Hexamerization by the N-terminal domain and intersubunit phosphorylation by the C-terminal domain of cyanobacterial circadian clock protein KaiC

BBRC Biochemical and Biophysical Research Communications 348 (2006) 864–872 www.elsevier.com/locate/ybbrc

Hexamerization by the N-terminal domain and intersubunit phosphorylation by the C-terminal domain of cyanobacterial circadian clock protein KaiC Fumio Hayashi b

a,b,1

, Ryo Iwase

a,c

, Tatsuya Uzumaki

a,b

, Masahiro Ishiura

a,b,c,*

a Center for Gene Research, Nagoya University, Furo, Chikusa, Nagoya 464-8602, Japan Bio-oriented Technology Research Advancement Institution (BRAIN), Nagoya University, Furo, Chikusa, Nagoya 464-8602, Japan c Division of Biological Science, Graduate School of Science, Nagoya University, Furo, Chikusa, Nagoya 464-8602, Japan

Received 28 June 2006 Available online 31 July 2006

Abstract Cyanobacterial clock protein KaiC has a hexagonal, pot-shaped structure composed of six identical dumbbell-shaped subunits. The opposing spherical regions of the dumbbell-shaped structures correspond to the N-terminal and C-terminal domains of KaiC. Previously, we hypothesized that the N-terminal domain of KaiC is responsible for the ATP-induced hexamerization of KaiC while the C-terminal domain is responsible for the phosphorylation of KaiC (Hayashi et al. 2004, J. Biol. Chem. 279, 52331–52337). Here, we tested that hypothesis using the purified protein of each domain. We prepared N-terminal and C-terminal domain proteins (KaiCN and KaiCC, respectively), examined their function by analyzing their ATP- or 5 0 -adenylylimidodiphosphate (AMPPNP; an unhydrolyzable ATP analog)-induced hexamerization, interactions with KaiA, and phosphorylation, and we demonstrated the following: (1) KaiCN had higher ATP- or AMPPNP-induced oligomerization activity than KaiCC. (2) KaiCc had phosphorylation activity as KaiCWT whereas KaiCN had no activity. (3) KaiCC interacted with KaiA whereas KaiCN did not. (4) The interactions of KaiCC with KaiA did not require that KaiC has a hexamer structure. (5) The interactions of KaiCC with KaiA enhanced the phosphorylation of KaiCC. Furthermore, we presented evidence for the intersubunit phosphorylation of KaiC. KaiCCatE2 , which lacks KaiC phosphorylation activity due to mutations of the catalytic Glu residues, was phosphorylated when it was co-incubated with KaiCC. We propose that the KaiC hexamer consists of a rigid ring structure formed by six N-terminal domains with hexamerization activity and a flexible structure formed by six C-terminal domains with intersubunit phosphorylation activity. Ó 2006 Elsevier Inc. All rights reserved. Keywords: Cyanobacteria; Circadian; Clock; KaiC; ATP; Intersubunit phosphorylation; Hexamerization

Circadian rhythms—24-h biological oscillations of metabolic and behavioral activities observed ubiquitously in prokaryotes and eukaryotes—are endogenously regulated by circadian clocks. Several clock and clock-related genes from Drosophila, Neurospora, Arabidopsis, mice, and cyanobacteria have been cloned and analyzed [1]. *

Corresponding author. Fax: +81 52 789 4526. E-mail address: [email protected] (M. Ishiura). 1 Present address: Department of Nano-Material Systems, Graduate School of Engineering, Gunma University, 1-5-1 Tenjin, Kiryu, Gunma, 376-8515, Japan. 0006-291X/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2006.07.143

Cyanobacteria are the simplest organisms that exhibit circadian rhythms. We previously cloned and analyzed the kaiABC circadian clock gene cluster in the cyanobacterium Synechococcus sp. strain PCC 7942 (hereafter called Synechococcus), which is essential for the generation of circadian rhythms in cyanobacteria [2]. The gene cluster consists of two operons—kaiA, whose product (probably indirectly) enhances kaiBC promoter activity, and kaiBC, which is repressed (probably indirectly) by KaiC protein itself [2]. KaiC is phosphorylated by itself [3–5], and the phosphorylation is enhanced by interactions with KaiA [4–6]. The KaiA effect is achieved by the KaiA C-terminal

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

clock-oscillator domain (KaiAC) [5,6]. The level of KaiC phosphorylation oscillates circadianly not only in cyanobacterial cells [7,8] but also in an in vitro system consisting of KaiA, KaiB, and KaiC [9]. KaiC also interacts with SasA, a sensory histidine kinase that enhances kaiBC promoter activity [10]. KaiC has a duplicative structure [2,11], and each half has a set of ATPase motifs (Walker motifs A and B, and a pair of deduced catalytic carboxylate Glu residues (named CatEs)) [2,11]. We previously determined the three-dimensional structure of KaiC in the thermophilic cyanobacterium Thermosynechococcus elongatus BP-1 by single particle analysis of cryo-electron microscopic images and demonstrated that the protein has a hexagonal, pot-shaped structure composed of six identical dumbbellshaped subunits [11]. The spherical regions of the dumbbell-shaped structures correspond to the N-terminal and C-terminal domains of KaiC [11], as confirmed by the crystal structure of Synechococcus KaiC [12]. We also demonstrated the different roles of the N-terminal and C-terminal ATPase motifs of KaiC using KaiC proteins carrying mutant ATPase motifs [13]. The KaiC subunit has two types of ATP-binding sites—a high-affinity site in the N-terminal ATPase motifs and a low-affinity site in the C-terminal ATPase motifs. The N-terminal motifs are responsible for hexamerization, whereas the C-terminal motifs are involved in KaiC phosphorylation. These functions, however, have not been confirmed with purified domain proteins. In the present work, we prepared the N- and C-terminal domain proteins of the KaiC subunit, examined their role in hexamerization and their interactions with KaiA and KaiA-enhanced KaiC phosphorylation, and determined the function of those domains. We also discuss the dynamics of KaiC hexamer structure. Materials and methods Plasmid construction. We amplified DNA fragments encoding KaiCN (residues 1–251) and KaiCC (residues 252–518) of T. elongatus KaiC by the polymerase chain reaction (PCR) using T. elongatus genomic DNA as a template, and we inserted them into pGEX-6P-1 (Amersham). We introduced the resulting plasmids (pTe KaiCN and pTe KaiCC) into Escherichia coli DH5a and BL21, and propagated them. We grew the E. coli in Luria–Bertani broth (LB) or Terrific Broth (TB) supplemented with 50 lg/ml ampicillin and on LB plates containing 1.5% agar. We constructed plasmids expressing mutant KaiCs by the in vitro mutagenesis method by PCR. Production and purification of KaiCN and KaiCC in E. coli. We carried out the production and purification of KaiCN and KaiCC in E. coli BL21 as described previously [4] with minor modifications. Briefly, we purified KaiCN and KaiCC, which had been overproduced in E. coli cells as soluble GST-fusion proteins, with glutathione–Sepharose-4B (Amersham). We excised KaiCN and KaiCC from the GST-fusion proteins with PreScission Protease (Amersham) and purified them by ion exchange chromatography on a Q Sepharose HP 26/10 column (Amersham) followed by gel filtration chromatography on a Superdex75 HR10/30 column (Amersham). Each protein eluted as a single symmetric peak at the position expected for a globular protein of about 30 kDa (data not shown). We prepared T. elongatus wild-type KaiA (KaiAWT), the C-terminal clock-oscillator domain of KaiA (KaiAC), wild-type KaiC (KaiCWT), and

865

KaiCCatE2 (carrying a double mutation E318Q and E319Q) [13] as described previously [5,13]. We carried out protein determination using Coomassie brilliant blue (CBB) (Bio-Rad Protein Assay Dye Reagent Concentrate, Bio-Rad) and bovine serum albumin as a standard, SDS–PAGE using mini-gels, NativePAGE using a PhastSystem (Amersham), and gel staining using Simply Blueä SafeStain (Invitrogen) or PhastGel Blue R (Amersham) solution as described previously [4]. Preparation of dephosphorylated KaiCC. We prepared dephosphorylated KaiCC as described previously [4] with minor modifications. Briefly, we incubated KaiCC (4.5 nmol) with 10,000 U k-protein phosphatase (kPPase, New England Biolabs) at 25 °C for 2 h in 1 ml of 50 mM Tris–HCl buffer (pH 7.5) containing 2 mM MnCl2, 0.1 mM EDTA, 5 mM DTT, and 0.01% Brij-35. We separated the dephosphorylated KaiCC and the kPPase by ion-exchange chromatography on a MonoQ HR5/5 column (Amersham). Assay for the oligomerization of KaiCN, KaiCC, and KaiCC S431A&T432A by gel filtration chromatography. We incubated KaiCN, KaiCC, and KaiCCS431A&T432A at 25 °C for 0.5 h in 0.3 ml of 20 mM Tris–HCl buffer (pH 7.5) containing 0.1 mM ATP or 1 mM AMPPNP, 5 mM MgCl2, and 150 mM NaCl, respectively, and then applied them to a Superdex200 HR10/30 column (Amersham) equilibrated with the same buffer, respectively. We monitored proteins by absorbance at 280 nm and estimated the relative amounts of oligomers and a monomer using the elution profiles of the proteins. We used apoferritin (443 kDa), b-amylase (200 kDa), alcohol dehydrogenase (150 kDa), BSA (66 kDa), and carbonic anhydrase (29 kDa) as size markers. Assay for the oligomerization of KaiCN and KaiCC by Native-PAGE. We incubated monomeric KaiCN or KaiCC (180 pmol) or their mixture in the absence of ATP and AMPPNP or presence of 1 mM ATP or 1 mM AMPPNP in 10 ll of 20 mM Tris–HCl buffer (pH 7.5) containing 5 mM MgCl2 at 25 °C for 30 min and then subjected 4 ll-aliquots to NativePAGE. Measurement of the circular dichroism (CD) spectra and thermostability of KaiCN, KaiCC, and KaiCWT. We dissolved KaiCN, KaiCC, and KaiCWT in 20 mM Tris–HCl buffer (pH 7.5) containing 150 mM NaCl, 5 mM MgCl2, and 1 mM AMPPNP at a concentration of 10 lM, and measured their CD spectra at 200–270 nm at 25 °C using a spectropolarimeter equipped with a thermally jacketed quartz cuvette with a 1-mm path length (Jasco JA-720 W, JASCO). Then, we gradually (0.4 °C min1) increased the temperature of the protein solution to 95 °C while monitoring molar ellipticity at 222 nm. We used the deconvolution program of CDNN by Gerhard Bo¨hm [14] to calculate the relative amount of secondary structure (a-helix, parallel, and anti-parallel b-strands, and random coil). Immunoblotting of KaiA–KaiC complexes. We mixed KaiA (KaiAWT or KaiAC) and KaiC (KaiCN or KaiCC) monomers (180 pmol each, Fig. 4) in all combinations and incubated them at 25 °C for 30 min in the absence or presence of 1 mM AMPPNP in 10 ll of 20 mM Tris–HCl buffer (pH 7.5) containing 150 mM NaCl, 5 mM MgCl2, and 1 mM DTT, and subjected 1-ll aliquots of the reaction mixture to Native-PAGE. Similarly, we incubated KaiC (KaiCN or KaiCC) (180 pmol each in monomer, Fig. 5) with various amounts of KaiA (KaiAWT and KaiAC), and subjected aliquots of the mixtures to Native-PAGE. We stained the gels with PhastGel Blue R or immunoblotted them with rabbit anti-KaiA (diluted to 1/ 10,000) and anti-KaiC (diluted to 1/400,000) antisera, as described [4]. Assay for the phosphorylation of KaiCN, KaiCC, KaiCWT, and KaiCCatE2 . We assayed KaiC phosphorylation as described previously [4] with minor modifications. Briefly, we incubated 360 pmol of monomeric KaiCN, KaiCC, KaiCWT, or KaiCCatE2 , in the absence or presence of 360 pmol KaiA dimer, with 50 lM [c-32P]ATP (0.33 nCi/mM, Amersham) in 120 ll of 20 mM Tris–HCl buffer (pH 7.5) containing 5 mM MgCl2, 150 mM NaCl, and 1 mM DTT at 25 °C. We removed 15-ll aliquots from the reaction mixture at the times indicated to terminate the reaction by mixing with SDS-sample buffer and then subjected the samples to SDS–PAGE. We washed the polyacrylamide gels with Milli Q four times, dried up the gels, and then exposed them to PhosphorImager screens (Amersham) for 15 h. We also exposed filter papers containing various amounts of

866

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

[c-32P]ATP spotted as radioactivity standards, together with the gels, to the same PhosphorImager screens. We visualized the radioactivities by STORM 820 PhosphorImager (Amersham) and estimated the amount of 32 P incorporated into KaiC. In this phosphorylation assay, the amounts of the 32P incorporated into KaiCWT and KaiC derivatives (KaiCC and KaiCCatE2 ) may be underestimated because of difficulties in the quantification. The amount of the 32P incorporated into KaiCWT incubated for 8 h in the presence of KaiA was 0.13 pmol per 1 pmol of monomeric KaiC. Assay for the intersubunit phosphorylation between KaiC CatE2 and KaiCC. We incubated KaiCCatE2 monomer with KaiCC monomer (360 pmol each) and analyzed their phosphorylation as described above.

Results Preparation and properties of KaiCN and KaiCC To clarify the functions of the N-terminal and C-terminal domains of the KaiC subunit, we prepared both proteins—KaiCN (N-terminal domain of T. elongatus KaiC, residues 1–251) and KaiCC (C-terminal domain, residues 252–518). The two proteins showed an identity of 21% and a similarity of 32%. Each protein contained a set of ATPase motifs (one Walker motif A and a pair of CatEs and one Walker motif B) (Fig. 1). The KaiCN and KaiCC purified by gel filtration chromatography were monomers (data not shown) and showed a single and a double band, respectively, on SDS–polyacrylamide gels (Fig. 2A, lanes 1 and 2). When we treated the KaiCC with 10,000 U/ml kPPase at 25 °C for 2 h, the upper band disappeared (Fig. 2A, lane 3), suggesting that it was a phosphorylated form of KaiCC (p-KaiCC) similar to KaiCWT [4]. We confirmed the bands corresponding to the phosphorylated and non-phosphorylated forms of KaiCC by phosphorylation experiments using [c-32P]ATP (Fig. 7). p-KaiCC and a

Fig. 1. Alignment of KaiCN (residues 1–251 of T. elongatus KaiC) and KaiCC (residues 252–518). Each protein contains a set of ATPase motifs (Walker motifs A and B, and two catalytic carboxylate Glu residues (CatEs)). The identical and similar residues are shaded yellow and green, respectively.

Fig. 2. Electrophoretograms of KaiCN, KaiCC, and dephosphorylated KaiCC. (A) SDS–PAGE of KaiCN, KaiCC, and dephosphorylated KaiCC. Lanes: M, molecular weight markers; 1, purified KaiCN; 2, purified KaiCC; 3, KaiCC dephosphorylated with k-PPase. (B) Native-PAGE of KaiCC. Lanes: 1, purified KaiCC; 2, KaiCC dephosphorylated with kPPase. Bands: (a,d), p-KaiCC; (b,c), non-phosphorylated form of KaiCC.

non-phosphorylated form of KaiCC were also distinguished by Native-PAGE (Fig. 2B). Oligomerization activities of KaiCN and KaiCC in the presence of ATP or AMPPNP KaiCWT forms a hexamer by binding to triphosphate nucleotides, including ATP and AMPPNP [11]. To determine whether KaiCN or KaiCC was responsible for KaiC hexamerization, we examined the ATP- or AMPPNP-induced oligomerization activities of KaiCN and KaiCC by gel filtration chromatography (Fig. 3). In the presence of 0.1 mM ATP, the apparent molecular weight of the KaiCN was estimated to be 254 (corresponded to hexamer), 154 (tetramer), and 42 (monomer) kDa. Most KaiCN was hexamer (11% of total KaiCN) or tetramer (87%). The apparent molecular weight of KaiCC was estimated to be 120 (trimer or tetramer) and 40 (monomer) kDa. In contrast to KaiCN, we did not detect KaiCC hexamer whereas we detected a significant amount of KaiCC monomer (24% of total KaiCC). In the presence of 1 mM AMPPNP, KaiCN existed as tetramer (18%) and monomer (82%), whereas KaiCC existed as mainly monomer (100%). These results indicate that both KaiCN and KaiCC had a nucleotide-induced oligomerization activity and that the activity of KaiCN was higher than that of KaiCC. Therefore, the N-terminal domain of KaiC is mainly responsible for ATP-induced KaiC hexamerization. In KaiCWT, it was reported that the phosphorylations of Ser431 and/or Thr432 located on KaiC C-terminal domain were not responsible for KaiC hexamerization [15]. To determine whether the phosphorylations of the Ser431 and/or Thr432 are responsible for the hexamerization of KaiCC, we examined the hexamerization of KaiCCS431A&T432A by gel filtration chromatography. While

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

867

Fig. 3. Gel filtration chromatography elution profiles of KaiCN, KaiCC, and KaiCCS431A&T432A. We incubated KaiCN (blue), KaiCC (green), and KaiCCS431A&T432A (black) separately at 25 °C for 30 min in the presence of 0.1 mM ATP, and then applied to a Superdex200 column equilibrated with buffer containing 0.1 mM ATP, respectively (upper panel). We also incubated KaiCN and KaiCC similarly in the presence of 1 mM AMPPNP, and then applied to a Superdex200 column equilibrated with buffer containing 1 mM AMPPNP, respectively (lower panel). We monitored proteins by absorbance at 280 nm. The arrowheads indicate the positions of the molecular size markers.

in the absence of ATP, KaiCCS431A&T432A was eluted as a single symmetric peak corresponding to the size of monomer (data not shown), in the presence of 0.1 mM ATP, it was eluted as oligomer and monomer, indicating that KaiCCS431A&T432A had an oligomerization activity (Fig. 3). Therefore, although the lack of the phosphorylations of Ser431 and Thr432 affected KaiCC oligomerization, the phosphorylations were not essential for the oligomerization. When KaiCWT monomer and hexamer are electrophoresed on Native-polyacrylamide gels, the monomer migrates a long distance whereas the hexamer migrates only a short distance [11]. According to this principle, we analyzed ATP- or AMPPNP-induced oligomerization activities of KaiCN and KaiCC also by Native-PAGE (Fig. 4A). While the KaiCN incubated without ATP and AMPPNP migrated a long distance, the KaiCN incubated with 1 mM ATP or 1 mM AMPPNP migrated varying shorter distances (Fig. 4A), suggesting that KaiCN formed different oligomers. The band pattern of KaiCN was almost the same, irrespective of whether KaiCN was incubated with ATP or AMPPNP (Fig. 4A), indicating that the Native-PAGE assay could not distinguish the difference of oligomerization induced by ATP and AMPPNP that could be detected

Fig. 4. Native-PAGE, CD spectra, and heat-denaturation curves of KaiCN, KaiCC, KaiCN–KaiCC mixture, and KaiCWT. (A) Native-PAGE of the reaction products of KaiCN and KaiCC incubated in the absence or presence of 1 mM ATP or 1 mM AMPPNP. We incubated KaiCN and KaiCC monomers (180 pmol each) separately at 25 °C for 30 min in the absence of ATP and AMPPNP (lanes 1,2), the presence of 1 mM AMPPNP (lanes 4,5) or 1 mM ATP (lanes 7,8). We also incubated an equimolar mixture of KaiCN and KaiCC monomers (180 pmol each) similarly in the absence of ATP and AMPPNP (lane 3) or the presence of 1 mM AMPPNP (lane 6) or 1 mM ATP (lane 9). (B) CD spectra of 10 lM KaiCN (blue), KaiCC (green), KaiCN–KaiCC mixture (orange), and KaiCWT (red) in the presence of 1 mM AMPPNP. (C) heat-denaturation curves of 10 lM KaiCN (blue), KaiCC (green), KaiCN–KaiCC mixture (orange), and KaiCWT (red) in the presence of 1 mM AMPPNP.

by the gel filtration assay described above. The CD spectra of KaiCN and KaiCWT were almost the same, demonstrating that the various oligomers of KaiCN formed in the presence of AMPPNP were not denatured (Fig. 4B). The thermostability assay with the CD spectropolarimeter also demonstrated that KaiCN denatured gradually as the temperature of the solution increased gradually (from 55 to 95 °C) (Fig. 4C), suggesting that KaiCN formed various oligomers in the presence of AMPPNP.

868

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

In contrast, the KaiCC oligomers could not be detected by the Native-PAGE method. KaiCC migrated similarly on Native-polyacrylamide gels irrespective of whether KaiCC was incubated without ATP and AMPPNP or incubated with either 1 mM ATP or 1 mM AMPPNP (Fig. 4A), suggesting that the KaiCC oligomers formed were more unstable than the KaiCN oligomers formed. In consistency with this, even in the presence of 1 mM AMPPNP, KaiCC denatured rapidly at around 45 °C (Fig. 4C). These results suggest that the N-terminal domain of KaiC is more responsible for ATP-induced KaiC hexamerization but is not sufficient for the formation of a stable hexamer. Next, we examined the effects of KaiCC on AMPPNPinduced KaiCN hexamerization. When we incubated KaiCN with KaiCC in the presence of AMPPNP and analyzed the reaction products by Native-PAGE, we did not detect any new bands. The denaturation curve of the KaiCN and KaiCC mixture in the presence of AMPPNP differed greatly from that of KaiCWT (Fig. 4C), indicating that any interactions between KaiCN and KaiCC were weak and suggesting that a peptide bond connecting the N- and C-terminal domains of a KaiC subunit may be essential for the formation of a stable KaiC hexamer. Formation of KaiA–KaiCC complexes KaiCWT directly interacts with KaiAWT in vitro via the KaiA C-terminal clock-oscillator domain (KaiAC) [5]. To determine whether KaiCN or KaiCC is involved in the interactions with KaiA (KaiAWT and KaiAC), we assayed KaiA (KaiAWT and KaiAC)–KaiC (KaiCN and KaiCC) interactions by Native-PAGE. We used AMPPNP as a nucleotide because stable KaiA–KaiC complexes can be

detected in the presence of AMPPNP, but not in the presence of ATP, by Native-PAGE [5]. We first determined the mobility of KaiAWT, KaiAC, KaiCN, and KaiCC after each was incubated in the absence of AMPPNP (Fig. 5, lanes 1–4, bands a, b, c1, c2, d1, and d2). When KaiCC was incubated with KaiAWT or KaiAC in the absence of AMPPNP, we detected new CBB-stained bands that probably corresponded to KaiA–KaiCC complexes (Fig. 5, lanes 6 and 8, bands e and f). We confirmed that the new bands were recognized by both anti-KaiA and anti-KaiC antisera, although the signals obtained with the anti-KaiC antiserum were very faint (Fig. 5, lanes 14, 16, 22, and 24, bands e and f). The KaiCC purified by gel filtration chromatography in the absence of AMPPNP was found to be a monomer because it eluted as a symmetric peak at the position corresponding to monomer size by gel filtration chromatography (data not shown). AMPPNP did not affect its mobility on Native gels (Fig. 4A, lanes 2 and 5, and Fig. 5, lane 4, bands d1, d2, j1, and j2), suggesting that KaiCC was also a monomer in the presence of AMPPNP. Thus, KaiCC monomer directly interacted with KaiAWT and KaiAC, suggesting that KaiA–KaiC interactions occur between the KaiA C-terminal clock-oscillator domain, and the KaiC C-terminal domain and that the interactions do not require hexameric KaiC. On the other hand, we did not detect any interactions between KaiCN (KaiCN oligomers formed in the presence of AMPPNP, as well as the KaiCN monomer, existed in the absence of AMPPNP) and KaiAWT or KaiAC. To further examine KaiAWT–KaiCC and KaiAC–KaiCC interactions, we incubated KaiCC with various amounts of KaiAWT or KaiAC and analyzed the reaction products (Fig. 6). When KaiCC was incubated with KaiAWT in the

Fig. 5. Native-PAGEs and immunoblots of the reaction products of KaiA–KaiC interactions in the absence or presence of AMPPNP. We incubated KaiAWT or KaiAC dimer (90 pmol) with KaiCN or KaiCC monomer (180 pmol) in the absence or presence of AMPPNP at 25 °C for 30 min. Lanes: 1–8, PhastGel Blue R staining of native gels; 9–16, Immunoblots with anti-KaiA antiserum; 17–24, Immunoblots with anti-KaiC antiserum. Bands: (a,g), KaiAWT; (b,h), KaiAC; (c1,c2), KaiCN; (d1,j1), non-phosphorylated form of KaiCC; (d2,j2), p-KaiCC; (e,k), KaiAWT–KaiCC complexes; (f,l), KaiAC– KaiCC complexes; i1–i5, various oligomers of KaiCN.

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

869

Fig. 6. Native-PAGEs and immunoblots of KaiA–KaiC complexes formed in the absence or presence of AMPPNP. We incubated monomeric KaiCN or KaiCC (180 pmol) with various amounts of KaiAWT or KaiAC in the absence or presence of 1 mM AMPPNP at 25 °C for 30 min. Lanes 1–7, PhastGel Blue R staining of native gels; lanes 8–14, Immunoblots with anti-KaiA antiserum; lanes 15–21, Immunoblots with anti-KaiC antiserum. Bands: a, KaiAWT; b1, non-phosphorylated form of KaiCC; b2, p-KaiCC, c, KaiAWT-KaiCC complexes; d, KaiAC; e, KaiAC–KaiCC complexes.

absence of AMPPNP, we detected a new band corresponding to a KaiAWT–KaiCC complex (Fig. 6, top panel, band c) that was recognized by both anti-KaiC and anti-KaiA antisera, although the band c overlapped with the KaiAWT band detected by CBB staining and anti-KaiA antiserum. We also detected a band corresponding to KaiAWT–KaiCC complexes in the presence of AMPPNP (Fig. 6, 2nd panel, band c) and a new band corresponding to a KaiAC–KaiCC complex (Fig. 6, 3rd and 4th panels, band e) in the products of the reaction between KaiAC and KaiCC irrespective of whether AMPPNP was present. In this case, the band was clearly distinguished from KaiAC and KaiCC by CBB staining and immuno-staining with anti-KaiA and anti-KaiC antisera. The results indicate that when they were present together in the reaction mixture, KaiA (KaiAWT or KaiAC) and KaiCC formed KaiA(KaiAWT or KaiAC)–KaiCC complexes. The amount of KaiA–KaiC complex formed in the reaction mixture should increase as the amount of KaiA increases, until a saturation level is reached. In three reactions—those between KaiAWT and KaiCC in both the absence and presence of AMPPNP, and between KaiAC and KaiCC in the presence of AMPPNP—the amount of KaiA–KaiC complex increased as expected (Fig. 6, 1st, 2nd, and 3rd panels). However, in the reaction between KaiAC and KaiCC in the absence of AMPPNP, the amount of KaiAC–KaiCC complex gradually decreased; we cannot explain this. In all the reaction mixtures we examined, the non-phosphorylated form of KaiCC (Fig. 6, band b1) disappeared earlier than the phosphorylated form p-KaiCC (Fig. 6, band b2), and the disappearance was accompanied by an increase in the amount of both KaiA and the KaiA–KaiCC

complex, suggesting that the non-phosphorylated form of KaiCC preferentially interacts with KaiA. This result was consistent with our previous hypothesis that KaiA interacts with KaiC, enhances KaiC phosphorylation, and dissociates from p-KaiC [4]. Phosphorylation activity of KaiCC The phosphorylated and non-phosphorylated forms of KaiC are distinguished by their different mobilities on an SDS–polyacrylamide gel [4]. KaiCC showed a doublet band on SDS–polyacrylamide gels, and the upper band corresponded to p-KaiCC, as described above (Fig. 2A), suggesting that KaiCC had phosphorylation activity as KaiCWT [4]. We examined the phosphorylation activity of KaiC in detail by phosphorylation experiments using [c-32P]ATP (Fig. 7). We carried out the phosphorylation reaction of KaiC at 25 °C because the Tm value of KaiCC in the presence of AMPPNP was 45 °C (Fig. 4C), and more than half of KaiCC was denatured at 50 °C. KaiCWT showed a phosphorylation activity even at 25 °C, and the activity was enhanced about 1.5 times by the presence of KaiA (Fig. 7C). KaiCC also showed a low phosphorylation activity, which was 0.5% of the activity of KaiCWT, suggesting that the N-terminal domain of KaiC assists its C-terminal domain express full KaiC phosphorylation activity. This low activity was enhanced 25 times by KaiA (Fig. 7C). On the other hand, we could not detect any phosphorylation activity in KaiCN even in the presence of KaiA (Fig. 7A). These results indicate that KaiCC, not KaiCN, had phosphorylation activity and that interactions of KaiA with C-terminal domain of KaiC (Figs. 5 and 6) resulted in the enhancement of KaiCC phosphorylation. KaiA

870

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

Fig. 7. Time course of the phosphorylation of KaiCWT, KaiCN, or KaiCC. We incubated monomeric KaiCWT, KaiCN, or KaiCC (360 pmol) with 50 lM (0.33 nCi/mM) [c-32P]ATP at 25 °C for the times indicated in the absence or presence of 360 pmol KaiA. (A) Autoradiograms visualized by STORM 820. (B) SDS–polyacrylamide gels stained with PhastGel Blue R. Bands: a, non-phosphorylated form of KaiCWT; b, p-KaiCWT; c, non-phosphorylated form of KaiCN; d, KaiA; e, non-phosphorylated form of KaiCC; f, p-KaiCC. (C) The amounts of the 32P incorporated into KaiCWT (red triangle), KaiCN (blue square), or KaiCC (green circle) calculated from their autoradiograms. The open and the closed symbols indicate their phosphorylation activities in the absence and presence of KaiA, respectively. Error bars indicate SD (n = 3).

enhanced the phosphorylation activities of KaiCWT and KaiCC 1.5 and 20 times, respectively, suggesting that KaiA complemented partially the lack of the N-terminal domain in KaiCC.

Intersubunit phosphorylation KaiC is phosphorylated in a reaction catalyzed by itself [4,5]. We determined whether KaiC phosphorylation occurs within or between subunits using KaiCCatE2 , which lacks KaiC phosphorylation activity ([13]. and Fig. 8). Although the phosphorylation activity of KaiCCatE2 was not detected in our previous experiments [13], here we could detect a very low phosphorylation activity in KaiCCatE2 , which was 0.2% of the activity of KaiCWT, using a high-sensitive imager. When we incubated monomeric KaiCCatE2 in the presence of ATP with monomeric dephosphorylated (by treatment with k-PPase) KaiCC, we detected an increase and a decrease in the amounts of the 32P incorporated into KaiCCatE2 and KaiCC, respectively (Fig. 8A). This result indicates that the phosphorylation reaction of KaiC occurred between subunits. Discussion

Fig. 8. Phosphorylation of KaiCCatE2 with KaiCC. We incubated KaiCCatE2 with KaiCC monomers (360 pmol each) as described in the legend for Fig. 7. (A) Autoradiograms visualized by STORM 820. The upper and lower bands in the panel of ‘‘Co-incubation’’ represent KaiCCatE2 and KaiCC, respectively. (B) The amounts of the 32P incorporated from [c-32P]ATP into KaiCC (s) and KaiCCatE2 (D) when the two kinds of KaiC were incubated separately and those of KaiCC (d) and KaiCCatE2 (m) when they were incubated together. Error bars indicate SD (n = 3).

Previously, we demonstrated that the two sets of ATPase motifs—one on the N-terminal and one on the C-terminal domain of a KaiC subunit—play different roles [13]. In the present work, using protein preparations of the two domains, we directly demonstrated those roles and we proposed that the KaiC hexagonal pot-shaped structure consists of a rigid part formed by six N-terminal domains and a flexible part formed by six C-terminal domains (Fig. 9A). We explain our model based on the atomic structure of Synechococcus KaiC hexamer [12]. The N-terminal domain of a KaiC subunit is mainly responsible for ATPinduced KaiC hexamerization via KaiCN hexamerization. In N-terminal domain, the b-phosphate group of the ATP bound to the high-affinity ATP-binding site in the domain [13] is held tightly in place by N-terminal Walker

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

871

Fig. 9. Functions of the N-terminal and C-terminal domains of KaiC. (A) Model of a KaiC hexamer consisting of a rigid ring formed by six N-terminal domains and a flexible portion formed by six C-terminal domains whose conformation is regulated by phosphorylation state of each subunit. Each dumbbell along the vertical axis corresponds to a KaiC subunit, and the upper and lower spheres of the dumbbell correspond to the N-terminal and Cterminal domains, respectively. Purple circles represent the location of ATPase motifs. Green spheres represent phosphate groups. Interactions between Nterminal domains induced by the binding of ATP result in KaiC hexamerization [13]. Glu318 and/or Glu319 in the C-terminal domain catalyze release of the c-phosphate group from ATP [13], and the group is transferred to a phosphorylation site (Ser431 and/or Thr432) of the neighboring subunit. (B–E) Models of various conformations of the six C-terminal domains in the KaiC hexamer. Bottom views (views from the C-terminal domain side) are shown. The dashed lines represent the rigid ring formed by the six N-termini. Pink circles numbered I–VI represent the C-terminal domain. The green circles on them represent phosphorylated Ser431 and/or Thr432 residues [15,16]. Possible conformations of a KaiC hexamer when no domain (B), one domain (C), two domains (D), or three domains (E) are phosphorylated are represented.

motif A, and the c-phosphate group interacts with residues Lys224 and Arg226 [12] located on the neighboring subunit. Surface residues Glu14, Glu113 and Glu116, and Glu214 locate on the side, top, and underside of the same subunit, respectively, and their side chains reach over to the neighboring subunit (counterparts are residues Lys85, Arg166 and Arg162, and Arg217 on the neighboring subunit, respectively [12]), locking the hexamer structure. Residue Ile185 of an ATP-binding subunit and residue Phe199 of the neighboring subunit localize near the interface of the two subunits, and Phe199 projects into the ATP-binding subunit, interacting hydrophobically with Ile185 [12]. This interaction is also involved in KaiC hexamerization. In the C-terminal domain, there are interactions of the c-phosphate group of ATP bound to the C-terminal ATP-binding site with residues Lys457 and Arg459 of the neighboring subunit, and Phe419 of the ATP-binding subunit with Phe456 of the neighboring subunit [12]. These interactions are also presumably involved in KaiC hexamerization via KaiCC hexamerization. Thus, N-terminal domain has more multiple intersubunit interactions than C-terminal domain. We believe that the interactions of the c-phosphate group of ATP bound to the N-terminal ATP-binding site with residues Lys224 and Arg226 (described above) are probably crucial for the initiation of ATP-induced KaiC hexamerization, because the affinity for ATP of N-terminal ATP-binding site is higher than that of C-terminal ATPbinding site [13]. The reason ADP and AMP cannot induce KaiC hexamerization [11], presumably, is that they cannot

interact with those residues. We proposed that six N-terminal domains of a KaiC hexamer are responsible for ATPinduced KaiC hexamerization and a rigid ring structure of a KaiC hexamer (Fig. 9A). In a KaiC hexamer, the nucleotide bound to the low-affinity ATP-binding site on the C-terminal domain in the subunit [13] should exist in two states, ATP or ADP, because it is the substrate for KaiC phosphorylation (Fig. 7). When ATP binds to the C-terminal low-affinity site, residues Lys457 and/or Arg459 on the neighboring subunit interact with the c-phosphate group of the ATP [12], and this binding probably strengthens the interactions between the two adjacent subunits (Fig. 9B). When ATP is hydrolyzed to ADP, and ADP occupies the site, it can no longer interact with Lys457 and/or Arg459 and this probably weakens the intersubunit interactions (Fig. 9C–E). We propose that Lys457 and/or Arg459 act as sensors of the phosphorylation state (ATP vs. ADP) of the nucleotide bound to the C-terminal ATP-binding site of each subunit of a KaiC hexamer and that the presence or absence of interactions between ATP and Lys457/Arg459 in the six C-terminal domains in the hexamer causes multiple conformation changes in the C-terminal structure of KaiC hexamer (Fig. 9B–E). In both T. elongatus and Synechococcus, the KaiC phosphorylation sites on the C-terminal domain of a KaiC subunit are Ser431 and Thr432 [13,15,16], which locate at the interface of two adjacent subunits in a KaiC hexamer [12]. Using purified KaiCCS431A&T432A preparations, we demonstrated that the mutant’s deficiency in the phosphor-

872

F. Hayashi et al. / Biochemical and Biophysical Research Communications 348 (2006) 864–872

ylation of Ser431 and Thr432 caused a great reduction in the hexamerization activity of KaiCC (Fig. 3), suggesting that the phosphorylation state of Ser431 and Thr432 is also important for determining the C-terminal structure of KaiC hexamer (Fig. 9). The mode and strength of the interactions between the two adjacent subunits probably depend on how many (0, 1, or 2) residues are phosphorylated. We propose that the phosphorylation state of the six KaiC subunits in a KaiC hexamer, in addition to the ATP-Lys457/Arg459 interactions described above, is responsible for multiple conformation changes—a flexible structure—of the C-terminal domains. We hypothesize that the stepwise changes of the nucleotide state (including binding and unbinding) in each C-terminal domain of a KaiC hexamer, and the resulting stepwise changes in the conformation of the C-terminal domains, are critical for generating circadian clock oscillations. The phosphorylation state of the 12 Ser431 and Thr432 residues in a hexamer may also be important for generating normal clock oscillations because mutant KaiCK294H, which lacks KaiC phosphorylation activity, can generate clock oscillation but causes rhythm irregularity [13]. Recently, in ring-shaped hexameric ATPases such as T7 gp4 helicase [17] and the bacteriophage /12 P4 RNA packing motor [18], the continuous and coordinated conformational changes induced by continuous nucleotide state changes (ATP-binding, hydrolysis, ADP-binding, and dissociation of ADP, including its unbinding) were proposed as the translocation mechanism of ssDNA and ssRNA. Such nucleotide state changes in a KaiC hexamer may be also important for generating normal clock oscillations. To elucidate the clock oscillation mechanism, it will be necessary to demonstrate circadian changes in the conformation, nucleotide state, and phosphorylation state of each subunit coupled to circadian changes in the conformation of a KaiC hexamer. Acknowledgments We thank Satoko Ogawa and Kumiko Tanaka for technical support and Dr. Miriam Bloom (SciWrite Biomedical Writing & Editing Services) for professional editing. This work was supported by grants from the Japanese Ministry of Education, Science and Culture (MEXT), Program for Promotion of Basic Research Activities for Innovative Biosciences, Research for the Future of Japan Society for the Promotion of Science ‘‘Novel Gene Function Involved in Higher-Order Regulation of Nutrition-Storage in Plants,’’ ’’Ground-based Research Announcement for Space Utilization’’ promoted by the Japan Space Forum, ‘‘National Project on Protein Structural and Function Analyses’’ promoted by MEXT, and ‘‘Promoting Cooperative Research Project’’ promoted by the Aichi Science and Technology Foundation (to M. I.). The Division of Biological Science, Graduate School of Science, Nagoya University, was supported by a 21st Century Center of Excellence grant from MEXT.

References [1] J. Dunlap, Molecular bases for circadian clocks, Cell 96 (1999) 271– 290. [2] M. Ishiura, S. Kutsuna, S. Aoki, H. Iwasaki, C.R. Andersson, A. Tanabe, S.S. Golden, C.H. Johnson, T. Kondo, Expression of a gene cluster kaiABC as a circadian feedback process in cyanobacteria, Science 281 (1998) 1519–1523. [3] T. Nishiwaki, H. Iwasaki, M. Ishiura, T. Kondo, Autophosphorylation of the clock protein KaiC as a circadian timing process of cyanobacteria, Proc. Natl. Acad. Sci. USA 96 (2000) 495–499. [4] F. Hayashi, H. Itoh, M. Fujita, R. Iwase, T. Uzumaki, M. Ishiura, Stoichiometric interactions between cyanobacterial clock proteins KaiA and KaiC, Biochem. Biophys. Res. Commun. 316 (2004) 195– 202. [5] T. Uzumaki, M. Fujita, T. Nakatsu, F. Hayashi, H. Shibata, N. Itoh, H. Kato, M. Ishiura, Role of KaiA functional domains in circadian rhythms of cyanobacteria as revealed by crystal structure, Nat. Struct. Mol. Biol. 11 (2004) 623–631. [6] S.B. Williams, I. Vakonakis, S.S. Golden, A.C. LiWang, Structure and function from the circadian clock protein KaiA of Synechococcus elongatus: a potential clock input mechanism, Proc. Natl. Acad. Sci. USA 99 (2002) 15357–15362. [7] H. Iwasaki, T. Nishiwaki, Y. Kitayama, M. Nakajima, T. Kondo, KaiA-stimulated KaiC phosphorylation in circadian timing loops in cyanobacteria, Proc. Natl. Acad. Sci. USA 99 (2002) 15788–15793. [8] Y. Xu, T. Mori, C.H. Johnson, Cyanobacterial circadian clockwork: roles of KaiA, KaiB and the kaiBC promoter in regulating KaiC, EMBO J. 22 (2003) 2117–2126. [9] M. Nakajima, K. Imai, H. Ito, T. Nishiwaki, Y. Murayama, H. Iwasaki, T. Oyama, T. Kondo, Reconstitution of circadian oscillation of cyanobacterial KaiC phosphorylation in vitro, Science 308 (2005) 41415. [10] H. Iwasaki, S.B. Williams, Y. Kitayama, M. Ishiura, S.S. Golden, T. Kondo, A KaiC-interacting sensory histidine kinase, SasA, necessary to sustain robust circadian oscillation in cyanobacteria, Cell 101 (2000) 223–233. [11] F. Hayashi, H. Suzuki, R. Iwase, T. Uzumaki, A. Miyake, J.R. Shen, K. Imada, Y. Furukawa, K. Yonekura, K. Namba, M. Ishiura, ATPinduced hexameric ring structure of the cyanobacterial circadian clock protein KaiC, Genes Cells 8 (2003) 287–296. [12] R. Pattanayek, J. Wang, T. Mori, Y. Xu, C.H. Johnson, M. Egli, Visualizing a circadian clock protein: crystal structure of KaiC and functional insights, Mol. Cell 15 (2004) 375–388. [13] F. Hayashi, N. Itoh, T. Uzumaki, R. Iwase, Y. Tsuchiya, H. Yamakawa, M. Morishita, K. Onai, S. Itoh, M. Ishiura, Roles of two ATPase-motif-containing domains in cyanobacterial circadian clock protein KaiC, J. Biol. Chem. 279 (2004) 52331–52337. [14] G. Bo¨hm, R. Muhr, R. Jaenicke, Quantitative analysis of protein far UV circular dichroism spectra by neural networks, Protein Eng. 5 (1992) 191–195. [15] T. Nishiwaki, Y. Satomi, M. Nakajima, C. Lee, R. Kiyohara, H. Kageyama, Y. Kitayama, M. Temamoto, A. Yamaguchi, A. Hijikata, M. Go, H. Iwasaki, T. Takao, T. Kondo, Role of KaiC phosphorylation in the circadian clock system of Synechococcus elongatus PCC 7942, Proc. Natl. Acad. Sci. USA 101 (2004) 13927–13932. [16] Y. Xu, T. Mori, R. Pattanayek, S. Pattanayek, M. Egli, C.H. Johnson, Identification of key phosphorylation sites in the circadian clock protein KaiC by crystallographic and mutagenetic analyses, Proc. Natl. Acad. Sci. USA 101 (2004) 13933–13938. [17] M.R. Singleton, M.R. Sawaya, T. Ellenberger, D.B. Wigley, Crystal structure of T7 gene 4 ring helicase indicates a mechanism for sequential hydrolysis of nucleotides, Cell 101 (2001) 589–600. [18] E.J. Mancini, D.E. Kainov, J.M. Grimes, R. Tuma, D.H. Bamford, D.I. Stuart, Atomic snapshots of an RNA packaging motor reveal conformational changes linking ATP hydrolysis to RNA translocation, Cell 118 (2004) 743–755.