ATP hydrolysis by KaiC promotes its KaiA binding in the cyanobacterial circadian clock system

ATP hydrolysis in the KaiC hexamer triggers the exposure of its C-terminal segments into the solvent so as to capture KaiA, providing mechanistic insights into the circadian periodicity regulation.


Introduction
Endogenous, entrainable oscillation with periods of~24 h, known as the circadian rhythm, is found in many organisms. In cyanobacteria, the circadian rhythm is regulated by three clock proteins  (i.e., KaiA, KaiB, and KaiC) that autonomously undergo periodic assembly and disassembly irrespective of transcriptional and translational feedback systems (Tomita et al, 2005;Rust et al, 2007). KaiC is an AAA+ ATPase consisting of CI and CII domains and forms a hexameric ring structure, which harbors six ATP-binding sites in the CI ring and another six ATP-binding sites in the CII ring (Pattanayek et al, 2004). The KaiC hexamer experiences autophosphorylation and dephosphorylation cycles in a 24-h period through interactions with KaiA and KaiB in the presence of ATP . Two phosphorylation sites at Ser431 and Thr432 in the KaiC CII domain (simply denoted as S and T, respectively) go through a cycle as follows: S/T → S/pT → pS/pT → pS/T → S/T, where "p" represents the phosphorylated residue (Nishiwaki et al, 2004). Phosphorylation oscillation is controlled via the interplay among KaiA, KaiB, and KaiC. KaiA and KaiB interact with the CII and CI domains of KaiC, respectively, so that its phosphorylation is up-regulated/downregulated. Dephosphorylated KaiC interacts with KaiA, followed by an increase in the KaiC phosphorylation (Iwasaki et al, 2002;Williams et al, 2002). By contrast, KaiB interacts with phosphorylated KaiC and, thereby, accelerates its dephosphorylation (Kitayama et al, 2003;Xu et al, 2003).
A series of phospho-mimicking KaiC mutants have been widely used for characterizing the Kai protein complex formation in a phosphorylation state-dependent manner (Pattanayek et al, 2009(Pattanayek et al, , 2011Lin et al, 2014;Sugiyama et al, 2016;Tseng et al, 2017;Mori et al, 2018). In these studies, it has been suggested that the interaction of KaiC with KaiB depends not only on the KaiC phosphorylation states but also on the nucleotide states in KaiC. For example, a phosphorylation-mimicking KaiC mutant was reactive to KaiB in the presence of ATP but loses the KaiB-binding affinity when adenylyl imidodiphosphate AMPPNP, a nonhydrolyzable ATP analog, was used instead of ATP (Phong et al, 2013;Mukaiyama et al, 2018). However, little is known about the relationship between the nucleotide state of KaiC and its KaiA-binding activity. Herein, we address this issue through characterizing ATP hydrolysis dependence of the KaiA-KaiC interaction by native mass spectrometry (MS), providing mechanistic insights into the binding of KaiA to the C-terminal segment of KaiC based on nuclear magnetic resonance (NMR) data. Kondo, 2012;Nishiwaki-Ohkawa et al, 2014). To control the phosphorylation states, we used two KaiC mutants, KaiC DD and KaiC AA (in which Ser431 and Thr432 were both substituted with aspartate and alanine residues, respectively), mimicking the phosphorylated and dephosphorylated states of KaiC, respectively. It has been reported that KaiC AA shows enhanced complex formation with KaiA in comparison with KaiC DD (Lin et al, 2014;Tseng et al, 2014). To examine the possible dependence of KaiA-KaiC interaction on KaiCbound nucleotide states, ATP and its nonhydrolyzable analog, adenylyl imidodiphosphate (AMPPNP), were used for the formation of the mutated KaiC hexamers, which were subjected to native MS analysis.
The native MS data confirmed that both KaiC AA and KaiC DD formed hexamers upon the addition of either ATP or AMPPNP (Figs 1A and B, S1A, and B). Whereas the KaiC AA hexamer mediated by AMPPNP had a constant molecular mass of 353,857 ± 10 D holding 12 AMPPNP molecules ( Fig 1A and Table 1), the ATP-mediated KaiC AA hexamer preincubated for 5 h in the presence of 1 mM ATP exhibited two series of ion peaks ( Fig 1B). The major peaks corresponded to the KaiC AA hexamer containing seven ATP and five ADP molecules, whereas the minor peaks corresponded to the KaiC AA hexamer containing seven ATP and three ADP molecules (Table 2). Native MS data of AMPPNP-or ATP-mediated KaiC AA hexamer incubated for 5 h in the presence of ATP and ADP at varying ratios, indicated that the nucleotide states of KaiC AA did not depend on the external ATP/ADP condition ( Fig S2). These data indicate that the KaiC AA hexamer mediated by ATP spontaneously became asymmetric in terms of nucleotide state and that the resultant ADP molecules were releasable. By contrast, the KaiC DD hexamer held at least 11 nucleotides, suggesting its low ATPase activity, which is consistent with a previous report that the ATPase activity of KaiC AA is six times higher than that of KaiC DD (Mutoh et al, 2013). Furthermore, we conducted tryptic fragmentation into CI and CII of the KaiC AA hexamer, followed by native MS analysis. The results revealed that the CI hexamer retained six nucleotides as prehydrolyzed ATP molecules, suggesting that the observed ATP hydrolysis exclusively occurred in the CII ring ( Fig S3). This was supported by the observation that a KaiC AA mutant in which a catalytic glutamate (Glu77) in the CI ATPase domain (designated as KaiC AA/E77Q ) was substituted by glutamine exhibited virtually identical properties in terms of the nucleotide state and the KaiA binding to those of the KaiC AA hexamer ( Fig S4 and Table 2).
Upon the addition of KaiA, both KaiC AA and KaiC DD formed~420-kD complexes, indicating that two KaiA molecules bind one KaiC hexamer (Fig 1C and D,and Tables 1 and 2,and Fig S1C and D). This stoichiometry is consistent with the previously reported smallangle X-ray scattering and electron microscopy data, indicating that the KaiA dimer is tethered to the KaiC hexamer through its flexible C-terminal region (Pattanayek et al, 2006(Pattanayek et al, , 2011. Even in the presence of excess amounts of KaiA, the KaiA-KaiC complex were formed primarily in a 2:6 stoichiometry and in a 4:6 stoichiometry as minor complex ( Fig S5). The occurrence of the complex significantly depended on the nucleotide state of KaiC and its phosphorylation mutation. The ATP-mediated KaiC AA hexamer almost exclusively formed the complex with KaiA, whereas the complex formation was compromised in the AMPPNP-mediated KaiC AA hexamer and both the AMPPNP-and ATP-mediated KaiC DD hexamers. These data indicate that the nonphosphorylated KaiC hexamer becomes most reactive with KaiA after ATP hydrolysis. Interestingly, in most cases, the KaiA-KaiC complexes lost one or two nucleotides (Tables 1 and 2). Noteworthy, the AMPPNP-mediated KaiC AA hexamer in complex with the KaiA dimer lacked one and two nucleotides, strongly suggesting that KaiA binding promotes nucleotide release from KaiC.
On the basis of our data and previous reports that KaiA binding stimulates the ATPase activity of KaiC (Terauchi et al, 2007;Murakami et al, 2008), we conclude that ATP hydrolysis of KaiC promotes its interaction with KaiA, which is coupled with ADP release and, in turn, enhances the KaiC ATPase activity.

KaiC ATP hydrolysis triggers exposure of its KaiA-binding segments
Using NMR-based structural characterization, we attempted to address how ATP hydrolysis in KaiC affects its binding to KaiA. Because of the fact that KaiA was reported to interact with the A-loop (Glu487-Ile497) and C-tail (Ser498-Ser518) of KaiC (Vakonakis & LiWang, 2004;Pattanayek et al, 2006Pattanayek et al, , 2011Pattanayek & Egli, 2015), we focused on these C-terminal segments. We compared the 1 H-15 N HSQC spectra of KaiC AA and its mutant with the deletion of the C-terminal region 487-518 in the presence of AMPPNP or ATP ( Fig  S6). Despite its high molecular mass (347 kD), the KaiC AA hexamer yielded~30 observable peaks originating from the backbone amide groups with high flexibility, most of which disappeared in the spectrum of the truncated mutant and, therefore, were assigned to the C-terminal region. The signals derived from the C-terminal region were assigned to individual amino acid residues located in the A-loop and C-tail by a series of 1 H/ 13 C/ 15 N multinuclear NMR experiments (Fig 2A and B). In the KaiC AA hexamer mediated by AMPPNP, these peaks demonstrated enhanced peak broadening in comparison with those from the ATP-mediated hexamer and, importantly, Gly488, Ile489, and Ile497 yielded no observable peak, indicating that its A-loop became mobile upon ATP hydrolysis (Fig 2D and F). Furthermore, the peaks originating from the C-terminal segments, including Gly488 and Ile489 in the A-loop, were attenuated or even disappeared upon the addition of KaiA in the presence of ATP, indicating that both the A-loop and the C-tail are involved in the interaction with KaiA (Fig 2C and E). This result is consistent with the X-ray crystal structure, which shows that the full-length Synechococcus elongatus KaiA homodimer is in complex with two KaiC C-tail peptides (Asp500-Ser519) (Pattanayek & Egli, 2015) and also that the NMR solution structure of the C-terminal domain of Thermosynechococcus elongatus KaiA (residues 180-283) interacting with a C-terminal peptide includes the A-loop and C-tail (residues 488-518) of KaiC. In the latter case, the KaiC-derived peptide binds a KaiA surface through its eight hydrophobic side chains in the A-loop (Ile490, Thr493, and Thr495) and in the C-tail (Val499, Leu505, Ala506, Ile508, and Met512) (Vakonakis & LiWang, 2004).
To date, the crystal structures of the wild-type KaiC hexamer and its phosphorylated mutants have been solved as AMPPNP-bound states (Pattanayek et al, 2004(Pattanayek et al, , 2009(Pattanayek et al, , 2014. In these crystal structures, the A-loop forms a U-shaped hairpin, which is accommodated in a cleft proximal to the nucleotide-binding site and is followed by the C-tail, which is exposed to the solvent and gives no interpretable electron density for the segment beyond E504 (Fig 2F) (Pattanayek & Egli, 2015). Under this circumstance, the interaction of KaiA with the KaiC hexamer is compromised because of the inaccessibility of the A-loop. Our MS and NMR data indicate that, upon ATP hydrolysis, the A-loop is released and becomes reactive with KaiA, leading to the enhanced KaiA-KaiC interaction.
The crystal structure revealed that AMPPNP molecules are located at the intersubunit interfaces of the KaiC hexamer (Pattanayek et al, 2014). In the CII ring, a loop comprising residues 415-433 (termed "422-loop") makes contact with both AMPPNP and the A-loop (Fig 2F) (Vakonakis & LiWang, 2004;Kim et al, 2008;Egli et al, 2013). Based on our findings in conjunction with the crystallographic data, we propose a mechanistic model in which ATP hydrolysis in the CII ring triggers a conformational rearrangement of the 422-loop, resulting in "leaping out" of the A-loop so as to interact with KaiA. Reciprocally, it is possible that KaiA pulls out the A-loop and, thereby, causes a microenvironmental rearrangement surrounding the nucleotide-binding site of the KaiC CII ring, facilitating ADP release from it. This is supported by a recently reported molecular dynamics simulation of the KaiC hexamer (Hong et al, 2018). The present study experimentally reveals that ATP hydrolysis in the KaiC hexamer triggers the exposure of its C-terminal segments into the solvent so as to capture KaiA, which in turn facilitates ADP release. These findings imply that the ATPase-dependent KaiA interaction promotes ADP/ATP turnover on KaiC, leading to the upregulation of its autophosphorylation (Fig 3). Our findings provide mechanistic insights into the circadian periodicity mediated by the unique AAA+ ATPase.

Protein expression and purification for native MS
KaiA and KaiC mutants originating from thermophilic cyanobacteria, T. elongatus BP-1, were expressed in Escherichia coli and purified as previously described (Ishii et al, 2014). The expression plasmids of the KaiC mutants (KaiC AA , KaiC AA/E77Q , and KaiC DD ) were also constructed according to a previous study (Murakami et al, 2012;Mutoh et al, 2013).

Protein expression and purification for NMR spectroscopy
The expression plasmids of KaiC AA and its mutant with the deletion of the C-terminal segment 487-518 were constructed according to a previous study (Ishii et al, 2014). For NMR analyses, the protein was expressed in M9 minimal medium containing [ 15 N]NH 4 Cl and/or [ 13 C]glucose. The protein was purified according to a previous report (Ishii et al, 2014). The NMR samples were prepared by dissolving the mutated KaiC (53 μM final concentration) in 10% D 2 O containing 20 mM Tris-HCl (pH 7.0), 150 mM NaCl, 5 mM MgCl 2 , 0.5 mM EDTA, and 1 mM DTT in the presence of 1 mM ATP or AMPPNP.

Native MS analyses
For KaiC hexamerization, the purified monomeric KaiC mutants (KaiC AA , KaiC AA/E77Q , and KaiC DD ) (20 μM) were incubated in the presence of 1 mM nucleotide (ATP or AMPPNP) for 30 min on ice. The KaiC solutions were incubated at 37°C for 5 h in the presence of ATP or AMPPNP. In the examination of possible effects of the external ATP/ADP variation, the AMPPNP-or ATP-mediated KaiC AA hexamer was incubated in the presence of ATP and ADP at varying ratios. Subsequently, these KaiC solutions, in the absence or presence of KaiA (10 μM), were buffer-exchanged into 150 mM ammonium acetate (pH 6.8) by passing the proteins through a Bio-Spin 6 column (Bio-Rad) and were then immediately subjected to nanoflow electrospray ionization MS using gold-coated glass capillaries made in-house (~2-5 μl sample loaded per analysis). For the tryptic fragmentation into CI and CII of the KaiC AA hexamer, KaiC AA incubated at 37°C in the presence of 1 mM ATP for 5 h was bufferexchanged into 150 mM aqueous ammonium acetate and then digested with 0.02 mg/ml trypsin for up to 60 min, followed directly by nanoflow electrospray ionization MS measurements. Spectra Relative quantity of two ion series are shown.  ) and (E), the residues that yielded no observable peaks under the AMPPNP condition are highlighted in red, whereas the asterisks indicate the proline residues and residues whose chemical shift perturbation data could not be obtained because of severe peak overlapping. (F) Crystal structure of two KaiC protomers in cartoon and surface representation, respectively, in the KaiC homohexameric ring mediated by AMPPNP (PDB ID code: 4O0M). In the crystal structure, the C-terminal region comprises a U-shaped A-loop (Glu487-Ile497) (orange) and a solvent-exposed C-tail (S498-S518), in which only the Ser498-Glu504 part (green) was modeled. The three residues (i.e., Gly488, Ile489, and Ile497) located in the A-loop, whose HSQC peaks were unobserved under the AMPPNP condition, are colored blue. The A-loop and AMPPNP molecule (red) are mediated by a loop comprising residues 415-430 (termed 422-loop, magenta).
were recorded on a SYNAPT G2-Si HDMS mass spectrometer (Waters) in positive ionization mode at 1.33 kV with a 150 V sampling cone voltage and source offset voltage, 0 V trap and transfer collision energy, and 5 ml/min trap gas flow. Spectra were calibrated using 1 mg/ml cesium iodide and analyzed using MassLynx software (Waters) (Sugiyama et al, 2016).

NMR analyses
All NMR spectra were acquired at 37°C using Bruker DMX-500 and Bruker AVANCE 800 US spectrometers. The chemical shifts of KaiC peaks were assigned on the basis of 2D 1 H-15 N HSQC, 3D 1 H-13 C-15 N HSQC, 3D HNCO, and 3D HNCA spectral data in the presence of AMPPNP or ATP. The purified monomeric KaiC with isotope labeling (53 μM) was buffer-exchanged into 20 mM Tris buffer (pH 7.0) containing 150 mM NaCl, 5 mM MgCl 2 , 0.5 mM EDTA, 1 mM nucleotide (ATP or AMPPNP), and 10% (vol/vol) D 2 O and was then incubated at 4°C for 30 min for hexamerization. The KaiC hexamers formed by AMPPNP were used for the NMR assignment experiment. After 9 h of incubation at 37°C, the KaiC solutions were subjected to 1 H-15 N HSQC measurements in the absence and presence of a KaiA dimer (53 μM). All NMR data were processed using NMRPipe software and analyzed with Sparky software (Delaglio et al, 1995;Lee et al, 2009).