G-protein coupled Receptor Kinase 2 is required for rhythmic olfactory responses in Drosophila

Summary

Background

The Drosophila circadian clock controls rhythms in the amplitude of odor-induced electrophysiological responses that peak during the middle of night. These rhythms are dependent on clocks in olfactory sensory neurons (OSNs), which suggests that odorant receptors(ORs) or OR-dependent processes are under clock control. Since responses to odors are initiated by heteromeric OR complexes that form odor-gated and cyclic-nucleotide-activated cation channels, we tested whether regulators of ORs were under circadian clock control.

Results

The levels of G-protein coupled receptor kinase 2 (Gprk2) mRNA and protein cycle in a circadian clock-dependent manner with a peak around mid-night in antennae. Gprk2 overexpression in OSNs from wild-type or cyc⁰¹ flies elicits constant high amplitude electroantennogram (EAG) responses to ethyl acetate, whereas Gprk mutants produce constant low amplitude EAG responses. Odorant receptors (ORs) accumulate to high levels in the dendrites of OSNs around mid-night, and this dendritic localization of ORs is enhanced by Gprk2 at times when ORs are primarily localized in the cell body.

Conclusion

These results support a model in which circadian clock-dependent rhythms in Gprk2 abundance control the rhythmic accumulation of ORs in OSN dendrites, which in turn control rhythms in olfactory responses. The enhancement of OR function by GPRK2 contrasts with the traditional role of Gprks in desensitizing activated receptors, and suggests that GPRK2 functions through a fundamentally different mechanism to modulate OR activity.

Introduction

Many sensory systems are regulated by the circadian clock. Various insects including flies, moths, and cockroaches show circadian rhythms in odor-dependent electrophysiological and behavioral responses [1–5]. In mammals, the firing rate of isolated mouse olfactory bulb neurons is regulated by the circadian clock [6, 7], as is odor-evoked brain activity waves (e.g. event-related potentials or ERPs) in humans [8]. Daily rhythms in neuronal activity or sensitivity have been reported for other sensory systems such as the visual and auditory systems [9–12].

We previously reported that the circadian clock modulates olfactory responses in Drosophila: robust EAG responses are seen during mid-night and weak EAG responses are seen during mid-day [2]. These rhythms in EAG responses are controlled by the OSNs in Drosophila, which act as independent peripheral circadian oscillators [13]. Colocalization of the circadian oscillator and a rhythmic output to the OSNs indicates that the abundance and/or activity of ORs and/or OR-dependent processes are under clock control. Drosophila ORs are seven transmembrane domain proteins that share some structural similarities with G-protein coupled receptors (GPCRs) [14]. However, recent studies demonstrate that Drosophila ORs have an inverted membrane topology compared to canonical GPCRs [15, 16], and function as odor-gated and cyclic-nucleotide-activated cation channels [17, 18]. To understand how the clock modulates odor-dependent responses, we determined whether factors that modulate ORs were regulated by the circadian clock.

G-protein coupled receptor kinases (GPRKs) and arrestins act to terminate GPCR signaling in mammalian systems, thereby protecting cells from receptor overstimulation. GPRK-phosophorylated GPCRs are internalized by arrestin and subsequently degraded or recycled [19]. Two Gprk genes, Gprk1 and Gprk2, have been reported in Drosophila [20]. Gprk1 is enriched in photoreceptor cells, and expressing a Gprk1 dominant negative mutant in photoreceptors increases the amplitude of electroretinogram (ERG) responses [21]. Gprk2 is required for egg and wing morphogenesis, and embryogenesis in Drosophila [22, 23]. In mammals, seven Gprk genes are divided into 3 subfamilies based on sequence homology: the rhodopsin kinase or visual Gprk subfamily (Gprk1 and Gprk7), the β-adrenergic receptor kinase subfamily (Gprk2 and Gprk3), and the Gprk4 subfamily (Gprk4, Gprk5 and Gprk6) [24]. Gprk3 knockout mice are unable to mediate odor-induced desensitization of odorant receptors [25]. In contrast, loss of Gprk2 function in C. elegans olfactory sensory neurons results in reduced chemosensory behavior, suggesting that Gprk2 is necessary for GPCR signaling [26]. These results suggest that GPRKs play different roles in vertebrate and invertebrate olfaction.

Here we report that Gprk2 expression is regulated by circadian clocks in antennae, and that Gprk2 drives circadian rhythms in olfactory responses by enhancing OR accumulation in the dendrites of basiconic sensilla. Gprk2 mRNA and protein expression levels were high around mid-night, which is coincident with the peak of olfactory responses. Flies that overexpress Gprk2 in OSNs show constant high electroantennogram responses to ethyl acetate during 12h light: 12h dark (LD) cycles and accumulate high levels of ORs in OSN dendrites, whereas hypomorphic Gprk2 mutants show constant low EAG responses to ethyl acetate during LD. Based on these results, we propose that GPRK2 mediates cycles of OR accumulation in OSN dendrites to generate rhythms in EAG responses.

Results

Drosophila GPRK2 is most similar to GPRK4 from mammals

We compared Drosophila GPRK2 to members of the three mammalian Gprk subfamilies in humans and found the highest amino acid identity in the kinase catalytic domain, with lower levels of similarity in the flanking N-terminal and C-terminal regions. GPRK2 was most similar to human GPRK4, with 87% identity in the kinase domain and 45 % and 47 % identity in the N- and C-terminal regions, respectively. GPRK2 displays substantially less sequence identity to mammalian GPRK3, which is necessary to desensitize odorant receptors [25]. The N-terminal region of GPRK2 includes a unique stretch of amino acids (Gly124-Gly261, 138 amino acids) containing asparagine rich and glycine rich clusters (Supplemental Fig. 1). This unique amino acid region is not present in other Gprk subfamilies and is not homologous to sequences in other proteins based on BLAST searches. This data suggests that GPRK2 is a member of the GPRK4 family from mammals, but may have additional functions compared to other GPRK4 family members.

Two GPRK2 isoforms are expressed in olfactory sensory neurons

We used a newly generated anti-Drosophila GPRK2 antibody to demonstrate that GPRK2 protein is expressed in antennae (Fig. 1A; Fig. 2A, B). GPRK2 mobility in SDS-PAGE is 95KDa in antennae (Fig. 1A), though the estimated molecular weight of the 714aa GPRK2 open reading frame is ~80KDa. To determine if these ~95KDa bands corresponded to GPRK2, a V5 epitope tagged GPRK2 was expressed in S2 cells and probed with anti-V5 antibody and anti-GPRK2 antibodies. Both V5 and GPRK2 antibodies detected the same ~95KDa bands (Fig. 1B), thereby demonstrating that the 95KDa bands correspond to GPRK2.

Figure 1.

Identity and expression of Drosophila GPRK2. (A) GPRK2 expression in antennae from wild-type (WT), Gprk2 mutant (Gprk2⁶⁹³⁶, Gprk2^EY09213 and Gprk^pj1), and GPRK2 overexpression (Or83b-Gal4;UAS-Gprk2) flies, or in S2 cells transfected with plasmids containing actin promoter-driven Gprk2 ORF (pAc-Gprk2) or actin promoter-driven Gprk2 ORF + 5’ and 3’ UTRs (pAc-Gprk2+UTRs). GPRK2 runs as an α and β isoform in wild-type antennae. Molecular weight markers, in kilodaltons (KDa), are on the left. GPRK2 isoforms migrate at an apparent molecular weight of 95 KDa. (B) GPRK2 expression in S2 cells transfected with plasmids containing metallothionine promoter-driven V5 epitope-tagged GPRK2 (pMT-Gprk2) or metallothionine promotor alone (pMT) under induced (+) or non-induced (−) conditions, and probed with V5 antibody (Anti-V5) or GPRK2 antibody (Anti-GPRK2). (C) Diagram of the Gprk2 promoter region in Gprk2^EY09219 and Gprk2^pj1 mutants. Hatched box, Gprk2 exon 1; white boxes, non-duplicated portion of Gprk2 intron 1; white box with arrow, duplicated portion of Gprk2 intron 1; black boxes, P-element inverted repeats; gray box, cg11337 transcribed sequence; gray box with arrow, duplicated sequence in cg11337; bent arrows, start of transcription; thin line, intergenic region; gray line, insertion site of P-element; dashed lines, region deleted in Gprk2^pj1; spotted box, internal P-element sequences; numbers, distances relative to the Gprk2 transcription start.

Figure 2.

GPRK2 is expressed in OSNs. GPRK2 immunostaining in antennae from wild-type (A) and Gprk2^pj1 (B) flies. Scale bars in A and B are 50µm. GPRK2 immunostaining (C), Or22a-driven GFP fluorescence (D) or both GPRK2 immunostaining and Or22a-driven GFP (Merge) detection (E) in an antenna from a Or22a-Gal4;UAS-GFP fly. Scale bars in C-E are 10µm.

Two GPRK2 isoforms are detected in antennae, a higher molecular weight α isoform and a lower molecular weight β isoform (Fig. 1A). The Gprk2 gene is predicted to generate a single mRNA [27], which suggests that the two GPRK2 isoforms are due to post-translational modifications or use of alternate translation starts. The Gprk2 open reading frame, with and without the 5’ and 3’ untranslated sequences, was expressed in S2 cells to determine if either or both produced the α and β isoforms. The open reading frame produced only the α isoform, whereas the open reading frame plus the 5’ and 3’ untranslated sequences produced both the α and β isoforms (Fig. 1A). This result suggests that these isoforms are not due to post-translational modifications, but how the Gprk2 untranslated regions generate the smaller β isoform is not known.

Two P-element insertions in the 5’-UTR of Gprk2, Gprk2⁶⁹³⁶ and Gprk2^EY09213, were identified previously [23, 28]. Reduced levels of the GPRK2 β isoform were observed in both Gprk2 mutants (Fig. 1A), which shows that these mutants are hypomorphs and suggests that the 5’UTR is important for generating the β isoform. To obtain a more severe Gprk2 mutant, we mobilized the P-element of Gprk2^EY09213. One of the resulting mutants, designated Gprk2^pj1, is a complex rearrangement that deletes 193bp of Gprk2 genomic DNA (from −181 to +12nt), and at the site of this 193bp deletion adds an inverted portion of Gprk2 intron 1 (from +1442 to +7782nt) and a 22bp sequence upstream of Gprk2 (from −1644 to −1622) that is flanked by 32bp inverted repeats from the P-element (Fig. 1B). Compared to wild-type flies, little GPRK2 protein was detected in antennae of Gprk2^pj1 mutants either on western blots or by immunostaining (Figs. 1A; Fig. 2A, B), suggesting that Gprk2^pj1 is a strongly hypomorphic allele.

Gprk2 expression is under clock control

To determine whether Gprk2 expression is under circadian clock control, mRNA and protein levels were measured in antennae from flies collected every 4 hours during 12:12 light/dark (LD) conditions. Gprk2 RNA and protein levels oscillate in a diurnal manner with a peak at ZT17 and a trough at ZT1 (Fig. 3A, 3C). The levels of GPRK2 α and β isoforms cycled with a similar phase and amplitude (Fig. 3C). Cyclic Gprk2 expression was abolished in per⁰¹ and tim⁰¹ mutants, where Gprk2 mRNA and protein remained at wild-type trough levels (Fig. 3B, 3D). Likewise, constant low levels of GPRK2 were present in cyc⁰¹ mutant flies (Supplemental Fig. 2). These results demonstrate that Gprk2 mRNA and protein levels are under clock control in antennae.

Figure 3.

Daily rhythms in Gprk2 expression are regulated by the circadian clock. (A) Relative levels of Gprk2 mRNA at the indicated times during an LD cycle. Relative RNA abundance refers to the Gprk2/rp49 RNA ratio, where rp49 serves as a control for RNA content in each sample. The white and black bars indicate times when lights were on and off, respectively. Representative data of three independent experiments is shown. The overall effect of time of day is significant by one-way ANOVA (P<0.0001). (B) Gprk2 RNA levels in per⁰¹ and tim⁰¹ flies at ZT5 and ZT17. Asterisk denotes a significant (P<0.02) difference between wild-type at ZT17 and wild-type at ZT5, per⁰¹ at ZT5 and ZT17, and tim⁰¹ at ZT5 and ZT17. (C) Western blot showing the levels of GPRK2 α and β isoforms and ACTIN at the indicated ZT times in wild-type flies. The levels of GPRK2 α and β isoforms were quantified at the indicated times during an LD cycle. Relative protein abundance refers to the GPRK2/Actin ratio. The GPRK2/Actin values at ZT1, ZT9, ZT13, ZT17 and ZT21 are relative to the value at ZT5, which was set to 1.0. The mean ± SEM is shown for each data point from four independent experiments. The overall effect of time of day is significant for both α and β isoforms by one-way ANOVA (P<0.005). (D) Western blot showing the levels of GPRK2 α and β isoforms and Actin in wild-type, per⁰¹, and tim⁰¹ flies at ZT5 and ZT17. The mean ± SEM is shown for each data point from three independent experiments. Asterisk denotes a significant (P<0.03) difference between wild-type at ZT17 and wild-type at ZT5, per⁰¹ at ZT5 and ZT17, and tim⁰¹ at ZT5 and ZT17.

Cycling of Gprk2 expression is necessary for EAG response rhythms

We previously showed that the circadian clock modulates olfactory responses to ethyl acetate in basiconic sensilla, where robust responses are seen during mid-night and weak responses are seen during mid-day [2]. GPRK2 is expressed in the Or22a positive sensory neuron in large ab3 basiconic sensillae, which respond to ethyl acetate (Fig. 2C–E). Given that Gprk2 expression cycles in antennae, we examined EAG responses in Gprk2⁶⁹³⁶ mutants and flies that overexpress Gprk2 in OSNs (Or83b-GAL4 driven UAS-Gprk2). Flies that overexpress Gprk2 in OSNs displayed constant high amplitude EAG responses that were near the wild-type peak value in basiconic sensilla, whereas Gprk2⁶⁹³⁶ mutant flies showed constant low amplitude EAG responses (Fig. 4A). Overexpression of Gprk2 is sufficient to induce robust EAG responses in cyc⁰¹ flies, indicating that GPRK2 levels determine the magnitude of EAG responses in the absence of a functional clock (Fig. 4B). These results demonstrate that the levels of GPRK2 determine the amplitude of EAG responses, and suggest that clock-dependent rhythms in GPRK2 confer rhythms in EAG responses.

Figure 4.

Gprk2 expression levels control rhythms in EAG responses. (A) Diurnal changes in mean EAG responses are plotted for ethyl acetate (10⁻⁴ dilution) on day 4 of LD. Wild-type (WT), open circles; Gprk2⁶⁹³⁶, closed circles; Gprk2 overexpression (Gprk2 OE), open triangle. Each point represents the mean ± SEM of at least eight females. The white and black bars indicate times when lights were on and off, respectively. The overall effects of time of day, genotype and their interaction is statistically significant (P<0.001) by two-way ANOVA. The asterisk denotes significant (P<0.01) increases in EAG responses at ZT17 in wild-type flies compared to at all other times of day. The cross indicates significant (P<0.01) differences in EAG responses between wild-type flies and the Gprk2⁶⁹³⁶ mutant at the same times of day. The caret indicates significant (P<0.01) differences in EAG responses between wild-type flies and Gprk2 OE flies at the same times of day. Posthoc analysis show no significant differences (P>0.9) as a function of time in either Gprk2 OE flies or in the Gprk2⁶⁹³⁶ mutant. (B) EAG responses in cyc⁰¹ flies and cyc⁰¹ flies overexpressing Gprk2 (Gprk2 OE;cyc⁰¹) at ZT5 and ZT17. Asterisks denote a significant (P<0.0001) differences between cyc⁰¹ at ZT5 or ZT17 and Gprk2 OE;cyc⁰¹ flies at ZT5 and ZT17.

The levels of GPRK2 in Gprk2^pj1 antennae are lower than those in Gprk2⁶⁹³⁶ antennae (Fig. 1A). Given the correspondence between GPRK2 levels and EAG responses, EAG responses should be further reduced in Gprk2^pj1 flies. Indeed, EAG responses in Gprk2^pj1 flies at ZT5 were less than half the amplitude of Gprk2⁶⁹³⁶ or Gprk2^EY09213 flies, and were more than 3-fold lower in amplitude than wild-type flies (Table 1). The low responses in Gprk2^pj1 flies are consistent with this allele retaining a small amount of Gprk2 activity.

Table 1.

Gprk2 levels control the amplitude of EAG repones. EAG reponses to a 10⁻⁴ dilution of ethyl acetate were recorded at ZTS in wild-type (WT), Gprk2^pj1, Gprk2⁶⁹³⁶, and Gprk2^EY09213 flies that were entrained for at least 3 days in LD cycles. The mean EAG reponse ± SEM is shown. Asterisk denotes significant difference (P<0.0001) between Gprk2^pj1 and WT, Gprk2^EY09213 or Gprk2⁶⁹³⁶ EAG values. Cross denotes a significant difference (P<0.0006) between WT and Gprk2^EY09213 or Gprk2⁶⁹³⁶ EAG values.

Strains	EAG reponses(mV)	Number tested
WT	†10.40 ± 0.74	N=8
Gprk2pj1	*2.68 ± 0.19	N=15
Gprk26936	6.87 ± 0.32	N=8
Gprk2EY09213	6.77 ± 0.51	N=8

Cycling odorant receptor accumulation in dendrites is driven by Gprk2

Although many GPRKs function to reduce GPCR dependent signal transduction by inducing endocytosis of GPCRs [19], Drosophila Gprk2 enhances olfactory responses. Such an enhancement of olfactory responses is also seen in a C. elegans Gprk2 mutant, suggesting that GPRK may function through a different mechanism in invertebrate olfactory systems. To assess whether GPRK2 affects the number of ORs in the dendrites, we measured levels of myc-tagged ORs driven by Or83b GAL4 in the dendrites of OSNs. Two myc-tagged ORs, Or7a and Or22a, were expressed in olfactory sensory neurons (Fig. 5). In large basiconic sensilla, high levels of both ORs were detected in dendrites at ZT17, but not ZT5. When these myc-tagged ORs were co-expressed with GPRK2, they accumulated in the dendrites of large basiconic sensilla at ZT5, indicating that GPRK2 mediates their dendrite localization. No staining with anti-Myc antibody was detected in the antennae of flies that contain the Or83b Gal4 driver or the UAS-Or7a and UAS-Or22a responders alone (data not shown). The correspondence between GPRK2 levels and OR localization in basiconic sensillae suggests that rhythms in GPRK2 levels control the rhythmic accumulation of ORs in OSN dendrites in basoconic sensillae.

Figure 5.

GPRK2 levels control the accumulation of ORs in OSN dendrites. Myc antibody was used to detect Or83b-Gal4 driven UAS-myc-tagged Or7a in wild-type antennae at ZT5 (A) and ZT17 (B) and in antennae from Gprk2 overexpression (Gprk2 OE) flies at ZT5 (C), and to detect Or83b-Gal4 driven UAS-myc-tagged Or22a in wild-type antennae at ZT5 (D) and ZT17 (E) and in antennae from Gprk2 overexpression (Gprk2 OE) flies at ZT5 (F). Each sample was co-immunostained with ELAV, a non-rhythmically expressed nuclear antigen that serves as a reference to quantify myc-OR levels. The images are representative data from 7–9 antennae for each genotype and time point. Arrows, large basiconic sensillae. Scale bar is 5.0µm. (G) Quantification of myc-ORs from 15–20 OSNs for each genotype and timepoint. Relative myc-tagged OR levels were calculated as the ratio of myc immunofluoresence intensity in the dendrite to ELAV immunofluoresence intensity in the nucleus (myc dendrite/ELAV). Error bars denote ±SEM. Asterisks denote a significant (P<0.0001) difference in myc dendrite/ELAV ratio of Or7a and Or22a immunostaining between wild-type (WT) at ZT5 and Gprk2 overexpression (OE) at ZT5 or wild-type at ZT17.

Discussion

Cycling of Gprk2 expression

We demonstrate that Gprk2 is a circadian output gene whose mRNA and protein peak during the middle of the night in antennae. This phase of mRNA expression is similar to that of per, tim and other genes driven directly by CLK-CYC binding to E-box regulatory sequences [29–31]. However, CLK-CYC-dependent genes are expressed at constitutively high levels in per⁰¹ and tim⁰¹ mutants and constitutively low levels in Clk^Jrk and cyc⁰¹ mutants [30, 32, 33], whereas Gprk2 is expressed at low levels in per⁰¹, tim⁰¹, and cyc⁰¹ mutants (Fig. 3D; Supplemental Fig. 2). Several rhythmically expressed transcripts identified by microarray analysis of heads have low levels of expression in per⁰¹ and Clk^Jrk mutants [34, 35], but the mechanism governing their rhythmic expression has not been explored.

Analysis of arrhythmic clock mutants indicate that cycling levels of Gprk2 mRNA give rise to rhythms in GPRK2 protein. The levels of GPRK2 protein cycle in phase with Gprk2 mRNA, and correspond to rhythms in EAG rhythms. Other kinases such as DOUBLE-TIME (DBT), Shaggy (SGG)/GSK3, and Casein kinase 2 (CK2) in Drosophila are constitutively expressed proteins [36–40], whereas Gprk2 is the first example of a rhythmically expressed kinase. However, other kinases such as Erk-MAP kinase and Calcium/calmodulin-dependent protein kinase II in the chicken retina are rhythmically activated due to phosphorylation, and control cGMP-gated ion channels in cone photoreceptors [41].

Gprk2 dependent rhythms in EAG responses

Cycling levels of GPRK2 are coincident with rhythms in EAG responses; GPRK2 levels and EAG responses peak around mid-night and are at their lowest levels during mid-day. When GPRK2 levels are constitutively low, as in per⁰¹, tim⁰¹, and cyc⁰¹ mutants, EAG responses are also low. In addition, levels of GPRK2 are at or below the normal wild-type trough in Gprk2⁶⁹³⁶ and Gprk2^EY09213 mutants (Fig. 1A), and generate EAG responses that are at or below those at the wild-type trough (Fig. 4; Table 1). GPRK2 levels are barely detectable in Gprk2^pj1 antennae (Fig. 1A, 2A), and produce weak EAG responses well below the wild-type trough level (Table 1). These results suggest that Gprk2^pj1 is not a null allele, and raises the possibility that a Gprk2 null mutant will lack EAG responses altogether. Such a result would demonstrate that Gprk2 is required for olfactory responses per se. In contrast, constitutive overexpression of Gprk2 produces constant high EAG responses in both wild-type and cyc⁰¹ flies, demonstrating that high levels of GPRK2 can effect high amplitude EAG responses independent of other clock-dependent factors. Taken together, these results argue that Gprk2 levels control the amplitude of EAG responses. If so, this would imply that low levels of GPRK2 present in the Gprk2⁶⁹³⁶ mutant do not cycle in abundance.

Given that GPRK2 levels regulate the amplitude of EAG responses, what is the mechanism through which Gprk2 controls EAG response amplitude? The traditional targets of GPRKs are GPCRs [19]. In the mammalian olfactory system, GPRK3 desensitizes ORs by triggering their internalization [25]. Our results suggest that Drosophila Gprk2 is necessary for EAG responses, and taken together with C. elegans Gprk2 function, they indicate that Gprks play a different role in invertebrate olfaction than in vertebrate olfaction. The subcellular localization of ORs is high in dendrites of basiconic sensilla at ZT17 and low at ZT5, but the abundance of ORs in these dendrites at ZT5 can be driven to high levels by increasing GPRK2 expression (Fig. 5). These results support a model in which the circadian clock generates a rhythm of Gprk2 expression, which in turn generates rhythms in the amplitude of EAG responses by promoting OR accumulation (and consequently odor-gated cation channel formation) in OSN dendrites from basiconic sensillae. Gprk2-dependent rhythms in the amplitude of spontaneous spikes are also seen in OSNs [42], thus demonstrating that the clock controls basic (i.e. odor-independent) properties of the OSN membrane. It is possible that the rhythmic localization of odor-gated cation channels to OSN dendrites accounts for rhythms in the amplitude of spontaneous spikes. Our results can’t exclude the possibility that cyclic expression of other genes also contribute to rhythms in EAG responses. We have tested mRNA cycling for several genes that could potentially modulate EAG responses including arrestin 2, G-protein coupled receptor kinase 1 (Gprk1), and kurtz arrestin [20, 43, 44], and find that arrestin 2 mRNA levels cycle, but neither Gprk1 or kurtz arrestin mRNA levels cycle (data not shown). Given that microarray analysis was done on fly heads depleted of antennae, microarray analysis of antennae may reveal other rhythmically expressed genes that contribute to EAG rhythms.

Myc-tagged ORs did not accumulate to high levels in the dendrites of trichoid sensilla at ZT17 (Fig. 5). Trichoid sensilla have different functions than basiconic sensilla; T1 trichoid sensilla detect the pheromone 11-cis-vaccenyl acetate (cVA) whereas the basiconic sensilla recognize food and plant odors [45, 46]. It could be that the circadian clock regulates OSN activity differently in basiconic sensilla and trichoid sensilla, although we can not exclude the possibility that detection of myc-tagged ORs in dendrites failed due to low expression levels in trichoid sensillae, poor permeability of anti-myc antibody into trichoid sensilla, or the long, thin geometry of trichoid sensillae.

In summary, we show here that Drosophila Gprk2 mRNA and protein expression is under clock-control in antennae. The levels of GPRK2 protein determine the amplitude of EAG responses to ethyl acetate in basiconic sensillae; high levels generate high amplitude EAGs and low levels produce low amplitude EAGs. This result suggests that GPRK2 directly or indirectly enhances OR activity, in contrast to the inhibition of olfactory signaling by Gprk3 in mammals. Given that the most severe Drosophila Gprk2 mutant still produces low amplitude EAG responses, a complete loss of Gprk2 function may lack EAG responses altogether and be required for olfaction. High levels of GPRK2 enhance OR localization to dendrites of basiconic sensillae, and support a model in which rhythms in GPRK2 levels drive rhythms in OR localization to dendrites that ultimately mediates rhythms in EAG responses.

Materials and Methods

Fly strains and P-element excisions

The wild-type strain was Canton-S. Mutant and transgenic strains used in these studies include per⁰¹ [47], tim⁰¹ [48], cyc⁰¹ [33], Or83b-Gal4 [49], Or22a-Gal4 [50], UAS-GFP [51], UAS-Myc-Or22a [50], and UAS-Myc-Or7a [52]. Two P-element inserts in exon 1 of Gprk2, P{PZ}Gprk2⁰⁶⁹³⁶ (Gprk2⁶⁹³⁶) and P{EPgy2}CG11337^EY09213 (Gprk2^EY09213), are hypomorphic mutants [27]. To generate additional Gprk2 mutants via imprecise excision, P{EPgy2}CG11337^EY09213, a w⁺ transposable element insert in the 5’UTR of Gprk2, was excised as described [53]. A total of 22 w⁻ excision lines were produced including Gprk2^pj1, which contains a rearrangement that causes a more severe loss of Gprk2 function (see text).

Transgenic flies

To generate UAS-Gprk2 transgenic flies, Gprk2 EST LE65371 (coding region and 5’ and 3’ UTRs) was first digested with Bam HI and ligated into pBlueScript SK(+). The resulting pBS-Gprk2 plasmid was digested with Eco RI and Xba I and subcloned into the pUAST transformation vector [54], and used to generate transgenic animals at the Duke University Non-mammalian Model Systems Flyshop. Transgenic UAS-Gprk2 flies were balanced using w; Sco/CyO; TM2/TM6b.

Antibody generation and western blotting

A C-terminal fragment of Gprk2 (Met563-Ser714) was amplified from cDNA via PCR (5’ primer: 5’-AATTAATCATATGCTGGAGCCACCCT TTGTG 3’; 3’ primer: 5'-CAGGAAACAGCTATGAC 3’). The PCR amplification products were digested with Nde I and Xho I, subcloned into pET28b (EMD Biosciences, Inc.), and expressed as a histidine tagged protein in the E. coli BL21 (DE3) after induction using 0.3mM IPTG. His-tagged GPRK2 protein was purified using nickel column chromatography (Qiagen Inc.) according to the manufacturers instructions. The protein preparation was then used for antibody production in Guinea pigs (Cocalico Biologicals, Inc.).

For western blots, 200–300 3rd antennal segments were dissected from flies entrained to at least three 12h light: 12h dark (LD) cycles and homogenized in protein extraction buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.3% Triton X-100, 1µg/ml Leupeptin, 1µg/ml Aprotinin, 1 mM PMSF). Western blots were prepared using 400ng of protein for each sample as described [55]. Western blots were probed with anti-GPRK2 diluted 1:5000, anti-V5 diluted 1:5000, or anti-actin (Sigma-Aldrich) diluted 1:200, and visualized using ECL (Amersham). Western blots were repeated at least three times with independent samples. The levels of GPRK2 α and β isoforms were quantified relative to the α isoform at ZT5, which was set to 1.0.

Immunostaining

Flies were collected after entrainment to at least three LD cycles. Consecutive 14µm cryosections were collected, fixed in 4% formaldehyde in PBST (0.1% Triton X-100 in PBS) and blocked in 10% goat serum for 1-hr as described [50]. Cryosections were incubated overnight with primary antibody at 4°C, rinsed with PBST, incubated for 3hr with secondary antibody at room temperature, rinsed with PBST, and mounted on a glass slide with Vectashield (Vector Laboratories, Inc.) mounting medium. For cryosections co-immunostained with ELAV antibody, after the first primary and secondary antibodies were incubated and washed as above, ELAV antibody was incubated for 3hr at room temperature, rinsed with PBST, incubated for 3hr with secondary antibody at room temperature, and rinsed with PBST before mounting. Specimens were imaged with an Olympus FV1000 confocal microscope. The primary antibodies used were rat anti-ELAV (Developmental Studies Hybridoma Bank) diluted 1:200, mouse anti-MYC (Sigma-Aldrich) diluted 1:500, and Guinea pig anti-GPRK2 diluted 1:500. The secondary antibodies used were Cy3-conjugated anti-mouse, Cy3-conjugated anti-rat, Cy3-conjugated anti-Guinea pig, Alexa 488-conjugated anti-rat, and Alexa 488-conjugated anti-Guinea diluted 1:200. Levels of GPRK2 were quantified as the ratio of GPRK2 signal intensity in the cell body to ELAV intensity in the nucleus in 15–20 OSNs using Image J (NIH). Levels of Myc-tagged ORs were quantified as the ratio of Myc signal intensity in the dendrite to ELAV intensity in the nucleus in15–20 OSNs using Image J (NIH).

qPCR

Flies (0–5 days old) were entrained for at least three LD cycles at 25 °C and collected during LD. For each collection, RNA was purified from 150–200 dissected 3^rd antennal segments and used to generate cDNA for qPCR analysis as described [13]. The following probe and primers for detecting Gprk2 mRNA were designed using ABI primer Express software: 5’ primer, TGCTGGAGCCACCCTTTG, 3’ primer, CGAGCACATCTTTGGCGTAA; probe CCAGACCCGCACGC. The Gprk2 probe was labeled with FAM (6-carboxyfluorescein) and 3’ labeled with TAMRA (6-carboxytetramethylrhodamine). To eliminate the possibility of contaminating genomic DNA amplification, the probe sequence was designed to across exon junctions in the cDNA sequence.

S2 cell experiments

To inducibly express V5 epitope-tagged GPRK2 in S2 cells, the ORF of Gprk2 from the UAS-Gprk2 vector was amplified via PCR with the Gprk2 forward (5’ GGTCGGAATTCATGGAATAGAGAAT 3’) and reverse (5’ GCTTACTCGAGCTTTCGACCGTCGTG 3’) primers, digested with Eco RI and Xho I, and inserted into the pMT/V5-His B vector (Invitrogen, Carlsbad, CA) to generate pMT-Gprk2. To constitutively express GPRK2 from the Gprk2 ORF in S2 cells, the UAS-Gprk2 vector was amplified via PCR with the same Gprk2 forward and reverse primers listed above, digested with Eco RI and Xba I, and inserted into the pAc5.1/V5-His B vector (Invitrogen, Carlsbad, CA) to generate pAc-Gprk2. To constitutively express GPRK2 from the Gprk2 ORF + 5’ and 3’ UTRs in S2 cells, the Eco RI-Xba I fragment from UAS-Gprk2 was subcloned into pAc5.1/V5-His B to generate pAc-Gprk2+UTRs. All Gprk2 constructs were verified by DNA sequencing. 500ng of pAc-Gprk2 or pAc-Gprk2+UTR plasmids were transfected into S2 cells and cultured as described (13). 500ng of either pMT-Gprk2 or pMT vector alone was transfected into S2 cells and induced by the addition of CuSO₄ as described [56]. Transfected cells were harvested via centrifugation (3000 × g, 5 min, 4 °C), homogenized in protein extraction buffer (50mM Tris-HCl (pH7.5), 150mM NaCl, 1mM EDTA (pH 8.0), 1mM PMSF, 1mg/ml Leupeptin, 1 mg/ml Pepstatin), and used for Western blot analysis.

EAG measurements

EAG measurements were performed as described [57]. A 10⁻⁴ dilution of ethyl acetate was used in all EAG experiments.

Sequence comparisons

Amino acid sequence comparisons between Drosophila GPRK2 and human GPRKs or C. elegans GPRK2 were performed using Clustal W analysis. Sequence identity was determined within the GPRK N-terminal domain, kinase doman and C-terminal domain.

Statistical analysis

Statistical analysis was done using MS-EXCEL (Microsoft) and Statistica (Statsoft). ANOVA analysis was done using Statistica and MS-EXCEL. Posthoc comparisons were done using Scheffe’s test (α = 0.05). Student’s T-test was used to compare values at peak and trough time points.