Publication Cover
Chronobiology International
The Journal of Biological and Medical Rhythm Research
Volume 40, 2023 - Issue 8
352
Views
1
CrossRef citations to date
0
Altmetric
Research Article

Transcriptional regulation of the mouse EphA4, Ephrin-B2 and Ephrin-A3 genes by the circadian clock machinery

ORCID Icon, , , ORCID Icon, , , & ORCID Icon show all
Pages 983-1003 | Received 21 Mar 2023, Accepted 11 Jul 2023, Published online: 08 Aug 2023

ABSTRACT

Circadian rhythms originate from molecular feedback loops. In mammals, the transcription factors CLOCK and BMAL1 act on regulatory elements (i.e. E-boxes) to shape biological functions in a rhythmic manner. The EPHA4 receptor and its ligands Ephrins (EFN) are cell adhesion molecules regulating neurotransmission and neuronal morphology. Previous studies showed the presence of E-boxes in the genes of EphA4 and specific Ephrins, and that EphA4 knockout mice have an altered circadian rhythm of locomotor activity. We thus hypothesized that the core clock machinery regulates the gene expression of EphA4, EfnB2 and EfnA3. CLOCK and BMAL1 (or NPAS2 and BMAL2) were found to have transcriptional activity on distal and proximal regions of EphA4, EfnB2 and EfnA3 putative promoters. A constitutively active form of glycogen synthase kinase 3β (GSK3β; a negative regulator of CLOCK and BMAL1) blocked the transcriptional induction. Mutating the E-boxes of EphA4 distal promoter sequence reduced transcriptional induction. EPHA4 and EFNB2 protein levels did not show circadian variations in the mouse suprachiasmatic nucleus or prefrontal cortex. The findings uncover that core circadian transcription factors can regulate the gene expression of elements of the Eph/Ephrin system, which might contribute to circadian rhythmicity in biological processes in the brain or peripheral tissues.

Introduction

Organisms have developed endogenous circadian rhythms to adapt their biological functions to daily changes of the environment. Internal rhythmicity lasts approximately 24 h and is orchestrated by a molecular clock. In mammals, this clock comprises a transcriptional-translational feedback loop, in which the core proteins CLOCK and BMAL1 (circadian locomotor output cycles kaput 1 and brain and muscle ARNT [aryl hydrocarbon receptor nuclear translocator]-like protein 1) heterodimerize to activate the transcription of, among others, the clock genes Period (Per) and Cryptochrome (Cry) (Gekakis et al. Citation1998; Takahashi Citation2017). Once translated, PER and CRY translocate to the nucleus, and inhibit the activity of CLOCK and BMAL1, therefore repressing their own transcription (Dardente et al. Citation2007; Kume et al. Citation1999; Takahashi Citation2017). CLOCK and BMAL1 (or their homologs NPAS2 [neuronal PAS domain protein 2] and BMAL2) activate transcription by binding to regulatory DNA elements called E-boxes (CANNTG) (Gekakis et al. Citation1998; Kiyohara et al. Citation2008; Leclerc and Boockfor Citation2005; Maemura et al. Citation2000; Reick et al. Citation2001).

The CLOCK:BMAL1 heterodimer also activates the expression of other components of the clock such as Nr1d1/2 (coding for REV-ERBα/β), which generate an additional negative feedback by binding to retinoic acid-related orphan receptor response element (RORE) found in the Clock and Bmal1 genes (Crumbley and Burris Citation2011; Guillaumond et al. Citation2005; Liu et al. Citation2008; Preitner et al. Citation2002). In addition, rhythmic posttranslational modifications regulate the activity, transport, and degradation of core clock elements (Bellet and Sassone-Corsi Citation2010; Hirano et al. Citation2016). For example, glycogen synthase kinase 3β (GSK3β) phosphorylates BMAL1, CRY2, and REV-ERBα, and controls, notably, their degradation or nuclear location (Harada et al. Citation2005; Sahar et al. Citation2010; Yin et al. Citation2006). Importantly, core clock components also control the expression of a variety of “clock-controlled genes” in a rhythmic manner (via binding to E-boxes or RORE) to adapt physiological functions such as lipid/glucose metabolism or neuronal activity (Doi et al. Citation2010; Ikeda and Ikeda Citation2014; Pan et al. Citation2010).

Most (if not all) mammalian cells have a functional molecular clock (Mure et al. Citation2018; Yamazaki et al. Citation2000; Zhang et al. Citation2014). In the brain, the suprachiasmatic nuclei of the hypothalamus (SCN) act as a main circadian oscillator by synchronizing the internal time to the environmental day-night time via receiving direct excitatory input from retinal ganglion cells (Abrahamson and Moore Citation2001; Hastings et al. Citation2018; Yamazaki et al. Citation2000). SCN output signals are, among others, driven by clock-controlled genes contributing to 24-h changes in neuronal activity/firing (Hastings et al. Citation2018). Elsewhere in the brain, the molecular clock (and clock-controlled genes) also contributes to daily variations in behavior and neuroplasticity. Indeed, studies in mice have shown that the transcriptional control of tyrosine hydroxylase or monoamine oxidase by the molecular clock could underly time-dependent neuronal firing in the striatum and mood alterations (Chung et al. Citation2014; Hampp et al. Citation2008). Genes coding for cell/synaptic adhesion molecules are also candidates in bridging the molecular clock to rhythmic neuronal function given their E-box content and the roles in neurotransmission and neuroplasticity of their protein products (El Helou et al. Citation2013; Freyburger et al. Citation2016; Giroldi et al. Citation1997; Hannou et al. Citation2018; Li et al. Citation2016; Meighan et al. Citation2015). For instance, the Neuroligin-1 gene, which codes for a postsynaptic adhesion protein involved in glutamatergic signalling, was shown to be bound and transcribed by CLOCK and BMAL1 and to be expressed in a rhythmic manner in the mouse forebrain (El Helou et al. Citation2013; Hannou et al. Citation2018). Nevertheless, the potential for other cell/synaptic adhesion proteins to act as an output signal of the molecular circadian clock largely remains to be defined (Hannou et al. Citation2020).

Ephrins (Efns) and their Eph receptors represent a large family of cell adhesion molecules highly expressed in brain cells (e.g., neurons, glia) (Chen et al. Citation2012; Goldshmit et al. Citation2006; Murai and Pasquale Citation2011). The interaction between EPHA4 and its ligand EphrinA3 (EFNA3) regulates glutamate uptake (via astrocytic glutamate transporters), dendritic spine plasticity (via ACTIN remodelling), cell proliferation, and cortical development (Filosa et al. Citation2009; Murai et al. Citation2003; Steinecke et al. Citation2014; Tanasic et al. Citation2016; Zhu et al. Citation2021). Ephrin-B2 (EFNB2), another ligand of EPHA4, has roles in vascular and cortical development, and in the regulation of neuronal plasticity and N-methyl-D-aspartate (NMDA) receptors (Bouzioukh et al. Citation2007; Essmann et al. Citation2008; Ghori et al. Citation2017; Hu et al. Citation2014; Slack et al. Citation2008; Xing et al. Citation2019). We have previously reported the presence of E-boxes in the EphA4 gene, together with a decreased mRNA expression of EphA4, EfnA3 and EfnB2 in ClockΔ19 mice (Freyburger et al. Citation2016). Moreover, we found relatively high expression of EphA4 in the SCN of both mice and rats (Freyburger et al. Citation2016), and altered circadian phenotypes in EphA4 knockout (KO) mice (Kiessling et al. Citation2018). These phenotypes notably included a longer endogenous period of wheel-running activity under constant darkness and reduced phase-shift and number of c-FOS+ cells in the SCN after a delaying light pulse (Kiessling et al. Citation2018). Despite these observations suggesting a role for EphA4 in circadian clock functions and the likelihood of it representing a clock-controlled gene, little is known concerning its transcriptional regulation and that of its protein partners.

The aim of this research was to determine whether EphA4, EfnB2 and EfnA3 are regulated by the circadian clock machinery. Firstly, in vitro assays investigating direct transcriptional activation by CLOCK and BMAL1 (or their respective homologs NPAS2 and BMAL2) of gene sequences upstream of EphA4, EfnB2 and EfnA3 transcription start sites (TSS) were conducted. Secondly, the effect of E-box mutations in EphA4, and the impact of GSK3β were assessed using similar transcriptional assays. Thirdly, protein levels were measured at six different times of the day in the SCN and prefrontal cortex (PFC), and gene expression at two times of the day in multiple brain regions to verify daily changes in the targeted Eph/Ephrin. CLOCK:BMAL1 and/or NPAS2:BMAL1 were found to induce transcriptional activation via putative promoter regions of EphA4, EfnB2 and EfnA3, which was not linked to significant rhythms in protein level in the SCN or PFC. These findings provide support to a transcriptional regulation of elements of the Eph/Ephrin system by the circadian clock molecular machinery, and could suggest that this regulation primarily serves non-circadian roles.

Methods

Promoter analysis

Gene sequences for EphA4, EfnB2 and EfnA3 and upstream sequences were obtained from mm9 in the UCSC (University of California Santa Cruz) genome browser. Sequences were aligned and compared with the gene ID 13 838 (EphA4, chr. 1, 77343819–77491763, complement), gene ID 13 642 (EfnB2, chr. 8, 8667235–8711242, complement) and gene ID 106 644 (EfnA3, chr. 3, 89221200–89231359, complement) in NCBI (National Centre of Biotechnology Information), and with sequences ENSMUSG00000026235 (EphA4), ENSMUSG00000001300 (EfnB2) and ENSMUST00000028039 (EfnA3) in the Ensembl genome browser. The number and location of exons, introns, and TSS were extracted and compared with genomes mm10 and mm39. The A plasmid Editor (ApE) (Davis W.; accessed in Citation2017) was used to screen the putative promoter regions, identified from 3000 bp upstream of the TSS to the TSS, for cis-regulatory elements related to the molecular clock: canonical E-boxes (CACGTG), non-canonical E-boxes (CANNTG, CACGNG), RORE, cAMP-response element (CRE), glucocorticoid response element (GRE), etc. The identified regulatory elements are listed in .

Table 1. Regulatory elements found in the 3kb upstream of EphA4, EfnB2 and EfnA3 transcription start site.

The EphA4 gene (also known as 2900005C20Rik, Cek8, Hek8, rb, Sek, Sek1, Tyro1) contains 18 exons in rodents (19 in human), and one general TSS (beginning of exon 1). Another TSS after exon 11 has also been suggested (Zhao et al. Citation2017). The 3 kb upstream of the initial TSS in the mouse includes 13 non-canonical E-boxes (11 CANNTG and two CACGNG; ). For the mouse EfnB2 and EfnA3 genes, the upstream 3 kb shows 16 and 18 CANNTG sequences, respectively (). In addition, the EfnA3 upstream region contains an E-box-like element (E’-box: CACGTT), and both EfnB2 and EfnA3 contain one CACGNG sequence very close to the TSS (similar to the CACGNGs in EphA4). RORE was observed in EphA4 and EfnA3 putative promoters and a half PPRE (peroxisome proliferator-activated receptors [PPAR] response element) in EphA4 and EfnB2 upstream regions (). The regions were also screened for binding sites not related to the molecular clock, in particular sites linked to transcription factors that have been proposed to regulate EphA4, EfnA3 or EfnB2 transcription (such as EGR2 and Sp1 binding GC-boxes or FOXO response elements [FRE]) with fewer number of elements observed in comparison to E-boxes ( and ).

Figure 1. EphA4, EfnB2 and EfnA3 putative promoter regions contain circadian- and sleep-related regulatory elements. (a) Schematic representation of the 3 kb upstream of EphA4, EfnB2 and EfnA3 transcription start sites (TSS). Black boxes indicate exon 1 (also in b). Arrows indicate the TSS considered in different genome assemblies: grey in mm9, black in both mm10 and mm39, red in the NCBI tool. Colour bars indicate the regulatory elements found in these regions. ARE: antioxidant response element; CRE: cAMP response element; C/EBP: CCAAT/enhancer-binding protein site; EGR2: early growth response (Krox20) site; FRE: FOXO-recognized element; NF-κB: nuclear factor kappa light chain enhancer of activated B cells; PPRE: proliferator-activated receptor response element; RORE: retinoic acid-related orphan receptor response element. See Table 1 for representative sequences of the regulatory elements. (b) Alignment and comparison of E-boxes in the 3 kb upstream of the TSS for EphA4, EfnB2 and EfnA3 in the human, rat, and mouse. Arrows indicate the TSS. Salmon indicates CANNTG sequences, dark grey indicates CACGTG sequences (canonical E-boxes), light grey indicates CACGTT sequences, white indicates CACGNG, and light green bars indicate tata boxes. Horizontal black bars indicate the position of forward and reverse primers used for cloning of the mouse sequences.

Figure 1. EphA4, EfnB2 and EfnA3 putative promoter regions contain circadian- and sleep-related regulatory elements. (a) Schematic representation of the 3 kb upstream of EphA4, EfnB2 and EfnA3 transcription start sites (TSS). Black boxes indicate exon 1 (also in b). Arrows indicate the TSS considered in different genome assemblies: grey in mm9, black in both mm10 and mm39, red in the NCBI tool. Colour bars indicate the regulatory elements found in these regions. ARE: antioxidant response element; CRE: cAMP response element; C/EBP: CCAAT/enhancer-binding protein site; EGR2: early growth response (Krox20) site; FRE: FOXO-recognized element; NF-κB: nuclear factor kappa light chain enhancer of activated B cells; PPRE: proliferator-activated receptor response element; RORE: retinoic acid-related orphan receptor response element. See Table 1 for representative sequences of the regulatory elements. (b) Alignment and comparison of E-boxes in the 3 kb upstream of the TSS for EphA4, EfnB2 and EfnA3 in the human, rat, and mouse. Arrows indicate the TSS. Salmon indicates CANNTG sequences, dark grey indicates CACGTG sequences (canonical E-boxes), light grey indicates CACGTT sequences, white indicates CACGNG, and light green bars indicate tata boxes. Horizontal black bars indicate the position of forward and reverse primers used for cloning of the mouse sequences.

The number and position of E-boxes were also mapped for the putative promoter regions upstream of rat and human gene sequences for the three targets (ENST00000281821.7, ENST00000646441.1, ENST00000368408.4 for human; NM_001162411.1, NM_001107328, XM_039103763.1 for rat). This was done to identify potential regions of higher relevance for transcriptional assay in the mouse genome. E-box position and number in the 3 kb upstream of the TSS were relatively well conserved between species (). Interestingly, for EphA4, there was an apparent clustering of E-boxes around a more distal and a more proximal region of the putative promoter, which seemed particularly conserved in the mouse, rat and human genomes (). Accordingly, primers were designed to clone these two regions, identified as EphA4D and EphA4P, respectively.

Cloning

Five different regions of the putative promoter of EphA4, EfnB2 and EfnA3 were selected to generate reporter constructs: a 989 bp proximal sequence of EphA4 (EphA4P) at location −1603 to −615 bp from TSS; a distal 998 bp region at location −2981 to −1984 (EphA4D); a 1092 bp proximal sequence of EfnB2 (EfnB2P) at location −1392 to −301; a distal 1039 bp region at location −2973 to −1935 (EfnB2D); and a 1081 bp distal sequence of EfnA3 at location −2961 to −1881 (EfnA3D). Specific restriction enzymes were chosen to avoid cutting in the cloned sequence using NEBcutterv2.0 (New England Biolabs Inc.), and forward and reverse primers were designed for EphA4P, EphA4D, EfnB2P, EfnB2D and EfnA3D using the Oligo Analysis Tool of Eurofins Genomics. The primer sequences are provided in .

Table 2. Forward (fw) and reverse (rv) primers used for cloning with annealing temperature used for each pair.

DNA was purified from ear pieces of C57BL/6J mice with the DNeasy Blood & Tissue Kit according to manufacturer’s instructions (Qiagen, Germany). For PCR amplification, 25 µL of Master Mix (2.5 µL PCR Buffer 10X, 1 µL of dNTP 10 mM, 1.25 µL forward primer 20 µM, 1.25 µL reverse primer 20 µM, 0.25 µL Taq HotStart [5 U/µL] (Qiagen), 16.08 µL dH2O) were mixed with 50 ng of DNA, and amplification was done using the following program: 5 min at 95°C (hot start); 30 cycles of 30 s at 94°C, 30 s at 53–67°C (see ), and 1 min at 72°C; 4 min at 72°C. Two and a half µL of PCR products (with 0.5 µL of loading buffer) were run on an agarose gel 1% at 120 V during 40 min for size verification, and PCR products were purified with QIAquick PCR Purification Kit according to manufacturer’s protocol (Qiagen).

Purified amplicons and plasmid pGL3-basic (Promega, US) were digested with the restriction enzymes XhoI and HindIII (Thermo Fisher Scientific, US). Digestion mixes for plasmid (6 µg of plasmid DNA, 3 µL of each enzyme, 6 µL 10X Fast Digest Green Buffer and water up to 60 µL) were incubated for 10 min at 37°C. Digestion mixes for insert DNA (0.4 µg DNA, 1 µL of each enzyme, 4 µL 10X Fast Digest Green Buffer, and water up to 60 µL) were incubated for 20 min at 37°C. Then, 30 µL of digested samples were run on a 1% agarose gel at 120 V for 45 min, and DNA was purified using the QIAquick Gel Extraction Kit (Qiagen). Purified samples were ligated by mixing 50 ng of digested plasmid with digested insert (ratio plasmid:insert 1:1 or 1:4), 4 µL of ligase buffer 5X, 1 µL of T4 DNA ligase and up to 20 µL of dH2O, and incubated overnight (4°C). The final cloned plasmids are all 5.8 to 5.9 kb and were all verified using Sanger sequencing (Genome Quebec, Montreal, Canada). Due to PCR constraints, EphA4P plasmid contained two mutations in cytosines (C T at positions 198 and 837 of the insert) and EfnA3D contained a mutation in the fifth E-box (CAGTTG to CGGTTG), but the other constructs did not contain any nucleotide change in comparison to sequences in NCBI/Ensembl databases.

Design of EphA4D with mutated E-boxes

The gene sequence of EphA4D was then designed in silico with mutated E-boxes and ordered, already cloned in pGL3-basic, from Biomatik (Canada). From a literature review on E-box mutations impacting transcriptional activation and binding by CLOCK and BMAL1 (see ), four or five E-boxes (CANNTG) were mutated to GCTAGT. The same restriction sites used for EphA4D are flanking the mutated inserts (EphA4Dmut4 and EphA4Dmut5). The sequence of these commercial constructs was also verified using Sanger sequencing (Genome Quebec, Montreal, Canada).

Table 3. Mutated E-boxes used in previous research.

Cell culture and transfection

Cell culture, transfection and luciferase assay were conducted similar to those described previously (Dardente et al. Citation2007; Mongrain et al. Citation2008; Travnickova-Bendova et al. Citation2002). COS-7 cells were cultured in a humidified atmosphere at 37°C with 5% CO2 in COS-7 media (HyClone Dulbecco’s Modified Eagle Medium [DMEM]/High Modified; GE Healthcare Life Sciences, Thermo Fisher Scientific] with 10% fetal bovine serum [FBS; Life Technologies] and 1% glutamine [Life Technologies]). For luciferase assays (see below), cells were plated on 24-well plates at 105 cells/well with 0.5 mL of COS-7 media. After overnight incubation (37°C, 5% CO2; to reach 80–90% confluence), cells in each well were transfected with plasmid mixes containing 50 ng of reported constructs for selected targets (EphA4P, EphA4D, EfnB2P, EfnB2D, EfnA3D or EphA4Dmut) or 25 ng of positive control pGL3-mPer1 (a 1.8-kb promoter region of mPer1; Travnickova-Bendova et al. Citation2002), 25 ng of transfection normalizer pCR3-LacZ, 200 ng of pSG5-mCLOCK or pSG5-empty, 200 ng of pCS2-MTK-mBMAL1 or pCS2-MTK-empty, and completed to a total of 700 ng of plasmids using pBluescript (Stratagene). In some experiments, pSG5-NPAS2 and pCS2-MTK-BMAL2 were used to replace CLOCK and BMAL1 expressing vectors, respectively. CLOCK, NPAS2, BMAL1 and BMAL2 expressing vectors were similar to those previously described (Dardente et al. Citation2007; Hannou et al. Citation2018; Travnickova-Bendova et al. Citation2002). For transfection, each well was treated with 0.7 µL of Plus Reagent (Thermo Fisher Scientific) diluted in 50 µL of OPTI-MEM (Gibco, Life Technologies), and incubated 5 min at room temperature. Cells were then immediately transfected using 2 µL of Lipofectamine LTX (diluted in 50 µL OPTI-MEM) and incubated for 30 min at room temperature. After 5 h of incubation at 37°C, 0.5 mL of COS-7 media was added per well, and plates were incubated overnight (37°C, 5% CO2) before luciferase assay. Transfection conditions were always conducted in triplicates (i.e., 3 wells per condition per plate). Individual wells were considered as different replicates contributing to the sample size, given that they were independent cell cultures. Rare wells with substantial cell death or outlier values were discarded.

The implication of the circadian clock machinery in the transcriptional control of EphA4 and EfnB2 was further investigated using luciferase assays conducted with the addition of the negative regulator of the clock machinery GSK3β. Assays were performed with a wild-type form of GSK3β (GSK3β-WT) or with a constitutively active form to prevent inactivation by intracellular mechanisms (GSK3β-S9A, which inactivation via serine-9 phosphorylation is rendered impossible by a substitution to alanine; Beaulieu Citation2012; Stambolic and Woodgett Citation1994). Transfection conditions and plasmid mixes were similar as described above but included 50 ng of pcDNA3-HA-GSK3β-WT (Addgene, Cambridge, MA; Jim Woodgett, Mont Sinai Hospital, Toronto, ON) or pcDNA3-GSK3β-S9A (Hannou et al. Citation2018) or pcDNA3.1(+)-empty (#V790–20, Invitrogen).

Luciferase assays

Media was removed, and cells were rinsed with PBS 1X. One hundred fifty µL of lysis buffer (25 mM Tris, 2 mM EDTA, 1 mM dithiothreitol [DTT], 10% (v/v) glycerol, and 1% Triton X-100) was added to each well, and plates were incubated at room temperature in a Rocking Shaker (model 55, Reliable Scientific) at half its maximum speed for 10 min. Cell lysates were scratched and centrifuged at 13 000 rpm for 2 min to precipitate debris. For each well, 12 µL of supernatant was transferred into a white 96-well plate. An EnSpire Multimode plate reader (PerkinElmer) or a Synergy 4 plate reader (with dispenser module PN 7090568; Biotek) was used to inject 50 µL of luciferase buffer (20 mM Tris/Phosphate pH 7.8, 1 mM MgCl2, 2.7 mM MgSO4, 0.1 mM EDTA, 33.3 mM DTT, 530 μM ATP, 270 μM Co-enzyme A, 470 μM D-Luciferin) per well, immediately followed by reading of luminescence counts at 560 nm.

Luminescence counts were normalized to the total amount of protein and the transfection efficiency using, respectively, a DC (Lowry) protein assay (Bio-Rad) and a β-galactosidase assay. The DC protein assay was performed according to manufacturer’s instructions (Bio-Rad Laboratories, US), and absorbance reads were done at 750 nm using the EnSpire or Synergy 4 plate reader. For the β-galactosidase assay, 30 µl of the lysate supernatant was mixed with 750 µL of β-Mercaptoethanol in buffer Z (60 mM Na2HPO4.7 H2O, 40 mM NaH2PO4.H2O, 10 mM KCl, 1 mM MgSO4.7 H2O) and incubated 5 min at 37°C. Then, 150 µL of buffer Z containing 4 mg/mL o-nitrophenyl α-D-galactopyranoside (ONPG; Sigma Aldrich) was added to each condition and incubated at 37°C until a yellow coloration appeared. A volume of 375 µL of 1 M NaCO3 was added to stop the reaction, and absorbance was measured at 420 nm using the EnSpire or Synergy 4 plate reader. Luminescence (luciferase) counts normalized with DC protein and β-galactosidase assay absorbances were finally expressed as relative values over the respective negative control condition (i.e., empty plasmids).

Brain tissue punches and protein extraction

Males and females were studied for in vivo experiments given the reported sex differences in circadian rhythms (Dib et al. Citation2021), including in gene expression rhythms in the rodent brain (Chun et al. Citation2015; Kuljis et al. Citation2013, Citation2016). Thirty-six male and 36 female C57BL/6J mice were habituated to individual housing, a 12:12-h light:dark cycle, and food/water available ad libitum for 2 weeks. Then, to unmask endogenous circadian rhythmicity, animals were exposed to constant darkness for two complete days and on the third day of constant darkness, they were sacrificed under dim red light at six different times: CT0, CT4, CT8, CT12, CT16 and CT20 (six mice per sex per time-point; age between 10 and 16 weeks at sacrifice), with CT12 representing the beginning of the active period. Full brains were immediately frozen and kept at −80°C. This experiment was performed in accordance with guidelines of the Canadian Council on Animal Care and approved by the Comité d’éthique de l’expérimentation animale of the CIUSSS-NIM.

Protein levels were measured for the SCN and PFC. The SCN was targeted given its roles in the circadian system, and because EphA4 is expressed in this area (Freyburger et al. Citation2016), and the PFC given the reported rhythmic levels of BMAL1 and PER1 (Angeles-Castellanos et al. Citation2007; Coria-Lucero et al. Citation2016). PFC and SCN regions were sampled by tissue punches done with magnifying glasses following brain slicing using a cryostat (HM525 NX, Thermo Scientific, lame S35 – Feather®). A 20 G needle, cut to obtain a flat end with a diameter <0.9 µm, was kept in the cryostat (−12 to −13°C) and used to collect SCN and PFC punches. Brains were first sectioned in 500 μm coronal slices at −12 to −13°C (starting +2 mm (anterior) from the Bregma for PFC and −0.3 mm (posterior) from Bregma for SCN). For PFC, five punches from the same brain section were sampled per animal, while for the SCN, five punches each from a different mouse were pooled per time point. Each punch was immediately released in 40 µL of ice-cold modified RIPA buffer (50 mM HEPES, 10 mM EDTA, 0.1% SDS, 1% IGEPAL, 0.5% sodium deoxycholate, protease and phosphatase inhibitors [Sigma-Aldrich]). After the addition of five punches to the 40 µL cold RIPA buffer, tissues were mechanically homogenized on ice with a Pellet Pestle (Sigma Aldrich) until translucid (one 30-s and one 20-s interval). Samples were immediately centrifuged at 8 000 rpm for 40 min (4°C), and the supernatants were kept at −80°C for subsequent analysis.

Immunoblotting and protein quantification

Twenty μg of protein were loaded on 8% polyacrylamide gels and separated by SDS-PAGE using a 65 min migration at 100 V. Proteins were then transferred to a PVDF membrane (Bio-Rad) for 60 min at 100 V. Membranes were blocked with blocking buffer (5% dry milk diluted in Tris-buffered saline [TBS: 15 mM Tris-HCl, 5 mM Tris base, 150 mM NaCl]) for 1 h at room temperature, and then incubated overnight at 4°C with primary antibodies against EPHA4 (1:1000; Invitrogen #37–1600) and EFNB2 (1:1000; R&D Systems Inc. #AF496) diluted in 5% dry milk in TBS-T (TBS with 0.1% Tween 20). After TBS-T washes, membranes were incubated for 1.5 h at room temperature with secondary antibodies (1:15000 IRDye® 680RD goat anti-mouse IgG (H+L) #926–68070, IRDye® 800CW donkey anti-goat IgG (H + L) #926–32214; LI-COR) diluted in 5% dry milk TBS-T. Membranes were revealed using an Odyssey CLx imaging system (LI-COR). After image acquisition, membranes were stripped with 10% NaOH for 30 min, washed in TBS-T, and blocked again with 5% dry milk TBS-T for 1 h at room temperature. Membranes were then incubated overnight at 4°C with a second set of primary antibodies, namely targeting PER2 (1:1000; Novus Biologicals #100–125) together with ACTIN (1:8000; Sigma Aldrich #A5441), diluted in 5% dry milk TBS-T. After TBS-T washes, membranes were incubated for 1.5 h at room temperature with secondary antibodies (1:15000 IRDye® 680RD goat anti-mouse IgG (H+L) #926–68070, IRDye® 800CW goat anti-rabbit IgG (H+L), #926–32211; LI-COR) diluted in 5% dry milk TBS-T. Membranes were again revealed using an Odyssey CLx imaging system. Bands were quantified using ImageJ (NIH) (Schneider et al. Citation2012). Values were normalized to ACTIN, and to the mean CT0 level for each tissue.

Spatial gene expression quantification

The gene expression of EphA4, EfnB2 and EfnA3 was compared between the early rest period (Zeitgeber time 4: ZT4; with ZT0 representing light onset, and ZT12 light offset) and the early active period (ZT14) for different mouse brain regions using the Visium Spatial Gene Expression kit (10X Genomics). Coronal 10 μm brain slices (1.5 mm posterior to Bregma) were prepared from 12 to 13 weeks C57BL/6J mice injected with saline 3–4 h before sacrifice (control samples from Ballester Roig et al. Citation2023) and processed for spatial transcriptomics according to manufacturer’s instructions. Briefly, slices were fixed, stained with hematoxylin-eosin and permeabilized, following which cDNA libraries were prepared using the Visium Spatial Gene Expression kit. Libraries were sequenced using a NovaSeq6000 platform (Illumina) at Genome Québec. Sequencing reads in the FASTQ format were aligned to the mouse genome and compared between ZT4 and ZT14 using Space Ranger (pipeline ”‘spaceranger aggr’”) and Loupe Browser 5.0 (10X Genomics) separately for males and females. Differentially expressed genes were defined as having a false discovery rate (FDR) <0.05 (Benjamini and Hochberg Citation1995). The dataset included one male and one female brain sampled at ZT4 and one male and one female brain sampled at ZT14, and only the gene expression of EphA4, EfnB2 and EfnA3 is graphically represented here together with that of the circadian control gene Per1, Per2, and Dbp. Data are publicly available at GEO (accession numbers GSE218537 and GSE217058). This experiment was also performed in accordance with guidelines of the Canadian Council on Animal Care and approved by the Comité d’éthique de l’expérimentation animale of the CIUSSS-NIM.

Statistical analyses

Except for the spatial transcriptomic dataset, Prism 7 (GraphPad Software Inc., US) was used to conduct statistical analyses and prepare figures. One-way analyses of variance (ANOVA) were used for comparisons of luciferase data between conditions, and of protein levels between time points. Post hoc Tukey’s tests were used to decompose significant effects. Data are presented as means ± SEM, and the threshold for statistical significance was set to 0.05. Twenty-four h curve fit was conducted for protein levels with the cosine analysis from GraphPad, and the significance of the fit was calculated for period length between 21 and 27 h with CircWave Batch v3.3.

Results

Transcriptional activation of EphA4 by clock transcription factors

The transcriptional activation via proximal (EphA4P) and distal (EphA4D) regions of the putative promoter of EphA4 by CLOCK, BMAL1 and homologs NPAS2 and BMAL2 was first investigated. When transfected alone, none of the core clock transcription factors was driving transcriptional activation, which applies to both the proximal and distal regions (). Co-transfection of CLOCK and BMAL1 induced a significant transcriptional activation via EphA4P and EphA4D (EphA4P F8,96 = 11.7, p < 0.0001; EphA4D F8,107 = 31.4, p < 0.0001). More precisely, CLOCK and BMAL1 caused a 1.6-fold induction via EphA4P and a 3.7-fold induction via EphA4D. Moreover, NPAS2 and BMAL1 also resulted in a 1.3- and a 1.9-fold induction via EphA4P and EphA4D, respectively (F8,107 = 31.4, p < 0.0001). No significant induction was found for combinations with the BMAL1 homolog BMAL2.

Figure 2. Circadian clock transcription factors activate transcription via EphA4 putative promoter sequences. (a) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2 and BMAL2 via EphA4P (left) and EphA4D (right). Upper schematics represent the inserts cloned inside pGL3 vector and used for assays. The number of replicates (i.e., independent cultures) is between 3 and 24 per condition. (b) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2, GSK3β-S9A and GSK3β-WT via EphA4P (left) and EphA4D (right). N between 7 and 14 per condition. (c) Left schematic illustrates mutated E-boxes of EphA4D in grey in comparison to the original sequence in salmon. Right graph shows transcriptional activation by CLOCK and BMAL1 via EphA4D and two mutated EphA4D constructs (EphA4Dmut4 and EphA4Dmut5). N between 15 and 27 per condition. + indicate transfection of plasmids containing circadian clock transcription factor or luciferase reporters (absence of + indicates transfection with corresponding empty plasmids). Transcriptional activation is expressed relative to the negative control shown with white bars. *: p < 0.05, **: p < 0.01 and ***: p < 0.001 between indicated bars (post hoc comparisons).

Figure 2. Circadian clock transcription factors activate transcription via EphA4 putative promoter sequences. (a) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2 and BMAL2 via EphA4P (left) and EphA4D (right). Upper schematics represent the inserts cloned inside pGL3 vector and used for assays. The number of replicates (i.e., independent cultures) is between 3 and 24 per condition. (b) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2, GSK3β-S9A and GSK3β-WT via EphA4P (left) and EphA4D (right). N between 7 and 14 per condition. (c) Left schematic illustrates mutated E-boxes of EphA4D in grey in comparison to the original sequence in salmon. Right graph shows transcriptional activation by CLOCK and BMAL1 via EphA4D and two mutated EphA4D constructs (EphA4Dmut4 and EphA4Dmut5). N between 15 and 27 per condition. + indicate transfection of plasmids containing circadian clock transcription factor or luciferase reporters (absence of + indicates transfection with corresponding empty plasmids). Transcriptional activation is expressed relative to the negative control shown with white bars. *: p < 0.05, **: p < 0.01 and ***: p < 0.001 between indicated bars (post hoc comparisons).

The implication of the core circadian clock molecular loop in the control of EphA4 gene expression was further assessed with luciferase assays in the presence of GSK3β, a negative regulator of the CLOCK:BMAL1 heterodimer. These assays using an additional plasmid showed significant transcriptional activation by CLOCK and BMAL1 and by NPAS2 and BMAL1 for both EphA4P and EphA4D (F6,65 = 28.7, p < 0.0001 and F6,54 = 14.6, p < 0.0001, respectively; ). These activations were completely abolished by the constitutively active form of GSK3β (i.e., GSK3β-S9A), whereas GSK3β-WT was not impacting CLOCK:BMAL1- and NPAS2:BMAL1-driven transcriptional activation ().

To verify the implication of specific E-boxes of EphA4D in the CLOCK:BMAL1-driven transcriptional activation, luciferase assays were conducted with reporter constructs containing four or five mutated E-boxes in the EphA4D sequence (EphA4Dmut4 and EphA4Dmut5; ). CLOCK:BMAL1-mediated transcriptional activation via mutated EphA4D constructs showed an approximately 1.7-fold induction relative to control, which was significantly lower than the transcriptional activation via the original EphA4D (F5,118 = 21.1, p < 0.0001; ). In fact, in comparison to the WT EphA4D sequence (2.7-fold induction), transcriptional activation via EphA4Dmut showed a 37% reduction. Given the similar induction with EphA4Dmut4 and EphA4Dmut5, the findings suggest an implication of at least one of the first four mutated E-boxes in the transcriptional activation of EphA4 by core clock transcription factors.

Transcriptional activation of EfnB2 and EfnA3 by clock transcription factors

To assess whether the circadian clock molecular machinery also activates the gene expression of EPHA4 ligands, luciferase assays were conducted using proximal and distal sequences of EfnB2 (EfnB2P, EfnB2D) and a distal sequence of EfnA3 (EfnA3D). Significant inductions were found for all three sequences (EfnB2P F4,28 = 49.7, p < 0.0001; EfnB2D F4,56 = 46.3, p < 0.01; EfnA3D F4,53 = 5.2, p < 0.005; ). More precisely, simultaneous transfection of CLOCK and BMAL1 resulted in a 4.3-fold induction via EfnB2P, while NPAS2:BMAL1 and NPAS2:BMAL2 co-transfections led to 4.8-fold and 2.6-fold inductions, respectively, via the same sequence. Transcriptional activations via EfnB2D were more modest, although significant, reaching 1.4-fold for CLOCK:BMAL1 and 1.8-fold for NPAS2:BMAL1. Finally, transcriptional activation via EfnA3D was only induced by NPAS2:BMAL1 (2.8-fold; ).

Figure 3. Circadian clock transcription factors activate transcription via EfnB2 and EfnA3 putative promoter sequences. (a) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2 and BMAL2 via EfnB2P (left), EfnB2D (middle) and EfnA3D (right). Upper schematics represent the inserts cloned in pGL3 vector and used for assays. The number of replicates (i.e., independent cultures) is between 6 and 14 per condition. (b) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2, GSK3β-S9A and GSK3β-WT via EfnB2P (left) and EfnB2D (right). N between 2 and 9 per condition. + indicates transfection of plasmids containing circadian clock transcription factors or luciferase reporter (absence of + indicates transfection with corresponding empty plasmids). Transcriptional activation is expressed relative to the negative control (white bars). *: p < 0.05 and **: p < 0.01 between indicated bars (post hoc comparisons).

Figure 3. Circadian clock transcription factors activate transcription via EfnB2 and EfnA3 putative promoter sequences. (a) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2 and BMAL2 via EfnB2P (left), EfnB2D (middle) and EfnA3D (right). Upper schematics represent the inserts cloned in pGL3 vector and used for assays. The number of replicates (i.e., independent cultures) is between 6 and 14 per condition. (b) Transcriptional activation by different combinations of CLOCK, BMAL1, NPAS2, GSK3β-S9A and GSK3β-WT via EfnB2P (left) and EfnB2D (right). N between 2 and 9 per condition. + indicates transfection of plasmids containing circadian clock transcription factors or luciferase reporter (absence of + indicates transfection with corresponding empty plasmids). Transcriptional activation is expressed relative to the negative control (white bars). *: p < 0.05 and **: p < 0.01 between indicated bars (post hoc comparisons).

Assays with EfnB2 promoter sequences were also conducted in the presence of GSK3β-S9A and GSK3β-WT. The constitutively active GSK3β-S9A (but not GSK3β-WT) blocked the effect of CLOCK:BMAL1 and NPAS2:BMAL1 on EfnB2P (F6,24 = 23.0, p < 0.01; ). Intriguingly, GSK3β-WT induced a 7.9-fold transcriptional activation via EfnB2P when co-transfected with CLOCK:BMAL1 (significantly more than co-transfecting CLOCK:BMAL1 without GSK3β), and the transcriptional activation of NPAS2:BMAL1 via EfnB2P was also potentiated by GSK3β-WT (9.3-fold with GSK3β-WT compared to 4.7-fold without). Concerning EfnB2D, the transcriptional activation by CLOCK:BMAL1 and NPAS2:BMAL1 were both abolished by the constitutively active GSK3β-S9A (F6,38 = 16.3, p < 0.01), but not by GSK3β-WT, which is reminiscent of observations made for EphA4 promoter sequences. In sum, these results support that core clock transcription factors can activate the transcription via putative promoter sequences of EfnB2 and EfnA3, which is modulated by GSK3β.

Absence of EPHA4 and EFNB2 rhythm in the PFC and SCN

To assess whether the transcriptional regulation of EphA4 an EfnB2 by core clock transcription factors results in a circadian rhythm of their respective protein product, EPHA4 and EFNB2 protein levels were measured at six different circadian times in the SCN and PFC in male and female mice. For both males and females, no significant circadian oscillation of EPHA4 or EFNB2 was found in the PFC (R2 < 0.13, p > 0.24; ). Similar observations were made for EPHA4, EFNB2 as well as the control protein PER2 when pooling PFC punches of five different animals per time (). In the SCN, circadian time did not appear to change the expression of EPHA4 for both males and females (no statistical comparison since n = 1 per time point; bottom and 4c). A potential circadian variation in PER2 was observed in the female SCN (no statistical comparison), but not for males. The level of EFNB2 in the SCN was too low to allow a reliable quantification. These results suggest that, although core clock transcription factors act on EphA4 and EfnB2 putative promoter regions, EPHA4 and EFNB2 proteins are not showing robust variations at a circadian scale in the PFC and SCN when mice are kept in constant darkness.

Figure 4. EPHA4 and EFNB2 protein levels in the mouse PFC and SCN do not show circadian oscillations. (a) Representative blots of PFC samples of individual male and female mice (top), of PFC samples of pooled female mice (middle), and of SCN samples of pooled female mice (bottom). (b) Quantifications of EPHA4 and EFNB2 in PFC punches from individual mice (dots of the same colour are from the same blot). Cosine waves fitted to the data are also shown (same in panel c); indicated p-values were calculated by fitting data to a 24-h curve. (c) Quantification of PER2, EPHA4 and EFNB2 for pooled PFC punches and pooled SCN punches (each point represents a pool of 3 to 5 mice for PFC and of 5 mice for SCN).

Figure 4. EPHA4 and EFNB2 protein levels in the mouse PFC and SCN do not show circadian oscillations. (a) Representative blots of PFC samples of individual male and female mice (top), of PFC samples of pooled female mice (middle), and of SCN samples of pooled female mice (bottom). (b) Quantifications of EPHA4 and EFNB2 in PFC punches from individual mice (dots of the same colour are from the same blot). Cosine waves fitted to the data are also shown (same in panel c); indicated p-values were calculated by fitting data to a 24-h curve. (c) Quantification of PER2, EPHA4 and EFNB2 for pooled PFC punches and pooled SCN punches (each point represents a pool of 3 to 5 mice for PFC and of 5 mice for SCN).

EphA4 and Efn expression pattern in different brain regions

We previously reported that diurnal variations in the gene expression of components of the Ephrin/Eph system are dependent on the brain region (Freyburger et al. Citation2016). We here investigated whether the mRNA expression of EphA4, EfnB2 and EfnA3 differed between the early light (ZT4) and the early dark (ZT14) period using a spatial transcriptomic strategy applied on a mouse coronal slice comprising the cerebral cortex, hippocampus, thalamus, hypothalamus, amygdala, and striatum (). The expression of EphA4 and EfnA3 was found to be particularly high in the pyramidal layer of hippocampal CA1-CA3 regions and granular layer of the dentate gyrus. EphA4 expression was also generally elevated in the thalamic region, and that of EfnA3 in cortical regions. In contrast, the expression of EfnB2 was lower, especially in the hippocampus, thalamus and hypothalamus. Globally, these expression patterns matched with those reported by the Allen Brain Atlas and previous literature (Allen Brain Atlas; Liebl et al. Citation2003). EphA4, EfnB2 and EfnA3 were not comprised in the list of differentially expressed genes (DEGs, FDR < 0.05) between ZT4 and ZT14 when considering the complete brain slice or the specific regions of interest targeted (), although there could be a trend for a higher EphA4 expression at ZT14 in layers I–IV of the auditory and entorhinal cortices. For comparison, the known cycling transcripts Per1, Per2 and Dbp are shown in (see also full DEG list in Table S1).

Figure 5. EphA4, EfnB2 and EfnA3 gene expression in the mouse brain at ZT4 and ZT14. (a) Spatial pattern of gene expression in the mouse brain for a selected coronal slice sampled at ZT4 (left) or ZT14 (right). Upper panels are from female mice, and lower panels from male mice (also in c). Cold to hot color scale indicates minimum to maximum gene expression, respectively (also in c). (b) Violin plots of EphA4, EfnB2 and EfnA3 gene expression in groups of spatial spots representing specific regions of interest: layers I-IV or layers V-VI of auditory and entorhinal cortices (A-EC), piriform cortex and amygdala (PC-Amy), hippocampus (Hipp), pyramidal and granular layers of the hippocampus (Hipp P/G), thalamus and hypothalamus. Orange denotes expression at ZT4, and grey at ZT14. (c) Spatial pattern of Per1, Per2 and Dbp gene expression in the mouse brain for the coronal slices sampled at ZT4 or ZT14.

Figure 5. EphA4, EfnB2 and EfnA3 gene expression in the mouse brain at ZT4 and ZT14. (a) Spatial pattern of gene expression in the mouse brain for a selected coronal slice sampled at ZT4 (left) or ZT14 (right). Upper panels are from female mice, and lower panels from male mice (also in c). Cold to hot color scale indicates minimum to maximum gene expression, respectively (also in c). (b) Violin plots of EphA4, EfnB2 and EfnA3 gene expression in groups of spatial spots representing specific regions of interest: layers I-IV or layers V-VI of auditory and entorhinal cortices (A-EC), piriform cortex and amygdala (PC-Amy), hippocampus (Hipp), pyramidal and granular layers of the hippocampus (Hipp P/G), thalamus and hypothalamus. Orange denotes expression at ZT4, and grey at ZT14. (c) Spatial pattern of Per1, Per2 and Dbp gene expression in the mouse brain for the coronal slices sampled at ZT4 or ZT14.

Discussion

The present study provides support to a role for the circadian clock molecular machinery in the regulation of the gene expression of elements of the Eph/Ephrin system. More precisely, CLOCK and BMAL1 or NPAS2 and BMAL1 were found to induce transcription via sequences upstream of the TSS of EphA4, EfnB2 and EfnA3 using in vitro assays. This transcriptional activation was found to be repressed by the circadian clock regulator GSK3β, when a constitutively active form of this kinase was used. For EphA4 in particular, mutating E-boxes in the promoter sequence reduced CLOCK:BMAL1-driven transcriptional activation. Although robust rhythms in protein level or gene expression were not detected in different brain regions, current in vitro findings combined to our previous observations of modified Eph/Ephrin gene expression in Clock mutant mice and altered circadian phenotypes in EphA4 KO mice (Freyburger et al. Citation2016; Kiessling et al. Citation2018) are supporting a relationship between the molecular circadian clock and components of the Eph/Ephrin system.

Our findings emphasize the importance of non-canonical E-boxes in the transcriptional control by core clock transcription factors. Indeed, CLOCK:BMAL1- and NPAS2:BMAL1-induced transcriptional activation occurred via sequences of the EphA4 and EfnB2 putative promoter with the only presence of CANNTG non-canonical E-boxes (i.e., without the presence of the canonical sequence CACGTG). CANNTG sequences are predicted to randomly appear every 256 bp in the genome but have nevertheless been reported to be important for transcriptional regulation in previous studies (Hannou et al. Citation2018; Kiyohara et al. Citation2008; Leclerc and Boockfor Citation2005). For instance, CATGTG could robustly activate the transcription of the Dbp gene (Kiyohara et al. Citation2008), whereas CATTTG appeared a key sequence in the transcriptional control of Prolactin by CLOCK and BMAL1 (Leclerc and Boockfor Citation2005). One CATGTG was present amongst the six E-boxes of EphA4P, and four CATTTG amongst the nine E-boxes of EfnB2D. However, some of the E-boxes included in our targeted gene sequences (e.g., CAGATG) were reported not to be bound by core clock transcription factors in another study (Oishi et al. Citation2005). It is also important to underline that the functionality of E-boxes can depend on the flanking DNA or other regulatory elements, as already evoked previously (Nakahata et al. Citation2008). Aiming at considering the cooperative nature of E-boxes, four or five were mutated in the distal region of EphA4, which led to a significant decrease in transcriptional activation by CLOCK and BMAL1. This result supports a role for at least one of these elements or their overall combination in the observed effect. Mutating each E-box separately will be required to pinpoint the exact DNA sequence(s) responsible for the transcriptional activation of Eph/Ephrin genes by clock transcription factors.

GSK3β was able to shape CLOCK:BMAL1- and NPAS2:BMAL1-mediated transcriptional activation of EphA4 and EfnB2. Given the roles of GSK3β in the regulation of the molecular clock (Besing et al. Citation2015; Sahar et al. Citation2010), this further supports an implication of core clock mechanisms in the gene regulation of Eph and Ephrin. Interestingly, EphA4 KO mice and GSK3β haploinsufficient mice have both been shown to express a longer endogenous period of wheel-running activity under constant darkness conditions (Kiessling et al. Citation2018; Lavoie et al. Citation2013). Thus, the downregulation of GSK3β and EphA4 could impact the circadian system similarly, at least at the level of output pathways affecting the locomotor activity rhythm. Besides roles in the circadian system, GSK3β is notably regulated by neurotransmission and insulin pathways (Beurel et al. Citation2015; Patel and Woodgett Citation2017) and has been associated with neurodegeneration and neurological conditions (Inoki et al. Citation2006; Liu and Klein Citation2018; Patel and Woodgett Citation2017). Our findings therefore imply that the regulation of Eph/Ephrin gene expression by GSK3β could serve cellular responses in “non-circadian” contexts.

A limitation of the present dataset could reside in luciferase assays conducted only using COS-7 cells. Previous research from our group has supported the suitability of this system for transcriptional studies related to the circadian clock (Hannou et al. Citation2018; Mongrain et al. Citation2008), but the cloned sequences do not recapitulate the complexity of in vivo transcriptional regulation notably because plasmids lack surrounding DNA as well as long distance regulatory elements, which both contribute to DNA folding and can impact the binding by the circadian clock molecular machinery. In fact, gene sequence positioned 7 to 11 kb upstream of EphA4 TSS was shown to have roles in its transcriptional regulation (Nakajima et al. Citation2006; Theil et al. Citation1998). Relatedly, a second limitation concerns the use of relatively short DNA fragments in the current study (~1 kb), and the lack of coverage of the most proximal promoter given the exclusion of the initial part of the 5’ untranslated region (UTR). In parallel, the promoter analysis highlighted potential sites of interaction with other transcription factors that have not been interrogated with the used experimental design. Only few studies have investigated the transcriptional regulation of Eph and Ephrin genes. On the one hand, the transcription factors with roles in development EGR2 (Krox20), MESP2, and PAX/FOXO1a (PAX/FKHD) were shown to bind EphA4 promoter regions and activate transcription (Begum et al. Citation2005; Nakajima et al. Citation2006; Theil et al. Citation1998), while stimulating protein 1 (SP1) binds EphA4 promoter to reduce mRNA and protein expression in the context of cell proliferation regulation (Huang et al. Citation2016). On the other hand, the EfnB2 promoter was shown to be bound by Meis homeobox 1 (MEIS1), Myc-associated zinc finger protein (MAZ) and nuclear factor-Y (NF-Y), and SP1 was found to activate EfnB2 gene expression (Obi et al. Citation2009; Sohl et al. Citation2009, Citation2010). The cooperative regulation of Eph/Ephrin gene expression by different transcription factors should be considered in the future research, even when focusing on the circadian clock machinery.

The level of EPHA4 and EFNB2 measured in the mouse SCN and PFC did not show 24 h variations under constant darkness. For the SCN, rhythmic levels of these targets and of PER2 in males could have been masked by sampling a large rostro-caudal region, because substantial differences in the peak time of PER2 are reported between rostral and caudal SCN (Evans et al. Citation2015). It is possible that measurements performed under light–dark conditions could have resulted in rhythmic levels given the presence of CRE in gene sequences upstream of TSS, and the role of CRE in the entrainment of clock gene expression to light (Ikegami et al. Citation2020; Tischkau et al. Citation2003). Additionally, cell type-specific rhythms could be missed in the present study sampling total proteins in the tissue. A study interrogating cell-type-specific gene expression in a microdissection of the mouse SCN region reported a significant circadian rhythm of EphA4 expression in astrocytes and neuronal populations in the immediate surroundings of the SCN (Wen et al. Citation2020). Astrocytes have crucial roles in the maintenance of circadian rhythms via crosstalk with SCN neurons (Barca-Mayo et al. Citation2017; Brancaccio et al. Citation2017), and the Eph/Ephrin system is well recognized for its involvement in astrocyte-neuron communication (Murai and Pasquale Citation2011). Moreover, SCN cholecystokinin/complement C1q-like 3 (Cck+/C1ql3+) neurons were shown to rhythmically express EfnB3, and EfnB1 and EfnA2 were rhythmically expressed in neurons outside/surrounding the SCN (Wen et al. Citation2020), which provides further support to an implication of the Eph/Ephrin system in circadian physiology.

Our spatial transcriptomic approach did not identify EphA4, EfnB2 and EfnA3 as genes significantly changed between the early light and early dark period in areas covered by the targeted coronal slice. This finding is in line with our previous observations of rhythmic expression being significant only for EfnA3 when considering the cerebral cortex, hippocampus and a region covering the thalamus and hypothalamus (Freyburger et al. Citation2016). Moreover, other transcriptomic studies targeting the mouse hippocampus or forebrain did not detect diurnal rhythms in the expression of EphA4, EfnA3 and EfnB2 (Debski et al. Citation2020; Noya et al. Citation2019), which also applies to the post-mortem human dorsolateral PFC (Seney et al. Citation2019). This likely implies a lack of robust circadian oscillation for the expression of these genes in multiple brain regions. Nevertheless, transcriptional regulation by clock transcription factors may drive rhythmic gene expression of these Eph/Ephrins in other brain regions or peripheral tissues, such as the cerebellum or cardiovascular tissues that show high levels of EphA4 protein and mRNA (Goldshmit and Bourne Citation2010; Li et al. Citation2021; Liebl et al. Citation2003; Martone et al. Citation1997). In fact, rhythmic EphA4 expression was reported in baboon muscle and cornea and in human tibial artery and heart atrial tissue and of EfnB2 in baboon muscle and adipose tissue (Mure et al. Citation2018; Ruben et al. Citation2018). Tissue-specific (and cell type-specific) rhythmic expression of elements of the Eph/Ephrin system is reminiscent of reports that 6–10% of mRNA are expressed rhythmically in a given tissue, whereas 80% are showing rhythms of expression in at least one tissue (Maret et al. Citation2007; Menet et al. Citation2012; Mure et al. Citation2018; Panda et al. Citation2002; Storch et al. Citation2002; Zhang et al. Citation2014). In sum, the transcriptional regulation of EphA4 and EfnB2 by core clock transcription factors does not appear to translate into strong gene expression rhythms for most brain regions, and could thus mainly fulfill non-circadian biological functions.

In conclusion, we reported that putative promoter regions of EphA4, EfnB2 and EfnA3 can be activated by core transcription factors from the circadian system. The Eph/Ephrin system has important roles in cell–cell communication and cytoskeleton remodelling, which have implicated EphA4 (as well as EfnA3 and EfnB2) in multiple diseases/pathological conditions, including cancer, injury/stroke and Alzheimer’s disease (Fu et al. Citation2014; Chen et al. Citation2012, Citation2022; Goldshmit and Bourne Citation2010; Huang et al. Citation2016; Lemmens et al. Citation2013). Therefore, understanding the transcriptional regulation of these transmembrane molecules by the circadian clock molecular machinery should help reveal the contribution of the circadian system to cell adhesion, intracellular signalling and plasticity, and could also contribute to understand mechanisms underlying diseases.

Author contribution

M.N.B.R., E.B.-N., and V.M. designed the experiments. M.N.B.R., L.H., and E.B.-N. performed cloning. M.N.B.R., P.-G.R., T.-A.S.G., L.H., B.D.-L., and E.B.-N. conducted experiments in COS-7 cells. M.N.B.R. conducted experiments related to protein measurement. M.N.B.R. and J.D.-G. conducted experiments related to spatial gene expression quantification. M.N.B.R. analysed the data, produced figures, and wrote the first draft of the manuscript. M.N.B.R. and V.M. revised and edited the manuscript. M.N.B.R. and V.M. contributed to funding acquisition. V.M. supervised the project.

Supplemental material

Supplemental Material

Download MS Excel (55.8 KB)Supplemental Material

Acknowledgments

The authors are thankful to Romina López Urbina for comments on the manuscript, and to Clément Bourguignon for help regarding cosine analyses.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

RNA-sequencing data have been deposited in the GEO database (https://www.ncbi.nlm.nih.gov/geo) and are available under accession numbers GSE217058 and GSE218537.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/07420528.2023.2237580.

Additional information

Funding

This work was supported by a Vanier Canada Graduate Scholarship (M.N.B.R.), a J.A. De Sève fellowship from the Recherche CIUSSS-NIM (M.N.B.R.), a recruitment fellowship from the Université de Montréal (M.N.B.R.), Discovery grants from the Natural Sciences and Engineering Research Council of Canada (V.M.), and the Canada Research Chair in Sleep Molecular Physiology (V.M.).

References

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.