|Home | About | Journals | Submit | Contact Us | Français|
Using a forward genetics ENU mutagenesis screen for recessive mutations that affect circadian rhythmicity in the mouse, we isolated a long period (~26 h) circadian mutant named Overtime (Ovtm). Positional cloning and genetic complementation reveal that Ovtm is encoded by the F-box protein FBXL3 a component of the SKP1-CUL1-F-box-protein (SCF) E3 ubiquitin ligase complex. The Ovtm mutation causes an isoleucine to threonine (I364T) substitution leading to a loss-of-function in FBXL3 which interacts specifically with the CRYPTOCHROME (CRY) proteins. In Ovtm mice, expression of the PERIOD proteins PER1 and PER2 is reduced; however, the CRY proteins CRY1 and CRY2 are unchanged. The loss of FBXL3 function leads to a stabilization of the CRY proteins, which in turn leads to a global transcriptional repression of the Per and Cry genes. Thus, Fbxl3Ovtm defines a molecular link between CRY turnover and CLOCK/BMAL1-dependent circadian transcription to modulate circadian period.
The mechanism of circadian oscillators in mammals is generated by a cell autonomous autoregulatory transcription-translation feedback loop (Lowrey and Takahashi, 2004; Reppert and Weaver, 2002). In the primary negative feedback loop, the bHLH-PAS transcription factors, CLOCK (and its paralog NPAS2) and BMAL1 (ARNTL) dimerize and activate transcription of the Period (Per1, Per2) and Cryptochrome (Cry1, Cry2) genes (Bunger et al., 2000; Gekakis et al., 1998; King et al., 1997; Kume et al., 1999). As the PER proteins accumulate, they form complexes with the CRY proteins, translocate into the nucleus, and interact with the CLOCK/BMAL1 complex to inhibit their own transcription (Lee et al., 2001). This leads to a fall in the inhibitory complex through turnover, and the cycle starts again with a new round of CLOCK/BMAL1- activated transcription. Additional pathways in the circadian gene network such as the second negative feedback loop (involving Rev-erbα) in the positive limb of the oscillator are thought to add robustness to the circadian mechanism (Preitner et al., 2002; Sato et al., 2004). Finally, post-translational modifications play critical roles in regulating the turnover, cellular localization and activity of circadian clock proteins (Eide et al., 2005; Gallego and Virshup, 2007; Lowrey et al., 2000).
Despite this progress, it is clear that a significant number of genes that strongly influence and regulate circadian rhythms in mammals remain to be discovered and identified (Shimomura et al., 2001; Takahashi, 2004). Forward genetic screens have been one of the most effective tools for circadian gene discovery (Takahashi, 2004; Takahashi et al., 1994; Vitaterna et al., 1994), and we have used this approach to screen the mouse genome for circadian rhythm mutants generated in the Neurogenomics Project in the Center for Functional Genomics at Northwestern University (Vitaterna et al., 2006).
Using a forward genetic approach in mice, we report here the isolation and positional cloning of a novel circadian period mutant in mice. This mutation reveals a defect in an orphan member of the F-box protein family, FBXL3, which leads to a stabilization of CRY protein levels and a global repression of Per and Cry gene transcription.
In an N-ethyl-N-nitrosourea (ENU) recessive screen using the BTBR T+ tf/J (BTBR/J) inbred mouse strain (Siepka and Takahashi, 2005a), we identified a long period mutant with a 25.8-h period length, which is more than 10 standard deviations from the mean in this screen (Figure 1). This mutant, named Overtime, is semidominant and autosomal. In crosses with C57BL/6J and C3H/HeJ mouse strains, the homozygous mutant phenotype is distinct with an average circadian period of 26.2 h and a range of about 25–28 h, which is indistinguishable from the original mutant (Figure 2A–D). Ovtm maps to a 1.7 cM interval on chromosome 14 between D14Mit265 (5 recombinants/642 meioses) and D14Mit197 (6 recombinants/642 meioses) (Figure 2E). This region corresponds to a 4 Mb interval and contains 18 open reading frames (Assembly build NCBI m36 December 2005, Gene build June 2006), none of which corresponds to previously known circadian clock genes (Figure 3A) (Lowrey and Takahashi, 2004).
Several genes in the nonrecombinant interval were likely candidates for Ovtm. Protein degradation machinery has as an important role in the clock mechanism (Gallego and Virshup, 2007) and there are several ubiquitin-pathway related genes in this interval (Uch13, Lmo7, Fbxl3, Phr1, 2610206B13Rik and Ndfip2). Because the entire 4 Mb Ovtm interval has a low polymorphism rate and shares the identical haplotype with the mapping strains, the resolution of meiotic mapping could not be increased. This region of the genome is part of the piebald deletion complex which contains several developmentally important genes (Peterson et al., 2002). This may have caused selection of a limited number of ancestral haplotypes in this region of chromosome 14 which would in turn lead to a low polymorphism rate.
We sequenced all annotated exons for the 18 candidate genes in the Ovtm interval and found only a single nonsynonymous point mutation within the coding region of Fbxl3. There is a single base transition from A to G in exon 5 of Fbxl3 in Ovtm mice as compared to wildtype BTBR/J mice. This mutation co-segregated perfectly with the long-period phenotype of Ovtm/Ovtm mice (0 recombinations out of 642 meioses, Figure 2D, E). The point mutation converts amino acid residue 364 from isoleucine to threonine in FBXL3 (Figure 3A, B). This isoleucine residue is highly conserved in FBXL3 from vertebrates and in the mouse paralog FBXL21 (Figure 3B). FBXL3 is a member of the F-box protein family with leucine rich repeats (LRR) which is defined by its founding member, SKP2 (S-phase kinase-associated protein-2) (FBXL1) (Jin et al., 2004). SKP2 is the F-box protein moiety in the SKP1-CUL1-F-box-protein (SCF)SKP2 E3 ubiquitin ligase complex which mediates the recognition and ubiquitination of the CDK2 inhibitor, p27Kip1, to target it for proteasomal degradation (Cardozo and Pagano, 2004). Structural studies show that CUL1 provides a rigid scaffold upon which SKP1 and RBX1 subunits assemble: SKP1 interacts with SKP2 via its N-terminal F-box motif (Zheng et al., 2002). Recent work shows that the LRR region and the C-terminal tail of SKP2 interact with the substrate, p27Kip1, and the accessory protein, CKS1 (Hao et al., 2005). FBXL3 which has 11 LRRs can be aligned with SKP2 which has 10 LLRs based on its protein structure; however the C-termini of SKP2 and FBXL3 are not conserved likely due to the recognition of different substrates (Hao et al., 2005). The Ovtm I364T mutation occurs in the C-terminus of FBXL3 between LRR10 and LRR11 where the alignment with SKP2 becomes divergent (Figure 3B). Because the LRR domains of F-box proteins are involved with substrate recognition with the SCF complex, we hypothesized that the I364T mutation could alter the interaction of FBXL3 with its substrates.
Because we isolated only one mutant allele of Fbxl3 and either a second independent allele, rescue, or functional evidence is required for proof in positional cloning (Takahashi et al., 1994), we used genetic complementation tests to confirm that Ovtm was allelic with Fbxl3. A search of the International Gene Trap Consortium (IGTC) database (http://www.genetrap.org/) revealed three gene traps in Fbxl3, and we obtained one gene trap ES cell line (S13–12G1) from Philippe Soriano (Chen et al., 2004). ES cells were microinjected into blastocysts to produce chimeras and 19 out of 19 mice were chimeric. Because the ES cell contribution in these chimeras was >95%, they were mated directly to Ovtm/Ovtm mice to test for germline transmission and genetic complementation simultaneously. The crosses show that Ovtm and the Fbxl3 gene trap (GT) fail to complement each other, thus providing independent and definitive evidence that Ovtm is an allele of Fbxl3 (Figure 3C). Interestingly, the period length of GT/Ovtm mice is indistinguishable from Ovtm homozygotes suggesting that the Ovtm mutant allele is likely a hypomorphic or loss-of-function allele.
The mRNA expression of Fbxl3 itself in the mouse is ubiquitous, but enriched in the brain (data not shown). Northern blot analysis of Fbxl3 from liver in wildtype (WT) and Ovtm mice did not reveal any significant differences in transcript size or abundance between the genotypes. In addition, there was no obvious circadian rhythm in Fbxl3 mRNA expression in the liver. In situ hybridization of Fbxl3 shows clear RNA expression in the suprachiasmatic nucleus (SCN) and throughout the brain (see Supplemental Material).
Because FBXL3 is likely a component of an SCF E3 ubiquitin ligase complex, we examined the in vivo expression patterns of circadian clock proteins in mouse tissues to explore whether Ovtm might alter their abundance by affecting degradation. WT and Ovtm mice were housed in running wheel cages to measure activity rhythms and were transferred to constant darkness (DD) for 10–14 days. Animals were killed under infrared illumination every 3 h in DD based on circadian time (CT) to compensate for the differences in circadian period between WT and Ovtm mice. Figure 4A shows the expression patterns of the clock proteins, CRY1, CRY2, PER1, PER2, CLOCK and BMAL1 in liver and cerebellum. In WT mice, there were low amplitude rhythms of CRY1 and CRY2 and high amplitude rhythms of PER1 and PER2 as reported previously (Lee et al., 2001). In Ovtm liver tissue, CRY1 and CRY2 protein patterns were not significantly altered; however, PER1 and PER2 levels were significantly reduced (Figure 4B). In the cerebellum, the effects of Ovtm were more striking. Although CRY1 levels were not different, CRY2 levels were significantly elevated in Ovtm mice consistent with the hypothesis that CRY degradation is impaired. In addition, there were very clear reductions in the levels of PER1 and PER2. The levels of CLOCK and BMAL1 were not detectably changed in either tissue in Ovtm mice (Figure 4A). The reduction of PER1 and PER2 levels in Ovtm mice is significant in both liver and cerebellum, and this finding is unexpected and counterintuitive. We would have expected to see an increase rather than a decrease in protein abundance if the PER proteins were targets of FBXL3 because the Ovtm mutation is a loss-of-function mutation. This suggests that it is unlikely that the PER proteins are targets of FBXL3 and that the reduction in PER levels could occur as a consequence of the negative feedback on CLOCK/BMAL1-dependent transcription.
To explore the reasons for the reduction in PER1 and PER2 protein abundance, we profiled the in vivo circadian mRNA expression patterns for Cry1, Cry2, Per1, Per2 and Dbp in the liver and cerebellum of mice maintained in DD. As shown in Figure 4C, the Ovtm mutation caused significant reductions in the mRNA abundance of all of these cycling transcripts with the strongest effects being seen with Cry1 and Per2 in the cerebellum. At the mRNA level, both a delay in the peak time and a reduction in abundance can be seen. Importantly, although CRY1 and CRY2 protein levels were not lower in Ovtm mice, the corresponding mRNA levels for Cry1 and Cry2 are significantly reduced in both tissues. Similar reductions in the mRNA levels for Cry1 and Per2 in Ovtm mice could also be observed in the SCN using in situ hybridization (see Supplemental Material). In addition, mRNA levels for the cycling CLOCK target gene, Dbp (Ripperger and Schibler, 2006), were very strongly reduced in Ovtm mice. Thus, the mRNA profiling experiments point to an interesting and unexpected consequence of the Ovtm mutation: a reduction in steady-state mRNA expression of Cry1, Cry2, Per1, Per2 and Dbp which are all transcriptional targets of the CLOCK/BMAL1 complex (Gekakis et al., 1998; Kume et al., 1999; Ripperger and Schibler, 2006; Yoo et al., 2005).
Comparison of the effects of Ovtm on protein vs. mRNA abundance suggests that there are two different effects on the expression of the CRY and PER proteins. The PER protein levels appear to be reduced as a consequence of reduced transcript levels. By contrast, the CRY protein levels are not reduced even in the face of reduced transcript levels. This suggests that potential reductions in CRY protein levels caused by reduced Cry transcript levels could be compensated by a reduction in protein degradation.
The Pagano laboratory has found that FBXL3 targets CRY proteins for ubiquitination and degradation (personal communication). To confirm these results and determine whether the Ovtm mutation affects interactions with CRY, we examined the interaction of FBXL3 or OVTM with circadian clock proteins by immunoprecipitation assays. First, we used tagged Fbxl3, Ovtm and βTrCP1 expression constructs to transfect NIH 3T3 cells which contain circadian oscillators and express clock proteins (Nagoshi et al., 2004). Cell lysates were immunoprecipitated with anti-FLAG or anti-V5 antibodies, and interactions with CRY1, CRY2, PER1 and PER2 native proteins were probed on Western blots (Figure 5A). Both FBXL3 and OVTM interacted strongly with native CRY1 and CRY2 proteins. Very weak or no interaction of FBXL3 was seen with PER1 and PER2, especially in comparison to that seen between the PERs and βTrCP1, an F-box protein known to interact with the PERs (Eide et al., 2005; Shirogane et al., 2005) (Figure 5A). In all experiments, there was a discernibly stronger interaction of the CRY proteins with FBXL3 relative to OVTM, but the difference was subtle.
We also used tagged proteins in co-immunoprecipitation assays in 293A cells which are easily transfected and express relatively low levels of clock proteins. Both FBXL3 and OVTM interacted strongly with CRY1 and CRY2 (Figure 5B). To explore the weak interaction of FBXL3 with PER proteins, tagged Per constructs were also tested. All three PER proteins showed interactions with FBXL3 and OVTM, however, the strongest interactions were seen with PER2. Because PER2 interacts very strongly with CRY1 (Griffin et al., 1999; Kume et al., 1999; Lee et al., 2001), it is likely that the interactions seen here with FBXL3 may be indirect via CRY1.
To determine whether OVTM is less efficient than FBXL3 in inducing the degradation of CRY1, we compared the effects of FBXL3 and OVTM on the stability of CRY1 following cycloheximide treatment to prevent de novo protein synthesis in transfected cells (Figure 5C). In 293A cells, transfected CRY1 is relatively stable with a half-life of 6.4 h (Figure 5D). Addition of FBXL3 or OVTM increased the turnover of CRY1 significantly with FBXL3 leading to a 1.7 h CRY1 half-life and OVTM leading to a 2.5 h CRY1 half-life (Figure 5D). Thus, FBXL3 expression leads to the degradation of CRY1, and this degradation is blocked by the 26S proteasomal inhibitor MG132 (Figure 5C). OVTM is less effective than FBXL3 in causing CRY1 degradation under these conditions, and the half-life parameter, K, is significantly different (p=0.016) between FBXL3 and OVTM in the model-based nonlinear exponential decay analysis with GraphPad Prism 4. Interestingly, the turnover of FBXL3 is also affected by the OVTM mutation (Figure 5C, bottom). FBXL3 is relatively stable with a half-life of greater than 7 h; whereas, OVTM has a much shorter half-life of 2.7 h (Figure 5D). Therefore, there are two effects of the OVTM mutation: a reduction in proteasome-mediated CRY1 degradation and a decreased stability of the OVTM protein itself, both of which could contribute to a loss-of-function phenotype.
To determine whether these changes in CRY1 stability seen in transfected cells are physiological, we used fibroblasts prepared from either WT or Ovtm mice and determined the half-lives of native CRY1 and PER2 proteins in these cells. Fibroblast cultures were synchronized with 10 μM forskolin and treated with cycloheximide. The half-lives of CRY1 and PER2 were then determined by Western blot against native proteins. As shown in Figure 5E and F, the half-life of CRY1 in Ovtm fibroblasts is extremely long (9 h) as compared to WT cells (half-life = 5.2 h). The overall levels of PER2 in Ovtm fibroblasts were very low similar to that seen in the cerebellum. When the half-life of PER2 was determined, however, there was no detectable difference in the half-life of PER2 in WT and Ovtm cells (Figure 5F). Thus, these experiments in fibroblasts from WT and Ovtm mice show that native CRY1, but not PER2, turnover is specifically altered by the Ovtm mutation. These experiments strongly suggest that OVTM is selectively defective in its ability to target CRY1 for degradation by the proteasome.
To understand why the steady state transcript levels of Per and Cry are reduced in Ovtm mice, we estimated transcription rates by measurement of pre-mRNA levels and mRNA stability by half-life experiments. In both liver and cerebellum, Cry1 and Per2 pre-mRNA levels are strongly correlated with steady-state mRNA abundance levels with lower pre-mRNA levels observed in Ovtm tissue (Figure 6A). As seen for steady state mRNA levels, the reduction in Cry1 and Per2 pre-mRNA levels was most clear in the cerebellum. Thus Ovtm strongly reduces the transcription of Cry1 and Per2 in vivo consistent with the hypothesis that transcription rates account for the reduction in steady state mRNA abundance. To reinforce this hypothesis, we also measured the half-lives of mRNAs for Cry1, Cry2, Per1 and Per2. WT and Ovtm fibroblasts were treated with Actinomycin D to inhibit de novo transcription, and mRNA half-life was determined by quantitative PCR. As shown in Figure 6B, although there are differences in the initial mRNA levels of Cry1 and Per2 as seen in the cerebellum, the half-lives of all four transcripts are not significantly different in WT and Ovtm cells using model-based nonlinear exponential decay analysis with GraphPad Prism 4. Thus, taken together these experiments argue that the reduced mRNA levels of the Cry and Per genes in Ovtm mouse tissues are due to a reduction in transcription of these genes.
To address the functional consequences of the mutation, we assessed whether Fbxl3 or Ovtm could affect Clock/Bmal1-mediated transcriptional activation. We asked whether Fbxl3 or Ovtm could modulate the inhibitory effects of Cry1 on Clock/Bmal1-induced transcription using a Per2 E2 enhancer driving luciferase (Yoo et al., 2005). In Figure 6C, Clock/Bmal1 transfection activates luciferase activity ~6-fold. As recently reported (Kwon et al., 2006), treatment with the proteasomal inhibitor MG132 significantly reduces the magnitude of Clock/Bmal1 activation. As expected, Cry1 co-transfection strongly inhibited Clock/Bmal1 activation (Griffin et al., 1999; Kume et al., 1999). The addition of Fbxl3 strongly diminished the Cry1-dependent inhibitory effects on Clock/Bmal1 activation; while addition of Ovtm was much less effective than that seen with Fbxl3. Western blots confirmed that the expression levels of FBXL3 and OVTM were comparable. However, Fbxl3 and Ovtm transfection reduced the levels of CRY1 consistent with the CRY1 degradation experiments in Figure 5C. Thus, FBXL3 and OVTM both reduced CRY1 expression levels and the reduction in CRY1 is correlated with a reduction in CRY1-dependent inhibition of CLOCK/BMAL1 activation (linear regression analysis of CRY1 level vs. PER::LUC activity: slope is significant, p = 0.0181, Pearson’s correlation coefficient R = −0.67). Thus, one possible explanation for the difference in CLOCK/BMAL1-dependent transcription in WT and Ovtm tissues could be that FBXL3 normally provides a steady state turnover of CRY via proteasomal degradation that is reduced by the Ovtm mutation. This could then lead to an accumulation of CRY to suppress CLOCK/BMAL1-dependent transcription.
We have shown that the ENU-induced Overtime mutant is caused by an I364T mutation in the mouse FBXL3 protein, a member of the F-box protein with leucine rich repeats family (Jin et al., 2004). The OVTM protein is less efficient than FBXL3 in degrading CRY1 thus providing genetic evidence that FBXL3 appears likely to be a primary F-box protein within an SCF E3 ubiquitin ligase complex (Cardozo and Pagano, 2004) that targets the CRY proteins for degradation in the proteasome. The I364T OVTM mutation lengthens circadian periodicity ~2.5 h in mice. We propose that the phenotypic effects of the Ovtm mutation occur primarily through two mechanisms: (1) loss of FBXL3 function leading to stability of CRY1 protein; and (2) repression of CLOCK/BMAL1-dependent transcriptional activation. These two processes lead to a striking reduction in the expression of the PER proteins which is caused by a reduction in transcription of the Per genes. By contrast, the levels of CRY are not reduced by the Ovtm mutation despite lower rates of Cry transcription. Indeed in the cerebellum, the level of CRY2 protein is significantly higher as expected (perhaps due to the slightly higher relative rate of transcription of Cry2 compared to Cry1 in the cerebellum in Ovtm vs. WT mice). However, the levels of CRY1 protein were not elevated in Ovtm mice. Because the majority of CRY protein is cytoplasmic (Lee et al., 2001), we also examined levels of CRY1 in the nucleus but did not detect an increase in CRY levels in nuclear fractions from the liver of Ovtm mice in comparison to WT (data not shown). In retrospect, because the transcription of the Cry1 gene is strongly attenuated by the Ovtm mutation, it is surprising that CRY1 protein levels are not lower. The low rate of CRY1 protein degradation in Ovtm tissues must offset the lower synthesis of CRY1 so that the steady state abundance of CRY1 protein is similar in Ovtm and WT mice. Importantly however, because the turnover rate of CRY1 is reduced, the clearance of CRY1 will be prolonged even if the initial steady-state abundance levels are comparable. This would then lead to a prolongation of the CRY-dependent repression phase of the circadian cycle. If such a prolongation extended CRY repression for 2–3 hours, the period lengthening phenotype seen in Ovtm mice would follow as a consequence. It will be of great interest to test existing biochemical models of the circadian oscillator (Forger and Peskin, 2003; Leloup and Goldbeter, 2004) for these phenotypic effects. In particular, it will be important to determine whether global changes in the set point for transcription rates of the clock genes occur as a result of decreases in the degradation of the CRY proteins and whether the changes in CRY turnover rates can account for the period lengthening phenotype of Ovtm mice. In addition, it will be interesting to explore the reasons underlying the tissue-specific differences observed in Ovtm mice.
These results highlight the significance of an SCFFBXL3 E3 ubiquitin ligase complex in regulating the stability and kinetics of CRY degradation. The specificity of the FBXL3 interaction with the CRY proteins is striking and suggests that additional F-box proteins may regulate other circadian clock components. The first example of an F-box protein playing a role in circadian rhythmicity was the Arabidopsis gene ZEITLUPE (ZTL) which encodes an F-box protein with an N-terminal LOV domain and C-terminal kelch repeats (Somers et al., 2000). ZTL targets the Arabidopsis clock protein, TOC1, for degradation by the proteasome and is thought to regulate circadian period by controlling TOC1 stability (Mas et al., 2003). In addition, the F-box protein, FKF1, mediates the cyclic degradation of CDF1, a repressor of the photoperiodic gene CONSTANS (Imaizumi et al., 2005), and the F-box protein, AFR, is a positive regulator of phytochrome A-mediated light signaling in Arabidopsis (Harmon and Kay, 2003). FBXL3 is the second example of a mammalian F-box protein regulating the circadian clock proteins. The first example is βTrCP which has been shown to interact directly with the PER proteins (Eide et al., 2005; Shirogane et al., 2005). Evidence for βTrCP in the circadian pathway first emerged from Drosophila in which it was shown that Slimb, the ortholog of βTrCP, regulated circadian expression of PER and TIM (Grima et al., 2002; Ko et al., 2002). Interestingly, in Neurospora, the ortholog of βTrCP, FWD1, regulates the degradation of the clock protein, FREQUENCY (He et al., 2003). More recently, JETLAG, the Drosophila ortholog of Fbxl15, has been shown to play a critical role in light-induced TIM degradation by the proteasome (Koh et al., 2006). In Drosophila two different SCF complexes appear to control TIM levels: a circadian pathway involving Slimb and a light-dependent pathway involving JET (Koh et al., 2006). It will be interesting see whether similar types of mechanisms are conserved in mammals. Because of the differences in the roles of the PER and CRY proteins in Drosophila and in mammals (Allada et al., 2001; Young and Kay, 2001), where CRY is primarily a circadian repressor (not a photoreceptor), FBXL3 appears to function in a circadian SCF complex-mediated pathway. Unlike Drosophila PER, the PER1 and PER2 proteins in mammals are transcriptionally induced by light in the SCN. It will be interesting to see whether βTrCP functions in a circadian or in a light-dependent SCF pathway for PER degradation (analogous to the TIM protein in Drosophila).
Using a forward genetics approach in mice, we have identified, FBXL3, as a new molecular component of the negative feedback loop that generates circadian rhythmicity. Our results both confirm biochemical studies from the laboratory of Pagano (personal communication) showing that the SCFFBXL3 E3 ubiquitin ligase complex targets CRY1 for ubiquitination and provide in vivo evidence that links CRY stability to CLOCK/BMAL1-mediated transcriptional activation. In Ovtm mice the fine equilibrium between activation and repression of clock gene transcription is disturbed due to the mutation. Future studies will further clarify how this equilibrium between activators and repressors leads to the generation of circadian periodicity.
BTBR T+ tf/J (Stock #002282), C57BL/6J (Stock # 000664), C3H/HeJ (Stock # 000659) and 129S1/SvImJ (Stock #002448) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). CD-1 (Strain code 022) female mice were purchased from Charles River Laboratories (Wilmington, MA). All mice were housed under LD12:12 unless otherwise noted. All animal care and experimental treatments were in accordance with Northwestern University guidelines for animal care and use.
N-Ethyl-N-nitrosourea (ENU) (Sigma cat #N3385) was prepared as described previously (Siepka and Takahashi, 2005a). Six-week-old male BTBR/J mice were injected with 250 mg/kg body weight of ENU. After a 6-week recovery period, the ENU treated mice were mated with wildtype BTBR/J females to generate generation 1 (G1) males. G1 males were mated with wildtype BTBR/J females to produce G2 females. Four G2 females were backcrossed to their G1 fathers to produce G3 mice for phenotyping. Five G3 mice from every G2 backcross (20 mice per G1 pedigree) were phenotyped to ensure an 85% probability of detecting a recessive mutation. 3198 G3 mice from 216 G1 pedigrees were successfully screened from a total of 3609 G3 mice produced.
Mice (6–10 weeks of age) were placed in individual running wheel cages and activity recorded using the ClockLab data collection system (Actimetrics, Wilmette, IL). After one week on LD12:12, the mice were released into constant darkness (DD) for an additional 3 weeks. Free running period was measured by linear regression analysis of activity onsets from data collected during the DD portion of the assay using ClockLab software (Siepka and Takahashi, 2005b).
Ovtm/Ovtm male mice were mated to C3H/HeJ female mice to create [Ovtm/+ x C3H/HeJ] F1 mice. The F1 females were backcrossed to produce 156 [(Ovtm/+ x C3H/HeJ) x Ovtm/Ovtm] N2 mice for mapping. Alternately, Ovtm/Ovtm female mice were mated to C57BL/6J male mice to create [Ovtm/+ x C57BL/6J] F1 mice. These F1 siblings were intercrossed to produce 1480 [Ovtm x C57BL/6J] F2 mice for mutation mapping. Wheel-running behavior of all mapping mice was collected and analyzed as described above. Only presumptive Ovtm homozygous mice (321 genotyped for a total of 642 meioses) were used for genetic mapping as described in the Supplemental Materials.
Sequencing was performed either in our laboratory or at the Harvard Partners Genome Center, Cambridge MA. Genomic DNA was extracted from tail tips from 2 wildtype BTBR/J mice and two Ovtm/Ovtm mice. All annotated exons from MGSC 36 within the 4 MB interval between D14Mit265 and D14Mit197, which contains Ovtm, were amplified by PCR using genomic DNA templates. Most exons were small enough to be amplified and sequenced in a single amplicon (~600 bp). Some larger exons however, were sequenced by multiple overlapping 600 bp amplicons. The following genes and the number of exons and amplicons sequenced were as follows: Tcb1d4 (19 exons, 20 amplicons), Commd6 (4 exons, 3 amplicons), Uch13 (9 exons, 8 amplicons), Q8C3P1 (1 exon, 2 amplicons), Lmo7 (27 exons, 31 amplicons), Irg1 (5 exons, 8 amplicons), Cln5 (4 exons, 5 amplicons), Fbxl3 (5 exons, 7 amplicons), Phr1 (86 exons, 88 amplicons), AK190093 (3 exons, 0 amplicons—due to the highly repetitive nature of the sequence), Scel (31 exons, 30 amplicons), Slain1 (7 exons, 9 amplicons), Ednrb (8 exons, 8 amplicons), D130079A08Rik (2 exons, 3 amplicons), Pou4f1 (3 exons, 4 amplicons), 2610206B13Rik (6 exons, 9 amplicons), 1700009P03Rik (22 exons, 21 amplicons), Ndfip2 (8 exons, 7 amplicons). A table of primers sequences used in this project is available on request. Note that all primers also contained M13 tags (M13 Forward 5′-TGTAAAACGACGGCCAGT-3′ and M13 reverse 5′-AACAGCTATGACCATG-3′) at their 5′ ends. M13 universal primers were used for all sequencing reactions. Sequencing data were analyzed using Sequencher 4.5 software (Gene Codes Corporation, Ann Arbor, MI).
Mice were genotyped for the Ovtm mutation using real-time PCR to detect single nucleotide polymorphisms (SNPs). PCR reactions were carried out in 6 μl volumes using 10–25 ng template genomic DNA with 1x Master Mix for SYBR® Green (Part # 4309155, Applied Biosystems Inc, Foster City, CA) 5 pmoles forward (wildtype forward primer = 5′-GTTGCAAAAATTTGTCAGCaAt-3′ or Ovtm forward primer = 5′-GTTGCAAAAATTTGTCAGCaAc-3′) and 5 pmoles reverse primer (5′-CCCCACACATCTTCACAAACT-3′) in MicroAmp 96- or 384-well Optical Reaction Plates (Applied Biosystems, Inc.). Thermo-cycling reaction conditions were as follows: 40 cycles of 95°C for 15 seconds and 60°C for 30 seconds. Data was collected and analyzed with either an ABI Prism 7700 Sequence Detector or an ABI 7900HT Fast Real-Time PCR System (Applied Biosystems, Inc.).
Mouse embryonic stem cell gene trap line S13–12G1 was obtained from the Soriano Lab Gene Trap Resource at the Fred Hutchinson Cancer Research Center, Seattle WA (http://www.fhcrc.org/science/labs/soriano/trap.html). The cell line contains a ROSAFARY gene trap vector (Chen et al., 2004) that disrupts expression of Fbxl3 to create a null allele of Fbxl3. The maintenance of ES cells, production of chimeric mice and genotyping of the Fbxl3 gene trap are described in the Supplemental Materials.
Immunoblotting was performed as described previously (Yoo et al., 2004). Antibodies against PER1, PER2, CRY1, CRY2, CLOCK, BMAL1 were made as described previously (Lee et al., 2001). Immunoprecipitation was performed 32 h after transfection including an 8 h treatment of 20 μM MG132 (EMD Biosciences). Harvested cells were homogenized in EB (20mM HEPES pH7.5, 100mM NaCl, 0.05% TritonX-100, 1mM EDTA, 20mM NaF, 1mM Na3OV4, Complete mini, Roche) and centrifuged at maximum speed for 10 min at 4 °C. The supernatants were transferred to fresh tubes and incubated with 2 μg of anti-FLAG or anti-V5 antibodies for 2 h at 4°C. 10 μl 50% protein-A slurry (GE) was added, and the incubation continued for an additional 1.5 h.
Supernatants were discarded after centrifugation at 3000 rpm at 4 °C for 1 min and protein-A beads were washed 3 times with 1 ml EB. Pellets were resuspended in 20 μl of 2 SDS sample buffer and boiled for 3 min. Protein samples were separated by 10% or 6% SDS-polyacrylamide gel then transferred to a nitrocellulose membrane (NEN). 5% nonfat dry milk in Tris-buffered saline containing 0.05% Tween20 was used for blocking buffer. Anti-HA (Roche), anti-FLAG (Sigma), anti-V5 (Invitrogen) were used to detect CRY1, FBXL3 (or OVTM) and βTRCP1, respectively.
Total RNA was isolated from frozen tissues with Trizol reagent (Invitrogen) then treated with DNase (Ambion). 0.5 μg of DNase-treated total RNA was reverse-transcribed using Taqman reverse transcription reagent (Roche). The cDNA equivalent to 10 ng of total RNA was PCR-amplified in either an ABI Prism 7700 Sequence Detector or an ABI 7900HT Fast Real-Time PCR System (Applied Biosystems, Inc.). RNA expression was quantified by using SYBR green real-time PCR analysis. Analysis was performed as described previously (Wilsbacher et al., 2002). Primer sequences are described in the Supplemental Material.
Primary fibroblasts were isolated from the ear tissue of wildtype and Ovtm mutant mice. Briefly, minced ear tissue was incubated with collagenase III (1 mg/ml) and trypsin (0.05%). After 10 min incubation at 37°C minced tissue was allowed to adhere onto a 100 mm culture dish. Approximately 10 days after culture in DMEM (Mediatech) supplemented with 10% fetal bovine serum, ear fragments were removed and outgrowing fibroblasts were replated into T75 flasks by trypsinization. NIH 3T3 and 293A (ATCC) cells were cultured in DMEM (Mediatech) supplemented with 10% fetal bovine serum. For immunoprecipitations, 1 × 106 cells were plated into the 100 mm dishes one day before transfection, Effectene reagent (Qiagen) was used to transfect DNA according to the manufacturer’s protocol. For the luminescence assay, 293A cells were plated the day before transfection at 2 × 105 cells per well in 6-well plates. Cells were transfected with indicated vectors by using Effectene reagent. 48 h after transfection, cells were lysed and luminescence was measured from 20 μl of lysate in the Luciferase Assay System (Promega) using a luminometer (AutoLumet Plus; Berthold Technologies).
We thank Michele Pagano and Pat Nolan for communicating unpublished results, Steven Reppert for clock gene expression plasmids, Lawrence H. Pinto for help with mutagenesis; Renee McGurk, Min Cheng, Dawn Olson and Jennifer Wakowiak for mouse production, phenotyping, genotyping and DNA sequencing; Ethan Buhr, Kazuhiro Shimomura and Ming-Lee Chow for assistance with experiments; Andrew Schook for in situ hybridization experiments; Kate Montgomery and Alex Tamburino at Harvard Partners Genome Center for DNA sequencing and analysis of candidate genes. Research supported by NIH grants U01 MH61915, P50 MH074924 and R01 MH078024 to J.S.T. and R01 NS053616 to C.L. J.S.T. is an Investigator in the Howard Hughes Medical Institute.