Primary cilia mediate early life programming of adiposity through lysosomal regulation in the developing mouse hypothalamus

Abstract

Hypothalamic neurons including proopiomelanocortin (POMC)-producing neurons regulate body weights. The non-motile primary cilium is a critical sensory organelle on the cell surface. An association between ciliary defects and obesity has been suggested, but the underlying mechanisms are not fully understood. Here we show that inhibition of ciliogenesis in POMC-expressing developing hypothalamic neurons, by depleting ciliogenic genes IFT88 and KIF3A, leads to adulthood obesity in mice. In contrast, adult-onset ciliary dysgenesis in POMC neurons causes no significant change in adiposity. In developing POMC neurons, abnormal cilia formation disrupts axonal projections through impaired lysosomal protein degradation. Notably, maternal nutrition and postnatal leptin surge have a profound impact on ciliogenesis in the hypothalamus of neonatal mice; through these effects they critically modulate the organization of hypothalamic feeding circuits. Our findings reveal a mechanism of early life programming of adult adiposity, which is mediated by primary cilia in developing hypothalamic neurons.

Fig. 1. Adult-onset ciliary dysgenesis in POMC neurons does not change energy and glucose metabolism.

a Experimental scheme in mice with POMC-specific IFT88 depletion that was induced by tamoxifen injections at 7 weeks. b Representative confocal images of AC3 (adenylyl cyclase-3, cilia marker) and tdTomato double staining in the ARH of POMC-cre/ERT2;;IFT88^f/f;;tdTomato mice and POMC-cre/ERT2;;tdTomato mice (n = 5). Arrowheads indicate cilia of POMC^tdTomato neurons. The graph depicts the average cilia lengths of about 100 POMC and non-POMC neurons. *p < 0.001. Scale bars: 20 μm. c Body weights in POMC-cre/ERT2;;IFT88^f/f mice and their IFT88^f/f littermates on a chow diet (n = 10 for IFT88^f/f mice, n = 8 for POMC-cre/ERT2;;IFT88^f/f mice). d Lean mass and fat mass measured at 12 weeks (n = 10 for IFT88^f/f mice, n = 7 for POMC-cre/ERT2;;IFT88^f/f mice). e Average weekly food intake during 4–16 weeks (n = 5). f Energy expenditure measured at 12 weeks (n = 5). g Glucose and insulin tolerance tests performed at 15 weeks (n = 7). h Body weights and cumulative food intake during a high-fat diet (HFD) challenge (n = 7). Data are presented as the mean ± SEM values. Statistics were performed using two-sided Student’s t test (b, d) and one-sided two-way ANOVA (c, e–h) followed by post hoc least significant difference (LSD) test. ns not significant.

Fig. 2. Inhibition of ciliogenesis in the developing POMC neurons causes obesity and glucose dysregulation in adulthood.

a Experimental scheme in mice with POMC-specific IFT88 depletion that commenced from E11–E12. b Representative cilia images and measurement of cilia lengths in the ARH of POMC-cre;;IFT88^f/f;;tdTomato mice and age-matched POMC-cre;;tdTomato mice (n = 5). Arrowheads indicate cilia of POMC^tdTomato neurons. Scale bars: 20 μm. c Body weights and body lengths of POMC-cre;;IFT88^f/f and IFT88^f/f male and female mice (body weights: n = 6 for POMC-cre;;IFT88^f/f males and n = 4 for the other 3 groups, body lengths: n = 7). d Lean mass and fat mass measured at 8 and 20 weeks (n = 4 for IFT88^f/f males, n = 6 for the other 3 groups). e Energy expenditure determined at 8 weeks (n = 4). f Average values of weekly food intake during 6–20 weeks (n = 6). g Glucose and insulin tolerance tests performed at 5 and 20 weeks (n = 5). h Patch clamp recordings showing the frequency and mean amplitude of the miniature excitatory postsynaptic currents (mEPSCs) and the miniature inhibitory postsynaptic currents (mIPSCs) in the POMC neurons lacking KIF3A or IFT88 (EPSC recording: n = 30 for POMC-cre mice, n = 15 for POMC-cre;;KIF3A^f/f mice, n = 23 for POMC-cre;;IFT88^f/f mice; IPSC recording: n = 29 for POMC-cre mice, n = 14 for POMC-cre;;KIF3A^f/f mice, n = 21 for POMC-cre;;IFT88^f/f mice). Data are presented as the mean ± SEM values. Statistics were performed using two-sided Student’s t test (b, d–f), one-sided two-way ANOVA (c, g) and one-sided one-way ANOVA (h) followed by post hoc LSD test and. *p < 0.05, **p < 0.01, ***p < 0.001 vs. IFT88^f/f controls. ns not significant.

Fig. 3. Ciliogenesis in POMC neurons during the early postnatal period is critical for adulthood energy balance.

a Cilia (AC3) staining in the developing hypothalamus of C57 mice (n = 4 for P0 and P28, n = 5 for E12.5, E15.5, P14, and P60, n = 6 for E18.5 and P7). The graphs depict the average length of about 100 ARH cilia and the ciliated cell percentage (the cilia numbers divided by the DAPI numbers) in each mouse. Scale bars: 20 μm. 3V third cerebroventricle. b IFT88 immunoblotting in the developing hypothalamus of C57 mice (n = 4). c AC3 and β-endorphin double staining in the ARH of C57 mice showing POMC neuron ciliogenesis during development (n = 4 for E18.5, P7, and P60, n = 5 for P1 and P14). Scale bars: 20 μm. d Experimental scheme in mice with POMC-specific IFT88 depletion that was induced by tamoxifen injections during the postnatal period (P1–P14). e Representative cilia images showing ciliary dysgenesis in POMC neurons (n = 4). Arrowheads indicate cilia of POMC neurons. Scale bars: 20 μm. f Body weights in mice with postnatal ciliary dysgenesis in POMC neurons and their IFT88^f/f litters on a chow diet (n = 5 for males, n = 6 for females). g Lean mass and fat mass measured at 15 weeks (n = 5 for males, n = 6 for females). h Average values of weekly food intake during 10–15 weeks (n = 5 for males, n = 6 for females). i Energy expenditure measured at 12 weeks (n = 4). Data are presented as the mean ± SEM values. Statistics were performed using one-sided one-way ANOVA (a–c), one-sided two-way ANOVA (f) followed by post hoc LSD test and two-sided Student’s t test (e, g–i). *p < 0.05, **p < 0.01, ***p < 0.001 vs. IFT88^f/f controls or between the indicated groups.

Fig. 4. Altered neurogenesis in the developing POMC neurons with ciliary dysgenesis.

a The numbers of tdTomato-labeled POMC neurons in the ARH of POMC-cre;;KIF3A^f/f;;tdTomato mice compared with POMC-cre;;tdTomato littermates (n = 5 for POMC-cre;;IFT88^f/f mice at both ages, n = 4 for IFT88^f/f mice at 3 weeks, n = 3 for IFT88^f/f mice at 8 weeks). Measurements were conducted at the indicated ages. *p = 0.0367. Scale bars: 100 μm. 3V third cerebroventricle. b The numbers of β-endorphin-expressing POMC neurons in the ARH of POMC-cre;;IFT88^f/f mice and IFT88^f/f littermates at the indicated ages (n = 5 for POMC-cre;;IFT88^f/f mice and n = 6 for IFT88^f/f mice). **p = 0.002. Scale bars: 50 μm. c 5-Bromouridine (BrdU) and POMC (β-endorphin) double staining showing the reduced POMC neurogenesis during E10.5–E12.5 in POMC-cre;;IFT88^f/f mice (n = 4 for POMC-cre;;IFT88^f/f mice, n = 3 for IFT88^f/f mice). The experimental schedule was presented. Arrowheads indicate BrdU⁺ POMC neurons. *p = 0.023. Scale bars: 50 μm. d BrdU/β-endorphin double staining showing the increased neurogenesis of POMC neurons in POMC-cre;;IFT88^f/f mice during the post-weaning period (3–6 weeks) (n = 5). Arrowheads indicate BrdU⁺ POMC neurons. *p = 0.027. Scale bars: 50 μm. Data are presented as the mean ± SEM values. Statistics were performed using two-sided Student’s t test (a–d). ns not significant.

Fig. 5. Impaired neuronal circuit formation in POMC neurons with defective cilia.

a Axonal projections of tdTomato-labeled POMC neurons in hypothalamic paraventricular nucleus (PVH), dorsomedial nucleus (DMH) and lateral hypothalamic area (LH) of POMC-cre;;KIF3A^f/f;; tdTomato and POMC-cre;;tdTomato male mice at 8 weeks (n = 5). Representative images and the graph depicting the axonal fiber density of POMC neurons in each hypothalamic area. Scale bars: 100 μm. b, c β-Endorphin-immunoreactive axonal projections of POMC neurons in the PVH and DMH of POMC-cre;;IFT88^f/f and POMC-cre/ERT2;;IFT88^f/f mice and their control littermates at 8–10 weeks (n = 6). Scale bars: 100 μm. d The primary neuronal process of tdTomato⁺ POMC neurons in POMC-cre;;KIF3A^f/f;;tdTomato and POMC-cre;;tdTomato mice (n = 6). Representative images and graph depicting the average number of the primary process of 100 POMC neurons. Scale bars: 20 μm. ARH hypothalamic arcuate nucleus. e MAP2 (dendrite marker), neurofilament (axon marker), and tdTomato (POMC neuron) triple immunostaining in primary cultured hypothalamic neurons at 21 days in vitro (DIV), which were obtained from POMC-cre;;IFT88^f/f;;tdTomato and POMC-cre;;tdTomato embryos, respectively. Sholl analysis was conducted for the assessment of dendritic arborization in 20 POMC neurons from POMC-cre;;tdTomato embryos and 12 neurons from POMC-cre;;IFT88^f/f;;tdTomato embryos. The full lengths of axons were measured for 19 cells from POMC-cre;;tdTomato embryos and 8 cells from POMC-cre;;IFT88^f/f;;tdTomato embryos. Scale bars: 20 μm. f NPY immunostaining in the ARH, PVH, and DMH of 3-week-old POMC-cre;;IFT88^f/f mice and IFT88^f/f mice (n = 6 for IFT88^f/f mice, n = 5 for POMC-cre;;IFT88^f/f mice). Scale bars: 100 μm. Data values are presented as mean ± SEM. Statistics were performed using two-sided Student’s t test (a–d, e—axon, f) and one-sided two-way ANOVA followed by post hoc LSD test (e—dendrite). **p < 0.01, ***p < 0.001 between the indicated groups.

Fig. 6. Lysosomal dysfunction in the POMC neurons with defective ciliogenesis.

a LC3B (autophagy marker), β-endorphin (β-END), and DAPI staining for autophagy analysis in the ARH of POMC-cre;;IFT88^f/f and IFT88^f/f neonates at P11 (n = 5 for saline, n = 4 for leupeptin each genotype). Scale bars: 20 μm. b p62 (autophagy substrate) and tdTomato costaining in the ARH of POMC-cre;;IFT88^f/f;;tdTomato and POMC-cre;;tdTomato neonates at P14 (n = 5). The graph depicts the average fluorescence intensity of p62 in 100 cells per mouse. Scale bars: 20 μm. c Double staining of DQ-BSA, indicative of lysosomal protein degradation, and LAMP1 in the ARH of C57 mice (n = 6). Magnified images of (i) ARH cell with DQ-BSA signals merged with lysosomes, (ii) cell with high-intensity lysosomal and extralysosomal DQ-BSA signals, and (iii) cell with lack of DQ-BSA signals. Scale bars: 10 μm. d Double staining of DQ-BSA and β-END in the ARH of POMC-cre;;IFT88^f/f mice and IFT88^f/f littermates (n = 5). The average DQ-BSA intensity values in 100 cells and the numbers of DQ-BSA⁺ cells in the ARH are presented. Scale bars: 20 μm. e LAMP1 and β-END double staining showing lysosomal mass and sizes in POMC and non-POMC ARH cells at P14 (n = 4 for POMC-cre;;IFT88^f/f mice, n = 5 for IFT88^f/f mice). The graph depicts the average values of LAMP1 intensity (lysosomal mass) and LAMP1⁺ puncta size (lysosomal size) in 100 cells per mouse. Scale bars: 20 μm. f Double immunostaining of DQ-BSA and LAMP1 in the ARH cells of C57 mice (n = 6). Magnified images of (i) ARH cell with larger lysosomes and lower DQ-BSA fluorescence intensity and (ii) cell with smaller lysosomes and higher DQ-BSA intensity. The graph depicts the correlation between the average lysosomal size and DQ-BSA intensity in ARH cells. Scale bar: 5 μm. Data values are presented as mean ± SEM. Statistics were performed using two-sided Student’s t test (a—autophagy flux, b, d, e), one-sided one-way ANOVA followed by post hoc LSD test (a—LC3 puncta), and linear regression analysis (f). ***p < 0.001 between groups. ns not significant.

Fig. 7. Interplay among cilia, lysosomal protein degradation, and axonal projection in the hypothalamic neurons.

a Confirmation of ciliary dysgenesis in N1 hypothalamic neuron cells expressing IFT88 small hairpin RNA (shIFT88)-GFP. ARL13B (cilia marker), GFP, and DAPI staining in N1 cells transfected with shIFT88-GFP-expressing AAV (adeno-associated virus) or GFP-AAV (n = 4 wells). The average cilia lengths and ciliated cell percentages of about 50 infected cells and 50 uninfected cells are presented in each treatment group. Scale bars: 20 μm. b DQ-BSA intensity in N1 hypothalamic neuron cells infected with either shIFT88-GFP-AAV or GFP-AAV (n = 5 wells). The average DQ-BSA intensity of 50 GFP⁺ cells per well was presented. Scale bars: 20 μm. c Immunostaining of lysosomes using LysoTracker dye in N1 hypothalamic neuron cells (n = 5 wells). The average values of about 50 infected cells per well are presented. Scale bar: 10 μm. d Fluorescent-labeled albumin (Fl-albumin), GFP, and DAPI staining in N1 hypothalamic neuron cells transfected with shIFT88-GFP-AAV or GFP-AAV (n = 5 wells). The average numbers of Fl-albumin⁺ puncta in 50 GFP-expressing cells were measured per well. Scale bars: 20 μm. e β-END immunostaining in the hypothalamus of C57 neonates injected with saline or leupeptin during P8–P13 (n = 4). Scale bars: 100 μm. f POMC (β-END) and cilia (AC3) double immunostaining in neonates with saline or leupeptin injections (n = 4). Scale bars: 100 μm. Data are presented as mean ± SEM. Statistics were performed using two-sided Student’s t test (a–f). **p < 0.01 and ***p < 0.001 between groups. ns not significant.

Fig. 8. The neonatal leptin surge stimulates ciliogenesis and neural circuit formation in the hypothalamus.

a Cilia (AC3) staining in the hypothalamic ARH and hippocampal dentate gyrus of C57 neonates at P14 that received either saline or leptin antagonist SHLA injections from P4 to P13. The average lengths of 100 cilia and the ciliated cell percentage in each area per mouse are presented (n = 5 for hypothalamus, n = 4 for hippocampus). Scale bars: 20 μm. b Cilia staining in the hypothalamus (Hypo) and hippocampus (Hippo) of ob/+ mice, ob/ob mice, and ob/ob mice with leptin treatment (P10–P14) (n = 5). The average lengths of 100 cilia in each area per mouse are presented. Scale bars: 20 μm. c Upper panel: cilia (AC3) and POMC (β-END) costaining in the hypothalamic ARH in 7-week-old ob/+ mice, ob/ob mice, and ob/ob mice with adulthood leptin replacement (10 mg/kg/day for 7 days; n = 5). Lower panel: axonal projections of POMC neurons in the three groups (n = 4). Scale bars: 50 μm. d Ciliogenesis in the ARH and POMC axonal projection to the PVH in ob/ob neonates injected with saline, leptin alone, or leptin + shIFT88-GFP-AAV (n = 4 for AAV on-target group, n = 5 for the other 3 groups). The mice with off-target AAV injection were used as a control. Scale bars: 50 μm. e AC3 (cilia), functional leptin receptor (LepRb), GFP (AAV transfection), and DAPI costaining in the ARH of ob/ob neonates injected with saline, leptin, or leptin + shIFT88-GFP-AAV. The percentages of ARH cells and AAV-infected cells with periciliary leptin receptor accumulation were counted (n = 5 for the ob/ob-saline group and n = 4 for the other 2 groups). Scale bars: 10 μm. Data values are presented as mean ± SEM. Statistics were performed using two-sided Student’s t test (a) and one-sided one-way ANOVA (b–e) followed by post hoc LSD test. **p < 0.01, ***p < 0.001 between the indicated groups. ns not significant.

Fig. 9. Leptin stimulates lysosomal protein degradation in the hypothalamic neurons via cilia-dependent mechanism.

a Representative images of DQ-BSA/β-END staining and LAMP1/DAPI staining in the hypothalamic ARH of ob/+ and ob/ob mice at P14 (n = 5). The graphs depict the average values of DQ-BSA fluorescence intensity (lysosomal protein degradation), LAMP1 intensity (lysosomal mass) and LAMP1⁺ puncta size (lysosomal size) in 100 ARH cells per mouse. Scale bars: 50 μm. 3V third ventricle. b DQ-BSA/β-END and LAMP1/DAPI staining in the ARH of ob/ob mice injected with saline, leptin alone, or leptin + shIFT88-GFP-AAV (n = 5). The average values of DQ-BSA and LAMP1 intensity and puncta size in 100 ARH cells per mice are presented. Scale bars: 50 μm. c DQ-BSA fluorescence intensity in N1 cells transfected with GFP-AAV or shIFT88-GFP-AAV and treated with leptin (100 nM for 1 h) (n = 4 wells for GFP-AAV, n = 5 wells for shIFT88-GFP-AAV). The DQ-BSA fluorescence intensity of 50 GFP⁺ cells per well was analysed. Scale bars: 50 μm. d TFEB/DAPI double staining, as a measure of lysosomal biogenesis, in the ARH of ob/+ mice with saline and ob/ob mice with saline or leptin treatment (n = 4 for ob/+ saline, n = 5 for ob/ob saline, and n = 6 for ob/ob leptin). Asterisks indicate cells with nuclear TFEB expression. Scale bars: 10 μm. e Lysosomal staining using LysoTracker and pH-sensitive LysoSensor dyes in N1 cells with or without leptin treatment (n = 4 wells). The average fluorescent intensity of 100 cells per well was analysed. Scale bars: 50 μm. f The vacuolar H⁺-ATPase (v-ATPase) activity in N1 cells treated with leptin alone or leptin with lysosomal v-ATPase inhibitor bafilomycin (n = 4 wells). g The v-ATPase activity in N1 cells transfected with either control small inhibitory RNA (siRNA) or IFT88 siRNA and then treated with leptin (n = 4 wells). Data values are presented as mean ± SEM. Statistics were performed using two-sided Student’s t test (a, e, g) and one-sided one-way ANOVA (b–d, f) followed by post hoc LSD test. *p < 0.05 and ***p < 0.001 between the indicated groups. ns not significant.

Fig. 10. Maternal malnutrition disrupts the leptin regulation of hypothalamic cilia–lysosome–neuronal circuit formation in neonatal mice.

a Hypothalamic cilia images in the neonatal offspring (at P14) of dams fed a high-fat diet (HFD) or low protein diet (LPD) during gestation and lactation or during lactation (n = 5 for M-CD, n = 4 for M-HFD and M-LPD). The average lengths of 100 ARH cilia per mouse are presented. Scale bars: 50 μm. b Changes in body weights, fat and lean mass, cumulated food intake, and energy expenditure (at 4 weeks) during the post-weaning period in the offspring of CD- or LPD-fed dams (n = 4 for food intake monitoring, n = 8 for other measurement). c DQ-BSA/β-END and LAMP1/DAPI double staining in the hypothalamic ARH of offspring of dams on a HFD or LPD (n = 5). The average values of 100 ARH cells per mouse are presented. Scale bars: 50 μm. d Plasma leptin concentrations in the offspring of CD-, HFD-, or LPD-fed dams during lactation at P10 (n = 7). e Hypothalamic cilia staining (AC3), lysosomal protein degradation (DQ-BSA/β-END staining) in the ARH, and POMC axonal projection in the PVH in the pups nourished by dams on a HFD or LPD and received either saline or leptin injections during P4–P13 (n = 5 for DQ-BSA study and n = 6 for all other analyses). Scale bars: 20 and 100 μm (axonal projections). f Hypothalamic leptin receptor (Leprb) mRNA expression in the offspring of CD-, LPD-, or HFD-fed dams at P14 (n = 6 for M-CD, n = 5 for the other 2 groups). Data values are presented as mean ± SEM. Statistics were performed using one-sided one-way ANOVA (a, c, d, f) and one-sided two-way ANOVA (b) followed by post hoc LSD test and two-sided Student’s t test (b—energy expenditure, e). *p < 0.05, **p < 0.01, and ***p < 0.001 between the indicated groups. ns not significant.