The E3 ubiquitin ligase UBE3A is an integral component of the molecular circadian clock through regulating the BMAL1 transcription factor

Abstract

Post-translational modifications (such as ubiquitination) of clock proteins are critical in maintaining the precision and robustness of the evolutionarily conserved circadian clock. Ubiquitination of the core clock transcription factor BMAL1 (brain and muscle Arnt-like 1) has recently been reported. However, it remains unknown whether BMAL1 ubiquitination affects circadian pacemaking and what ubiquitin ligase(s) is involved. Here, we show that activating UBE3A (by expressing viral oncogenes E6/E7) disrupts circadian oscillations in mouse embryonic fibroblasts, measured using PER2::Luc dynamics, and rhythms in endogenous messenger ribonucleic acid and protein levels of BMAL1. Over-expression of E6/E7 reduced the level of BMAL1, increasing its ubiquitination and proteasomal degradation. UBE3A could bind to and degrade BMAL1 in a ubiquitin ligase-dependent manner. This occurred both in the presence and absence of E6/E7. We provide in vitro (knockdown/over-expression in mammalian cells) and in vivo (genetic manipulation in Drosophila) evidence for an endogenous role of UBE3A in regulating circadian dynamics and rhythmic locomotor behaviour. Together, our data reveal an essential and conserved role of UBE3A in the regulation of the circadian system in mammals and flies and identify a novel mechanistic link between oncogene E6/E7-mediated cell transformation and circadian (BMAL1) disruption.

Figure 1.

Stable integration and expression of E6/E7 immortalized and transformed MEFs and disrupted the circadian clock. (A) Phase contrast micrograph showing E6/E7 cells were capable of growing beyond passage 10 with morphological change while WT primary cells had become senescent. Scale 100 μm. (B) Representative PER2::Luc bioluminescence rhythms in WT (n = 3) and E6/E7-positive (n = 5) cells following three synchronization protocols (dex pulse, forskolin and 50% horse serum). Rhythm dampened rapidly in all cases and period was irregular and significantly longer (28.44 ± 0.77 h compared to 24.83 ± 0.84 h in primary cells; dex synchronized, P < 0.05). Y-axis value x1000 = cpm (photon counts per minute). (C) Immunoblotting (IB) showing BMAL1 protein rhythm in cells synchronized with dex (100 nM). (D) Bmal1 gene expression (qPCR) in dex-synchronised cells, n = 3. ***P < 0.001; ns, not significant.

Figure 2.

E6/E7 expression enhanced BMAL1 ubiquitination and proteasomal degradation through UBE3A. (A) Left panel: IB of steady state BMAL1 in WT and E6/E7 cells. Right panels: integrated density measurement of BMAL1 (n = 3, P < 0.05) and steady state Bmal1 gene expression by qPCR (n = 3, P > 0.05). **P < 0.01, see underlined. (B) Top panel: BMAL1 protein level following cycloheximide (20 μg/ml). Bottom panel: protein degradation rate. (C) Ubiquitination assay of V5-BMAL1 in HEK 293T cells by nickel affinity pull-down. Total levels of V5-BMAL1 in the input lysates are shown in the panel below. MG132 (5 μM) was applied in all cases. (D) V5-BMAL1 levels in HEK 293T cells, transfected with E6/E7, co-transfected with either HA-UBE3A or a dominant negative mutant (C833A). Representative, n = 3.

Figure 3.

UBE3A interacts with BMAL1 both in the presence and absence of E6/E7. (A) Co-IP of HA-UBE3A and V5-BMAL1 in HEK 293T cells. IP was performed using an anti-HA antibody and precipitated V5-BMAL1 or HA-UBE3A was detected by IB using an anti-V5 or anti-HA antibody (top and middle panels). (B) V5 Co-IP showing that V5-BMAL1 could pull down Myc-UBE3A. MG132 (5 μM) was applied in all cases. (C) Co-IP of V5-BMAL1 and myc-UBE3A in HEK 293T cells in the presence or absence of MG132. IP was performed using an anti-V5 antibody and precipitated myc-UBE3A or V5-BMAL1 was detected by IB (top two panels). (D) Endogenous Co-IP between UBE3A and BMAL1 in SW1353 human chondrocytes in the presence of MG132. IP was performed using an anti-UBE3A antibody and the precipitated BMAL1 was detected by IB (top two panels).

Figure 4.

Knockdown of endogenous UBE3A disrupts circadian rhythms in mammalian cells. (A) Representative raw (left) and detrended (right) bioluminescence traces of Bmal1::Luc NIH3T3 cells treated with shRNA #1 (top) and #2 (bottom) against Ube3a. (B) Representative traces of Per2::luc SW1353 cells treated with human Ube3a SiRNA constructs at 10 (mid-grey) and 20 nM (black) doses. Scrambled SiRNA was used as control. (C) Left panels: circadian amplitude analysis (n = 3, shRNA #1 P < 0.01; shRNA #2 P < 0.001; RNAi 10 nM, P < 0.05; RNAi 20 nM, P < 0.01). Right panel: circadian period analysis (n = 3, t-tests: RNAi 10 nM, P < 0.05; RNAi 20 nM, P < 0.05). Note: the rapid dampening of circadian oscillation in shRNA-treated NIH3T3 cells prevented proper evaluation of periodicity. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 5.

Drosophila dUBE3A is necessary for circadian rhythmic behaviour in central clock neurons. (A) IB of total protein from L3 larvae brain, showing complete lack of protein in the dUBE3A-null (DUBE3A³⁵), efficiency of dUBE3A-RNAi knockdown (RNAi) and over-expression (OE) by the pan-neuronal driver elav-Gal4. Two independent blots showed similar results. (B) Expression of GFP driven by the pdf-Gal4 driver, showing restriction to the ventrolateral neuron (LN_v) subgroups of central clock cells in the fly brain that coordinate daily sleep-wake cycles. (C) Upper panels: representative double-plotted actograms for locomotor activity data in flies with Pigment dispersing factor (PDF) neuron-specific dUBE3A RNAi or dUBE3A over-expression, along with relevant controls. LD: white background; DD: grey background. Light and dark bars at the top indicate day (ZT0–12) and night (ZT12–24) during LD, or, subjective day (CT0–12)/subjective night (CT12–24) for DD, respectively. Lower panels: representative periodograms during LD and DD for each genotype. (D) Relative percentage of rhythmic flies for each genotype; n = 10–33.