The C-terminal Kinase and ERK-binding Domains of Drosophila S6KII (RSK) Are Required for Phosphorylation of the Protein and Modulation of Circadian Behavior*

Background: S6KII (RSK) kinase is important for neuronal plasticity and behavior. Results: Using genetic approaches, we identified functional S6KII domains relevant for circadian behavior. Conclusion: ERK binding to S6KII and activation of its C-terminal kinase modulate circadian period whereas the N-terminal kinase is dispensable for circadian function. Significance: This is the first in vivo structure/function analysis of S6KII (RSK) in the context of behavior. A detailed structure/function analysis of Drosophila p90 ribosomal S6 kinase (S6KII) or its mammalian homolog RSK has not been performed in the context of neuronal plasticity or behavior. We previously reported that S6KII is required for normal circadian periodicity. Here we report a site-directed mutagenesis of S6KII and analysis of mutants, in vivo, that identifies functional domains and phosphorylation sites critical for the regulation of circadian period. We demonstrate, for the first time, a role for the S6KII C-terminal kinase that is independent of its known role in activation of the N-terminal kinase. Both S6KII C-terminal kinase activity and its ERK-binding domain are required for wild-type circadian period and normal phosphorylation status of the protein. In contrast, the N-terminal kinase of S6KII is dispensable for modulation of circadian period and normal phosphorylation of the protein. We also show that particular sites of S6KII phosphorylation, Ser-515 and Thr-732, are essential for normal circadian behavior. Surprisingly, the phosphorylation of S6KII residues, in vivo, does not follow a strict sequential pattern, as implied by certain cell-based studies of mammalian RSK protein.

p90 ribosomal S6 kinase (RSK) 2 is a well-conserved serine/ threonine kinase and a target of the Ras-mitogen-activated protein kinase (MAPK)/extracellular-signal-regulated kinase (ERK) signaling cascade (1). RSK was first identified as a kinase that phosphorylates the 40 S ribosomal subunit protein S6 in Xenopus larval extracts (2), although it has since been determined that S6 protein may not be a relevant substrate in vivo (3)(4)(5). As with other elements of the Ras-ERK pathway, RSK kinase activity has been linked to a wide range of critical cellular processes including transcriptional and translational regulation, cell proliferation, survival, growth, and cell motility (6), reviewed in Refs. 7, 8. RSK is unique among serine/threonine kinases in that it contains two distinct and functional kinase domains joined by a linker region (9,10). The kinases are thought to be sequentially activated by a series of phosphorylation events, beginning with phosphorylation by activated ERK. However, recent studies indicate that kinase activation may occur in the absence of this full sequential series of modifications (11,12). As the majority of these studies were conducted in vitro, there is insufficient evidence as to whether sequential phosphorylation/kinase activation is required in vivo. The sole known purpose of the RSK carboxyl (C)-terminal kinase is to activate the amino (N)-terminal kinase through autophosphorylation. In contrast, the N-terminal kinase is responsible for phosphorylation of other protein substrates (9,13,14). N-terminal kinase activity was thought to be essential for RSK function until a recent study demonstrated an alternative role for RSK in fly eye development as a non-catalytic, scaffolding protein (15).
The mammalian genome encodes four isoforms of RSK (RSK 1-4) while only a single isoform has been described for Drosophila melanogaster (dRSK or S6KII) that has ϳ60% amino acid identity with RSK1 (16). Whereas there have been extensive studies of RSK structure and function using mammalian cell-based assays, detailed studies of fly S6KII functional domains have not been reported even though the kinase is known to be important for memory functions and circadian behavior (17)(18)(19). The use of Drosophila for such an analysis will permit critical domains and phosphorylation sites of S6KII to be delineated, in vivo, in the context of behavior and plasticity, with obvious relevance for mammalian RSK function.
Even though S6KII is ubiquitously expressed throughout Drosophila development and in all embryonic tissues (15), S6KII null mutant flies are viable (15,19). Interestingly, S6KII null (S6KII ignorant or S6KII ign ) flies exhibit circadian molecular and behavioral phenotypes previously described by our laboratory (17). Specifically, the mutant exhibits a short-period phenotype that appears to result from alteration of the PERIOD (PER)-based circadian oscillator. Based on those and other studies, we have proposed a role for S6KII in modulating the Drosophila circadian molecular oscillator that involves interaction and cooperation with the known clock kinase casein kinase 2 (CK2) (17). Given that S6KII also interacts with numerous other partners in a variety of ERK pathway roles, it is possible that S6KII modulation of oscillator function is controlled by ERK signaling. In addition, it is not known whether S6KII serves as a kinase or alternatively as a scaffolding protein in the circadian system. Finally, we wondered whether the sequence of RSK phosphorylation and kinase activation observed in mammals is relevant, in vivo, for S6KII modulation of fly circadian behavior.
To further understand the role of S6KII in the circadian system, we set out to evaluate the importance of conserved protein domains and phosphorylation sites for kinase regulation of locomotor activity rhythms. We report that S6KII's circadian function does not require its N-terminal kinase activity, similar to findings reported for Drosophila eye development (15). In contrast, C-terminal kinase activity, previously thought to be responsible only for N-terminal kinase activation, is required for normal circadian behavior. Our studies also suggest that ERK binding to and phosphorylation of S6KII threonine 732 (T732) within clock neurons is essential for normal rhythmicity. Whereas S6KII was shown to negatively regulate ERK in the fly eye (15) and at the neuromuscular junction (20), our work indicates that activation of S6KII by ERK is required for modulation of the circadian clock. Further, we show that both ERK binding and C-terminal kinase activity are important for autophosphorylation of S6KII serine 515 (S515) and T732 phosphorylation, whereas phosphorylation at S357, which activates the N-terminal kinase, is not dependent on these activities. Phosphorylation of S6KII S515 or T732 is not required for normal phosphorylation of the protein, but it is required for wild-type circadian behavior. These studies provide novel insights about the function of S6KII, in vivo, and support a model in which ERK regulates S6KII in clock cells, thereby, modulating circadian behavior.

EXPERIMENTAL PROCEDURES
Stock Maintenance and Generation of Fly Strains-Drosophila cultures were reared at 25°C and 60% relative humidity in an LD 12:12 cycle on a modified standard medium containing wheat germ. For genetic crosses and behavioral experiments, flies were collected using C0 2 anesthesia.
Analysis of Behavioral Rhythmicity-Locomotor activity data from individual flies were collected using the Drosophila Activity Monitor (DAM) system (Trikinetics, Waltham, MA). Flies were entrained at 20°C to an LD 12:12 cycle for 4 days and then transferred to constant darkness (DD) at the same temperature for approximately 2 weeks. Our previous work (17) shows that the S6KII mutant phenotype is most severe at 20°C. To estimate period and visualize actograms, we used a MATLAB (MathWorks)-based signal processing toolbox (21). We employed a time series analysis called autocorrelation to look for periodicity in the activity data and generate a correlogram (with peaks representing harmonics in the data). In accordance with the standard in the field, period was estimated from the third peak of the correlogram. Differences in circadian period were assessed for statistical significance using a nonparametric ANOVA (Kruskal-Wallis Test) with Dunn's Multiple post-hoc comparisons (GraphPad InStat).
Western Analyses-Fly heads were collected and homogenized in 3 volumes of Head Extraction Buffer (50 mM KCl, 10 mM HEPES, 5 mM Tris-HCL, 10% glycerol, 2 mM EDTA, 1% Triton X-100) with 1 mM DTT, 0.4% Nonidet P-40, 0.5 mM PMSF, 10 mM pNPP, and a 1:100 dilution of Halt protease inhibitor mixture (Pierce). Extract buffer for pS6KII blots also contained a 1:10 dilution of PhosSTOP phosphatase inhibitor mixture (Roche). Standard immunoblotting techniques were used to examine protein abundance for S6KII and phosphorylated S6KII (pS6KII). Protein lysates to be examined with antibodies against phosphorylation sites were only heated to 55°C for 10 min while other proteins were heated to 95°C for 5 min. To generate gels for blotting, 50 g or 75 g of protein per time point were loaded for the detection of S6KII or pS6KII, respectively. The following antibodies were used: guinea pig anti-S6KII, 1:1000 (19); rabbit anti-pSer380-RSK, 1:1000 (Cell Signaling); rabbit anti-pSer221-RSK, 1:1000 (Abcam); rabbit anti-pThr573-RSK, 1:200 (Cell Signaling); rat anti-Drosophila (D) N-Cadherin, 1:500 (Developmental Studies Hybridoma Bank). Bands were visualized using the ECL or ECL Plus Western blotting detection systems (Amersham Biosciences). Band density and background were measured using NIH ImageJ. After background subtraction, total S6KII was normalized to N-Cadherin level to account for differences in loading. pS6KII signal was expressed as a fraction of mean total S6KII amount for each genotype. Quantitative differences in S6KII phosphorylation status and total abundance were evaluated using an ANOVA with Dunnett's post test.

Transgenic Expression of Mutant Proteins for the Assay of S6KII Functional Domains in Vivo-
To discern the role S6KII plays in circadian rhythms, we examined the behavioral and molecular effects of mutations that either block the function of a particular conserved domain (kinase or ERK-binding domains) or alter phosphorylation status of specific residues of the protein. Fig. 1A presents a current model for the sequential activation of S6KII, based on studies of mammalian RSK. The indicated phosphorylation sites are conserved in all mammalian RSK isoforms, and in the Drosophila and Caenorhabditis elegans homologs (22). ERK is thought to initiate the sequential phosphorylation of multiple S6KII residues by binding to a specific docking domain at the terminus of the protein, resulting in phosphorylation of residue T732 (and perhaps also S498) (T573 and S363 in RSK, respectively) and subsequent activation of the C-terminal kinase (22)(23)(24). The C-terminal kinase autophosphorylates S515 (RSK S380), creating a binding site for PDK1, which then docks and phosphorylates S6KII S357 (RSK S221), activating the N-terminal kinase domain and enabling the phosphorylation of S6KII target substrates (14,(25)(26)(27).
We utilized multiple site-directed mutants of S6KII, summarized in Fig. 1B, to determine which conserved domains and phosphorylation events are required for the protein modulation of circadian oscillator function. All mutant S6KII transgenes carried single amino acid changes that were produced in vitro (see "Experimental Procedures").
An N-terminal kinase-dead mutant called S6KII K231R carries a Lys-231 to Arg change while C-terminal kinase-dead S6KII K597M contains a Lys-597 to Met alteration (Fig. 1B). Arg-902 was mutated to Ala to obtain an ERK binding-deficient allele (S6KII R902A ). It was previously demonstrated that S6KII K231R and S6KII K597M are kinase-dead alleles, based on labeled phosphate incorporation into the S6 ribosomal protein, a known substrate of S6KII, and that S6KII R902A lacks ERKbinding activity (15). The S6KII K231R/R902A line contains a combination of the K231R (N-terminal kinase-dead) and R902A (ERK binding-deficient) mutations in the same S6KII transgene. Pseudo-phosphorylated (S6KII T732E ) or unphosphorylatable (S6KII T732A ) mutants of S6KII were generated in our laboratory by Thr to Glu and Thr to Ala changes, respectively. Similarly, we generated pseudo-phosphorylated (S6KII S515D ) or unphosphorylatable (S6KII S515A ) S515 mutants by Thr to Asp and Thr to Ala changes, respectively (Fig. 1B). Mutant proteins were expressed specifically in clock cells, for immuno- , are joined by a linker region. S6KII also contains an ERK1/2-binding domain at its C terminus. Schematic depicts the series of events thought to lead to S6KII kinase activation (right to left), beginning with ERK binding to the C terminus and phosphorylation of T732. Arrows indicate phosphorylation events. Homology between S6KII and human RSK1 phosphorylation sites is shown. B, point mutations generated to disrupt N-terminal kinase activity (lysine, K to arginine, R), C-terminal kinase activity (lysine, K to methionine, M) and ERK-binding (arginine, R to alanine, A) are shown below the S6KII protein schematic. Serine and threonine phosphorylation sites mutated to be pseudophosphorylated (aspartate, D or glutamate, E) or unphosphorylatable (alanine, A) are depicted above the schematic. chemical and behavioral studies, using the Gal4/UAS binary expression system (28). A summary of all S6KII mutants employed in this study with abbreviations is shown in supplemental Table S2.
Mutant Forms of S6KII Can Be Detected in Head Tissues and Clock Neurons of the Fly Brain-Prior to behavioral testing, we verified that mutant S6KII proteins could be detected in whole head or brain tissues. Detection of proteins was accomplished using an antibody specific for fly S6KII (19). To distinguish transgene-encoded S6KII from the endogenous protein, we performed immunoblotting experiments in an S6KII proteinnull background using Df(1)ign ⌬58/1 , a deletion of the S6KII gene (hereafter referred to as S6KII ign ) (19). As expected, S6KII protein was detected in head lysates from S6KII ϩ genetic background control flies (w 1118 ), but not in lysates from S6KII ign mutants ( Fig. 2A). Importantly, control lysates from flies, which carried only the Gal4 driver (S6KII ign ;timUASGal4) or the UAS responder transgene (S6KII ign ;UAS-S6KIIϩ) did not exhibit detectable S6KII protein. However, transgene-encoded wildtype and domain mutant (K231R, R902A, or K231R/R902A) S6KII proteins, when expressed using timUASGal4, were easily detected in an S6KII ign background, at higher than endogenous levels in head lysates. S6KII S515 phosphorylation-mutant proteins (S515A, S515D) were also detected at comparable, high levels, and not significantly different from the transgene-encoded wild-type S6KIIϩ (Fig. 2B). Using two different antibodies, T732 phosphorylation-mutant proteins (T732A, T732E) were detected at levels higher than endogenous S6KII (see w 1118 control), but less than that observed with timUASGal4-driven S6KII ϩ expression ( Fig. 2B and data not shown). We note, however, that only T732A is statistically different from the S6KIIϩ control (Fig.  2B). Reduced abundance may be caused by genomic position effects on transgene expression or reduced stability of the mutant proteins. Nonetheless, S6KII T732A and S6KII T732E proteins are expressed at a level higher than endogenous S6KII, which we know is sufficient for normal circadian behavior (17).
The pdfGal4 driver expresses in 16 critical pacemaker neurons (a subset of the TIM-containing clock cells) and was utilized in certain behavioral experiments. Using immunoblotting techniques, we failed to detect S6KII protein in head extracts of pdfGal4ϾUAS-S6KII transgenic flies. Thus, immunostaining procedures were employed to detect the S6KII K597M , S6KII T732A and S6KII T732E mutant proteins in these pacemaker neurons. Using whole mounts of the brain (supplemental Fig.  S1), both wild-type and mutant S6KII proteins (K597M, T732A, T732E) were detected in the PDF neurons, although the level of expression varied slightly among strains. Nevertheless, studies described below (Figs. 3 and 5) demonstrated that certain mutant proteins can rescue behavior of S6KII ign flies even when expressed at levels lower than that of transgene-encoded wild-type protein. In addition, a very high level of wild-type protein did not cause behavioral alterations ( Fig. 3; Ref. 17), indicating that S6KII abundance is not limiting for circadian behavior.
S6KII N-terminal Kinase Activity Is Dispensable for Circadian Behavior-Cell-based assays suggest that mammalian RSKs function primarily as kinases while in Drosophila S6KII enzymatic activity is thought to be dispensable for the protein function in eye and wing development (15). To establish whether S6KII kinase activities are relevant for the modulation of circadian period, we determined if expression of various mutant S6KII proteins in clock cells was sufficient to rescue the S6KII ign short-period locomotor-activity phenotype (Fig. 3, A  and B; supplemental Table S1). The short period phenotype of S6KII ign was described in detail in a previous publication from our laboratory (17). Briefly, the wild-type period for the locomotor activity rhythm is ϳ24.0 h whereas S6KII ign mutants have an average period of ϳ22.8 h. As shown previously (17) and in Fig. 3, expression of UAS-S6KII ϩ in TIM-containing  MAY 11, 2012 • VOLUME 287 • NUMBER 20 JOURNAL OF BIOLOGICAL CHEMISTRY 16751 clock cells (timUASGal4) or PDF neurons (pdfGal4) is able to completely rescue the S6KII ign short-period phenotype (Fig. 3,  A and B, supplemental Table S1). Similarly, timUASGal4driven expression of S6KII K231R , an N-terminal kinase-dead allele, restored normal circadian period to S6KII ign mutants, demonstrating that this kinase domain is not required for circadian behavior (Fig. 3, A and B, supplemental Table S1).

S6KII Structure and Function in Circadian Behavior
S6KII C-terminal Kinase Activity Is Required for Normal Circadian Period-We next asked whether C-terminal kinase activity is required for modulation of circadian behavior using expression of the S6KII K597M mutant transgene. As shown in Fig. 3, A and B and supplemental Table S1, expression of the S6KII K597M mutant in PDF neurons of S6KII ign flies did not significantly rescue circadian period, i.e. circadian periods were similar in S6KII ign ;pdf-Gal4ϾUAS-S6KII K597M flies and control mutants carrying only the UAS transgene. We note, however, that activity bouts are longer in flies expressing the K597M mutant, relative to S6KIIϩ expression, although we do not know the cause of this effect. Nonetheless, there is no difference in percent rhythmicity (not shown) or circadian period (supplemental Table S1) between the genotypes. Lack of rescue was also observed for a mutant in which both N-terminal and C-terminal kinase activities are eliminated (29), presumably due to lack of C-terminal kinase activity. These results demonstrate that C-terminal kinase activity is important for the circadian function of S6KII, and they represent the first link between the S6KII C-terminal kinase and behavior.
ERK Binding to S6KII Is Required for Normal Circadian Behavior-ERK binding to S6KII is essential for Drosophila eye and wing development (15). Thus we asked whether the same is true for circadian behavior. As shown in Fig. 3, A and B, and supplemental Table S1, expression of an ERK-binding mutant (S6KII R902A ) in TIM-containing clock cells did not rescue mutant behavior; i.e. circadian period of S6KII ign ; timUASGal4ϾS6KII R902A flies is similar to that of mutants carrying only the UAS transgene (the genetic background control). This implies that the ability of S6KII to bind to ERK is important for its circadian function. We have also shown that a double S6KII mutant (S6KII K231R/R902A ) does not rescue the short-period phenotype, attributable to its ERK-binding deficit, as N-terminal kinase activity is not required for normal circadian function.
S515 Phosphorylation Correlates with S6KII Rescue of Short Period-In mammals, RSK T732, S515, and S357 are sequentially phosphorylated in a cascade of phosphorylation events that involve ERK and the C-terminal kinase domain (see Fig.  1A) (9,13,14). To determine which domains might regulate S6KII phosphorylation, we employed anti-phospho-RSK (pRSK) antibodies to monitor phosphorylation at several different residues. As these commercially available pRSK antibodies were generated against the mammalian protein, we validated the specificity of the serine antibodies for fly phospho-S6KII (pS6KII) (supplemental Fig. S2). Using these anti-phospho-RSK antibodies, S6KII pT732, pS515, and pS357 were examined in S6KII ign flies that expressed wild-type or mutant isoforms of the protein. pS6KII signal was normalized to total S6KII levels for wild type and the mutants to account for possible differences in S6KII abundance. Although ERK is also known to phosphorylate T359/S363 of RSK (22), there are no anti-phospho antibodies that recognize the homologous residues of fly S6KII (S494/S498).
As shown in Fig. 4A, there was robust phosphorylation of residue S515 on the S6KII ϩ and S6KII K231R proteins. For mutant proteins that did not rescue circadian period (S6KII R902A , S6KII K231R/R902A , and S6KII K597M ) we observed significantly decreased levels of pS515 immunoreactivity. Decreased pS515 signal in the S6KII K597M mutant strongly suggests that the C-terminal kinase autophosphorylates at this site, as in mammals. A similar result for S6KII R902A and S6KII K231R/R902A mutants indicates that ERK binding to S6KII is an important regulator of autophosphorylation. Interestingly, however, the pS515 proteins in these mutants exhibited an apparent mobility shift (migrating faster), and this might be due to reduced overall phosphorylation of the S6KII protein. Analogous to the model for RSK, ERK binding may initiate a series of events that leads to autophosphorylation at S515.
T732 Phosphorylation Also Correlates with Ability to Rescue Circadian Period-S6KII T732 is homologous to the site within the C-terminal region of mammalian RSK that is phosphorylated by ERK (23,24). Both S6KII ϩ and S6KII K231R rescued circadian period and also displayed high levels of pT732 immunoreactivity (Fig. 4A). On the contrary, very low levels of pT732 immunoreactivity were observed for mutant proteins that did not rescue behavior (S6KII R902A , S6KII K231R/R902A , and S6KII K597M ). Lower pT732 signal in the S6KII R902A and S6KII K231R/R902A mutant lines suggests that ERK binding is required for wild-type levels of phosphorylation at this site. While T732 phosphorylation is thought to be "upstream" of C-terminal kinase activation (see model in Fig. 1A), low levels of pT732 on S6KII K597M suggests that phosphorylation of this residue is dependent, at least in part, on C-terminal kinase activation.

Phosphorylation of S357 Is Not Altered by S6KII Mutations Affecting ERK Binding or N-terminal Kinase Activity, but Is
Dependent upon C-terminal Kinase Activity-Based on the RSK activation model, phosphorylation of S6KII S515 modulates "downstream" phosphorylation events including pS357, a postulated target of the PDK1 kinase (26,27). PDK1 phosphorylation is thought to be the last step in the activation of the N-terminal kinase. To determine if phosphorylation of S357 is altered by the S6KII R902A , S6KII K231R/R902A , and S6KII K597M mutations, we stripped and reprobed the pS515 blots shown in Fig. 4, A and B with anti-pS357 antibody. Wild-type and mutant S6KII proteins all had substantial (and not significantly different) immunoreactive signals, relative to total S6KII, indicative of phosphorylation at S357, with the exception of S6KII K597M (Fig. 4A). Thus, a deficiency in ERK binding does not affect phosphorylation of S357. However, S357 phosphorylation is dependent upon upstream C-terminal kinase activity, as evidenced by the low pS357 immunoreactivity in S6KII K597M . Given that pS357 levels do not correlate with ability to rescue, we suggest that S357 phosphorylation and the subsequent activation of S6KII N-terminal kinase is not critical for the modulation of circadian period.
Phosphorylation of S6KII T732 Is Essential for Rescue of Circadian Behavior-Given that ERK binding is important for S6KII circadian function, and pT732 levels in the domain mutants correlate with circadian period rescue, we wondered whether S6KII phosphorylation by ERK directly affects circadian behavior. We mutated the ERK phosphorylation site (T732) on S6KII to create a pseudo-phosphorylated mutant (S6KII T732E ) and an unphosphorylatable mutant (S6KII T732A ). UAS transgenes encoding these proteins were expressed in the PDF clock cells of the S6KII ign mutant.
We observed that S6KII T732E fully rescued the S6KII ign short period phenotype while S6KII T732A did not (Fig. 5, A and B, supplemental Table S1; compare both to their respective UAS transgene controls). Thus, phosphorylation at T732 appears to be required for rescue of the S6KII ign circadian phenotype. We examined pT732 immunoreactivity on the S6KII T732A and S6KII T732E proteins with the intent of verifying the presence or absence of phospho-signal in flies expressing the mutant proteins. However, pT732 signal could not be detected on either mutant protein (Fig. 4B), suggesting that the amino acid change rendered the epitope unrecognizable to the phosphospecific antibody. Nevertheless, opposite effects of the two mutations on circadian behavior suggest that S6KII T732E likely functions as a pseudo-phosphorylated protein. We had previously wondered whether ERK binding and phosphorylation of S6KII was under circadian control, yet rescue by the constitutively pseudo-phosphorylated S6KII T732E suggests that phosphorylation at this site need not be rhythmic for the modulation of circadian behavior. These experiments further implicate ERK binding and phosphorylation at S6KII residue T732 in clock function.
Rescue of Circadian Behavior Requires Phosphorylation of S6KII S515-The requirement of a functional S6KII C-terminal kinase for rescue of circadian period motivated us to manipulate phosphorylation at the C-terminal autophosphorylation site. Thus, we created pseudo-phosphorylated (S6KII S515D ) and unphosphorylatable (S6KII S515A ) mutants of this residue. When expressed under control of timUASGal4, a UAS transgene encoding S6KII S515A (S6KII ign ; timUASGal4ϾS6KII S515A ) and control (S6KII ign ; S6KII S515A ) flies had similar average periods (Fig. 5, A and B, supplemental Table S1), although period was slightly longer than flies carrying only S6KII ign (we attribute this period lengthening to an insertional effect of the UAS transgene). Thus, we conclude that clock cell expression of S6KII S515A does not rescue mutant behavior. Indeed, neither did expression of S6KII S515D , which we thought might behave as a pseudo-phosphorylated protein. Most likely, the mutant aspartate on S6KII S515D does not mimic a phosphorylated residue as intended (Unlike replacement of T732 with glutamate which seemed to better mimic a pseudo-phosphorylated state). An antibody against pS515 did not recognize either S6KII S515A or S6KII S515D , similar to our attempt to detect pT732 in mutant T732 flies (Fig. 4B). Nonetheless, the inability of S6KII S515A to rescue circadian period, suggests that phosphorylation of S6KII S515 is required for normal circadian behavior, and this result supports a role for autophosphorylation of this site by the C-terminal kinase.
Neither S515 nor S357 Phosphorylation Is Greatly Altered by Upstream Phosphorylation Events while pT732 Is Unaffected by Downstream S515 Phosphorylation-To ask about the interdependence of S6KII phosphorylation events, we examined the phosphorylation status of T732A, T732E, S515A, and S515D mutant proteins in vivo. T732 mutant (T732A or T732E) and S515 mutant (S515A or S515D) proteins all exhibited high levels of pS357 when signals were normalized to S6KII abundance (Figs. 4B and 2B). However, the S515A and S515D mutants had slightly but significantly decreased signal at this residue relative to S6KIIϩ (Fig. 4B). The presence of high levels of pS357 in the T732 and S515 mutants argues that phosphorylation of the latter residues does not contribute significantly to modification of S357. pT732 levels are similar to S6KIIϩ in both the unphosphorylatable S515A and pseudo-phosphorylated S515D mutants, indicating that T732 phosphorylation is independent of or upstream in the cascade leading to S515 phosphorylation (Fig. 4B). The unphosphorylatable T732A and pseudo-phosphorylated T732E proteins both exhibit levels of pS515 that are comparable to S6KIIϩ (Figs. 4B). Thus, while phosphorylation of T732 and S515 are both required for normal circadian behavior (Fig. 5, A and B), modification of S515 alone is not sufficient for rescue (i.e. in the absence of T732 phosphorylation).
Altogether, these immunoblotting experiments indicate that ERK binding is required for normal phosphorylation of S515, although phosphorylation of T732 (presumably by ERK) is not essential for downstream events of the cascade. Similarly, C-terminal kinase activity is necessary for robust phosphorylation of S357 while phosphorylation at S515 (regulated by the C-terminal kinase as demonstrated in Fig. 4A) is not required. Hence, the C-terminal catalytic and ERK-binding domains play a major role in regulating S6KII protein phosphorylation while the influence of the individual phosphorylation sites on each other's status is negligible (see summary in supplemental Table S2).

DISCUSSION
This study utilized wild-type and mutant forms of S6KII in genetic rescue experiments to identify domains that are critical for the protein's function in circadian behavior. To our knowledge, it represents the first study to identify domains of S6KII (RSK) that are required, in vivo, for a behavioral function. Although in many cases S6KII isoforms were expressed at  Table S1. B, histogram shows mean circadian period (h) Ϯ S.E. for three independent experiments. *, p Ͻ 0.001. higher than normal levels in transgenic flies, we do not think our results can be attributed to overexpression of the protein.
Expression of wild-type S6KII at high levels has no discernable effects on circadian behavior or the phosphorylation pattern of S6KII. For example, as discussed below, high level expression of a C-terminal kinase-dead mutant (S6KII Km ) does not rescue behavior nor phosphorylation defects observed at several sites including S357, a postulated PDK1 docking site within the N kinase domain. We recognize, however, that there may be effects of S6KII overexpression that are not discernable in our molecular and behavioral assays.
The C-terminal Kinase of S6KII Is Required for Normal Circadian Behavior while N-terminal Kinase Activity Is Dispensable-In agreement with a previous study of fly development (15), we show that the S6KII N-terminal kinase is dispensable for its circadian function. This result contrasts with previous studies showing that RSK functions in the Ras/MAPK pathway as a kinase (1); it suggests that phosphorylation of downstream targets by the N-terminal kinase is not essential for modulation of the circadian clock. In support of a noncritical role for the N-terminal kinase, certain mutants that fail to rescue behavior nonetheless exhibit phosphorylation of S357, an event thought to activate RSK kinase activity. In addition, an S6KII mutant (S6KII ign⌬24Ϫ3 ) missing a large portion of the N-terminal region, including the N-terminal kinase, was shown to partially rescue an S6KII-null mutant in a previous study (30).
In contrast, our studies emphasize the importance of S6KII C-terminal kinase activity for modulation of the Drosophila circadian clock. This is the first evidence, in either vertebrate or invertebrate systems, of a function for the S6KII Cterminal kinase that is independent of activation of the N-terminal kinase. It is also the first direct link between the C-terminal kinase and behavior. Heretofore, the only known function of the RSK C-terminal kinase was autophosphorylation, which leads to activation of the N-terminal kinase (9,13,14). Our results suggest that either autophosphorylation serves an independent purpose (such as altering protein-protein interactions) or that the C-terminal kinase phosphorylates other proteins.
We show that the C-terminal kinase promotes phosphorylation of S515, a presumed autophosphorylation site and a residue within the hydrophobic motif site of AGC-type kinases. This region is important for stabilization of the catalytic domain of such kinases, including RSK, cAMP-dependent kinase and protein kinase C. We note that there is residual pS515 signal in a S6KII C-terminal kinase-dead mutant (S6KII K597M ) which may indicate that other kinases phosphorylate the site or that the mutant retains an undetectable amount of activity.
Our work also suggests that S6KII S357 and T732 phosphorylation events are modulated by the C-terminal kinase. The N-terminal kinase is dispensable for circadian regulation, but nevertheless our data suggests that it is activated by the C-terminal kinase via phosphorylation of S357; in agreement with cell-based studies of RSK. The C-terminal kinase may promote S357 phosphorylation through recruitment of PDK1 or another factor. While the mechanism for C-terminal kinase modulation of T732 phosphorylation is unknown, it is possible that kinase activity that is stimulated by ERK binding feeds back to activate ERK phosphorylation of T732. This and other alterations may also involve the actions of phosphatases as there is undoubtedly a dynamic interplay between the two types of modifying enzymes. Of interest, pS515 levels are not altered in the T732A/ T732E mutants, indicating that T732 phosphorylation is not a prerequisite for S515 phosphorylation.
There is a positive correlation between S6KII variants that rescue S6KII ign mutant behavior and robust phosphorylation of S515; this suggests that phosphorylation of this residue is essential for normal circadian behavior. We note, however, that while S515 phosphorylation is correlated with rhythmicity, it is not sufficient for normal circadian behavior. Therefore, pS515 may simply be indicative of a functional C-terminal kinase whose kinase activity is necessary for modulating circadian behavior via phosphorylation of other unknown targets. In addition, phosphorylation of this residue does not affect the phosphorylation of other S6KII domains; instead C-terminal kinase activity and modification of S515 may serve to alter S6KII protein conformation and relevant protein-protein interactions. While we show that C-terminal kinase activity is important for rhythmicity, our experiments do not exclude the idea that S6KII functions as a scaffold in the circadian system, analogous to its role in Drosophila eye and wing development (15).
ERK Binding to and Phosphorylation of S6KII Modulates Circadian Function-We have shown that ERK binding to S6KII is required for transgenic rescue of circadian behavior, as it is for rescue of Drosophila eye development phenotypes (15). Consistent with a role for ERK in this pathway, we show that phosphorylation of S6KII at T732 (a known ERK phosphorylation site on RSK) is required for rescue of behavioral rhythms. ERK phosphorylation of T732, previously demonstrated for RSK in mammalian cell-based assays, was verified in the fly by the observation that pT732 is reduced in ERK binding-deficient mutants. ERK binding may promote S6KII function by facilitating activation of the C-terminal kinase, as evidenced by the decreased autophosphorylation of ERK-binding mutants. Alternatively, ERK binding may alter S6KII localization and/or binding to clock-related proteins such as CK2, similar to S6KII's regulation of ERK in fly eye development. Whatever the precise mechanism, phosphorylation of T732 and S515 are likely to be important for ERK's interaction with S6KII and the regulation of circadian period.
It was demonstrated in fly photoreceptor cells that S6KII negatively regulates ERK by retaining it in the cytoplasm. Using immunostaining procedures, however, we have shown that the localization pattern of ERK and diphosphorylated (activated) ERK are the same within PDF clock neurons (primarily cytoplasmic) in wild-type flies (w 1118 ), S6KII ign -null mutants, and ERK-binding mutants (S6KII ign ;timUG4ϾS6KII R902A ) (29). Hence, ERK may bind to and activate S6KII in clock cells, but there is no evidence that S6KII regulates ERK localization in this cell type.
Elements of Sequential Phosphorylation/Activation Occur in Vivo, but May Not Be Essential-RSK protein is thought to be activated by a sequence of protein binding and phosphorylation events, based on cell-based investigations of the protein. More recent cell-based assays question the validity of this model (12,31,32) and give added relevance to our studies as the first to examine this model in vivo.
We provide the first evidence that phosphorylation/activation of Drosophila S6KII can occur in the absence of a strict sequence of binding and phosphorylation events, but we note that there is some dependence of certain events on others. Although there is extremely low pS515 immunoreactivity in C-terminal kinase and ERK-binding mutants, indicating that S515 phosphorylation is a downstream event, there is residual phospho-signal on this residue in such flies. Thus, ERK binding and C-terminal kinase activation may not be the only events contributing to S515 phosphorylation. Consistent with this idea, in vitro analysis of mammalian RSK has demonstrated that S380 phosphorylation (S515 in S6KII) and C-terminal kinase activation can occur in the absence of RSK-ERK interactions (12). Similarly, ERK binding and C-terminal kinase activity are not the only contributors to S6KII T732 phosphorylation because residual pT732 signal exists in mutants lacking these functions. Our results also indicate that neither ERK binding nor phosphorylation at S515 or T732 is essential for phosphorylation S357 although C-terminal kinase activity influences this event. This result is in agreement with mammalian cell-based studies demonstrating that N-terminal kinase activation is not fully dependent upon C-terminal kinase activity (31,32). Altogether, ERK-binding and C-terminal catalytic activity appear to play an important role in regulating phosphorylation of the S6KII protein, but the phosphorylation of individual sites is not absolutely required for the downstream phosphorylation of others.
Model for Circadian Clock Modulation-Our previous work indicated that S6KII modulates circadian function by negatively regulating the activity of the clock kinase CK2, via physical interaction with the CK2␤ subunit (17). Thus, it is possible that a prerequisite for the S6KII-CK2 interaction is activation of the S6KII C-terminal kinase or a conformational change in the protein resulting from ERK binding. CK2␤ may be a phosphorylation target of the S6KII C-terminal kinase (although there is no evidence of this), and this would provide a mechanism by which S6KII could regulate CK2 activity. Alternatively, a change in S6KII conformation might regulate interaction with CK2, thus modulating the previously documented effects of the kinases on the PER-based clock (17,33). Finally, we cannot discount the possibility that S6KII regulates circadian clock function through a CK2-independent pathway. Further analysis of the S6KII binding partners and substrates may yield insights about the precise role of the C-terminal kinase and ERK-binding domains in circadian regulation.