Abstract
A rod-shaped appendage called a primary cilium projects from the soma of most central neurons in the mammalian brain. The importance of cilia within the nervous system is highlighted by the fact that human syndromes linked to primary cilia dysfunction, collectively termed ciliopathies, are associated with numerous neuropathologies, including hyperphagia-induced obesity, neuropsychiatric disorders, and learning and memory deficits. Neuronal cilia are enriched with signaling molecules, including specific G-protein-coupled receptors (GPCRs) and their downstream effectors, suggesting that they act as sensory organelles that respond to neuromodulators in the extracellular space. We previously showed that GPCR ciliary localization is disrupted in neurons from mouse models of the ciliopathy Bardet–Biedl syndrome (BBS). Based on this finding, we hypothesized that mislocalization of ciliary GPCRs may impact receptor signaling and contribute to the BBS phenotypes. Here, we show that disrupting localization of the ciliary GPCR dopamine receptor 1 (D1) in male and female mice, either by loss of a BBS protein or loss of the cilium itself, specifically in D1-expressing neurons, results in obesity. Interestingly, the weight gain is associated with reduced locomotor activity, rather than increased food intake. Moreover, the loss of a BBS protein or cilia on D1-expressing neurons leads to a reduction in D1-mediated signaling. Together, these results indicate that cilia impact D1 activity in the nervous system and underscore the importance of neuronal cilia for proper GPCR signaling.
SIGNIFICANCE STATEMENT Most mammalian neurons possess solitary appendages called primary cilia. These rod-shaped structures are enriched with signaling proteins, such as G-protein-coupled receptors (GPCRs), suggesting that they respond to neuromodulators. This study examines the consequences of disrupting ciliary localization of the GPCR dopamine receptor 1 (D1) in D1-expressing neurons. Remarkably, mice that have either an abnormal accumulation of D1 in cilia or a loss of D1 ciliary localization become obese. In both cases, the obesity is associated with lower locomotor activity rather than overeating. As D1 activation increases locomotor activity, these results are consistent with a reduction in D1 signaling. Indeed, we found that D1-mediated signaling is reduced in brain slices from both mouse models. Thus, cilia impact D1 signaling in the brain.
Introduction
Primary cilia are restricted compartments and numerous mechanisms regulate the complement of ciliary proteins (Nachury and Mick, 2019). This partitioning of signal transduction pathways allows cilia to mediate specialized sensory and signaling functions that are important for proper development and cellular homeostasis (Anvarian et al., 2019). Cilia dysfunction results in a class of human diseases, termed ciliopathies, which impact many organ systems (Reiter and Leroux, 2017). Ciliopathies can manifest neuropathologies, including structural malformations, obesity, and intellectual disability (Suciu and Caspary, 2021), underscoring the importance of cilia for normal brain development and function.
Neuronal cilia are enriched for certain G-protein-coupled receptors (GPCRs), including the following: melanocortin 4 receptor (MC4R; Wang et al., 2021); somatostatin receptor 3 (Sstr3; Händel et al., 1999); melanin-concentrating hormone receptor 1 (Mchr1; Berbari et al., 2008b); kisspeptin receptor 1 (Kiss1r; Koemeter-Cox et al., 2014); Neuropeptide Y receptor 2 (NPY2; Loktev and Jackson, 2013); and dopamine receptor 1 (D1; Domire et al., 2011). The GPCR effectors type 3 adenylyl cyclase (AC3; Bishop et al., 2007) and β-arrestin (Green et al., 2016) also localize to neuronal cilia. Recent studies suggest that neuronal cilia respond to factors in the extracellular space. Disruption of cilia on Kiss1r-expressing neurons in mice reduces kisspeptin-mediated increases in firing rate (Koemeter-Cox et al., 2014). Sstr3 ciliary signaling modulates excitatory synapses in rat neocortical neurons (Tereshko et al., 2021). Disruption of Mchr1 ciliary localization in mouse hypothalamic neurons leads to decreases in MCH-mediated cAMP and extracellular signal-regulated kinase signaling (Hsiao et al., 2021). Inhibition of MC4R signaling in cilia in the mouse hypothalamus induces hyperphagia and obesity (Wang et al., 2021). Thus, neuronal ciliary localization is critical for normal signaling of several GPCRs.
GPCR ciliary localization is regulated by the proteins mutated in the ciliopathy Bardet–Biedl syndrome (BBS). BBS is a heterogeneous autosomal-recessive disorder that presents with a wide range of phenotypes, including obesity (Forsythe and Beales, 2013). A subset of BBS proteins (BBS1, 2, 4, 5, 7, 8, 9, and 18) form a complex called the BBSome (Nachury et al., 2007; Loktev et al., 2008). Neural defects in BBS knock-out (KO) mice may be caused by the loss of ciliary localization of several GPCRs (Sstr3, Mchr1, and NPY2; Berbari et al., 2008a; Loktev and Jackson, 2013). Yet, D1 abnormally accumulates in neuronal cilia in BBS KO mice (Domire et al., 2011). In this regard, the impact of BBSome disruption on D1 is an aberration among ciliary GPCRs, and it is unknown how this altered cellular localization might affect D1 activity in neurons.
D1 is one of five dopamine receptors that mediate the various physiological functions of the neurotransmitter dopamine, including voluntary movement, feeding, and reward (Beaulieu and Gainetdinov, 2011). D1 is not detected in cilia in brain sections from wild-type (WT) mice but shows abundant ciliary localization in BBS KO mice (Domire et al., 2011), suggesting that D1 transiently localizes to cilia and becomes “trapped” by loss of the BBSome. In support of this, D1 is observed in cilia on cultured WT neurons and is recruited to the ciliary membrane in response to increased cAMP levels and exported from cilia in response to agonist binding (Domire et al., 2011). Yet, the impact of ciliary localization on D1 signaling in vivo is unknown.
Understanding the physiological consequences of D1 ciliary localization is confounded in BBS KO mice by the mislocalization of multiple GPCRs. Here, we addressed this limitation by crossing a mouse line carrying a conditional Bbs1 allele with a mouse line expressing Cre recombinase under the control of the D1 promoter. These mice have abundant D1 ciliary accumulation in D1-expressing neurons. As a complementary approach. we used a conditional knock-out mouse line to disrupt cilia selectively on D1-expressing neurons. Interestingly, mice that have either abnormal accumulation of D1 in cilia or loss of D1 cilia, have lower activity levels, develop obesity, and show a reduction in D1-mediated signaling. These results demonstrate the importance of cilia for D1 signaling on central neurons.
Materials and Methods
Mice and tissue preparation.
All mouse studies were conducted in accordance with institutional guidelines based on National Institutes of Health standards and were performed with approval of the Institutional Animal Care and Use Committee at The Ohio State University (OSU). All animals were maintained in a temperature- and humidity-controlled vivarium with 12 h light/dark cycle at a maximum of five mice per cage and given access to normal chow and water ad libitum. The germline Bbs1 deleted (null) allele was generated by crossing Bbs1fl/fl mice (Carter et al., 2012) with B6.C-Tg(CMV-cre)1Cgn/J mice from The Jackson Laboratory (stock #006054). Bbs1wt/null mice were crossed with C57BL/6J mice to remove the CMV-Cre and then crossed with B6.FVB(Cg)-Tg(Drd1-cre)EY262Gsat/Mmucd mice from the University of California, Davis, Mutant Mouse Regional Resource Center (stock #030989) to introduce the D1-Cre. Bbs1wt/null male mice hemizygous for D1-Cre were crossed with Bbs1fl/fl female mice to generate Bbs1wt/fl::D1-Cre (control) and Bbs1null/fl::D1-Cre (experimental) mice for use in the behavioral and biochemical experiments. Bbs1 control and experimental mice were on a congenic background (C57BL/6J). For electrophysiology experiments, Bbs1wt/null male mice hemizygous for D1-Cre were crossed with Bbs1fl/fl female mice homozygous for R26R-EYFP to generate Bbs1wt/fl::D1-Cre::R26R-EYFP (control) and Bbs1null/fl::D1-Cre::R26R-EYFP (experimental) mice. R26R-EYFP mice were from the Jackson Laboratory (#006148). Constitutive Bbs1null/null mice were generated by intercrossing Bbs1wt/null male and female mice. Ift88cko mice, a gift of Brad Yoder (University of Alabama at Birmingham, Birmingham, AL), were on a mixed background (C57BL/6J × 129P2/OlaHsd). Ift88wt/null mice were crossed with B6.FVB(Cg)-Tg(Drd1-cre)EY262Gsat/Mmucd mice to introduce the D1-Cre. Ift88wt/null male mice hemizygous for D1-Cre were crossed with Ift88fl/fl female mice homozygous for R26R-EYFP to generate Ift88wt/fl::D1-Cre::R26R-EYFP (control) and Ift88null/fl::D1-Cre::R26R-EYFP (experimental) mice. Ift88 control and experimental mice were on a mixed background. Fixed mouse brains were generated and processed as previously described (Berbari et al., 2008a). For D1 protein quantification, striatum was dissected out and membrane-enriched proteins were isolated as previously described (Domire et al., 2011).
Immunofluorescence and histology.
Mouse monoclonal anti-D1 (SG2-D1a; Santa Cruz Biotechnology), which we previously validated with D1 knock-out tissue (Domire et al., 2011), was used at 1:250. Rabbit polyclonal anti-AC3 (C-20; Santa Cruz Biotechnology), which has been validated with AC3 knock-out tissue (Wong et al., 2000), was used at 1:350, and rabbit polyclonal anti-AC3 (EnCor Biotechnology) was used at 1:1000. Secondary antibodies included Alexa Fluor 488-conjugated donkey anti-mouse IgG and Alexa Fluor 546-conjugated donkey anti-rabbit IgG (Thermo Fisher Scientific). Nucleic acids were stained with DRAQ5 (Thermo Fisher Scientific). Immunofluorescence procedures have been previously described (Green et al., 2012). Samples were imaged on a laser-scanning confocal microscope (model TCS SP8, Leica) at the Hunt-Curtis Imaging Facility in the Department of Neuroscience at OSU. Multiple consecutive focal planes (z-stack), spaced at ∼0.3 µm intervals, were captured. Brain slices were stained with the Hematoxylin and Eosin Stain Kit (Vector Laboratories), according to the manufacturer instructions.
Weight and feeding assessment.
At weaning, mice of the same sex and genotype (Bbs1wt/fl::D1-Cre or Bbs1null/fl::D1-Cre, = Ift88wt/fl::D1-Cre::R26R-EYFP, or Ift88null/fl::D1-Cre::R26R-EYFP) were doubly housed in regular cages. An excess of normal mouse chow was weighed and placed in the wire tops of the cages. The remaining amount of food was weighed weekly, mice were moved into a fresh cage, and any uneaten pieces of chow visible in the bottom of the cage were combined with the uneaten food. The weight of the uneaten food was subtracted from the previous amount then divided by 14 to determine the average daily food intake for each mouse. Mice were weighed every other week on a balance.
Mouse behavioral analysis.
Activity of age- and weight-matched control and experimental mice was tracked using the Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments). After acclimatization to the cages, the activity of each mouse was monitored using infrared photocells located in the X and Z direction. Total activity was a sum of the total number of infrared beam breaks detected in both the X and Z direction over 24 h. The rotating rod (Rotarod) paradigm was performed by the OSU Rodent Behavior Core. Age- and weight-matched control and experimental mice were placed on an accelerating rod (4–40 rpm over the course of a 300 s trial). A Five Station Mouse Rota-Rod (Med Associates) was used in these experiments. Latency to fall was automatically recorded by the apparatus, and mice that did not fall were scored as 300 s. After an initial training day, experiments were performed on 3 consecutive days. Mice were subjected to three trials per day.
Preparation of striatal slices.
For biochemical analysis, ex vivo slices of the striatum were prepared from 8- to 12-week-old sex-matched control and experimental littermates. Mice were killed by decapitation, and whole brains were removed and immediately submerged in the following ice-cold artificial CSF (aCSF) solution for 3 min (in mm): 10 glucose, 26 NaHCO3, 1.5 CaCl2, 1.25 KH2PO4, 1.5 MgSO4, 4 KCl, and 123 NaCl, and oxygenated (95% O2, 5% CO2), at pH 7.4. Cerebellar tissue was then removed, and 350 µm coronal slices were cut from the tissue block containing the striatum using a vibratome (catalog #VT-1200s, Leica Microsystems) in oxygenated ice-cold aCSF. The striatum was dissected out of the slices using 20 gauge needles. Striatal slices were then immediately transferred to a chamber containing oxygenated aCSF supplemented with 10 µg/ml adenosine deaminase (AD; Roche) at 32°C and incubated for 30 min. After 30 min, the slices were transferred to a secondary chamber containing fresh oxygenated aCSF at 32°C without AD and incubated for an additional 30 min. For electrophysiological analysis, acute coronal brain slices (300 µm) containing the dorsolateral striatum were obtained. Briefly, mice were anesthetized with isoflurane and brains were rapidly extracted and placed into ice-cold cutting solution containing the following (in mm): 250 sucrose, 24 NaHCO3, 25 glucose, 2.5 KCl, 1.25 NaH2PO4, 1.5 MgSO4, 2 CaCl2, and 1 kynurenic acid, at pH 7.35. Slices were cut and transferred to a holding chamber containing aCSF as follows (in mM): 124 NaCl, 3 KCl, 2 CaCl2, 1 MgSO4, 24 NaHCO3, 1.25 NaH2PO4, and 10 glucose, at pH 7.35 and continuously bubbled with carbogen (95% O2 and 5% CO2). Slices were incubated at 32°C for 20 min and then moved to room temperature for at least 1 h before recording.
Drug preparation and ex vivo signaling on striatal slices.
The D1-selective full agonist A-77 636 hydrochloride (Conroy et al., 2015; Tocris Bioscience) was dissolved in methanol to a final concentration of 10 mm and stored at −20°C between uses. R(+)-SCH-23 390 hydrochloride (MilliporeSigma) was dissolved in water to a final concentration of 100 mm immediately before each experiment. Following the 1 h incubation, striatal slices were transferred to a chamber containing fresh oxygenated aCSF at 32°C with either vehicle (methanol) or 10 µm A-77636 and incubated for 5 or 10 min. For antagonist experiments, slices were pretreated in 100 µm SCH-23390 for 15 min before treatment with 10 µm A-77636 for 5 min. Slices were then collected in Microfuge Tubes, excess solution was removed, and slices were immediately frozen in liquid nitrogen and stored at −80°C. Proteins were isolated as previously described (Tanda et al., 2009). Briefly, the frozen slices were sonicated for 10 s in boiling 1% SDS and boiled for an additional 10 min. The tubes were then centrifuged at 16–18,000 × g for 10 min at room temperature to pellet debris. The supernatants were transferred to fresh tubes, and protein concentrations were determined by a Bradford assay (BIO-RAD).
Western blots.
Equal amounts of protein were loaded on a denaturing 4–15% gradient polyacrylamide gel (BIO-RAD). Proteins were transferred to a PVDF-FL membrane (BIO-RAD). Primary antibodies used were as follows: rabbit anti-phosphor-threonine-34 DARPP-32 (D27A4; Cell Signaling Technology) used at 1:1000; rabbit anti-DARPP-32 (19A3; Cell Signaling Technology) used at 1:2000; mouse anti-D1 used at 1:400; and mouse anti-β actin (8H10D10; Cell Signaling Technology) used at 1:20,000. Secondary antibodies used were HRP-linked goat anti-rabbit IgG (Cell Signaling Technology) used at 1:2000, and HRP-linked horse anti-mouse IgG (Cell Signaling Technology) used at 1:2000. Membranes were incubated in primary antibodies for 16–24 h at 4°C in TBS containing 0.2% Tween-20 and 5% BSA). Following washes in TBS containing 0.2% Tween-20, membranes were probed with secondary antibodies for 2 h at room temperature in TBS containing 0.2% Tween-20 and 5% BSA. Following washes, the membranes were incubated in SuperSignal West Pico PLUS chemiluminescent substrate (Thermo Fisher Scientific) and imaged on a ChemiDoc XRS+ Gel Imaging System (BIO-RAD). Following imaging for P-DARPP-32, membranes were incubated in Restore PLUS Western Blot Stripping Buffer (Thermo Fisher Scientific) for 15 min, and then probed for total DARPP-32. Western blots were quantified using Image Studio Lite (LI-COR). For each sample, the P-DARPP-32 signal was divided by the total DARPP-32 signal to give a relative P-DARPP-32 level.
Electrophysiology recording.
Individual slices were transferred into the recording chamber perfused with aCSF and continuously oxygenated with carbogen. Recordings were made at 32°C with patch electrodes (3.5–5 MΩ) pulled from thin-walled borosilicate glass (catalog #TW150-3, World Precision Instruments) using a P-97 puller (Sutter Instrument). The internal solution consisted of the following (in mm): 55 KCl, 75 K2SO4, 8 MgCl2, and 10 HEPES, at pH 7.27. Nystatin was prepared in DMSO at 1 mg/10 µl. Patch pipettes were front filled with the internal solution and then backfilled with the same solution containing nystatin (250 µg/ml). After attaining a giga-seal, perforated-patch recordings were initiated when the access resistance reached and stabilized at 30–40 MΩ. Enhanced yellow fluorescent protein (EYFP)-positive neurons within the dorsolateral striatum were visualized with infrared differential interference contrast optics equipped with an upright microscope (model BX51WI, Olympus). A MultiClamp 700B amplifier, Axon Digidata 1550B digitizer, and Clampex 11 acquisition software (Molecular Devices) were used to conduct recording. The signal was filtered at 1 kHz and digitized at 10 kHz. Cell and electrode capacitances, and serial resistance were monitored and compensated electronically as necessary. Recordings were obtained in current clamp, with the resting membrane potential determined by basal measurements with no current injection. To measure evoked action potentials (APs), a series of 1 s injected current pulses were delivered that increased in 30 pA increments. The threshold current to evoke the first action potential was determined when the current injection induced at least one action potential. This same current injection protocol was performed to evaluate the impact of D1 activation. A-77636 was dissolved in methanol to a concentration of 20 mm and diluted to a final concentration of 10 µm in aCSF. Note that the cells were exposed to vehicle (methanol) at the same concentration as during perfusion of A-77636 before the application of A-77636. The relationship between current injection and action potential number was determined in 5 min increments following the application of A-77636. Threshold data reported represent the point at which the minimal current injection evoked an action potential. The average time to maximal response was ∼12 min, but the bath perfusion rate was slow, and maximal responses were observed between 5 and 35 min in different recordings.
Neuron and cilia analysis.
Coronal brain sections from 5- to 6-week-old Ift88 control and experimental mice were labeled for AC3 and stained with DRAQ5. Three to six regions of the striatum were imaged from each animal and the number of EYFP-positive neurons, AC3-positive cilia, and nuclei were counted in each image.
Experimental design and statistical analyses.
Quantification of Western blots and cilia images was performed by a person blinded to the experimental conditions. For mouse behavioral and slice electrophysiological experiments, the person performing the experiments was blinded to the genotype of the animals, and the key was not broken until after the analysis was completed. Data were analyzed using Prism 9 software (GraphPad Software), and results were expressed as the mean ± SEM. Differences between two groups were examined using Student's unpaired t test, with the exception of relative P-DARPP-32 levels, which were examined using Student's paired t test. Average Rotarod performance across days and between genotypes was analyzed using a two-way ANOVA. A p value ≤ 0.05 was considered significant.
Results
Disruption of D1 ciliary export is associated with obesity
To test the physiological consequences of BBSome disruption in D1-expressing neurons, we crossed mice carrying a conditional Bbs1 allele (Carter et al., 2012) with mice expressing Cre recombinase under the control of the D1 promoter (D1-Cre; Gong et al., 2007). Loss of Bbs1 protein has been shown to disrupt BBSome formation and function (Guo et al., 2016), and Cre recombinase expression in D1-Cre mice matches the expression of endogenous D1 in the brain (Gong et al., 2007; Bateup et al., 2010). Control animals were heterozygous for a wt and floxed (flox) Bbs1 allele (Bbs1wt/flox) and positive for D1-Cre. Experimental animals were heterozygous for a deleted (null) and floxed Bbs1 allele (Bbs1null/flox) and positive for D1-Cre. Thus, in response to Cre expression, the BBSome in D1-expressing neurons would be unaffected in control animals (D1Bbs1+) because of the presence of the wild-type allele but would be disrupted in experimental animals (D1Bbs1–). Consistent with our previous results in constitutive BBS knock-out mice (Domire et al., 2011), we did not detect D1 ciliary localization in the brains of adult D1Bbs1+ mice (Fig. 1A–C). However, we observed abundant D1 ciliary localization in several regions of adult D1Bbs1– mice, including the striatum, amygdala, and olfactory tubercle (Fig. 1D–F). Interestingly, we consistently observed a reduction in AC3 signal in D1-positive cilia compared with D1-negative cilia. D1 ciliary enrichment may be either preventing AC3 ciliary localization or interfering with binding of the AC3 antibody. To test whether D1 ciliary accumulation was because of differences in D1 expression levels, we immunoblotted striatal lysates from D1Bbs1+ and D1Bbs1– brains. As with constitutive BBS knock-out mice (Domire et al., 2011), loss of BBSome function did not impact D1 protein levels in D1Bbs1– mice (t = 0.864, df = 10, p = 0.4078; Fig. 1G,H).
Disruption of Bbs1 in D1-expressing neurons causes D1 ciliary accumulation in vivo without impacting D1 expression or causing dilated ventricles. A–F, Representative images of the striatum in adult Bbs1wt/flox::D1-Cre (D1Bbs1+; A–C) and Bbs1null/flox::D1-Cre (D1Bbs1–; D–F) mice. Labeling for D1 (green) reveals the absence of D1-positive cilia in the D1Bbs1+ section (A) but abundant D1-positive cilia in the D1Bbs1– section (D). Labeling for AC3 (red) reveals a similar distribution of AC3-positive cilia between D1Bbs1+ (B) and D1Bbs1– (E) sections. Merged images confirm an absence of D1 ciliary labeling in the D1Bbs1+ (C) section and colocalization of D1 and AC3 in the D1Bbs1– (F) section. Note that the AC3 signal is noticeably reduced in D1-positive cilia. Nuclei are stained with DRAQ5 (blue). Scale bars, 10 µm. G, Representative images of a Western blot of membrane-enriched striatal lysates from D1Bbs1+ (+) and D1Bbs1– (–) mice labeled with antibodies against D1 (top blot) and β-actin (bottom blot). H, Quantification of D1 protein levels relative to β-actin protein in membrane-enriched striatal lysates from D1Bbs1+ and D1Bbs1– mice. Note that relative D1 protein levels in D1Bbs1+ and D1Bbs1– striatal lysates were not significantly different (ns; n = 5–7 animals/genotype). Values are expressed as the mean ± SEM. I, J, Representative images of brain slices from adult D1Bbs1+ (I) and D1Bbs1– (J) mice stained with hematoxylin and eosin show comparable ventricle volumes. K, L, Representative images of brain slices from 6-week-old constitutive Bbs1w/w (K) and Bbs1null/null (L) mice stained with hematoxylin and eosin show lateral ventricle dilation in the Bbs1null/null section.
Previous research has shown that knock-in mice homozygous for the human BBS1 mutation M390R develop neonatal hydrocephalus because of increased apoptosis and reduced proliferation of neural progenitors in the periventricular region (Carter et al., 2012). Histologic analysis of brains from adult D1Bbs1+ and D1Bbs1– mice showed that ventricles were not enlarged in the brains of D1Bbs1– mice (Fig. 1I,J). However, mice constitutively homozygous for a null Bbs1 allele (Bbs1null/null) showed obvious ventricle enlargement, compared with Bbs1w/w mice (Fig. 1K,L). These results indicate that the loss of Bbs1 throughout the brain causes hydrocephalus, whereas loss of Bbs1 specifically in D1-expressing neurons does not.
Interestingly, we noticed that adult D1Bbs1– mice appeared obese compared with their D1Bbs1+ littermates (Fig. 2A). Monitoring of mouse weights revealed that D1Bbs1+ and D1Bbs1– mice weighed the same at weaning but at 8 or 10 weeks of age D1Bbs1– female or male mice, respectively, weighed significantly more than D1Bbs1+ mice (t = 5.871, df = 20, p < 0.0001; and t = 3.541, df = 28, p = 0.0014, respectively; Fig. 2B,C). In constitutive BBS knock-out mouse models, the weight gain was also more pronounced in female mice (Nishimura et al., 2004; Fath et al., 2005; Eichers et al., 2006). The obesity phenotype in constitutive BBS knock-out mice is associated with an increase in food intake (Nishimura et al., 2004; Fath et al., 2005; Davis et al., 2007; Rahmouni et al., 2008; Zhang et al., 2013; Cognard et al., 2015). To determine whether D1Bbs1– mice are hyperphagic, we monitored the amount of food consumed each week by D1Bbs1+ and D1Bbs1– mice. We found that preobese D1Bbs1– female and male mice did not eat more than D1Bbs1+ mice (Fig. 2D,E), suggesting that the weight gain was not because of overeating. As constitutive BBS knock-out mice show decreased locomotor activity (Fath et al., 2005; Davis et al., 2007; Rahmouni et al., 2008) and reductions in activity levels without hyperphagia can lead to obesity in mice (Coyle et al., 2002; 2008; Zhang et al., 2014), we assayed basal locomotor activity levels in 6- to 8-week-old weight-matched D1Bbs1+ and D1Bbs1– mice. Interestingly, D1Bbs1– female and male mice showed significantly lower levels of basal locomotor activity, compared with D1Bbs1+ mice (t = 4.116, df = 11, p = 0.0017; and t = 3.875, df = 9, p = 0.0038, respectively; Fig. 2F,G). To test whether the lower levels of locomotor activity were because of motor deficits, we subjected preobese D1Bbs1+ and D1Bbs1– mice to the rotarod test. D1Bbs1– mice performed comparably to D1Bbs1+ mice in this paradigm (F(2,33) = 0.08261, p = 0.9209; Fig. 2H), suggesting that the reduction in activity was not because of a defect in gross motor skill. Overall, these results demonstrate that loss of BBSome function in D1-expressing neurons results in D1 ciliary accumulation that is associated with reduced basal locomotor activity and obesity.
Obesity in D1Bbs1– mice is associated with reduced locomotor activity without hyperphagia. A, Image of 12-week-old male D1Bbs1+ (left) and D1Bbs1– (right) mice demonstrating higher body mass in the D1Bbs1– mouse. B, C, Biweekly average weights of D1Bbs1+ and D1Bbs1– female (B) and male (C) mice beginning at 4 weeks of age and ending at 12 weeks of age. Note that, on average, D1Bbs1– female and male mice weighed significantly more than D1Bbs1+ female and male mice after 8 and 10 weeks of age, respectively. (n > 7 animals/genotype for each sex). D, E, Average daily food intake of D1Bbs1+ and D1Bbs1– female (D) and male (E) mice beginning at 4 weeks of age and ending at 12 weeks of age. Note that female D1Bbs1– mice had a significantly lower daily food intake, compared with female D1Bbs1+ mice, in weeks 4, 6, and 7. There was no significant difference in food intake between male D1Bbs1+ and D1Bbs1– mice at any age (n = 4 animals/genotype for each sex). F, G, Number of beam breaks recorded over a 24 h time period for 6- to 8-week-old weight-matched D1Bbs1+ and D1Bbs1– female (F) and male (G) mice. Note that, on average, D1Bbs1– mice had significantly lower numbers of beam breaks compared with D1Bbs1+ mice, which is indicative of a lower activity level (n > 5 animals/genotype for each sex). Values are expressed as the mean ± SEM. *p <0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. H, Results from Rotarod testing in 6- to 8-week-old weight-matched D1Bbs1+ and D1Bbs1– mice showing average latency to fall times on 3 consecutive testing days. Note that two-way ANOVA showed no significant interaction between day and genotype (F(2,33) = 0.08261, p = 0.9209; n = 6–7 animals/genotype; n = 5 males, n = 8 females).
Disruption of D1 ciliary export in striatal medium spiny neurons results in reduced D1 signaling
In the striatum, dopamine plays a major role in neuronal activity. Medium spiny neurons (MSNs) make up >90% of the neurons within the striatum and are evenly divided into the following two major classes: D1 expressing and D2 expressing. D1-expressing neurons are part of the “direct” pathway, which ultimately excites the thalamus and initiates behavior. In mice, the activation of D1-expressing MSNs promotes locomotion (Kravitz et al., 2010; Freeze et al., 2013), and reduced D1 signaling in the striatum is associated with lower locomotor activity (Bateup et al., 2010; Zhang et al., 2014). In the striatum, D1 couples to Gαs/olf to activate adenylyl cyclase and increase cAMP production (Zhuang et al., 2000). The increased cAMP levels stimulate activity of the cAMP-dependent protein kinase A (PKA) by binding to the regulatory (R) subunits, which triggers the release of active catalytic (C) subunits. PKA-C subunits then phosphorylate target proteins, such as dopamine- and cAMP-dependent phosphoprotein of 32 kDa (DARPP-32). DARPP-32 is a major downstream target of D1 activation, and D1 agonist treatment causes an increase in phosphorylation at Thr34 in DARPP-32 in the striatum (Nishi et al., 2011). To determine whether D1 signaling is altered in D1Bbs1– MSNs, we assayed phosphorylation at Thr34 in DARPP-32 (P-DARPP-32) in striatal slices from D1Bbs1+ and D1Bbs1– mice. Based on previous results showing that P-DARPP-32 levels peak in striatal slices ∼5–10 min after D1 agonist treatment (Tanda et al., 2009), slices were treated with vehicle or a D1 agonist for 5 and 10 min. Interestingly, the levels of P-DARPP-32 were significantly lower in vehicle-treated D1Bbs1– striatal slices, compared with vehicle-treated D1Bbs1+ slices (t = 4.399, df = 6, p = 0.0046; Fig. 3A,B), suggesting that basal levels of P-DARPP-32 may be lower in D1Bbs1– mice. Importantly, we observed significantly lower levels of P-DARPP-32 in response to 5 min of agonist treatment in D1Bbs1– striatal slices, compared with striatal slices from D1Bbs1+ littermates (t = 3.4, df = 6, p = 0.0145; Fig. 3A,B). The levels of P-DARPP-32 were similar between D1Bbs1+ and D1Bbs1– slices after 10 min of agonist treatment (t = 0.6773, df = 6, p = 0.5234; Fig. 3B). Pretreatment with a D1 antagonist prevented the agonist-mediated increase in P-DARPP-32 in D1Bbs1+ and D1Bbs1– slices (Fig. 3C), confirming that increases were the result of D1 activation. These results indicate that the loss of BBSome function in D1-expressing striatal neurons reduces D1-dependent signaling.
Disruption of Bbs1 leads to reduced D1-mediated DARPP-32 phosphorylation in striatal slices but does not affect intrinsic or D1-mediated neuronal excitability. A, Representative images of a Western blot showing lysates from D1Bbs1+ and D1Bbs1– striatal slices treated with vehicle (V) or D1 agonist (Ag) for 5 min and labeled with antibodies against Thr34 P-DARPP-32 (top blots) and total DARPP-32 (bottom blots). B, Quantification of P-DARPP-32 levels relative to DARPP-32 levels in striatal slices from D1Bbs1+ and D1Bbs1– mice treated with vehicle or D1 agonist. Note that phosphorylated DARPP-32 levels are significantly lower in D1Bbs1– striatal slices, compared with striatal slices from D1Bbs1+ littermates, after 5 min of vehicle or agonist treatment (n = 7 animals/genotype). Values are expressed as the mean ± SEM. *p < 0.05, **p < 0.01 C, Representative images of a Western blot showing lysates from D1Bbs1+ striatal slices treated with vehicle (V), D1 agonist (Ag), or D1 antagonist followed by agonist (Ant + Ag) for 5 min and labeled with antibodies against P-DARPP-32 (top blot) and total DARPP-32 (bottom blot). Note that the agonist-mediated increase in P-DARPP-32 level is blocked by pretreatment with the D1 antagonist. D, Membrane potential in D1-positive neurons as measured in current clamp under basal conditions (n = 72 D1Bbs1+ cells from 15 animals; n = 31 D1Bbs1– cells from 6 animals). E, Threshold current injection necessary to evoke an action potential (n = 72 D1Bbs1+ cells from 15 animals; n = 31 D1Bbs1– cells from 6 animals). F, Representative traces from perforated patch-clamp recordings of D1-positive neurons from D1Bbs1+ and D1Bbs1– slices. Responses are current injections to 210 pA (gray) and 240 pA (black) before (left) and after (right) A-77636 addition. G, Change in current necessary to induce an action potential following A-77636 addition. The value represents the difference between the current necessary to evoke action potentials before and after the application of A-77636 (n = 26 D1Bbs1+ cells from eight animals; n = 17 D1Bbs1– cells from four animals). H, Fold change in the number of action potentials induced with injection of current at the threshold level measured after application of A-77636 (n = 27 D1Bbs1+ cells from eight animals; n = 15 D1Bbs1– cells from four animals). Values are expressed as the mean ± SEM. ns, Not significantly different.
To test whether this reduction in D1-dependent DARPP-32 phosphorylation also affected D1-dependent changes in neuronal excitability, we used the whole-cell perforated patch-clamp technique to determine the impact of BBSome disruption on the neuronal response to D1 agonist treatment in ex vivo striatal slices. To identify D1-positive neurons, we introduced a mouse line carrying the EYFP gene downstream of a loxP-flanked stop sequence inserted into the Gt(ROSA)26Sor locus (R26R-EYFP; Srinivas et al., 2001). Consequently, EYFP is only expressed in cells expressing Cre recombinase, which is controlled by the D1 promoter. We found that there was no difference in the basal membrane potential or the current necessary to evoke an action potential between D1Bbs1+ and D1Bbs1– MSNs (Fig. 3D,E). Similarly, there was no difference between the number of action potentials induced at threshold between D1Bbs1+ and D1Bbs1– MSNs (from eight animals: 4.52 ± 1 0.70, n = 27; from four animals: 2.73 ± 0.61, n = 15; p = 0.095, t test). We determined that the application of D1 agonist resulted in a reduction in the current necessary to evoke action potentials (Fig. 3F). However, the impact of D1 agonist on threshold for action potential induction was not different between D1Bbs1+ and D1Bbs1– MSNs (Fig. 3G). The number of action potentials induced at the original threshold following A-77636 administration were also not different between D1Bbs1+ (from eight animals: 18.6 ± 2.84, n = 27) and D1Bbs1– MSNs (from four animals: 14.73 ± 2.90, n = 15; p = 0.38, t test). Finally, the fold change in AP number at the original threshold was not different between D1Bbs1+ and D1Bbs1– MSNs (Fig. 3H). Thus, the loss of BBSome function in D1-expressing striatal neurons does not affect the impact of D1 activation on neuronal excitability.
Disruption of cilia on D1-expressing neurons is associated with obesity
To further test the impact of cilia on D1 signaling, we disrupted cilia on D1-expressing neurons by crossing D1-Cre mice with a mouse line carrying a conditional allele of Ift88. Intraflagellar transport (IFT) is a ciliary transport process that is required for the formation and maintenance of cilia (Rosenbaum and Witman, 2002; Pedersen and Rosenbaum, 2008) and the disruption of Ift88 effectively ablates cilia on central neurons (Chizhikov et al., 2007; Davenport et al., 2007; Berbari et al., 2013; Koemeter-Cox et al., 2014). We also introduced the R26R-EYFP mouse line to identify D1-positive neurons. Control animals were heterozygous for a wt and floxed Ift88 allele (Ift88wt/flox) and positive for D1-Cre and EYFP. Experimental animals were heterozygous for a null and floxed Ift88 allele (Ift88null/flox) and positive for D1-Cre and EYFP. Confirming the loss of cilia on D1-expressing neurons is complicated by the fact that D1 ciliary localization is not detected unless BBSome function is disrupted, so labeling for D1 will not be informative. Moreover, D2-expressing neurons are also ciliated, and in slices it is difficult to determine whether a cilium originates from a particular cell body. Thus, to confirm the absence of cilia on D1-expressing MSNs, we labeled brain sections from 5- to 6-week-old D1Cilia+ and D1Cilia– mice with an antibody against AC3, which is enriched in neuronal cilia throughout the brain (Bishop et al., 2007), and quantified the number of cilia and EYFP-expressing neurons in the striatum (Fig. 4A–F). We found that the ratio of cilia to EYFP-expressing neurons was reduced by >50% in the striatum of D1Cilia– mice (Fig. 4G), indicating that the depletion of Ift88 in D1-expressing neurons effectively ablates cilia.
Disruption of Ift88 in D1-expressing neurons causes cilia loss without impacting the viability of neurons or D1 expression. A–F, Representative images of the striatum in 5- to 6-week-old Ift88wt/fl::D1-Cre::R26R-EYFP (D1Cilia+; A–C) and Ift88null/fl::D1-Cre::R26R-EYFP (D1Cilia–; D–F) mice. EYFP expression (green) reveals comparable numbers of D1-expressing neurons in D1Cilia+ (A) and D1Cilia– (D) sections. Labeling for AC3 (red) reveals a reduction in AC3-positive cilia in the D1Cilia– (E) section, compared with the D1Cilia+ (B) section. Nuclei are stained with DRAQ5 (blue). G, Quantification of AC3-postive cilia relative to EYFP-positive neurons in sections from D1Cilia+ (981 AC3-positive cilia/687 EYFP-positive neurons) and D1Cilia– (489 AC3-positive cilia/876 EYFP-positive neurons) sections indicates a reduction in the relative number of AC3-positive cilia by >50% in D1Cilia– sections (n = 5 animals/genotype). ****p < 0.0001. H, Quantification of EYFP-positive neurons relative to nuclei reveals a similar fraction between D1Cilia+ (687 EYFP-positive neurons/1316 nuclei) and D1Cilia– (876 EYFP-positive neurons/1668 nuclei) sections (n = 5 animals/genotype). Values are expressed as the mean ± SEM. ns, Not significantly different. I, Representative images of a Western blot of membrane-enriched striatal lysates from D1Cilia+ (+) and D1Cilia– (–) mice labeled with antibodies against D1 (top blot) and β-actin (bottom blot). J, Quantification of D1 protein levels relative to β-actin protein in membrane-enriched striatal lysates from D1Cilia+ and D1Cilia– mice. Note that relative D1 protein levels in D1Cilia+ and D1Cilia– striatal lysates were not significantly different (ns; n = 6–10 animals/genotype). Values are expressed as the mean ± SEM.
Recent reports have suggested that cilia provide a neuroprotective role in response to environmental insults (Bae et al., 2019; Ishii et al., 2021) and the disruption of Ift88 can be associated with neuronal loss (Bae et al., 2019; Bowie and Goetz, 2020). However, the ratio of EYFP-expressing neurons to nuclei was not reduced in the striatum of 5- to 6-week-old D1Cilia– mice (Fig. 4H), indicating that the loss of Ift88 does not impact survival of D1-expressing MSNs in younger adult mice. Moreover, immunoblotting striatal lysates from D1Cilia+ and D1Cilia– mice showed that the loss of cilia did not impact D1 protein levels in D1Cilia– mice (t = 0.2201, df = 14, p = 0.829; Fig. 4I,J). Similar to Bbs1 disruption, we discovered that by 8 weeks of age D1Cilia– female and male mice weighed significantly more than D1Cilia+ mice (t = 2.622, df = 23, p = 0.0152; and t = 2.77, df = 21, respectively; p = 0.0115; Fig. 5A-C). As opposed to Bbs1 disruption, however, there was no difference in weight gain between female and male D1Cilia– mice (Fig. 5B,C). Our results are consistent with a previous study that found knocking out Ift88 in D1-expressing cells is associated with weight gain without impacting D1 expression (Mustafa et al., 2019). D1Cilia– mice do not eat more than D1Cilia+ mice (Fig. 5D,E). However, 6- to 8-week-old weight-matched female and male D1Cilia– mice show significantly lower levels of basal locomotor activity compared with their D1Cilia+ littermates (t = 6.14, df = 8, p = 0.0003; and t = 4.978, df = 13; p = 0.0003, respectively; Fig. 5F,G). Our results differ from a recent study that found disrupting Ift88 in GAD2-expressing GABAergic medium spiny neurons in the striatum results in a decrease in body weight without a change in basal locomotor activity (Ramos et al., 2021). The reason for the difference with our results is likely because of the fact that cilia were disrupted on both D1- and D2-expressing neurons in the previous study. D1Cilia– mice performed comparably to D1Cilia+ mice in the rotarod test (F(2,45) = 1.303, p = 0.2818; Fig. 5H), suggesting that the reduction in activity was not because of gross motor deficits. Overall, these results demonstrate that the loss of cilia on D1-expressing neurons is associated with reduced basal locomotor activity and obesity.
Obesity in D1Cilia– mice is associated with reduced locomotor activity without hyperphagia. A, Image of 12-week-old male D1Cilia+ (left) and D1Cilia– (right) mice demonstrating higher body mass in the D1Cilia– mouse. B, C, Biweekly average weights of D1Cilia+ and D1Cilia– female (B) and male (C) mice beginning at 4 weeks of age and ending at 12 weeks of age. Note that female and male D1Cilia– mice weighed significantly more than female and male D1Cilia+ mice after 8 weeks of age (n = 11–13 animals/genotype for each sex). D, E, Average daily food intake of D1Cilia+ and D1Cilia– female (D) and male (E) mice beginning at 4 weeks of age and ending at 12 weeks of age. Note that male D1Cilia– mice had a significantly higher food intake compared with D1Cilia+ mice only at 11 weeks of age. There was no significant difference in food intake between female D1Cilia+ and D1Cilia– mice at any age (n = 10 animals/genotype for each sex). F, G, Number of beam breaks recorded over a 24 h time period for 6- to 8-week-old weight-matched D1Cilia+ and D1Cilia– female (F) and male (G) littermates. Note that, on average, D1Cilia– mice had significantly lower numbers of beam breaks compared with D1Cilia+ mice, which is indicative of a lower activity level (n > 5 animals/genotype for each sex). Values are expressed as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. H, Results from Rotarod testing in 6- to 8-week-old weight-matched D1Cilia+ and D1Cilia– mice showing average latency to fall times on 3 consecutive testing days. Note that two-way ANOVA showed no significant interaction between day and genotype (F(2,45) = 1.303, p = 0.2818; n = 9–10 animals/genotype; n = 7 males, 12 females).
Disruption of cilia on D1 striatal MSNs results in reduced D1 signaling
To determine whether D1 signaling is altered in D1Cilia– MSNs, we measured D1-dependent phosphorylation of DARPP-32 in striatal slices from D1Cilia+ and D1Cilia– mice. Levels of P-DARPP-32 were not significantly different in D1Cilia+ and D1Cilia– slices treated with vehicle or agonist for 5 min (Fig. 6B). However, we observed significantly lower levels of P-DARPP-32 after 10 min of agonist treatment in D1Cilia– striatal slices, compared with striatal slices from D1Cilia+ littermates (t = 2.632, df = 6, p = 0.0389; Fig. 6A,B). Pretreatment with a D1 antagonist prevented the agonist-mediated increase in P-DARPP-32 in D1Cilia+ and D1Cilia– slices (Fig. 6C), confirming that increases were the result of D1 activation. These results indicate that the disruption of cilia on D1-expressing striatal neurons reduces D1-dependent signaling.
Disruption of cilia leads to reduced D1-mediated DARPP-32 phosphorylation in striatal slices but does not affect intrinsic or D1-mediated neuronal excitability. A, Representative images of Western blots of lysates from D1Cilia+ and D1Cilia– striatal slices treated with vehicle (V) or D1 agonist (Ag) for 10 min and labeled with antibodies against P-DARPP-32 (top blots) and total DARPP-32 (bottom blots). B, Quantification of P-DARPP-32 levels relative to DARPP-32 levels in lysates from D1Cilia+ and D1Cilia– striatal slices treated with vehicle or D1 agonist. Note that P-DARPP-32 levels are significantly lower in D1Cilia– striatal slices, compared with striatal slices from D1Cilia+ littermates, after 10 min of agonist treatment (n = 7 animals/genotype). Values are expressed as the mean ± SEM. *p < 0.05. C, Representative images of Western blots of lysates from D1Cilia+ striatal slices treated with vehicle (V), D1 agonist (Ag), or D1 antagonist followed by agonist (Ant + Ag) for 5 min and labeled with antibodies against P-DARPP-32 (top blot) and total DARPP-32 (bottom blot). Note that the agonist-mediated increase in P-DARPP-32 level is blocked by pretreatment with the D1 antagonist. D, Membrane potential in D1-positive neurons as measured in current clamp under basal conditions (n = 54 D1Cilia+ cells from nine animals; n = 42 D1Cilia– cells from eight animals). E, Threshold current injection necessary to evoke an action potential (n = 54 D1Cilia+ cells from nine animals; n = 42 D1Cilia– cells from eight animals). F, Representative traces from perforated patch-clamp recordings of D1-positive neurons from D1Cilia+ and D1Cilia– slices. Responses are current injections to 180 pA (gray) and 210 pA (black) before (left) and after (right) A-77636 addition. G, Change in current necessary to induce an action potential following A-77636 addition. The value represents the difference between the current necessary to evoke action potentials before and after application of A-77636 (n = 29 D1Cilia+ cells from six animals; n = 27 D1Cilia– cells from six animals). H, Number of action potentials induced with injection of current at the threshold level measured after application of A-77636 (n = 29 D1Cilia+ cells from six animals; n = 25 D1Cilia– cells from six animals). Values are expressed as the mean ± SEM. ns, Not significantly different.
We also used a whole-cell perforated patch-clamp technique to determine the impact of cilia disruption on the neuronal response to D1 agonist treatment in ex vivo striatal slices. We found that there was no difference in the basal membrane potential or the current necessary to evoke an action potential between D1Cilia+ and D1Cilia– MSNs (Fig. 6D,E). We determined that the application of D1 agonist resulted in a reduction in the current necessary to evoke action potentials (Fig. 6F). However, the impact of D1 agonist on the threshold for action potential induction was not different between D1Cilia+ and D1Cilia– MSNs (Fig. 6G). Similarly, the number of action potentials induced at the original threshold was not different between D1Cilia+ and D1Cilia– MSNs (from six animals: 5.48 ± 0.75, n = 29 cells; from six animals: 6.64 ± 0.89, n = 25 cells, respectively; p = 0.3882, t test). The number of action potentials induced following A-77636 addition was also not different between D1Cilia+ and D1Cilia– MSNs (from six animals: 13.96 ± 2.15, n = 29 cells; from 6 animals: 12.96 ± 2.15, n = 25, respectively; p = 0.94, t test). It should be noted that the fold change in action potential number following A-77636 addition was less robust in the D1Cilia+ MSNs compared with the D1Bbs1+ MSN (3.85 ± 0.97 vs 8.71 ± 2.2, respectively; p = 0.033, t test). Yet, the fold change in action potential number evoked at the original threshold following A-77363 was not different between D1Cilia+ and D1Cilia– MSNs (Fig. 6H). Thus, the loss of cilia on D1-expressing striatal neurons does not affect the impact of D1 activation on neuronal excitability.
Discussion
In the brain, dopamine acts on neuronal circuitry through a relatively slow modulation of neurotransmission and impacts numerous critical physiological functions. Consequently, dopaminergic dysfunction is associated with multiple human disorders, including Parkinson's disease, schizophrenia, depression, and drug addiction (Beaulieu and Gainetdinov, 2011). To develop therapeutics to treat these disorders it is essential to understand the molecular mechanisms governing dopaminergic signaling. Our results indicate that neuronal cilia play a critical role in dopaminergic signaling. Specifically, disrupting either D1 ciliary export (D1Bbs1–) or D1 ciliary localization (D1Cilia–) in mice results in a reduction in D1 signaling, lower locomotor activity levels, and obesity. Although constitutive D1 KO mice exhibit increased locomotor activity (Xu et al., 1994), studies have shown that mice with lower levels of D1 expression or signaling exhibit reduced locomotor activity (Bateup et al., 2010; Zhang et al., 2014; Chiken et al., 2015) and obesity (Zhang et al., 2014). Although we cannot rule out the possibility that disruption of D1 ciliary localization may result in developmental changes that contribute to downstream behavioral changes, our results are consistent with previous findings and indicate that D1 ciliary localization is important for normal D1 signaling in neurons. This finding is somewhat at odds with the fact that little D1 is observed in neuronal cilia under normal conditions and suggests that transient ciliary localization facilitates D1 signaling. Yet, not all effects of activation of D1 were affected by the disruption of primary cilia function, as D1-dependent changes in neuronal excitability were unaffected. These results suggest that cilia play a specific and modulatory role in D1 signaling. Further, it is intriguing that both the accumulation of D1 in cilia or the loss of D1 ciliary localization led to a reduction in D1 signaling. There are several possible explanations for this apparently discordant finding.
One possibility is that D1-initiated cAMP signaling within the cilium is an important component of normal D1 activity and that this signaling is compromised with loss of cilia or BBSome function. This possibility is supported by the fact that D1 and AC3 colocalize within neuronal cilia. An important area of future study will be to determine the precise complement of the D1 ciliary signalosome. The abnormal accumulation of D1 observed in cilia on D1Bbs1– neurons may simply crowd out other proteins and limit the availability of ciliary signaling partners. Indeed, we consistently see a reduction in AC3 signal in D1-enriched cilia. As constitutive AC3 KO mice exhibit adult-onset obesity associated with lower locomotor activity and hyperphagia (Wang et al., 2009), it is tempting to speculate that D1 enrichment could impact normal ciliary cAMP signaling by preventing AC3 ciliary localization. This has also been observed with 5-HT6 expression, which can reduce or prevent ciliary localization of AC3 and the ciliary GTPase Arl13b (Guadiana et al., 2013; Hu et al., 2017). Moreover, ciliary localization could differentially impact the downstream consequences of D1 activation. A recent study showed that ciliary cAMP, but not cytoplasmic cAMP, inhibits Hedgehog signaling, indicating that different subcellular pools of cAMP can convey different information (Truong et al., 2021). Thus, D1-mediated cAMP signaling in the cilium may be quantitatively or qualitatively different then D1 signaling at the plasma membrane.
Another possibility is that normal D1 signaling requires endocytosis of D1 and its effectors from cilia and signaling from endosomes. This endosome-based signaling would be lost if D1 ciliary export is disrupted or D1 is unable to localize to cilia. D1 is known to facilitate cAMP signaling from early endosomes after internalization in striatal slices (Kotowski et al., 2011). As opposed to GPCR signaling on the plasma membrane, endosomal-based cAMP signaling encodes discrete spatial information and is required for efficient regulation of downstream transcription by regulating PKA-C nuclear entry (Tsvetanova and von Zastrow, 2014; Peng et al., 2021). Endosomes derived from cilia, which are located on the soma of the neuron, would provide additional spatial constraints and could theoretically increase the efficiency of transmitting signals to the nucleus. It is also possible that D1 does not generate a signal in the ciliary compartment per se, but D1 trafficking into and out of the cilium is an important step in establishing the signaling potential of D1. Conceivably, D1 trafficking through the cilium could provide some modification or facilitate an association with another protein that is critical for proper D1 signaling. Future studies using live-cell imaging and genetically encoded signaling reporters should reveal whether D1 signals on the ciliary membrane and how this signaling is altered in D1Bbs1– neurons.
Finally, the observed reduction in signaling in D1Bbs1– neurons may be because of a requirement for the BBSome to traffic proteins that comprise the D1 ciliary transduction pathway. For example, we have observed ciliary colocalization of D1 and Mchr1 in a subset of cultured mouse striatal neurons (Green et al., 2012). Serotonin receptor 6 (Htr6) also localizes to ∼80% of cilia on cultured mouse striatal neurons (Brodsky et al., 2017), raising the possibility of Htr6 signaling in cilia on D1-expressing neurons. Ciliary signaling from either of these receptors may be required for normal D1 signaling through long-term modulation of D1-dependent signaling cascades. Alternatively, ciliary GPCRs can heteromerize (Green et al., 2012) and potentially generate a unique signal. D1 can form heteromers with a variety of GPCRs and channels (Beaulieu and Gainetdinov, 2011), which has the potential to modify receptor properties, such as ligand affinity, cellular trafficking, and signaling. Trafficking of D1 and signaling partners to and from the cilium could regulate such interactions. Cilia on D1Bbs1– neurons would be expected to lack this signal because of the lack of Mchr1 (or possibly Htr6) ciliary localization. Indeed, it is possible that the disruption of ciliary localization of receptors other than D1 contributes to the observed phenotypes. Further work into the role of these receptors within the striatum will lend insight into this possibility.
Our results also provide insight into the role of BBS proteins in body weight regulation. Constitutive BBS knock-out mice or mice lacking Bbs1 throughout the brain are hyperphagic (Nishimura et al., 2004; Fath et al., 2005; Davis et al., 2007; Rahmouni et al., 2008; Zhang et al., 2013; Cognard et al., 2015), suggesting that the BBS proteins are required for the regulation of appetite. Yet, constitutive BBS knock-out mice that are fed the same amount of food as their wild-type littermates still have increased adiposity, indicating that low energy expenditure contributes to obesity in BBS animals (Rahmouni et al., 2008). Indeed, constitutive BBS knock-out mice also show decreased locomotor activity (Fath et al., 2005; Davis et al., 2007; Rahmouni et al., 2008). Our results suggest that a reduction in D1-mediated signaling in the striatum contributes to the reduced locomotor activity in constitutive BBS knock-out mice. In BBS patients, obesity is a consistent phenotype and an important cause of morbidity (Grace et al., 2003). Yet, the etiology is unclear. Although anecdotal reports suggest that the appetites of BBS patients are difficult to satiate (Tobin and Beales, 2007), a study comparing BBS patients with body mass index-matched control subjects reported similar basal metabolic rates and energy intake, but BBS patients had lower levels of physical activity (Grace et al., 2003). Our findings suggest that dopaminergic regulation of movement may contribute to the development of obesity in BBS patients.
We did observe some phenotypic differences between D1Bbs1– and D1Cilia– mice. Specifically, the onset of weight gain was more prominent in female D1Bbs1– mice compared with male D1Bbs1– mice but was equivalent in female and male D1Cilia– mice. Additionally, levels of phosphorylated DARPP-32 were significantly lower in D1Bbs1– striatal slices, compared with D1Bbs1+ striatal slices, at 5 min of vehicle or agonist treatment, whereas levels of phosphorylated DARPP-32 were significantly lower in D1Cilia– striatal slices, compared with D1Cilia+ striatal slices, at 10 min of agonist treatment. These differences may reflect differential impacts on D1 signaling when D1 ciliary export is disrupted versus the loss of D1 ciliary localization in D1Bbs1– and D1Cilia– mice, respectively. Alternatively, given that Bbs1 mice are on a congenic background and the Ift88 mice are on a mixed background, these differences may be because of genetic variability. Indeed, the fact that the fold change in action potential number following A-77636 addition was less robust in D1Cilia+ MSNs compared with D1Bbs1+ MSNs may also reflect strain differences.
In summary, we have shown that disruption of either export of D1 from cilia or localization of D1 to cilia causes a reduction in D1-mediated signaling and obesity that is associated with lower locomotor activity, but not increased food intake, in mice. Thus, cilia play an important, hitherto unappreciated, role in dopaminergic signaling in the brain.
Footnotes
This work was supported by National Institutes of Health (NIH)/National Institute of Mental Health Research Project Grants R21-MH-107021 and MH-121744 (to K.M.), and NIH/National Institute of Neurological Disorders and Stroke Center Core Grant P30-NS045758. We thank Sarah Kelley, Josh Omlor, and Ivan Santiago for contributions to this project.
The authors declare no competing financial interests.
- Correspondence should be addressed to Kirk Mykytyn at mykytyn.1{at}osu.edu or Candice C. Askwith at askwith.1{at}osu.edu
