Drosophila DBT Autophosphorylation of Its C-Terminal Domain Antagonized by SPAG and Involved in UV-Induced Apoptosis

Abstract

Drosophila DBT and vertebrate CKIε/δ phosphorylate the period protein (PER) to produce circadian rhythms. While the C termini of these orthologs are not conserved in amino acid sequence, they inhibit activity and become autophosphorylated in the fly and vertebrate kinases. Here, sites of C-terminal autophosphorylation were identified by mass spectrometry and analysis of DBT truncations. Mutation of 6 serines and threonines in the C terminus (DBT^C/ala) prevented autophosphorylation-dependent DBT turnover and electrophoretic mobility shifts in S2 cells. Unlike the effect of autophosphorylation on CKIδ, DBT autophosphorylation in S2 cells did not reduce its in vitro activity. Moreover, overexpression of DBT^C/ala did not affect circadian behavior differently from wild-type DBT (DBT^WT), and neither exhibited daily electrophoretic mobility shifts, suggesting that DBT autophosphorylation is not required for clock function. While DBT^WT protected S2 cells and larvae from UV-induced apoptosis and was phosphorylated and degraded by the proteasome, DBT^C/ala did not protect and was not degraded. Finally, we show that the HSP-90 cochaperone spaghetti protein (SPAG) antagonizes DBT autophosphorylation in S2 cells. These results suggest that DBT autophosphorylation regulates cell death and suggest a potential mechanism by which the circadian clock might affect apoptosis.

INTRODUCTION

The circadian clock produces daily changes in a wide range of physiological activities, as exemplified by the sleep-wake cycle, and is also essential for seasonal changes in response to changing photoperiods. It is found in single-cell organisms such as cyanobacteria as well as multicellular organisms such as humans (1). The disruption of the clock can cause many health problems, including metabolic disease, sleep disorders, or even cancers in humans (2), so it is important to understand its mechanism from both basic and clinical perspectives.

Circadian clocks are the result of oscillations of several circadian clock proteins, including those of the period protein (PER) (3). In flies and humans, the casein kinase I ortholog (called the doubletime or dbt protein in flies) is essential for the oscillations of PER because it phosphorylates PER during the day and early evening to cause PER degradation (4–8). During the late night, DBT phosphorylation of PER is reduced, and PER accumulates in the nucleus as a consequence of its interaction with the timeless protein (TIM), which antagonizes phosphorylation of PER by DBT to confer rhythmic regulation of the per and tim genes (9, 10). Hence, the rhythmic phosphorylation of PER by of DBT is essential for the rhythmic accumulation of PER protein and transcriptional feedback that underlie the circadian clock (4–7).

In such a mechanism, anything that confers temporal regulation to DBT activity would contribute to the oscillations. Clearly, one such regulator is the timeless protein (TIM), which is degraded in response to light (11–13) but whose accumulation at night leads to a PER/TIM heterodimer in which DBT activity is antagonized (10, 14). Recently, another regulator has been shown to be a noncanonical FK506-binding protein (BDBT) that binds with DBT and increases its activity toward PER (15). Another possible regulator of DBT activity is phosphorylation of DBT itself, which is elevated in flies with reduced BDBT activity (15). It was first demonstrated that mammalian CKIδ autophosphorylates its C-terminal domain to inhibit its activity (16), while a series of mutants with mutations in the C-terminal domain of CKIε were generated to identify specific phosphorylation residues that mediate inhibition (17). Others have proposed a model in which the phosphorylated C-terminal domain physically interacts with the catalytic domain to inhibit the kinase activity (18).

From the evolutionary standpoint, the catalytic domains of DBT and CKIδ/ε are highly conserved. They are over 86% identical in amino acid sequence in their N-terminal domains (7, 19, 20). While the noncatalytic C-terminal regions show no sequence homology, we have recently shown that the C-terminal domain inhibits DBT kinase activity and that DBT is autophosphorylated (21), suggesting that the C-terminal domain of DBT may regulate DBT in the same manner that vertebrate CKIδ/ε are regulated.

Here, we demonstrate phosphorylation of the DBT C terminus and analyze its biological function. The C terminus of DBT exhibited progressive phosphorylation when a phosphatase inhibitor was added. Catalytically inactive forms of DBT and C-terminally truncated forms did not exhibit this phosphorylation, demonstrating that that it involves autophosphorylation of the C terminus (as is the case with vertebrate CKIε/δ), and mutation of 6 C-terminal serines and threonines, including one shown to be phosphorylated by mass spectrometry (MS), greatly reduced autophosphorylation-induced DBT electrophoretic mobility shifts. However, unlike the case for the vertebrate kinases, Drosophila DBT kinase activity was not inhibited by the autophosphorylation in S2 cells. When expressed in fly circadian cells, neither wild-type DBT (DBT^WT) nor the mutant DBT that does not autophosphorylate extensively shows strong daily changes in phosphorylation state, and neither strongly affects the behavioral rhythms, suggesting that the circadian clock can prevent DBT autophosphorylation.

Hence, we analyzed DBT's effects on apoptosis (22, 23) to determine if these might be affected by autophosphorylation. In S2 cells, wild-type DBT overexpression protected cells from apoptosis while becoming degraded, and the mutant of DBT which cannot be extensively autophosphorylated in its C-terminal domain was not protective or degraded. Finally, the Hsp-90 cochaperone spaghetti protein (SPAG) (24) was shown to antagonize DBT autophosphorylation in S2 cells. Our results suggest that the circadian clock does not require DBT autophosphorylation to produce circadian rhythms but that elevated phosphorylation of DBT does have potential consequences for other processes such as apoptosis that are affected by DBT autophosphorylation.

MATERIALS AND METHODS

Fly strains.

The wild-type strain was the Canton S strain. The timGAL4 and ActinGAL4 driver lines were obtained from the Bloomington stock center. Generation of the UAS-DBT^WT has been previously described (5), while generation of the UAS-DBT^C/ala is described below.

S2 cell transfection and expression analysis.

pMT-DBT^WT-MYC-HIS, -DBT^K/R-MYC-HIS, and -CKI-V5 and pMT expressing various DBTs with deletions of the DBT C terminus (296, 332, and 387 amino acids remaining) have been described previously (5, 21). Mutations of 6 serines/threonines in the DBT C terminus to alanine were generated by two successive site-directed mutageneses using a QuikChange kit (Stratagene, CA), with the following primers for rounds I and II: DBTC/alaIF, CAACATGGACGACGCGATGGCGGCCGCCAACGCGGCGAGACCGCCGTAC; DBTC/alaIR:, GTACGGCGGTCTCGCCGCGTTGGCGGCCGCCATCGCGTCGTCCATGTTG; DBTC/alaIIF, CCGCCGTACGACGCGCCGGAGCGTCGGCCTGCGATACGGATGCGG; and DBTC/alaIIR, CCGCATCCGTATCGCAGGCCGACGCTCCGGCGCGTCGTACGGCGG.

Constructs containing all 6 mutations were identified by DNA sequencing. The constructs were cotransfected with a puromycin resistance plasmid into Drosophila S2 cells and stable transformants selected in the presence of puromycin as described by the supplier (Invitrogen, CA). Expression of DBT with CuSO₄ induction and okadaic acid treatment with or without MG132 (50 μM) were performed as previously described (5, 21). Cells were either lysed directly in SDS buffer for SDS-PAGE analysis or processed for immunoprecipitation as described below.

Immunoprecipitation of DBT or CKIδ.

S2 cells were collected, rinsed with phosphate-buffered saline (PBS), homogenized with 1% NP-40, 40 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM leupeptin, 1 mM aprotinin, and 1 mM pepstatin A, and centrifuged at 10,000 rpm for 10 min, after which the supernatants were collected. These were nutated with gamma-bind Sepharose beads for an hour, and after a brief centrifugation the supernatants were then incubated with anti-MYC antibody (1:150) (Covance, PA) or anti-V5 antibody (Invitrogen, CA) and gamma-bind Sepharose for 2 h, followed by brief centrifugation, three washes with 0.1% NP-40–50 mM Tris-HCl (pH 8.0)–150 mM NaCl, and a final rinse with 50 mM HEPES (pH 7.5). The beads were resuspended in a 50% slurry and either used immediately for assays or stored at −80°C with 5% glycerol in 50 mM HEPES (pH 7.5).

Phosphatase treatment.

One aliquot of the immunoprecipitated or cell lysate sample was treated with λ phosphatase (New England Biolabs, MA). The reaction was done according to the manufacturer's instructions, and the mixture was incubated in a 30°C water bath for 2 h and inactivated at 65°C for 30 min with 1 mM EDTA. For S2 cell lysates, the phosphatase was added by itself or with 1.5 mM Na₃VO₄ and 1.5 mM NaF for the reactions that received inhibitor. Subsequently, the sample was resuspended in a 1× Laemmli SDS buffer, followed by SDS-PAGE analysis.

Kinase assay.

The DBT or CKI samples purified by immunoprecipitation were subjected to a kinase activity assay using the reaction conditions described previously (5, 6) with either dephosphorylated casein (Sigma, MO), HIS-tagged PER isolated from bacteria by Ni-agarose chromatography (5), or no added protein substrate and [γ-³²P]ATP (NEN, MA). The reaction mixtures were incubated at room temperature for 30 min, and reactions were terminated by the addition of SDS-PAGE loading buffer; pilot studies indicated that the incorporation of ³²P was linear for times longer than 30 min. Half of the sample was subjected to immunoblot analysis for the detection of the DBT C terminus or the V5 epitope, while the other half was subjected to SDS-PAGE and autoradiography for the detection of PER, casein, or DBT phosphorylation. The immunoblot signals for DBT or CKI were quantified by chemifluorescence as previously described (5, 6), while phosphorylation was quantified by phosphorimager analysis (Molecular Dynamics, CA). Phosphorylation signals were normalized to DBT amounts as described in the Fig. 3 legend.

FIG 3.

Autophosphorylation of DBT's C terminus does not reduce its kinase activity. MYC-tagged Drosophila DBT (DBT^WT or DBT^K/R) or V5-tagged Xenopus CKIδ was induced with CuSO₄ in S2 cells transfected with the appropriate transgene in the presence of the indicated concentration of okadaic acid (OA) and then immunoprecipitated from S2 cell lysates with an antibody to MYC epitope tag or the V5 epitope tag. The immunoprecipitated kinase was incubated with the indicated substrates for 30 min at room temperature (pilot studies indicated that the incorporation of ³²P was linear for times longer than 30 min), and the reaction products were then subjected to SDS-PAGE for autoradiography (lower panels) (to detect phosphorylation of the substrate) or immunoblot analysis (upper panels) (to quantify the amount and mobility of the kinase with anti-DBTC or anti-V5). (A) Phosphorylation of casein by DBT^WT-MYC-HIS and not by DBT^K/R-MYC-HIS (establishing that the activity arises from DBT and not a contaminant). (B) Phosphorylation of recombinant PER expressed and purified from E. coli by DBT^WT and not by DBT^K/R. (C) Phosphorylation of casein by Xenopus CKIδ from S2 cells not treated with okadaic acid but not from cells treated with okadaic acid (+OA) (60 nM); 1.6 times the amount of kinase was used in lane 2 as in lane 1, and the “no tag” lane shows a reaction with beads from S2 cells that expressed an untagged DBT. Note the electrophoretic mobility shift for CKIδ produced by OA and the lack of phosphorylation of casein by the OA-treated kinase. (D) The amount of recovered DBT-MYC-HIS in an anti-MYC immunoprecipitate from Drosophila S2 cells was quantified with anti-DBTC, while the amount of ³²P incorporation into casein (top panel) or DBT itself without added exogenous substrate (bottom panel) was detected with autoradiography. Mutant DBT^K/R protein did not incorporate any ³²P into casein or itself, while DBT^C/ala incorporated comparable amounts of ³²P into casein but reduced levels into itself, consistent with the loss of autophosphorylation targets due to the mutations. (E) The amount of ³²P incorporation into casein was quantified by phosphorimager analysis for three OA-treated (+OA) or untreated (−OA) DBT^WT-MYC-HIS-containing reaction mixtures, and the amount of DBT^WT-MYC in each reaction mixture was quantified by chemifluorescence scans of immunoblots probed with anti-DBTC. The amount of ³²P incorporation was normalized to the amount of DBT detection, and each individual reaction was further normalized to the DBT-normalized activity for the reaction without OA in each experiment (so the value without OA is one). The amount of ³²P incorporation produced by OA-treated DBT^WT was significantly higher than the incorporation produced by untreated DBT (star, P < 0.05 by a t test for single mean, tested against a value without OA of 1; t value = 6.4).

MS analysis.

Drosophila S2 cells that were stably transfected with pMT-DBT^WT-MYC-HIS were harvested and centrifuged at 1,000 × g for 3 min. Cell pellets were stored at −80°C for at least 1 day and were then resuspended in lysis buffer (200 mM Tris-Cl [pH 8], 0.4 M ammonia sulfate, 10 mM MgCl₂, 10% glycerol, and 20 mM NaF plus protease inhibitors) and incubated on ice for 30 min, followed by sonication. DBT was purified using Talin metal affinity resin (Clontech), followed by anti-MYC chromatography (9E10 monoclonal antibody, affinity matrix from Covance). Purified DBT was eluted with Laemmli SDS loading buffer, separated by 10% SDS-PAGE, and Coomassie blue stained. Excised Coomassie blue-stained DBT gel bands were reduced, alkylated, and subjected to in-gel trypsin digestion by standard methods, and extracted peptides were analyzed by nanoLC-tandem mass spectrometry (MS) as described previously (25, 26). DBT peptide identifications were made using Mascot 2.4 (Matrix Science, Ltd.) or Proteome Discoverer (Thermo Fisher) with trypsin specificity and using the doubletime sequence only as a database (UniProtKB accession O76324), considering differential modification (phosphorylation) of Ser/Thr residues.

Transgenic fly and circadian analysis.

The mutated dbt^C/ala described above was subcloned from pMT into a PUAST-attB vector with EcoRI and PmeI, and the DNAs were then injected into a w⁴⁴⁸ strain with the attP2 integration site by the Duke Model Systems Center. The injected flies were crossed to flies containing the IIIrd chromosome balancer TM3SbSer, since the attP2 site is located on the III chromosome, and several lines were established. The transgenic flies were later mated with the circadian clock-specific expression driver timGAL4. After at least 3 days of entrainment to a 12-h/12-h light-dark (LD) cycle, the flies were released into constant darkness (DD) and locomotor activity monitored and analyzed as previously described (15, 27). Heads were collected at the indicated times in a light-dark cycle as previously described (5, 15).

Immunoblot analysis of DBT levels.

An aliquot of S2 cell extract, immunoprecipitate, kinase assay product, or fly head homogenate suspended in 1× Laemmli SDS loading buffer was subjected to SDS-PAGE analysis on a 10% acrylamide gel, transferred onto a nitrocellulose membrane (Whatman), incubated with rabbit anti-DBT-C antibody (1:2,000), anti-MYC (Invitrogen, CA), anti-V5, antitubulin, or antiactin (Hybridoma Bank, IA), and detected by chemiluminescence or chemifluorescence with ECL Plus reagent (GE Healthcare, MA) as previously described (5). Immunoblots are representative of those observed in 2 or 3 experiments.

Assays of S2 cell viability and DBT expression after UV treatment.

Drosophila S2 cells with no DNA (“mock”) or stably transfected with either pMT-DBT^WT-MYC-HIS or pMT-DBT^C/ala-MYC-HIS were plated at a density of 1 × 10⁶ cells per 35-mm well and allowed to attach for at least 2 h. S2 cells were then washed with Schneider's medium supplemented with 10% fetal bovine serum (FBS) and maintained in 2 ml of medium, at which point cells were induced for DBT expression using 0.5 mM copper sulfate. Cells were induced for 48 h, followed by UV irradiation of cells for 10 min using a UV transilluminator (dual-intensity transilluminator, model TM-15 [UVP, Upland, CA]). After UV treatment, cells were incubated for 24 h, and cell viability was determined using a hemocytometer to count unlysed cells. The numbers were normalized, with the value for mock-treated cells (no UV) set at 100% viability.

To assay expression of DBT, S2 cells were harvested at either 4 or 18 h after UV treatment or at the same time from cells not treated with UV in SDS loading buffer. Samples were run on a 10% SDS-polyacrylamide gel and assayed by immunoblotting as described above for DBT and actin.

To confirm proteasome involvement in the UV-induced degradation of DBT, S2 cells that expressed DBT (induced for 48 h) were treated with or without the proteasome inhibitor MG132 (50 μM) from 1 h before UV treatment to the end of the experiment and were harvested 18 h after UV treatment. Samples were run on a 10% SDS-polyacrylamide protein gel, in some cases after treatment with lambda phosphatase as described above, and probed with anti-DBT-C antibody as described above.

For the larval viability experiments, the indicated transgenic DBT was expressed with the timGAL4 or actinGAL4 driver, and third-instar larvae were subjected to UV irradiation (UV irradiation of larvae for 10 min in 50-mm plates without fly food using the UV transilluminator noted above) and harvested at the indicated times after UV irradiation. DBT-MYC was detected by immunoblot analysis with anti-MYC antibody and actin with antiactin. The percentage of larvae that survived until pupae was tabulated as viability for each genotype. Almost no larvae emerged as adults for any of the genotypes.

RNA interference (RNAi) protocol.

Drosophila S2 cells were plated at a density of 1 × 10⁶ cells per well and allowed to adhere for at least 2 h. S2 cells were then washed 2 times with serum-free Schneider's medium, followed by addition of 1 ml of serum-free medium to the cells. spaghetti double-stranded RNA (dsRNA) was prepared from PCR fragments carrying spag nucleotides (nt) 735 to 1242 with T7 promoters at each end, and it was then transfected into S2 cells by the procedures of the Perrimon lab (http://www.flyrnai.org/DRSC-HOME.html). The primers were as follows: forward, 5′-TAATACGACTCACTATAGGGCAAAAGTGGGCCAAACTTTAC-3′; reverse, 5′TAATACGACTCACTATAGGGTTCTGGGCTGCGTTCTAT-3′. dsRNA (80 μg per well) was added to the cells and shaken vigorously to mix. Cells were incubated without serum for 5.5 h, at which point 10% serum was added to the cells. For the experiment involving DBT-MYC, DBT was induced with CuSO₄ for 48 h, followed by spag dsRNA addition. At the indicated times, cells were harvested in their own medium and centrifuged at 13,000 rpm for 5 min, and the pellet was resuspended in SDS loading buffer and analyzed by immunoblot analysis as described above.

RESULTS

Wild-type DBT autophosphorylates its C-terminal domain.

Drosophila DBT and vertebrate CKIε and -δ are highly conserved in their N-terminal domains but completely divergent in their C-terminal domains (7, 19). It has been shown that both vertebrate CKIδ and CKIε can autophosphorylate their C-terminal domains when cellular phosphatases are inhibited, leading to inhibition of their kinase activity (16, 17). Previously, we showed that Drosophila DBT has autophosphorylation activity (21), which is characterized further here. DBT expressed from the endogenous genomic locus in Drosophila S2 cells exhibits retarded electrophoretic mobility when the cells are treated with the phosphatase inhibitor okadaic acid (Fig. 1A), as does MYC-tagged DBT expressed from a transgene (Fig. 1B). In order to demonstrate that the mobility shift is due to phosphorylation, immunoprecipitated DBT-MYC proteins were treated with λ phosphatase, which converts the slow-mobility protein to the fast form, so the mobility shift is indeed due to phosphorylation of DBT (Fig. 1B). The lack of a mobility shift for a catalytically inactive form of DBT (Fig. 1B, DBT^K/R-MYC) shows that the phosphorylation is autophosphorylation (produced by the intrinsic kinase activity of DBT).

FIG 1.

Inhibition of cellular phosphatases with okadaic acid induces autophosphorylation of DBT in its C-terminal domain. Unmodified S2 cells or S2 cells expressing the indicated MYC-tagged DBT were treated 41 h after induction by CuSO₄ with the indicated concentration of okadaic acid for 17 h and harvested for immunoblot analysis. (A) DBT expressed from the endogenous S2 cell genome (ENDO-DBT) and detected with an anti-DBTC antibody exhibits a range of retarded electrophoretic mobilities after treatment with okadaic acid (OA). Tubulin was detected as a loading control. (B) DBT^WT-MYC-HIS or catalytically inactive DBT^K/R-MYC-HIS was induced in S2 cells, treated with the indicated concentration of okadaic acid, and then immunoprecipitated from cell lysates with anti-MYC antibody. The immunoprecipitates were treated with λ phosphatase that was then washed from the beads, and the bound DBT proteins were subjected to SDS-PAGE and detected with an anti-DBTC antibody. DBT^WT produces an electrophoretic mobility shift that is eliminated by treatment with phosphatase, establishing that it is due to phosphorylation. Catalytically inactive DBT^K/R does not produce any mobility shift, establishing that the phosphorylation arises from DBT catalytic activity (autophosphorylation). (C) S2 cells were transfected with plasmids expressing DBT^WT-MYC-HIS, DBT^K/R-MYC-HIS, or DBT mutants truncated at the indicated amino acid. DBT^WT is 440 amino acids long, and the region of homology to vertebrate CKIδ/ε ends at amino acid 296. The cells were treated with the indicated concentrations of okadaic acid, and DBT was detected with anti-MYC antibody. Only DBT^WT produces a stoichiometric mobility shift in response to okadaic acid treatment, demonstrating that the target sites lie mostly in amino acids C terminal to 387. The mobilities of phosphorylated full-length DBT (p-DBT^WT-MYC) or of phosphorylated DBT-387 MYC (p-387-DBT-MYC) (a minor isoform) are indicated by arrows. Tubulin (Tub) was detected as a loading control. (D) MG132 treatment of S2 cells does not elevate detection of phosphorylated DBT in OA-treated cells. pMT-DBT^WT-MYC-HIS, pMT-DBT^K/R-MYC-HIS, pMT-DBT^C/ala-MYC-HIS (see Fig. 2 for description of this construct), and pMT-387DBT-MYC-HIS were transfected into S2 cells, induced with CuSO₄, and then treated without or with 60 nM okadaic acid (OA) for 17 h in the absence or presence of 50 μM MG132. Cells were then collected and resuspended in 1× SDS buffer for electrophoresis and immunoblotting with anti-MYC antibody or antitubulin as a loading control.

In order to determine whether the C-terminal domain of DBT is the target for autophosphorylation, the electrophoretic mobilities of various C-terminally truncated mutants of DBT were assayed in Drosophila S2 cells. While full-length wild-type DBT-MYC (DBT^WT) (440 amino acids) exhibited robust mobility shifts in the presence of okadaic acid, none of the mutant proteins truncated within the nonconserved C-terminal domain at the indicated amino acids exhibited robust mobility shifts (Fig. 1C), demonstrating that the sites leading to mobility shifts are likely to be in the C-terminal domain (note that we have previously shown that these truncated forms are active kinases, so the lack of mobility shift does not arise from lack of kinase activity [21]). In addition, assessment of cells treated with both okadaic acid and the proteasome inhibitor MG132 did not produce extensive mobility shifts for the mutant DBTs (Fig. 1D), suggesting that their absence in okadaic acid-treated samples is not due to their rapid degradation by the proteasome.

Further analysis mapped the target serines and threonines more precisely. BLAST alignment of the DBT C-terminal sequence from Drosophila melanogaster with those from other Drosophila species showed strongest conservation of amino acid sequence in a region C terminal to amino acid 387 (Fig. 2A), suggesting that the 6 conserved threonines and serines in this region might be the sites of autophosphorylation. To determine if these and other potential sites are phosphorylated in DBT, we purified DBT-MYC-HIS from Drosophila S2 cells by sequential Ni-agarose chromatography and anti-MYC immunoprecipitation and analyzed tryptic peptides by mass spectrometry. Using this approach, we identified one phosphopeptide in DBT from S2 cells not treated with okadaic acid (the low recovery of DBT-MYC-HIS from okadaic acid-treated cells precluded an analysis from these cells.). This was RPSIR with phosphorylation occurring on Ser408. Figure 2B presents the MS/MS scans from both the unphosphorylated and phosphorylated forms of the indicated tryptic peptide (trypsin does not cleave the arginine immediately N terminal to proline), which exhibits the expected difference in the mass of the modified peptide as well as the typical neutral loss (H₃PO₄) from the peptide during peptide MS/MS. Unfortunate placement of useful endopeptidase specific sites in this region hindered additional phosphopeptide detection.

FIG 2.

Identification of candidate phosphorylation sites in the C terminus of DBT. (A) BLAST alignments showed a region of strong sequence conservation in the DBT C termini of the indicated Drosophila species, with 6 conserved serines and threonines that were all mutated to alanine in the DBT^C/ala mutant. Phosphorylation of the indicated serine (star) was detected by mass spectrometry. (B) The peptide RPSIR was detected both unmodified and phosphorylated in the liquid chromatography (LC)-tandem MS data, shown here in the tandem MS spectra of the two identified forms from their doubly charged state with some of the more abundant b-ion and y-ion MS/MS fragments indicated. In addition, the expected common residual ion (m/z ∼305.8 [+2]) due to neutral loss of either phosphoric acid from the phosphopeptide or water from the unmodified serine during peptide MS/MS is indicated in both spectra. (C) DBT^WT-MYC-HIS and DBT^C/ala-MYC-HIS were induced in Drosophila S2 cells with CuSO₄ and then treated with 60 nM okadaic acid (+OA) in dimethyl sulfoxide (DMSO) or without okadaic acid (DMSO only) (−OA) for 17 h, followed by immunoblot analysis of the MYC-tagged transgenic protein with anti-MYC antibody. DBT^WT exhibits phosphorylation-dependent mobility shifts, while the DBT^C/ala protein is largely unaffected. Note that DBT^C/ala exhibits a faster mobility than DBT^WT in both S2 cells and fly heads when expressed with a timGAL4 driver (HE), either because of differences in residual phosphorylation under the condition without OA or because of the amino acid mutations. CTL, extracts from fly heads not expressing transgenic DBT. Extracts were aliquoted in amounts that produced roughly equal overall signals for DBT-MYC in all lanes (otherwise OA-treated extracts would contain much less DBT^WT).

The detection of phosphorylation of Ser408 by mass spectrometry confirmed the phosphorylation of DBT in the conserved region from position 389 to 408 (Fig. 2A), and therefore all 6 of the serines and threonines in this region were converted to alanines to determine if these changes eliminated okadaic acid-induced electrophoretic mobility shifts. Indeed, S2 cells expressing this mutant form of DBT (DBT^C/ala-MYC-HIS) exhibited a greatly reduced prevalence of shifted forms (Fig. 2C), as did DBT truncated at amino acid 387 (Fig. 1C). Phosphorylation of Ser333 or Ser334 detected in previous studies (28, 29) could produce the small amount of DBT that is slightly shifted in the 387 mutant (Fig. 1C and 2C); the strongly shifted species detected with DBT^WT, presumably produced by multiple phosphorylations, were not detected for the 387 or C/ala mutant DBTs expressed in S2 cells. Therefore, we conclude that most of the residues producing these electrophoretic mobility shifts have been mutated in the DBT^C/ala mutant, although not all of the 6 mutated residues are necessarily phosphorylated. It is possible that the DBT^C/ala mutant also reduces phosphorylation of Y400 by another kinase, due to a lack of normal surrounding sequences.

Autophosphorylation does not reduce DBT kinase activity in vitro.

Autophosphorylation has shown to inhibit the kinase activity of vertebrate CKIδ/ε toward substrates in vitro (16, 17), and we also find that it leads to a strong inhibition of casein phosphorylation by V5-tagged vertebrate CKIδ expressed in and immunoprecipitated from Drosophila S2 cells (Fig. 3C). However, in the same kinds of assays there is no inhibition produced by okadaic acid treatment of the kinase activity of MYC-tagged Drosophila DBT toward casein (Fig. 3A) or PER (Fig. 3B) despite the persistence of slow-mobility forms of DBT in vitro. In addition, DBT^C/ala-MYC-HIS exhibits activity comparable to that of DBT^WT-MYC-HIS but reduced autophosphorylation in vitro, presumably because of the absence of C-terminal autophosphorylation sites (Fig. 3D). The in vitro autophosphorylation activity is additional confirmation of DBT's autophosphorylation activity. Note that both kinase activities (phosphorylation of a substrate and phosphorylation of the DBT C-terminal domain) arise from the DBT in the immunoprecipitate rather than from a contaminating kinase, because they are absent in immunoprecipitates of the catalytically inactive DBT^K/R. A small peptide was consistently phosphorylated by DBT not treated with okadaic acid and not by DBT treated with okadaic acid in the PER phosphorylation reactions (left arrowhead in Fig. 3B), and it is not clear why this was the case. Perhaps this is a proteolytic fragment of PER that contains a phosphorylation site not recognized by the okadaic acid-treated DBT. On the other hand, it may be a proteolytic fragment that is produced only by non-okadaic acid-treated immunoprecipitates, which could include other proteolytic factors besides DBT.

The results of multiple assays of DBT kinase activity toward casein showed a small but statistically significant (P < 0.05) increase in the phosphorylation of casein by DBT from okadaic acid-treated cells relative to phosphorylation by DBT from vehicle-treated S2 cells (Fig. 3E), although a caveat to this interpretation is that the C-terminally directed DBT antibody may not have recognized the phosphorylated C terminus as well as the unphosphorylated C terminus (thereby leading to higher normalized values for these reactions). We have previously shown that the presence of the C-terminal domain in Drosophila DBT reduces its kinase activity (21), but the inhibition is shown here not to be increased when the C-terminal domain is phosphorylated. However, we cannot rule out that some phosphorylation produced in other cell types or under other conditions (perhaps more extensive or qualitatively different than what has been produced here; see analysis of fly heads below) might lead to reduced DBT kinase activity.

Endogenous phosphorylated DBT is detected at low levels in wild-type flies.

In wild-type flies, DBT levels and primary electrophoretic mobility remain mostly constant throughout the day in LD (6, 10, 15) (Fig. 4A), although weak detection of shifted bands sometimes occurs (e.g., Fig. 4A, ZT13). In order to address the biological relevance of DBT's C-terminal phosphorylation, the dbt^C/ala mutant was cloned into a vector which allows Phi-C31-mediated site-specific integration and expression with tissue-specific GAL4 drivers. When expressed in all circadian clock cells with the timGAL4 driver, DBT^WT-MYC-HIS and DBT^C/ala-MYC-HIS exhibited high mobilities, with DBT^C/ala migrating slightly faster than DBT^WT (Fig. 2C and 4B). This suggests that DBT^WT-MYC-HIS is not extensively autophosphorylated in circadian cells, perhaps because cellular phosphatases maintain it in the unphosphorylated state, as they do in S2 cells without okadaic acid treatment. Neither transgenic protein shows any evidence of a circadian oscillation in electrophoretic mobility (Fig. 4B).

FIG 4.

Endogenous DBT, DBT^WT, and DBT^C/ala are not extensively autophosphorylated in vivo. (A) Head extracts from flies of the Canton S (CS) wild-type genotype were collected at the indicated time (ZT1, 1 h after lights on; ZT7, 7 h after lights on; ZT13, 1 h after lights off; ZT19, 7 h after lights off) and subjected to immunoblot analysis for β-tubulin and endogenous DBT. The slow-mobility band is putative autophosphorylated DBT (p-DBT) but is only weakly detected. (B) Head extracts collected from timGAL4>UAS-dbt^WT-myc-his, -dbt^C/ala-myc-his, or nontransgenic (CTL) progeny at the indicated times of an LD cycle were immunoblotted with anti-MYC or antitubulin. The mobilities of transgenic DBT do not change over the course of the day, and DBT^C/ala-MYC is expressed comparably to DBT^WT-MYC with a slightly higher electrophoretic mobility (See also Fig. 2C for electrophoretic mobility differences).

Expression of the DBT^C/ala mutant protein in circadian cells does not disrupt behavioral circadian rhythms.

We have previously shown that it is possible to recapitulate circadian phenotypes of mutants by overexpressing the mutant proteins with the timGAL4 driver, in a manner which suggests that the transgenic protein can titrate the endogenous DBT protein in clock protein complexes (5, 21). If the altered phosphorylation state of the DBT C terminus has biological relevance for the circadian clock mechanism, then expression of DBT^WT and DBT^C/ala in circadian clock cells might perturb the clock differently. However, there were no clear difference between DBT^C/ala-MYC-HIS overexpression and DBT^WT-MYC-HIS expression; analysis of two of the three DBT^C/ala lines suggested slightly longer periods (P < 0.05) than with the wild type (nontransgenic controls), but DBT^WT expression tends to produce slightly longer periods too (Table 1) (5, 30). While it is possible that mutation of the DBT C terminus produces a null mutant that cannot titrate endogenous DBT from circadian complexes, this possibility is unlikely because there are weak effects on circadian period and rhythmicity that resemble those of DBT^WT overexpression (Table 1) (5, 30). It is likely that DBT^C/ala is equivalent to DBT^WT because DBT^WT is maintained largely unphosphorylated in circadian complexes, so the absence of phosphorylated residues in the mutant does not produce a state any different from overexpression of the wild-type one. Our previous work has suggested that overexpression of transgenic DBT in circadian cells replaces endogenous DBT with the transgenic protein, which for DBT^WT produces fairly small effects on rhythms because the endogenous wild-type protein is replaced with transgenic wild-type protein (5).

TABLE 1.

Activity rhythms of flies expressing DBT^WT or DBT^C/ala in circadian cells^a

Genotype	Line	Avg period (h) ± SEM	% rhythmic (no. of flies assayed)
timGAL4/+		23.9 ± 0.07	94 (31)
timGAL4/+>;UAS-dbtWT-myc-his/+	45F2B	24.4 ± 0.2	31 (29)*
	21M1C	24.3 ± 0.3	100 (22)
timGAL4/+>;UAS-dbtC/ala-myc-his/+	13M1A	24.9 ± 0.3*	62 (76)
	13M1A only	23.5 ± 0.05	93 (30)
	17M1A	25.7 ± 0.2*	49 (45)*
	17M1A only	23.8 ± 0.1	96 (26)
	17M2A	24.9 ± 0.3	47 (32)*

^aLines with the indicated UAS-dbt gene were crossed to the timGAL4 driver line, progeny flies hemizygous for both the driver and responder were assayed in locomotor activity assays, and periods were determined by chi-square periodogram analysis. In addition, responder-only and driver-only controls were tested. Rhythmic flies exhibited a single dominant peak in the periodogram analysis and clearly rhythmic actograms. A one-way analysis of variance (ANOVA) indicated a statistically significant effect of line on period (F[8, 188] = 9.9, P < 0.0001), and a post hoc Tukey test showed that the periods of the indicated lines (*) differed significantly from those of the timGAL4/+ and responder-only flies (P < 0.01) but not from the periods of the other timGAL4>UAS-dbt lines (P > 0.05). A Kruskal-Wallis nonparametric ANOVA demonstrated significant differences in rhythmicity between lines of different genotypes (H[8, N = 291] = 71.0, P < 0.0.05), but all of these were differences from one or more of the driver-only or responder-only controls (*). None of the timGAL4>dbt genotypes differed significantly from any of the other timGAL4>dbt lines with a different genotype (P > 0.3).

The conserved serines and threonines in the DBT C terminus are required for inhibition of apoptosis with DBT overexpression, which is accompanied by proteasome-dependent degradation of DBT^WT but not DBT^C/ala.

Since mutation of the DBT C terminus phosphorylation sites does not have effects on the circadian rhythm but the circadian clock bdbt mutant produces elevated phosphorylation of DBT (15), we suspected that the suppression of autophosphorylation by the circadian clock might affect other processes but not the clock itself. In work to be published elsewhere, we demonstrate that reductions in DBT activity in circadian cells lead to caspase activation at ZT7 and not at other times of day (23). Moreover, previous work from others has demonstrated that reductions in DBT activity induce apoptosis in imaginal disks (22). Hence, we tested whether elevated DBT might inhibit apoptosis in S2 cells and whether this inhibition would involve changes in phosphorylation of the DBT C terminus. When subjected to high-intensity UV light, S2 cells underwent apoptosis, resulting in reduced cell number (Fig. 5, mock). Overexpression of DBT^WT-MYC-HIS prior to UV irradiation prevented loss of cells (Fig. 5). This protection from apoptosis requires the phosphorylated amino acid(s) in the C terminus, as the DBT^C/ala-MYC-HIS mutant does not confer protection (Fig. 5).

FIG 5.

DBT^WT but not DBT^C/ala protects S2 cells from UV-induced apoptosis. (A) Photomicrographs of S2 cells taken 24 h after UV treatment (+UV) or at the same time but without UV treatment (−UV). Cells were stably transfected with plasmids expressing the indicated form of DBT (DBT^WT-MYC-HIS or DBT^C/ala-MYC-HIS induced with CuSO₄) or without any plasmid (“mock” indicates S2 cells without a transgene that were treated with CuSO₄ along with the transgenic lines.). (B) Cell viability was determined by an observer blinded to the condition of the cells 24 h after UV treatment for the cells described for panel A. The individual was unaware of the treatment and counted 3 random fields of view from each well. The experiment was performed three times, and for each experiment 2 wells were set aside for each treatment. Unlysed cells were counted, and the number of cells in the mock-transfected group without UV treatment was set to 100%, with the others normalized to that number. Each data point represents the mean from three experiments ± standard error of the mean (SEM).

DBT^WT-MYC-HIS was degraded after UV treatment (Fig. 6A) in a proteasome-dependent manner, since the proteasome inhibitor MG132 prevented the reduction in DBT^WT-MYC-HIS and resulted in slower-mobility forms of DBT that are typically indicative of ubiquitination and phosphorylation. A good part of the shift is due to phosphorylation, as treatment of the lysates with lambda phosphatase eliminated most of the mobility shift (Fig. 6A). Interestingly, DBT^C/ala-MYC-HIS protein levels, which were similar to those of DBT^WT-MYC-HIS before UV exposure, were less affected by UV treatment and were not shifted in the presence of MG132 by UV treatment and subsequent phosphatase treatment, suggesting that phosphorylation of the amino acids altered in DBT^C/ala is required for the degradation of DBT^WT in response to UV (Fig. 6A).

FIG 6.

DBT^WT but not DBT^C/ala is targeted for phosphorylation and degradation after UV irradiation. (A) DBT-MYC-HIS or actin (a loading control) was detected by immunoblot analysis of lysates from S2 cells expressing the indicated transgenic protein (DBT^WT-MYC-HIS or DBT^C/ala-MYC-HIS, induced with CuSO₄ and detected with anti-MYC) collected without UV treatment (−UV) or 18 h after UV treatment (+UV). The samples on the right were harvested from cells treated with the proteasome inhibitor MG132, and the lysates were either treated or not with lambda phosphatase before electrophoresis. (B) Cells expressing DBT from an endogenous gene were treated or not treated with MG132 prior to UV irradiation (+UV) and harvested after 4 to 18 h for immunoblot analysis. DBT with an electrophoretic mobility shift was detected at 4 h, and at 18 h from cells treated with MG132 during UV treatment; the 18-h shift was eliminated by lambda phosphatase treatment but not by lambda phosphatase with phosphatase inhibitors (i).

Similar effects were also produced by UV on endogenous DBT in S2 cells. By 4 h after UV treatment, endogenous DBT exhibited electrophoretic mobility shifts, and by 18 h, endogenous DBT was gone (Fig. 6B). However, if the cells were treated with MG132 before UV, DBT persisted in a slow-electrophoretic-mobility state that was largely (but not completely) reversed by lambda phosphatase treatment (Fig. 6B). These results suggest that endogenous DBT also becomes phosphorylated and eventually degraded by the proteasome in response to UV treatment.

In order to determine if overexpression of DBT^WT could antagonize UV-induced cell death in vivo, we expressed both DBT^WT-MYC-HIS and DBT^C/ala-MYC-HIS with the timGAL4 and actinGAL4 drivers and subjected the larvae to UV exposure. None of the larvae so exposed emerged from pupae. However, overexpression of DBT^WT allowed a significantly higher proportion of the larvae to pupate, while overexpression of DBT^C/ala did not (Fig. 7C; viability plotted is the percentage of irradiated larvae that survived for 48 h after irradiation and pupated.). In addition, DBT^WT exhibited electrophoretic mobility shifts and disappeared after UV treatment, while DBT^C/ala did not (Fig. 7A and B), consistent with the S2 cell response. Therefore, DBT^WT is shown to promote survival with overexpression in a phosphorylation-dependent manner in vivo.

FIG 7.

Expression of DBT^WT-MYC-HIS in circadian cells or all cells provides some protection from death and phosphorylation of DBT, while expression of DBT^C/ala-MYC-HIS does not. (A) The indicated transgenic DBT was expressed with the timGAL4 driver, and larvae were subjected to UV irradiation and harvested at the indicated times after UV irradiation. DBT-MYC was detected by immunoblot analysis with anti-MYC antibody and actin with antiactin. (B) The experiment was conducted as for panel A but with the actinGAL4 driver. (C) The percentage of larvae that survived 48 h after UV treatment was tabulated as viability for each genotype. Almost no larvae eclosed for any of the genotypes. The viability results were obtained in 3 experiments, and the data are plotted as mean ± SEM.

The spaghetti cochaperone targets the DBT C-terminal domain.

In work to be published elsewhere, we show that reductions in the spaghetti cochaperone (SPAG) (CG13570), isolated in a screen for proteins that affect DBT electrophoretic mobility, trigger autophosphorylation and degradation of DBT in fly heads at ZT7, with subsequent activation of caspases that do not trigger immediate apoptosis (23). Hence, we addressed whether RNAi knockdown of SPAG in S2 cells would lead to increased phosphorylation of DBT. A slower-mobility form of DBT was produced with spag RNAi for DBT^WT (Fig. 8A), and DBT^WT-MYC-HIS levels were eventually reduced while those of the phospho-mutant DBT^C/ala-MYC-HIS were not (Fig. 8B). Hence, SPAG antagonizes DBT autophosphorylation and stabilizes DBT. The effect of SPAG on DBT is likely due to an interaction between SPAG and DBT complexes, as we detect SPAG-hemagglutinin (HA) in coimmunoprecipitates with DBT-MYC-HIS from S2 cells (Fig. 8C).

FIG 8.

spaghetti dsRNA causes a DBT phosphorylation-dependent electrophoretic mobility shift and eventual degradation, and SPAG associates with DBT. (A) Endogenous DBT levels were detected by immunoblotting using anti-DBT-C antibody from S2 cell lysates after treatment with spaghetti dsRNA (lane numbers indicate hours after RNA addition). (B) Anti-MYC immunoblot of S2 cell lysate at 24 h after spaghetti dsRNA addition. DBT^WT-MYC-HIS is degraded but not DBT^C/ala-MYC-HIS. (C) HA-tagged SPAG is detected in S2 cell lysates as a coimmunoprecipitate with MYC-tagged DBT.

DISCUSSION

DBT directly phosphorylates PER (5, 6) and contributes to CLK phosphorylation (31, 32) to coordinate the timing of PER degradation and CLK activity during the circadian cycle. It is essential for circadian rhythmicity to restrict the timing of DBT's effects to the late day and early evening, when it triggers the degradation of nuclear PER and delays the accumulation of cytosolic PER to convert a simple negative feedback loop to a circadian oscillation of clock gene expression with a long (24-h) period (4). This temporal regulation could be effected by regulation of DBT activity, its targeting of PER, or the proteolysis that is a consequence of PER phosphorylation. The work described here was initiated to investigate the possibility that autophosphorylation of DBT's C terminus confers circadian regulation of its activity, although we found no evidence for such regulation.

Because most mutations of mammalian CKIδ/ε shorten period (30, 33) while most mutations of Drosophila DBT lengthen the circadian period (4, 34), it was possible that mammalian CKI and Drosophila DBT have somewhat different mechanistic roles in the clock. However, our previous work established that period-altering mutations introduced into Drosophila DBT and vertebrate CKIδ produce the same period-shortening effects in the context of either protein and in both flies (21) and (in the case of the tau mutation) mammals (33). Moreover, despite the sequence divergence of their C termini, our current results demonstrate that the C terminus of DBT is the target of autophosphorylation (Fig. 1 and 2), as it is in mammalian CKIδ/ε. While phosphorylation of the C terminus is required to produce electrophoretic mobility shifts of DBT, we cannot rule out phosphorylation of DBT outside the C terminus.

Our current analysis does suggest a difference in the consequences of autophosphorylation of Drosophila DBT and mammalian CKIδ/ε. In S2 cells, Drosophila DBT activity is not reduced in response to okadaic acid-induced autophosphorylation, while vertebrate CKIδ activity is (Fig. 3) (16, 17). It has been proposed that the C terminus of mammalian CKIε undergoes a phosphorylation-dependent interaction with the kinase domain to prevent substrate binding and phosphorylation (18). Since the positions of serines and threonines in the DBT and CKIδ/ε termini are not well conserved, it is not expected that this phosphorylation-dependent regulatory interaction would be conserved. It is also possible that the DBT recovered in our experiments is not fully enough phosphorylated in S2 cells to exhibit reduced kinase activity, while in fly circadian tissues DBT phosphorylation is more extensive or different (although there is no evidence for highly phosphorylated forms of DBT in wild type flies [Fig. 4]). However, another possible role of DBT phosphorylation would be to destabilize DBT, since UV treatment produces phosphorylation in S2 cells (Fig. 6) and larvae (Fig. 7), and proteasome-dependent degradation requires the wild-type phosphorylation sites (Fig. 6) of DBT in S2 cells, as does treatment of S2 cells with spag RNAi (Fig. 8).

The lack of strong mobility shifts for DBT during the circadian cycle and the equivalent effects of DBT^C/ala-MYC-HIS and DBT^WT-MYC-HIS overexpression on circadian behavior suggest that rhythmic phosphorylation of DBT is not essential for the circadian clock mechanism but leave open the possibility that the clock can regulate DBT autophosphorylation. bdbt dsRNAi increases the amount of DBT phosphorylation (15). In this scenario, the clock is predicted to suppress any pathway that is triggered by DBT autophosphorylation.

Here, it is shown that one effect of DBT that is linked to its autophosphorylation is DBT's antagonism of apoptosis induced in response to UV light (Fig. 5 to 7). While DBT^WT-MYC-HIS can prevent apoptosis, the DBT^C/ala-MYC-HIS mutant will not. While DBT^WT is phosphorylated and degraded after UV treatment and DBT^C/ala is not, the protective effect of DBT^WT implies that its overexpression (relative to that of endogenous DBT) and its phosphorylation confer the protection. Expression of DBT^C/ala, while more persistent than that of DBT^WT, does not confer protection, potentially because it does not get autophosphorylated. The cochaperone SPAG, which was identified in our screen of DBT regulators as one which strongly lengthens circadian period and is required for high DBT expression at ZT7 (23), is shown here to be required in Drosophila to maintain DBT in a mostly unphosphorylated state. SPAG contributes to assembly of HSP-90-dependent complexes such as RNA polymerase and ribonucleoprotein particles (24, 35, 36). It contains tetratricopeptide repeats (TPRs), as does BDBT, which is likewise essential for normal circadian rhythms, DBT phosphorylation, and DBT activity (15). Taken together, our data suggest that normal DBT circadian complexes are important for maintaining it in an unphosphorylated state and that disruption of circadian complexes can contribute to hyperphosphorylation of DBT.

Our additional work has shown that reduced levels of SPAG or DBT trigger caspase activation (23), which typically induces apoptosis, so it is curious that the autophosphorylation of DBT^WT observed here is associated with protection from cell death, as it is followed by elimination of DBT from S2 cells. It is possible that DBT's inhibition of cell death requires at least transient overexpression of a phosphorylatable form of DBT. It is also possible that DBT targets a regulatory step of the cell death pathway that can be either inhibitory or activating depending on cellular context. While DBT has previously been suggested to repress cell death in Drosophila (22), its role in the regulation of mammalian cell death remains unclear (37). In any event, our results show a potential role for DBT autophosphorylation in the regulation of apoptosis. Roles for autophosphorylation have been demonstrated for signaling by the metabotropic glutamate receptor (38) and the Wnt receptor (39). It will be important to determine the mechanisms that trigger DBT autophosphorylation to induce apoptosis and to assay the other numerous signal transduction pathways in which CKIε/δ have been implicated for any effects mediated by autophosphorylation.