The After-Hours Mutant Reveals a Role for Fbxl3 in Determining Mammalian Circadian Period

See allHide authors and affiliations

Science  11 May 2007:
Vol. 316, Issue 5826, pp. 897-900
DOI: 10.1126/science.1141138


By screening N-ethyl-N-nitrosourea–mutagenized animals for alterations in rhythms of wheel-running activity, we identified a mouse mutation, after hours (Afh). The mutation, a Cys358Ser substitution in Fbxl3, an F-box protein with leucine-rich repeats, results in long free-running rhythms of about 27 hours in homozygotes. Circadian transcriptional and translational oscillations are attenuated in Afh mice. The Afh allele significantly affected Per2 expression and delayed the rate of Cry protein degradation in Per2::Luciferase tissue slices. Our in vivo and in vitro studies reveal a central role for Fbxl3 in mammalian circadian timekeeping.

Circadian rhythms, oscillations with a period of ∼24 hours, are vital for physiological and behavioral homeostasis in multi- and unicellular organisms. In mammals, these biological rhythms are generated by several genetic elements that form autoregulatory transcriptional and translational feedback loops (14). For the clock to function effectively, however, the phase and extent of protein expression must also be tightly regulated. Posttranslational modifications necessary for the accurate timekeeping of this molecular machinery have been described. For example, Per levels depend on phosphorylation by Doubletime in Drosophila (5) or casein-kinase Iϵ in mammals (6), and sumoylation is a regulator of Bmal1 function (7). Circadian roles for ubiquitin E3 ligases containing F-box motifs (8) have been described in Drosophila for Slimb (9) and JETLAG (10) and in Arabidopsis for ZEITLUPE (11, 12) and FKF1 (13). Large families of mammalian F-box proteins have been identified (1416), but, apart from mammalian orthologs of Drosophila Slimb (β-TRCP) in cultured cells (17), none are known to degrade clock proteins in mammals.

To identify previously unknown genetic factors affecting mammalian circadian behavior, we conducted N-ethyl-N-nitrosourea (ENU) screens for alterations in circadian wheel-running activity in mice (18). We identified one mouse with a circadian period (τDD) of ∼24 hours, significantly longer than the population mean (23.63 hours). The phenotype was inherited in a dominant fashion, with τDD ranging from 23.9 to 24.3 hours. Intercrosses revealed an additional phenotype with a τDD of ∼26.5 hours and a delay in the entrainment phase angle (Fig. 1A). Frequency distribution plots of backcross (n = 264) and intercross progeny (n = 73) indicated semidominant inheritance of the phenotype, with the latter revealing a group of animals with τDD ranging from 25.3 to 27.6 hours (Fig. 1B). We named the mutant “after hours (Afh) and mapped the dominant phenotype at low resolution (18), selecting 13 mice displaying the most extreme τDD phenotypes for a genome scan. A single region of linkage was found on mouse chromosome 14, flanked by D14Mit239 (66.65 Mb) and D14Mit267 (114 Mb). Using additional polymorphic markers, we refined the nonrecombinant region to 6.41 Mb between 89.65 and 96.06 Mb (Fig. 1C) and confirmed the position by genotyping homozygotes. This region is extremely gene-poor, containing only 25 annotated genes (19). We scanned all Ensembl-annotated genes for mutations using denaturing high-performance liquid chromatography and sequencing. Sequencing revealed a point mutation in the coding region of an F-box gene, Fbxl3, resulting in a T→A transversion at position 1430 (Fig. 1D). The Afh mutation results in the substitution of serine for Cys358 (C358S) in Fbxl3 (Fig. 1E). Genotyping allowed us to characterize the behavioral phenotype of mutants systematically (table S1 and fig. S1).

Fig. 1.

Mutant phenotype and cloning of the Afh mutation. (A) Representative activity records of +/+ (τDD = 23.68 hours), Afh/+ (τDD = 24.23), and Afh/AfhDD = 26.73) mice. Images show double-plotted actograms of wheel-running activity under light-dark (LD) conditions (days 0 to 7) followed by constant darkness (DD, days 8 to 21). Short, vertical bars represent bouts of wheel-running activity. Periods of light and dark are indicated by horizontal bars above actograms, and the transition from LD to DD is indicated by an arrow. (B) Frequency distribution of τDD in Afh/+ backcross and intercross progeny. (C) Mapping of the Afh mutation to chromosome 14 by haplotype analysis. The number of haplotypes analyzed (n) and τDD (means ± SEM) are indicated. (D) Sequence analysis revealing a T→A transversion at nucleotide position 1430 of mouse Fbxl3 in the Afh mutant. (E) (Top) Schematic representation of Fbxl3 indicating the positions of F-box and LRR-cc regions. The position of the C358S substitution is marked by an asterisk. (Bottom) Protein sequence alignment of the C-terminal end of Fbxl3 in mutant and wild-type mouse and other vertebrate species. The shaded vertical box indicates the position of the C358S substitution, and the bold horizontal line indicates the consensus LRR-cc sequence.

The physiological function of Fbxl3 is unknown. The N-terminal F-box domain implicates this protein in ubiquitination and proteasomal degradation, whereas the C-terminal leucine-rich repeat (LRR) domain is likely to be involved in recognition of phosphorylated targets. Analysis of conserved functional domains (20) suggests that C358 is an essential residue in a subclass of LRR, the cysteine-containing LRR (LRR-cc). The Fbxl3 sequence is highly conserved in vertebrates, but there are no invertebrate orthologs. Fbxl3 is ubiquitously expressed, and the protein is predominantly, although not exclusively, nuclear (14). In vitro transfection of either wild-type or mutant Fbxl3 tagged with green fluorescent protein (GFP) into COS7 cells confirmed its predominantly nuclear localization with no differences in protein stability (fig. S2A). Furthermore, there was no circadian or genotype effect on gene or protein expression (fig. S2, B and C). Similarly, protein localization was not affected.

To investigate the molecular basis of period lengthening in homozygotes, we examined circadian expression profiles using in situ hybridization, immunohistochemistry, and Western blotting. In the suprachiasmatic nucleus (SCN), the central pacemaker driving circadian rest-activity cycles, Afh affected steady-state levels of the principal negative-feedback regulators of the clock, Period genes Per1 and Per2, Cryptochrome gene Cry1, and the positive regulator Bmal1 (Fig. 2A). Although rhythmic with a prolonged period, peak mRNA levels of all four genes were reduced relative to wild-type. Moreover, during late circadian night [circadian time (CT)18 to CT03], Per1, Per2, and Cry1 mRNA levels were suppressed for longer than in wild-type mice. These mRNA changes were echoed in cycles of protein expression. For Per1 and Per2, the number of immunoreactive (-ir) nuclei in SCN was severely reduced across circadian time, particularly at the peak around CT12 to CT15. Similarly, the nocturnal peak of Bmal1-ir was lower in mutants. Despite the marked decline in Cry1 mRNA, however, Afh homozygotes showed no apparent alteration in the number of Cry1-ir nuclei (fig. S3). We confirmed and extended the observations in the SCN by quantifying liver mRNA and protein in wild-type and homozygous mice. Peak mRNA levels of Per1, Cry1, Reverbα, and the clock-controlled D site albumin promoter–binding protein (Dbp) were reduced in homozygotes. Fbxl3 and Clock mRNA levels were nonperiodic and equivalent in wild-type and homozygous mutant liver (fig. S4). Quantification of liver proteins by Western blotting confirmed the suppression of Per1 and Per2 in mutants, whereas Cry1 showed intermediate levels of expression at all time points (Fig. 2, B and C). The levels of Cry protein over the circadian cycle can explain the effect of Afh on the clock machinery. Elevated levels of Cry at CT9 to CT12 can enhance the suppression of Clock-Bmal1–dependent transcriptional activation in mutants, whereas reduced levels of Cry at CT18 to CT24 are a consequence of this.

Fig. 2.

Suppression of clock gene and protein oscillations in Afh/Afh mice. (A) Levels of Cry1, Per1, Per2, and Bmal1 mRNA in SCN of +/+ (solid circles) and Afh/Afh (open circles) mice was measured by in situ hybridization. Each value represents the mean ± SEM (n = 3). (B) Levels of Cry1, Per1, and Per2 protein in liver of +/+ (solid circles) and Afh/Afh (open circles) mice was measured by densitometric analysis of Western blots. Each value represents the mean ± SEM (n = 3). Values are normalized to an internal control (actin) and expressed as the fold change from the value at CT3. (C) Representative Western blots of liver from +/+ and Afh/Afh mice showing stabilization of Cry1 and suppression of Per oscillations in mutants (22).

To investigate directly the effect of Fbxl3 on molecular timekeeping within the SCN, the Afh mutant was crossed into the Per2::Luciferase (Per2::Luc) reporter line (21), and we recorded bioluminescence from SCN slices using a photomultiplier assembly (22). SCN from neonatal mice exhibited sinusoidal rhythms of luciferase activity, with a significant lengthening of period in heterozygotes and homozygotes (Fig. 3A). In adult mice, τDD was determined in vivo by wheel-running, and the period of the SCN molecular cycle was tested in vitro. In slices from wild-type mice, the in vitro period was slightly shorter than the in vivo, but overall, the two methods gave comparable results (Fig. 3, A and B). Peak amplitude in homozygotes was low and variable; there was no significant effect of genotype (Fig. 3C). Finally, alignment of SCN waveforms by peak expression revealed that the principal effect of the homozygous mutation was a prolongation of the circadian nadir of Per2::Luc expression (Fig. 3D). This extension of the negative-feedback phase is consistent with in situ hybridization and immunostaining data and is sufficient to explain the lengthened circadian period. Real-time recordings of Per2::Luc from wild-type and heterozygous kidney, lung, and liver all exhibited well-defined circadian cycles for 4 days (fig. S5). Recordings confirmed that the molecular clock in peripheral tissues is sensitive to the Afh allele.

Fig. 3.

Per2::Luc expression in SCN slices reveals a delay in oscillator protein accumulation in Afh/Afh mice. (A) Lengthening of the mutant period of Per2::Luc expression in postnatal and adult SCN slices. Wild-type (blue), heterozygote (red), and homozygote (green), n = 14, 18, and 7 (postnatal) and n = 3, 4, and 3 (adult), respectively, means ± SEM. (**P < 0.01, ***P < 0.001). (B) Comparison of the circadian period observed in vivo (actogram) and in vitro using photomultiplier tube (PMT) analysis: wild-type (solid circles, n = 3), heterozygote (open diamonds, n = 4), and homozygote (open circles, n = 3), means ± SEM. (C) Amplitude of Per2::Luc expression in wild-type (n = 14), heterozygous (n = 18), and homozygous (n = 7) postnatal slices, means ± SEM. (D) (Top) Representative bioluminescence rhythms (cps, counts per second) recorded from organotypic SCN slices in wild-type (blue), heterozygous (red), and homozygous (green) mice. (Bottom) Alignment of waveforms by peak expression (22).

In the accompanying paper (23), Busino et al. found that Fbxl3 binds and drives the ubiquitination and consequent degradation of Cry1 and Cry2. Thus, we investigated the dynamics of Cry degradation in wild-type and mutant tissue. We treated lung tissue slices from Per2::Luc mice with the protein synthesis inhibitor cycloheximide (CHX) and followed the time course of protein degradation (Fig. 4A). Although peak levels of Per2::Luc were lower in Afh/Afh slices (coinciding with the time of CHX treatment), the rate of degradation of Per2::Luc was similar to that of wild-type, with minimal levels detected after 4 hours of CHX treatment. Per2::Luc bioluminescence half-life was 0.54 ± 0.09 hours for wild-type and 0.46 ± 0.07 hours for Afh/Afh (means ± SEM, n = 5). Conversely, Cry1 levels, although lower in wild-type samples after 3 hours of treatment, remained high in Afh/Afh samples after up to 8 hours of treatment. In a further experiment, we looked at the rate of S-tagged Cry2 degradation in CHX-treated COS7 cells cotransfected with wild-type Fbxl3 or Fbxl3Afh protein (Fig. 4B). Although the dynamic of protein degradation differs from that of native tissue, we detected a significant delay in Cry2 degradation in cells transfected with Fbxl3Afh.

Fig. 4.

Increased Cry stability in organotypic tissue slices from Afh/Afh mice and in mammalian cells expressing Fbxl3Afh. (A) Real-time recording of Per2::Luc (top panel) in isolated lung from +/+ (solid circles) and Afh/Afh (open circles) mice, means ± SEM (n = 3). Cycloheximide (CHX, 20 μg/ml) was included in the medium at the peak of Per2::Luc expression, and bioluminescence was recorded after 0, 1, 2, 4, and 6 hours of treatment. Tissue samples were collected after 0, 3, 5.5, and 8 hours of treatment, and protein levels were determined by Western blotting (bottom, WT, wild-type, Afh, homozygous mutant). Blots represent pooled samples from five tissue slice recordings and were repeated to confirm results. (B) (Top) Western blots against S-tagged Cry2 expressed in COS7 cells in the presence of either wild-type Fbxl3 or Fbxl3Afh protein. Cultures were treated with CHX (20 μg/ml) and sampled after 0, 2.5, or 5 hours. Group data (bottom, means ± SEM, n = 3) reveal progressive degradation of Cry2 in presence of wild-type Fbxl3. Degradation of Cry2 was significantly delayed in the presence of Fbxl3Afh (22).

These data provide a role for Fbxl3 in modulating mammalian circadian homeostasis and a biological mechanism whereby this can occur. Fbxl3Afh lengthens τDD by 3 hours, delaying the rate of Cry protein degradation and prolonging the phase of Cry-mediated negative feedback. As a consequence, peak levels of Per, Cry, and Bmal1 mRNA are reduced in homozygotes, as are the corresponding proteins. Despite reduced levels of Cry mRNA, protein levels are stabilized in mutants, with an intermediate level of expression. Consequently, the circadian Cry profile in Afh mutants reflects a compensatory combination of reduced Cry transcription and Cry proteasomal degradation.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1


References and Notes

Stay Connected to Science

Navigate This Article