GPR45 modulates Gα_s at primary cilia of the paraventricular hypothalamus to control food intake

Abstract

The melanocortin system centrally regulates energy homeostasis, with key components like melanocortin-4 receptor (MC4R) and adenylyl cyclase 3 (ADCY3) in neuronal primary cilia. Mutations in MC4R and ADCY3 as well as ciliary dysfunction lead to obesity, but how melanocortin signaling works in cilia remains unclear. Through mouse random germline mutagenesis, we identified two missense mutations in Gpr45 that lead to obesity owing to hyperphagia. GPR45 was expressed in paraventricular nucleus of the hypothalamus (PVH), where it localized to cilia and recruited Gα_s to increase ciliary cAMP via ADCY3. GPR45 colocalized with MC4R in PVH cilia and promoted ciliary MC4R activation. Loss of GPR45 in PVH/MC4R+ neurons caused obesity. These findings establish GPR45 as a key regulator of the ciliary melanocortin system, bridging MC4R and ADCY3.

Graphical Abstract

GPR45 transports Gα_s into primary cilia to regulate food intake

In wild type mice, GPR45 relies on TULP3 to transport Gα_s into the primary cilia of PVH neurons. Within the cilia, Gα_s supports MC4R and possibly other Gα_s-coupled GPCRs to activate ADCY3, leading the production of cAMP from ATP, which signals to suppress appetite. In Gpr45 mutant mice, whether due to missense mutations that prevent its ciliary localization or complete knockout, this process is disrupted. As a result, ciliary signaling is impaired, leading to increased food intake and obesity.

Summary

Introduction

The discovery of obese mice and the positional cloning of the obese gene leptin have significantly increased our understanding of the endocrine control of energy balance. Secreted by adipose tissue, leptin primarily acts on its receptor in the arcuate nucleus of the hypothalamus (ARH) to transduce signals that either promote or suppress appetite to melanocortin-4 receptor (MC4R) neurons in the paraventricular nucleus of the hypothalamus (PVH). This leptin-melanocortin signaling pathway centrally regulates energy homeostasis while its dysfunction causes obesity in both mice and humans.

Rationale

To accelerate the discovery of genes involved in energy balance, we performed a forward genetic screen using random mutagenesis in mice, followed by automated meiotic mapping. This approach identified multiple genes associated with primary cilia – small, hair-like structures extending from the surface of most neurons. Genetic disorders that disrupt ciliary function frequently cause obesity. Both MC4R and its downstream signaling mediator adenylyl cyclase 3 (ADCY3) are localized to neuronal cilia and their mislocalization from cilia leads to obesity. However, the mechanisms by which melanocortin signaling functions within cilia and how ciliary dysfunction contributes to obesity are not well understood.

Results

Our genetic screen identified two obesity-associated alleles, expansive and extensive, which result from distinct missense mutations in an orphan G protein-coupled receptor (GPCR), Gpr45. Knockout of Gpr45 in mice led to obesity, confirming that both mutations cause a loss of function. Mice lacking Gpr45 consumed more food, and pair-feeding them the same amount as control mice rescued the obesity phenotype. Gpr45 mRNA was highly expressed in the PVH, and its loss in PVH or MC4R-expressing neurons led to obesity. We found that the GPR45 protein was exclusively localized to primary cilia in both cultured cells and PVH neurons, with its ciliary localization mediated by TUB-like protein 3 (TULP3), a key adaptor for ciliary trafficking. Overexpression of GPR45 caused the accumulation of the stimulatory G protein subunit Gα_s in cilia, where it is normally present at low levels. In contrast, Gpr45 knockout mice exhibited reduced ciliary Gα_s levels in the PVH. Furthermore, the Gα_s transported by GPR45 was active in stimulating ADCY3, thereby increasing localized cyclic AMP (cAMP) levels in the cilia, which is distinct from the cytoplasmic cAMP pool. Most MC4R cilia in the PVH also showed GPR45 presence while co-expression of GPR45 with MC4R enhanced Gα_s translocation and ciliary cAMP production. GPR45 and ADCY3 are likely working in the same signaling pathway to regulate feeding since loss of Gpr45 did not further increase obesity in Adcy3-deficient mice. Moreover, both expansive and extensive mutations disrupted the ciliary localization of GPR45 and impaired its ability to transport Gα_s and increase ciliary cAMP levels.

Conclusion

This study identifies GPR45 as a critical ciliary GPCR that regulates food intake by modulating ciliary Gα_s signaling in the PVH. Differing from typical GPCRs, GPR45 transports Gα_s into cilia to be utilized by other Gα_s-coupled GPCRs, such as MC4R. GPR45 functions as a gatekeeper for ciliary Gα_s to sustain a localized cAMP pool essential for melanocortin signaling. Dysregulated cilia or ciliary targeting of key leptin-melanocortin members disrupt such GPR45-maintained ciliary cAMP pool and thus cause obesity. Given that GPCRs are highly druggable, these findings have potential therapeutic implications for developing anti-obesity medications.

The rising prevalence of obesity has become a global public health problem (1). Obesity develops through a dysregulated energy intake and energy expenditure, which is primarily regulated by the central nervous system (CNS) (2). Recent genomic and molecular studies have identified the hypothalamus as a central hub governing energy metabolism (3, 4). Different nuclei of the hypothalamus integrate circulating and neuronal signals to assess the current energy state and coordinate with other brain regions to control feeding behavior and energy usage. In particular, the melanocortin system plays a critical role in the central regulation of food intake and energy balance (5). This system includes pro-opiomelanocortin (POMC)-expressing neurons that suppress appetite and agouti-related protein (AgRP)/neuropeptide Y (NPY)-expressing neurons that stimulate appetite in the arcuate nucleus of the hypothalamus (ARH), and MC4R-expressing neurons as receptors in the PVH and other brain regions (6). Much of our understanding of energy homeostasis regulation stems from studying mouse mutants. Since the discovery of obese (ob) mice (7) and the positional cloning of the ob gene leptin (8), we have gained numerous insights into the endocrine control of energy balance (9). Despite this, random mouse mutations relevant to energy homeostasis are limited.

Most cell types within the CNS are equipped with primary cilia, antenna-like sensory organelles protruding from the cell surface. Over the years, mounting evidence has highlighted the critical role of hypothalamic cilia in regulating energy metabolism in mammals (10, 11). Disruption of cilia in adult mouse models leads to hyperphagia-induced obesity (12), while mutations affecting cilia function result in ciliopathies often characterized by severe obesity (13). A detailed localization screen of GPCRs associated with energy homeostasis in the hypothalamus identified seven ciliary GPCRs, including neuropeptide Y receptor Y2 (NPY2R) and Y5 (NPY5R) (14). Proper ciliary localization of these NPY receptors is essential for maintaining energy balance in mice (14). Notably, MC4R also localizes to primary cilia with the assistance of melanocortin receptor-associated protein 2 (MRAP2), and disruption of its ciliary localization is associated with obesity in both mice and humans (15-18). As a GPCR, MC4R is known to couple with the guanine nucleotide-binding proteins alpha subunit stimulating (Gα_s) to activate adenylyl cyclase (ADCY) for cAMP release (19). ADCY3 also localizes to primary cilia throughout the brain (20) and has been implicated in obesity in mice (21) and humans (17, 22, 23). However, it remains unclear whether and how Gα_s functions simultaneously in the cytoplasm and/or primary cilia in the regulation of cAMP production with relation to central obesity phenotypes.

Here, we performed a large-scale forward genetic screen to identify an orphan GPCR, GPR45, as a key regulator of the melanocortin system in the PVH/MC4R neurons and uncover a non-canonical role of GPR45 in transporting Gα_s into primary cilia to maintain a ciliary cAMP pool.

Forward genetic screen reveals two obesity alleles caused by missense mutations in Gpr45

To uncover mechanisms underlying obesity, we employed a forward genetic screening platform coupled with automated meiotic mapping (24) to identify gene mutations influencing body weight in mice. First-generation (G1) mice mutagenized with N-ethyl-N-nitrosourea (ENU) were bred for two generations, producing pedigrees with an average of 40 third-generation (G3) germline mutant mice, each carrying approximately 45 mutations inherited from the same pedigree. 160,968 G3 mice harboring 253,464 coding/splicing mutations were weighed at least once during their lifetimes, achieving an estimated 60.29% autosomal saturation – the fraction of autosomal genes with protein products damaged or destroyed and studied for phenotypic effects twice or more in the homozygous state.

Among these mutations, a phenotype termed expansive, characterized by increased body weight on a normal chow diet, was identified in pedigree R5319 (Fig. 1A). The expansive phenotype was mapped as a quantitative trait, with body weight scaled by age and sex in 18 G3 mice (Table S1). Automated meiotic mapping implicated a missense allele of Gpr45 as the causative mutation, displaying the strongest linkage in a recessive inheritance model (p=6.777x10⁻¹⁶) among all 63 mutations present in the G1 founder (Fig. 1B and Table S2). The expansive mutation was a single-nucleotide transition from T to C, resulting in the substitution of serine for proline at position 214 (S214P) in the GPR45 protein. Subsequently, a second Gpr45 allele, termed extensive, was identified in pedigree R5667 (Fig. 1C). The extensive phenotype, also characterized by increased body weight, was mapped as a quantitative trait using scaled body weight data from 53 G3 mice (Table S3). Similar to expansive, the extensive phenotype exhibited a recessive inheritance pattern (p=2.602x10⁻⁷) among 52 mutations identified in the G1 founder (Fig. 1D and Table S4). The extensive mutation involved a single-nucleotide transition from A to G, leading to the substitution of tyrosine for cysteine at position 287 (Y287C) in the GPR45 protein.

Fig. 1. Identification of expansive and extensive obesity alleles and mapping of the causative gene as Gpr45.

(A-B) Scaled body weight data (A) and Manhattan plot (B) of the expansive phenotype. (C-D) Scaled body weight data (C) and Manhattan plot (D) of the extensive phenotype. Mean (μ) and SD (σ) are indicated. REF, homozygous for the reference allele; HET, heterozygous for the reference allele and for the expansive/extensive allele; VAR, homozygous for the expansive/extensive allele. The −Log10 (p value) (y axis) are plotted versus the chromosomal positions of mutations. Horizontal dark red and pink lines represent thresholds of p = 0.05 with or without Bonferroni correction, respectively. (E-G) Body weight (E), fat weight (F), and lean weight (G) of male Gpr45^−/− mice (n = 6) and +/+ mice (n = 6) from 5 to 8 weeks of age. (H) Body weight of 12-week-old male mice. (I) Representative photograph of male +/+ and Gpr45^−/− mice at 12 weeks of age. (J) Representative photographs of eWAT, iWAT, iBAT, and liver from 12-week-old male mice. (K) H&E staining of sections from different adipose tissues of 12-week-old male mice. Scale bars, 100 μm. (L and M) Blood glucose (L) and insulin (M) of 12-week-old male mice after a 6-h fast. (N) Glucose tolerance test. Blood glucose was measured at indicated times after i.p. glucose injection in 12-week-old male mice (n = 6). (O) Insulin tolerance test. Blood glucose was measured at indicated times after i.p. insulin injection in 12-week-old male mice (n = 6). (P and Q) Blood cholesterol (P) and triglycerides (Q) of 12-week-old male mice after a 6-h fast. (R and S) Liver weight (R) and liver triglycerides (S) of 12-week-old male mice. (T-U) H&E staining (T) and Oil Red O staining (U) of liver sections from 12-week-old male mice. Scale bars, 100 μm. Data are presented as means ± SD. p values were determined by Student’s t test (E–H, L-S). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001; ns, not significant with p > 0.05. Data points represent individual mice (A, C, H, L-M, P-S). Data are representative of two independent experiments (E-U) or one experiment (A-D).

By CRISPR/Cas9 gene targeting, a null allele of Gpr45 was made. Both male and female Gpr45 knockout (Gpr45^−/−) mice gained more weight from 6 weeks of age, and the obesity was fully developed at 12 weeks of age (Fig. 1, E-I and S1, A-D). Necropsy revealed that adipose tissues of 12-week-old Gpr45^−/− mice were increased in size relative to those in wild-type (WT, +/+) littermates, including epididymal white adipose tissue (eWAT), inguinal white adipose tissue (iWAT), and interscapular brown adipose tissue (iBAT) (Fig. 1J). Hematoxylin and eosin (H&E) staining revealed enlarged adipocytes, indicative of hypertrophy (Fig. 1K). At 12 weeks of age, fasting blood glucose levels were increased in Gpr45^−/− mice, while fasting insulin levels were comparable between Gpr45^−/− and WT mice (Fig. 1, L-M and S1, E-F). Glucose intolerance (Fig. 1N and S1G) and insulin resistance (Fig. 1O and S1H) were observed in 12-week-old Gpr45^−/− mice. Gpr45^−/− mice had increased cholesterol in the serum with no change in blood triglycerides (Fig. 1, P-Q and S1, I-J). In addition, 12-week-old Gpr45^−/− mice had large, pallid livers (Fig. 1J) with increased liver weight (Fig. 1R and S1K) and liver triglycerides (Fig. 1S and S1L). H&E staining (Fig. 1T) and Oil Red O (ORO) staining (Fig. 1U) showed an abundance of stored lipids in the liver. These data confirm that Gpr45 mutations in expansive and extensive strains were causative of the observed obesity phenotype and rule out gain-of-function as a mechanism.

The obesity phenotype of Gpr45^−/− mice is caused by hyperphagia

Metabolic cage experiments were conducted to measure the energy input and output of 5-week-old Gpr45^−/− mice and WT littermates, at the point when the obesity phenotype of Gpr45^−/− was just beginning to manifest (Fig. 2, A-C). Significantly increased food intake was observed in Gpr45^−/− mice compared with WT littermates (Fig. 2, D and E). In line with the hyperphagia phenotype, Gpr45^−/− mice exhibited an elevated respiratory exchange rate (RER), suggesting a consistent preference for carbohydrates as a fuel source, even during light cycles (Fig. 2F). Indirect calorimetry analyses found no difference in heat production and physical activities between Gpr45^−/− and WT mice (Fig. 2, G-J). The normal housing temperature (~23°C) is below the thermoneutral zone (~30 °C) of mice, which might limit the use of mouse models to study human obesity (25). Thus, Gpr45^−/− and WT mice were housed at either 23°C or 30°C for 8 weeks starting at 4 weeks of age to monitor the development of obesity phenotypes. Housing at thermoneutrality did not blunt, but rather further aggravated, the extent of obesity in Gpr45^−/− mice compared to those housed at room temperature (Fig. S2, A-D). Different from previously reported (26), Gpr45^−/− mice maintained their core body temperature very well during acute cold stress in the absence of food throughout the experiment (Fig. 2K), with a normal expression of POMC mRNA and protein in the hypothalamus (Fig. S2, E-G). Thus, loss of GPR45 mainly causes increased food intake along with the development of obesity.

Fig. 2. The obesity phenotype of Gpr45^−/− mice is caused by increased food intake.

(A-C) Body weight (A), fat weight (B), and lean weight (C) of 5-week-old male mice used for metabolic cage experiments in (D)-(J). (D) Average daily food intake measured by metabolic cages of 5-week-old male +/+ mice and Gpr45^−/− mice housed at 23 °C. (E-J) Metabolic cage measurements of cumulative food intake (E), respiratory exchange ratio (F), energy expenditure (G), fine movement (H), ambulatory movement (I), and rearing movement (J) of 5-week-old male Gpr45^−/− mice (n = 6) and +/+ mice (n = 6) housed at 23 °C. One Gpr45^−/− mouse was excluded from (H) and (I) due to aberrant values; final n = 5. Dark color indicates dark phase when light was off. (K) Internal temperature of 8-week-old male mice housed at 6 °C in the absence of food (n = 5 +/+, 6 Gpr45^−/−). (L-N) Daily food intake (L), cumulative food intake (M), and body weight (N) of male mice recorded daily from 5 weeks of age to 8 weeks of age (n = 9 +/+, 6 Gpr45^−/−). (O) Body weight of pair-fed male mice recorded daily from 5 weeks of age to 8 weeks of age (n = 5 +/+, 5 Gpr45^−/−). (P) Representative photograph of male +/+ and Gpr45^−/− mice pair-fed for 3 weeks since 5 weeks of age. (Q) Representative photographs of eWAT, iWAT, iBAT, and liver from male +/+ and Gpr45^−/− mice pair-fed 3 weeks since 5 weeks of age. (R) Blood glucose of male +/+ and Gpr45^−/− mice pair-fed 3 weeks since 5 weeks of age after a 6-h fast. Data are presented as means ± SD. p values were determined by Student’s t test (A-D, K-O, R). A linear correlation with a two-tailed comparison of slope and intercept was calculated and compared between different mouse groups (G). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001; ns, not significant with p > 0.05. Data points represent individual mice (A-D, G, R). Data are representative of two independent experiments.

To further confirm the hyperphagia phenotype, Gpr45^−/− mice and WT littermates were singly housed to monitor food consumption and body weight daily from 5 weeks of age. The daily food intake of Gpr45^−/− mice fluctuated, with a consistent increase observed from 40 days of age, and the cumulative food intake of Gpr45^−/− mice was significantly increased from 37 days of age, when body weight was similar between Gpr45^−/− mice and WT littermates at these times (Fig. 2, L-N). After the onset of hyperphagia, body weight began to increase at 42 days of age (Fig. 2N). To check if hyperphagia causes the obesity phenotype, Gpr45^−/− mice were pair-fed with the same amount of food as WT littermates from 5 weeks of age. There was no difference in body weight between Gpr45^−/− mice and control mice after 3 weeks of pair feeding (Fig. 2O). Compared with ad libitum-fed mice, pair feeding rescued both the obesity phenotype and diabetes/fatty liver phenotypes in Gpr45^−/− mice (Fig. 2, P-R and S2H). Thus, hyperphagia directly contributes to the development of obesity in GPR45-deficient mice.

GPR45 is expressed and functions within the paraventricular hypothalamus to regulate body weight

Gpr45 mRNA was highly expressed in the brain and was barely detectable by real-time PCR in all other major mouse tissues (Fig. 3A). We tested all commercially available anti-GPR45 antibodies, and none could detect endogenous levels of GPR45 protein in the brain. To better check the expression pattern of the GPR45 protein, we generated a mouse with a GFP tag knock-in at the C-terminus of GPR45 (Gpr45-GFP) with CRISPR/Cas9 technology (27) (Fig. 3B). Individual founder lines with the correct GFP sequence inserted at the desired site were bred to homozygosity after two generations. The insertion of a GFP tag at the C-terminus of the GPR45 protein did not disrupt its function, as evidenced by the absence of an obesity phenotype in homozygous Gpr45-GFP mice (Fig. S3, A-C). Furthermore, real-time PCR analysis using a primer set targeting Gfp mRNA demonstrated an expression pattern consistent with that of Gpr45 mRNA across various mouse tissues, which further validated the Gpr45-GFP knock-in mouse model (Fig. S3D). Compared with WT control mice, three independent Gpr45-GFP lines showed a very specific expression of GPR45 protein in whole brain lysates (Fig. 3C). The brain of the Gpr45-GFP mouse was further dissected into different regions. The GPR45 protein was expressed highest in the hypothalamus with much lower expression in the brain stem, cortex, olfactory bulb, hippocampus, and cerebellum (Fig. 3D), with both Gpr45 mRNA and Gfp mRNA showing a similar expression pattern (Fig. S3E). Consistent with the Gpr45 mRNA profile, no GPR45 protein was detected outside the brain (Fig. S3F). We focused our study on GPR45 in the hypothalamus owing to the high GPR45 protein expression and the importance of this region in regulating energy metabolism. Within the hypothalamus, in situ hybridization (RNAscope) experiments detected Gpr45 mRNA in the PVH (Fig. 3E), with no clear enrichment of Gpr45 mRNA in Pomc expressing neurons of the ARH (Fig. 3F).

Fig. 3. GPR45 is highly expressed in the PVH neurons to regulate energy homeostasis.

(A) Relative Gpr45 mRNA level in different mouse tissues normalized by Polr2a (n = 3, 8-week-old C57BL/6J male mice). (B) Generation of GFP tagged Gpr45 knock-in mice by CRISPR. (C) Immunoblot analysis of Gpr45-GFP protein expression in +/+ and homozygous Gpr45-GFP knock-in mice (n = 3, 8-week-old male, # indicates different lines). (D) Immunoblot analysis of Gpr45-GFP protein expression in different brain regions from 8-week-old male homozygous Gpr45-GFP knock-in mice. (E-F) Representative RNAscope image of the PVH with Gpr45 probe in green (E) and the ARH with Gpr45 probe in green and Pomc probe in red (F). DAPI was used to visualize nuclei in blue. Scale bar, 50 μm. (G) Generation of Gpr45-flox mice by CRISPR. (H) Illustration of different Cres used to delete Gpr45 in PVH and ARH. (I-K) Body weight (I), fat weight (J), and lean weight (K) of 10-week-old Gpr45^flox/flox mice and Gpr45^flox/flox; Pomc-Cre mice fed HFD for 4 weeks. (L-N) Body weight (L), fat weight (M), and lean weight (N) of 10-week-old Gpr45^flox/flox mice and Gpr45^flox/flox; Agrp-Cre mice fed HFD for 4 weeks. (O-Q) Body weight (O), fat weight (P), and lean weight (Q) of 10-week-old Gpr45^flox/flox mice and Gpr45^flox/flox; Sim1-Cre mice fed HFD for 4 weeks. (R-T) Body weight (R), fat weight (S), and lean weight (T) of 10-week-old Gpr45^flox/flox mice and Gpr45^flox/flox; Mc4r-Cre mice fed HFD for 4 weeks. (U) Illustration of stereotaxic injection of AAV in the PVH of 8-week-old Gpr45^flox/flox mice. AP, anterior-posterior; ML, medial-lateral; DV, dorsal-ventral. (V) Body weight change of Gpr45^flox/flox mice after AAV injections. (W-Y) Body weight (W), fat weight (X), and lean weight (Y) of Gpr45^flox/flox mice at 7 weeks after AAV injections. Data are presented as means ± SD. p values were determined by Student’s t test (I-T). *p ≤ 0.05, ***p ≤ 0.001, and ****p ≤ 0.0001; ns, not significant with p > 0.05. Data points represent individual mice (I-T, W-Y). Data are representative of two independent experiments (A, C-D, I-T) or one experiment (E-F, V-Y).

To investigate the loss of GPR45 function in specific hypothalamic regions and neurons, we generated Gpr45-flox mice using a CRISPR-Cas9 approach, as previously described (28) (Fig. 3G). Founders that contained the correct insertion in both left and right LoxP sites were bred into homozygosity to establish lines. Gpr45-flox mice were crossed with various Cre lines targeting distinct hypothalamic neurons/regions (Fig. 3H): Pomc-Cre was used to knockout Gpr45 in Pomc-expressing neurons in ARH, Agrp-Cre was used to knockout Gpr45 in Agrp-expressing neurons in ARH, Sim1-Cre was used to knockout Gpr45 in PVH, and Mc4r-Cre was used to knockout Gpr45 in Mc4r-expressing neurons in PVH. Gpr45^flox/flox; Pomc-Cre mice had similar fat weight and body weight compared with control mice, even at 16 weeks of age (Fig S4, A-C). Similarly, Gpr45 deletion via Agrp-Cre did not alter body weight or fat weight in 16-week-old mice (Fig. S4, D-F). In contrast, both Gpr45^flox/flox; Sim1-Cre and Gpr45^flox/flox; Mc4r-Cre mice exhibited significantly increased body weight, fat weight, and lean weight by 16 weeks of age (Fig. S4, G-L). Given the absence of an obesity phenotype in Pomc-Cre and Agrp-Cre mice, we used a high-fat diet (HFD) to try to accelerate the onset of mild obesity phenotypes. All Gpr45^flox/flox; Cre and Gpr45^flox/flox control mice were placed on a HFD starting at 6 weeks of age and maintained for 4 weeks. Even under HFD conditions, Gpr45^flox/flox; Pomc-Cre and Gpr45^flox/flox; Agrp-Cre mice showed no obesity phenotype (Fig. 3, I-N). However, Gpr45^flox/flox; Sim1-Cre and Gpr45^flox/flox; Mc4r-Cre mice exhibited significantly increased body weight and fat weight (Fig. 3, O-T).

While both Sim1-Cre and Mc4r-Cre target PVH neurons, they are also expressed in other brain regions (29, 30). To achieve PVH-specific Gpr45 deletion and avoid confounding effects from Cre-driven deletions during hypothalamic development, we performed stereotaxic injections of adeno-associated virus (AAV)-Cre into the PVH of Gpr45^flox/flox mice (Fig. 3U). The Cre expression, driven by the neuron-specific human synapsin 1 (hSyn) promoter (31), ensured exclusive deletion of Gpr45 in neurons. GFP co-expression was used to identify Cre-expressing cells (Fig. S5, A and B), and the successful deletion of Gpr45 in the PVH was confirmed by reduced Gpr45 mRNA levels (Fig. S5C). Two weeks following AAV injections, Gpr45^flox/flox mice injected with AAV-hSyn-GFP-Cre exhibited greater body weight compared to control mice injected with AAV-hSyn-GFP (Fig. 3V). By 7 weeks post-injection, significant increases in body weight, fat weight, and lean weight were observed in the AAV-hSyn-GFP-Cre group (Fig. 3, W-Y). However, injection of the same AAV-hSyn-GFP-Cre into the PVH of WT mice did not result in obesity or noticeable toxicity (Fig. S5, D-I), further confirming that the obesity phenotype was due to Cre-mediated Gpr45 deletion. Thus, GPR45 in PVH neurons plays a direct role in the regulation of body weight.

GPR45 is specifically localized in primary cilia

To further understand the intracellular role of GPR45, we examined its subcellular localization. GPR45 was exclusively localized to the primary cilia of N11 hypothalamic cells, as marked by red acetylated α-Tubulin (Ac-Tub) signals (Fig. 4A). This ciliary localization was not unique to N11 cells, because GPR45 was also detected in the primary cilia of IMCD3 cells, an inner medullary collecting duct cell line derived from the kidney (Fig. 4B). The tubby family proteins TUB bipartite transcription factor (TUB) and TUB like protein 3 (TULP3) are required for ciliary targeting of cilia localized GPCRs (14, 32-34). TUB and TULP3 function as adapters of the highly conserved intraflagellar complex-A (IFT-A) complex in ciliary trafficking (34, 35). Indeed, GPR45 interacted with TULP3 when co-expressed in 293T cells (Fig. 4C). In Tulp3 knockout (Tulp3^−/−) IMCD3 cells (36), the ciliary localization of GPR45 was totally disrupted (Fig. S6A). Because the complete knockout of Tulp3 reduces ciliary length and potentially disrupts cilia structure, we used Tulp3-specific siRNA to knock down Tulp3 expression in N11 cells to assess its effect on GPR45 localization (Fig. 4D). Tulp3 knockdown did not alter ciliary length but significantly reduced the ciliary localization of GPR45 (Fig. 4, E-H). We then mapped the ciliary localization sequence (CLS) within GPR45 that mediates TULP3 binding and its ciliary trafficking. Previous studies suggest that intracellular loop 3 (IC3) of GPCRs (32, 37-39) and the charged residues of amphipathic helices (36) often contain CLS. Using hydrophobic moment analysis of the α-helix in the IC3 of GPR45, we identified two predicted amphipathic helices (Fig. S6B). Targeted deletions removing charged residues in these helices revealed that deleting one helix partially reduced ciliary localization, while deleting both helices completely disrupted GPR45’s ciliary localization (Fig. S6, C-G). Thus, TULP3 plays a critical role in transporting GPR45 to the primary cilia by recognizing the CLS within the IC3 of GPR45.

Fig. 4. GPR45 is exclusively localized in the primary cilia.

(A) N11 cells expressing Gpr45-3xFlag were immunostained with Ac-Tub (red) to visualize cilia, Flag (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. (B) IMCD3 cells expressing Gpr45-3xFlag were immunostained with Ac-Tub (red) to visualize cilia, Flag (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. (C) Immunoblots of immunoprecipitates (top and middle) or lysates (bottom) of 293T cells expressing HA-tagged TULP3 and Gpr45-3xFlag or pBOB control. (D) Relative Tulp3 mRNA level in N11 cells transfected with control siRNA or Tulp3 siRNA. (E and F) Quantification of the cilia length (E) and percentage of GPR45+ cilia (F) in Gpr45-3xFlag expressing N11 cells transfected with control siRNA or Tulp3 siRNA. (G and H) Gpr45-3xFlag expressing N11 cells transfected with control siRNA (G) or Tulp3 siRNA (H) were immunostained with Ac-Tub (red) to visualize cilia, Flag (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. (I) PVH regions of brain coronal sections from homozygous Gpr45-GFP mice were immunostained with ADCY3 (red) to visualize cilia, GFP (green), and Hoechst (blue) to visualize nuclei. GFP positive cilia are marked with arrowheads. Scale bar, 10 μm. (J) PVH regions of brain coronal sections from WT mice were immunostained with ADCY3 (red) to visualize cilia, GFP (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. Data are presented as means ± SD. p values were determined by Student’s t test (D-F). ****p ≤ 0.0001; ns, not significant with p > 0.05. Data are representative of two independent experiments.

We used Gpr45-GFP knock-in mice to examine the subcellular localization of endogenous GPR45 in the hypothalamus and various brain regions. In the PVH, GPR45 was exclusively localized to primary cilia, identified by co-localization with red ADCY3 signals (Fig. 4I and S7B). The GFP signal was detectable only in Gpr45-GFP mice and was absent in WT control mice (Fig. 4J and S8B). Notably, GPR45 signal was weak or undetectable in the primary cilia of other major hypothalamic nuclei, including the ARH, anterior hypothalamic nucleus (AHN), suprachiasmatic nucleus (SCH), dorsomedial nucleus of the hypothalamus (DMH), ventromedial hypothalamic nucleus (VMH), lateral hypothalamic area (LHA), and tuberal nucleus (TU) (Fig. S7 and S8). Similarly, GPR45-GFP signal was minimal or undetectable from the primary cilia in other brain regions, including the cerebral cortex (CTX), hippocampus, midbrain (MB), cerebellum (CB), medulla (MY), and pons (P) (Fig. S9 and S10). These findings align with the high expression of Gpr45 mRNA in the PVH (Fig. 3E) and the obesity phenotype reported in both Gpr45^flox/flox; Sim1-Cre mice (Fig. 3, O-Q and S4, G-I) and those with PVH-specific deletion of Gpr45 via AAV-Cre (Fig. 3, U-Y). The selective ciliary localization of GPR45 in the PVH suggests its functional role in PVH neurons for appetite regulation.

GPR45 translocates Gα_s into primary cilia

In unstimulated cells, GPCR interacts with various Gα subunits across the cell membrane, including the Gα_s, Gα_i, Gα_q, and Gα₁₂ families (40). Among the major Gα subunits tested, GPR45 exhibited the strongest interaction with Gα_s (Fig. S11). In IMCD3 cells, overexpressed HA-tagged Gα_s (HA-Gαs) was scattered in the cytoplasm with no localization in primary cilia (Fig. S12, A and B). Surprisingly, co-expression of 3xFlag tagged Gpr45 (Gpr45-3xFlag) caused almost all HA-Gα_s to translocate to primary cilia, as indicated by the colocalization of Gpr45-3xFlag with HA-Gα_s (Fig. S12C) and HA-Gα_s with ciliary marker Ac-Tub (Fig. S12D). Given the role of TULP3 in the ciliary localization of GPR45, we assessed Gα_s translocation by GPR45 in Tulp3^−/− IMCD3 cells. While Gα_s and GPR45 partially colocalized in the cytoplasm, neither protein localized to the cilia in the absence of TULP3 (Fig S12, E-F). Similar translocation of Gα_s into primary cilia was observed in N11 cells (Fig. 5, A-C). To check the translocation of endogenous Gα_s, we used an antibody specifically recognizing the Gα_s protein (Fig. S13). In N11 cells, only a small fraction of cilia contained low levels of Gα_s, with the majority showing no detectable signal (Fig. 5D). However, overexpression of GPR45 led to the accumulation of endogenous Gα_s in all cilia (Fig. 5, E and I). Next, we sought to determine which region of GPR45 is critical for the ciliary translocation of Gα_s. Using AlphaFold 3 (41), we modeled the interaction between GPR45 and Gα_s (Fig. S14). Although the C-terminal tail (C-tail) of GPR45 showed a low predicted local distance difference test (pLDDT) score, it appeared in close proximity to Gα_s (Fig. S14). This is atypical, because Gα_s coupling primarily relies on transmembrane domains (TM3 and TM5) with some contributions from intracellular loops (IC2 and IC3) in most GPCRs (42). To evaluate the role of the GPR45 C-tail in Gα_s translocation, we performed a domain swap experiment with GPR161, a well-characterized Gα_s-coupled ciliary GPCR (39). GPR161 overexpression alone did not induce Gα_s translocation into cilia (Fig. 5, F, H, and I). Furthermore, replacing the GPR161 C-tail with that of GPR45 (GPR45-161) failed to translocate Gα_s into cilia, despite GPR45-161 still localizing to cilia (Fig. 5, G, H, and I). Thus, the C-tail of GPR45 plays a critical role in mediating the ciliary translocation of Gα_s.

Fig. 5. GPR45 targets Gα_s into primary cilia.

(A and B) N11 cells expressing HA-Gα_s (A) or HA-Gα_s and Gpr45-3xFlag (B) were immunostained with Flag (red), HA (green), and Ac-Tub (blue) to visualize cilia. Scale bar, 10 μm. (C) Quantification of the percentage of Gα_s+ cilia in A and B. (D-G) N11 cells expressing control, Gpr45-3xFlag, Gpr161-3xFlag, or Gpr45 with C-tail swapped with Gpr161 (Gpr45-161-3xFlag) were immunostained with Flag (red), Gα_s (green), and Ac-Tub (blue) to visualize cilia. Scale bar, 10 μm. (H and I) Quantification of the percentage of Flag+ cilia (H) or Gα_s+ cilia (I) in D-G. (J) PVH regions of brain coronal sections from homozygous Gpr45-GFP mice were immunostained with Gα_s (red), GFP (green), and ADCY3 (blue) to visualize cilia. Cilia with coexpression of Gα_s and GFP are marked with arrowheads. Scale bar, 10 μm. (K and L) PVH regions of brain coronal sections from +/+ mice (K) or Gpr45^−/− mice (L) were immunostained with ADCY3 (red) to visualize cilia, Gα_s (green), and Hoechst (blue) to visualize nuclei. Gα_s positive cilia are marked with arrowheads. Scale bar, 10 μm. (M) Quantification of the percentage of Gα_s+ cilia in K and L. Data are presented as means ± SD. p values were determined by Student’s t test (C, H, I). ****p ≤ 0.0001; ns, not significant with p > 0.05. Data are representative of two independent experiments.

To further explore the relevance of Gα_s translocation by GPR45, we examined the distribution of Gα_s in the PVH of Gpr45-GFP knock-in mice. Gα_s colocalized with GPR45 in the cilia of PVH neurons (Fig. 5J). In the PVH of WT mice, Gα_s+ cilia were observed in 47.8% (± 5.0%) of ADCY3+ cilia (Fig. 5, K and M). In contrast, Gα_s+ cilia accounted for only 14.6% (± 8.7%) of ADCY3+ cilia in the PVH of Gpr45^−/− mice (Fig. 5, L and M). Thus, the ciliary localization of Gα_s in the PVH is dependent on the role of GPR45.

GPR45 coordinates MC4R and ADCY3 signaling via ciliary Gα_s targeting to regulate cAMP levels in the cilia

Activation of Gα_s-coupled GPCRs leads to the release of Gα_s, which subsequently activates ADCYs to produce cAMP. Among the different ADCYs, ADCY3’s importance at neuronal primary cilia has been linked to monogenic causes of human obesity (17, 22, 23). In N11 cells, GPR45 and ADCY3 colocalized in primary cilia (Fig. 6A). Because GPR45 recruits Gα_s into cilia, we measured cAMP levels in cells overexpressing GPR45. GPR45 overexpression did not increase cAMP levels but instead lead to a decrease in the total intracellular cAMP (Fig. 6B). Nevertheless, this reduction was dependent on Gα_s, as GPR45 failed to decrease the intracellular cAMP levels with Gnas knockdown (Fig. 6C). This observation suggests that GPR45 may not act as a typical GPCR. Instead, it binds to Gα_s, facilitating its translocation into the cilia. Overexpression of GPR45 may sequestrate Gα_s, preventing its interaction with ADCY3 and subsequent activation. Supporting this hypothesis, overexpression of GPR45 significantly reduced the interaction between Gα_s and ADCY3 (Fig. 6D). To explore whether the translocated Gα_s could potentially activate ADCY3 in the cilia, we used a nanobody (NB35) (43) which binds to the Gα_s-Gβγ interface and was fused with GFP (NB35-GFP) to monitor the formation of active Gα_s-Gβγ heterotrimers. In N11 cells, NB35-GFP was predominantly found in the cytoplasm (Fig. 6E). However, overexpression of GPR45 caused NB35-GFP to accumulate in the cilia, with little to no signal in the cytoplasm (Fig. 6, F-H). Because the primary cilium has been described as a signaling hub that concentrates various molecules, including cAMP, the translocation of Gα_s by GPR45 suggests that cAMP levels in the cilium and cytoplasm might be differentially regulated. To test this, we employed a Förster resonance energy transfer (FRET)-based cAMP sensor (Arl13b-H187) (44) to measure cAMP levels in the cilia. Indeed, GPR45 overexpression significantly increased cAMP levels in the cilia (Fig. 6I), indicating that Gα_s transported by GPR45 is active in stimulating ADCY3 and elevating cAMP levels in the cilia.

Fig. 6. GPR45 targets Gα_s to cilia to promote MC4R activation and cAMP production via ADCY3.

(A) N11 cells expressing Adcy3-HA and Gpr45-3xFlag were immunostained with HA (red), Flag (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. (B) Relative cAMP level of Neuro-2a cell lysates expressing pCMV control or Gpr45. (C) Relative cAMP level of Neuro-2a cell lysates expressing pCMV control or Gpr45 and control siRNA or Gnas siRNA. (D) Immunoblots of immunoprecipitates (top and middle) or lysates (bottom) of 293T cells expressing indicated plasmids. (E and F) N11 cells expressing NB35-GFP (E) or NB35-GFP and Gpr45-3xFlag (F) were immunostained with Flag (red), GFP (green), and ARL13B (blue) to visualize cilia. Scale bar, 10 μm. (G and H) Quantification of NB35-GFP to ARL13B ratio (G) or NB35-GFP density in cytoplasm (H) in E and F. (I) Qualification of ciliary cAMP level using FRET-based cAMP sensor in N11 cells expressing control or Gpr45. Relative cAMP level was calculated as the ratio of 480 nm to 535 nm. (J) Representative photograph of male mice at 12 weeks of age. (K and L) Body weight (K) and food intake (L) of 8-week-old male mice. (M and N) PVH regions of brain coronal sections from homozygous Gpr45-GFP knock-in mice (M) or +/+ mice (N) were immunostained with MC4R (red), GFP (green), and Hoechst (blue) to visualize nuclei. Cilia with coexpression of MC4R and GPR45 are marked with arrowheads. Scale bar, 10 μm. (O) Qualification of ciliary cAMP level using FRET-based cAMP sensor in N11 cells expressing control or different genes. Relative cAMP level was calculated as the ratio of 480 nm to 535 nm. Data are presented as means ± SD. p values were determined by Student’s t test (B-C, G-I, K-L, O). **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001; ns, not significant with p > 0.05. Data points represent individual mice (K-L). Data are representative of two independent experiments.

Our cell-based assays suggest that GPR45 functions to translocate Gα_s into primary cilia, where it activates ADCY3. To test the genetic interaction between GPR45 and ADCY3 in vivo, we crossed Gpr45^−/− and Adcy3^−/− mice to generate single and double knockout animals. However, Adcy3^−/− mice exhibit anosmia and have a high mortality rate within 48 hours of birth (45). During our genetic screening, we identified a recessive allele of Adcy3, named magnificent_frigatebird, which was associated with severe obesity (Fig. S15, A and B). The L278H missense mutation disrupted both the adenylate cyclase activity of ADCY3 (Fig. S15, C and D) and its ciliary localization (Fig. S15, E and F). However, unlike Adcy3^−/− mice, Adcy3^L278H/L278H mice were viable and survived well. Although both Gpr45^−/− and Adcy3^L278H/L278H single mutant mice are obese and hyperphagic, Gpr45^−/−; Adcy3^L278H/L278H double mutant mice did not show more severe obesity or hyperphagia compared to the single mutants (Fig. 6, J-L). Similarly, hyperglycemia and liver steatosis were comparable in Gpr45^−/−, Adcy3^L278H/L278H single mutants, and double mutant mice (Fig. S16, A-D). Thus, GPR45 and ADCY3 likely function within the same signaling pathway to regulate food intake.

Because MC4R couples with Gα_s and both MC4R and ADCY3 are localized in the neuronal cilia of the PVH (17, 18), and knockout of Gpr45 using Mc4r-Cre caused an obesity phenotype (Fig. 3, R-T and S4, J-L), it is plausible that GPR45 modulates the activity of MC4R by targeting Gα_s into the cilia. To test this idea, we used an MC4R-specific antibody (18) to examine the localization of MC4R+ and GPR45+ cilia in Gpr45-GFP knock-in mice. The majority of MC4R+ cilia in the PVH expressed GPR45 (Fig. 6, M and N). In N11 cells, overexpression of MC4R, along with its cilia-targeting accessory protein MRAP2 (15), resulted in exclusive MC4R localization in the cilia (Fig. S17C). However, despite the presence of ciliary MC4R, Gα_s localization in the cilia was not increased (Fig. S17C), and no change in ciliary cAMP was observed (Fig. 6O), as indicated by the FRET-based cAMP sensor. Co-expression of GPR45 with MC4R promoted the translocation of Gα_s into the cilia (Fig. S17, A-E), leading to a further increase in ciliary cAMP compared to GPR45 overexpression alone (Fig. 6O). GPR45-induced MC4R activation resulted in ciliary cAMP levels comparable to those observed upon MC4R stimulation by α-MSH (Fig. S17F). However, α-MSH did not further elevate ciliary cAMP in GPR45-overexpressing cells (Fig. S17F), suggesting that ligand-induced MC4R activation might occur independently of GPR45-mediated MC4R activation. Indeed, administration of the MC4R agonist setmelanotide reduced food intake in Gpr45 knockout mice (Fig. S17G). Thus, GPR45 operates in a distinct pathway to translocate Gα_s into cilia, where it is used by MC4R to activate ADCY3 and elevate ciliary cAMP.

Both expansive and extensive mutations disrupt the ciliary localization of GPR45

GPR45 is predicted to be a seven-transmembrane GPCR. The expansive mutation S214P and the extensive mutation Y287C are located in the fifth and sixth transmembrane regions of GPR45, respectively (Fig. 7A). Neither the Y287C nor the S214P mutation affected the protein’s stability, as indicated by similar expression levels and patterns when overexpressed in 293T cells (Fig. 7B). Like WT GPR45, both Y287C and S214P mutant forms still interacted with Gα_s (Fig. 7B), and overexpression of GPR45-S214P or GPR45-Y287C resulted in a significant reduction in cAMP levels in whole-cell lysates (Fig. 7C). However, in contrast to WT GPR45, GPR45-S214P and GPR45-Y287C failed to increase cAMP levels in the cilia (Fig. 7D). Additionally, both the Y287C and S214P mutants of GPR45 localized exclusively in the cytoplasm, with no ciliary localization observed in N11 cells (Fig. 7, E-H). Consistent with the disruption of ciliary localization, HA-tagged Gα_s failed to localize in the cilia when GPR45-S214P or GPR45-Y287C was overexpressed. Instead, both Gα_s and GPR45 mutants were colocalized in the cytoplasm (Fig. 7, I-L). Moreover, endogenous Gα_s also failed to be translocated to the cilia by overexpression of either GPR45-S214P or GPR45-Y287C mutants (Fig. S18, A-E). Thus, while the mutants sequestrate Gα_s in the cytoplasm, they fail to direct Gα_s to the cilia, thereby impairing the activation of ciliary cAMP signaling, which highlight the critical role of ciliary localization in GPR45-mediated regulation of energy homeostasis.

Fig. 7. Expansive and extensive mutations disrupt the ciliary localization of GPR45.

(A) Topology of mouse GPR45 protein with the mutation sites S214P (expansive) and Y287C (extensive) colored in red. (B) Immunoblots of immunoprecipitates (top and middle) or lysates (bottom) of 293T cells expressing indicated plasmids. (C) Relative cAMP level of Neuro-2a cell lysates expressing pCMV control, Gpr45-S214P, or Gpr45-Y287C. (D) Qualification of ciliary cAMP level using FRET-based cAMP sensor in N11 cells expressing pBOB control, Gpr45-S214P, or Gpr45-Y287C. Relative cAMP level was calculated as the ratio of 480 nm to 535 nm. (E-G) N11 cells expressing Gpr45-3xFlag (E), Gpr45-S214P-3xFlag (F), or Gpr45-Y287C-3xFlag (G) were immunostained with Ac-Tub (red) to visualize cilia, Flag (green), and Hoechst (blue) to visualize nuclei. Scale bar, 10 μm. (H) Quantification of the percentage of GPR45+ cilia in E-G. (I-K) N11 cells expressing HA-Gα_s with Gpr45-3xFlag (I), Gpr45-S214P-3xFlag (J), or Gpr45-Y287C-3xFlag (K) were immunostained with Flag (red), HA (green), and Ac-Tub (blue) to visualize cilia. Scale bar, 10 μm. (L) Quantification of the percentage of Gα_s+ cilia in I-K. Data are presented as means ± SD. p values were determined by Student’s t test (C-D, H, L). **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001; ns, not significant with p > 0.05. Data are representative of two independent experiments.

Discussion

Many breakthroughs in understanding human obesity have been achieved through studies on mouse mutants, leveraging the physiological similarities between mice and humans. However, the inherently low rate of random mutations and labor-intensive positional cloning techniques once hindered the speed of discoveries. In recent years, we have deliberately induced random mutations in mice and utilized a forward genetic screen platform (24) that incorporates highly automated meiotic mapping to identify mutations causing obesity in an unbiased and rapid manner. Here we describe obese mutants caused by two missense mutations of Gpr45, named expansive and extensive. Knockout of Gpr45 reproduced the obesity phenotype observed in these mutants, suggesting these are loss-of-function mutations. Nonetheless, our mechanistic studies revealed that both mutations disrupt the ciliary localization of GPR45 and the translocation of Gα_s. These Gpr45 mutants reveal the importance of Gα_s ciliary transportation by GPR45 in the regulation of food intake and energy homeostasis.

Primary cilia have emerged as clinically important sensory organelles, with ciliary dysfunction causing ciliopathies (13, 46). Many patients with ciliopathies are associated with pediatric obesity, suggesting the importance of the structure and function of cilia in energy homeostasis (11). Disruption of cilia in adult mouse models leads to hyperphagia induced obesity (12). Additionally, several obesity-associated gene mutations identified in humans are linked to primary cilia, including MC4R (17), ADCY3 (20-23, 47), fat mass and obesity-associated (FTO) (48-50) regulated retinitis pigmentosa GTPase regulator-interacting protein-1 like (RPGRIP1L) (51, 52), and TUB (53, 54). Our finding of the ciliary localization of GPR45 adds another regulator of obesity to this tiny organelle. Coincidentally, both GPR45 and MC4R are class A rhodopsin-like GPCRs. The ciliary localization of GPR45 requires TULP3, while the ciliary localization of MC4R requires MRAP2 (15). Both expansive and extensive missense mutations disrupted the ciliary localization of GPR45; these mutations might interfere the association of GPR45 with TULP3/TUB or the association of GPR45 with other unknown ciliary trafficking chaperones.

The canonical role of GPCRs is to couple the binding of agonists with the activation of specific heterotrimeric G proteins, ultimately modulating downstream effector proteins (55). However, our findings revealed a non-canonical role of GPR45 in transporting Gα_s into primary cilia. It remains unclear whether these transported Gα_s proteins in the cilia are in an active form with GTP bound or in an inactive form with GDP bound, and whether GPR45 binds Gα_s in the cytoplasm and releases Gα_s in the cilia. Regardless, GPR45 increases the local concentration of Gα_s in the cilia, and these Gα_s could activate ADCY3 to increase the local cAMP levels in the cilia. Sequestration of a Gα_i coupled receptor GPR88 in cilia provides an insulating function of cilia in specifically inhibiting catecholamine signaling in cilia while inhibiting extraciliary catecholamine signaling in non-ciliated cells (56). GPR88 expression also inhibits MC4R ciliary signaling in the PVH of mice and induces obesity (17). The Omega-3 fatty acid activated GPCR, free fatty acid receptor 4 (FFAR4), also promotes ciliary cAMP levels in controlling peripheral adipogenesis (57). These studies, along with ours, suggest the importance of ciliary cAMP in mediating ciliary GPCR signaling. The activation of G proteins represents a rate-limiting step in GPCR activation (58). In addition to MC4R, increased ciliary Gα_s may facilitate the activation of other Gα_s-coupled ciliary GPCRs, which remains to be explored. Most GPCRs exhibit a basal level of GTP exchange activity in the absence of ligands. The basal activity of GPR45 might be essential for the transportation and shuttling of Gα_s from the cytoplasm to the cilia. However, GPR45 might be fully activated by ligands, which could also affect the ciliary transportation of Gα_s. Searching for ligands/agonists of GPR45 will help to understand this non-canonical role of GPR45.

While MC4R is known to couple with Gα_s (19), and cAMP signaling in PVH^MC4R neurons directly regulates satiety (59)_, it has also been reported that MC4R couples with Gα_q/11 (60), β-arrestin (61, 62), mitogen-activated protein kinase (MAPK) (63, 64), and Kir7.1 potassium channel (65). Here, we found that the MC4R agonist setmelanotide still reduced food intake in Gpr45 knockout mice, suggesting the possibility that various MC4R downstream signaling pathways might compensate for each other upon strong ligand activation. Nevertheless, our observation that overexpression of GPR45 was sufficient to activate ciliary MC4R, which could not be further increased by α-MSH, suggests that GPR45-mediated ciliary MC4R activation does not necessarily require ligands. However, this observation is limited by the fact that GPR45 signaling was driven by overexpression due to the lack of GPR45 ligand. The physiological levels of ciliary melanocortin signaling may be much more dynamic and balanced. Furthermore, while our study focused on the regulation of energy homeostasis by the ciliary melanocortin system, our work does not negate the extraciliary roles of MC4R and other associated signaling pathways.

Our results establish hyperphagia as the primary cause of obesity in Gpr45^−/− mice. However, it has been reported that decreased energy expenditure also contributes to the development of obesity in Gpr45^−/− mice (26). This apparent discrepancy may stem from the use of different mouse strains. In our study, all mouse models were bred in the C57BL/6J background, the strain most commonly used in obesity research owing to its susceptibility to diet-induced obesity and severe insulin resistance when fed HFD (66). In contrast, the previous study used FVB/NJ mice, which are known for their resistance to diet-induced obesity and comparatively lower weight gain on HFD (66).

The mouse GPR45 protein shares 89% identity with its human counterpart, with highly similar homologs also found in other vertebrates. Similar to the expression profile of mouse Gpr45, the human GPR45 gene exhibits its highest expression in the hypothalamus, along with significant expression in other brain regions. Notably, there is also high expression in the testis, although its functional significance remains unknown (GTEx Analysis Release V8, dbGaP Accession phs000424.v8.p2). Considering the conserved role of the melanocortin system in energy homeostasis, it is likely that the role of GPR45 in body weight maintenance is conserved in humans. MC4R agonists have demonstrated efficacy in reducing body weight in obese patients with rare genetic mutations in the leptin-melanocortin pathway (67). Investigating GPR45 mutations in humans with early-onset obesity could elucidate its potential role in human obesity and facilitate research into GPR45 as a potential target for anti-obesity drug development.

Materials and Methods

Mice

C57BL/6J mice (stock# 000664), Pomc-Cre mice (68) (stock# 010714), Sim1-Cre mice (30) (stock# 006395), and Mc4r-Cre mice (29) (stock# 030759) are available from The Jackson Laboratory. The expansive (Gpr45^S214P), extensive (Gpr45^Y287C), and magnificent_frigatebird (Adcy3^L278H) mice were created through ENU mutagenesis and are detailed at Mutagenetix (http://mutagenetix.utsouthwestern.edu). The Gpr45^−/− mice, Gpr45-flox mice, and Gpr45-GFP mice were developed in our laboratory using CRISPR/Cas9 system with the sgRNAs and DNA templates listed in Table S5. Tissue-specific knockout of Gpr45 was accomplished through the breeding of Gpr45-flox mice with mice carrying each specific Cre strain. Gpr45^−/−; Adcy3^L278H mice were generated by mating heterozygous Gpr45 knockout mice with Adcy3^L278H mice. All mice were provided with a standard chow diet (Teklad Global 16% Protein Rodent Diet) or a HFD (60 kcal% fat, Research Diets) and were housed at room temperature (23 °C) unless indicated. The mice were maintained at The University of Texas Southwestern Medical Center, and all studies were conducted in compliance with institutionally approved protocols. Approval for all experiments in this study was obtained from The University of Texas Southwestern Medical Center Institutional Animal Care and Use Committee.

Stereotaxic injection

Stereotaxic injection of AAV into the PVH of the mouse brain was performed according to a published protocol (69). Briefly, 8-week-old Gpr45^flox/flox mice or +/+ mice were first anesthetized with isoflurane and maintained throughout the procedure. The mice were placed into a stereotaxic frame, and a small opening was made in the skull directly over the injection site (bregma: AP, −0.5 mm; ML, ± 0.22 mm; DV, −4.8 mm). A total of 100 nL of AAV-hSyn-GFP or AAV-hSyn-GFP-Cre (2x10¹² vg/ml, AAV serotype 8, UNC viral core) were pressure injected into the PVH bilaterally. After the injection, all animals were allowed to recover for at least two weeks before being used in the experiments.

Metabolic analysis of mice

Mice underwent a 6-hour fasting period (from 7:00 AM to 1:00 PM) for insulin tolerance tests. Blood glucose levels were assessed using the AlphaTRAK glucometer and test strips. Following the initial blood glucose measurement, the insulin tolerance test was initiated with an intraperitoneal injection of human insulin (0.75 U/kg; Sigma-Aldrich), and blood glucose was monitored at designated intervals over the next 2 h. For glucose tolerance tests, mice were subjected to a 14-hour fast (from 6:00 PM to 8:00 AM). After the initial blood glucose measurement, a 10% glucose solution (1 g/kg; Sigma-Aldrich) was administered intraperitoneally, and blood glucose levels were measured at specific time points over the subsequent 2 h. Magnetic resonance spectroscopy (MRS) of live mice was conducted using EchoMRI Body Composition Analyzers with default settings. Internal temperatures of mice were recorded by implanting a temperature transponder (IPTT-300) under the skin and measured using a portable reader (DAS-8007-IUS, BioMedic Data Systems). Metabolic cage measurements were carried out using the TSE PhenoMaster system. For acute cold exposure, mice were individually housed in 6°C cold chambers without access to food, and body temperature was monitored at the specified time points. For testing the effect of setmelanotide on food intake, mice were first fasted for 24 h, then injected with saline, and basal food intake was recorded over 2 h. After a one-week recovery period, the mice underwent another 24-hour fast, followed by setmelanotide injection (5 mg/kg, i.p.), with food intake measured over 2 h. The data are presented as the percentage of food intake after setmelanotide injection relative to food intake after saline injection for each mouse.

Serum chemistries and liver triglycerides measurement

Mice underwent a 14-hour fasting period (from 6:00 PM to 8:00 AM) for blood collection. Blood glucose levels were assessed using the AlphaTRAK glucometer and test strips. Insulin levels in the serum were quantified using ELISA kits (Crystal Chem) in accordance with the manufacturer’s instructions. Additionally, triglyceride and cholesterol levels were measured separately using Infinity Triglycerides Reagent and Infinity Cholesterol Reagent, respectively (Thermo Fisher Scientific). To quantify liver triglycerides, frozen liver samples were weighed and homogenized in homogenization buffer (10 mM Tris-Cl pH 7.4, 0.9% NaCl, and 0.1% Triton X-100) using a Tissue Lyser (QIAGEN). Liver homogenates were diluted and incubated at 37 °C with 1% deoxycholate. Liver triglycerides were measured using Triglycerides Liquid Stable Reagent (Thermo Fisher Scientific, TR22421), following a similar procedure as serum triglyceride measurement. The obtained liver triglyceride values were normalized with either total protein (mg) or the weight of the liver sample.

Tissue immunohistochemistry, immunostaining, and RNAscope

Tissue samples designated for routine hematoxylin and eosin (H&E) staining were fixed in 10% neutral-buffered formalin for 48 h, while samples intended for Oil Red O (ORO) staining were fixed in methanol-free 4% paraformaldehyde for the same duration before being equilibrated in 30% sucrose. Subsequent processing involved paraffin embedding for H&E-stained sections and cryoembedding for ORO-stained sections as previously described (70). For hypothalamus staining, mice were perfused with PBS, followed by fixation in a 4% paraformaldehyde solution (except for Gpr45-GFP staining, which used 1% paraformaldehyde to enhance signal detection). Subsequently, brains were dissected and post-fixed in the fixation solution overnight at 4°C. The fixed brains were then immersed in 18% and 30% sucrose solutions overnight for cryoprotection. Coronal sections of 25-μm thickness were obtained using a microtome after freezing the brains with dry ice. The sections were initially blocked for 1 h in blocking buffer containing 3% normal goat serum in PBS, 0.4% Triton X-100, and 0.2% sodium azide. Following blocking, sections were incubated overnight at 4°C with primary antibodies. Subsequently, slides were washed with PBS four times for 15 min each and then incubated with secondary antibodies along with Hoechst 33342. After incubation, slides were washed with PBS four times for 5 min each and mounted in mounting medium (20 mM Tris-Cl, pH 8.0, 90% glycerol, 0.5% N-propyl gallate). The following primary antibodies were used in this study: chicken anti-GFP (1:1000, Aves Labs, GFP-1020, RRID:AB_10000240), rabbit anti-ADCY3 (1:500, LSBio, LS-C804038), guinea pig anti-MC4R (18) (2 μg/ml), and mouse anti-Gα_s (1:500, Thermo Fisher Scientific, MA5-27115, RRID:AB_2724378). The second antibodies (Alexa Fluor 488, Rhodamine Red-X, and Alexa Fluor 647) were from Jackson ImmunoResearch. RNAscope was performed using the RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics, 323270) following the manufacturer’s protocol with the following probes: Mm-Gpr45 (Advanced Cell Diagnostics, 318211), Mm-Pomc (Advanced Cell Diagnostics, 314081-C2). Immunostaining and RNAscope images were taken by Zeiss LSM 880 or LSM 980 inverted confocal microscopes using ZEN software.

Cell culture and transfection

The 293T cells (CRL-3216) and Neuro-2a cells (CCL-131) were purchased from American Type Culture Collection (ATCC). N11 cells were kindly gifted by Dr. Xiaoyong Yang (Yale University). IMCD3 and IMCD3 Tulp3^−/− cells were described before (71). All cells were grown in culture medium [DMEM (Gibco), 10% FBS (ATCC), Pen-Strep antibiotics (Gibco)] at 37 °C with 5% CO₂. Transfection of plasmids or siRNAs (DsiRNA, Integrated DNA Technologies) was carried out using Lipofectamine 2000 or Lipofectamine RNAiMAX (Invitrogen), respectively, according to the manufacturer’s instructions. The siRNAs used are listed in Table S5. Cells were harvested between 36 and 48 h post-transfection.

Sample preparation, immunoprecipitation, and western blot analysis

For regular immunoprecipitation, cells were lysed in Nonidet P-40 lysis buffer (50 mM Tris-Cl, pH 8.0, 0.1 M NaCl, 10 mM sodium fluoride, 1 mM sodium vanadate, 1% Nonidet P-40, 10% glycerol, 1.5 mM EDTA, and Protease Inhibitor Mixture) for 30 min on ice. After centrifugation, lysates were incubated with Flag antibody-conjugated beads (Sigma-Aldrich, M8823) for 4 h at 4 °C. Beads were washed three times with 1 mL of Nonidet P-40 lysis buffer and then eluted with 3xFlag peptides for 1 h at 4 °C. For Western blot, samples were resolved by NuPAGE 4-12% Bis-Tris gels (Thermo Fisher Scientific), transferred to NC membranes (Bio-Rad), blotted with the primary antibody at 4 °C overnight and the secondary antibody for 1 h at room temperature, and then visualized by chemiluminescent substrate (Thermo Fisher Scientific). The following primary antibodies were used in this study: mouse anti-HA (1:5000, Sigma-Aldrich, H9658, RRID:AB_260092), rabbit anti-HA (1:2000, Cell Signaling Technology, 3724, RRID:AB_1549585), mouse anti-Flag (1:500, Sigma-Aldrich, F1804, RRID:AB_262044), rabbit anti-GAPDH (1:5000, Cell Signaling Technology, 5174, RRID:AB_10622025), rabbit anti-GFP (1:2000, Abcam, ab13970, RRID:AB_300798), rabbit anti-POMC (1:1000, Cell Signaling Technology, 23499, RRID: AB_2716565), and mouse anti-Gnas (1:1000, Thermo Fisher Scientific, MA5-27115, RRID:AB_2724378). The second antibodies (HRP-based) were from Jackson ImmunoResearch.

RNA isolation, reverse transcription, and quantitative PCR

RNA purification was conducted using the TRIzol Plus RNA Purification Kit (Thermo Fisher) following the manufacturer's instructions. For reverse transcription, 1 μg of RNA was used with the SuperScript III First-Strand Synthesis SuperMix (Thermo Fisher) following the standard protocol. Quantitative PCR was performed with Applied Biosystems QuantStudio 6 with Powerup SYBR Green Master Mix (Life Technologies). The 2^−ΔΔCt method was used for relative quantification. The primers used are listed in Table S5.

Cell immunostaining

For immunostaining of culture cells, IMCD3 or N11 cells were transfected with corresponding plasmids. After 24 h, cells were cultured in serum-free medium for 48 h to induce ciliation. Cells were fixed with 4% PFA for 20 min at room temperature, followed by permeabilization and blocking in blocking buffer (1xPBS, 5% normal goat serum, and 0.3% Triton X-100) for 1 h at room temperature. After blocking, cells were incubated with primary antibodies diluted in antibody dilution buffer (1xPBS, 1% BSA, and 0.3% Triton X-100) overnight at 4 °C, then washed with 1xPBS and incubated with secondary antibodies along with Hoechst 33342 diluted in antibody dilution buffer for 1 h at room temperature. Finally, cells were mounted in mounting medium (20 mM Tris-Cl, pH 8.0, 90% glycerol, 0.5% N-propyl gallate). The following primary antibodies were used in this study: Mouse anti-HA (1:500, Sigma-Aldrich, H3663, RRID:AB_262051), rabbit anti-HA (1:500, Cell Signaling Technology, 3724, RRID:AB_1549585), mouse anti-Flag (1:500, Sigma-Aldrich, F1804, RRID:AB_262044), rabbit anti-DYKDDDDK (1:500, Cell Signaling Technology, 14793, RRID: AB_2572291), mouse anti-Gnas (1:1000, Thermo Fisher Scientific, MA5-27115, RRID:AB_2724378), rabbit anti-Acetyl-α-Tubulin (1:1000, Cell Signaling Technology, 5335, RRID:AB_10544694), and mouse anti-ARL13B (72) (1:500; Addgene, 180085; RRID:AB_2750771). The second antibodies (Alexa Fluor 488, Rhodamine Red-X, and Alexa Fluor 647) were from Jackson ImmunoResearch. Immunostaining images were taken with Zeiss LSM880 or LSM980 inverted confocal microscopes using ZEN software.

cAMP measurement in cell lysates and cilia

For measuring total cAMP levels, cells were transfected with indicated plasmids. After 24 h, 1x10⁴ cells were seeded into 96-well plates. Subsequently, the cells were treated with DMSO, IBMX (500 μM), or IBMX (500 μM) + forskolin (2 μM) for 1 h. Following treatment, the Cyclic AMP XP Assay Kit (Cell Signaling Technology, 4339) was used to measure cAMP levels according to the manufacturer’s instructions. The Arl13b-H188 plasmid (kindly gifted by Dr. Aldebaran M. Hofer at Harvard Medical School) was used to measure ciliary cAMP by live cell imaging as previously described (73). N11 cells stably expressing corresponding plasmids were transfected with the Arl13b-H188 plasmid. Following a 24-h post-transfection period, the cells were subjected to serum-free medium for 24 to 48 h. Subsequently, the medium was replaced with live cell image solution (Thermofisher). For experiments with/without ligand stimulation, imaging was first conducted without the ligand. Subsequently, 1 μM α-MSH was added to the cells, and after a 5-min incubation, image acquisition was initiated. cAMP levels in cilia were represented by FRET emission ratios (480 nm/535 nm; 440 nm excitation). and FRET imaging experiments were conducted at 37 °C, 5% CO2, utilizing Zeiss LSM880 or LSM980 inverted confocal microscopes with ZEN software.

Supplementary Material

Figs. S1 to S18

Tables S1 to S5