Report

SCFFbxl3 Controls the Oscillation of the Circadian Clock by Directing the Degradation of Cryptochrome Proteins

See allHide authors and affiliations

Science  11 May 2007:
Vol. 316, Issue 5826, pp. 900-904
DOI: 10.1126/science.1141194

Abstract

One component of the circadian clock in mammals is the Clock-Bmal1 heterodimeric transcription factor. Among its downstream targets, two genes, Cry1 and Cry2, encode inhibitors of the Clock-Bmal1 complex that establish a negative-feedback loop. We found that both Cry1 and Cry2 proteins are ubiquitinated and degraded via the SCFFbxl3 ubiquitin ligase complex. This regulation by SCFFbxl3 is a prerequisite for the efficient and timely reactivation of Clock-Bmal1 and the consequent expression of Per1 and Per2, two regulators of the circadian clock that display tumor suppressor activity. Silencing of Fbxl3 produced no effect in Cry1–/–;Cry2–/– cells, which shows that Fbxl3 controls clock oscillations by mediating the degradation of CRY proteins.

SCF (Skp1/cullin/F-box protein) ubiquitin ligase complexes mediate the timely proteolysis of important eukaryotic cellular regulators (1). In mammals, there are more than 70 SCF ligases, each characterized by a different F-box protein subunit that binds and recruits substrates. Despite the large number of F-box proteins, only five human SCF ubiquitin ligases have been matched to substrates. To identify biologically important substrates of SCF ligases, we have combined immunopurification with analysis by mass spectrometry (2, 3). Here we describe work on the orphan F-box protein Fbxl3, which we originally identified as an interactor of Skp1 by yeast two-hybrid screening (4).

Fbxl3 was expressed in HeLa-S3 cells and immunopurified (fig. S1). The mass spectrometry analysis of copurified endogenous proteins revealed the presence of two peptides corresponding to Cry1 and three peptides corresponding to both Cry1 and Cry2.

The Cryptochrome proteins, Cry1 and Cry2, are evolutionarily conserved proteins that control the circadian clock (5, 6). A light-regulated master circadian pacemaker in the suprachiasmatic nucleus of the hypothalamus controls sleep-wake cycles and other rhythms (6, 7). In addition, an intrinsic, cell-autonomous circadian clock (light-insensitive) with a rhythm of about 24 hours exists in peripheral tissues to coordinate the timing of basic cellular functions (for example, cell cycle progression and checkpoint activation) (8, 9). The clock machinery is driven by two proteins, Clock and Bmal1, which heterodimerize to form an active transcription complex. Among several clock genes, the Clock-Bmal1 heterodimer drives the transcription of the Cry and Period (Per) genes. In turn, Cry1 and Cry2 inhibit Clock-Bmal1–dependent transcription, creating a negative-feedback loop that is central to the oscillation of the clock. In mammals, CRY proteins are the most potent repressors of Clock-Bmal1 (1013). An integral component of this feedback loop is the requirement for degradation of CRY repressor proteins, without which the amplitude of the circadian clock oscillations would be inhibited. Although CRY proteins are known to be degraded via the ubiquitin system (14), the specific ubiquitin ligase that directs this event has remained unknown.

To investigate the specificity of binding between CRY proteins and Fbxl3, we screened 10 human F-box proteins. FLAG-tagged versions of these F-box proteins were retrovirally expressed in NIH 3T3 cells and then immunoprecipitated to evaluate their interaction with three endogenous clock proteins. We found that the only F-box protein able to coimmunoprecipitate Cry1 and Cry2 was Fbxl3 (Fig. 1A). The Fbxl3-CRY interaction was further confirmed by a complementary experiment in which FLAG-tagged versions of eight clock proteins were expressed in human embryonic kidney (HEK) 293T cells, immunoprecipitated, and tested for their ability to coimmunoprecipitate seven endogenous F-box proteins. Only Cry1 and Cry2 interacted with endogenous Fbxl3, but not with Skp2, Fbxl10, Fbxl11, βTrcp1, Emi1, or cyclin F (Fig. 1B).

Fig. 1.

CRY proteins specifically bind Fbxl3. (A) NIH 3T3 cells were infected with retroviruses encoding the indicated FLAG-tagged F-box proteins (FBPs). During the last 6 hours before harvesting, cells were treated with the proteasome inhibitor MG132. Exogenous proteins were immunoprecipitated (IP) from cell extracts with an anti-FLAG resin, and immunocomplexes were probed with antibodies to the indicated endogenous proteins. Lane 1 shows a whole-cell extract (WCE) from cells infected with an empty virus (EV). (B) HEK293T cells were transfected with constructs encoding the indicated FLAG-tagged circadian clock proteins (CCPs). Exogenous proteins were immunoprecipitated from cell extracts with an anti-FLAG resin, and immunocomplexes were probed with antibodies to the indicated endogenous proteins. Lane 1 shows a whole-cell extract from cells transfected with an empty vector.

These results prompted us to test whether Fbxl3 is involved in regulating the stability of CRY proteins. To this end, we transfected HEK293T cells with Cry2 and either Fbxl3 or Skp2. Whereas the expression of Skp2 had no observable effect on Cry2 stability, the expression of Fbxl3 resulted in a decrease in the half-life of Cry2, which was efficiently counteracted by the addition of the proteasome inhibitor MG132 (Fig. 2A and fig. S2). We also used two small hairpin RNA (shRNA) constructs to reduce the expression of Fbxl3 in HEK293T cells. Although both constructs targeted Fbxl3, construct 2 was more efficient in silencing Fbxl3 expression (fig. S3A). Accordingly, both constructs induced an increase in the half-life of Cry2, but construct 2 produced a more robust effect (Fig. 2B and fig. S3B). Stabilization of both Cry1 and Cry2 was also observed as an effect of Fbxl3 knockdown in NIH 3T3 cells (fig. S4).

Fig. 2.

Fbxl3 controls the degradation and ubiquitination of CRY proteins. (A) Fbxl3 promotes the degradation of Cry2. The graph shows quantification of Cry2 half-life in HEK293T cells transfected with Cry2 alone or in combination with either Fbxl3 or Skp2 (with or without MG132). Error bars represent ±SD (n = 3). A representative experiment is shown in fig. S2. (B) Knockdown of Fbxl3 stabilizes Cry2. The graph shows quantification of Cry2 half-life in HEK293T cells infected with lentiviral constructs that direct the synthesis of two different shRNAs to Fbxl3 or an shRNA targeting LacZ. Error bars represent ±SD (n = 3). A representative experiment is shown in fig. S3. (C) Fbxl3 ubiquitinates Cry2. In vitro ubiquitination assays of recombinant Cry2 protein were conducted in the presence of the Skp1-Cul1-Roc1 complex plus one of the following recombinant F-box proteins: Fbxl3, Skp2, or βTrcp1, as indicated. Samples were analyzed by immunoblotting with an antibody to Cry2. Thebracket at the left marks a ladder of bands corresponding to polyubiquitinated Cry2. (D) HEK293T cells were transfected with Cry2, Skp1, Cul1, and Roc1 in the absence or presence of either FLAG-tagged Fbxl3 or a FLAG-tagged Fbxl3(ΔF-box) mutant. After immunopurification with an anti-FLAG resin, in vitro ubiquitination of Cry2 was performed. Samples were analyzed by immunoblotting with an antibody to Cry2. The lower panel shows levels of Fbxl3, Fbxl3(ΔF-box), Skp1, and Cul1 in the immunoprecipitates.

To test whether CRY proteins are ubiquitinated via the SCFFbxl3 ubiquitin ligase, we reconstituted the ubiquitination of Cry2 in vitro. Cry2 was efficiently ubiquitinated only when Fbxl3 was present (Fig. 2C). Different recombinant F-box proteins, including Skp2 and βTrcp1, were unable to trigger the ubiquitination of Cry2. Addition of methylated ubiquitin to the reaction inhibited the formation of the highest-molecular-weight forms of Cry2 (fig. S5), demonstrating that the high-molecular-weight forms of Cry2 are polyubiquitinated species of the protein. We also used a complementary in vitro ubiquitination strategy using immunopurified Fbxl3 proteins. Fbxl3, but not an inactive Fbxl3(ΔF-box) mutant (4), induced the ubiquitination of Cry2 (Fig. 2D), which supports the notion that the effect of Fbxl3 on Cry2 is direct.

To study the biological importance of SCFFbxl3-mediated proteolysis of CRY proteins, we investigated the effect of Fbxl3 knockdown in cells in which the circadian clock is synchronized. A 90-min pulse with 50% horse serum synchronizes clock oscillators in confluent fibroblasts (15, 16). By synchronizing mouse embryonic fibroblasts (MEFs), we observed periodic oscillations in the levels of Cry1, Per1, and Per2 (Fig. 3A). Silencing of Fbxl3 abolished the oscillations in the levels of Cry1 and produced a decrease in the expression of Per1 and Per2 (Fig. 3A). Similar results were obtained in NIH 3T3 cells with the use of two different shRNA constructs (figs. S6 and S7).

Fig. 3.

Fbxl3 controls oscillations of the clock by mediating the degradation of CRY proteins. (A) Fbxl3 promotes Cry1 oscillations and the accumulation of PER proteins in Cry1+/+;Cry2+/+ cells but not in Cry1–/–;Cry2–/– cells. MEFs were infected with lentiviral constructs that direct the synthesis of shRNAs targeting LacZ or Fbxl3, as indicated. Cells were asynchronously grown to confluence in a medium containing 10% serum (Asynch.) and then shifted to a medium containing 50% horse serum (50% HS) for 90 min. After this period, the serum-rich medium was replaced with 0.5% fetal bovine serum–containing medium and cells were collected at the indicated hours after serum shock. Cell extracts were analyzed by immunoblotting with antibodies to the indicated proteins. The graphs on the right illustrate the quantification by densitometry of triplicate experiments, including that shown in the left panels. The value given for the amount of Cry1, Per1, and Per2 present in asynchronous cells was set as 1. Error bars represent ±SD. (B) Fbxl3 promotes the accumulation of Bmal1-Clock–regulated mRNAs in Cry1+/+;Cry2+/+ cells but not in Cry1–/–;Cry2–/– cells. Triplicate experiments were performed as in (A), except that whole-cell RNA was prepared to determine the levels of the indicated mRNA via quantitative reverse transcription polymerase chain reaction. The value given for the amount of mRNA present in asynchronous cells was set as 1. Error bars represent ±SD.

We propose that when Fbxl3 is down-regulated, stabilized Cry1 persistently inhibits Clock-Bmal1, which is responsible for the induction of Per and Cry genes. In support of this hypothesis, we found that the knockdown of Fbxl3 inhibited the induction and oscillations of Clock-Bmal1–regulated mRNAs (Per1, Per2, and Cry1) (Fig. 3B and fig. S8). Moreover, chromatin immunoprecipitation experiments showed that silencing of Fbxl3 increased the abundance of Cry1 protein at the promoters of the Per1, Per2, and Cry1 genes (fig. S9). Because the levels of Fbxl3 do not oscillate (Fig. 3A), it is possible that posttranslational modifications of CRY proteins regulate their recognition by Fbxl3, as is the case for most substrate–F-box protein interactions (1).

To determine whether the effects of Fbxl3 knockdown on the circadian clock are dependent on the stabilization of CRY proteins, we silenced Fbxl3 expression in Cry1–/–;Cry2–/– MEFs (12). No substantial differences in the protein and mRNA levels of Per1 and Per2 were observed in Cry1–/–;Cry2–/– MEFs after down-regulation of Fbxl3 (Fig. 3).

In the accompanying report (17), Godinho et al. used a forward genetic screen to induce mutations in mice that increase the length of the circadian clock and identified a mouse mutation in Fbxl3 [a Cys358 → Ser (C358S) substitution] that results in a circadian rhythm of about 27 hours in homozygotes. To investigate whether this mutation interferes with Fbxl3 activity, we generated a Fbxl3(C358S) mutant, as well as a Fbxl3(C358A) mutant (containing a Cys358 → Ala substitution). FLAG-tagged versions of Fbxl3 and the two mutants were retrovirally expressed in NIH 3T3 cells and then immunoprecipitated to evaluate their interaction with endogenous Cry1 and Skp1. Whereas all three proteins were able to coimmunoprecipitate Skp1, only wild-type Fbxl3 interacted with Cry1 (Fig. 4A). In a complementary experiment, MYC-tagged Cry2 was expressed in HEK293T cells together with one of the following F-box proteins: Fbxl3, Fbxl3(C358S), Fbxl15, and βTrcp1. After immunoprecipitation, Fbxl3(C358S) was found to bind Cry2 less efficiently than did Fbxl3 (Fig. 4B). Accordingly, expression of Fbxl3(C358S) had no effect on the stability of Cry1 (fig. S10), and Fbxl3(C358S) was less efficient than Fbxl3 in ubiquitinating Cry2 in vitro (Fig. 4C). These results provided evidence that the phenotype observed in mice (17) is due to a decreased ability of Fbxl3(C358S) to bind to and induce the proteolysis of CRY proteins.

Fig. 4.

The Fbxl3(C358S) mutant binds to and ubiquitinates CRY proteins less efficiently than wild-type Fbxl3. (A) NIH 3T3 cells were left uninfected (UI) or infected with retroviruses encoding Fbxl3, Fbxl3(C358S), or Fbxl3(C358A) (all FLAG-tagged). During the last 6 hours before harvesting, cells were treated with the proteasome inhibitor MG132. Proteins were immunoprecipitated from cell extracts with anti-FLAG resin, and immunocomplexes were probed with antibodies to the indicated proteins. (B) HEK293T cells were transfected with MYC-tagged Cry2 together with vectors encoding the indicated proteins. Cry2 was immunoprecipitated from whole-cell extracts with an antibody to MYC, and immunocomplexes were probed with antibodies to the indicated proteins. (C) The experiment was performed as in Fig. 2C, except that Fbxl3(C358S) was also assayed.

Our study demonstrates that the SCFFbxl3 ubiquitin ligase controls the oscillations of the circadian clock. Fbxl3-mediated degradation of Cry1 and Cry2 is a prerequisite for the efficient and timely reactivation of Clock-Bmal1 and the consequent transcription of target genes, including Per1 and Per2, two putative tumor suppressors that control fundamental processes such as the timing of cell cycle progression and checkpoint activation (1821). Because silencing of Fbxl3 produces no effects in Cry1–/–;Cry2–/– cells, we conclude that the effects of Fbxl3 on the clockare largely mediated via regulation of Cry1 and Cry2 stability.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1141194/DC1

Materials and Methods

Figs. S1 to S10

References

References and Notes

View Abstract

Navigate This Article