NEWS AND VIEWS Visualizing a biological clockwork’s cogs Carl H Johnson & Martin Egli Recent structural studies of KaiA and KaiB, two bacterial biological clock proteins, mark the beginning of a new phase in the analysis of circadian clock mechanisms. Have you ever looked inside an old-fashioned I clock or wristwatch to see its oscillating gears II and escapement in an attempt to infer how it KaiC monomers keeps time? Biological clock researchers have P-KaiC hexamer the same goal when it comes to the self- ATP + KaiA dimer sustained biochemical oscillators known as "dome" circadian clocks. These biological clocks regu- uto-phosphorylation late an enormous variety of processes, ranging A from cell division to sleeping and waking. I "waist" uto-dephosphorylation Their remarkable properties include temp- II A Complex of stable P-KaiC hexamer + KaiA erature compensation, a time constant of KaiC hexamer ∼24 hours and high precision, but these prop- + KaiB dimer erties are difficult to explain by the currently http://www.nature.com/natstructmolbiol known biochemical reactions. The ultimate Dephosphorylation explanation for the mechanism of these unusual oscillators will require characterizing Rephosphorylation the structures, functions and interactions of KaiC hexamer Complex of unstable + KaiA + KaiB P-KaiC hexamer + the molecular components of circadian clocks. KaiA + KaiB In the past 20 years, many clock components have been identified in a variety of organisms, Other components but the structures of these components have (SasA, others?) been unknownthat is, until now. Four recent reports, including that of Uzumaki Figure 1 A hypothetical model for establishing Active KaiABC 1 et al. in this issue of Nature Structural & an active multiprotein clock complex in complex (helicase, Molecular Biology, reveal for the first time the cyanobacteria. DNA pump, other function?) three-dimensional structure of two essential 1–4 © 2004 Nature Publishing Group circadian clock proteins, KaiA and KaiB . The Kai proteins come from the simplest cells that are known to exhibit circadian phe- in both phosphorylated and non-phosphory- KaiA is very unlikely to act as a true receiver nomena, the prokaryotic cyanobacteria, lated forms11–13, and its phosphorylation domain, because it lacks the highly conserved where genetic and biochemical studies have status is correlated with the period of the aspartyl residues of this family14. Therefore, been productive5. A mutational analysis from clock in vivo13. In vitro, KaiC can both auto- the N-terminal domain of KaiA has been the cyanobacterium Synechococcus elongatus phosphorylate11 and auto-dephosphory- labeled a ‘pseudo-receiver’ domain that sug- revealed that its circadian system is regulated late13. KaiA and KaiB modulate the gests an alteration of function3,14. Little is by a cluster of three essential clock genes, phosphorylation status of KaiC in vitro and known about the exact function of the KaiA kaiA, kaiB and kaiC6. A fourth gene encodes in vivo: KaiA enhances KaiC phosphorylation N-terminal domain, as it does not promote the histidine kinase SasA that is not essential (and/or inhibits its dephosphorylation), KaiC autophosphorylation in vitro1,14. for rhythmicity but significantly enhances the whereas KaiB antagonizes the effects of Furthermore, KaiA from another cyanobac- robustness of the oscillation7. We already KaiA12–15. Addition of ATP to purified KaiC terium (Anabaena sp. PCC 7120) doesn’t even know some things about the interactions and in vitro initiates KaiC phosphorylation and have the N-terminal part of KaiA1,4, suggesting modifications of the proteins encoded by the also stimulates the formation of the hexa- that it might be dispensable. However, muta- kai genes (Fig. 1). KaiA, KaiB and KaiC inter- meric KaiC ring complex16,17. This last find- tions in the N-terminal domain of KaiA from act with each other8,9 to form large complexes ing is consistent with the observation of KaiC Synechococcus elongatus lengthen the circadian in vivo, with KaiC at the core10. KaiC can exist in high-molecular-mass complexes in vivo10. period, suggesting that it does play an impor- What have the structural studies added to tant role18. Indeed, Williams and co-workers14 C.H.J. is in the Department of Biological this picture? Let’s begin with KaiA. KaiA has propose that this pseudo-receiver domain Sciences and M.E. is in the Department of two major domains1,3,14. The N-terminal transduces input signals to the clock complex. Biochemistry, Vanderbilt University, Nashville, domain of KaiA shares structural, but not The C-terminal domain of KaiA is the Tennessee 37235, USA. sequence, similarity with the receiver domain ‘business end’ in relation to KaiC auto- e-mail: [email protected] or of bacterial two-component response regula- phosphorylation. The isolated C-terminal [email protected] tors14. Nonetheless, the N-terminal domain of domain is completely active in promoting 584 VOLUME 11 NUMBER 7 JULY 2004 NATURE STRUCTURAL & MOLECULAR BIOLOGY NEWS AND VIEWS KaiC auto-phosphorylation in vitro1,14, and it Therefore, there are two topologically distinct filamentous Anabaena. Even though the KaiA is the part that binds to KaiC1. The structure and separated regions of KaiC with which and KaiB sequences for these species align of the C-terminal domain has been solved by KaiA interacts. These interactions almost cer- well (with the exception of the N-terminal both NMR2 and crystallography1,3,4. The tainly underlie the influence of KaiA on the region of KaiA that varies), their biochem- agreement among various KaiA structures phosphorylation state of KaiC. Notably, only istry might not be directly comparable. That determined by two different techniques and one KaiA dimer is enough to enhance the caveat aside, these structures allow us to by the several groups (and from three different phosphorylation of one KaiC hexamer to an refine our analyses of circadian mechanisms. species) is excellent. The C-terminal domain almost saturated level19. Considering that As with all good science, however, the new adopts a four-helix bundle that dimerizes and there is a six- to seven-fold excess of KaiC results raise many new questions. What is can interact with KaiC. The full-length struc- hexamers over KaiA dimers in vivo10, how can the three-dimensional configuration of the ture of KaiA shows that the N-terminal so few KaiA dimers maintain the phosphory- entire KaiABC complex? Are other proteins domains are ‘swapped’ with respect to the lation status of the KaiC hexamers? Probably besides SasA also members of this complex? C-terminal domains in the KaiA dimer3. part of the answer is that not all the KaiC is What is the enzymatic activity of the KaiABC Although Uzumaki et al.1 suggest that KaiA phosphorylated in vivo12. complex? The sequence similarity between may have three domains1, the crystallographic Less is known about the structure and inter- KaiC and DNA helicases/recombinases sug- analysis of the complete KaiA favors the inter- actions of KaiB. The crystal structure of the gested a hypothesis that the KaiABC complex pretation that there are two predominant cyanobacterium Anabaena KaiB4 now adds to mediates rhythmic changes in chromosomal domains connected by a linker whose absence our understanding. Like KaiA, KaiB from topology that lead to global orchestration of has functional consequences3. Anabaena purifies and crystallizes as a dimer. gene expression5. The structural studies of How does this C-terminal domain interact An important question to be answered from the Kai proteins will undoubtedly provide with the core KaiC hexamer? KaiC is an inter- the structural studies is how KaiB antagonizes useful tests to distinguish the various possi- nally duplicated protein, with its two halves the effects of KaiA on KaiC phosphorylation. ble mechanisms of the circadian clockwork, named KaiCI and KaiCII8. Taniguchi et al.9 Garces et al.4 note that there are some com- including its precision, 24-hour time con- http://www.nature.com/natstructmolbiol found that there were two widely separated mon surface features between KaiA and KaiB stant and temperature compensation. regions in the KaiC sequence—one in each from Anabaena and suggest that KaiA and half—that interacted with KaiA. On the basis KaiB may compete for a common binding site 1. Uzumaki, T. et al. Nat. Struct. Mol. Biol. 11, 623– 3 631 (2004). of this observation and their assumption that on KaiC. On the other hand, Ye et al. pro- 2. Vakonakis, I. et al. Proc. Natl. Acad. Sci. USA 101, KaiA and KaiC will interact at a single site, posed on the basis of KaiA’s structure that the 1479–1484 (2004). these authors proposed that the two inter- action of KaiB might be to interact directly 3. Ye, S., Vakonakis, I., Ioerger, T.R., LiWang, A.C. & 3 Sacchettini, J.C. J. Biol. Chem. 279, 20511–20518 action regions wrapped back on themselves in with KaiA. In particular, Ye et al. proposed (2004). the KaiC monomer so that they were in that KaiB binds to a site in the C-terminal 4. Garces, R.G., Wu, N., Gillon, W. & Pai, E.F. EMBO J. 23, 1688–1698 (2004). approximately the same relative spatial posi- domain of KaiA that is either ‘open’ (available 5. Johnson, C.H. Curr. Issues Mol. Biol. 6, 103–110 2 tion. Vakonakis et al. amalgamated the sug- for KaiB binding) or ‘closed’ (masked by the (2004). gestion of Taniguchi et al.9 into a model using N-terminal domain of KaiA). Therefore, the 6. Ishiura, M. et al. Science 281, 1519–1523 (1998). 7. Iwasaki, H. et al. Cell 101, 223–233 (2000). electron micrographs of the KaiC hexamer N-terminal pseudo-receiver domain was pro- 8. Iwasaki, H., Taniguchi, Y., Kondo, T. & Ishiura, M. and the NMR structure of KaiA. In this posed to allosterically regulate the binding of EMBO J.