BTBD9 and dopaminergic dysfunction in the pathogenesis of Restless Legs Syndrome

Abstract

Restless Legs Syndrome (RLS) is characterized by an urge to move the legs, usually accompanied by uncomfortable sensations. RLS symptoms generally happen at night and can be relieved by movements. Genetic studies have linked polymorphisms in BTBD9 to a higher risk of RLS. Knockout of BTBD9 homolog in mice (Btbd9) and fly results in RLS-like phenotypes. A dysfunctional dopaminergic system is associated with RLS. However, the function of BTBD9 in the dopaminergic system and RLS is not clear. Here, we made use of the simple Caenorhabditis elegans nervous system. Loss of hpo-9, the worm homolog of BTBD9, resulted in hyperactive egg-laying behavior. Analysis of genetic interactions between hpo-9 and genes for dopamine receptors (dop-1, dop-3) indicated that hpo-9 and dop-1 worked similarly. Reporter assays of dop-1 and dop-3 revealed that hpo-9 knockout led to a significant increase of DOP-3 expression. This appears to be evolutionarily conserved in mice with an increased D₂ receptor (D₂R) mRNA in the striatum of the Btbd9 knockout mice. Furthermore, the striatal D₂R protein was significantly decreased and the dynamin I was increased. Overall, activities of DA neurons in the substantia nigra were not altered, but the peripheral D₁R pathway was potentiated in the Btbd9 knockout mice. Finally, we generated and characterized dopamine neuron-specific Btbd9 knockout mice and detected active-phase sleepiness, suggesting that dopamine neuron-specific loss of Btbd9 is sufficient to disturb the sleep. Our results suggest that increased activities in the D₁R pathway, decreased activities in the D₂R pathway, or both may contribute to RLS.

Introduction

Restless legs syndrome (RLS) is a common neurological disorder characterized by a strong urge to move the legs, with or without unpleasant sensations in lower limbs, which can generally be relieved by movements (Garcia Borreguero et al., 2017). As these symptoms predominately occur in the evening or at night, the disease often leads to a disruption of sleep and a poor quality of life (Abetz et al., 2004; Trenkwalder et al., 2005). Prescription medications for RLS patients include D₂/D₃ dopaminergic agonists (Garcia-Borreguero et al., 2016; Ferre et al., 2017; Trenkwalder et al., 2018).

Several studies have suggested the involvement of the dopaminergic system in the RLS. A higher concentration of BH₄ (Earley et al., 2001, 2006), which is a rate-limiting cofactor for tyrosine hydroxylase (TH) (Meiser et al., 2013), and an increased level of 3-ortho-methyldopa (Earley et al., 2006; Allen et al., 2009), a metabolite of L-DOPA (Allen et al., 2009), have been found in the cerebrospinal fluid of RLS patients. Autopsy studies show decreased (D₂ dopamine receptor) D₂R expression in the putamen and increased phosphorylated TH in both putamen and substantia nigra (SN) (Connor et al., 2009). Brain imaging studies show decreased dopamine transporter (DAT) (Earley et al., 2011) and D₂R binding potential (Michaud et al., 2002; Earley et al., 2013), and a strong correlation between the loss of D₂R and the severity of RLS (Connor et al., 2009).

Dopamine (DA) binds to the postsynaptic DA receptors or presynaptic DA autoreceptors after released into the synaptic cleft (Meiser et al., 2013; Koblinger et al., 2014). There are two classes of DA receptors, D1-like and D2-like. The D₁ DA receptor (D₁R) is involved in the direct pathway, while D₂R participates in the indirect pathway (Keeler et al., 2014). The classic model indicates that direct pathway enables movement and is pronociceptive, while the indirect pathway inhibits movement and is antinociceptive (Ben-Sreti et al., 1983; Rooney and Sewell, 1989; Zarrindast and Moghaddampour, 1989; Verma and Kulkarni, 1993; Hore et al., 1997; Gao et al., 2000; Wall et al., 2013). Therefore, direct and indirect pathways work in a coordinated manner in motor control and nociception (Klaus et al., 2019).

Genome-wide association studies (GWAS) have implicated that up to 19 risk loci, including single nucleotide polymorphisms (SNPs) in BTBD9 intron 5, increase the susceptibility to RLS (Stefansson et al., 2007; Winkelmann et al., 2007; Schormair et al., 2017). As a leading gene associated with RLS, BTBD9 encodes a protein belonging to the BTB (POZ) protein family, which is ubiquitously expressed and modulates cytoskeleton arrangement, transcription repression, and protein ubiquitination (Stefansson et al., 2007; Winkelmann et al., 2007). A loss of the BTBD9 homolog, Btbd9, in mice is linked to altered hippocampal synaptic plasticity and enhanced learning and memory (DeAndrade et al., 2012b). Furthermore, both mice and fruit flies with knockouts of BTBD9 homologs show RLS-like phenotypes. Specifically, Btbd9 complete knockout mice develop motor restlessness, disrupted sleep, and altered sensory perception (DeAndrade et al., 2012a). The sensory deficit of knockout mice can be relieved by ropinirole, a DAergic agonist widely used for RLS treatment (DeAndrade et al., 2012a). Similarly, loss of the BTBD9 homolog in Drosophila melanogaster, CG1826, results in increased motor activity, decreased DA levels, and disrupted sleep patterns (Freeman et al., 2012). The sleep phenotype can be reproduced by RNAi-mediated knockdown of CG1826 in a subset of DA neurons (Freeman et al., 2012). In conclusion, these observations suggest that loss-of-function in BTBD9 can lead to abnormal DAergic function and RLS pathogenesis; however, the mechanisms through which mutations of BTBD9 lead to DAergic dysfunction and RLS-like phenotypes are not known.

To determine if the loss of BTBD9 in DA neurons alone can cause RLS-like phenotypes in mice, we characterized the egg-laying and locomotor behaviors of a mutant Caenorhabditis elegans (C. elegans) strain, in which the BTBD9 homolog, hpo-9, was knocked out. Because of the simple nervous system and the ease of making genetic crosses in worms, we were able to investigate the relationship of the BTBD9 homolog and DA receptors. This was followed by complementary studies in Btbd9 complete knockout mice. Furthermore, we generated DA neuron-specific Btbd9 knockout mice (Btbd9 dKO) using the Cre-loxP system and analyzed their sensorimotor behaviors related to RLS.

Materials and methods

C. elegans strains

C. elegans strains were maintained on nematode growth media (NGM) plates seeded with Escherichia coli strain OP50 at 20°C using standard methods unless otherwise noted. The wildtype (WT) strain used in this study was Bristol N2. The hpo-9(tm3719) strain was obtained from the National BioResource Project (Japan). The hpo-9(tm3719) strain lacks nucleotides 12131/12132 to 12892/12893 (761 bp deletion) of cosmid C05C8 (Figure 1A). The hpo-9(tm3719) mutant was backcrossed four times to the N2 background. PCR was used to genotype the tm3719 allele (KO1: 5’-ACAAATCTGTTGTACAACATCTT-3’ and KO2: 5’-GATAGTGTGGAATTATATTCGTGT-3’, Figure 2A) and the wild-type (WT1: 5’-CAGCAATAAGCGAATATTTCAAGG-3’ and WT2: 5-AATCTCTCGCAAGAAGCTCC-3’, Figure 1A). Strains obtained from Caenorhabditis Genetics Center (University of Minnesota, Twin Cities) are dop-1(vs100), dop-3(vs106), dop-1(vs100); dop-3(vs106), and vsIs33; lin-15B&lin-15A(n765); vsIs28. To generate hpo-9(tm3719);dop-1(vs100), hpo-9(tm3719);dop-3(vs106) double knockouts, hpo-9(tm3719);dop-1(vs100); dop-3(vs106) triple knockouts, and hpo-9(tm3719); dop-1::gfp; dop-3::rfp strain, we induced hpo-9(tm3719) mutant males by heat shock (34° C) and crossed them into dop-1(vs100), dop-3(vs106), dop-1(vs100); dop-3(vs106) and vsIs33; lin-15B&lin-15A(n765); vsIs28 respectively using standard method. Double and triple mutants were confirmed by fluorescence and PCR (dop-1 forward: 5’-GTGTGCTGGCTTCCGTTCTTC-3’; dop-1 reverse: 5’-GATTCAGGCGAGTTGCATTCG-3’; dop-3 forward: 5’-CCAGAAAGCGTCCCATCTTCC-3’; dop-3 reverse: 5’-GACCGCGCTGAACCAAAGTATG-3’), with target size of 291 bp for dop-1 knockout and 384 bp for dop-3 knockout, respectively.

Figure 1.

Generation of hpo-9(tm3719) worm and egg laying behavior. (A) hpo-9 gene structure and tm3719 mutation. Black rectangular boxes denote coding exons. Redline above shows the genomic region deleted. The sites of the PCR primers are indicated by arrowheads, with black for N2 and red for hpo-9(tm3719) mutant. The PCR result is presented at right, confirming the successful deletion of the hpo-9 allele. (B, C) Egg retention assay. (D) Egg-laying assay. *, p < 0.05; ***, p < 0.005. Bars represent the means plus SEs for 7–20 animals of each strain. (E) Representative pictures of N2 and hpo-9(tm3719) worms 22 hrs after the L4 stage. Scale bar represents 50 μm. (F, G) Embryo stages of unlaid and laid eggs.

Figure 2.

Egg retention assay with or without the pretreatment of DA. The numbers of eggs retained were increased with DA treatment in both N2 and hpo-9(tm3719) strains. hpo-9(tm3719) had fewer eggs than N2 under either situation. Bars represent the means plus SEs for 12 animals for each strain. *, p < 0.05; ***, p < 0.005. hpo-9, hpo-9(tm3719) worms.

Egg retention assay

The number of eggs retained in utero was measured as previously described (Chase and Koelle, 2004). Briefly, larval stage 4 (L4) worms (n=10–20 for all strains) were isolated and allowed to mature at 20°C for 22 hrs. Next, young adult worms were placed individually in a solution of sodium hypochlorite, which dissolves the worm but leaves the eggs intact. The number of eggs sustained was counted within 10 to 20 min, and the developmental stages of the eggs were identified and scored (Ringstad and Horvitz, 2008).

Egg-laying assay

To count the number of eggs laid in 5 days, we obtained synchronized young adult worms (n=12 for both strains) as described above. Worms were isolated and placed on individual NGM plate seeded with OP50 for 24 hrs. Subsequently, each adult worm was moved to a new plate, and the number of eggs left on the previous plate was counted. This step was repeated 5 times, and the total number of eggs laid in 5 days was counted for each worm. To identify and score developmental stages of eggs laid, we isolated and placed synchronized young adult worms (n=8–12 for both strains) on NGM plate seeded with OP50 for 1 hr. The scores were given according to others (Ringstad and Horvitz, 2008).

RNA interference

RNAi against hpo-9 (hpo-9 RNAi) (Kamath et al., 2003), or an empty vector (EV) was used according to a standard feeding method with HT115 bacterial strain (Kamath et al., 2001). Briefly, adult worms laid eggs on a plate covered with bacteria containing the hpo-9 RNAi (n=8) or the EV (n=7). Then the adult worms were removed, and the progeny grew on that plate until the L4 stage. The worms were then transferred to another plate seeded with the HT115 containing the same RNAi or EV. After 22 hrs, eggs retained in utero were counted as described above.

DA modulation of egg retention

Synchronized young adult worms (n=11–16) were placed in 8 mM DA solution freshly dissolved in NGM buffer or NGM buffer alone for 1 hr, after which they were dissolved by sodium hypochlorite and the number of eggs unlaid was counted. The protocol was modified from (Weinshenker et al., 1995).

WormLab

Worms used for filming were all at the age of 24 hrs following L4. Movies for behavioral analyses were obtained using the WormLab system (MBF Bioscience) and were acquired at a rate of 14 frames/s at room temperature (22–24°C).

For locomotor assays on agar, an individual worm was transferred to fresh 5 cm NGM agar plates thinly seeded with OP50 and allowed to acclimate for 1–2 min before filming. In all cases, 2 min digital video of the individual worm was captured and then analyzed using the WormLab software. All parameters were kept constant at the manufacturer’s recommended settings except that the threshold was adjusted individually to maximize the optimal recognition of worms.

Imaging and fluorescence quantification

Worms were mounted on a 1% agarose pad and analyzed with a CCD camera (Qimaging optiMOS) on an Olympus BX51 microscope at 40× magnification. GFP or RFP intensity was measured in mean gray value (backgrounds were subtracted with rolling ball radius of 200 pixels) from the head region right behind the support cells and ended after the nerve ring using ImageJ (Thompson et al., 2014). The experiment was repeated twice. For the first batch, 10 dop-1::gfp; dop-3::rfp worm and 20 hpo-9(tm3719); dop-1::gfp; dop-3::rfp were used. For the second batch, 30 animals were quantitated for each strain. The representative pictures were captured at 10× magnification.

Mice

All experimental procedures complied with the National Institutes of Health guidelines for animal care and the Institutional Animal Care and Use Committees at East Carolina University and the University of Florida. Btbd9 loxP mice were imported from the European Mouse Mutant Archive (EMMA) (EMMA ID: 05554) in which the 4th exon of the Btbd9 gene is flanked by loxP sites. Generation of the Btbd9 complete knockout mice was performed as described (Lyu et al., 2019b). Most of the experiments were conducted with males to minimize the variations caused by estrous cycles in females.

Dat-cre mice were imported from Jackson’s Lab (stock no. 6660) and used to breed with Btbd9 loxP mice to obtain Dat-cre and Btbd9 loxP double heterozygous mice. When cre and loxP sites are present in the offspring, Cre-mediated recombination occurs around the 4th exon, resulting in the loss of the 4th exon and a frameshift mutation, thereby effectively leading to the inactivation of Btbd9 gene specifically in DA neurons (Figure 9A). Double heterozygous mice (Dat-cre+/-Btbd9 loxP+/-) were used for breeding with hetero- or homozygous Btbd9 loxP mice to generate the experimental mice in which cre is expressed and loxP sites exist in both alleles (Dat-cre+/-Btbd9 loxP−/−, Btbd9 dKO) and control groups, including WT littermates, Dat-cre+/- mice, and animals only having loxP sites in one allele (Btbd9 loxP+/-) or both of alleles (Btbd9 loxP−/−). PCR was used for genotyping the Dat-cre (common forward: TGG CTG TTG GTG TAA AGT GG; WT reverse: GGA CAG GGA CAT GGT TGA CT; mutant reverse: CCA AAA GAC GGC AAT ATG GT) and the loxP sites (forward: ACA TCA CCC ATT ACT TAG AAC CTC; reverse: CAC AGC TAT TTC CTG TCA TTC TGG ACA). To confirm the specific deletion of Btbd9 in DA neurons, we dissected out brain regions following the protocol (Spijker, 2011). PCR was used to detect null alleles with primers (forward: CAT GTG AAG TGG AGC AAA GGA; reverse: GAA CCT CTA TGT GTG CTA CCT) (Figure 9B). The mice were housed under normal 12 hrs light and 12 hrs dark condition (12 LD).

Figure 9.

Generation of Btbd9 dKO mice and behavioral tests. (A) Schematic diagram of the generation of the Btbd9 dKO mice. Filled boxes represent exons. Filled triangles indicate loxP sites. Open triangles indicate the FRT sites that were incorporated to remove the neo cassette. In Btbd9 dKO mice, exon 4 is deleted in DA neurons because cre is expressed specifically in DA neurons, and the recombination occurs in the cells. The gray arrows indicate the sites of the PCR primers used to detect the knockout. (B) Tissue-specific deletion of Btbd9 exon 4 in Btbd9 dKO mice was confirmed by PCR using DNA isolated from each brain region. The deletion was detected only in the midbrain (includes SN) of Btbd9 dKO mouse as predicted (Δ). (C) 30-min open field test (Btbd9 dKO, n=7; controls, n=8). (D) Wheel running test (Btbd9 dKO, n=10, controls, n=10). (E-F) Continuous open filed test (Btbd9 dKO, n=5; controls, n=8). (G) Tail-flick experiment (Btbd9 dKO, n=9; controls n=12). The data in figure D-G were analyzed after log transformation. GEE model normalized the control group to 1 without the error bar (see Method). Bars represent means plus standard errors (SEs). The data in Figures D, E and G were presented as median with 95% CIs. *, p < 0.05.

Open field

Seven naïve Btbd9 dKO and 8 naïve male control littermates at an average age of 4 months were used in the open field analysis as previously described (Dang et al., 2005) during midday. Briefly, each mouse was placed in the center of a VersaMax Legacy open field apparatus connected to a computerized Digiscan System (Accuscan Instruments, Inc. OH) and continuously monitored for 30 min. Bright illumination (approximately 1 k lux at the center by a 60 W white bulb) was focused on the center of each field.

Continuous open field

Five naïve Btbd9 dKO and 8 untested male control littermates were used in the experiment. The animals have an average age of 4 months and were maintained on 12 LD for 7 days. As described previously (Meneely et al., 2018; Lyu et al., 2019a; Lyu et al., 2019b; Lyu et al., 2019c), each mouse was placed in the center of a VersaMax Legacy open field apparatus with enough corncob bedding, food, and water. The apparatus contains infrared sensors along the walls that detect any breaks in the beams, which are then decoded by VERSDATA version 2.70–127E (AccuScan Instruments INC.) into behavioral patterns. The data was recorded every 15 min (15 min bin) throughout the experiment. Only the last 4 days of data were analyzed. To compare the total distance traveled between Btbd9 dKO and WT, we separated the data into the light cycle and the dark cycle. Each cycle contains 4 periods, from day 4 to day 7, or night 4 to night 7. The distance traveled during each period was added up from the 15 min bins. To compare the probability of waking between the two groups, we recoded the data according to the total distance traveled during each 15 min bin. If the total distance traveled during the 15 min was 0, the mouse was considered as sleeping, and the data were coded as 0; otherwise, the mouse was considered as awake, and the data were coded as 1.

Wheel running

Ten Btbd9 dKO male mice and 10 male control littermates at an average age of 4 months were maintained under a 12 LD condition for 7 days. Most of the animals had been tested in the short open field experiment except 3 dKO and 2 controls, which were naive. Wheel running activity was recorded as the number of wheel revolutions occurring during 5 min bins and analyzed using Lafayette Instrument Activity Wheel Monitor software. The activities from the 4th to the 7th day were included in the data analysis, grouped by light phase and dark phase.

Tail flick

Nine male Btbd9 dKO mice and 12 male control littermates at an average age of 7 months were tested for the perception of warm stimuli using the Tail Flick Analgesia Meter (San Diego Instruments) as previously described (DeAndrade et al., 2011). Seven dKO and 8 control mice had been tested with short open field followed by wheel running. Two dKO and 2 control mice had been tested with wheel running only. Two additional new naive controls were used in the experiment. Briefly, each mouse was placed in an acrylic restrainer with the distal end of its tail protruding under a heat lamp. The lamp, together with a timer, was turned on, both of which stopped automatically when the mouse flicked its tail away from the light. The latency to respond was limited to 30 s to avoid injury to the mouse.

Western blot

Western blot was performed as previously described (Yokoi et al., 2015). The striata were dissected from 7 Btbd9 complete knockout and 6 WT with an average age of 8 months at midday, and 6 Btbd9 complete knockout and 7 WT with an average age of 8 months at midnight. The mice did not have any behavioral testing history. The lumbar spinal cords were dissected from 5 Btbd9 complete knockout and 3 WTs with an average age of 5 months. Tissues were homogenized in 200 μl of ice-cold lysis buffer (Tris/HCL 50mM, pH=7.4; NaCl 175 mM; EDTA 5mM, pH=8.0) containing protease inhibitor cocktail (Roche). 22 μl of ice-cold 10% Triton X100 was added in the homogenate. The mixtures were incubated for 30 min on ice and centrifuged at 10,000 × g for 15 min at 4°C. The supernatant was used as protein samples for Western blot. The protein concentration of the supernatant was measured by protein assay reagent (Bio-Rad). An aliquot of the supernatant corresponding to 30 μg of protein was mixed with 2× loading buffer containing 2-mercaptoethanol and boiled for 5 min, chilled on ice and spun down. The proteins were separated on a 10% SDS-PAGE gel and transferred to Millipore Immobilon–FL transfer membranes (PVDF). The PVDF membranes were washed in 0.1M PBS for 5 min and blocked with LI-COR Odyssey blocking buffer for 1 hr. The membranes were incubated overnight at 4°C with rabbit polyclonal D₁R antibody (Abcam, ab20066) at 1: 2000 dilution (for striata) or mouse monoclonal D₁R antibody (Abcam, ab78021) at 1: 500 dilution (for spinal cords), mouse monoclonal D₂R antibody (Santa Cruz, sc-5303) at 1:500 dilution, goat polyclonal dynamin I (Santa Cruz, sc-6402) at 1:1000 dilution or goat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (Santa Cruz, sc-20357) at 1:2000 dilution in the blocking buffer. The membranes were washed with 0.1M PBS containing 0.1% Tween 20 for 4 times at 5 min each, then treated for 1 hr with LI-COR IRDye 680RD donkey anti-rabbit IgG (H+L), LI-COR IRDye 800CW donkey anti-mouse IgG (H+L), or LI-COR IRDye 800CW donkey anti-goat IgG (H+L) at 1:15,556 dilution. After being washed 4 times with 0.1M PBS containing 0.1% Tween 20 for 5 min each and 0.1M PBS 3 times for 5 min each, the membranes were dried, and the signals were detected and quantified by an LI-COR Odyssey imaging system.

Quantitative RT-PCR

To determine whether mRNA levels of D₁R and D₂R were altered, we performed quantitative PCR by using CFX real-time PCR detection system (Bio-Rad) with SYBR Select Master Mix for CFX (life technologies) and PCR primer sets as described before (Dang et al., 2012; Yokoi et al., 2015). In brief, 3 Btbd9 complete knockout and 3 WT naïve mice at an average age of 4 months were sacrificed during midday. A separate cohort of 4 Btbd9 complete knockout and 5 WT naïve mice at an average age of 3 months were sacrificed at midnight. Their striata were dissected out and flash-frozen in liquid nitrogen. RNA was extracted using an RNAeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Next, cDNA was made using SuperScript III reverse transcriptase (Invitrogen). The relative quantity of cDNA for D₁R and D₂R to that of β-actin was calculated by CFX Manager™ Software #1845000.

In Vitro recording of DA neurons in SN

The experiment was conducted with 5 Btbd9 complete knockout and 3 WT naïve male littermates with an average age of 8 months. The electrophysiological recordings were performed by investigators blind to the genotypes. Mice were deeply anesthetized by the inhalation of isoflurane and then decapitated. The brains containing the midbrain part were rapidly removed and cut coronally into 250 μm-thick slices in ice-cold, oxygenated cutting saline (in mM: 180 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 10 D-glucose, 2 CaCl₂, 10 MgCl₂, and 10 glucose) using a Vibratome (Leica VT 1000s). The slices were recovered in a holding chamber for 60 min at 35°C with artificial cerebrospinal fluid (ACSF). Final concentrations of ACSF (in mM): 126 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 1 MgCl₂, 2 CaCl₂, and 10 glucose. The slices were then incubated at room temperature until use. Slices were placed in a recording chamber and continuously perfused with ACSF, which was bubbled with 95% O₂ / 5% CO₂ at 34.5–35.5°C at a rate of 1.5 ml/min and visualized with an upright microscope (Zeiss, Germany) using a 40× water-immersion objective with infrared optics. The DA neurons were identified by large cell body and triangular multipolar shape in the SN pars compacta.

Cell-attached recording electrodes were filled with a K-gluconate-based solution containing the following (in mM): 130 K-gluconate, 10 HEPES, 0.6 EGTA, 5 KCl, 4 MgCl₂·6H₂O, 3 Na₂ATP, 0.3 Na₃GTP and 10 Na₂-phosphocreatine, pH 7.3 with KOH (270–280 mOsm/l) and had resistances of 5–10 MΩ. While approaching the cell, the patch electrode was applied with positive pressure. The seal (> 5 GΩ) between the recording pipette and the cell membrane was obtained by applying suction to the electrode. Action potential currents were recorded in a voltage-clamp mode, which maintained an average 0 pA holding current.

After breaking through the membrane, cell properties (capacitance, input resistance, and time constant) were obtained while the membrane potential was held at −70 mV. Resting membrane potentials were recorded in current clamp mode. Depolarizing and hyperpolarizing incremental current steps (25pA) of 1000 ms were delivered to trigger action potentials for current step recording. Then hyperpolarizing current steps of 500 ms were delivered to produce a prominent time-dependent sag in the voltage deflection; this process was repeated at 9 increasingly hyperpolarized potentials with incremental current steps (25pA).

All experiments were recorded at 32±0.5°C by a dual automatic temperature controller (TC-344B). Cell-attached recording and whole-cell recording were obtained from DA neurons using Axopatch 1D Amplifier (Molecular Devices), and data were acquired using pCLAMP 10 software (Molecular Devices, USA). Signals were filtered at 5 kHz, digitized at 10 kHz with a DigiData 1440 (Molecular Devices, Union City, CA). Cell firing activity was recorded in the form of action potential currents, which were detected by the Mini Analysis Program.

Recording of spinal reflex amplitudes (SRAs)

Experimental procedures have been described in detail previously (Clemens and Hochman, 2004; Keeler et al., 2012). Neonatal WT and Btbd9 complete knockout mice (P7–14, WT: n=8, Btbd9 heterozygous knockout: n=6, Btbd9 homozygous knockout: n=4) were anesthetized and decapitated, with the spinal cord rapidly dissected out in a Sylgard-lined Petri dish containing aerated (95% O₂ / 5% CO₂) artificial cerebrospinal fluid (ACSF) (in mM): 125 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 25 glucose, 1.25 NaH₂PO₄, and 26 NaHCO₃, pH 7.4. After the opening of the dura mater, spinal cords were hemisected, and pairs of suction electrodes were attached to corresponding dorsal and ventral lumbar roots on each hemisect. After a resting phase of ~ 60 min at room temperature, reflex responses were elicited with a constant current stimulator (Iso-Stim 01D, NPI Electronics, Tamm, Germany) with pulses of 100–500 μA, 50–250 μs, at intervals of 30–60 s. Spinal reflex responses were recorded from corresponding ventral roots with a 4-channel differential AC amplifier (Model 1700, A-M Systems, Sequim, WA), digitized with a Digidata 1440A, and analyzed with the pClamp software package (Molecular Devices, Sunnyvale, CA). Reflex responses were recorded and analyzed by rectified integration. After establishing stable baseline recordings, we tested for the effects of DA D₁R agonist, SKF 38393 (Tocris, Ellisville, MO, 10 μM), on SRAs. SKF 38393 was bath-applied for durations of 30–60 min and subsequently washed out for equally long epochs. Using SigmaPlot (Systat Software, Inc., San Jose, CA), we compared SRA responses during the last 10 min of the drug perfusion protocol with those of the last 10 min in ACSF alone. All experiments were performed under “double-blind” conditions, with the experimenters at ECU unaware of the genetic makeup of the animals studied. At the end of the dissection process, tails were harvested and sent to the Li-lab for genotyping. Only after successful genotyping were the electrophysiological datasets matched to the respective WT and the 2 Btbd9 knockout groups.

Statistical analysis

Data were tested for normality using the SPSS statistical package before analysis. The open field data and fluorescence intensity were normally distributed and analyzed by mixed model ANOVA (SAS statistical package). Data of continuous open field, tail flick, wheel running, and electrophysiological recording of DA were not normally distributed and analyzed by generalized mixed linear model (GENMOD) using a GEE model for repeated measurement (SAS), which log-transformed the data and then normalized the WT or the control group to 0 without the error bar. The age of mice was used as covariates in both ANOVA and GENMOD. For the spontaneous activity of DA, cell ID was nested within the animal ID. For the current step recording of DA, both cell ID and the intensity of currents were nested within the animal ID. The number of laid or unlaid eggs, WormLab data, qRT-PCR and the Western blot results were analyzed by Student’s t-test. The distribution of the stages of unlaid and laid eggs was analyzed by Wilcoxon Mann-Whitney rank-sum test, a nonparametric test.

Results

Hyperactive egg-laying behavior in hpo-9(tm3719) mutant C. elegans

Compared with mice, C. elegans have a much simpler nervous system and are a strong tool for studies about genetic interactions. We obtained a C. elegans strain with the Btbd9 homolog gene, hpo-9, knocked out. We first measured the egg-laying behavior because it is known to be influenced by DA (Schafer and Kenyon, 1995; Schafer, 2005). Egg-laying behavior is quantified by counting either the number of eggs laid or the number of eggs retained in the uterus, which is determined by the rates of egg production and egg laying. Egg-laying occurs about every 20 min (Waggoner et al., 1998), while the number of unlaid eggs or egg retention assay is a measurement of egg-laying behavior over many hours. An egg retention assay removes any variations in egg-laying rates and provides an indirect readout of egg-laying behavior that is both sensitive and consistent (Chase and Koelle, 2004; Gardner et al., 2013). The egg production rate can be quantified by the determination of brood size over the lifetime of a worm (Hodgkin and Barnes, 1991). Finally, the egg-laying deficit can also be detected by examining the developmental stages of the eggs laid and retained in the uterus (Trent et al., 1983; Ringstad and Horvitz, 2008).

hpo-9(tm3719) worms retained significantly fewer eggs in utero compared to N2 worms (Figure 1B, p<0.0001). Similarly, knocking down hpo-9 using RNA interference (RNAi) in N2 worms resulted in a significantly reduced number of eggs in utero (Figure 1C, p=0.001), suggesting that hpo-9(tm3719) worms exhibit hyperactive egg-laying behavior. We investigated further by measuring the stages of eggs held in the uterus and those that were freshly laid. There were significantly more unlaid eggs at earlier embryonic stages in the uterus of hpo-9(tm3719) than N2 worms (Figure 1F). In addition, hpo-9(tm3719) laid significantly more eggs at early development stages compared to N2 worms (Figure 1G). These results suggest a hyperactive egg-laying behavior in the hpo-9(tm3719). Finally, egg production was determined by the number of eggs produced over five days. We found that hpo-9(tm3719) worms had a significantly increased brood size than the N2 worms (Figure 1D, p=0.02).

DAergic modulation of egg-laying in hpo-9(tm3719) mutants

To determine if hpo-9 knockout can cause an altered response to exogenous DA, we applied the drug and quantified the egg retention and locomotion behaviors with hpo-9(tm3719) and N2 worms. Synchronized adult worms were placed in either NGM solution or DA dissolved in NGM (8mM), and the egg retention assay was conducted after 1 hour. Application of DA significantly increased the number of eggs retained in both N2 and hpo-9(tm3719) worms (Figure 2, N2, p=0.003, hpo-9 (tm3719), p=0.04), which is consistent with previous findings that DA and DAergic receptor agonists could inhibit egg-laying behavior in C. elegans (Schafer and Kenyon, 1995). The hpo-9(tm3719) worms held significantly fewer eggs than N2 worms either with or without DA (Figure 2, without DA, p<0.0001, with DA, p=0.0007). The results suggest that the DA is insufficient to normalize the egg retention deficit observed in hpo-9(tm3719) mutants and the inhibitory effect of DA on egg-laying is intact in the hpo-9(tm3719) worms.

The relationship between hpo-9, dop-1, and dop-3

We then investigated the genetic interactions among hpo-9 and genes for DAergic receptors. To date, four DA receptors have been identified in C. elegans, including D₁-like DA receptor DOP-1 and D₂-like DA receptor DOP-3 (Chase and Koelle, 2007). The functions of these DA receptors in egg-laying behavior have not been previously reported. Therefore, we performed an egg retention assay to determine whether the dop-1 or dop-3 mutation could impact the egg-laying behavior by egg retention assay. We found that dop-1(vs100) mutants retained a similar number of eggs as N2 animals (Figure 3A, p=0.07), while dop-3(vs106) mutants retained significantly more eggs (Figure 3A, p=0.003). This suggests that dop-3, but not dop-1, is directly involved in the egg-laying behavior. Next, we found that knocking out dop-1 in dop-3(vs106) worms decreased the number of unlaid eggs to the N2 level (Figure 3A, p=0.02), indicating that dop-1 antagonizes with dop-3 in the regulation of the egg-laying behavior. Moreover, application of DA did not change the numbers of unlaid eggs in both dop-1(vs100) and dop-3(vs106) (Figure 3A, dop-1(vs100), p=0.94, dop-3(vs106), p=0.54), which suggests that the inhibitory effect of DA on egg-laying behavior depends on both DOP-1 and DOP-3.

Figure 3.

Egg retention assay and WormLab recording showing relationship among hpo-9, dop-1 and dop-3. Bars represent the means plus SEs for 10–20 animals of each strain. *, p < 0.05; **, p < 0.01; ***, p < 0.005. hpo-9, hpo-9(tm3719) worms. dop-1, dop-1 (vs100) worms. dop-3, dop-3(vs106) worms. dop-1; dop-3, dop-1(vs100); dop-3(vs106) worms. hpo-9; dop-1, hpo-9(tm3719); dop-1(vs100) worms. hpo-9; dop-3, hpo-9(tm3719); dop-3(vs106) worms. hpo-9; dop-1; dop-3, hpo-9(tm3719); dop-1(vs100); dop-3(vs106) worms.

Next, we included hpo-9(tm3719) and double knockouts in the assay and observed that hpo-9(tm3719);dop-1(vs100) worms retained significantly more eggs than hpo-9(tm3719) mutants (Figure 3B, p<0.0001) but fewer eggs than dop-1(vs100) mutants (Figure 3B, p<0.0001) or N2 worms (Figure 3B, p=0.006). Therefore, the dop-1 mutation partially rescues the egg retention deficit caused by the loss of HPO-9. In contrast, hpo-9(tm3719);dop-3(vs106) worms had the same number of eggs in utero as N2 worms (Figure 3B, p=0.48), more eggs than hpo-9(tm3719) mutants (Figure 3B, p<0.0001), but fewer eggs than dop-3(vs106) worms (Figure 3B, p=0.04). These results demonstrate that dop-3 and hpo-9 have opposing effects on the number of eggs retained and counteract each other. Thus, the dop-3 mutation completely suppresses the hyperactive egg-laying behavior of hpo-9(tm3719).

Finally, we used a locomotor study to explore the relationship between hpo-9, dop-1, and dop-3. It has been previously shown that dop-3 knockout increases resistance to DA-induced paralysis and dop-1 knockout reverses the defects resulting from the lack of DOP-3 (Chase et al., 2004). We recorded the movement speed of different strains with WormLab and found that dop-3(vs106) worms moved significantly faster than N2 worms (Figure 3C, p=0.002) while hpo-9(tm3719) or dop-1(vs100) mutants moved at a speed equivalent to N2 worms (Figure 3C, hpo-9(tm3719), p=0.68, dop-1(vs100), p=0.53). Therefore dop-3, but not dop-1 or hpo-9, affects the worm movement speed. Knocking out hpo-9 in either dop-1 (Figure 3C, p=0.04) or dop-3 single mutant worms significantly reduced the movement speed (Figure 3C, p=0.02), indicating that dop-1(vs100) and dop-3(vs106) worms, in which the DAergic systems have been disturbed, are super sensitive to the additional hpo-9 mutation. Knocking out dop-1 or hpo-9 in dop-3 knockout background reduced the movement speed of mutant worms to the N2 level (Figure 3C), suggesting that both dop-1 and hpo-9 antagonize with dop-3. Moreover, in the dop-1(vs100); dop-3(vs106) worms, the hpo-9 mutation led to a significant increase in the movement speed (Figure 3C, p=0.002). In hpo-9(tm3719); dop-3(vs106) worms, the dop-1 mutation also stimulated movement speed instead of being inhibitory, as in the dop-3 single mutants (Figure 3C, p=0.007). These results suggest a functional similarity between HPO-9 and DOP-1. Finally, the effect of the hpo-9 mutation on movement speed is highly genetic background specific. The hpo-9 mutation did not affect the WT background, reduced movement speed in either the dop-1 or dop-3 single mutant background, and interestingly, significantly increased movement speed in the dop-1 and dop-3 double mutant background.

HPO-9 deficiency led to increased transcription of DOP-3

With two C. elegans behavioral assays indicating that hpo-9 interacts with both dop-1 and dop-3, we further tested the influence of hpo-9 knockout on expression levels of these DA receptors. We crossed hpo-9(tm3719) with strains harboring gfp under the control of dop-1 promoter and rfp under the control of the dop-3 promoter. Consistent with other literature (Chase et al., 2004; Ezak and Ferkey, 2010), dop-1 and dop-3 expressed throughout the body, including neurons of the head, the ventral cord and the tail (Figure 4A, left panel). Expression patterns of dop-1 and dop-3 in the hpo-9 mutant were similar to N2 (Figure 4A, right panel). We quantified the fluorescent intensity in the head region (Figure 4B) and found that the hpo-9 mutation led to a significantly higher fluorescent intensity of RFP (DOP-3) (Figure 4D, p=0.02), but not GFP (DOP-1) (Figure 4C, p=0.17). The result suggests that hpo-9 influences the transcription of DOP-3.

Figure 4.

Altered expression levels of DA receptors in the hpo-9 strain. (A) Representative fluorescent micrographs of animals expressing the dop-1::gfp; dop-3::rfp alone (left panel) and with hpo-9 knocked out (right panel). Scale bar represents 100 μm. (B) The image depicts the anterior region of C. elegans, and the boxed region indicates the head area where measurements of fluorescence were taken in all animals. Scale bar represents 25 μm. (C) Bar chart representing GFP intensity values (DOP-1) measured in the boxed region shown in B. (D) Bar chart representing RFP intensity values (DOP-3) measured in the boxed region shown in B. Bars represent the means plus SEs for 30 animals of each strain. ***, p < 0.005. hpo-9, hpo-9(tm3719) worms.

BTBD9 deficiency led to increased transcription of D₂R during midnight

Similarly, we tested the D₁R and D₂R mRNA levels in the striatum of Btbd9 complete knockout mice and their WT littermates. Symptoms of RLS patients usually occur or become worse in the evening or at night (Garcia Borreguero et al., 2017). Therefore, RLS is a disease with a circadian component. We collected the mouse samples two times a day, which were midday and midnight, respectively. It should be noticed that mice have opposite day-night rhythms to humans. They usually are active during the night but are sleeping during the day. We found a similar increase of D₂R mRNA expression (Figure 5B, p=0.03; C, p=0.04), but not D₁R (Figure 5A, p=0.75), in the striatum of Btbd9 complete knockout mice sacrificed during midnight. However, there were no significant differences in D₁ and D₂ mRNA expression during midday between WT and Btbd9 complete knockout mice (Figure 5A, p=0.41; 5B, p=0.55; 5C, p=0.27), indicating that the influence of Btbd9 complete knockout on mRNA level of D₂R is circadian-dependent. Interestingly, the mRNA levels of DA receptors were decreased in WT mice during the night or active phase (Figure 5A, p=0.02; 5B, p=0.04; 5C, p=0.02), but not in knockout mice (Figure 5A, p=0.12; 5B, p=0.44; 5C, p=1.00). These results suggest that BTBD9 and its homolog strongly influence the transcription of DA receptors in both worms and mice, with the later have additional circadian rhythm-dependent controls.

Figure 5.

Increased D2R mRNA levels during midnight (B, C) and an absence of circadian variations in the Btbd9 KO mice (A-C). Bars represent the means plus SEs. Midday: WT, n=3, Btbd9 KO, n=3; Midnight: WT, n=5, Btbd9 KO, n=4. *, p < 0.05. KO, knockout.

BTBD9 deficiency caused decreased protein level of D₂R during midday

The half-life of mRNA is generally short, and the mRNA level is not always consistent with the actual protein levels (Greenbaum et al., 2003). Therefore, we did a Western blot with mouse striatum to determine the D₁R and D₂R protein levels. D₁R protein was not altered (Figure 6A, p=0.83; Figure 6D, p=0.95), which is consistent with the RT-PCR data. Surprisingly, the D₂R protein level was significantly decreased during midday (Figure 6B, p=0.01). At midnight, the D₂R protein level showed a trend of decrease (Figure 6E, p=0.23). The result indicates that the BTBD9 mutation affects not only the D₂R mRNA level but also the D₂R protein level. The changes in the D₂R protein level may cause incoordination between D₁-mediated and D₂-mediated DAergic pathways, leading to RLS-like phenotypes.

Figure 6.

Western blot analysis of D₁R, D₂R, and DNM-1 protein levels in the striata of Btbd9 complete knockout and their WT littermates. (A) D₁R protein levels were not changed in Btbd9 complete knockout mice compared with WT littermates at midday. (B) D₂R protein levels were significantly decreased in Btbd9 complete knockout mice at the midday. (C) DNM-1 protein levels were not changed in Btbd9 complete knockout mice compared with WT littermates at midday. (D) D₁R protein levels were not changed in Btbd9 complete knockout mice compared with WT littermates at midnight. (E) D₂R protein levels were not significantly decreased in Btbd9 complete knockout mice at midnight. (F) DNM-1 protein levels were significantly increased in Btbd9 complete knockout mice at midnight. Target protein levels were normalized to the GAPDH. Blots were cropped to show representative bands. Bars represent the means plus SEs (A-E) or means with 95% confident intervals (CIs, F). Midday: WT, n=6, Btbd9 KO, n=7; midnight: WT, n=7, Btbd9 KO, n=6. *, p < 0.05. KO, knockout.

BTBD9 deficiency caused increased protein level of dynamin 1 in mouse striatum

The internalization of D₂R is under the regulation of dynamin I (DNM-1) (Iwata et al., 1999). The previous study found an increased level of DNM-1 in the hippocampus of Btbd9 complete knockout mice (DeAndrade et al., 2012b). Here, we determined the DNM-1 level in the striatum. There was no change in the level of the protein between Btbd9 complete knockout and WT mice euthanized during midday (Figure 6C, p=0.33). However, the level of DNM-1 was higher in the Btbd9 complete knockout compared with the WT euthanized at midnight (Figure 6F, p=0.01). The decreased level of D₂R may be caused by an elevated level of DNM-1.

BTBD9 deficiency did not influence the activity of DA neurons in mouse

To determine if the presynaptic part of the striatal DAergic system has altered by BTBD9 deficiency, we did an electrophysiological study with DA neurons in SN, which fire spontaneously in vitro in a single-spike, pacemaker pattern without bursts (Paladini et al., 2003), and are involved in motor control and reward-based learning (Berretta et al., 2010). Using patch-clamp recording in brain slices, we found that neither spontaneous firing frequency (Figure 7B, left panel, p=0.74) nor regularity (Figure 7B, right panel, p=0.27) of DA neurons was altered by the lack of BTBD9. In addition, the responses to external stimuli of DA neurons were also the same between Btbd9 complete knockout and their WT littermates (Figure 7E, p=0.60). The result indicates that the activity of DA neurons in SN is not altered.

Figure 7.

In vitro recording of DA neurons (WT, n=17; Btbd9 KO, n=35) in SN. (A) A representative spontaneous activity trace of a DA neuron. (B) Both spontaneous firing frequency and regularity were not significantly different between Btbd9 KO and WT mice. (C) Identification of DA neurons in SN pars compacta from current-clamp recordings. DA neurons exhibit a prominent time-dependent sag in the voltage deflection in response to hyperpolarizing current injection. (D) A representative trace of DA neurons in response to current steps at 100pA (top), and -50 pA (bottom). (E) In response to the step current injection, DA neurons of Btbd9 complete knockout mice fired a similar number of action potentials compared with the WT mice. Data in B and C were presented as median with 95% CIs. Data in figure E were analyzed after log transformation. GEE model normalized the WT to 1 without the error bar (see Method). KO, knockout.

The response of a D₁R agonist on SRAs was diminished with BTBD9 deficiency

Both central and peripheral DAergic systems have been shown to play a role in RLS (Yokota et al., 1991; Bara-Jimenez et al., 2000; Tings et al., 2003; Bachmann et al., 2010; Marconi et al., 2012; Ferri et al., 2015). Lesions of the hypothalamic descending A11 DAergic system leads to RLS-like phenotypes (Qu et al., 2007). To test the influence of loss of BTBD9 in the spinal DAergic system, we bath-applied the SKF 38393 to the isolated spinal cords and recorded the SRAs of WT and Btbd9 complete knockout mice. Application of SKF 38393 significantly increased SRAs of the WT mice (Figure 8A, p=0.03) as we reported earlier (Keeler et al., 2012). This stimulative function of SKF 38393, however, was absent in both the Btbd9 heterozygous (Figure 8A, p=0.45) and homozygous knockout mice (Figure 8A, p=0.93), indicating a potentiated D₁R-mediated DAergic pathway. To test if the increased activity in D₁R-mediated DAergic pathway is caused by expression change of DA receptors, we did Western blot and found that both D₁R and D₂R protein levels in the lumbar spinal cord remained the same in Btbd9 complete knockout mice (Figure 8B, left panel, p= 0.99; right panel, p=0.85).

Figure 8.

The effects of a D₁R agonist, SKF 38393, on SRAs and Western blot analysis of D₁R and D₂R protein levels in the lumbar spinal cord. (A) SKF increased the SRAs in WT mice (n=8) but did not influence the SRAs in Btbd9 heterozygous (n=6) and homozygous knockout mice (n=4). (B) Btbd9 complete knockout mice (n=5) had the same level of D₁R and D₂R proteins in the lumbar spinal cord as WT mice (n=3). Target protein levels were normalized to the GAPDH. Blots were cropped to show representative bands. Bars represent the means plus SEs. *, p < 0.05. KO, knockout.

No motor restlessness and thermal sensory alteration in Btbd9 dKO mice

To determine if BTBD9 deficiency specifically in the DA neurons is enough for mice to develop RLS-like phenotypes, we generated Btbd9 dKO mice as shown in Figure 9A (see Method). To confirm the tissue specificity of the cre-mediated recombination, we dissected and analyzed different parts of the brains. Only the midbrain showed the band with the right size for the recombination, indicating Btbd9 deletion was restricted to the midbrain DAergic neurons (Figure 9B).

The principal feature of RLS is the urge to move (Garcia Borreguero et al., 2017). Previous phenotypic mouse or fruit fly models of RLS have shown increased activity levels (Ondo et al., 2000; Clemens and Hochman, 2004; Esteves et al., 2004; DeAndrade et al., 2012a; Freeman et al., 2012). Therefore, to assess the total activity levels of the Btbd9 dKO mice, we did both short- and long-term open field experiments. In the 30-min open field test, Btbd9 dKO mice exhibited no change in total distance traveled (Figure 9C, left panel, p=0.75), as well as clockwise and counterclockwise circling compared with the control group (Figure 9C, right panel, CW, p=0.66, CCW, p=0.78). The continuous open field experiment did not detect any difference in total distance traveled between mutant and control mice either in the light phase (Figure 9E, light, p=0.07), when the animals are usually sleeping or resting, or during the dark phase, when the animals are active (Figure 9E, dark, p=0.67). However, Btbd9 dKO mice showed a decreased probability of waking in the dark phase (Figure 9F, light, p=0.04), which could suggest that the mutant mice showed sleepiness during the active phase. The abnormality in the active phase suggests a decreased sleep quality in the rest phase. This result is consistent with the symptoms of RLS patients that usually begin or worsen during the rest phase and therefore, these patients usually have disrupted sleep at night and show day-time fatigue (Garcia Borreguero et al., 2017). Next, we measured the voluntary activity of these mice using a wheel running setup. Btbd9 dKO mice showed unchanged levels of activity during both the light phase and dark phase (Figure 9D, light, p=0.33, dark, p=0.22). These data, taken together, suggest that Btbd9 dKO mice did not exhibit alteration in motor activity but have active phase sleepiness.

RLS patients have uncomfortable sensations in legs that are usually associated with the urge to move (Garcia Borreguero et al., 2017). In our previous study, Btbd9 complete knockout mice exhibited a sensory deficit (DeAndrade et al., 2012a) in the tail-flick test, so we tested the sensory system of Btbd9 dKO mice with the same method (DeAndrade et al., 2012a). The mutant mice had the same level of response to the heat stimuli as the control group (Figure 9G, p=0.72), indicating that the Btbd9 dKO mice did not exhibit altered thermal sensation. Therefore, except for the active-phase sleepiness, the Btbd9 dKO did not have RLS-like behaviors as observed in other BTBD9 homolog mutant animal models (DeAndrade et al., 2012a; Freeman et al., 2012).

Discussion

DA has been extensively implicated in RLS, and DA agonists targeting D₂ and D₃ receptors have been used to treat RLS patients. In this study, we employed two model systems to study the role of BTBD9 in the DAergic system and RLS. With C. elegans, we found HPO-9 and DOP-1 functioned similarly in both egg-laying and locomotor behaviors. HPO-9 deficiency led to increased transcription of dop-3. Similarly, Btbd9 complete knockout mice had a midnight-specific elevation of striatal D₂R mRNA, but not D₁R mRNA, compared with their WT littermates. The protein level of D₂R was reduced, which may be caused by an increased level of DNM-1. Presynaptic DA activities did not change in the Btbd9 complete knockout mice. The weakened D₂R-mediated central DAergic pathway was accompanied by the enhanced D₁R-mediated peripheral DAergic pathway. Furthermore, knocking out BTBD9 specifically in DA neurons led to active-phase sleepiness, indicating that the loss of Btbd9 only in DA neurons was enough to induce a sleep deficit.

Our results contribute to the understanding of the DA signaling in the egg-laying behavior of C.elegans. Increased number of unlaid eggs in dop-3 but not dop-1 mutants suggests that dop-3, but not dop-1, directly regulates egg-laying behavior. Additionally, both dop-1 knockout and dop-3 knockout worms showed no response to exogenous DA. Thus, DA acted through both DOP-1 and DOP-3 to regulate egg-laying behavior. Moreover, knocking out dop-1 recovered the number of unlaid eggs of dop-3(vs106) to the N2 level, indicating that dop-1 counteracted with dop-3 in the regulation of the egg-laying behavior.

Btbd9 homolog gene, hpo-9, worked similarly with dop-1. First, both hpo-9 knockout and dop-1 knockout reduced egg retention in dop-3(vs106). Second, the WormLab locomotion data implied that the dop-1 or hpo-9 mutation alone did not induce changes in locomotion. However, the dop-1 or hpo-9 mutation in dop-3(vs106) reduced the higher moving speed observed in dop-3 single mutant worms to the N2 level, which is consistent with the antagonistic function of dop-1 over dop-3 found previously (Chase et al., 2004). Finally, in the hpo-9 and dop-3 double knockout worms, the dop-1 mutation increased the movement speed, which was opposite from its function in dop-3 single knockout worms. In the dop-1 and dop-3 double knockout worms, the HPO-9 deficiency increased the movement speed as well.

BTBD9 and its homolog participated in the transcriptional regulation of D₂ or D₂-like DA receptors. Worm reporter gene assays demonstrated an increase in dop-3 transcription with the hpo-9 knockout. In mice, D₂R mRNA was increased in the striatum of Btbd9 complete knockout at least at midnight, which suggests that the influence of BTBD9 deficiency on the D₂R mRNA level is under the control of circadian rhythm. DAergic system of rodents, at least the extracellular level of DA, is under the control of circadian rhythm (Khaldy et al., 2002; Castaneda et al., 2004; Akhisaroglu et al., 2005). Here, we found that in the WT mice, D₁R, D₂L, and D₂S mRNA levels were all decreased during the active phase compared with the rest phase. This transcriptional reduction of DA receptors in the active phase was absent in the Btbd9 complete knockout mice, which may be the reason for higher levels of mRNAs at night in the knockout mice. Btbd9 complete knockout mice showed sleep disturbance (DeAndrade et al., 2012a), and symptoms of RLS in patients are known to be more common at night (Garcia Borreguero et al., 2017). The disruption of circadian oscillations in the transcription of DA receptor mRNAs may contribute to these results.

The D₂R protein level was decreased and significantly reduced at midday in the Btbd9 complete knockout mice, during the peak of the sleep cycle. Regulation of mRNA translation, post-translational modifications, protein stability, and protein recycling combined dictate the final protein level in the cell. After binding with DA, D₁R and D₂R, which belong to G-protein coupled receptors (GPCRs), are transferred to intracellular compartments from the plasma membrane, a process known as internalization (Koenig and Edwardson, 1997). The internalization of D₂R is modulated by DNM-1 (Iwata et al., 1999), which showed a significant increase in the Btbd9 complete knockout mice. It is possible that an increased level of DNM-1 in the active phase accelerated the endocytosis of D₂R, which is usually followed by the degradation in the lysosome (Thompson et al., 2010), and caused the observed D₂R reduction in the rest phase (Figure 10). Combined with the mRNA data, the loss of transcriptional reduction of DA receptors during the active phase in Btbd9 complete knockout may be a compensatory result in response to a decreased level of D₂R protein level (Figure 10). It is also possible that BTBD9 protein, which contains a BTB domain, participates directly in the regulation of transcription (Stogios and Prive, 2004; Stogios et al., 2005).

Figure 10.

Model of regulation of D2R by BTBD9. BTBD9 acts as an adaptor for E3 ubiquitin ligase, which is necessary for the degradation of DNM-1. DNM-1 participates in the endocytosis of D2R. Therefore, BTBD9 deficiency leads to increased DNM-1 level and decreased D2R level. BTBD9 may also regulate the transcription of D2R.

We found that knocking out Btbd9 led to reduced activity in the D₂R-mediated indirect pathway in the striatum and increased activity of D₁R-mediated responses in the spinal cord. In the striatum of Btbd9 complete knockout mice, there was a significant decrease in D₂R but not D₁R. In addition, short term corticostriatal synaptic plasticity onto the D1 medium spiny neurons (MSNs) was increased (Lyu et al., 2019c). It has been reported that RLS patients show decreased D₂R binding potential in the striatum (Michaud et al., 2002). Both the symptoms of RLS patients and the thermal sensory deficit of Btbd9 complete knockout mice can be relieved by D₂/D₃ agonists (DeAndrade et al., 2012a). Therefore, our results are consistent with previous clinical and animal studies and help to explain the effectiveness of D₂/D₃ agonists in RLS. In addition, we also found that the isolated spinal cord of Btbd9 complete knockout mice lost the response to SKF 38393, a D₁R agonist. The result indicates that, with BTBD9 deficiency, the activity of the D₁R-mediated DAergic system may have reached an apex so that it cannot be increased anymore. Previously, D₃-receptor knockout mice (D3KO) and iron deprived (ID) mice have been used as models for RLS (Erikson et al., 2001; Clemens and Hochman, 2004; Zhao et al., 2007). In both D3KO and ID mice, there is a significant increase of the D₁R protein expression in the lumbar spinal cord (Zhao et al., 2007; Brewer et al., 2014). In contrast, Btbd9 complete knockout mice did not show the alteration in the D₁R protein level in the lumbar spinal cord, suggesting that the functional increase in the peripheral D₁R-mediated DAergic system in Btbd9 complete knockout mice is not caused by protein level change. The basal ganglia output modulates the spinal cord through feedback to the cortex. Moreover, a microcircuit has been suggested to exist between the corticostriatal tract and the corticospinal tract, starting in the striatum and ending in the spinal cord (Kiritani et al., 2012). This supports the existence of a weakened D₂R-mediated pathway but an enhanced D₁R-mediated pathway in RLS. We speculate that mutations of BTBD9 might result in incoordination between the two DAergic pathways, which should normally work in regulated harmony controlling motor and sensory outputs.

Btbd9 dKO mice did not have RLS-like phenotypes, as observed in Btbd9 complete knockout mice (DeAndrade et al., 2012a). Previously, we have found that the striatal MSN-specific Btbd9 knockout mice exhibit significantly increased total distance traveled in 30-min open field test, increased probability of waking during the light phase but not the dark phase in the continuous open field test, elevated level of activity during the light phase but not the dark phase in the wheel running test, and increased sensitivity to the heat stimuli in the tail-flick test (Lyu et al., 2019b). On the other hand, mice with Btbd9 specifically knocked out in cholinergic interneurons (ChI) do not show any difference in total distance traveled in 30-min open field test, and have a decreased probability of waking during the light phase but increased probability of waking during the dark phase in the continuous open field test (Lyu et al., 2019b). Therefore, we concluded that the loss of BTBD9 protein only in MSNs, but not ChIs, can cause RLS-like phenotypes in mice. In parallel, we also knocked out Btbd9, specifically in the cerebral cortex. The conditional knockout mice show a significant increase in their activity level and the probability of waking during the light phase, but not during the dark phase, in the continuous open field test. The mutant mice also show decreased sensitivity to the heat stimuli in the tail-flick test (Lyu et al., 2019c). The results indicate that the cerebral cortex also participates in the generation of RLS-like phenotypes. Overall, the data suggest that striatal MSNs and the cerebral cortex, but not DA neurons, play central roles in the pathogenesis of RLS.

In summary, hpo-9 resembled dop-1 functionally and participated in the regulation of DOP-3 expression. BTBD9 was involved in the regulation of both mRNA and protein levels of D₂R. Loss of BTBD9 and its homolog, HPO-9, caused altered DA-regulated outputs, which may be the underlying mechanism for RLS.