Setting the pace of the Neurospora circadian clock by multiple independent FRQ phosphorylation events Chi-Tai Tanga,1, Shaojie Lia,b,1, Chengzu Longc, Joonseok Chaa, Guocun Huanga, Lily Lia,2, She Chend,3 and Yi Liua,4 aDepartment of Physiology, The University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390; bKey Laboratory of Systematic Mycology and Lichenology, Institute of Microbiology, Chinese Academy of Sciences, Beijing 100080, China; cNational Institute of Biological Sciences, 7 Life Science Park Road, Changping District, Beijing 102206, China; and dCollege of Life Science, Beijing Normal University, Beijing 100875, China Communicated by Melanie H. Cobb, University of Texas Southwestern Medical Center, Dallas, TX, May 5, 2009 (received for review April 15, 2009) Protein phosphorylation plays essential roles in eukaryotic circa- that FRQ is phosphorylated at multiple sites. Although some dian clocks. Like PERIOD in animals, the Neurospora core circadian putative FRQ phosphorylation sites and domains have been protein FRQ is progressively phosphorylated and becomes exten- identified by deletion and mutation analyses, no FRQ phosphor- sively phosphorylated before its degradation. In this study, by ylation sites have been confirmed in vivo (22, 24, 25, 27). How using purified FRQ protein from Neurospora, we identified 43 in FRQ phosphorylation is temporally regulated by different ki- vivo FRQ phosphorylation sites by mass spectrometry analysis. In nases is not known. addition, we show that CK-1a and CKII are responsible for most To understand the function and regulation of FRQ phosphor- FRQ phosphorylation events and identify an additional 33 phos- ylation, we identified 43 in vivo phosphorylation sites by MS phorylation sites by in vitro kinase assays. Whole-cell metabolic analyses using purified FRQ from Neurospora. In vitro kinase isotope labeling and quantitative MS analyses suggest that circa- assays showed that CK-1a and CKII are responsible for most dian oscillation of the FRQ phosphorylation profile is primarily due FRQ phosphorylation, and they identified 33 additional putative to progressive phosphorylation at the majority of these newly FRQ phosphorylation sites. By performing whole-cell metabolic discovered phosphorylation sites. Furthermore, systematic muta- isotope labeling and quantitative MS analyses, we demonstrate tions of the identified FRQ phosphorylation sites led to either long that a majority of the phosphorylation sites are preferentially or short period phenotypes. These changes in circadian period are phosphorylated when FRQ becomes hyperphosphorylated. Our GENETICS attributed to increases or decreases in FRQ stability, respectively. systematic mutagenesis of these phosphorylation sites reveals Together, this comprehensive study of FRQ phosphorylation re- that phosphorylation at different regions along FRQ regulates veals that regulation of FRQ stability by multiple independent the period of the clock by either promoting or inhibiting FRQ phosphorylation events is a major factor that determines the degradation. period length of the clock. A model is proposed to explain how FRQ stability is regulated by multiple phosphorylation events. Results Mapping In Vivo FRQ Phosphorylation Sites. To purify FRQ from mass spectrometry ͉ casein kinase ͉ frequency Neurospora for mapping phosphorylation sites, we created an expression construct in which a 5ϫ c-Myc tag and a 9ϫ His tag ukaryotic circadian oscillators from fungi to mammals are were inserted into the C-terminal end of the FRQ ORF. This Econtrolled by autoregulatory negative feedback loops (1–4). construct was transformed into a frq-null strain (frq10) (28). In the filamentous fungus Neurospora crassa, 2 protein com- Transformants exhibited robust circadian rhythms (Fig. S1A), plexes function in the core circadian negative feedback loop (5, indicating that epitope-tagged FRQ functions as the endogenous 6). WHITE COLLAR complex (WCC), formed by WC-1 and FRQ protein. The Myc-His-FRQ construct was also transformed WC-2, activates transcription of the frequency (frq) gene by into an fwd-1RIP strain (20). In the fwd-1 mutant, FRQ levels are binding to its promoter (7–13). On the other hand, FFC (con- elevated and FRQ exists as hyperphosphorylated forms. Myc- sisting of FRQ and the FRQ-interacting RNA helicase, FRH) His-FRQ was purified to near homogeneity from either a inhibits WCC activity by promoting the phosphorylation, and frq10,Myc-His-FRQ strain or a fwd-1RIP,Myc-His-FRQ strain by consequently repression, of frq transcription (12, 14–18). tandem affinity purification (Fig. S1B) (21). Protein bands Like the animal PERIOD (PER) proteins, FRQ is progres- corresponding to FRQ were excised and subjected to tryptic sively phosphorylated after its synthesis and becomes extensively digestion followed by MS analyses. phosphorylated before its disappearance, resulting in a robust MS analyses using purified Myc-His-FRQ from either the oscillation of its phosphorylation profile (19). One role of FRQ frq10,Myc-His-FRQ strain or the fwd-1RIP,Myc-His-FRQ strain phosphorylation is to promote FRQ degradation through the identified 43 in vivo FRQ phosphorylation sites at Ser and Thr ubiquitin-proteasome pathway mediated by ubiquitin E3 ligase residues. As shown in Fig. 1C, phosphorylation sites (amino acids SCFFWD-1. FWD-1 acts as the substrate-recruiting subunit that labeled in red) are located throughout FRQ. In many instances, recognizes and binds phosphorylated FRQ (20–22). Under normal conditions, FRQ is phosphorylated by CK-1a, CKII, and PKA (12, 16, 19, 23–25). In the ck-1a (casein kinase 1a), cka Author contributions: C.-T.T., S.L., C.L., S.C., and Y.L. designed research; C.-T.T., S.L., C.L., J.C., G.H., L.L., and S.C. performed research; C.-T.T., S.L., C.L., J.C., G.H., L.L., S.C., and Y.L. (catalytic subunit of CKII), and ckb-1 (regulatory subunit of analyzed data; and C.-T.T. and Y.L. wrote the paper. CKII) mutants, FRQ is hypophosphorylated and more stable The authors declare no conflict of interest. relative to the wild type, resulting in arrhythmia or long-period 1 phenotypes (12, 23, 25). These results suggest that CK-1a and C.-T.T. and S.L. contributed equally to this study. 2 CKII phosphorylate and promote FRQ degradation. In contrast, Present address: Department of Dermatology, The University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390. PKA counters the role of casein kinases by stabilizing FRQ (12, 3To whom correspondence may be addressed. E-mail: [email protected]. 16). FRQ is also dephosphorylated and stabilized by protein 4To whom correspondence may be sent at: Department of Physiology, ND13.214A, Uni- phosphatases PP1 and PP4 (17, 26). versity of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX FRQ exists as many isoforms, with mobilities ranging from 120 75390-9040. E-mail: [email protected]. Ϸ kDa to 200 kDa when analyzed by SDS/PAGE (19). These This article contains supporting information online at www.pnas.org/cgi/content/full/ variations are due to differences in phosphorylation, suggesting 0904898106/DCSupplemental. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0904898106 PNAS Early Edition ͉ 1of6 Downloaded by guest on October 5, 2021 Fig. 1. Map of FRQ. FRQ ORF with the identified phosphorylation sites and known FRQ domains are indicated. Amino acids in red are identified in vivo FRQ phosphorylation sites. The asterisks above the residues indicate in vitro FRQ phosphorylation sites by CK-1a, CKA, or CK-1a and CKA combined. Short and long blue lines indicate sites and phosphopeptides preferentially phosphorylated in hyperphosphorylated FRQ samples detected by quantitative MS experiments, respectively. Underlined sequences (M1–19) indicate mutated phosphorylation sites. 2 or more of these sites are clustered, suggesting sequential also perform distinct roles in FRQ phosphorylation. On the phosphorylation events. Despite the large number of sites un- other hand, 10 phosphorylation sites were detected only when covered, it is certainly an underestimate. First, tryptic peptides both CK-1a and CKII were present in the kinase reaction, either too small or too large are difficult to detect by MS. Thus, indicating that they also cooperate with each other to phosphor- peptide coverage of FRQ by trypsin was limited (44–63%) ylate FRQ. When compared to the 43 identified in vivo sites, 30 depending on the experiments. For example, the regions around were phosphorylated by CK-1a, CKII, or when combined, indi- the FRQ-CK-1a interaction domain, the FRQ-FRH interaction cating that these 2 kinases play a major role in phosphorylating domain, the PEST-1 and PEST-2 region, and the C-terminal tail FRQ. Additional kinases, such as PKA and CHK2 (16, 29), may had very poor or no MS coverage. Second, although their phosphorylate at sites independent of casein kinases. Together, unphosphorylated forms could be detected, peptides with mul- our MS analyses have led to the identification of 76 confirmed tiple phosphorylation events are known to ionize poorly in MS and potential FRQ phosphorylation sites. experiments and are often difficult to detect. Finally, limited The in vitro phosphorylation by CKI and CKII revealed 33 amounts of purified Myc-His-FRQ (Ϸ1–2 ␮g) and dephosphor- additional potential FRQ phosphorylation sites. These addi- ylation in the cell during purification procedures also make tional sites include 12 sites spanning amino acids 211–289, S513, comprehensive identification difficult. Nonetheless, the 43 in and 519 (2 previously identified sites) (22), and a phosphoryla- vivo sites identified on FRQ represent the highest number of tion site in the PEST-1 domain. These results suggest that these phosphorylation events mapped for a clock protein. In contrast additional in vitro sites may also play important roles in FRQ to FRQ, only 1 phosphorylation site (Ser-21) was identified on phosphorylation in vivo. FRH despite its 60% peptide coverage by MS. Quantitative MS Identified Preferentially Phosphorylated Sites and FRQ Phosphorylation by CK-1a and CKII.