The RNA-binding protein hnRNP Q represses translation of the clock gene Bmal1 in murine cells

Most living creatures have a circadian rhythm that is generated by a precisely regulated transcriptional–translational feedback loop of clock genes. Brain and muscle ARNT-like 1 (BMAL1) is one of the core clock genes and transcription factors that represents a positive arm of this autoregulatory circadian clock system. Despite the indispensable role of BMAL1 in the circadian rhythm, the molecular mechanisms underlying translational control of BMAL1 are largely unknown. Here, using murine NIH-3T3 cells, gene constructs, and a variety of biochemical approaches, including RNAi- and luciferase reporter gene–based assays, along with immunoblotting, in vitro transcription, quantitative real-time PCR, and real-time bioluminescence experiments, we show that translation of Bmal1 is negatively regulated by an RNA-binding protein, heterogeneous nuclear ribonucleoprotein Q (hnRNP Q). Interestingly, we found that hnRNP Q rhythmically binds to a specific region of the Bmal1 mRNA 5′ UTR and controls its time-dependent expression. Moreover, we demonstrate that knockdown of hnRNP Q modulates BMAL1 protein oscillation amplitude without affecting mRNA rhythmic patterns. Furthermore, hnRNP Q depletion increases the mRNA oscillation amplitudes of BMAL1-regulated target genes. Together, our results suggest that hnRNP Q plays a pivotal role in both Bmal1 translation and BMAL1-regulated gene expression.

The majority of living organisms have a 24-h circadian clock that is responsible for physiological and behavioral rhythmic phenomena. These rhythmic processes are precisely maintained by transcriptional-translational feedback loops of several circadian clock genes (1). A master clock of the circadian system is in the hypothalamic suprachiasmatic nucleus of the brain (2), and this central pacemaker synchronizes various peripheral clocks via a combination of neural and humoral signaling (3). Peripheral tissues also harbor their own clock, and most cells, even cultured fibroblasts (4,5), possess a self-sustained and cell-autonomous circadian rhythm oscillator. This endogenous cellular oscillator is composed of an autoregulatory feedback loop that includes positive elements of the Bmal1 and Clock heterodimer that activates transcription of Period (Per), 2 Cryptochrome (Cry), Rev-erb␣, and Ror␣. In addition, there is a negative arm of this internal oscillator, a Per and Cry heterodimer that suppresses the function of Bmal1 and Clock complex. Tightly regulated expression of these core clock genes is essential for maintaining the biological rhythm of the organism.
Among the core clock genes, Bmal1 receives much attention because it has critical roles in circadian rhythms as well as various physiological processes. For example, absence of Bmal1 results in immediate and complete impairment of circadian rhythmicity and reduced locomotor activity (6). This reduced activity is explained by a report showing that Bmal1-deficient mice develop joint ankylosis because of ossification in tendons and ligaments of the hind limbs (7). In addition, both glucose metabolism and insulin signaling are abolished in Bmal1 mutant mice, leading to hypoinsulinemia and diabetes (8,9). It has been reported that embryonic fibroblasts of Bmal1 knockout mice fail to differentiate into adipocytes. Also, Bmal1 overexpression in adipocytes up-regulates lipid synthesis activity (10). Moreover, loss of Bmal1 results in a shortened life span and several early aging phenotypes, including cataracts, sarcopenia, organ shrinkage, and less subcutaneous fat (11,12). Furthermore, Bmal1 deletion affects the degeneration of synaptic terminals and diminishes cortical connectivity, which promotes neurodegeneration (13).
Given the importance of Bmal1 in circadian timing and other physiological phenomena, the expression of Bmal1 needs to be precisely regulated. Previous studies have described diverse regulatory mechanisms of Bmal1 expression, including transcriptional and posttranslational regulation. For instance, Rev-erb␣ and Ror␣ suppress and enhance Bmal1 transcription, respectively (14 -17). Also, Bmal1 protein is known to be phosphorylated by casein kinase 1 ⑀ (CK1⑀) (18), mitogen-activated protein kinase (19), CK2␣ (20), and glycogen synthase kinase 3 ␤ (GSK3␤) (21) to mediate the timekeeping mechanism of the clock system. In addition, it has been reported that SUMOylation of Bmal1 regulates circadian rhythmicity (22). Moreover, ubiquitin protein ligase E3A (UBE3A) has been described to interact with and degrade Bmal1 protein (23). However, to date, translational regulation of Bmal1 mRNA remains poorly understood. Therefore, we sought to find the mechanism of Bmal1 protein synthesis. Emerging evidence reports that RNAbinding proteins play a central role in translational modulation (24). We have demonstrated previously that various RNAbinding proteins have an important function in translational control of several clock genes (25)(26)(27)(28)(29)(30). Therefore, we hypothesized that a specific RNA-binding protein may have a critical role in Bmal1 mRNA translation. In this study, we determine that an RNA-binding protein, hnRNP Q, has a critical role in translational control of Bmal1 mRNA. We also show that hnRNP Q binds to a specific region of the Bmal1 mRNA 5Ј UTR and negatively regulates Bmal1 mRNA translation. Moreover, we demonstrate that this translation is controlled by rhythmic interaction between hnRNP Q and Bmal1 mRNA. Furthermore, hnRNP Q-mediated translational repression of Bmal1 increases the mRNA oscillation amplitude of Bmal1 target genes, suggesting that this translational control is critical for the circadian clock system.

hnRNP Q negatively regulates Bmal1 translation
Maintenance of circadian rhythm depends on oscillation of biological clock gene expression. Posttranscriptional regulation of rhythmic genes is one of the important regulatory mechanisms. In this regulation, RNA-binding proteins have a critical role in governing the life of mRNAs. Although a recent report demonstrated that more than 1,500 proteins can interact with RNAs (31), Bmal1 mRNA-binding proteins are poorly understood. In addition, the molecular mechanisms of the posttranscriptional regulation of Bmal1 mRNA are still unknown. We therefore focused on the posttranscriptional control of Bmal1 mRNA. First, we performed in silico analysis of which RNAbinding proteins could interact with Bmal1 mRNA. Several hnRNPs were expected to bind to Bmal1 mRNA. To identify specific RNA-binding proteins that control Bmal1 expression, we performed RNAi screening by using three different siRNAs targeting hnRNP Q, hnRNP K, or hnRNP I (Fig. S1A). These hnRNPs are known to have a critical role in the posttranscriptional regulation of specific mRNAs (25,26,(32)(33)(34)(35)(36). Of the three siRNAs, only hnRNP Q knockdown caused an increase in the Bmal1 protein level (Fig. 1A). To determine whether this result comes from the up-regulated Bmal1 mRNA level or stability, we analyzed the Bmal1 mRNA level and decay kinetics after hnRNP Q depletion. The result showed that reduction of hnRNP Q does not significantly alter the Bmal1 mRNA level and stability (Fig. 1, B and C), suggesting that hnRNP Q may inhibit the translation of Bmal1 mRNA.
Because the 5Ј UTR of mRNA has a critical role in translational regulation (37), the Bmal1 5Ј UTR could be associated with protein synthesis inhibition mediated by hnRNP Q. By utilizing a luciferase reporter system, we sought to verify whether hnRNP Q directs translational repression of Bmal1 via its 5Ј UTR. The Bmal1 5Ј UTR sequences were inserted into the upstream of firefly luciferase (Fluc) coding sequence of the reporter vector. NIH-3T3 cells were transfected with the reporters after hnRNP Q depletion, and the luciferase activities were analyzed. Reduction of hnRNP Q significantly increased Bmal1 5Ј UTR-harboring reporter activity compared with the control reporter (Fig. 1D). To investigate whether the up-regulated reporter expression came from translational activation, we analyzed the mRNA level of reporters by using qPCR. The results showed unaltered reporter mRNA levels (Fig. S1B), suggesting that hnRNP Q mediates the Bmal1 5Ј UTR when it negatively regulates Bmal1 translation. Next, to further confirm the importance of hnRNP Q in Bmal1 protein synthesis, we performed a polysome profiling analysis with or without knockdown of hnRNP Q. The overall ribosome distributions in the sucrose gradient were not changed after hnRNP Q depletion (Fig. 1E). This observation was validated by results showing that GAPDH mRNA levels were not significantly altered in monosome or polysome fractions after hnRNP Q silencing (Fig. 1F). When hnRNP Q was reduced, however, the distribution of Bmal1 mRNA was shifted from the monosome to the polysome fraction, indicating up-regulation of Bmal1 mRNA translation (Fig. 1G). These results indicate that hnRNP Q mediates downregulation of the Bmal1 protein level by inhibiting Bmal1 mRNA translation through its 5Ј UTR.

hnRNP Q binds to a specific region of the Bmal1 mRNA 5 UTR
To determine whether hnRNP Q could bind to the Bmal1 5Ј UTR, we conducted an in vitro binding assay using the biotinconjugated Bmal1 5Ј UTR and cell lysates. Although the biotinylated Bmal1 5Ј UTR alone pulled down hnRNP Q protein, this binding affinity was dramatically reduced in combination with a competitor ( Fig. 2A), suggesting that it is a specific interaction between the Bmal1 5Ј UTR and hnRNP Q. Moreover, we further confirmed this interaction with GAPDH coding sequences (CDS). Biotin-conjugated GAPDH CDS could not promote hnRNP Q binding, and the interaction between hnRNP Q and the Bmal1 5Ј UTR was not inhibited after addition of the GAPDH CDS as a nonspecific competitor (Fig. S2A). Because most RNA-binding proteins interact with stem-loop elements of mRNA, we predicted possible secondary structures of Bmal1 5Ј UTR by in silico analysis. Assuming that hnRNP Q may bind to specific secondary structures of the Bmal1 5Ј UTR, we generated serially deleted Bmal1 5Ј UTR constructs that excluded each secondary structure to potentially abolish the binding of hnRNP Q (Fig. 2B). To discover binding sites of hnRNP Q in the Bmal1 5Ј UTR, we performed an in vitro binding assay with a biotin-incorporated, partially deleted Bmal1 5Ј UTR. The result demonstrated that hnRNP Q binds to fulllength and 182-nt-deleted (⌬182) constructs of the Bmal1 5Ј UTR. However, the 399-nt-truncated (⌬399) mutant showed significantly reduced affinity for hnRNP Q (Fig. 2C), suggesting that the binding sites of hnRNP Q reside in residues 1-399 of hnRNP Q regulates Bmal1 translation the Bmal1 5Ј UTR. We then asked whether only 1-182 residues of the Bmal1 5Ј UTR could facilitate hnRNP Q interaction. The result showed that the minimal region of 1-182 could promote this binding (Fig. S2B). Next we wondered whether this binding pattern affects the translational efficiency of Bmal1. To test this possibility, we performed a reporter assay with mutant constructs after reduction of hnRNP Q. Interestingly, the result showed an increase in Fluc level in control cells transfected with the ⌬399 construct. Moreover, hnRNP Q depletion significantly enhanced Fluc expression in the Bmal1 5Ј UTR full and hnRNP Q regulates Bmal1 translation ⌬182 constructs, but this up-regulation was not observed in the ⌬399 construct, which previously showed weak binding affinity for hnRNP Q (Fig. 2D). These results suggest that hnRNP Q interacts with residues 1-399 of the Bmal1 5Ј UTR to inhibit the translation process.

hnRNP Q plays a critical role in Bmal1 protein oscillation
Because Bmal1 plays a major role in the circadian system, we speculated that hnRNP Q-mediated translational control may lead to crucial consequences for circadian rhythmicity. To investigate the physiological significance of Bmal1 translational repression mediated by hnRNP Q, we reduced hnRNP Q and synchronized circadian time by dexamethasone treatment to analyze the Bmal1 oscillation profile. Silencing of hnRNP Q did not alter the Bmal1 mRNA oscillation profiles (Fig. 3A); however, interestingly, the amplitude of Bmal1 protein oscillation was significantly enhanced by hnRNP Q depletion (Fig. 3, B-D). This implies a critical role of hnRNP Q in Bmal1 protein syn-  A, interaction between hnRNP Q and the Bmal1 mRNA 5Ј UTR was determined by an in vitro binding assay. The biotin-conjugated Bmal1 5Ј UTR was incubated with cytoplasmic cell extracts and subjected to immunoblotting using hnRNP Q antibody. The biotin-unconjugated Bmal1 5Ј UTR was used as a competitor. B, schematic of serially deleted Bmal1 5Ј UTRs. C, confirmation of the interaction between hnRNP Q and the Bmal1 5Ј UTR. Biotin-conjugated full-length and serially deleted Bmal1 5Ј UTRs were incubated with cytoplasmic cell lysate. D, translation up-regulation mediated by reduction of hnRNP Q in full-length and the serially deleted Bmal1 5Ј UTR reporter. Knockdown of hnRNP Q was confirmed by immunoblotting. Data represent mean Ϯ S.D. (n ϭ 3; *, p Ͻ 0.05; **, p Ͻ 0.001). Rluc, Renilla luciferase.

hnRNP Q regulates Bmal1 translation
thesis. To further confirm this conclusion, we tested polysome profiles with or without hnRNP Q knockdown at certain circadian time points. The overall ribosome profiles were not signif-icantly altered by hnRNP Q knockdown at the indicated circadian time points (Fig. S3A). This was confirmed through control GAPDH mRNA (Fig. S3B). However, Bmal1 mRNAs hnRNP Q regulates Bmal1 translation were significantly shifted from monosome to polysome under hnRNP Q-depleted conditions. Also, as shown above, Bmal1 protein reaches peak level 12 h after synchronization. In accordance with this, Bmal1 mRNAs were highly associated with polysome fractions with or without hnRNP Q depletion 12 h after synchronization (Fig. S3C). To further examine whether hnRNP Q negatively regulates Bmal1 circadian expression, we monitored Bmal1 rhythmic oscillations in realtime mode using a reporter vector in which luciferase was fused with the Bmal1 promoter (Bmal1::Luc) (38). We inserted the Bmal1 5Ј UTR downstream of the promoter (Bmal1-5ЈUTR::Luc) to test the function of hnRNP Q in circadian rhythmicity of Bmal1. NIH-3T3 cells were transfected with reporters in the presence or absence of hnRNP Q siRNA and synchronized by dexamethasone treatment. We found that hnRNP Q silencing did not affect the oscillation pattern in Bmal1::Luc reporter-transfected cells (Fig. 3, E-G). On the other hand, the oscillation amplitudes were significantly enhanced in Bmal1 ϩ 5ЈUTR::Luc reporter-expressing cells under hnRNP Q reduction (Fig. 3, H-J). These results demonstrate that hnRNP Q is a decisive factor in the translation as well as circadian oscillation of Bmal1.
Given that the protein level of hnRNP Q is relatively constant after synchronization, we wondered how hnRNP Q demonstrates the regulation of oscillatory translation of Bmal1. We postulated that hnRNP Q might rhythmically interact with Bmal1 mRNA to exhibit translational control. To examine our hypothesis, we performed RNA-IP with an hnRNP Q antibody after synchronization. The results showed that the immunoprecipitated hnRNP Q level and total hnRNP Q protein level remained relatively constant during the experimental time period (Fig. 4A). Also, Bmal1 mRNA was pulled down by the hnRNP Q antibody compared with the IgG control (Fig. 4B). Interestingly, co-immunoprecipitated Bmal1 mRNA had its

hnRNP Q regulates Bmal1 translation
lowest level 12 h after synchronization (Fig. 4C), which was the peak time of Bmal1 protein expression (Fig. 3D). These results suggest that hnRNP Q interacts with Bmal1 mRNA in a timedependent manner to mediate translational repression.
It is well known that Bmal1 activates the transcription of several target genes, including Cry, Per, Rev-erb␣, and Ror␣. We asked whether hnRNP Q-induced Bmal1 translational inhibition affects the transcriptional activity of those target mRNAs. To test this possibility, we observed the mRNA oscillation profiles of Bmal1 target genes with or without hnRNP Q silencing. Remarkably, Cry1, Per1, Rev-erb␣, and Ror␣ mRNA oscillation amplitudes were enhanced under hnRNP Q reduction (Fig. 5, A-D). Together, these results suggest that hnRNP Q-mediated translational repression of Bmal1 affects the oscillation patterns of Bmal1 target genes.

Discussion
To date, most molecular studies of circadian rhythmicity have focused on the transcription and translation feedback loops of circadian genes to illustrate their rhythmic expression. Interestingly, however, a recent study has demonstrated that only 22% of cycling mRNAs are driven by de novo transcription (39), suggesting that posttranscriptional regulation has a major role in the circadian system. In line with this report, accumulating evidence shows that several clock genes are controlled by posttranscriptional regulation, such as mRNA splicing, decay, and translation (40,41). We demonstrated previously that RNA-binding proteins play major roles in these regulatory mechanisms. For instance, hnRNP R, hnRNP Q, and hnRNP L cooperatively destabilize arylalkylamine N-acetyltransferase mRNA (42). In addition, the internal ribosome entry site (IRES)-mediated translation of Rev-erb␣ mRNA is modulated by hnRNP Q and polypyrimidine tract-binding protein (32). In this study, we found a novel function of the RNA-binding protein hnRNP Q in translational control of the core clock gene Bmal1.
Although our results present several lines of evidence demonstrating hnRNP Q-mediated translational inhibition of the core clock gene Bmal1, we could not specify a clear molecular mechanism of how hnRNP Q suppresses Bmal1 protein synthesis. Unlike previous studies performed with rat pineal glands (25,28), hnRNP Q protein levels were relatively constant in the mouse fibroblast cell line. The binding between hnRNP Q and Bmal1 mRNA was rhythmic, and this interaction anti-correlated with the Bmal1 protein oscillatory pattern. The mechanisms underlying this rhythmic binding process need to be investigated further. For instance, posttranslational modifications of hnRNP Q, including phosphorylation, may influence the affinity of hnRNP Q for Bmal1 mRNA. Although it is not known which kinase can phosphorylate hnRNP Q, it will be interesting to see whether a specific kinase has circadian oscillatory activity. Because the secondary structures of mRNA are important in interaction with RNA-binding proteins, the stem-loop structure of Bmal1 mRNA may vary in a time-dependent manner that allows a different affinity for hnRNP Q. Also, it will be more interesting to see whether hnRNP Q down-regulates translation efficiency through competition with translation initiation factors. Moreover, scanning by preinitiation complex may be stalled in Bmal1 mRNA by strong binding between hnRNP Q and Bmal1 mRNA.

hnRNP Q regulates Bmal1 translation
hnRNP Q is known to be ubiquitously expressed in the heart, brain, lung, liver, and many other organs (43), and currently, hnRNP Q knockout mice are not available. Although generating mutant mice may be technically challenging, it will be valuable to test whether this hnRNP Q-mediated translational repression of Bmal1 could affect the circadian rhythm of the organism.
It is well known that Bmal1 is a main component of the circadian clock system and works as a transcription factor alongside the protein Clock. Interestingly, as expected, hnRNP Q knockdown enhanced the mRNA oscillation of Bmal1 target genes. Further investigation will be needed to find out whether this result comes from increased Bmal1 and Clock heterodimer transcription activity. In addition, we could not exclude the possibility that other RNA-binding proteins can interact with Bmal1 mRNA to regulate its protein synthesis. When and how Bmal1 expression is controlled by the RNA-binding protein repertoire remains to be determined. Moreover, tissue-specific hnRNP Q knockout mice will be instrumental to understand the function of hnRNP Q in the overall circadian system.
In conclusion, our results show a novel function of hnRNP Q in translational repression of Bmal1 that may be an important regulatory stage in circadian clock system. Given the importance of Bmal1, this study may help us to understand rhythmic translational regulation and give insights into the precisely controlled circadian clock system.

DNA constructs
To construct reporter plasmids, full-length and serially deleted mouse Bmal1 5Ј UTRs were amplified from full-length mouse Bmal1 complementary DNA (NM_007489.4). These PCR products were digested with HindIII and SmaI and inserted upstream of the Fluc coding sequence. The PCR products were also digested with EcoRI and XbaI and then subcloned into the pSK vector for in vitro binding assays. For the real-time bioluminescence luciferase reporter, the Bmal1::Luc plasmid (38) and Bmal1 5Ј UTR plasmid were used as a template. To construct the Bmal1-5ЈUTR::Luc plasmid, in the first round of PCR, the forward primer (5Ј-AAGAGCTCGC-AGAGTCCGCAACGCAGTGGCCTCAGCG-3Ј) and reverse primer (5Ј-AGGCCCGGCCCCCCGCCCACCAATCGCTGT-3Ј) of the Bmal1 promoter region and forward primer (5Ј-GTGGGCGGGGGGCCGGGCCTGGGCCGGCGG-3Ј) and reverse primer (5Ј-AACTCGAGTGGAAGGAAGGTGCTT-GCAAGGGACTCTAACT-3Ј) of the Bmal1 5Ј UTR were used to create two fragments. The two PCR products were mixed together for a second round of PCR. The final product of the Bmal1 promoter and 5Ј UTR sequences were digested with SacI and XhoI and inserted into the pGL3 basic plasmid.

Cell culture and drug treatment
NIH-3T3 cells were cultured in DMEM (Hyclone) containing 10% FBS (Hyclone), 100 units/ml penicillin, and 100 g of streptomycin (Welgene) and maintained in a humidified atmosphere of 5% CO 2 at 37°C. To synchronize the circadian oscillation, NIH-3T3 cells were treated with 100 nM dexamethasone. After 2 h, the medium was replaced with complete medium, and cells were harvested at the indicated time points for further experiments. To block transcription or translation, NIH-3T3 cells were treated with actinomycin D (5 mg/ml) or cycloheximide (100 g/ml), respectively.

Transfection and RNAi
Reporter plasmid expression and siRNA transfection for transient knockdown in cell lines were performed using Lipofectamine 2000 (Invitrogen) or the Neon transfection system (Invitrogen) according to the manufacturer's recommendations. Cells were harvested 24 h after transfection for subsequent experiments. siRNA targeting hnRNP Q was used as described previously (29).

Luciferase assay
NIH-3T3 cells were transfected with luciferase reporters and harvested 24 h post-transfection. Cells were lysed with reporter lysis buffer, and Renilla and firefly luciferase activities were analyzed using the Dual-Luciferase reporter assay system (Promega) according to the manufacturer's instructions. Firefly luciferase activities were normalized by Renilla luciferase activities.

Western blotting
NIH-3T3 cells were lysed with radioimmune precipitation assay buffer containing a protease inhibitor tablet (Thermo Scientific), and protein samples were resolved by SDS-PAGE, transferred to nitrocellulose membranes (Pall Corp.), and immunoblotted with specific antibodies. Blocking was performed with 5% nonfat dry milk in TBS and 0.1% Tween 20 (TBST) for 0.5-1 h. Membranes were incubated with primary antibodies in blocking solution overnight at 4°C. The primary antibodies used in this study were as follows: anti-Bmal1 (ab3350, Abcam), anti-hnRNP Q (ab10687, Abcam), anti-hn-RNP K (ab70492, Abcam), anti-hnRNP I (homemade), and anti-GAPDH (A300-641A, Bethyl). After several washes with TBST, the membranes were incubated with secondary HRPconjugated mouse (Thermo Scientific) or rabbit (Promega) secondary antibodies for 1-2 h and then visualized with SUPEX ECL reagent (Neuronex) and a LAS-4000 system (Fujifilm) according to the manufacturer's instructions. Acquired images were further analyzed with the Image Gauge program (Fujifilm).

Polysome profile analysis
NIH-3T3 cells were transfected with control siRNA or hnRNP Q-targeting siRNA, treated with cycloheximide (100 g/ml) for 30 min on ice, and harvested. Cell extracts were subjected to sucrose density gradient analysis as described previously (27). Briefly, cells were lysed in polysome lysis buffer (300 mM KCl, 5 mM MgCl 2 , 10 mM HEPES (pH 7.4), 0.5% NP-40, 5 mM DTT, and 100 g/ml cycloheximide). Cell lysates were loaded on 15-45% sucrose gradients in polysome buffer (300 mM KCl, 5 mM MgCl 2 , and 10 mM HEPES (pH7.4)) and centrifuged at 32,000 rpm in a SW41Ti rotor at 4°C for 3 h. The gradients were collected using a Brandel gradient density fractionator and analyzed by an Econo UV monitor (Bio-Rad). Total RNAs in each fraction were extracted using TRI-Solution hnRNP Q regulates Bmal1 translation (Bio Science Technology). RNA levels were quantified by quantitative PCR.

In vitro transcription, in vitro binding assay, and RNA immunoprecipitation
pSK vectors containing the Bmal1 5Ј UTR or GAPDH CDS were linearized with Xba1 and in vitro transcribed in the presence of biotin-UTP using T7 RNA polymerase (Promega) for the in vitro binding assay. Cytoplasmic lysates were prepared from NIH-3T3 cells using hypotonic lysis buffer containing 10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl 2 , 1 mM DTT, 0.2 mM PMSF, and 2.5% NP-40. The cytoplasmic fraction of NIH-3T3 cells was incubated with in vitro transcribed, biotin-labeled mRNA and subjected to streptavidin-agarose beads (Thermo Scientific). Bead-bound proteins were pulled down and analyzed by SDS-PAGE followed by immunoblotting. For RNA-IP, NIH-3T3 cells were lysed with RNA-IP buffer (10 mM HEPES (pH 7.5), 100 mM KCl, 5 mM MgCl 2 , 0.1% NP-40, protease inhibitor, and RNase inhibitor). Mouse normal IgG or hnRNP Q antibody was incubated with NIH-3T3 cell lysates at 4°C overnight and then incubated with protein G beads (Thermo Scientific) at 4°C for 4 h. The beads were washed three times with RNA-IP buffer, and RNA was isolated using TRI-Solution (Bio Science Technology). RNA levels were quantified by quantitative PCR.

RNA isolation and quantitative PCR
Total RNAs were extracted from NIH-3T3 cells using TRI-Solution (Bio Science Technology), following the manufacturer's instructions. Isolated RNAs were reverse-transcribed with the ImProm-II TM reverse transcription system (Promega) according to the manufacturer's recommendations. RNA levels were detected by quantitative PCR using the StepOnePlus realtime system (Applied Biosystems) with FastStart Universal SYBR Green Master (Roche) according to the manufacturer's instructions. Specific primer pairs used are shown in Table S1.

Real-time bioluminescence measurements
NIH-3T3 cells were plated in 35-mm dishes and allowed to grow to 80% confluence. The Bmal1::Luc (38) or Bmal1-5ЈUTR::Luc reporters were co-transfected with control siRNA or hnRNP Q-targeting siRNA to NIH-3T3 cells using Lipofectamine 2000 (Invitrogen) according to manufacturer's recommendations. Cells were synchronized with 100 nM dexamethasone for 2 h, and the medium was changed to 1 mM luciferin (Promega) containing phenol red-free DMEM (Hyclone). 35-mm dishes were then transferred to LumiCycle 32 (Actimetrics) kept in the 37°C incubator, and photons were measured for 6 days. LumiCycle analysis software (Actimetrics) was used for LumiCycle data analysis.

Statistical analysis
All quantitative data are presented as the mean Ϯ S.D. Statistical analyses were performed using Student's t test, one-way analysis of variance, or two-way analysis of variance with post hoc Tukey's multiple comparisons test. p values of less than 0.05 were considered statistically significant. GraphPad Prism 8 was used to perform statistical analyses and to generate figures.