Cryptochromes modulate E2F family transcription factors

Early 2 factor (E2F) family transcription factors participate in myriad cell biological processes including: the cell cycle, DNA repair, apoptosis, development, differentiation, and metabolism. Circadian rhythms influence many of these phenomena. Here we find that a mammalian circadian rhythm component, Cryptochrome 2 (CRY2), regulates E2F family members. Furthermore, CRY1 and CRY2 cooperate with the E3 ligase complex SKP-CULLIN-FBXL3 (SCFFBXL3) to reduce E2F steady state protein levels. These findings reveal an unrecognized molecular connection between circadian clocks and cell cycle regulation and highlight another mechanism to maintain appropriate E2F protein levels for proper cell growth.


circadian rhythms and the cell cycle
Circadian rhythms allow organisms to anticipate daily environmental changes in accordance with the Earth's rotation. In mammals, nearly every cell has a circadian clock 1 . Mammalian circadian rhythms are generated by a core molecular clock composed of a transcription translation feedback loop (TTFL); the heterodimer circadian locomotor output cycles kaput (CLOCK) and Brain and muscle ARNT-like protein 1 (BMAL1) drives transcription of their own repressors Period (PER 1-3) and Cryptochrome (CRY1-2). Circadian clocks have been found to influence cell cycle progression, possibly through transcriptional regulation of c-Myc, Ccnd1, and/or Wee-1 2,3 . It is becoming increasingly clear that the circadian clock and the cell cycle can interact dynamically, but the molecular mechanisms connecting these two biological oscillators are largely unknown 2,4,5 .
TTFL components have been postulated to have cellular functions outside circadian rhythm generation 2,6,7 . CRY1/2 bind many thousands of unique sites in the genome independent of CLOCK and BMAL1 8 , and interact with several non-circadian transcription factors including nuclear hormone receptors [9][10][11][12] and c-MYC 13 . In addition, CLOCK, BMAL1, PERs, and CRYs have been implicated in cell cycle control. CLOCK and BMAL1 overexpression in a human colon cancer cell line prevented entry into S phase possibly through decreased CYCLIN D1 protein levels 14 . In wildtype mice subjected to partial hepatectomy (PH), hepatocytes entered M phase faster when the surgery was performed in the afternoon compared to those that underwent PH at night, possibly due to lower Wee1 expression in the afternoon, suggesting a link between time of day and cell cycle progression 3 . PER has been implicated in cell cycle control via p16-Ink4A, which is rhythmically expressed throughout the day and is suppressed by PER2 6 . PER2 also modulates the stability of P53 15,16 , which could influence the cell cycle in unstressed conditions and help the cell anticipate genotoxic stress. Another important protein involved in cell cycle control is c-MYC -a widely known proto-oncogene. We recently discovered that CRY2 acts as a co-factor for the SKP-CULLIN-FBXL3 (SCF FBXL3 ) complex to promote degradation of phosphorylated substrates, including c-MYC 13 and Tousled-like kinase 2 (TLK2) 17 .

The E2F family
The Early 2 Factors (E2Fs) are a family of eight winged-helix transcription factors that are key to regulating cell cycle progression from G1 to S phase among other functions 18,19 . E2F1, E2F2, and E2F3a are expressed in early G1 through S phase and are believed to promote S phase initiation by activating genes involved in the G1 to S phase transition 18 . E2F3b, E2F4, E2F6, E2F5, E2F7, and E2F8 act as repressors and are expressed later in the cell cycle 18 . Since E2Fs are crucial in mediating cell cycle progression their protein levels must by tightly regulated. SCF SKP2 interacts with the N-terminus of E2F1 and promotes ubiquitination and degradation of E2F1 20 . The anaphase-promoting complex (APC) promotes E2F1 degradation during M phase and this requires interaction between the C-terminus of E2F1 and the APC adaptor protein CDH1 21 . APC/C CDH1 also stimulates degradation of E2F3, E2F7, and E2F8 22,23 . Recently, CYCLIN F has been shown to promote ubiquitination of E2F1-3a 24 . Here, we demonstrate that CRY1/2 and SCF FBXL3 reduce the steady state protein levels of some E2F family members.
These findings highlight a more widespread substrate repertoire of CRY2 and SCF FBXL3 mediated degradation and further supports the interconnection between circadian clocks and cell cycle progression.

Results
E2F target genes are upregulated in Cry2 −/− compared to Wildtype Mefs. To determine how the global transcriptome is altered in wildtype (WT) and Cry2 −/− cells in an unbiased manner, we performed gene set enrichment analysis (GSEA) on RNA-sequencing (RNA-seq) data from WT and Cry2 −/− mouse embryonic fibroblasts (MEFs). Following synchronization of circadian rhythms with the synthetic glucocorticoid agonist dexamethasone 13 . When analyzed as a group 25 , expression of 200 transcripts defined as hallmark 26 E2F target genes (Table S1) exhibits a small but highly consistent elevation in Cry2 −/− MEFs compared to WT (Fig. 1A,B). We used RNA samples isolated from independently synchronized WT and Cry2 −/− cells to evaluate the reproducibility of  Table S1. (C) Expression of indicated transcripts in primary WT (black) and Cry2 −/− (gray) MEFs at indicated times (hours) after dexamethasone treatment. Data represent mean ± s.e.m. of 2-3 biological triplicates each analyzed in triplicate. *p < 0.05, **p < 0.01, ****p < 0.0001 by two-way ANOVA with Tukey's multiple comparisons for effect of genotype.
Steady state levels of E2F1, E2F4, and E2F8 are decreased in the presence of CRY2 and FBXL3. CRY2 can act as a co-factor to enhance turnover of c-MYC and TLK2 through interaction with FBXL3 13,17 . To determine whether CRY2 and FBXL3 similarly influence E2F family proteins, we measured overexpressed E2F protein levels in the presence or absence of overexpressed human CRY2 (hCRY2.1) and FBXL3. The steady state protein levels of all three E2F proteins studied (E2F1, E2F4, and E2F8) were markedly decreased in the presence of CRY2 and FBXL3 ( Fig. 3A-C). This decrease of E2F1, E2F4, and E2F8 protein levels in the presence of human CRY1/CRY2 and FBXL3 was partially mitigated with treatment of MG-132, a reversible proteasome inhibitor. The impact of MG-132 treatment seems to be greatest for E2F4 in the presence of FBXL3 combined with either CRY1 or CRY2 and for E2F8 in the presence of CRY1 and FBXL3 (Fig. S2). The robust effect on steady state protein levels makes it difficult to interpret effects on the turnover of overexpressed proteins, but the turnover of overexpressed E2F1 appears to increase in the presence of overexpressed FBXL3 and CRY2 (Fig. S3).

Deletion of endogenous Cry1/2 impacts E2F protein levels.
To determine whether endogenous CRY1 and CRY2 impact E2F protein levels, we stably expressed tetracycline-inducible FLAG-tagged human E2F1, E2F4, or E2F8 in WT or Cry-deficient adult mouse ear fibroblasts (AMEFs) (Figs. 4 and Fig. S4). Since E2F proteins are involved in cell cycle regulation, we induced their expression at low and high plating densities (20% and 100%) to capture both proliferating and growth-arrested conditions. E2F1 protein levels are elevated when Cry1 is deleted, but not when Cry2 or both Cry1 and Cry2 are deleted at both plating densities (Fig. 4A-D). Even though E2F1 protein abundance is significantly increased in the absence of Cry1 (Fig. 4B,D), the E2F1 protein levels are highly variable, and loss of Cry1 also seems to impact expression of exogenous E2F1 mRNA (Fig. S5) making us less confident in the biological significance of the increased E2F1 in the Cry1 −/− genotype. E2F4 protein is robustly increased in cells lacking either CRY1 or CRY2, regardless of confluency ( Fig. 4E-H). E2F8 protein levels are dramatically increased in the absence of Cry1, regardless of cell confluency ( Fig. 4I-L) suggesting that CRY1 may play a more important role than CRY2 in the regulation of E2F8 protein abundance. We tried to detect endogenous mouse E2F1, E2F4, and E2F8 protein levels in AMEFs; however, we could not confidently detect endogenous mouse E2F1, E2F4, or E2F8. Although we were not able to confirm its specificity with shRNA or E2F8 −/− cells, the anti-E2F8 antibody detects a protein at the correct molecular weight (indicated by arrow), which is increased in all Cry1/2-deficient genotypes, further supporting our findings regarding E2F8 (Fig. S6). Cry1/2 genotype had little to no effect on the expression of endogenous E2f1 or E2f4 mRNA (Fig. S5). Interestingly, as we have observed for other doxycycline-inducible systems 17 , doxycycline-induced expression of human E2F1 or E2F4 mRNA was sometimes elevated in Cry-deficient cells compared to WT (Fig. S5). Because the upregulation of doxycycline-induced expression of human E2F4 mRNA across the genotypes did not follow the same trends observed at the protein level (Fig. 4), CRYs seem to impact E2F4 protein abundance post-translationally. The robust impact on E2F8 protein level suggests that it may also be a target of CRY1/2-dependent post-translational regulation. All in all, these data support our hypothesis that CRY1 or CRY2 can decrease E2F4 and E2F8 protein levels by recruiting them to SCF FBXL3 .

Discussion
We find that E2F target genes are slightly but consistently upregulated in Cry2 −/− MEFs compared to matched WT control cells. Cry2 −/− MEFs proliferate faster than WT MEFs 13 , which may be effected by higher expression of c-MYC as we have previously documented 13 and could also be influenced by enhanced E2F activity. CRY2 could also influence cell growth through its partner PER2, which has been found to modulate the stability and nuclear translocation of p53 15,27 , and/or through the actions of the core clock activating complex of CLOCK and BMAL1. It is likely that CRY2 acts in multiple ways to influence cell growth; dissecting the relative contributions of each of these mechanisms to increased proliferation in CRY2-deficient cells will require extensive additional investigation.
Here, we find that both activator and repressor E2F family transcription factors interact with FBXL3 and these interactions are enhanced by CRY1 and/or CRY2. Intriguingly, the representative members of the repressor subfamilies, E2F4 and E2F8, interact more strongly with CRYs and FBXL3 than does the canonical activator E2F1. While we did not identify the region(s) of the E2F proteins that interact with FBXL3 and/or CRYs in this study, these differences likely reflect preferential interaction with amino acid sequences that are conserved in the repressor subfamilies. These observations are also consistent with the greater impact of CRY1/2 overexpression (2020) 10:4077 | https://doi.org/10.1038/s41598-020-61087-y www.nature.com/scientificreports www.nature.com/scientificreports/ or genetic deletion on protein levels of overexpressed or endogenous repressor E2F family members, E2F4 and E2F8, compared to the activator E2F1.
The observed greater sensitivity of steady state protein levels of repressor E2F family members to the loss of Cry2 is counterintuitive in the context of the significant elevation of E2F target gene expression in Cry2 −/− MEFs. However, we only assessed the impact of CRY regulation of a select subset of the E2F family; it is possible that E2F2 and/or E2F3 are significantly stabilized in the absence of CRY2 and contribute to the observed effects on gene expression. Conversely, other E2F repressors may be more responsive to the genetic manipulation of Crys than E2F4 and E2F8 are. Given that CRYs repress the transcriptional activity of CLOCK-BMAL1 28 and of several nuclear hormone receptors 9 , CRYs may suppress the transcriptional activity of activator E2Fs or directly mediate transcriptional repression conferred by repressor E2Fs. If they do, that could also explain the increased expression of E2F target genes in Cry2 −/− cells. Indeed, analyzing Chromatin-immunoprecipitation (ChIP) sequencing data from mouse livers across circadian time (GSE39860) 8 reveals that endogenous CRY1 and CRY2 are associated with the genomic loci containing many of the 200 "hallmark" E2F target genes in mouse livers, and at least some of these sites are not bound by other circadian transcription factors, supporting the idea that CRYs could independently regulate E2F family members bound to chromatin (Table S2). Additionally, E2Fs could have unexpected transcriptional activities in the context of CRY-deficient cells. Finally, some evidence suggests that E2F family members cannot be easily classified into activators and repressors 29 , and the increased expression of E2F target genes in Cry2 −/− AMEFs may reflect increased target gene activation by E2F family members that are thought to act as repressors.
A recent study demonstrated that CYCLIN F interacts with and promotes the degradation of activator E2Fs to restrict E2F activity to the S phase of the cell cycle 24 . Our finding that CRYs modulate E2F protein abundance suggests a mechanism by which their abundance and activity could be regulated by circadian cycles. Disruption of circadian rhythms, such as that experienced by shift workers, increases the risk of cancer and other pathologies 30 . E2Fs have been implicated in cancer 18,[30][31][32] , and disruption of their modulation by CRYs could contribute to the elevated cancer risk caused by circadian disruption. Methods cell culture. Cell culture methods were the same as in 17 . In brief, all primary mouse embryonic (MEF) and adult mouse ear (AMEF) fibroblasts cells were prepared from embryos collected at E15.5 or from ear biopsies of adult mice respectively, and were passaged no more than 10 times as in 17 . Cells were grown in complete Dulbecco's Modified Eagle Medium (DMEM) (Invitrogen cat #10569) supplemented with 10% fetal bovine serum (FBS) (HEK 293T cells) or 15% FBS (MEFs and AMEFs), and 1% penicillin and streptomycin. HEK 293T cells were grown in a 37 °C incubator maintained at 5% CO 2 and 20%O 2 (high oxygen) and MEF cells were grown in a 37 °C incubator maintained at 5% CO 2 and 3% O 2 (low oxygen). Transfections were carried out using polyethylenimine (PEI; Polysciences Inc catalog #23966-2) by standard protocols. After 24 hours, the HEK 293T cell media was again replaced and protein extracts were isolated 48 hours post transfection. Cycloheximide (CHX) (Fisher  www.nature.com/scientificreports www.nature.com/scientificreports/ cat# 50255724) was used at a concentration of 100 µg/mL as indicated. Cells were treated with 10 μM MG-132 (Sigma cat# C2211-5MG) for 4-5 hrs prior to lysis.

Generation of viruses and stable cell lines.
Methods were the same as in 17 except lentiviruses expressing pCW-2xFLAG-E2F1 E2F4, or E2F8 were used.
Quantitative RT-PCR (qPCR). Methods were the same as in 17 .