures were approved by the local ethical committee in Uppsala (permit C 488/12) and follows international guidelines.

Pairwise linkage disequilibrium

All SNPs in ETV5 or close to ETV5, and that have been associated with BMI and/or dipolar disorder in genome-wide association studies (GWAS) were selected: rs1356292, rs1516725, rs1516726, rs4234589, rs4686728, rs4686730, rs6777779, rs6809651, rs7433760, rs9824882, rs7647305 and rs9816226. P-values of association with BMI and/or bipolar disorder were retrieved from GWAS central [39]. Pairwise linkage disequilibrium (LD) between the SNPs were retrieved from SNAP proxy [83], using “1000 Genome Pilot 1” as the SNP data set, and CEU as the population panel. Fig 9C was generated using the R package LDheatmap [84].

Statistical analysis

Mean and standard error from all replicates of each experiment was calculated. All analysis was performed with GraphPad Prism 4, and used ANOVA with appropriate post hoc analysis for multiple comparisons (See figure legends for which analysis was performed for each assay).