Abstract
To remain synchronous with the environment, plants constantly survey daily light conditions using an array of photoreceptors and adjust their circadian rhythms accordingly. ZEITLUPE (ZTL), a blue light photoreceptor with E3 ubiquitin ligase activity, communicates end-of-day light conditions to the circadian clock. To function properly, ZTL protein must accumulate but not destabilize target clock transcription factors before dusk, while in the dark ZTL mediates degradation of target proteins. It is not clear how ZTL can accumulate to high levels in the light while its targets remain stable. Two deubiquitylating enzymes, UBIQUITIN-SPECIFIC PROTEASE 12 and UBIQUITIN-SPECIFIC PROTEASE 13 (UBP12 and UBP13), which have opposite genetic and biochemical functions to ZTL, were shown to associate with the ZTL protein complex. Here we demonstrate that the ZTL light-dependent interacting partner, GIGANTEA (GI), recruits UBP12 and UBP13 to the ZTL photoreceptor complex. We show that loss of UBP12 and UBP13 reduces ZTL and GI protein levels through a post-transcriptional mechanism. Furthermore, the ZTL target protein TOC1 is unable to accumulate to normal levels in ubp mutants, indicating that UBP12 and UBP13 are necessary to stabilize clock transcription factors during the day. Our results demonstrate that the ZTL photoreceptor complex contains both ubiquitin-conjugating and -deconjugating enzymes, and that these two opposing enzyme types are necessary for the complex to properly regulate the circadian clock. This work also shows that deubiquitylating enzymes are a core design element of circadian clocks that is conserved from plants to animals.
Main
Circadian clocks in all organisms rely on photoreceptors to sense light and entrain the central oscillator. The exact timing of the light-to-dark transition (dusk) is especially important for plants, as this indicates the length of the day and provides seasonal timing information necessary for the adjustment of plant developmental processes (Carre, 2001; Yanovsky and Kay, 2002; Imaizumi et al., 2003; Salome and McClung, 2004; Imaizumi and Kay, 2006; Nozue et al., 2007; Mizoguchi and Yoshida, 2009; Ito et al., 2012). One way that plants sense the end of the day is by using a unique photoreceptor called ZEITLUPE (ZTL) to control the stability of circadian clock transcription factors differentially in the light and the dark (Somers et al., 2000). ZTL contains an N-terminal light-oxygen-voltage sensing (LOV) domain which senses blue light. Adjacent to the LOV domain are the F-box domain, which allows ZTL to function as an E3 ubiquitin ligase, and a Kelch-repeat domain. ZTL mediates degradation of transcription factors that are at the core of the plant circadian clock including TIMING OF CAB2 EXPRESSION 1, PSEUDO-RESPONSE REGULATOR 5, and CCA1 HIKING EXPEDITION (TOC1, PRR5, and CHE) (Mas et al., 2003; Han et al., 2004; Kiba et al., 2007; Fujiwara et al., 2008; Baudry et al., 2010; Lee et al., 2018). In the light, ZTL accumulates to high levels but is unable to mediate degradation of the clock transcription factors (Kim et al., 2003; Kim et al., 2007). The accumulation of ZTL protein during the day is dependent on interaction with the co-chaperone protein GIGANTEA (GI) (Kim et al., 2013; Cha et al., 2017; Cha et al., 2017). GI interacts with ZTL through the LOV domain in the light and dissociates from ZTL in the dark, allowing ZTL to mediate degradation of its target proteins and then be degraded by the ubiquitin proteasome system, likely through autocatalytic activity (Kim et al., 2003; Mas et al., 2003; Somers et al., 2004; Kiba et al., 2007; Kim et al., 2007; Kim et al., 2011; Kim et al., 2013). One of the roles of GI is to recruit HSP70/HSP90 for maturation of the ZTL protein in the light, but ZTL is unable to mediate ubiquitylation and degradation of target proteins until dark (Mas et al., 2003; Kiba et al., 2007; Fujiwara et al., 2008; Cha et al., 2017; Pudasaini et al., 2017). It was proposed that GI can promote maturation of ZTL and block or counteract ZTL activity; however, this second role for GI has not been investigated in depth (Fujiwara et al., 2008; Pudasaini et al., 2017).
We previously identified ZTL-interacting proteins using immunoprecipitation followed by mass spectrometry (IP-MS) with a “decoy” ZTL that lacks E3 ubiquitin ligase activity and stably binds interacting proteins (Lee et al., 2018). We identified UBIQUITIN-SPECIFIC PROTEASE 12 and 13 (UBP12 and UBP13) as high confidence ZTL-interacting proteins which were shown previously to have an unspecified role in clock function (Cui et al., 2013; Lee et al., 2018). UBP12 and UBP13 also interact with GI in IP-MS experiments (Krahmer et al., 2019), suggesting that either the UBPs interact with ZTL and GI independently or that ZTL, GI, and the UBPs exist together in a complex. UBP12 and UBP13 are closely related deubiquitylating enzymes that can cleave lysine 48-linked mono- or poly-ubiquitin from substrates (Ewan et al., 2011; Cui et al., 2013), interestingly, a biochemical role opposite to that of ZTL. In addition to regulating the circadian clock, they are also involved in flowering time, pathogen defense, root differentiation, and hormone signaling (Ewan et al., 2011; Derkacheva et al., 2016; Jeong et al., 2017; Zhou et al., 2017; An et al., 2018). We performed yeast two-hybrid assays and found that UBP12 and UBP13 interacted with GI but not with ZTL or the ZTL target proteins TOC1, PRR5, or CHE (Fig.1a). We next tested the interaction between GI and UBP12 and UBP13 in planta via bimolecular fluorescence complementation (BiFC) in Arabidopsis protoplasts (Fig.1b). GI, UBP12, and UBP13 are localized in the cytoplasm and nucleus (Cui et al., 2013; Kim et al., 2013), and our BiFC results show that UBP12 and UBP13 interact with GI in both compartments with strong signal in the nucleus and weaker but detectable signal in the cytoplasm. The interacting complexes of UBP12 and GI formed nuclear foci, similar to the localization of GI alone (Kim et al., 2013). UBP12 and UBP13 contain a MATH-type (meprin and TRAF homology) protein interaction domain and a ubiquitin-specific protease (USP) domain (Fig. S1). The MATH domains of UBP12 and UBP13 were necessary for interaction with GI while the protease domain and the C-terminal portions did not mediate GI-interaction (Fig. 1c). This suggests that the interaction between GI and UBP12 or UBP13 is not dependent on the UBP USP domains binding to poly-ubiquitylated GI protein.
We next determined whether GI was necessary to bridge the interaction between UBP12 or UBP13 and ZTL in vivo by performing IP-MS on wild type (Col-0) and gi-2 mutant transgenic lines expressing the decoy ZTL protein (Fig. S2). We collected samples at 9 hours after dawn from plants grown in 12h light/12h dark cycles to capture the time when ZTL and GI are normally interacting. We found that UBP12 and UBP13 were enriched in the Col-0 samples (p-value= 3.58E-5 and 0.0113 for UBP12 and UBP13 respectively), but not in the gi-2 mutant (p-value= 1 for both) (Fig.1d and Table S1). These results indicate that GI is required for UBP12/UBP13 to form a complex with ZTL, substantiating our interaction studies in heterologous systems. Notably, LKP2, a known ZTL interacting partner, associated with ZTL in the presence or absence of GI and suggests that the decoy ZTL is able to form biologically relevant protein complexes even in the gi-2 mutant (Takase et al., 2011). Together these results suggest that the GI protein physically bridges the interaction between UBP12 or UBP13 and ZTL in vivo.
As a complementary approach to the IP-MS (Fig.1d) we co-expressed FLAG-UBP12 or FLAG-UBP13 with HA-GI and Myc-ZTL in N. benthamiana leaves. We then performed immunoprecipitation with anti-FLAG antibody and detected the presence of FLAG-UBP12, FLAG-UBP13, HA-GI, and Myc-ZTL using western blotting (Fig.1e). In the FLAG immunoprecipitation samples, HA-GI was always detected when co-expressed with FLAG-UBP12 or FLAG-UBP13, showing that UBP12 and UBP13 interact with GI independently of the presence of Myc-ZTL. Furthermore, Myc-ZTL was undetectable in the FLAG immunoprecipitation samples unless co-expressed with HA-GI showing that the interaction between UBP12 or UBP13 and ZTL is dependent on GI. These assays support our previous results (Fig.1a-d) and show that a trimeric complex between full-length ZTL, GI, and UBP12 or UBP13 can form in vivo (Fig. 1f).
Our physical interaction model (Fig. 1f) led us to hypothesize that UBP12 and UBP13 regulate the circadian clock through the same genetic pathway as ZTL and GI. We tested this via epistasis analyses with loss-of-function mutants in ZTL, GI, UBP12, and UBP13. Previously, it was shown that knockdown of UBP12 and UBP13 results in shortened clock periods (Cui et al., 2013). We first determined the period of a series of mutant alleles in UBP12 and UBP13 by crossing them to the pCCA1::LUC clock reporter transgenic line and measuring luciferase activity (Fig. 2a-d). We found that single mutations in either UBP12 or UBP13 shortened the clock period with period lengths that varied from 0.4 to 1 hour shorter than wild type. We next generated ubp12-1/gi-2 and ubp13-1/gi-2 double mutants and measured the expression of the core clock gene CCA1 during a 2-day circadian time course in constant light using qRT-PCR (Fig. 2e-f and Table S2). LS Periodogram analysis using the Biodare2 platform [biodare2.ed.ac.uk (Zielinski et al., 2014)] showed that the ubp12-1/gi-2 double mutant had a similar phase and amplitude of CCA1 expression to the gi-2 mutant alone and a period more similar to ubp12-1 (Table S3). These results show a non-additive interaction and suggest they function in the same circadian genetic pathway. The ubp13-1/gi-2 double mutant had a similar amplitude to the gi-2 mutant but had a more similar phase and period to the ubp13-1 mutant (Table S3). This again shows a non-additive genetic interaction but also suggests that the roles of UBP12 and UBP13 have slightly diverged with respect to clock function. We also crossed the gi-2 mutant with the ubp12-2w mutant which had reduced expression of both UBP12 and UBP13 and the shortest clock period of the tested ubp mutant alleles (Fig.S3, 2a-d). The pattern of CCA1 expression in the ubp12-2w/gi-2 double mutant was nearly identical to the gi-2 mutant, further confirming that the effects of the UBPs and GI are not additive (Table S3). These results indicate that UBP12 and UBP13 work in the same pathway as GI to control clock function.
ZTL functions downstream of GI to regulate the circadian clock (Kim et al., 2007). Thus, we hypothesized that ZTL would function downstream of UBP12 and UBP13 as well. To test the genetic interaction between UBP12 or UBP13 and ZTL, we crossed ubp12-1 and ubp13-1 to the ztl-4 null mutant (Fig. 2g and h). The daily expression patterns of CCA1 in the ubp12-1/ztl-4 and ubp13-1/ztl-4 double mutants were nearly identical to the ztl-4 mutant alone in phase and amplitude (Table S3). Interestingly, the period data showed that the ubp12-1/ztl-4 was more similar to ztl-4 than ubp12-1, but the ubp13-1/ztl-4 is more similar to ubp13-1. This data suggests that ZTL is epistatic to UBP12 and UBP13 but that UBP13 has diverged in function from UBP12. It is important to note that the qRT-PCR data is below the suggested resolution for Biodare2 analysis which can result in inaccurate period calls (i.e. ubp13-1 period is estimated by Biodare2 as the same period as wild type in this experiment). These results corroborate our physical interaction studies and suggest that UBP12 and UBP13 regulate the circadian clock upstream of ZTL.
UBP12 and UBP13 are functional deubiquitylases that can cleave poly-ubiquitin from generic substrates (Ewan et al., 2011; Cui et al., 2013). We tested whether this deubiquitylation activity is necessary for their role in circadian clock function. To do this we performed complementation studies with wild-type UBP12 and mutant UBP12C208S. UBP12C208S has a mutation in the cysteine-box of the USP enzymatic core (Fig. S1) that renders it non-functional as a deubiquitylase (Cui et al., 2013; Derkacheva et al., 2016; Jeong et al., 2017). We transformed UBP12-YFP or UBP12C208S-YFP driven by the UBP12 native promoter into the ubp12-1 mutant and analyzed a population of T1 transgenic lines. In this experiment we consider a line to have rescued the ubp12-1 mutant clock phenotype if it has a period length longer than the average period length of the ubp12-1 plus one standard deviation. Using this criteria, 10 of 32 transgenic lines (31%) transformed with catalytically active UBP12 rescued the short period defect of the ubp12-1 mutant. Strikingly, only one transgenic line transformed with the inactive UBP12C208S was able to rescue the short period phenotype of ubp12-1 (Fig. 2i-j). As reference, approximately 13% of the ubp12-1 plants themselves and 62% of the wild type plants fell into the rescue category. This is likely due to normal variations in population level data of this type. We further confirmed that UBP12-YFP and UBP12C208S-YFP were both localized to the cytoplasm and nucleus, suggesting that differences in the clock phenotypes are not due to mislocalization of the UBP12C208S protein (Fig S4). These results indicate that the deubiquitylating functions of UBP12 are necessary for its role in regulating the circadian clock.
By cleaving poly-ubiquitin from proteins, deubiquitylase enzymes can regulate protein stability and accumulation (Komander et al., 2009; Jeong et al., 2017; Mevissen and Komander, 2017; An et al., 2018). The physical and genetic interactions shown for UBP12, UBP13, GI and ZTL prompted us to hypothesize that the UBP12 and UBP13 regulate GI or ZTL protein levels, allowing for accumulation of the proteins in the end of the day. We measured the level of HA-tagged GI under the control of the GI native promoter (pGI::GI-HA) in the ubp12-1 and ubp13-1 mutants during a 12h light/12h dark time course (Fig. 3a). GI protein levels were approximately 50% lower in the ubp12-1 and ubp13-1. mRNA expression of GI-HA was also approximately 25% lower than wild type at the peak of GI mRNA expression, ZT8 (Fig. 3b). This suggests that GI protein accumulation is partially dependent on UBP12 and UBP13, but that altered transcription of GI could also have an effect on GI protein.
Next, we measured ZTL protein levels in the ubp12-1 and ubp13-1 mutants (Fig. 3c). ZTL protein levels were substantially decreased in the ubp12-1 and ubp13-1 mutants throughout the entire day/night cycle. Overexposure of the immunoblot showed that a small amount of ZTL protein can still accumulate in the ubp mutants (Fig. 3c). The expression of ZTL mRNA was largely unaffected in these lines (Fig. 3d), suggesting that the decrease in ZTL protein levels was caused by a post-transcriptional mechanism. This is similar to the post-transcriptional control of ZTL reported in gi loss-of-function mutants (Kim et al., 2007), and indicates that UBP12 and UBP13 are necessary for robust accumulation of the ZTL protein.
Interestingly, the ubp12-1 and ubp13-1 mutants caused severe reduction in the levels of the ZTL protein but had a short period phenotype, opposite to the long period phenotype of ztl loss-of-function mutants. Normally, loss of ZTL causes aberrantly high levels of TOC1 protein while overexpression of ZTL causes low levels of TOC1 protein (Mas et al., 2003; Kiba et al., 2007; Pudasaini and Zoltowski, 2013; Pudasaini et al., 2017). To determine if UBP12 and UBP13 affect TOC1 protein levels, we crossed a transgenic line expressing TOC1 fused to YFP under the TOC1 promoter (TMG) to the ubp12-1 and ubp13-1 mutants and measured TOC1 protein levels (Fig. 3e). TOC1 protein levels were severely reduced in the ubp12-1 and ubp13-1 mutants while mRNA expression of the TOC1-YFP transgene was similar in the wild type and mutant backgrounds, suggesting that the decrease in TOC1 protein levels was caused by a post-transcriptional mechanism (Fig. 3f). Notably, TOC1 protein was unable to accumulate to high levels in the light in the ubp mutants (Fig. 3e at 12 hours after dawn). This is similar to the effects of the gi-2 mutant, where TOC1 protein levels never accumulate to full wild-type levels (Kim et al., 2007). Lowered levels of the TOC1 protein result in shortened period, suggesting this was the cause of the short period phenotype of the ubp12 and ubp13 mutants.
We have shown that UBP12 and UBP13 are components of the ZTL-GI photoreceptor complex that are necessary for accumulation of the proteins in the end of the day. UBP12 and UBP13 can remove poly-ubiquitin from targets non-specifically (Ewan et al., 2011; Cui et al., 2013). Thus, we hypothesize that UBP enzymes are recruited by GI to the ZTL photoreceptor complex to prevent formation of poly-ubiquitin chains, resulting in increased stability of the protein complex (Fig. S5). Interestingly, ZTL protein levels were severely damped in the ubp12 and ubp13 mutants, but counterintuitively the ZTL target, TOC1, also had reduced levels (Fig. 3c-f). This effect is similar to what was observed in a gi loss-of-function mutant, and suggests that GI and UBP12 and UBP13 can counterbalance the activity of ZTL during the day, allowing TOC1 to accumulate to high levels before being degraded (Kim et al., 2007). Although ZTL levels were decreased in the ubp mutants, there was still a small amount that could potentially decrease TOC1 levels in the light (Fig. 3c long exposure). This is different than what was seen when HSP90 activity was inhibited, resulting in lower ZTL levels but higher TOC1 levels. This suggests that HSP90 is necessary for ZTL protein maturation and to promote its activity (Kim et al., 2011). This data in combination with our results suggest that GI performs two roles in the ZTL photoreceptor complex: (1) acting as a co-chaperone that recruits HSP proteins to facilitate ZTL maturation (Cha et al., 2017; Cha et al., 2017), and (2) counterbalancing ZTL’s role in ubiquitin conjugation with UBP12 and UBP13 present to deconjugate ubiquitin. The light-regulated nature of the ZTL-GI interaction also indicates that light is controlling the balance of ubiquitin conjugation and deconjugation that allows the ZTL photoreceptor complex to accurately degrade proteins at the correct time of day. It was previously shown that mammalian and insect circadian clocks utilize deubiquitylation to regulate stability and subcellular localization of clock proteins (Scoma et al., 2011; Luo et al., 2012; Yang et al., 2012). In light of this, our results further demonstrate that deubiquitylation activity is an evolutionarily conserved architectural design feature of the clocks of higher eukaryotes Furthermore, the mammalian orthologue of UBP12 and UBP13, USP7, impacts clock function in response to environmental stress (Papp et al., 2015; Hirano et al., 2016) suggesting that these deubiquitylases are conserved clock regulators across evolution.
Materials and Methods
See supplemental information
Author contribution
JMG and CML conceived of the project. CML, MWL, AF, AMS and WL conducted the experiments and analyzed the data. JMG and CML wrote the manuscript.
Competing interests
The authors declare no competing financial interests.
Acknowledgement
We thank Dr. David E. Somers for providing the anti-ZTL antibody. We would like to thank Dr. Nicole Clay and Dr. Jimi Miller for kindly sharing pABind vectors and the assistance with transient expression in the Nicotiana benthamiana experiments, the Keck Proteomics Facility at Yale for processing samples and analyzing proteomic data, Dr. Shirin Bahmanyar, Dr. Marshall Delise and Dr. Joseph Wolenski for assistance with confocal microscopy experiments. Also, Suyuna Eng Ren, Chris Adamchek, Chris Bolick, Christine Ventura, Denise George, and Sandra Pariseau for technical and administrative support. We would like to thank Dr. Vivian Irish, Dr. Mark Hochstrasser and Dr. Eric Bennet for helpful comments on experimentation and the manuscript. This work was supported by NSF EAGER grant 1548538 (JMG), NIH R35GM128670 (JMG), Rudolph J. Anderson Fund Fellowship (CML), Forest B.H. and Elizabeth D.W. Brown Fund Fellowship (CML and WL), NIH GM007499 (AF), The Gruber Foundation (AF), and NSF GRFP DGE-1122492 (AF).